Objective. To assess the relationship between maternal depression and 4 parent-based prevention practices (use of car seats and electrical plug covers, presence of syrup of ipecac in the home, and reading to their child), using a large nationally representative follow-back sample.
Methods. The maternal self-report components of 2 databases were used for this study, the 1988 National Maternal and Infant Health Survey and the linked companion 1991 Longitudinal Follow-Up Survey. A total of 7537 mothers with newborns in 1988 served as the subjects. Measures of the 4 prevention practices were extracted from the 1991 survey. Depressive symptom measures were derived from both surveys using the Center for Epidemiologic Studies-Depression Scale. Weighted bivariate and multivariate logistic analyses were used to assess the relationship between maternal depressive symptoms (trichotomized to depression at both time points, at 1 time point, and at neither time point) and parental prevention practices, while controlling for a wide variety of sociodemographic variables.
Results. Mothers reporting a high level of depressive symptoms (Center for Epidemiologic Studies-Depression Scale score ≥16) reported significantly poorer prevention practices for car seat use, covering electrical plugs, and having syrup of ipecac in the home. High depressive symptoms were also related to a lower likelihood of daily reading, but only for those mothers presently living with a male partner. Engagement in all prevention practices, except having syrup of ipecac in the home, were less likely if the mother reported high levels of depressive symptoms at both time points versus a single time point.
Conclusion. Maternal depression may significantly impede parental prevention practices. As maternal depression is a treatable condition, screening and treating this disorder may contribute to improvement in childhood prevention practices and ultimately child health.
Improving parental health promotion and disease prevention practices for young children may have significant impact in decreasing child morbidity and mortality. Identifying factors that impede parents from engaging in prevention practices would aid in designing appropriate interventions. Studies on parental engagement in these prevention practices are limited and are weighted toward the assessment of sociodemographic correlates.1–3 Although sociodemographic factors may suggest target groups, they provide only limited information on potential impeding mechanisms. Maternal depression is 1 potential impeding mechanism that has been understudied. Maternal depression is known to impact on mother–child interactions and has a high prevalence in the population.4–7 Hence, it would be an appropriate factor to investigate with regard to its potential impact on prevention behavior.
Few papers directly address the relationship between maternal depression and the engagement in prevention practices. The study by Watson and Kemper8 of 202 low-income urban mothers of young children did not find a relationship between substance abuse or depression and 2-month delays in immunization rates. However, this study's sample was drawn from a single clinic with a limited sociodemographic distribution. A Swedish study found no difference in well-baby clinic care use during the preschool period between a sample of mentally ill mothers and a comparison population.9However, the sample size was again limited and differences in health care delivery in Sweden may limit its generalizability to the United States. With regard to parental prevention practices outside of health services, there has been some investigation into psychological factors associated with car seat usage. Association has been found between lower rates of car seat use and low internal locus of control, forgetfulness, inconvenience, and long-term family stress.10–13 Some of these factors may be manifestations or correlates of depression.
Data from the 1988 National Maternal and Infant Health Survey and 1991 Longitudinal Follow-up Survey provide an opportunity to assess the relationship between maternal depression and several prevention practices. These large surveys contain nationally representative data on mothers and their young children and address the limitations of earlier studies that used small and homogenous samples. Parent-based prevention behaviors in these surveys included: use of car seats and electrical plug covers, having syrup of ipecac in the home, and reading to their child. In addition, a well-recognized depression measure, the Center for Epidemiologic Studies-Depression Scale (CES-D), was included in both surveys.14
The 4 practices to be studied reflect important prevention activities. Car seat use has received important emphasis in health promotion, as well as legislation, given that motor vehicle accidents are 1 of the principal causes of childhood morbidity and mortality and car seat use has reduced these rates.15,,16 Syrup of ipecac is important in that accidental poisoning in the home is a significant problem for young children and is 1 of the prevention targets of the American Academy of Pediatrics Injury Prevention Program.17Although the use of electrical plug covers has received little study, it was 1 of the most frequently recommended prevention strategies in a study of prevention counseling by pediatric residents.18Finally, considering that exposure to children's books contributes to a preschooler's emerging literacy,19,,20 reading to a child can be considered a prevention behavior in the sense of decreasing suboptimal cognitive development and delaying school readiness.
The goal of this study was to investigate, in a large, nationally representative database, whether different levels of maternal depression are related to 4 childhood prevention practices, controlling for the effects of sociodemographic variables.
