Objective. To examine changes in antidepressant (ATD) prevalence and the sociodemographic and clinical correlates of ATD use among youths who are treated in community practice settings.
Methods. A retrospective study was undertaken using large data sets from 3 US sites. Outpatient prescription and clinical service records of youths who were aged 2 to 19 and enrolled in Midwestern Medicaid (MWM) and mid-Atlantic Medicaid (MAM) state programs and a group-model health maintenance organization (HMO) were organized into seven 1-year cross-sectional data sets from 1988 through 1994 to evaluate ATD utilization patterns.
Results. In 1994, ATD prevalence per 1000 youths was 19.10 (MWM), 17.78 (MAM), and 12.85 (HMO), which represented a consistent increase in prevalence from 1988–1994: 2.9-fold (MWM), 4.6-fold (MAM), and 3.6-fold (HMO). Despite the rapidly expanding use of selective serotonin reuptake inhibitors prescribed mainly for depression, more than half of ATD use in 1994 was still attributable to tricyclic antidepressants prescribed mainly for attention-deficit/hyperactivity disorder. ATD prevalence was generally predominant among 10- to 14-year-old boys and among 15- to 19-year-old girls. In the Medicaid populations, 42% (MAM) and 72% (MWM) of ATD-treated youths had primary care services, whereas the bulk of the remainder had psychiatric services. Attention-deficit/hyperactivity disorder followed by depression led the physician-reported primary care diagnoses associated with ATD use, whereas that diagnostic rank order was reversed for youths who received psychiatric services.
Conclusions. ATD treatments among youths substantially increased in the 1990s. This was generated primarily by primary care providers, and thus evaluations of the outcome of ATD treatment need to target primary care in addition to psychiatric providers. Longitudinal study designs are needed to evaluate the use of ATDs in youths in regard to the duration of treatment, combination medications, and the reasons for treatment.
- childhood mental disorders
- prescribing practices
Marketing data indicate that the use of antidepressant (ATD) medications in the United States rose from 40 million prescriptions in 1988—when fluoxetine was first introduced—to 120 million in 1998.1 For adults, ATDs are indicated for the treatment of a wide variety of emotional disorders,1 whereas for youths there are few evidenced-based indications for their use.2–4 Nonetheless, survey data from 1985 to 1993/1994 indicate that increases in ATD treatment have been far greater proportionally for youths than for adults.5 Although the prominent increase in fluoxetine prescriptions for youths in the United States has been featured in the lay press,6–8 little is known about community ATD treatment patterns of youths. Physician office visit surveys, eg, the National Ambulatory Care Survey, estimate national medication use patterns, but the sample sizes for treated youths are insufficient for a detailed and reliable analysis.9 For more comprehensive analysis, medication prevalence can be developed from population-based data sources, eg, health maintenance organization (HMO) and Medicaid records. Consequently, we undertook a prevalence study that was based on medical service utilization from 2 geographically separated state Medicaid programs and an HMO database to expand our knowledge of community treatment patterns of ATD use.
Data sets were assembled from 2 types of health care systems. The first 2 are outpatient data sets from 2 geographically distinct Medicaid populations, 1 in a midwestern state and 1 in a mid-Atlantic state. The third set of data comes from a group-model HMO that serves a predominantly employed population in the Northwest region of the United States. The total enrollment (continuous and noncontinuous) for youths ages 2 through 19 in 1988 and 1994, respectively, was as follows: Midwestern Medicaid (MWM), 542 431 and 578 401; mid-Atlantic Medicaid (MAM), 147 372 and 200 566; and group-model HMO (HMO), 110 900 and 121 332. Nonwhites were overrepresented in the Medicaid populations and were underrepresented among HMO enrollees according to general statistical profiles of the settings.10 The data were derived from the fee-for-service payment category that in the study years represented the vast majority of Medicaid-enrolled youths. The study was given an expedited exempt classification by the University of Maryland institutional review board.
