In the search for interventions to improve outcomes in patients with multimorbidity, self-management education programs are increasingly recognized as an important component [14, 23, 24]. As multimorbidity is more prevalent in older patients and in those with lower socioeconomic status, [25] it is expected to find these characteristics among multimorbid patients included in self-management education programs.
In this present study, we observed that about one fourth of the participants who received the intervention reliably improved their heiQ scores, suggesting a modest effect of the intervention. Although the proportion of subjects in the intervention group with a reliable improvement was modest, the results are not far from those previously observed in courses for chronic diseases self-management in which, on average, one third of participants reported substantial benefits in the heiQ domains at the end of a course. [18] In the control group, we observed a number of participants who improved spontaneously in a few domains, suggesting a Hawthorne effect. [26] In addition, we should consider that patients accepting to participate in the study were volunteers and motivated to make changes to improve their health. Nonetheless, a significantly larger proportion of participants of the intervention group improved in six of the eight domains of the heiQ as compared to the control group.
Compared to the initial study (PR1MaC), our results show that patients with three or more chronic conditions improved the same domains of the heiQ (6 out of 8). The OR of improvement were comparable and slightly superior in patients with multimorbidity. Given the high prevalence of multimorbidity in primary care, it is reassuring to see that this subgroup has the potential to improve. When controlling for SEF, only four domains remained statistically significant. In one domain (Health-directed behavior), the OR stayed the same. In the three other domains, after controlling for socioeconomic factors, the OR improved slightly. Overall, these findings suggest that SEF has little impact on the results.
Although our study studied the effect of SEF on patients, it was not targeted specifically to a vulnerable population with low score of SEF. In an exploratory randomized-controlled study, Mercer et al. used a whole-system approach to improve the quality of life of patients with multimorbidity, targeting areas of high socioeconomic deprivation [15]. The intervention included patient-centered care and supported self-management. Quality of life improved at six months in the intervention group but was similar at 12 months. One of the three domains of well-being improved in the intervention group at 12 months. Although the results were promising, the effect of the intervention was positive in the short-term, but not significant at 12 months. This study showed promising results but also reminds us of the difficulties in developing successful interventions for multimorbid patients.
Previous studies have reported that financial constraints and low socioeconomic status in general are barriers to effective self-management. [23, 27, 28] In contrast, one study reported that controlling for education level of the participants did not influence the results of an intervention supporting self-management in patients with chronic conditions . [29] This is in contrast with a study that included patients with depression and type 2 diabetes or chronic obstructive pulmonary disease and found that education influenced the results of a self-management intervention. [30]
In this study, after adjusting for family income, education, and self-perceived financial status, in addition to age and sex, the improvement observed in the heiQ scores was no longer significant in the domains Emotional wellbeing and Social integration and support, suggesting that, at least, the results of the educational activity in these two domains are influenced by socioeconomic factors. However, there was an important overlap in the OR 95% confidence intervals before and after controlling for socioeconomic factors, suggesting a lack of power to detect the differences in this secondary analysis. More research is needed to evaluate the impact of SEF on the outcomes of interventions among multimorbidity patients. Further studies should be powerful enough to include subgroups or more vulnerable patients with low score of SEF.
Limitations of the study
One limitation of this study is a possible selection bias. In this pragmatic intervention, the primary care providers were invited to refer patients with one of the selected chronic diseases. They may have referred only participants in whom they believed the intervention could benefit the most. All patients referred (n = 481) were then offered to participate in the trial. Among them, 30% (144/481) refused to participate. Altogether, these factors should have equally affected both groups.
Social desirability is a potential bias for any self-report questionnaire, but it is consistent with use for a patient-centered approach. Since participant selection could not be blinded, it is possible that part of the effect is explained by a desirability bias in the intervention group. But the heiQ has been shown to have low social desirability bias [31]. This bias was also reduced by having one designated research assistant conducting standardized interviews.
The duration of chronic diseases and multimorbidity was not assessed. Duration of disease could play a role on self-management. In one study of diabetic patients, those who had diabetes for less than two years seemed to show better improvement in self-management [32]. Also, motivation of the patients was not assessed in the initial study and could have had an impact on the results.
During the intervention, the cluster effect was limited by the fact that interventions were provided by the same health professional team and were delivered in the clinic where the subject was recruited. It is possible that external factors contributed to individual practices, but it was impossible for the research team to measure it.
A total of 481 subjects were referred to the initial study, but only 332 were randomized and 281 of them had multimorbidity. Except for age and gender, no other characteristics could be collected for those not participating. The analysis was limited to the outcome at three months. Sustainability of the changes on this subgroup of multimorbid patients is unknown.
Health literacy has been identified as potentially playing a part in explaining the link between low education and low health status [33] and it is said to play a crucial role in self-management [34]. Health literacy was not measured in this study and its effect on the results of the intervention is unknown.
It would have been interesting to receive the feed-back of the patients on how SEF affected them during the study.