The database for this study was derived from the National Center for Health Statistics' 1988 National Maternal and Infant Health Survey (NMIHS) and 1991 Longitudinal Follow-up Survey. The 1988 survey is a stratified nationally representative sample of calendar year 1988 births from 48 states, the District of Columbia, and New York City.21 Blacks and low and very low birth weight infants were oversampled. There were 13 417 certificates of live births of which 9953 women (74.2%) responded to the questionnaire.22 Mothers were more likely to respond if they were over 30 years old, white, married, and/or had at least a high school education.23 Records were given poststratified weights to be representative of all 1988 US births. Mothers were surveyed at a mean of 17 months (standard deviation = 5.0) postpartum. Details on the sampling scheme, contact protocols, and response rates are documented more fully elsewhere.22
A subsequent longitudinal survey was attempted on all women in the 1988 NMIHS sample who had agreed to be followed up and whose child was alive at the time of the 1991 interview (n = 9440). Eighty-eight percent of this sample (n = 8285) completed the second survey.24 Mothers were surveyed at a mean of 35 months (standard deviation = 4.6) postpartum. Details on the sample characteristics are presented more fully elsewhere.24 The final follow-up sample was again reweighted to be nationally representative and is the sample used in this analysis.
Both surveys included the CES-D. For this study, our sample includes mothers who completed the CES-D in both surveys. Excluded were those that had more than 2 items missing on the depression scale for the first survey (n = 320) or more than 4 items imputed on the second survey (n = 329). The first study imputed all missing items using the hot-deck method. The second study used prorating for cases with less than 5 items missing. The final sample size is 7537.
Maternal depression at the time of each interview (NMIHS and Longitudinal Follow-up Survey) was measured with the CES-D.14 This is a self-report instrument with 20 items, each rated as to the duration and/or frequency to which they experienced them in the previous week. A higher score indicates more frequent and severe depressive symptoms with a maximum score of 60. A score of 16 or greater is used here to indicate depression. This score has been frequently used as a cutoff score indicating possible depression. It represents approximately the 80th percentile of 1 large community sample.25 For this study, we combined the CES-D data from both surveys, creating 3 groups: those never depressed, those depressed at 1 time point, and those depressed at both time points. Although a diagnosis of depression cannot be made using this instrument, for convenience, we use the term depression rather than substantial depressive symptoms.
The 4 parent prevention practice questions were each reduced to dichotomous variables. “How often does (child) use a child car seat or booster seat when riding in a motor vehicle?” was dichotomized to all the time versus less than always (collapsing most of the time, some of the time, rarely, and never). We excluded those who reported that the child did not ride in a motor vehicle (n = 172). The response options of yes or no were retained for: “Do you have Ipecac or another drug in your household which causes vomiting?” Don't know responses were excluded (n = 50). The responses to “Of the electrical outlets in your home within the child's reach that don't have something plugged in them, how many are covered with plastic safety covers, tape, or other coverings?” were dichotomized to all within reach versus less than all (collapsing most, some, or none). “About how often do you or another household or family member read stories to your child?” was dichotomized to every day versus less than every day (collapsing at least 3 times a week, once a week, several times a month, several times a year, or never).
Potential confounding variables that we included in our analyses were race, educational attainment, income, maternal age, cohabitation, and birth order. Race was divided into whites and blacks with others dropped (n = 283). Educational attainment was trichotomized to <12th grade, 12th grade, and >12th grade. Income was dichotomized at $18 000 because those below this level are often eligible for public entitlement programs. Maternal age was trichotomized to teenagers, those 20 to 29, and those 30 and over. Cohabitation was dichotomized based on the mother's positive or negative report of a significant male partner living in her home (child's father, child's step-father, or her boyfriend). Birth order was dichotomized to first birth versus greater than first birth.
All variables included in the study were first examined using univariate analysis with both unweighted and weighted data. All subsequent analyses used only weighted data to allow for national estimates. The unadjusted bivariate relationships between maternal depression and the 4 maternal prevention practices were assessed using the χ2 statistic. In addition, although not shown, stratified analyses of independent, dependent, and control variables were examined to ascertain for effect modifications and interactions.
Multivariate logistic regression analyses were used to assess the relationship between maternal depression and each of the 4 maternal prevention practices, adjusting for significant control variables. All control variables that demonstrated a significant relationship (P < .05) with either the dependent or independent variables were entered into the final multivariate model. In addition, interaction terms were added into the final model if a Breslow-Day value of >.20 was found on stratified analysis and the interaction term remained in a stepwise reduction procedure, where the significance level was set at .05.
