 |
 |
|
|
|
More than a decade ago, a review of epidemiologic studies of children and adolescents found diagnostic comorbidity to occur at higher-than-chance expectations (1). Despite the frequent cooccurrence of psychiatric disorders, only a handful of studies have examined diagnostic comorbidity in general population samples that include adolescents (2–12). There is much variability in the comorbidity rates reported in these studies, mainly arising from methodological differences that include variations in sample age, study design, and diagnostic criteria; reliance on different informants and assessment instruments; the inclusion or exclusion of a functional impairment criterion; and the time frame used to establish diagnoses. Nonetheless, findings from these studies indicate that ANX and DEP cooccurred at rates ranging from 12.9% to 75.0%. Despite the wide variability, most of the studies reported high rates of comorbid ANX and DEP. Within the externalizing disorders, the cooccurrence of ADHD and CD–ODD ranged from 4.7% to 93.0%, although again, most of the studies reported a high comorbidity rate. Finally, cooccurring internalizing and externalizing disorders varied from 2.5% (ADHD and DEP) to 83.3% (CD–ODD and DEP). A metaanalysis of general population studies found that comorbidity risk in youth was highest within the externalizing (ADHD and CD–ODD median OR 10.7) and internalizing (DEP and ANX median OR 8.2) disorders (13). The relation between DEP and CD–ODD was almost as strong as that between DEP and ANX, and DEP rather than ANX had a stronger relation to the externalizing disorders. Several explanations have been offered for the comorbidity of psychiatric diagnoses (1,13–16). First, detection artifacts exist. These include referral bias (that is, clinical vs general population samples), screening and surveillance factors (that is, the procedure for choosing general population samples), rater expectations and assessment strategies (that is, structured vs unstructured interviews), and the use of multiple informants describing the same disorder in different terms. However, a review of recent community-based comorbidity studies argued against methodological biases as the principle cause for findings of diagnostic comorbidity among youth (13). Second, there are nosological explanations for comorbidity that include categories or dimensions (that is, the way a disorder is defined), overlapping diagnostic criteria (that is, similar symptoms for different disorders), artificial subdivision of syndromes (that is, dividing disorders that revolve around one primary symptom complex into various subcategories), and arguments that one disorder represents an early manifestation of the other and that one disorder is part of the other (that is, the presence of one disorder excludes the diagnosis of a second disorder when the latter disorder forms part of the symptomatology of the first disorder). A review of recent community-based comorbidity studies presented data challenging some of these arguments but also acknowledged that nosological considerations probably do contribute to some aspects of comorbidity (13). The authors also noted that this possibility does not negate research aimed at understanding why certain symptoms commonly occur together and why there is diagnostic overlap. Third, it has been suggested that there are possible substantive causes of diagnostic comorbidity, such as shared risk factors (that is, risk factors associated with an increased vulnerability to one disorder are identical to those associated with vulnerability to another disorder); that overlap between risk factors (that is, risk factors for 2 disorders may be distinct and different while also being correlated with each other); that the comorbid pattern constitutes a meaningful syndrome (that is, the syndrome of the comorbid disorder is distinct from that of the 2 disorders in isolation); and that one disorder creates an increased risk for the other. Study Goals Past comorbidity studies using general population samples have generally been limited in 2 ways. First, they have either not examined sex-specific comorbidity rates or they have not presented these results. One exception is the Ontario Child Health Study, which found comparable cooccurring ADHD and CD rates for girls and boys aged 12 to 16 years. The rate of cooccurring emotional and externalizing disorders was higher for girls than for boys, but this effect was not tested statistically (11). Another exception was a study of youths aged 9 to 16 years that found no statistically significant sex differences in comorbidity rates for the various combinations of ANX, DEP, ADHD, and CD–ODD (2). A second limitation of past comorbidity studies is that they have typically combined data from adolescents and parents to arrive at diagnostic comorbidity rates. One exception is a community-based study that was based solely on adolescent reports (6). The goal of our study was to estimate ANX, DEP, ADHD, and CD–ODD comorbidity rates for an adolescent community sample. Prevalence estimates for these diagnostic categories are presented elsewhere (17). We were particularly interested in examining sex and informant effects on diagnostic comorbidity. From our prevalence estimates, we hypothesized that adolescent girls would report higher comorbid ANX and DEP rates than would adolescent boys and that boys would report higher comorbid ADHD and CD–ODD rates than would girls. Regarding informant differences, we hypothesized that comorbidity rates based on adolescent reports would be higher than those based on mothers’ reports, given that mothers would have less access to their children’s inner experiences (for example, feelings of depression) and delinquent activities outside the home (for example, conduct problems such as vandalism). MethodSample and Procedure In 1986–1987, mothers and teachers completed questionnaires for 4488 kindergarten children from French