You are here

Article Text

Article menu
Original research
Socioeconomic circumstances, health-related behaviours and paediatric infections: a mediation analysis
  1. Samuel Videholm1,2,
  2. Sven Arne Silfverdal1,
  3. Per E Gustafsson3
  1. 1 Department of Clinical Sciences, Pediatrics, Umeå University, Umea, Sweden
  2. 2 Unit of Research, Development, and Education, Östersund Hospital, Ostersund, Sweden
  3. 3 Department of Epidemiology and Global Health, Umeå University, Umea, Sweden
  1. Correspondence to Mr Samuel Videholm, Department of Clinical Sciences, Pediatrics, Umeå University, Umea, Sweden; samuel.videholm{at}umu.se

Abstract

Objective To examine how the effect of disadvantaged socioeconomic circumstances on the risk of paediatric infections is mediated by pregnancy smoking, excess weight during pregnancy and breast feeding in children under 5 years of age.

Design A population-level, retrospective cohort study. The Swedish Medical Birth Register was combined with the National Patient Register, the longitudinal integration database for health insurance and labour market studies, the Cause of Death Register and a local register held by the Child Health Care Unit in Uppsala Region.

Setting Uppsala Region, Sweden.

Patients 63 216 term and post-term singletons born to women who resided in Uppsala Region, Sweden between 1997 and 2015.

Main outcome measures Number of hospital admissions for infectious diseases. Secondary outcomes were the number of hospitalisations for respiratory and enteric infections, respectively.

Results The effect of disadvantaged socioeconomic circumstances, that is, low maternal education on the overall risk of paediatric infections was mediated to a considerable (33%–64%) proportion by pregnancy smoking, excess weight during pregnancy and breast feeding.

Conclusions Pregnancy smoking, excess weight during pregnancy and breast feeding mediate a considerable proportion of the association between deprived socioeconomic circumstances and paediatric infectious diseases. Therefore, inequalities in paediatric infectious diseases may be reduced by public health policies addressing these health-related behaviours.

  • infectious diseases
  • epidemiology
  • child health
  • healthcare disparities
http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/archdischild-2023-325850

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS KNOWN ABOUT THIS TOPIC

    • It is well known that infectious diseases disproportionately affect children living under deprived socioeconomic circumstances.

    • Less is known about how the effect of deprived socioeconomic circumstances on paediatric infectious diseases is mediated by health-related behaviours including pregnancy smoking, excess weight during pregnancy and low breast feeding.

    WHAT THIS STUDY ADDS

    • Pregnancy smoking, excess weight during pregnancy and low breast feeding mediate around half of the association between deprived socioeconomic circumstances and paediatric infectious diseases.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • Inequalities in paediatric infectious diseases may be reduced by public health policies targeting these health-related behaviours.

    Background

    Paediatric infections disproportionately affect children living under deprived socioeconomic circumstances. Population-based studies have shown substantial socioeconomic inequalities in common and severe paediatric infections.1–3 In particular, children born to parents with low socioeconomic position seem to be at a higher risk of gastroenteritis and respiratory tract infections.1 4 5

    Pregnancy smoking, excess weight during pregnancy and early cessation of exclusive breast feeding are health-related behaviours that may link deprived socioeconomic circumstances to paediatric infectious diseases; they are all associated with both deprived socioeconomic circumstances and paediatric infections.6–8 Both smoking and excess weight during pregnancy are associated with unfavourable perinatal conditions including low birth weight and congenital malformations.9 10 Low birth weight and congenital malformations are in turn associated with paediatric infections.11 12 It is well known that breast feeding reduces the risk of paediatric infections, as breast milk contains IgA antibodies, antibacterial proteins and growth factors that stimulate maturation of the infant’s immune system.13 Both duration and intensity of breast feeding are important, in particular, exclusive breast feeding <4 months seems to increase the risk of paediatric infections.8

    Previous studies have examined how the effect of disadvantaged socioeconomic circumstances on the risk of paediatric infections is mediated by health-related behaviours using traditional mediation analysis.14 15 In contrast, counterfactual mediation can decompose the mediation effect into different exposure and different susceptibility.16 Importantly, this may influence how preventive strategies to tackle health inequalities are designed and implemented.17

    Therefore, this study aimed to examine how the effect of disadvantaged socioeconomic circumstances on the risk of paediatric infections may be mediated by different exposure and different susceptibility to pregnancy smoking, excess weight during pregnancy and breast feeding in children under 5 years of age.

