Predicting Health-Related Quality of Life Among Chinese Residents: Latent Class Analysis Based on Panel Survey Data

Introduction

Human health is increasingly influenced by factors such as population structure, socioeconomic development, environmental conditions, and changes in lifestyle and dietary habits. The cumulative impact of these factors has increased the prevalence of chronic diseases such as obesity,¹ hypertension,² and cardiovascular diseases,³ as well as the risk of cancer,⁴ thereby affecting individuals’ physical health status. Moreover, these factors contribute to stress, fatigue, sleep disorders, anxiety, and depression,^5,6 reducing psychological and social well-being. Human health faces threats from multiple sources, resulting in significant impacts on quality of life. Accordingly, health-related quality of life (HR-QoL) is a crucial concern.

HR-QoL is a multidimensional measure, reflecting the aspects of physical, mental, emotional, and social health,⁷ and is a critical predictor for health care utilization,⁸ mortality,⁸ morbidity,⁹ and poor health.¹⁰ For individuals to lead fulfilling lives and for society to thrive, it is crucial to maintain high HR-QoL.¹¹ HR-QoL is a pressing issue in China due to a combination of factors, including an aging population, high rates of chronic disease, and environmental challenges.¹² These issues are compounded by disparities in access to healthcare services, with rural areas and low-income individuals facing greater barriers to care.¹³ The health status of residents in developing countries has improved in recent years,¹⁴ as both individuals and governments have become increasingly focused on health.¹⁵ China’s new “Healthy China” strategy and other health-related policies have improved residents’ health status and quality of life and promoted research on health, health services, and influencing factors.^16,17 However, due to increasing social competition and the ongoing effects of the COVID-19 pandemic, the overall QoL of the Chinese population decreased between 2008 and 2020.¹⁸ The generalizability of findings regarding HR-QoL is inconsistent, as studies have been conducted on different populations and at different times. China is the largest developing country in the world, and the HR-QoL of Chinese residents is constantly changing. Therefore, it is necessary to re-evaluate the changing trends in the HR-QoL of Chinese residents.

HR-QoL is an important indicator of the comprehensive health of a population and may be affected by many factors. One study found that need factors, such as self-reported health, had the most significant impact on HR-QoL among older adults in Korea.¹⁹ The predisposing factors were age and education level, and the enabling factors were income and daily living activities. One study in China found that poor HR-QoL was associated with lower socioeconomic status, lack of family support, and lack of social support.²⁰ Several studies have shown that HR-QoL is affected by various demographic characteristics.^19,21 Other study have found that social capital factors have less of an impact on HR-QoL than do demographic characteristics.²² Thus, demographic characteristics can be used to predict the HR-QoL of Chinese residents, and intervention strategies based on demographic characteristics can help improve it.

Most studies on HR-QoL have used cross-sectional surveys.^23–25 The cross-sectional nature of these studies precludes the effect of time on HR-QoL, making it difficult to obtain information on the health trends of the population. Moreover, studies in this area have focused on particular groups, such as older adults,¹⁰ pregnant women,²⁶ adolescents,²⁷ children,²⁸ college students,²⁹ patients with cancer,³⁰ patients with cardiopathy,³¹ patients with sarcoidosis,³² and stone formers.³³ By contrast, longitudinal studies assessing HR-QoL across entire populations remain relatively scarce. Moreover, few studies have sought to identify latent classes in HR-QoL domains and factors that distinguish these classes. Thus, studies should use data obtained from random sampling of the entire population, and panel survey data should be employed to analyze and predict the health trends and influencing factors of various population groups. This study aimed to identify distinct trends in HR-QoL among Chinese residents and to analyze the demographic characteristics contributing to these trends.