Data for the study were obtained on the basis of the occurrence of a prescription claim for a psychotropic medication during seven 1-year intervals from 1988 through 1994. Psychotropic medications included ATDs, stimulants, antipsychotics, lithium, anxiolytics, hypnotics, anticonvulsants, and antiparkinsonians. ATDs were categorized into 3 subclasses: tricyclic antidepressants (TCAs); selective serotonin reuptake inhibitors (SSRIs); and “other ATDs,” which included trazodone, bupropion, maprotiline, and venlafaxine.
Measurements and analyses fall into 3 broad categories and were operationalized as follows.
Prevalence and Patterns
Total ATD prevalence was defined as the frequency of persons with 1 or more HMO pharmacy records or Medicaid prescription claims for an ATD medication per 1000 continuously and noncontinuously enrolled youths during the study year. ATD subclass prevalence was defined similarly for persons with a medication exposure to a TCA, SSRI, or other ATD. To examine changes in utilization, we constructed time trends based on cross-sectional annual data from 1988 through 1994.
For the 3 health sites, age-specific ATD prevalence rates were established for the 4 age strata defined by the US census categories (aged 2–4, 5–9, 10–14, and 15–19). Gender-specific ATD prevalence was reported in terms of the male:female prevalence ratios; also, the interactive effect of age and gender on ATD prevalence was assessed.
Clinical diagnoses and provider specialties associated with ATD utilization were available for the 2 state Medicaid populations. Diagnostic information was obtained from all claims generated by providers in office-based, clinic, and outpatient settings during the study year in which the prescription was recorded. Psychiatric diagnoses, recorded as International Classification of Diseases, Ninth Revision, diagnosis codes, are consistent with those of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, and were classified into 15 broad groups. From the frequency listing of diagnoses, the 5 leading diagnostic groups were attention-deficit/hyperactivity disorder (ADHD), depression, anxiety, conduct disorders, and adjustment disorders. Youths who were prescribed an ATD and given only a medical nonpsychiatric diagnosis were listed separately in the analyses. On the basis of service use recorded in the claims data file, youths who received ATDs were categorized by provider specialty into 4 mutually exclusive groups: any psychiatry service, primary care alone or with some other specialty service, medical specialty other than psychiatry or primary care, or missing (ie, no service claims in the study year or no provider specialty code). Any psychiatric services included youths who received solely psychiatric services and those who received psychiatric in addition to primary care or other specialty services. Prevalence rates were established for youths with any psychiatry service and for those with no psychiatry services in terms of ATD (total and subclass).
The population-based prevalence and 95% confidence intervals (CI) for total, subclass, and age- and gender-specific ATD use were calculated according to health care site, and time trends for the 7-year period (1988–1994) were calculated. Prevalence ratios are presented according to gender (3 sites) and ethnicity (2 Medicaid sites). A multivariable logistic regression model was used to estimate the likelihood of ATD use as a function of provider specialty (any psychiatry service group versus the no psychiatry service group), with adjustment for age, gender, ethnicity, and state Medicaid program. The regression analysis was restricted to the 2 state Medicaid databases because provider specialty and diagnostic information were not provided in the HMO data and to white versus black youths because of the small number of other minorities. The significance of the interactions of provider specialty with each state Medicaid program and with age was assessed using the log-likelihood ratio test, which approximates a χ2 distribution. An analysis was performed for the unadjusted and adjusted odds ratios and 95% CI for the main effect, covariates, and the interaction of provider specialty with age and state.
Prevalence and Patterns
The data across all 3 sites illustrate that between 1988 and 1994, many more youths were receiving ATD treatment (total as well as subclass). In 1994, ATD prevalence among 2- to 19-year-olds ranged from a low of 12.85 per 1000 for the HMO youths to 17.78 and 19.10 per 1000 for MAM and MWM, respectively (Table 1). TCAs were the most commonly used ATD subclass in 1988, but by 1994 SSRIs nearly equaled the prevalence of TCAs. Time trend analysis of the total ATD prevalence rates revealed a 3- to nearly 5-fold increase for the 7-year period beginning in 1988 (Table 1). The TCA increase was modest relative to the far more prominent increase in SSRIs and other ATDs. Nevertheless, in 1994 TCA use still represented the majority of the total ATD use across the 3 health care sites. Specifically, in 1994, 3 of the top 5 ATDs in rank order were TCAs: imipramine, nortriptyline, desipramine (MAM) and imipramine, amitriptyline, and nortriptyline (MWM).
