This study was conducted in Denmark with data collected yearly in the period 2002–2022. In Denmark, every family with a newborn child is offered a series of free home visits from a community health nurse. A community health nurse is a nurse with a 1.5-year further education in child health and development. The community health nurses visit approximately 95% of all families with a newborn in Denmark [25]. The Danish Health Authority recommends five home visits during the first year. At these visits, the community health nurses systematically register data on child health and development including weight and length.

This study is based on data from the Child Health Database, which is a collaboration between community health nurses from municipalities in Denmark. The collaboration was initiated by community health nurses in 2002 where 16 municipalities participated and since 2015 33 of the 98 municipalities in Denmark have been a part of the database. Eight municipalities have participated with data on length and weight throughout the entire study period (2002–2022). These eight municipalities are mainly located in suburban Copenhagen and comprise citizens from diverse socioeconomic backgrounds. A Danish report from 2020 shows that the prevalence of infant overweight and obesity in these eight municipalities is similar to the national average [26]. In the Child Health Database collaboration, data collected by community health nurses were registered based on the application of a common manual, securing comparability, and increasing validity of the records. Two record systems, Solteq Sund and Novax, were used to collect the data. All Danish citizens receive a unique personal identification number at birth, enabling exact individual-level linkage between national registers in Denmark as well as linking the infant’s identification number with its parents’ identification numbers. The two record systems, Solteq Sund and Novax, included the infant’s unique personal identification number, by which it was possible to link the data to Danish registers. Data was linked with national health and sociodemographic registers with complete data from the entire population.

The study design involved continuous and yearly cross-sectional data collections among infants from birth to 10 months. Infants enrolled in the study were born between 1st January 2002 and 31st December 2022 and included all infants from the eight municipalities with data on weight and length in the Child Health Database collaboration. The specific study population for this study was infants aged 6–10 months. The age range of 6–10 months was chosen because, at this stage infants in Denmark are no longer fully breast-fed or fully bottle-fed and begin to become more active. A total of 77,245 infants from the eight municipalities were registered, of which 63,100 (81.7%) had complete data on weight and length. Missing data analyses showed that infants with missing data on weight and length were more likely to not live with both parents, have mothers under the age of 30, have unemployed parents, and have parents with non-Danish origin (data not shown). Despite being statistically significant, the differences were relatively minor and are therefore not expected to substantially impact the main findings.

Different indicators of SEP, such as neighborhood deprivation, parental income, educational attainment of the parents, and parental employment, have been used for studying social inequalities in childhood overweight and obesity. Parental income and/or education are recommended [27], and in Denmark, these factors are closely related. In this study, we applied parental education because we study trends over time and education tends to change more slowly compared to income, which can fluctuate more in a short period. Furthermore, education may influence cognitive functioning, making a person more receptive to health education messages, and better equipped to access and communicate with health services [28]. Parental education was measured by parents’ highest ongoing or completed educational attainment at the infant’s birth, and this information was retrieved from the Population’s Educational Register from Statistics Denmark [29]. The highest educational attainment of the parents was categorized into five hierarchical levels from the highest level of education (5 or more completed years at university) to the lowest level (primary school, 9 years at school). Infants were categorized according to the highest educational level of their parents. If the education level of only one parent was available, it was used to categorize the infant.

Measurements of weight and recumbent length between 6–10 months were assessed by the community health nurse. The infant was weighed without diaper and only light clothes using a handheld hanging scale. Recumbent length was measured on either a measuring board or with measuring tape. The Danish Health Authority recommends the weight to be rounded to one decimal place and the recumbent length to be measured to nearest half centimeter [30]. The outcome estimate of BMI status was based on weight and length measured at age 6–10 months combined with information on infant sex and age. A recent article recommends using BMI z-scores from birth and forward [31]. BMI z-scores were calculated using sex- and age-specific standard values from the WHO [32]. Infants at age 6–10 months were classified as overweight, when BMI z-score was between +2 SD and +3 SD and obese, when BMI z-score was greater than +3 SD.