    Methods

    Study design and population

    This population-level, retrospective cohort study includes term and post-term singletons born to women who resided in Uppsala Region, Sweden between 1997 and 2015. We linked national and local health and administrative registers using the unique personal identification number assigned to every Swedish resident at birth or immigration. The Swedish Medical Birth Register contains information on prenatal, delivery and neonatal care, with almost complete coverage.18 The National Patient Register holds information on nearly all hospitalisations in Sweden including International Classification of Disease, Tenth Revision (ICD-10) codes for primary diagnosis.19 The longitudinal integrated database for health insurance and labour market studies (LISA), contains socioeconomic data including education.20 The Cause of Death Register holds information on all deaths in Sweden since 1952.21 The child health service database in Uppsala Region contains information on child health indicators including breast feeding, collected by child health nurses during routine health check-ups. The database holds information on virtually all children under 5 years of age living in Uppsala Region, including children living in deprived families.22 The data show high concordance with medical records.23 Registers were linked and subsequently de-identified by the Centre for Epidemiology at the Swedish National Board of Health and Welfare.

    Exposures

    In our study, socioeconomic circumstances were captured through maternal education level obtained from LISA. Maternal education was categorised as secondary school or less (≤9 years), upper secondary school (10–12 years), short postsecondary education (13–14 years) and long postsecondary education (≥15 years).

    Potential mediators and confounders

    Information on pregnancy smoking, pregnancy weight, parity, maternal age, maternal country of birth and year of birth was retrieved from the Medical Birth Register. Pregnancy smoking was reported at the first antenatal care visit, normally between 8 and 12 weeks of gestation, and categorised as yes/no.24 Parity was categorised as 1, 2, 3, 4 and ≥5. Pregnancy maternal body mass index (BMI) was calculated using weight and length obtained at the first antenatal care visit and categorised as excess weight if BMI ≥25 kg/m2. Maternal age was categorised as ≤25, 25–29, 30–34 and ≥35 years. Maternal country of birth was categorised as Sweden, other high-income countries, upper-middle-income countries, lower-middle-income countries and low-income countries using the World Bank Atlas method.25 Information on current breast feeding (exclusive, partial breast feeding or no breast feeding) was obtained from the child health service database in Uppsala Region. Breast feeding was categorised as low exclusive breast feeding (exclusive breast feeding <4 months) if partial breast feeding and/or no breast feeding was reported at 1 week, 2 months and/or 4 months. Data on disposable household income were obtained from LISA and divided into annual quintiles (Q1–Q5), with Q1 representing the highest household income quintile.

    Outcomes

    The main outcome was the overall number of hospitalisations with a principal diagnosis of infectious disease recorded in the National Patient Register. Secondary outcomes were the number of hospitalisations for respiratory and enteric infections, respectively. ICD-10 codes were identified using a previously developed coding scheme. This coding scheme was developed by a panel of infectious disease clinicians, microbiologists and disease coding experts.2 Online supplemental table A includes a list of ICD-10 codes.

    Statistical analysis

    We examined associations between maternal education, pregnancy smoking, excess weight during pregnancy and low exclusive breast feeding and number of hospitalisations for paediatric infections using Poisson regression models. Two models were fitted to each outcome (overall, respiratory and enteric infections). In the first model, analyses were adjusted for maternal age, maternal country of birth and year of birth; in the second model, analyses were additionally adjusted for maternal education, pregnancy smoking, excess weight during pregnancy and low exclusive breast feeding. Log follow-up time was used as offset. Results were presented as incidence rate ratios (IRR) with 95% CIs.

