Maternal psychiatric illness is a well-documented risk factor for child psychiatric disorders. Most generational studies of psychiatric illness employ "top-down" sampling—that is, identifying ill parents and then assessing their children. Researchers have found that children of depressed or anxious parents are themselves at a substantially increased risk (two- to five-fold) of psychiatric disorders (1,2,3,4,5,6). A relatively small number of "bottom-up" studies (for example, sampling from ill children) demonstrate that first-degree relatives of children with psychiatric disorders are themselves at increased risk of depression and anxiety (7,8,9,10). Little is known, however, about rates of psychiatric diagnoses among mothers who bring their children to pediatric mental health clinics or the impact of maternal illness on child treatment outcomes.
The relationship between maternal and child psychiatric illness is complex and multifactorial. Contributing risk factors include shared genetic vulnerabilities (11) and shared environmental factors such as poverty (12) and domestic violence (13). Risk may be moderated by factors such as the father-child relationship (14) and child temperament (15). Untreated maternal psychiatric illness is likely to affect children with psychiatric illness both directly (for example, by impaired parenting and decreased empathy) and indirectly (for example, by affecting the quality of mental health treatment received by their children). Notably, depressed and anxious mothers tend to overreport psychiatric symptoms among their children (16), leading to misinformation about child psychopathology with the attendant risks of overtreating the child and undertreating the mother. Untreated maternal illness, in turn, has been associated with poor outcomes among children receiving psychiatric treatment (17,18). In addition, impairment in maternal functioning (psychological well-being, emotional control, and social support) is associated with increased use of child health services (19). Thus untreated maternal psychiatric illness not only may lead to worse outcomes for children treated in mental health settings but simultaneously may increase child health care utilization.
To our knowledge, there is only one previous study of psychiatric illness among mothers who bring their depressed offspring for psychiatric treatment (20). In that study, investigators from Columbia University found elevated rates of current psychiatric diagnoses among mothers who brought their children to an urban research mental health clinic and low rates of maternal psychiatric treatment. Their bottom-up sampling procedures did not permit the investigators to elucidate the relationship between maternal and child illness by examining such questions as: Which comes first? Does parenting an ill child contribute to increased maternal psychopathogy? Do children become ill because of genetic predispositions for these disorders? From a public health perspective, however, this approach approximated the screening process one might use in pediatric mental health settings to identify—and ultimately treat—ill mothers of children who are receiving psychiatric treatment. Thus bottom-up methods are especially relevant to clinicians and health care policy makers interested in improving care for families in pediatric clinical settings.
To further elucidate the relationship between maternal and child psychiatric illness, we reported results from a study designed to assess rates of lifetime and current psychiatric diagnoses by using clinician-administered diagnostic interviews among mothers who brought their school-age children to a rural community pediatric mental health clinic. As in the Columbia study (20), we elected to use bottom-up sampling techniques so that the results might inform health care interventions in this setting. We also assessed the relationship between maternal diagnosis and occurrence of child psychopathology and characterized the functional impairments of mothers with psychiatric diagnoses.
Potential participants included biological or adoptive mothers, aged 18 years and older, bringing their nonpsychotic children, aged six to 17 years, for evaluation to a rural community pediatric mental health clinic. Mothers were excluded from the study if they did not have custody of the child, were not currently living with the child, or were at serious risk for child abuse or neglect. Mothers were also excluded if their child required current inpatient hospitalization or needed services for mental retardation only. On the basis of these criteria, 496 participants were identified as eligible for inclusion during the study period (June 30, 1998, to February 1, 2001). Of the 496 eligible mothers, 222 (44 percent) ultimately entered the study. Fifty-one (10 percent) refused permission to contact them. The remainder failed to schedule or to attend the research interview.
We had no information on the mothers who chose not to participate. Very limited data on 140 children of mothers who were eligible but nonparticipating showed no significant differences in gender or race between participant and nonparticipant children, although nonparticipant children were significantly older (mean age=12.8) than participants (mean age=11.8) (t=-2.79, df=376, p=.006) (unpublished data, Greeno C, Sales E, Anderson C, et al, 2005).
All procedures were approved by the institutional review board of the University of Pittsburgh. Mothers were approached initially by their children's clinicians. After describing the study to participants, written informed consent was obtained from mothers and informed consent or assent was obtained from children. Mothers who completed the assessments received a $50 check. Child participants received a $20 gift voucher to a local mall.