Demographic information on infant sex, maternal and paternal age at the time of childbirth (<20, 20–29, 30–39, and ≥40), parental origin (two, one, or no parents of Danish origin), parental occupational attachment (two, one, or no parents in work or education), as well as family type (infant lives with both parents, yes vs. no) were included to describe the characteristics of the study population. Information on these variables were sourced from national population registers: the Danish Civil Registration System [33] and the Employment Classification Module [34].

The statistical analyses were conducted in SAS version 9.4. In all analyses, level of significance was set to p < 0.05. The study period 2002–2022 was divided into seven time periods, each consisting of 3 years. Cochran-Armitage test for trend was used to assess trends in prevalence of overweight and obesity from 2002 to 2022. Prevalence differences within each of the seven time periods were tested with chi-square test. Both relative and absolute social inequality are reported [24]. To estimate the relative social inequality for each educational level within the seven time periods, we calculated odds ratio (OR) for overweight and obesity with simple logistic regression analysis with highest level of education as reference group. To account for different distribution of parental educational attainment within and across the time periods, both Slope Index of Inequality (SII) and Relative Index of Inequality (RII) were calculated. SII predicts the difference between the top and bottom of the social hierarchy, (e.g., those with the highest and lowest level of education), of a health indicator, whereas RII predicts the ratio [23, 35]. Both measures are obtained by regression-based weighted analysis. For SII, the dependent variable (the outcome variable: “overweight/obesity at 6–10 months” weighted by the population in each group) was regressed against the midpoint scores of the five educational levels within each of the seven time periods. The RII was estimated similarly, with the exception that the dependent variable in the regression model was logit-transformed, followed by taking the exponential function of the parameter estimate [23, 36]. SII positive values indicate that infant overweight/obesity is more prevalent among the more educated groups, negative values indicate that infant overweight/obesity is more prevalent among the less educated groups, and zero indicates no inequality [23, 36]. Values of the RII greater than one indicate a higher prevalence in the most educated groups and values less than one illustrate a higher prevalence in the less educated groups [23]. Lastly, we conducted trend analyses to test for change in social inequality over time. Trend test reports whether the changes in SII and RII over time are stable or there has been a significant change. Trend test was performed by a simple linear regression of SII and RII.

The data collection was approved by the Research and Innovation Organization at the University of Southern Denmark (registration number 11,667) and the Danish Regional Council (registration number R-22030405) and complies with national regulations of data protection and consent. Data was passed on from the municipalities to the National Institute of Public Health as described in the Data Protection Legislation [37] and comply with the Danish codex for integrity in research [38].

Table 1 provides an overview of the distribution of sociodemographic characteristics of the study population. A total of 63,100 infants born between 2002 and 2022 were enrolled in the study. The total study population was characterized by 62% of the infants’ parents having a higher education or ≥5 years of university education, 82% of the parents were in job or pursuing education, and 64% of the infants had two parents of Danish origin. About 55% of the infants were born by parents being 30–39 years old at the time of childbirth and 90% lived with both parents. During the study period (2002–2022), a clear shift in the distribution of parental educational attainment occurred with an increasing proportion of parents having a ≥5 years university education. Minor variations in the distribution of the other sociodemographic variables were observed across the study period.

Table 2 shows the prevalence of infants with overweight or obesity at age 6–10 months across levels of parental educational attainment by time period, as well as OR (95% CI) for overweight and obesity across levels of parental educational attainment. Table 3 shows RII and SII for overweight and obesity by time period and trend test of RII and SII.

