To our knowledge, WellNet study is the first study to evaluate the changes in HRQoL among patients with one or more chronic conditions in Australian primary care settings based on the principles of PCMH model. Findings of this study are consistent with the growing body of evidence showing strong association between patients’ HRQoL and several core elements of the PCMH such as involvement of a MDT [36, 37], continuity of care [38, 39], and shared decision making and patient-provider communication [22, 23]. Previous Australian studies by McCaffrey et al [24] and others [40] have reported on health utilities and HRQoL on general population norms using cross-sectional data. However, studies reporting on disease-specific, high risk sub-group population using GP data are relatively less, which is of interest, as primary care is the forefront of care delivery in Australia with at least 85% of Australians consulting a GP every year [3]. In view of this, the WellNet study is novel as it closely examines the outcome of integrating care delivery on HRQoL at two different time points whilst determining predictors of change using GP data.
In this study, the use of EuroQol EQ-5D-5L over other instruments owes to its simplicity in accruing several aspects of an individual’s self-perceived health status in a relatively short duration through use of a short 5-item questionnaire [41]. Moreover, EQ-5D-5L has also been reported as one of the sensitive instruments in terms of better discriminative power in effectively detecting changes in HRQoL [42]. In addition, it is reported to have better known-group validity where subjective patient scores are shown to be in accordance with the objective investigator findings of changes in HRQoL [43].
For studies measuring the impact of treatment outcome/s, the minimal clinically important difference (MCID) reports on the smallest change in the outcome of interest that is considered to be clinically significant or meaningful [44]. A comprehensive review of 18 studies by Coretti et al [45] estimated the overall MCID for EQ-5D range to be between 0.03 and 0.54. In view of this, findings of our study showed both statistical significance whilst also meeting the bare minimal threshold of clinical significance in EQ-5D index scores after adjusting for baseline covariates. However, considering that our sample is chronically ill with many patients having multiple diseases, MCID may not even be a significant indicator on population level. In this population, we would typically expect that many patients would have progressed in their disease, so even small change or no change in EQ5D may be a positive outcome for the program. The effectiveness of PCMH model on improving patients’ HRQoL is consistent with studies by Schuttner et al [13] and Hynes et al [14].
Of the five dimensions of EQ-5D, WellNet patients reported substantial improvement particularly on two domains of pain/discomfort and usual activities in terms of a 33% and 28% increase, respectively, in the ‘no problem’ level at follow-up. This could be attributed to the primary objective of the WellNet program in improving self-management behaviour among patients to effectively manage symptoms associated with their chronic conditions [27]. Improved self-management behaviours are strongly associated with improved HRQoL [46, 47].
Findings of the multivariable regression models (Model 1 and 2) show that baseline index value and positive diagnosis of respiratory disease were significantly associated with EQ-5D index at 12 months. Higher baseline EQ-5D index value as significant predictor of increased follow-up index scores is consistent with other study findings by Van Eck et al [35]. This could be because patients who already reported better HRQoL at baseline benefitted through the patient education and self-management from the WellNet care team.
A positive history of respiratory disease was negatively associated with HRQoL at follow-up compared to those without prior respiratory disease. The poor HRQoL reported among patients with respiratory disease due to several reasons of duration and severity of the condition supplemented with or without harmful lifestyle behaviours is well documented [48, 49]. Furthermore, lack of PHI was associated with poor HRQoL at follow-up in the imputed model. This could be because patients without PHI coverage are less likely to receive appropriate and timely care leading to poor health outcomes and subsequently poor HRQoL [50, 51].
KOOS and HOOS assessments were recorded in parallel with EQ5D instrument in the WellNet study. Changes in KOOS and HOOS scores were supplemented with the primary outcome of EQ5D changes. Besides statistical significance, the scores also met the MCID rendering them clinically relevant for change in patient management. The favourable changes in this study is consistent with findings of other studies of collaborative care [52, 53].
Our study has several strengths and limitations. This is the first study in Australia to evaluate the outcome of a PCMH model on HRQoL among patients in primary care setting. The study includes an effectively targeted sample with longitudinal measurements at two different time intervals enabling determining predictors of change in HRQoL scores. This study also adds to the relatively less than adequate research conducted using GP-data. Although the aim of this study was to evaluate changes in HRQoL after the 12-month WellNet intervention, this study was not designed as an effectiveness study, but rather as a proof-of-concept study.
In regard to study limitations, although WellNet program comprises an effectively matched comparison group, the EuroQol EQ-5D-5L was recorded only among treatment group, thereby limiting to within-group analysis. The lack of control group means that the possibility of potential bias cannot be excluded, and we cannot be sure that improvement in EQ5D scores may have occurred anyway without the enhanced PCMH intervention. However, that seems unlikely based on research conducted with use of control groups reporting similar outcomes [54, 55]. Additionally, some key socio-economic variables such as annual income were unavailable due to privacy concerns, which may also have impacted prediction of index scores over time. Finally, consistent with other originally designed programs, reproducibility of findings is constrained by potential barriers in the form of uniqueness of data and by patient and provider-level determinants [27].