Sequential screening for psychosocial and behavioural risk during pregnancy in a population of urban African Americans


  • M Kiely,

    Corresponding author
    1. Division of Epidemiology, Statistics and Prevention Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, Rockville, MD, USA
    • Correspondence: M Kiely, Division of Epidemiology, Statistics and Prevention Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, 6100 Executive Blvd, Rockville, MD 20852-7510, USA. Email

    Search for more papers by this author
  • MG Gantz,

    1. Statistics and Epidemiology Unit, RTI International, Rockville, MD, USA
    Search for more papers by this author
  • MN El-Khorazaty,

    1. Statistics and Epidemiology Unit, RTI International, Rockville, MD, USA
    Search for more papers by this author
    • Deceased.
  • AAE El-Mohandes

    1. College of Public Health, University of Nebraska Medical Center, Omaha, NE, USA
    Search for more papers by this author



Screening for psychosocial and behavioural risks, such as depression, intimate partner violence, and smoking, during pregnancy is considered to be state of the art in prenatal care. This prospective longitudinal analysis examines the added benefit of repeated screening, compared with a single screening, in identifying such risks during pregnancy.


Data were collected as part of a randomised controlled trial to address intimate partner violence, depression, smoking, and environmental tobacco smoke exposure in African American women.


Prenatal care sites in the District of Columbia serving mainly women of minority background.


A cohort of 1044 African American pregnant women in the District of Columbia.


Mothers were classified by their initial response (acknowledgement of risks), and these data were updated during pregnancy. Risks were considered new if they were not previously reported. Standard hypothesis tests and logistic regression were used to predict the acknowledgment of any new risk(s) during pregnancy.

Main outcome measures

New risks: psychosocial variables to understand what factors might help identify the acknowledgement of additional risk(s).


Repeated screening identified more mothers acknowledging risk over time. Reported smoking increased by 11%, environmental tobacco smoke exposure increased by 19%, intimate partner violence increased by 9%, and depression increased by 20%. The psychosocial variables collected at the baseline that were entered into the logistic regression model included relationship status, education, Medicaid, illicit drug use, and alcohol use during pregnancy. Among these, only education less than high school was associated with the acknowledgement of new risk in the bivariate analyses, and significantly predicted the identification of new risks (OR 1.39, 95% CI 1.01–1.90).


It is difficult to predict early on who will acknowledge new risks over the course of pregnancy, and thus all women should be screened repeatedly to allow for the identification of risks and intervention during prenatal care.


Psychosocial problems among pregnant women, such as poverty, mental health problems, including depression, substance abuse, violence, and social isolation, have an adverse impact not only on pregnancy outcome, but also on the child's health, behaviour, and development.[1, 2] When women are seen for prenatal care, they should be screened for psychosocial problems. Although many are not remediable to change within the clinical setting, identification of such risk factors can be helpful in targeting anticipatory guidance as well as referral to other healthcare or social service settings. Recommendations to screen for such risks are considered to be state of the art in peri- and prenatal care.[3]

The exposure to risks considered in this study (depression, intimate partner violence [IPV], smoking, and environmental tobacco smoke exposure [ETSE]) have all been causally associated with poor pregnancy outcomes. Depression during pregnancy is common, with rates ranging between 10 and 30%.[4-6] Depressive symptoms can lead to an increased risk for low birthweight (LBW) and preterm delivery (PTB),[6, 7] poor mother–child relationships, and poor psychosocial child development.[8, 9] These findings are particularly relevant to the lives of African American women, as research has consistently shown that they experience multiple sources of stress in their lives,[10, 11] and that greater exposure to stressors is associated with increased depressive symptoms.[12] Exposure to IPV increases the likelihood of poor physical health, physical disability, psychological distress, mental illness, including depression, and heightened substance use, including alcohol and illicit drugs.[13, 14] Abused women have higher rates of sexually transmitted diseases, vaginal bleeding or infection, and urinary tract infections.[15] Abuse during pregnancy has been associated with significantly higher rates of depression, suicide attempts, as well as the use of tobacco, alcohol, and illicit drugs,[16-21] LBW, very LBW, PTB, very PTB, and neonatal death.[14, 22-24] Smoking is known to increase the likelihood of LBW,[25, 26] PTB,[25, 27, 28] as well as infant mortality and morbidity.[28-30] Adverse effects of ETSE during pregnancy exist,[31] and are similar to those for active smoking.[31-33]

This prospective longitudinal study examines the added benefit of repeated screening, compared with a single screening, in identifying psychosocial and behavioural risks during pregnancy.