Demographic data and psychiatric treatment history were recorded on standardized research forms. Maternal lifetime and current psychiatric diagnoses were assigned by using the Structured Clinical Interview for DSM-IV, Clinician Version (SCID) (21). Maternal depressive and anxiety symptoms were assessed with the Beck Depression Inventory (BDI) (22) and Beck Anxiety Inventory (BAI) (23), respectively. Individuals who score 10 to 18 on the BDI (24) or BAI (23) are considered to have symptoms in the mild to moderate range, and individuals who score 19 to 29 are considered to have symptoms in the moderate to severe range. Maternal functioning and disability were assessed by using the SF-36 Medical Outcomes Study Questionnaire (MOS) (25) and Sheehan Disability Scale (SDS) (26). Because the participants were a nonclinical sample, the physical and mental component summary scores of the MOS are reported as T scores having a mean of 50 and a standard deviation of 10.
We administered the Interpersonal Support Evaluation List (ISEL) (27), a 48-item instrument that measures four dimensions of support: tangible (instrumental or material support), appraisal (availability of someone to talk to about problems), self-esteem (positive comparison of oneself with others), and belonging (people one can do things with). Norms for the four subscales (12 items each) range from 32.9±5.0 to 34.4±6.0 (28). We also administered the 50-item Partner Abuse Scale (PAS) (29) to detect current domestic violence, a risk factor associated with mental illness (30). Reported mean±SD scores for a general population are .90±5.13 for physical and 6.94±12.29 for verbal abuse. If participants were not in a relationship over the past two years, they did not complete this form.
Child psychiatric diagnoses were assigned by using the Schedule for Affective Disorders and Schizophrenia for School Age Children, Present and Lifetime version (K-SADS-PL) (31,32), a semistructured psychiatric interview that yields DSM-IV diagnoses and has been widely used in studies of childhood psychiatric disorder. The interview is administered separately to both parent and child with the interviewer combining parent and child information. We assessed child behaviors and social-emotional functioning with the Child Behavior Checklist (CBCL) (33). The CBCL is a parent-report measure that produces nine clinical problem scales that are, in turn, combined to produce both internalizing and externalizing problem scales. These scales are normed separately for boys and girls aged four to 11 years and 12 to 18 years. The measure contains 118 items that parents rate for their child as 0, not true in the past six months; 1, somewhat or sometimes true; or 2, very true or often true. We report scores on the problem scales as T scores having a mean of 50 and a standard deviation of 10 (33).
Mother and child interviews were administered by master's-level clinicians trained and certified in the administration of the SCID and the K-SADS-PL in accordance with the standards of the biometrics division of the New York State Institute. Interviewers achieved 100 percent concordance with a certified SCID rater for primary diagnosis and presence or absence of comorbidity on four consecutive, taped interviews.
Descriptive approaches were used to present data characterizing frequencies of diagnoses and demographic variables. Chi square analyses (for categorical variables) or independent-sample t tests (for continuous variables) were used to compare mothers with and without a diagnosis on measures of symptom severity, impairment, social support, and the diagnoses and CBCL scores of their children. Two-tailed significance alpha was set at .05.
Demographic characteristics and referral source
Demographic characteristics of mothers are shown in t1. Participants were mostly Caucasian, reflecting the ethnic composition of the rural county where this research was conducted. The mean±SD age of the children at entry to the study was 11.8±3.4 years. Of the 222 children, 92 (41 percent) were female.
The most common source of referral to the mental health clinic was the family itself (82 participants, or 37 percent). Other sources of referral included the school system (40 participants, or 18 percent), juvenile justice (four participants, or 2 percent), child protective services (16 participants, or 7 percent), and a primary care provider or pediatrician (27 participants, or 12 percent). Thirty-four participants (15 percent) endorsed the category of "other" source of referral; information on referral source was missing for 19 participants (9 percent).
Lifetime and current maternal psychiatric diagnoses
As can be seen from t2, 79 percent of mothers interviewed met SCID criteria for at least one lifetime axis I diagnosis; 61 percent met criteria for at least one current axis I diagnosis. A quarter of the sample (N=57) met criteria for both a current depressive disorder and a current anxiety disorder.
