Background. Maternal domestic violence (MDV) screening by pediatricians is not well-studied.
Objectives. To determine the practicality and dynamics of routine MDV screening in a private pediatric office and to determine the rate of MDV in Upper Cape Cod, Massachusetts.
Setting. A 3-pediatrician, private pediatric office in Falmouth, Massachusetts.
Participants. Mothers of children aged 1 month to 10 years scheduled to undergo a well-child visit between February 7 and July 7, 2000.
Intervention. Completion of an 11-item questionnaire related to violence.
Results. Seven hundred sixty-six families were scheduled for well visits. Five hundred ninety-two eligible mothers presented to the office. Five hundred fifty-three completed questionnaires were returned (71.2%). The rate of MDV was 2.5% in current relationships (95% confidence interval [CI]: 1.4–4.3), 14.7% in past relationships (95% CI: 11.9–18.0), and 16.5% overall (95% CI: 13.5–19.9).
Increased incidence of MDV was associated with the following variables: 1) harm to a child (odds ratio [OR]: 57.3, 95% CI: 7.3–1232.4), 2) being in a relationship other than a first marriage (OR: 4.6, 95% CI: 2.7–7.8), 3) having been previously asked about MDV (OR: 3.5, 95% CI: 2.1–6.1), 4) having 4 or more children (OR: 3.1, 95% CI: 1.6–6.1), 5) Women, Infants, and Children's program eligibility (OR: 3.0, 95% CI: 1.8–5.0), 6) having public insurance (ie, Medicaid or Children's Medical Security Plan) (OR: 2.2, 95% CI: 1.3–3.7), 7) a history of failure to present for a scheduled well-child visit (no-show; OR: 2.0, 95% CI: 1.0–4.2) and 8) anonymous questionnaire completion (OR: 1.7, 95% CI: 1.0–2.9).
Thirty-two and one-half percent (32/91, 95% CI: 25.6–46.0) of mothers with a history of MDV recall having previously been asked about this by a health professional, compared with 16.9% overall (93/551, 95% CI: 13.9–20.3). Eighty-two and eight tenths (82.8) percent (457/552, 95% CI: 79.3–85.8) of mothers favored pediatricians asking about MDV.
Discussion. This information was gathered within the context of normal work hours in a busy office. No additional staff were required. Hence, routine MDV screening appears feasible.
The results suggest that a documented history of child abuse in a family makes it very likely that the mother has also been abused. However, child abuse among abused mothers is probably underreported.
Furthermore, because most mothers favor domestic violence screening, concerns about lack of acceptance of maternal screening at pediatric visits seem to be unfounded. Screening may actually increase satisfaction with care.
In addition, families who do not show up for appointments are at higher risk. Therefore, screening only at well visits will miss an important group.
Conclusions. Maternal domestic violence screening at well-child visits is practical in a private pediatric office setting. Current rates of screening are low; however, most mothers favor such screening. Furthermore, MDV screening should also be offered on a catch-up basis for those who miss well-child visits, as is currently recommended for immunizations.
- MDV =
- maternal domestic violence •
- WIC =
- Women, Infants, and Children •
- OR =
- odds ratio •
- CI =
- confidence interval
Maternal domestic violence (MDV) is alarmingly common. Current literature suggests that in the general population, 13% to 30% of women have been involved in an abusive relationship at some time in their lives.1
The American Academy of Pediatrics recommends MDV screening at the time of well-child visits.2 However, information about such screening is limited. There is currently 1 published study of routine MDV screening during pediatric visits. The rates of MDV were 31% overall and 17% in the last 2 years.3 This study was conducted in 2 hospital-owned suburban practices. It involved only 154 mothers, and 65% were on Medicaid. Before this, Wissow et al4 conducted interviews of mothers presenting to pediatric primary clinic in a large, urban teaching center; 69% of families were on Medicaid. The incidence of family violence was 40% over the previous year.
A large, recent study assessed the ability of pediatricians to predict MDV among families presenting for care5; the rate of MDV was significantly underestimated by pediatricians. This was conducted by researchers at the pediatricians' offices rather than by the pediatricians themselves during the visit. This suggests that if MDV is to be detected by pediatricians, then it is likely that indirect observations alone are an insufficient means of screening.
