The DSM-IV Rates of Child and Adolescent Disordersin Puerto Rico:  Prevalence, Correlates, Service Use, and the Effects of Impairment FREE

The formulation of explicit diagnostic criteria in the DSM-III¹ and its revision, DSM-III-R,² facilitated several large-scaleepidemiological studies of mental disorder in the United States on the basisof psychiatric diagnosis.^3- 7 Manyof the revisions in DSM-IV⁸ followedfrom results of epidemiological and clinical studies suggesting that disorderspreviously thought to develop in adulthood were also evident in childhoodand that children and adolescents could experience a wide variety of problemsand manifestations of distress that fail to satisfy criteria of a mental disorder.^9- 10 These studies also provided new insightsinto the prevalence and comorbidity of mental disorders, the role of impairmentin arriving at meaningful diagnoses, and the discrepancies between rates ofneed for services and actual service provision. When DSM-IV was developed, epidemiological information based on DSM-III and DSM-III-R study results was criticalin fine-tuning the diagnostic criteria and integrating clinical significancein the formulation of the disorders.

Starting with DSM-III, all versions of the DSM stipulated that to qualify for a psychiatric diagnosis,a disorder must generally meet a "clinical significance" criterion that involvesfeatures of impairment or distress. However, in DSM-IV,this criterion was made more explicit. The DSM-IV includesclinical significance among the essential criteria for most specific disorders,with only the most obvious and impairing conditions (eg, schizophrenia andother psychotic disorders) excepted.¹¹ Theimplications of these revisions in the DSM have onlybegun to be explored empirically for both adult¹² andchildhood^9,13 disorders, althoughthe revisions have generated considerable theoretical debate.^14- 15

The explicit clinical significance requirement has led to the developmentof epidemiological diagnostic instruments based on the DSM-IV that for the first time make clinical significance a prerequisitefor each of the disorders measured. When impairment in functioning was nottaken into account, a number of epidemiological studies that applied DSM-III and DSM-III-R criteria^4- 5,16 reported prevalencerates higher than 30% for diagnosable mental disorders. Although such highrates were not universal findings, they suggested that the DSM-III and DSM-III-R, as operationalizedin certain structured diagnostic instruments, tended to be overly inclusive.^17- 19 It is still not clearthe extent to which these findings are caused by characteristics of the taxonomyor to the way in which the taxonomy is operationalized in instruments.

New measurement procedures have been developed to implement the clinicalsignificance and other criteria changes in the DSM-IV,and several of the procedures are designed for assessing persons in communitysamples and/or in primary care settings.^20- 21 Althoughthere are still a number of substantial and unresolved issues regarding validand reliable assessment of psychiatric status and functioning in childrenand adolescents,^8,10- 22 itis important to inform the discussion of these issues by presenting systematicdata collected by using the latest measures and the latest diagnostic criteria.Few epidemiological studies of prevalence rates of child psychiatric disordersand their correlates based on DSM-IV criteria havebeen reported.²³

To add to the pool of data regarding how DSM-IV criteriaapply to general population samples, we present findings about the prevalenceand correlates of child and adolescent disorders defined according to the DSM-IV in a representative sample of children aged 4 to17 years on the island of Puerto Rico who were interviewed between 1999 and2000. The DSM-IV diagnoses were operationally determinedby using the Diagnostic Interview Schedule for Children, version IV (DISC-IV).²⁴ The DISC-IV has added questions to assess the clinicalsignificance (impairment or distress) to the diagnostic schedules of eachof the disorders it measures. To illustrate the importance of including clinicalsignificance as 1 of the criteria that defines each psychiatric disorder,we highlight the issue of clinical significance by reporting DSM-IV disorders with and without disorder-specific impairment-distresscriteria. The issue of impairment is further examined by considering informationfrom a global impairment rating, the Parent Interview–Children's GlobalAssessment Scale (PIC-GAS).¹⁸

