Mental Health and Substance Abuse Treatment Utilization among Individuals Served by Multiple Public Agencies in Three States Jeremy W. Bray, PhD Keith L. Davis, MA RTI International* Linda Graver Don Schroeder, PhD Thomson Medstat Jeffrey A. Buck, PhD Joan Dilonardo, PhD Rita Vandivort, MSW SAMHSA July 14, 2004 *RTI International is a trade name of Research Triangle Institute. Key words: MH/SA service utilizat ion, Medicaid, state MH/SA agencies Address correspondence to: Jeremy W. Bray, PhD Behavioral Health Economics Program RTI International 3040 Cornwallis Road Research Triangle Park, NC 27709 Phone: (919) 541-7003 Fax: (919) 541-6683 E-mail: bray@rti.org Acknowledgments: Funding for this study was pr ovided by the Substance Abuse and Mental Health Services Administration (SAM HSA). Outstanding comments and suggestions on this research were pr ovided by members of the Integrated Database Technical Expert Panel, which comprises a wide variety of researchers and treatment providers from several states and federal agencies. Excellent research assistance on this study was provided by Janet Cummings of RTI. 1 Author Information Jeremy W. Bray, PhD Senior Research Economist Behavioral Health Economics Program RTI International 3040 Cornwallis Road Research Triangle Park, NC 27709 Phone: (919) 541-7003 Fax: (919) 541-6683 E-mail: bray@rti.org Keith L. Davis, MA Research Economist Behavioral Health Economics Program RTI International 3040 Cornwallis Road Research Triangle Park, NC 27709 Phone: (919) 541-1273 Fax: (919) 541-6683 E-mail: kldavis@rti.org Linda Graver Senior Research Leader The MEDSTAT Group 5425 Hollister Avenue, Suite 140 Santa Barbara, CA 93111-2348 Phone: (805) 681-5879 Fax: (805) 681-5888 E-mail: linda.graver@medstat.com Don Schroeder, PhD Senior Programmer The MEDSTAT Group 5425 Hollister Avenue, Suite 140 Santa Barbara, CA 93111-2348 Phone: (805) 681-5868 Fax: (805) 681-5888 E-mail: don.schroeder@medstat.com 2 Jeffrey A. Buck, PhD Director of Organi zation and Financing Center for Mental Health Services Substance Abuse and Mental He alth Services Administration 5600 Fishers Lane, Room 15-87 Rockville, MD 20857 Phone: (301) 443-0588 Fax: (301) 480-8296 E-mail: jbuck@samhsa.gov Joan Dilonardo, PhD Social Science Analyst Center for Substance Abuse Treatment Substance Abuse and Mental Health Services Administration 5515 Security Lane, Room 7-206 Rockville, MD 20852 Phone: (301) 443-0555 Fax: (301) 480-3045 E-mail: jdilonar@samhsa.gov Rita Vandivort, MSW Public Health Analyst Center for Substance Abuse Treatment Substance Abuse and Mental Health Services Administration 5515 Security Lane, Room 7-198 Rockville, MD 20852 Phone: (301) 443-0789 Fax: (301) 480-3045 E-mail: rvandivo@samhsa.gov 3 Mental Health and Substance Abuse Treatment Utilization among Individuals Served by Multiple Public Agencies in Three States 4 Abstract Patterns of mental health (MH) and substa nce abuse (SA) treatm ent utilization among populations receiving services through multiple public programs are not well known. This study examines the extent to which populations with MH and/or SA conditions utilize treatment services through Medicaid and St ate MH/SA Agencies. Data are from the SAMHSA Integrated Database, a multi-year file for three states combining Medicaid and State MH/SA Agency administrative data into a uniform database. Although populations with co-occurring conditions and those served by both Medicaid and State MH/SA Agencies have substantial contact with the public treatment system, a majority of the MH/SA populations examined here utilize few services over brief periods of tim e. Utilization is most limited among individuals with MH-only conditions and those served exclusively by Medicai d. While a lack of data on clinical outcomes prevents us from drawing conclusions about the e ffectiveness of MH/SA serv ices, results of this analysis do indicate that public pr ograms in the states examined here do not provide services that are primarily utilized on a frequent or chronic basis. 5 Introduction A common belief among many health care professionals is that individuals with mental health and/or substance abuse (MH/SA) conditions utilize treatme nt services frequently over long periods of time. Studies suggest, however, that the majority of privat ely insured individuals utilize relatively few behavioral health se rvices over brief, disc rete periods of time. 1 -8 Cohen and Cohen9 refer to the discrepancy between the percei ved and actual use of MH/SA services as the 223clinician222s illusion,224 whereby long-term patients dominate clinicians222 time, use the vast majority of services, and thus create an unr epresentative impression of the general MH/SA population with regard to treatm ent frequency and duration. More over, characterizations that MH/SA populations remain in pub lic treatment for long periods of time do not support the recovery-based approach in both the MH and SA fields227that persons seeking MH/SA services often receive effective treatment a nd do not need treatment chronically.10, 11 Although limited service utiliza tion has been shown for Medicaid beneficiaries with MH conditions,12 few studies have examined the use of MH/SA services among populations covered by