Introduction
It is widely accepted that exposure to maternal smoking during pregnancy (MSDP) may have deleterious effects on health outcomes in children including stillbirth (Flenady et al. Reference Flenady, Koopmans, Middleton, Froen, Smith, Gibbons, Coory, Gordon, Ellwood, McIntyre, Fretts and Ezzati2011), lowered birth weight (Jaddoe et al. Reference Jaddoe, Troe, Hofman, Mackenbach, Moll, Steegers and Witteman2008), obesity (Gorog et al. Reference Gorog, Pattenden, Antova, Niciu, Rudnai, Scholtens, Splichalova, Slotova, Voko, Zlotkowska and Houthuijs2011), and externalizing disorders (Thapar et al. Reference Thapar, Fowler, Rice, Scourfield, van den Bree, Thomas, Harold and Hay2003; Obel et al. Reference Obel, Linnet, Henriksen, Rodriguez, Jarvelin, Kotimaa, Moilanen, Ebeling, Bilenberg, Taanila, Ye and Olsen2009). Animal and human studies have suggested that MSDP can disrupt neurodevelopment via effects on maturing neurotransmitter systems and brain architecture in regions associated with stress and mood regulation. Despite these findings the debate continues with regard to whether these associations represent causal relationships (Langley et al. Reference Langley, Heron, Smith and Thapar2012; D'Onofrio et al. Reference D'Onofrio, Lahey, Turkheimer and Lichtenstein2013; Skoglund et al. Reference Skoglund, Chen, D'Onofrio, Lichtenstein and Larsson2014). MSDP is known to be associated with numerous social and environmental factors (e.g. teenage motherhood, lower maternal education, increased single motherhood) that influence childhood outcomes (Gilman et al. Reference Gilman, Gardener and Buka2008; Ellingson et al. Reference Ellingson, Rickert, Lichtenstein, Langstrom and D'Onofrio2012). In addition genes associated with the likelihood of MSDP may also affect childhood outcomes through maternal–child genetic inheritance (Agrawal et al. Reference Agrawal, Knopik, Pergadia, Waldron, Bucholz, Martin, Heath and Madden2008; Chang et al. Reference Chang, Lichtenstein and Larsson2012). For this reason studies utilizing sibling-control and quasi-experimental designs (such as children from in vitro fertilization) (Agerbo et al. Reference Agerbo, Mortensen and Munk-Olsen2013) have been undertaken in an attempt to control for unmeasured genetic and environmental confounding (Knopik, Reference Knopik2009; D'Onofrio et al. Reference D'Onofrio, Lahey, Turkheimer and Lichtenstein2013). These studies have generally demonstrated attenuation of previously observed associations (Obel et al. Reference Obel, Olsen, Henriksen, Rodriguez, Jarvelin, Moilanen, Parner, Linnet, Taanila, Ebeling, Heiervang and Gissler2011; Skoglund et al. Reference Skoglund, Chen, D'Onofrio, Lichtenstein and Larsson2014).
In contrast to the effort expended in exploring the MSDP-externalizing behaviour association, relatively few studies have explored the association between MSDP and internalizing disorders. Studies that have investigated this association report both positive and null results (Ashford et al. Reference Ashford, van Lier, Timmermans, Cuijpers and Koot2008; Carter et al. Reference Carter, Paterson, Gao and Iusitini2008; Robinson et al. Reference Robinson, McLean, Oddy, Mattes, Bulsara, Li, Zubrick, Stanley and Newnham2010; Moylan et al. Reference Moylan, Gustavson, Overland, Karevold, Jacka, Pasco and Berk2015). Limitations of these studies such as small sample size, limited controls for potential confounders, and differential reporting of smoking, among other issues, may have contributed to these inconsistencies. Internalizing