Child well-being is a key public health priority (1), and child mortality is commonly used as an economic and social welfare indicator (2, 3). Although the deaths of children and adolescents [defined as individuals aged 0–19 years (4)] have fortunately become rare in high-income countries (5), these deaths are far from insubstantial and often preventable (5, 6). Endeavors to further reduce child and adolescent mortality should be grounded upon a thorough understanding of the underlying factors that, in turn, could be exploited in the planning of preventive measures and resource allocation. The reduction of child and adolescent mortality remains a crucial public health priority also in high-income countries (7, 8).

The social determinants of health, as defined by the World Health Organization, comprise a wide range of factors that constitute the circumstances one lives in (9). Notably, the list includes common sociodemographic traits such as education level, employment rate, income level, neighborhood characteristics and crime rate. While the association between social inequality and child mortality is widely prevalent in low- and middle-income countries (10), a similar relationship seems to persist also in high-income countries such as Canada (11), the USA (12), and England (13).

Studies from high-income countries have previously reported several sociodemographic indicators of child and adolescent mortality. In a study from Quebec, Canada with data from the years 1990–2005, an areal socioeconomic deprivation score (comprised of employment, low education, and income) was associated with pediatric mortality in several age groups (11). Similarly, a study from the US with data from the years 1969 to 2000 showed that a county-level deprivation index (composite of 17 indicators, including educational, occupational, economic, and housing-related factors) was associated with child mortality (12). A study comprising two birth cohorts with data from 2003 to 2013 found a clear association between lower socioeconomic status and higher child mortality in Sweden (socioeconomic status proxied by income) and England (socioeconomic status proxied by a composite score of income, employment, disability, education, housing, living environment and crime) (13). However, the association between socioeconomic indicators and child mortality is still considered complex and relatively poorly understood (8), warranting further emphasis on the subject. In order to disentangle the effects of sociodemographic traits from each other, a comprehensive approach to the indicator variables and their intercorrelations is essential.

In Finland, 0.5% of all decedents are aged 0–19 years (14). Exploiting a nationwide dataset from Finland, the present study aimed to elucidate the sociodemographic predictors of child and adolescent mortality at the municipality level. The present dataset included a wide range of sociodemographic indicators which were modeled against child and adolescent mortality by means of generalized estimating equations (GEE). The results of this study were expected to provide municipality-level tools for the identification of geographical areas and/or population groups that have higher child and adolescent mortality.

The analysis was based on 310 municipalities with a median population size of 6,191 (interquartile range 2,901–14,689). The general sociodemographic characteristics of the municipalities are presented in Table 1. On average, 49.5% of the population were women, 2.0% were foreign language speakers, 34.9% only had primary education, and 11.9% were unemployed. A total of 2,371 child and adolescent deaths occurred, yielding an average annual mortality rate of 26.7 per 100,000 individuals. During the study period 2011–2018, the annual number of child and adolescent deaths fluctuated between 247 and 358 cases (Supplementary Table 2; Supplementary Figure 1). On average, both the number of deaths and mortality rate showed a decline of 3% per year (95% CI 1–5%) (Supplementary Figure 1; Table 2).

Univariate and full multivariable GEE models for the association between sociodemographic indicators and child and adolescent mortality are shown in Table 2; intermediate models are presented in Supplementary Table 3. Effect sizes (i.e., RRs for mortality) are presented relative to one standard deviation change in the predictor variable. In the preliminary univariate models, higher mortality was associated with population mean age, unemployment, rurality, and higher percentage of individuals living in their birth municipality. Correspondingly, lower mortality was associated with higher income, family size, and higher percentage of overcrowded households. In the full multivariable model, where the internal dependencies within the dataset were controlled for, population density was associated with higher child and adolescent mortality (rate ratio 1.03, 95% confidence interval 1.01–1.06), and the percentage of foreign language speakers was associated with lower child and adolescent mortality (0.96, 0.93–0.99).

Using a comprehensive dataset on 310 Finnish municipalities from the period 2011–2018, this study aimed to explore the association between sociodemographic indicators and child and adolescent mortality at the municipality level. Despite a fluctuating trend, the present analysis estimated child and adolescent deaths to have declined at an average rate of 3% per year over the study period. After accounting for the intercorrelations between the sociodemographic variables, child and adolescent mortality was associated with higher population density and lower percentage of foreign language speakers. The present findings are expected to advance the identification of geographical areas and population groups with higher child and adolescent mortality.

