Background

Self-rated health is a widely used indicator of overall health and well-being [1,2,3]. It is routinely employed by health and medical professionals, psychologists, and social scientists seeking to identify disease risk factors in large and diverse populations [2, 4,5,6]. Self-rated health has been shown to predict morbidity and mortality [7,8,9], mental health [10, 11], medical conditions and functional limitations [3, 12], and healthcare utilization [13, 14]. It has also been linked with many possible antecedents including socioeconomic status (SES), social capital and support, diet and exercise, and a sense of mastery, as well as age, gender, and race/ethnicity [4, 5, 15, 16].

The current study builds on this work by addressing three limitations in the literature. First, there is evidence that self-rated health varies around the world [12, 16,17,18,19,20], but few studies have utilized representative samples in developing and non-Western nations, and comparisons of self-rated health and its predictors across countries and cultures remain rare [12, 18, 21, 22]. Second, very little research has examined how early-life experiences shape adult self-rated health in diverse countries [23], and the findings reported here are among the first for several nations. And third, most research has relied on self-reports of overall health (e.g., “Would you say your health in general is excellent, very good, good, fair, or poor?”) [8]. While useful for assessing broad perceptions of different aspects of health combined, it is not clear what roles mental versus physical health play in reporting. Disentangling these two unique components of self-rated health may be useful [9, 23,24,25,26,27], so the current study works toward this goal by examining a measure of self-rated physical health. Future research should compare these findings with self-reported mental health and seek to identify explanatory factors that may work in unique ways for each aspect of overall health.

The life course perspective argues that individuals are embedded in powerful social contexts, and that developmental processes must be examined within the full breadth of the life span, beginning at conception and continuing throughout the later years [28, 29]. One implication of this view is that adult outcomes must be understood in the context of early-life experiences, where habits, practices, relationships, and commitments are initially formed and then reverberate over time [30]. This means that early-life experiences such as parent–child relations, family structure, socioeconomic status, abuse, childhood health, immigration experiences, and religious participation may shape adult outcomes, including self-rated health [6, 23, 31].

To date, most research in this area has used data from economically developed and Western countries, and relatively little work has been conducted in other nations [23, 27, 32, 33]. Further, only a few studies have examined the associations between early-life experiences and self-rated physical health specifically [23, 27, 33,34,35,36,37], and even fewer have been published using data from countries outside the USA [23, 27, 33]. Since this outcome has rarely been examined in the context of early-life experiences, and research specifically on self-rated physical health is more sparse than overall general health, the following review documents studies of self-rated health broadly defined in order to help build a conceptual framework. The studies that have been done specifically on self-rated physical health are noted.

To begin with, considerable research shows that parent–child relations during childhood have important implications for self-rated health in adulthood [35, 38]. Early-life family structure (e.g., two-parent, step, and single-parent) also matters [39]. Individuals raised in families with both parents appear to have better self-rated health, as well as more desirable outcomes on smoking, alcohol consumption, nutrition, physical activity, and health complaints; in contrast, experiencing a divorce during childhood may be a risk factor for poor outcomes [40, 41]. We do not fully understand the mechanisms underlying parent–child variables, but these studies suggest that offspring educational attainment, household income, social support, and poor health behaviors such as cigarette smoking may play key roles in mediating these long-term associations. Most of our knowledge in these areas is based on general self-rated health, but a few promising studies specifically on self-rated physical health have been published, primarily examining the effects of composite measures of family life that include indicators of parent marital status and family experiences [27, 35,36,37].

Early-life SES also shapes self-rated health in adulthood, and findings have been reported around the world including Brazil [42], several European nations [43], China [33], and the USA [6], among others [23]. The health consequences of SES are not confined to the individual level (e.g., personal education, income, occupation); findings also suggest that neighborhood conditions are significant [44, 45]. There are many reasons why early-life SES is linked with adult health, and the studies cited above suggest that social capital, exposure to violence, family conflict and stress, health behaviors like smoking and dietary practices, access to health care, and various psychological characteristics including a sense of personal control, optimism, and pessimism may play important roles. Even though most research on these topics has used overall self-rated health, a couple of recent studies of self-rated physical health have been published, and importantly, these include data from diverse countries around the world [23, 33].

Child abuse and other adverse early-life experiences also shape self-rated health in adulthood [27, 31, 36, 46, 47]. Childhood adversity can take many forms including physical and emotional abuse, sexual mistreatment, witnessing parental domestic violence, and experiencing parental divorce, and these can lead to poor self-rated health, as well as functional limitations, diabetes, and heart disease, among other outcomes in adulthood [48, 49]. Early-life abuse and adversity may work through various indirect pathways including self-compassion [50], attachment styles [51], socioeconomic attainment [52], epigenetic mechanisms for disease risk [53], psychological distress and subsequent adverse experiences [54], and relationship problems including negative beliefs about others [55]. Several papers in these areas have reported results specifically for self-rated physical health [27, 34,35,36,37], but additional work is warranted since most studies use general self-rated health.

