Skip to main content
Download PDF

Prognostic factors for premature cardiovascular disease mortality in Malaysia: a modelling approach using semi-parametric and parametric survival analysis with national health and morbidity survey linked mortality data

BMC Public Health volume 24, Article number: 2745 (2024) Cite this article

Abstract

Background

Cardiovascular disease (CVD) is the leading cause of premature mortality worldwide. Despite existing research on CVD risk factors, the study of premature CVD mortality in Malaysia remains limited. This study employs survival analysis to model modifiable risk factors associated with premature CVD mortality among Malaysian adults.

Method

We utilised data from Malaysia’s National Health and Morbidity Survey (NHMS) conducted in 2006, 2011, and 2015, linked with mortality records. The cohort comprised individuals aged 18 to 70 during the NHMS interview. Follow-up extended to 2021, focusing on CVD-related premature mortality between ages 30 and 70. We employed six survival models: a semi-parametric Cox proportional hazard (PH) and five parametric survival models, which were Exponential, Weibull, Gompertz, log-normal and log-logistic distributions using R software. The age standardized incidence rate (ASIR) of premature CVD mortality was calculated per 1000 person-years.

Results

Among 63,722 participants, 886 (1.4%) experienced premature CVD mortality, with an ASIR of 1.80 per 1000 person-years. The best-fit models (based on AIC value) were the stratified Cox model by age (semi-parametric) and the log-normal accelerated failure time (AFT) model (parametric). Males had higher risk (Hazard Ratio, HR = 2.68) and experienced 49% shorter survival time (Event Time Ratio, ETR = 0.51) compared to females. Compared to Chinese ethnicity, Indians, Malays, and other Bumiputera had higher HR and lower survival times. Rural residents and those with lower education also faced increased HRs and reduced survival times. Diabetes (diagnosed: HR = 3.26, ETR = 0.37; undiagnosed: HR = 1.63, ETR = 0.63), hypertension (diagnosed: HR = 1.84, ETR = 0.53; undiagnosed: HR = 1.46, ETR = 0.68), and undiagnosed hypercholesterolemia (HR = 1.31, ETR = 0.80) increased risk and decreased survival times. Additionally, current smoking and abdominal obesity elevated risk (HR = 1.38, 1.60) and shortened survival (ETR = 0.81, 0.71).

Conclusion

The semi-parametric and parametric survival models both highlight the considerable impact of socioeconomic status and modifiable risk factors on premature CVD mortality, underscoring the imperative for targeted interventions to effectively mitigate these effects.

Peer Review reports

Introduction

Cardiovascular disease (CVD), particularly ischemic heart disease (IHD) and stroke, remains the leading cause of mortality [1] and premature mortality worldwide [2]. The World Health Organization (WHO) defines premature mortality as unconditional probability of death between ages 30 and 70 years [3] from non-communicable diseases (NCDs) including CVDs, cancer, diabetes, and chronic respiratory diseases. Globally, NCDs account for 17 million premature deaths [4], with 38% of those attributable to CVDs [1]. In low- and middle-income countries (LMICs), CVD significantly adds to the growing burden of NCDs, accounting for approximately 86% of all premature NCD deaths globally [4]. In Asia, CVD emerged as the leading cause of death, accounting for approximately 10.8 million deaths, which constituting around 35% of the total deaths in the region in 2019. Notably, nearly 39% of these CVD deaths were categorized as premature mortality [5]. Similarly, in Malaysia, CVD has been the leading cause of death since the 1990s and remains a major threat, contributing to around 35% of premature mortality in the country [6, 7]. In 2018, IHD continues as the predominant cause of premature mortality among Malaysians in both sexes, constituting 17.7% of Years of Life Lost (YLL), while stroke positioned as the fourth-ranking, contributes 8% of YLL [8]. Premature mortality from NCDs, including CVD, has implications for productivity loss and economic impact. To address this global burden, the WHO set targets to reduce the probability of premature mortality due to NCDs by 25% by 2025 [9]. Effectively managing these NCDs, including CVD, and their associated complications is crucial for safeguarding of this public health issue.

The burden of premature CVD mortality is influenced by various risk factors, which can be broadly categorized as non-modifiable and modifiable. Non-modifiable risk factors, such as age, sex, and ethnicity, play a role in determining an individual’s susceptibility to CVD but are beyond our control [10,11,12,13]. In contrast, modifiable risk factors offer the potential for targeted interventions and lifestyle modifications to reduce the risk of CVD-related morbidity [10, 11] and mortality [14,15,16]. Among the key modifiable risk factors, metabolic factors such as diabetes mellitus (DM), hypertension (HPT), and hyperlipidaemia (HPL) and obesity stand out as significant contributors to premature CVD mortality [14, 16, 17]. In addition to these modifiable risk factors, other behavioural risk factors significantly contribute to the rising prevalence of premature CVD mortality. Smoking, for instance, is a major risk factor for CVD [14, 17], while excessive alcohol consumption further influences CVD mortality [18]. Moreover, physical inactivity and sedentary lifestyles significant contributors to CVD mortality [19]. These behaviours linked to modifiable risk factors correlate with increasing trends in CVD mortality [20,21,22,23]. Without mitigation efforts targeting these risk factors, it is projected that nearly 23.6 million people will die from CVDs by 2030 [24].

The risk factors for premature CVD mortality can vary across regions and income levels. A study of 155,722 participants from 21 countries (PURE project) investigate the association of modifiable risk factors with CVD and mortality, revealing diverse impacts based on income levels [16]. In middle-income countries (MICs), HPT was the leading risk factor for CVD, alongside influential metabolic and behavioural factors, low education had a greater impact in MICs compared to high-income countries (HICs). In low-income countries (LICs), the major CVD risks were linked to metabolic factors, household air pollution, and poor diet [16]. Additionally, an analysis of a cohort of 33,583 individuals aged 35–70 from South Asia indicated a smaller impact from metabolic risk factors but a more substantial influence from low education, weak physical strength, and poor diet quality [14]. Similarly, in China, HPT continued to be the primary population-level risk factor, while low education ranked as the second-largest risk factor for both CVD and mortality [15]. The diversity in CVD risk profiles is likely associated with difference CVD mortality rate among the Asian region and countries. Disparities in age-standardized CVD mortality within different Asian regions have been documented in several published studies [25,26,27,28]. Notably, high-income Asian countries, like Japan (11%) and Israel (15%), exhibit lower proportions of premature CVD mortality, while many low- and middle-income Asian countries including Malaysia, show considerably higher proportions of premature mortality from CVD [27].

Malaysia’s diverse population encompasses various ethnicities, each with distinct genetic predispositions and lifestyle habits that can influence CVD risk profiles. Despite existing research on CVD risk factors, there remains a scarcity of studies investigating premature CVD mortality specifically within Malaysia. This information is useful to inform policy formulation and public health strategies for controlling and preventing CVDs in Malaysia. Hence, our study aims to investigate the prognostic factors associated with premature CVD mortality in Malaysia, with a particular focus on modifiable risk factors. Specifically, we will use both semi-parametric survival analysis (Cox Proportional Hazard regression model) and parametric survival analysis (Exponential, Weibull, Gompertz, log-normal and log-logistic regression model) to analyse data from the National Health Morbidity Survey (NHMS), a national population-based survey in Malaysia linked with mortality files. Additionally, we will calculate the latest age standardized incidence rate (ASIR) of premature CVD mortality. By focusing on easily accessible and modifiable risk factors, our research aims to provide practical and actionable recommendations that can effectively reduce premature CVD mortality and improve overall population health.

Method

Study design and subjects

This study utilized data from the NHMS, a large population-based survey conducted nationwide in Malaysia since 1986. NHMS is a cross-sectional survey carried out by the Institute for Public Health (IPH), National Institute of Health (NIH), Malaysia to provide community-based data for the Ministry of Health Malaysia to review health programs and priorities. For this study, we used data from three NHMS surveys conducted in 2006, 2011, and 2015, focusing on the prevalence of NCDs and risk factors and linked with mortality data. The longitudinal design was chosen to link baseline risk factors with mortality outcomes, allowing us to identify both modifiable and non-modifiable risk factors associated with premature CVD mortality. The NHMS surveys used a multistage, stratified probability cluster design to select a representative sample of the noninstitutionalized population in Malaysia. A detailed description of NHMS surveys is provided elsewhere [29,30,31]. We included individuals between 18 and 70 years who were interviewed in the NHMS surveys as this age group represents the prime adulthood demographic within the country. During NHMS, personal interviews and physical examinations were conducted by trained research assistants and nurses from the Ministry of Health Malaysia. The information gathered include provided baseline data. Next, the NHMS sample was linked to the mortality data through 2021 to enable a longitudinal study of the participants with a follow-up time of up to 15 years. By linking these survey data with death registry, we can comprehensively investigate the long-term impact of these risk factors on premature CVD mortality.

