Health status is a key indicator that reflects the extent of health loss from disease, injury, or other factors affecting physical and mental health [1,2]. In the Global Burden of Diseases study, disability weight was used to quantify the severity of health states from disease sequela and was a pivotal parameter for disease burden calculation [3]. However, it is limited by the complexity of measurement methodology and by the inconsistency between clinical classification description and disease sequela description. For this purpose, series multi-attribute utility instruments based on health-related quality of life (HRQOL) are used to classify health status [1,4]. HRQOL, a multidimensional concept of broad physical and mental health, was used to evaluate overall health and track population health status, health needs, and disparities [5,6]. HRQOL instruments do not assess health status based on diseases or sequelae, but rather on the general population, which is rapidly gaining acceptance as a measurable outcome [7,8].

Based on a synthesis of the scientific literature and advice from its public health partners, the US Centers for Disease Control and Prevention (CDC) defined HRQOL as “an individual’s or group’s perceived physical and mental health over time” and developed a set of Healthy Days measures to track and assess population health status and HRQOL in states and communities [9]. This tool includes four core questions (core Health Days measures [HRQOL-4]) and an additional 10-item set of health perception and activity limitation questions. The CDC HRQOL-4 aims to assess a person’s self-rated general health, poor physical health, poor mental health, and activity limitations. These questions have been well validated in a variety of populations and used fairly extensively in some national surveys in the United States (eg, the Behavioral Risk Factor Surveillance System [BRFSS] and National Health and Nutrition Examination Survey) [9-12]. In addition, the HRQOL-4 has been included in the China Chronic Disease and Risk Factor Surveillance (CCDRFS) since 1996 [13,14]. Given that self-care and usual activities are defined as two different dimensions in some commonly used HRQOL measuring tools (eg, EQ-5D and 36-Item Short Form Survey [SF-36]), the fourth question on activity limitations in the HRQOL-4 was split into two items corresponding to usual activity and self-care limitations, which is called the Chinese version of the core Healthy Days measures (HRQOL-5).

The inclusion of the HRQOL-5 in the CCDRFS has become one of the greatest strengths of the instrument compared to others, as this survey is conducted once every 3 years and provides health-related data from a national sample [15]. No other HRQOL instrument, like the SF-36 or EQ-5D, is accessible in an annual sample as large as that in the CCDRFS. In addition, the government needs to capture poor health statuses with a concise measure due to faster aging and more severe subreplacement fertility in China [16]. The HRQOL-4 differs from other instruments like EQ-5D and SF-36 by not using preference-based measures and not establishing a standardized evaluation system [17]. Previous studies found good internal consistency in the four measures, suggesting that the HRQOL-4 may be suitable for combining a summary index [18]. The US CDC recommended a summary index of “unhealthy days,” using the sum of days of poor physical health (Q2) and days of poor mental health (Q3). However, this index has not been fully validated as a summary of overall health [19]. Some researchers have proposed quantifying the presumed latent HRQOL-4 construct by factor analysis. Horner-Johnson et al [20] and Mielenz et al [21] found that the CDC HRQOL instrument (HRQOL-4 plus the five optional HRQOL module questions) could be reduced to two latent factors that correspond conceptually to the physical and mental health construct. Given that the optional HRQOL module questions were only available for a few national surveys, Yin and colleagues [22] conducted a factor analysis to create a summary score (factor score) by using the CDC HRQOL-4. This summary score showed good validity, stability, and measurement invariance over time in BRFSS data sets, but the feasibility of developing a summary measure based on the HRQOL-5 has not been validated among the Chinese population.

Local levels of population HRQOL may vary from population to population due to differences in socioeconomic status, religion, and lifestyle. Several factors, including age [9,23], sex [9,23], BMI [24], race/ethnicity [9], socioeconomic status [23], tobacco use [25], substance abuse [25], physical activity [26], injury and violence [9], and chronic disease [9,23,25] have previously been studied as correlates of healthy days. However, these studies usually focused only on one or two dimensions of the HRQOL-4 or used univariate analysis that did not take into account the effects of potential confounders. Few studies have evaluated risk factors that may influence population HRQOL based on the summary score of the HRQOL-4.

