Table 4 Health equity impact assessment. SES: Socioeconomic status. FTT: Failure to thrive. HC: Head circumference.

Ahn ¹⁶

None reported.

Exclusion of “socially vulnerable” premature infants. Inclusion of infants born at an urban University hospital, and exclusion of infants transferred from medical centres. Characteristics of infants lost to follow-up at 6 months (if any) are not reported.

Context provided regarding lack of official pregnancy leave in Korea, sociocultural factors which place childcare burden on mothers, and the 2011 OECD report highlighting that most Korean women have low quality jobs with the greatest income inequality by sex in any OECD nation.

Impact (Ahn) ¹⁶

Neutral.

No explicit efforts have been specified to engage with communities regarding participation in this study, as the sample was recruited from inborn preterm infants at a University Hospital and underwent a 6-month follow-up.

Negative.

Those experiencing socioeconomic minoritisation, and those from rural areas with lower SES ³⁸would have been disproportionately excluded.

Positive.

The study allows interpretation of the adverse effect of maternal employment in the context of structural barriers placed on employed Korean mothers, rather than considering “employment” a proxy for socioeconomic security.

Bocca-Tjeertes ¹⁷

Not reported, despite the study cohort being a community-based sample.

Recruitment from community healthcare centres covering 25% of all children in the Netherlands at a routine visit attended by 97% of children. However, mothers in the questionnaire non-response group more likely to be of non-Dutch ethnicity and have a lower education level.

Authors mention that low maternal education may be associated with low maternal HC, thereby suggesting a genetic mechanism for HC growth restraint for.

Impact (Bocca-Tjeertes) ¹⁷

Likely negative. Building community partnerships for engagement may have increased the questionnaire response rate of socioeconomically minoritised people, including those of non-Dutch origin.

Negative.

Data was disproportionately missing from those of non-Dutch ethnicities, known to also be socioeconomically minoritised⁷⁸, and mothers with a lower education level. This may have underestimated the effect of socioeconomic minoritisation on growth restraint.

Negative.

The study subscribes to biological essentialism by not offering any non-biological explanations for the relationship between lower education level and HC growth restraint have been offered. Those experiencing socioeconomic and ethnic minoritisation have been suggested to have restricted access to specialised healthcare services in the Netherlands ⁷⁸.

Sammy ¹⁸

Not reported.

The sample was recruited at discharge from a District Hospital newborn unit serving a rural population in Kenya. 3 of 115 infants screened for inclusion were excluded due to congenital anomaly and lack of guardian consent. 4 of 112 infants were lost to follow-up, and there were 2 neonatal deaths.

Authors state that “education always empowers the mothers in all aspects and should therefore be emphasised”. Specific mechanisms by which educational empowerment may contribute to better health outcomes, and structural barriers to education for mothers are not discussed.

Impact (Sammy) ¹⁸

Neutral.

Building community partnerships for engagement may have facilitated attendance at follow-up for the 4 infants lost to follow-up, although these were a small proportion of the total sample (3.5%).

Positive.

This study focused on preterm infants in rural Kenya, where a lower proportion of households are in the wealthiest national quintile compared to urban areas⁴⁰. The exclusion criteria are not specifically associated with socioeconomic minoritisation. Although loss to follow-up ¹⁰⁴and neonatal death ¹⁰⁵are associated with socioeconomic minoritisation in Kenya, only 5% of infants were lost to follow-up in total.

Negative.

The study contributes to a deficit-based narrative of mothers with a low education level. In identifying the mechanisms by which health outcomes may be worsened for those experiencing minoritisation due to their education level, nor acknowledging structural barriers to education in rural Kenya, the study misses an opportunity to identify specific targets for intervention.

Teranishi ¹⁹

Not reported.

UK birth cohort of white, singleton children born in 1958 was studied.

Brief mention of social determinants of health, e.g. education, nutrition, healthcare, which may have contributed to greater catch-up growth among preterm infants in the upper social classes. Authors acknowledge that “improved social conditions during gestational, infant and childhood periods” at the end of the Discussion.

Impact (Teranishi ¹⁹)

Likely negative.

No efforts to engage with communities regarding research participation and attendance at follow-up time points are specified. Given the long follow-up duration of 23 years, lack of community engagement is likely to have resulted in non-random attrition among socioeconomically minoritised participants ³⁷.

Negative.

In addition to the likely exacerbated attrition of socioeconomically minoritised participants over the 23-year follow-up period, the exclusion of children from non-white race/ethnicities would have disproportionately excluded participants from disinvested communities. Poverty is known to be associated with race/ethnicity in the UK ¹⁰⁶.