multiple public agencies. The lack of resear ch on these populations is due primarily to limited data. Because state organizations ma naging the delivery of MH/SA services often operate in isolation of one another, informati on about MH/SA service ut ilization resides with each individual agency.13 Databases containing informati on on individuals receiving MH/SA services through multiple public agencies are th erefore rare and typically incomplete. Fragmented data have impeded the efforts of re searchers and policymakers to determine whether service utilization varies between public agenci es and between individuals with single and co- occurring MH/SA conditions. Such information may greatly benefit state policy makers in 6 making difficult decisions about the distribution of scarce resour ces for the provision of MH/SA services. The purpose of this study is to describe patterns of utilization of mental health and/or substance abuse (MH/SA) treatment services provided through Medi caid and State MH/SA Agencies. A unique data source, the Integrated Database (IDB), is used to examine the length of time MH/SA patients in three states remain in th e public treatment system, how often they utilize services, and through which agencies (Medicaid, St ate Agencies, or both) they receive services over a three-year study period. This study also examines the extent to which serv ice utilization varies between individuals with single or bot h MH and SA conditions. To the authors222 knowledge, this study is the first to present this information for populations receiving MH/SA services through multiple public ag encies over a multiyear period. Data and Methods Overview To address the lack of complete information on populations receiving public MH/SA services, the Substance Abuse and Mental Health Services Admi nistration (SAMHSA) initiated an effort in 1996 to integrate disparate sources of data on MH/SA services. The result of this effort, the IDB, assembles information from three types of state organiza tions: State Medicaid programs, State MH Agencies, and State SA Agencies. The IDB links service record information on MH/SA treatment utilization for e ach person into a uniform database. Because the IDB combines information for individuals who receive services under multiple public programs, the IDB thus provides a more complete picture of the MH/SA clients seen in more than one part of the state-supported MH/SA trea tment system. The IDB contains person- and 7 service-level data for all such cl ients within a state. For a fu ll description of the methodology used to link IDB service records across state organizations, see Whalen et al.14 The IDB contains administrati ve service records for indivi duals receiving public MH/SA services through Medicaid and/or State MH/SA Agencies and en compasses three full calendar years (19962261998) for three states: Delaware , Oklahoma, and Washington. The three participating states were chosen based on their availability of elec tronic data, the ability of their data systems to link clients across agencies, and state interest in the IDB project. The IDB also contains information on patient demographics, such as age, sex, race, and urban/rural location, as well as information on Medicaid eligibility st atus, MH/SA diagnosis codes, providers, and Medicaid drug prescriptions and othe r Medicaid medical records. Study Population The study population for this analysis consists of individuals who had a primary MH or SA diagnosis or who received any MH or SA service during the study period. MH/SA diagnoses are defined using codes based on the Internati onal Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-C M). MH/SA diagnoses are iden tified using ICD-9-CM codes listed in Coffey et al.13 Clients with missing diagnoses were selected for the study population based on evidence of having received an MH/SA service. MH/SA service categories were created using several criteria, including source of record and service description. 13 After identifying the study population based on diagnosis or use of service, persons older than 64 and persons who changed age category (youth to adult or adult to elderly) du ring the study period are excluded. Excluding persons older than 64 elim inates 7.9% of all clients in the IDB study population while exclusion of persons who changed age category eliminates an additional 1.2%. 8 After making these exclusions, roughly 70 percen t of the study population in each state is classified as adult (ages 18 to 64), with the remaining 30 percen t classified as youth (ages less than 18). Beginning in 1998, State MH Agency reco rds from Washington did not include information on specific outpatient service dates, but rather only the month of service and the number of service encounters within a month. As a result, encounter date s are evenly assigned to individuals within each month to approximate service use pa tterns similar to those seen in 1996 and 1997. Although this method does not reflec t the true date-speci fic service use of Washington service users in 1998, it is more r ealistic than the alternative of assigning all observed encounters within a month to a single date. Service encounter dates created in this manner account for roughly 