STRENGTHS AND LIMITATIONS OF THIS STUDY

The large sample size involving urban and rural localities in all 13 states and 3 federal territories in Malaysia enables the generation of findings representative of Malaysian adults.

Differences in our study, in comparison to others, are possibly due to methodological variations.

Despite the smaller number of conditions, data were harnessed from a national survey representative of the population, with a list of conditions that have been noted to have a substantial burden on the Malaysian population.

Introduction

Multimorbidity has been recognised as a global public health issue, particularly in high-income countries; nevertheless, evidence also points out its presence and burden in low- and middle-income countries. A recent meta-analysis of 126 studies reported an estimated pooled prevalence of 38.6% (95% CI=35.3–41.9%) in high-income countries, 38.7% (95% CI=35.5–41.9%) in upper-middle-income countries (UMICs), and 32.1% (95% CI=24.3–40.0%) in low- and lower-middle-income countries.1 Globally, multimorbidity has long been reported to be high among older adults.1–3 At present, it has been recognised that multimorbidity is not exclusive to older adults, and younger age groups have also been found to be afflicted with multiple health conditions.4 By definition, multimorbidity refers to coexisting chronic conditions, which can comprise physical or mental conditions or long-standing infectious diseases in endemic areas.4 The most common cut-off point of number conditions is at least two conditions; different cut-offs have also been explored, such as the presence of at least three conditions.5 At present, a gold standard list of conditions has not yet been established, and conditions included to operationalise multimorbidity are those which are found to be prevalent in the population, obtained from a large sample and with clinical importance.5 6

Multimorbidity posits that coexisting conditions should be considered holistically, such that on consultation with an individual with multimorbidity, all conditions are taken into account when healthcare management plans are discussed. Approaches to disease and its management are single-focused in nature, where treatment focuses on individual conditions, posing the person to possible fragmented care processes.2 3 Hence, from the person-centred perspective, which includes the individual and household members, living with multiple non-communicable diseases (NCDs) means numerous appointments, tests and medication schedules.7 Compounding this, multimorbidity has been linked with non-favourable health outcomes such as premature death and poor quality of life.6

From a health systems perspective, individuals with multimorbidity have been found to have high healthcare utilisation. A recent systematic review of the UK literature found that multimorbidity is related to higher use across the spectrum of services, including primary care, dental care, emergency department and inpatient care.8 A study of 16 countries in Europe showed that an increasing number of chronic diseases was associated with more visits to the doctor, and the average annual visits for multimorbid patients were twice more than patients without any chronic conditions.9 Specific to primary care utilisation, a positive relationship was also found across all included countries.9 Focusing on South Asia,10 via a systematic review, it was found that healthcare utilisation was the most common outcome studied, with similar findings of increased use of healthcare services. Despite the growth in literature on multimorbidity, the knowledge gap has been highlighted in terms of evidence from middle- and low-income countries.11 In Malaysia, findings from national surveys using data from 2011, 2015 and 2019 consistently show that the prevalence of inpatient service utilisation among individuals with multimorbidity was twice as high as that among those without multimorbidity.12

Internationally, many studies have reported the determinants of multimorbidity4 13; in the local setting, this issue has not yet been extensively explored. A previous study on the prevalence and determinants of multimorbidity was conducted, however, among Malaysia’s older adults only.14 In addition, another local study focused on adults who reside in semiurban and rural areas of one state in Malaysia and found that age, sex, education level and employment status were associated with multimorbidity.15 Hence, there is relevance in obtaining further insight to drive efforts in understanding multimorbidity in Malaysia.