    In counterfactual mediation analysis, we quantify the effects of maternal education level on paediatric infections acting through health-related behaviours. For each outcome (overall, respiratory and enteric infections), we fitted a natural effect model for the joint mediating effect of pregnancy smoking, excess weight during pregnancy and low exclusive breast feeding. All analyses were adjusted for the potential confounding effect of maternal age, maternal country of birth and year of birth (figure 1). All models used log follow-up time as offset. The natural effect models used robust SEs. In our analysis, the total effect (TE) of maternal education level on childhood infections was decomposed into three components, the pure natural direct effect (PNDE), the pure natural indirect effect (PNIE) and the mediated interactive effect (MIE). The PNDE describes the effect of maternal education on childhood infections that is neither mediation nor interaction. The PNIE describes the effect of maternal education on childhood infections that is due to just mediation, not interaction (differential exposure). The MIE describes the effect of maternal education on childhood infections that is due to an additive interaction between maternal education level and mediators (differential susceptibility).26

    Figure 1
    Figure 1

    Directed acyclic graph representing the assumed causal structure. E: exposure (maternal education), Y: outcomes (paediatric infections), M: mediators (pregnancy smoking, excess weight during pregnancy and low exclusive breast feeding), C: confounders (maternal age, maternal country of birth and year of birth), L: exposure-induced mediator-outcome confounders (parity and poverty), D: unmeasured confounders (maternal childhood circumstances and genetic factors) and H: unmeasured exposure-induced mediator-outcome confounders (paternal smoking, unhealthy diet, maternal employment, paternal education and paternal employment).

    Mediation analysis was conducted in two steps. In the first step, we imputed counterfactual outcomes for each potential level of maternal education. The dataset was expanded so it contained four observations per child, one observation for each potential level of maternal education. Mean values for the counterfactual outcomes were then imputed using a Poisson regression model that included maternal education, mediators, confounders and an interaction term between maternal education and mediators. In the second step, we estimated the PNDE, PNIE, TE and MIE. The PNDE, PNIE and a multiplicative interaction term were estimated by fitting a natural effect Poisson regression model to the expanded dataset. The TE was estimated as the linear combinations of PNDE, PNIE and the multiplicative interaction term. The MIE was estimated as the relative excess risk due to interaction (RERI)=IRRTE−IRRPNDE−IRRPINE+1. In models with a TE of IRR >1.05, we calculated the overall percentage mediated using the following formula: ((IRRPNIE−1)/(IRRTE−1)+RERIMIE/(IRRTE−1))×100.26 In sensitivity analysis, we examined if the potential confounding effect of parity by restricting the analysis to the firstborn child and the potential confounding effect of poverty by excluding children born to mothers in the two lowest household income quintiles. Missing data were imputed using multiple imputation. We used a chained-equation approach. The predictive model included all variables including outcomes. Ten imputed datasets were generated. Percentages and medians were calculated as averages over all imputations; regression results were pooled across all imputed datasets using Rubin’s rules.

    Data preparation was performed in Stata Statistical Software: Release 17 (College Station, Texas, USA). Analyses were conducted using R V.4.2.2, the medflex package (V.0.6–7) was used for mediation analysis, the mice package (V.3.15.0) was used for multiple imputation and daggity (V.0.3–1) was used to illustrate the assumed relationships between variables.27–29

    Results

    Study population

    The Medical Birth Register included 63 216 term and post-term (gestational age ≥37 weeks) singletons born alive to mothers who resided in Uppsala Region between 1 January 1997 and 31 December 2015. Children were followed until 5 years of age, death (n=70), emigration (n=1059) or 31 December 2016. The total follow-up was 283 119 person-years. The study included 6977 infectious disease hospitalisations, 3184 admissions for respiratory infections and 1455 admissions for enteric infections. Data were missing on one or more variables for 26% (n=16 219) of children.

    Table 1 presents the background characteristics of the study population. In comparison with children born to mothers with long postsecondary education, children born to mothers with shorter education were more likely to be exposed to pregnancy smoking, low exclusive breast feeding and, to a lesser extent, excess weight during pregnancy. Additionally, the mother was more likely to be foreign-born.