disorders such as depression and anxiety contribute significantly to the global burden of disease (Murray et al. Reference Murray, Vos, Lozano, Naghavi, Flaxman, Michaud, Ezzati, Shibuya, Salomon, Abdalla, Aboyans, Abraham, Ackerman, Aggarwal, Ahn, Ali, Alvarado, Anderson, Anderson, Andrews, Atkinson, Baddour, Bahalim, Barker-Collo, Barrero, Bartels, Basanez, Baxter, Bell, Benjamin, Bennett, Bernabe, Bhalla, Bhandari, Bikbov, Bin Abdulhak, Birbeck, Black, Blencowe, Blore, Blyth, Bolliger, Bonaventure, Boufous, Bourne, Boussinesq, Braithwaite, Brayne, Bridgett, Brooker, Brooks, Brugha, Bryan-Hancock, Bucello, Buchbinder, Buckle, Budke, Burch, Burney, Burstein, Calabria, Campbell, Canter, Carabin, Carapetis, Carmona, Cella, Charlson, Chen, Cheng, Chou, Chugh, Coffeng, Colan, Colquhoun, Colson, Condon, Connor, Cooper, Corriere, Cortinovis, de Vaccaro, Couser, Cowie, Criqui, Cross, Dabhadkar, Dahiya, Dahodwala, Damsere-Derry, Danaei, Davis, De Leo, Degenhardt, Dellavalle, Delossantos, Denenberg, Derrett, Des Jarlais, Dharmaratne, Dherani, Diaz-Torne, Dolk, Dorsey, Driscoll, Duber, Ebel, Edmond, Elbaz, Ali, Erskine, Erwin, Espindola, Ewoigbokhan, Farzadfar, Feigin, Felson, Ferrari, Ferri, Fevre, Finucane, Flaxman, Flood, Foreman, Forouzanfar, Fowkes, Fransen, Freeman, Gabbe, Gabriel, Gakidou, Ganatra, Garcia, Gaspari, Gillum, Gmel, Gonzalez-Medina, Gosselin, Grainger, Grant, Groeger, Guillemin, Gunnell, Gupta, Haagsma, Hagan, Halasa, Hall, Haring, Haro, Harrison, Havmoeller, Hay, Higashi, Hill, Hoen, Hoffman, Hotez, Hoy, Huang, Ibeanusi, Jacobsen, James, Jarvis, Jasrasaria, Jayaraman, Johns, Jonas, Karthikeyan, Kassebaum, Kawakami, Keren, Khoo, King, Knowlton, Kobusingye, Koranteng, Krishnamurthi, Laden, Lalloo, Laslett, Lathlean, Leasher, Lee, Leigh, Levinson, Lim, Limb, Lin, Lipnick, Lipshultz, Liu, Loane, Ohno, Lyons, Mabweijano, MacIntyre, Malekzadeh, Mallinger, Manivannan, Marcenes, March, Margolis, Marks, Marks, Matsumori, Matzopoulos, Mayosi, McAnulty, McDermott, McGill, McGrath, Medina-Mora, Meltzer, Mensah, Merriman, Meyer, Miglioli, Miller, Miller, Mitchell, Mock, Mocumbi, Moffitt, Mokdad, Monasta, Montico, Moradi-Lakeh, Moran, Morawska, Mori, Murdoch, Mwaniki, Naidoo, Nair, Naldi, Narayan, Nelson, Nelson, Nevitt, Newton, Nolte, Norman, Norman, O'Donnell, O'Hanlon, Olives, Omer, Ortblad, Osborne, Ozgediz, Page, Pahari, Pandian, Rivero, Patten, Pearce, Padilla, Perez-Ruiz, Perico, Pesudovs, Phillips, Phillips, Pierce, Pion, Polanczyk, Polinder, Pope, Popova, Porrini, Pourmalek, Prince, Pullan, Ramaiah, Ranganathan, Razavi, Regan, Rehm, Rein, Remuzzi, Richardson, Rivara, Roberts, Robinson, De Leon, Ronfani, Room, Rosenfeld, Rushton, Sacco, Saha, Sampson, Sanchez-Riera, Sanman, Schwebel, Scott, Segui-Gomez, Shahraz, Shepard, Shin, Shivakoti, Singh, Singh, Singh, Singleton, Sleet, Sliwa, Smith, Smith, Stapelberg, Steer, Steiner, Stolk, Stovner, Sudfeld, Syed, Tamburlini, Tavakkoli, Taylor, Taylor, Taylor, Thomas, Thomson, Thurston, Tleyjeh, Tonelli, Towbin, Truelsen, Tsilimbaris, Ubeda, Undurraga, van der Werf, van Os, Vavilala, Venketasubramanian, Wang, Wang, Watt, Weatherall, Weinstock, Weintraub, Weisskopf, Weissman, White, Whiteford, Wiebe, Wiersma, Wilkinson, Williams, Williams, Witt, Wolfe, Woolf, Wulf, Yeh, Zaidi, Zheng, Zonies, Lopez, AlMazroa and Memish2012). MSDP is potentially preventable in contrast to many potential risk factors associated with childhood outcomes. Based on this background a greater understanding of the MSDP-internalizing behavior association is highly relevant for public health.