Mothers with and without a current diagnosis
t3 compares mothers who met criteria for at least one current axis I disorder (N=135) with mothers who did not meet criteria for any current axis I diagnosis (N=87) on measures of functioning, social support, and symptoms. No significant differences were found on any demographic variable (age, number of children, education, marital status, ethnicity, employment status, or household composition) between mothers with a diagnosis and mothers without a diagnosis. As shown in t3, the mothers with a diagnosis had significantly worse scores on all measures of symptoms, functioning, and disability. Depression (BDI) and anxiety (BAI) symptom scores of mothers with a diagnosis were in a clinically significant range (23,24). Among the mothers with a current diagnosis, only 44 (33 percent) were currently receiving psychiatric treatment. Three of the mothers without a diagnosis (3 percent) were receiving psychiatric treatment.
Children of mothers with and without a diagnosis
t4 compares children of mothers with a diagnosis with children of mothers without a diagnosis. Groups did not differ in age or gender. Of 221 children, 167 (76 percent) of those who completed the K-SADS-PL met criteria for at least one current axis I disorder. Compared with children of mothers with no diagnosis, children of mothers with a diagnosis were more likely to meet criteria for at least one current axis I disorder (Fisher's exact test, two-sided, p=.02). They also met criteria for significantly more diagnoses on the K-SADS-PL and had significantly higher scores on several problem scales of the CBCL.
t5 compares rates of child psychopathology in offspring of mothers with and without psychiatric illness. A trend toward a higher rate of internalizing disorders was found among offspring of mothers with any diagnosis compared with offspring of mothers with no diagnosis (44 percent compared with 28 percent; χ2=3.6, df=1, p=.57). By contrast, no significant difference in rates of externalizing disorders was found between offspring of mothers with or without a psychiatric diagnosis (56 percent compared with 65 percent, respectively). Mothers with a diagnosis of depression or anxiety were significantly more likely to have children with internalizing disorders than mothers without a diagnosis of depression or anxiety (χ2=6.0, df=1, p=.01), but no significant difference was found between the two groups of offspring in rates of externalizing disorders.
This study had many limitations. It is impossible to determine causality from these data: did the child become ill because of the mother's psychiatric disorder, or do some vulnerable mothers become ill in the face of caring for sick offspring? We also did not assess fathers in this study. Fathers are clearly important, and a good father-child relationship may mitigate the effects of maternal depression on child outcomes (14). We were able to enroll less than half those eligible for this study, and we have few data available on individuals who did not participate. A 50 percent "no show" rate is typical in community clinics, and our research recruitment reflects this broad clinical problem. Nevertheless, low participation rates raise questions about the estimates of illness prevalence in this sample.
Maternal psychiatric illness itself is likely to affect the quality of the data. Child assessments such as the CBCL rely on maternal report of child symptoms, and previous reports show that depressed and anxious mothers overendorse symptoms in their children (16). However, in our study reports on the CBCL converged with the relatively objective K-SADS-PL assessment, which suggests that the higher rate of child psychopathology in offspring of ill mothers was not simply an artifact of maternal reporting. Eighteen percent of children had no diagnosis on the K-SADS-PL, regardless of whether or not their mother had a diagnosis of psychiatric illness. The same interviewer evaluated both the mother and her child, so assessments were not blind to the clinical status of the other member of the dyad. This fact represents another potential source of bias in our study. In addition, the number of tapes (four) used to establish interrater reliability on the SCID and the KSADS may have been insufficient to guarantee agreement among interviewers.
Despite these limitations, we found an exceptionally high prevalence of psychiatric disorders in a nonclinical population of adult females. More than half the mothers (61 percent) assessed met DSM-IV criteria for at least one current axis I disorder, and more than three-fourths (79 percent) met criteria for lifetime psychiatric illness. Interestingly, our group found almost twice the rate of maternal psychiatric illness as detected in the Columbia study by Ferro and colleagues (61 percent compared with 31 percent) (20). This discrepancy may be explained by prominent differences in inclusion criteria of the two studies (children with any nonpsychotic disorder in our study compared with children with depression in the Columbia study), recruitment setting (community clinic versus research clinic), diagnostic instrument (SCID versus the self-report PRIME-MD Patient Problem Questionnaire), and demographic composition (rural Caucasian compared with urban Latina). Despite differences in samples and methods, it is notable that both groups found high rates of untreated depression and anxiety among the mothers of children who had psychiatric illnesses.