We considered it important to include both the measure of specific impairmentand the global measure for several reasons. First, the assessment of specificimpairment, such as that obtained with the DISC-IV, is a subjective ratingrequiring that the respondent establish a causal connection between the symptomsof a specific diagnosis and resulting impairment or distress. Although theseratings have demonstrated reliability,²⁵ theaccuracy with which parents or children can make the attribution of impairmentto the specific symptoms has never been validated and is therefore open toquestion.²⁶ We wanted to provide differentalternatives for assessing impairment and elected to use the global measure,PIC-GAS, which has demonstrated reliability and validity.¹⁸

Another consideration for including both measures was that the DISCimpairment measures also incorporate an item that relates to distress causedby the symptoms. In attempting to adhere to DSM-IV criteria,we considered it important to include the notion of distress, as well as thatof impairment. We also wanted to measure need for mental health services byusing a definition of need for services that includes meeting criteria fora disorder and substantial impairment in functioning. Because clinical significanceis such an important issue, we did not want to rely exclusively on a singlemeasure. Moreover, the rating of global impairment takes into account allaspects of the child's overall functioning, whereas the DISC schedules usedin the present study include only a selected group of DSM-IV disorders. It is conceivable, for example, that a child with autismor another serious developmental disorder not included in the DISC could haveresults negative for the DISC diagnoses being considered, as well as for diagnosis-specificimpairment on the DISC-IV, and still be substantially impaired in functioning.

Besides reporting last-year prevalence rates of selected DSM-IV disorders, with and without clinical significance criteria,we report rates of mental health service use among those with and those withouttargeted DSM-IV diagnoses. The patterns of serviceuse not only provide useful information for policy makers who use prevalenceestimates, but they also provide insights into the interpretation of bothdiagnostic and impairment information.

There are several advantages in studying DSM-IV/DISC-IVdiagnoses in a representative island-wide sample of the Puerto Rican population.This population includes the full spectrum of socioeconomic status, with childrenliving in different contexts of economic advantage and poverty in both urbanand rural areas. The sample obtained for the present study represents thefull range of diversity of the child population. These factors enhance thegeneralizability of the findings. Although Puerto Rico clearly differs frommainland American populations in language, culture, and distribution of riskfactors, prevalence studies of both adult and child psychiatric disorder performedin Puerto Rico have shown remarkable similarity to overall trends in the UnitedStates.^16,27- 30

Most child psychiatric epidemiological studies have focused on a narrowage range or on specific age groups.¹⁷ Ourstudy includes children from a broad age range of 4 to 17 years. We are ableto analyze patterns of association between rates of disorder and sex, parentaleducation, family income, and urban vs rural residence to verify if relationsbetween psychiatric disorder and these demographic factors are similar tothose found elsewhere, as well as to findings of an island-wide child studyof DSM-III disorders performed in 1985.¹⁶

An island-wide household probability sample of children aged 4 to 17years was drawn from 4 strata: Puerto Rico's health reform areas, urban vsrural areas, participant age, and participant sex. Block groups defined inthe US Bureau of the Population and Housing 1990 Census for Puerto Rico werethe primary sampling units. These groups were classified according to economiclevel and size, grouped into block clusters, and further classified as urbanor rural.

Clusters of households were randomly selected from each stratum, householdswith children aged 4 to 17 years were selected from the clusters, and 1 childwas selected from each household by using Kish tables³¹ adjustedfor age and sex. From 2102 eligible households, 1886 parent-child dyads wereinterviewed for a total response rate of 90.1%.

The sample was weighted to represent the general population of childrenin Puerto Rico in the year 2000. The weights correct for differences in theprobability of selection because of the sampling design and adjust for nonresponse.To account for the complex sampling design, SEs were estimated (SUDAAN softwarerelease 8.0; Research Triangle Institute, Durham, NC).

