Social Network Characteristics and Smoking Index Among Smoking Cessation Outpatients in Kunming’s Grade-A Tertiary Hospitals in China: A Cross-Sectional Observational Study

Aim We aimed to identify the associations between the social network characteristics of smoking cessation outpatients and their smoking indexes. The association was analyzed with participants in Grade-A Tertiary Hospitals in the capital of Yunnan province in A multicenter cross-sectional survey was conducted in Kunming in six randomly sampled Grade-A tertiary hospitals. Participants included 351 smoking cessation outpatients who provided data on cigarette smoking and social networks. Multivariate logistic regression was used to examine the association between social network characteristics and smoking index across outpatients. Strong associations were identied using adjusted odds ratios and a 95% condence interval. the the of cessation outpatients' social networks on their provides current data on social network and functional We found multiple associations between social networks and the smoking indexes of smoking cessation outpatients at Kunming’s Grade-A Tertiary Hospitals.


Introduction
In China, the smoking rate is high. The Chinese report states that there are more than 300 million smokers throughout the nation [1] , and the World Health Organization reports that cigarette smoking causes approximately 1 million Chinese deaths annually [2] . Smoking damages people's health and is also one of the main risk factors driving the growing prevalence of non-communicable diseases globally [3] .
Smoking cessation clinics is the foremost institution that provides smoking cessation services in China. As of 2019, there were 366 hospitals and primary health care centers in 31 provinces with smoking cessation clinics in China [4] . According to the latest statistics from Yunnan Province, 26 smoking cessation clinics have been established in that region [5] . Smoking cessation clinics have been shown to have a positive impact on tobacco control. Bai Hongmei et al. found that smokers' continuous smoking cessation rates over periods of seven days, one month, and three months were 63.4%, 29.7%, and 22.8%, respectively [6] . Previous literature reported that the success cessation rate of Chinese smoking cessation clinics is 28.0-30.3%, which is lower than the 57% reported abroad.
Scholars have found that conventional smoking cessation services and/or medications cannot wholly solve patients' Fagerstrom Test for Nicotine Dependence score, especially those involving social networks [7,8] . Therefore, it is necessary for physicians treating patients to help them quit smoking to comprehensively evaluate and understand outpatients' smoking behavior in conjunction with social networks. Social networks is composed of a person's social interaction and a personal relationship network [9] . It is not a simple onedimensional plane network but a complex and multidimensional spatial network. Many observational studies have established that social networks are related to healthy behaviors. Social network interventions have been used to reduce weight, improve schizophrenics' health, and treat HIV infection [10,11] . Social networks also play an important role in smoking behavior [12][15] [16] . Pallav Pokhrel et al. found that social network size, perceived social support, and smoking rates were negatively correlated [13] . Bonnie Khanh HaBuı et al. con rmed that for non-smokers who do not discuss health issues with network members, the risk of becoming a heavy smoker was 1.097 times higher; for smokers with more intimate relationships, it was 1.531 times higher [14] .
In short, there has been research on the association between social networks and smoking behavior (smoking amount, smoking rate, smoking status, willingness to quit, and motivation to quit), but, across the board, there is little or no information about associations with the smoking index. The lack of research is relevant because the smoking index is an important measurement used in smoking cessation clinics [17,18] . At present, there are few domestic Chinese studies on social networks and smoking behavior, especially in Yunnan Province. Yunnan Province's tobacco industry was the central economic pillar. Thus, the number of smokers remains high, and smoking cessation clinics are di cult to manage. However, there is a lack of research in this area. Therefore, it is necessary to study the in uence of smoking cessation outpatients' social networks on the smoking index in Grade-A Tertiary Hospitals Kunming, the capital city of the province, to provide a basis for formulating personalized smoking cessation intervention strategies.

Research Design and Procedures
This cross-sectional study was conducted in Kunming in 2020. Baseline data of outpatients from six Grade-A tertiary hospitals. They were randomly selected between October 2020 and December 2020. Exclusion criteria included (1) outpatients who went to the smoking cessation clinic by mistake and had not come to quit smoking; (2) outpatients who were unwilling to complete the questionnaire; (3) those with mental illness who could not answer correctly or complete the questionnaire; and (4) not a rst-time outpatient. Twelve welltrained research assistants conducted face-to-face interviews with smoking cessation outpatients in six smoking cessation clinics. They were Master's students from Kunming Medical University. They rst used questionnaires to collect general information, smoking behavior, social networks, and other information. The questionnaire required approximately 30 minutes to complete.