Mothers with diagnoses showed marked impairments in functioning across domains, typically in clinically significant ranges (25,26). They also experienced significantly higher levels of partner abuse (both physical and verbal) and lower levels of social support than the mothers without a diagnosis. Studies indicate that marital conflict is highly stressful for women, even when they are in good relationships (34), and social support is important to women (35). It is likely that mothers with depression or anxiety will have difficulty navigating interpersonal problems (for example, extricating themselves from abusive relationships) and that the interpersonal problems they experience (especially marital conflict and absent social support) will exacerbate their psychiatric symptoms. Thus these mothers experience a range of difficulties that might have an impact on their capacity to optimally manage children suffering from their own psychiatric illnesses.
In support of the hypothesis that maternal psychiatric illness is associated with greater levels of child illness in this population, we found that children of mothers with current axis I disorders were more likely to experience higher levels of psychopathology themselves: they met criteria for significantly more diagnoses on the K-SADS-PL, experienced higher levels of anxiety and depressive symptoms, were more withdrawn, and complained of more somatic symptoms than children of mothers without a current axis I disorder. This finding is especially remarkable given the fact that all children in this sample were being brought to the clinic for psychiatric evaluations. Thus, even among a group of children seen in a mental health clinic, having a mother with psychiatric illness predicts greater symptom burden on the child. Interestingly, offspring of mothers with a psychiatric illness (typically depression or anxiety in this sample) were at greater risk of internalizing—but not externalizing—disorders. This may reflect the well-documented genetic link between internalizing disorders across generations.
Despite clinically significant levels of symptoms and impairment, two-thirds of mothers with a diagnosis were receiving no psychiatric treatment. In addition to contributing to needless suffering, the economic costs of untreated psychiatric illness are high (36). Untreated maternal illness leads to lost productivity of mothers (37) and higher utilization of public resources by their children (19). In addition, having a mother with psychiatric symptoms predicts poor response to child psychiatric treatment (17,18).
Because this was not an epidemiologic study, we cannot conclude that rates of maternal psychiatric illness detected in this sample are representative. In fact, on the basis of evidence that suggests that mothers with depression and anxiety are more likely to seek treatment for their children than nondepressed mothers with anxiety (19), we suspect that the rates of psychiatric illness in our sample may be higher than in mothers of ill children who are not brought in for treatment. From a clinical and services perspective, pediatric mental health clinics may represent an underutilized setting in which to identify and possibly treat mothers with undetected psychiatric illness.
In a pediatric mental health clinic, we identified very high rates of untreated psychiatric illness and a high degree of functional impairment among mothers who brought their children for evaluation. We also found greater levels of psychopathology in the offspring of ill mothers. Although their own depression and anxiety may have driven mothers to bring their offspring for treatment, the majority of these mothers did not obtain treatment for themselves. Untreated maternal illness may in turn jeopardize treatment outcomes in children. These findings underscore the importance of developing integrated treatment and service structures that conjointly address the needs of children with psychiatric disorders and the mothers with psychiatric disorders who care for them.
This study was supported by grants R24-MH-56848, P30-MH-30915, and K23-MH-64518 from the National Institute of Mental Health. The authors thank Howard Stein, B.A., Susan Wheeler, R.N., Barbara Kumer, M.S., Gloria Klima, M.A., Seth Duncan, B.S., Patricia Houck, M.S.H., and Genevieve Barrow, M.S., for their assistance.
Dr. Swartz, Dr. Shear, Ms. Sullivan, and Dr. Ludewig are affiliated with the department of psychiatry of the University of Pittsburgh School of Medicine, 3811 O'Hara Street, Pittsburgh, Pennsylvania 15213 (e-mail, firstname.lastname@example.org). Dr. Wren is affiliated with the department of child and adolescent psychiatry at Stanford University in Palo Alto, California. Dr. Greeno and Dr. Sales are affiliated with the School of Social Work at the University of Pittsburgh. This paper was presented in part as a poster at the annual meeting of the American Psychiatric Association held May 5 to 9, 2001, in New Orleans.
Demographic characteristics of 222 mothers who brought their children for mental health treatment
Lifetime and current psychiatric disorders among 222 mothers who brought their children for mental health treatment
Scores on clinical measures of mothers who brought their children for mental health treatment, by whether or not they had a current axis I disorder
Clinical and demographic characteristics of children of mothers with and without a current axis I diagnosis
Relationship between maternal and child psychiatric diagnoses among mothers who brought their children for mental health treatmenta
aPercentages were calculated by using the number of children in each diagnostic category as the numerator and the number of mothers in each category as the denominator: that is, 44 percent of mothers with at least one diagnosis had a child with an internalizing disorder.