The Impact of a Health Empowerment Program on Self-Care Enablement and Mental Health among Low-Income Families: Evidence from a 5 Year Cohort Study in Hong Kong

Health empowerment can be an effective way to reduce health inequities. This prospective cohort study evaluated the 5 year impact of a health empowerment program (HEP) on health outcomes among adults from low-income families. The Patient Enablement Instrument version 2 (PEI-2), Depression, Anxiety and Stress Scale 21 (DASS-21), and 12 item Short-Form Health Survey version 2 (SF-12v2) were administered at baseline and follow-up for both intervention and comparison groups. A total of 289 participants (n = 162 for intervention group, n = 127 for comparison group) were included in the analysis. Most of the participants were female (72.32%), and aged from 26 to 66 years old (M = 41.63, SD = 6.91). Linear regressions weighted by inverse probability weighting using the propensity score showed that, after follow-up of 5 years, the intervention group demonstrated significantly greater increases in all items and total scores for the PEI-2 (all B > 0.59, p < 0.001), greater decreases in the DASS depression score (B = −1.98 p = 0.001), and greater increases in the Mental Component Summary score of the SF-12v2 (B = 2.99, p = 0.027) than the comparison group. The HEP may be an effective intervention enabling adults from low-income families to manage their health-related issues and improve their mental health, as evidenced by our study.


Introduction
Poverty is a global problem linked to poor health outcomes [1]. In addition to difficulties in accessing healthy food, clean water, and safe shelter, limited healthcare recourses are also a common problem for people from low-income families [2]. Indeed, a study found that people who received low incomes and/or lived in poverty had poorer health than the age-sex-matched individuals from the general population [3]. Moreover, healthcare expenses can further divert already limited disposable income from the educational, social, and other needs of families, hindering children's development and resulting in transgenerational poverty [4][5][6]. Thus, the close link between poverty and poor health forms a around 40% of residents live in poverty [40]. There was only one public primary care clinic in Tung Chung [41] serving 78,000 local residents [42] when this study began in 2012. In 2013, there was a public regional hospital established in Tung Chung. However, it only has primary care, psychiatric, emergency, internal medicine, and allied health services. The public healthcare services in Hong Kong do not provide regular health assessments, and residents have to self-finance for this in the private section. However, Hong Kong tops the world in terms of costs of living, including medical costs [43], and the combination of low-income and limited public healthcare services puts the health of Tung Chung residents at risk.
In 2012, the Trekkers Family Enhancement Scheme (TFES) was initiated by a local philanthropic group, the Kerry Group Kuok Foundation (Hong Kong) Limited (KGKF). The TFES offer supports related to health, education, employment, and environmental harmony to 200 low-income families in Tung Chung. A health empowerment program (HEP) with intercalated annual health assessments, health literacy and self-care enablement courses, and health ambassador training was delivered regularly to support the health of the TFES families. This study aimed to evaluate the 5 year effectiveness of the HEP for the health of low-income families. We examined whether the HEP was associated with greater health enablement, better mental health, and higher health-related quality of life over a 5 year follow-up.

Study Subjects and Data Collection Procedure
The present study was a prospective, comparative cohort study. It involved two groups of low-income families with young children studying in grades 1-3 (aged 7 to 11). All TFES families were invited to enroll in the HEP (herein referred to as "intervention families"), and low-income families who had not participated in the TFES were also recruited in Tung Chung and Kwai Chung as the comparison group. Kwai Chung is also a developing satellite residential district with similar sociodemographics and public healthcare facilities as Tung Chung [44]. Families were recruited between July 2012 and September 2015 if they satisfied all the inclusion criteria: (1) there was at least one family member working (full-time or part-time); (2) there was at least one dependent child in the family who studied in grades 1-3; (3) the monthly income of the family did not exceed 75% of the Hong Kong population median household income; and (4) written consent was provided. Participants in both groups completed a comprehensive health assessment and a telephone questionnaire survey at baseline upon enrollment and at around 5 years after the baseline assessment. All adults and children aged 7-11 years old at the initiation of the study from each family were included in the study.
There were 369 adults in total invited from July 2012 to September 2015 (n = 191 for the intervention group, n = 178 for the comparison group), and 357 adults (n = 190 for the intervention group, n = 167 for the comparison group) provided consent and completed the baseline assessment. There were 68 participants who did not complete the follow-up assessment at 5 years, representing a drop-out rate of 19% (n = 28 for the intervention group, n = 40 for the comparison group). Overall, 78% of participants (n = 289, n = 162 for the intervention group, n = 127 for the comparison group) completed the follow-up assessment. These participants were included in the analysis. The mean and median durations of these participants' follow-ups were both five years. Most of participants were females (72.32%), and their averaged age was 41.63 years old (SD = 6.91). The subject recruitment and follow-up flowchart is presented in Figure 1.
females (72.32%), and their averaged age was 41.63 years old (SD = 6.91). The subject recruitment and follow-up flowchart is presented in Figure 1.

