Join ISPOR  | Sign up for mailing List  | Follow Us: LinkedIn Twitter Facebook YouTube
 


ARTICLE SUPPLEMENTARY INFORMATION
(Supporting information is provided by the author to support the content
of the parent article)


Modeling complex treatment strategies: construction and validation of a discrete event simulation model for glaucoma

Aukje van Gestel, Johan L. Severens, Carroll A. B. Webers, Henny J. M. Beckers, Nomdo M. Jansonius, Jan S. A. G. Schouten

Value in Health Supplemental Information


Appendix

 

 

Modeling complex treatment strategies: construction and validation of a discrete event simulation model for glaucoma

 

Appendix

 

 

This appendix has been compiled to provide supporting information for the manuscript “Modeling complex treatment strategies: construction and validation of a discrete event simulation model for glaucoma”, published in Value in Health, 2010.

 

Aukje van Gestel, MSc.1

Johan L. Severens, Professor, PhD2,3 Carroll A. B. Webers, MD, PhD1 Henny J. M. Beckers, MD, PhD1 Nomdo M. Jansonius, MD, PhD4

 

Jan S. A. G. Schouten, MD, PhD1

 

1University Eye Clinic, Maastricht University Medical Center

 

2Department of Clinical Epidemiology and MTA, Maastricht University Medical Center

 

3Department of Health Organisation, Policy, and Economics, Maastricht University Maastricht, The Netherlands

 

4 Department of Ophthalmology, University Medical Center Groningen

Groningen, The Netherlands

 

Contact:

Mrs. Aukje van Gestel, scientific researcher,

University Eye Clinic, Maastricht University Medical Center

P.O. Box 5800, 6202 AZ, Maastricht, The Netherlands.

Telephone

31 43 3875346

Fax

 

31 43 3875343

E-mail

 

a.van.gestel@mumc.nl

 

Dr. Jan Schouten, clinical epidemiologist, ophthalmologist.

 

University Eye Clinic, Maastricht University Medical Center

P.O. Box 5800, 6202 AZ, Maastricht, The Netherlands.

Telephone

31 43 3875346

Fax

31 43 3875343

E-mail

j.schouten@mumc.nl

 

Contents

 

----------

 

 

1.

Introduction.............................................................................................................

6

2.

Discrete Event Simulation model for glaucoma ......................................................

7

 

2.1

| Conceptualization of glaucoma disease progression.....................................

7

 

2.2

| Event-based time progression ..........................................................................

8

 

2.3

| Summary of parameter estimates .....................................................................

8

3.

Treatment schedule..............................................................................................

11

4.

Visit schedule .......................................................................................................

14

5.

Drug effectiveness................................................................................................

15

 

5.1

| Pre-surgery medication....................................................................................

15

 

 

5.1.1 | Monotherapy .....................................................................................................

15

 

 

5.1.2 | Combination therapy: first addition ...................................................................

18

 

 

5.1.3 | Combination therapy: second addition .............................................................

20

 

5.2

| Post-surgery medication..................................................................................

21

 

5.3

| LT pressure lowering effect.............................................................................

22

 

5.4

| Prevalence of Timolol contraindications........................................................

24

 

5.5

| Side-effects with medication ...........................................................................

25

6.

Surgery effectiveness ...........................................................................................

26

 

6.1

| Intra-ocular pressure after trabeculectomy ...................................................

26

 

6.2

| Intra-ocular pressure after Baerveldt implantation .......................................

28

7.

Conversion risk.....................................................................................................

30

 

7.1

| Baseline risk......................................................................................................

30

 

7.2

| Relative risks.....................................................................................................

30

 

 

7.2.1 | Relative risk of intra-ocular pressure ................................................................

30

 

 

7.2.2 | Relative risk of age ...........................................................................................

31

 

 

7.2.3 | Other prognostic factors ...................................................................................

31

 

7.3

| Equations ..........................................................................................................

33

8.

Progression ..........................................................................................................

34

 

8.1

| Systematic review of glaucomatous progression (and rate of progression)

 

....................................................................................................................................

 

34

 

8.2

| Baseline rate of progression ...........................................................................

42

 

8.3

| Relative risk of intra-ocular pressure .............................................................

46

 

 

 

8.4 | Equations ..........................................................................................................

47

 

8.5 | Criterion for progression in the model ...........................................................

47

9. Cataract

................................................................................................................

49

 

9.1 | Baseline risk of cataract formation.................................................................

49

 

9.2 | Relative risk of trabeculectomy for cataract formation ................................

50

 

9.3 | Cataract extraction ...........................................................................................

50

10.

Utility outcomes ..................................................................................................

51

 

10.1

| Visual Functioning Questionnaire ................................................................

51

 

10.2

| Health Utilities index ......................................................................................

52

 

10.3

| EQ-5D utility ....................................................................................................

52

11.

Survival...............................................................................................................

53

12.

Average OHT/POAG population.........................................................................

55

 

12.1

| Ocular hypertension population ...................................................................

55

 

 

12.1.1 | Age..................................................................................................................

55

 

 

12.1.2 | Gender ............................................................................................................

55

 

 

12.1.3 | Baseline IOP ...................................................................................................

55

 

 

12.1.4 | MD after conversion........................................................................................

55

 

12.2

| Primary open-angle glaucoma population ...................................................

56

 

 

12.2.1 | Age..................................................................................................................

56

 

 

12.2.2 | Gender ............................................................................................................

56

 

 

12.2.3 | Baseline IOP ...................................................................................................

56

 

 

12.2.4 | MD at baseline ................................................................................................

56

 

 

12.2.5 | Response to trabeculectomy ..........................................................................

57

13.

Costs ..................................................................................................................

 

59

14.

Medication Costs ................................................................................................

60

15.

Ophthalmologist, procedures and interventions Costs .......................................

61

 

15.1

| Ophthalmologist visit .....................................................................................

61

 

15.2

| Visual field measurement ..............................................................................

62

 

15.3

| Laser trabeculoplasty (LT).............................................................................

62

 

15.4

| Trabeculectomy ..............................................................................................

63

 

15.5

| Re-trabeculectomy .........................................................................................

63

 

15.6

| Baerveldt implant............................................................................................

63

 

15.7

| Cataract extraction .........................................................................................

64

16.

Costs of low-vision rehabilitation services ..........................................................

65

 

 

 

 

16.1

| Resource utilization........................................................................................

65

 

16.2

| Cost prices ......................................................................................................

66

17.

Costs of low-vision aids ......................................................................................

67

 

17.1

| Resource utilization........................................................................................

67

 

17.2

| Cost prices ......................................................................................................

68

18.

Costs of homecare, grooming and nursing .........................................................

70

 

18.1

| Resource utilization, nursing home..............................................................

70

 

18.2

| Resource utilization homecare......................................................................

71

 

18.3

| Cost prices ......................................................................................................

72

19.

Costs of transportation........................................................................................

73

 

19.1

| Resource utilization........................................................................................

73

 

19.2

| Cost prices per unit ........................................................................................

74

 

19.3

| Total costs of transportation .........................................................................

74

20.

Costs of informal care.........................................................................................

76

 

20.1

| Resource utilization........................................................................................

76

 

20.2

| Cost prices ......................................................................................................

76

21.

Costs of productivity loss ....................................................................................

78

 

21.1

| Resource utilization........................................................................................

78

 

21.2

| Cost prices ......................................................................................................

78

22.

Summary of MD-related costs ............................................................................

79

23.

Abbreviations......................................................................................................

81

24.

References .........................................................................................................

82

 

1. Introduction

 

----------

 

In the manuscript “Modeling complex treatment strategies: construction and validation of a discrete event simulation model for glaucoma” we have presented the basic structure of the health economic model for ocular hypertension and primary open -angle glaucoma. In this appendix we present the sources and methods of the derivations of the most important structural relationships and the sources, best estimates and distributions of the main parameter estimates in the base case model, as well as graphical presentations of several aspects of the model design. Only distributions pertaining to patient variability and heterogeneity are described here (first order uncertainty). The distributions that were created to represent uncertainty in the estimates of population means (second order uncertainty) are not discussed here. Future reports that present the outcomes of probabilistic sensitivity analyses will be accompanied by a description of the distributions used to represent second order uncertainty.

 

2. Discrete Event Simulation model for glaucoma

 

----------

 

 

2.1 | Conceptualization of glaucoma disease progression

 

We have conceptualized glaucoma and its treatment from a clinical perspective. This means that we have not necessarily simulated the actual pathogenetic processes themselves, but rather how they manifest themselves in clinical practice. An elevated intra-ocular pressure (IOP) is the most important known risk-factor for primary open- angle glaucoma. As long as the IOP is elevated without signs of retinal nerve fiber loss, the condition is termed ocular hypertension (OHT). However, when nerve fiber loss occurs to a level that causes optic nerve cupping and/or visual field loss, the condition is termed primary open -angle glaucoma (POAG). The transition from OHT to POAG is termed ‘conversion’. If nerve fiber loss continues (progression), the visual field deteriorates and a patient may progress to blindness. Lowering the IOP by treatment reduces both conversion and progression.[1, 2] This information has been translated into the model as shown in Figure 1.

 

 

 

 

OHT and POAG represent two distinct disease states. Conversion is modeled as an event upon which the disease state changes from OHT to POAG. Visual field damage is a proxy for glaucoma severity and is expressed as Mean Deviation ranging from 0 (no damage) to -35 (severe damage) decibel (dB) .[3] Below a certain MD threshold, patients are considered blind. Progression is modeled by means of an intrinsic rate at which the visual field decreases annually. The effect of treatment is that it lowers IOP, which in turn affects the conversion risk and the intrinsic progression rate of the simulated patient.

 

2.2 | Event-based time progression

 

Events in a discrete event simulation (DES) model represent relevant moments in time. At an event the attributes of the entity are reevaluated and (if needs be) adjusted (Figure 2). In our model, time -progression is event-based, which means that the model ‘jumps’ from one event to the next. The timing of future events may be conditional upon the new values of the attributes.

 

 

 

2.3 | Summary of parameter estimates

 

In later paragraphs of this appendix, the derivation of the parameter estimates used in the base case setting of the model will be addressed. Here we present a summary of the parameter estimates.

 

 

Table 1: Parameter estimates

Parameter

Best estimate

Distribution

Source

Hazard rate for conversion

0.02/year

n.a.

[4]

Hazard ratio age for conversion

1.26 per decade older than 55 years.

n.a.

[5]

Hazard ratio IOP for conversion

1.09 per mmHg higher than 24 mmHg.

n.a.

[5]

MDR

-0.34 dB/year

Gamma

[6-10]

Relative Risk IOP for MDR

1.13 per mmHg higher than 15.5 mmHg

n.a.

[10]

Medication Mean effect / Incidence side-effectsa)

 

 

Β-blocker

26 % / 8%

Beta

[11, 12]

Prostaglandin analogue

29.5 % / 8%

Beta

[11, 12]

Carbonic-anhydrase inhibitor

19.5 % / 14%

Beta

[11, 12]

α2-adrenergic agonist

21 % / 23%

Beta

[11, 12]

Mean effect LT

34 %

Beta

[10, 13-17]

Mean IOP after surgery (TE)

12.5 mmHg

Gamma

[18-22]

Mean IOP after surgery (tube)

15.0 mmHg

Gamma

[19, 22, 23]

IOP = Intraocular Pressure, MDR = Mean Deviation Rate, LT = Laser Trabeculoplasty, TE = Trabeculectomy.

a) Side-effects that lead to a treatment switch.

 

 


Table 2: Costs associated with attributes and events in the simulation model

Resource

Costs

Source

Β-blocker

€ 6.00/month

[24, 25]

Prostaglandin analogue

€ 20.20/month

[24, 25]

Carbonic-anhydrase inhibitor

€ 13.90/ month

[24, 25]

α2-adrenergic agonist

€ 14.00/month

[24, 25]

Ophthalmologist consultation

€ 65

[26, 27]

Visual field measurement

€ 133 (€ 266 in case of progression)

[26, 27]

LT

€ 75

[27, 28]

Trabeculectomy

€ 1,214 (+ 1 ophthalmologist consultation)

[26, 27]

Implant surgery

€ 1,714 (+ 1 ophthalmologist consultation)

[26, 27]

Cataract surgery

€ 1,400

[26], HA

Paid household help

€ 37 / month (if MD < -10 dB)

[26, 29]

Homecare nursing

€ 159 / month (if MD < -10 dB)

[29];[26]

Family help

€ 56 / month (if MD < -15 dB)

[26, 29]

Homecare grooming

€ 103 / month (if MD < -15 dB)

[26, 29]

Retirement home

€ 80 / month (if MD < -20 dB)

[26, 29]

Nursing home

€ 130 / month (if MD < -20 dB)

[26, 29]

Informal care

€ 20 / month (if MD < -5dB)

[26, 29]

Low-vision services

€ 1-5 /month

[29, 30]

Transport to ophthalmologist

€ 4.90 / visit (if MD > -10 dB)

€ 8.90 / visit (if MD < -10 dB)

[26, 29]

Transport to pharmacy

€ 1.50 / visit (if MD > -10 dB)

€ 2.60 / visit (if MD < -10 dB)

[26, 29]

Low-vision aids

€ 325 (once) if MD moves below -15 dB

[29, 31]

Productivity loss

€ 3,029 (once) if MD moves below -15 dB

[26, 29]

Costs for LT (Laser Trabeculoplasty) and surgery are doubled to account for the same procedure in the other (i.e. worse) eye. Transport costs to the pharmacy are incurred once in three months if the patient receives medication, and transport costs to the ophthalmologist/hospital are incurred for each visit and for each procedure (LT, surgery). HA = Hospital Administration.

 

3. Treatment schedule

 

----------

 

The choice for the various treatment options in the model is made based on the two flow-charts presented in Figure 3 (between treatment types) and Figure 4 (within the medication blocks shown in Figure 3).

 



I

Figure 3: Intervention for OHT and POAG in the model; the order of treatment types. Reasons to change treatment are A) side-effects, B) insufficient effectiveness and C) IOP above the target IOP.

 

The main ‘route’ through the various treatment types are shown in Figure 3 by the black arrows, but there are several detours built into the schedule as well (grey arrows):

 

LT is skipped if a patient has received cataract surgery in the past (attribute).

 

Surgery (i.e. trabeculectomy and implant surgery) is skipped if a patient is too old. A second trabeculectomy is not performed if there was immediate failure of the first

 

trabeculectomy.

 

OHT patients are only treated with medication block 1 and/or laser treatment. They can never move to trabeculectomy or medication block 3.

 

Trabeculectomy is not performed if no visual field progression has been observed. If trabeculectomy is indicated due to an IOP that is higher than the target IOP, but progression has not been observed (either because no visual field measurement has been performed, or because the visual field measurement did not indicate progression), the medication the patient was previously taking is continued until progression is observed.

 

Detours are also possible in the medication flowchart (Figure 4). If a patient suffers from side-effects or low effectiveness on the current medication, the model finds the next medication by moving one step downward in the flowchart. However, if that next medication is contraindicated (fixed attribute) or has given rise to side-effects in the past (attribute), the model makes another step downward. If the current medication has good effectiveness and does not give side-effects, but the resulting IOP is nonetheless higher than the target IOP, the model make one step rightward.

 

 

In the default model the order of monotherapies is timolol (Mono 1), latanoprost (Mono 2), dorzolamide (Mono 3) and brimonidine (Mono 4).

 

If a patient moves to LT by a rightward step, all medication is continued. If a patient moves to LT by a downward step, all medication is stopped. However, if in the latter case the patient does not reach the target pressure three months after LT, medication is added again. The model chooses the last medication not causing side-effects the patient received before the LT.

 

The definition of the target IOP constitutes a part of the treatment strategy. Before the analyses, the target pressures used in the model can be defined by the model user. A target pressure can be entered for four different situations (Table 3).

 

Table 3: Example of a look-up table for target IOP depending on disease status and the occurrences of disease progression.