Table 1 shows the demographiccharacteristics of the sample before poststratification. The distributionby age and sex of the sample obtained is similar to that of the Puerto Ricanpopulation as described in the 2000 US census. The final sample of 1897 childrenconstituted a sampling fraction of approximately 2.2 per 1000 children inthe population.

Last-year DSM-IV psychiatric disorders wereassessed by using the latest translation into Spanish of the DISC-IV,²⁵ with parallel youth and parent interview versions.The test-retest reliability of the DISC-IV has been reported in both Spanish-speakingand English-speaking clinic samples and yielded comparable results.^24- 25

The DISC-IV inquires about the level of impairment and distress associatedwith each diagnosis through probes that determine the degree to which thesymptoms of a given diagnosis have caused distress to the child or affectedhis or her school functioning or relations with caretakers, family, friends,or teachers. The test-retest reliability of the impairment and distress probeswere fair to moderate; most κ values ranged from 0.42 to 0.80 for mostdisorders assessed.²⁵

Children younger than 11 years were not interviewed with the DISC becausethere is evidence that their reports would not be reliable.³² Thesubstance abuse disorders schedules were administered in children aged 11to 17 years but not to their parents, whereas oppositional defiant disorder(ODD) was excluded from the child protocol. Parents tend to be unaware ofthe use of substances in their adolescent children and are not consideredgood sources of information for substance use disorders in their children.Results of prior studies in which Puerto Rican and mainland samples were includedhave also shown poor reliability for ODD in child reports, as well as poorconcordance with clinical diagnosis.³³

The official DISC-IV scoring algorithms use data from parent and childinformants and allow the ascertainment of the presence of a diagnosis, withor without impairment as measured with the DISC impairment scales. In thisarticle, we use the DISC impairment algorithm that refers to moderate impairmentin at least 1 area of functioning. The rates reported include both parentand child informants for children aged 11 to 17 years, and only parent informantsfor children younger than 11 years. A case is considered positive if it meetsfull DSM-IV diagnostic criteria according to eitherthe parent or child DISC-IV.

Overall global impairment in functioning was measured by using the PIC-GAS.The PIC-GAS is the lay version of the Children's Global Assessment Scale.¹⁸ The concurrent, discriminant, and construct validityof the PIC-GAS have been reported by using combined data from 4 communities,including data from Puerto Rico.¹⁸ Althoughthe PIC-GAS is presented in 10 deciles that describe different levels of functioning,the instructions for assigning scores on the PIC-GAS, as well as on the DSM-IV Global Assessment of Functioning Scale, specificallyencourage raters to "use intermediary levels." Previous psychometric workwith the clinician Children's Global Assessment Scale³⁴ revealedthat a score lower than 65 was the best cutoff to distinguish between thosedefinitely impaired and those not impaired. Similar psychometric work withthe PIC-GAS shows that with lay raters, a score lower than 69 is optimal toidentify children who are impaired in functioning.¹⁸

To our knowledge, none of the publications available have shown empiricallyderived cutoff scores to distinguish those who are mildly, moderately, orseverely impaired. Therefore, we have opted to use a cutoff lower than 69on the PIC-GAS coupled with diagnosis-specific impairment ratings to operationalizethe classification of serious emotional disturbance that is considered bythe US federal government in reimbursing states for mental health services.³⁵ This definition requires the presence of a DSM-IV diagnosis with substantial impairment in functioning.

The Spanish-language version of the Service Assessment for Childrenand Adolescents (SACA)³⁶ was used to ascertainthe types of services and treatments used by children for emotional, alcohol,and drug problems. The instrument inquires about lifetime and last-year useof 25 specific service settings that are divided among inpatient-residential,outpatient, and school settings.³⁷ The outpatientsetting is disaggregated into mental health outpatient and general healthoutpatient. The mental health outpatient includes patients who attended amental health or substance abuse clinic or saw a professional in a privateoffice. The Spanish-language SACA has shown fair to moderate reliability,with most κ values ranging from 0.41 to 0.87 for most services reportedby parents and children, as well as moderate to substantial sensitivity whenparental reports were compared without medical records.³⁶ TheSACA has 2 versions (parent and child), and use of services was consideredpositive if a positive response was given by either parent or child.