Neutral.

The study identifies social determinants of health as being targets for intervention throughout the lifecourse. However, despite the fact that social class was studied as the exposure in this study, the authors document health inequity and do not expand on how “social conditions” could be improved.

Holmqvist ²⁰

Not reported.

Exclusion of infants with maternal and foetal complications. Follow-up was based at an urban University hospital.

Authors state there may be “possible socioeconomic influences” contributing to the favourable growth of infants of well educated mothers from birth to 7 months.

Impact (Holmqvist ²⁰)

Likely negative.

No efforts have been specified to engage with communities regarding participation in this study. Lack of community engagement may have affected differential attrition by SES over the 48-month follow-up³⁵, especially as the follow-up visits required in-person attendance

Negative.

Maternal and foetal complications are both associated with socioeconomic minoritisation in Sweden^103,107, resulting in disproportionate exclusion of socioeconomically minoritised participants. The study’s urban setting may have impacted the follow-up attendance of infants from rural communities.

Neutral.

Despite acknowledging “socioeconomic influences”, this study does not discuss the mechanistic pathways by which named influences may have affected the postnatal growth of preterm infants, thereby missing an opportunity to identify targets for intervention.

Ghods ²¹

Not reported.

Infants lost to follow-up before 66 months were excluded. Over 40% of infants were from immigrant families. Follow-up was based at a University hospital in Vienna.

No context provided for the criteria in the Home Facilities (light, ventilation, accommodation etc) and Finance (education level, occupation and social facilities) assessments to justify their use, nor for the scale used to determine parental SES (Good, Adequate, Inadequate). The language used in the interview is not specified. Mention previous literature which shows an association between “family socioeconomic situation” and HC growth. Breastfeeding is suggested as a potential mediator for the relationship between higher maternal education and catch-up growth, but breastfeeding was not associated with maternal education in this study.

Impact (Ghods ²¹)

Likely negative.

No efforts to engage community members in research participation have been discussed; such efforts may have reduced the 11% loss to follow-up.

Negative.

Exclusion of infants lost to follow-up would have disproportionately excluded socioeconomically minoritised people who are more impacted by loss to follow-up^35,36, and may have disadvantaged those from rural areas. The proportion of infants from immigrant families reflects the proportion of the population with a “migrant background” in Vienna in ¹⁰⁸.

Negative.

The use of a compound measure of parental SES with a non-validated scale may obscure the health equity impact of certain factors (e.g. occupation versus education versus “social facilities”).

Lack of use of appropriate interpretation services during interviews to determine parental SES will likely have resulted in disproportionate misclassification among those who do not speak the language used, given 40% of the sample were immigrants. Further consideration of the association between the Home Facilities and Financial Situation ratings and HC catch-up growth may have identified targets for intervention.

Ni ²²

Not reported.

Participants lost to follow-up before 19 years were excluded, and participants with data at 19 years were more likely to have mothers with higher education and parents with higher occupational SES. This study included children born across maternity units in the UK and Ireland (276 centres in total).

Authors posit that pre-pregnancy maternal weight may mediate the relationship between lower parental SES, birthweight and BMI at 19 years.

Impact (Ni ²²)

Likely negative.

The authors do not specify measures taken to engage community members in this research. These measures may have helped to address research mistrust and encouraged more minoritised participants invited to participate in the assessment at 19 year ¹⁰⁹.

Negative.

Participants were recruited from maternity units based in urban and rural areas. Loss to follow-up over the extended 19 year follow-up period disproportionately affected socioeconomically minoritised participants ^35,36.

Negative.

Although authors identify pre-pregnancy maternal weight as a potential mediator, they do not acknowledge the structural determinants of health which underlie the association between low parental SES and pre-pregnancy weight. An individualist explanation is offered without identifying wider determinants amenable to policy and/or social change.

Liang ²⁵

Not reported.

Only infants of parents with “basic reading and comprehension skills” were included. Infants with severe illness were excluded. Follow-up was based at a teaching hospital.

No context or explanation offered for the relationship between household income and growth outcomes at 12 months.

Impact (Liang ²⁵)

Likely negative.

Given that the authors aim to study the feasibility and impact of family integrated care, community partnerships may have facilitated better informed consent from parents, particularly those from minoritised communities, for study participation.

Negative.

The study excludes infants of parents who experience minoritisation due to their education level, for whom family integrated care may have been particularly beneficial due to the paucity of accessible information and training on caring for preterm infants. Excluded infants with severe illness are more likely to be from socioeconomically minoritised communities¹¹⁰. Those from rural communities, who are more likely to have a lower SES³⁸, may have faced additional barriers to follow-up attendance.