Age-Specific ATD Prevalence
Figure 1 illustrates the 7-year trend according to age group and site. Older adolescents (ages 15–19) generally represented the major ATD utilizers as shown in the 1994 prevalence data for the HMO (25.6/1000; 839/32 734) and the MWM (38.3 per 1000; 4050/105 810). The exception was the MAM population, whose 10- to 14-year-olds who received ATDs predominated over the 15- to 19-year-olds during the 7-year period including 1994 (35/1000 and 1480/42 413 vs 26/1000 and 981/38 133).
ATD Subclass Utilization According to Age
Figure 2 illustrates the proportional use by ATD subclass and age group in 1994 for MWM. Because analyses from the MAM and HMO sites were similar to MWM and because of space limitations, only MWM data are presented. SSRI use increased with age, whereas TCA use diminished with age. Four products (trazodone, bupropion, maprotiline, and venlafaxine) constituted the “other ATD” subclass, but the 1994 proportional use of other ATDs remained small (14.0%) compared with TCAs (58.7%) and SSRIs (42.0%), regardless of age group.
Gender by Age-Specific ATD Prevalence
Overall, boys in the Medicaid program were more likely than girls to receive an ATD in 1994, whereas girls predominated in the HMO population. However, for all 3 sites, the gender effect was influenced by age. Specifically, the male-to-female (M:F) ratio of ATD prevalence rates showed a male predominance among 10- to 14-year-olds (M:F ratios of 1.6 for MWM and MAM and 1.4 for HMO). However, in the 15- to 19-year-old youths, girls predominated among MWM (M:F = 0.8) and HMO (0.4) youths, whereas the gender ratio favored boys slightly in MAM (1.2).
Diagnostic Data for ATD-Treated Youths in a Medicaid Population
In the 1994 MWM data set, 10 441 ATD youths with service claims were reviewed and 7.1% (n = 744) were excluded because no diagnosis was coded. Among the remaining youths (n = 9697), the physician specialty was missing (n = 11) or pertained to “other specialty” for too few cases (n = 39) to permit additional analysis. Consequently, data on 9647 ATD-treated youths were available to characterize the relationship between physician specialty (primary care or psychiatric services) and International Classification of Diseases, Ninth Revision diagnosis. Of these, 72.1% received primary care services and 27.9% received psychiatric services as defined in this study. In 1994, a nearly equal proportion of primary care ATD-treated youths had only medical (nonpsychiatric) diagnoses coded (51.7%) as had a psychiatric diagnosis (48.3%), whereas most of those who received psychiatric services (97%) had a psychiatric diagnosis coded and 3% had only medical diagnoses. Youths who had primary care services and only a medical diagnosis compared with those who had a psychiatric diagnosis showed no difference in mean age (12.194 vs 12.188; not significant) or in the proportion of white ethnicity (86.4% vs 83.2%; not significant), but because of the large sample size, a modest, statistically significant difference in male gender (58.0% vs 49.5%; P < .001) was observed. When those with only a medical diagnosis were compared with the psychiatric diagnosis for those who received psychiatric services, the same pattern of similarity in age, gender, and ethnicity was observed. This suggests that the absence of psychiatric diagnoses for a proportion of youths with primary care services would not alter the overall characterization of these data.
Leading Psychiatric Diagnoses
Among youths with a psychiatric diagnosis, 54.9% had only 1 such diagnosis coded during the study year, 26.8% had 2, and the remaining 18.3% had 3 or more diagnoses (range: 3–10). Table 2 shows the most common psychiatric diagnoses in relation to total and subclass ATD use for youths who received either primary care services or psychiatric services in MWM in 1994. ATD use among youths who received primary care services was more likely to be associated with ADHD, whereas youths who received psychiatric services were more likely to carry a diagnosis of depression. Regardless of specialty, depression and anxiety were the highest diagnoses among SSRI users, whereas ADHD and conduct disorder diagnoses were highest among TCA users. These data suggest that primary care use of TCAs is for behavioral problems, whereas psychiatric service provider use is more likely to be for emotional problems. It is interesting that TCAs were prescribed in association with a diagnosis of nocturnal enuresis in <1% (n = 28) of the youths in 1994.