The current study was undertaken to investigate the practicality and dynamics of MDV screening in a private pediatric office at the time of well visits, as well as to determine the rate of MDV in our region.
The study was conducted at Falmouth Pediatric Associates, a 3-pediatrician group practice in Falmouth (Cape Cod), Massachusetts during a 5-month period between February and July 2000.
Patients and families seen at this office are almost entirely year-round residents of the Upper Cape Cod area. This practice cares for approximately half of the children in the area. The local population is predominantly white (of a broad range of European descent), with a combined minority of <10% black (primarily of Cape Verdian descent) and Native American (Wampanoag) peoples.6
Mothers were considered eligible if they were scheduled for a well-child visit with a child aged 1 month to 10 years of age inclusive during the study period. The 1-month visit is the second well-child visit at our office; the first visit was omitted so that new parents would have a chance to orient to the practice without other distractions. Ten years was used as the upper age-limit because mothers are often not present in the examination room after this age.
Permanent, nonbiologic mothers, such as adoptive parents, were also considered eligible.
Mothers were excluded if they were accompanied in the examination room by the father, if they had previously completed the questionnaire, if they were not the permanent guardian (eg, stepmother in split custody situations), or if they were unable to read/write in English.
Children within the age criteria who were scheduled for a well-child visit were identified before the office visit. A list of those eligible was posted at the nurses' station. Charts were marked with a sticker to flag the child's eligibility. After the visit, the list was updated depending on whether or not the questionnaire was a) handed out, b) deferred (because of maternal absence, paternal presence, or no-show) or c) refused. Visits rescheduled before the time of the appointment were not considered no-show.
The mother was given the questionnaire when she and her child were brought into the examination room. To avoid the creation of difficult situations for mothers, care was taken to try to avoid speaking the words “domestic violence” when a child was present who might later question this subject. Mothers had the option of filling out the questionnaire anonymously and leaving it in a locked box in the examination room. They could also leave the questionnaire out for the doctor to review when he came in the room. A questionnaire was considered to have been completed anonymously if the last name of the mother and child was absent and the questionnaire was not given to the doctor to discuss.
All mothers who received a questionnaire also received a “personalized safety plan”7 (Appendix 2), regardless of whether the questionnaire was completed. This describes available local MDV resources and steps to be considered before leaving an abusive relationship.
Those who had been deferred on their scheduled day for whatever reason were placed on a second list, which was posted at the receptionist's desk and at the nurses' station. If a staff member noted that this mother brought any of her children in for a visit (sick or well) during the remainder of the study period, the questionnaire was offered. This was done to avoid the bias that would have been present if only those who arrived for their well appointment (without the father) were questioned.
No change in standard office booking or staffing took place to facilitate the study. Families do not get a reminder notice before their appointment, nor is there a computerized reminder system to detect appointments that are not booked or immunizations that are delayed. Families are contacted by telephone and, if necessary, by mail if appointments are missed. The timing of rescheduled routine well visits varies with age; infants are generally seen within 2 weeks, whereas for older children, the wait would usually be at least a month.
There were 2 components to the questionnaire: demographics and violence-related questions (Appendix 1). Demographics included child and mother's names and dates of birth, number of children in the family, insurance type, and eligibility for the Women Infants and Children's (WIC) program. The violence questions were modified from recommended questions from the American Medical Association8 and were similar to those of Siegel et al3 for ease of comparison. The questions were as follows:
In your current relationship, have you ever been harmed or felt afraid of your partner?
In a previous relationship, have you ever been harmed or felt afraid of your partner?
Has your current or past partner harmed any of your children?
Are there any guns in your house?
Mothers were then asked if they had previously been asked about MDV by any health care provider. They were also asked to respond, using a 5-point Likert scale, to the statement: “I think it is a good idea for pediatricians to ask mothers about domestic violence.” (1 = strongly agree, 3 = neutral, 5 = strongly disagree).
Data were entered into and analyzed in EpiInfo Version 6.04 (Atlanta, GA)9 and Systat Version 9.0 (Chicago, IL)10 with data cleaning at the time of entry and duplicate data review to reduce error. The relationship between risk factors and MDV was examined initially using 2 × 2 tables, with χ2 analysis and Mantel-Haenszel correction where appropriate.