Childhood health and related health behaviors also have strong associations with adult health [56, 57]. These relationships are robust and powerful, and one study found that individuals who reported poor childhood health had three times greater odds of poor self-rated health in adulthood [58]. This study also found increased risk for disability and chronic health problems among individuals with poor childhood health. These associations may function through several mechanisms including general practitioner consultations among adults [59], self-perception as an unwell or stigmatized person [60], and many other consequential adult outcomes including anxiety, substance abuse, criminal behavior, social functioning, and socioeconomic attainment [61,62,63].

Immigration experiences are also important for self-rated health [64, 65]. These associations arise for a variety of reasons including differences among immigrants compared with native-born individuals in health behaviors (e.g., diet, alcohol and tobacco consumption, exercise), utilization of health care services, human and social capital, living and working conditions, real and perceived discrimination, fear of deportation, and family separation, among others [66, 67]. Relatively few studies have directly compared self-rated health among immigrants and non-immigrants using nationally representative samples from nations as diverse as those included in the GFS, so the findings described here will provide unique insight into this important aspect of social life. There do not appear to be any published studies specially on self-rated physical health, so the current findings will address this shortcoming in the literature.

Religious service attendance is another key predictor of self-rated health [22, 68, 69]. Despite debates about causal order, there is evidence for a pathway from religiosity to self-rated health, at least among adults in the USA [70], and there is some evidence for childhood religious upbringing as well [71]. Various mechanisms for the effects of religious attendance have been proposed including social support, self-regulation, health behaviors and other lifestyle factors, and psychological characteristics such as a sense of personal control and meaning in life [68, 72,73,74]. A few studies have been published on early-life service attendance and health in adulthood [75], but very little work has been done specifically on self-rated health [71]. Even less has examined service attendance and self-rated health from an international or cross-national perspective [22], and there do not appear to be any studies of self-rated physical health specifically.

Finally, two key demographic factors—year of birth and gender—are likely to shape adult self-rated health because they structure health advantages and disadvantages from the day people are born [4, 21, 76]. Many factors related to year of birth are relevant, including economic conditions at the time, political stability, peace and conflict, educational opportunities, media and advertising, infectious diseases, and dietary and physical fitness practices, among many others [77]. These vary across time and place and may shape self-rated health beginning in childhood and continuing across the entire life course. The literature also confirms that gender is an important predictor of self-rated health [78, 79]. Considerable research has been devoted to the developmental origins of gender identities, roles, and stereotypes, and this work demonstrates that gender-related outcomes begin taking shape during the first few years of life [80]. Through interactions with others—including parents, other family members, peers, and teachers—children learn the expectations associated with their gender, and these may shape self-rated health in part because they often involve differences in key roles, status, norms for behavior, and power [81]. Gender differences in health-related outcomes such as personality, mental health, and social behaviors and orientations likely play a role as well [82]. Together, these factors may shape self-rated health across the entire life course, starting early and continuing into the later years. Some research has been done specifically on self-rated physical health [83, 84], but this work is rare compared with self-rated health in general, and additional studies are needed.

These issues are addressed here by analyzing data from the first wave of the Global Flourishing Study (GFS) [85]. Three main research questions, which were revised slightly from versions preregistered with the Center for Open Science (COS) (https://osf.io/4zp7y), guide this study. First, are childhood experiences and characteristics (i.e., parent–child relations, parent marital status, financial conditions, abuse and adversity, early-life health, immigration experiences, religious participation, year of birth, and gender) associated with self-rated physical health in adulthood? Second, do the strengths of these associations vary by country, possibly reflecting country-specific influences and contexts? If so, this cross-national variation may illuminate the role of broad societal factors in shaping the relationships between childhood characteristics and adult self-rated physical health. And third, are these associations robust against potential unmeasured confounding? This will be assessed through E-values, which gauge how strong confounding by an unmeasured variable would need to be to explain away the observed associations. The following hypotheses are examined: First, childhood characteristics will be associated with self-rated physical health in adulthood. Second, the strength of these associations will vary by country. And third, these associations will be robust against potential unmeasured confounding.

Methods

The description of the methods below has been adapted from VanderWeele et al. [85]. Further methodological detail is available elsewhere [86,87,88,89,90,91,92,93,94,95,96], including the GFS resources webpage (https://www.cos.io/gfs-resources).

Data

Data come from the GFS, which examines the distribution and determinants of well-being across a sample of 202,898 participants from 22 geographically, economically, and culturally diverse countries. Wave 1 collected nationally representative data from the following countries and territories: Argentina, Australia, Brazil, Egypt, Germany, Hong Kong (Special Administrative Region of China, with mainland China also included from 2024 onwards), India, Indonesia, Israel, Japan, Kenya, Mexico, Nigeria, the Philippines, Poland, South Africa, Spain, Sweden, Tanzania, Turkey, the UK, and the USA. These countries were chosen to (a) maximize coverage of the world’s population; (b) ensure geographic, cultural, and religious diversity; and (c) prioritize feasibility and existing data collection infrastructure. Gallup Inc. conducted the data collection primarily in 2023, although some regions began in 2022; timing varied by country, and more information can be found elsewhere [90]. The precise sampling design to ensure nationally representative samples varied by country [90]. Face-to-face, telephone, and/or web-based interviews were conducted to obtain the most representative samples possible based on local conditions and infrastructure (see [90] for details). Each country/territory survey underwent rigorous question development, pretesting, cognitive testing, language translation assessments, and quality checks for consistent meanings in survey questions across contexts, national representativeness, reliability, and validity, and participants were ensured that confidentiality would be maintained [86,87,88, 90,91,92, 94, 95]. Plans are in place to collect four additional waves of annual panel data on the participants from 2024 to 2027. The data are publicly available through COS (https://www.cos.io/gfs). The translation process followed the TRAPD model (translation, review, adjudication, pretesting, and documentation) for cross-cultural survey research (ccsg.isr.umich.edu/chapters/translation/overview) [90, 92]. Additional details are available in the GFS Questionnaire Development Report and scholarly paper [86, 88], Methodology [93,94,95], Codebook (https://osf.io/cg76b) [96], and Translations documents [92].