Disease status

IOPtarget

OHT

24 mmHg

POAG, without observed progression

21 mmHg

POAG, and one observed progression

18 mmHg

POAG, and two or more observed progressions

15 mmHg

IOPtarget = target Intraocular Pressure, OHT = Ocular Hypertension, POAG = primary open-angle glaucoma


4. Visit schedule

 

The frequency of visits to the ophthalmologist in the model is part of a treatment strategy and is therefore adjustable by the user. In the default model the schedule presented in Table 4 is used. The length of the time interval between two visits depends on two factors: 1) whether or not there has been a treatment change, and 2) the type of new treatment. The number of visits since that last treatment change is counted in the leftmost column of Table 4, while the new treatments are listed in the top row. For example, a patient that is not treated at all will visit the ophthalmologist every 36 months (3 years). The first visit after a change in medication will take place 3 months after the change, but the next visits will occur every 6 months as long as the treatment remains unaltered. After LT or surgery a series of short visit intervals follows to be able to monitor the patient closely. The visit frequency gradually returns back to the normal interval length.

 

Table 4: Periods between visits in base case model (months)

Visit number

No treatment

Medication

LT

Surgery

1

36

3

0.23

0.1

2

36

6

1.15

0.1

3

36

6

6

0.1

4

36

6

6

0.1

5

36

6

6

0.23

6

36

6

6

0.23

7

36

6

6

0.23

8

36

6

6

0.5

9

36

6

6

0.5

10

36

6

6

1

> 10

36

6

6

6

5. Drug effectiveness

 

----------

 

 

5.1 | Pre-surgery medication

 

There is a wide variety of pressure-lowering eye-drops for the treatment of elevated intra-ocular pressure. Four classes of pharmaceuticals are commonly used nowadays: beta-adrenergic antagonists, prostaglandin analogues (or hypotensive lipids), carbonic anhydrase inhibitors and alpha-2 adrenergic agonists. In most cases several analogues exist within each of these classes. The pressure- lowering eye drops can be applied individually (as monotherapy) or in combination with each other (combination therapy). Oral pressure-lowering medication is also in use (e.g. acetazolamide), but since this medication is often only used temporarily, it was not included in the present model.

 

The model offers the possibility to define four monotherapies, one within each class, which will be used throughout the treatment strategy. In the default scenario the models uses a representative medication from each class. These are timolol, latanoprost, dorzolamide and brimonidine respectively. The effectiveness of each medication is expressed as the pressure lowering relative to the intra-ocular pressure before treatment was started. From literature it is known that the relative pressure -lowering effect of medication that is added to existing medication is lower than the effect if the same medication is applied as a monotherapy.

 

In order to account for this variation in effectiveness depending on the existing treatment, the effectiveness of medication is estimated for three separate situations:

 

  1. medication is applied as a monotherapy
  2. medication is added to one other medication
  3. medication is added to two or more other medications.

 

5.1.1 | Monotherapy

 

Default estimates for drug effectiveness as monotherapy were derived from a meta-analysis of ‘all commonly used glaucoma drugs’ in 2005.[11]This meta-analysis included studies that compared pressure-lowering eye-drops monotherapy to placebo in POAG and/or OHT patients, and that used IOP as the primary endpoint of the study.

 

An excerpt of the results of the meta-analysis is reported in Table 5.

 

 

Table 5: meta-analysis of pressure lowering drug effects [11]

 

Absolute change (mmHg)

(95% confidence limits)

Relative change (%)

(95% confidence limits)

Timolol, trough

-6.9 (-7.4; -6.5)

-26 (-28; -25)

Timolol peak

-6.9 (-7.5; -6.3)

-27 (-29; -25)

Latanoprost, trough

-6.8 (-7.6; -6.1)

-28 (-30; -26)

Latanoprost, peak

-7.9 (-8.3; -7.4)

-31 (-33; -29)

Dorzolamide, trough

-4.5 (-5.0; -4.0)

-17 (-19; -15)

Dorzolamide, peak

-5.9 (-6.5; -5.2)

-22 (-24; -20)

Brimonidine, trough

-4.5 (-5.2; -3.8)

-18 (-21; -14)

Brimonidine, peak

-6.1 (-6.7; -5.4)

-25 (-28; -22)

 

In the model, the average of the reported effectiveness for trough and peak was used as an estimate of the average drug effectiveness.

 

In order to estimate a distinct effectiveness of MONO1 through MONO4 in individual patients in the model, random distributions were used. However, there is no information in the literature regarding the distribution of medication effectiveness in OHT or POAG population samples. Only the mean (and sometimes the standard deviation or the standard error of the mean) is reported. A beta distribution was used to describe the medication effectiveness, because the beta distribution has the characteristics that it is limited to values between 0 and 1 (or in this case, 0 and 100% pressure lowering).

 

The beta distribution is defined by two parameters, alpha and beta. Using the method of moments, alpha and beta could be estimated from observed means and variances by the formulas in Table 6. Unfortunately, most articles reporting the results of RCT’s for drug effects do not report the variances in relative pressure reductions. From the reported variances, standard deviations and or standard errors of the mean surrounding IOP’s and absolute pressure lowering in the individual studies used in the meta-analysis, the variance of the relative pressure lowering effect of monotherapies was estimated at around 1-1.7%. [11]

 

 

With a variance of 1% (s2) and a mean effect of 27% (µ), the method of moments would lead to an estimate of α = 5 and β = 14, and a distribution of the timolol effectiveness in the OHT/POAG population as presented in Figure 5. In this distribution, 26% of the population has a relative effectiveness lower than 20%, 11% has a relative effectiveness higher than 40% and 2% has a relative effectiveness higher than 50%.

.

 

 

 

Figure 5: Beta distribution for timolol effectiveness in simulated population,

alpha = 5, beta = 14.

 

According to clinical experts (CW, HB, JS), the probability of an effectiveness higher than 40% with timolol are smaller than the 11% that is the result of this theoretical beta distribution, which means that the proposed distribution is too wide. This may be due to the fact that the effectiveness in literature is reported with parameters assuming a normal distribution (µ and s2), whereas the distribution of effectiveness may deviate from normal. In the absence of any other information, the theoretical distribution was fine-tuned to the experts’ expectations. The estimates for alpha and beta values are presented in Table 7 and the distributions are drawn in Figure 6.

 

 

Table 7: default estimates of medication effectiveness in the model (Beta distribution)

 

Average

Standard deviation

Estimated Alpha

Estimated Beta

% below 20%

% above 40%

Timolol

27%

8%

8

22

20%

6%

Latanoprost

29.5%

8%

9

22

11%

10%

Dorzolamide

19.5%

8%

5

19

57%

1%

Brimonidine

21%

8%

5

20

49%

2%

 

 

 

5.1.2 | Combination therapy: first addition

 

The relative pressure-lowering effectiveness of the medications when added as the second drug to previously initiated medication, was estimated based on empirical research in the University Eye Clinic Maastricht (DURING study).[11] This research prospectively included patients with ocular hypertension (OHT) and glaucoma that were treated with pressure lowering medication. The initiation of therapy and any change in therapy were registered either prospectively or retrospectively from the medical files. The resulting change in intra-ocular pressure was calculated from the intra-ocular pressure measurements before and after the adjustment in therapy. The results for the four representative medications are presented in Table 8.

 

 

Table 8: additive effectiveness observed in the DURING study

 

N

Average IOP before addition (mmHg)

Pressure lowering

(95% CI)

SE

(derived)

SD

(derived)

Timolol

32

25.5

14.1% (7.6; 20.6)

3.3%

19%

Latanoprost

127

23.2

17.7% (14.7; 20.7)

1.5%

17%

Dorzolamide

20

21.6

8.8% (2.0; 15.6)

3.5%

16%

Brimonidine

39

22.9

16.6% (12.2; 21.0)

2.2%

14%

 

In a recent large systematic review Webers et al. investigated the pressure-lowering effectiveness of second-line glaucoma medication.[33] The authors included studies that investigated the additional pressure lowering effectiveness of adding a medication to further lower intra- ocular pressure. They distinguished the additional effectiveness at peak, trough and on the day-curve. A summary of the results is presented in Table 9. This overview provided a large amount of information regarding the pressure lowering of added medication, but as the authors of the article state, the results from these studies may be biased towards higher effectiveness estimates. In the clinical trials included in the systematic review, the study designs included a run-in phase with the initial medication and no further selection of the patients based on their response to the run-in medication. In clinical practice (and also in the disease progression model), a second medication is only added if the target pressure is

not achieved despite a sufficient response to the initial medication (i.e. more than 20% pressure lowering). If a proportion of the patients in the studies included in the systematic review were in fact non-responders to the run-in medication, the treatment effect that was measured may have been a combination of both initial and additive effectiveness. And since initial effectiveness can be expected to be higher, the treatment effects in the systematic review may have been overestimated.

 

 

Table 9: Summarized results of a systematic literature review of the additional pressure lowering-effectiveness of second line glaucoma medications.[32]

 

Number of included studies

Minimal additional effectiveness found in included studies (%)

Maximal additional effectiveness  found in included studies (%)

Timolol

 

 

 

Trough

-

-

-

Peak

2

19.2 ± 14.1

20.2 ± 7.5

Day-curve

2

10.6 ± 7.6

15.0 ± 11.3

Latanoprost

 

 

 

Trough

3

15.7 ± 10.5

26.0 ± 12.0

Peak

4

17.0 ± nr

24.6 ± 14.1

Day-curve

8

12.0 ± 14.7

29.9 ± 10.2

Dorzolamide

 

 

 

Trough

17

11.3 ± 12.5

23.1 ± 8.3

Peak

18

10.2 ± nr

23.1 ± 12.6

Day-curve

8

11.6 ± 7.7

26.9 ± 12.3

Brimonidine

 

 

 

Trough

7

7.3 ± 12.5

19.7 ± nr

Peak

11

13.4 ± 9.1

27.6 ± nr

Day-curve

1

12.5 ± 11.4

-

 

Non-responder bias was not an issue in the DURING study because the data were drawn from clinical practice. It is unlikely that a medication would have been added to an ineffective medication. For that reason the estimates of the additive effectiveness of medication in the model were based on the results in de DURING study, and the hierarchy in effectiveness in the monotherapies was maintained (Table 7) . The estimated effectiveness of brimonidine from the DURING study (16.6%) was slightly adjusted downward to the effectiveness of Timolol (14%), because as a monotherapy timolol is more efficacious than brimonidine.[11]

 

The distributions of the effectiveness of added medication in a heterogeneous OHT/POAG population were simulated with a beta distribution. The distribution parameters alpha and beta were based on the means and variances. The standard deviation in second -line effectiveness was based on the results of the DURING study and the systematic review, and fine-tuned to 8% in order to obtain a distribution that met the experts’ expectations. The resulting parameters and distributions are presented in Table 10 and Figure 7.

 


Table 10: Default estimates for the relative effectiveness of medication if added to one other medication (Beta distribution).

 

Average

Standard deviation

Alpha

Beta

% below 20%

% above 40%

Timolol

14%

8%

2

15

79

1

Latanoprost

18%

8%

4

18

64

1

Dorzolamide

9%

8%

1

11

90

0

Brimonidine

14%

8%

2

15

79

1

 

 

 

 

5.1.3 | Combination therapy: second addition

 

If two medications still do not suffice to reach the target pressure, a third medication may be added to the therapy. This only applies to the monotherapies that are the third and the fourth option in the treatment strategy (Figure 4).In the default situation the order of monotherapy options is timolol, latanoprost, dorzolamide and brimonidine. Therefore, in the default situation only the effectiveness of dorzolamide and brimonidine as the third medication needs to be estimated. The literature on the effectiveness of the second added medication is very scarce and there are no formal clinical trials that have investigated it. To make estimates for the model the proportional effectiveness of the second-line medications was calculated relative to the monotherapies, and this factor was applied to the effectiveness of the second - line medications. For example, the estimated effectiveness of dorzolamide as a second line medication was 9%, and as monotherapy 20%. The factor 0.09/0.20 was applied to 0.09, and the resulting estimate is 4%. Similarly the resulting estimate for the effectiveness of brimonidine if added to two other medications is 9%. The estimated standard deviation is 4%. These estimates are presented in Table 11.

 

Table 11: Default estimates for the relative effectiveness of medication if added to two or more other medications (Beta distribution).

 

Average

Standard deviation

Alpha

Beta

Dorzolamide

4%

4%

1

22

Brimonidine

9%

4%

5

46

 

 

5.2 | Post-surgery medication
If patients undergo surgery resulting in a functional filter or tube, the relative pressure lowering effectiveness of medications no longer applies. There is no quantitative information from literature about the effectiveness of medication after successful surgery, but expert opinion is that the effectiveness is much lower than in eyes without previous surgery. Also, the differences between drugs (compounds) are less pronounced. Based on expert opinion (CW, HB, JS) the average pressure lowering effectiveness of all medication in eyes with functional filters or tubes was estimated at 2 ± 0.5 mmHg. A normal distribution truncated at 0 mmHg was used for the OHT/POAG population distribution. The draw from the distribution however is based on the same random number as the draw from the beta distributions for the effectiveness before surgery. This ensures that the effectiveness of the medications before and after surgery are correlated (r = 1).

 

The truncated normal distribution of the post-surgery effectiveness of all medications is shown in Figure 8.

 

In combination therapy after surgery, the absolute pressure lowering effect of each added monotherapy to the combination is adjusted in order to prevent that the effectiveness of the combination therapy in the model becomes too high. The factor for adjustment is based on the relationship between the drugs’ estimated effectiveness in monotherapy, first addition and second addition (Table 7, Table 10).

 

The correction for the first added medication is 0.5, for the second added medication 0.25 and for the third added medication 0.1. For example, the average absolute pressure lowering effectiveness of timolol (2 mmHg) and latanoprost (2 mmHg) is 2 + 0.5*2=3 mmHg.

 

If a simulated patient receives surgery, the effectiveness of all subsequent medication will be considered in terms of an absolute pressure lowering as long as the filter or tube is functioning. If the latter is no longer the case, e.g. after failed trabeculectomy, the effectiveness of any subsequent medication will be considered in terms of the relative pressure lowering effect again (as in the pre-surgery situation).

 

 

5.3 | LT pressure lowering effect

 

The literature does not provide a systematic review of the pressure lowering effectiveness of laser trabeculoplasty in patients with OHT or primary open-angle glaucoma (POAG). Since October 2007 there is a systematic review for ‘Laser trabeculoplasty for open angle-glaucoma’ in the Cochrane library [34], but this has not investigated the pressure lowering efficacy of laser treatment. The committee on Ophthalmic Procedure Assessments of the American Academy of Ophthalmology has issued an Ophthalmic Procedure Assessment regarding ‘Laser trabeculoplasty for primary open-angle glaucoma’ in 1996. In this report the committee states that “The ocular hypotensive effect of laser trabeculoplasty is usually apparent within 1 month after treatment. Most studies have shown an initial reduction in intraocular pressure of approximately 20% to 30% or 6 to 9 mmHg.”[35]

 

We have conducted a literature review for articles reporting the pressure lowering effect of laser trabeculoplasty in OHT of POAG patients. Many of the identified articles did not report the pressure lowering effect of the procedure, but rather used the success rate as the primary (and only) outcomes measure, defined as the proportion of patient with a pressure under a certain threshold value. These articles were excluded because they did not provide information that could be translated into model input. The included articles and their results are presented in Table 12.

 

In previously untreated patients a pressure reduction of 31-35% was seen, in medically treated patients the reduction was 13-19%. In the model, LTP is either applied as a ‘monotherapy’ or it can be added to the medication to further lower intra-ocular pressure. Based on the review results we have estimated an average effectiveness for LTP in monotherapy of 34%, and an average effectiveness for LTP when added to medications of 16%. As far as the distribution of LTP effectiveness in the OHT/POAG population is concerned, there is less information from literature. Only one of the reviewed studies gave a standard deviation of the effectiveness, but did not report the shape of the distribution.

 

For the model, the variance in the effectiveness of LTP was assumed to be similar to that of latanoprost, since the estimate of the average effectiveness of LTP is also very similar to that of latanoprost. The final estimated distributions and their parameters are presented in Figure 9 and Table 13.

 

 

5.4 | Prevalence of Timolol contraindications

 

Timolol contraindications are asthma and severe chronic obstructive pulmonary disease, sinusbradycardia, second- or third degree atrioventricular block, and latent or uncontrolled heart failure.[24] In the DURING study, the prevalence of respiratory contraindications (which is the most evident and directive contraindication in clinical practice) was 123/1273.[36] This was rounded up to a default estimate of 10%.

