A large health economic evaluation of a medication-related clinical decision support system (CDSS) used in German general practitioner practices has found no statistically significant effect on total healthcare costs, and substantial uncertainty about whether the technology represents good value for money. The study, published in BMC Digital Health, was conducted alongside a stepped-wedge cluster-randomized controlled trial known as AdAM, which enrolled patients with polypharmacy in primary care practices in Westphalia-Lippe, Germany, between June 2017 and March 2021.
Polypharmacy, defined in the study as the simultaneous use of five or more medications, is increasingly common worldwide as populations age and multimorbidity rises. Older individuals with polypharmacy face an elevated risk of medication errors, potentially inappropriate prescribing, and adverse drug events. Potentially inappropriate prescribing is significantly associated with all-cause mortality, hospital readmission, and increased healthcare costs. In Germany, costs of adverse drug events emerging from outpatient care have been estimated at 816 million euro per year, making medication management a substantial economic burden for health systems.
The intervention tested in the trial was a web-based CDSS that gave general practitioners access to patient information drawn from statutory health insurance claims data, including diagnoses, ambulatory services, hospitalizations, and pharmacy prescription data. Physicians could also enter or update patient-related information such as weight and kidney function. The system assessed medication appropriateness and alerted GPs to drug-drug interactions, duplicated medications, and potentially inappropriate medications, providing supporting information on adverse drug reactions. GPs in the control phase had no access to the system. Patients were eligible if they were insured by BARMER, were at least 18 years old, and had prescriptions for at least five distinct medications in at least one quarter of the previous year.
The economic evaluation used claims data from BARMER, a large nationwide German statutory health insurer covering more than eight million individuals, and measured outcomes at the patient-quarter level for up to 14 quarters per individual. Quarterly costs included outpatient services, inpatient hospital services, pharmaceuticals, medical aids and remedies, rehabilitation, nursing services, nursing home care, and incapacity for work. Intervention costs were captured through three special billing codes introduced to reimburse the intervention. The analysis adopted a payer’s perspective, in line with the methods of the Institute for Quality and Efficiency in Health Care, and costs and effects were not discounted.
The study sample comprised 15,328 patients in the intervention group and 11,912 in the control group, with closely matched characteristics. Mean age was 71.9 years in the intervention group and 71.8 years in the control group, women constituted roughly 62 percent of both groups, and most patients had no nursing care level. In the descriptive comparison, mean costs per patient per quarter were slightly lower in the intervention group for outpatient services, inpatient hospital services, pharmaceuticals, home nursing care, and sick pay, while costs for medical aids and remedies and additional nursing services were slightly lower in the control group.
After adjusting for confounding factors including age, sex, nursing care level, previous year’s total healthcare costs, and a medication-based chronic disease score, the intervention was associated with an estimated change of minus 7.2 euro in total costs per patient per quarter, with a 95 percent confidence interval from minus 109.7 to 95.3 euro and a p-value of 0.891, indicating no statistically significant difference. Outpatient costs showed a small but statistically significant increase of 9.3 euro (95 percent CI: 1.9 to 16.6; p = 0.014), while inpatient costs (minus 19.9 euro; p = 0.581) and pharmaceutical costs (minus 15.8 euro; p = 0.238) were non-significantly lower in the intervention group.
Three sensitivity analyses tested the robustness of these findings. Including inactive GP practices yielded an estimate of minus 23.3 euro, restricting the sample to the period before the first COVID-19 lockdown in early 2020 yielded 0.6 euro, and limiting the sample to participants with a nursing care level of two or below yielded minus 55.9 euro. None of these analyses was statistically significant, and the authors concluded that none altered the main finding that the intervention was not associated with a significant change in total healthcare costs.
The cost-effectiveness analysis combined the cost estimates with the trial’s primary endpoint, a composite of all-cause mortality and all-cause hospitalizations per quarter, which had shown no significant difference between groups, with a risk ratio of 1.00. Because no accepted willingness-to-pay thresholds exist for the cost per percentage point reduction in the risk ratio, the researchers explored a range of hypothetical thresholds using non-parametric bootstrapping with 1,000 replications. Most bootstrap replications clustered near the origin of the cost-effectiveness plane, and 49.2 percent fell in the lower-left quadrant, indicating the intervention tended to be slightly less effective and less costly than usual care. The probability of cost-effectiveness rose with higher willingness-to-pay thresholds and stabilized around 60 percent above 250 euro per percentage point reduction in the risk ratio, never exceeding 62 percent at the highest threshold tested, which the authors interpreted as indicating considerable remaining uncertainty.
The trial itself had not demonstrated a statistically significant effect on its combined primary outcome, and secondary outcomes such as hospitalizations, mortality, and high-risk prescriptions also did not differ between groups. The authors point to limited uptake of the CDSS as a likely constraint on effectiveness. A quantitative process evaluation conducted alongside the trial found that while GPs did address inappropriate prescriptions identified by the system, the frequency and intensity of use was lower than expected. The authors suggest that the COVID-19 pandemic, which extensively disrupted workflows in GP practices, and the inability to integrate the CDSS with practices’ electronic medical records due to technical constraints within the German healthcare system likely impeded implementation. Organizational readiness for change has also been shown to influence physician adoption behavior.
The authors conclude that the intervention generated neither measurable improvements in clinical outcomes nor clear economic advantages, and that these findings are consistent with recent evidence showing uncertain economic benefits of CDSS interventions. They note that the study relied on administrative claims data collected primarily for billing purposes, adopted a payer perspective that excluded implementation costs such as software purchase, training, and maintenance, and used an incremental cost-effectiveness ratio that limits interpretability for real-world decision-making. Future research, they argue, should integrate clinical, economic, and implementation perspectives, following the Medical Research Council guidance for complex interventions, to better understand the contextual factors and mechanisms under which CDSS interventions can generate their full impact in primary care.
Subject of Research: Costs and cost-effectiveness of a medication-related clinical decision support system for patients with polypharmacy in German primary care
Article Title: Costs and cost-effectiveness of a medication-related clinical decision support system in patients with polypharmacy in a primary care setting: a health economic evaluation alongside a stepped-wedge cluster-randomized controlled trial
Article References: Lampe, D., Surmann, B., Grosser, J., Greiner, W., AdAM Study Group, Kellermann-Mühlhoff, P., Düvel, L., Beckmann, T., Hammerschmidt, R., Jachmich, J., Leicher, E., Brandt, B., Richard, J., Meyer, F., Flume, M., Müller, T., Muth, C., Gerlach, F. M., Müller, B. S., … Grandt, S. (2026). Costs and cost-effectiveness of a medication-related clinical decision support system in patients with polypharmacy in a primary care setting: a health economic evaluation alongside a stepped-wedge cluster-randomized controlled trial. BMC Digital Health, 4(1), Article 73. https://doi.org/10.1186/s44247-026-00308-2
Image Credits: AI Generated
DOI: 10.1186/s44247-026-00308-2
Keywords: clinical decision support system, polypharmacy, primary care, cost-effectiveness, healthcare costs, medication review, cluster-randomized trial, Germany, statutory health insurance, potentially inappropriate prescribing, general practitioners, health economics
News Source: Denise Maddox. (October 28, 2026). Digital medication alerts in German primary care show no clear cost savings, trial finds. Scienmag.
