Effects of Unhealthy Lifestyle Factors on Healthcare-seeking Behavior in China

Background: People with unhealthy lifestyles may experience omitted or delayed healthcare, leading to severe sickness and higher healthcare expenditures in the future. Hence, the current study aims to ascertain the effects of current smoking, regular drinking, and physical inactivity on healthcare-seeking behavior among the sick in China. Methods: The data used in this study was obtained from the China Family Panel Studies (CFPS). The nal sample consisted of 44,362 individuals in all ve waves of data collection. Logistic regression models were used for the analysis. Results: Based on the tests, the xed effects estimation was the preferred method for panel data in this study. The current study found that sick adults who currently smoked cigarettes were 0.73 times less likely to seek healthcare than those who never smoked or stopped smoking. Compared to non-drinkers, sick adults who regularly drank alcohol decreased the likelihood of seeking healthcare. Sick adults who never participated in physical exercise decreased the odds of seeking healthcare by 24% compared to those who participated in physical exercise. Conclusions: Current smoking, regular drinking, and physical inactivity decreased the probability of seeking healthcare among sick adults. Therefore, screening and brief advice programs should be delivered by primary level care and pay more attention to those who have unhealthy lifestyles, reducing the burden of diseases.


Background
In 2015, a Chinese national nutrition and chronic disease report presented that the prevalence of current smoking, harmful drinking, and physical inactivity among adults were 28.1%, 9.3%, and 71.3% [1], and these preventable risk factors have contributed to the increased rise in chronic diseases. Chronic diseases now account for an approximated 80% of total deaths and 70% of total disability-adjusted life years lost in China [2].
Unhealthy lifestyles have proven to be independently or synergistically a cause of diseases, such as hypertension, dyslipidemia, diabetes, and obesity [3]. Most people make unhealthy lifestyle changes due to the health concerns and illnesses they experienced. For example, among Chinese smokers who make a quit attempt in the past 12 months, 38.7% of smokers worried about their future health status, and 26.6% smokers who experienced severe sickness in 2018 [4]. Therefore, ex-smokers are associated with higher healthcare utilization and increased healthcare expenditures [5][6][7][8]. People with unhealthy lifestyles may not care about their health status or maybe risk-tolerant individuals. They may experience omitted or delayed healthcare, leading to severe sickness and higher healthcare expenditures in the future.
To our knowledge, little is known about the association between unhealthy lifestyle factors and healthcare-seeking behavior. Hence, the current study aims to ascertain the effects of current smoking, regular drinking, and physical inactivity on healthcare-seeking behavior among the sick in China using the ve-waves of longitudinal panel dataset. This knowledge will help better understand the underlying causes of healthcare-seeking behavior by current smoking, regular drinking, and physical inactivity, thus helping the government make healthcare resource allocation decisions and improve health education programs for target populations.

Data Source and Study Sample
The data used in this study was obtained from the China Family Panel Studies (CFPS), conducted by the

Measures Dependent Variable
Healthcare-seeking behavior was set as a binary variable, denoting the decision to consult a doctor (or not) of the adults reporting themselves to have health problems in the past two weeks. The two questions in CFPS that represent this variable was: 'During the past two weeks, have you ever had health problems?' and 'Have you seen a doctor?'.

Independent Variables
The respondents completed a face to face interview about unhealthy lifestyle factors, including current smoking, regular drinking, and physical inactivity. In the CFPS, each respondent was asked: 'Have you smoked in the past month?'. The adults reporting 'Yes' were considered current smoking. Second, regular drinking was de ned as a binary variable. The CFPS question supporting this variable was: 'Have you often drunk alcohol (at least three times a week) in the past month?'. The respondents reporting 'Yes' were coded as 1 and 'No' coded as 0. In the CFPS, respondents were asked how often did you participate in physical exercise in the past week. The adults were categorized as physical inactivity if they answered "Never".
The other independent variables were selected based on the previous studies, including the respondent's age, gender, marital status, urban residency, household income, medical insurance, educational attainment, employment status, self-reported health status, and chronic conditions. De nitions of all relevant variables are provided in Table 1. Bivariate analyses were used to examine differences between adults who consulted a doctor (seeking healthcare) and those who did not consult in each wave. Pearson's chi-square test was used to analyze the categorical independent variables.
Logistic regression was performed to analyze the association of the unhealthy lifestyle factors on healthcare-seeking behavior in China. The rst step in the analysis could be to pool all available information from 2010 to 2018 for all adults and treat them as independent information for N = 44,362 adults. Hence, pooled logistic regression could be employed here. This study treated the data as a panel structure and made a choice between the xed effects and random effects logistic model. In this study, a possible unobserved variable is health literacy, which is correlated with the time-varying explanatory variables in the model (e.g., regular drinking, smoking, or physical inactivity). With such correlated heterogeneity, the xed effects logistic model should be preferred over the random effects logistic. However, when estimating the xed effects logistic model, many pieces of information are lost. The random effects logistic model was also presented in this study [10,11].