 

5.5 | Side-effects with medication

 

The prevalence of side -effects with each of the medications in the model was based on the results of the DURING study.[12] In this study, previously untreated patients starting pressure lowering medication, and patients that switched medication, were followed for the next three visits. When the initiated treatment was stopped due to side -effects this was registered. The estimate of the incidence of side-effects was based on the proportion of patients on a certain treatment that stopped the medication due to side-effects (as judged by the ophthalmologist) within one or two follow-up visits.

 

In the model, the occurrence of side-effects indicates that the patient suffers from side-effects that are severe enough to warrant a switch to another therapy. There are a number of randomized controlled trials that have investigated pressure lowering monotherapies and reported on the occurrence of adverse events, but the occurrence of side -effects that warrant a treatment switch cannot be derived from these numbers. Therefore the estimates of the incidence of side-effects with the four base case medications were based solely on our own observational data. The original data from this research was used as input to the beta distributions. Alpha is the number of patients stopping treatment, and beta is the number of patients not stopping treatment.

Table 14: point-estimate of the incidence of side-effects, and parameters for second-order distribution (beta distribution)

 

Average

Alpha

Beta

Timolol

8%

10

109

Latanoprost

8%

22

258

Dorzolamide

14%

2

12

Brimonidine

23%

5

17

 

6. Surgery effectiveness

 

----------

 

 

In the model, the effect of glaucoma surgery, i.e. trabeculectomy (with mitomycin C) and Baerveldt implantation, is expressed as a new intra-ocular pressure. The average intra-ocular pressure after trabeculectomy and after Baerveldt implantation was estimated from literature.

 

 

6.1 | Intra-ocular pressure after trabeculectomy

 

A literature review was performed to estimate the average intra-ocular pressure after a successful trabeculectomy (i.e. the cases of immediate failure are not included). The included studies and their results are presented in Table 15.

 

The average intra-ocular pressure after trabeculectomy was estimated to be 12.5 mmHg based on this review and a weighted averaging of the outcomes according to sample size. In literature the average IOP’s after surgery were reported with a standard deviation, but the distribution of IOP’s after surgery cannot be normally distributed with e.g. 9.9 ± 5.0, because that would mean that a vast amount of patients reached an intra-ocular pressure below the physiological limit (± 6 mmHg) and some patients reached a negative intra-ocular pressure. The distribution of post-operative IOP’s in the model was simulated with a gamma distribution, because this distribution can take the shape of a normal distribution but always remains higher than 0.

 

The parameters of the gamma distribution (alpha and beta) are related to each other through the formulas in Table 16. However, completing these formulas with the values for means and standard deviations reported in literature would lead to very wide distributions of the postoperative IOP. Therefore we first set the boundaries of plausible values of a post-surgical IOP in discussions with clinical experts, which were approximately 6 mmHg at the lowest and approximately 20 mmHg at the highest. This latter (maximum) value is based on the fact that in the model it is assumed that a proportion of the patients does not respond to the trabeculectomy and keeps his/her preoperative IOP. This proportion should not be represented by the distribution of IOP’s after successful surgery.

 

The value of beta was varied while the expected value was kept at 12.5 mmHg. Alpha was adjusted such that the last formula in Table 16 was always valid. The value of beta was fine-tuned to reach the desired distribution. The result is presented in Figure 10.

 

 

6.2 | Intra-ocular pressure after Baerveldt implantation

 

Literature about the post-operative IOP after Baerveldt implantations is scarce. The point estimate of the IOP after a Baerveldt implantation was based on expert opinion and two studies in literature.

 

Table 17: literature review for implantation results

Article

Year

Patients

N

IOP baseline (mmHg)

IOP after surgery (mmHg)

Follow-up

Wilson et al. [19]

2003

POAG or PACG

59

25.9 ± 7.6

16.2

6 months

Gedde et al. [22]

2007

Glaucoma

107

25.6 ± 5.3

12.4 ± 3.9

(1.3 ± 1.3 co-medications)

12 months

Goulet et al.[23]

2007

Glaucoma

62

35.3 ± 12.9

14.7 ± 6.8

(0.8 ± 0.9 co-medications)

12 months

Wilson et al. studied the effects of Ahmed rather than Baerveldt devices.

 

The estimate of two ophthalmologists that frequently perform implantation surgery in glaucoma patients (HB, NJ) was that the post-operative intra-ocular pressure was 12 to 14 mmHg in the presence of some co-medication. Based on these estimates and the literature results (which were also in the presence of co-medication) a point -estimate of 15 mmHg without co-medication was made. In order to make an estimate of the distribution of IOP’s after Baerveldt implantation we used a gamma distribution with the same value for beta that was used in the distributions of IOP’s after trabeculectomy. The resulting distribution is presented in Figure 11.

7. Conversion risk

 

----------

 

 

7.1 | Baseline risk

 

The estimate of the baseline risk of conversion is based on the Kaplan-Meier results of the Ocular Hypertension Treatment Study where 78 untreated patients converted to POAG within 5 years and 741 patients did not convert, with an average IOP during follow-up of 23.9

 

± 2.9 mmHg and an average age of 54.9 ± 9.5 years.[4] In 50-57% of the patients the diagnosis of conversion was based on the appearance of the optic disc, in 32-42% it was based on visual field measurements, and in 8-10% it was based on both the appearance of the optic disc and visual field measurements.

 

A recent systematic review of randomized controlled trials in untreated OHT patients found a higher conversion risk in several trials other than in the OHTS (10% -37% in approximately 5 years), but all these trials differed in the definition of conversion, follow-up time and population base.[1] It is therefore hard to compare the results. However, in the systematic review the reported cumulative risks were based on the total number of events in the total follow-up period, while our model uses the Kaplan-Meier estimates at 60 months exactly. For example, in the systematic review the cumulative risk reported for the OHTS study was 10.9% in an average follow-up of 60 (and median 76.5) months, while our model uses 9.5% after 60 months exactly. This could partially contribute to the higher cumulative risks found in literature compared to the estimate used in the model.

 

 

7.2 | Relative risks

 

The investigators of the OHTS and the EGPS study have joined the data of untreated patients in both studies to develop a prediction model for the development of POAG.[5] The results of the final multivariate hazard model are presented in Table 18.

 

 

Table 18: Multivariate hazard ratios (HR) in the pooled OHTS and EGPS control groups.[5]

Variables

Hazard ratio

95% confidence interval

Age (per decade higher)

1.26

1.06; 1.50

Mean IOP (per mmHg higher )

1.09

1.03; 1.17

Mean CCT (per 40 μm higher)

2.04

1.70; 2.45

Mean vertical C/D ratio (per 0.1 higher)

1.19

1.09; 1.31

Mean PSD (per 0.2 dB higher)

1.13

1.04; 1.24

IOP = intra-ocular pressure, CCT = central corneal thickness, C/D ratio = cup/disc ratio, PSD = pattern standard deviation.

 

In our model, the relative risk of age and intra-ocular pressure are updated continuously, and are based on the actual age and intra- ocular pressure of the simulated patient. The other prognostic factors are aggregated into one additional factor in the model (‘Other risk factors’)

 

7.2.1 | Relative risk of intra-ocular pressure

 

The point-estimate for the relative risk of IOP above 23.9 mmHg is 1.09, based on the results of the pooled OHTS/EGPS risk model.[5] This estimate was further supported by a meta-analysis performed in the University Eye Clinic Maastricht.[2]

7.2.2 | Relative risk of age

 

The point-estimate for the relative risk of age above 54.9 years is 1.26, based on the results of the pooled OHTS/EGPS risk model.

 

7.2.3 | Other prognostic factors

 

The presence of other risk factors (or the relative risk of the other prognostic factors in Table 18) is in the model simulated by a single variable. As a model input an estimate is needed for the distribution of the ‘additional risk’ in the typical population of OHT patients. The average of this value is 1, since the average population of the OHTS and the EGPS studies has the average risk of conversion, i.e. no additional risk (or risk reduction).The distribution of the value of this variable however could not be derived from the OHTS or the EGPS studies, because the risk model has never been applied to the actual study populations themselves. The OHTS investigators have however used a cohort of patients of the “Diagnostic innovations in glaucoma study” (DIGS) to validate their predication model for the development of POAG.[37] They present the distribution of predicted probabilities for the 5-year risk of glaucoma development among the 126 untreated patients with OHT.

 

We have used the risk distribution in the DIGS cohort to deduct the distribution of additional risk in an OHT population. The distribution of the natural logarithm of the additional risk was assumed to be normal with an average 0 (since e0=1). A population was simulated with an age and IOP distribution similar to that reported for the DIGS population, and hazard ratio’s for age and IOP as reported by the OHTS/EGPS investigators.[37]. Subsequently a normal distribution was used to simulate the additional risk, and the standard deviation of this distribution was fine-tuned in such a way that the resulting distribution of risk in the population resembled the distribution reported for the DIGS cohort. With a standard deviation of 0.7 (Figure 12), the resulting distribution of predicted risk of conversion resembled the DIGS cohort best (Figure 13).

 

7.3 | Equations

 

The baseline risk and the relative risks of age, IOP, and other risk factors are aggregated in the model into the following equation for the current hazard rate for conversion from OHT to POAG of individual i.

 

 

 

The individual hazard is entered into the survival function

P =1 S =1 e

h  t

i

, from which a random draw is made to arrive at the new value for time-to-conversion. Figure 14 shows two examples of such cumulative probability distributions for conversion.

 

 

8. Progression

 

----------

 

 

Progression of POAG in the model is simulated via the level of the Mean Deviation (Humphrey Field Analyzer, Carl Zeiss Meditec, Jena, Germany) of the simulated patient. A decrease in the Mean Deviation signifies a decrease in the quality of the visual field, which is the pivotal feature of functional glaucomatous progression.

 

Glaucomatous progression is the mechanism where nerve fibers in the optic nerve continue to degenerate, which leads to a loss of functional nerve fibers, which in turn leads to a desensitization of specific areas of the retina, resulting in localized ‘gaps’ in the visual field, which can ultimately lead to a loss of visual functioning. The translation of glaucoma progression into a value of the Mean Deviation in the model is not straightforward. Mean Deviation is calculated as the average deviation of the retinal sensitivity from the age - corrected normal values. The sensitivity is the threshold value of light intensity that is no longer perceived by the patient, and is expressed in terms of apostilbs. A 10- fold decrease in average sensitivity equals a loss of 10 decibels in the Mean Deviation, and a 1000- fold decrease in average sensitivity equals a loss of 30 decibels in the Mean Deviation. The consequence of this conversion is that Mean Deviation is actually a logarithmic parameter and this might be reflected in the natural course of the Mean Deviation in glaucomatous patients.

 

In a systematic review the available data on the characteristics of glaucomatous progression in time was investigated, and no evidence was found that indicated that progression measured by the Mean Deviation is not linear in time. The model therefore assumes the existence of a baseline progression rate (in dB per month) which can be influenced by the level of the intra-ocular pressure.

 

 

8.1 | Systematic review of glaucomatous progression (and rate of progression)

 

Several of the assumptions made in the model concerning the progression and speed of progression, were based on a systematic literature review. The method and results of this review will be briefly described here.

 

The purpose of the systematic review was to collect information regarding the progression of visual fields in time and to estimate the rate of progression in glaucoma patients.

 

Sources of references:          Embase

 

Pubmed

Bibliography of included articles

 

Search Terms:                       “Open-angle glaucoma”  AND

 

“Progression”  AND

“Visual field” OR “optic disc” OR “optic nerve”

 

Restrictions:                            English or Dutch

 

Adult population

 

Clinical Trial, Meta-Analysis, Randomized Controlled Trial, Case Reports, Classical Article, "Clinical Trial, Phase I", "Clinical Trial, Phase III", "Clinical Trial, Phase IV", Controlled Clinical Trial, Journal Article, Multi center Study

 

Exclusion criteria during title and abstract screening:

 

  1. Not OAG
  2. Technical report about visual field measurement
  3. Cross-sectional study
  4. Follow-up < 4.5 years AND no reference to progression rates (xx/month) in the abstract

 

  1. Does not concern OAG progression (e.g. IOP monitoring, screening, prevalence etc.)
  2. Review or case report.

 

Exclusion criteria during full-text screening:

 

  • The rate of visual field progression is not quantified and reported directly, AND
  • The rate of visual field progression cannot be derived from the change in score and mean follow-up time.

 

 

 

Table 19: Included studies in systematic literature review for glaucomatous progression rate

 

Population

N

FU (yrs)

A

B

Intervention

Outcomes

Chen et al., 2002 [6]

POAG, NTG, PEX, PDG

152 pts

7.5

N

N

Usual care

MD/year ± SD, PSD/year ± SD, AGIS, c/d ratio.

Chen et al., 2000 [7]

POAG, NTG, PEX, PDG

36 pts

5.6

N

N

Usual care

MD/year ± SD, PSD/year ± SD, AGIS, c/d ratio.

Feiner et al., 2003 [37]

POAG, PDG, PEX

607 pts

5

Y

N

1) Medical

2) Surgical

CIGTS

Heijl et al., 2002 [10]

POAG, NTG, PEX

255 pts

5.7

Y

N

1) LT

2) No Tx

MD/year ± SD

Lee et al., 2004 [38]

POAG, PACG

48 eyes

11.7

N

N

Usual care

%/year ± SD, c/d ratio

Mayama et al., 2004 [39]

OAG

105 eyes

7.5

N

Y

Usual care

MD

Mikelberg et al., 1986 [40]

OAG (not NTG)

45 eyes

7.6

N

N

Usual care

Scotoma mass/month

Oliver et al., 2002 [41]

POAG, PEX, PDG, NTG

290 pts

15

N

Y

Usual care

Stages/year

Pereira et al., 2002 [42]

POAG

40 eyes

16.2

N

N

Usual care

%/year ± SD, c/d ratio

Rasker et al., 2000 [43]

POAG, NTG, OHT

227 pts

9

Y

N

Usual care

%/year ± SD

Schwartz et al., 2004 [44]

OAG

30 eyes

6.7

N

N

Usual care

MS/year ± SD

Soares et al., 2003 [45]

OAG

44 eyes

4.9

Y

N

Usual care

MD/year

Vesti et al., 2003 [46]

OAG

76 pts

7

N

N

Usual care

MD

Wilson et al., 2002 [47]

Glaucoma, glaucoma suspect

287 eyes

10

Y

N

No Tx

AGIS, CIGTS

Zink et al., 2003 [48]

OAG

29 eyes

2.8

N

?

Usual care

CPSD/year ± SD

Kwon et al., 2001[49]

POAG

40 eyes

15.2

N

N

Usual care

%/year ± SD, c/d ratio

Smith et al., 1996[8]

Glaucoma

191 eyes

7.1

N

N

Usual care

MD/year ± SD, CPSD/year ± SD.

O’Brien et al., 1991 [50]

OAG (not NTG)

40 eyes

3.7

N

N

Usual care

MS/year ± SD

Katz et al., 1997 [9]

OAG from OHT

67 eyes

7

N

N

Usual care

MD/year ± SD, CPSD/year ± SD.

A: Prospective?, B: Selection based on progression? N = No, Y = Yes.

MD=Mean Deviation (HFA), MS=Mean Sensitivity (Octopus), %=Percentage of maximum sensitivity.


 

Glaucomatous progression (or glaucoma severity) was measured with several different methods in the included studies, as can be seen in the rightmost column of Table 19. The reported rates of progression with each of these methods are presented in Table 20 to Table 27. Some articles did not report the actual observed rate of progression. In these cases the rate was derived from the difference between the average final and the average baseline visual field parameter divided by the average length of follow-up. Most authors reported results divided in groups based on the occurrence of progression.