Data sources

Mortality status was ascertained by matching NHMS participants with the Malaysia death registry database up to 31 December 2021, using matching identifiers (National Registration Identity Card, NRIC or passport number). Both NHMS data and mortality data were obtained from the Sector for Biostatistics and Data Repository, under NIH Malaysia. The matching process was conducted by the Sector for Biostatistics and Data Repository to ensure the confidentiality of participants’ identifiers. This repository unit serves as a specialized data storage facility for both research data by the Ministry of Health Malaysia and the death database directly obtained from the authoritative National Registration Department (NRD) to support research activities under the Ministry of Health Malaysia.

In Malaysia, the law mandates that all deaths be registered with the NRD, which is responsible for issuing death certificates [32]. Deaths in the country are divided into two main categories: medically certified deaths, which occur in health facilities and are determined by medical officers based on symptoms and examination, and non-medically certified deaths, which occur outside health facilities. While death registration quality is an issue in many countries [33], Malaysia stands out as one of the few Asian countries with a functioning vital registration system. Analysing trends from 1995 to 2010, medically certified deaths increased over time, while non-medically certified deaths remained stable [34]. In 1995, non-medically certified deaths were 55%, surpassing medically certified deaths at 45% [34]. In 2021, DOSM reported an improvement, with medically certified deaths at 70.0% and non-medically certified deaths at 30.0% [35]. To ensure study quality and enhance the accuracy of cause-of-death information, we matched the NHMS sample with medically certified deaths. The underlying causes of death were classified according to the International Classification of Diseases, 10th Revision, with codes I00 to I99 considered as CVD causes of death [36].

Baseline data collection

These NHMS surveys involved trained data collectors conducting face-to-face interviews to get information on demographic characteristics and NCDs history such as DM, HPT, HPL, smoking, alcohol use, and physical activity at baseline using a standardized questionnaire. While, clinical nurses performed anthropometric measurements, including weight, height, and waist circumference, as well as clinical assessments, such as measurements of finger prick blood glucose, cholesterol levels, and blood pressure. A detailed description of the data collection method including anthropometric and clinical measurements had been reported elsewhere [29,30,31].

Definition of terms and study variables

Outcome variable

The outcome variables were survival time (time to event) and survival status. Survival times (measured in years) were calculated as the difference between the entered study date (date of NHMS interview) and the date of event (death from CVD) up to December 31, 2021. Survival status was categorized as “event” (coded as 1) for individuals who died from CVD between the ages of 30 and 70 (defined as premature mortality [3]) and “censored” (coded as 0) for those who did not experience the event during the follow-up time or reached the age of 70, whichever came first. Data on individuals who died from CVD before the age of 30 or after the age of 70 were considered censored, as well as those with deaths from causes other than CVD, or with missing or unmatched records in the death registry database.

Prognostic variables

The prognostic variables modelled for this study included socio-demographic factors such as sex, age at entry, ethnicity, location, education level, household income level, occupation, and marital status. For modifiable risk or NCD risks, this study included metabolic risk factors such as DM, HPT, HPL, obesity, abdominal obesity and behavioural risk factors including current smoking and physical inactivity. To ensure consistency in the NHMS survey data, we performed harmonization of key variables due to variations in the questionnaire across different survey years. This process was essential to ensure that the data collected from different time points could be effectively compared. For socio-demographic factors, the variables for marital status, educational level, ethnicity, and occupation status were found to be reported relatively consistently during the study period, which allowed us to make simple adjustments to create consistent categories. The most substantive harmonization involved adjusting household income variables, considering the impact of inflation over time. We also standardized the definitions for modifiable risk factors across all NHMS surveys. Detailed explanations on the categorization, definition, and adjustments made for each factor are listed in Table 1.

Table 1 List of prognostic variables
Full size table

However, for variable current alcohol use, we encountered major differences in definitions. NHMS 2011 and NHMS 2015 defined current alcohol use as the consumption of any alcoholic beverage in the past 12 months, whereas NHMS 2006 limited the period for current use to the past 1 month. Additionally, we observed a substantial amount of missing data (21.95%) for the variable current drinker, with NHMS 2006 contributing the majority of missing data (46.4%). As this variable is crucial but affected by significant missing data, we decided to exclude it from our analysis to prevent potential biases.

Statistical analyses

The data management and statistical analyses were performed using R language in RStudio integrated development environment (IDE) version 2023.03.0 [47]. We created a comprehensive cohort by merging data from three NHMS datasets using rbind function in R. Participants with missing or invalid NRCI or passport number (n = 4,863 or 6.7%) and those aged over 70 years at baseline (n = 4,144 or 5.7%) were excluded, resulting in a final dataset of 63,722 individuals (out of 72,729) aged between 18 and 70 years at baseline. This dataset was linked with mortality data up to December 31, 2021, for further analysis. Details of the data management steps can be found in Fig. 1.

Fig. 1
figure 1

Flow chart of the data cleaning process

Full size image

Incidence rate

We calculated the incidence of premature mortality from CVD in our study population, expressed as rates per 1000 person-years. Person-years were calculated based on the total sum of the number of years that each sample of the study population has been under observation. The age-specific incidence rates were obtained based on observed deaths and person-years for the overall population and sex strata. Subsequently, we used the direct standardization method by multiplying the age-specific death rate with the appropriate weight in the WHO standard population [48] to estimate the age-standardized rate for each age group/sex strata. Then, the overall ASIR was obtained from the summation of these age groups of age-standardized rate. This step of calculation follows the WHO Pan American Health Organization (PAHO) approach [49]. The details of the ASIR calculation are available in the supplementary material (Table S1-S2).

Survival analysis

This study used both semi-parametric survival analysis (Cox Proportional Hazard regression model) and parametric survival analysis (Exponential, Weibull, Gompertz, log-normal and log-logistic regression model) to model risk factors associated with premature CVD mortality. Parametric models were used alongside semi-parametric Cox regression in our study due to their advantages in providing enhanced validity and accuracy in parameter estimates, particularly when dealing with data where the proportional hazards (PH) assumption may be violated [50]. Unlike Cox regression, parametric models do not require checking the assumption of proportional hazards, making them valuable tools for more robust analysis [50].

The semi-parametric Cox PH model was first employed for univariable and multivariable regression analyses of premature CVD mortality modelling using survival package in R software. The Cox PH model is widely used for modeling hazard rates without making assumptions about the distribution of survival times [50]. The hazard ratio (HR) and their respective 95% confidence intervals (95% CI) derived from the Cox PH model were used to quantify the strength and direction of the association between potential risk factors and the risk of premature CVD mortality. By comparing the hazard rates of different groups, we can identify which factors significantly increase or decrease the risk of premature CVD mortality. For a detailed description of the semi-parametric survival analysis, including the formula for the Cox PH model and the HR, please refer to the Supplementary Materials (Supplementary Method Section). The model building involved several steps [51]: (1) univariable Cox regression (unadjusted analysis), (2) variable selection using the backward method with Akaike’s information criteria (AIC) to determine the best-fitting model, (3) assessing the assumption of hazard proportionality for the chosen model, and (4) checking for interactions between prognostic factors. The PH assumption necessitates a constant relationship between the outcome and the covariate vectors over time [50]. To assess the PH assumption, we used the cox.zph function from the survival package. This function calculates scaled Schoenfeld residuals for each covariate and includes a global test to assess their correlation with time. A significant p-value from the global test indicates a violation of the PH assumption in our model. The Schoenfeld residual plots were also used to visualized the pattern of hazard ratios over time for each covariate. If the PH assumption was violated, we proceeded to conduct the stratified Cox regression model. Stratified Cox models extend standard Cox models to accommodate covariates with non-proportional hazards.