Materials and Methods

Design

Building on a pilot study conducted in 2015, the Statistical Information Center of the National Health and Family Planning Commission of China initiated the second phase of monitoring of residents’ healthcare service utilization in 2016. This study used data from a survey of healthcare service utilization behavior among the residents of Hubei Province, China. From July to December 2016, residents’ healthcare service utilization was monitored monthly. The monitoring process primarily consisted of field investigators and auditors visiting households. They conducted one-on-one interviews using standardized monitoring questionnaires, including the Baseline Monitoring Form and Monitoring Form, developed by the Chinese National Health and Family Planning Commission’s Statistical Information Center.³⁴ The Baseline Monitoring Form includes information on the general situation of the family and the personal situation of the family members, with a primary focus on basic demographic characteristics and the prevalence of chronic diseases. The Monitoring Form includes changes in family members, their health status, disease surveillance, medical visits and hospitalizations during the monitoring cycle. It primarily focuses on the residents’ HR-QoL, diseases, and medical behavior.

The investigators were trained to use the instruments and interviewed the participants prior to the survey. During the survey, the investigators provided guidance for all questionnaire responses. After data collection, households were contacted to verify data accuracy, and the data were then anonymized. All participants provided informed consent before participating in the survey.

Participants

The survey was a long-term follow-up investigation requiring a high level of trust and commitment from the participants. Fostering trust and long-term cooperation among the surveyed households required obtaining a firm understanding of their fundamental circumstances prior to home visits. A pilot that was conducted in two places (urban city and rural town) in Hubei Province monitored residents’ healthcare service utilization behavior from July to December 2016. Five communities and townships within each of the two sample areas were selected as sampling points. The survey used stratified random sampling to select the initial set of participants, and then tracked their health and other related data on a monthly basis. Initially, 1597 participants were selected and monitored. Complete basic information was obtained from 1517 participants, resulting in an effective response rate of 94.99% (1517/1597). Missing data were obtained using interpolation methods.

Instruments

Demographic characteristic measures included respondents’ district, gender, presence of chronic diseases, family annual income, age, education level, marital status, and employment status.

The European Quality of Life Instrument (EuroQol; later named “EQ-5D”) is a five-item, patient-based utility instrument that measures HR-QoL.³⁵ The EQ-5D is a brief self-reported questionnaire that assesses the respondents’ general health. Respondents were asked to rate their health on the day that the questionnaire was administered. The EQ-5D scale has good reliability and validity when used to assess HR-QoL among various populations,^18,36–39 including the Chinese population.^18,39 The EQ-5D questionnaire contains two parts. The first is a descriptive component comprising five dimensions: mobility (MO), self-care (SC), usual activities (UA), pain/discomfort (PD), and anxiety/depression (AD). This is commonly used to calculate the EQ-5D index in economic analyses using a country-specific value set. The second is the EQ-visual analog scale (EQ-VAS), which measures the respondents’ self-rated health on a 20-cm vertical line (VAS), with endpoints labeled “worst imaginable health state” at 0 and “best imaginable health state” at 100.⁴⁰ The EQ-VAS scores range from 0 to 1 and are calculated by dividing the number marked on the scale by 100. To more accurately reflect changes in the HR-QoL of Chinese residents, facilitate the comparison of various results, and minimize the impact of respondents’ cultural background and comprehension on the accuracy of the findings, we selected a utility index specifically suited for the Chinese population. This index was used to calculate health utility values, providing a clearer reflection of the HR-QoL in this demographic. Health utility values were divided into scores ranging from 0 (death) to 1 (full health) using the Chinese version of the time trade-off (TTO).⁴¹

Analysis Methods

The data were presented as frequencies. Longitudinal latent class analysis (LLCA) was used to analyze participants with similar HR-QoL trajectories. Multinomial logistic regression was used to examine the association between the demographic characteristics and class membership. LLCA was used to obtain the main types of HR-QoL, which were then matched with the demographic characteristics of the participants to obtain the characteristics of each class and predict the tendency of health changes based on specific features.

