Association between companion animal ownership and overall life satisfaction in Seoul, Korea

This study aimed to analyze the association between companion animal ownership, the sub-factors of this ownership (the species and number of owned pets), and overall life satisfaction (OLS). Data was obtained from the publicly available responses to the 2017 Seoul Survey, conducted among Seoul-based Korean locals aged ≥ 15 years (N = 42,687; pet owners = 8,708, non-owners = 33,979). Propensity score was calculated by performing logistic regressions with covariates and data was matched using the nearest-neighbor method. Further, multiple linear regression was performed to analyze this association using the matched data. Additionally, survey-weighted multiple regressions were performed: 1) within pet owners, and 2) after stratifying owners with the number of pets owned. Pet owners in Seoul, South Korea reported higher levels of OLS than non-owners, even after controlling for covariates—age, sex, marital status, family size, family income, job, education, types of housing, housing tenure. Owners with both dogs and cats showed the highest average OLS scores (owners with 2 pets: Mean [M] = 58.05, Standard Deviation [SD] = 0.67; owners with ≥ 3 pets: M = 59.03, SD = 1.02), followed by single pet owners of either a cat (M = 56.64, SD = 0.37) or a dog (M = 56.14, SD = 0.13). Single pet owners reported significantly higher levels of OLS than those with 2 or ≥ 3 pets when pet types were adjusted for. When owners had a single pet, pet types (dog or cat) did not generate a significant difference in OLS scores. Among owners with 2 or ≥ 3 pets, however, owners with both dogs and cats had higher OLS scores than dog owners. This research has significant implications for promoting future study on companion animal effects for improving human health and well-being. Mechanisms of the effect, including cultural factors, should be further investigated.


Introduction
Recent structural changes in the population due to aging, increasing single-person households, and lifestyle changes have subsequently led to a steady increase in pet-owning households and data both from both local Koreans and foreigners, but we used the data obtained from the Korean sample only. This sample included households and their members aged 15 years or older, who lived in Seoul in September 2017.
Since 2003, the Seoul Survey has been conducted to investigate 12 sectors in order to aid policy-making decisions, including: population and household, economy, housing and living sectors (asked every year); personal health and healthcare, safety and disaster, governance, Seoul welfare services, a set of values and social awareness sectors (asked alternate years); and education and childcare, environment, transportation, culture and leisure sectors (asked every second year in place of the second set of sectors). Companion animal-related items were added to the interview questionnaire in 2013 and asked every year. In 2017, this survey examined the sectors of personal health and healthcare, safety and disaster, governance, Seoul welfare services, and a set of values and social awareness.
A total of 20,000 households, with 42,687 people, were surveyed from September 1-30, 2017. The survey was conducted by a trained interviewer through face-to-face interviews. In the absence of an interviewee, the interviews were rescheduled and examined. Participants were recruited using stratified cluster sampling. Detailed information about the 2017 Seoul Survey methodology is available elsewhere [27].

Variables
Companion animal ownership data were obtained from the 2017 Seoul Survey. The respondents were asked whether they owned companion animals (1) or not (0). The responses to "How many companion animals do you presently own: dog/cat/others?" were categorized to indicate the species and number of owned pets. The companion animal species were classified as "dogs (1; owning only dogs or owning both dogs and pets other than dogs and cats)," "cats (2; owning only cats or owning both cats and pets other than dogs and cats)," "dogs and cats (3; owning both dogs and cats or owning dogs, cats, and other pets)," and "others (4; owning pets other than dogs and cats)." Dummy variables were created by using the case of "dogs (1)" as the reference group. Similarly, number of pets owned was classified as follows: "one pet (1)," "two pets (2)," and "three or more pets (3)." Dummy variables were created using the case of "one pet (1)" as the reference group.
The 2017 Seoul survey enquired respondents' degree of satisfaction with their: standard of living, life achievements, personal relationships, sense of safety, feeling of belongingness toward a community, future security, amount of time spent doing what they like, and local environment quality. Each item was scored from 0 (not satisfied at all) to 10 (very satisfied). The 2017 Seoul Survey evaluated overall life satisfaction by employing the domain evaluations module, which is recommended as a measure of subjective well-being by the Organization for Economic Co-operation and Development (OECD), across diverse life aspects based on the policy needs of each institute [28]. The questions incorporated in this module were derived mainly from the Personal Wellbeing Index-Adult (PWI-A), which showed adequate construct and convergent validity [29,30]. Additionally, experts reported sufficient content validity of the life satisfaction measures in the 2017 Seoul Survey [27]. The Cronbach's alpha coefficient for the eight life satisfaction sections was 0.919. The overall life satisfaction was evaluated as the sum of the eight life satisfaction parameters. The overall life satisfaction scores ranged between 2 and 80, where higher scores indicated higher levels of overall life satisfaction.
Age, sex, marital status, family size, family income, job, education, types of housing, and housing tenure type were considered as the potential confounding factors, associated with overall life satisfaction, pet ownership, and sub-factors of ownership (the species and number of owned pet).