Study Intervention-The HEP
The HEP consisted of intercalated annual health assessments, health talks, self-care enablement courses, and health ambassador training. The health assessment program included an annual telephone health and health service use survey, clinical health assessments, and a health hotline. Based on the telephone surveys and clinical health assessments, those with significant health risks or abnormalities were counseled by a nurse or doctor from the project team or referred to appropriate services for further management. Regular health talks and seminars targeted common problems identified in the health assessments, which included healthy eating, weight management, the health benefits of exercise, liver diseases, nutrition, stress management, psychosomatic illnesses in children, and child development. Self-care enablement courses included stress management, nutrition, dancing and exercise training, and hiking. The courses contained multiple sessions and emphasized participants' participation. The health talks and enablement courses were all delivered by specialists. The details can be found in Appendixes A and B. During the nutrition and exercise training courses, we encouraged some participants to become the group leaders of the classes and coordinate group practices after class. This group of adults became health ambassadors of their families and peers.

Outcome Measures and Study Instruments
The primary outcome was self-care enablement as assessed by the Chinese version of the Patient Enablement Instrument version 2 (PEI-2), which has been found to be valid and reliable among the local Chinese people [45]. It includes six items on perceived abilities to cope with life, understand and manage illness, maintain health, and help oneself. An example item is, "In the past four weeks, how much have you felt able to cope with life". Responses to all the items of the PEI were based on a five-point Likert scale, in which 1 meant "not at all" and 5 meant "extremely well". The scores for each item of the PEI were summated to form the total score, which ranged from 6 to 30. A higher score indicated greater enablement.

Study Intervention-The HEP
The HEP consisted of intercalated annual health assessments, health talks, self-care enablement courses, and health ambassador training. The health assessment program included an annual telephone health and health service use survey, clinical health assessments, and a health hotline. Based on the telephone surveys and clinical health assessments, those with significant health risks or abnormalities were counseled by a nurse or doctor from the project team or referred to appropriate services for further management. Regular health talks and seminars targeted common problems identified in the health assessments, which included healthy eating, weight management, the health benefits of exercise, liver diseases, nutrition, stress management, psychosomatic illnesses in children, and child development. Self-care enablement courses included stress management, nutrition, dancing and exercise training, and hiking. The courses contained multiple sessions and emphasized participants' participation. The health talks and enablement courses were all delivered by specialists. The details can be found in Appendices A and B. During the nutrition and exercise training courses, we encouraged some participants to become the group leaders of the classes and coordinate group practices after class. This group of adults became health ambassadors of their families and peers.