The data were coded, revised, entered, and double-checked by investigators. The gathered information was loaded into EpiData 3.2 version for sorting and subsequently extracted into Excel 2019 for filtering and cleaning. Data were analyzed using SPSS 26.0 (IBM Corp., Armonk, NYC, USA), Mplus 8.5 (Muthen & Muthen, Los Angeles, NY, USA), and Stata 12 (STATA Corp). First, the general characteristics of the participants were analyzed to determine frequencies and percentages. Second, a latent class analysis was performed for HR-QoL. The Akaike information criterion (AIC), Bayesian information criterion (BIC), sample size-adjusted BIC (saBIC), Lo–Mendel–Rubin likelihood ratio test (LMR), and bootstrap likelihood ratio test (BLRT) were used to assess the goodness of fit for the selection of latent classes. The lowest absolute values of the AIC and BIC indicate excellent model fitness.⁴² The AIC and BIC are affected by sample size, whereas saBIC is not.⁴³ The LMR and BLRT provide significance probabilities for the class models. The null hypothesis posits that the latent class model will be rejected, whereas the alternative hypothesis posits that the latent class model will be accepted. The alternative hypothesis was accepted if probable significance was less than 0.05. Third, a multinomial logistic regression analysis was used to identify the demographic characteristics that predicted the HR-QoL in individual latent classes. In this analysis, the dependent variable was the HR-QoL class of residents, and various demographic factors served as independent variables to assess their predictive influence on HR-QoL classes. The low HR-QoL class, representing individuals in poorer health who may require additional resources and attention, was used as the reference class. By using the low HR-QoL class as a reference, the model offers a more nuanced understanding of how demographic factors are associated with different health trajectories, thus providing a foundation for targeted health interventions. Statistical significance was set at P < 0.05.

Results

Baseline Characteristics

A total of 1597 participants were included in the analysis, of whom 1517 had complete baseline variables (Table 1). The percentages of participants who were urban residents, women, chronic disease sufferers, in a relationship, and employed were 53.1, 52.5, 38.9, 80.9, and 89.6%, respectively. Among the participants, 45.6% had an annual family income below 50,000 RMB. Furthermore, 27.8% of the participants were 15–44 years of age, and 17.8% had a higher education background.

Table 1 Baseline Characteristics of Participants in the Complete Data Group by HR-QOL Trends During a Six-Month Follow-Up (N = 1517)

Latent Class Analysis

LLCA was used to investigate the different tendencies of the EQ-5D (Table 2). The latent class analysis model fit statistics indicated that a three-class solution was optimal for the EQ-5D, which was thus selected as the final model. The analysis was performed using both complete cases and those containing missing values, and the classifications of the results for the two cases were similar. All data and complete data groups were defined based on the conditional response probabilities, as illustrated in Figures 1 and 2, respectively. Changes in EQ-5D scores were divided into three classes: the low HR-QoL class, which indicates a downward trend (Class 1); the medium HR-QoL class, indicating an upward trend (Class 2); and the stable HR-QoL class, indicating high scores (Class 3). As the complete data were consistent with the classification results of “all” data, descriptive statistics were conducted using complete baseline data for testing. The descriptive statistics of the demographic characteristics for each class were presented in Table 1. The low HR-QoL class was the smallest of the sample (5.5%), followed by the medium HR-QoL class (12.1%), while the stable HR-QoL class accounted for the majority (82.4%).

Table 2 Fit Indices of Latent Class Analysis

Figure 1 Latent classes of HR-QoL of participants in all data group (N = 1597).

Figure 2 Latent classes of HR-QoL of participants in complete data group (N = 1517).

Predictors of Latent Classes of HR-QoL

To investigate the influence of the demographic characteristics on HR-QoL, multinomial logistic regression analysis was conducted using complete baseline data. As depicted in Table 3, for the medium HR-QoL class, the analysis revealed that district, chronic disease, age, and employment status were significant (P < 0.05). Participants without chronic diseases (OR = 3.323, 95% CI = 1.342–8.227), those aged 45–64 years (OR = 2.196, 95% CI = 1.114–4.328), and employed individuals (OR = 3.422, 95% CI = 1.596–7.340) were more likely to be in the medium HR-QoL class. Conversely, urban participants (OR = 0.215, 95% CI = 0.094–0.492) were more likely to be in the low HR-QoL class. However, the influences of education, annual family income, and marital status were not significant (P > 0.05). For the stable HR-QoL class, district, chronic disease, age, and employment status were significant (P < 0.05). Specifically, participants without chronic disease (OR = 12.832, 95% CI = 5.580–29.512), those aged 15–44 years (OR = 7.849, 95% CI = 2.133–28.886) or 45–64 years (OR = 3.834, 95% CI = 2.093–7.024), and employed individuals (OR = 8.164, 95% CI = 4.034–16.520) were more likely to be in the stable HR-QoL class. Conversely, urban participants (OR = 0.419, 95% CI = 0.202–0.866) were more likely to be in the low HR-QoL class. Similar to the medium HR-QoL class, the impacts of education, annual family income, and marital status were not significant (P > 0.05).