Statistical analysis
Descriptive statistics and chi-squared tests were used to analyze the demographic characteristics of the population by companion animal ownership. Additionally, descriptive analysis, independent samples t-test, analysis of variance (ANOVA), and Scheffe's test as a post-hoc analysis, were employed for examining the overall life satisfaction levels among the total population, the pet owners' group, and the non-owners' group. Moreover, descriptive analysis was performed for each combination of the two sub-factor variables-the species and number of pets owned.
Multiple linear regression analyses were performed using propensity score matching to evaluate the associations between companion animal ownership and overall life satisfaction [31,32]. Propensity score is a methodological adjustment, which, here, accounts for the differences in socio-demographic traits between pet owners and non-owners that could influence mental and physical health outcomes [33]. Propensity score-the probability that an individual owns a pet-was computed by performing logistic regression analyses including the following covariates: age, sex, marital status, family size, family income, job, education, types of housing, housing tenure. This study employed the nearest-neighbor method with a ratio of 1:2 (petowners to non-owners) and a caliper width of 0.01 to match the data. Shorter caliper distance gives stricter threshold for matching [32]. Thus, the matched data generated by this method only contained non-owners (control group) and owners (treatment group) who were matched. Further, we performed multiple linear regression analyses using the matched data to examine the associations between companion animal ownership and overall life satisfaction. Significant differences in overall life satisfaction (outcome) due to pet-ownership may indicate a causal impact of pet-ownership on the outcome variable, because of the reduction in unobserved heterogeneity through matched data.
Additionally, survey-weighted multiple linear regression analyses were performed to analyze the associations between the sub-factors of ownership and overall life satisfaction among the pet owners' group. Propensity score matching was not conducted among pet owners because of the small number of whom both owned dog and cat (n = 265, 3.0%) and whom owned three or more pets (n = 191, 2.2%). However, while examining this group, we excluded the "others (4)" category from the species of pets owned to clarify the differences between each group of species, which are "dogs (1)," "cats (2)," and "dogs and cats (3)". Moreover, in order to clarify the effect of the species owned, we stratified each study subject with the number of pets owned and conducted survey-weighted multiple linear regression considering the interaction between the species and the number of pets owned.
All data were statistically analyzed using SPSS version 19 (IBM), and R version 4.0.3. Statistical significance was set at p < 0.05, and multicollinearity was examined using the Variance Inflation Factor (VIF < 4). This study was exempted from ethical approval by the Institutional Review Board (IRB) of Korea University (IRB exemption number: KUIRB-2018-0079-01). Table 1 presents the demographic characteristics of the study participants (N = 42,687). Of these, 8,708 participants (20.4%) owned companion animals. The chi-squared test results showed that age, sex, marital status, family size, family income, education, types of housing, housing tenure type (all p < 0.001), and job (p = 0.013), differed significantly by ownership status.

