Screening for partner violence in the early postpartum period: are we missing families most at risk of experiencing violence?

Abstract

Objectives

In Manitoba, government policy is for public health nurses to screen families with newborns within 1-week post-discharge for risk factors associated with poor child developmental health. The purpose of this study was to compare the characteristics of families who are screened for intimate partner violence (IPV) with families without a documented response to an IPV screen item. This information can be used to help identify and target families in need of support whose needs are not being met within the current system.

Methods

Manitoban women giving birth to a live singleton in the province from January 1, 2003 to December 31, 2006 were included in the analyses (N = 52,710). Data were part of a larger research study following these families for several years to examine longer-term developmental outcomes. Administrative databases from the Manitoba Centre for Health Policy provided data for the study. Descriptive statistics and logistic regression were used to examine relationships between IPV screen status and socio-demographic covariates and birth outcomes.

Results

In the study population, 66.7% of the sample were screened for IPV. Women less than 20 years of age, not in married or common-law unions, and living in lower income areas were less likely to have a documented response to the IPV screen item. A low number of prenatal care visits, prenatal mental health problems, and prenatal substance use, as well as premature and low birthweight delivery, were associated with a decreased likelihood of having a documented response to the IPV screen item.

Conclusion

Incorporating violence screening into routine prenatal and postnatal care, rather than only screening women after birth, may help to better identify families with unmet needs and ensure more timely referrals to positive strengths-based supports and services.

Introduction

Violence against women is a global public health concern that has devastating consequences not only for individual victims (World Health Organization 2010) but for society as a whole (Zhang et al. 2013). The World Health Organization estimates that 1 in 5 women will experience some form of physical and/or sexual violence in their lifetime (World Health Organization 2010). It is estimated that between 3 and 11% of pregnant women will experience physical violence during pregnancy (Taillieu and Brownridge 2010). Violence during pregnancy can have a negative impact on the physical and mental health of the mother (Sharps et al. 2007; Howard et al. 2013; Alhusen et al. 2015; Howell et al. 2017), a number of adverse pregnancy outcomes (Sharps et al. 2007; Alhusen et al. 2015; Howell et al. 2017; Meuleners et al. 2011; Shah and Shah 2010; Silverman et al. 2006), and devastating consequences for the developing child (Howell et al. 2017).

The pregnancy period offers a unique opportunity to intervene in the lives of women experiencing violence because prenatal care encompasses multiple visits, reaches almost all pregnant women, and targets the age group most at risk for violence (Durant et al. 2000). Several health organizations in the United States (American College of Obstetricians and Gynecologists 2012; American Academy of Pediatrics, Committee on Child Abuse and Neglect 1998; Thackery et al. 2010; United States Preventive Services Task Force 2013) and the United Kingdom (National Health Services Executive 2001; Royal College of Obstetricians and Gynaecologists 2004) have recommended that health care providers routinely screen all women for intimate partner violence (IPV) during pregnancy; however, most health care providers do not follow these recommendations (Alhusen et al. 2015; Bunn et al. 2009; Chamberlain and Perham-Hester 2000; D’Avolio et al. 2001). To date, similar recommendations for universal screening have not been endorsed by Canadian professional health care organizations (Wathen and MacMillan 2003; Society of Obstetricians and Gynaecologists of Canada 2005). As well, several barriers to IPV screening have been identified in the literature, including time constraints, lack of training, inadequate resources, personal discomfort, and uncertainty regarding how to handle disclosures or how to access resources (Howell et al. 2017; Chamberlain and Perham-Hester 2000; Buck and Collins 2007; Mezey et al. 2003; Lauti and Miller 2008). The inability to meet with the women privately (Mezey et al. 2003), the need to ensure women’s safety (García-Moreno 2002), and concerns about privacy, confidentiality, and mandatory reporting requirements (Lauti and Miller 2008) have also been identified as barriers to IPV screening.