Ethics
The study was approved by the Medical School Review Board of Kunming Medical University(J-2020-89). The outpatients all signed informed consent, they were given detailed information about completing the form and were informed as to the purpose of the study. All collected data were kept con dential. Con rm that all methods were performed in accordance with relevant guidelines/regulations.

Smoking Index
Smoking behavior was derived from the questionnaire responses and used to assign smoking indexes. The calculated scores the average number of cigarettes smoked per day × years of tobacco smoking [19] . The international medical standard use, the smoking index is divided into four levels: 0 = non-smoking, 1 = index score of score ≤100, 2 = a score of 101-200, and level 3 applies to a score > 200. A high smoking index score refers to a smoking index level of 3 (> 200).

Social Network
The questionnaire collected data from the perspective of the personal-centered social network using the four primary aspects of social network structure: social network relationship types, social support methods, social in uence, and closeness. First, the outpatients are guided to identify all the people they know well and everyone who knows them well. Then, they identify the people with whom they had contact in the past month through face-to-face communication, phone calls, or any other channel. Finally, from this list, each outpatient identi ed the ve most connected people in their social networks. The detailed characteristics of each network person, including their relationship with the outpatient, whether they smoke, and other demographic characteristics; the type of social support and the degree of in uence or closest to the patient, such as whether they support smoking, share cigarettes, and discuss health problems together.
The portion concerning the social network includes the following open-ended questions: (1) Mental support: "If you want to talk about something very personal and private, would you talk to him/her? (Yes/No);" (2) Financial support: "If you want to borrow 1,000 yuan, will he/she lend it to you? (Yes/No);"(3) Health advice: "If you need advice on health issues, will he/she provide advice? (Yes/No);" (4)"What is your relationship with the person?(Spouse, children, other relatives, friends who drink alcohol, friends who smoke, friends (nonsmoking, non-drinking), colleagues, classmates, medical staff, leaders);"(5) "(What is the ) education level of this person?(Illiterate,elementary school, junior high school, high school, university and above);"(6) "How close are you to this person?(Not close at all, a little close, close, very close);"(7) "How much in uence does this person have on you?"scored from 1 to 5 points (1 = not affect me at all; 5 = can change my decision); (8) "Do you discuss smoking with this person?(No, sometimes, always discuss);" (9) "Does this person share cigarettes with you? (No, yes)"; (10)Does this person support you smoking? (No support, support);"and (11)"Does this person think that smoking is harmful to your health? (De nitely not, maybe not, maybe yes, must be)."

Data Analysis
All data analyses were performed using R software (version4.0.3). Descriptive analysis was applied to describe the demographic data, smoking index, and social network characteristics of smoking cessation outpatients, and the network size was de ned as the number of members in a person's social network. We recognized that the characteristics of smoking cessation outpatients determined the degree of connection between them and their social networks. The median of discrete variables was used to describe the structural variables in the social network characteristics according to the characteristics of the index; the homogeneity of gender and education level (the number of people with the same characteristics as the surveyed patients/total number of people) were estimated. The score range was 0-1where"1"means that 50% or more of the social network members had the same education level and gender(i.e., the gender and education level are similar among st all of the social network members);"0"means that more than 50% of social network members and the surveyed person had different educational levels and genders (i.e., the social network did not have the same gender and educational level as the respondent). Regarding the closeness and in uence of a social network, if any network member could provide a higher level of intimacy or in uence (intimacy level 3, in uence level 4 or 5),the network was de ned as having a "high level of closeness or in uence." Regarding social network function variables, if any network member provided emotional support, nancial support, health advice, and evaluation support sources, the network was marked as having that speci c function. A logistic regression model was used to test the in uence of social network characteristics on the smoking index (≤ 200 or >200). Adjusted demographic data in the multivariate model included gender, ethnicity, marital status, occupation, and economic income. The analysis excluded records with more than 1% missing values for any variable. Those variables found to have p-values less than 0.05 were considered important predictors in the nal model. Table 1 presents the demographic distribution of smoking cessation outpatients. In this study, 360 questionnaires were sent out, and 351 valid questionnaires were returned. The recovery rate was 97%. The average smoking index of the sample was 216.77 ±240.94. The majority were married (51.85%), Han (70.94%), and male (97.45%). Most outpatients were not religious (66.39%) and had a high level of education (67.24%). The percentage of participants engaged in non-service industries was the largest (81.20%). The four income levels were nearly equally represented among the participants with the smallest (lowest income) group making up 19.66% of the sample and the largest (highest income) group accounting for 29.34%. Table 1 Demographic distribution data of smoking cessation outpatients by hospital; n(% Social Network Characteristics Regarding the degree of closeness and in uence between outpatients and network members, 98.58% of outpatients had at least one close or very close member, 86.04% of outpatients had at least one very in uential member, 94.02% of outpatients had members who discussed smoking with them, and 89.17% of outpatients had members who shared cigarettes and supported their smoking. Regarding social support, the majority believed that at least one of their members would provide nancial support (93.45%), emotional support (97.72%), health advice (87.18%), and evaluation support (98.29%).