The present study aimed to investigate trends in social inequality in infant overweight and obesity between 2002 and 2022. The study revealed no absolute or relative social inequality between 2002 and 2004. From 2005 to 2019, we found social inequality in overweight and obesity with increasing odds for overweight and obesity among infants of parents with lower educational attainment. However, changes occurred in 2020–2022 where increased odds for overweight and obesity were only observed for infants of parents with vocational education and higher education. Across the entire study period (2002–2022), we found an increased absolute social inequality (measured by SII) in overweight and obesity among infants aged 6–10 months, whereas no upward or downward trend was observed for the relative social inequality (measured by RII) in overweight and obesity. To our knowledge, no previous studies have investigated trends in social inequality in infant overweight and obesity. Previous studies have investigated trends in social inequality in childhood overweight and obesity (aged >2 years) showing a persistent social inequality in overweight and obesity over time [21, 22]. Further, some studies have found an increasing prevalence or less improvements in prevalence of overweight and obesity among children with low SEP, whereas levelling off or improved overweight and obesity prevalence are found in children with high SEP [12, 39].

The present study found an overall increasing prevalence of infants with overweight and obesity from 2002 to 2022. These findings are not in line with existing literature showing a stagnating prevalence in overweight and obesity among infants in the US and Denmark [5, 6]. This discrepancy might be caused by different time periods studied. The present study covered the period from 2002 to 2022, whereas the two other studies followed infants until 2011 and 2012, respectively [5, 6].

It is important to highlight that the most recent time period (2020–2022) in this study was influenced by the COVID-19 lockdowns. During this period, the prevalence of overweight and obesity decreased for some educational levels, while it increased for others. The reasons for these changes are difficult to interpret. Visits from community health nurses may have been canceled, which potentially have affected the observed changes in prevalence of overweight and obesity. However, other underlying factors may also have an impact as fluctuating prevalence of overweight and obesity within each educational level was observed in other time periods, such as among infants of parents with primary school between time period 2008–2010 and 2011–2013. However, it is relevant to closely examine how the COVID-19 lockdowns may have affected the prevalence of overweight and obesity, and to identify the factors contributing to variations in prevalence across different levels of educational attainment. Supplementary analyses excluding the most recent time period including the COVID-19 lockdowns (2020–2022) were performed. These analyses showed a significant decrease in both RII and SII from 2002 to 2019, suggesting an increase in relative and absolute social inequality in overweight and obesity among infants from 2002 to 2019. These findings emphasize the need for further research on how the social inequality in overweight and obesity among infants will develop in the coming years. Furthermore, given that social inequality in child health emerges through a wide range of complex pathways [14], it would be interesting to explore how other socioeconomic factors, such as parental origin or parental occupational status, influence the development of overweight and obesity.

The RII and SII led to different conclusions which may be explained by the sensitivities of the two measures. During the study period, the overall prevalence of overweight and obesity increased and there was a shift in the distribution of parental educational attainment, with an increasing proportion of parents having a ≥5 years university education. Also, minor variations occurred in the distribution of the other sociodemographic variables. For example, a small increase in infants with no parents of Danish origin was observed. The SII is sensitive to changes in population distribution among different socioeconomic categories as well as to changes in the frequency of the health problem being studied [36]. For example, if the overall health level improves proportionately in all the socioeconomic categories, the SII will increase, while relative differences will remain constant [36]. The RII is sensitive to changes in educational attainment levels. It may show larger value in one population compared to another (e.g., across different time periods) due to a larger inequality in the distribution of individuals across the socioeconomic categories [36]. Also, the RII may decrease when the health problem being studied improves due to an upward shift in the educational levels [40]. Renard et al. [40] cautions against using RII and SII for assessing inequality when socioeconomic structures change suggesting instead showing the absolute changes within the socioeconomic categories. We therefore chose to present both the absolute prevalence of overweight and obesity for each educational attainment level as well as OR for each educational level along with RII and SII.

