Exploring the Impact of the Prescription Automatic Screening System in Health Care Services: Quasi-Experiment (Preprint)

Abstract

Background: Hospitals have deployed various types of merging technologies (e.g., intelligent diagnosis systems, mobile tablets). However, the impacts of these technologies on behavior change are rather vague, especially the computer-aided diagnosis systems. Although the critical roles of information systems have been demonstrated in previous research, there still remains a substantial amount of unsatisfying results after the implementation of information systems in hospitals, such as the high rate of overrides relating to alerting information generated from Prescription Automatic Screening System (PASS). Objective: The objective of this study is to systematically explore the imperative role of PASS, an aided diagnosis system, on physicians' behavior change in order to clarify the various factors relative to the effective use of PASS. Methods: In order to systematically examine the various impacts of PASS, we adopted a quasiexperiment by using a two-year observation dataset from two hospitals. We then analyzed the data related to physicians' prescription behavior both before and after the deployment of PASS and eliminated the influences from a variety of perplexing factors by utilizing a control hospital which does not use a PASS system. In total, 754 physicians were included in this experiment composed of 11,054 patients, 400 physicians in the treatment group and 354 physicians in the control group, respectively. Moreover, this study was preceded by a series of interviews, which was employed to identify the moderators and we adopted propensity score matching (PSM) integrated with the difference-in-differences (DID) so as to isolate the impacts of PASS. Results: The impacts of PASS on prescription errors and medical treatment costs were all significant (Error: 95% CI -0.40-0.11; P=0.001; Costs: 95% CI -0.75--0.12; P=0.007), and help explain physician behavior over performance. Moreover, institutional pressure decreased the impact of PASS on the error of prescription (95% CI 0.07-0.55; P=0.012) and workload, respectively, thus reducing the impact of PASS on medical treatment costs (95% CI 0.18-0.39; P<0.001). We also suspect that other pressures (e.g., title and risk of illness) also impair physicians' attention to the intelligent alerting from PASS. However, the impacts of PASS did not change among physicians having a higher title or experience or when treating diseases demonstrating high risk. The researcher posits that this is because these physicians will focus more attention on their patients in these situations, regardless of having access to an intelligent system. Conclusions: Auxiliary diagnosis systems (such as PASS) do play a critical role in physicians' behavior change. However, workload and organizational pressure remain problematic in this context, since they will impair the positive impacts of auxiliary diagnosis systems on performance. This again highlights the importance of considering both technical and organizational issues to obtain the highest level of effectiveness when deploying IT in...