A multivariate model using stepwise logistic regression was subsequently fitted from the risk factors that had demonstrated significant associations (P < .05) within the univariate statistics.
Each result is expressed as an odds ratio (OR)–the odds of MDV with a certain risk factor compared with the odds of MDV without the same risk factor. An OR >1 indicates a higher risk of MDV.
Study recruitment and completion data are shown in Fig 1. A total of 766 families were identified as scheduled to present on at least 1 occasion with 1 or more children for well-child visits during the study period. Four hundred sixty-six eligible mothers completed the questionnaire at the initial visit. Two hundred ninety-two mothers were not studied on the first scheduled visit; of these, 93 did eventually complete it, giving a total of 553. Hence, the overall response rate was 71.2% (553/776), including 93.4% (553/592) of those who were eligible and arrived at the office. 47.6% (263/553) of completed questionnaires were done anonymously.
Demographic information is shown in Table 1. The mean maternal age was 32.6 (standard deviation: ± 6.5) years and median number of children was 2.0 (range: 1–11).
A total of 65.3% (359/550)FNa of eligible respondents were currently married and 63.5% (346/545) had private insurance, whereas 32.5% (177/545) had some form of public insurance. WIC eligibility was used as an indirect measure of income; 39.0% were WIC-eligible currently or in the past (214/497, + 52 unsure of eligibility).
The overall rates of MDV are shown in Table 2.
Fourteen and seven tenths (14.7) percent (81/552, 95% confidence interval [CI]: 11.9–18.0)a of mothers acknowledge having ever been abused in a previous relationship; 91.3% (73/80)a of this abuse had taken place more than a year earlier. Two and a half percent (14/552, 95% CI: 1.4–4.3) were currently in a relationship in which they had been abused; 71.4% (10/14) had taken place more than a year earlier. In total, 16.5% (91/552, 95% CI: 13.5–19.8) were abused by a partner at some time in their life.
A history of child abuse was acknowledged in 2.0% (11/550) of families. Ten and nine tenths percent (10.9%; 60/552) of households owned at least 1 gun.
Several variables were associated with a history of MDV (Table 3); these included: 1) a history of harm to a child (OR: 57.3, 95% CI: 7.3–1232.4), 2) being in a relationship other than a first marriage (OR: 4.6, 95% CI: 2.7–7.8), 3) having been previously asked about a history of MDV (OR: 3.5, 95% CI: 2.1–6.1), 4) having 4 or more children (OR: 3.1, 95% CI: 1.6–6.1), 5) being WIC-eligible (currently or previously; OR: 3.0, 95% CI: 1.8–5.0), 6) having public insurance (Medicaid or Children's Medical Security Plan; OR: 2.2, 95% CI: 1.3–3.7), 7) having previously “no-showed” for a well-child visit (OR: 2.0, 95% CI: 1.0–4.2), and 8) filling out the survey anonymously (OR: 1.7, 95% CI: 1.0–2.9). Additional analysis of these risk factors is shown inTable 4.
Multivariate analysis suggests a persistent role of MDV and the following risk factors: harm to a child, being in a relationship other than a first marriage, having been previously asked about MDV, having 4 or more children, and being WIC-eligible.
Sixteen and nine tenths percent (16.9%; 93/551, 95% CI: 13.9–20.3) of respondents recall having previously been asked by a health professional about MDV. Even among those with a history of MDV, only 35.2% (32/91, 95% CI: 25.6–46.0) had ever been asked about MDV by a health care professional.
Using the 5-point Likert scale, participants were asked to comment on the statement “I think it is a good idea for pediatricians to ask mothers about domestic violence” (1 = strongly agree, 3 = neutral, 5 = strongly disagree; Table 5). The mean score was 1.6 (standard deviation ± 1.0). 82.8% (457/552, 95% CI: 79.3–85.8) favored being asked about domestic violence, 12.1% (67/552, 95% CI: 9.6–15.2) were neutral and 5.1% (24/552, 95% CI: 3.5–7.3) opposed such questioning. Response to this question did not differ for those who had/did not have a history of MDV.