Within the context of this study, Malaysia is a UMIC currently facing demographic and epidemiological transitions. In 2022, the population of older adults was estimated at 7.3%, corresponding to the United Nation’s definition of an ageing society.16 Further, trends of rising NCDs have been established, along with the endemicity of certain infectious diseases. NCDs accounted for 63.9% of premature deaths in Malaysia and were reported as the primary contributors to the overall burden of disease in 2017.17 The healthcare system is a dichotomous system comprising a tax-funded public sector alongside a private sector, with an established primary healthcare system that had begun before the country attained independence. Healthcare services for NCDs are provided by both sectors, across the spectrum of preventive, curative and rehabilitative care, with subsidised fees in the public sector. Healthcare utilisation of outpatient services shows a preponderance of the public sector (64.6%, 95% CI=59.9–68.9), with an overall mean of 2.74 (95% CI=2.4–3.0) visits to outpatient visits annually.18 Cognisant of the threat of NCDs to population health and the role of the health system, strategic plans have been formulated with an emphasis on strengthening the role of primary healthcare, which is provided in outpatient settings, and people-centred care.19 Further, among the objectives of the latest National Strategic Plan for NCDs is to monitor the trends and determinants of NCDs, hence supporting national capacity for research related to NCDs and highlighting evidence needs.19 Nevertheless, coexisting NCDs have yet to be assessed widely in the Malaysian context, particularly in terms of evidence on the relationship between multimorbidity and healthcare utilisation. Consequently, this study aims to estimate the prevalence of NCDs among adults in Malaysia, identify factors associated with multimorbidity and assess the association between the number of NCDs and outpatient services utilisation.

MethodsStudy population, setting and data

This study was a retrospective secondary data analysis using data from the National Health and Morbidity Survey (NHMS) 2019, a cross-sectional household survey among the population in Malaysia who were non-institutionalised and residing in the selected households for at least 2 weeks prior to the data collection. The NHMS was conducted every 4 years to collect data on health needs and healthcare services utilisation; health needs comprised questions on the presence of three types of NCDs that have been found to be prevalent in Malaysia, namely diabetes, hypertension and hypercholesterolemia, whereas healthcare services utilisation covered visits to health facilities. A two-stage stratified cluster sampling was applied to ensure national representativeness. The primary stratum is comprised of all states in Malaysia including the Federal Territories, and the secondary stratum is comprised of urban and rural strata within the primary stratum. Data collectors were trained before the survey to familiarise them with the questionnaire, strengthen their interpersonal skills and emphasise the importance of teamwork in obtaining good-quality data. The entire survey was subjected to quality control in various aspects such as debriefing sessions among the data collection team members.18 For data collection, face-to-face interviews were conducted using a validated questionnaire20 by trained research assistants. The response rate for NHMS 2019 was 83.4%. A detailed methodology of the survey is described elsewhere.18

This study used data of adults aged 18 years and above. NHMS 2019 had 11 674 respondents aged 18 years and above, and only respondents with complete data on outcome variables and the covariates were included in this study (remaining n=11 347 of 11,674; dropped 2.8%). Complete case analysis was acceptable for the missing data proportion of less than 5%.21

Study variablesOutcome variables

The two outcome variables were multimorbidity and healthcare utilisation. Multimorbidity is defined as the presence of at least two NCDs (yes, no). The number of NCDs self-reported by each individual was assessed from the questions ‘Have you ever been told by a doctor or assistant medical officer that you have diabetes?’ ‘Have you ever been told by a doctor or assistant medical officer that you have high blood pressure?’ ‘Have you ever been told by a doctor or assistant medical officer that you have high cholesterol?’ Only those who answered yes to any two of these three questions were categorised as ‘yes’ for multimorbidity and used as predicting variables for subsequent analysis. Healthcare utilisation refers to the outpatient services utilisation and mean number of outpatient visits in the past 2 weeks prior to the interview. Outpatient services included health clinics in community or hospital settings, T&CM providers and ambulatory care providers, which did not incur overnight stays.

