Skip navigation

Predicting non-elective hospital re-admissions

 

Among hospital patients who are discharged home, a high proportion (31-50%) of older patients is readmitted to hospital within 90 days of discharge. Some of these may reflect the natural course of disease or the poor quality of care. But, can ways be found to identify and intervene with high-risk patients to prevent such readmission? The issue is particularly relevant to the Veterans Health Administration in the USA that provides care to older and vulnerable patients. It is also an issue for all health care services. A recent study has sought to identify clinical and patient-centred factors that predict non -elective hospital readmission and thus provide the basis for the development of new interventions to reduce these events.

Reference

David M. Smith and colleagues. Predicting non-elective hospital readmissions: a multi-site study. Journal of Clinical Epidemiology2000 53: 1113-1118.

How was the question tackled?

The study was a secondary analysis of a multi-site randomised clinical trial of increasing access to primary care for patients at 9 Veterans Health Administration Medical Centres. The sites were chosen to provide diversity of location and academic affiliation - not on the basis of readmission rates. The study chose three diseases that are prevalent among older people and where appropriate care might reduce readmission, i.e. diabetes mellitus, chronic obstructive pulmonary disease (COPD) or congestive heart failure (CHF).

The teams at the 9 sites agreed a study protocol and criteria to identify suitable patients. Research assistants at each site screened all patients and randomised them into one of two groups - a control (usual care) group or a primary care intervention group (which provided enhanced access to a primary care nurse and physician). Baseline data was assembled to illustrate the hospital and patient characteristics.

The key outcome measure in the study was time to first non-elective readmission within 90 days of discharge. 90 days was chosen because a plot of readmission frequency over time showed a drop in frequency after 90 days: it had also been used in several previous similar studies.

What did the study find?

During the study, 10,129 patients were screened; 3209 met all the criteria for eligibility and 1396 (43%) were enrolled in the trial. The common reasons for non-enrolment were patient's decision not to participate (they were concerned about losing access to 'their' specialist's care) or discharge before enrolment was completed. A small number of patients died or withdrew from the study leaving a cohort of 1378, of these 321 (23%) were non-electively readmitted within 90 days. This rate is consistent with other similar studies. The percentage of patients readmitted varied significantly across the 9 sites, ranging from 15 to 37%.

Analysis of the data related to patients readmitted showed that they were older, more likely to be Caucasian and unemployed. Readmitted patients had a higher burden of illness (as reflected by higher BUN, presence of anemia and leukocytosis) and they had a longer initial hospital stay. Lower physical and mental health scores (from SF-36) were also evident for those readmitted. Patients assigned to the intervention (primary care) group were more likely to be readmitted.

Readmission rates for patients with CHF increased consistently from 23 to 35%. The situation was similar for patients with COPD, here readmission increased from about 28 to 38%. However for patients with diabetes mellitus readmission increased only for those with end organ damage.

What has the study told us?

This is a helpful study that adds to our knowledge about managing patients' discharge carefully and assessing risk of readmission.

Five variables previously identified were reaffirmed:

The study identified two other variables that were significantly and independently related to non-elective readmission.

First, the mental component summary scores from the SF-36 . The influence of mental health problems on risk for increased hospitalisation is increasingly being recognised. Notably, more hospital services are used by older patients with congestive heart failure and depression than by patients with congestive heart failure but without depression. There is support for lower mental health scores being a risk factor for readmission.

Second, the patient satisfaction scale . Increased satisfaction with access to emergency care was associated with increased risk of readmission. The intervention that increased or improved access to primary care resulted in increased patient satisfaction with access to both emergency and primary care - but also increased readmission. This was particularly the case for patients who have more severe conditions but without a specific intervention.

Previous studies had shown that clinical trials without targeting specific risk factors had not been successful in reducing readmission. Identification of those factors that best predict readmission provides direction for potential interventions and/or further research for reducing readmission. However, few of the risk factors identified in the study are modifiable. Making sure that mental health status and patient satisfaction are included in such assessments will be an important step forward.