A large number of interventions aiming to prevent overweight and obesity in children have been developed. However, only a small number of studies have focused on weight development in infancy [41], and few have shown to be efficient in reducing social inequalities in early childhood overweight/obesity [42]. The lack of effective interventions may be caused by a limited understanding of the complexity of factors contributing to social inequality in overweight and obesity. It is proposed that children and adolescents develop obesity through interactions of genetic, behavioural, psychosocial, and environmental factors, interactions which are all influenced by SEP [43, 44]. In a new model for obesity development, Hemmingsson and colleagues (2023) explores the complex interplay of various factors contributing to obesity. For example, the model emphasizes that prolonged social adversity within the family (e.g., parents having a low education, low social position, poverty or uncertain accommodation) increases the risk of parental chronic stress and psychological strain which in turn contribute to an insecure and disharmonious family environment. According to the model, failures to meet the child’s needs may lead to disrupted energy balance homeostasis, resulting in weight gain and obesity [43], a situation that might be more prevalent in low SEP families. The higher absolute number of infants with overweight or obesity among parents with high levels of educational attainment compared to infants of parents with lower educational attainment underline the need for universal prevention rather than solely relying on interventions targeted at high-risk groups. A recent Danish review of the international literature concludes that more comprehensive and multicomponent interventions are needed to address the challenge of childhood overweight and obesity [41, 45]. Many of the previous interventions have followed a one-size-fits all approach, but it is likely that different target groups (e.g., fathers/partners, families of low SEP, and ethnic minorities) need different modes of delivery [46].

The present study population included almost complete populations of infants in eight Danish municipalities affiliated to the Child Health Database collaboration with data on weight and length during the entire study period (2002–2022). The exposure variable combined the level of education of both parents, categorizing the infant by the parent with the highest educational level. We chose this approach because, in most families in Denmark, both parents are highly involved in raising children, and the duration of paternity leave has increased in recent years. While women tend to have slightly higher levels of education [47], men still earn more than women in comparable similar job positions [48]. By linking community health nurses’ records with national population registers this study had high level of data coverage and is thus less likely subject to selection bias. Given the extensive experience and training of the community health nurses, we expect their measurements of weight and recumbent length to be less prone to measurements errors, compared to those reported by parents. We chose to use BMI z-scores instead of weight for length z-scores. In a recent publication, Roberge and colleagues found no substantial differences between weight for length z-scores and BMI z-scores in predictive efficacy of adiposity and cardiometabolic measures [31]. Thus, they advocate for the convenience of using the BMI z-scores from birth eliminating the need to switch between methods when reaching the age of 2 years where BMI z-scores are the recommended standard [31]. Further, in contrast to weight for length, BMI reflects age-dependent variations in weight and length during infancy and can be adjusted for gestational age [31].

The missing analyses suggest that the analyses might have underestimated the prevalence of overweight and obesity and have weakened the correlation between exposure and outcome. This inference arises from the observation that the infants with incomplete data were characterized by for example, being more likely to not live with both parents and having unemployed parents and these factors might serve as risk factors for childhood overweight [49]. The study population covers only eight out of the 98 municipalities in Denmark, with most of these eight municipalities located in suburban Copenhagen and the generalizability of the study may therefore be limited. However, a Danish report from 2020 observed that the prevalence of infant overweight and obesity in the eight municipalities, which were included in the present study, was similar with the national prevalence [26].

Within current efforts towards reducing the risk of developing childhood overweight or obesity, there is generally a concurrent and strong focus on reducing social inequalities [41]. Continued monitoring of changes in both prevalence levels and social inequalities in infant and child overweight and obesity are therefore essential. From these indications of the efficiency of especially population level actions and changes, initiatives and interventions can be drawn. It is therefore essential that future overweight and obesity prevention is accompanied with prioritized monitoring initiatives.

From 2002 to 2022, the prevalence of infants with overweight or obesity at age 6–10 months increased. No relative social inequality was observed in 2002–2004, but from 2005 to 2019 relative social inequality in overweight and obesity with increasing odds for overweight and obesity among infants of parents with lower educational attainment were found. The relative index of inequality (RII) suggested a persistent relative social inequality in overweight and obesity among infants aged 6–10 months from 2002 to 2022, however, the slope index of inequality (SII) indicated an increase in absolute social inequality in overweight and obesity from 2002 to 2022.