Validity of the Parent Young Mania Rating Scale in a Community Mental Health Setting

Abstract

The objective of this study was to examine the potential of using the Parent Young Mania Rating Scale (P-YMRS) to distinguish pediatric bipolar disorder from other psychiatric conditions. The design of the study was a retrospective chart review. The setting of the study was community mental health. Participants included 130 children and adolescents. Measurements were based on P-YMRS scores. An ANOVA revealed a significant difference between mean scores of patients with and without bipolar disorder. A receiver operating characteristic (ROC) analysis revealed that the P-YMRS is very good at predicting group membership and suggested a cutoff of 18 or higher for the presence of bipolar disorder. The authors conclude that the P-YMRS should be useful to differentiate pediatric bipolar disorder from other mental health conditions.

INTRODUCTION

Child and adolescent bipolar disorder is a serious illness. It can be associated with school failure, aggression, a high suicide rate, engagement in high-risk behaviors, substance abuse, and can impair a child's developmental and emotional growth.^1,2 Accurate assessment of this disorder is essential to prevent delay in the initiation of appropriate treatment. However, many authors have reported that it is frequently difficult to differentiate bipolar disorder from other psychiatric illness in children and adolescents. One explanation for this difficulty is that the presentation of pediatric bipolar disorder is frequently different than adult bipolar illness.^3,4 Another problem is that there is frequent comorbidity and symptom overlap with attention-deficit/hyperactivity disorder (ADHD) and other childhood psychiatric disorders.^5–11 An accurate diagnostic scale could potentially aid in the evaluation process of children referred for psychiatric treatment. Such an instrument could be especially useful in a community mental health setting where the difficulty of making accurate diagnoses can be increased by lack of information about genetic risk of patients in foster care, multiple comorbid diagnoses, and family dysfunction.

Recently, Gracious and colleagues reported on the use of a parent version of the Young Mania Rating Scale (P-YMRS) as a potential assessment instrument for child and adolescent bipolar disorder.¹² Their data suggest that the P-YMRS is very promising as a potential assessment instrument for pediatric bipolar disorder. However, the authors note that one limitation of their study is that it was performed in a specialized research setting and results may not generalize to other settings. They also state the need to cross-validate results in order to determine robust cutoff scores for both clinical and research purposes. We are aware of no studies reporting on the use of the P-YMRS in a community mental health setting.

After that report was published, the P-YMRS was used as an assessment instrument for children referred for psychiatric treatment in a community mental health center. The aim of this retrospective chart review was to investigate the validity and operating characteristics of this instrument in a community mental health population.

Methods

The authors reviewed the medical records of all patients between the ages of 3 and 17 years old whose parents or guardians completed the P-YMRS between December 2002 and April 2003. Demographic and clinical data obtained included age, sex, race, and psychiatric disorders. No patients were excluded from the study. All parents or guardians read English adequately to complete the instrument.

The P-YMRS is an 11-question, multiple-choice instrument.¹² Questions inquire about mood, energy level, sexual interest, sleep, irritability, rapid speech, changes in thought patterns, hallucinations and/or grandiosity, aggression, changes in appearance, and insight about needing treatment. Each question has five multiple-choice answers. Parents mark zero if the child is having no symptoms. Positive answers range from 1 through 4 on seven questions and 2 through 8 on the remaining four questions. The scores from each question are added to yield a total score. Higher scores indicate greater symptom severity. Completion time for parents is 3 to 5 minutes. In cases where patients were already stabilized on medications or in a euthymic phase between mood episodes, parents were instructed to complete the instrument based on a past manic or hypomanic episode.

Psychiatric diagnoses were determined by a board-certified psychiatrist using a semistructured interview assessment. This evaluation was part of the routine treatment process and was conducted to determine the patient's diagnosis as well to establish whether there was an indication for psychotropic medication. In all cases, collateral information was obtained from the patient's guardian. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition Text Revision (DSM-IV-TR) criteria were used to make all diagnoses.¹³ Patients were only given a bipolar diagnosis if they clearly met all required criteria. Some patients received the diagnosis of bipolar disorder NOS. This category was only used if patients met DSM-IV-TR criteria for a hypomanic episode but had not had a depressive episode. All diagnoses were reviewed and confirmed as part of the chart review process and cases of diagnostic uncertainty were excluded from the study.

Results

In this section, we will first review the demographic characteristics of our sample and then examine the discriminative validity of the P-YMRS. The a priori significance level for all statistical tests was p<0.05.

Demographic characteristics. The sample consisted of 130 children and adolescents whose parents or legal guardians completed the P-YMRS. Sixty-four subjects previously had been diagnosed with bipolar disorder, and 66 previously had been diagnosed with non-bipolar disorders. Ages ranged from 3 to 17, with a mean age of 10.65 (SD=3.49).