Utilizing the rich Danish population-based registers, we aim to examine the putative effect of MSDP on the risk of severe depression and anxiety disorders at the population level and within and between families. Specifically, we investigate, whether offspring exposed to MSDP have a higher risk of developing depression or anxiety disorders than offspring not exposed to MSDP, and whether familial factors account for this potential link. In pursuing this, we established a nationwide population-based cohort of prospectively collected data on prenatal maternal smoking within and between families and individual onset of depression and anxiety disorders.
Method
Data sources
We utilized data from a record linkage of six Danish population-based registries: the Danish Civil Registration System (Pedersen, Reference Pedersen2011), the Danish Psychiatric Central Register (Mors et al. Reference Mors, Perto and Mortensen2011), the Danish National Hospital Registry (Lynge et al. Reference Lynge, Sandegaard and Rebolj2011), the Danish Medical Birth Register (Knudsen & Olsen, Reference Knudsen and Olsen1998), the Danish Education Registers (Jensen & Rasmussen, Reference Jensen and Rasmussen2011), and the Registers on Personal Income and Transfer Payments (Baadsgaard & Quitzau, Reference Baadsgaard and Quitzau2011).
All residents of Denmark including immigrants have a unique personal identification number that is used in all national registers, which enables data to be linked across registers at an individual level. The Danish Civil Registration System was computerized in 1968 and gathers information on gender, date of birth, and vital status (continuously updated) of all persons, who have lived in Denmark since 1968 (Pedersen, Reference Pedersen2011). The Danish Psychiatric Central Register includes data on all people admitted to a psychiatric hospital for assessment, treatment, or both in Denmark from 1969 onwards, or people who had appointments with psychiatric outpatient services from 1995 onwards (Mors et al. Reference Mors, Perto and Mortensen2011). In the Danish National Hospital Registry all inpatient treatments at non-psychiatric facilities have been recorded from 1977 onwards, whereas outpatient and emergency-room contacts have been recorded from 1995 onwards (Andersen et al. Reference Andersen, Madsen, Jorgensen, Mellemkjoer and Olsen1999). Diagnoses are based on the International Classification of Diseases – eighth (ICD-8) and tenth (ICD-10) revisions. The Danish Medical Birth Registry was established in 1968 and was computerized in 1973, it provides data on antenatal and delivery care services and health of newborns (Knudsen & Olsen, Reference Knudsen and Olsen1998). Common to Danish Education Registers is individual-level information, which links education and educational institutions of students enrolled in Denmark, but the oldest information goes back to a full population census in 1970. For each year individual-level information on enrolment status, and completed levels of education and examinations is available (Jensen & Rasmussen, Reference Jensen and Rasmussen2011). From 1980 onwards the Income Statistics Register includes information on salaries, entrepreneurial income, taxes, public transfer payments, capital income, private pension contributions, and pay-outs (Baadsgaard & Quitzau, Reference Baadsgaard and Quitzau2011).
Study population
We identified all persons born in Denmark between 1991 and 2007 (N = 1 185 152) with complete linkage available for both parents. After the exclusion of persons with missing values on MSDP (N = 118 023), multiple births (N = 40 223), death, emigration, or diagnosis of depression or an anxiety disorder before 5 years of age or before 1996 (N = 18 625), and those with serious congenital malformations (N = 50 646) the study population included 957 635 persons, covering 770 315 siblings nested within 331 396 families (see Fig. 1). In this study all the individuals were followed up from their 5th birthday until the diagnosis of interest (depression or anxiety disorders), or until censoring due to death, emigration or end of study (31 December 2012), whichever occurred first. The study was approved by the Danish Data Protection Agency. The investigators were blind to the identity of study members. According to Danish legislation informed consent was not required.
Sample selection.
