Claims for adverse events: a predictive algorithm

Glòria Galvez

Strategies focused on encouraging patients' participation in the health system, and more specifically those related to quality and safety, have seen some a great deal of progress in recent years. A person-centred health system should promote active patient participation and use the complaints handled by patient care services as a specific instrument of participation. When the patient expresses the disagreement with the attention received, he or she is providing us with valuable information that is very useful in the continuous monitoring and improvement of quality. It doesn’t seem that there are many health institutions that use complaints and claims as a learning tool, but they rather use it as a mere descriptive statistic in the annual report of the organization, thus losing the opportunity for improvement that their analysis and monitoring would provide.

Dr. Gallagher, who, as someone with extensive experience in issues related to patient safety and disclosure of medical errors, has published an article in BMJ Quality & Safety: “Taking complaints seriously: using the patient safety lens” in which he proposes analysing complaints from a point of view of patient safety and treating them as if they were adverse events, in the same way as with the more traditional ones, such as those related to safe surgery or the appropriate use of medications. This is an innovative approach that will provide relevant information when proposing proactive interventions.

On the other hand, Spittal et al. (Dr. Spittal is a lecturer at the Melbourne School of Population and Global Health at the University of Melbourne) in an article recently published in the journal, “The PRONE score: an algorithm for predicting doctors' risks of formal patient complaints using routinely collected administrative data.” focuses on the study of complaints of the Australian health system. It points out that half of the claims processed in 10 years have been related to 3% of doctors. This relevant fact alerts us to the importance of analyzing the cases of professionals who receive complaints repeatedly, and to promote interventions that address the risk that some doctors have to go through a lawsuit.

In their study, 13,849 complaints were made by Australian health service patients over a 12-year period involving 8,424 doctors. Researchers rank claims based on certain risk factors that have shown the high likelihood that a doctor will receive a complaint. Using the multivariate logistic regression analysis, they have created a predictive risk algorithm that will allow us to estimate, with some reliability, the probability that a particular professional will receive a complaint in the two years following an initial complaint taken as reference.

There have been previous attempts to use simple algorithms that would allow an alert system to predict the risk of malpractice claims, such as that developed by Hickson et al: Patient at Risk Score (PARS), which indicates that using this tool represents not only a benefit for the organization but also for patient satisfaction.

The PRONE system (Predict Risk of New Event) is based on four variables: a) the doctor's specialty, b) the sex, c) the number of previous complaints, and d) the time elapsed since the last complaint. The results show that it can be a valid method for assessing the risk of recurrent complaints in doctors (70% of claims registered in the two years following a related complaint were related to the same doctors), easily extrapolated to other health professionals and other organizations but the most important is the feasibility of implementing interventions aimed at avoiding the adverse event by modifying the professionals’ work environment and behaviour. Spittal poses three possible interventions: 1) informing doctors who are at risk of receiving a future complaint, 2) training them in issues that usually arise in their complaint profile, and 3) citing the doctor to a medical oversight body who may assess the appropriate professional intervention for each situation.

The first step in reducing adverse events is to consider that there is a preventable harm. Getting to know the patients’ needs and systematically analyzing their complaints from a safety point of view will allow us to design predictive models that are essential in health management as they allow us to optimize resources, reduce costs and improve the health care system.