Covariates

This study adapted Andersen’s Behavioral Model of Health Care Utilisation due to its popularity and ease of use in modelling studies concerning healthcare accessibility and utilisation, which emphasises three core factors: predisposing, enabling and health need factors.22 Predisposing factors represent the tendency of an individual to use healthcare services such as sex (male, female), age (18–39, 40–59, 60+ years), ethnicity (Malay, non-Malay), marital status (single, married, widow(er)/divorcee), locality (urban, rural), education level (no formal education, primary education, secondary education, tertiary education) and working status (yes, no). Enabling factors represent the available resources that can be used to access health services and facilities, such as the presence of any health coverage (yes, no) and wealth group (low, moderate, high). The wealth variable was calculated based on the total annual household expenditure,23 adjusted on an adult equivalent scale, taking into account the composition and size of the household and economies of scale.24 The household expenditures were then classified according to quintiles, with the first quintile (Q1) as the poorest and the fifth (Q5) as the wealthiest quintile. Subsequently, the quintiles were grouped into three income categories: low (Q1 and Q2), moderate (Q3 and Q4), and high (Q5). The perceived need for healthcare services is represented by health need factors such as perceived health status (excellent and good, fair, poor and very poor).

Patient and public involvement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Statistical analysis

Characteristics of respondents were described using descriptive statistics adjusted for the complex survey design. We presented the prevalence of individuals without NCDs, with one NCD and with multimorbidity. We also presented the prevalence of outpatient services utilisation and the mean number of outpatient visits stratified by the number of NCDs. Logistic regression analyses were also performed using the complex survey design to account for the survey’s sampling weights and clustering. We used multivariable logistic regression to determine the factors associated with multimorbidity. We then used a multivariable logistic model to assess the association between the number of NCDs (coded as a discrete variable beginning from zero) and outpatient services utilisation (binary response; yes, no). To analyse the relationship between the number of NCDs and the number of outpatient visits, we applied a marginalised zero-inflated negative binomial model.

Based on Andersen’s Behavioral Model of Health Care Utilisation,22 we adjusted for covariates listed above to assess the association between the number of NCDs and outpatient services utilisation. Our analysis employed both theoretical foundations and statistical criteria for selecting the variables included in the regression models.25 26 Variables which had shown a significant association with the outcomes in the published literature were considered relevant and were forced into the model, for example, age and marital status (for multimorbidity regression model); any health coverage and income categories (for outpatient services utilisation regression model). Following the purposeful selection approach proposed by Bursac et al, we used a two-stage process for variable selection.26 In the first stage, we included variables with a p-value threshold of 0.25, ensuring that potential predictors with a moderate association were retained for further evaluation. In the second stage, variables that were not significant at the 0.1 α level and did not lead to a 15–20% change in any remaining parameter estimates were removed from the model.

We presented crude OR (COR) and adjusted OR (AOR) with 95% CIs from the logistic model and coefficient from the marginalised zero-inflated negative binomial model. The multicollinearity was examined using variance inflation factor (VIF). A VIF of >10 indicates a potential problem with multicollinearity.27 The goodness-of-fit model was tested using Archer-Lemeshow statistics, and a p-value >0.05 was considered a good fit.28 We performed the analyses using StataCorp Stata 14 (StataCorp, College Station, TX, USA) and at a 5% level of statistical significance.

Results

Overall, 11 347 respondents representing 21.7 million population were included in the analysis. The characteristics of the respondents are shown in table 1. The prevalence of NCDs by sociodemographic characteristics is presented in online supplemental table 1.

Table 1

Characteristics of respondents (n=11 347)

This study found a prevalence of 11.4 (95% CI=10.43–12.39) for multimorbidity among Malaysian adults aged 18 years and above. Higher prevalence was found among older adults aged 60 years (39.9%, 95% CI=37.01–42.93) in contrast to middle-aged and pre-old adults (16.4%, 95% CI=14.69–18.28) (see table 2). The increasing age gradient is depicted by more specific age groups in figure 1. Additionally, the presence of at least two coexisting conditions could be seen among adults in their 30s, and among adults aged 75 and above, approximately half were multimorbid (figure 1). By sex, a female preponderance was notable, and a slightly higher prevalence was found for rural localities in contrast to urban ones. Next, in our study, multimorbidity was found to be significantly associated with older age (AOR 34.43, 95% CI=23.59–50.24) (see table 3). Marital status was also found to be a predisposing factor, with AOR 2.00, 95% CI=1.29–3.08 for being widowed/divorced. Unemployment showed an increased likelihood of multimorbidity (AOR 1.88, 95% CI=1.53–2.31).