Modeling ATD Use in 2 Medicaid Populations
Prevalence rates in 1994 revealed that ATD use and physician services differed by state Medicaid program and by age group. For example, 58% of ATD-treated youths in MAM received psychiatric services and the other 42% received primary care services, whereas in MWM, 27.9% of ATD-treated youths received psychiatric services and the other 72.1% received primary care services. Thus, psychiatric services in MAM were greater than in MWM, which is less likely to be related to medication issues than to the availability of specialty providers because there were 2.5 times more psychiatrists in MAM than in MWM.11 In addition, ATD use and the receipt of psychiatric services differed by age for each state. Logistic regression analyses were performed to assess the association between physician specialty and ATD use. The unadjusted odds ratio suggests that ATD use was 3.37 times greater (95% CI: 3.20–3.54) for those who received psychiatric services than for those who received primary care services. After adjusting for age, gender, ethnicity, and the interaction of physician specialty with age and state, the odds of ATD use was 2.48 times greater (95% CI: 1.90–3.24) among those who received psychiatric services relative to primary care services.
The study findings offer empirical evidence to support 3 important points. First, from 1988 to 1994, there was a 3- to 5-fold increase in the prevalence of ATD treatment of US youths who were younger than 20 years. The prevalence of all 3 major subclasses of ATD medication rose during this period, although the increase was greatest for SSRIs (>19-fold) and for other ATDs (4- to 11-fold) compared with TCAs (2- to 3-fold). Second, the continuing frequent use of TCAs for children and adolescents has clinical implications regarding appropriateness. Third, Medicaid service utilization data suggested that for youths who received primary care services, ATD treatment was associated more with ADHD than with depression. By comparison, ATD use among youths who received psychiatric services had the reverse pattern. Each of these major findings is considered in more detail below.
Prevalence Findings: Retrospect and Prospect
Certain findings corroborate or augment those of previous studies. For example, previous studies of youths show that the prevalence of psychotropic treatment in HMO enrollees tends to be less than in Medicaid enrollees,10,12,13 probably because more chronic cases are eligible for Medicaid insurance.14,15 Second, the prevalence of ATD treatment is greater with each successive age group of youths,16 but nearly equivalent use is found for 10- to 14-year-olds as for 15- to 19-year-olds in Medicaid populations. Also, the age-specific prevalence is influenced by gender such that boys who are younger than 15 years are prescribed more ATDs than are girls, whereas the reverse is the case for those who are older than 15 years.10 These findings indicate that younger youths are being identified as candidates for ATD use, suggesting the need for more verification of the reasons for this use. The Medicaid diagnostic patterns reported here suggest that some of this use is likely to be depression co-occurring with ADHD, but de novo treatment of ADHD is also plausible as well as the expanded identification and acceptance of treatment for depression among 10- to 14-year-olds. The research implications of these utilization patterns are addressed later.
Utilization patterns for older adolescents tend to resemble adult ATD treatment patterns. For example, physician surveys of the ATD treatment visits by US adults based on a national probability sample in 1985 and 1993/19945,17 revealed trends similar to those reported herein for adolescents aged 15 to 19 years. The similarities between adults and older adolescents in ATD use include a 2-fold or greater increase in ATD prevalence during the study periods, an SSRI proportion approximating at least 50% of ATD use by 1994, and a preponderance of females receiving SSRIs related primarily to a diagnosis of depression.