Measures
Outcomes
Using the Danish Psychiatric Central Register and the Danish National Patient Register, we identified all persons diagnosed with depression (ICD-10 codes: F32.00-F33.99 F34.10-F34.90 F38.00-F39.99), or anxiety disorders (ICD-10 codes: F40.00-F40.20 F41.00-F41.10 F42.00-F43.10); including acute stress reaction, agoraphobia, generalized anxiety disorder, obsessive-compulsive disorder, panic disorder, post-traumatic stress disorder, specific phobia, and social phobia. For each individual the date of first psychiatric contact leading to the diagnosis of interest was designated as the date of onset.
Exposures
MSDP reported at the first antenatal visit was derived from the Danish Medical Birth Register from 1991 onwards (more detailed from 1997 onwards). All women were asked by the midwife at their first antenatal visit (13–15 weeks of gestation), whether they had ever smoked during the present pregnancy. For analyses the following variables for maternal smoking were constructed: (i) never smoker (women who at the visit to the midwife stated that they had never smoked during the present pregnancy) and (ii) ever smoker (women who at the first visit to the midwife stated that they had stopped smoking during the first trimester, stopped smoking at the beginning of the second trimester, or were current smokers at the first visit to the midwife) (Knudsen & Olsen, Reference Knudsen and Olsen1998).
Confounding and mediating factors
Based on previous research measured covariates included sex, calendar year, mother's parity (1st, 2nd, 3rd, 4th or ⩾5th), parental age at childbirth (⩽20, 21–25, 26–30, 31–35, >35 years), parental psychiatric history (yes/no), substance abuse (ICD-8 codes: 291.xx, 303.xx, 304.xx, 571.09, 571.1x; ICD-10 codes: F10-F16, F18, F19, I85, K70), divorce, abuse (ICD-8 codes: E960-E969; ICD-10 codes: T74.xx, X85.00-Y09.99), parental highest education at time of birth (categorized as unknown, elementary school, above elementary school), and parental income at time of birth (annual gross income in tertiles). Maternal somatic illness was assessed using the Charlson Comorbidity Index (Charlson et al. Reference Charlson, Pompei, Ales and Mackenzie1987). The Charlson Index is an indicator of the somatic disease burden based on 19 severe chronic diseases, each assigned a weight from 1 to 6 corresponding to the severity of the disease.
As low birth weight, early gestational age, and a low Apgar score 5 min after birth might mediate the association of MSDP with depression or anxiety disorders, we chose to not adjust for these measures. However, we conducted a sensitivity analysis with the cohort restricted to individuals with gestational age 37–44 weeks, birth weight >2500 g, and Apgar score of 10 at 5 min.
Statistical analyses
We used Cox proportional survival analysis to estimate the effect of MSDP on the risk of depression or anxiety disorders at the population level. The models calculated hazard rate ratios (HRRs) for time to depression or anxiety diagnosis using age as underlying time-scale. Robust standard errors adjusted the 95% confidence intervals (CIs) for the presence of familial clustering in the analyses at the population level. We further adjusted the crude model for the above-mentioned measured covariates. Analyses were conducted at the population level and the cohort restricted to at least one maternal sibling. In the sibling cohort, we followed the suggestions of Begg & Parides (Reference Begg and Parides2003) in order to disentangle familial- and individual-level effects of MSDP. We assessed the effect of the mother smoking during one specific pregnancy (individual-level effect of MSDP) and adjusted for how often the same mother was smoking during all her pregnancies (familial mean exposure to MSDP). We therefore added to the model the familial mean exposure to MSDP and a ‘centred’ form of individual MSDP (MSDP
ij
–
$\overline {\rm MSDP} _i $
), where the i and j indexes represent the families and the individuals, respectively, and where
$\overline {\rm MSDP} _i $
is the average over individuals in the ith family. Since the individual measurement is replaced by its deviation from the familial level mean, this new version of the individual score represents, how much larger or smaller the individual measurement is compared to other individuals in its family. We also tested the deviation of individual MSDP from familial mean exposure as the sole predictor of depression and anxiety disorders.
We supplemented the analyses exploring the effect of unmeasured familial confounding. We hereby applied stratified Cox regression models with a separate stratum for each set of maternal siblings. In the sibling sample, there were 82 041 siblings discordantly exposed to MSDP, nested in 34 984 nuclear families. Sibling comparisons adjust for all unmeasured factors that are shared and constant within the nuclear family. The stratified Cox regression models using sibling data were adjusted for the same covariates as in the models on the population level. All statistical analyses were conducted in SAS software v. 9.4 (SAS Institute Inc., USA).