Table 2

Prevalence of the number of NCDs (n=11 347)

Table 3

Multivariable logistic regression of multimorbidity (n=11 347)

Figure 1

Prevalence of multimorbidity, by age group (n=11 347).

In terms of utilisation of outpatient services, adults with multimorbidity were high users of services (20.4%, 95% CI=17.5–23.7), approximately threefold of adults with no NCDs (table 4). In the final model adjusted for predisposing, enabling and other health need factors, multimorbidity showed AOR 3.28 (95% CI=2.48–4.32) for outpatient services utilisation. To explore the relationship further, the effect of multimorbidity on the number of visits was also assessed; however, it was not statistically significant (online supplemental table 2).

Table 4

Effect of the number of NCD on utilisation of outpatient services (n=11 347)

Discussion

The prevalence of multimorbidity in our study is lower than that reported by Chowdhury et al,1 who reported a global prevalence of 37.2% (95% CI=34.9–39.4) based on 54 countries and 35.0% (95% CI=31.4–38.5) for Asia. It is possible that the smaller number of conditions included in our operational definition influenced the prevalence obtained. The same authors also reported that a prevalence of 25.0% (95% CI=22.3–27.8) was noted for studies with 5–9 diseases.1 In contrast, our study included a list of three diseases. A local study based on a cohort located in the southernmost state of Malaysia found that 28.5% of their study population (n=18 101 adults) had multimorbidity, with a preponderance among females (60.5%).15 The higher prevalence could be due to the fact that it was based on a list of 13 conditions, hence casting a wider net; a greater number of conditions have been found to yield a higher prevalence of multimorbidity.1 An increasing gradient across age was also notable in our study, as widely reported in the literature.4 Nevertheless, multimorbidity is no longer a condition that is deemed exclusive for older adults, and our study also corresponded to this pattern. This implies many upcoming decades of potentially living in poor health or having an increased risk of poor health outcomes. This also hints at the role of lifestyle habits that predispose towards the development of chronic conditions, in contrast to age-related changes, lifestyle habits that probably have begun even in early adulthood.

Associated factors for multimorbidity in our study included age, marital status and working status. Age as a predisposing factor for multimorbidity has been established for some time across many countries and settings. In an assessment of European countries, an increasing likelihood with age was found (AOR 1.36, 95% CI=1.17–1.59 for the 60–69-year group; AOR 2.79, 95% CI=2.34–3.32 for the 70+-year group, compared with the 50–59-year group), in all but two countries.9 Findings based on a cohort study in the local context of semiurban and rural areas also found a higher likelihood of having multimorbidity in older age (AOR 1.05, 95% CI=1.05–1.05); in this study, 9.6% among those younger than 35 years had multimorbidity, increasing to almost half in the older adults (46.8%).15 In addition, the authors also found that employment was associated with less likelihood of multimorbidity (AOR 0.70, 95% CI=0.60–0.82) compared with being retired. Our findings corresponded to this, where not working showed a greater likelihood of multimorbidity. A plausible explanation is that being unemployed leads to barriers to access highly nutritious diets as well as to access health services, in addition to vulnerability to being involved in unhealthy lifestyle habits. Illustrating this further, Alvarez-Galvez et al reported that non-working or being unemployed was linked to certain combinations of health conditions, for example, never working and increased risk of a combination of multimorbid cardiometabolic conditions.13

Next, in our study, widowhood and divorce were associated with an increased risk of multimorbidity (AOR 2.00, 95% CI=1.29–3.08). A possible explanation here is that the presence of a spouse is often the closest social network and social support an individual receives and can be an enabler of positive lifestyle changes, which then promotes good health outcomes. In their assessment of determinants of multimorbidity among community-dwelling older adults, based on the 2021 Shenzhen Healthy Ageing Research, Ni et al found that marital status was a protective factor.29 Adding to this, Kim et al found that individuals living without a spouse had a higher likelihood of poor health behaviours and illnesses.30 Spouses often are the main caregivers of individuals with any illness and thus have a central role in the adjustment to medical diagnoses, which, in the context of multimorbidity, are multiple and long term. Therefore, added support for individuals with multimorbidity who face challenges in terms of limited social network, be it spouse, other family members or friends, can be deemed necessary to facilitate good management of their conditions.