 

 

Table 20: Reported rate of MD change (dB/year)

Study

Patients

N

MD baseline (dB)

IOP (mmHg)

Mean ± SD

Chen et al., 2002

Progressed

54

Better: -3.6 ± 4.1

Worse: -7.6 ± 7.3

17

-0.64 ± 0.52

 

Stable

98

-

-0.39 ± 0.53

Chen, 2000

Progressed

6

Better: -6.0 ± 4.8

Worse: -17.9 ± 6.0

17.6

-2.20 ± 1.40

 

Stable

30

16.5

0.10 ± 0.70

Smith, 1996

Progressed

24

-8.7 ± 3.8

-

-1.26 ± 0.60

 

Stable

167

-

0.06 ± 0.60

Katz, 1997

Progressed

12

-7.4

-

-0.96 ± 0.44

Heijl, 2002

Control

(no treatment)

126

-4.4 ± 3.3

20.8

-0.60 ± 0.84

 

Treatment

129

-5.0 ± 3.7

15.5

-0.36 ± 0.60

 

 

Table 21: Derived rate of MD change (dB/year)

Study

Patients

N

MD baseline (dB)

IOP (mmHg)

Mean ± SD*

Mayama, 2004

Progressed

105

-4.2 ± 3.7

-

-0.72 ± 1.03

 

Stable

355

-9.2 ± 7.2

-

0.03 ± 1.52

Soares, 2003

All

44

-4.2 ± 2.9

-

-0.30 ± 0.40

Vesti, 2003

All

76

-8.5 ± 4.7

-

-0.31 ± 0.46

*Estimates of SD

 

 

Table 22: Reported rate of PSD change (dB/year)

Study

Patients

N

Mean ± SD

Chen, 2000

Progressed

6

0.61 ± 0.59

 

Stable

30

0.02 ± 0.51

Chen, 2002

Progressed

54

0.40 ± 0.30

 

Stable

98

0.30 ± 0.52

 

 

Table 23: Reported rate of CPSD change (dB/year)

Study

Patients

N

Mean ± SD

Zink, 2003

High tension glaucoma

13

-0.01 ± 0.81

Smith, 1996

Progressed

27

0.71 ± 0.34

 

Stable

164

-0.01 ± 0.39

Katz, 1997

Progressed

5

0.91 ± 0.25

Table 24: Reported rate of change in mean threshold values (dB/year)

Study

Patients

N

IOP (mmHg)

Mean ± SD

Schwartz, 2004

All

30

18 ± 2

-0.38 ± 0.56

O’Brien

Progressed

10

16.5 ± 1.8

-1.39 ± 0.78

 

Stable

30

16.8 ± 2.5

-0.07 ± 0.43

Schwartz, 2004

All

30

18 ± 2

-0.38 ± 0.56

 

Table 25: Reported rate with Grids methods, (%/year)

Study

VF method

Patients

N

IOP (mmHg)

Mean ± SD

Lee, 2004

Goldman, I2th and I4th isopter

% of normal field (full field)

All

48

17.2 ± 2.8

-0.81 ± 1.00

Pereira, 2002

Goldman, I4th isopter

% of normal field (full field)

All

40

18.4 ± 2.5

-1.29 ± 1.37

Rasker, 2000

Peritest, screening.

% of maximum possible sensitivity loss (full field)

Progressed

27

-

-2.5 ± 1.8

 

 

Stable

41

-

-0.45 ± 2.23

Kwon, 2001

Goldman, I4th isopter

% of normal field (full field)

All

40

18.4 ± 2.5

-1.28 ± 1.37

 

 

Table 26: Derived rate of change in AGIS score (points/year)

Study

Patients

N

IOP (mmHg)

Mean (change points / fu)

Chen, 2002

Worse eye

152

-

0.4 / 7.5 = 0.05 points / year

Chen, 2000

Worse eye

36

-

2.1 / 5.6 = 0.38 points / year

Wilson, 2002

Untreated

287

-

5.4 / 10 = 0.54 points / year

 

 

Table 27: Derived rate of change in CIGTS score (points/year)

Study

Patients

N

IOP (mmHg)

Mean (change points / fu)

Feiner, 2003

All

607

-

0.3 / 5 = 0.06 points / year

Wilson, 2002

Untreated

287

-

7.5 / 10 = 0.75 points / year

 

 

Although the methods to measure the severity of glaucoma or the rate of visual field loss varied across the included studies, most methods have in common that they are based on the sensitivity measurements from the Humphrey or Octopus automated perimeters. Some authors have commented on the linearity of the relationship between the visual field and time.

 

Smith, 1996 (MD/year) [8]

 

“We determined the type of regression function based on visual inspection of the data. This gave no indication that a better fit would be obtained with a nonlinear rather than a linear function. Because visual field data were recorded in decibels, a linear decline represents an exponential decay in retinal sensitivity.”

 

Katz, 1997 (MD/year) [9]

 

“A linear fit appeared to adequately describe the changes occurring in the visual field over time as evidenced by a lack of relationship between the residuals and time”.

 

Mikelberg et al. report that the majority of patients with progression had a linear progression of scotoma mass in time (Figure 16), and Kwon et al. showed that a linear model resulted in good fits for the visual field score versus time (Figure 17).

 

In addition to these data in literature, we inspected the visual field data of the patients included in our observational data from the quality-of- life study.[29, 30] Similarly to the conclusion of Smith et al. we found no indication that the relationship between Mean Deviation and time would be any other than linear. Some examples of patients with more than three visual field measurements are presented in Figure 18.

 

 

 

 

An evaluation of the assumption that in the natural course of disease MD decreases linearly in time, is hampered by the fact that there are no records of long-term MD progression in untreated POAG patients. The scarce data that are available typically concern short follow-up data of treated early POAG patients, with (presumably) more intensive treatment upon progression. The fact that MD decrease seems linear in these data may suggest that the actual progression in untreated patients is exponential, with an increasingly fast MD decrease as the visual field worsens. Some literature reports indeed suggest that baseline disease severity is an independent risk factor for progression [52], but this is contradicted by other reports [48, 50, 53]. The latter study results may even suggest that a more severe baseline visual field defect is associated with a lower risk of progression. However, this result may be observed due to the fact that there is a limit to the amount of visual field a person can loose. Patients with severe visual field defects have less to loose than patients with early visual field defects. Given the current level of evidence we concluded that it would not be implausible to assume that an individual glaucoma patient in the model has a constant rate at which the Mean Deviation decreases each year, and that the height of this rate can be influenced by the IOP level.

 

In order to make an estimate of the height of this natural rate, a meta-analysis was performed in Review Manager (Cochrane) with the studies from the systematic review that reported the rate of progression in MD/year with variance data (95% confidence interval, variance, SE or SD) and the reported rates were compared against a hypothetical group with no change in the Mean Deviation. The results of this meta-analysis are presented in Figure 19. The average rate of MD change per year in treated patients was -0.33 dB/year (95% CI: - 0.38; -0.28).

 

In 2009, results from the Groningen Longitudinal Glaucoma study were reported, which showed show that in an unselected cohort of OAG patients with an average IOP of 14.9 mmHg during a mean follow-up of 5.3 years, the annual change in MD was -0.25 dB/year.[54] This is in good agreement with the results from the meta-analysis and further support the estimates made for the model input.

 

 

8.2 | Baseline rate of progression

 

In the model, each individual patient is ‘assigned’ a reference progression rate (MDRref) that represents the rate with which MD would decrease annually if IOP and additional risk were as in the referent POAG population. There is no information on the actual distribution of the rate of MD progression in the POAG population. For the sake of the model consistency, it was assumed that the decrease in the Mean Deviation in glaucoma patients is always larger than zero (i.e. there are no patients with an improving visual field). We therefore chose to use a gamma distribution, because values in the Gamma distribution are always higher than zero while the distribution is flexible in its shape trough the shape parameter.

 

The estimate of the distribution of progression rates in the POAG population was initially based on the treated patient population in the meta-analysis (Figure 19). The average progression rate was 0.33 dB/year (which corresponds to 0.028 dB/month), and the standard deviation was derived from the individual studies in Table 20 and estimated at 0.63 dB/year (which corresponds to 0.053 dB/month). The formulas in Table 16 and the estimated mean and standard deviation of 0.028 ± 0.053 dB/month for treated patients, lead to an initial estimate for alpha 0.36 and an estimate for Beta 0.08 and a population distribution as presented in Figure 20.

Since we did not know the actual shape of the distribution of progression rates in a POAG population, we validated the initial estimates for the distribution in Figure 20 by using the only other source of information we had. The initial distribution was translated into an expected survival curve by means of simulation in order to compare it with the survival curve reported in the EMGT study. The estimated average progression rate of 0.028 ± 0.053 is very similar to the results reported for the EMGT study.[10] In the EMGT study, progression was defined as either visual field progression or optic disc progression. Visual field progression was defined as at least 3 test-points significantly progressing at the same location in the EMGT Pattern Change Probability Maps (based on the PSD) in 3 consecutive tests. Optic disc progression was defined as a clear and progressive change located at the same optic disc clock-hour, confirmed in photographs 6 months later. Optic disc progression was determined by 2-3 independent graders. The authors of the EMGT study have reported that, in retrospect, progression was associated with an MD change of -2.26 dB and that progression was based on the visual field measurements in 82 -91% of the cases. In Figure 21 the curve resulting from our simulation was merged with the progression curve reported for the EMGT study.

 

 

From Figure 21 it is clear that the expected incidence of progression with the initial Gamma distribution in Figure 20 was much lower than the incidence observed in the EMGT study. Moreover, the shape of the survival curve did not match the observed curve. Therefore the parameters of the Gamma distribution were adjusted, restricted by the mean of 0.028 dB/month, in order to fine-tune the survival curve towards one that would resemble the results of the EMGT study. The result is presented in Figure 22 which shows a Gamma distribution with alpha = 2 and Beta = 0.014. This distribution has an expected value of 0.028

 

± 0.02. It was not possible to find a Gamma distribution that lead to survival curves that matched the EMGT results exactly, but the distribution in Figure 22 lead to survival curves that matched the shape of the EMGT survival curves best. We valued a match in the shape of the curve higher than a match (over a small range) in the absolute values of the incidence of progression with a curve shape that does not match. The reason for this is that the difference between two curves with a similar shape corresponds with a time-shift, or a delay in the onset of progression in the simulated cohort relative to the observed cohort (in this case approximately 18 months), whereas the difference between two curves with a different shape corresponds to a different relationship between time and risk.

 

 

8.3 | Relative risk of intra-ocular pressure

 

Lowering intra-ocular pressure reduces the risk of progression in POAG patients.[10, 55, 56] However, the magnitude of risk reduction per unit of pressure lowering has not been established in systematic reviews or meta- analyses. The only source for an estimate of the relative risk of the intra-ocular pressure on progression is the EMGT study.

 

The EMGT authors have performed a multivariate Cox proportional hazards models and found a hazard ratio of 1.13 (95% CI: 1.07; 1.19) per mmHg higher during a median follow-up of 6 years, corrected for age, baseline IOP, exfoliation, number of eligible eyes and MD.[10] The reference value of the intra-ocular pressure in the model is the average value of the intra-ocular pressure in the treated EMGT group (15.5 mmHg).

 

The relative risk (odds ratio) of progression per mmHg was derived from logistic regression analysis, with the occurrence of progression as a dichotomous outcome. However, in the model progression is not modeled as an event but rather as a continuous process. The relative risk of IOP works on the progression rate, not on progression risk. Still, in the absence of any other data on the relative risk of IOP on the rate with which MD decreases in time, we have applied the reported odds ratios to the MDR directly. This means that, forexample, an odds ratio of 2 results in an MD-change that is twice as fast as the MDRref. We have checked this assumption by creating survival curves from hypothetical distributions of

MDR and calculating the relative risk of progression (i.e. a total MD decrease of 2.3 dB or more) compared to the reference MDR distribution. The reference MDR distribution was the Gamma(2, 0.014) distribution described in the previous paragraph. The hypothetical distributions were created by increasing the expected value with a factor 2, 4, and 6. The relative risks of progression derived from these distributions depend on time, because the cumulative event curves converge and the relative risks ultimately all approach 1. However, when we looked at the time-period from which the odds ratio in the EMGT study was derived (i.e. 6 years), the relative risks of progression in the hypothetical distributions were 2.6, 4.8 and 6.3 respectively. These numbers resemble the multiplication factors of the distributions (2, 4 and 6), and we concluded that the direct extrapolation of relative risks into MDR multiplication factors in the model does not lead to unlikely results.

The construction described above, where each simulated patient is assigned a random reference MDR which is subsequently adjusted to the actual IOP of the patient, has the consequence that the MDR in the model can approach zero, but it can never become zero. After all, the reference MDR drawn from a Gamma distribution is always higher than 0, and so is the relative risk, even at low IOP values. As a result of this, a POAG patient in the model will never reach a stable disease stage. Some reports in literature indicate that in POAG the rate of progression might decrease to zero if the intra-ocular pressure is low enough, although it remains an issue of some dispute.[55, 57, 58] In the AGIS study the investigators found that very few of the patients with an intra-ocular pressure that was always below 18 mmHg in the first six years of follow-up (on average 12.3 mmHg) showed progression in ten years.[55] Shirakashi et al. found that patients without progression in a follow-up of fifteen years had an average intraocular pressure of 13.4 ± 1.3 mmHg.[57] It may therefore be reasonable to assume that there is something like a threshold IOP below which no progression occurs. In the base case model we assumed that the MDR will become 0 dB/months if the intraocular pressure is lower than 13 mmHg.

8.4 | Equations

The baseline rate of progression and the relative risks of IOP are aggregated in the model into an the following equation for the current MDR of individual i.

8.5 | Criterion for progression in the model

The disease progression of POAG patients is modeled via the gradual decrease of MD in time. Whether or not a simulated patient experiences ‘clinical progression’ in the sense that it calls for a treatment adjustment therefore depends on the definition of progression in any particular analysis. The occurrence of progression is not an outcome of the model, but rather it is a tool to guide treatment decisions throughout the simulated life of POAG patients. When progression occurs in a simulated patient, the target pressure will be adjusted, and (if possible) a treatment change will occur.

In real clinical practice the establishment of progression is a process that involves multiple measurements and clinical judgment. It is not only important to establish objectively whether the optic nerve or the visual field has worsened, but also whether this worsening calls for a treatment change. Not only the absolute change may play a role, but also the time-frame in which this absolute change was observed relative to the life-expectancy of the patient.

In the base case model presented here, an absolute MD decrease of 2 dB was set as the criterion for progression, irrespective of the time-frame and life-expectancy of the patient. The threshold of 2 dB was based on the literature reports by Wesselink et al. and Heijl et al., where POAG patients with confirmed progression had an average MD decrease of 2.4 dB and 2.3 dB respectively.[10, 54] The criterion for progression in this base case model may be rather stringent compared to clinical practice.

It is important to note that the base case model does not directly take inter-test variability of visual field measurements into account. The modeled MD value is the ‘real’ MD value, and it is assumed that the ophthalmologist can measure this value with a 100% sensitivity and specificity. In that context, the progression criterion of 2 dB defines the threshold for the absolute reduction in MD that will, in the model, call for a treatment change, given that it has been established that the real MD decrease is indeed more than 2 dB. The 2 dB threshold should not be confused with the threshold that will in clinical practice trigger a suspicion of progression and call for extra visual field measurements to confirm or invalidate this suspicion.

 

The base case model adds the costs of an extra visual field measurement to a visit in which progression is observed, to take account of the extra visual field measurement that is in clinical practice performed to confirm a suspected progression.

 

In the model, the absolute MD decrease is measured relative to either the first MD measurement in the model, or the MD value at the moment the previous progression was observed.

 

For example:

 

Visit

MD measured

Progression?

 

 

1

- 3.0

No

3.0 dB

 

2

- 4.5

No

 

 

 

3

- 6.0

Yes

 

 

4

- 7.0

No

2.2 dB

 

5

- 8.2

Yes

 

 

9. Cataract

 

----------

 

 

POAG and cataract are related by the fact that the risk of developing cataract is higher in patients that received a trabeculectomy in the past. Conversely, a laser trabeculoplasty is contraindicated if a patient has undergone a cataract extraction in the past. For this reason, the occurrence and treatment of cataract is included in the disease progression model of OHT and glaucoma.

 

 

9.1 | Baseline risk of cataract formation

 

The age -related incidence of cataract was derived from medical statistics provided by the National Institute for Public Health and the Environment (RIVM).[59] The incidence numbers were used to estimate the age-related hazard of cataract formation for the patients in the model.