Relationship Between Social Network and Smoking Index
As shown in Table 3, social network characteristics signi cantly associated with smoking indexes were large network size, having children, having drinking friends, and having non-smoking and non-drinking friends in-network. Other factors that increased the association were having at least one very in uential person in one's social network and getting health advice from network members. In addition, large network size and having children in the network were risk factors for a high smoking index. However, after adjusting for gender, ethnicity, marital status, occupation, and personal income, the association between the large network size and a high smoking index decreased, but the correlation between having children and a high smoking index increased. Beyond that, having at least one very in uential person in the network changed the association between the social network and a high smoking index signi cantly.
Having in-network friends who drink alcohol, friends who do not smoke, friends who do not drink alcohol, and network members who provide health advice were all protective factors against having a high smoking index. Similarly, after adjusting for the above factors, the associations were stronger between higher smoking index scores and having in-network friends who drink alcohol; the index and having friends who neither smoke nor drink were also strongly correlated. However, the association was weaker between smoking index scores and getting health advice from network members (Table 3). Note: a Adjusted for demographic characteristics: gender, ethnicity, marriage, occupation, and personal income. Among: *** p < 0.001; ** p < 0.01; * p < 0.05.

Discussion
This study was the rst to evaluate the impact of smoking cessation outpatient social networks on the smoking index in Kunming, the capital and largest city of Yunnan province in China. This article reports the in uence of smoking cessation outpatients' social networks on their smoking indexes and provides the most current data on social network structure and functional characteristics. We found multiple associations between social networks and the smoking indexes of smoking cessation outpatients at Kunming's Grade-A Tertiary Hospitals.