In contrast to other studies in the literature, our analysis did not find a significant association for education or socio-economic status. To illustrate, Palladino et al reported that higher levels of education showed less likelihood of being multimorbid (primary completed: AOR 1.52, 95% CI=1.32–1.75; no formal education: AOR 1.55, 95% CI=1.26–1.91, with completion of at least a second level of education being the reference group).9 A similar pattern was notable for the wealth group, where the most affluent quintile showed less likelihood of multimorbidity (AOR 0.64, 95% CI=0.54–0.75, compared with the poorest quintile). Although our study showed a similar tendency, that is, a ‘protective’ effect for higher education and wealth groups, these were not statistically significant. Possibly, higher education and economic status enable easier access to healthy food and health promotion initiatives which subsequently leads to better health status. Nevertheless, a systematic review focusing on Southeast Asia reported mixed results on the association between education and multimorbidity, whereby differences between results could also be attributed to variations in how the variable is defined or categorised across studies.31 Possible reasons posited for a link between education and multimorbidity are in relation to its link to subsequent occupational opportunities and income, in addition to higher health literacy.

Focusing on the healthcare system, our findings concur with the literature in terms of a positive relationship between multimorbidity and healthcare utilisation. Lehnert et al reported, based on a systematic review of 35 studies, that not only is multimorbidity commonly associated with more health services use, but there is also a positive relationship with healthcare costs.32 In addition, this financial burden is compounded by added costs with each additional disease.33 Globally, the literature has reported multimorbid individuals as more frequent users of primary healthcare services. To begin with, analysis of the WHO Study on Global Ageing and Adult Health conducted in India showed that mean outpatient visits in the preceding 12 months for multimorbid patients were approximately three times higher compared with non-multimorbid (6.2 vs 2.2).34 A more recent study in China also showed a similar pattern; multimorbid patients were predominant users of outpatient services (73.1%; AOR 1.89, 95% CI=1.65–2.17) after controlling for sociodemographic factors, health behaviour and health status.35 Soley-Bori et al focused on studies in the UK and reported a doubling of expected primary care utilisation for multimorbid individuals compared with those with nil or one condition (AOR 2.56, 95% CI=2.48–2.64).8 Our estimates echoed these findings, with the prevalence of outpatient utilisation at a ratio of 3:1 for multimorbid individuals in contrast to non-multimorbid, as well as being three times more likely to use outpatient services.

Further, multimorbidity is linked to a higher number of healthcare visits. In an analysis of 32 583 patients in 36 general practices, van Oostrom et al found that the mean number of contacts in patients with multimorbidity was significantly higher than in patients with one or no chronic diseases (18.3 vs 11.7 and 6.1 contacts, respectively).36 Another study that analysed health insurance claims data found that the mean number of healthcare provider consultations per year was 15.7 in the multimorbid compared with 4.4 in the non-multimorbid counterparts; further, every additional condition was associated with an additional 3.2 consultations with healthcare providers.37 For primary care providers specifically, this corresponded to 7.4 versus 1.9 for multimorbid compared with non-multimorbid counterparts. De Souza and Braga in Brazil examined the trends for mean annual medical appointments for 1998, 2003, 2008 and 2013 and observed that the use for multimorbid individuals was consistently at least 1.5 times more compared with non-multimorbid.38 Although an increasing trend of mean outpatient visits was noted in our study, the statistical association was not significant.