Among the 64 patients who met DSM-IV-TR criteria for bipolar illness, 22 (34.4%) were bipolar I, 29 (45.3%) were bipolar II, 4 (6.3%) had cyclothymia, and 9 (14.1%) were bipolar NOS. All were rapid cycling. Twenty-one (32.8%) had a history of one or more episodes of mixed symptoms. Sixteen (25.0%) had psychotic symptoms at sometime. Fifty-five (85.9%) had one or more comorbid psychiatric disorders (M=2.56, SD=1.48). The most common comorbid diagnoses were ADHD (n=49, 76.6%), relational problem (n=22, 34.4%), oppositional defiant disorder (n=12, 18.8%), conduct disorder (n=10, 15.6%), and post-traumatic stress disorder (n=5, 7.8%).

For the 66 non-bipolar patients, most had more than one psychiatric disorder (M=3.09, SD=1.54). The most frequent diagnosis was ADHD (n=46, 69.7%). Other common diagnoses were relational problem (n=35, 53.0%), learning disability (n=28, 42.4%), major depression (n=21, 31.8%), any anxiety disorder (n=20, 30.3%), oppositional defiant disorder or conduct disorder (n=15, 22.7%), and any pervasive developmental disorder (n=10, 15.2%). In order to have adequate numbers for statistical analysis, the non-bipolar patients were divided into two subgroups: those with non-bipolar mood disorders (n=27) and those with other disorders (n=39).

A one-way ANOVA with diagnosis as the independent variable and age as the dependent variable revealed a significant difference in the mean ages of the three groups, F(2, 127)=5.02, p=0.01, η²=.07. The mean ages of the groups were, from oldest to youngest, non-bipolar mood disorder group (M=12.33, SD=2.95), bipolar disorder group (M=10.55, SD=3.66), and other disorders group (M=9.67, SD=3.19). A post-hoc Tukey HSD revealed that the non-bipolar mood disorder group was significantly different from the other disorders group (p=0.01). The non-bipolar disorder group was not significantly different from the bipolar disorder group (p=0.56) and the bipolar disorder group was not significantly different from the other disorder group (p=0.41).

The ethnic make up of the sample matched that of the larger community, with a majority of the participants being non-Hispanic Caucasian (n=114) and the largest minority group being Hispanic (n=11). The sample included 89 men (68.5%) and 41 women (31.5%).

Additional demographic data was collected on whether the participant was learning disabled or in special education, whether the participant had been in foster care or adopted, whether the participant had been the victim of neglect or abuse, and whether the participant's parents were together or separated/divorced. A chi-square for independence was performed for each of these demographic variables to assess whether the groups differed on these dimensions. The only significant difference between the diagnostic groups was whether the participant's parents were together or separated/divorced, χ²(df=2, n=104)= 7.13, p=0.03, V_c=0.26. The statistical significance stems largely from the parents of bipolar patients being separated/divorced more than would be expected and the parents of other disorder patients being together more than would be expected if the variables were independent. The parents of the non-bipolar mood disorder patients basically matched the expectations for being together or separated/divorced.

Analyses were performed to see if any of the demographic variables would predict P-YMRS scores. Aside from diagnostic category, none did.

Discriminative validity of the P-YMRS. A one-way ANOVA was performed with diagnostic category as the independent variable and P-YMRS score as the dependent variable. The analysis revealed a significant difference between the groups, F(2, 127)=114.67, p=0.00, η²=0.64. The mean P-YMRS scores for the groups were (Table 1), from highest to lowest, bipolar disorder group (M=30.67, SD=9.31), non-bipolar mood disorder (M=11.52, SD=7.74), and other disorders group (M=8.03, SD=5.60). A post-hoc Tukey HSD revealed that the bipolar disorder group differed from the non-bipolar mood disorder group (p=0.00) and from the other disorders group (p=0.00). The non-bipolar mood disorder group did not differ from the other disorders group (p=0.19).

Table 1.

P-YMRS mean scores by diagnostic categories

Group	n	M	SD
Bipolar disorder	64	30.67	9.31
Non-bipolar mood disorders	27	11.52	7.74
Other disorders	39	8.03	5.60
Note: P-YMRS = Parent version of the Young Mania Rating Scale. A Tukey HSD revealed that the bipolar group was significantly different from the non-bipolar mood disorder group and the other disorders group (p=0.00 for both). The non-bipolar mood disorder group and the other disorders group were not significantly different from each other (p=0.19).