Measures

Dependent variable

Self-rated physical health was measured with a single indicator [97]. Respondents were asked: “In general, how would you rate your PHYSICAL health?” They were instructed to rate their physical health on a 0–10 scale, where 0 = poor and 10 = excellent. This measure was treated as continuous in all analyses. The vast majority of previous research has asked about overall health more broadly, so the current study provides insight into the predictors of self-rated physical health specifically.

Independent variables

Relationship with one’s mother during childhood was assessed with the question: “Please think about your relationship with your mother when you were growing up. In general, would you say that relationship was very good, somewhat good, somewhat bad, or very bad?” Responses were dichotomized to very/somewhat good versus very/somewhat bad. An analogous variable was used for the relationship with father. Does not apply was treated as a dichotomous control variable for respondents who did not have a mother or father due to death or absence. Parental marital status during childhood was assessed with responses of married, divorced, never married, and one or both had died. Family financial status was measured with: “Which one of these phrases comes closest to your own feelings about your family’s household income when you were growing up, such as when you were around 12 years old?” Responses were lived comfortably, got by, found it difficult, and found it very difficult. Abuse was assessed with yes/no responses to: “Were you ever physically or sexually abused when you were growing up?” Participants were separately asked: “When you were growing up, did you feel like an outsider in your family?” Childhood health was assessed by: “In general, how was your health when you were growing up? Was it excellent, very good, good, fair, or poor?” Immigration status was assessed with: “Were you born in this country, or not?” Religious service attendance during childhood was assessed with: “How often did you attend religious services or worship at a temple, mosque, shrine, church, or other religious building when you were around 12 years old?” Responses were at least once/week, one-to-three times/month, less than once/month, and never. Childhood religious tradition had response categories of Christianity, Islam, Hinduism, Buddhism, Judaism, Sikhism, Baha’i, Jainism, Shinto, Taoism, Confucianism, Primal/Animist/Folk religion, Spiritism, African-Derived, some other religion, or no religion/atheist/agnostic; response categories varied by country [87]. When the category no religion/atheist/agnostic had more than 5% of the within-country sample size, this was used as the reference category; otherwise, the most prominent religious group was used. Additionally, all religious categories endorsed by less than 3% of the within-country sample size were collapsed into a single religious category of “other.” For inclusion in the childhood predictor regression analyses, race/ethnic identity was collapsed as a binary variable of whether an individual was in the most prominent group versus a minority group (race plurality).

Analytic strategy

A weighted linear regression model with complex survey-adjusted standard errors was fit within each country for self-rated physical health on all of the aforementioned childhood variables simultaneously. A Wald-type test was conducted to obtain a global (joint) test of the effect of all categories within a childhood variable resulting in a single global p-value of the effect of each childhood characteristic. In the primary analysis, a random effects meta-analysis of the regression coefficients was estimated [98, 99] along with confidence intervals, lower and upper limits, 95% prediction intervals, heterogeneity (τ), and I2 for evidence concerning variation within a given variable across countries [100]. Forest plots of estimates are available in Additional file 1. Religious tradition and race/ethnicity were used within-country as control variables when available, but these coefficients themselves were not included in the meta-analyses because categories/responses varied by country. The meta-analysis was conducted in R using the metafor package [101]. Within each country, a global test of association of each childhood variable with the outcome was conducted, and a pooled p-value across countries was reported concerning evidence for association within any country [102]. Bonferroni-corrected p-value thresholds were provided based on the number of childhood variables [103, 104]. For each indicator, E-values were calculated to evaluate the sensitivity of results to unmeasured confounding. An E-value is the minimum strength of the association an unmeasured confounder must have with both the outcome and the predictor, above and beyond all measured covariates, for an unmeasured confounder to explain away an association [105]. As a supplementary analysis, a population-weighted meta-analysis of the regression coefficients was estimated. All analyses were pre-registered with COS prior to data access, with only slight subsequent modification in the regression analyses due to multicollinearity. Code to reproduce the analyses is openly available in an online repository (https://osf.io/vbype/overview) [89].