Results
A descriptive summary of all variables over time is shown in Table 2 Table 3 compares sick adults who sought healthcare and those who did not seek across a variety of unhealthy lifestyle factors. Adults who currently smoke had a lower proportion of seeking healthcare from 2010 to 2018, and a similar trend was observed for adults who regularly drink alcohol. Physically inactive adults had a lower ratio of seeking healthcare between 2014 and 2018. There were statistical differences between sick adults who sought healthcare and those who did not seek in current smoking, regular drinking, and physical activity in partial or all the ve waves. These results further demonstrated that the xed effects logistic model should be preferred over random effects logistic. The results of the logistic regression analysis are showed in Table 3 as odds ratios. An odds ratio (OR) greater than one indicates a positive effect on the likelihood of seeking healthcare; an odds ratio less than one indicates a negative effect. than those who never smoked or stopped smoking (OR = 0.73, p < 0.01). Compared to non-drinkers, sick adults who regularly drank alcohol decreased the likelihood of seeking healthcare (OR = 0.77, p < 0.01).
Sick adults who never participated in physical exercise decreased the odds of seeking healthcare by 24% compared to those who participated in physical exercise (OR = 0.76, p < 0.01).
Irrespective of the estimation method, sick adults who currently smoked cigarettes, regularly drank alcohol, and never participated in physical exercise were less likely to seek healthcare (see Column (i)-(iii) of Table 4).

Discussion
In the current study, we examined the association between unhealthy lifestyle factors and healthcareseeking behavior among a Chinese adult general population using a ve-wave of longitudinal dataset.
This study found that the proportions of seeking healthcare increased from 68.76% in 2010 to 76.05% in 2018. This result is not unexpected due to easy access and availability of healthcare facilities and improved insurance coverage since the launch of the new health reform of 2009. However, people with unhealthy lifestyle (current smokers, regular drinkers, and physical inactivity) had a lower proportion of seeking healthcare compared to those with health healthy lifestyle.
The xed effects logistic regression model was used to identify the unhealthy lifestyle factors affecting healthcare-seeking behavior. The results indicate that among the sick, adults who currently smoked cigarettes, regularly drank alcohol, and never participated in physical exercise were less likely to seek healthcare. Similar results have been discovered in China, England, and Australia. For example, Zhou et al. [12] presented that people who are physically inactive decrease the probability of seeking healthcare in China. Smith et al. [13] found that smokers are less likely to seek help than non-smokers in England. Feng et al. [14] uncovered that people with unhealthy lifestyles are less likely to see general practitioners in Australia.
Three possible reasons may explain the inverse relationship between unhealthy lifestyles and healthcareseeking behavior: rst, people with unhealthy lifestyles are more risk-tolerant, and they may more willingly bear disease risk [15,16]. Higher willingness to bear risk decreased the probability of seeking healthcare [17]. Second, unhealthy lifestyles, like smoking, drinking, and physical inactivity are linked with poor health conditions. People in poor health conditions are more likely to have negative experiences in the healthcare system and be less satis ed with it [18]. Lower patient satisfaction tends to have decreased the probability of seeking healthcare [19]. Last, unhealthy lifestyles contribute to productively losses at work and reduced ability to work [20,21]. People with unhealthy lifestyles may sacri ce some of their leisure time to perform un nished work. Time constraints may delay some people from seeking healthcare [22]. Reduced ability to work may affect people's ability to afford healthcare.
Although the present study used a national survey to analyze the unhealthy lifestyle factors affecting healthcare-seeking behavior among the sick, several limitations should be emphasized. First, the CFPS survey does not collect information on self-medication practice, so this study de ned healthcare-seeking behavior as professional help sought. Previous studies have shown that individuals who perceive themselves to have mild health problems are more likely to select self-medication in China [23,24].
Second, data were obtained via a nationally representative survey, and thus shares the limitations of all self-reported data: recall bias and unreliability of responses under pressure. Last, we could not exclude exsmokers who quit smoking in the past 30 days from the group of current smokers and ex-drinkers from non-regular drinkers.

Conclusions
Current smoking, regular drinking, and physical inactivity decreased the probability of seeking healthcare among sick adults. People with unhealthy lifestyles experience omitted or delayed healthcare, leading to serious health problems and higher healthcare costs on society in the near future. Therefore, screening and brief advice programs should be delivered by primary level care and pay more attention to those who have unhealthy lifestyles, reducing the burden of diseases.

Declarations
Ethics approval and consent to participate For this research, we have used publicly available secondary data set with all individual identi ers removed prior to making the data set available publicly. No ethical approval was required due to the type and nature of data set used.

Consent for publication
No applicable.

Availability of data and materials
The datasets generated and/or analysed during the current study are available in the Peking University Open Research Data Platform repository, https://opendata.pku.edu.cn/dataset.xhtml? persistentId=doi:10.18170/DVN/45LCSO. JS designed the study and made important contributions to the revision of the manuscript. CL led the data analysis and wrote the manuscript. All authors read and approved the nal manuscript.