35 percent of all MH /SA service dates from Washington across the three-year study period. Client Classification Individuals included in the an alysis are classified and examined on two major domains: (1) service agency and (2) MH/SA category. Servi ce agency refers to the data source (Medicaid or State MH/SA Agency) from which each IDB reco rd was obtained and allows us to generally identify individuals who receive MH/SA services through Medicai d only, through State Agencies only, or through both Medicaid and State Agencies. I ndividuals classified as having MH/SA service records in both Medicaid and State Agency databases, however, do not necessarily receive services th rough both auspices concurrently. A client with one Medicaid record at the beginning of the study period and one State Agency record at the end of the study period, for example, is classified as ha ving received services through both auspices. In some 9 cases, the same MH/SA service record appears on both the Medicaid and State Agency databases. Overlapping records may occur if Me dicaid reimburses a bill but the State Agency provides the service. To avoid ove rstating utilization rates, only one service date is counted for cases in which a service user, provider, servi ce, and service date ar e reported on both the Medicaid and State Agency databases. Additiona lly, individuals with these types of records are classified as receiving services through bot h Medicaid and State Agencies. For further information on the reconciliation of overlapping se rvice records in the IDB, see Coffey et al.13 The second domain on which individuals are cl assified and analyzed is MH/SA category, which is used to identify indi viduals who had services for onl y MH conditions (MH-only), only SA conditions (SA-only), or co-occurring (both MH and SA) conditions during the study period. Service users are assigned to MH-only and SA-only categories based on primary diagnosis, but secondary diagnostic information was considered for co-occurring conditions. Individuals are classified as having co-occurring conditions if th ey had any of the following within the three- year study period: (1) both a primary MH and SA diagnosis, (2) a primary MH and secondary SA diagnosis, or (3) a primary SA and secondary MH diagnosis. In th e absence of diagnosis information, MH-only and SA-only classifications were assigned based on the type of service received during the study period. For cases in which diagnosis information was not available, individuals were classified as having co-occurring conditions based on evidence of receiving both a MH and SA service. Individuals classified as having co-occurring conditions did not necessarily have MH and SA conditions concurre ntly. A client with a MH record at the beginning of the study period and a SA record at the end of the study period, for example, is classified as having co-occurring conditions. 10 Individuals served exclusively by Medicaid make up 20 to 40 percent of the study population across all three states, while 45 to 66 percent are served exclusively by State Agencies. The proportion of individuals serv ed by both Medicaid and St ate Agencies varies from 12 to 36 percent across the th ree states. Most service users (55 to 70 percent across all three states) in the study populati on are classified as MH-only, while a much smaller proportion of clients (16 to 28 percent) ar e classified as SA-only. Indivi duals classified as having co- occurring conditions make up 10 to 17 percent of the study population across all three states. Utilization Measures Medians and frequency distribut ions of individuals222 length of service window, number of total MH/SA encounter dates, and frequency of MH/SA service use are presented to examine the level of contact individuals in the study population have with th e public treatment system. A service window is defined as the number of days between an individual222s first and last observed MH/SA service record during the three-year study period. An individual222s total number of service encounter dates is defined as the count of unique date s over the entire study period on which they had an administrative record with at least one MH/SA diagnosis or service. Service encounter dates occurring within a single inpatient stay are considered to be distinct and separate encounter dates. To present a more comprehensive picture of public MH/SA service utilization, the concepts of service window lengt h and number of encounter date s are combined to create four mutually exclusive categories of service utilization: (1) single encounter date, (2) short-term, (3) occasional, and (4) frequent ut ilizers of the public MH/SA treatment system. Single encounter date utilizers are defined as persons with only one encounter date during the entire study period. 