Missing data

Missing data on all variables were imputed using multivariate imputation by chained equations, and five imputed datasets were used [106, 107]. To account for variation in the assessment of certain variables across countries (e.g., religious tradition and race/ethnicity), the imputation process was conducted separately in each country. This within-country imputation approach ensured that the imputation models accurately reflected country-specific contexts and assessment methods. Sampling weights were included in the imputation models to account for specific-variable missingness that may have been related to the probability of inclusion in the study.

Accounting for complex sampling design

The GFS used different sampling designs across countries based on the availability of existing panels and recruitment needs [90]. All analyses accounted for the complex survey design components by including weights, primary sampling units, and strata. Additional methodological detail, including accounting for the complex sampling design, is provided elsewhere [93].

Multicollinearity

Some degree of multicollinearity was expected because many of the childhood variables are correlated to some degree. The magnitude was expected to be minimal, however, since there are important conceptual distinctions between all of the measures included in the original study preregistration. Exploratory analyses revealed potentially problematic levels of multicollinearity with four variables that were originally intended to be included in the study: religious service attendance of mother, religious service attendance of father, love from mother, and love from father. These variables were removed from the study due to substantial correlations with childhood religious service attendance and relationships with mother and father. For additional information, see [90].

Results

Descriptive statistics

Table 1 provides descriptive statistics for all measures for the 22 countries combined. A majority of participants reported that (a) they had a very good relationship with their mother (63%) and father (53%) when growing up; (b) their parents were married when they were about 12 years of age (75%); (c) their family either got by financially or lived comfortably when they were around age 12 (76%); (d) they had very good or excellent health while growing up (64%); and (e) they attended religious services at least once a month when they were around age 12 (57%). Most participants were born in the country where data were collected (94%). A minority of participants indicated they experienced abuse (14%) or felt like an outsider in their family when growing up (14%). Age ranged the entire adult lifespan (18–80 +), and the largest cohort (20%) represented individuals born between 1983 and 1993. Gender was almost equally divided among women (51%) and men (49%), with a small number of other gender identities. Nationally representative descriptive statistics for each individual country are provided in Tables S1a–S22a in Additional file 1.

Table 1 Nationally-representative descriptive statistics of the observed sample*
Full size table

Childhood characteristics and adult self-rated physical health: meta-analysis

Results from the random effects meta-analysis are provided in Table 2. Across all 22 countries combined and controlling for all of the other variables in the table, there were positive associations between self-rated physical health and very/somewhat good relationships (compared with very/somewhat bad) for both the mother (β = 0.10, 95% CI 0.02, 0.17) and the father (β = 0.07, 95% CI 0.01, 0.13). The effects of parents being divorced (β = − 0.04, 95% CI − 0.14, 0.06), single/never married (β = − 0.10, 95% CI − 0.20, 0.00), and deceased (β = − 0.10, 95% CI − 0.20, 0.00) were all negative compared with having parents who were married, though with less evidence for divorce with its smaller estimate and 95% confidence interval. There was a strong association between financial status growing up and self-rated physical health in adulthood: lived comfortably (β = 0.18, 95% CI 0.11, 0.26), found it difficult (β = − 0.12, 95% CI − 0.17, − 0.07), and found it very difficult (β = − 0.30, 95% CI − 0.39, − 0.21) were all significantly different compared with got by. Both abuse (β = − 0.40, 95% CI − 0.49, − 0.31) and feeling like an outsider growing up (β = − 0.18, 95% CI − 0.23, − 0.13) were significantly different compared with not experiencing these deleterious conditions. Similarly, fair (β = − 0.46, 95% CI − 0.58, − 0.33) and poor (β = − 0.82, 95% CI − 1.12, − 0.53) health growing up were associated with lower levels of self-rated physical health in adulthood compared with good. In contrast, excellent (β = 0.70, 95% CI 0.52, 0.87) and very good (β = 0.39, 95% CI 0.30, 0.48) health were positively associated with adult self-rated physical health. There was less definitive evidence for immigration status. However, there was stronger evidence for religious service attendance: less than monthly (β = 0.10, 95% CI 0.05, 0.15), monthly (β = 0.21, 95% CI 0.13, 0.29), and weekly (β = 0.21, 95% CI 0.12, 0.30) attendance were all positive compared with never. There was a somewhat linear and negative association between self-rated physical health and year of birth. Some neighboring categories were not significantly different from each other based on overlapping confidence intervals, but there were significant differences between younger and older cohorts, with self-rated physical health being worse among older groups. Compared with men, women reported significantly worse health (β = − 0.17, 95% CI − 0.25, − 0.09). Overall, these results provide support for the first hypothesis for most variables.

Table 2 Random effects meta-analysis of regression of self-rated physical health on childhood predictors
Full size table

There was considerable heterogeneity in how some early-life conditions were associated with adult self-rated physical health across the 22 countries. Heterogeneity estimates were relatively low (< 0.20) for several variables including relations with both mother and father, parent marital status, financial well-being, abuse, feeling like an outsider, religious service attendance, and gender. This indicates that these variables have somewhat similar associations with self-rated physical health across countries, though for abuse, weekly service attendance, and feeling like an outsider, there was more substantial heterogeneity. In contrast, there was more heterogeneity for childhood health, immigration status, and year of birth. These results provide some support for the second hypothesis. Additional details regarding country-specific relationships can be found in Tables S1b–S22b in Additional file 1. These results, which provide additional support for the second hypothesis, are summarized in the “Discussion” section.

