Can a structured electronic medical record with decision-making support improve nursing home quality? Healthcare administration through structured records

Background: Nursing homes face challenges in the coming years due to the increased number of elderly. A new law in force from Jan 2012 (“Samhandlingsreformen”) places more responsibilities on the counties running the nursing homes. Quality will come under pressure, expectations of services will rise and clinical complexity will grow. New strategies are needed to meet this situation. Modern clinical information systems with decision-making support may be part of that. In addition, knowledge about the prevalence of clinical conditions among long-term patients in nursing homes is poor, and research on this population is needed. Objectives: We wished to define the clinical and practical parameters needing to be improved among long-term patients in nursing homes, which could then be used as endpoints in an intervention study. We then wanted to test if a structured electronic medical record system with decision-making support improved the quality of the endpoints. Methods: First we performed a literature search study on structured medical records. We then developed a full-scale, semi-structured, interdisciplinary electronic medical record system with extensive decision support options and conducted four studies, three to define endpoints and gain medical knowledge about the nursing home population: “Hospitalizations from nursing homes”, “Psychoactive drugs in 7 nursing homes” and “Atrial fibrillation and heart failure in nursing homes”. Then we performed an intervention study, “Can electronic tools improve nursing home quality?” Results: Installing the information system in seven new nursing homes proved easier than expected. After four months’ training the nursing homes switched to the new system and used it as the only medical record system on a daily basis for the next 12 months (February 2008 – February 2009). We discovered a seriously low warfarin treatment rate (14%) to patients with atrial fibrillation (N = 90), considerable treatment rate differences between institutions regarding use of neuroleptics (18 – 55%) and the proportion of patients not weighed for the last 30 days was 72.6%. The proportion of patients taking neuroleptics was reduced from 33.0% to 21.5% (N = 183 before/205 after, chi-square test, p = 0.015), i.e. a difference of 11.5% (95% CI: 2.3 to 20.6%). Warfarin increased from 3.0% to 9.8% (p = 0.013), i.e. a difference of 6.8% (95% CI: 1.6 to 12.1%). The internal controls did not change: use of digitoxin did not increase significantly (8.0% vs. 8.5%; p = 0.1), thyroxin was not reduced (10.0% vs. 8.6%, p = 0.765) and antidiabetics did not increase (10.0% vs. 10.5%; p = 0.996). The proportion of patients not weighed for the last 30 days was reduced from 72.6% to 16.0% (p < 0.001), i.e. a difference of 56.6% (95% (CI: 47.5 to 64.5%)). Conclusions: There exist treatment differences among nursing homes. Research showing consequences for patients is pending. The electronic medical record system with integrated decision-making support may be a way to improve quality. The present material is too small for firm conclusions however. The application should be tested in multiple medical settings, and may provide a route from pure economic to more scientific healthcare governance, as management data can be produced through daily work without time-consuming and costly additional projects and can be monitored electronically on a continuous basis. This may have relevance to New Public Management, which so far have had shortcomings regarding valid quality parameters. We introduce the idea of "health administration through structured records" (HATS)