    View this table:
    Table 1

    Study population characteristics and their association with maternal education level

    Regression analyses

    The associations between maternal education and mediators, and overall, respiratory and enteric infections hospitalisations are shown in table 2. Compared with children born to mothers with ≥15 years of education, children born to mothers with shorter education were more likely to be hospitalised for overall infections and respiratory infections; children born to mothers with ≤9 years of education were also more likely to be hospitalised for enteric infections. The association between maternal education and all outcomes decreased considerably when mediators were added to the models. Pregnancy smoking and low exclusive breast feeding were associated with all outcomes. Excess weight during pregnancy was associated with overall and respiratory infections.

    View this table:
    Table 2

    Associations between maternal education, pregnancy smoking, pregnancy overweight, low exclusive breast feeding and hospitalisations for all respiratory and enteric infections

    Mediation analyses

    The total and decomposed effects on paediatric infections due to counterfactual changes in maternal education level are presented in table 3. For overall infections, mediation by health-related behaviours accounted for a large proportion of the TE (total percentage mediated 33%–64%, depending on exposure level). For respiratory and enteric infections, health-related behaviours mediated a substantial proportion of the additional risk observed in the lowest education group (total percentage mediated 57% and 40%, respectively). The mediation effects were due to higher exposure to health-related behaviours and we found no evidence of increased susceptibility. In sensitivity analysis, we obtained similar results in an analysis restricted to the firstborn child and in an analysis excluding children born to mothers in the two lowest household income quintiles. This indicates no major confounding effect due to parity or poverty. Online supplemental table B includes results from sensitivity analysis.

    View this table:
    Table 3

    Total and decomposed effects on all respiratory and enteric paediatric infections due to counterfactual changes in maternal education level

    Discussion

    In this large population-based study, we show that pregnancy smoking, excess weight during pregnancy and breast feeding mediate around half of the association between maternal education level and paediatric infections.

    Mediation analyses

    We hypothesised that health-related behaviours mediated the relationship between deprived socioeconomic circumstances and paediatric infections, based on previous research; pregnancy smoking, excess weight during pregnancy and low exclusive breast feeding are all associated with paediatric infections and more frequent in deprived groups.6–8 Similar to our results, a Swedish cohort study reported that pregnancy smoking and family size were important mediators for the effect of social deprivation on respiratory infections in young children.14 Additionally, an Australian cohort study found that pregnancy smoking and family size mediated the effect of maternal education on a combined measurement of child health that included infectious diseases.15 In comparison to these studies, we used counterfactual mediation analysis which allowed us to decompose the mediating effect into different exposure and different susceptibility.

    Public health policy implications

    Inequalities in child health may be reduced by policies that (1) target social stratification, (2) decrease exposure to health-related factors, (3) decreases vulnerability to health-related factors and (4) reduces social consequences of ill-health.30 In our study, a substantial proportion of the association between maternal education and paediatric infections was mediated by increased exposure to health-related behaviours; we found no evidence of increased susceptibility. Consequently, our findings support strategies that reduce smoking, overweight and obesity and promote breast feeding. Smoking can be reduced by different tobacco control policies, for example, smoke-free legislation, tobacco taxation and smoking cessation programmes; importantly, some evidence suggests that tobacco taxation and smoke-free legislation can reduce absolute socioeconomic inequalities.31 Overweight and obesity can be addressed by food policies including taxation of unhealthy foods, for example, soft drinks, subsidies of healthy foods and enforcement of understandable nutrition labelling.32 Currently, Sweden has no specific tax on unhealthy foods. Breast feeding can be promoted by breastfeeding support offered by professionals and lay people together.33 These policies should be applied universally to ensure that health inequalities are reduced across the entire social gradient.30