Study population

The population included in these analyses was recruited to a randomised controlled trial (RCT): the District of Columbia Healthy Outcomes of Pregnancy Education (DC–HOPE), which was part of the National Institutes of Health – District of Columbia Initiative to Reduce Infant Mortality in Minority Populations. This RCT evaluated the efficacy of an integrated cognitive behavioural intervention targeting cigarette smoking, ETSE, depression, and IPV during pregnancy. Women were eligible if they were at least 18 years of age, self-identified as belonging to an ethnic minority, were at less than 29 weeks of gestation, English speaking, a Washington, DC resident, and acknowledged at least one of the four targeted risks. Women were recruited and followed between July 2001 and July 2004 at six prenatal care sites. Women were screened using an audio computer-assisted self-interview (ACASI; see El-Khorazaty et al.[34] for details). For women who were eligible based on their screening, baseline interviews were conducted an average of 9 days after screening. Institutional Review Board approval was obtained from all participating institutions.

Of the 2913 women who were screened, 1398 were eligible and 1070 were from a minority background. These women were consented, completed the baseline questionnaire, and were randomised to either the intervention or to normal care. Of these, 1044 women self-identified themselves as African American and were still pregnant at the time of the baseline interview.

The intervention that was delivered as part of the RCT was conducted during routine prenatal care (PNC) visits at the clinics by interventionists (master's level social workers or psychologists), who were trained specifically to deliver this intervention. The intervention was evidence-based and specific to each of the psychobehavioural risks.[35] At each intervention session the woman identified which of the four risks she was experiencing, and the intervention was targeted to address all of the reported risks, regardless of what the woman had reported previously. For example, the intervention for IPV emphasised safety behaviours, provided information about types of abuse and the cycle of violence, provided a Danger Assessment Component to assess risks, and preventive options women might consider (e.g. filing a protection order), as well as the development of a safety plan. The women also received a list of community resources.[14] The intervention was designed to help women address the targeted psychobehavioural risks. Eight women (six randomised to the intervention and two to normal care) were identified as suicidal during intervention or data collection. These women were immediately referred to mental health care, and were excluded from further participation in the study.

Data collection

Data on sociodemographic and behavioural risk were collected by telephone interview at baseline, and during the second and third trimesters (at 22–26 and 30–34 weeks of gestation, respectively). Interviewers were blinded to randomisation group. Smoking was based on self-reported cigarette smoking in the past week. ETSE was assessed by women's report of their partner, household members, or family/friends smoking, and their estimated household exposure for the past 7 days as well as personal ETSE on a typical day at or away from home in the past week. Depression was measured for the last month using the 20–item Hopkins Symptom Checklist Depression Scale, and IPV was measured using the physical assault and sexual coercion subscales of the Revised Conflict Tactics Scale. Both victims and perpetrators of IPV were classified as having IPV risk. The reference period for baseline IPV was the previous year, and the reference period at each follow-up time point was the time since the previous interview.

Statistical analysis

Risks acknowledged during the second and third trimester interviews were classified as new if they had not been reported at a previous interview. At each follow-up time point, the number of women with each new risk (smoking, ETSE, depression, or IPV) was divided by the number of women who acknowledged the risk at baseline to calculate the percentage increase in the risk, compared with baseline. In order to understand which factors at baseline might help identify who was likely to acknowledge additional risk(s) moving forwards through pregnancy, standard hypothesis tests compared women who acknowledged a new risk with those who did not acknowledge a new risk, based on demographic and psychosocial variables measured at baseline, including age, parity, gravidity, relationship status, education, Medicaid, illicit drug use, and alcohol use during pregnancy. Student's t–tests were used to compare the groups with respect to continuous variables, and chi-square tests were used for comparisons with respect to categorical variables. A logistic regression model was constructed to predict acknowledging any new risk at either follow-up interview. Predictors included in the model were those variables that were statistically significant at the P < 0.10 level in the bivariate analysis.