Table 3 Prediction of Latent Class Based on Demographic Characteristics (N = 1517)a

Discussion

This study identified three distinct classes of HR-QoL among Chinese residents. District, chronic disease, age, and employment status predicted these classes. A 2010 study on HR-QoL among the Chinese population found that 6.3% of the participants reported poor or very poor health status.¹³ This finding may reflect the quality and effectiveness of China’s healthcare system.

Influence of District

This study found that the health of urban residents was more likely to deteriorate than that of rural residents. The results contradicted recent findings on older populations in China.^44,45 This discrepancy may be attributed to differences in study groups and levels of economic development, indirectly demonstrating the impact of economic growth on public health. This study used baseline survey data from the entire Chinese population in 2016, providing a comprehensive overview. It found that rural China experienced slower economic development than urban China during this period. Most rural residents have better subjective experiences of their living conditions, such as goals, expectations, and standards, than urban residents do. Consequently, mental pressure and competition among rural residents are lower than among urban residents.^46,47 Moreover, production and lifestyle in rural areas provide more opportunities for rural residents to interact with each other; whereas inconvenience, insecurity, and numerous obstacles to building in urban areas hinder communication between urban residents.⁴⁸ In addition, population density is higher in urban areas of China than in rural areas; consequently, crime, mortality, social isolation, air pollution, and noise pollution are higher in urban areas, and social and environmental problems are more common.⁴⁸ Therefore, Chinese urban residents experience worse the physical, mental, and social health levels than those of rural residents, and their HR-QoL is more likely to deteriorate. Thus, policies must ensure the provision of facilities that can be used for social public services based on individual needs, such as sports facilities, leisure activity centers, lifelong education centers, and volunteer work centers. This would improve the environment for urban participation in such facilities and establish a safe living space with appropriate activities for urban residents. Furthermore, policy makers should also focus on developmental differences between urban and rural areas and make timely policy adjustments to provide appropriate support for key groups and ensure fairness, coordination, and sustainability between urban and rural areas.

Influence of Chronic Diseases

This study found that chronic diseases had a significant impact on the HR-QoL of Chinese residents, finding that was consistent with those of most other studies.^49,50 Patients with chronic diseases are more likely to suffer from pain and reduced physical function, limiting their ability to perform daily activities and exercise, and leading to a range of psychological problems that cause stress, anxiety, and depression.^50,51 In addition, some patients with chronic diseases require long-term care, which imposes physical, psychological, and financial burdens on their primary caregivers and affects the patients’ HR-QoL.^50,52 Therefore, healthcare providers should incorporate these insights to enhance the early detection and continuous management of chronic diseases, which can improve patient outcomes and reduce healthcare costs over time. Additionally, policymakers should consider these factors when designing health services, highlighting the importance of integrated care that encompasses medical treatment, psychological and social support, and health education.

Influence of Employment Status

We found that employed participants had higher levels of HR-QoL, which was in line with a study of the Chinese population¹⁸ but contradicted the findings of another study on Nigerian maternity,⁵³ which found that employed women had lower HR-OoL than unemployed women. This discrepancy may be explained by differences in research objects. Generally, employment provides socioeconomic status and wealth.⁵⁴ Unemployed people with a relatively low socioeconomic status and material wealth may be more vulnerable to impairments in their social functioning and psychological resources than employed people. This vulnerability influences health outcomes through associations with behavioral and psychological risk factors. For instance, people with low socioeconomic status may experience more pressure and engage in poorer health behaviors than those with higher socioeconomic status. These factors can lead to depression, withdrawal, inattention, and aggressive behavior, which can impair quality of life.⁵³ Therefore, policymakers should explore strategies to enhance employment opportunities and support systems, particularly for individuals at risk of job loss. Such strategies may include job training programs, financial incentives for employers, and targeted interventions aimed at improving job retention and career development. These measures would foster a more inclusive and supportive socio-economic environment, ultimately enhancing the health and well-being of the residents.

