DEVELOPMENT OF AN ALGORITHM THAT DIFFERENTIATES AMONG RISK FACTORS, COMORBIDITIES, AND CONSEQUENCES OF DISEASE IN PEER-REVIEWED PUBLICATIONS
Author(s)
Parker G1;Carman W2, Zarotsky V*1 1Optum, Eden Prairie, MN, USA, 2Optum, Ann Arbor, MI, USA
Presentation Documents
OBJECTIVES: A common limitation of the peer-reviewed literature is failure to establish whether a condition precedes or follows disease diagnosis. If disease diagnosis comes first, the condition is either a comorbidity or consequence of the disease, whereas if the condition precedes disease diagnosis, it may be a risk factor. Our objective was to develop an algorithm for appropriate classification of risk factors, comorbidities, and consequences of disease to enable accurate assessment of the literature. METHODS: We established the following procedure to identify risk factors in articles retrieved through our literature search: 1) exclude cross-sectional studies as potential sources of risk factor data because they cannot establish the necessary temporal sequence; 2) if the terms ‘risk factor’ or ‘incidence’ are present in the remaining articles, include only those that: a) report on a population followed over time; b) contain baseline data indicating absence of signs/symptoms of the disease of interest at study onset; and c) have conducted statistical analyses demonstrating associations between individual baseline factors and the disease. If these requisites are not met, signs/symptoms present at study onset should be classified as comorbid conditions. Conditions arising as a result of the condition may be classified as disease consequences. RESULTS: We systematically applied our algorithm to 100 peer-reviewed articles in clinically focused journals retained for inclusion after screening. The algorithm allowed for accurate classification of risk factors (ie, underlying conditions that predispose a patient to development of the disease), comorbidities (ie, conditions that complicate disease progress), and consequences of the disease (ie, events shown to be statistically related to the disease and occurring after disease onset). CONCLUSIONS: We have developed an algorithm that accurately differentiates among risk factors, comorbidities, and consequences of a disease. This tool will aid in the accurate assessment of clinical literature when conducting systematic reviews.
Conference/Value in Health Info
2013-05, ISPOR 2013, New Orleans, LA, USA
Value in Health, Vol. 16, No. 3 (May 2013)
Code
PRM19
Topic
Methodological & Statistical Research
Topic Subcategory
Confounding, Selection Bias Correction, Causal Inference
Disease
Multiple Diseases