Identifying Depression in a National Sample of Caregivers Investigated in Regard to Their Child’s Welfare

Parental depression disproportionately affects economically and socially disadvantaged populations and poses significant risks to family functioning and child well-being (1–4). Depressed parents and other caregivers are more likely to behave aggressively toward their children and use more disengaged, withdrawn, or intrusive parenting styles (5–8). They are also less likely to adhere to or benefit from recommended treatment programs and to bring their children for recommended preventive health services (9–12). The cumulative negative effect on child well-being can be severe. Compared with children whose caregivers are not depressed, children of depressed caregivers exhibit decreased social competence, experience more emotional and behavioral problems in adolescence and early adulthood, and incur greater health expenditures (13–16).

Children involved with child protective services (CPS) are at particularly high risk of being adversely affected by parental depression because their parents often experience domestic violence or other prior trauma that can exacerbate effects of parental depression on child safety and well-being (17,18). For many of these families, CPS involvement represents an opportunity to connect with needed services. Evidence-based treatment can reduce or eliminate symptoms of depression (19–22). Remission of parental depression has in turn been linked to improvements to children’s behavioral health symptoms and functioning (23–25). Unfortunately, studies of community samples suggest that only 30% of depressed adults receive any treatment for their illness (26). Among parents and other caregivers involved with child welfare, treatment rates may be as low as 20% (27).

Identifying caregiver depression is the first step in provision of treatment across service settings but can be challenging for child welfare caseworkers. Depression can manifest as anxiety or somatic conditions, which caseworkers generally receive little or no training in how to identify (28–31). Fear of losing custody of their children may also affect caregivers’ willingness to report mental health needs (32,33). As a result of these and other factors, caregiver depression is often underidentified in a child welfare investigation or an assessment process (34).

To provide policy makers and administrators with information useful in improving the frontline investigation, this study examined agency practices, caseworker attributes, and case characteristics related to child welfare caseworker identification of parental depression. Child welfare agency practices hypothesized to positively affect caseworker identification of depression include use of standardized mental health assessments in the investigation process and strength of collaboration with local mental health providers. Standardized mental health screening and assessment tools can improve diagnosis of parental depression and aid in developing appropriate treatment plans (35). Although the effect on caregivers has not been previously examined, child welfare agency collaboration with local mental health providers has been shown to improve children’s access to mental health services (36,37). Prior research has also found that colocation of child welfare caseworkers and credentialed alcoholism and substance abuse counselors can facilitate identification of caregiver substance use and subsequent treatment referrals (38).

Caseworker attributes hypothesized to affect identification of caregiver depression include education, job tenure (experience), and caseload (balancing competing demands). Although many caseworkers lack the clinical and communication competencies necessary to detect caregiver depression (28), caseworkers with a formal education in social work or psychology may be better prepared to investigate complex family needs than caseworkers without such training (39). Job tenure and caseload may also influence identification of caregiver needs through their effects on caseworkers’ ability to effectively engage with families and respond to their needs (40). Manifestation of depression can be subtle, and eliciting candid information about mental health status can be challenging. These skills may evolve over time as caseworkers gain experience working with families; hence, investigative caseworkers’ years in child welfare are hypothesized to be positively associated with detection of caregiver depression. In contrast, caseworkers with excessive caseloads may overlook caregiver depression because they are forced to conduct superficial investigations (41). Thus, in this study, investigative caseworker caseload was hypothesized to be negatively associated with caseworker identification of caregiver depression. Finally, at the case level, maltreatment reports initiated by health care providers were hypothesized to increase identification of caregiver depression because such professionals are more likely to routinely screen for depression and to communicate concerns about caregiver mental health needs to caseworkers (42–44).