Influence of Age

The analysis revealed that residents under 65 years old had better HR-QoL, indicating that this age was an important turning point for the HR-QoL of residents in China. This result was generally consistent with population health studies conducted in other countries and population health studies conducted in China.^13,55,56 However, some differences are worth noting. For instance, one study showed that age 55 was an important turning point for HR-QoL among Chinese urban residents in 2015.⁵⁷ Another study conducted in Zhejiang, China, in 2018 showed that individuals over the age of 75 had worse HR-QoL than those younger.⁴⁵ Although differences in the study population may influence the comparison of results, differences in results could also reflect an overall improvement in the health of the Chinese population and development of healthcare services in China. Older adults may gradually lose vestibular function as they age, affecting not only their expected areas of vulnerability, such as vision and walking, but also “healthy” areas, such as speech, dexterity, and emotions.⁵⁵ Vestibular symptoms (especially dizziness and vertigo) are associated with social isolation, reduced autonomy, and difficulties performing daily living activities among older adults.^58,59 The emotional burden caused by these symptoms can lead to adverse reactions such as depressive symptoms, anxiety, and panic, which can significantly reduce their quality of life along multiple dimensions of well-being.⁵⁵ Moreover, older adults face threats from loneliness, chronic diseases, and low health service satisfaction.⁶⁰ Loneliness impairs the immune and cardiovascular systems of older adults, affecting their physical and mental health and contributing to behaviors that lead to poor health outcomes and ultimately impair their quality of life.⁶¹ Older adults also have a high incidence of chronic diseases, which have a negative impact on their daily lives and affect their physical and mental health, resulting in reduced HR-QoL.⁶² Therefore, policymakers should prioritize expanding access to geriatric care and implementing community-based programs that promote healthy aging. These programs could include regular health check-ups, mental health support, and social engagement activities designed to reduce loneliness and isolation. Additionally, healthcare professionals should receive enhanced training to better meet the unique needs of older residents, providing them with more personalized and effective care, which would support their independence and quality of life.

Limitations

This study had several limitations. First, we used the EQ-5D to measure HR-QoL rather than the more generic HR-QoL, as measured by older people’s quality of life questionnaire (OPQOL) or self-enumerated scale of quality of life (CASP-19). The HR-QoL trajectory and its predicted value may differ from the more generic QoL. Therefore, this study may have underestimated changes in HR-QoL. Furthermore, although a representative population area of China was sampled, the findings cannot be generalized, as the sampling points were selected based on the last national pilot, and variations exist in social environments across areas of China. Moreover, as this study used secondary data, we could not control for all related factors. In addition, the non-random selection of the reference group in the multinomial logistic regression analysis could have introduced a selection bias, which could limit our ability to draw causal inferences. Finally, the data lag stemming from the analysis of potentially outdated data may limit the ability of the study to fully capture current conditions. Despite these limitations, this study was significant due to the use of large-scale panel and monitoring data for six consecutive months, which effectively reflected trends in the health of Chinese residents over time, identified latent classes of HR-QoL among Chinese residents and assessed their predictors.

Conclusion

HR-QoL is closely related to health outcomes. This study identified classes with different HR-QoL outcomes among Chinese residents and factors influencing these trends. Three latent classes of HR-QoL scores were confirmed: those with a downward trend, those with an upward trend, and those with stable and high scores. Among them, a small number of residents experienced low and declining HR-QoL, whereas the majority maintained a consistently high HR-QoL. This study found different long-term HR-QoL trajectories for Chinese residents with different lifestyles and health conditions, and that residents without chronic diseases, residents under age 65, and employed residents had better HR-QoL. Conversely, urban residents had poor HR-QoL. However, education level, income, and other demographic characteristics did not significantly affect long-term health. These classes and their predictors revealed the importance of considering multiple factors when seeking to understand and address HR-QoL disparities among Chinese residents. Future research should build on these findings and develop targeted interventions to improve the HR-QoL of specific subgroups of Chinese residents.