 

Table 28: Estimated prevalence and incidence of cataract in The Netherlands

 

Year prevalence

Incidence

 

 

(per 1000)

 

(per 1000 per year)

Age

Male

Female

Male

Female

 

 

 

 

 

0-4

0,74

0,00

0,27

0,00

5-9

0,37

0,16

0,00

0,00

10-14

0,63

0,53

0,00

0,00

15-19

0,69

0,19

0,00

0,00

20-24

0,65

0,18

0,00

0,00

25-29

0,36

0,73

0,00

0,00

30-34

0,43

0,90

0,00

0,13

35-39

0,95

0,62

0,00

0,47

40-44

2,11

0,99

0,20

0,00

45-49

2,28

4,13

0,56

0,77

50-54

2,73

4,91

0,61

1,67

55-59

7,29

8,52

2,44

3,01

60-64

16,95

19,27

3,67

4,82

65-69

40,45

52,20

9,95

14,62

70-74

84,84

113,91

18,65

22,46

75-79

153,00

183,46

32,16

33,16

80-84

205,98

237,24

35,85

35,78

85+

283,92

296,89

20,21

22,14

At baseline the model establishes whether a patient has experienced cataract formation in the past, based on the cumulative risk at the baseline age of the patient. If a patient has not developed cataract in the past, the model uses the age- related hazard of cataract formation to simulate cataract development during the simulation.

 

9.2 | Relative risk of trabeculectomy for cataract formation

 

Trabeculectomy may enhance cataract formation. In the Collaborative Initial Glaucoma Treatment Study (CIGTS) the authors found a relative risk of cataract extraction of approximately 3.0 after trabeculectomy relative to medication only.[56] The CIGTS study was also referred to in a Cochrane systematic review.[60] The authors of the review reported a relative risk of cataract extraction after surgery of 2.72 at up to three years follow-up (95% CI: 1.51; 4.89).

 

In the Advanced Glaucoma Intervention Study (AGIS), the authors reported an increased risk of cataract surgery in the group that was initially treated with trabeculectomy versus the group that was initially treated with laser trabeculoplasty (1.1 to 1.3) after ten years.[61] It should be noted however that 50% of the patients in the laser trabeculoplasty group also received trabeculectomy within the first 10 years (hence perhaps the lower relative risk).

 

In the model the default estimate of the relative risk of cataract formation after trabeculectomy was 2.7.

 

 

9.3 | Cataract extraction

 

Cataract is often surgically removed. However, the model takes account of a small percentage of patients that can, for any reason, not undergo cataract extraction. A baseline attribute (yes/no) informs the model whether a patient will undergo cataract extraction if cataract develops during the model, or whether the patient has undergone cataract extraction if the baseline attribute ‘Cataract in the past’ is positive.

 

10. Utility outcomes

 

----------

 

 

Estimates for the relationship between disease, treatment and utility were derived from observational research among 531 OHT and glaucoma patients in the University Eye Clinic Maastricht and five other Dutch ophthalmology centers. The methods and results of this observational research are described elsewhere.[29, 30] In this document only the results relevant for the parameter estimates of utilities are described.

 

 

10.1 | Visual Functioning Questionnaire

 

A multivariable linear regression model was used with the measured value of the patients’ score on the National Eye Institute Visual Functioning Questionnaire (NEI VFQ-25).

 

 

Table 29: Results of multivariable linear regression analysis with VFQ-score.

 

Unstandardized coefficient

Standard error

Significance

Constant

94.246

1.189

.000

Co-morbidities (yes vs. no)

-2.014

.424

.000

Side-effects score (per point)

-.194

.034

.000

MD in worse eye (per dB)

.496

.099

.000

MD in better eye (per dB)

1.050

.152

.000

Cataract in worse eye (yes vs. no)

-6.891

3.196

.032

Cataract in better eye (yes vs. no)

-2.301

2.921

.431

 

Parameters that occur in the model are: side-effects, MD and cataract. The number of co-morbidities was included in the regression model because it had an effect on the coefficients of cataract. Side-effects in the model represent the occurrence of side-effects that necessitate a change in medical therapy. In the data from the observational study, the participants were divided into two groups based on their answer to the question “How much do side-effects from medication impact your quality- of-life”. One group consisted of patients who answered “Not at all”, “Hardly”, “Somewhat” or “Quite a bit”, and the other group consisted of patients who answered “Much” or “very much”. The average side -effects score of the first group was 13, and the average score of the second group was 49. It was therefore assumed that the occurrence of side-effects severe enough to warrant a change in medication was associated with a side-effects score of 50. This is subsequently associated with a loss of 50*0.194=9.7 VFQ points. The model only simulates the MD in the better eye, but it is assumed that the disease progression is reasonably symmetric, and the coefficient of both MD in the worse eye as well as MD in the better eye was applied to the current MD in the model. The same goes for the presence of cataract.

 

Therefore, the current VFQ- score in the model is calculated with the formula: VFQ = 94 – 9.7*side-effects + 1.54*MD – 9.2*cataract.

 

The VFQ-25 scores is a value on a scale from 0 (worst) to 100 (best). This scale was converted directly into a utility scale from 0 to 1 by dividing the VFQ-25 score by 100. The life-years in the model adjusted for the VFQ-25 score therefore represent ‘visual functioning quality adjusted life-years (VFQaly).

 

10.2 | Health Utilities index

 

To estimate the relationship between disease, treatment and utility by the Health Utilities Index Mark 3 (HUI3), the same linear regression model that was used for the VFQ scores was employed for HUI. The resulting coefficients for the model parameters are presented in Table 30.

 

Therefore, the current HUI utility in the model is calculated with the formula: HUI3 = 0.88 - 0.1*side-effects + 0.01*MD - 0.059*cataract.

 

 

10.3 | EQ-5D utility

 

To estimate the relationship between disease state and utility by the EuroQol 5 dimensions questionnaire (EQ-5D), the same linear regression model that we used for the VFQ scores was employed. The resulting coefficients for the model parameters are presented in Table 31.

 

 

Therefore, the current EQ-5D utility in the model is calculated with the formula: EQ-5D = 0.97 + 0.05*side-effects + 0.004*MD.

 

The coefficients for cataract surgery are not included because they were positive. It is however highly unlikely that cataract would lead to a higher quality-of-life, so the cataract parameters were excluded for reasons of face validity.

 

11. Survival

 

----------

 

 

The survival of patients entering the model is based on their age at entry and the statistics on life-expectancy in The Netherlands from Statistics Netherlands (Centraal Bureau voor de Statistiek) (Table 32).[62]

 

 

 

For each simulated patient the model determines at baseline at which age he or she will die. The method to determine this final age was the following. For each age in Table 32 and Table 33 a random draw is made from a Bernoulli distribution with p equal to the risk of death at that age (depending on the gender). The result (0 or 1) is added to the table in an additional column. Subsequently the model searches for the first occurrence of the value 1 in this column starting at the baseline age of the patient in the table. The age that corresponds to this first occurrence of ‘1’ is the age-at-death.

 

 

12.  Average OHT/POAG population

 

----------

 

 

The cost-effectiveness analyses that were performed with the model focused on the average OHT/POAG population. In order to simulate the average OHT/POAG population, several estimates were made of the expected values and distributions of patient characteristics in this population. The sources of these estimates are described below.

 

 

12.1 | Ocular hypertension population

 

12.1.1 | Age

 

The age distribution of patients with OHT in the disease progression model was derived from the Ocular Hypertension Treatment Study and the European Glaucoma Prevention Study.[4, 63] The average age of the patients with OHT in those studies was 55 ± 12 years (skewed to the right) and 57 ± 10 (skewed to the left) respectively. In the disease progression model the age distribution of OHT patients was assumed to be normal with average 55 and standard deviation 10.

 

12.1.2 | Gender

 

The gender distribution of patients with OHT in the disease progression model was derived from the Ocular Hypertension Treatment Study and the European Glaucoma Prevention Study.[4, 63] The percentage of men in these studies was 43% and 46% respectively. In the disease progression model the gender distribution of OHT patients was assumed to be dichotomous with a 40% probability of the male gender.

 

12.1.3 | Baseline IOP

 

The distribution of intra-ocular pressure of new patients in the model was based on the average intraocular pressure found in the OHTS (25 mmHg), the EGPS (24 mmHg) and the Groningen longitudinal glaucoma study (27 mmHg), and on the reported distribution of IOP’s in the patient population in the European Glaucoma Prevention Study (EGPS).[4, 63, 64] The average IOP in the EGPS population was 23.6 ± 1.7 mmHg, but the distribution was strongly skewed to the right and truncated at 29 mmHg. In the disease progression model the distribution of intraocular pressure at baseline in the average population of OHT patients was assumed to be a normal distribution with average of 22 mmHg and standard deviation 4, but truncated on the left at 22 mmHg. The resulting distribution has an average IOP of 25 mmHg is skewed to the right, and includes intra-ocular pressures up to the high thirties.

 

12.1.4 | MD after conversion

 

The value of the Mean Deviation after conversion in the disease progression model was based on the Groningen longitudinal glaucoma study, where the average MD in recently converted patients was -3.6 dB with a range of -0.8 dB to -7.6 dB (personal communication).[64]

 

The OHT population distribution of MD after conversion in the model was represented by a (negative) gamma distribution, which cannot take a value higher than zero. The latter restriction was built into the model because it was precluded that POAG patients can have MD values higher than zero. The parameters of the distribution were iterated to obtain a gamma distribution with an average of 3 dB and a standard deviation of 1 dB. The parameters of the final distribution were - Gamma (6, 0.5), which has an average of -3 dB and a range of -0.5 to -7.5 dB.

 

Note: the MD value in an OHT patient only becomes relevant at (and after) a conversion event. This does not mean that the model assumes that the establishment of conversion was

based on the visual field only. Rather, conversion is modeled as an event, as a given fact, that is not necessarily observed by the ophthalmologist. The model’s determination of the MD value in the converted patient is a consequence of the fact that the patient has converted, because the model needs an MD value to be able to further simulate the disease progression. The chosen distribution of MD values in newly converted patients includes MD values that are close to zero, which represent the patients with glaucomatous changes in optic disc but without apparent defects in their visual field.

 

 

12.2 | Primary open-angle glaucoma population

 

A problem with trial based averages for POAG patients is that the study population usually represents only a small selection of patients (based on the in- and exclusion criteria). It is therefore actually quite difficult to estimate the baseline clinical characteristics for glaucoma patients. Here we report the parameters and the reasoning behind the parameter derivation that we used in the base case model presented in the article.

 

12.2.1 | Age

 

The age distribution of the average population of POAG patients was based on the study population in the Early Manifest Glaucoma Trial.[10] The average age was 68 ± 5 years and the distribution was slightly skewed to the left. In the disease progression model a normal distribution was used with average 68 and standard deviation 5. Several other sources were consulted for a typical age distribution of the average population of new POAG patients. The DURING study included 518 new POAG patients with an average age of 62 ± 11, while in the CIGTS it was 57 ± 11.[36, 56]

 

12.2.2 | Gender

 

The gender distribution of the average POAG population in the model was based on the EMGT population.[10] In this population, 34% were men.

 

12.2.3 | Baseline IOP

 

In the EMGT population the average intra-ocular pressure at baseline was 21 ± 4 mmHg, whereas the CIGTS population had an average intra-ocular pressure of 28 ± 6 at baseline.[10, 56] The unselected POAG population (including normal tension glaucoma patients) in the Groningen longitudinal glaucoma study had a baseline IOP of 30.3 ± 9.5 mmHg.[54] The differences were likely to be caused by the eligibility criteria of the trials: the EMGT excluded patients with an average IOP (in both eyes) higher than 30 mmHg, while the CIGTS excluded patients with an IOP lower than 20 mmHg. In the disease progression model, the baseline IOP in the POAG population was described by a normal distribution with mean 28 mmHg and standard deviation 3 mmHg, truncated on the left at 22 mmHg. The resulting distribution has an average of 29 ± 3 mmHg.

 

12.2.4 | MD at baseline

 

The value of the Mean Deviation in the average (newly diagnosed) POAG population in the disease progression model was based on the baseline MD value of the participants in the Early Manifest Glaucoma Trial (EMGT) and the Collaborative Interventional Glaucoma Treatment Study (CIGTS) .[10, 56] In both trials, early glaucoma patients were included. The average baseline MD in the EMGT study was -4.7 ± 3.5 dB, in the CIGTS -5.5 ± 4.2 dB, and the distributions were skewed to the left. Considering that the distribution of MD values was skewed to the left, and the fact that the MD in a converted patient cannot take on a positive value, the population distribution of MD in the average population of POAG patients in the model was based on a (negative) gamma distribution. The distribution was truncated at -3 dB because patient with POAG at the first presentation to an ophthalmologist will generally not

have MD values higher than – 3 dB. The base case distribution (- Gamma (2, 2.5, truncated at 3 dB) has an average of -7.2 ± 3.2 dB and ranges from -4 to -20 dB.

 

 

12.2.5 | Response to trabeculectomy

 

An attribute of the simulated patient is the type of reaction this patient has to trabeculectomy. There are three options:

  1. Immediate failure
  2. Late failure
  3. Never failure

 

In order to estimate the incidence of these three types of responses among patients with a primary trabeculectomy, a review of the literature was performed. From the articles thus considered it was apparent that the necessary information for the model could not directly be retrieved from the reported results. The articles generally report the failure rates of trabeculectomy, but ‘failure’ is not uniformly defined. Usually ‘failure’ of trabeculectomy is defined as an intra-ocular pressure persistently over a certain threshold (e.g. 15 mmHg, 18 mmHg or 21 mmHg) despite co-medication, whereas in the disease progression model, failure of trabeculectomy signifies a return to the pre-surgical intra-ocular pressure in the absence of co-medication. A second issue is the duration of follow-up. Long follow -up data are very scarce, and short term failure is reported only after 6 months (rather than a more immediate term such as 6 weeks). Still, several articles were consulted to inform our estimate of the incidence of trabeculectomy responses. The included articles are listed in Table 34.

 

 

 

The estimated intra-ocular pressure after trabeculectomy in the disease progression model is 12.5 mmHg with a distribution such that 94% of the cumulative function is lower than 15 mmHg (see paragraph 6.1 | ). The only results on short-term failure rates with a threshold of 15 mmHg are from Beckers et al. and Wudunn et al . which were 13% and 12% respectively. In the base-case model, an incidence of ‘immediate failure’ of 12% was assumed.

 

Due to the lack of long-term data of the results of trabeculectomy without co-medication, an estimate of the incidence of ‘never failure’ was made by two glaucoma surgeons (HB, NJ) at 40%. This estimate implies that in the model it is assumed that 60% of the patients who receive trabeculectomy will go back to the pre-surgical intraocular pressure within 10 years if they would not receive additional medication or laser treatment: 12% within 6 weeks (immediate failure), the remaining 48% gradually during the ten years after surgery.

 


13. Costs

 

----------

 

 

Costs were calculated in 2006 euro’s. Cost prices retrieved from older sources were indexed to 2006 euro’s with the percentages listed inTable 35.

 


14. Medication Costs

 

----------

 

The cost prices of eye-drops were collected from the Pharmacotherapeutic Compass, which listed the monthly costs of eye-drops based on the defined daily dosis (DDD) and the prices listed in the Z-index on 1 November 2007 (table 2).[24] The cost prices represent the declaration costs including claw-back, and excluding VAT (19%) and dispensing fee. None of the registered pharmaceuticals for glaucoma treatment are subject to co-payment in the Netherlands.

 

A pharmacist receives € 6.10 per dispensed drug, irrespective of the quantity.[25] The average annual amount of recipes for a glaucoma patient in the Netherlands ranged from 3.2 to 4.0 (depending on the type of drug) in 2006.[32] The average frequency of recipe collection was assumed to be every 3 months for all patients. The monthly cost for the pharmacists’ fee was therefore estimated at € 2.

 

The all-in monthly costs of medication were calculated by adding VAT and the pharmacists’ fee to the costs listed in the Pharmacotherapeutic Compass (Table 36). The cost price of timolol was based on Timoptol eye drops 0.5%. The costs for combination therapy and triple therapy were calculated by a summation of the cost prices of the monotherapies of the medications in the combination. The differences between prices of fixed combinations and the sum of the separate monotherapies are negligible.

15. Ophthalmologist, procedures and interventions Costs

 

----------

 

Three sources were identified for the estimation of costs associated with ophthalmologist visits, procedures and interventions.