In Manitoba, as part of the government-funded Families First (FF) program, public health nurses screen families with newborns within one week post-discharge for risk factors associated with poor child developmental health (Brownell et al. 2011). Information on biological risks is obtained from the hospital record, and nurses are trained to inquire about additional risk factors (including a question pertaining to a past or current history of violence between parenting partners) in a sensitive and non-judgemental manner (Brownell et al. 2011). Families deemed to be at higher risk for poor child developmental outcomes (i.e., 3 or more risk factors indicated on the 38-item FF screen) may be referred to additional services (e.g., home visitation, parenting programs, financial services, mental health programs). Due to jurisdictional issues related to health care delivery in Manitoba, women living in First Nations (i.e., reserve) communities are under federal (vs. provincial) jurisdiction and, therefore, are not screened as part of the provincial FF program (Brownell et al. 2011; Healthy Child Manitoba 2010). All other Manitoban families with newborns, including Indigenous families living off-reserve, are eligible for the FF screen. It is estimated that approximately 83% of all eligible Manitoba births are screened (Healthy Child Manitoba 2010). Therefore, although the government policy is to screen all eligible families with newborns, a substantial proportion of eligible families are not being screened at all (Brownell et al. 2011). Importantly, non-screened families appear to represent a particularly high-risk subset of families in Manitoba as children from these families are twice as likely to enter child protective care than children from screened families (Brownell et al. 2011). Generating information on the characteristics of these families and their children may help to identify and better target for supports and services high-risk families with unmet needs.

The primary objective of this study was to compare the characteristics of families who were screened for a history of IPV (past or current) between parenting partners relative to families who did not have a documented response to the IPV screen item using a population-based sample of all Manitoban women eligible for the FF screen. It was hypothesized that missing data on the FF screen would be non-random, and that families who did not have a documented response to the IPV screen item or were missing the FF screen form would represent a high-risk subset (e.g., young maternal age, low income, poor pregnancy outcomes) and potentially low-risk (e.g., not screened as not perceived as being at risk for IPV) subset of Manitoba families compared with families who were screened for IPV. Thus, we also examined differences in the characteristics of families based on whether missingness on the IPV screen was due to a missing IPV screen item (i.e., no documented response to the IPV screen item on the FF form) or a missing FF screen form (i.e., not screened at all/missing FF form).

Methods

Population and data source

Administrative databases housed at the Manitoba Centre for Health Policy in the Manitoba Population Research Data Repository provided the data for this study. Virtually all contacts by Manitoba residents are captured in these systems (Brownell et al. 2011). Information from Manitoba Health (e.g., hospitalizations, physician visits) and Healthy Child Manitoba (i.e., FF Screen) were linked to provide data for the current study. All Manitoba women with a live singleton birth (weighing at least 400 g and born at a minimum of 18 weeks’ gestation) in the province from January 1, 2003 to December 31, 2006 were included in the analyses (N = 52,710). Data were part of a larger research study that is following the same cohort of women and children for several years to examine longer-term outcomes based on IPV screen response in the early postpartum period. Women living on reserve were excluded (based on postal code at time of birth) from analyses due to jurisdictional issues related to health care delivery in Manitoba (i.e., women living in First Nations communities are not eligible for the FF screen) (Brownell et al. 2011; Healthy Child Manitoba 2010). This resulted in a final sample of N = 45,896 births. This study was approved by the Manitoba Health Information Privacy Committee (HIPC No. 2015/2016-31) and the University of Manitoba Health Research Ethics Board (Ethics File No. H2015:355 (HS18922)).

Measures

Violence screen

An item from the FF screen form was used to assess IPV screen status. Public health nurses document whether or not there was a “current or history of violence between parenting partners” (yes or no). Mothers and their children were divided into three groups based on whether the mother was screened for a history of IPV on the FF screen form: (1) screened for IPV; (2) screened (i.e., FF screen form filled out) but missing a documented response to the IPV screen question; and (3) not screened at all/no FF screen form.

Socio-demographic characteristics

Socio-demographic covariates included the following: maternal age at birth (ordinal), marital status (i.e., registered legal marriage or common-law union in health insurance registry vs. no registered union in health insurance registry), area-level income quintile (based on postal code at the time of birth), total number of children in home (including the newborn), and provincial region of residence. Low maternal education (i.e., less than a grade 12) and social assistance receipt/financial difficulties (yes or no) were only assessed on the FF screen form; therefore, these variables were only considered for assessment in comparisons involving mothers with a completed FF screen form.