Negative.

Although this study was focused on family integrated care as the intervention, the authors miss an opportunity to discuss why monthly household income is significantly associated with all three measured growth outcomes (HC, weight and body length), and identify targets for intervention.

Fu ²⁴

Not reported.

Infants without data at the 4-7 year follow-up visit due to lack of attendance or loss to follow-up within the birth cohort were excluded.

No context offered regarding the findings on the lack of relationship between maternal occupation or education and overweight/obesity at 4-7 years.

Impact (Fu ²⁴)

Likely negative.

Building community partnerships may have decreased the number of children for whom “data essential for defining childhood overweight/obesity” were unavailable at 4-7 years in this birth cohort study (4823 of 8269 eligible infants, 58%).

Negative.

Socioeconomically minoritised infants would have been overrepresented among those who did not attend the 4-7 year visit or were lost to follow-up ^35,36,37.

Negative.

The authors adjust for maternal education and occupation when examining the association between modifiable feeding practices and BMI trajectories, suggesting they consider these to be important confounders of this association. However, they do not comment on why maternal education or occupation was not associated with overweight/obesity at 4-7 years – it may be that these variables were not granular enough or contextually appropriate to capture the impact of SES in this study, and consequently there may be residual confounding due to socioeconomic minoritisation in the analyses.

Sices ²⁶

Not reported.

Infants without at least two consecutive growth measurements were excluded. 18% of infants who were discharged from the NICU did not attend all 3 follow-up visits. Follow-up was based at an urban teaching hospital.

No context offered for the lack of association between maternal education and growth failure during the three study periods. Maternal education is presented as part of a composite “social risk” score along with “African=American/Black” ethnicity and “unmarried” maternal status.

Impact (Sices ²⁶)

Likely negative.

Community engagement in research participation may have facilitated greater attendance at follow-up visits up to 20 months’ corrected age.

Negative.

Loss to follow-up or lack of attendance would have occurred disproportionately among socioeconomically minoritised infants^35,36. Infants from rural areas³⁹, may have faced additional barriers to follow-up attendance.

Negative.

The authors do not explore mechanisms by which “social risk”, composed of non-modifiable factors, may contribute to postnatal growth failure, thereby failing to identify strategies for intervention. Having less than a high school education is positioned as being equivalently risky to “African-American/Black” ethnicity and “unmarried” status, ignoring the pathways of structural oppression and minoritisation associated with these individual factors, and the importance of intersection between these factors.

Peterson ²⁷

Not reported.

Infants with an unreliable or missing HC measurement at school age were excluded. Follow-up was based at an urban teaching hospital and two other tertiary centres.

No context regarding lack of association between maternal education and subnormal HC at school age. Maternal education less than high school, “non-white” ethnicity and “unmarried” marital status constitute a composite “sociodemographic risk” score.

Impact (Peterson ²⁷)

Likely negative.

Building relationships with communities may have facilitated greater participation from minoritised communities at the school-age visit.

Negative.

Socioeconomically minoritised infants would have been more likely to have not attended the school-age follow-up visit^35,36,37. Unreliable or missing HC measurements may have been more likely in infants from rural communities ³⁹.

Negative.

Similar to the study by Sices et al, the authors do not discuss how “sociodemographic risk” may result in subnormal postnatal HC growth. The equivalation of unmarried status, “non-white” ethnicity and less than high school education again misses the tangible mechanisms by which structural oppressions operate and intersect.

Kelleher ²³

Not reported.

Exclusion of infants lost to follow-up before 30 months, infants with mothers who could not adequately communicate in English, infants with mothers reporting drug or alcohol abuse or psychiatric hospitalisation, infants requiring intensive medical intervention or severe neurodevelopmental abnormalities.

No context provided to explain the bimodal distribution of maternal education between the FTT and non-FTT infants, nor for the finding that advanced maternal education was associated with the development of FTT. No explanation offered for the lack of association between family income and FTT.

Impact (Kelleher ²³)

Likely negative.

Building community relationships may have supported more minoritised participants to attend follow-up until 30 months,

Negative.

This disproportionately excludes socioeconomically minoritised participants ^33,34,35,36.

Negative.

The authors do not account for their significant selection bias in explaining the association of advanced maternal education and FTT. They also do not discuss factors which may have led to disparate findings for the two indicators of SES, namely family income and maternal education. There is no discussion of contextual factors which may lead to one measure of SES being more suitable than the other, nor the different mechanisms by which maternal education level and family income may act to influence postnatal growth.