Improving quality of care for older medical patients: Identifying patients at risk of readmission and factors influencing implementation of a readmission prevention program

Introduction
Provision of high quality health care for older medical patients with multifaceted health problems has faced challenges due to fragmentation of the health care system. Improvement of health care quality for older medical patients should include precise targeting of interventions and implementation of methods to integrate care. One way to measure quality of care in the transition from hospital to home among older medical patients is to focus on the frequency of acute readmissions. Prediction models to predict all-cause acute 30-day readmission among older medical patients have almost all reported low performance. Programs to improve health care quality and reduce/prevent readmissions often entail integration of care across health care providers. Implementation of integrated care programs faces various challenges in terms of organizational and cultural differences among the inter-organizational program stakeholders. Hence, the literature has called for more knowledge on adherence to integrated care programs and on the influence of contextual factors in the process of implementation.

Aims
This PhD thesis aims to develop and internally validate a model to predict 30-day readmission among older medical patients. Moreover, it seeks to explore the factors that influence implementation of integrated care programs targeting older patients who risk readmission to hospital following discharge. The thesis is divided into three sub-studies with the following aims:
- To develop and internally validate a model for predicting 30-days readmission among older medical patients (sub-study I).
- To examine the degree to which adherence to an interdisciplinary, post-discharge follow-up program targeting older patients is associated with patient-specific factors and organizational factors (sub-study II).
- To explore how contextual factors, perceived as external demands, influence the implementation of an inter-organizational program. Hence, the goal here is to show the contradictions between the demands imposed by program implementation and everyday work routines in health care (sub-study III).

Methods and material
The research for the three sub-studies was based on data derived from testing and regional implementation of a post-discharge follow-up program in Region Zealand, Denmark, which aimed to improve the quality of care and reduce readmissions among older medical patients. The program provided for joint visits by GP and municipal nurse in the patients’ home within one week after discharge. Sub-studies I and II were quantitative observational studies, whereas sub-study III was qualitative and based on focus group data. Sub-study I (n=770) was based on data from a clinical screening of consecutive patients who were aged 65 years or older and who had been discharged from the medical department at Holbæk University Hospital in 2012. Sub-study II was based on the cohort of patients who were screened and found eligible for post-discharge follow-up visits at a hospital in Region Zealand in 2014 (n=1,659). Level of adherence was measured according to two essential steps of the post-discharge follow-up program: referral of the screened patient to the municipality (step 1) and successfully completed post-discharge follow-up visits (step 2). The list of candidate predictors in sub-study I and co-variables in sub-study II was derived from data provided by various Danish administrative registers. Sub-study III was based on data from five interdisciplinary focus group interviews with a total of 24 health professionals engaged with implementation of the post-discharge followup program.

Results
In sub-study I, a readmission prediction model was developed with acceptable performance and no indication of overfitting in the internal validation. The model showed that demographic, social, organizational and health-related factors predicted acute allcause 30-day readmissions; i.e. being male, low education, prior contact with emergency physician, diagnosis related to the respiratory system, not being diagnosed due to atypical symptoms, cognitive problems, high CCI, longer hospital stays and three different pharmacy groups related to chronic conditions. Sub-study II showed a low level of adherence to the post-discharge follow-up program in both step 1 (69% adherence) and step 2 (54% adherence). Moreover, adherence to referral from hospital to municipality was associated with the particular hospital from which the patient had been discharged, and whether the patient had received nursing- or homecare assistance from the municipality prior to admission. Level of adherence in terms of successfully completed post-discharge  follow-up visits was associated with gender, municipality of residence and whether the GP operated alone or in shared practice. Sub-study III confirmed that implementation of the post-discharge follow-up program was intensively influenced by contextual factors related to the inter-organizational work of the program. Thus, the post-discharge follow-up program introduced new demands related to patient enrollment and interdisciplinary work that came into conflict with the professionals’ existing work routines in their health care work. The study further found that the conflicting demands were embedded in the institutional logic of the organization.

Conclusion
Based on the studies in this PhD thesis, a comprehensive model for prediction of acute allcause 30-day readmission among older medical patients was developed. The thesis further showed that adherence to an integrated care program that operates across organizational levels is heavily dependent on organizational factors. Implementation of a new integrated care program is potentially subject to contradictory demands that originate in the very institutional context in which health professionals work.