Additionally, we explored an alternative survival analysis approach by conducting a fully parametric survival analysis based on selected covariates from a multivariable Cox PH model. To determine the optimal functional form for survival time, we constructed five types of parametric survival models including the Exponential, Weibull, Gompertz, log-normal and log-logistic distributions. We used the flexsurvreg function from flexsurv package in R to perform parametric regression model models. Model selection was based on AIC scores, with the model having the lowest AIC score considered the best fit. For a comprehensive explanation of the parametric survival analysis, including equations and interpretations of HRs and event time ratios (ETRs), please refer to the Supplementary Materials (Supplementary Method Section). In brief, in the PH model, HRs greater than 1 indicate an increased risk of premature CVD mortality, while HRs less than 1 indicate a reduced risk. Conversely, in the AFT model, an ETR below 1 suggests a shorter survival time (higher risk), and an ETR above 1 suggests a longer survival time (lower risk).

Results

The analysis included 63,722 respondents with a mean age of 41 years at baseline, of whom 45.8% were male. The median follow-up duration was 10.7 years, during which 886 (1.4%) respondents experienced premature CVD mortality. This equated to a cumulative follow-up time of 729,242 person-years. The crude incidence rate for premature CVD mortality was 1.71 per 1000 person-years and the ASIR was 1.80 per 1000 person-years. The ASIR for male and female were 2.59 and 1.15 per 1000-person year respectively. Our key findings from both semi-parametric and parametric survival models identified that demographic profiles and modifiable risk factors such as sex, ethnicity, education, DM, HPT, HPL, smoking, and abdominal obesity are significantly associated with premature CVD mortality.

The baseline characteristics of participants and the status of premature CVD mortality are shown in Table 2. Regarding demographic attributes, certain factors exhibited a higher proportion of premature CVD mortality in comparison to their counterparts. These factors include male sex (2.0%), rural residence (1.7%), Indian ethnicity (1.9%), Malay ethnicity (1.5%), the lowest income group (B40%: 1.8%), and specific marital statuses (married: 1.6%; divorced: 1.9%). Furthermore, NCD risk factors also displayed a higher proportion of premature CVD mortality. These factors encompass diagnosed DM (5.4%), diagnosed HPT (3.7%), and diagnosed HPL (3.2%). Additionally, other NCD risk factors such as GO (1.7%), AO (1.8%), current smoking status (2.1%), and physical inactivity (1.6%) were observed with a higher proportion of premature CVD mortality. The detailed distribution of premature CVD mortality status for each NHMS (NHMS 2006, 2011 and 2015) survey by listed risk factors was presented in the supplement file (Table S3).

Table 2 Characteristics of Malaysia adults aged 18–70 years at baseline sampled from National Health and morbidity surveys (2006, 2011, 2015) and 15 years follow-up of premature cardiovascular mortality
Full size table

Parametric survival analysis model

The prognostic factors for premature CVD mortality based on semi-parametric Cox PH regression, are presented in Table 3. The unadjusted model, which involved simple Cox regression, provided an initial evaluation of the individual impacts of all variables listed. Notably, significant associations were observed for all factors, except for occupation and physical activity. For the adjusted model, we followed a meticulous variable selection process. We compared the full adjusted model (comprising all listed variables) with reduced models derived through the backward elimination method based on the log-likelihood test. We found no significant difference between these models. Consequently, we selected the preliminary adjusted model derived from the reduced model from the backward selection method for further analysis. This model was adjusted with variables including sex, age of entry, ethnicity, education level, DM status, HPT status, HPL status, AO, and current smoking status.

Table 3 Semi-parametric Cox Proportional Hazard regression model for premature cardiovascular disease mortality in Malaysia (N = 63,722)
Full size table

We then assessed the PH assumption on the preliminary adjusted model and found the significant p-value from the Global test. This result demonstrated the violation of the PH assumption in our model for multiple covariates. Specifically, we observed that the Schoenfeld residual test for the age of entry variable was significant (p < 0.05), with a noticeable pattern in its Schoenfeld residual plot indicating non-constant hazard ratios over time. In light of these observations, we implemented a stratified Cox PH model, employing age of entry as the stratification variable, to address this potential violation. Following the analysis of the stratified Cox model by age of entry, the model fitness significantly improved, with the AIC value reducing from 16,152 to 1412. Additionally, the PH assumption of other covariates was met as indicated by the insignificance of the Global test, and improvements were observed in the Schoenfeld residual plots for all covariates. The Schoenfeld residual test and plot for each covariate before and after the stratified Cox model were presented in the supplement file, Figure S1 and Figure S2. Table 3 presents outcomes derived from the stratified Cox PH model, revealing multiple independent factors with statistically significant associations to premature CVD mortality.

For parametric survival analysis, we first explore the parametric distributions. The different shapes of parametric distributions are illustrated in Fig. 2. Specifically, Fig. 2(a) depicts the distribution for the survival model, while Fig. 2b) shows the distribution for the hazard model. These distributions were fitted to our premature CVD mortality data, estimating intercept-only parametric regression models without covariates. Notably, the best-performing models were those that support an arc-shaped pattern, including the log-normal (AIC = 13624) and log-logistic (AIC = 13623) distributions. Additionally, models with a monotonic trend, such as the Weibull (AIC = 13623) and Gompertz (AIC = 13635) distributions, also demonstrated strong performance. In contrast, the Exponential distribution, which only supports a constant hazard, exhibited the highest AIC value (13656).

Fig. 2
figure 2

Parametric distribution fits to premature CVD mortality data by estimating baseline parametric models (without covariates)

Full size image

Table 4 presents the results of different parametric survival analysis models with covariates. The PH model includes distributions such as Exponential, Weibull, and Gompertz. On the other hand, the AFT model covers distributions like log-normal, log-logistic, and Weibull. Among parametric models, a log-normal distribution has the lowest AIC value, indicating its superior fit. We conclude that the log-normal distribution, which exhibits the lowest AIC value, provides the best model for the survival pattern of premature CVD mortality.

Table 4 Parametric survival analysis models of premature cardiovascular disease mortality in Malaysia
Full size table

Based on the adjusted semi-parametric stratified Cox PH model and the adjusted parametric log-normal AFT model, our study identified significant associations between premature CVD mortality and various demographic profiles as well as modifiable risk factors. Males exhibited a markedly elevated risk compared to females (HR = 2.68, 95% CI: 2.08, 3.44) and experienced a 49% shorter estimated survival time (ETR = 0.51). Each additional unit of age led to a 3% decrease in estimated survival time (ETR = 0.97). Rural residents had an increased risk (HR = 1.29, 95% CI: 1.03, 1.60) and a 13% decrease in estimated survival time (ETR = 0.87) compared to urban residents. Ethnicity displayed significant associations with heightened risks and shorter survival times, with Indians (HR = 1.93, 95% CI: 1.27, 2.94; ETR = 0.62), Malays (HR = 1.60, 95% CI: 1.19, 2.15; ETR = 0.69), and other Bumiputera (HR = 1.57, 95% CI: 1.03, 2.40; ETR = 0.75) exhibiting elevated risks compared to Chinese individuals. Lower education levels were associated with higher risks and shorter survival times: no formal education (HR = 2.21, 95% CI: 1.24, 3.93; ETR = 0.67), primary education (HR = 1.95, 95% CI: 1.24, 3.06; ETR = 0.55), and secondary education (HR = 1.89, 95% CI: 1.23, 2.91; ETR = 0.65) compared to tertiary education.

Diagnosed DM emerged as a substantial predictor (HR = 3.26, 95% CI: 2.48, 4.30; ETR = 0.37), with undiagnosed DM also showing a significant association (HR = 1.63, 95% CI: 1.04, 2.58; ETR = 0.63). Both diagnosed (HR = 1.84, 95% CI: 1.39, 2.44; ETR = 0.53) and undiagnosed HPT (HR = 1.46, 95% CI: 1.13, 1.88; ETR = 0.68) exhibited significant associations with elevated risks. Undiagnosed HPL was linked to an increased risk (HR = 1.31, 95% CI: 1.04, 1.65; ETR = 0.80), but no significant association was observed for diagnosed HPL. Current smoking (HR = 1.38, 95% CI: 1.11, 1.72; ETR = 0.81) and abdominal obesity (HR = 1.60, 95% CI: 1.23, 2.08; ETR = 0.71) significantly impacted premature CVD mortality.