Pregnancy-related characteristics

Pregnancy-related covariates included the following: low number of prenatal care visits (i.e., less than 5 visits prior to delivery), parity (primaparous or multiparous), and maternal prenatal mood or anxiety disorder (yes or no). Maternal prenatal mood or anxiety disorder was based on whether the mother had a diagnosed mood and/or anxiety disorder (as indicated in hospital discharge abstracts, medical claims records, and/or prescription drug information) in the year prior to delivery (Heaman et al. 2012). Dichotomous assessments (yes or no) of maternal smoking, alcohol use, and drug use during pregnancy were also documented by public health nurses on the FF screen form; therefore, these variables could not be assessed in comparisons involving mothers who were missing the FF screen form (i.e., not screened at all group).

Birth outcomes

Birth outcomes included the following: child sex (male or female), preterm birth (i.e., less than 37 weeks’ gestation), low birthweight (i.e., less than 2500 g), high birthweight (i.e., more than 4500 g), small-for-gestational age (i.e., birthweight at or below the 10th percentile based on Canadian standards for sex and gestational age) (Kramer et al. 2001), and large-for-gestational age (i.e., birthweight at or above the 90th percentile based on Canadian standards for sex and gestational age) (Kramer et al. 2001).

Statistical analyses

Descriptive analyses involved computing the prevalence of screening for a history of violence (past or current) between parenting partners in the study population. Group differences (i.e., screened; screened but missing a documented response to IPV item; not screened at all/missing FF screen form) on socio-demographic covariates, pregnancy-related variables, and birth outcomes were examined using cross-tabulations with chi-square tests of association. Significant differences between groups indicated that missing values were likely non-random. Three series of multivariable logistic regression models were computed to examine the extent to which each covariate was associated with each category of missingness: (1) screened but missing documented response to IPV screen question vs. screened for IPV; (2) not screened at all/missing FF form vs. screened for IPV; and (3) screened but missing documented response to IPV screen question vs. not screened at all/missing FF form. Multivariable logistic regression models (i.e., all models in Table 2) adjusted for socio-demographic covariates available for all three groups (i.e., maternal age, marital status, total number of children in home, area-level income quintile, and provincial region of residence). To account for the increased risk of error associated with multiple comparisons, results at a p value of 0.01 were considered statistically significant. All data management, programming, and analyses were performed using SAS® version 9.4.

Table 2.

Relationship between categories of missingness on an intimate partner violence screen item and socio-demographic, pregnancy-related, and birth outcome covariates