Sensitivity analysis

Table 3 shows E-values and E-value limits for the regression findings shown in Table 2. These results suggest that some relationships are quite robust to unmeasured confounding. Several variables including finding it very difficult to get by financially while growing up, abuse, childhood health, and age had E-values of 1.50 or higher. As an example, to explain away the association between childhood abuse and adult self-rated physical health (β = − 0.40), an unmeasured confounder associated with both variables by risk ratios of 1.62 each (above and beyond all other covariates included in the model) could suffice, but a weaker joint confounder association could not. Additionally, in order to explain away the 95% confidence interval for the association, an unmeasured confounder would need to be associated with both abuse and self-rated physical health by risk ratios of 1.52 each to shift the confidence interval to include the null, but weaker joint confounder associations could not. In contrast, other variables including parent–child relations, parental marital status, feeling like an outsider, immigration status, religious service attendance, and gender have somewhat smaller E-values and may therefore be more susceptible to confounding. These results provide some support for the third hypothesis, but the robustness of each variable varied.

Table 3 Sensitivity of meta-analyzed childhood predictors to unmeasured confounding
Full size table

As an additional sensitivity analysis and alternative to the primary random effects meta-analysis, Table S23 in Additional file 1 shows results from a fixed effects meta-analysis using 2023 population size weights reflecting differences in the number of people in each country. In other words, this analysis was weighted by individuals in each country rather than by country, which elevated the influence of more populous countries such as India. Compared with Table 2, the associations for parent–child relationships and marital status were weaker. Lived comfortably remained important compared with got by, but found it difficult or very difficult did not. Childhood abuse and feeling like an outsider were similar, and the same was true for childhood health. Frequent religious attendance remained important. The general decline in self-rated physical health across birth cohorts and lower levels of self-rated physical health among women compared with men were still present. In general, the overall pattern of findings remained somewhat consistent with the primary analyses, suggesting the findings are fairly robust.

Discussion

According to the life course paradigm, individuals are embedded in social contexts from the day they are born, development is a lifelong process, and many different habits, practices, relationships, and commitments formed during childhood have consequences that extend well into the later years of life [28,29,30]. Drawing on this insight, the current study examined the associations between a variety of childhood characteristics and self-rated physical health in adulthood. Findings from a random effects meta-analysis of all 22 countries combined showed that parent–child relations, financial status growing up, abuse, and feeling like an outsider during childhood, early-life health, religious service attendance around age 12, year of birth, and gender were all associated with self-rated physical health in adulthood. There was considerable variation in how some early-life conditions were associated with adult self-rated physical health across countries, however. Heterogeneity estimates were relatively low for relations with mother and father, parent marital status, financial well-being, abuse, feeling like an outsider, religious service attendance, and gender, indicating that these variables had somewhat similar associations with self-rated physical health across countries. There was more variation across countries for childhood health, immigration status, and year of birth. Separate regression models for each individual country are available in Tables S1b–S22b in Additional file 1, and are discussed in detail below.

To begin with, existing research suggests that relationships with one’s parents during childhood may shape self-rated health well into adulthood [35, 38]. In the current data, a somewhat or very good relationship with one’s mother growing up was positively associated with self-rated physical health compared with somewhat or very bad in Germany, Japan (marginally—i.e., significant without a Bonferroni correction), and Mexico. A somewhat or very good relationship with one’s father was associated with better self-rated physical health in Argentina, Brazil, and Japan. Family structure has been linked with adult self-rated health as well [39, 40]. Here, parental divorce was negatively associated with self-rated physical health compared with married in Egypt (marginally), Japan (marginally), Kenya, the Philippines (marginally), and Poland. There was an inverse association with self-rated physical health among those whose parents were never married in Egypt (marginally), Israel (marginally), Kenya, and Sweden, and among those with parents who had died in India (marginally) and Nigeria. These findings add to research in this area, specifically on self-rated physical health, but more work is needed on this topic [27, 35,36,37]. Overall, these findings were weaker than one might expect based on existing research, and there are at least two possible explanations. First, the sample included only adults, and the average age was roughly 46 years, so the time between these influences and the measurement of health is large in some cases. Second, it is possible that the effects of the parent variables were explained by other variables in the model, particularly financial status, abuse, feeling like an outsider, and childhood health. Future research should disentangle the complex interconnections between the different childhood characteristics examined here in order to identify both direct and indirect pathways. Further, family structure may have a weaker association with health in contexts where generous welfare policies buffer economic hardship and where cultural norms—such as high rates of shared physical custody and comparatively stable parental unions—provide additional support, as exemplified by several Scandinavian countries [108, 109]. Thus, local contexts are obviously important for these associations.