At the baseline interview, 198/1044 (19.0%) women acknowledged smoking, 742/1025 (72.4%) acknowledged ETSE, 463/1044 (44.3%) women were depressed, and 464/1041 (44.6%) women acknowledged IPV as a victim, perpetrator, or as both. In total, 591 women participated in the first follow-up interview, 717 participated in the second follow-up interview, and 458 participated in both. Overall 850/1044 (81.4%) had at least one follow-up interview during pregnancy (first or second). Figure 1 provides a diagram of the numbers of women screened, their eligibility, and follow-up in project DC–HOPE. At the follow-up interviews in the second and third trimesters, each woman was questioned again about each of the risks. Women acknowledging active smoking increased by 5.1% at the first follow-up visit, and by 5.6% at the second follow-up visit. Women acknowledging exposure to ETSE increased by 11.9% at the first follow-up visit, and by 7.1% at the second follow-up visit. Women acknowledging IPV increased by 3.7% at the first follow-up visit, and by 5.0% at the second follow-up visit. Women acknowledging depression increased by 8.6% at the first follow-up visit, and by 11.7% at the second follow-up visit (Table 1). The total number of risks acknowledged increased from 1867 at baseline to 2163 after the last follow-up interview, an overall increase of 15.9%. Because the RCT was designed to reduce risks, Table 2 reports results only by care group. Women in the intervention group had an overall 14.7% increase of reported risks, whereas women in the normal care group had a 17.1% increase of reported risks. Looking at the results by women rather than by risk, 13.4% of women randomised to the intervention acknowledged additional risks at the first follow-up, and 9.6% acknowledged additional risks at the second follow-up; in the control group 12.6% of women acknowledged new risks at the first follow-up, and 12.2% acknowledged new risks at the second follow-up.

Table 1. Acknowledgement of risk at baseline and at follow-up interviews during pregnancy
Risk factorBaseline (4–28 weeks of gestation)Follow-up 1 (22–26 weeks of gestation)Follow-up 2 (34–38 weeks of gestation)
  1. Data given in parentheses are expressed as percentage.

Active smoking198+10 (5.1)+11 (5.6)
ETSE742+88 (11.9)+53 (7.1)
IPV464+17 (3.7)+23 (5.0)
Depression463+40 (8.6)+54 (11.7)
Table 2. Acknowledgement of risk at baseline and at follow-up interviews during pregnancy by care group
Risk factorCare groupBaseline (4–28 weeks of gestation)Follow-up 1 (22–26 weeks of gestation)Follow-up 2 (34–38 weeks of gestation)
  1. Data given in parentheses are expressed as percentage.

Active smokingIntervention106+4 (4.0)+6 (6.0)
Normal care92+6 (7.0)+5 (5.0)
ETSEIntervention365+44 (12.0)+25 (7.0)
Normal care377+44 (12.0)+28 (7.0)
IPVIntervention229+18 (8.0)+23 (10.0)
Normal care234+22 (9.0)+31 (13.0)
DepressionIntervention241+8 (3.0)+10 (4.0)
Normal care223+9 (4.0)+13 (6.0)
Figure 1.

Profile of the project DC–HOPE randomised controlled trial.

In the bivariate analyses, only education of less than high school level was associated with an acknowledgement of new risks at the P < 0.10 level (Table 3). As the only independent variable in the logistic regression model, an education of less than high school level significantly predicted the acknowledgement of new risks (OR 1.39, 95% CI 1.01–1.90).

Table 3. Bivariates of women reporting versus not reporting new risks at follow-up interviews during pregnancy
CharacteristicValueNew risks after baseline (n = 256)No new risks after baseline (n = 594) P Total (= 1044)
  1. Data given in parentheses are expressed as percentage.