Outcome Measures and Study Instruments
The primary outcome was self-care enablement as assessed by the Chinese version of the Patient Enablement Instrument version 2 (PEI-2), which has been found to be valid and reliable among the local Chinese people [45]. It includes six items on perceived abilities to cope with life, understand and manage illness, maintain health, and help oneself. An example item is, "In the past four weeks, how much have you felt able to cope with life". Responses to all the items of the PEI were based on a five-point Likert scale, in which 1 meant "not at all" and 5 meant "extremely well". The scores for each item of the PEI were summated to form the total score, which ranged from 6 to 30. A higher score indicated greater enablement.
The secondary outcomes included negative emotional states and health-related quality of life (HRQOL). We used the Chinese version of the Depression, Anxiety and Stress Scale 21 (DASS-21) to capture participants' negative emotional states. This measure has shown good reliability in the Chinese setting [46]. DASS-21 includes three seven-item subscales regarding several negative emotional states, including depression, anxiety, and stress. The responses to all the items were provided on a four-point Likert scale, in which 0 meant "did not apply" and 3 meant "very much or most of the time". An example item is, "I found it hard to wind down". We first added the scores of the seven items in a subscale and then multiplied the sum by 2 so that they could be compared to the DASS normative data and to other publications on DASS [47]. Each transformed subscale score ranged from 0 to 42. Higher scores indicated greater emotional disturbance.
We utilized the Chinese 12 item Short-Form Health Survey Version 2 (SF-12v2) to measure HRQOL. The Chinese SF-12v2 has shown good validity and reliability among Chinese populations [48,49]. The measure consists of 12 items covering eight domains: physical functioning (PF), general health (GH), bodily pain (BP), physical role-functioning (RP), emotional role-functioning (RE), social functioning (SF), vitality (VT), and mental health (MH). An example item is, "During the past 4 weeks, how much did pain interfere with your normal work (including both work outside the home and housework)?". These eight domains were weighted into two summary scores: a physical component summary (PCS) score and a mental component summary (MCS) score. The scores for each domain ranged from 0 to 100. The PCS and MCS scores for the SF-12v2 were norm-based. Specifically, the population mean for the two scores was 50 and the standard deviation for the two was 10. A higher score indicated better HRQOL.
Socioeconomic status, general state of health, and physician factors were found to be associated with patient enablement [50]. Given that, confounding factors (covariates), including age, gender, highest education level obtained, household income (monthly), working status, marital status, smoking status, alcohol consumption, obesity status, chronic morbidity, reception of government assistance, and use of a regular family doctor, were collected with a structured questionnaire.
The Chinese PEI-2, DASS-21, SF-12v2, and the covariates questionnaire were administered by a trained interviewer in person or by telephone. All outcome and covariate data were self-reported.

Data Analysis
Data analyses were based on the complete-case analysis and only participants without missing values were included. We used STATA version 16.0 (StataCorp LP, College Station, TX, USA) to conduct all the statistical analyses. The presented tests of significance were two-tailed, with p values lower than 0.05 indicating statistical significance.
Descriptive statistics were used to present participants' baseline characteristics. Inverse probability weighting based on propensity scores was used to account for residual confounding bias and minimize differences in the characteristics of the two groups. We first used a logistic regression model to calculate each participant's propensity score, with adjustment for the aforementioned baseline covariates. The balance of baseline covariates between the two groups before and after the inverse probability weighting was assessed according to the p value, with p > 0.05 indicating an optimal balance between two groups.
Cronbach's α coefficient was measured to test the reliability for internal consistency for each measure, with values larger than 0.7 representing good reliability. Linear regressions weighted by inverse probability weighting using propensity scores were applied to identify the independent effects of the HEP on the participants' changes in PEI-2, DASS-21, and SF-12v2 scores from baseline to the 5 year follow-up. We first conducted preliminary tests to check the assumptions of the multiple linear regression (e.g., normality, homoscedasticity, and multicollinearity). We checked that there were no cases of outliers (i.e., Cook's distance < 1 [51]). Although the assumption of normality was violated, it did not substantively affect the results because of the large sample size (i.e., the number of observations per variable was greater than 10 [52]) in the present study. For each model, the R 2 and F-test of overall significance are reported; the unstandardized coefficients (B), 95% confidence level, and p-value are reported to indicate the effects of the intervention on each dependent outcome variable; furthermore, a power analysis for a two-sample means test was applied to calculate the post hoc power. To assess the robustness of the results, we conducted a sensitivity analysis of the multiple linear regressions, adjusting for baseline covariates, without inverse probability weighting.