The survey was performed from January 1999 through December 2000. Thechild's biological mother was the adult informant in 89.4% of the cases. Interviewstook place in the subject's home, with different interviewers for parent andchild, and interviewers were blinded to the results of each other's interviews.Interviews were audio taped, and 15% were spot-checked for quality control.

Table 2 presents the last-yearprevalence rates of specific DSM-IV/DISC-IV disordersfor children aged 4 to 17 years. Prevalence is reported in 4 ways. The firstcolumn shows the presence of DISC diagnostic criteria in either parent orchild report, without taking clinical significance (impairment and distressresulting from symptoms) into account. The rates reported in the second columnqualify the column 1 diagnoses by requiring at least 1 level of DISC-IV impairmentor distress according to either parent or child. Therefore, the values inthe second column refer to disorders that meet full DSM-IV criteria. The third column presents rates based on the presence ofcriteria for diagnosis in either parent or child without the DISC impairmentcriteria but with global impairment as measured with the PIC-GAS with a scorelower than 69. The rates appearing in the fourth column require the presenceof DISC-IV criteria for diagnosis according to either parent or child, withDISC-IV impairment and a global impairment score lower than 69 on the PIC-GAS.

The proportion of the sample that appeared to have met all DSM-IV criteria except for the impairment criterion is 19.8% (firstcolumn of Table 2). Prevalencerates are all greater without impairment than when impairment is considered.The second column is a more faithful representation of DSM-IV criteria, in that it requires both diagnosis and some degreeof specific impairment or distress to be present. When all diagnoses assessedwith some impairment are considered (second column), 16.4% of the sample has1 or more DSM-IV diagnoses. The specific diagnosiswith the highest prevalence is attention-deficit/hyperactivity disorder (ADHD)(8.0%), followed by ODD (5.5%), separation anxiety (3.1%), and major depression(3.0%).

In contrast, the third column shows the prevalence rates when globalimpairment (PIC-GAS score lower than 69) is substituted for the DISC-IV disorder-specificimpairment criteria. The rates in this column are not consistent with thefull DSM-IV criteria, in that they take into accountclinical significance of mental disorder as measured with global impairmentinstead of impairment due to specific symptoms. The rate for any diagnosisis reduced from 16.4% to 7.6% (third column). For both overall prevalenceand the diagnostic groupings, the effect of the global impairment qualification,as compared with the standard DSM-IV/DISC-IV definitionthat includes the DISC impairment scales, tends to reduce the prevalence ratemuch more (about half). In the fourth column, both global impairment and theDISC-IV disorder-specific impairment criteria are included in the designationof a case, and the rate is reduced slightly from 7.6% to 6.9%.

Prior to the development of the DSM-IV, a numberof demographic factors were shown to relate to child psychiatric disorder.^30,38- 39 To determine whichof these correlates of the diagnoses remain after the application of DSM-IV/DISC-IV criteria, we report in Table 3 associations of the more prevalent disorders with age, sex,parental education, household income, parental civil status, and urban vsrural residence. Odds ratios (ORs) were estimated by using logistic regressionmodels that weighted the data for the sample design poststratified to the2000 census results (SUDAAN).