Covariate	Missing item on FF form vs. screened for IPV1 (AOR (99% CI))	Missing FF form vs. screened for IPV1 (AOR (99% CI))	Missing FF form vs. missing item on FF form2 (AOR (99% CI))
Socio-demographic covariates
Maternal age at birth
Less than 20 years	1.00	1.00	1.00
20 to 24 years	0.92 (0.80, 1.06)	0.60 (0.52, 0.70)**	0.67 (0.56, 0.80)**
25 to 29 years	0.92 (0.80, 1.05)	0.53 (0.46, 0.62)**	0.61 (0.51, 0.73)**
30 to 34 years	0.88 (0.76, 1.01)	0.47 (0.40, 0.55)**	0.56 (0.46, 0.67)**
35 to 39 years	0.85 (0.72, 0.99)*	0.48 (0.40, 0.58)**	0.58 (0.47, 0.73)**
40 years and older	0.99 (0.79, 1.26)	0.51 (0.37, 0.71)**	0.51 (0.36, 0.73)**
Marital status
Married/common-law	1.00	1.00	1.00
No registered union	1.15 (1.07, 1.22)**	1.35 (1.24, 1.48)**	1.11 (1.001, 1.23)*
Area-level income quintile
1 (lowest)	0.88 (0.80, 0.98)*	1.48 (1.28, 1.71)**	1.56 (1.33, 1.84)**
2	0.84 (0.76, 0.94)**	1.29 (1.12, 1.49)**	1.39 (1.18, 1.64)**
3	0.84 (0.76, 0.93)**	1.23 (1.07, 1.42)**	1.41 (1.19, 1.67)**
4	0.95 (0.86, 1.05)	1.19 (1.03, 1.37)*	1.23 (1.04, 1.45)*
5 (highest)	1.00	1.00	1.00
Total children in home
0	2.14 (1.33, 3.43)**	5.73 (3.91, 8.63)**	2.74 (1.73, 4.33)**
1	1.00	1.00	1.00
2	1.14 (1.06, 1.22)**	1.11 (1.01, 1.23)*	0.98 (0.87, 1.09)
3 or more	1.29 (1.19, 1.41)**	1.44 (1.29, 1.60)**	1.08 (0.95, 1.22)
Health region of residence
Winnipeg	1.00	1.00	1.00
Southern Health	0.13 (0.11, 0.14)**	0.39 (0.34, 0.45)**	2.77 (2.32, 3.31)**
Interlake-Eastern	0.22 (0.19, 0.26)**	0.78 (0.66, 0.92)**	3.11 (2.53, 3.83)**
Prairie Mountain	0.19 (0.16, 0.21)**	0.54 (0.47, 0.61)**	2.78 (2.34, 3.31)**
Northern	0.12 (0.09, 0.16)**	1.70 (1.44, 2.00)**	13.10 (9.79, 17.54)**
Low maternal education
No	1.00	NA	NA
Yes	1.12 (1.002, 1.24)*	NA	NA
Social assistance receipt/financial difficulties
No	1.00	NA	NA
Yes	1.56 (1.41, 1.74)**	NA	NA
Pregnancy-related covariates
Low number of prenatal care visits
No	1.00	1.00	1.00
Yes	1.34 (1.19, 1.52)**	1.61 (1.41, 1.84)**	1.22 (1.04, 1.43)*
Parity
Primaparous	1.00	1.00	1.00
Multiparous	1.28 (1.10, 1.48)**	1.48 (1.24, 1.76)**	1.18 (0.97, 1.43)
Maternal prenatal mood or anxiety
No	1.00	1.00	1.00
Yes	1.04 (0.95, 1.14)	1.07 (0.96, 1.21)	1.03 (0.90, 1.18)
Maternal smoking during pregnancy
No	1.00	NA	NA
Yes	1.42 (1.30, 1.55)**	NA	NA
Maternal alcohol use during pregnancy
No	1.00	NA	NA
Yes	1.16 (1.04, 1.29)**	NA	NA
Maternal drug use during pregnancy
No	1.00	NA	NA
Yes	1.68 (1.42, 1.98)**	NA	NA
Birth outcomes
Child sex
Male	1.00	1.00	1.00
Female	1.00 (0.94, 1.06)	0.99 (0.91, 1.07)	1.00 (0.91, 1.10)
Preterm birth
No	1.00	1.00	1.00
Yes	1.21 (1.06, 1.37)**	1.46 (1.25, 1.70)**	1.13 (0.95, 1.35)
Low birthweight
No	1.00	1.00	1.00
Yes	1.19 (1.02, 1.38)*	1.52 (1.27, 1.83)**	1.27 (1.03, 1.57)*
High birthweight
No	1.00	1.00	1.00
Yes	1.12 (0.93, 1.35)	1.11 (0.88, 1.40)	1.03 (0.78, 1.34)
Small for gestational age
No	1.00	1.00	1.00
Yes	1.05 (0.93, 1.17)	1.01 (0.87, 1.18)	0.97 (0.81, 1.16)
Large for gestational age
No	1.00	1.00	1.00
Yes	1.00 (0.91, 1.09)	0.99 (0.88, 1.12)	1.00 (0.87, 1.14)

FF Families First screening form, IPV intimate partner violence, NA variable was not available in data due to missing FF screen form, AOR adjusted odds ratio (i.e., all models in this table are adjusted for maternal age, marital status, total number of children in home, area-level income, and provincial region of residence)

¹Screened for IPV is the reference group for the dependent variable in logistic regression models

²Missing a response to the IPV screen question on the FF form is the reference group for the dependent variable in logistic regression models

*p < 0.01; **p < 0.001

Results

In the study population, 66.7% of the sample were screened for a history (past or current) of violence between parenting partners (among screened women, 2.1% screened positive for a history of past or current IPV); 22.7% were missing a documented response to the violence screen question on the FF form; and 10.7% were not screened at all as part of the FF screening program.