Clinical Practice Implications of TCA Utilization
In contrast to 15- to 19-year-old ATD treatment patterns in the present study, youths who were younger than 15 years and treated with ATDs had a different profile of ATD treatment. TCA use was most prominent of the 3 ATD subclasses, the recipients were mostly male, the leading associated diagnoses were ADHD and conduct disorder, and multiple diagnoses were less common. Even with the surge in SSRI utilization, TCA prevalence among youths rose 2- to 3-fold from 1988 through 1994; furthermore, 3 of the 5 most commonly prescribed ATDs in 1994 were TCAs. As to indications for use, TCAs were seldom associated with the diagnosis of nocturnal enuresis but were largely associated with ADHD, being a second-line treatment recommendation for this disorder.18 Notably, 40% to 45% of TCA use was associated with a diagnosis of depression, although clinical trial data do not support their use in youths for that indication.19–21
Primary Care Use of ATDs Among Medicaid Youths
This study highlights the role of primary care in the delivery of health services among the Medicaid youths. The majority of ATD use occurred in youths who were receiving primary care services and did not have contact with a mental health provider during the study year. A smaller proportion received ATDs along with psychiatric services. The findings are consistent with recent US primary care treatment trends showing the important role that primary care providers have in the prescription and management of ATDs.22 Between the late 1970s and the mid-1990s, the identification and diagnosis of psychosocial problems in US youths by pediatricians increased 2.8-fold.23 During that span of time, there was a similar and presumably related increase in psychotropic medication (primarily “off label”24) prescribed for youths largely through primary care settings.25,26 This study likewise documents a very prominent increase in the prevalence of ATD treatment of youths in primary care during the period from 1988 through 1994.
It should be noted that a substantial proportion of the service claims of youths who received primary care treatment had only medical diagnostic codes, which limits knowledge of the reasons for treatment. Such findings are not uncommon, as indicated in a recent analysis from National Ambulatory Medical Care Survey that showed that >30% of physician office visits for youths involving a psychotropic drug mention do not include a psychiatric diagnosis.27 The absence of a psychiatric diagnosis in youths suggests that the main reason for the primary care visit was medical and that the recording of a secondary diagnosis was neglected. Other explanations include the physician’s reluctance to diagnose a mental disorder to avoid stigma or uncertainty about the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition diagnosis. In the present study, because the data are not from a longitudinal cohort, it is possible that a psychiatrist’s diagnostic assessment occurred in a previous year and therefore would not appear in the study data set.28
Several limitations of the study data should be considered in interpreting the findings. First, data from the 2 Medicaid states may not generalize to Medicaid populations from other regions of the country. However, the remarkable consistency across these 2 large geographic areas gives some confidence that the findings generalize to other regions. Also, the absence of a more representative sample of youths with employer-based insurance makes it impossible to know whether those with Preferred Provider Organization- and Independent Practice Association-based insurance have similar treatment patterns as in the group model HMO treatment setting. Second, there is no variable in the claims data to link directly the prescription claims with the prescribing physician specialty, so their co-occurrence during the study year was used to define the linkage, which is less than optimal.
Although the diagnoses of community providers tend to have low interrater reliability, we believe that physician-based diagnostic data provide researchers with vital information needed to understand community practice patterns. Furthermore, to reduce diagnostic variability, we grouped the coded diagnoses into broad diagnostic groups as suggested by Carson et al.29
Several findings have substantial implications for additional research. First, the evidence base for SSRI use to treat depression among youths who are younger than 15 years is meager,2–4,30,31 and must be strengthened in view of prominent utilization (41% were prescribed an SSRI in 1994 in association with the diagnosis of depression). Second, research shows that a substantial proportion of SSRI use occurs in combination with stimulants,32 a combination for which there are no effectiveness or safety data. Furthermore, distinguishing between the behavioral adverse effects of SSRIs, eg, agitation, precipitation of mania,2,33 and comorbid or de novo psychopathology, is a major research challenge. Third, because of the major role of primary care providers in the ATD treatment of youths and the overall increase in the prevalence of such treatment in this population, evaluations of outcome will need to include both primary care and psychiatric services. Fourth, efforts to establish effectiveness and long-term safety of ATD treatments for youths will require the development and inclusion of epidemiologic methods to assess outcomes better. For these outcomes, a longitudinal study of cohorts in the usual practice settings is needed, after which additional drug evaluation standards for effectiveness and long-term safety should evolve. Recent meetings sponsored by the Food and Drug Administration and the National Institute of Mental Health to promote discussion along these lines have involved policy makers, pharmaceutical industry leaders, and academic researchers.34 If these meetings lead to creative developments, then it is likely that a broader, more comprehensive model of drug development will emerge with rigorous phase 4 (postmarketing surveillance) activities, including research, monitoring, and training, to address prescription drug use in a community-based treatment model that reflects a public health perspective. Undoubtedly, the collaboration of large health systems, experienced physicians in the usual treatment settings, academic research, and public funding will be essential to move this agenda forward.