Boys had more disruptive disorders (OR = 1.6), ADHD (OR = 2.0), andODD (OR = 1.6), while girls had more depressive disorders in general (OR =0.3) and specifically major depression (OR = 0.4). Age had a mixed patternof associations with these childhood disorders. Rates tended to increase withage for depressive disorders (OR = 5.2), anxiety disorders (OR = 1.7), socialphobia (OR = 2.4), and major depression (OR = 4.6). Rates tended to decreasewith age for ADHD (OR = 0.5), and there was no reliable age association withany disruptive disorder, specifically conduct disorder and ODD, or with separationanxiety disorder. Neither parental education nor income was related to anyof the disorders assessed. Children whose parents were not married (single,separated, widowed, or divorced) were more likely to meet criteria for anystudied DISC-IV disorder and major depressive disorder. In addition, childrenwho lived in urban areas were more likely to meet criteria for any diagnosis(OR = 1.9), any disruptive disorder (OR = 1.9), and ADHD (OR = 2.1).

In Table 4, we show last-yearservice use rates for groups of children cross classified according to whetherthey had 1 or more DSM-IV/DISC-IV diagnosis and whetherthey had global impairment scores lower than 69. As expected, children with DSM-IV diagnoses and impairment used the most services,but only half of the children (49.6%) in this group received any type of service.Overall, school services were more frequently used than any other serviceand were used by children who were impaired regardless of whether they met DSM-IV/DISC-IV criteria.

To further assess the relative power of DSM-IV/DISC-IVand global impairment (PIC-GAS score lower than 69) to predict service use,we performed a series of logistic regression analyses. The results (Table 5) showed that for all service categoriesexcept for any school service, use could be described in terms of main effectsof both diagnosis (present/absent) and global impairment (present/absent).The DISC diagnosis and global impairment were both strongly associated withall types of service use, but the effects of global impairment were generallystronger than the effects of diagnosis, with the exception of use of the specialtymental health sector (any mental health professional), for which diagnosishad a larger OR than did global impairment.

For any school service, the interaction term was significant. As shownin Figure 1, about one third ofchildren who had PIC-GAS scores lower than 69 received school services, regardlessof the presence of a diagnosis. Among those with no impairment and no diagnosis,school services were used by 6.8%, and use increased to 17.6% among thosewith a diagnosis.

Our results are generally consistent with previous DSM-III prevalence rates reported for the island,¹⁶ despitechanges in the nosology and the new version of the DISC that was used. Inboth the survey performed in 1985 and the current study, the relative frequencywith which the disorders occurred was similar, even though the prevalencerates of the specific disorders were somewhat different. In both surveys,ADHD and ODD were approximately twice as common as were major depression,separation anxiety, social phobia, generalized anxiety, and conduct disorder.

Nevertheless, we identified differences in prevalence rates across thesestudies that cannot be attributed solely to the application of a clinicalsignificance criterion such as was used in our present data set. These differencesin prevalence rates are likely to be influenced by differences in the methodsused in the surveys. The 1985 DSM-III survey inquiredabout 6-month prevalence rates rather than the year rates of the current survey.It also combined information from child and parent informants in all ages,whereas information from both informants in the present survey was combinedonly in the children aged 11 to 17 years. Furthermore, the 1985 rates werebased on clinical judgment by clinicians who used a prior version of the DISCto structure the interview, whereas in this study diagnoses were estimatedwith a structured diagnostic interview used by lay interviewers.

The DSM-IV/DISC-IV prevalence rates we reportare not fully consistent with those from a recently reported DSM-IV study performed among black and white youth in North Carolina.In that study, Angold et al²³ reported lowerrates of ADHD, ODD, and major depression than those obtained in the presentstudy, but they found higher rates of conduct disorder and substance use.Several factors may help explain these differences. The lower rates couldbe explained by the fact that the Angold et al²³ surveyused an interviewer-based structured interview¹⁹ differentfrom the one we used. Moreover, they reported a 3-month prevalence rate ratherthan a 1-year prevalence rate, and the children assessed were aged 9 to 17years and living in rural areas. Their study also included a wider range ofpsychiatric disorders than does the present study. Nevertheless, Angold andcolleagues²³ performed further analyses oftheir data with the same diagnoses evaluated in the present study (oral communication,November 2002) and found rates of any DSM-IV disorderof 17.7% (95% confidence interval, 15%-20%) that are similar to the rate inthis study of 17.3% (95% confidence interval, 14.5%-20.7%), when we excludedchildren aged 4 to 8 years from our rates to make the analyses comparable.