    Strengths and limitations

    Our study used high-quality registers with individual-level data and counterfactual mediation analysis. However, a large proportion (26%) of children had missing data on one or more variables. Assuming that data were missing at random, we used multiple imputations to impute missing data thereby reducing the risk of selection bias.34 Maternal education may induce other factors that in turn may confound the association between health-related behaviours and paediatric infections. In sensitivity analysis, we found no confounding effect of poverty or parity. However, other socioeconomic factors, for example, maternal employment and paternal education may still be confounders. The measured health-related behaviours may be associated with other unmeasured behaviours, for example, pregnancy smoking may be associated with smoking after pregnancy as well as paternal smoking. Therefore, our findings support broad public health policies reaching the whole population over specific interventions targeting, for example, pregnant women. We were unable to adjust for maternal childhood circumstances and genetic factors. Even though we do not expect these factors to be major confounders, they may affect our estimates. Finally, our results may not be generalisable to other settings with different rates of excess weight during pregnancy, pregnancy smoking and breast feeding, at least not without considering the distribution of health-related behaviours by socioeconomic circumstances in these settings.

    Conclusions

    We found that pregnancy smoking, excess weight during pregnancy and low breast feeding mediate around half of the association between maternal education level and paediatric infectious diseases. This mediation was due to increased exposure to health-related behaviours in children born to low-educated mothers, we found no evidence of increased susceptibility. Consequently, our results suggest that inequalities in paediatric infectious diseases can be targeted by public health policies addressing exposure to these health-related behaviours.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study was approved by the Regional Ethics Board in Umeå (reference numbers 2012- 265-31M and 2017-399-32M) and by the Swedish Ethical Review Authority (reference number 2021-06337-02).

    Acknowledgments

    The authors would like to thank Håkan Sjöberg, MEc at Statistics Sweden, for help in compiling the dataset and Thomas Wallby, PhD Uppsala University, for providing data from the Child Health Care Unit in Uppsala.