Ethical Approval
The current study received ethical approval (UW 12-517) from the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster. Table 1 presents the baseline characteristics of the subjects by group before and after weighting. Before weighting, the marital and working statuses of the two groups were significantly different. Nevertheless, all baseline characteristics were balanced after propensity score weighting. In the weighted sample, most subjects were between 35 and 50 years old (72.79%), and they were predominantly women (71.26%). The measures showed good reliability with the data from the current study, including the PEI-2 (Cronbach's α = 0.90), DASS-21 (Cronbach's α = 0.84 for depression, Cronbach's α = 0.79 for anxiety, Cronbach's α = 0.85 for stress), and SF-12v2 (Cronbach's α = 0.73 for PCS, Cronbach's α = 0.81 for MCS). The PEI-2, DASS-21, and SF-12v2 scores at the baseline and follow-up assessments before and after weighting are presented in Table 2 for both the intervention and comparison groups. Table 3 presents the results for the regression of the HEP's impact on the 5 year changes in the PEI-2, DASS-12, and SF-12v2 scores after propensity score weighting. The HEP intervention group showed significantly greater increases in all the items and the total score for the PEI-2 (B ranged between 0.59 and 5.22, all p < 0.001). Additionally, the HEP was significantly associated with a greater decrease in the DASS depression score (B = −1.98, p = 0.001). Nevertheless, the differences in the changes in the DASS anxiety and stress scores between the two groups were not significant. There was a positive association between the HEP and increases in MCS scores (B = 2.99, p = 0.027), but there was no difference in the changes in PCS scores between groups. The results consistently showed significantly greater increases in all the items and the total score for the PEI-2 in the HEP group than in the comparison group in the sensitivity analysis using multiple linear regression to adjust for baseline covariates (Appendix C).

Discussion
As far as we know, the current longitudinal study is the first to examine the longterm health effects of a complex HE intervention in low-income families. The results with inverse probability weighting showed that participants in the HEP demonstrated greater increases in self-care enablement (PEI-2 scores) and mental HRQOL (SF-12v2 MCS score) and greater decreases in depressive symptoms (DASS depression score) than those without involvement in the HEP intervention.
Previous studies have demonstrated the effectiveness of various HEPs in improving participants' health-related outcomes, but most of them were short-term studies lasting for only a few months with follow ups of less than one year [53]. In particular, studies have shown that involvement in a short-term HEP can modify health-related attitudes [13][14][15][16]. For example, an HEP with a combination of teaching sessions, discussions, role playing, field tours, and so forth enhanced self-efficacy in self-care for 94 community-dwelling older adults in Taiwan after 12 weeks [16]. A HEP with 5 month problem-based learning sessions increased the perceived health control and sense of capacity to take health action among 63 immigrant women in Taiwan [14]. Furthermore, HEPs can improve psychological health [12,19]. For instance, by delivering a series of education courses over four months, an HEP increased psychological well-being and health-related quality of life for 54 Italians with diabetes [54]. An HEP with six weekly home visits by a trained nurse was found to improve psychological well-being for 32 homebound older adults in the United States [17]. As indicated above, although most HEPs only intervene for a few months and focus on smallscale individual programs, the effectiveness of these HEPs has been demonstrated across various populations, including for different ages [15][16][17]19], genders [14], ethnic groups [18], immigration statuses [14], health statuses [13,15,55], and regions [16,56]. However, research on the long-term effectiveness of HEPs and whether a family-based complex intervention is feasible and applicable in low-income families is limited.
The findings of the present study complement those of previous studies in affirming the long-term effectiveness of HEPs in improving participants' self-care enablement and mental health. One explanation regarding the mechanism of how HEPs increase enablement and mental health is that regular health assessments engage participants and raise their health awareness [57]; the health talks, self-care enablement courses, and advice on appropriate management empower them to take control of their health and cope with health problems [58]. This sense of control has the potential to inhibit the triggering of negative emotions and help with their management [17], therefore increasing the MCS score and decreasing the DASS depression score. The results indicate the feasibility and applicability of HEPs with interrelated components for the improvement of health care enablement and self-reported mental health among parents in low-income families. In particular, Hong Kong is one of the regions that has the highest Gini coefficients and significant income inequity [37]. The effectiveness of HEPs in Hong Kong targeting low-income families suggests the possibility of using HEPs to reduce health equity.
Additionally, previous studies have indicated that HEPs can stimulate the adoption of healthy habits. For example, a Korean study showed that utilizing 8 week lifestyle improvement education, group discussions, and exercise training enabled 27 hypertensive older adults to increase physical activity and reduce sedentary behavior [15]. In contrast, we did not find any significant association between the HEP and increases in physical HRQOL (SF-12v2 PCS score). In general, we observed a decrease in the PCS scores in both groups over the 5 year follow up, which coincided with the outbreak of the COVID-19 pandemic. During the peak of the COVID-19 outbreak, the Hong Kong government released mandates to close indoor sports facilities and limit outdoor activities for over five months in 2020 [59]. The outbreak of the COVID-19 pandemic and limited physical activities could have reduced physical fitness, leading to worsening of physical functioning and general health, which are the major determinants of the SF-12v2 PCS [60]. This suggests that the failure of the current HEP to improve SF-12v2 PCS may have been related to the outbreak of the COVID-19 pandemic and its influence on individuals' physical activity and physical health. Additionally, during the pandemic, the formats of the intervention components in the current HEP changed because the face-to-face intervention was not feasible. We therefore delivered the intervention via real-time video meetings, distribution of videos, a health app, telephone consultations, and so forth. While the hybrid mode of the HEP was successfully in improving self-reported self-enablement and mental health, the nonsignificant association between the HEP and the SF-12v2 PCS score suggested that the effect of the hybrid mode on physical health may have been limited.
The comparison between the current HEP and other HE interventions involving lowincome families [24][25][26][27][28][29][30][31] showed two main differences in terms of intervention design. First, our HEP included four interrelated components that covered various aspects of health by providing health knowledge, encouraging regular exercise, promoting family activities, facilitating regular health assessments and cues to take health action, increasing social interaction, and offering health consultations. Second, the complex intervention involved whole families, including children and parents. The association between family members' health was considered in the design of the intervention components to maximize the effectiveness of the HEP. The involvement of the whole family and various components likely increased participation and participants were seldom lost during follow up, which also indicate the success of the present intervention.
The findings of the current intervention contribute to the current health empowerment literature in three key ways. First, health empowerment among low-income families is feasible and has sustainable effectiveness. Second, both adults and children should be involved in the regular health assessment and self-care enablement activities in light of the close links among family members' health and the mutual influence on health-seeking behavior. Third, multi-dimensional but intercalated components can be utilized to enhance not only health promotion knowledge and practice but also accessible professional support to solve health problems. The various components provided insights into the design of future complex health interventions in real-world settings.
There are four potential limitations worthy of further discussion. First, participants in the HEP were those who volunteered to join and may therefore have had a prior interest in receiving health information and resources. Additionally, the intervention was limited to people who were currently residing in one district in Hong Kong, so the sample cannot be viewed as representative. Non-randomization and district limitations may affect the generalizability of the findings and their external validity. Second, since it was not a randomized control trial, the observed and unobserved confounding variables could not be fully accounted for in the analysis. However, our main analysis adjusted for residual confounding by applying inverse probability weighting based on propensity score. Third, we used face-to-face or telephone interviews to collect data. Furthermore, data regarding the variables of interest were based on participants' self-reports. The different data collection methods and the self-reported data could have had the potential to cause measurement errors, but this was equally applicable to both the intervention and comparison groups. Finally, the long recruitment period and the loss of participants during follow up could have led to bias in the results. Nevertheless, the drop-out rates in this cohort study in both groups were lower than the upper limit of the acceptable rate (i.e., 50%) [61], which supported the reliability and validity of the results.

