MARGINAL STRUCTURAL MODEL FOR STUDYING THE CAUSAL EFFECT BETWEEN VASO-OCCLUSIVE CRISES AND OCCURENCE OF DEATH OR COMPLICATIONS IN THE SICKLE-CELL DISEASE PATIENTS
Author(s)
Herquelot E1, Morgan G2, Lamarsalle L1, Raguideau F1, Burke T3, Brown S4, Bonner A4, Bailey M5, Disher T4, Abioye A5
1HEVA, Lyon, France, 2HCD Economics, Daresbury, WRT, UK, 3HCD Economics, Daresbury, UK, 4Cornerstone Research Group Inc., Burlington, ON, Canada, 5Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA
Presentation Documents
OBJECTIVES : Sickle-cell disease (SCD) is one of the most prevalent inherited diseases. Previous research suggests a relationship between Vaso-Occlusive Crises (VOC), acute painful episodes, and death. However, the time-varying effect of VOC on these outcomes prevents the establishment of a causal relationship using classically adjusted models. This study assessed the effect of VOC in the previous year on occurrence of death and the frequently reported complications using an Inverse Probability Weighted Marginal Structural Model (IPW-MSM). METHODS : Prevalent patients with SCD between 2008 and 2017, their vital status and complications were identified using ICD-10 diagnosis codes recorded in NHS England's Hospital Episode Statistics database linked to the Office for National Statistics Mortality Data. The causal effect of VOC categorized into 3 groups (0,1-2,3+) on each complication and death was estimated using IPW-MSM Cox models approximated by weighted pooled logistic regressions. The weights were estimated using the probability of being in each VOC category with a multinomial logistic model, and the probability of being censored with a logistic model. These models were adjusted for age, gender, ethnicity, complications and comorbidities. RESULTS : A total of 15,076 SCD patients were included with the occurrence of 1,141(8%) deaths, 4,097(27%) Acute Chest Syndrome, 1,571(10%) gallstones, 1,295(9%) Avascular Necrosis, 1,098(7%) sepsis and 1,037(7%) cardiomegaly during the follow-up. The Hazard Ratios of 1-2 VOCs versus 0 VOC and 3+VOCs versus 0 VOC were estimated at 1.44(95% CI:1.15;1.81) and 4.50(2.91;6.97) for death, at 2.98(2.64;3.37) and 5.51(3.67;8.27) for ACS, at 2.60(1.78;3.81) and 2.20(1.43;3.41) for gallstones, at 1.86(1.33;2.58) and 2.07(1.05;4.09) for AVN, at 1.91(1.45;2.52) and 2.11(1.21;3.67) for sepsis and at 1.36(1.04;1.77) and 2.48(1.54;3.99) for cardiomegaly. CONCLUSIONS : Experiencing 3+ VOCs per year are associated with a higher risk of death and complications. Reducing the number of VOCs experienced or preventing patients from reaching 3 VOCs may significantly reduce the occurrence of death and most complications.
Conference/Value in Health Info
2019-11, ISPOR Europe 2019, Copenhagen, Denmark
Code
PSY38
Topic
Epidemiology & Public Health, Methodological & Statistical Research, Real World Data & Information Systems
Topic Subcategory
Confounding, Selection Bias Correction, Causal Inference, Health & Insurance Records Systems, Public Health
Disease
Systemic Disorders/Conditions