Ethical statement
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Results
A total of 957 635 persons born between 1991 and 2007 were followed from their 5th year birthday for the development of severe depression and anxiety disorders. During the 49 148 258 person-years at risk, 6525 persons were diagnosed with depression and 6739 with anxiety disorders. In 13 484 (1.4%) cohort members follow-up was ended before the end of the study, 927 died, 12 354 emigrated from Denmark, and 203 were lost to follow-up. Table 1 shows the distribution of offspring and maternal covariates. In families with MSDP, parents were more likely to be mentally ill (p < 0.0001), of younger age (p < 0.0001), lower education (p < 0.0001), and lower income (p < 0.0001).
Baseline characteristics of individuals exposed to maternal smoking during pregnancy
MSDP and offspring's risk for internalizing disorders
The crude association showed that offspring exposed to MSDP were at increased risk for both severe depression (HRR 1.39) and anxiety disorders (HRR 1.36). This association was marginally attenuated after adjustment for measured covariates (depression: HRR 1.29; anxiety disorders: HRR 1.26; see Table 2). Of the covariates included parental psychopathology particularly had an effect on risk estimates. Mental illness of the mother increased the risk for depression (HRR 1.66, 95% CI 1.55–1.77) and anxiety disorders (HRR 1.52, 95% CI 1.39–1.66) in offspring more than paternal mental illness (depression: HRR 1.29, 95% CI 1.20–1.39; anxiety disorders: HRR 1.18, 95% CI 1.07–1.30). In the cohort restricted to individuals having maternal siblings, the association of MSDP and risk for depression (HRR 1.34) and anxiety disorders (HRR 1.28) were comparable to the entire cohort. Sensitivity analyses restricting the cohort to individuals with gestational age 37–44 weeks, birth weight >2500 g, and Apgar score of 10 at 5 min resulted in very similar associations (depression: HRR 1.34, 95% CI 1.25–1.44; anxiety disorders: HRR 1.27, 95% CI 1.18–1.36).
Hazard rate ratios of internalizing disorders exposed to maternal smoking during pregnancy (MSDP)
a Hazard rate ratio adjusted for calendar year of birth and gender.
b Hazard rate ratio adjusted for calendar year of birth, gender, parity, parental age at time of birth, parental income, parental education, and parental psychiatric history.
Familial- and individual-level effects of MSDP and offspring's risk for internalizing disorders
After adjustment for the familial mean exposure to MSDP in the sibling cohort no differences in the risk of severe depression (HRR 1.11) or anxiety disorders (HRR 0.94) were observed for individual MSDP, whereas we observed strong familial-level effects of MSDP (depression: HRR 1.39; anxiety disorders: HRR 1.37; see Table 3). A total of 82 041 siblings were discordantly exposed to MSDP nested in 34 984 nuclear families (18 554 mothers smoked during the first pregnancy compared to 13 576 in the second, 8192 in the third). Stratified sibling comparisons showed similarly that associations observed at the population level were completely attenuated (depression: HRR 1.18; anxiety disorders: HRR 0.87). Exposed and unexposed siblings had nearly equivalent rates of depression and anxiety disorders, indicating that unmeasured familial factors, that are constant within nuclear families, explain the associations between MSDP and the later risk of depression and anxiety.
Individual and familial level effects of maternal smoking during pregnancy (MSDP)
a Hazard rate ratio of familial and individual MSDP adjusted for calendar year of birth, gender, parity, parental age at time of birth, parental income, parental education, and parental psychiatric history.
b Hazard rate ratio of the deviation of individual MSDP from familial mean exposure adjusted for calendar year of birth, gender, parity, parental age at time of birth, parental income, parental education, and parental psychiatric history.
c Hazard rate ratio of MSDP derived from stratified Cox regression models with a separate stratum for each set of maternal siblings adjusted for calendar year of birth, age, gender, parity, parental age at time of birth, parental income, parental education, and parental psychiatric history.
