Quality of Care for Medicaid-Covered Youth Treated With AntidepressantTherapy

By age 18 years, it is estimated that 20% of youth will have experiencedat least 1 episode of major depression.¹ Depressedyouth are at increased risk for suicide, school failure, substance abuse,nicotine dependence, obesity, early pregnancy, and social isolation.^{2 - 3} The results of recent studies^{4 - 7} suggestthat selective serotonin reuptake inhibitor antidepressants (SSRIs) may beeffective in reducing symptoms of depression in adolescents and children.However, information is limited on the efficacy and safety of antidepressantdrugs in this age group.

Despite the paucity of studies that have been conducted among childrenand adolescents, there has been a 3- to 8-fold increase in the use of antidepressantsin children and adolescents during the last decade.^{8 - 9} Althoughdata are limited regarding the source of this increased use, primary careproviders are more likely than mental health specialists to prescribe an antidepressantfor their pediatric patients with depression.¹⁰ Inaddition, in investigations using administrative data, most antidepressant-treatedyouth have no evidence of mental health specialist contact.⁸ Therefore,indirect evidence suggests that many prescriptions are being provided by non–mentalhealth specialists, who may lack in-depth training in the use of these medications.

Because antidepressants may be prescribed by providers with minimaltraining in their use or in the treatment of depression, it is important toevaluate the quality of care that antidepressant-treated youth receive. Mostof the quality-of-care indicators, such as the RAND guidelines for adolescentdepression,¹¹ are based on the 1993 recommendationsof the Depression Guideline Panel organized by the Agency for Health CarePolicy and Research.¹² These guidelines recommendweekly to biweekly follow-up in the first 6 to 8 weeks after starting pharmacotherapyfor depression.

The Health Plan Employer Data and Information Set (HEDIS) quality-of-careguidelines for depression are also based on the Depression Guideline Panelrecommendations, but were developed for use with administrative data.¹³ Given some of the limitations of administrative data,the HEDIS guidelines are less stringent than the Agency for Health Care Policyand Research depression treatment guidelines. The HEDIS guidelines recommendat least monthly follow-up appointments in the 3 months following antidepressantprescription fill, and at least 3 months of continuous treatment for acutephase medication treatment and 6 months for the continuation phase. The mainobjective of this study was to examine if youth who were treated with antidepressantsare receiving treatment in a manner that would meet these minimal quality-of-carestandards. As a secondary aim, we evaluated the association between the receiptof treatment meeting quality-of-care standards and characteristics of thefirst treatment episode and patient demographics.

This study was conducted using Washington State Medicaid claims datacollected from July 1, 1997, through December 31, 1998. During these years,the Washington State Medicaid program covered youth up to age 19 years infamilies with incomes up to 200% of the federal poverty level. Most youthwere enrolled in a Medicaid managed care plan unless they had complex chronicmedical conditions or were placed in foster care, in which case they werecovered by a fee-for-service plan. About 7% of enrolled youth were coveredby the fee-for-service plan. Services covered by Medicaid included outpatientand inpatient health care and mental health care, as well as prescriptionmedications. There were no formulary restrictions or co-pays for antidepressantmedications during this period. Outpatient mental health services were providedby community mental health centers or mental health providers within managedcare plans, depending on the preference of enrolled families. Although novisit limit existed for community mental health centers, visits with mentalhealth specialists in the managed care system were limited to 12 per year.Administrative data captured in claims format were available for managed careand fee-for-service plan enrollees.

Children and adolescents (aged 5-18 years) were eligible to be includedin this study if they were continuously enrolled in Medicaid from July 1,1997, to December 31, 1998, and had presented with a "new episode" of depression.

Consistent with methods used by others,¹⁴ youthwere classified as having a new episode of depression ifthey had a visit coded with an International Classificationof Diseases, Ninth Revision (ICD-9) code fora depressive disorder after a 6-month window with no depressive disorder diagnosisand no antidepressant medication fill. Depressive disorder diagnosis codeswere selected based on codes outlined in the HEDIS guidelines (ICD-9 codes 296.2, 296.3, 298.0, 300.4, 309.1, and 311).¹³ Therewere 9 ICD-9 code fields available for each claim,although few records had more than 2 or 3 coded diagnoses. A youth was consideredto have depression if a code was present in any ICD-9 codefield. Youth with a bipolar disorder diagnosis at any time during the study(n = 1336) were excluded from analysis, as their needs and treatment wereanticipated to differ from those of youth with depression alone.