public schools in the province of Quebec. A random subsample of 2000 children was followed annually into adolescence. Trained interviewers separately administered a structured psychiatric interview to mothers and their children between 1995 and 1997, when most children (68.9%) were aged 15 years (range 14 to 17 years). Before the interview, a letter describing the study’s procedures and aims was sent to mothers. Research assistants then telephoned mothers to arrange interview times for those who, along with their children, agreed to participate in the study. At the time of the interview, adolescents and their mothers each signed a written consent form stating that they agreed to be interviewed, that their responses would remain confidential, and that they could end their participation in the study at any time. Interviews lasted approximately 90 minutes and occurred in participants’ homes. Data were collected for 1201 of the initial 2000 participants. The most common reasons for missing data were refusal to participate (87%) and inability to contact participants (13%). A detailed comparison of respondents and nonrespondents is presented elsewhere (17). In brief, nonrespondents scored lower than respondents on mother’s age at the birth of her first child, parents’ years of schooling, parents’ socioeconomic status, and teacher-rated kindergarten prosocial behaviour. As well, nonrespondents scored higher than respondents on family adversity. While these differences were statistically significant, the magnitude of differences was small. There were no significant differences in father’s age at the birth of his first child, mother-rated kindergarten prosocial behaviour, and mother- and teacher-rated kindergarten hyperactive, oppositional, inattentive, aggressive, and anxious behaviour. To better ensure sample representation, we implemented weighting procedures, using SPSS for Windows, Version 10.0 (18). Using the forward stepwise procedure, we conducted a series of logistic regressions in which demographic and kindergarten-rated behavioural variables, along with the adolescent’s sex, were entered separately and in combination as predictor variables. The dependent variable was response vs nonresponse to the psychiatric interview. Sex (OR 1.35), mother’s educational level (OR1.17), and teacher-rated prosocial behaviour (OR 1.04) were significant predictors of responding to the psychiatric interview. We entered these 3 predictor variables into a logistic regression analysis to calculate each participant’s probability of responding to the psychiatric interview. We then used an inverse probability weighting scheme to assign weights. Instrument Diagnostic Interview Schedule for Children-2.25. A French translation (19) of the DISC-2.25 (20) examined adolescent psychiatric functioning over the previous 6 months. The DISC-2.25 is a structured interview for youth aged 6 to 18 years that evaluates DSM-III-R diagnoses. We also used DISC-2.25 questions to assess the level of impairment from symptoms in regard to the adolescent’s functioning at home, at school or work, and with peers. If there were problems in at least one area, we scored the response as positive for impairment. Adolescents aged 12 to 14 years from a general population sample completed the French DISC-2.25, and mean test–retest reliability over a 2-week period was k = 0.37 (disruptive disorders), k = 0.49 (ANX), and k = 0.55 (DEP). Parents of children aged 6 to 14 years also completed the French DISC-2.25, and mean test–retest reliability was k = 0.32 (DEP), k = 0.60 (disruptive disorders), and k = 0.69 (ANX) (21). Data Analyses To calculate comorbidity rates, we used a bidirectional approach that takes into account disorder base rates and provides a common comorbidity metric (22). The formula is as follows: number of cases with both disorders A and B / (number of cases with both disorders A and B + number of cases with only disorder A + number of cases with only disorder B). The diagnostic categories were ANX (comprising simple phobia, social phobia, agoraphobia, separation anxiety disorder, overanxious disorder, and generalized anxiety dis-order), DEP (comprising major depression and dysthymia), ADHD, and CD–ODD (comprising CD and ODD). Analyses were conducted separately according to sex and informant. To increase comparability with past studies, we also presented comorbidity rates for adolescent and mother reports combined at the diagnostic level. We calculated ORs to measure comorbidity strength and Yates continuity-adjusted chi-square to test for sex differences. ResultsComorbidity Within Internalizing Disorders Tables 1 to 6 present diagnostic comorbidity rates for adolescent and mother reports, both separately and combined. According to adolescent reports for the total sample, close to 1 in 4 (22.7%) met diagnostic criteria for both ANX and DEP (Table 1). The OR indicates that the risk of suffering from DEP was 13.2 times higher among adolescents suffering from ANX, compared with adolescents without ANX (and vice versa for the risk of having anxiety among adolescents suffering from DEP). Mother reports for the total sample showed an ANX–DEP comorbidity rate (14.8%) that was lower than the adolescent-reported rate (22.7%). There remained, however, a greater-than-chance risk of having one of the disorders if adolescents met diagnostic criteria for the other disorder (OR 11.0). Combining adolescent and mother reports showed that almost 1 in 4 (23.0%) adolescents met diagnostic criteria for both ANX and DEP and that the risk of DEP was 9.6 times higher among adolescents with ANX, compared with adolescents without ANX (and vice versa for the risk of having anxiety among adolescents with DEP). Finally, the chi-square results for adolescent reports only, mother reports only, and mother and adolescent reports combined consistently indicated no statistically significant sex difference in the cooccurrence of ANX and DEP.