Maternal ageMean ± SD24.1 ± 5.124.5 ± 5.40.316124.6 ± 5.4
Pregnancies (including current)Mean ± SD3.5 ± 2.23.7 ± 2.40.21453.7 ± 2.4
Previous live birthsMean ± SD1.4 ± 1.61.4 ± 1.50.79401.4 ± 1.6
Education level< High school88 (34.4)163 (27.4)0.0420251 (29.5)
Relationship statusSingle/separated/widowed/divorced193 (75.4)457 (76.9)0.6260650 (76.5)
Married or living with partner63 (24.6)137 (23.1) 200 (23.5)
MedicaidYes194 (76.1)460 (77.7)0.6052654 (77.2)
Alcohol use in this pregnancyYes51 (19.9)135 (22.8)0.3579186 (21.9)
Illicit drug use in this pregnancyYes28 (10.9)71 (12.0)0.672099 (11.7)
Care groupIntervention123 (48.1)300 (50.5)0.5108423 (49.8)
Usual care133 (52.0)294 (49.5)427 (50.2)

The data reported here do so both in an overall sense, not considering whether the women recruited to this study were randomised to the intervention or normal care, and by randomisation group. It should be noted that the intervention was designed to intervene on women's risks. The intervention was successful in significantly reducing IPV and ETSE, but not depression or active smoking.[14, 36] The overall effect of the intervention on all risks significantly reduced the occurrence of severe prematurity.[37]


Main findings

It is evident from our results that sequential screening for psychosocial and behavioural risks will assist healthcare providers in identifying a larger percentage of women impacted upon by such risks. As noted by Harrison et al.,[38] such screening allows providers to make a better assessment of multiple co-occurring risks and their impact on an individual patient. Despite this, such screening is not uniform in the USA or abroad. Additionally, interventions to all four risk factors are available, and have shown efficacy in improving pregnancy outcomes, either singularly or in combination.[14, 36, 37, 39-41]

Some risks, such as depression, actually do wax and wane. It is quite common with mood disorders such as depression to observe variances over time, from depressed to normal or hypomanic moods, or other variations.[42] Additionally, there are risks, such as smoking, from which women may abstain from when they realise that they are pregnant; however, women who quit smoking during the first trimester voluntarily, or because of a physical aversion, may be likely to resume smoking during the latter part of pregnancy.[43-45]

The women in this sample brought with them many challenges to their pregnancies, in addition to the risks for which they were screened, including poverty and other forms of substance use. Although they were willing to participate in the interviews (the data presented here), a portion of the women randomised to the intervention did not participate, although they represented a minority of the women.

It is generally accepted that longitudinal data are preferable to cross-sectional data, and will provide a researcher with a richer data set. It has also been shown that socially desirable responses (e.g. answering negatively to questions about smoking during pregnancy) will decrease over time.[46] In this study we did not measure social desirability, although this was likely to decrease over the repeated interviews.

We can only speculate why women with lower educational status were more likely to report new risks during later stages of their pregnancy. Although the questionnaire was designed for a low literacy level, participants may not have clearly understood the questions during the initial interview(s). Although it is possible that the women did not understand the questions, this may not be the most likely explanation. Women with a lower educational attainment may also have the perception of being less empowered, from a sociocultural perspective. These women may have issues of trust with the healthcare providers, and may be unwilling to share information that they perceive may expose them to judgment or further disempowerment. It is also possible that women were reluctant to share information about themselves to an unfamiliar interviewer, or that additional stressors in their lives impacted the expression of risk directly or indirectly over the course of their pregnancy. All of these possibilities could have contributed to our findings and warrant further study, particularly a more in-depth study of the correlates of emergent risks.

Strengths and weaknesses

The main strengths of this study include that the data were collected as part of a prospective, controlled trial. Women were followed through their pregnancies. Additional strengths include that the sample is longitudinal and targeted high-risk expectant mothers, hence can thoroughly assess the research objective: to examine whether the repeated screening of risks might encourage certain mothers to acknowledge the presence of risks. A limitation of the study includes its restriction to high-risk African American women. Although it is likely that these results would apply to other high-risk pregnant women from minority backgrounds, there is a potential lack of generalisability to a broader population.