Antidepressant use was defined as having atleast 1 pharmacy claim for an SSRI, tricyclic, or other antidepressant (eg,bupropion hydrochloride and trazodone hydrochloride) in the 6 months followingpresentation with a new episode of depression. Antidepressants were identifiedusing national drug codes for antidepressant medications. A list of antidepressantsand the dosing guidelines used to calculate minimum doses are provided in Table 1.

Follow-up care was assessed by the number of claims for a health providercontact in the 3 months following SSRI medication fill. Codes accepted includedevaluation and management visits and mental health specialty visits. As physicianshave been noted not to code for depression in health care visits,¹⁵ we accepted any evidence of provider contact as apotential follow-up for antidepressant use. In a subanalysis, we requiredthat 1 of the 9 ICD-9 codes for the follow-up visitinclude a depression diagnostic code.

Mental health specialty visits were also assessed in the 6 months followingpresentation with a new episode of depression. The billing provider specialtywas not available for all records. In defining mental health visits, we decidedto take a conservative approach and be overinclusive rather than underinclusiveof potential mental health specialty visits. Youth were defined as havinga mental health specialty visit if they had (1) an evaluation and managementvisit coded by a psychiatrist or psychologist, (2) a visit coded with a Current Procedural Terminology code for psychotherapy orpsychiatric assessment (includes care provided by master's-level therapists,psychiatric nurse practitioners, psychologists, and psychiatrists), or (3)a state-specific code for a mental health visit in the capitated mental healthsystem. We assumed that non–mental health providers would not use mentalhealth–specific billing codes.

The number of days that antidepressant use was prescribed was calculatedby multiplying the number of pills on each prescription by the pill dose andthen dividing by the minimal acceptable daily dose for each medication, asdefined by the Depression Guideline Panel¹² andnewer guidelines for antidepressants released after the depression guidelineswere published.¹⁶ The full number of days wascounted for each prescription, even if a refill occurred in the middle ofa prior prescription, or if a new medication was prescribed before the completionof a previous prescription's supply. This is consistent with the way othershave calculated medication days.¹⁷ Medicationdays were added within classes of medications (eg, SSRI, tricyclic, or otherantidepressant) to prevent double-counting days that might be due to dualtreatment (eg, treatment with an SSRI and trazodone).

Covariates included sex, subject age, race/ethnicity, and rural or urbanresidence, as each of these covariates has been shown to be associated withthe likelihood of receiving treatment for depression.^{18 - 19}

Rural and urban residence status was determined using the rural andurban commuting area coding system developed for the Washington State areain conjunction with the Federal Office of Rural Health Policy and the Departmentof Agriculture's Economic Research Service.²⁰

Differences between managed care and fee-for-service coverage were notexamined because of complexities related to an individual's ability to changecoverage status during the 18-month study and the possibility of a managedcare beneficiary receiving certain services in the fee-for-service sector.

Data preparation was performed using SAS for Windows,²¹ andall analyses were performed using STATA 7 statistical software.²² Descriptiveanalyses were performed to assess the characteristics of youth with new episodesof depression who were treated with antidepressants, compared with those whowere not treated with antidepressants. χ² Statistics and t tests were used to examine for significant differencesamong youth who were treated with antidepressants and those who were not treatedwith antidepressants.

To examine the degree to which youth were receiving care that met HEDIScriteria standards, summary statistics were generated to describe the numberof follow-up visits each youth had received in the 3 months following fillingtheir first antidepressant prescription, as well as the number of youth whoreceived at least 3 months or 6 months of antidepressants at a guideline-leveldose. Subsequently, logistic regression methods were used to evaluate demographicand diagnosing-provider characteristics that were most associated with receivingcare that met the HEDIS standards. Finally, to assess the association betweenthe number of follow-up visits and duration of antidepressant treatment, logisticregression methods were used to assess the odds of receiving at least 3 monthsof treatment based on the number of follow-up visits received.