Cross-tabulations indicated a number of significant differences between screen response groups (see Table 1), suggesting that missingness on the IPV screen question was likely non-random. Women less than 20 years of age, not in a registered marital/common-law union, living in a lower income area, or living in Winnipeg or the Northern provincial region were less likely to have a documented response to the IPV screen item. Women whose FF forms indicated that they were receiving social assistance and/or having financial difficulties were also less likely to have a documented response to the IPV screen item. Pregnancy-related factors also appeared to be related to whether or not a history of IPV was documented on the FF form. Multiparous women, and women with less than five prenatal care visits, with diagnosed prenatal mood and/or anxiety disorders, and who smoked or used drugs while pregnant were less likely to have a documented response to the IPV screen item. Women with premature births and low-birthweight infants were also less likely to have a documented response to the IPV screen item.

Table 1.

Prevalence of socio-demographic, pregnancy-related, and birth outcome covariates associated with categories of missingness on an intimate partner violence screen item

Covariate	Screened for IPV N = 30,558 (66.6%)	Not screened: missing item on FF form N = 10,379 (22.6%)	Not screened: missing FF form N = 4959 (10.8%)	Chi-square
	% (n)	% (n)	% (n)
Socio-demographic covariates
Maternal age at birth
Less than 20 years	6.1 (1869)	6.8 (708)	13.0 (646)
20 to 24 years	20.8 (6352)	18.8 (1953)	23.3 (1155)
25 to 29 years	31.6 (9657)	30.0 (3112)	29.0 (1438)
30 to 34 years	28.0 (8560)	29.4 (3056)	23.0 (1138)
35 to 39 years	11.4 (3474)	12.3 (1276)	9.8 (486)
40 years and older	2.1 (646)	2.6 (274)	1.9 (96)	420.94**
Marital status
Married/common-law	48.1 (14,682)	41.3 (4281)	34.4 (1675)
No registered union	51.9 (15,835)	58.7 (6084)	65.6 (3191)	396.14**
Area-level income quintile
1 (lowest)	19.2 (5844)	24.7 (2556)	26.7 (1319)
2	20.1 (6139)	19.9 (2061)	20.6 (1018)
3	21.3 (6486)	17.9 (1858)	18.6 (916)
4	20.2 (6169)	19.2 (1985)	18.3 (901)
5 (highest)	19.2 (5847)	18.3 (1898)	15.8 (780)	277.81**
Total children in home
0	0.3 (87)	0.6 (59)	1.8 (88)
1	42.5 (12,961)	41.8 (4328)	41.3 (2009)
2	33.9 (10,336)	34.8 (3608)	31.1 (1514)
3 or more	23.4 (7133)	22.9 (2370)	25.8 (1255)	219.66**
Provincial region of residence
Winnipeg	49.6 (15,144)	85.7 (8896)	62.7 (3107)
Southern	22.3 (6800)	5.0 (517)	10.1 (499)
Interlake-Eastern	8.4 (2555)	3.4 (352)	7.3 (360)
Prairie Mountain-Western	15.4 (4708)	4.9 (504)	10.7 (530)
Northern	4.4 (1351)	1.1 (110)	9.3 (463)	4772.09**
Low maternal education
No	79.2 (23,481)	80.2 (6106)	NA
Yes	20.8 (6178)	19.8 (1510)	NA	3.72
Social assistance receipt/financial difficulties
No	85.1 (25,552)	78.1 (6375)	NA
Yes	14.9 (4486)	21.9 (1787)	NA	226.51**
Pregnancy-related covariates
Low number of prenatal care visits
No	93.6 (28,590)	92.0 (9543)	87.7 (4348)	221.35**
Yes	6.4 (1965)	8.1 (836)	12.3 (610)
Parity
Primaparous	41.0 (12,542)	39.7 (4119)	40.2 (1991)
Multiparous	59.0 (18,015)	60.3 (6259)	59.9 (2968)	6.45
Maternal prenatal mood or anxiety disorder
No	87.2 (26,644)	85.6 (8879)	85.3 (4152)
Yes	12.8 (3914)	14.5 (1500)	14.7 (715)	26.12**
Maternal smoking during pregnancy
No	81.4 (24,524)	76.1 (6903)	NA
Yes	18.6 (5612)	23.9 (2169)	NA	122.51**
Maternal alcohol use during pregnancy
No	87.7 (26,311)	87.1 (7427)	NA
Yes	12.3 (3688)	12.9 (1098)	NA	2.10
Maternal drug use during pregnancy
No	96.7 (28,918)	94.2 (7856)	NA
Yes	3.3 (977)	5.8 (485)	NA	115.00**
Birth outcomes
Child sex
Male	51.1 (15,627)	51.3 (5325)	51.8 (2553)
Female	48.9 (14,931)	48.7 (5054)	48.5 (2406)	0.25
Preterm birth
No	94.4 (28,838)	92.9 (9640)	91.1 (4516)
Yes	5.6 (1720)	7.1 (739)	8.9 (443)	93.05**
Low birthweight
No	96.4 (29,459)	95.5 (9911)	93.7 (4646)
Yes	3.6 (1099)	4.5 (468)	6.3 (313)	85.90**
High birthweight
No	97.2 (29,688)	97.0 (10,068)	96.8 (4799)
Yes	2.9 (870)	3.0 (311)	3.2 (160)	2.43
Small for gestational age
No	92.7 (28,328)	91.9 (9536)	92.2 (4563)
Yes	7.3 (2227)	8.1 (842)	7.8 (385)	8.08
Large for gestational age
No	85.8 (26,221)	86.3 (8957)	85.7 (4241)
Yes	14.2 (4334)	13.7 (1421)	14.3 (707)	1.74