 

The Dutch Manual for costing research lists several integral standard cost prices in 2003 euro’s.[26] These were based on bottom-up cost research in twenty Dutch hospitals and include materials, equipment, housing, wages, and overhead costs such as interest and depreciation costs.

 

Oostenbrink et al. have investigated the resource utilization and associated costs during the first two years after diagnosis of OHT or glaucoma in 200 to 500 patients in 5 to 10 Dutch hospitals.[27, 67] Cost prices in this study were based on detailed micro-costing studies in two participating (one peripheral and one university) hospitals, and can be regarded as integral cost-prices.

 

Peeters et al. have based their cost-estimates for glaucoma treatment on bottom-up costing research in the University Hospital Maastricht.[28] It is uncertain whether the authors have included overhead costs in their cost calculations, and the reported cost-prices may therefore not be integral.

 

The cost prices listed in the manual for costing research were the primary sources for the cost estimates in the OHT/POAG disease progression model, because they were derived from micro-costing studies, based on a large number of hospitals, and were transparently described in the manual. In addition, the standard cost prices listed in the manual are frequently used in Dutch health economic evaluations, increasing the comparability of our study results with others.

 

15.1 | Ophthalmologist visit

According to the manual for costing research the ratio of peripheral and university hospitals is 84:16. This ratio was applied to the manual’s cost prices and the resulting average estimate of € 65 was used in the base case model.

.

15.2 | Visual field measurement

 

The cost prices in the various sources range from € 46 to € 133. The estimate based on the manual for costing research (€ 133) was used in the base case model.

 

 

15.3 | Laser trabeculoplasty (LT)

 

The estimate of the cost price for laser trabeculoplasty (LT) must include all resources that are needed to perform the procedure:

 

Medical staff Equipment Housing

 

Day stay or hospital admission (usually not required)

 

 

 

There is a very big difference between the cost price estimates, which may be the result of a difference in the resources that were included in the estimate. The sources do however not contain a detailed account of the resources that were included in their estimates of the cost price for LT. We reasoned that the costs of medical staff and housing for LT are comparable to a regular ophthalmologist consultation, that there are additional costs for the equipment and that the costs of hospital admission are negligible since LT is performed on an outpatient basis. Since the base case estimate for an ophthalmologist consultation was € 65, the base case estimate for LT was kept at € 75.

 

15.4 | Trabeculectomy

 

The estimate of the cost price for trabeculectomy must include all resources that are needed to perform the procedure:

Medical staff Equipment Housing

Day stay (usually) or hospital admission (occasionally)

 

 

The estimate for the cost price of trabeculectomy in the base case model was based on the standard cost prices in the manual for costing research. It was assumed that all patients are admitted on a day stay basis. Therefore the base case estimate for the cost price of trabeculectomy is € 1214. Follow-up visits after trabeculectomy are modeled as separate events starting on the third day after surgery. However, in practice the first check -up after surgery will occur on the first post-operative day. To account for the costs of this consultation, the cost price of an ophthalmologist consultation is added to the cost price of trabeculectomy (€ 65) to reach a total of € 1279. This represents the cost price of trabeculectomy including day stay and a next day check.

 

 

15.5 | Re-trabeculectomy

 

The estimated cost price of a second trabeculectomy in the same eye is similar to the cost price of the first trabeculectomy.

 

 

15.6 | Baerveldt implant

 

The implantation of a filtering device is a surgical procedure (like trabeculectomy) that usually involves a day stay, and very occasionally an overnight stay. The cost price of an implantation procedure includes the costs of:

 

Medical staff Equipment Device

 

Housing

Day stay (usually) or hospital admission (occasionally)

 

 

The cost price of a Baerveldt implantation procedure has not been reported in any of the previously consulted sources.[26-28, 67] It is likely that the costs of medical staff and housing for implantation surgery are slightly higher than with trabeculectomy because the procedure takes more time. It is however unclear how much more time is required. Therefore the estimate of the cost price for implantation surgery was based on the estimate for trabeculectomy (€ 1214) and € 500 was added to account for the implanted device (personal communication with a glaucoma specialist, HB).

The final estimate for the integral cost price of implantation surgery, including day stay and a next day check is € 1779.

 

 

15.7 | Cataract extraction

 

Cataract surgery is performed on an outpatient basis with local anesthetics. The cost price for cataract surgery consists of the procedure itself (medical staff, equipment, housing) and the costs of two post-surgery follow-up visits.

 

Table 41: Cost prices of cataract extraction in various sources

 

In source

In 2006

Manual for costing research, 2003. [26]

€ 1525

€ 1584

 

 

 

University hospital Maastricht (personal communication with hospital financial administration)

€ 1100

 

Cataract surgery is a procedure for which the costs have been calculated quite precisely by hospital administrations, in view of the new billing system based on diagnose related groups (Diagnose Behandel Combinatie, DBC). The cost price of cataract surgery in the university hospital Maastricht has been communicated to us in 2008. Considering the recentness of this information the cost price of cataract surgery in the base case model was based on the estimate of € 1100.

 

16. Costs of low-vision rehabilitation services

 

----------

 

Low-vision rehabilitation services entail the services that are available to the visually impaired and blind to help them cope with their visual impairment, both on a physical, social and mental level. There is no scientific literature on the use of such services by glaucoma patients, nor do the institutions providing the services have information on the degree of service utilization by glaucoma patients. Therefore we have asked over 500 OHT and glaucoma patients to complete a questionnaire collecting information on resource utilization related to (impaired vision as a consequence of) glaucoma.[29, 30]

 

 

16.1 | Resource utilization

 

The questionnaire included a question asking about the utilization of services provided by revalidation institutions for the visually impaired or blind, e.g. Sensis, Vision and Bartiméus, during the last three months. The results, stratified by the Mean Deviation averaged over both eyes, are presented in the table below. This table indicates the number of patients in each stratum, the number (and %) of patients reporting the utilization of services during the last three months, and the type of service the patients received.

 

Table 42: Utilization of low-vision rehabilitation services in he last three months

Average MD in both eyes (dB)

Total n

Utilized services

SEM

95% CI

Type of service

MD ≥ 0

68

1 (1%)

1.2%

0%; 3.4%

Low vision investigation

-5 ≤ MD < 0

204

2 (1%)

0.7%

0%; 2.4%

Habits of living investigation

Habits of living investigation

-10 ≤ MD < -5

74

2 (3%)

2.0%

0%; 6.9%

Low vision examination

Habits of living investigation

-15 ≤ MD < -10

60

1 (2%)

1.8%

0%; 5.5%

Mobility instruction

-20 ≤ MD < -15

46

3 (7%)

3.8%

0%; 9.4%

Low vision examination

Low vision examination

Habits of living investigation

-25 ≤ MD < -20

29

1 (3%)

3.2%

0%; 9.3%

Audio book

Daisy player

Computer course

MD < -25

21

0 (0%)

0%

0%; 0%

-

Total

502

10 (2%)

0.6%

 

 

SEM = Standard error of the mean, 95% CI = 95% confidence interval

 

The translation of this information to cost price estimates is hampered by the fact that the questionnaire only asked for service utilization during the past three months, whereas these services are typically offered only once during a patients disease progression. It is virtually impossible to translate the three-month incidences to life-time incidences, so the model uses monthly costs for rehabilitation services based on the observed three month incidence numbers

 


16.2 | Cost prices

 

Cost prices for low-vision rehabilitation services were derived from the maximal tariffs set by the Dutch Healthcare Authority (NZa) in 2007. From personal communication with employees at Sensis we have learned that low- vision and habits of living examinations fall under ‘Basic treatment’ with a maximal tariff of € 96.20 per hour. Services such as independence training, mobility training, revalidation and social services fall under ‘Activating guidance, level 3’ with a maximal tariff of € 104.60 per hour. Low- vision examinations usually take 2 to 2.5 hours, the other examinations take approximately 1.5 hours and the activating guidance sessions take on average 2 hours.

 

The cost prices per hour and the average durations of the services were aggregated into an estimate of € 192 per low -vision rehabilitation service. This was multiplied with the observed three-month incidence of the utilization of services to obtain an estimate of the average costs of rehabilitation services per three months. Finally, the thus estimated costs were divided by three to obtain the monthly costs (Table 43).

 

Table 43: Calculation of average costs for low-vision rehabilitation services depending on MD in the better eye

MD in the better eye (dB)

Incidence in three months

Average cost per three months

Average cost per month

95% CI

MD ≥ 0

1%

€ 3.10

€ 1.03

0; 2.9

-5 ≤ MD < 0

1%

€ 2.10

€ 0.69

0; 2.0

-10 ≤ MD < -5

3%

 € 5.70

€ 1.89

0; 5.8

-15 ≤ MD < -10

2%

€ 3.50

€ 1.17

0; 4.6

-20 ≤ MD < -15

7%

€ 13.70

€ 4.57

0; 7.9

-25 ≤ MD < -20

3%

€ 7.20

€ 2.41

0; 7.8

MD < -25

0%

€ 0

€ 0

0

 


17. Costs of low-vision aids

 

----------

 

 

Low-vision aids for glaucoma patients entail both devices that aid the patient to see better, but also devices that aid to improve activities of daily living and mobility.

 

 

17.1 | Resource utilization

 

The degree of low -vision aid utilization in glaucoma patients was captured with a questionnaire.[29, 30] The participants were asked to indicate whether they currently used a specific aid, or whether specific adjustments were made to their house (e.g. lighting). The prevalence of optical aid utilization is presented in the next table, stratified by the average MD in both eyes.

 

Table 44: Prevalence of low-vision aid utilization in seven strata of average MD in both eyes

Type aid

MD ≥ 0

-5; 0

-10 ; -5

-15 ;-10

-20 ; -15

-25 ; -20

MD < -25

Total patients (n)

61

114

14

133

37

105

64

Glasses

49 %

49 %

69 %

63 %

72 %

75 %

62 %

Hand loupe

3 %

5 %

16 %

18 %

24 %

36 %

24 %

TV reading loupe

2 %

1 %

0

2 %

4 %

4 %

10 %

Loupe lamp

0

1 %

1 %

3 %

4 %

14 %

10 %

Loupe glasses

0

1 %

0

2 %

3 %

8 %

6 %

Filter glasses

3 %

0

2 %

5 %

0

0

0

Daisy player

2 %

0

0

0

0

17 %

0

Contacts

0

0

0

2 %

2 %

0

0

Night glasses

0

0

0

2 %

0

0

0

Adjusted lighting

0

0

3 %

0

0

7 %

5 %

White cane

0

1 %

0

2 %

9 %

18 %

24 %

Telephone

0

1 %

0

0

2 %

14 %

14 %

Software

0

0

3 %

0

0

4 %

0

Monitor

0

0

1 %

0

2 %

4 %

14 %

Monitor magnifier

0

0

1 %

0

0

0

0

Dictaphone

0

0

0

0

0

3 %

0

Walking stick

0

0

0

0

0

0

5 %

Keyboard

0

0

0

0

0

0

5 %

                                                                                                                                               

The utilization of low-vision aids other than glasses is generally low. For the purpose of the model it was important to establish whether the utilization of a specific aid differed between groups based on glaucoma severity (MD). Such a difference was seen with:

 

  1. Glasses
  2. Loupe
  3. TV reading loupe
  4. Loupe lamp
  5. Loupe glasses
  6. Daisyplayer
  7. White cane
  8. Telephone
  9. Monitor

 

Patients with the worse average MD also indicated a higher utilization of adjusted lighting. However, the cost price of adjusted lighting was estimated to be negligible.

 

In order to translate the survey results to model input we reasoned that the purchase of a low-vision aid is usually a one-time event that occurs when glaucoma severity has crossed a certain threshold. Over the whole of the low-vision aids, the largest increase in the prevalence of utilization was seen at MD values lower than -15 dB. Therefore the MD threshold to incur low-vision aid costs in the model was set at -15 dB. Next the study population was divided in two groups based on the average MD in both eyes: higher than - 15 dB and lower than -15 dB. The difference in the observed prevalence of aid utilization was assumed to be an estimate of the incidence of glaucoma-related low-vision aid utilization.

 

 

Table 45: Prevalence of low-vision aid utilization in two strata of average MD in both eyes

Low-vision aid

MD > -15

MD ≤ -15

Difference

Total patients (n)

405

97

 

Glasses

55%

71%

16%

Loupe

9%

27%

18%

TV reading loupe

1%

5%

4%

Loupe lamp

1%

8%

7%

Loupe glasses

1%

5%

5%

Daisy player

0%

5%

5%

White cane

1%

15%

14%

Telephone

0%

8%

8%

Monitor

0%

5%

5%

 

17.2 | Cost prices

 

Various sources were consulted to obtain estimates of the cost prices of the most important low-vision aids for glaucoma patients.

 

Ergra Low vision catalogue 2007

 

Price

White cane

€ 20.50

Telephone

€ 35 - € 150 (wireless)

 


 

Internet

 

Source

Price

Reading loupe with lamp

www.seniorenthuiszorgwinkel.nl

€ 50 - € 125

Daisy player

www.lexima.nl

€ 300 - € 400

Loupe glasses

www.lvbc.nl/produkt/view/607/print

€ 80 - € 270

TV reading loupe

http://kobavision.be/nl/prijzen.html

€ 3000 - € 4000

Monitor

http://kobavision.be/nl/schermpc.html

€ 750 - €1000

Portable loupe

http://kobavision.be/nl/hulpmid.html

€ 70

 

The cost prices per item were multiplied by the estimated incidence of glaucoma related low-vision aid utilization. The resulting total costs of low-vision aids (Table 46) were incurred in the model when a simulated patient’s better eye progressed to an MD value lower than -15 dB.

 

 

Table 46: Calculation of the average costs of low-vision aids

Low-vision aid

Prevalence

Cost price

Costs

Glasses

16%

€ 500

€ 80

Loupe

18%

€ 70

€ 13

TV reading loupe

4%

€ 3500

€ 140

Loupe lamp

7%

€ 75

€ 5

Loupe glasses

5%

€ 175

€ 9

Daisy player

5%

€ 350

€ 18

White cane

14%

€ 21

€ 3

Telephone

8%

€ 150

€ 12

Monitor

5%

€ 900

€ 45

Total

 

 

€ 325

 


18. Costs of homecare, grooming and nursing

 

----------

 

 

The degree to which progression of glaucoma leads to costs related to homecare or nursing homes was estimated based on the results of the questionnaire survey among OHT and glaucoma patients.

 

 

18.1 | Resource utilization, nursing home

 

The questionnaire asked whether the patient had ever needed to move as a result of OHT or glaucoma.

 

Table 47: Incidence of moving as a result of OHT or glaucoma in seven strata of MD in both eyes

Average MD in both eyes (dB)

Total patients (n)

Moved to other house

Moved to service flat

Moved to retirement home

Moved to nursing home

MD ≥ 0

68

1 (1%)

0

0

0

-5 ≤ MD < 0

204

0

1 (0%)

0

0

-10 ≤ MD < -5

74

0

1 (1%)

1 (1%)

0

-15 ≤ MD < -10

60

0

1 (2%)

1 (2%)

1 (2%)

-20 ≤ MD < -15

46

2 (4%)

0

1 (2%)

0

-25 ≤ MD < -20

29

0

1 (3%)

2 (7%)

0

MD < -25

21

1 (5%)

0

0

1 (5%)

Total

502

4 (1%)

4 (1%)

4 (1%)

2 (0%)

 

The percentage of patients indicating that they have had to move as a result of OHT or glaucoma was very low (Table 47). However, since long-term stay in retirement homes and nursing homes can be associated with high costs, we have calculated how much of the habituation of nursing homes and retirement homes can be attributed to progressing glaucoma. The total population was divided in two groups based on the average MD in both eyes: higher than -20 dB and lower than -20 dB. The difference in prevalence of nursing- or retirement home habituation was assumed to be attributable to glaucoma progression to a visual field with an MD lower than -20 dB (Table 48).

 

 

Table 48: Incidence of moving as a result of OHT or glaucoma in two strata of MD in both eyes

 

MD > -20

MD ≤ -20

Difference

Total patient in group (n)

452

50

 

Moved to retirement home

1%

4%

3%

Moved to nursing home

0%

2%

2%

 


18.2 | Resource utilization homecare

 

The questionnaire asked the participants how many hours a week (on average) they had received homecare in the last three months. We distinguished the following types of homecare:

 

  1. Family help
  2. Household help
  3. Grooming
  4. Nursing
  5. Other paid help

 

If patients had indicated not to have received homecare, the amount of hours per week was set at 0. The average time the study population had received each type of homecare, stratified by the average MD in both eyes, is presented in the next table.