Previous reports of rates of specific diagnoses based on earlier versionsof the DSM and in which the same assessment instrumentswere used have shown few differences in rates between Puerto Rico and mainlandlocalities, with 1 salient exception. The rates of conduct disorder in bothour 1985 survey⁴⁰ and in the Methods for theEpidemiology of Child and Adolescent Mental Disorders study²⁶ havebeen compared with those obtained in mainland populations and were lower amongisland Puerto Rican children and adolescents. Lower rates of substance usedisorders in both adult and adolescent populations have also been reportedfor island Puerto Ricans, as compared with findings in mainland populations.^41- 42 The lower rates appear to be relatedto the extent to which correlates of these disorders appear on the island,the most salient being better family relations and social support.²⁶

Similar to results of other surveys,^16,43- 44 ourstudy results showed many of the expected associations of ADHD, ODD, anxietydisorders, and depressive disorders with age and sex.^10,43,45 Unlikeresults of many studies of mental disorder among children and adolescents,^46- 48 including the earlierstudy¹⁶ in Puerto Rico, we found no associationsbetween rates of disorder and indicators of socioeconomic status, such asparents' education and family income. This unexpected finding led to furtherexploration. In one analysis, we constructed levels of income with 3 categoriesused by Costello et al⁴⁶ to define relativepoverty: lowest third, middle third, and upper third. In another analysis,we combined parental income and education to define a familial high-risk groupthat was low in both income and education. Neither of these additional analysesrevealed an association. However, our failure to find that prevalence varieswith traditional measures of socioeconomic factors is consistent with resultsof a handful of studies that focused on impoverished families.^23,49

Beyond absolute and relative measures of poverty, we considered perceptionsof what constitutes poverty. Forty-eight percent of Puerto Ricans live belowthe poverty level.⁵⁰ Our survey showed thatonly 22.2% of parents with an annual income below $12 000 consideredthat they lived poorly or very poorly, while 42.8% of those thought they livedwell or very well. Of parents in the $12 001 to $25 000 income level,13.2% thought they lived poorly. When perception of poverty was substitutedfor actual income, we found that disruptive disorders (ADHD and ODD) weremore frequent among those who reported that they lived poorly. Perceptionof poverty was not related to either anxiety or depressive disorders. Thispattern is consistent with results in other studies^51- 52 thatreported stronger associations between poverty and externalizing disordersthan with internalizing disorders. The findings suggest that absolute or relativepoverty indexes may not be the most appropriate measures to use in populationswhere most persons are of low income. Further development of measures of perceivedpoverty or the inclusion of measures of "social capital,"⁵³ suchas the existence of job opportunities, community organization, and other communityresources, may be more fruitful.

We found that global impairment as indicated by a PIC-GAS score lowerthan 69 was an important predictor of service use, even when adjusting forthe presence of a DSM-IV/DISC-IV disorder. Exceptfor mental health specialty services, global impairment appeared to be asimportant, if not more important, than diagnosis in predicting service. Forschool service use, presence of a diagnosis was important only for childrenwith PIC-GAS scores of 69 or higher. Service use among children with globalimpairment was the same for those with a diagnosis as for those without adiagnosis as measured in our survey. It is possible that some of the childrenreceiving school services had disorders, such as learning, motor skills, andlanguage disorders, that were not assessed in our study. Such disorders couldhave been perceived by counselors and teachers and therefore could have beenaddressed in school. Children with impairment without diagnosis may not havebeen eligible for specialty mental health treatment because they did not meetthe eligibility criteria of severe emotional disturbance established by moststates and insurance companies in both the mainland United States and PuertoRico (ie, meeting criteria for DSM-IV disorder andsubstantial impairment).³⁵

The importance of impairment in our study is not simply a function ofour use of logistic regression. Alegría et al⁵⁴ foundthat impairment was the most important predictor of service use when parentdecisions to seek service were modeled by using classification and regressiontree approaches.⁵⁵ Given that 13.3% of thechildren in Puerto Rico were impaired (data not shown) and that this seemsto be such an important predictor of service use, an important strategy forprevention interventions and policy makers should be the identification ofchildren with impairment and intervening with parents and educators so thatthey can link impairment with the need for mental health service.