Comparisons with Prior Work
We found that the risk of a high smoking index increased 1.79 times for every increase in the size of the close social network for the smoking cessation outpatients in this study. This nding was consistent with those of Ennett et al., who also reported that social network size was signi cantly positively correlated with the number of smokers who had smoked in the prior three months [20] . First, we speculated that a more extensive social network would be accompanied by a broader range of smoking information and/or more smoking-related resources, which some prior studies have supported [21] . In our study, 351 smoking cessation outpatients reported an average network size of 4.85, of which the proportion of smokers was 89.17%. Second, a larger network potentially provides more social network relationship types; thus, larger networks had a relatively greater impact on healthy behaviors. For example, social network relationships can hinder healthy lifestyles [22] . According to a previous study, female smokers who were closely related to other smokers had a 51.52 times higher risk of developing heavy smoking habits, and their odds of quitting smoking were signi cantly reduced (OR 0.74; 95% CI = 0.63, 0.87 ) [23] .
We also found that outpatients with children in the network were at 6.346 times greater risk of having a high smoking index than those without children, which seems to correspond to ndings of previous studies that have shown that raising children is one of the main sources of parental pressure, especially since the implementation of China's second-child policy [24,25] . For parents, smoking provides a way to manage bad moods and relieve stress. Eriksen testi ed that parents with more than one child in their social network smoked 10 more cigarettes (or more)per day than parents without children [26] . A comparative study of 69 parents of children with developmental disabilities and 137 parents of healthy children showed that the pressure level of the two parents increased, and having children was positively correlated with them being smokers [27] .From the above research, we reasonably speculated that, within the network, smoking cessation outpatients who have children would have higher nancial and other child-related pressures. Therefore, they would be likely to relieve the pressures of parenting by smoking or increasing the number of cigarettes smoked, leading to a higher smoking index, suggesting that physicians treating patients for smoking cessation should be aware of outpatients with children.
We found that if there was at least one very in uential person in someone's social network, the risk of a high smoking index was 2.738 greater than if there was no very in uential person. Our research results supported those of Mercken, Dong-Chul, Khalil, Georges, and others [28][29][30] that highlight the impact of social in uence as an integral element of networks. In other words, the premise that social in uence encourages network members to adopt or adapt to certain behaviors was supported by the research showing that people changed their behaviors and shared activities after spending time together. Many studies have reported that social in uence is related to both positive and negative health behaviors, such as smoking, drinking, drug use, and diet/weight-related behaviors [31][32][33][34] . Our study highlighted the importance of having at least one very in uential person in the network to prevent a high smoking index in smoking cessation clinics.
Interestingly, we also found that having drinking friends, non-smoking and non-drinking friends in the network were all associated with a lower probability of a high smoking index. By contrast,other studies have reported that having drinking friends in the network was a risk factor for smoking behavior [35,36] . In a study of 630 students with an average age of 16.2 years, Ritchey et al. found that having drinking friends increased the possibility of these students became smokers (β= 0.23), and the peer pressure to drink alcohol was the main reason for the students becoming smokers (β= 0.14) [37] . We observed that the current research object in the relationship between alcohol-drinking friends and smoking behavior in the network is primarily adolescents. Previous studies have also suggested that personal values play a key role in adolescents' smoking behavior.
Instead of adults having relatively stable values, they obtained more information about unhealthy behaviors from different sources and were relatively less affected by drinking friends. Regarding whether alcoholdrinking friends provided more health advice that protected outpatients against smoking (reducing the smoking index of smoking cessation outpatients), we did not further stratify research on drinking friends. It was recommended that the population a strati ed study conducted with this population would be better for illustrating this problem and should include non-smoking and non-drinking friends in the network sharing to examine good social relationships.
Some studies have shown that good social relationships have a positive impact on the health-related behaviors of respondents [38] . Healthy behaviors among network members help people have better and more active health behaviors, thereby reducing both the risk of smoking and the number of cigarettes [39] . In our study, having alcohol-drinking friends, non-smoking friends, and non-drinking friends were all related to a lower probability of a high smoking index, indicating that the smoking behaviors of in-network friends can profoundly impact smoking behavior. This result is consistent with the research conclusions of Rostila et al. They deemed that friends' behavior played an important role in the network; in particular, friends with healthy behaviors have a positive impact on the network [40] .
Moreover, we found that the ability to get health advice from network members was also associated with a lower probability of a high smoking index, corresponding with the ndings of Khanh HaBuı and colleagues who reported that being able to discuss health issues with network members helps reduce smoking (OR = 0.555, p <.05) [41] . Health advice is considered one of the most cost-effective interventions for smoking cessation interventions [42] . Ralph et al. found that 25% of outpatients accepted their doctor's advice, which reduced frequent smokers (36%) and frequent cannabis users (27%) [43] . This nding may help improve the effectiveness of health advice in smoking cessation clinics. In the future, smoking cessation clinics can add health education courses focusing on health advice to bene t smoking cessation clinics.

Limitations and prospects
There are several limitations to this study. First, this was a cross-sectional survey, so the results can only be used to consider the possibility of causality, and we cannot make any causal inferences. Second, their social networks are undoubtedly larger than the ve people they were limited to for nomination in the questionnaire. Third, the measure is also limited to evaluating cigarette use among those in smoking cessation clinics; the use of other tobacco products (such as e-cigarettes and water pipes) is not examined. Future research will bene t from further strati cation studies with a longitudinal study that can follow up on the in uence of social networks on smoking cessation clinic patients over time. Future studies should also investigate the impact of social networks on the use of other tobacco products (such as e-cigarettes and water pipes) in smoking cessation clinics.

Conclusion
The results of this study also demonstrate the association between social network characteristics and the smoking indexes of smoking cessation outpatients at Kunming's Grade-A Tertiary Hospitals. Physicians helping patients quit smoking should be aware of the complexity of the in uence of social networks. The ndings of this study can serve as a rationale for developing individualized smoking cessation programs and interventions that focus on strengthening social networks. For example, including emotional module courses and interpersonal relationship processing skills in health education for smoking cessation outpatients could help them manage the pressures of raising children; it may also help them reduce negative impacts and increase the positive effects of in uential people in their networks.