Comorbidity Within Externalizing Disorders According to adolescent reports for the total sample, the comorbidity rate for ADHD and CD–ODD was 5.8% (Table 2). The OR shows that the risk of having CD–ODD was 17.6 times higher among adolescents with ADHD, compared with adolescents without such a disorder. Mother reports for the total sample showed a higher comorbidity rate for ADHD and CD–ODD (13.7%) than did the adolescent reports (5.8%), and there was a twelvefold increase in the risk of having one of the disorders if adolescents met diagnostic criteria for the other disorder. Combining adolescent and mother reports showed that 13.4% of adolescents met diagnostic criteria for both ADHD and CD–ODD and that the risk of having CD–ODD was 8.1 times higher among adolescents with ADHD, compared with adolescents without ADHD. Finally, results from the chi-square analysis showed no statistically significant sex difference in comorbidity rates of ADHD with CD–ODD for adolescent and mother reports, either separately or combined.
Comorbidity Between Internalizing and Externalizing Disorders ANX–ADHD. Adolescent reports for the total sample indicated little overlap in ANX and ADHD (0.9%), and the OR suggested a chance-level risk of one disorder in adolescents who met diagnostic criteria for the other disorder (Table 3). The ANX–ADHD comorbidity rate for the total sample was higher in mother reports (5.5%), compared with adolescent reports. In addition, mothers indicated that the risk of having ADHD was 2.8 times higher among adolescents with ANX, compared with adolescents without ANX. Combining adolescent and mother reports showed that 7.1% of adolescents met diagnostic criteria for both ANX and ADHD. The OR indicated that the risk of meeting diagnostic criteria for ADHD was 2.9 times higher among adolescents with ANX, compared with adolescents with no such disorder. Turning to sex differences, the findings were not statistically significant for adolescent and mother reports separately. According to the combined reports of adolescents and mothers, however, there was a statistically significant sex difference (c2 = 5.71, df 1; P < 0.05) in the comorbidity rate for ANX and ADHD. Specifically, adolescent boys had significantly higher comorbidity rates than did adolescent girls (13.1% and 2.7%, respectively).
ANX and CD–ODD. Table 4 shows that, according to adolescent reports for the total sample, there was little overlap in ANX and CD–ODD (2.5%), and the OR suggested a chance-level risk of one disorder in adolescents who met diagnostic criteria for the other disorder. In contrast, mother reports for the total sample showed a higher comorbidity rate (6.4%) and a greater-than-chance risk of having one of the disorders if adolescents met diagnostic criteria for the other disorder (OR 3.3). The combination of adolescent and mother reports resulted in a comorbidity rate of 5.7% and a chance-level risk of one disorder in adolescents who met diagnostic criteria for the other disorder. Finally, the chi-square results for adolescent reports only, mother reports only, and mother and adolescent reports combined consistently indicated no statistically significant sex difference in the cooccurrence of ANX and CDD–ODD.
DEP and ADHD. According to adolescent reports for the total sample, only 2.9% of adolescents met diagnostic criteria for both DEP and ADHD (Table 5). However, the risk of having ADHD was 6.6 times higher among adolescents with DEP, compared with adolescents with no such disorder. Mothers also reported little overlap in DEP and ADHD (1.3%), and there was a chance-level risk of one disorder in adolescents who met diagnostic criteria for the other disorder. According to adolescent and mother reports combined, the cooccurrence of DEP and ADHD was 4.0%, and there was a chance-level risk of one disorder in adolescents who met diagnostic criteria for the other disorder. Turning to sex differences, adolescent reports showed a trend (c2 = 3.76, df 1; P < 0.10). In particular, adolescent boys had higher rates of cooccurring DEP and ADHD than did adolescent girls (14.2% and 0%, respectively). There were no statistically significant sex differences according to mother reports alone or according to adolescent and mother reports combined.
DEP and CD–ODD. Based on adolescent reports for the total sample, the comorbidity rate for DEP and CD–ODD was 7.2% (Table 6). The OR shows that the risk of having a CD–ODD was 3.5 times higher among adolescents with DEP, compared with adolescents not suffering from DEP. For the total sample, the mother-reported rate for comorbid DEP and CD–ODD was higher (10.8%) than the adolescent-reported rate (7.2%). The risk of meeting diagnostic criteria for CD–ODD was 8.6 times higher among adolescents with DEP, compared with adolescents without such a disorder. Combining adolescent and mother reports showed that 11.3% of adolescents met diagnostic criteria for both DEP and CD–ODD and that the risk of having CD–ODD was 3.9 times higher among adolescents with DEP, compared with adolescents not suffering from DEP. Finally, results from the chi-square analysis showed no statistically significant sex difference in DEP and CD–ODD comorbidity rates for adolescent and mother reports, both separately and combined.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||