There were no new social services consultations required, and only 1 mother admitted to any harm to a child in a current relationship; this had been >2 years earlier and social services was already involved.
This is the largest study to date of MDV screening incorporated into routine well-child examination and the first in a private pediatric office. The overall incidence of MDV is within the range of currently quoted rates.1 The rate is probably an underestimate of the true incidence in this community, as the families surveyed did at least present for care at some point.
This incidence of MDV was lower than that the only previous routine screening study reported by Siegel et al,3 in which 31% of mothers had been abused. Their population had a substantially higher population of Medicaid. A study of mothers presenting to a pediatric emergency departments in an urban center showed a 52% rate of adult physical abuse.11 The same group found that abused adult women are more likely to use emergency departments12; it is not known if this is also true for pediatric emergency departments.
The option of anonymous completion of the questionnaire in this study is hoped to have helped achieve accurate reporting; this is supported by the increased rate of MDV among mothers that remained anonymous (OR: 1.7). This might contradict a previous study,13 which has suggested that women with a history of MDV will tell their physicians if asked. It may be, however, that some women with a history of MDV preferred to remain anonymous, but still would have disclosed if this option were not available.
Mothers were aware that a current history of child abuse would prompt mandatory follow-up; for this reason, child abuse may well have been underreported. In addition, the wording of the question related to child abuse specifically asked about harm by a partner to a child; this unfortunately offered a loophole for mothers who didn't want to report personal abuse to a child.
Because no change in routine booking or staffing took place, this study supports the practicality of routine office screening. Time required for questionnaire completion was not formally measured, but seemed to take only about 5 minutes and was done while waiting.
Several variables were associated with MDV in this study. As previously described, child abuse was more common in homes in which there had been domestic violence. In fact, failure to report personal abuse made it extremely unlikely that child abuse would be reported. This should serve to further heighten the index of suspicion of child abuse in families where mothers are known/suspected to be victims themselves, but have not disclosed.
Variables associated with low income (WIC and public insurance eligibility) showed a higher likelihood of MDV, as did a history of having “no-showed” for a well-visit. Being in a first marriage was associated with a lower likelihood of a history of MDV, as was having <4 children.
The area of risk factors for MDV has been well-studied in the past, but not from the perspective of the pediatrician's office. The pediatrician needs to be able to make informed decisions about the following questions: 1) should I screen for maternal domestic violence?; 2) who should I screen?; and 3) when should I screen? An approach to these questions will be discussed in turn, in light of available information.
Clearly, MDV screening by health care providers is important. The American Academy of Pediatrics is strongly in favor of such screening by pediatric caregivers.2 The regular access of pediatricians to mothers, the known impact of MDV on children, and the link between MDV and child abuse all make MDV screening by pediatricians seem logical. Furthermore, screening should generally improve overall satisfaction with care, as >82% of mothers clearly also approved of it. This is in agreement with previous studies of domestic violence screening in nonpediatric encounters.14 ,15
One might also ask what the “yield” of MDV screening has been. In this study, no new cases of child abuse were uncovered, and so, no mandatory social services consultations were made. Several families already were (currently or in the past) involved with social services. Several mothers currently in abusive relationships chose to remain anonymous. We were impressed that 80 women had left abusive relationships, and only 4 of these had currently returned to an abusive relationship. Of course, it is possible that others had in fact reentered an abusive relationship without being able to recognize it.
The lack of formal reports here should not take away from the value of screening. To begin, it is striking that only 35.2% of women with a known abuse history recall having ever discussed it with a health care provider. Under other circumstances, these women could still have been in such relationships. Furthermore, every mother who participated (and those who did not return the survey) was given an instruction sheet about how to leave an abusive relationship (Appendix 2). Making support available may provide benefits in time. In addition, the study by Siegel et al3 in a different population did yield many new social services referrals.
The next question is who and when to screen; the answer seems to be “all families intermittently”. This study shows that, in terms of easily-accessible demographic information, that those families with public insurance, those that are WIC-eligible, and those with 4 or more children are at higher risk. However, to selectively screen these populations would be both discriminatory and incomplete; this is supported by the low positive predictive value of these variables (24, 24, and 33%, respectively). In addition, Kerker et al5showed that physicians are poor predictors of which families are at risk for domestic violence; one could speculate that this is attributable to an overreliance on stereotypes.