Conclusions
This study shows the effectiveness of a longitudinal HEP in enabling self-care and improving mental health among adults from low-income families. It offers insights into the feasibility and applicability of HEPs in real-world community settings. Given the increasingly wide income inequality and the close link between poverty and poor health, similar HEPs may help to break the vicious cycle. This opens a new research agenda regarding how HEP care models can be more widely implemented to enhance health equity. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patient(s) to publish this paper.
Data Availability Statement: Data from this study are available upon request.

Acknowledgments:
We are most grateful to the Kerry Group Kuok Foundation (Hong Kong) for their funding support. We thank the Neighborhood Advice-Action Council, Hong Kong Sheng Kung Hui Lady MacLehose Centre, and Hong Kong Outlying Island Women's Association for their help in recruitment of the comparison families. We also thank the Neighborhood Advice-Action Council, Tung Chung Safe and Healthy City, HKFEW Wong Cho Bau School, and Tung Chung Catholic School for providing venues for our HEP activities. A special thanks to Versitech for granting us a complimentary license to use the OPine Software for data collection in our study. The timely completion of the telephone surveys by the HKU Social Science Research Centre was much appreciated. We are also grateful to both the intervention and comparison families for their participation. The hard work of our research staff in data collection and analysis must be acknowledged.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results. Common female health problems