As was shown in other research,^7,44 ourdata showed that the inclusion of the DSM-IV clinicalsignificance criterion with the diagnostic criteria of most Axis I diagnosesreduced the prevalence rates of disorders. The use of specific impairmentitems reduced by 18% the prevalence from that obtained through criteria withoutimpairment and distress. Rates were reduced by an additional 58% when a globalimpairment measure was included. Several investigators^5,7,56 notedthat prevalence rates vary according to the impairment measure used. Thesefindings suggest the need for a consensus about how clinical significanceis conceptualized, operationalized, and measured for both research and clinicalpurposes. Although progress has been made in the conceptualization of impairmentby the World Health Organization in its International Classificationof Functioning, Disability and Health,⁵⁷ furtherwork is needed to examine the applicability of this classification to children.Some of the discrepancies observed in prevalence rates among studies mightbe the result of such a lack of consensus; it may also contribute to the lowlevel of concordance frequently observed among clinicians on clinical diagnosticassessments.

The reduction in prevalence rates observed when the criterion of impairmentand distress is introduced to the diagnostic criteria does not necessarilychallenge the validity of the DSM-IV. Some have arguedthat the inclusion of this criterion in most DSM-IV (AxisI) diagnoses is unnecessary¹¹ because manydisorders already include in their symptom criteria impairment in functioning.Nevertheless, the DSM-IV criteria, as implementedin epidemiological interview measures that incorporate impairment and distress,such as the DISC-IV, led to prevalence rates in Puerto Rico that serve asreasonable working hypotheses about the burden of mental disorders among childrenand adolescents. The information provided by trained lay interviewers in asingle assessment session can only begin to approximate a full professionalassessment that uses behavioral observations, multiple informants, and, whereavailable, biological tests.

Economical methods to assess mental health and impairment due to mentaldisorders, such as those used in field surveys, provide the opportunity todetermine the commonness of the signs and symptoms of disorders that are proposedbecause of the nosology and to plan longitudinal studies of onset, course,and treatment of mental disorders. Perhaps more strikingly, these assessmentsare important tools for planning mental health services and for assessingchildren with serious emotional disturbance in the educational system. Althoughthe inclusion of specific clinical severity criteria in DSM-IV reduces prevalence estimates, these lower rates do not automaticallytranslate into lower service needs. Results of a comprehensive assessmentof both diagnosable psychiatric disorder and global impairment suggest thatservices are needed for an important group of children who are below the thresholdneeded for a diagnosis but are nonetheless globally impaired. These findings,if replicated, have important implications for future revisions of the DSM-IV.

Corresponding author: Glorisa Canino, PhD, Behavioral Sciences ResearchInstitute, Medical Sciences Campus, PO Box 365067, San Juan, PR 00936-5067(e-mail: gcanino@rcm.upr.edu).

Submitted for publication February 25, 2003; final revision receivedMay 29, 2003; accepted June 2, 2003.

This study was supported by National Institute of Mental Health (Bethesda,Md) grant UO1-MH54827 and by National Institute of Mental Health grant PO1-MH59876as part of Dr Chavez's minority supplement.

We thank Lizbeth Fábregas, MA, director of the study, and JoséCabiya, PhD, for his helpful comments about the manuscript.

The views expressed in this article are those of the authors and shouldnot be construed as the official position of the National Institute of MentalHealth, National Institutes of Health, Bethesda.