Objective: The association of perinatal psychological adversity (ie, stressors and distress) with infant lung function (ILF) and development is not well studied in Africa and elsewhere. We determined the association between maternal perinatal psychological adversity and ILF in African infants. Design: Prospective longitudinal follow up of the Drakenstein Child Health Study birth cohort. Participants: Seven hundred and sixty-two infants aged 6 to 10 weeks and 485 infants who had data for both maternal perinatal psychological adversity and ILF (measured at 6 to 10 weeks and 12 months). Methods: The main analyses were based on cross-sectional measures of ILF at each assessment (6 to 10 weeks or 12 months), using generalized linear models, and then on the panel-data of both longitudinal ILF assessments, using generalised estimating equations, that allowed specification of the within-group correlation structure. Results: Prenatal intimate partner violence (IPV) exposure was associated with reduced respiratory resistance at 6 to 10 weeks (beta coefficient [β] = −.131, P =.023); postnatal IPV with reduced ratio of time to peak tidal expiratory flow over total expiratory time (tPTEF/tE) at 12 months (β = −.206, P =.016); and prenatal depression with lower respiratory rate at 6 to 10 weeks (β = −.044, P =.032) and at 12 months (β = −.053, P =.021). Longitudinal analysis found an association of prenatal IPV with reduced tPTEF/tE (β = −.052, P <.0001); postnatal IPV with decreased functional residual capacity (FRC; β = −.086, P <.0001); prenatal posttraumatic stress disorder with increased FRC (β =.017, P <.0001); prenatal depression with increased FRC (β =.026, P <.0001) and postnatal depression with increased FRC (β =.021, P <.0001). Conclusion: Screening for psychological adversity and understanding the mechanisms involved may help identify children at risk of altered lung development and inform approaches to treatment.
Mother‐infant dyads were enrolled into the Drakenstein Child Health Study, to investigate the epidemiology, aetiology, and determinants of child health with longitudinal follow‐up of children until they are at least 5‐years of age.13, 14 The study was conducted in a periurban, low socioeconomic community, about 60 km from Cape Town in SA. Pregnant mothers were enrolled in their second trimester through the local primary health care clinics. Psychological adversity was assessed before birth and soon after birth,13 environmental exposures, and clinical measures were obtained before birth,14 and ILF was measured at 6 to 10 weeks and 1 year of age.6, 9 Prenatal and postnatal psychological adversity included in this analysis were: (a) IPV, (b) posttraumatic stress disorder (PTSD), and (c) depressive symptoms. Prenatal assessments for psychological adversity were done before birth while postnatal assessments were evaluated either at 6 to 10 weeks after birth. IPV was assessed with an IPV tool, which was piloted in the local South African population and found to be reliable when used in this population.13, 15 The tool comprises three sub‐scales as well as an overall IPV score based on frequency of abuse: (emotional abuse [four questions], physical abuse [four questions], and sexual abuse [three questions]), both of which are scored on a Likert scale of 1‐4, representing the frequency an abuse happened (1 = never, 2 = once, 3 = few, and 4 = many). Lifetime or past 12 months exposure for each subscale is generated. Emotional abuse was assessed by asking the mother about being insulted or made to feel bad, being belittled or humiliated in public, and being purposefully scared or intimidated. Physical abuse included history of being beaten (eg, by fist), pushed by force, injured (eg, burns), or threatened by weapons. Sexual abuse was assessed by asking about history of forced sexual intercourse when unwilling or forced into sexual activity that was humiliating. The responses for each question and scale were summed up to generate a continuous total score based on the frequency of abuse. PTSD was assessed with the Modified PTSD Symptom Scale (MPSS) which has 18 questions, with responses organised into 4‐likert scale (0‐3).16 The scales measure the frequency of PTSD symptoms, with 0 representing absence of symptoms, 1 present once, 2 presence of two to four symptoms, and 3 presence of 5 or more symptoms.13 The final 18th item assesses for duration of symptoms, with response options including less than 1 month; 1 to 3 months; 3 months to 1 year; and more than 1 year. The PTSD scores were then organised into “no exposure,” “suspected exposure,” and “definite exposure i.e. suspected exposure to PTSD” as previously described.13, 15 Previous studies have used the MPSS due to its good diagnostic validity for PTSD and good psychometric properties including concurrent validity. The Edinburgh Postnatal Depression Rating Scale (EPDS), is a 10‐item self‐report measure of depressive symptoms in the past 1 week.17 Items are scored