11 Short-term utilizers are defined as individuals wi th a service window of three months or less but more than one encounter date. Occasional utiliz ers are defined as thos e with a service window greater than three months but fewer than 10 encounter dates. Frequent ut ilizers are defined as persons with a service window gr eater than three months and 10 or more encounter dates. Results are presented for each state side-by-si de to aid readers222 comprehension of state- specific results and to identify within-state trends that appear similar across the three states. However, comparisons of MH/SA utilization betw een states should not be made because state programs managing the delivery of MH/SA services differ in many dimensions (e.g., MH/SA program financing, organization, benefits, provider payment arrangements, available settings for care, and provider networks). For further in formation on the organizational framework of MH/SA service delivery in each state, see Coffey et al.13 Results Because the IDB spans a three-year period, indivi duals appear in the database in different years and for varying lengths of time. A basic but important result of this analysis is that the majority of MH/SA service users (between 60 an d 73 percent across all three states) appear in the IDB during one and only one year of the study period, while a much smaller proportion of individuals (11 to 17 percent) a ppear in all three years. Betw een 14 and 20 percent of service users across all three states ha ve MH/SA service records in tw o consecutive years of the study period, while a very small proportion of individua ls (less than three pe rcent across all three states) have service record s in 1996 and 1998 but no records in 1997. These results are consistent with patient turnover rates estimated in other studies of Medicaid populations (e.g., [15]). 12 Length of service window Table 1 presents the distribution of service windows and suggests th at the majority of MH/SA service users in the states examined here have relatively brief contact with the public treatment system. Half of all MH/SA service users in each state, for example, are present in the treatment system for 139 days or less over a three year period while one quarte r of all individuals has service windows of 8 days or le ss. Contact with the treatment sy stem is particularly brief for those served exclusively by Medica id, as one quarter of these individuals have a service window of only one day. Persons receiving servic es through both Medicaid and State Agencies, however, appear to have a substa ntially longer period of contact with the treatment system than those receiving services through Medi caid or State Agencies alone. Among all service users, for example, 50 percent of those receiving services through both auspices ha ve a service window of at least 344 days compared to only 135 days for individuals served by Me dicaid alone or by State Agencies alone. Table 1 also indicate s that service window length varies by MH/SA category. Individuals with co-occurring conditions, for example, ge nerally have a lengthy service window (ranging from 302 to 465 days at the median across all thre e states) and remain in treatment more than four times longer (at the median) than individu als with single MH or SA conditions. For individuals with MH-only conditions , however, contact with the treat ment system is particularly brief as one quarter of these individuals in each state are present in the treatment system for only one day. Finally, individuals w ith MH-only conditions in two st ates have a shorter period of contact with the treatment system at the median than SA-only service users. 13 Total service encounter dates Table 2 presents the distribution of total serv ice encounter dates. Half of all service users in each state have 12 or fewer MH/SA serv ice dates over the thre e-year study period. While individuals served by bot h Medicaid and State Agencies have the greatest number of encounter dates over the study period, those served exclusively by Medicaid appear to have the fewest. Low intensity of utilization among Medicaid-only service users is further pronounced in that 75 percent of these individu als across the three states have fewer than 23 encounter dates over the three-year study period. Table 2 also shows that the number of se rvice dates varies by MH/SA category. As expected, individuals with co-occurring conditio ns have a higher median number of service encounter dates than those with a single MH or SA condition. Individuals with co-occurring conditions in each state, for example, have at least 20 more service dates at the median than those with MH-only conditions an d at least 11 more at the median than those with SA-only conditions. Moreover, three quarter s of individuals with co-occurr ing conditions have at least 8 encounter dates over the study period, and in two states three quarters of individuals with SA- only conditions have at least 6 encounter dates. Additionally, MH-only patients in two states have at least 20 fewer encounter dates at the median than th ose with SA-only conditions. Limited utilization among MH-only service user s relative to those with both co-occurring and SA-only conditions is further pronounced in that one quarter of all MH -only service users in each state have only one service encounter date over the study period. When MH-only patients are served by both Medicaid and State Agencies, however, service encounte r dates for this group rise substantially (to at least 11 enco unter dates) in 2 of the 3 states. 