IPV intimate partner violence, FF Families First screening form, NA variable was not available due to missing FF screen form

*p < 0.01; **p < 0.001

The results of the multivariable logistic regression analyses examining the relationship between categories of missingness on the IPV screen question are provided in Table 2. Among women who were screened, differences emerged with regard to whether or not the public health nurse filled out the response to the IPV screen question (i.e., missing documented item on FF form vs. screened for IPV comparisons). In these models, adjusted odds ratios (AORs) greater than 1.0 indicated a higher likelihood of not having a documented response to the violence screen question among women with a FF screen form (and AORs less than 1.0 indicated that the covariate was associated with a higher likelihood of having a documented response to the violence screen question). For the most part, maternal age was not related to whether or not a woman had a documented response to the violence screen question among women who were screened. Women not in a registered marital/common-law union, with no children or more than one child living in the home at the time of birth, with lower education levels, or who were receiving social assistance and/or having financial difficulties were significantly less likely to have a documented response to the IPV screen question, despite having a completed FF form (AORs ranged from 1.12 to 1.56). Women who smoked, used alcohol or used drugs during pregnancy, had a low number of prenatal care visits, or who delivered a preterm or low-birthweight infant were also less likely to have a documented response to the IPV screen question, despite having a completed FF form (AORs ranged from 1.16 to 1.68). Conversely, women living in lower income quintile areas (below 60th percentile) and outside of Winnipeg were more likely to have a documented response to the IPV screen question than women living in the highest income quintile or in Winnipeg (as evidenced by AORs less than 1.0).

The comparisons of women missing the FF form (i.e., not screened at all) and women screened for IPV on the FF form revealed that women older than 20 years of age were less likely to be missing the form than women younger than 20 years (AORs ranged from 0.47 to 0.60). Women not in a registered marital/common-law union, with no children or more than one child living in the home at the time of birth, living in lower income quintile areas, and living in the Northern provincial region were significantly more likely to be missing the FF form (i.e., not screened at all) than women screened for IPV (AORs ranged from 1.11 to 5.73). Compared with women living in Winnipeg, women living in other provincial regions (except in the Northern region) were less likely to be missing the FF form than screened for IPV (AORs ranged from 0.39 to 0.78). Women with a low number of prenatal care visits and delivering preterm or low-birthweight infants were also more likely to be missing the FF form than screened for IPV (AORs ranged from 1.46 to 1.61).

In comparing characteristics of women with missing a FF form (i.e., not screened at all) with characteristics of women missing a documented response to the IPV screen question on the FF form, women less than 20 years of age were more likely to have a missing FF form (vs. missing item on the FF form) than women 20 years of age and older. Women not in a registered marital/common law union, with 0 children living in the home, from lower income quintile areas, living in provincial regions outside of Winnipeg, and who had a low number of prenatal care visits or who delivered a low-birthweight infant were also more likely to be missing the FF form (i.e., not screened at all; AORs ranged from 1.11 to 13.10).

