Maternal well-being and its association to risk of developmental problems in children at school entry

Participants in this study are part of a well-established cohort of women followed from early pregnancy to the time their children entered school. The original Community Perinatal Care (CPC) study began recruiting participants in 2000 through three family physician maternity clinics in the city of Calgary, inviting women at low medical risk to contribute to the study as they were seeking to begin prenatal care. In a randomized controlled trial, 1,737 women were assigned to three study groups, one receiving the current standard of prenatal care, a second receiving the same with the addition of consultation with a nurse trained in prenatal care, and the third receiving standard care, consultation with a trained nurse as well as home visits from a paraprofessional trained in non-medical prenatal care. Three computer-assisted telephone interviews followed over the course of the pregnancies in the first trimester, at 32-34 weeks gestation and eight weeks after delivery, and the women were also invited to participate in future research. Mothers who agreed to participate further formed the participant base for the first follow up study when their children were three years of age. With a focus on parenting and child development, the follow-up study at three years collected information from 791 women using telephone questionnaires and built on information from the cohort established during the original study. Detailed methods and full results from the CPC study and the first follow up study when children were three years of age are reported elsewhere [14, 32]. Two years later, when the children were aged four to six years, the participants from the follow up study when children were three years of age formed the cohort for this second follow up study and received a mailed questionnaire. Mothers were excluded if they could not speak English and refused participation in the first follow up study.

This questionnaire followed similar themes as were addressed in the follow up study when children were three years of age. In addition to demographic data, specific questions addressed the child's development, social competence, health care utilization, and referrals. Other questions addressed maternal physical health, emotional health, and social support as well as parenting, including morale, sense of competence, and supports and resources (see Additional file 1).

The questionnaire included six standardized measurement scales. One of these was the Parents' Evaluation of Development Status (PEDS), a 10 item parent-report screening measure to facilitate detection of risk for developmental and behaviour/emotional problems [34]. This was our primary outcome measure. Using the PEDS, parents report their concerns in 10 areas: global/cognitive, expressive language, receptive language, fine motor, gross motor, behaviour, social-emotional, self-help, school, and other issues (typically medical or sensory) [35]. The PEDS screens children by level of risk for developmental disabilities and behavioural/emotional problems (i.e. assigns PEDS 'paths') with a sensitivity and specificity that ranges between 70% and 80% [34]. At 4 1/2 to 6 years of age, concerns reported by parents in the global/cognitive, expressive language, receptive language, fine motor, gross motor, school, and other areas are predictive of developmental disabilities [35, 36]. Scoring of the PEDS categorizes children into one of fives paths:

Also included were the Child Social Competence Scale for parents to assess pro-social behaviour, communication and self-control in the child, the SF-8 to assess maternal physical and mental health, and the Medical Outcomes Study Social Support Scale to assess maternal social support [37‐39]. To examine parenting further, the Parenting Sense of Competence Scale and the Parenting Morale Index were included as well [40, 41]. The Parenting Morale Index measures "parenting zeal, enthusiasm, and capability to endure hardship" by measuring the frequency of ten emotional states of parents, and the internal consistency of this relatively new scale is strong (Cronbach's alpha = 0.84) [41]. Table 1 contains detailed information on the standardized measurement scales that were part of the questionnaire. The entire questionnaire was pilot tested with 10 mothers for length, flow, and comprehension and was revised as per comments and consultations.

Questionnaires were mailed to the 791 respondents from the first follow up study with a cover sheet detailing their voluntary participation, confidentiality and the links that would be made to previously collected data. The questionnaire took approximately 15-20 minutes to complete. It was requested that the questionnaire be mailed back within two weeks, after which time telephone calls were made to offer an alternative response mode. Women were considered unreachable if they could not be contacted by telephone due to a change in phone number and alternate contacts could not be reached. Data collection began in August 2007 and was completed by January 2008. Questionnaires were scanned and verified in Teleform, an electronic data capture and management system [42]. Ethics approval was granted to the study from the Conjoint Health Research Ethics Board at the University of Calgary.

The data collected through the questionnaires were linked to data from the original CPC study as well as to the first follow up study using unique study identification numbers. This resulted in data for mothers at five time points from the first trimester to 5 years post partum. Data were analyzed using the statistical package Stata/SE Version 10.0 [43]. The data analysis included descriptive methods for categorical data as well as bivariate and multivariate methods. Based on the PEDS scoring, children were categorized into one of the five PEDS paths (Path A, B, C, D, or E), which indicated the level of risk for developmental and behavioural/emotional problems as noted above. The variables based on the Child Social Competence Scale, SF-8, Medical Outcomes Study Social Support Scale, Parenting Sense of Competence Scale, and Parenting Morale Index were dichotomized at the value closest to the 25^th percentile. Scores equal to or below this cutoff were categorized as high risk while scores above the cutoff were categorized as low risk. During the prenatal period, women had been asked about abuse in the following way: "Abuse can take many forms: physical, emotional (including psychological or verbal), sexual, financial (e.g. withholding or controlling money) or neglect. We ask all participants in this study about abuse in their lives. Have you ever been physically abused, emotionally abused, sexually abused, financially abused, or neglected?" Women who indicated they had been abused in any form were considered to have a history of abuse.