14 Levels of MH/SA service utiliz ation presented in Table 2 ar e somewhat lower than those found at the national level in previous studi es. Two recent studies by Olfson et al.,16, 17 for example, examined national trends in outpatien t treatment for depressi on and found that adults and adolescents had on average 8 to 9 annual encounters for depression over the 1996-1999 period. While still brief in duration, the number of annual behavioral he alth encounters found by Olfson et al. is somewhat higher than the 3-year levels found in the IDB. MH/SA service utilization categories Table 3 combines the concepts of service window length and number of encounter dates to classify MH/SA patients in the study populati on into four mutually exclusive categories. These categories consist of single encounter utilizers (i ndividuals with a single encounter date), short-term utilizers (individuals with a servi ce window of three months or less), occasional utilizers (individuals with a se rvice window greater th an three months, but less than 10 encounter dates), and frequent utilizers (i ndividuals with a service window greater than three months and more than 10 encounter dates). Table 3 presen ts the percentage of individuals in the study population that fall into each category. Additional evidence presented in Table 3 s uggests that the majority of MH/SA service users in the states examined here do not rece ive frequent care over l ong periods of time. Specifically, a substantial proportion of all MH/SA service users (at least 18 per cent across all three states) have only one service encounter date over the entire three-year study period. Additionally, after combining singl e encounter and short-term u tilizers, roughly half of all persons (44 to 54 percent across al l three states) are in the pub lic treatment system for three months or less. 15 Table 3 also provides additional evidence that individuals served by both Medicaid and State Agencies have the most contact with the public treatment system while those served by Medicaid only have the least contact. Across all three states, for example, roughly 65 to 81 percent of individuals served by both auspices are classified as frequent utilizers while only 5 percent or less are classified as single encounter utilizers. In c ontrast, only 35 percent or less of individuals served by Medicaid only across all three states are fre quent utilizers while as many as 52 percent of individuals served ex clusively by Medicaid have only a si ngle encounter date. Finally, Table 3 provides further evidence suppo rting the finding that individuals with co- occurring conditions have substantially more c ontact with the public treatment system than individuals with single MH or SA conditions. Specifically, individuals with co-occurring conditions are less likely to have a single enc ounter date and are more likely to be frequent utilizers of public MH/SA services than indi viduals with MH-only or SA-only conditions. Results presented in Table 3 also support the fi nding that SA-only service users generally have greater contact with the public tr eatment system than MH-only serv ice users. In two states, for example, those with SA-only conditions are su bstantially less likely to have only a single encounter date over the three-ye ar study period than those with MH-only conditions. Moreover, roughly half of all individuals with SA-only cond itions in two states are classified as frequent utilizers compared to only one third of MH-onl y service users classified as frequent utilizers during the three-y ear study period. Sensitivity analysis Service window and encounter date distributions presen ted in Tables 1 and 2 are potentially inflated as a result of individuals who are institutiona lized or receive services in an inpatient or other long-term setting. Tables 1 and 2 were reproduced (available upon request 16 from the authors) excluding those who received treatment in long-term settings and found only a minimal decrease in the median and upper perc entiles of service window length and total encounter dates. The potentially confounding effect of long-term service users is therefore quite small. Tables 1-3 were also reproduced separately for youths (ages 0-17) and adults (ages 18- 64) to detect differences in service utilization by age. Serv ice window length and number of total encounter dates were found to be lower among youths than among adults. In two states, youths were also found to be more likely to have a single encounter date and less likely to be frequent utilizers of public MH/SA services. Utilization among individu als served by Medicaid only, however, was found to be higher for youths th an for adults, a result that may reflect the youth-specific focus of many outreach initiatives implemented by state Medicaid programs. Limitations Results presented in this paper should be interpreted with cau tion, as this study has several limitations. First, comparisons of servic e utilization between stat es should not be made because the organizational framework and policie s under which services are delivered varies considerably across the states. Second, the li mited time frame of the IDB prevents us from observing data on individuals who utilized MH/SA services either before 1996 or after 1998. As a result, it is possible that some individuals w ho