Consistent with previous research [23, 33, 42,43,44,45], financial status was associated with self-rated physical health in numerous countries. Compared with those who reported getting by financially during childhood, there was a positive association between self-rated physical health and living comfortably while growing up in Argentina, Australia, Brazil, Hong Kong, Indonesia, Japan, Kenya (marginally), Mexico, the Philippines, Sweden, the UK (marginally), and the USA. In contrast, there was a negative association among those who reported (a) having a difficult time getting by early in life in Egypt (marginally), Germany (marginally), Indonesia, Japan, Poland, and Spain; and (b) having a very difficult time in Germany, India, Japan, Kenya (marginally), Poland, Tanzania, and Turkey (marginally). Countries where significant associations occurred were diverse in terms of region, economic development, per capita income, life expectancy, and majority religious tradition. Given that financial strain has been linked with worse health in general [110, 111], it is somewhat surprising that this association was not found in every country. Heterogeneity estimates were somewhat low, but there was still variation across countries. One possible explanation is that macro-level contextual factors moderate these associations. For example, childhood material deprivation and its adult health consequences may be partially buffered in countries with generous welfare states, inclusive health-care systems, and egalitarian education regimes; in contrast, high income inequality, insecure labor markets, stratified schools, and lower social protections may lead to the accumulation of socioeconomic disadvantage that subsequently translates into poorer adult health through fewer opportunities for upward mobility, greater chronic stress, and reduced access to high-quality care [112,113,114]. Future research should carefully examine these possibilities and processes in each country to promote a deeper understanding of these complex associations.

Childhood abuse is likely to shape self-rated health in adulthood as well [27, 31, 36, 46, 47]. In the GFS data, abuse was negatively associated with self-rated physical health in Argentina, Australia, Brazil, Germany, India, Indonesia, Japan, Kenya, Mexico, the Philippines, Poland, South Africa, Spain, Sweden, Tanzania, the UK, and the USA. This question was not asked in Israel, and the results were not significant in the remaining countries. Feeling like an outsider growing up also had a negative association with self-rated physical health in several countries including Australia, Brazil, Germany, Israel (marginally), Mexico, Spain, Sweden (marginally), and the USA. A variety of adverse childhood experiences including mental, physical, and sexual abuse have been linked with self-rated health in previous research [46, 47], and the current findings are consistent with this work in most countries. In addition, most research on this topic has drawn on general health, not self-rated physical health, so the current study adds to research in this area [27, 34,35,36,37]. Surprisingly, neither abuse nor outsider status was associated with adult self-rated physical health in Egypt, Hong Kong, Nigeria, and Turkey, four diverse nations. They were, however, in the expected direction in Egypt and Nigeria, but not in Hong Kong or Turkey (abuse was not but outsider was in Turkey). Abuse was somewhat lower in Egypt (9%), Hong Kong (11%), Nigeria (13%), and Turkey (11%) compared with the overall average of 14% for all countries combined. Outsider status was lower in Egypt (5%), Nigeria (10%), and Turkey (11%), but substantially higher in Hong Kong (22%), compared with the 22-country average of 14%. Based on these findings, there do not appear to be any clear similarities among these countries (or differences with other countries) that might explain the null findings for self-rated physical health. Future research should attempt to discover factors that might be unique to these contexts.

Consistent with considerable previous research [56,57,58], childhood health status was associated with adult self-rated physical health in every country except Nigeria. There was a somewhat linear and positive trend of increasing self-rated physical health as categories of child health increased from poor to excellent in most countries. In some cases, not all categories were significantly different from the reference group, which was good health. Health status early in life may have lasting consequences across the entire life course because (a) many key biological and developmental processes occur early in life; (b) stressful conditions can have profound effects when the brain and body are developing; and (c) epigenetic processes may contribute to trajectories of health that begin early in life [115,116,117]. Importantly, many of these effects may be stronger in developing nations due to greater exposure to disease, pollution, war, and other adverse conditions across the entire life course [118].

Recent work has argued that immigration is a social determinant of health [67], and it was associated with self-rated physical health in nine of the 22 countries examined here. In Argentina, Australia (marginally), Germany, Japan, Spain, Turkey (marginally), and the UK, immigrants had better self-rated physical health compared with native-born individuals. These findings are consistent with previous research on the healthy immigrant effect, which may be due to selection whereby healthy people are more likely to migrate, desirable health behaviors being more common among migrants compared with the native population, and social and cultural factors such as strong family and community bonds [119, 120]. In contrast, immigrants had worse self-rated physical health in Israel and the Philippines. In some contexts, immigrants face challenges in their new home countries including finding employment and housing, gaining access to health care, negative religious experiences, and real or perceived discrimination [121], all of which could undermine their health and explain the findings for these two countries. Given that immigrants reported better self-rated physical health in some countries and worse in others, future research should carefully examine both cases. In addition to the healthy immigrant effect, other processes that may help us understand these findings include acculturation and adaptation to a new country [122], immigration policies and legal status [123], discrimination [124], and origin–destination differences in culture and economics [125]. These and other contextual factors likely operate in unique ways in different countries around the world and may contribute to differences in immigrant versus native-born self-rated physical health. It is also important to note that the timing of immigration was not known in the current data, so one reason why the findings were somewhat weak may be because the experience of immigration happened years ago, and thus, the health consequences may have diminished. These findings appear to be the first ones reported for self-rated physical health specifically, so more work in this area is warranted.