Therefore, there are two main aims of this study: (1) to propose a summary value based on the HRQOL-5, which could be used to assess overall health status and estimate population-level disability, calculating health-adjusted life expectancy in the future, and (2) to identify factors associated with the summary score of the HRQOL-5 in a Chinese population.

To our knowledge, this is the first study to test the underlying latent construct of the core Healthy Days measures and factors associated with HRQOL in major cities of mainland China. The primary aim of the study extends the results by first using factor analysis to obtain a modified one-factor structure and a summary score based solely on the HRQOL-5. The second aim was to identify possible predictors of the summary score of the HRQOL-5 in a Chinese population. More specifically, age, education, average income, drinking, physical activity, and self-reported disease or injury were significantly associated with the summary score of the HRQOL-5.

The HRQOL-5 distinguished between usual activities and self-care activities in activity limitations and added only one question compared to the HRQOL-4 [14]. The use of the HRQOL-5 might be one of the most cost-effective ways of tracking general health needs in China. The results of the reliability analysis also indicated that the HRQOL-5 had acceptable internal consistency (Cronbach α >0.70). One subtle difference is that in our sample, the Cronbach α based on standardized items increased if the first question on general health status was removed as in previous research [22]. However, because of a small increase (+0.02) in our study and the lack of a clear cutoff value for the increase of the α value, we cannot suggest the internal consistency between the general health status item and the four other questions.

Credible construct validity of the HRQOL-5 was established by conducting both exploratory and CFA for our sample, as evidenced by a series of model goodness-of-fit indicators (RMSEA 0.02, TLI 0.93, GFI 0.99, SRMR 0.01). Previous studies, which used data obtained from the HRQOL-4 questions on the BRFSS, developed and tested a one-factor model [22]. Our results showed similar constructs of HRQOL and item factor loadings. As expected from the CDC HRQOL-4, one distinctive dimension that encompasses both physical and mental health was extracted, thus making it possible to compare the CDC HRQOL-4 and the HRQOL-5. It was noteworthy that a positive error correlation path between the days of usual activity limitation item and the days of self-care limitation item was found. There are at least two theoretical reasons not based on statistical support. First, the format of the fourth question on usual activity limitation and the fifth question on self-care limitation is very similar. Some research has found that using a similar question format may contribute to the covariance between two items, which can affect survey responses [22]. Second, Clifford [36] suggested the possibility of correlated error whenever survey questions are connected physically. Our model may account for this relation by indicating a positive correlation path between the error terms in the measures of usual activity limitation and self-care limitation. Overall, the HRQOL-5 was found to be largely similar to the CDC HRQOL-4, which showed that the underlying structure of HRQOL is consistent across different cultures and languages, thereby lending support to a claim that summary scores of the HRQOL-5 would be simpler and more comprehensive than using the single Healthy Days question [20,22].

Understanding the associations between different HRQOL instruments is becoming a higher priority for government agencies so that they can interpret the potential benefits of different health policies from studies that use either of these instruments. A systematic review shows that the EQ-5D and SF-36 were the most commonly used measure instruments for assessing quality of life [1]. Jia and colleagues [37] developed a mapping algorithm linking the HRQOL-4 and the EQ-5D, and this mapping score might provide a good measurement of individuals’ overall health status [38]. A national study from the United States explored associations between the HRQOL-4 and the SF-36 [39]. In addition, these studies also show differences in health status between the HRQOL-4 and other instruments due to their respective descriptive systems and assessment methods. Thus, there is debate over which instrument can better assess the overall HRQOL in a population. From the cost and efficiency perspective, Healthy Days seems to be more advantageous, since the HRQOL-4 or HRQOL-5 has been included in the national surveillance system in China or the United States. Thus, the national factor scores can be calculated without extra surveys. Additionally, we evaluated the overall health status by considering the potential construct in 5 HRQOL-5 items. It does not need to map to other generic instruments. The summary score based on the HRQOL-5 can be directly applied to various scenarios in China, such as health status evaluation, risk factor identification, and burden of disease assessment in future studies.