All data analyses were performed using the SAS System under Microsoft Windows Release 6.12 (SAS Institute, Cary, NC).26 To adjust for the initial sampling design, SUDAAN was used to correct the confidence intervals and χ2values.27
At time 1, 23.8% of the weighted sample were depressed (ie, scored 16 or higher on the CES-D). This dropped to 16.6% at time 2. Combining these values we found that 23.5% were depressed at 1 of the 2 time points and that 8.5% were depressed at both time points (Table 1).
The implementation of prevention practices ranged from 40% to 55%. Seat belt usage was reported most frequently and having syrup of ipecac in the house was the least frequent practice (Table 1).
In the unadjusted bivariate analyses, mothers with depression were significantly less likely to engage in prevention practices for all 4 practices (Table 2). Those reporting depression at both time points were less likely to engage in prevention practices than those reporting it at only 1 time point, although this was significant only for car seat usage.
Only 43% of the weighted sample reported that they covered all accessible electrical plugs (Table 1). Depression at both 1 and 2 time points remained significantly related to poorer practices when other factors were taken into consideration (odds ratios [ORs] = .79 and .75, respectively; Table 3). Having more than 1 child, being black, being poor, and not having a significant male partner in the home were also associated with poorer compliance with use of electrical plugs. Maternal age and education were not related to this practice.
Only 40% of the weighted sample reported having syrup of ipecac in the home (Table 1). Having depression at 1 of the time points remained associated with its availability in the final model (OR = .83), although depression at both times did not add to the model (Table 3). Less than 12th grade education, having only 1 child, being black, being poor, and not cohabitating with a male partner were related to not having syrup of ipecac in the home. Again, maternal age was not related to this practice in the final model.
Just over half of the sample reported always using car seats (Table 1). Depression at both 1 and 2 time points remained significantly related to car seat use in the final model (ORs = .77 and .69, respectively; Table 3). The absence of a college education, being black, being poor, and not living with a significant male partner were also associated with poorer compliance with car seat use. In addition, maternal age remained in the model with age over 30 associated with more frequent compliance with car seat use.
Half of the weighted sample reported that their child was read to on a daily basis (Table 1). Because the interaction term, depression by cohabitation was significant in our stepwise reduction procedure, we chose to divide the final model for reading into those who were cohabitating and those who were not (Table 3). Depression at both 1 and 2 time points was significantly related to less daily reading, but only for those mothers who reported living with a male partner (ORs = .65 and .61, respectively). Having less education, having more than 1 child, being black, and being poor remained associated with poorer practice in the final models for both those cohabitating and not cohabitating. Finally, mothers 20 to 29 years of age were less likely to report daily reading but only for those mothers cohabitating.
As hypothesized, we found significant and consistent negative relationships between maternal depression and 4 parent-based prevention practices with only 1 subgroup exception. In general, depression at 2 time points demonstrated a stronger relationship with poor practice than 1 time point suggesting a type of dose response.
The rate of depression at time 1 is similar to those found for women in the original community studies using the CES-D (20.7%–29.6%),28 although the rate at time 2 falls below this range. It is unclear why there would be such a decrease in depressive symptoms with time. Perhaps, over time, the joys of parenthood overcompensate for the stress of initially being a parent for some mothers. Of note, the mean age of the child at the first time point corresponds to the beginning of the notorious terrible 2s, while the second point falls beyond this period.
Overall, the rates of prevention behaviors are poor, leaving much room for improvement. The worst of the 4 practices was having syrup of ipecac in the home (39.7%), although the rate was higher than the 25% value found in the National Health Interview Survey of 1985.1 (Note: charcoal may now be a preferred substance to have available in the home rather than syrup of ipecac.29,,30) Car seat use was found to be somewhat higher than a report of 46% in 1984 noted in literature from the National Highway Traffic Safety Administration.31 Little information about the frequency of use of safety plugs is available. One study of Head Start programs noted that <60% of parents reported that they use safety plugs on outlets.32 Comparable rates of frequency of daily reading to children were not identified.