Of 1205 youth who met criteria for having a new diagnosis of depression,42.1% (n = 507) were treated with antidepressants. Compared with youth whomet criteria for having a new episode of depression but were not treated withantidepressants, youth who were treated with antidepressants were more likelyto be older (P<.001), female (P = .02), and white (P<.001) (Table 2).

Among youth who received antidepressants, 80.9% received at least 1prescription for an SSRI, 22.7% received at least 1 prescription for anotherantidepressant, and 11.6% received at least 1 prescription for a tricyclicantidepressant. Fourteen percent (14.2%) of youth who were treated with antidepressantsreceived antidepressant medications from more than 1 class. Selective serotoninreuptake inhibitor and other antidepressant use increased with age, whilethere was no significant age variation for tricyclic antidepressant use (Figure 1).

Fifty-two antidepressant-treated youth were hospitalized for any causeduring the 3 months following antidepressant prescription fill. These youthwere excluded from the follow-up visit analysis, leaving 455 antidepressant-treatedyouth in the final sample. Twenty-nine percent (29.2%) of antidepressant-treatedyouth had no evidence of further provider contact in the 3 months followingfirst antidepressant prescription fill, 28.1% had 1 contact, 14.5% had 2 contacts,and only 28.1% had the number of visits recommended by HEDIS guidelines (Figure 2). Depression was infrequently codedon these follow-up appointment records. When limiting the analysis to visitswith a code for depression, only 3.5% of youth had at least 3 visits in 3months. Twenty-nine percent (29.2%) of youth who received antidepressantshad 1 or more mental health specialty visits during the 3-month follow-up,compared with 40.7% of youth who did not receive an antidepressant (P<.001).

Early discontinuation of antidepressants was common in our sample (Figure 3). Focusing on SSRI antidepressants,46.6% (191/410) of youth who had filled a new antidepressant prescriptionreceived at least 3 months of treatment, and only 26.3% (n = 108) receivedat least 6 months of treatment as recommended by the HEDIS criteria. Manyyouth did not receive a sufficient supply to have an adequate trial for effectivenessof the medication: 28.3% (n = 116) of youth received less than 6 weeks ofantidepressant therapy. Younger children were less likely than older childrento receive an adequate duration of SSRI antidepressants (Figure 4). This result should be viewed with caution, as administereddoses affect duration calculations and dosing guidelines are not well establishedfor younger children.

There was a significant association between the number of follow-upvisits received and duration of antidepressant treatment. Compared with thoseyouth who had no follow-up visits, youth with at least 3 documented follow-upencounters were significantly more likely to receive at least 3 months ofSSRI treatment (odds ratio, 2.02; 95% confidence interval, 1.27-3.22).

Only 1 factor was significantly associated with increased odds for havingat least 3 follow-up provider contacts in the 3 months after filling a newprescription for an antidepressant: having a mental health–related billingcode (psychiatric Current Procedural Technology codeor state-specific mental health code) on the index diagnostic visit (oddsratio, 1.90; 95% confidence interval, 1.21-3.26). Living in a rural regionwas significantly associated with increased likelihood of receiving at least3 months of an antidepressant medication at a guideline-level dose (odds ratio,1.89; 95% confidence interval, 1.12-3.00).

This is one of the first studies to examine duration of antidepressantuse and follow-up care among youth who have been treated with antidepressantsfor a new episode of depression. Many youth in our sample received antidepressants,but few received care that would meet the minimal standards of quality careas outlined by the HEDIS criteria.

Among adult primary care patients, 25% to 30% of patients discontinueantidepressants within 1 month of starting treatment, and 40% to 50% discontinuewithin 3 months of treatment.^{16 ,23} Twenty-fivepercent of adult patients initiating antidepressant treatment receive follow-upcare that meets the HEDIS standard.²⁴ Althoughour results are similar to those of adult studies, they are of concern inpediatric populations, for whom data on the effectiveness and safety of antidepressantsare limited. The finding that more than one half of the youth discontinuedantidepressants before receiving at least 3 months of treatment suggests thatwe may need to develop health services interventions that assess and endeavorto improve the effectiveness of these medication treatments in children.