Our study has several implications for policy dialogue. Despite a small number of conditions assessed, this study hints that the prevalence of multimorbidity in the population is considerable, and conditions are present even prior to old age; this warrants further attention. The burden of NCDs has been an area of priority in Malaysia, with a strategic plan developed in line with the Global Action Plan for the Prevention and Control of Non-Communicable Diseases 2013–2020.19 Strategies planned and implemented would largely benefit from the essential consideration that multimorbid individuals can be vulnerable to many health-related challenges, from polypharmacy to financial burden as well as carer burden. The Health White Paper for Malaysia, approved by Parliament in June 2023, underscores the importance of strengthening primary healthcare and adopting a person-centred approach.39 An ongoing initiative at public health clinics, Enhanced Primary Healthcare, includes an integrated care component that prioritises continuity of care across health facilities and effective care coordination.40 For individuals with multimorbidity, these components are essential to ensure their conditions are well-managed and their health outcomes are optimised.

From the perspective of healthcare utilisation and implications of service provision and delivery, existing services span a wide network of public health clinics in both rural and urban areas, as well as private general practitioner clinics, which are more numerous in urban areas.41 In addition, clinics are also available within hospitals where patients are seen for follow-up care without being admitted to the ward, as well as a range of traditional and complementary medicine services. It is likely that for many individuals with multiple health conditions, outpatient services are a common point of contact for healthcare and, consequently, will have many repeated episodes of consultations with healthcare providers. Subsequently, it is imperative that care be person-centred and quality care is consistently provided throughout repeated interactions with the health system.6 Traditional approaches to care are focused on a single-diseased perspective; hence, the premise of multimorbidity advocates for transitions towards integrated and coordinated care.42 Undoubtedly, health systems should be cognisant of the complexities of multimorbidity and in active pursuit of inclusivity in terms of care provision and delivery.

Strengths and limitations of the study

The large sample size involving both urban and rural localities in all 13 states and 3 federal territories in Malaysia enables the generalisation of the findings across Malaysian adults. Differences in our study, in comparison to others, are possibly due to methodological variations, that is, the operational definition of multimorbidity used, which includes the number of diagnoses considered (such as two or more, or three or more) and the conditions considered. Appropriate estimates for a population can rely on a large sample as well as consideration of conditions with a higher burden or effect in that population.6 Nevertheless, this study can contribute to the evidence base on multimorbidity in Malaysia. Next, the information obtained in our study was based on self-report data, which could be affected by recall bias. Hence, there could be under-reporting of the NCDs and the use of outpatient services. However, the data collection protocols of the study included training of data collectors as well as mechanisms of checking data quality. Finally, due to the cross-sectional nature of this study, no causal link could be established between outcome variables and associated factors.

Conclusion

Adults with multimorbidity were three times more likely to use outpatient services. Identifying factors associated with multimorbidity and understanding its significant impact on outpatient service use is crucial for guiding future healthcare system transformation and planning. The burden of NCDs in Malaysia is more than the increasing trend of individual diseases as multimorbidity impacts not only the afflicted individuals, their caregivers and healthcare providers, but also the health system. The potential complexities brought forth by having multiple NCDs emphasise the need for integrated care models with coordination across multiple levels of care. The recently launched Health White Paper for Malaysia has rightly emphasised primary care as a critical component to achieve aspirations of the healthcare system, which includes equity and responsiveness. Strengthening primary care services and improving patient navigation across healthcare levels are critical to supporting individuals with multimorbidity as outpatient facility settings are often the first point of contact with healthcare providers. Future research should explore a wider range of conditions and patterns of multimorbidity, and assess the longitudinal impact of multimorbidity on healthcare utilisation and health outcomes.

Data availability statement

Data are available upon reasonable request. The dataset that supports the findings of this article is not publicly available to protect participant privacy. Request for data can be obtained from the corresponding author on reasonable request and with permission from the Director General of Health, Malaysia.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

The Medical Research and Ethics Committee (MREC), Ministry of Health Malaysia, granted permission to carry out the National Health and Morbidity Survey 2019 (NMRR-18-3085-152 44207). The tenets of the Declaration of Helsinki were followed during the study. Written informed consent was obtained from all participants prior to the interviews.

Acknowledgments

We thank the Director General of Health, Malaysia, for his permission to publish this article. We would like to thank all research team members for their contributions to this study. We are thankful for the kind cooperation of all respondents of the survey.