Methods

Data sources

Data were drawn from the National Survey of Child and Adolescent Well-Being (NSCAW), the only national, longitudinal study of families subject to maltreatment investigations or assessments conducted by CPS (45). NSCAW includes two cohorts of children spaced approximately ten years apart. Given significant changes in child welfare practices over the past decade, we used data from only the second cohort (NSCAW II). A two-stage cluster-sampling approach was used in which primary sampling units were first selected, with each corresponding to the geographic area served by a single CPS agency. Within these sampling units, a total of 5,873 children from infancy to age 17.5 years were sampled from all CPS investigations or assessments conducted between February 2008 and April 2009.

Baseline (wave I) data collection occurred between March 2008 and September 2009, on average approximately four months after the completion of the child welfare investigation. Information on child welfare agency practices was obtained through structured interviews with directors of each of the 86 child welfare agencies in the NSCAW II sample. Detailed assessments of family context and well-being were collected through structured, face-to-face interviews with current caregivers, children, and their investigative caseworkers. To assess their mental health status, permanent primary caregivers whose children initially remained in their home after the investigation were administered a series of validated instruments via audio-computer-assisted self-interview, a methodology shown to increase reporting of sensitive behaviors (46).

Investigative caseworkers were separately asked to provide information on their background and work practices, including caseload, as well as their assessment of a given family’s behavioral health; this entailed referring to families’ case records to increase information accuracy (45). Finally, maltreatment report source and type were obtained from the National Child Abuse and Neglect Data System; these administrative data were linked to NSCAW with families’ permission (47). [Additional information on NSCAW procedures is provided in the online data supplement to this article and is extensively described elsewhere (45,48).]

Given our interest in identifying parental depression, we restricted the study sample to permanent, primary caregivers whose confidential, self-reported responses to a validated screening instrument indicated major depression within the past 12 months. A total of 889, or 23%, of permanent, primary caregivers met these criteria.

Measures

Caregiver depression.

Caregiver depression was assessed with the World Health Organization Composite International Diagnostic Interview Short Form (CIDI-SF) (49), a standardized interview that screens for mental disorders with criteria established in DSM-IV (50,51). The CIDI-SF has been validated for use with the general population, with an overall diagnostic classification accuracy of 93% for major depressive episodes (49). Consistent with the suggested CIDI-SF scoring method (52), caregivers were classified as having a major depressive episode if they reported a two-week period during the preceding 12 months in which either a dysphoric mood or anhedonia was experienced to a significant degree as well as at least three other symptoms of major depression.

Identification of caregiver mental health treatment needs.

Investigative caseworkers’ identification of caregiver mental health treatment needs was operationalized as a dichotomous variable, set to 1 if the investigative caseworker indicated that the permanent primary caregiver needed services for “an emotional, psychological, or other mental health problem like depression, bipolar disorder, schizophrenia, etc.” in the past 12 months and set to 0 otherwise.

Agency practices.

Agency use of standardized mental health assessments was indicated by a dichotomous variable, set to 1 if the director responded yes to a question asking whether the child welfare agency “has used a standardized mental health assessment for parents during the investigation process” and set to 0 if the agency director responded no. Consistent with literature suggesting that having a combination of different interorganizational arrangements reflects stronger overall collaboration between agencies (53,54), strength of ties was measured as the number of distinct relationships each child welfare agency director reported having with mental health providers. Examined relationships included having a memorandum of understanding or other formal interagency agreement, cross-training of staff, joint budgeting or resource allocation, and child welfare agency colocation with a mental health partner.

Caseworker attributes.

Caseworker education was indicated by whether the caseworker’s highest degree was a non–social work bachelor’s degree or less (referent) or a bachelor of social work (B.S.W.), a master of social work (M.S.W.), or other graduate degree. Caseworker experience was operationalized as the number of years the caseworker reported working in child welfare. Caseload was defined as the average number of new investigations per month assigned to the caseworker.

Case characteristics.