Childhood religious service attendance was positively associated with self-rated physical health in several countries. In Germany, Japan, and Poland, less than monthly attendance, monthly attendance, and weekly attendance were all associated with higher self-rated physical health compared with never attend (monthly was marginal in Germany and less than monthly was in Japan). In Egypt, Hong Kong, and Sweden, monthly and weekly reported higher self-rated physical health compared with never (monthly was marginal in Egypt and weekly was in Sweden), whereas self-rated physical health was higher in the UK for less than monthly and weekly compared with never. These findings are consistent with existing research showing salutary associations between attendance and health [22, 68, 69]. This relationship was even present in several “secular” nations including Germany, Hong Kong, and Sweden. Possible explanations for the effects of attendance include differences in health behaviors and other lifestyle factors among those who do and do not attend [73], as well as psychological characteristics such as a sense of personal control [126] and meaning in life [74]. It is important to note, however, that most previous research on this topic has focused on attendance during adulthood and examined correlations with adult health [68]. There is some evidence supporting the possibility that early-life religious service attendance is associated with health in adulthood [75], but research on this topic is rare and our knowledge remains limited, especially for self-rated health specifically [71]. The current study appears to be the first to report findings on early-life attendance and self-rated physical health in diverse countries, so the findings add much needed empirical evidence in this area. That said, there were null findings for numerous countries, so future research should examine each nation and culture in detail to determine why attendance matters in some countries but not others. Among other possible explanations, this could be due to some individuals attending during childhood but not as adults, or to a decline in the importance of religion over time. Childhood attendance may matter primarily for those who have remained religious as adults, although some residual effects of early-life participation may endure through private beliefs and practices.

Year of birth and gender may shape adult self-rated physical health in part because they structure health-related advantages and disadvantages from birth until death [4, 21, 76]. These operate through many potential mechanisms including economic and political conditions, peace and war, educational opportunities, media influences, prejudice and discrimination, exposure to infectious diseases, dietary practices, physical fitness, personality, emotions, and social roles and connections [77]. In the current data, year of birth had complex associations with self-rated physical health across countries. In some (Indonesia, Israel, Kenya, Nigeria, the Philippines, Poland, Tanzania, and Turkey), there was a somewhat linear negative association between year of birth and self-rated physical health. In several others (Argentina, Egypt, Germany, India, Spain, and the UK), the relationship was primarily linear, but with a small increase in self-rated physical health among the oldest cohort(s). Four countries had a U-shaped pattern. In Australia, Japan, and Sweden, the youngest and oldest cohorts had better self-rated physical health compared with the middle ones. In contrast, younger and older cohorts had worse health in Hong Kong. In the remaining countries (Brazil, Mexico, South Africa, and the USA), there were significant differences between the youngest birth cohort and some other categories, but there were no clear linear or curvilinear patterns to these associations. Consistent with some previous research [78, 79], women had worse self-rated physical health than men in Argentina, Brazil, Egypt, India, Israel, Kenya, Mexico, Poland, South Africa, Spain, Sweden, Tanzania, Turkey, and the United Kingdom. In contrast, women had significantly higher self-rated physical health in one country: Japan. Possible explanations for this unique finding may include interpreting long life expectancy as good health, relatively high socioeconomic status and financial conditions, and pressure to present oneself as healthy regardless of actual health status or challenges [79, 127]. Future research should examine these possibilities in detail, and more work specifically on self-rated physical health is needed.

Religious tradition and race/ethnicity were not included in the meta-analysis, but were examined in individual countries. There were a large number of categories in some countries, so describing the results was quite difficult. Detailed findings are provided in Tables S1b–S22b in Additional file 1, but here is a brief summary. Religious tradition had a complex relationship with self-rated physical health across countries. There were null findings for most countries, but in Egypt (marginally), Germany, and Poland, members of minority religions had worse self-rated physical health compared with members of the largest group. In contrast, self-rated physical health was higher among members of smaller religious traditions compared with the largest group in Hong Kong (marginally), Nigeria (marginally), and Sweden (Christianity was significant but other religion was only marginally significant). The current analyses may not capture all important religious group differences, however, and future cross-national research should examine aspects of religion and spirituality more closely [68]. The findings for racial/ethnic differences were weak or null in most countries. In Germany and Tanzania (marginally), members of minority groups had lower self-rated physical health than the largest group, but in Brazil, minorities had better self-rated physical health (but this finding was only marginally significant). Data were not available for Germany, Japan, Spain, or Sweden. Moving forward, religion and race scholars with expertise on each nation should examine these findings in detail and offer insights based on their knowledge of local cultures and contexts.

