Determinants of hospitalization in Chinese patients with type 2 diabetes receiving a peer support intervention and JADE integrated care: the PEARL randomised controlled trial

Abstract

Background: In a randomized controlled trial of 628 Chinese patients with type 2 diabetes receiving multidisciplinary care in the Joint Asia Diabetes Evaluation (JADE) Progam, 372 were randomized to receive additional telephone-based peer support (Peer Empowerment And Remote communication Linked by information technology, PEARL) intervention. After 12 months, all-cause hospitalization was reduced by half in the PEARL group especially in those with high Depression Anxiety and Stress Scale (DASS) scores.

Methods: We used stratified analyses, negative binomial regression, and structural equation modelling (SEM) to examine the inter-relationships between emotions, self-management, cardiometabolic risk factors, and hospitalization.

Results: Hospitalized patients were older, more likely to have heart or kidney disease, and negative emotions than those without hospitalization. Patients with high DASS score who did not receive peer support had the highest hospitalization rates. After adjustment for confounders, peer support reduced the frequency of hospitalizations by 48% with a relative risk of 0.52 (95% CI 0·35-0·79;p = 0·0018). Using SEM, improvement of negative emotions reduced treatment nonadherence (Est = 0.240, p = 0.034) and hospitalizations (Est=-0.218, p = 0.001). The latter was also reduced by an interactive term of peer support and chronic kidney disease (Est = 0.833, p = < 0.001) and that of peer support and heart disease (Est = 0.455, p = 0.001).

Conclusions: In type 2 diabetes, improvement of negative emotions and peer support reduced hospitalizations, especially in those with comorbidities, in part mediated through improving treatment nonadherence. Integrating peer support is feasible and adds value to multidisciplinary care, augmented by information technology, especially in patients with comorbidities.

Trial registration: NCT00950716 Registered July 31, 2009.

Keywords: Adherence; Hospitalizations; Integrated care; Negative emotions; Peer support; Quality improvement; Structural equation modelling.

Conflict of interest statement

This study was approved by the Joint CUHK-NTEC Clinical Research Ethics Committee. Agreeable patients gave written informed consent. This trial was pre-registered at clinicaltrial.gov under identifier NCT00950716 Registered July 31, 2009.Consent for publication of raw data was not obtained and the dataset could, in theory, pose a threat to confidentiality. The researchers did not specifically ask for anonymized data to be deposited in the public repository. All data are anonymized and captured using the portal implemented through the Asia Diabetes Foundation established as a knowledge transfer project in the form of a registered charitable research organization governed by the CUHK Foundation. A patient log was kept by the clinics for identification and linkage to the hospitalization data using the CMS governed by the Hong Kong Hospital Authority, which is the main public healthcare provider in Hong Kong.JCNC has received research grant and honorarium for consultancy or giving lectures from Eli Lilly. All research grants were awarded to the CUHK or ADF. All honoraria have been donated to the Chinese University of Hong Kong to support diabetes research and education. JCNC is also the Chief Executive Officer (pro bono) of ADF, a charitable organisation governed by the CUHK Foundation (www.adf.org.hk) which develops the web-based JADE program to conduct quality improvement and translational research programs through private-public partnerships. EF is the Global Director of the Peers for Progress, supported by the American Academy of Family Physicians Foundation and funded by Eli Lilly Company and Foundation. Other authors declared no conflict of interest relevant to this manuscript.Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures

Fig. 1

Proposed structural equation model with hypothesized explanatory and outcome variables (rectangle: measured variables; oval: latent variables) and interlinking paths to explain the multi-causality of hospitalization (15). Abbreviations: PEARL: peer support intervention, Frequency: number of hospitalization episodes, Stay: Length of hospital stay, PHQ-9: 9-item Patient Health Questionnaire assessing depressive symptoms, DASS-21: 21-item Depression Anxiety Stress Scale, CDDS-15: 15-item Chinese Diabetes Distress Scale, Medication Adherence: 4-item Morisky Medication Adherence Scale, CKD: Chronic Kidney Disease with eGFR 15–60 ml/min/1.73 m2, CVD: Cardiovascular disease including coronary heart disease (myocardial infarction, unstable angina, percutaneous coronary intervention, coronary bypass operation), stroke, peripheral vascular disease (lower extremity amputation, absent foot pulses with ankle:brachial ratio < 0.9 and/or lower limb revascularization), BP: Blood pressure, SBP (systolic) DBP (diastolic), TC/LDL-C: Total cholesterol/LDL-cholesterol, TG: Triglycerides, BMI: Body mass index, Waist: waist circumference, N.B. Baseline explanatory measures were analyzed

Fig. 2

Kaplan Meier curves showing the cumulative proportions of people with type 2 diabetes requiring hospitalization. a Stratified by assignment to peer support (PEARL). b Stratified by baseline negative emotions (DASS≥17). c Stratified by peer support and baseline negative emotions. Pearl denotes peer support intervention. Depression Anxiety Stress Scale are denoted as DASS, using ≥17 points as the cut off for those in the top quintile of subjects with negative emotions

Fig. 3

A structural equation model (SEM) showing the effects of change in negative emotions (DASS, PHQ9, CDDS) on hospitalizations (admission episodes and stay) and medication adherence as well as the interaction between peer support and CKD/CVD on hospitalizations. Goodness of fit indices of the model: pvalue of the Chi-squared test is 0.1330; the root mean square error of approximation is 0.019, the CFI value is 0.978. The figure represents the estimates of latent variable relationships to outcome variables with pvalues in parentheses (refer to Table 2 for full details of the SEM). Abbreviations: PHQ-9: 9-item Patient Health Questionnaire assessing depressive symptoms, DASS-21: 21-item Depression Anxiety Stress Scale, CDDS-15: 15-item Chinese Diabetes Distress Scale, Medication Adherence: 4-item Morisky Medication Adherence Scale, CKD: Chronic Kidney Disease as estimated glomerular filtration rate 15–60 ml/min/1.73 m , CVD: Cardiovascular disease including coronary heart disease (myocardial infarction, unstable angina, percutaneous coronary intervention, coronary bypass operation), stroke, peripheral vascular disease (lower extremity amputation, absent foot pulses with ankle:brachial ratio