Maltreatment report source was categorized according to whether the initial maltreatment allegation was made by medical or mental health personnel, or by another type of reporter, such as an educator, a social service worker, or a nonprofessional (47). We also controlled for several other factors that might affect caseworker ability to identify caregiver depression, including the most serious type of alleged maltreatment (6,55); the presence of a co-occurring substance use disorder (56–58); the caregiver’s race-ethnicity, including African American, Hispanic, or other, which included American Indian or Asian, Hawaiian, or Pacific Islander (59); the caregiver’s biological or functional relationship with the child (60); and child age (26). Caregivers were identified as having a co-occurring substance use disorder if their confidential self-reported responses on either of two validated instruments, the Alcohol Use Disorders Identification Test (AUDIT) or the Drug Abuse Screening Test (DAST-20), indicated harmful use or dependence (61,62). Caregivers whose AUDIT scores were 5 or higher or whose DAST-20 scores were 6 or higher were considered to meet this criterion (63–66).

Analyses

Although NSCAW data have a hierarchical structure, our sample did not meet criteria for use of multilevel modeling (67–69). Consequently, we conducted a single-level logistic regression model using the Stata 12.0 “svy” module, which accounts for probability weights and stratification as well as correlations in outcomes across families served by the same child welfare agencies. Phi and biserial correlations between independent variables were all less than .4 and variance inflation factors were less than 2.5, below the threshold at which multicollinearity might be a concern in logistic regression. Twenty multiply imputed data sets were created with the multivariate normal imputation method within the Stata 12.0 “MI” module to handle missing data. Weighted t tests conducted after imputation did not reveal any statistically significant differences between imputed and nonimputed variables; therefore, only imputed results are provided below.

Results

As shown in Table 1, although all of the caregivers in the sample had CIDI-SF scores indicating a major depressive episode in the past 12 months, investigative caseworkers identified mental health needs of only 38% of those caregivers. Twenty-five percent of caregivers were investigated by CPS agencies whose directors reported agency use of a standardized mental health assessment during investigation. For approximately 11% of cases in our sample, the initial maltreatment report was made by medical or mental health providers. Most of these cases (88%) were reported by medical rather than mental health providers, and over half involved children less than one year of age.

To supplement multiple logistic regression results in Table 1, we also calculated predicted probabilities for all statistically significant variables to illustrate the magnitude of the association with caseworker identification of caregiver depression (70). In keeping with our hypothesis, agency use of a standardized mental health assessment for parents in the investigation process was associated with a 45% higher probability that caseworkers would identify caregiver depression. Also as hypothesized, each additional case assigned per month to investigative caseworkers decreased by 1% their probability of identifying caregiver depression. The probability of identifying caregiver depression increased by 21% when the initial maltreatment report was made by medical or mental health providers rather than by social workers, law enforcement, educators, or nonprofessional sources (including relatives, friends, and neighbors). Contrary to our hypothesis, child welfare agency ties with mental health providers were not associated with detection of depression. Finally, among the measures included as covariates, caseworkers’ probability of identifying depression increased by 22% when caregivers were American Indian, Asian, Hawaiian, or Pacific Islander rather than non-Hispanic white.

Discussion

To the best of our knowledge, this study was the first to examine factors associated with child welfare caseworker identification of caregiver depression. Data from this large, national sample of children referred to CPS indicate that even among families deemed sufficiently low risk to retain custody of their children, rates of major depression are disproportionately high: 23% of caregivers had CIDI-SF scores indicating a major depressive episode within the past 12 months, a rate more than three times higher than the estimated 12-month prevalence of major depression in the general population (71). Identification of caregiver depression is a critical first step to connecting these vulnerable families to appropriate treatment. However, identifying depression may be challenging for caseworkers, who identified mental health needs for less than 40% of caregivers in our sample. This finding is consistent with prior research suggesting substantial underidentification of mental health needs among families involved with the child welfare system (27,72) and represents a missed opportunity for intervention.

