Planning for better care: the example of myocardial infarction

According to the World Health Organization, 17 million people die every year from heart attacks and strokes.

Jérôme Azé, University of MontpellierJessica Pinaire, University of MontpellierPaul Landais, University of Montpellier and Sandra Bringay, University of Montpellier

Care trajectories for patients with heart problems can be optimized. - Shutterstock

The institution points out that cardiovascular disease accounts for approximately 31% of all deaths worldwide, making it the leading cause of mortality. The burden of these diseases is set to rise to 11.06% by 2030, bringing the number of cases to around 36.2 million. Cardiovascular disease is a major public health issue, placing a heavy burden on the sector's resources.

In such a context, the fight against modifiable risk factors (smoking, sedentary lifestyle, overweight or obesity, high blood pressure, type 2 diabetes, excessively high cholesterol or triglyceride levels, etc.) is obviously essential. But there is also another important area for exploration: improving health planning, i.e. all the measures needed to improve the efficiency of the healthcare system.

Demonstrate this by characterizing the flow of patients hospitalized for myocardial infarction in France.




See also:
Cardiovascular risk factors, a revolutionary discovery not so long ago


Inequalities with serious consequences

The aim of health planning is not only to ensure the efficiency of the healthcare system, but also to reduce inequalities and geographical disparities in access to care.

These inequalities begin with territorial disparities in healthcare provision, the greatest of which concern the distribution of specialist physicians. The Île-de-France and PACA regions, for example, have high medical densities, whereas these are much lower in northern and central France. In rural areas in particular, healthcare provision is less extensive and unevenly distributed. What's more, 20% to 30% of hospital posts are vacant, and 30% of general practitioners are over 60 and can't find replacements.

These inequalities can have amajor impact on health: for example, one-third of French people who live a long way from health care services are forced to forego treatment. The greater the distance to healthcare, the greater the drop in the frequency of consultations, as well as the refusal to undergo certain tests. Difficulties are also encountered when emergency hospitalization is required (maternity, myocardial infarction, stroke, traumatology, etc.).

Tracing care trajectories for better planning

Inequalities also affect mortality risk. For example, France is the European country with the highest differences in risk of death before age 65 between people in manual and non-manual occupations.

One effective way of helping to improve this planning is to know more about the patient's care history: what care he or she has received, how often, etc. These "care trajectories" are seen as indicators of real patient needs. Considered as indicators of real patient needs, these "care trajectories" also provide information on the time and cost of care. We sought to gain a better understanding of these trajectories in the context of hospitalization for cardiac conditions.

The originality of our approach lies in the fact that we have studied national databases of medico-economic data from the Programme de médicalisation du système d'information (PMSI). The PMSI is managed by the Agence Technique de l'Information sur l'Hospitalisation. This public institution, which reports to the Ministry of Health, is a center of expertise in the four "fields" of hospital activity: medicine, surgery, obstetrics and dentistry; follow-up care and rehabilitation; psychiatry; and home hospitalization.

In France, each hospitalization entails the production of a stay summary. This document describes the patient's main and associated diagnoses, certain characteristics of the stay (mode of entry/exit, duration, etc.), and the procedures performed. These medico-administrative data are recorded in PMSI databases. These data are not intended for medical applications, but are invaluable for health planning.

Making information visible

In an era of digitized healthcare systems, many researchers are turning their attention to data mining methods. These methods have already proved their worth in the healthcare sector, whether to identify under-diagnosed patients, reduce costs in the health insurance system or detect insurance fraud.

One challenge associated with the use of such data is to design tools that can both process such a mass of data, and extract relevant information from it.

In the case of myocardial infarction, we have extracted spatio-temporal patterns characterizing care trajectories in order to better describe and understand patient flows. In concrete terms, using an artificial intelligence technique, we identified sets of individuals in the data who had experienced similar medical events over time.

The originality of the method consists in matching patients by considering the proximity of the pathologies present in the PMSI databases. For example, if one patient is admitted for an acute full-thickness (transmural) infarction of the lower myocardial wall, and another is admitted for an acute transmural but anterior wall myocardial infarction, they will be considered to have experienced a "close" medical event, as these two diagnoses belong to the same sub-group of the disease "acute myocardial infarction".

The patterns thus identified were then integrated into a visualization tool. In this way, it becomes possible to reconstruct the different possible evolutions of a disease. Here's an example of the flow of patients over 65 hospitalized for cardiovascular disorders.

Flow chart of patients with myocardial infarction (AP: angina pectoris; ISC: ischemia; MI: myocardial infarction; DC: death; TRY: cardiac rhythm and conduction disorders; IC: heart failure).
DR, Author provided

We have thus identified three flows, whose initial events are angina pectoris, ischemic heart disease and myocardial infarction respectively. These flows then split into several branches whose subsequent events are those already mentioned, plus death. From the third hospitalization onwards, two new events appear: cardiac rhythm disorders and heart failure. We also note that patient flows become increasingly reduced as time goes by (with a proportion of patients having died compared with the initial flow).

This type of visualization helps identify three key stages in patient flow patterns: myocardial infarction, angina pectoris and ischemic heart disease.

The majority of patients show signs of recurrent coronary insufficiency in the form of angina pectoris. Many also suffer a repeat infarction and/or develop ischemic heart disease. Others are affected by conditions such as aneurysms, ischemia, rhythm or conduction disorders, or heart failure. In addition, the data show that 21% of men and 32% of women hospitalized die, most often during their first hospitalization.

These findings are consistent with some of the possible outcomes of myocardial infarction.

A powerful forecasting tool

In addition, we have grouped these care trajectories to identify trends in inter-stay times and tariff trends. The combination of these time and rate profiles could help us to implement new care organization strategies.

At facility level, these trajectories are invaluable for bed and staff planning. On a national level, this method makes it possible to envisage a territorial analysis. Indeed, previous studies have highlighted a North-South gradient not only for hospital admissions and readmissions, but also for myocardial infarction mortality. Exploring flows by integrating the parameter of the patient's region of origin would make it possible to compare care and the evolution of the pathology, taking this context into account.

Finally, the results obtained from these flow analyses can now be integrated into a decision-support tool for clinicians. This will enable clinicians to guide their recommendations and warn patients of potential risks, by comparing their profile with that of patients with similar care trajectories.The Conversation

Jérôme Azé, University Professor, University of MontpellierJessica Pinaire, Research engineer, University of MontpellierPaul Landais, , University of Montpellier and Sandra Bringay, Professeur des Universités, University of Montpellier

This article is republished from The Conversation under a Creative Commons license. Read theoriginal article.