Because the non-bipolar mood disorder group did not differ from the other disorders group, we combine them into a single group for the purpose of conducting a receiver operating characteristic (ROC) analysis. An ROC analysis plots the sensitivity of a test (true positive rate) against the specificity of a test (false positive rate). The utility of a diagnostic test can be assessed by looking at the area under the curve (AUC). A diagnostic test with performance no better than chance would have an AUC=0.50, while a diagnostic test with perfect performance would have an AUC=1.00. Using P-YMRS scores to predict group membership (bipolar patients versus all non-bipolar patients), the P-YMRS had an AUC=0.96, p=0.00. This suggests that the P-YMRS is a diagnostic test with high accuracy.

An ROC analysis can also be used to suggest cutoff scores for clinical significance on a test. In this case, our objective was to maximize the true positive rate and minimize the false positive rate. With our sample, a cutoff score of 18 on the P-YMRS appears to be optimal. Using a score of 18 or greater as an indicator of bipolar disorder correctly identifies 97 percent of the bipolar patients (true positives) and incorrectly identifies only six percent of the non-bipolar patients as bipolar (false positives).

A t-test was performed to determine whether a family history of bipolar disorder would influence the scores. This was done for the subgroup of bipolar patients for whom we had genetic information (n=29). We found no significant difference between the scores of those with a first-degree relative with bipolar disorder and those without (t=0.63, p=0.54, α=0.05).

Discussion

Prompt diagnosis and treatment of pediatric bipolar disorder is critical to prevent the severe consequences of this illness, which include a high risk of suicide and impairment of functioning. Further, it is critical to avoid false positives which can expose children and adolescents to potentially serious side effects of treatment with mood stabilizers and atypical antipsychotics. Accurate assessment of childhood psychiatric illness can be especially difficult in community mental health settings where children often present with many factors, which can add diagnostic uncertainty. These include lack of genetic history for children adopted or in foster care, severe family dysfunction, and lack of medical records from previous treatment. Therefore, a valid diagnostic instrument for bipolar spectrum illness could be very valuable for community mental health clinicians. Such an instrument could potentially be useful both for screening and as an aid to confirm suspected bipolar disorder.

The major finding of this study is that the P-YMRS discriminated well between youths with bipolar spectrum disorders and those with other disorders, including non-bipolar mood disorders and ADHD in a community mental health setting. This was true even though the non-bipolar patients were quite ill with multiple psychiatric conditions (M=3.09). The mean P-YMRS scores were significantly different for the bipolar group (M=30.67) versus the non-bipolar mood disorder group (M=11.52) and the other disorder group (M=8.03, p=0.00).

An ROC analysis showed that the P-YMRS was an accurate predictor of the presence of bipolar disorder with an AUC=0.96, p=0.00. This finding is consistent with but better than the performance of the P-YMRS in the research of Gracious, et al.¹² Gracious, et al., found an AUC=0.85 when predicting group membership for bipolar against all other non-bipolar patients. Our data suggest that a score of 18 or higher on the P-YMRS can be used as a reliable cutoff to indicate the presence of bipolar disorder (97% true positives and 6% false positives).

Finally, this study suggests the P-YMRS can be used either to rate symptoms retrospectively or to rate current levels of symptomatology. This is an important characteristic of a bipolar diagnostic scale. Since bipolar disorder is a cycling disorder by definition, youths may often present for treatment when they are not currently having manic or hypomanic symptoms. Further, patients frequently present already on some psychotropic medications, which may result in symptom control; however, the need still exists to confirm the diagnosis. In this study, parents or guardians were asked to complete the P-YMRS either based on current symptom levels or in the case of patients who were not symptomatic based upon symptoms during a previous manic or hypomanic episode.

There are several limitations of this study. The primary limitations are that it is not a randomized and that it is retrospective in nature. Selection bias must also be considered a possibility because subjects were selected as a result of clinical decisions regarding diagnosis and treatment. Also, the treating psychiatrist and study investigators were not blind to the diagnosis of the subjects, which could have resulted in bias. Further, two important issues were not addressed due to the retrospective nature of the study. These were test-retest reliability and agreement between parents of the same individual. Both need to be evaluated in a future prospective study before the usefulness of the P-YMRS in a community mental health setting can be completely established. Finally, the subject population was predominately white and receiving treatment in a community mental health setting. Therefore, these findings may not generalize to minority youth or those receiving treatment in other settings.

Further studies are needed in other clinical settings to fully characterize the usefulness of the P-YMRS, including potential utility as a screening tool. Further studies are also needed to establish whether a P-YMRS cutoff of 18 is optimal for samples other than the current one. However, this work supports the findings of Gracious and colleagues who published the initial report of the validity of this instrument. Taken together, these studies suggest that the P-YMRS may be useful to differentiate pediatric bipolar disorder from other psychiatric conditions.