The expanded utilization of ATDs for the management of behavioral and emotional disorders of youth in the 1990s was prominent, and additional evaluation is needed to strengthen the evidence-base for their use. Toward this end, community-based descriptive studies should measure the duration of drug therapy and the use of both monotherapy and combinations. Outcomes of ATD treatment in the usual practice setting should include both primary care and psychiatric settings to assess the extent and quality of drug and psychosocial interventions as well as satisfaction with treatment.
This study was supported by funding from the National Institute of Mental Health Services Branch (R01 MH55259) and the George and Leila Mathers Foundation.
Richard E. Johnson, PhD, and Linda Phelps provided assistance at several stages in the design or analysis of this study.
- ↵Foote SM, Etheredge L. Increasing use of new prescription drugs: a case study. Health Aff (Millwood).2000;19 :165– 170
- ↵Strauch B. Use of antidepression medicine for young patients has soared. New York Times.1997; Aug 10:1, 24
- Crowley M. Do kids need Prozac? Newsweek.1997; Oct 20: 73–74
- ↵Chua-Eoan H. Escaping from the darkness. Time.1999; May 31: 44–49
- ↵Peterson BD, West J, Tanielian TL, et al. Mental health practitioners and trainees. In: Mandersheid RW, Henderson MJ, eds. Mental Health, United States, 1998. Washington, DC;1998:214– 246. DHHS Publ. No. [SMA] 99-3285
- ↵Shatin D, Levin R, Ireys HT, Haller V. Health care utilization by children with chronic illnesses: a comparison of Medicaid and employer-insured managed care. Pediatrics.1998;102(4) . Available at: http://www.pediatrics.org/cgi/content/full/102/4/e44
- ↵Jadad AR, Boyle M, Cunningham C, Kim M, Schachar R. Treatment of Attention-Deficit/Hyperactivity Disorder. Evidence Report/Technology Assessment No. 11. Rockville, MD: Agency for Healthcare Research and Quality (AHRQ);1999:50 . AHRQ Publ. No. 00-E005
- ↵Hazell P, O’Connell D, Heathcote D, Robertson J, Henry D. Efficacy of tricyclic drugs in treating child and adolescent depression: a meta-analysis. BMJ.1995;310 :897– 901
- ↵Kelleher KJ, McInerny TK, Gardner WP, Childs GE, Wasserman RC. Increasing identification of psychosocial problems: 1979–1996. Pediatrics.2000;105 :1313– 1321
- ↵Zito JM, Safer DJ, dosReis S, et al. Ten Year Prevalence Trends for Psychotropic Medication From Two Health Service Systems. Presented at the New Clinical Drug Evaluation Meeting; June 1–4,1999; Boca Raton, FL. Abstract
- ↵Safer DJ, Zito JM. Pharmacoepidemiology of methylphenidate and other stimulants for the treatment of ADHD. In: Greenhill LL, Osman BB, eds. Ritalin: Theory and Practice. 2nd ed. Larchmont, NY: MA Liebert, Inc;2000:7– 26
- ↵Carson JL, Ray WA, Strom BL. Medicaid databases. In: Strom BL, ed. Pharmacoepidemiology. New York, NY: John Wiley & Sons;2000:307– 324
- ↵Milin RP, Simeon J, Spenst BA. Double-Blind Study of Paroxetine in Adolescents With Unipolar Major Depression, #NR-67. Presented at the American Academy of Child and Adolescent Psychiatry Annual Meeting; October1999; Chicago, IL. Abstract
- ↵Pliszka SR, Carlson CL, Swanson JM. ADHD With Comorbid Disorders. New York, NY: Guilford Press;1999:115
- ↵Vitiello B. Psychopharmacology for young children: clinical needs and research opportunities. Pediatrics.2001;108 :983– 989
- Copyright © 2002 by the American Academy of Pediatrics