 

Table 49: Average utilization of home care in seven strata of MD in both eyes (hours/week)

Average MD in both eyes (dB)

Family (hrs/week)

Household (hrs/week)

Grooming (hrs/week)

Nursing (hrs/week)

Paid help (hrs/week)

Total (hrs/week)

MD ≥ 0

0.1 ± 1.2

0.3 ± 0.7

0

0

0.2 ± 1.2

0.3 ± 2.7

-5 ≤ MD < 0

0.0 ± 0.4

0.0 ± 0.4

0.1 ± 0.5

0

0.1 ± 0.7

0.2 ± 1.3

-10 ≤ MD < -5

0.1 ± 0.6

0.1 ± 0.4

0.1 ± 0.5

0.0 ± 0.1

0.0 ± 0.2

0.3 ± 1.1

-15 ≤ MD < -10

0.1 ± 0.4

0.1 ± 0.4

0

1.4 ± 10.8

0.5 ± 3.0

2.0 ± 11.2

-20 ≤ MD < -15

0.4 ± 2.2

0.1 ± 0.4

1.2 ± 6.2

0

0.3 ± 0.9

2.0 ± 7.1

-25 ≤ MD < -20

0.4 ± 1.1

0

0.5 ± 1.4

0.4 ± 1.9

0.5 ± 1.6

1.8 ± 4.1

MD < -25

1.1 ± 2.9

0

0

0

0.2 ± 0.9

0.4 ± 1.9

Total

0.2 ± 1.1

0.1 ± 1.0

0.2 ± 2.0

0.2 ± 3.8

0.2 ± 1.3

0.7 ± 4.8

 

Overall the utilization of paid help appeared to be higher in patients with an average MD lower then -10 dB. The total study population was divided in two groups based on the average MD, and the difference in the utilization of paid help was considered the to glaucoma attributable amount of paid help utilization.

 

Table 50: Average utilization of family help and grooming in two strata of MD in both eyes (hours/week)

 

MD > -15

MD ≤ -15

Difference

Total patients (n)

404

94

 

Family help (hours/week)

0.08 ± 0.03

0.55 ± 0.22

0.47

Grooming (hours/week)

0.04 ± 0.02

0.71 ± 0.45

0.67

 

 

Table 51: Average utilization of other paid help and nursing in two strata of MD in both eyes (hours/week)

 

MD > -10

MD ≤ -10

Difference

Total patients (n)

345

154

 

Other paid help (hours/week)

0.08 ± 0.04

0.39 ± 0.17

0.31

Nursing

0.003 ± 0.003

0.61 ± 0.55

0.61

 

18.3 | Cost prices

 

The cost price per unit of the various types of homecare, retirement homes and nursing homes were derived from the Manual for costing studies and multiplied by the utilization estimates (Table 52).[26]

 

Table 52: Calculation of the average costs of homecare

Type

MD threshold

Utilization per week

Utilization per month

Cost price

Cost per month

Other paid help

-10 dB

0.31 hrs

1.3 hrs

€ 27.70/hour

€ 37

Nursing

-10 dB

0.61 hrs

2.6 hrs

€ 61.20/hour

€ 159

Subtotal

-10 dB

 

 

 

€ 196

Family help

-15 dB

0.47 hrs

2.0 hrs

€ 27.70/hour

€ 56

Grooming

-15 dB

0.67 hrs

2.9 hrs

€ 35.40/hour

€ 103

Subtotal

-15 dB

 

 

 

€ 159

Retirement home

-20 dB

 

3%

€ 88/day

€ 80

Nursing home

-20 dB

 

2%

€ 214/day

€ 130

Subtotal

-20 dB

 

 

 

€ 210

 



19. Costs of transportation

 

----------

 

 

19.1 | Resource utilization

 

The questionnaire survey among patients with OHT and glaucoma collected information on the means of transportation to various types of caregivers. The results are presented in the next tables. The study population was stratified according to the average MD in both eyes (Table 53 to Table 55). For example: 21% of the patients with an average MD higher than 0 dB usually walk or take their bike to visit the ophthalmologist, and 12% uses public transportation.

 

Table 53: Usual means of transportation to ophthalmologist, pharmacy and hospital in seven strata of MD in both eyes

 

To the ophthalmologist

Average MD in both eyes (dB)

Walking / cycling

Car

Public transportation

Taxi

Came to the house

MD ≥ 0

21%

66%

12%

2%

0%

-5 ≤ MD < 0

17%

66%

13%

4%

0%

-10 ≤ MD < -5

16%

58%

19%

7%

0%

-15 ≤ MD < -10

12%

56%

15%

19%

0%

-20 ≤ MD < -15

11%

56%

22%

11%

0%

-25 ≤ MD < -20

4%

44%

26%

26%

0%

MD < -25

5%

48%

29%

19%

0%

 

To the pharmacy

Average MD in both eyes (dB)

Walking / cycling

Car

Public transportation

Taxi

Came to the house

MD ≥ 0

72%

27%

0%

0%

2%

-5 ≤ MD < 0

63%

32%

2%

0%

4%

-10 ≤ MD < -5

70%

25%

2%

0%

3%

-15 ≤ MD < -10

56%

31%

0%

4%

10%

-20 ≤ MD < -15

55%

31%

2%

2%

10%

-25 ≤ MD < -20

59%

14%

9%

0%

18%

MD < -25

50%

28%

11%

6%

6%


 

To the hospital

Average MD in both eyes (dB)

Walking / cycling

Car

Public transportation

Taxi

Came to the house

MD ≥ 0

12%

72%

16%

0%

0%

-5 ≤ MD < 0

18%

64%

14%

4%

0%

-10 ≤ MD < -5

13%

64%

13%

8%

2%

-15 ≤ MD < -10

14%

61%

8%

18%

0%

-20 ≤ MD < -15

15%

56%

18%

10%

0%

-25 ≤ MD < -20

4%

48%

20%

28%

0%

MD < -25

10%

57%

14%

19%

0%

 


19.2 | Cost prices per unit

 

Cost prices per unit for each type of transportation were derived on the Manual for costing research (Table 56).[26]

 

Table 54: Cost prices for transportation

 

In source

In 2006

Car (per km)

€ 0.16/km

€ 0.17/km

Parking

€ 2.50

€ 2.50

Public transport (per km)

€ 0.16

€ 0.17

Taxi (per km)

€ 2.80 + € 1.75/km

€ 2.90 + € 1.80/km

 

Average distance to hospital: 7 km

Average distance to general practitioner: 1.8 km

 

It was assumed that the ophthalmologist is located in the nearest hospital, and that the distance to the nearest pharmacy is equal to the distance to the nearest general practitioner. The total cost prices were based on a two-way journey, plus parking costs if the journey was made by care. If the caregiver paid a home-visit to the patient the costs of transportation were assumed to be were similar to a car-ride to the caregiver, minus the parking costs. The cost prices that were used for the various types of transportation are the following:

 

 

Table 55: Calculation of cost prices for transportation to ophthalmologist, pharmacy and hospital

 

Walking / cycling

Car

Public transportation

Taxi

Came to the house

Ophthalmologist

0

€ 2.4 + € 2.5 = € 4.9

€ 2.4

€ 31

€ 2.4

Pharmacy

0

€ 0.6

€ 0.6

€ 12

€ 0.6

Hospital

0

€ 2.4 + € 2.5 = € 4.9

€ 2.4

€ 31

€ 2.4

 

 

19.2 | Cost prices per unit

 

Cost prices per unit for each type of transportation were derived on the Manual for costing research (Table 56).[26]

 

Table 56: Average cost of transportation to ophthalmologist, pharmacy and hospital in seven strata of MD in both eyes

 

Ophthalmologist

Pharmacy

Hospital

MD ≥ 0

€ 4.14

€ 1.37

€ 3.91

-5 ≤ MD < 0

€ 4.79

€ 1.71

€ 4.71

-10 ≤ MD < -5

€ 5.47

€ 1.35

€ 5.98

-15 ≤ MD < -10

€ 8.99

€ 3.00

€ 8.76

-20 ≤ MD < -15

€ 6.68

€ 2.43

€ 6.28

-25 ≤ MD < -20

€ 10.84

€ 1.33

€ 11.51

MD < -25

€ 8.94

€ 3.64

€ 9.02

 

Average distance to hospital: 7 km

 

Average distance to general practitioner: 1.8 km

 

It was assumed that the ophthalmologist is located in the nearest hospital, and that the distance to the nearest pharmacy is equal to the distance to the nearest general practitioner. The total cost prices were based on a two-way journey, plus parking costs if the journey was made by care. If the caregiver paid a home-visit to the patient the costs of transportation were assumed to be were similar to a car-ride to the caregiver, minus the parking costs. The cost prices that were used for the various types of transportation are the following:

 

 

Table 57: Average cost of transportation to ophthalmologist, pharmacy and hospital in two strata of MD in both eyes

 

Ophthalmologist

Pharmacy

Hospital

MD > -10 dB

€ 4.8

€ 1.5

€ 4.9

MD ≤ -10 dB

€ 8.9

€ 2.6

€ 8.9

 

19.3 | Total costs of transportation

 

The observed percentage of patients using a particular type of transportation was multiplied by the cost price for that type of transportation to obtain an estimate of the average costs of transportation for a visit to the ophthalmologist, pharmacy and hospital.

 

 

Table 58: Utilization of informal care in seven strata of MD in both eyes (hours/week)

 

Informal care received (hours/week)

MD ≥ 0

0

-5 ≤ MD < 0

0

-10 ≤ MD < -5

0.5 ± 2.6

-15 ≤ MD < -10

0

-20 ≤ MD < -15

0.1 ± 0.4

-25 ≤ MD < -20

2.5 ± 8.2

MD < -25

0.6 ± 2.5

Total

0.2 ± 2.2

 

In order to reduce the number of categories with different transportation costs in the model, the final number of strata was reduced to two, based on the average MD in both eyes: higher than -10 dB and lower than -10 dB (Table 59).

 

Table 59: Average cost of transportation to ophthalmologist, pharmacy and hospital in two strata of MD in both eyes

Ophthalmologist

Pharmacy

Hospital

 

 

 

 

MD > -10 dB

€ 4.8

€ 1.5

€ 4.9

MD ≤ -10 dB

€ 8.9

€ 2.6

€ 8.9

 

20. Costs of informal care

 

----------

 

 

20.1 | Resource utilization

 

The degree to which relatives, friends and neighbors help out with small tasks that a patient is unable to perform himself due to OHT or glaucoma was investigated with the questionnaire survey. Participants were asked to indicate how much per week they have received informal care during the past three months. The results are presented in the next table. If patients had indicated that they had not received informal care, the amount of time per week was set at 0.


 

 

Table 60: Utilization of informal care in seven strata of MD in both eyes (hours/week)

 

 

Informal care received (hours/week)

MD ≥ 0

 

0

-5 ≤ MD < 0

 

0

-10

≤ MD < -5

0.5

± 2.6

-15

≤ MD < -10

 

0

-20

≤ MD < -15

0.1

± 0.4

-25

≤ MD < -20

2.5

± 8.2

MD < -25

0.6

± 2.5

 

 

 

Total

0.2

± 2.2

 

Based on these results the total study population was divided in two groups, one with the average MD in both eyes higher than -5 dB, and one with the average MD in both eyes lower than -5 dB. The difference in the utilization of information care between the groups was assumed to be attributable to glaucoma progression (Table 61).

 

Table 61: Utilization of informal care in two strata of MD in both eyes (hours/week)

 

MD > -5

MD ≤ -5

Difference

 

 

 

 

Total patients (n)

270

218

 

Informal care (hrs/week)

0.0

0.50

0.5 hrs/week

 

 

 

20.2 | Cost prices

 

The cost price for one hour of information care was derived from the Manual for costing research (Table 62).[26] The manual offers two possibilities to estimate the cost price of informal care: one is based on research that elicited how people would valuate (in monetary terms) time spent on informal care giving (willingness-to-accept). The other is based on the costs if the unpaid help would have been performed by a paid help.

 

 

Table 62: Cost price of informal care

 

In source

In 2006

 

 

 

Willingness-to-accept (per hour)

€ 9.80

€ 10.20

Shadow price (per hour)

€ 8.30

€ 8.60

 

An average price of € 9 per hour was assumed. The monthly costs of informal care attributable to glaucoma were calculated by multiplying the monthly resource utilization (2.2 hours) with the average unit price, to obtain a base case estimate of € 20 per month for informal care if the MD progresses to values below -5 dB.

 

 

21. Costs of productivity loss

 

----------

 

 

21.1 | Resource utilization

 

Productivity losses can be caused by either temporary productivity loss due to sick days, or permanent productivity loss due to partial or full disablement. The questionnaire survey included a question on both work disablement as well as on sick days due to OHT or glaucoma.

 

None of the survey participants indicated that they had had a sick day due to OHT or glaucoma during the past three months. Work disablement did however occur among the participants (Table 63). The questionnaire collected information on the degree of disablement and the age of onset. The average time since the work disablement was calculated from the current age of the participant and the age of disablement onset.

 

Table 63: Incidence of work disablement in seven strata of MD in both eyes

Average MD in

Total patients

Work disabled

Age of

Degree of

Time since

both eyes (dB)

(n)

(%)

onset

disablement

onset (yrs)

 

 

 

 

 

 

MD ≥ 0

65

2 (3.0%)

54 ± 4

48 ± 25%

8.5 ± 0.1

-5 ≤ MD < 0

200

0 (0%)

-

-

-

-10

≤ MD < -5

70

3 (4.1%)

50 ± 3

78 ± 38%

8.3 ± 3.4

-15

≤ MD < -10

55

4 (6.8%)

51 ± 5

90 ± 12%

9.3 ± 3.9

-20

≤ MD < -15

44

2 (4.3%)

48 ± 8

63 ± 53%

11.7 ± 1.3

-25

≤ MD < -20

25

3 (10.7%)

53 ± 4

93 ± 12%

10.8 ± 1.7

MD < -25

15

6 (28.6%)

39 ± 11

92 ± 20%

12.8 ± 7.1

 

 

 

 

 

 

Totaal

502

20 (3.8%)

48 ± 9

82 ± 27%

10.6 ± 4.6

 

The prevalence of work disablement was higher in the two strata with the lowest average MD. Therefore the prevalence of work disablement in participants with MD lower than -20 dB (18.4%) was compared to participants with MD higher than -20 dB (2.5%). The difference (15.9%) was assumed to be the prevalence of work disablement attributable to glaucoma. The time since the onset of work disablement was approximately 10 years. With an average decrease of 0.03 dB per month and calculating back from -20 dB, the threshold MD value for the onset of work disablement was estimated at -15 dB.

In summary, in the model it is assumed that when MD progresses to values lower than -15 dB, 16% of the patients will become work disabled.

 

 

21.2 | Cost prices

 

The costs of productivity losses was calculated according to the friction cost method, using the standards proposed in the Manual for costing research.[26] The average costs per working person in 2003 were € 34.98 per hour (in 2006: € 36.33) . The friction period is 22 weeks, which equals 651.4 working hours. The elasticity was set at 0.8. Therefore the estimated friction costs for a full work disablement were 651.4 * 0.8 * 36.33 = € 18,932.

 

The model assumes a one-time cost of 0.16 * € 18,932 = € 3,029 as soon as a simulated patient progresses to MD values below -15 dB.

22. Summary of MD-related costs

 

----------

 

 

During the simulation of the disease progression of an individual patient, the costs of medication, ophthalmologist consultations, procedures and interventions were added to the total based on the occurrence of visits and the treatment decisions. All other costs were calculated during the simulation based on the MD value of the simulated patient. In the previous paragraphs the derivation of the cost estimates has been described. Here an overview is presented of the costs attributed to a patient based on his MD value. Three cost-types can be discerned: direct medical costs, direct non- medical costs and indirect non-medical costs. In addition, the costs can be added to the total as one-time costs as soon as a threshold MD value is passed, or as continuous costs that are incurred as long as the MD value remains on a certain level.

