A cross-sectional study was conducted in August 2017 at Hougang Polyclinic, which is part of a bigger network of clinics under the National Healthcare Group Polyclinics (NHGP). Hougang polyclinic is a large public primary care centre with approximately 140 healthcare providers including physicians, nurses, pharmacists, and allied health professionals. It provides a comprehensive range of healthcare services, including health screening, treatment for acute and chronic medical conditions, women and child health services, as well as dental care. The polyclinic serves the Hougang Township, the fifth largest township in Singapore, with approximately 223,010 residents of which 40% are middle-aged.22
There is no strict definition of “middle-age” and studies have defined it as age between 40-65 years23 or 45-64 years.3,5,24 In our study, the lower age limit was defined as 40 years old as that is also the minimum age at which general health screening is recommended.6 According to the Singapore census those aged 65 years and above are considered elderly22 and hence the upper limit of our middle-age range was defined as 64 years old.
With assistance from the NHGP Office of Clinical Informatics and clinic operations staff, we obtained the daily lists of middle-aged patients (40-64 years old) with physician appointments. The appointments were either scheduled in advance for regular review of their chronic conditions or scheduled on the same day as a walk-in appointment. Potential participants were selected in a systematic randomized manner based on the time and type (regular review or as walk-in) of appointment. They were then approached and screened for eligibility. The inclusion criteria were a) aged 40-64 years, b) had at least one or more chronic condition(s) out of a predetermined list of 14 chronic conditions, c) consented to access of electronic medical records for data collection, and d) spoke and understood any one of the three main languages in Singapore: English, Mandarin or Malay. Participants were excluded if they were non-communicative, unable to give consent, or if they did not complete the survey. Our study focused on how having chronic conditions affected one’s HrQoL and we also excluded those with no chronic conditions as they often present at our polyclinic with acute self-limiting conditions25 which can transiently affect their HrQoL.
Recruitment and consent-taking were done by trained interviewers, including a research assistant, two medical students, and the principal investigator. The interviewer-administered questionnaires were conducted in a quiet area in the polyclinic.
Definition of Multimorbidity
We defined multimorbidity as the presence of three or more chronic conditions. Although many studies3,5,9-10 and organisations26,27 have used a cut-off of two or more chronic conditions to define multimorbidity, some studies11,18 have used a higher cut-off of three or more conditions. Holzer et al28 found a close relationship between the estimated prevalence of two or more conditions and that of three or more conditions, and that both definitions of multimorbidity also gave the same information on prevalence. In an unpublished cross-sectional study29 of 787, 466 primary care patients in Singapore, the prevalence of multimorbidity in the middle-aged patients was 45.3% when the cut-off was two or more conditions. When the cut-off was three or more conditions, the prevalence decreased to 28.5%.29 For our study, using a higher cut-off to define multimorbidity can better identify patients with increased needs2 and this is more meaningful in our setting.
Although there is no standardised definition of multimorbidity, using a list of at least 12 chronic conditions resulted in little variation in prevalence estimates of multimorbidity.2 Thus, we used a list of 14 chronic conditions to define multimorbidity: diabetes mellitus, hypertension, lipid disorder, neurological conditions, respiratory diseases, psychiatric conditions, cancer, chronic kidney disease, heart diseases, arthritis, back/neck problems, gastrointestinal diseases, thyroid disease and physical disability. This list was previously used by Quah et al16 to measure multimorbidity in elderly patients at a primary healthcare setting and was derived from the Singapore Mental Health Study.30 Participants were asked to report if they had any of the chronic conditions listed above, as told to them by a registered physician. In this study, the number of chronic condition(s) was categorized dichotomously, distinguishing those with one or two conditions from those with three or more conditions i.e. with multimorbidity.
Measurement of Health-related Quality of Life
HrQoL was measured by EQ5D-3L questionnaire,31 which has been validated locally,32-34 and is available in the three most spoken languages in Singapore- English, Mandarin and Malay. The EQ5Dconsists of two components. The first component is the health-state Utility Index (UI). It measures five dimensions of HrQoL (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) on a three-point severity scale (no problems, moderate problems or extreme problems). The Singapore time trade-off values were used to convert the information into UI scores, with -0.790 being the worst health state and 1.000 being the best health state. The second component of the EQ5D is the visual analogue scale (VAS) which consists of a scale from 0 to 100. It is used to assess self-perceived global levels of health, with 0 representing the worst imaginable health state and 100 the best imaginable health state. Participants were asked to select a number on the VAS, which best represented their global health state for that day.
The sociodemographic variables collected included age, sex, ethnicity, main spoken language, marital status, education level, work status, monthly household income (in Singapore dollars), type of dwelling, home ownership, and living arrangement. With regards to the type of dwelling, the options are more varied in view of Singapore’s unique housing landscape where the majority of the population live in subsidised housing provided by the Housing Development Board (HDB). The size and value of these apartments correspond to the number of rooms stated. In addition, there are hybrids of public and private housing such as the Executive Condominiums and Housing Urban Development Company apartments that cost more than the usual HDB apartments. The minority of the population stay in private housing that includes private condominiums and landed properties.22
Sample size was calculated by assuming a Pearson’s coefficient of -0.2, which was derived from the Spearman coefficient of the association between EQ5D UI and the count of chronic conditions reported in H Radner et al.35 With alpha of 0.05 and power (1-beta) of 80%, the estimated sample size was 194. Assuming 30% refusal and incomplete data, the final calculated sample size was 278.
The sociodemographic characteristics, number of chronic conditions, and EQ5D states of the study population were analysed descriptively. Means with standard deviations were calculated for continuous variables, while frequencies and percentages were computed for categorical variables.
A generalised linear model with log link function was used to analyse the associations between multimorbidity and each of the two components (UI and VAS) of the EQ5D, producing regression coefficients with 95% confidence intervals. Binary logistic regression was used to compare the responses i.e., “moderate or severe problems” to “no problem” for the sub-group analysis of each of the EQ5D domain. The analyses were adjusted for sociodemographic variables. A p-value of <0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS for Windows Version 24 (IBM Corporation, Armonk, New York, USA)