Bivariate analysis, using Fisher's exact test, was undertaken to compare PEDS paths on child and maternal characteristics. Statistical significance was set at p < 0.05. A multinomial logistic regression model was constructed to explore the relationship between risk of developmental problems and historical factors associated with this risk, yielding odds ratios and 95% confidence intervals. Variables in the bivariate analysis were considered eligible for inclusion in the regression modeling if they were significant at p ≤ 0.10, or if prior evidence or theoretical considerations supported their inclusion. Characteristics of the child were entered into the regression model first (e.g. gender, ear infections prior to age two), then maternal measures (e.g. parenting morale, social support, physical health, mental health), and lastly socio-demographic measures. In addition, variables were entered into the model from past to present (e.g. history of abuse reported at pregnancy was entered before parenting morale measured at 3 years post partum). When measurement scales had subscales, the overall index was used in the regression model. Selected predicted probabilities for screening in each PEDS path were also calculated. Although the current study was a follow-up of a cohort over a period of over 5 years, data on child development were only collected at two points, thus limiting our ability to conduct a longitudinal analysis of the child development data.

Of the 791 respondents in the three year follow up study, 491 returned questionnaires when their children were at school entry, resulting in a 62% response rate (Figure 1). The majority of mothers who participated were married or with a partner, educated at a post-secondary level and had a household income over $40,000 per year (Table 2). Almost half reported being a homemaker. The average age of the children was five years, and about half were male. The majority of previous referrals had been to speech and language pathologists. Thirty-two percent of women reported a history of abuse when asked during pregnancy. Of the total sample, 61 women reported physical abuse (12.4%), 110 emotional abuse (22.4%), 65 sexual abuse (13.2%), 20 financial abuse (4.1%), and 16 neglect (3.3%). Mothers who were lost to follow up and did not participate in this study were more likely to have reported the following in the previous study: smoking, not having attended parenting classes or been pregnant again, poor physical health ratings, incomes less than $40,000 per year, less than a high school education, being single or divorced and/or being less than 25 years of age (all p < 0.05, data not tabled). There was no association between the type of prenatal care women had received during the RCT and the risk of developmental problems in their children at school age.

Analysis of the PEDS revealed that 15% (n = 72) of the children screened at the highest risk of developmental problems (Path A). Of these 72, 53% (n = 38) had at least one referral for further assessment or intervention since birth with 33% (n = 24) referred within the last year. Thirty-one percent (n = 153) of the children screened at moderate risk of developmental problems (Path B). Nineteen percent (n = 93) of the children screened in Path C which suggests low risk for developmental problems but elevated risk for behavioural/emotional problems. The remaining 35% (n = 172) of children were screened to Path E which suggests a low risk of either developmental or behavioural/emotional problems. None of the respondents were screened in Path D for parental communication difficulties. Of concerns reported by parents that are predictive of developmental problems, the most commonly reported was expressive language. Of concerns that are not predictive of developmental problems, behavioural and social-emotional concerns were reported most often by the mothers.

An increased risk for developmental problems was associated with lower social competence scores for children (p < 0.001) (Table 3). Of the children at high risk for developmental problems or elevated risk for behavioural/emotional problems (Path A or C), more than 40% also scored in the lowest quartile of overall social competence. Male children were also significantly more likely to screen at risk of developmental problems (Table 3).

The children of mothers who reported poor mental health and low social support were significantly more likely to screen at high risk for developmental problems (Path A) (Table 3). Poor mental health was reported by 37% of mothers with children in Path A but by only 17% of mothers with children in Path E (p = 0.004) (Table 3). Of the mothers with children in Path A, 46% scored in the lowest quartile for current social support while just 19% of mothers with children in Path E scored in the same quartile (p < 0.001) (Table 3). Parenting sense of competence and morale were also associated with a high risk for developmental problems (Table 3). Forty percent of mothers with children in Path A scored in the lowest quartile of parenting efficacy and satisfaction with parenting (p < 0.035) compared to 20% of mothers with children in Path E, and 32% of mothers in Path A compared to 7% of mothers in Path E scored in the lowest quartile for parenting morale (p < 0.001) (Table 3).

Children who were screened at risk for developmental problems were significantly more likely to have mothers who, during pregnancy, had reported a history of abuse and poor social support (p < 0.05) (Table 3). They were also more likely to have mothers who reported poor physical health and poor parenting morale when their child was three years (p < 0.05) (Table 3). In the bivariate analysis, no significant differences were found in income, education or marital status between Paths.

Multinomial logistic regression analysis revealed that the most significant predictors of having a child who was screened at high risk (Path A) compared to low risk of developmental problems included a child who was male, a mother who had reported a history of abuse while pregnant and poor parenting morale when her child was three years old (Table 4). Based on the logistic regression model, a boy, whose mother had a history of abuse and poor parenting morale, had a 35% predicted probability of screening at high risk of developmental problems (Path A) (Table 5). However, a boy, whose mother had no history of abuse and a history of good parenting morale, had a 13% predicted probability of screening at high risk of developmental problems. Similarly, the predicted probability of screening at high risk for developmental problems for a girl, whose mother had a history of abuse and poor parenting morale, was 24%. A girl, whose mother had no history of abuse and a history of good parenting morale, had a 7% probability of screening at high risk for developmental problems (Table 5).