An unexpected finding was the conditional relationship between daily reading and depression depending on the cohabitation status. One might have suspected that depression would have had a more significant impact on the single mother, as a male partner would not be available to compensate for her lack of daily reading. However, in this analysis, the male partner does not seem to serve as a buffer for maternal depression. Perhaps he serves as a competitor with the child for the limited time and energy the depressed mother has for her family. This finding is consistent with the general trend found in stratified analysis (not shown) in which depression demonstrated a weaker relationship with the prevention practice under more difficult conditions (eg, lower income and lesser education). This pattern is not consistent with previous findings, where multiple stressors, including maternal mental illness measures, seem to have an additive or synergistic effect with each other.33,,34
The only other variables demonstrating a consistent relationship across all 4 prevention practices were income and cohabitation status, where being poor and being without a significant male partner were related to poorer practice. As expected, a greater amount of maternal education was related to better practice, although electrical plug use was an exception. Being black was consistently related to poorer practices despite controlling for several other sociodemographic factors. This was particularly striking for presence of syrup of ipecac in the home. Car seat use was an exception, however. In contrast, birth order demonstrated an inconsistent relationship. This may reflect a balance of first time mothers being less knowledgeable but also more cautious with some aspects of caring for her child. Hence, the first time mother may be more vigilant in providing a safe environment, such as through the use of electrical plugs, but be unaware of syrup of ipecac, with the latter perhaps being learned about by mothers with more children through personal experiences with accidental poisonings. Not surprisingly, having more than 1 child negatively impacted on whether the index child was read to on a daily basis.
Although there are important strengths to the study, such as the large and representative nature of the sample, the use of a well-validated and commonly used depression screen, and the assessment of a variety of prevention practices; there are also several limitations. First, the study relies on self-reported prevention practices. This might lead to an overreporting of behaviors in the socially desirable direction.35,,36 However, a large percentage of the sample was willing to report less than ideal practice patterns. Although this potential bias may impact on the accuracy of the overall prevalence of these practices, it is unclear whether this would preferentially impact on reports by mothers depending on their depression status. One could hypothesize that the depressed mother may underestimate her engagement in prevention behaviors as a part of an overall devaluation of her parenting practices. However, we are unable to assess this hypothesis with this dataset.
A second limitation is the dependence on the CES-D score as a measure of depression as opposed to a diagnostic instrument. A false-positive rate of 6.1% and a false-negative rate of 36.4% were found for this instrument in 1 study where major depression, diagnosed using a structured diagnostic interview, was used as the criterion.37 However, it is unclear how this pattern of false-positives and false-negatives would impact the accuracy of the results of the current study. In addition, a measure of depressive symptoms, as opposed to a diagnosis of depression, has been found to be related to impaired mother–child interactions,38 hence a nondiagnostic dimensional instrument may be appropriate in and of itself rather than serving solely as a proxy for diagnosis.
An additional limitation of the study is the cross sectional nature of the data. Despite having two data points, we are not able to determine the temporal relationship between the proposed criterion and predictor variables. Hence, we are not able to determine causality, that is, if maternal depression predicts poor prevention practices. However, that depression may impair engagement in prevention practice is plausible. Low energy and fatigue characterize depression,39 making it difficult to put in the “extra effort” required to engage in various prevention behaviors, especially those that need to be repeated on a regular basis, such as the use of car seats. In addition to the physical manifestations of depression, there are distortions of cognition and beliefs.40 People with depression often have outlooks that are colored by a sense of helplessness and hopelessness,41 which may discourage them from engaging in prevention behavior with their more distant outcomes. In addition, self-efficacy, a related construct, is often decreased in depressed persons.42,,43 One manifestation of this may be that the depressed mother feels that she cannot make a difference in her child's health and development. Finally, the decreased interaction noted between depressed mothers and their children44 might impact on prevention behaviors that require sustained interactions, such as reading to a child.
Finally, the selection of a variety of parent-based prevention practices was limited to only 4 items, because these were the only ones included in this survey. We had hoped to also assess the relationship between depression and the use of well-child visits and delays in immunization; however, this information is primarily available from a linked provider dataset, which is quite incomplete. Although these 4 items may not be representative of parent-based prevention practices in general, they do represent a broad spectrum of activities.
Overall, the frequency of engagement in prevention practices by mothers of young children is far from ideal. Maternal depression may be 1 factor among multiple factors contributing to this deficient pattern. Maternal depression should be considered in addition to more traditional barriers, such as knowledge deficits and low income. Child health practitioners should inquire about maternal depression when poor adherence to prevention practices has been detected, or better yet, they should address maternal depression prophylactically before poor adherence develops.