Discussion
This large prospective study of offspring born in Denmark explored the risk associated with MSDP for depression and anxiety disorders. Consistent with previous research offspring exposed to MSDP were more often diagnosed with depression or anxiety disorders at the population-level (Ashford et al. Reference Ashford, van Lier, Timmermans, Cuijpers and Koot2008; Carter et al. Reference Carter, Paterson, Gao and Iusitini2008; Brion et al. Reference Brion, Victora, Matijasevich, Horta, Anselmi, Steer, Menezes, Lawlor and Davey Smith2010). The association between MSDP and offspring internalizing disorders was generally robust to the use of measured statistical covariates such as parental education or income. However, after accounting for unknown but shared family-level factors, there remained no individual-level effect of MSDP on the offspring's risk of internalizing disorders. As such these results strongly suggest that unmeasured genetic factors or shared familial environment are likely to account for the increased risk of severe internalizing disorders among offspring exposed to MSDP, and not putative biological effects of MSDP. These conclusions are strengthened by the observation that siblings within the same family, who were differentially exposed to MSDP, did not differ in their risk of developing a severe internalizing disorder.
The results are further consistent with sibling-control studies exploring the contribution of unmeasured genetic and environmental confounds of MSDP and other offspring's outcomes such as poor academic achievement (Lambe et al. Reference Lambe, Hultman, Torrang, Maccabe and Cnattingius2006; D'Onofrio et al. Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman, Neiderhiser, Langstrom and Lichtenstein2010b ), low intellectual abilities (Lundberg et al. Reference Lundberg, Cnattingius, D'Onofrio, Altman, Lambe, Hultman and Iliadou2010), criminality (D'Onofrio et al. Reference D'Onofrio, Singh, Iliadou, Lambe, Hultman, Grann, Neiderhiser, Langstrom and Lichtenstein2010a ), and attention deficit hyperactivity disorder (Knopik et al. Reference Knopik, Sparrow, Madden, Bucholz, Hudziak, Reich, Slutske, Grant, McLaughlin, Todorov, Todd and Heath2005; Skoglund et al. Reference Skoglund, Chen, D'Onofrio, Lichtenstein and Larsson2014; Obel et al. Reference Obel, Zhu, Olsen, Breining, Li, Gronborg, Gissler and Rutter2015). One possible mechanism is that mothers transmit liability genes to offspring that influence behaviours in both generations (Kuja-Halkola et al. Reference Kuja-Halkola, D'Onofrio, Larsson and Lichtenstein2014).
Our study builds significantly on previous studies based observational data that have demonstrated inconsistent results (Ashford et al. Reference Ashford, van Lier, Timmermans, Cuijpers and Koot2008; Carter et al. Reference Carter, Paterson, Gao and Iusitini2008; Brion et al. Reference Brion, Victora, Matijasevich, Horta, Anselmi, Steer, Menezes, Lawlor and Davey Smith2010). For example the RAINE study including 2758 mother–child pairs reported that children displayed higher internalizing behaviours between ages 2 and 14 years, if their mother failed to quit smoking, even after controlling for a range of potential confounders (Robinson et al. Reference Robinson, McLean, Oddy, Mattes, Bulsara, Li, Zubrick, Stanley and Newnham2010). In the Norwegian Mother and Child Cohort Study, MSDP was similarly associated with increased internalizing behaviours at 18 and 36 months even after controlling for smoking in past pregnancies (Moylan et al. Reference Moylan, Gustavson, Overland, Karevold, Jacka, Pasco and Berk2015). These results contrast with outcomes from two other large cohorts, in which adjustment for confounders eliminated associations with MSDP. In the Generation R study (N = 4680) effects of MSDP on childhood behavioural problems at 18 months were strongly confounded by parental characteristics chiefly socioeconomic status and parental psychopathology (Roza et al. Reference Roza, Verhulst, Jaddoe, Steegers, Mackenbach, Hofman and Tiemeier2009). In the Avon Longitudinal study (N = 4394) MSDP was not associated with increased internalizing behaviours in children aged 4 years, after controlling for a range of potential confounders including socioeconomic status, parental psychopathology and alcohol consumption (Brion et al. Reference Brion, Victora, Matijasevich, Horta, Anselmi, Steer, Menezes, Lawlor and Davey Smith2010).
In our study a robust statistical association between MSDP and internalizing behaviours was found while controlling for confounding factors. However, the results of the sibling analyses revealed strong familial confounding indicating that familial factors not frequently measured (or measured well) in research protocols are actually responsible for the increased risk in offspring whose mothers smoke during pregnancy. In our study we were unable to capture these confounding effects by adjusting for parental psychology. The inconsistencies observed across studies might be explained by differences in sensitivity of the confounders included in these studies to account for such familial factors.
