Caregiving Attitudes and At-Risk Maternal Behavior Among Mothers With Major Mental Illness

Mental health and child welfare personnel are increasingly called on to assess the caregiving abilities of individual parents with major mental illness (1). Such assessments often occur in the context of custody hearings after allegations of child maltreatment. Although such evaluations can have far-reaching consequences, much of the current parenting assessment methodology has not been empirically linked to actual parenting behavior (2).

Optimal parenting risk assessments include evaluations of behavior, a direct correlate of maltreatment. Ideally, such assessments draw on multiple sources of information and evaluate interrelated domains of parenting as well as other factors, such as support and stress, that are known to contribute to parenting capability (3). For parents with psychiatric disorders, illness-related variables, such as symptom severity and treatment response, should also be examined. As part of the effort to improve parenting assessment, our study isolated one potentially important parenting domain—attitudes about caregiving—to see whether it measurably related to insensitive maternal behavior. We subsequently explored whether the association to insensitive maternal behavior was due to specific types of caregiving attitudes: role reversal, physical punishment, or intensified self-sufficiency.

Having attitudes about one's children that are grossly out of proportion to their capabilities has been posited to be a risk factor for maltreatment among parents without mental illness (4). Although unrealistic caregiving attitudes are more prevalent among individuals with major psychiatric disorders than among those without such disorders (5), their relationship to maternal behavior has not been studied in this population. Our primary objective in this study was to determine whether the Parenting Opinion Questionnaire (POQ) could be a useful tool to apply in evaluations of parenting competence among mothers with mental illness.

Methods

Participants included 44 mothers who were undergoing a multimodal parenting assessment. Mother-child dyads were included in the study if the mother was aged 20 years or older, if the mother had at least one hospitalization within the past five years for a DSM-IV psychiatric illness other than mental retardation or substance abuse or dependence, if there was a case of child maltreatment or risk of harm to the child by the mother for which there was enough evidence to warrant custody loss, if the mother did not have current custody of the child, if the child was between the ages of eight months and four years, and if the mother had at least weekly contact with the child for the past year. Mothers were excluded from the study if they met the DSM-IV criteria for mental retardation (6).

The study group reflects the current priorities of the child welfare system (7). Adult mothers were targeted because they may have different parenting issues than fathers and adolescent parents (2). Mothers who had at least one previous hospitalization were sampled because they struggle with more severe mental illness than women who have never been hospitalized. Young children were targeted because they make up a sizable proportion of victims of maltreatment (7). Weekly visits and a minimal child age of eight months were required to ensure that the mother and the child knew each other and had enough contact to maintain a relationship.

The POQ is used to assess maternal attitudes about children, from infancy to adolescence. The POQ has good validity and demonstrated adequate test-retest reliability (4). POQ items require parents to agree or disagree with different attitudes about caregiving and children. The outcome measure is a continuous score that assesses the total number of unrealistic attitudes that mothers endorsed on the POQ.

This study used an abbreviated version of the POQ, because the study included only young children. By excluding items that related to children older than age five, the original 80-item POQ was modified to a 32-item scale.

As our observational measure of maternal behavior we used the Crittenden CARE Index (8). This index correlates with childrearing status—for example, abusing, neglecting, and adequate caregiving—and has been validated with families from different ethnic backgrounds and social classes.

Mother-child dyads were videotaped for three minutes in a clinical setting where they were asked to play as they usually do. Maternal behaviors—for example, facial expression, vocal expression, and body contact—were scored as 0, insensitive; 1, partially sensitive; or 2, sensitive. Summing the scored items yielded an overall score; possible scores ranged from 0 to 14, with 14 indicating a high level of maternal sensitivity and 0 indicating a high level of maternal insensitivity. Two independent and trained raters scored the 44 videotapes. Interrater reliability was high (kappa=.73).

DSM-IV criteria (6) were used for the mothers' psychiatric diagnoses, which were coded by using written reports of psychiatric interviews that were conducted by the third author. To control for confounding variables, maternal educational level, age, ethnicity, and number of children were also coded from the same reports. Spearman correlation coefficients were used to determine the associations between variables.

Results

The mothers' age ranged from 19 to 44 years. Sixteen mothers (36 percent) had not completed high school, 11 (25 percent) had a high school degree, and 17 (39 percent) had some post-high-school education. Twenty-four participants (55 percent) were African American, 16 (36 percent) were Caucasian, three (7 percent) were Hispanic, and one (2 percent) was Asian American. Eighteen (41 percent) had been given a DSM-IV diagnosis of schizophrenia or schizoaffective disorder. Eight (18 percent) had bipolar disorder, seven (16 percent) had major depression, recurrent, and seven (16 percent) had a personality disorder. The remaining four participants (9 percent) had another diagnosis (psychosis not otherwise specified or an explosive disorder).

Children ranged in age from eight to 48 months and were divided evenly by sex (22 boys and 22 girls). In 19 cases (43 percent), the child was removed from the mother's custody because of founded allegations of child neglect. In 13 cases (30 percent), the mothers had physically harmed their child. In 12 cases (27 percent), the child was removed because his or her mother was judged to be at substantial risk of harming her child.

No relation was found between POQ scores and maternal diagnosis, age, ethnicity, education, or number of children. However, POQ scores were associated with maternal behavior (r=-.35, p<.01). Mothers with higher overall scores on the POQ, indicating a higher number of unrealistic caregiving attitudes, were more prone than other mothers to show insensitive behavior when interacting with their children.

We subsequently examined whether insensitive maternal behavior was associated with specific caregiving attitudes (9). To this end, we categorized the POQ questions into one of three clinically useful groups: role reversal: expecting young children to support a parent; intensified self-sufficiency: expecting young children to fend for themselves without support; and physical punishment: endorsing harsh physical punishment as a way to discipline children.

Insensitive maternal behavior was associated with role reversal (r=-.27, p<.04). The relationship between maternal behavior and self-sufficiency approached statistical significance. Insensitive maternal behavior was not associated with attitudes endorsing physical punishment. Because mothers who are involved with child protection services are told not to use physical punishment, it is possible that these items were not endorsed because they were seen as socially undesirable.

Discussion and conclusions

The findings indicate that the POQ may be a useful tool to apply in evaluations of parenting competence among mothers with mental illness. Some limitations of the measure should nonetheless be underscored. Most important, maternal caregiving attitudes are only a second-order correlate of maltreatment and should not replace first-order correlates (direct observations). It follows that the POQ should not be used as a sole indicator of at-risk parenting. Optimally, parenting evaluations should include both direct observations and other tools that assess interrelated domains of parenting, mental illness, and environmental factors that can influence parenting capability.

Another limitation to note is that the POQ measure, as used in this study, focused on caregiving attitudes about young children. More research is needed to establish whether caregiving attitudes about older children are also linked to insensitive maternal behavior. More research is also needed to determine whether scores on the POQ predict insensitive caregiving among mothers with mental illness who have not lost custody of their children.

Dr. Leventhal is affiliated with Integrated Development Services, 1025 West Glen Oaks Lane, Suite 207, Mequon, Wisconsin 53092 (e-mail: [email protected]). Dr. Jacobsen is with the School of Social Work at the University of Illinois at Urbana-Champaign. Dr. Miller is with the department of psychiatry and Dr. Quintana is with the department of public health at the University of Illinois at Chicago.