PURPOSE: Budget impact analyses (BIA) describe the changes in short-term intervention-related and disease-related costs of new health care technologies. BIA are increasingly required for budgetary planning by decision-makers. Our study systematically reviewed published BIA for cancer screening programs, specifically applied methods, and the degree to which international BIA guidelines are followed in studies evaluating cancer screening programs.
METHOD(S): A systematic literature search was conducted in MEDLINE and EconLit for BIA evaluating cancer screening programs, published in English language during 2010-2016. Standardized evidence tables were used to extract study characteristics as outlined in the ISPOR BIA Task Force Guidelines including cancer type, model structure and assumptions, definition of population size/characteristics, perspective, time horizon, included costs, source of epidemiologic and clinical data, consideration of health impact, validation, and uncertainty analysis.
RESULT(S): Ten studies were identified evaluating screening for cancer of breast (n=3), colorectal (n=2) cervical (n=2), lung (n=1), prostate (n=1) and the skin (n=1). Applied model designs varied from different types of decision-analytic models (60%) to simple cost calculators (40%). Simple cost calculators are recommended by ISPOR guidelines as long as important conditions are credibly captured. Complex multiple-cohort models were mainly applied in the reviewed studies that combined cost-effectiveness and BIA (n=3). The time horizon ranged from one year (n=6) to 20 years (n=3). BIA framework should allow for calculating shorter and longer time horizons. Reporting both, annual results and cumulative results may be useful for planning annual expenditures and for showing the net effects on total costs over time because of delays in offsetting disease costs savings. However, not all studies provided this information. All studies included direct condition-related costs and two studies additionally included indirect cost. Health impact on patients was reported in 40% of the studies. Uncertainty analysis was limited in most of the studies. Only 40% of the studies validated their results.
CONCLUSION(S): Our review highlights a considerable variability in the extent to which the BIA studies evaluating cancer screening programs followed recommended guidelines. For example, most BIA failed to report projections beyond one-year or to report the model validation. To ensure high quality and sound decision support, best practice recommendations should be followed more rigorously in all key aspects.