Although the suffering that mothers with depression experience warrants mental health treatment in and of itself, the potential impact on children broadens the net of professional responsibility to child health practitioners (eg, pediatricians). Zuckerman and Beardslee45 have outlined several reasons for pediatricians to be concerned about maternal depression. Impairment in prevention practices may be 1 more reason for child health providers to be involved in universal screening and appropriate referrals for maternal depression.
This study was funded in part by the Van Amerigen Foundation and the American Psychiatric Association through a Health Services Research Scholar Award (to J.D.M.).
We thank Dr Linda Mundle for providing helpful comments on an earlier draft of this manuscript.
- Received June 4, 1999.
- Accepted September 3, 1999.
Reprint requests to (J.D.M.) Canadian Centre for the Studies of Children at Risk, 215 Patterson Building, Chedoke-McMaster Hospitals, 1200 Main Street West, Hamilton, Ontario, L8N-3Z5, Canada. E-mail:
Dr McLennan was in the Department of Psychiatry at the University of North Carolina at Chapel Hill during the preparation of this manuscript.
- NMIHS =
- National Maternal and Infant Health Survey •
- CES-D =
- Center for Epidemiologic Studies-Depression Scale •
- OR =
- odds ratio
- McGee R,
- Williams S,
- Kashani J,
- Silva P
- Pless I,
- Roghmann L
- Radloff L
- Johnston C,
- Rivara F,
- Soderberg R
- ↵National Highway Traffic Safety Administration. Child Passenger Safety Resource Manual, 1991. Washington, DC: US Department of Transportation; 1991:1–51
- ↵American Academy of Pediatrics. The Injury Prevention Program (T. I. P. P.). Chicago, IL: American Academy of Pediatrics; 1994
- ↵Goldfield B, Snow C. Reading books with children: the mechanics of parental influence on children's reading achievement. In: Flood J, ed. Promoting Reading Comprehension. Newark, DE: International Reading Association; 1984:205–215
- ↵McCormick C, Mason J. Intervention procedures for increasing preschool children's interest in and knowledge about reading. In: Teale W, Sulzby E, eds. Emergent Literacy: Writing and Reading. Norwood, NJ: Ablex Publishing Corp; 1986:90–116
- ↵National Center for Health Statistics. 1988 National Maternal and Infant Health Survey: Public Use Data Tape Documentation. Hyattsville, MD: National Center for Health Statistics, Public Health Service; 1991
- Sanderson M,
- Gonzalez J
- ↵National Center for Health Statistics. 1991 Longitudinal Followup to the 1988 National Maternal and Infant Health Survey: Public Use Data Tape Documentation. Hyattsville, MD: Public Health Service; 1993
- ↵Radloff LS, Locke BZ. The community mental health assessment survey and the CES-D Scale. In: Weissman M, Myers J, Ross C, eds. Community Surveys of Psychiatric Disorders. New Brunswick, NJ: Rutgers University Press; 1986:177–189
- ↵SAS. SAS System, Release 6.12. Cary NC: SAS Institute; 1997
- ↵Shah B, Barnwell B, Bieler G. SUDAAN User's Manual, Release 7.5. Research Triangle Park, NC: Research Triangle Institute; 1997
- ↵Kahane C. An Evaluation of Child Passenger Safety: The Effectiveness and Benefits of Safety Seats. Washington, DC: Office of Standards Evaluation, National Highway Traffic Safety Administration; 1997:1–13. Available at: www.nhtsa.dot.gov
- ↵Rutter M. Protective factors in children's responses to stress and disadvantage. In: Kent M, Rolf J, eds. Primary Prevention of Psychopathology, III: Social Competence in Children. Hanover, NH: University Press of New England; 1979:49–74
- Sameroff A,
- Seifer R,
- Barocas R,
- Zax M,
- Greenspan S
- ↵Sudman S, Bradburn N. Response Effects in Surveys. Chicago, IL: Aldine Publishing Company; 1974:9–10
- Hebert J,
- Ma Y,
- Clemow L,
- et al.
- ↵American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994:320–327
- ↵Beck A, Rush J, Shaw B, Emery G. Cognitive Therapy of Depression. New York, NY: Guilford Press; 1979:1–33
- ↵Akiskal H. Mood disorders: clinical features. In: Kaplan H, Sadock B, eds. Comprehensive Textbook of Psychiatry. 6th ed. Baltimore, MD: Williams & Wilkins; 1995:1123–1151
- Zuckerman B,
- Beardslee W
- Copyright © 2000 American Academy of Pediatrics