We also investigated potential two-way interaction effects using the stratified Cox PH model. Table 5 presents the HR from the interaction model between DM status, HPT status, AO, and smoking. Notably, within individuals with diagnosed DM, the risk of premature CVD mortality displayed significant reductions of 58%, 51%, 54%, and 54% in the presence of diagnosed HPT (HR = 0.42, 95% CI: 0.23, 0.77), undiagnosed HPT (HR = 0.49, 95% CI: 0.24, 0.99), AO (HR = 0.46, 95% CI: 0.27–0.79), and current smoking (HR = 0.46, 95% CI: 0.25, 0.83), respectively. Furthermore, the combined effect of diagnosed HPT with current smoker and AO demonstrated a significant reduction in risk by 51% (HR = 0.49, 95% CI: 0.28, 0.88), and 43% (HR = 0.57, 95% CI: 0.34, 0.98), respectively. These findings underscore the potential moderating influence or other biases that may result in spurious associations of the risk of premature CVD mortality when including this interaction term in the model. Thus, the adjusted model without interaction (Table 3) was selected as the final Cox PH model.

Table 5 Hazard ratio of premature cardiovascular disease mortality for the interaction Cox models
Full size table

Discussion

To our knowledge, this is the first study in Malaysia to apply advanced analysis using both semi-parametric and parametric survival models to model premature CVD mortality based on large population-based surveys linked with mortality data. In our analysis, we found significant impacts from both demographic profiles and modifiable risk factors. Key findings include: (i) diagnosed DM emerged as a substantial modifiable risk factor for premature CVD mortality, followed by diagnosed HPT, current smoking, abdominal obesity, and undiagnosed HPL. (ii) among demographic profiles, males had a higher risk compared to females, and we observed ethnic and educational disparities.

Recent studies have investigated the efficiency of semi-parametric and parametric event-time models in survival analysis for various diseases [52,53,54]. Our analyses, employing both semi-parametric methods (stratified Cox PH model) and parametric models (Exponential, Weibull, Gompertz, log-normal and log-logistic distributions), consistently reveal the impact of covariates, with minor differences in HR or ETR for specific factors between these models. The best-fitting parametric model, which is the log-normal distribution (having the lowest AIC value), describes the survival pattern for premature CVD mortality. This finding is consistent with the results of other studies that have also found the log-normal distribution to be the best-fit model when comparing survival distributions [55, 56]. Both semi-parametric and parametric models consistently demonstrated that metabolic risk factors, particularly diagnosed DM, exerted the most significant influence on premature CVD mortality. Furthermore, our study also underscored the significance of demographic profiles in contributing to premature CVD mortality, with factors such as sex, education level, and ethnicity emerging as influential impacts in our findings.

Sex is linked to about 2.7-fold increased risk for premature CVD mortality and reducing survival time by half (50%). The ASIR of premature CVD mortality in our study (2.59 per 1000 person-years in males and 1.15 per 1000 person-years in females) are slightly lower than those reported in previous Asian studies, such as the PURE sub study in South Asia (6.42 per 1000 person-years in males and 3.91 per 1000 person-years in females) [14], the Tehran Lipid and Glucose Study (4.8 and 3.9 per 1000 person-years in males and females, respectively) [57], and the Framingham study in the USA (4.69 per 1000 person-years in males and 1.73 per 1000 person-years in females) [58]. However, our findings are consistent with these prior studies in which a higher incidence rate is observed in males compared to females. This observation remains consistent throughout our semi-parametric stratified and parametric survival models, emphasizing that sex is the second most significant risk factor for premature CVD mortality after diagnosed DM. Our findings align with previous meta-analyses on premature CVD mortality, which demonstrate a significantly higher age-standardized mortality rate for males compared to females [59] and consistent findings from global data from the GBD study and the WHO study, underscoring the universality of these observations [12, 60, 61].

Diagnosed DM stands out as the most influential metabolic risk factor, elevating the risk of premature CVD mortality by 3.3 times and reducing survival time by 63%, aligning with previous research highlighting its consistent importance [14, 16, 57]. Individuals with diabetes are known to face a higher risk of premature death [62], particularly among younger individuals [63]. In Malaysia, despite most diagnosed individuals seeking treatment, only a quarter effectively manage their diabetes. When considering undiagnosed DM (those unaware of their condition), all models consistently indicate a significant reduction in survival time and increased hazard risk for death. To combat this challenge, prioritizing public awareness of diabetes symptoms and risk factors, promoting regular health check-ups, and ensuring healthcare affordability and accessibility are essential steps.

Following closely behind, both diagnosed and undiagnosed HPT is linked increase in risk and reduction in survival time, further emphasizing its role in contributing to premature CVD mortality. Our findings support and extend the body of knowledge regarding the impact of HPT on CVD outcomes, consistent with earlier studies [14, 16, 57]. In Malaysia, despite a higher proportion of diagnosed HPT cases receiving treatment over the years, HPT control has remained below 40% from 2006 to 2015 [64]. This persistent challenge may be primarily attributed to poor medication adherence [65]. Achieving effective control of HPT is crucial, especially when it comes to preventing premature CVD mortality. According to GBD data [66], targeting a reduction in HPT prevalence provides the most significant benefit for reducing premature CVD-specific mortality in most regions, including Southeast Asia.

In both semi-parametric and parametric models, the presence of diagnosed HPL did not exhibit a significant impact. In contrast, consistently throughout our analysis, undiagnosed HPL emerged as an independent factor significantly associated with an elevated risk of premature CVD mortality. It’s important to note that our study used total cholesterol data obtained from finger prick measurements, rather than differentiating between specific cholesterol components like HDL (high-density lipoprotein), LDL (low-density lipoprotein), or triglycerides, as seen in some other studies [14,15,16, 57]. Nonetheless, our findings consistently support the connection between lipid profiles, including total cholesterol, and CVD mortality. This suggests that the crucial factor of being unaware (undiagnosed) of HPL may play a critical role in determining CVD mortality risk. Despite the low percentage of individuals diagnosed or aware of having HPL in our study (7.1%), a substantial proportion (68%) of them received treatment, with more than half achieving cholesterol level control (data not shown). This may explain why diagnosed HPL does not exhibit significance in all survival models.

The relationship of tobacco smoking to CVD is well established. Our survival model indicates that smoking ranks among the highest contributors to the risk of premature CVD mortality, following closely after DM and HPT, resulting in approximately a 1.6-fold increased risk and a 30% reduction in survival time. Multiple lines of evidence support the connection between tobacco smoking and the physiological, pathological, and metabolic factors that contribute to the atherosclerotic process and other mechanisms leading to CVD morbidity and mortality [14,15,16,17]. According to the GBD study [66], reducing tobacco smoking has contributed significantly to the reduction in premature CVD mortality, particularly in most regions for men. However, in South Asia, tobacco reduction alone has lowered CVD mortality for men, but it hasn’t been sufficient to prevent an expected increase in CVD mortality by 2025 [66]. They suggested large reductions in premature CVD mortality are possible by 2025 if multiple risk factor targets are achieved.

Our findings also highlight the importance of addressing AO to reduce the risk of premature CVD mortality. Abdominal obesity and fat mass have recently been linked to cardiovascular morbidity and death [67, 68]. It has been demonstrated that AO more accurately measures body fat distribution [68]. Beyond the role of GO, the importance of AO in the development of CVD and CVD death has been recognised in various research studies. Our findings support the literature since only AO was significant in our adjusted model, whereas in the crude analysis, both AO and GO (as measured by BMI) were significant. Despite the fact that BMI has been associated with death in a number of major studies, it does not, however, distinguish between fat and lean mass or take into account how fat is distributed throughout the body. Large observational studies on all-cause mortality found conflicting results on whether abdominal adiposity measures are more strongly predictive of mortality than BMI [69, 70]. However, one meta-regression study confirmed our conclusion focusing on premature mortality (younger age), reporting that the relative mortality risk associated with AO was lower in older (> 65 years old) people compared to younger adults (65 years old) [71].