 

 

From previous paragraphs it may be apparent that is uncertainty surrounding the estimates for the costs associated with increasing disease severity. The total cost estimates are the

product of estimates for resource utilization and for cost-prices, which are surrounded with uncertainty themselves. The fact that there is uncertainty in the cost-estimates is a given, resulting from the reality that there are few data on resource consumption and cost-prices in ocular hypertension and glaucoma patients. However, the impact of parameter uncertainty on the results of incremental cost-effectiveness analyses can be evaluated with sensitivity analyses.

 

The total of costs for low-vision rehabilitation, low-vision aids, grooming, nursing, informal care and productivity loss can be interpreted as the ‘costs of low- vision and blindness’. There are not many sources of literature to verify our estimates. Burr et al. have recently concluded the same in their research for the cost-effectiveness of screening for open-angle glaucoma, and refer to the article by Meads & Hyde in which the annual costs of blindness as a result of macular degeneration were estimated at ₤6569 in the first year and ₤6487 in later years.[68, 69] With a 2006 conversion rate of € 1.5 for ₤ 1 these estimates would equal € 9700 per year. In our model a patient is considered ‘blind’ when the Mean Deviation in the better eye drops below -25 dB. At that time the annual costs for grooming, nursing and informal care are € 7020. This is lower than the estimates reported by Meads & Hyde. On the other hand, the model starts attributing costs for low-vision earlier in the disease progression process rather than only in case of blindness. Therefore the cost-estimates that served as input to the base case model lead to cost estimates for low-vision and blindness that resemble the few published estimates.

 

23. Abbreviations

 

----------

 

 

CI                    Confidence interval

 

EQ-5D            EuroQol 5 dimensions questionnaire

 

HUI3                Health Utilities Index mark 3

 

IOP                 Intraocular pressure

 

OHT                Ocular hypertension

 

POAG             Primary open-angle glaucoma

 

RCT                Randomized controlled trial

 

SD                   Standard deviation

 

SEM                Standard error of the mean

 

VFQ-25           25-item Visual Functioning Questionnaire

 

LT        Laser trabeculoplasty (also LTP)

 

24. References

 

----------

 

 

  1. Maier P, Funk J, Schwarzer G, et al. Treatment of ocular hypertension and open angle glaucoma: meta-analysis of randomised controlled trials. BMJ 2005;331(7509):134.
  2. Peeters A, Webers C, Prins M, et al. Quantifying the effect of intra-ocular pressure reduction on the occurrence of glaucoma. Acta Ophthalmol 2010;88:5-11.
  3. Heijl A, Patella V. Essential perimetry; The field analyzer primer, Third ed. Dublin, California, USA: Carl Zeiss Meditec, 2002.
  4. Kass M, Heuer D, Higginbotham E, et al. The ocular hypertension treatment study: a randomized trial determines that topical ocular hypotensive medication delays or prevents the onset of primary open-angle glaucoma. Arch Ophthalmol 2002;120:701-13.
  5. Gordon MO, Torri V, Miglior S, et al. Validated prediction model for the development of primary open-angle glaucoma in individuals with ocular hypertension. Ophthalmology 2007;114(1):10-9.
  6. Chen PP. Correlation of visual field progression between eyes in patients with open-angle glaucoma. Ophthalmology 2002;109(11):2093-9.
  7. Chen P, Bhandari A. Fellow eye prognosis in patients with severe visual field loss in 1 eye from chronic open-angle glaucoma. Arch Ophthalmol 2000;118:473-8.
  8. Smith SD, Katz J, Quigley HA. Analysis of progressive change in automated visual fields in glaucoma. Invest Ophthalmol Vis Sci 1996;37(7):1419-28.
  9. Katz J, Gilbert D, Quigley HA, Sommer A. Estimating progression of visual field loss in glaucoma. Ophthalmology 1997;104(6):1017-25.
  10. Heijl A, Leske C, Bengtsson B, et al. Reduction of intraocular pressure and glaucoma progression; results from the Early Manifest Glaucoma Trial. Arch Ophthalmol 2002;120:1268-79.
  11. Van der Valk R, Webers C, Schouten J, et al. Intraocular pressure-lowering effects of all commonly used glaucoma drugs - a meta-analysis of randomized clinical trials. Ophthalmology 2005;112:1177-85.
  12. Beckers HJ, Schouten JS, Webers CA, et al. Side effects of commonly used glaucoma medications: comparison of tolerability, chance of discontinuation, and patient satisfaction. Graefes Arch Clin Exp Ophthalmol 2008;246(10):1485-90.
  13. McIlraith I, Strasfeld M, Colev G, Hitnik C. Selective laser trabeculoplasty as initial and adjunctive treatment for open-angle glaucoma. J Glaucoma 2006;15(2):124-30.
  14. Glaucoma Laser Trial Research Group. The Glaucoma Laser Trial (GLT) and Glaucoma Laser Trial Follow-up Study: 7. Results. Am J Ophthalmol 1995;120:718-31.
  15. Chung P, Schuman J, Netland P, et al. Five-year results of a randomized, prospective, clinical trial of diode vs argon laser trabeculoplasty for open-angle glaucoma. Am J Ophthalmol 1998;126:185-90.
  16. Damji K, Shah K, Rock W, et al. Selective laser trabeculoplasty versus argon laser trabeculoplasty: a prospective randomised clinical trial. Britisch Journal of Ophthalmology 1999;83:718-22.
  17. Juzych M, Chopra V, Banitt M, et al. Comparison of long-term outcomes of selective laser trabeculoplasty versus argon laser trabeculoplasty in open-angle glaucoma. Opthalmology 2004;111(10):1853-9.
  18. Beckers H, Kinders K, Webers C. Five-year results of trabeculectomy with mitomycin C. Graefes Arch Clin Exp Ophthalmol 2003;241:106-10.
  19. Wilson M, Mendis U, Paliwal A, Haynatzka V. Long-term follow-up of primary glaucoma surgery with Ahmed glaucoma valve implant versus trabeculectomy. Am J Ophthalmol 2003;136:464-70.
  20. Singh K, Mehta K, Shaikh N, et al. Trabeculectomy with intraoperative mitomycin C versus 5-fluorouracil; prospective randomized clinical trial. Ophtalmology 2000;107:2305-9.
  21. Wudunn D, Cantor L, Palanca-Capistrano A, et al. A prospective randomized trial comparing intraoperative 5-fluorouracil vs mitomycin C in primary trabeculectomy. Am J Ophthalmol 2002;134:521-8.
  22. Gedde S, Schiffman J, Feuer W, et al. Treatment outcomes in the Tube Versus Trabeculectomy study after one year of follow-up. Am J Ophthalmol 2007;143:9-22.
  23. Goulet RJ, 3rd, Phan AD, Cantor LB, Wudunn D. Efficacy of the Ahmed S2 Glaucoma Valve Compared with the Baerveldt 250-mm2 Glaucoma Implant. Ophthalmology 2007.
  24. Health Care Insurance Board (CVZ). Pharmacotherapeutic compass (Farmacotherapeutisch kompas). Available at: www.fk.cvz.nl. Accessed: December 2007
  25. Foundation for pharmaceutical statistics (Stichting Farmaceutische Kengetallen). Data and facts 2007 (Data en feiten 2007).
  26. Oostenbrink J, Bouwmans C, Koopmanschap M, Rutten F. Manual for costing research (Handleiding voor kostenonderzoek; Methoden en standaard kostprijzen voor economische evaluaties in de gezondheidszorg.). Diemen: Health Care Insurance Board (CVZ), 2004.
  27. Oostenbrink J, Rutten-van Mölken M, Opdenoordt T. The treatment of newly diagnosed patients with glaucoma or with ocular hypertension in the Netherlands: an observational study of costs and initial treatment success based on retrospective chart review. Doc Ophthalmol 2000;98:285-99.
  28. Peeters A, Schouten JS, Webers CA, et al. Cost-effectiveness of early detection and treatment of ocular hypertension and primary open-angle glaucoma by the ophthalmologist. Eye 2008;22(3):354-62.
  29. Van Gestel A, Webers C, Beckers H, et al. The relationship between visual field loss in glaucoma and health-related quality-of-life. Eye 2010;In press.
  30. Van Gestel A, Webers C, Beckers H, et al. Does every bit of visual field loss count? World Glaucoma Congress. Boston: World Glaucoma Association, 2009.
  31. Dutch Healthcare Authority (Nederlandse Zorgautoriteit). Maximum tarifs (Tariefbeschikking maximumtarieven extramurale zorg in het tweede en derde compartiment ). 2007.
  32. Drug information system. GIP database. Available at: www.gipdatabank.nl. Accessed: 2006
  33. Webers C, Beckers H, Nuijts R, Schouten J. Pharmacological management of primary open-angle glaucoma: second line options and beyond. Drugs Aging 2008;25(9):729-59.
  34. Rolim de Moura C, Paranhos A, Jr., Wormald R. Laser trabeculoplasty for open angle glaucoma. Cochrane Database Syst Rev 2007(4):Cd003919.
  35. Committee on Ophthalmic Procedure Assessments. Ophthalmic Procedure Assessment: Laser Trabeculoplasty for Primary Open-angle Glaucoma. Ophthalmology 1996;103(10):1706-12.
  36. Van der Valk R. Thesis: Glaucoma medication; evidence from clinical trials and effects in practice. University Eye Clinic. Maastricht: University of Maastricht, 2005.
  37. Medeiros F, Weinreb R, Sample P, et al. Validation of a predictive model to estimate the risk of conversion from ocular hypertension to glaucoma. Arch Ophthalmol 2005;123:1351-60.
  38. Feiner L, Piltz-Seymour J. Collaborative Initial Glaucoma Treatment Study: a summary of results to date. Curr Opin Ophthalmol 2003;14:106-11.
  39. Lee YH, Kim CS, Hong SP. Rate of visual field progression in primary open-angle glaucoma and primary angle-closure glaucoma. Korean J Ophthalmol 2004;18(2):106-15.
  40. Mayama C, Araie M, Suzuki Y, et al. Statistical evaluation of the diagnostic accuracy of methods used to determine the progression of visual field defects in glaucoma. Ophthalmology 2004;111:2117-25.
  41. Mikelberg FS, Schulzer M, Drance SM, Lau W. The rate of progression of scotomas in glaucoma. Am J Ophthalmol 1986;101(1):1-6.
  42. Oliver JE, Hattenhauer MG, Herman D, et al. Blindness and glaucoma: a comparison of patients progressing to blindness from glaucoma with patients maintaining vision. Am J Ophthalmol 2002;133(6):764-72.
  43. Pereira ML, Kim CS, Zimmerman MB, et al. Rate and pattern of visual field decline in primary open-angle glaucoma. Ophthalmology 2002;109(12):2232-40.
  44. Rasker M, Van den Enden A, Bakker D, Hoyng P. Rate of visual field loss in progressive glaucoma. Arch Ophthalmol 2000;118:481-8.
  45. Schwartz B, Takamoto T, Martin J. Increased Rate of Visual Field Loss Associated with Larger Initial Visual Field Threshold Values on Follow-Up of Open-Angle Glaucoma. J Glaucoma 2004;13(2):120-9.
  46. Soares AS, Artes PH, McCormick TA, et al. Retinal arterial diameter changes in progressive and nonprogressive glaucoma. J Glaucoma 2003;12(3):243-9.
  47. Vesti E, Johnson C, Chauhan B. Comparison of different methods for detecting glaucomatous visual field progression. Invest Ophthalmol Vis Sci 2003;44:3873-9.
  48. Wilson M, Kosoko O, Cowan C, et al. Progression of visual field loss in untreated glaucoma patients and glaucoma suspects in St. Lucia, West Indies. Am J Ophthalmol 2002;134:399-405
  49. Zink JM, Grunwald JE, Piltz Seymour J, et al. Association between lower optic nerve laser Doppler blood volume measurements and glaucomatous visual field progression. Br J Ophthalmol 2003;87(12):1487-91.
  50. Kwon Y, Kim C, Zimmerman B, et al. Rate of visual field loss and long-term visual outcome in primary open-angle glaucoma. Am J Ophthalmol 2001;132:47-56.
  51. O'Brien C, Schwartz B, Takamoto T, Wu DC. Intraocular pressure and the rate of visual field loss in chronic open-angle glaucoma. Am J Ophthalmol 1991;111(4):491-500.
  52. Leske M, Heijl A, Hussein M, et al. Factors for glaucoma progression and the effect of treatment. Arch Ophthalmol 2003;121:48-56.
  53. Nouri-Mahdavi K, Hoffman D, Coleman A, et al. Predictive factors for glaucomatous visual field progression in the Advanced Glaucoma Intervention Study. Ophthalmology 2004;111:1627-35.
  54. Wesselink C, Heeg G, Jansonius N. Glaucoma monitoring in a clinical setting: Glaucoma Progression Analysis versus Nonparametric Progression Analysis. Arch Ophthalmol 2009;127(3):270-4.
  55. The AGIS investigators. The Advanced Glaucoma Intervention Study (AGIS): 7. The relationship between control of intraocular pressure and visual field deterioration. Am J Ophthalmol 2000;130:429-40.
  56. Lichter P, Musch D, Gillespie B, et al. Interim clinical outcomes in the Collaborative Initial Glaucoma Treatment Study comparing initial treatment randomized to medications or surgery. Ophthalmology 2001;108:1943-53.
  57. Shirakashi M, Iwata K, Sawaguchi S, et al. Intraocular pressure-dependent progression of visual field loss in advanced primary open-angle glaucoma: a 15-year follow-up. Ophthalmologica 1993;207(1):1-5.
  58. Singh K, Shrivastava A. Early aggressive intraocular pressure lowering, target intraocular pressure, and a novel concept for glaucoma care. Surv Ophthalmol 2008;53 Suppl1:S33-8.
  59. Hendrikse F, La Heij E. Eye diseases. Scope of the problem. How often do eye diseases occur? Available at: http://www.rivm.nl/vtv/object_document/o1143n17763.html. Accessed: 18 May, 2006
  60. Burr J, Azuara-Blanco A, Avenell A. Medical versus surgical interventions for open angle glaucoma. The Cochrane Database of Systematic Reviews. 2004:Issue 2. Art.No.: CD004399.pub2. DOI: 10.1002/14651858. CD004399.pub2.
  61. Ederer F, Gaasterland D, Dally L, et al. The Advanced Glaucoma Intervention Study (AGIS):
    Comparison of treatment outcomes within race: 10-year results. Ophthalmology 2004;111:651-64.
  62. Statistics Netherlands (CBS). Mortality rates by age and gender. Available at: http://statline.cbs.nl/StatWeb/. Accessed: 12-07-2005
  63. Miglior S, Zeyen T, Pfeiffer N, et al. Results of the European Glaucoma Prevention Study. Ophthalmology 2005;112:366-75.
  64. Heeg GP, Jansonius NM. The groningen longitudinal glaucoma study III. The predictive value of frequency-doubling perimetry and GDx nerve fibre analyser test results for the development of glaucomatous visual field loss. Eye 2008.
  65. The AGIS investigators. The Advanced Glaucoma Intervention Study (AGIS): 11. Risk factors for failure of trabeculectomy and argon laser trabeculoplasty. Am J Ophthalmol 2002;134(4):481-98.
  66. Statistics Netherlands (CBS). Statline. Available at: http://statline.cbs.nl/StatWeb/. Accessed: 2010
  67. Oostenbrink J, Rutten-van Mölken M, Sluyter-Opdenoordt T. Resource use and costs of patients with glaucoma or ocular hypertension: a one-year study based on retrospective chart review in the Netherlands. J Glaucoma 2001;10:184-91.
  68. Meads C, Hyde C. How much is the cost of visual impairment: caveat emptor. Pharmacoeconomics 2006;24(2):207-9; discussion 10.
  69. Burr J, Mowatt G, Hernández R, et al. The clinical effectiveness and cost-effectiveness of screening for open angle glaucoma: a systematic review and economic evaluation. Health Technol Assess 2007;11(41).

 


Value in Health Article Supplementary Information
 

Contact ISPOR @ info@ispor.org  |  View Legal Disclaimer
©2014 International Society for Pharmacoeconomics and Outcomes Research. All rights reserved under International and Pan-American Copyright Conventions. 
Website design by Eagle Systems USA, Inc.