This study has several strengths. First, the sample includes economically developed and developing nations, religiously diverse countries (including nations with majorities of several major religious traditions including Christianity, Hinduism, Islam, and Judaism), and nations that are geographically dispersed across North America, South America, Europe, Asia, Africa, and Oceania. The 22 countries are also politically diverse. This means that the findings are relevant to diverse individuals and groups of people around the world. Second, all survey items were carefully chosen and evaluated by leading scholars from around the world, and then extensively pretested by Gallup personnel on the ground in each country [86,87,88, 91, 94, 95]. This helped to ensure that the data met the highest standards for relevance and reliability. Third, the GFS is a very large survey, with 202,898 participants in 22 diverse countries. Given the large and representative samples, findings based on GFS data should offer reasonable estimates of many key constructs, including self-rated physical health, compared with research based on small, non-random, or specialized samples. Fourth, many of the countries included in this study have rarely been examined, so the findings provide new insights into variations in self-rated physical health around the world. Fifth, this is one of the first studies to examine a specific measure of self-rated physical health in multiple nations and to identify associations with key early-life experiences. Future research should extend this work to include measures of self-rated mental health for comparison.

There are limitations as well. First, this study is cross-sectional and only analyzes the first wave of GFS data, though the childhood predictors were assessed retrospectively, creating a synthetic longitudinal study of sorts. The baseline survey data were released on 2/13/24, and data collection for the second wave concluded at the end of 2024. Second, all variables were entered into the models together, so the findings for each variable reported here are net of all other variables. It is possible that some non-significant variables in the models are actually important, but are mediated or otherwise explained by one or more other variables. Follow-up research should address this by examining the interconnections between the different childhood variables examined here. Third, the childhood predictors were assessed retrospectively and may be subject to recall bias. However, for recall bias to completely explain away the observed associations, the effect of adult self-rated physical health on biasing retrospective assessments of the childhood predictors would essentially have to be at least as strong as the observed associations themselves [128], and some of these were quite substantial. And fourth, the adult self-rated physical health question is part of a larger, multi-item measure of flourishing [129] that will be asked in each wave of the GFS, whereas the childhood measure was adapted from other measures in the literature and will only be measured once (with a brief intake survey). The adult question specifically focused on physical health and had response choices ranging from 0 to 10, whereas the child question asked about overall health and had five categories. While it is possible that some participants may have noticed these differences and been confused about how to respond, they did not receive these questions back to back and may have even answered them at different times, so this is not likely to be a source of major bias or confusion in the study.

A note on cross-cultural research is also necessary here. Conducting surveys in diverse nations is difficult because questions must be translated into many different languages while maintaining comparable meanings, and norms surrounding personal issues like health (including speaking publicly about them) vary around the world [92, 130,131,132]. As described in the GFS documentation [85,86,87,88, 91, 92, 94, 95], Gallup conducted extensive survey pretesting, cognitive interviews, and translation work in hopes of obtaining comparable meanings of survey items across countries, but this was difficult because some words and concepts simply do not have clear analogs in different languages and countries. For this reason, questions in each country are best viewed as closely related but not necessarily identical assessments. The analytic strategy was formulated with this in mind. Each country was treated as its own cohort study, and meta-analytic techniques were used to aggregate separate within-country analyses of 22 closely related cohort studies. This approach did not assume items were interpreted identically across countries, only as closely related constructs. For interested readers, Gallup created a 922-page “Compendium of Global Flourishing Study Translations,” which can be downloaded from COS (https://osf.io/47jyf). It provides detailed information on all translations, as well as examples of the process Gallup used to obtain the most comparable meanings in survey items across countries. For example, there are several words in Mandarin Chinese that are commonly translated as “happy” (e.g., kuài lè, xìng fú, gāo xìng, and kāi xīn), but each has a slightly different meaning. The Compendium describes the translation process for this variable and a few others in detail, but the process was similar for all GFS items. The GFS documentation provides additional information about the methodology and pre-survey cognitive interview process [91, 94, 95].

Future research should expand upon these findings reported here in several ways. First, scholars should attempt to determine why self-rated physical health is relatively high in some countries and low in others (and not always correlated with GDP), and why some variables relate to higher self-rated physical health in some nations but not in others. While differences in socio-cultural context likely explain some of these variations among countries, this study has not explicitly tested that proposition. Second, future research with at least two waves of GFS data should examine how childhood characteristics shape longitudinal trends in self-rated physical health. Third, researchers working in each country should compare their findings with those reported here in hopes of better understanding the dynamics that shape self-rated physical health. Fourth, more research is needed on developing countries and nations that are not predominantly or historically Christian. This study is among the first to report findings from large nationally representative samples in several developing, non-Western, and non-Christian countries. And fifth, as noted above, there was considerable variation across countries in the associations between self-rated physical health and childhood health, immigration status, and year of birth. Contextual factors and potential country-level moderators like levels of government involvement in population health, social safety nets, gross domestic product (GDP), income inequality, social capital, individualism/collectivism, immigration policies and levels, and various cultural norms may offer fruitful areas of inquiry for future research seeking to understand the causes of cross-national variation in the correlates of self-rated physical health [133,134,135,136].

Conclusions

An analysis of data from 22 geographically, economically, and culturally diverse countries showed that numerous early-life factors including parent–child relations, financial status growing up, abuse, and feeling like an outsider during childhood, health, religious service attendance, year of birth, and gender were all associated with self-rated physical health in adulthood. There was considerable variation across countries, however, warranting further exploration. These baseline findings lay the foundation for future longitudinal studies on the causes and correlates of self-rated physical health in a global context.