Our model is supported by previous studies, which have shown that education level, along with metabolic risk factors, significantly impacts CVD outcomes with a greater effect observed in LMICs compared to HICs [14,15,16]. This could be attributed to the greater support available for individuals with low education in HICs or the wider disparities in education levels in LMICs. Moreover, low education levels are associated with a higher clustering of adverse health-related behaviours, and this association persists even after adjusting for these behaviours [72]. We also identified disparities in the burden of premature CVD mortality among individuals with lower socioeconomic status, particularly among rural residents compared to their urban counterparts in Malaysia. This finding is consistent with other studies [73, 74], which have shown that individuals living outside major cities experience a significantly increased burden of CVD outcomes. These disparities arise from the interaction of various factors at both the individual and community levels, including socioeconomic disadvantage, education levels, access to primary healthcare services, and health literacy [75].

Ethnicity is another significant demographic factor that contributes to differences in the risk of premature CVD mortality. Numerous previous studies have consistently shown that CVDs present with variations among different ethnic groups [73, 76, 77]. It is widely recognized that distinct genetic predispositions associated with different ethnicities can lead to varying effects on CVD mortality [78]. Beyond genetics, cultural and lifestyle habits also play a substantial role in influencing the risk of CVD mortality. Our study findings strongly suggest that the Chinese ethnic group in Malaysia faces a lower risk of CVD-related mortality when compared to individuals of Indian, Malay, and other Bumiputra ethnicities. In Malaysia, the Chinese population consistently reports the lowest prevalence of metabolic risk factors including DM, HPT, HPL and obesity, which further supports our findings [29, 79, 80].

The interaction effect of diabetes status with other covariates in relation to premature CVD mortality is of interest. Our findings suggest that when DM is present in combination with HPT, AO, or smoking, an inverse relationship is observed, indicating a reduced CVD mortality risk compared to models without the interaction term. This finding raises questions about a potential treatment effect or other unmeasured confounding associated with DM or HPT. For example, a meta-analysis of randomized studies involving comorbid individuals with coexisting DM and HPT indicated that reducing hemoglobin A1C (HbA1c) by 0.9% may reduce CVD events by 9% [81], and lowering systolic blood pressure can reduce the risk of myocardial infarction and stroke [82]. The interaction effect in our study presents an intriguing avenue for future research using causal inference analysis, particularly through the Directed Acyclic Graph (DAG) method. While our study focused on additive modeling, investigating potential causal pathways and interaction effects could shed light on the mechanisms at play. For instance, examining whether the observed inverse relationship between DM and CVD mortality in combination with HPT and other risk factors is indeed due to a protective effect or a result of treatment strategies could be a crucial direction.

Tackling premature CVD mortality is crucial due to its significant impact on public health, economics, and quality of life. According to GBD study [66], in many LMICs, a 25% reduction in premature CVD mortality may only be achievable if all relevant risk factor targets are met. To meet the UN target of a 25% reduction in premature CVD mortality by 2025, it’s crucial to address all risk factors comprehensively. Our findings resonate with the goals outlined in the current National Strategic Plan for Non-Communicable Diseases (NSP-NCD) in Malaysia [83], which aims to reduce the burden of NCDs in the country. For example, targeting modifiable risk factors such as smoking cessation programs, diabetes management, salt reduction strategies, and obesity prevention initiatives can effectively contribute to reducing premature CVD mortality. Additionally, addressing demographic factors such as sex, ethnicity, and education level in intervention strategies can help mitigate disparities and ensure equitable access to healthcare services. This aligns with one of the NSP-NCD objectives [83] to strengthen and orient health systems toward addressing the prevention and control of NCDs and underlying social determinants through people-centered primary health care and universal health coverage. By incorporating these findings into policy and practice, we can make substantial strides toward reducing premature CVD mortality and improving population health outcomes.

Limitations and strengths

While our study offers valuable insights into the factors influencing premature CVD mortality, it does come with certain limitations. First, we used self-reported data for some key variables such as known DM, HPT and HPL, smoking and physical activity, which may introduce recall bias or underreporting, potentially impacting the accuracy of our risk factor measurements. However, during the NHMS data collection, clinical assessments such as capillary blood sugar and total cholesterol levels, along with anthropometric measurements for obesity status, were also conducted. This strengthens our study by allowing us to capture both known or diagnosed and undiagnosed statuses for NCDs. Additionally, our findings for these variables show significant results consistent with existing literature. Second, as NHMS is a cross-sectional study capturing data at one point in time, this study focused solely on baseline assessments of exposures, omitting the consideration of changes over time and lifetime exposures. For example, individuals who smoked might have quit, or those with obesity might have experienced weight changes during the follow-up period. However, our use of real-world data collected from the general population in Malaysia enhances the external validity of our findings, making them highly applicable to real-life scenarios. Furthermore, although the relatively short follow-up period (15 years) in this study might not fully capture the long-term effects of certain risk factors on premature CVD mortality, our analysis included a large sample size of 63,722 participants aged 18 to 70. This provided substantial statistical power, enabling us to detect associations and draw meaningful conclusions. Additionally, although the study faced the challenge of a relatively small number of CVD events (1.4%), sensitivity analyses were conducted to address data imbalances. These analyses demonstrated that imbalanced data did not significantly affect the conclusions drawn from the analysis. For detailed information on the sensitivity analysis comparing treated imbalanced data with the original data for all models, please refer to the supplementary file (see Figure S3, Table S4, and Table S5). Other than that, this study employed a diverse range of survival models, including both semi-parametric Cox proportional hazard models and various parametric models, which adds strength by allowing for a comprehensive exploration of the relationships between risk factors and premature CVD mortality while considering different model assumptions. Finally, our study lays the groundwork for future research, as our findings hold the potential for causal mediation or causal inference analysis, offering a deeper understanding of how risk factors influence premature CVD mortality through various pathways.

Conclusion

Our study addresses the significant knowledge gap regarding premature cardiovascular disease mortality in Malaysia. By employing survival analysis on extensive data, we identified demographic profile and modifiable risk factors, such as sex, ethnicity, education, diabetes mellitus, hypertension, hypercholesterolemia, smoking, and abdominal obesity, play a crucial role in premature CVD mortality. These findings provide valuable insights for targeted interventions and policies to mitigate the impact of these risk factors and reduce the burden of premature CVD mortality in the Malaysian adult population. Future research should investigate the longitudinal impact of early lifestyle interventions on reducing these risk factors. Additionally, studies should explore the role of healthcare accessibility and quality in managing and preventing CVD among high-risk populations. Furthermore, it is essential to examine the effectiveness of community-based health promotion programs and policies aimed at reducing modifiable risk factors at a population level.

Data availability

The datasets analysed during the current study are available from the corresponding author on reasonable request. Additionally, the R code for this analysis can be accessed via the following link: https://github.com/shakirarodzlan/PhD_Survival_Analysis_PMCVD.git.

Abbreviations

CVD:

Cardiovascular disease

IHD:

Ischemic heart disease

WHO:

World Health Organization

NCDs:

Non-communicable diseases

YLL:

Years of Life Lost

DM:

Diabetes mellitus

HPT:

Hypertension

HPL:

Hyperlipidaemia

LMICs:

Low- and middle-income countries

MICs:

Middle-income countries

HICs:

High-income countries

LICs:

Low-income countries

PURE:

Prospective Urban Rural Epidemiology study

NHMS:

National Health and Morbidity Survey

IPH:

Institute for Public Health

NIH:

National Institute of Health

NRIC:

National Registration Identity Card

NRD:

National Registration Department

DOSM:

Department of Statistics Malaysia

CPI:

Consumer price indices

BMI:

Body Mass Index

GO:

General obesity

AO:

Abdominal obesity

IPAQ:

International Physical Activity Questionnaire

GPAQ:

Global Physical Activity Questionnaire

MET:

Metabolic equivalent task

IDE:

Integrated development environment

ASIR:

Age-standardized incidence rate

PAHO:

Pan American Health Organization

PH:

Proportional hazard

AIC:

Akaike’s information criteria

ASIR:

Age standardized incidence rate

AFT:

Accelerated failure time

HR:

Hazard ratio

ETR:

Event Time Ratio

CI:

Confidence Interval

References

  1. WHO. Cardiovascular diseases (CVDs). Key Facts. Word Health Organization (WHO). 2021 [cited 2022 Oct 9]. https://www.who.int/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds).