In contrast to most previous research, our study covers a longer follow-up period of 21 years and internalizing disorders can be identified at any time point during this period. Internalizing disorders were defined as diagnoses made at inpatient and outpatient facilities, for which the offspring also received treatment. In contrast to the parents’ reports of children's problem behaviour employed in other studies, this constitutes a more severe outcome definition. Further, we accounted for range of parental somatic and mental disorders, but we were unable to explore the effects of lifestyle, such as breastfeeding, diet, physical activity or alcohol consumption.
Our results should be interpreted in the context of some limitations. Although sibling comparison will not be confounded by factors shared by siblings, the estimates might be more sensitive to bias due to non-shared confounders than the unpaired estimates (Frisell et al. Reference Frisell, Oberg, Kuja-Halkola and Sjolander2012). The strict control for shared family factors further limits the analyses to a quite small subset of the population, namely those women who managed to change smoking habits from one pregnancy to another, whose change in smoking is assumed to be independent of their offspring's traits. Despite the obvious limitation in extrapolation to all smokers and especially heavy smokers it may, however, from a public health point of view be the most interesting group to focus on. In addition using the method of Begg & Parides (Reference Begg and Parides2003), which enabled us to include the entire sibling cohort, also revealed that familial effects account for the effect of MSDP on internalizing behaviour in offspring. Finally, it should be noted that our sample is rather young, which could imply that MSDP may have an effect on late onset of internalizing disorders.
The study made use of register-based diagnoses of severe depression or anxiety. As patient registers record contacts with clinics and psychiatric outpatient services, but not contacts with general practitioners, offspring with transitional or mild symptoms of depression and anxiety may have been missed. Thus our strategies probably could not avoid producing false negatives, while we consider bias due to false positives more unlikely. Primary findings were further in line with other representative studies in terms of prevalence of depression and anxiety disorders. Moreover, the use of a nationwide cohort minimized the risk for selection bias and allowed this longitudinal follow-up with minimal attrition. As in all observational studies we were not able to rule out residual confounding due to the lack of intact information on the exposure variable and other potential confounders. As pregnant women may also conceal their smoking habits there is a possibility of misclassification of exposure (Lindqvist et al. Reference Lindqvist, Lendahls, Tollbom, Aberg and Hakansson2002). However, previous studies have repeatedly shown support for a causal association between MSDP and low birth weight suggesting that the effect of exposure misclassification is small in magnitude (Cnattingius, Reference Cnattingius2004; Obel et al. Reference Obel, Zhu, Olsen, Breining, Li, Gronborg, Gissler and Rutter2015). Further information on smoking quantity and timing (in the first trimester, throughout pregnancy) was only available in the minority of our study members, which prevented us from studying these factors. Finally, information on smoking habits of fathers and other family members would have been desirable. Interestingly, we have recently shown strong consistency between the risk estimates obtained using self-reports of MSDP, as in this study, or biomarkers such as maternal cotinine levels (Reference Meier, Mors and ParnerMeier et al. in press).
The conclusions drawn from this study will need to be replicated in other studies, including more precise measures of MSDP and make use of other designs (Dolan et al. Reference Dolan, Geels, Vink, van Beijsterveldt, Neale, Bartels and Boomsma2016) to further rule out alternative processes. In summary our data suggest that the previously observed association between MSDP and internalizing disorder can be attributed to unmeasured familial confounding. Although MSDP is known to be harmful in many ways (e.g. low birth weight, and infant mortality) and pregnant women should still be encouraged to stop smoking, our study does not support MSDP as an independent risk factor for internalizing disorders. It is essential for clinicians, researchers, and policy makers to focus on true and amendable causal risk factors and MSDP is most probably not one of those.
Acknowledgements
This work was supported by the Lundbeck Foundation within the context of the Lundbeck Foundation Initiative for Integrative Psychiatric Research and Mental Health in Primary Care. Dr Meier received further funding from the Mental Health Services Capital Region Copenhagen Denmark and Dr Mortensen received funding from the Stanley Medical Research Institute. The funders had no role in the design and conduct of the study; collection management analysis or interpretation of the data; preparation review or approval of the manuscript; and decision to submit the manuscript for publication.
Declaration of Interest
None.