appear in the treatment system briefly at the beginning or end of the study period are in fact high utilizers of MH/SA services but are not captured as so in the three-year window. A th ird limitation of this study is that information on prescription drug utilization is not considered. It is thus possibl e that some individuals in the study population have few encounters because they are receiving treatment in the form of a medication-based maintenance program. Finally, because this analysis focuses on only discrete 17 events of service utilization without respect to clinical MH/SA outco mes or prevalence and severity of MH/SA conditions, conc lusions about the adequacy of treatment services provided in the states examined here cannot be drawn. Desp ite these limitations, results presented in this study have important implications that may aid states in the de livery and management of public MH/SA services. Implications for Behavioral Health Services Previous studies have shown that privately insured MH/SA patients generally receive few treatment services over brief periods of time. 1 -8 Results of this analysis indicate a similar pattern for individuals receiving MH/SA services through multiple public agencies. These results may support those from previous studies that prom ote a recovery model227that persons seeking MH/SA services often receive effective treatm ent and therefore do not need treatment on a continual basis. While a lack of data on clinical outcomes preven ts us from drawing conclusions about the effectiveness of MH/SA services, the results do indicate that public treatment programs in the states examined here do not primarily provi de services that are utilized on a frequent or chronic basis. The results presented in this study also indi cate that MH/SA service use varies across funding agencies, as individuals served by both Medicaid and State Agencies have substantial contact with the treatment system while thos e served only by Medica id have very limited contact. Specifically, in dividuals served by both Medicaid an d State Agencies are generally the most likely to be frequent utilizers and the least likely to have a single en counter date while those served by Medicaid alone are the most likely to have a single en counter date and the least likely to be frequent utilizers. 18 A major implication of the results presented here relates to the general finding that the majority of the MH/SA populations in the states examined here display limited use of public MH/SA services over brief periods of time. Given such transitory patterns of service use, it is likely that state organizations managing the delivery of MH/SA serv ices are not funding treatment of the same individuals from year to year. While most individuals do not remain engaged in public treatment from year to year, they appear much more likely to do so when they have co-occurring conditions. Service use among clients with MH-only conditions was shorter than for those with co-occurring conditions but still of significant duration. With little data on the severity of client conditions, it is difficult to determine whether the level of service use observed here is adequate for favorable client outcomes. Analysis results also indicate that treatment utilization for individuals with SA-only conditions was more intense for a brief initial period of time, but continuity of services after the initial time period was relatively absent. Greater intensity of in itial service use is encouraging, as it may indicate successful treatm ent engagement. The lack of s ubsequent treatment utilization after an initial period of intense service use among clients with SA-only conditions may be the result of several factors that ca nnot be detected in the IDB data , including patient follow-up with non-billing services such as Alcoholics Anonymous . Given the relapsing nature of addictions, however, the lack of con tinuing care observed in this study may raise the possibili ty that needed services are not being utilized a nd further study is thus warranted. The generally limited level of treatmen t utilization among the MH/SA populations examined here may be the result of several factors that have not been accounted for, such as state-specific managed care restrictions18 or participation in Aid to Families with Dependent Children (AFDC) or Temporary Assistance for Needy Families (TANF), criminal justice 19 programs, nonbilling programs such as Alcoholics Anonymous, or other Federal programs that provide resources for the use of me dical services. It is also unclear what effect evaluation and consultation visits have on utilization rates. Pr eliminary analyses suggest that evaluation visits may account for a substantial number of the singl e encounter date utilizers observed in this study. It may also be possible that some MH /SA patients require fe wer service encounters because of participation in medication-based ma intenance programs in addition to therapy. To the extent that a combination of medication and therapy is more efficacious than either treatment alone,19-21 short treatment durations and few encount er dates may be the preferred scenario among clinicians and policy makers. Investigating the impact of managed care