Of particular concern is the finding that almost one third of the youthwho received and filled a prescription for an antidepressant had no evidenceof any follow-up visits in the subsequent 3 months. In adults, increased frequencyof follow-up following a new antidepressant prescription is associated withlonger treatment duration and improved depression outcomes.²⁵ Inour study, we did not have information on depression outcomes; however, youthwho had more follow-up visits were significantly more likely to receive atleast 3 months of medications, although some of these visits may have beenwith a nonprescribing mental health or primary care provider. Youth with amental health specialty code on the first visit were also more likely to havereceived adequate follow-up. However, this finding does not necessarily indicatethat mental health specialists are providing better follow-up care for medicationuse. The presence of mental health specialty codes may also indicate a groupof youth who are receiving mental health therapy or who have an increasedseverity of depressive symptoms or other mental health concerns. We did nothave information regarding depressive symptom severity or mental health comorbiditiesto examine this further.

The failure to deliver adequate follow-up or duration of care may resultfrom barriers at the level of the patient, the provider, or the health system.For example, patients might discontinue medications because of inadequateeducation regarding their use, concerns about the stigma of depression diagnosisand treatment, adverse effects, or insufficient resources to follow-throughon recommended care. At the same time, providers might not be aware of guideline-leveltreatment recommendations or may think that newer antidepressants are so safethat youth do not require close follow-up. Finally, at the health system level,visit time constraints might contribute to inadequate patient education, andthe lack of systems to track nonadherence to medication use or follow-up visitsmay lead to infrequent visits or patient dropout.

In adult studies, barriers to care have been noted to occur on multiplelevels, and interventions have been developed that address barriers at thepatient, provider, and health system levels. These multimodal interventionsimprove patient education, increase frequency of follow-up, track outcomesand adherence, and facilitate return appointments to primary care providersin patients with adverse outcomes. These strategies have been successful inimproving not only treatment adherence but also depression outcomes.^{26 - 32}

It is important to note the limitations of this study. First, our datawere from a Medicaid program in a single state and, as such, may not be generalizableto other populations. In Washington State, many mental health professionalswill not accept Medicaid insurance, and the state behavioral health systemcan be difficult to access. This may result in a higher rate of antidepressantprescriptions by primary care physicians and a lack of access to the mentalhealth system and psychotherapy options.

A second limitation of these data is that we did not have informationregarding the actual dose of antidepressants taken. In counting days of antidepressantuse, we chose to use standards for minimally effective doses. This is a commonpractice in the use of administrative data, but will likely overestimate durationof treatment for youth who are taking higher doses or will underestimate theduration of treatment if youth are receiving less than guideline-level doses,which may be a particular problem for younger children, for whom there arefew dosing guidelines.

A third limitation is that we may have misclassified some youth withexisting depression as having a new episode. We tried to prevent this by includinga sufficiently long exclusion window during which youth were not receivingtreatment, but to the extent that youth were misclassified, it may have resultedin underestimates of follow-up rates and treatment duration. Finally, we didnot have access to information regarding treatment that was prescribed butnever obtained. Therefore, based on these data, we cannot further elucidateat what level barriers to receiving adequate follow-up care or treatment durationare occurring.

Despite these limitations, this article provides a new and importantglimpse into the quality of treatment that we provide to depressed youth.As the use of antidepressants increases, we need to examine the way in whichwe deliver care for these youth and to design strategies that ensure treatmentsare delivered in a manner that is safe and effective.

Corresponding author: Laura P. Richardson, MD, MPH, Child HealthInstitute, University of Washington, 6200 NE 74th St, Suite 210, Seattle WA(e-mail: lpr@u.washington.edu).

Submitted for publication June 23, 2003; final revision received December23, 2003; accepted January 20, 2004.

This study was supported in part by funding from the Nesholm FamilyFoundation, Seattle, and by the Klingenstein Third Generation Foundation Fellowshipin Child and Adolescent Depression, New York, NY (Dr Richardson), the PoncinFamily Scholarship, Seattle (Dr Richardson), and The Robert Wood Johnson FoundationDepression in Primary Care Initiative, Princeton, NJ (Dr Richardson).

This study was presented at the 2003 Annual Meeting of the PediatricAcademic Societies; May 6, 2003; Seattle.