Results
Among the owners' group, a total of 7,464 participants (85.7%) were dog owners, followed by 924 cat owners (10.6%), 265 owners of both dogs and cats (3.0%), and 55 other pet owners  Table 2 presents the results of descriptive analysis, independent samples t-test, and ANOVA on overall life satisfaction within total population, owners' group, and non-owners' group. Overall life satisfaction scores of the total population ranged from 2-80. Owners In the overall population, males, singles, white-collar workers, and those having higher education, living in an apartment, and owning a house showed higher levels of overall life satisfaction. Additionally, overall life satisfaction generally decreased with age, while it increased with growth in household income. However, single-member households displayed the lowest average overall life satisfaction.
Age, sex, marital status, family size, family income, job, education, type of housing, and housing tenure type were associated with significant differences in the overall life satisfaction levels among the pet owners' group and among non-owners' group (all p <0.05).
Average overall life satisfaction among dog owners decreased with an increase in the number of pets owned (single pet owners: M = 56.14, SD = 0.13). Similarly, among cat owners, single cat owner reported the highest overall life satisfaction levels (M = 56.64, SD = 0.37). However, among owners who raised both dogs and cats, those with three or more pets reported higher overall life satisfaction levels (M = 59.07, SD = 1.02) than those with two pets (M = 58.05, SD = 0.67). In other words, owners of both dogs and cats showed the highest average overall life satisfaction levels, followed by single pet owners of either a cat or a dog.
Propensity score matching reduced the imbalance of covariates (S1 Table). The matched data comprised 8,654 pet-owners and 17,293 non-owners. Table 3 presents the results of the multiple linear regression analyses conducted using matched data to identify the association between pet ownership and overall life satisfaction.
The model indicated no multicollinearity among the independent variables (VIF < 4). Pet ownership status (Coefficient = 0.510, 95% Confidence Interval [CI] = 0.263-0.757) was found to be positively associated with overall life satisfaction, even after accounting for the unobserved heterogeneity between owners and non-owners. Thus, pet owners reported higher overall life satisfaction levels than non-owners.  Table 4 presents the results of survey-weighted multiple linear regression analysis among the owners' group, as stratified by the number of pets owned. The overall life satisfaction level among owners with only one pet was 1.455 points higher (95% CI = -2.288 --0.628) than those with two pets, and 2.279 points higher than those with three or more pets (95% CI = -3.940 --0.618) under controlled conditions. Thus, overall life satisfaction decreased as the number of pets increased when pet types were adjusted for.
Among single pet owners, the difference in overall life satisfaction scores was not significant between dog and cat owners (Coefficient = 0.315, 95% CI = -0.617-1.247). Only those owners with both dogs and cats had higher life satisfaction scores than dog owners among owners of two pets (Coefficient = 0.2449, 95% CI = 0.812-4.087) and three or more pets (Coefficient = 5.923. 95% CI = 3.484-8.362).

Discussion
This study analyzed the associations between companion animal ownership, sub-factors of this ownership, and overall life satisfaction, using the data of local Seoul citizens obtained from the 2017 Seoul Survey. The results revealed that pet ownership had significant effects on generating higher levels of overall life satisfaction. Pet owners with both dogs and cats had the highest average overall life satisfaction scores, followed by single pet owners of either a cat or a dog.
Most of the companion animal owning participants in Seoul, South Korea had dogs (85.7%), while only 10.6% owners had cats. This finding is in contrast to the results of both a United States study [34] and New Zealand study [25], which showed that cats were the more common pet type. One reason that Koreans may have less interest in owing cats as pets could be the historical depiction of cats as wicked in Korea [35]. However, the reasons behind Seoul citizen having this unique pet type proportion still needs further investigation.
Companion animal ownership was associated with higher levels of overall life satisfaction. This finding is consistent with previous findings suggesting that pet owners display better well-being indices, higher self-esteem, and better exercise performance than non-owners [36]. However, this result was inconsistent with a New Zealand study which reported that pet ownership was not associated with life satisfaction [25]. The differences in the study findings could have been because of the different characteristics of the study samples. For instance, while the majority of participants reported that they had pets in New Zealand study, the majority of participants in our study reported not having pets. Also, pet owners had lower levels of education than non-owners in New Zealand-based study, while pet owners were more likely to have a higher level of education than non-owners in our study. The shorter history in Korea of keeping animals as companions, compared to that in western society [35], might also explain the difference. Furthermore, the cross-sectional study design of both studies could have set limits on comparisons between them. It is recommended that future studies that examine the effects of pet ownership in the population take into account the characteristics and cultural differences of the population using a longitudinal study design.
According to the results of this study's survey-weighted multiple linear regression to examine the association between the number of pets owned and overall life satisfaction, single pet ownership was associated with higher levels of overall life satisfaction than ownership of two and three or more pets, when pet species were adjusted for. This finding was in line with existing studies showing that single pet owners experienced higher subjective well-being and average overall quality of life than owners with two or more pets [19,20]; however, previous  findings were not statistically significant. Small sample size in previous studies might have effect on the statistical insignificance in previous studies. Furthermore, the results of our survey-weighted multiple linear regression within single pet owners revealed that overall life satisfaction levels were not significantly different between dog and cat owners. This result was inconsistent with earlier findings indicating that mental health and minor health issue improvements lasted for about 10 months among dog owners, while no improvement continued among cat owners [24]. This inconsistency might result from the different inclusion criteria of each study and different population base. Specifically, although the previous study excluded people who had owned pets before [24], the current study could not exclude them as it did not have data regarding the history of pet ownership. Future studies should clarify the causal relationship between the species of pets owned and overall life satisfaction, in accordance with the history of pet ownership.