Efforts to improve identification of depression in this population must take into account the current child welfare context. Caseworkers face significant pressure to close cases quickly, which may limit their ability to detect and meaningfully respond to families’ service needs (41,73). Their training in mental health is also often inadequate, further limiting their ability to identify parental depression (28,73). Given these constraints, validated screening instruments may help caseworkers quickly identify families in need of further assessment. Our findings indicate that caseworkers were more likely to identify caregiver depression when agencies used standardized mental health assessments during the investigative process. These findings are consistent with research in health care settings indicating that clinicians are more effective at detecting depressive symptoms when using structured clinical interviews or screeners (74–76) and reinforce the importance of such tools in identifying mental disorders. Most screening and assessment tools currently used during child welfare investigation do not directly address caregiver mental health or do so only through a single, binary item (77,78). Brief, valid, and easily scored and interpreted measures of depression, such as the Patient Health Questionnaire (79–81), have been developed that could be readily incorporated into the investigation process and administered by caseworkers and other personnel during office, clinic, or home visits. Inclusion of such measures in agency policies may play an important role in their use.

Although prior research has found that the strength of child welfare agency ties with local providers is associated with families’ service access and outcomes (36,37,82), such ties were not significant in this study. It is possible that practices such as cross-training and colocation are important for facilitating service access and coordinating care only after need has already been identified. NSCAW data do not include information on the purpose of collaborative ties; further research is needed to examine the extent to which collaboration with local mental health providers may be of use to caseworkers during their investigations.

The findings do, however, suggest the important role that medical and mental health professionals can play in identifying parental depression (42). These professionals are more likely to have the clinical competencies necessary to identify depression and may communicate this information when reporting child maltreatment. Although such professionals report a relatively small proportion of maltreatment cases (8% nationally) (83), it is also possible that they encounter the highest-risk families. Further research is needed to understand associations between maltreatment report source and caseworker identification of parental depression.

Although the magnitude of association between caseload and caseworker identification of depression was small, findings are also consistent with prior research suggesting that high caseloads can compromise casework practices (41). Constrained resources and high turnover make it difficult for agency managers to reduce the number of cases assigned to each worker; however, the high potential costs of not identifying families’ needs argue for making adequate staffing a high priority.

The findings also indicate the need for attention to potential effects of caregiver race-ethnicity on caseworker identification of mental health needs. In this study, investigative caseworkers were significantly more likely to identify depression among American Indians, Asians, Hawaiians, or Pacific Islanders than among non-Hispanic whites. Although the explanation for differential identification of depression in this subgroup is unclear, prior research has demonstrated disparities in mental health access among Asians as well as non-Latino whites, African Americans, and Latinos (84). Additional research is needed to better understand this phenomenon and its potential effect on service access.

Several limitations constrain this study. First, we relied on caregiver responses to the CIDI-SF to identify depression. Although it is validated for use with the general population (49), this instrument was developed for screening rather than for diagnostic purposes, and it has not been specifically tested with child welfare or minority populations. Contextual or cultural differences may affect the sensitivity and specificity of results; for example, caregivers involved with child welfare may conceal certain behaviors, warranting the use of lower cutpoints than those applied here. In addition, only the depression scales of the CIDI-SF were included in NSCAW, precluding the ability to examine a wider range of mental disorders. Given that caseworkers were asked whether caregivers had any “mental health problem like depression, bipolar disorder, schizophrenia, etc.” but caregivers were asked only about depression, we were also unable to determine the accuracy with which caseworkers identified caregiver depression. Nonetheless, we believe that focusing on major depression is still critical given its high prevalence among caregivers involved in child welfare and its impact on both caregivers and children. Finally, NSCAW did not ask caseworkers about training in the use of standardized mental health assessments, which could also have affected how effectively they used these tools.

Conclusions

Child welfare investigations focus on child safety but also represent important opportunities to identify and address family needs. Parental depression is underidentified in this high-risk population. Use of standardized screening and assessment tools may improve caseworker identification of caregivers’ mental health needs. Inclusion of such measures in agency practice directives could help instantiate their use.