  2. Roth GA, Abate D, Abate KH, Abay SM, Abbafati C, Abbasi N, et al. Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980–2017: a systematic analysis for the global burden of Disease Study 2017. Lancet. 2018;392(10159):1736–88.

    Article  Google Scholar 

  3. WHO. Indicator Metadata Registry List. Premature mortality from noncommunicable disease. World Health Organization (WHO). 2023 [cited 2023 Mar 30]. https://www.who.int/data/gho/indicator-metadata-registry/imr-details/3411

  4. WHO. Noncommunicable diseases: Key facts. World Health Organization (WHO). 2022 [cited 2023 Aug 21]. https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases

  5. Roth G. Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2017 (GBD 2017) Results. Seattle, United States: Institute for Health Metrics and Evaluation (IHME), 2018. The Lancet. 2018;392:1736–88.

  6. WHO. Noncommunicable diseases country profiles 2018: Malaysia [Internet]. 2018. https://apps.who.int/iris/handle/10665/274512

  7. Noor NM, Muzafar PMM, Khalidi JR. Health of the Nation: Measuring Burden of Disease. Kuala Lumpur: Khazanah Research Institute. License: Creative Commons Attribution CC BY 3.0.; 2020. (Page 9).

  8. Khaw WF, Chan YM, Nasaruddin NH, Alias N, Tan L, Ganapathy SS. Malaysian burden of disease: years of life lost due to premature deaths. BMC Public Health. 2023;23(1):1–11.

    Article  Google Scholar 

  9. WHO. Global action plan for the prevention and control of noncommunicable diseases 2013–2020. World Health Organization (WHO); 2013.

  10. Mohammadnezhad M, Mangum T, May W, Lucas JJ, Ailson S, editors. others. Common modifiable and non-modifiable risk factors of cardiovascular disease (CVD) among Pacific countries. World J Cardiovasc Surg. 2016;6(11):153.

  11. Choudhury MSJH, Chowdhury MTI, Nayeem A, Jahan WA. Modifiable and non-modifiable risk factors of stroke: a review update. J Natl Inst Neurosciences Bangladesh. 2015;1(1):22–6.

    Article  Google Scholar 

  12. Zhang J, Jin Y, Jia P, Li N, Zheng ZJ. Global gender disparities in premature death from Cardiovascular Disease, and their associations with Country Capacity for Noncommunicable Disease Prevention and Control. Int J Environ Res Public Health. 2021;18:19.

    Google Scholar 

  13. Lotufo PA. Cardiovascular diseases in Brazil: premature mortality, risk factors and priorities for action. Comments on the preliminary results from the Brazilian National Health Survey (PNS), 2013. Volume 133. Sao Paulo Medical Journal. Brazil; 2015. pp. 69–72.

  14. Joseph P, Kutty VR, Mohan V, Kumar R, Mony P, Vijayakumar K, et al. Cardiovascular disease, mortality, and their associations with modifiable risk factors in a multi-national South Asia cohort: a PURE substudy. Eur Heart J. 2022;43(30):2831–40.

    Article  PubMed  Google Scholar 

  15. Li S, Liu Z, Joseph P, Hu B, Yin L, Tse LA, et al. Modifiable risk factors associated with cardiovascular disease and mortality in China: a PURE substudy. Eur Heart J. 2022;43(30):2852–63.

    Article  PubMed  Google Scholar 

  16. Yusuf S, Joseph P, Rangarajan S, Islam S, Mente A, Hystad P, et al. Modifiable risk factors, cardiovascular disease, and mortality in 155 722 individuals from 21 high-income, middle-income, and low-income countries (PURE): a prospective cohort study. Lancet. 2020;395(10226):795–808.

    Article  PubMed  Google Scholar 

  17. Eslami A, Naghibi Irvani SS, Ramezankhani A, Fekri N, Asadi K, Azizi F et al. Incidence and associated risk factors for premature death in the Tehran lipid and glucose study cohort, Iran. BMC Public Health. 2019;19(1).

  18. Leon DA, Saburova L, Tomkins S, Andreev E, Kiryanov N, McKee M, et al. Hazardous alcohol drinking and premature mortality in Russia: a population based case-control study. Lancet. 2007;369(9578):2001–9.

    Article  PubMed  Google Scholar 

  19. WHO. Physical activity. Key Facts. World Health Organization (WHO). 2021 [cited 2023 Aug 22]. https://www.who.int/news-room/fact-sheets/detail/physic

  20. Zhang YB, Pan XF, Chen J, Cao A, Xia L, Zhang Y, et al. Combined lifestyle factors, all-cause mortality and cardiovascular disease: a systematic review and meta-analysis of prospective cohort studies. J Epidemiol Community Health. 2021;75(1):92–9.

    Article  PubMed  Google Scholar 

  21. Ezzati M, Obermeyer Z, Tzoulaki I, Mayosi BM, Elliott P, Leon DA. Contributions of risk factors and medical care to cardiovascular mortality trends. Nat Rev Cardiol. 2015;12(9):508–30.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Jagannathan R, Patel SA, Ali MK, Narayan KMV. Global updates on Cardiovascular Disease Mortality trends and Attribution of traditional risk factors. Curr Diab Rep. 2019;19(7):44.

    Article  PubMed  Google Scholar 

  23. Yusuf S, Hawken S, Ôunpuu S, Dans T, Avezum A, Lanas F, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 2004;364(9438):937–52.

    Article  PubMed  Google Scholar 

  24. Mathers CD, Loncar D. Projections of global mortality and Burden of Disease from 2002 to 2030. PLoS Med. 2006;3(11):e442.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Huxley RR, Hirakawa Y, Hussain MA, Aekplakorn W, Wang X, Peters SAE, et al. Age-and sex-specific burden of cardiovascular disease attributable to 5 major and modifiable risk factors in 10 Asian countries of the Western Pacific Region. Circ J. 2015;79(8):1662–74.

    Article  PubMed  Google Scholar 

  26. Ohira T, Iso H. Cardiovascular disease epidemiology in Asia–an overview–. Circ J. 2013;77(7):1646–52.

    Article  PubMed  Google Scholar 

  27. Zhao D. Epidemiological features of cardiovascular disease in Asia. JACC: Asia. 2021;1(1):1–13.

    PubMed  PubMed Central  Google Scholar 

  28. Roth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, Abyu G, et al. Global, Regional, and National Burden of Cardiovascular diseases for 10 causes, 1990 to 2015. J Am Coll Cardiol. 2017;70(1):1–25.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Institute for Public Health Malaysia. The Third National Health and Morbidity Survey (NHM III). 2006: General Findings [Internet]. Institute for Public Health (IPH), National Institute of Health, Ministry of Health Malaysia. Kuala Lumpur, Malaysia; 2008. https://iku.gov.my/images/IKU/Document/REPORT/2006/GeneralFindings.pdf

  30. Institute for Public Health Malaysia. National Health and Morbidity Survey 2011. (NHMS 2011). Vol. 1: Methodology and General Findings [Internet]. Volume 1. Institute for Public Health, National Institutes of Health, Ministry of Health Malaysia. Kuala Lumpur, Malaysia; 2011. 13–19 p. https://iku.gov.my/images/IKU/Document/REPORT/NHMS2011-VolumeI.pdf

  31. Institute for Public Health Malaysia. National Health and Morbidity Survey 2015. (NHMS 2015). Vol I: Methodology and General Findings [Internet]. Volume I. Institute for Public Health, National Institutes of Health, Ministry of Health Malaysia, Kuala Lumpur, Malaysia; 2015. 16–21 p. https://iku.gov.my/images/IKU/Document/REPORT/nhmsreport2015vol1.pdf

  32. Malaysian Legislation, Births. and Deaths Registration Act 1957 (Revised 1983) [Internet]. Commissioner of Law Revision, Malaysia under the authority of the Revision of Laws Act 1968 in collaboration with Percetakan Nasional Malaysia Bhd. 2006 [cited 2023 Aug 21]. http://www.commonlii.org/my/legis/consol_act/badra19571983376/

  33. Mathers CD, Ma Fat D, Inoue M, Rao C, Lopez AD. Counting the dead and what they died from: an assessment of the global status of cause of death data. Bull World Health Organ. 2005;83:171–c177.