Species
However, the results of our survey-weighted multiple linear regression among owners who had two or more pets revealed that owning both dogs and cats was likely to generate higher levels of overall life satisfaction than owning either dogs or cats. This result was inconsistent with a previous New Zealand-based study that suggested that pet types (cats only; dogs only; cats and dogs) did not have association with life satisfaction, psychological distress, and selfesteem [25]. This inconsistency could have arisen because pet ownership is stratified by pet types as well as the number of pets owned in our study, and the previous study categorized pet ownership only with the type of pet. Also, different pet type preferences with previous study, which our studies having more dog owners than cat owners unlike previous study, could cause the inconsistency. Therefore, it is recommended that future studies include both pet types and the number of pets owned and conduct on various population to deepen our understanding of the interplay between pet ownership and human well-being.
Our study also has several limitations. Firstly, the current study design (i.e., cross-sectional) could not identify causal relationships. However, despite this design limitation, we established the causal relationship between pet ownership and overall life satisfaction by employing the propensity score matching method. Future studies can mitigate this limitation by incorporating companion animal-related items to the existing epidemiological panel studies, rather than devising new prospective studies [37]. Secondly, future research should examine the causal inference between the sub-factors of pet ownership and overall life satisfaction, which could not be studied in the current research because the number of participants was too small for: 1) those who were both dog and cat owners (n = 265, 3.0%), and 2) those who owned three or more pets (n = 191, 2.2%). Thus, propensity score matching was not conducted among these pet owners. Thirdly, this study did not consider the duration and history of pet ownership as covariates or as exclusion criteria. Previous studies revealed that the duration of pet ownership affects human health and well-being [19,21]. Additionally, participants with a history of pet ownership were excluded from a previous study [24]; in another previous study it was possible to infer a causal relationship between pet ownership and human health by evaluating the ownership status both at present and five years prior [10]. However, the current survey data did not provide information regarding the duration and history of pet ownership. Lastly, despite the advantages of the large sample for achieving statistical significance, the implementation of stratified sampling and the propensity score matching made the sample size smaller and even permitted us to adjust for confounders properly. Nevertheless, our results provided valuable insights into the association between pet ownership and overall life satisfaction among local Seoul citizens in 2017.
Therefore, despite these limitations, this study is insightful; it advocates a One Health approach called Zooeyia, which suggests that interaction with an animal, especially a companion animal, could positively affect human health [38]. Furthermore, this study provides evidence regarding the relationship between the sub-factors of pet ownership-species and the number of pets owned-and overall life satisfaction. By examining the association between companion animal ownership and overall life satisfaction, this research could assist in promoting the future research of the companion animal effect on improving human health and wellbeing. Mechanisms of the effect, including cultural factors, also should be investigated.
Supporting information S1 Table. Distribution of covariates by pet-ownership before and after matching.