(115) A Critical Methodological Appraisal of Studies Assessing Computerized Clinical Decision Support Systems (CDSS) Success in Improving Prescription Safety in Patients with Low Levels of Kidney Function

Background: Clinical decision support systems (CDSS) integrated into computerized physician order entry (CPOE) systems are considered primary interventions to reduce medication errors during prescribing. However, their efficacy in reducing inappropriate prescribing in patients with low kidney function remains unclear.

Objectives: To critically review publications assessing the impact of CDSS integrated into CPOE systems specifically designed to improve drug dosing and avoid medications in patients with low levels of kidney function.

Methods: We conducted a systematic search in Medline, using keywords related to low kidney function, CDSS intervention, and outcomes of drug toxicities or inappropriate prescribing. Studies focusing on the research question were included in the review and evaluated across five domains: study design, sampling, measurements, statistical analyses, and the risk of bias. The first domain assessed designs, patient population, intervention details, and outcomes. The second domain discussed sampling and generalizability. Evaluation of exposure and outcome measurements, as well as methods of statistical analysis, were done in the third and fourth domains. The risk of bias was assessed using ROB 2 for trials and ROBINS-I for non-randomized intervention studies.

Results: Ten studies addressing the research question were included, including four randomized clinical trials (40%) and the remaining before-after (40%) or interrupted time-series designs (20%). Patient populations included in all the studies were individuals with low estimated glomerular filtration rate (eGFR), containing both acute kidney injury (AKI) and chronic kidney disease (CKD) patients together, for which the considerations of drug dosing may not be the same. All CDSSs were rule and algorithm-based systems, primarily developed in-house (90%), which may limit the generalizability of results to other CDSSs. Outcomes focused on inappropriate prescribing according to dosing references, with no consideration of clinically important outcomes such as drug toxicities. Notably, none of the studies mentioned the validation process of the CDSSs during their design. Risk of bias assessment revealed concerns regarding the randomization process (selection bias) and reporting of results (reporting bias) in all trials and a high risk of selective reporting bias in all non-randomized studies.

Conclusions: Despite CDSS in CPOE systems potentially mitigating medication errors, evidence on their efficacy in reducing inappropriate prescribing in patients with low kidney function is inconclusive. Future research should prioritize rigorous study designs, CDSS validation, and clinically significant outcomes for better evidence.