Traditionally, health behavior screening has been done in conjunction with anticipatory guidance at the well-child checkup. However, by tracking those families who missed well-child appointments, we have demonstrated that not only do these families need MDV screening as much as other families, in fact, they need it more. It follows then that “catch-up” MDV screening after a missed well-child visit may be necessary in the same way as “catch-up” immunizations. Although this risk factor did not maintain a persistent association following the multivariate analysis (suggesting it is linked to another risk factor), it provides very useful information about a population that would otherwise be missed.
Furthermore, it leads to the question of whether other anticipatory guidance/health behavior issues may be more important in those that “no-show”. It may be that instead of a list of topics that are discussed at each well visit, there should be a running checklist kept, so that those who miss a well appointment could receive “catch-up” anticipatory guidance at other visits. More research is necessary to investigate the value of this approach.
There are several limitations to this study. First, it is a select population in that only those women who bring their children for care were screened (however, in this area the vast majority of children have a primary care provider). Furthermore, those mothers who had previously had their child(ren) removed from their custody would not have taken part. These mothers are at higher risk for MDV.1 This would lead to underreporting, as would the knowledge that a current history of child abuse would prompt a mandatory social services referral. Based on conservative estimates from the literature, at least 30% of homes in which there is MDV also have ongoing child abuse. So, one might expect about 5 active cases of child abuse (versus 1 reported). The numbers involved are small and this study was not designed to examine the reliability of maternal reporting of child abuse.
In addition, demographic information such as ethnic background and exact income, are not included. However, WIC eligibility is an estimate of the latter, and may more readily available to care providers. Furthermore, local demographics are different from many cities; this will limit generalizability.
Routine maternal domestic violence screening can be come a part of pediatric practice, and a large majority of women favor such screening. The specificity of acknowledged child abuse with respect to maternal domestic violence was extremely high. Screening candidates should include those that do not keep well-child appointments.
We would welcome the replication of our questionnaire in other pediatric office settings.
The authors wish to thank the staff of Falmouth Pediatric Associates, LLP, for their commitment to this project, Dr Kevin Gordon (Dalhousie University, Halifax, Canada) for manuscript review and advanced statistical analysis, and especially the mothers who participated for their cooperation.
- Received January 26, 2001.
- Accepted April 17, 2001.
- Address correspondence to Gregory W. Parkinson, MD, Falmouth Pediatric Associates, 2 Bramblebush Park, Falmouth, MA 02540. E-mail:
↵FNa Denominators differ slightly because of differences in response rate for individual questions.
- American Academy of Pediatrics, Committee on Child Abuse and Neglect
- Siegel RM,
- Hill TD,
- Henderson VA,
- Ernst HM,
- Boat BW
- ↵Wissow SW, Wilson MEH, Roter D, Larson S, Berman HI. Family violence and the evaluation of behavioral concerns in a primary pediatric clinic. Med Care. 1992;30:MS150–MS164
- ↵US Government Printing Office. Social and Economic Characteristics. Washington, DC: US Government Printing Office; 1990:347, 350, 359, 905
- ↵Peace At Home, Inc. Modified From Domestic Violence: The Facts. Boston, MA: Peace at Home, Inc; 1996:12–13
- ↵American Medical Association. Diagnostic and Treatment Guidelines on Domestic Violence. Chicago, IL: American Medical Association; 1992:4–19
- ↵Centers for Disease Control and Prevention. Epi Info Version 6.04 [Computer Program]. Atlanta, GA: Centers for Disease Control and Prevention; 2000
- ↵SSPS Inc. Systat Version 9.0 [Computer Program]. Chicago, IL: SSPS Inc; 1999
- Duffy SJ,
- McGrath ME,
- Becker BM,
- Linakis JG
- Richardson DM,
- Becker BM,
- McGrath ME,
- Duffy SJ,
- Minugh PA,
- St Angelo LM
- ↵Caralis PV, Musialowski R. Women's experience with domestic violence and their attitudes and expectations regarding medical care of abuse victims. South Med J. 1997;9:1075–1080
- Copyright © 2001 American Academy of Pediatrics