Discussion

Several novel findings emerged from this study. First, at the time of cohort selection, 89.4% of families eligible for screening were assessed by public health nurses with the FF screen. However, the question on a history of violence (past or current) between parenting partners was not routinely assessed with regard to the implementation of the FF screen. Of all the screened women, 34.0% were missing a documented response to the IPV screen question on the FF form. This is concerning given that violence during pregnancy is associated with poor pregnancy, maternal, and infant outcomes (Sharps et al. 2007; Howard et al. 2013; Alhusen et al. 2015; Howell et al. 2017; Meuleners et al. 2011; Shah and Shah 2010; Silverman et al. 2006). Second, patterns of missingness on the IPV screen question appeared to be non-random. Young age, low income, single marital status, and substance use are all known risk factors for both IPV (World Health Organization 2010; Taillieu and Brownridge 2010) as well as poor pregnancy outcomes (Feldman and Landry 2015). In this study, these same characteristics were related to categories of missingness with regard to the implementation of the FF screen, regardless of whether missingness was due to a missing IPV item or a missing FF form. Therefore, findings also suggested that a high-risk subset of the Manitoba population was not reached by the FF screen (i.e., not screened for any risk factors), and even when screened (i.e., participate in the FF screen), they were less likely to have a documented response to the IPV screen question on the FF form. It is also important to recognize that in families experiencing IPV, it may be difficult for public health nurses to screen for IPV without compromising the new mother’s safety.

Overall, women less than 20 years of age, not in a registered marital/common-law union, using substances during pregnancy, living in a lower income area, and who were receiving social assistance or having financial difficulties were less likely to have a documented response to the IPV screen. Thus, the FF screen, particularly the IPV screen item, does not seem to be reaching women who are at most risk of experiencing IPV (Taillieu and Brownridge 2010; Capaldi et al. 2012). Covariates associated with not having a documented response to the IPV screen item were similar regardless of whether missingness was due to a missing FF form (i.e., not screened at all) or a missing item on the FF form (i.e., no documented response to the violence screen question). Focusing attention on improving overall FF screening rates in lower income families may help to identify families living in more disadvantaged circumstances who would likely benefit from additional supports offered by the FF program.

Women living in Winnipeg had markedly lower IPV screening rates (55.9%) compared with women living in the other provincial regions (prevalence estimates ranged from 70.2% to 87.1% in most other areas). Among women receiving the FF screen, women living in Winnipeg also were significantly more likely to be missing a documented response to the IPV screen question on the FF form than women living in other provincial regions. It is difficult to speculate why the screening rate, both overall (with the exception of women living in Northern regions) and for IPV specifically, is much lower in Winnipeg than elsewhere. Research aimed at identifying barriers to the implementation of the FF screen in Winnipeg may help address ways of improving overall screening rates specific to the Winnipeg area.

The higher prevalence of missing FF forms in Northern regions of the province is likely due to the underestimation of women living (or receiving health services) on-reserve in northern Manitoba using administrative data. Using postal codes to identify First Nations communities likely underestimates the total number of families living on-reserve because, for a few communities, a single postal code represents residents from both First Nations and non-First Nations communities. In these cases, residents were coded as not living on-reserve, hence likely underestimating families living on-reserve in this study. It is likely that at least some of the “missing” FF forms in the Northern region (many reserve communities are located in northern regions of the province) represent families who were not eligible for FF screening. Nonetheless, the examination of health care services, service delivery, supports, and resource availability for pregnant women living in First Nations communities in Manitoba remains an important research priority.