Child and adolescent mortality has undergone a significant decrease in Finland during the previous century (21). Although the number of annual deaths fluctuated over the study period, the positive decline seems to have continued also in the 2010s, thanks to investments in pediatric healthcare, child welfare and traffic safety, among other advances. It is, however, worth acknowledging that a total of 2,371 deaths still occurred during the study period. As the majority child and adolescent deaths are generally considered preventable (5), a need clearly remains for further research and actions enhancing the preventive process. Correspondingly, the differences between univariate and multivariable associations of the sociodemographic indicators with child and adolescent mortality underline the intercorrelated nature of the sociodemographic indicators behind mortality (8). In the full multivariable model, child and adolescent mortality was independently associated with only two sociodemographic indices, namely population density (positive association) and percentage of foreign language speakers (inverse association). However, it is worth acknowledging that the effect sizes of all predictors tended to be relatively small, indicating that the sociodemographic indices only have mild predictive power over child and adolescent mortality.

The positive association between population density and mortality has been previously reported in adult populations from the Netherlands (22), Denmark (23) and Japan (24). A number of socioeconomic indicators such as unemployment, poor housing conditions and crime rate were suggested as potential effect mediators. Quite surprisingly, the present analysis of child and adolescent mortality identified population density as the only socioeconomic or housing-related indicator which had an independent association with the outcome. It is particularly interesting that indices such as income, crime rate, and car ownership rate showed no association with mortality. While the generally low population density in Finland, residual confounding, and municipality-level perspective may explain the differing findings, future studies are encouraged to characterize the factors underlying this discrepancy. We speculate that high population density may also reflect “relative poverty”; even in a country with a low level of income inequality such as Finland (25), our finding would highlight this aspect of poverty upon children's lives. Nonetheless, children and adolescents living in Finnish municipalities with high population density have higher mortality than those living in less densely populated areas, even when total child and adolescent population sizes are accounted for. It would therefore stand to reason to primarily focus efforts to reduce child and adolescent mortality on densely populated areas.

Both the previous evidence of foreign language speakers associating with higher birth rate (18) and the present association between foreign language speakers and lower child and adolescent mortality underline the positive contribution of this population group to the Finnish demography and child well-being. There are also previous reports of lower child mortality among foreign language speakers from Canada (26) and the USA (27). The immigrant background in the majority of foreign language speakers is likely to explain the differences to the general population. In particular, culture-related health behaviors have been suggested as a protective factor in certain areas of health (27). However, foreign language speakers encompass a greatly diverse and heterogeneous group of individuals and families in terms of ethnic backgrounds and cultural traditions, which complicates drawing detailed conclusions from this population group. It may be that municipalities which offer a warm welcome for foreign language speakers and take their convenience into account in public services, also perform better in the prevention of child and adolescent mortality. In a wider sense, we speculate that the finding might reflect the “migrant drive” (28) or “immigrant paradox” (27) that incomers bring to their host society.

The main strength of this study was the nationwide coverage of 310 municipalities and an 8-year study period. The dataset was comprised of official national statistics collected by Statistics Finland from public authorities and registers. Data collections were performed systematically across the nation without regional discrepancies. As the data originate from a public data portal, they are openly available for confirmatory analyses and future research openings. Numerous sociodemographic variables were covered, and their intercorrelations were accounted for in the GEE analysis.

There were also limitations to this study. As we addressed the relationship between sociodemographic traits and child and adolescent mortality only at the municipality level, our results may not be extrapolated to the individual level. There may also be variance in the coverage and quality of register-based data. All-cause mortality was modeled instead of cause- or age-specific scrutiny due to the low number of outcome events. We fully acknowledge that factors such as age group and cause of death may moderate the relationship between sociodemographic variables and mortality, but these could not be addressed in our dataset due to significant floor effects (i.e., most municipalities having zero events, and only a few having more than one). Unfortunately, data on health indicators such as smoking, physical activity or obesity were not available to us. The observational nature of the study and time-series approach prevented conclusions regarding cause-effect-relationships. Although we were able to incorporate a wide range of sociodemographic traits in our multivariable analysis, we cannot fully rule out residual confounding. We hope that future studies will be able to address and compensate the limitations of our study.

In conclusion, this nationwide study of 310 Finnish municipalities aimed to identify sociodemographic indicators of child and adolescent mortality at the municipality level. An average annual decrease of 3% in child and adolescent deaths occurred during the study period. Higher population density was associated with higher mortality, and higher percentage of foreign language speakers with lower mortality. Densely populated areas should be the primary focus of efforts to reduce child and adolescent mortality. Future studies are welcomed to scrutinize the mediating pathways and individual-level factors behind the detected associations.

Publicly available datasets were analyzed in this study. This data can be found here: https://www.stat.fi/tup/tilastotietokannat/index_en.html (StatFin database, Statistics Finland).

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. Written informed consent from the participants' legal guardian/next of kin was not required to participate in this study in accordance with the national legislation and the institutional requirements.

PO and AS contributed to conception of the study. PO organized the database, performed the statistical analysis, and wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.