Swiss claims data : opportunities and challenges for observational research

Health care utilization databases such as claims data are frequently used in a variety of settings to analyze the outcome of drug treatments. Because of their large scale these data allow to study real-world effectiveness and utilization patterns at relatively low costs. The Swiss healthcare system is notable for its high degree of innovative capacity, which also involves new developments and improvements in the pharmaceutical industry. In collaboration with Helsana, we have recently used claims data for descriptive analyses of health resource utilization in Switzerland. The aim of this thesis was to contribute to the understanding of the Helsana claims data provided by the Helsana Group and to show the opportunities and challenges of these data with the following three projects.
Using claims data from the Helsana Group, Project I examined the association between statin use and the risk of cholecystectomy in a case-control analysis between 2013 and 2014. We applied conditional logistic regression analyses to calculate odds ratios (ORs) with 95% confidence intervals (CIs) and adjusted the analyses for history of cardiovascular diseases and for use of estrogens, fibrates and other lipid-lowering agents. The study supports the previously raised hypothesis that long-term statin use was associated with a reduced OR (adjusted OR [aOR] 0.77, 95% CI: 0.65-0.92). However, neither short-term current use nor past statin use affected the risk of cholecystectomy.
In a second case-control study, Project II examined the association between previous benzodiazepine use and the risk of developing Alzheimer’s disease (AD) and identified 1,438 incident AD cases between 2013 and 2014. Because initiation of benzodiazepine use shortly before the AD diagnosis date may occur due to symptomatic treatment of prodromal symptoms of early major neurocognitive disorder, we introduced an induction period of two years before the AD diagnosis date. We applied conditional logistic regression analyses to calculate ORs with 95% CIs and adjusted for antidepressant use. After accounting for benzodiazepine use initiated during the prodromal phase, long-term benzodiazepine use was not associated with an increased risk of developing AD (aOR 0.78, 95% CI: 0.53-1.14).
In Project III, a retrospective descriptive study, we quantified use of oral and parenteral iron supplementation in Swiss data and compared it to data from the UK between 2012 and 2014. We further assessed the frequency of serum ferritin and hemoglobin tests prior to newly started iron therapy to see whether use was based on documented low iron levels or blood parameters, especially in the case of parenteral iron supplementation. The three-year prevalence of iron supplementation was 9.4% in Switzerland compared to 4.4% in the UK. Iron use increased slightly between 2012 and 2014 in both countries (CH +0.3%/UK +0.2%) and recorded parenteral iron administration was roughly a thousand times higher in Switzerland (1.9%) than in the UK in 2014. Hemoglobin values prior to a new parenteral iron therapy were relatively infrequent in Switzerland despite the required documentation of hemoglobin prior to therapy.
The Helsana claims data can be used for descriptive studies as well as to study pharmacoepidemiological hypotheses. Its strengths are the large size, the accurate documentation of the data and the low costs. On the other hand, data on important potential confounder’s variables such as Body mass index (BMI), smoking status or alcohol consumption but also on clinical diagnoses is largely missing and leading to bias and confounding.