Furthermore, the rates reported in our study are true for women receiving care at the same institution, with a certain degree of continuity, and interviewed by the same person. These findings may not be reproducible where care is fragmented or where patients interact with multiple providers over the course of their pregnancy; however, we believe that the results of this analysis and its importance can be extended to other populations of pregnant women. When the women were interviewed, they were queried about each of the risks. At each data collection time point, validated instruments were used. At baseline and during pregnancy, the Revised Conflict Tactics Scale was used to measure IPV, the Hopkins Symptom Checklist Depression Scale was used to measure depression, and ETSE and smoking were assessed by self-report. We have no way to differentiate between a woman's failure to report a risk and the actual absence of it.


Whether the results presented here are a reflection of new risk exposure in a population free of that risk at baseline, or whether they reflect an increased level of comfort in sharing risk status with the provider, deserves further investigation. The data here do not allow us to understand whether it is one situation or the other or both, depending on the participant. The ability to differentiate these responses would enhance a provider's ability to target anticipatory guidance. A cross-sectional approach towards risk evaluation at a particular moment in pregnancy may be ill suited to the dynamic and longitudinal trajectory of biological and psychobehavioural circumstances. In these situations, a single measurement may give a poor indication of risk at a later point in pregnancy. Thus, repeated measurements are considered desirable to improve risk assessment. Regardless of which situation is relevant, repeated screening allows the provider the opportunity to offer interventions that may have otherwise been unavailable to the patient. Intervening on such newly identified risks at the time of discovery is likely to be of benefit to mothers and their infants. It is difficult early in pregnancy to predict who will acknowledge new risks over the course of pregnancy, and thus all women should be screened repeatedly to allow for identification and intervention during prenatal care.


The results as reported support repeated psychobehavioural assessment over the duration of a pregnancy to become incorporated as a standard of obstetric care. This issue cannot be left to the judgment of the individual healthcare provider, as an initial negative screening may not be consistently predictive of psychobehavioural risk later in pregnancy. This is particularly true in women with lower educational attainment, as seen in this study. The exact reason of variation in risk expression over time needs further research, and may only be possible in situations where objective measures can be matched against patients' report. Smoking would be a perfect example. In other risks, such as IPV, an objective measure would be hard to obtain.

Disclosure of interests

None of the authors has any competing interests to declare.

Contribution to authorship

MK, as the NICHD Project Officer, oversaw all of the activities of the study while it was in the field. She participated in the analysis and interpretation of the results. MK did a significant amount of the original writing of the article, as well as revising it critically for important intellectual content. MK has given final approval of the article. MMG supervised data processing and the creation of the analysis data set, revised and contributed to the analysis plan, performed substantial statistical analyses, and participated in the writing of the article. MMG has given final approval of the article for publication. MNE, as principle investigator of the Data Coordinating Centre, made substantive contributions to the conception, planning, design, sample size determination, development of the instruments, development of the Data Management System (DMS), and conduct of the study. He also monitored recruitment, data collection, and follow-up activities, supervised data processing, developed the analysis plan, conducted interim analysis, performed substantial statistical analyses, and participated in the writing of the article. MNE passed away prior to the finalisation of the article. AAE was directly involved in the design and implementation of all aspects of this study. He monitored all activities related to recruitment and retention as the executive principle investigator of the study team. AAE also participated directly in the analysis plan and the interpretation of results. He participated in the authorship of the article: he reviewed and edited, when necessary, and approved the text as presented in its final form.

Details of ethics approval

This study was approved by the Human Subjects Committees at Howard University (for the clinical sites), RTI International (the data coordinating centre), and the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The data were collected as part of a clinical trial, registered at : NCT00381823.


This study was supported by grant nos 3U18HD030445, 3U18HD030447, 5U18HD31206, 3U18HD031919, and 5U18HD036104, and by the Eunice Kennedy Shriver National Institute of Child Health and Human Development and the National Center on Minority Health and Health Disparities. This research was supported, in part, by the intramural program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development.


The authors would like to acknowledge our many collaborators. We would also like to thank the participants who welcomed us into their lives in hopes of helping themselves and their children.

  1. Correction added on 12 August 2013, after first online publication: grant number 3U18HD03919 was corrected to 3U18HD031919 in the Funding section.