Figure. Graphic representation of data flow through the ISTOP monitoring system. (1) At the time of prescription, primary care physicians receive electronic prompt identifying eligible patients; this information is based on a query of the patient's prescription history within the clinical data repository. (2) When the patient's eligibility is confirmed and the patient consents, primary care physicians enroll consenting patients into the program. (3) The system's electronic health record (Medical Office of the 21st Century web-based electronic prescribing and integrated drug management [MOXXI] system6) passes data to the interactive voice response system (IVRS) to queue automated follow-up calls. (4) The IVRS calls patients and administers the questionnaire on days 3 and 17 after the prescription; the questionnaire consists of 4 simple questions soliciting “yes” or “no” answers (med indicates the subject prescription medication). (5) The IVRS passes patient response data to MOXXI to update the electronic health record. (6) The IVRS sends an e-mail to the study pharmacist if responses by the patient indicate that follow-up is required. (7) The pharmacist personally contacts patients and documents information in the MOXXI system.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Forster AJ, Auger C, ISTOP ADE Investigators FT. Using Information Technology to Improve the Monitoring of Outpatient Prescribing. JAMA Intern Med. 2013;173(5):382–384. doi:10.1001/jamainternmed.2013.2002
Author Affiliations: Performance Measurement, The Ottawa Hospital, Ottawa, Ontario, Canada (Dr Forster); Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa (Dr Forster); Department of Epidemiology and Community Medicine (Dr Forster) and Department of Medicine, Faculty of Medicine (Dr Forster), University of Ottawa, Ottawa; Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada (Dr Forster); Clinical and Health Informatics Research Group, Montreal, Ontario, Canada (Dr Auger); and Epidemiology and Biostatistics Department, Faculty of Medicine, McGill University, Montreal (Dr Auger).
Adverse drug events (ADEs) and medication nonadherence are common and reduce the potential benefit of medications. Adverse drug events, defined as poor health outcomes caused by medications,1 occur in up to 25% of ambulatory care patients prescribed a medication.2Medication nonadherence, defined as patients not taking their medications as directed,3 can occur in 25% of new prescriptions.2,4,5 Improved monitoring and communication could reduce ADEs and nonadherence to minimize medication-associated problems.
We designed the ISTOP-ADE system (Figure), an information technology–based approach to monitor ambulatory patients receiving incident prescriptions. This system automatically called patients 3 and 17 days following a prescription and allowed patients to request a pharmacist phone call. The purpose of this study was to determine this approach's potential effectiveness.
We conducted a prospective cohort study of patients receiving incident prescriptions from 1 of 76 primary care physicians in Montreal and Quebec City, Canada. Practices were selected if they used the Medical Office of the 21st Century web-based electronic prescribing and integrated drug management system.6 Eligible patients received an incident prescription for at least 1 new medication. We excluded patients who could not speak English or French; had a telephone number with an extension; or were deemed inappropriate for the study by their primary care physician.
Twenty-one days after their prescription, we contacted all patients and used a structured interview to elicit information about medication use, symptoms, changing health state, and health care visits. We then collated these responses with electronic health record and automated call data into case record forms. Using standard methods, 2 physicians independently reviewed cases to determine whether patients experienced ADEs or primary nonadherence.4,5,7,8
We recruited 628 patients from 76 practices (median patient age, 66 years; interquartile range, 57-73 years). The 3 most common medication classes were antilipemic agents (n = 146; 22.2%), antidiabetic agents (n = 122; 18.5%), and inhaled corticosteroids (n = 91; 13.8%). We completed the study protocol on 568 of 628 patients (90%) and found no statistically significant differences between completers and noncompleters.
On the day-3 and day-17 calls, the system successfully connected with 465 (82%) and 475 patients (84%), respectively. Of patients connecting to the system, one-third required follow-up from the pharmacist. On the day-3 contact, the most common reason for pharmacist assistance was the patient not starting to take the medication (n = 65; 44%); at day 17, the most common reason was new problems starting since the patient started taking the medication (n = 81; 50%).
Overall, we identified 125 ADEs experienced by 125 patients (22% [95% CI, 19%-26%]). The ISTOP-ADE system identified 58 of 125 ADES (46.4% [95% CI, 38%-55%]). In 23 of the 58 patients who experienced an ADE (40% [95% CI, 27%-53%]), the pharmacist modified the patient's medication regimen in response to their interaction. For patients whose ADE was identified by the ISTOP-ADE system, 20 of 58 had symptoms lasting 7 days or longer (35%); in contrast, patients whose ADE was not identified by the system, 33 of 67 had symptoms that lasted 7 days or longer (49%) (P = .04). Thirty-three patients reported primary nonadherence (5.8% [95% CI, 4%-8%]), of which the system identified and potentially influenced the management of 10 (30%).
The ISTOP-ADE system successfully connected with over 80% of patients, one-third of whom requested personal contact with a pharmacist. The system identified 46% of ADEs and influenced the management of 40% of these. Those ADEs identified by our system were of shorter duration than those not identified. Finally, the system identified one-third of all primary nonadherence events. It is possible that the identification of ADEs and nonadherence events led to modifications in treatment that had positive health benefits.
Our approach for improving safety was to monitor all patients. When we identified patients experiencing an ADE or being nonadherent, we were able to intervene in a timely manner by connecting them to a pharmacist who called to personally provide help. We believe that this approach reduced the duration of ADEs, most of which were not preventable.
It is notable that our intervention did not identify all ADEs and primary nonadherence. We believe that certain enhancements could improve the system, including an inbound function to allow patients to call, an online data entry form for patients, improved education, and improved scripting and dialogue design.
Because our study was nonexperimental, one should not conclude that our intervention will improve patient health status. However, these preliminary results suggest a potential benefit and warrant further study.
In conclusion, our intervention allowed successful monitoring of most patients and identified many medication-related problems. The system led to important treatment modifications in a significant proportion of patients, suggesting that it has the potential to positively impact health outcomes. Before widespread implementation, we recommend controlled clinical trials.
Correspondence: Dr Forster, The Ottawa Hospital, Civic Campus, 1053 Carling Ave, Box 684, Administrative Services Bldg, 1st Floor, Room 1019, Ottawa, ON K1Y 4E9, Canada (email@example.com).
Published Online: February 4, 2013. doi:10.1001 /jamainternmed.2013.2002. Corrected March 11, 2013.
Author Contributions:Study concept and design: Forster. Acquisition of data: Forster. Analysis and interpretation of data: Forster and Auger. Drafting of the manuscript: Forster and Auger. Critical revision of the manuscript for important intellectual content: Forster and Auger. Statistical analysis: Forster. Obtained funding: Forster. Study supervision: Forster and Auger; Research coordination: Auger.
The ISTOP ADE Investigators: Alan J. Forster, MD, MSc; Natalie Oake, MSc; Claudine Auger, PhD; Maria Ali, MD; David Buckeridge, MD, PhD; Tewodros Eguale, MD, PhD; Carl van Walraven, MD, MSc; and Robyn Tamblyn, PhD.
Conflict of Interest Disclosures: None reported.
Funding/Support: Research for this article was supported by grant MOP 97843 from The Canadian Institute for Health Research.
Role of the Sponsor: None of the funders were involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, and approval of the manuscript.