In this study, women who did not have a documented response on the IPV screen had significantly increased odds of delivering premature and low-birthweight infants. The purpose of the FF screen is to identify families at risk of poor child developmental outcomes in order to offer the supports necessary to achieve more positive outcomes (Healthy Child Manitoba 2010), but the implementation of the screen seems to be missing some families most at risk. It is not just that these families are less likely to be visited by public health nurses after the birth of their child, but also, even when visited, that they are less likely to have a documented response to the IPV screen question. Families living in more disadvantaged circumstances face more challenges, and thus may be less willing or able to make time available for a public health nurse visit after birth, particularly if their newborn infant’s health is compromised in some way. This could also help explain the lower screening rates among young, non-married, and lower-income women in this study. Public health nurses may be reluctant to inquire about IPV for many reasons, including a lack of training, discomfort, or their own personal history of IPV (Howell et al. 2017; Chamberlain and Perham-Hester 2000; D’Avolio et al. 2001; Buck and Collins 2007). Women’s safety should be paramount in all IPV screening programs, so it may be difficult for public health nurses to screen women living in abusive situations without compromising their safety. Safety concerns could also potentially explain the lower IPV screen rates found in higher-risk families. As well, child welfare policy in most jurisdictions in Canada would also mandate that a positive IPV screen in the early postpartum period be reported to child protection, resulting in a negative, punitive response for families. The provision of easily accessible strengths-based supports and services and working collaboratively with families in positive, health-promoting ways would likely be more effective in assisting families experiencing violence than a more punitive, child protection response.

The strengths of this study included the use of a population-based sample of all Manitoban women eligible for the FF screen, which increased generalizability of the findings. Note, however, findings are not generalizable to families living in First Nations communities (i.e., on-reserve) in Manitoba, as these families are not screened as part of the provincial FF program. As well, the linkage capabilities of the data housed in the Repository captured medical outcomes for virtually all women and their children residing off-reserve in Manitoba, and allowed one to follow the same sample of women and their children from the prenatal period to delivery, regardless of prenatal care access or IPV screen response. This study was also subject to a number of limitations. First, the single-item screening question on a history of violence (past or current) between parenting partners could not distinguish between different types of violence experienced, the exact timing of the violence, or the frequency and severity of the violence among women screening positive for IPV. More detailed assessments of IPV experiences on the FF screen form would be useful for a better understanding of IPV during pregnancy and maternal and child outcomes. Second, the exclusion of other important prenatal outcomes (e.g., miscarriage, perinatal death) could not be examined due to the nature of the screening policy in Manitoba (i.e., women are generally not screened until after the birth of a live infant). Third, high-risk families (e.g., young mothers, mothers using alcohol or drugs during pregnancy) may be more likely to have children apprehended at birth. In these cases, no FF screen would be completed. Fourth, maternal prenatal mental health problems were limited to the treatment-seeking population, and diagnoses were contingent upon accurate coding by physicians and other medical personnel. Fifth, many important risk and protective factors (e.g., pregnancy intention, race/ethnicity, household income, social support) were not available in the administrative databases. Finally, information on the FF screen was from 2003 through 2006, and may not accurately represent current FF screen practices.

Conclusion

The FF screen in Manitoba is largely successful at screening mothers of newborns for risk factors associated with poor child developmental outcomes; almost 90% of all mothers eligible for screening were screened. However, the IPV component of the FF screen was less routinely implemented, with 34% of women eligible for the FF screen missing a documented response to the IPV screen question (due to either a missing FF item or a missing FF form). This is important as patterns of missingness on the IPV screen question appear to be non-random, with adolescent mothers, non-partnered women, more socio-economically disadvantaged families, women using substances during pregnancy, and women with poorer pregnancy outcomes being less likely to have a documented response to the IPV screen item on the FF form than women at lower risk. Incorporating IPV screening into routine prenatal and postnatal care, rather than assessing IPV experiences only after birth (as with the current FF screen), may help better identify families in need of support and, ultimately, improve pregnancy outcomes and the longer-term trajectory of women and their children. The provision of strengths-based supports and services needs to be integrated into the continuum of care beginning in the prenatal period so that IPV can be addressed early and with ongoing collaborative care well beyond the postnatal period.

Funding information

This research was supported by a University of Manitoba Graduate Fellowship, Manitoba Graduate Scholarship (Doctoral), Graduate Student Thesis Research Award in the Area of Child Development, and the Evelyn Shapiro Award for Health Services Research awarded to T. Taillieu.

Compliance with ethical standards

This study was approved by the Manitoba Health Information Privacy Committee (HIPC No. 2015/2016-31) and the University of Manitoba Health Research Ethics Board (Ethics File No. H2015:355 (HS18922)).

Conflict of interest

The authors declare that they have no conflict of interest.