    References

    1. ↵
      2. Biering-Sørensen S ,
      3. Søndergaard G ,
      4. Vitting Andersen K , et al
      . Time trends in socio-economic factors and risk of Hospitalisation with infectious diseases in pre-school children 1985-2004: a Danish register-based study. Paediatr Perinat Epidemiol 2012;26:226–35. doi:10.1111/j.1365-3016.2011.01255.x
      OpenUrlCrossRefPubMed
    2. ↵
      2. Baker MG ,
      3. Barnard LT ,
      4. Kvalsvig A , et al
      . Increasing incidence of serious infectious diseases and inequalities in New Zealand: a national Epidemiological study. Lancet 2012;379:1112–9. doi:10.1016/S0140-6736(11)61780-7
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      2. Mitchell HK ,
      3. Reddy A ,
      4. Montoya-Williams D , et al
      . Hospital outcomes for children with severe sepsis in the USA by race or Ethnicity and insurance status: a population-based, retrospective cohort study. Lancet Child Adolesc Health 2021;5:103–12. doi:10.1016/S2352-4642(20)30341-2
      OpenUrl
    4. ↵
      2. Carville KS ,
      3. Lehmann D ,
      4. Hall G , et al
      . Infection is the major component of the disease burden in aboriginal and non-aboriginal Australian children: a population-based study. Pediatr Infect Dis J 2007;26:210–6. doi:10.1097/01.inf.0000254148.09831.7f
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      2. Lind J ,
      3. Schollin Ask L ,
      4. Juarez S , et al
      . Hospital care for viral gastroenteritis in Socio‐Economic and geographical context in Sweden 2006–2013. Acta Paediatr 2018;107:2011–8. doi:10.1111/apa.14429
      OpenUrlPubMed
    6. ↵
      2. Nafstad P ,
      3. Jaakkola JJ ,
      4. Hagen JA , et al
      . Maternal smoking and lower respiratory tract infections. Eur Respir J 1996;9:2623–9. doi:10.1183/09031936.96.09122623
      OpenUrlAbstract/FREE Full Text
    7. ↵
      2. Videholm S ,
      3. Silfverdal SA ,
      4. Reniers G
      . Maternal weight and infections in early childhood: a cohort study. Arch Dis Child 2019;104:58–63. doi:10.1136/archdischild-2017-314628
      OpenUrlAbstract/FREE Full Text
    8. ↵
      2. Videholm S ,
      3. Wallby T ,
      4. Silfverdal SA
      . Breastfeeding practice, Breastfeeding policy and Hospitalisations for infectious diseases in early and later childhood: a register-based study in Uppsala County, Sweden. BMJ Open 2021;11:e046583. doi:10.1136/bmjopen-2020-046583
    9. ↵
      2. Avşar TS ,
      3. McLeod H ,
      4. Jackson L
      . Health outcomes of smoking during pregnancy and the postpartum period: an umbrella review. BMC Pregnancy Childbirth 2021;21:254.:254. doi:10.1186/s12884-021-03729-1
      OpenUrlCrossRefPubMed
    10. ↵
      2. Ruager-Martin R ,
      3. Hyde MJ ,
      4. Modi N
      . Maternal obesity and infant outcomes. Early Hum Dev 2010;86:715–22. doi:10.1016/j.earlhumdev.2010.08.007
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      2. Hviid A ,
      3. Melbye M
      . The impact of birth weight on infectious disease hospitalization in childhood. Am J Epidemiol 2007;165:756–61. doi:10.1093/aje/kwk064
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      2. Zachariah P ,
      3. Ruttenber M ,
      4. Simões EAF
      . Hospitalizations due to respiratory syncytial virus in children with congenital malformations. Pediatr Infect Dis J 2011;30:442–5. doi:10.1097/INF.0b013e318201813b
      OpenUrlCrossRefPubMed
    13. ↵
      2. Hanson LA ,
      3. Silfverdal SA
      . The mother’s immune system is a balanced threat to the Foetus, turning to protection of the neonate. Acta Paediatr 2009;98:221–8. doi:10.1111/j.1651-2227.2008.01143.x
      OpenUrlPubMed
    14. ↵
      2. Hjern A ,
      3. Haglund B ,
      4. Bremberg S
      . Lower respiratory tract infections in an ethnic and social context. Paediatr Perinat Epidemiol 2000;14:53–60. doi:10.1046/j.1365-3016.2000.00231.x
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      2. Davey TM ,
      3. Cameron CM ,
      4. Ng S-K , et al
      . The relationship between maternal education and child health outcomes in urban Australian children in the first 12 months of life. Matern Child Health J 2015;19:2501–11. doi:10.1007/s10995-015-1771-5
      OpenUrl
    16. ↵
      2. VanderWeele TJ
      . Mediation analysis: a practitioner’s guide. Annu Rev Public Health 2016;37:17–32. doi:10.1146/annurev-publhealth-032315-021402
      OpenUrlCrossRefPubMed
    17. ↵
      2. Diderichsen F ,
      3. Hallqvist J ,
      4. Whitehead M
      . Differential vulnerability and susceptibility: how to make use of recent development in our understanding of mediation and interaction to tackle health inequalities. Int J Epidemiol 2019;48:268–74. doi:10.1093/ije/dyy167