The analysis of associations between the summary scores of the HRQOL-5 with the other collected variables potentially affecting HRQOL among the Chinese population has allowed us to confirm data already available in the literature. Consistent with previous studies, age was a significant predictor summary score of the HRQOL-5 [21,40,41]. The health problems became increasingly serious with age, resulting in lower summary scores. Socioeconomic status appeared to be associated with HRQOL as indicated in this study and others [40,41]. Participants who were wealthier and had a higher level of education had higher HRQOL-5 summary scores. In addition, our results showed a positive association between physical activity and the HRQOL level. A national survey in the United States also found a dose-response relationship between unhealthy days and physical activity level [26]. Participants who reported diseases had worse HRQOL than those without. The summary scores of the HRQOL-5 were substantially decreased by self-reported diseases. Substantial evidence showed that self-reported diseases had been considered the main risk factor impairing HRQOL [42,43]. Another interesting finding is that drinking can increase the summary scores of the HRQOL-5. One possible explanation for this phenomenon is that Chinese drinking culture was ingrained in the Chinese culture [44]. Some Chinese people think Chinese alcohol could increase pleasure and arousal or decrease pain caused by injury or diseases, which may improve HRQOL and increase summary scores of the HRQOL-5. Of course, due to a range of residual confounding and the design of cross-sectional studies, the observed associations cannot be considered causal associations. Contrary to previous studies [45,46], our results show no statistically significant association between smoking and the summary score of the HRQOL-5. One possible explanation is the limited impact of mentally unhealthy days on the summary score due to its lower factor loading. Some studies found smoking was a predictor only for mental health indicators in healthy days measures [45], and smoking prevalence was higher in participants with poor mental health than those with better mental health [47,48]. Thus, it is essential to use a comprehensive index to evaluate the HRQOL of the population, which can avoid confusion caused by different determinants across different dimensions.

The Healthy China 2030 plan has made population health the ultimate goal of economic development and political reform including justice and equity as one of four core principles [49]. Health equity at the subnational level can be achieved if equipped with a monitoring system for population health status and sufficient data to guide investments. However, designing surveillance instruments must balance collecting needed data with survey costs by including only questions needed to answer research questions [50]. The Chinese version of the HRQOL-5 could substantially decrease participant burden and survey costs by only asking 5 questions. Our study developed a summary index based on the HRQOL-5 to estimate the health status of the population. A multi-data source and nationally representative sample were the greatest strengths of this instrument compared to others, like the SF-36 or EQ-5D. In considering potential intercultural differences, we found the HRQOL-5 could rapidly assess the severity of health states relevant to the Chinese population, and one summary score could capture overall HRQOL via factor analysis, which is important for further development of disease burden research. In addition, the government needs to track changes in health status in different subgroups and identify their determinants due to faster aging and the changing disease spectrum in China. Though Healthy Days measures have been used fairly extensively in various national surveys, previous studies only focused on determinants of a subset of items from the HRQOL-4. If one study found an improvement in one HRQOL-4 item and a decline in another or these HRQOL-4 items were influenced by different determinants, it could be difficult to draw an overall conclusion about the impact on health status. Our results exploratively evaluated risk factors that may influence the summary score of the HRQOL-5 in a Chinese population, which provided a more cohesive picture of HRQOL at the population level. It was particularly useful for setting priorities for health planning to improve primary care in China.

This study had three limitations. First, this study used data from household surveys, which is subject to selection bias because of the exclusion of hospital patients. Second, due to the design of cross-sectional studies, we cannot rule out the potential influence of unmeasured confounders; more studies are needed to draw causal inferences. Finally, since all the participants are from the same city in China, the limited national representation could affect the generalizability of our results. More studies from other Chinese regions are needed to confirm our findings.

In conclusion, this study suggests that it is feasible to summarize 5 HRQOL-5 items in a summary score via factor analysis. The results of the summary score could be used to estimate population-level disability and calculate health-adjusted life expectancy in the future. Our study also showed that age, educational attainments, income, regular exercise, alcohol use, self-reported disease, and self-reported injury were significantly associated with the summary score of the HRQOL-5.