on a frequency scale, ranging from 0 to 3. A continuous score was obtained by summing the individual items; with the lowest scores representing absent or nonsevere depressive symptoms and vice versa.13 It has been piloted and used in SA and found to possess good psychometric properties.18 ILF was tested at 6 to 10 weeks and 1 year of age, with the child in quiet natural sleep and included tidal breathing, multiple breath washout measures, and the forced oscillation technique as previously described.6, 9 Measures of ILF were first validated on normative data and found to be reliable before their application in this cohort.5 The following ILF parameters were obtained: tidal volume (mL), ratio of time to peak tidal expiratory flow over total expiratory time (tPTEF/tE), respiratory rate (per minute), functional residual capacity (FRC, mL), respiratory system resistance (cmH2O·L·s−1), and compliance (cmH2O·mL−1). All lung function measurements conformed to American Thoracic Society/European Respiratory Society guidelines, as previously published.5 Tidal breathing and flow volume loops (TBFVL) and multiple breath washout (MBW), performed using 4% SF6 as a tracer gas, were collected using the Exhalyzer D with ultrasonic flow meter (Ecomedics AG, Duernten, Switzerland) and mean measures calculated with acquisition and analysis software (Wbreath v3.28.0, Ndd Medizintechnik AG). The forced oscillation technique (FOT) measurement was made with purpose‐built equipment (University of Szeged, Hungary). Composite medium frequency signal (8‐48 Hz) was delivered to the infants via a wave‐tube through a facemask covering the mouth and nose. Socioeconomic status was based on a composite score for education, employment, income, assets and market access, and organised into four quartiles ranging from high to low. Maternal smoking during pregnancy was based on maternal urine cotinine (IMMULITE 1000 Nicotine Metabolite Kit; Siemens Medical Solutions Diagnostics, Glyn, Rhonwy, UK), with levels more than 500 ng/mL considered active smokers, 10 to 500 passive, and less than 10 nonsmokers.19 Benzene was considered present if household levels more than 5 μg/m3.20 Maternal alcohol was assessed with the Alcohol, Smoking and Substance Involvement Screening Test self‐reported questionnaires. Maternal respiratory illnesses were considered present if there was history of asthma, chronic cough, or recurrent wheeze in previous 12 months and/or low forced expiratory volume.6, 9 The study was approved by the Ethics Committee of the Faculty of Health Sciences, University of Cape Town, by Stellenbosch University and the Western Cape Provincial Research committee. Written informed consent was obtained from parents and is renewed annually. All the analyses were done using R statistical software (version 3.1.0 [2014‐04‐10], http://www.r-project.org) and Stata version 13 (Stata Corp, TX). Some psychosocial adversity variables (IPV and depression) were continuous and skewed and thus logarithmic (natural) transformation was performed to try to reduce the skewness of the residuals from regression models. The main analyses were based on cross‐sectional measures of ILF at each assessment (at 6‐10 weeks and at 12 months), using gamma regression, a type of generalized linear model (GLM) that assumes a gamma distribution for the outcome. This is more appropriate than a Gaussian distribution since our outcomes can only take on positive values, are slightly skewed and are log‐transformed. In the cross‐sectional analysis, all children at each assessment were included to allow comparisons in future studies since those missed at 12 months assessments may reinter the study in future follow‐ups. For our panel‐data of two ILF assessments, we used generalised estimating equations (GEEs), that allowed specification of the within‐group correlation structure, using only those children included in both 6 to 10 weeks and 12 months; the response variables were measures of ILF, while assessments of maternal perinatal psychological adversity were the explanatory variables. As with the GLM models, we specified a gamma distribution for our outcomes and an exchangeable correlation structure. In these GLM and GEE analyses, we first constructed unadjusted models, then adjusted models (accounting for potential confounders in particular, sex, socioeconomic status, population group, exposure to benzene, perinatal complications and maternal age, height‐for‐age z scores, respiratory illness, and HIV), and finally tested if there was interaction in the adjusted model between psychological adversity and maternal health behaviour variables (smoking, alcohol consumption, and breastfeeding). Height‐for‐age z scores added into the GLM and GEE models were available for both assessments (at 6‐12 weeks and at 12 months assessments). Other comparisons between two groups were done with the Wilcoxon signed‐rank test for non‐normal continuous scores and Pearson's χ 2 tests for frequency distributions.