      OpenUrlCrossRefPubMed
    18. ↵
      1. Socialstyrelsen
      . The Swedish medical birth register - A summary of content and quality; 2003.
    19. ↵
      2. Ludvigsson JF ,
      3. Andersson E ,
      4. Ekbom A , et al
      . External review and validation of the Swedish National inpatient register. BMC Public Health 2011;11:450. doi:10.1186/1471-2458-11-450
    20. ↵
      2. Ludvigsson JF ,
      3. Svedberg P ,
      4. Olén O , et al
      . The longitudinal integrated database for health insurance and labour market studies (LISA) and its use in medical research. Eur J Epidemiol 2019;34:423–37. doi:10.1007/s10654-019-00511-8
      OpenUrlCrossRefPubMed
    21. ↵
      2. Brooke HL ,
      3. Talbäck M ,
      4. Hörnblad J , et al
      . The Swedish cause of death register. Eur J Epidemiol 2017;32:765–73. doi:10.1007/s10654-017-0316-1
      OpenUrlCrossRefPubMed
    22. ↵
      2. Wallby T ,
      3. Hjern A
      . Child health care uptake among Low‐Income and immigrant families in a Swedish County. Acta Paediatr 2011;100:1495–503. doi:10.1111/j.1651-2227.2011.02344.x
      OpenUrlCrossRefPubMed
    23. ↵
      2. Wallby T
      . Lika för alla?: Social position och etnicitet som determinanter för amning, föräldrars rökvanor och kontakter med BVC. Acta Universitatis Upsaliensis, 2012.
    24. ↵
      2. Mattsson K ,
      3. Källén K ,
      4. Rignell-Hydbom A , et al
      . Cotinine validation of self-reported smoking during pregnancy in the Swedish medical birth register. Nicotine Tob Res 2016;18:79–83.:tv087. doi:10.1093/ntr/ntv087
      OpenUrlCrossRefPubMed
    25. ↵
      2. Fantom NJ ,
      3. Serajuddin U
      . The World Bank’s classification of countries by income. In: World Bank Policy Research Working Paper. 2016. doi:10.1596/1813-9450-7528
    26. ↵
      2. VanderWeele TJ
      . A three-way decomposition of a total effect into direct, indirect, and interactive effects. Epidemiology 2013;24:224–32. doi:10.1097/EDE.0b013e318281a64e
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      2. Steen J ,
      3. Loeys T ,
      4. Moerkerke B , et al
      . Medflex: an R package for flexible mediation analysis using natural effect models. J Stat Softw 2017;76:46. doi:10.18637/jss.v076.i11
    28. ↵
      2. Van S ,
      3. mice G-OK
      . Multivariate imputation by chained equations in R. J Stat Softw 2011;45:1–67. doi:10.18637/jss.v045.i03
      OpenUrlCrossRef
    29. ↵
      2. Textor J ,
      3. van der Zander B ,
      4. Gilthorpe MS , et al
      . Robust causal inference using directed Acyclic graphs: the R package ‘Dagitty Int J Epidemiol 2016;45:1887–94. doi:10.1093/ije/dyw341
      OpenUrlCrossRefPubMed
    30. ↵
      2. Pearce A ,
      3. Dundas R ,
      4. Whitehead M , et al
      . Pathways to inequalities in child health. Arch Dis Child 2019;104:998–1003. doi:10.1136/archdischild-2018-314808
      OpenUrlAbstract/FREE Full Text
    31. ↵
      2. Faber T ,
      3. Kumar A ,
      4. Mackenbach JP , et al
      . Effect of tobacco control policies on perinatal and child health: a systematic review and meta-analysis. Lancet Public Health 2017;2:e420–37. doi:10.1016/S2468-2667(17)30144-5
      OpenUrl
    32. ↵
      2. Hawkes C ,
      3. Smith TG ,
      4. Jewell J , et al
      . Smart food policies for obesity prevention. The Lancet 2015;385:2410–21. doi:10.1016/S0140-6736(14)61745-1
      OpenUrl
    33. ↵
      1. Cochrane Pregnancy and Childbirth Group,
      2. Gavine A ,
      3. Shinwell SC , et al
      . Support for healthy Breastfeeding mothers with healthy term babies. Cochrane Database Syst Rev 1996;2022. doi:10.1002/14651858.CD001141.pub6
    34. ↵
      2. Sterne JAC ,
      3. White IR ,
      4. Carlin JB , et al
      . Multiple imputation for missing data in Epidemiological and clinical research: potential and pitfalls. BMJ 2009;338:b2393. doi:10.1136/bmj.b2393

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors SV conceptualised and designed the study together with PEG and SAS, carried out analyses and drafted the initial manuscript. SAS conceptualised and designed the study together with SV and PEG, coordinated and supervised the database and reviewed and revised the manuscript. PEG designed the study together with SV and SAS, and reviewed and revised the manuscript. SV is guarantor.

    • Funding This study was supported by grants from a regional agreement between Umeå University and the Västerbotten County Council (ALF) RV-933162 (to SAS) and by the Unit of Research, Development and Education, Östersund Hospital, Östersund, Sweden JLL-930202 (to SV).

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.