    PubMed  PubMed Central  Google Scholar 

  34. Adnan TH, Bujang MA, Supramaniam P, Mustaâ N, Lazim SSMAT, Haniff J et al. Trend Analysis of medically certified death in Malaysia, 1995–2010. J Health Inf Dev Ctries. 2012;6(1).

  35. DOSM. Statistics on Cause of Death, Malaysia. 2022 [Internet]. Copyright ©2023 Department of Statistics Malaysia (DOSM). 2023 [cited 2023 Aug 21]. https://www.dosm.gov.my/site/downloadrelease?id=statistics-on-causes-of-death-malaysia-2022&lang=English

  36. National Center for Health Statistics. International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) [Internet]. National Center for Health Statistics, CDC. 2022 [cited 2022 Mar 26]. https://www.cdc.gov/nchs/icd/icd10cm.htm

  37. WorldData.info. Inflation rates in Malaysia: Historical inflation rates in comparison [Internet]. © WorldData.info. 2023 [cited 2023 Aug 23]. https://www.worlddata.info/asia/malaysia/inflation-rates.php

  38. DOSM. Consumer Price Index Malaysia [Internet]. Copyright ©2023 Department of Statistics Malaysia (DOSM) Official Portal. 2023 [cited 2023 Aug 23]. https://www.dosm.gov.my/portal-main/release-content/14da79c9-81c6-11ed-96a6-1866daa77ef9

  39. WHO. Definition and diagnosis of diabetes mellitus and intermediate hyperglycaemia: report of a WHO/IDF consultation. World Health Organization (WHO). World Health Organization; 2006.

  40. Working Group on Hypertension CPG. Clinical Practice Guideline: Management of Hypertension – 5th Edition. © Copyright 2018 by Malaysian Society of Hypertension. 2018. 28 p.

  41. National Cholesterol Education Program (US). Third report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III). Expert Pan. The Program; 2002.

  42. WHO. The Asia-Pacific perspective: redefining obesity and its treatment. World Health Organization (WHO) [Internet]. 2000;17–20. https://apps.who.int/iris/handle/10665/206936

  43. Craig CL, Marshall AL, Sjöström M, Bauman AE, Booth ML, Ainsworth BE, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003;35(8):1381–95.

    Article  PubMed  Google Scholar 

  44. Armstrong T, Bull F. Development of the world health organization global physical activity questionnaire (GPAQ). J Public Health (Bangkok). 2006;14:66–70.

    Article  Google Scholar 

  45. WHO. Global physical activity questionnaire (GPAQ) analysis guide. World Health Organization (WHO). Surveillance and Population-Based Prevention Prevention of Noncommunicable Diseases Department World Health Organization. 20 Avenue Appia, 1211 Geneva 27, Switzerland; 2012. https://www.who.int/docs/default-source/ncds/ncd-surveillance/gpaq-analysis-guide.pdf

  46. Committee IR. others. Guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ)-short and long forms. http://www ipaq ki se/scoring pdf. 2005.

  47. RStudio Team. RStudio: Integrated Development for R, RStudio, PBC, Boston. MA [Internet]. © 2023 Posit Software, PBC formerly RStudio, PBC. 2020. http://www.rstudio.com/

  48. Ahmad OB, Boschi-Pinto C, Lopez AD, Murray CJL, Lozano R, Inoue M et al. Age standardization of rates: a new WHO standard. Geneva: World Health Organization. 2001;9(10):1–14.

  49. PAHO. PAHO. Age standardized mortality rates per 100,000 population deaths less 70 years, 2009 [Internet]. Pan American Health Organization (PAHO). 2009 [cited 2023 Sep 7]. https://www.paho.org/en/documents/paho-age-standardized-mortality-rates-100000-population-deaths-less-70-years-2009

  50. Kleinbaum DG, Klein M. Survival analysis a self-learning text. Springer; 1996.

  51. Hosmer DW Jr, Lemeshow S, May S. Applied survival analysis: regression modeling of time-to-event data. Volume 618. John Wiley \& Sons; 2011.

  52. Wang SJ, Kalpathy-Cramer J, Kim JS, Fuller CD, Thomas CR Jr. Parametric survival models for predicting the benefit of adjuvant chemoradiotherapy in gallbladder cancer. In: AMIA Annual Symposium Proceedings. 2010. p. 847.

  53. Pourhoseingholi MA, Hajizadeh E, Moghimi Dehkordi B, Safaee A, Abadi A, Zali MR, et al. Comparing Cox regression and parametric models for survival of patients with gastric carcinoma. Asian Pac J Cancer Prev. 2007;8(3):412.

    PubMed  Google Scholar 

  54. Orbe J, Ferreira E, Núñez-Antón V. Comparing proportional hazards and accelerated failure time models for survival analysis. Stat Med. 2002;21(22):3493–510.

    Article  PubMed  Google Scholar 

  55. Alexopoulos G, Zhang J, Karampelas I, Patel M, Kemp J, Coppens J, et al. Long-term time series forecasting and updates on Survival Analysis of Glioblastoma Multiforme: a 1975–2018 Population-based study. Neuroepidemiology. 2022;56(2):75–89.

    Article  PubMed  Google Scholar 

  56. Ebrahimi V, Khademian MH, Masoumi SJ, Morvaridi MR, Ezzatzadegan Jahromi S. Factors influencing survival time of hemodialysis patients; time to event analysis using parametric models: a cohort study. BMC Nephrol. 2019;20(1):1–9.

    Article  Google Scholar 

  57. Eslami A, Mozaffary A, Derakhshan A, Azizi F, Khalili D, Hadaegh F. Sex-specific incidence rates and risk factors of premature cardiovascular disease. A long term follow up of the Tehran lipid and glucose study. Int J Cardiol. 2017;227:826–32.

    Article  PubMed  Google Scholar 

  58. Vasan RS, Song RJ, van den Heuvel ER. Temporal trends in incidence of premature Cardiovascular Disease over the past 7 decades: the Framingham Heart Study. J Am Heart Assoc. 2022;11(19):e026497.

    Article  PubMed  PubMed Central  Google Scholar 

  59. Hasani WSR, Muhamad NA, Hanis TM, Maamor NH, Chen XW, Omar MA, et al. The global estimate of premature cardiovascular mortality: a systematic review and meta-analysis of age-standardized mortality rate. BMC Public Health. 2023;23(1):1561.

    Article  PubMed  PubMed Central  Google Scholar 

  60. Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global burden of cardiovascular diseases and risk factors, 1990–2019: update from the GBD 2019 study. J Am Coll Cardiol. 2020;76(25):2982–3021.

    Article  PubMed  PubMed Central  Google Scholar 

  61. Masaebi F, Salehi M, Kazemi M, Vahabi N, Azizmohammad Looha M, Zayeri F. Trend analysis of disability adjusted life years due to cardiovascular diseases: results from the global burden of disease study 2019. BMC Public Health. 2021;21(1):1268.

    Article  PubMed  PubMed Central  Google Scholar 

  62. Mulnier HE, Seaman HE, Raleigh VS, Soedamah-Muthu SS, Colhoun HM, Lawrenson RA. Mortality in people with type 2 diabetes in the UK. Diabet Med. 2006;23(5):516–21.

    Article  CAS  PubMed  Google Scholar 

  63. Kerr D, Partridge H, Knott J, Thomas PW. HbA1c 3 months after diagnosis predicts premature mortality in patients with new onset type 2 diabetes. Diabet Med. 2011;28(12):1520–4.

    Article  CAS  PubMed  Google Scholar 

  64. Ab Majid NL, Omar MA, Khoo YY, Mahadir Naidu B, Ling Miaw Yn J, Rodzlan Hasani WS, et al. Prevalence, awareness, treatment and control of hypertension in the Malaysian population: findings from the National Health and Morbidity Survey 2006–2015. J Hum Hypertens. 2018;32(8–9):617–24.

    Article  PubMed  PubMed Central  Google Scholar 

  65. Ramli A, Ahmad NS, Paraidathathu T. Medication adherence among hypertensive patients of primary health clinics in Malaysia. Patient Prefer Adherence. 2012;613–22.

  66. Roth GA, Nguyen G, Forouzanfar MH, Mokdad AH, Naghavi M, Murray CJL. Estimates of global and regional premature cardiovascular mortality in 2025. Circulation. 2015;132(13):1270–82.

    Article  PubMed  Google Scholar 

  67. Yusuf S, Hawken S, Ounpuu S, Bautista L, Franzosi MG, Commerford P, et al. Obesity and the risk of myocardial infarction in 27 000 participants from 52 countries: a case-control study. Lancet. 2005;366(9497):1640–9.

    Article  PubMed  Google Scholar 

  68. Snijder MB, Van Dam RM, Visser M, Seidell JC. What aspects of body fat are particularly hazardous and how do we measure them? Int J Epidemiol. 2006;35(1):83–92.

    Article  CAS  PubMed  Google Scholar 

  69. Katzmarzyk PT, Janssen I, Ardern CI. Physical inactivity, excess adiposity and premature mortality. Obes Rev. 2003;4(4):257–90.

    Article  CAS  PubMed  Google Scholar 

  70. Baik I, Ascherio A, Rimm EB, Giovannucci E, Spiegelman D, Stampfer MJ, et al. Adiposity and mortality in men. Am J Epidemiol. 2000;152(3):264–71.

    Article  CAS  PubMed  Google Scholar 

  71. Carmienke S, Freitag MH, Pischon T, Schlattmann P, Fankhaenel T, Goebel H, et al. General and abdominal obesity parameters and their combination in relation to mortality: a systematic review and meta-regression analysis. Eur J Clin Nutr. 2013;67(6):573–85.

    Article  CAS  PubMed  Google Scholar 

  72. Skalamera J, Hummer RA. Educational attainment and the clustering of health-related behavior among US young adults. Prev Med (Baltim). 2016;84:83–9.

    Article  Google Scholar 

  73. Chen Y, Freedman ND, Albert PS, Huxley RR, Shiels MS, Withrow DR, et al. Association of Cardiovascular Disease with premature mortality in the United States. JAMA Cardiol. 2019;4(12):1230–8.

    Article  PubMed  PubMed Central  Google Scholar 

  74. Khan SU, Javed Z, Lone AN, Dani SS, Amin Z, Al-Kindi SG, et al. Social vulnerability and premature cardiovascular mortality among US counties, 2014 to 2018. Circulation. 2021;144(16):1272–9.

    Article  PubMed  Google Scholar 

  75. Wu JR, Moser DK, DeWalt DA, Rayens MK, Dracup K. Health literacy mediates the relationship between age and health outcomes in patients with heart failure. Circ Heart Fail. 2016;9(1):e002250.

    Article  PubMed  PubMed Central  Google Scholar 

  76. Lanza GA. Ethnic variations in acute coronary syndromes. Vol. 90, heart. BMJ Publishing Group Ltd; 2004. pp. 595–7.

  77. Anand SS, Yusuf S, Vuksan V, Devanesen S, Teo KK, Montague PA, et al. Differences in risk factors, atherosclerosis, and cardiovascular disease between ethnic groups in Canada: the Study of Health Assessment and Risk in ethnic groups (SHARE). Lancet. 2000;356(9226):279–84.

    Article  CAS  PubMed  Google Scholar 

  78. Poulsen P, Vaag A, Kyvik K, Beck-Nielsen H. Genetic versus environmental aetiology of the metabolic syndrome among male and female twins. Diabetologia. 2001;44:537–43.

    Article  CAS  PubMed  Google Scholar 

  79. Institute for Public Health Malaysia. National Health and Morbidity Survey (NHMS) 2019: vol. I: NCDs–Non-Communicable diseases: risk factors and other Health problems. Ministry of Health Malaysia Putrajaya, Malaysia; 2020.

  80. Mohamud WNW, Ismail A al, Sharifuddin S, Ismail A, Musa IS, Kadir KI. Prevalence of metabolic syndrome and its risk factors in adult malaysians: results of a nationwide survey. Diabetes Res Clin Pract. 2011;91(2):239–45.

    Article  PubMed  Google Scholar 

  81. Turnbull FM, Abraira C, Anderson RJ, Byington RP, Chalmers JP, Duckworth WC, et al. Intensive glucose control and macrovascular outcomes in type 2 diabetes. Diabetologia. 2009;52:2288–98.

    Article  CAS  PubMed  Google Scholar 

  82. Emdin CA, Rahimi K, Neal B, Callender T, Perkovic V, Patel A. Blood pressure lowering in type 2 diabetes: a systematic review and meta-analysis. JAMA. 2015;313(6):603–15.

    Article  PubMed  Google Scholar 

  83. Ministry of Health Ministry Malaysia. National Strategic Plan for Non-Communicable Disease (NSPNCD) 2016–2025. Section NCD (NCD), Division, Disease Control, Ministry of Health Putrajaya Malaysia, editors. Copyright © Ministry of Health Malaysia. 2016. https://www.moh.gov.my/moh/resources/Penerbitan/Rujukan/NCD/National Strategic Plan/FINAL_NSPNCD.pdf.

Download references

Acknowledgements

We would like to thank the Director-General of Health Malaysia for his permission to publish this article.

Funding

No funding was obtained for this study.

Author information

Authors and Affiliations

  1. Department of Community Medicine, School of Medical Sciences, Universiti Sains Malaysia, Kubang Kerian, 16150, Kelantan, Malaysia

    Wan Shakira Rodzlan Hasani, Kamarul Imran Musa & Tengku Muhammad Hanis

  2. Institute for Public Health, National Institutes of Health, Ministry of Health Malaysia, Setia Alam, 40170, Selangor, Malaysia

    Wan Shakira Rodzlan Hasani, Shubash Shander Ganapathy, Muhammad Fadhli Mohd Yusoff & Noor Ani Ahmad

  3. Sector for Biostatistics and Data Repository, National Institutes of Health, Ministry of Health Malaysia, Setia Alam, 40170, Selangor, Malaysia

    Mohd Azahadi Omar

  4. Biostatistics and Research Methodology Unit, School of Medical Sciences, Universiti Sains Malaysia, Kelantan, 16150, Malaysia

    Yee Cheng Kueh

Authors
  1. Wan Shakira Rodzlan Hasani
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Kamarul Imran Musa
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Mohd Azahadi Omar
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Tengku Muhammad Hanis
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Yee Cheng Kueh
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Shubash Shander Ganapathy
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Muhammad Fadhli Mohd Yusoff
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Noor Ani Ahmad
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

Conceptualization: WSRH and KIM. Data curation: WSRH, KIM, TMH, MAO and SSG. Formal analysis: WSRH. Investigation: WSRH, KIM and TMH. Methodology: WSRH, KIM, MAO, TMH, YCK and MFMY. Project administration, resources and software: WSRH. Writing – original draft: WSRH. Writing – editing: WSRH. All authors critically revised the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Kamarul Imran Musa.

Ethics declarations

Ethics approval and consent to participate

The study protocol and approvals for the use of NHMS data and the linkage with mortality files were obtained from the Medical Research and Ethics Committee (MREC) of the Ministry of Health Malaysia and registered with the National Medical Research Register (NMRR ID-22-00231-MOX). Additionally, this study received approval from the Human Research Ethics Committee of Universiti Sains Malaysia (ID: USM/JEPeM/22030181). It’s important to note that all data were anonymized before use in this study. As the data utilized in this study was originally collected during NHMS surveys where informed consent had been obtained from participants for each individual survey, the requirement for informed consent was waived by both the Medical Research and Ethics Committee (MREC) of the Ministry of Health Malaysia and the Human Research Ethics Committee of Universiti Sains Malaysia for the purpose of this secondary data analysis.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hasani, W.S.R., Musa, K.I., Omar, M.A. et al. Prognostic factors for premature cardiovascular disease mortality in Malaysia: a modelling approach using semi-parametric and parametric survival analysis with national health and morbidity survey linked mortality data. BMC Public Health 24, 2745 (2024). https://doi.org/10.1186/s12889-024-20104-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12889-024-20104-9

Keywords

BMC Public Health

ISSN: 1471-2458

Contact us