penetration, participation in other Federal programs, the use of medications, and other factors on MH/SA service use is an important di rection for future studies. An additional implication relates to the in terpretation given to the lower levels of treatment utilization f ound among individuals served by Medica id only. Specifically, differences in utilization between Medicaid and State Agencies may in part be accounted for by differences in the populations covered. For instance, Medi caid populations have a large number of TANF- eligible families that access MH/SA services at a lower rate and may need only one or a few service encounter dates in the public specialty treatment system . Low utilization among those served by Medicaid only may also be the resu lt of Medicaid providers engaging State MH/SA Agencies for individuals with more intensive treatment needs. Low utilization among Medicaid- only users may also reflect differences in the types of services covered under Medicaid. Moreover, low utilization among Medicaid-only serv ice users may be even less of a concern considering the relatively high rate of utilization found among i ndividuals receiving services through both Medicaid and State Agencies. 20 Finally, results of this analys is also indicate that individu als with co-occurring conditions have a higher level of contact wi th the public MH/SA treatment system than those with only MH or only SA conditions, and those with SA-only cond itions have higher levels of service use than those with MH-only conditions. It is reassuring to find, howev er, that individuals who are potentially the most severely ill (e.g., those with co-occurring conditions), have more extensive contact with the public treatment system than i ndividuals with a single MH or SA condition. Although several important findings are presente d in this paper, further research is needed to gain a more complete understanding of the delivery, financing, and utilization of public sector MH/SA service utilization. In li ght of previous resear ch documenting the cost offsets associated with both MH and SA treat ment ([22] and [23] respectively), one area of future research would be to examine the subseque nt general health care utilization of clients in the IDB. The IDB is a unique and rich data source that may support such studies. 21 References 1. Taube CA, Goldman HH, Burns BJ, et al. High users of outpatient mental health services, I: Definition and characteristics. American Journal of Psychiatry . 1988;145(1):19-24. 2. Howard KI, Davidson CV, O222Mahoney MT, et al. Patterns of psychot herapy utilization. American Journal of Psychiatry . 1989;146(6):775-778. 3. Bland RC, Newman SC, Orn H. Health care utilization for emotional problems: Results from a community survey. Canadian Journal of Psychiatry . 1990;35:397-400. 4. Kisch J. Utilization of mental health services: Attrition versus aggregation. HMO Practice . 1992;6(2):33-38. 5. Widman M, Platt JJ, Lidz V, et al. Pattern s of service use and tr eatment involvement of methadone maintenance patients. Journal of Substance Abuse Treatment . 1997;14(1):29-35. 6. Wu L, Ringwalt CL, Williams CE. Use of s ubstance abuse treatment services by persons with mental health and substance use problems. Psychiatric Services . 2003;54(3):363-369. 7. Wall MM, Stromberg KD, Pothoff S, Kane RL . Alcoholism treatment episodes validly defined using mental health care utilization records. Journal of Clinical Epidemiology. 2004;57(4):373-380. 8. Schoenbaum M, Zhang W, Sturm R. Costs and utilization of substance abuse care in a privately insured population under managed care. Psychiatric Services . 1998;49(12): 1573-1578. 22 9. Cohen P, Cohen J. The clinician222s illusion. Archives of General Psychiatry . 1984;41:1178-1182. 10. Anthony WA. Recovery from me ntal illness: The guiding vi sion of the mental health service system in the 1990222s. Psychosocial Rehabilitation Journal . 1993;16(4):11-23. 11. Sullivan WP. A long and winding road: The process of recovery from severe mental illness. In: Spaniol L, Gagne C, and Koehler M (Eds.) Psychological and Social Aspects of Psychiatric Disability . Boston, MA: Center for Psyc hiatric Rehabilitation. 1997; 14- 24. 12. Rothbard AB, Schinnar AP, Adams K. The utilization of Medicaid mental health services. Administration and Policy in Mental Health . 1996;24(2):117-128. 13. Coffey RM, Graver L, Schroeder D, et al. Mental Health and Substance Abuse Treatment: Results from a Study Integrating Data from State MH, SA, and Medicaid Agencies. SAMHSA Publication No. SM A-01-3528. Rockville, MD: Center for Substance Abuse Treatment and Center for Me ntal Health Services, Substance Abuse and Mental Health Services Administration; 2001. 14. Whalen D, Pepitone A, Graver L, et al. Linking Client Records fr om Substance Abuse, Mental Health, and Medicaid State Agenci es. SAMHSA Publication No. SMA-01-3500. Rockville, MD: Center for Substance Abuse Treatment and Center for Mental Health Services, Substance Abuse and Mental H ealth Services Administration; 2001. 15. Hadley TR, Schinnar A, Rothbard A. Manage d mental health care in the public sector. In: Feldman S, ed. Managed Mental Health Services . Springfield, IL: Charles C. Thomas; 1992. 23 24 16. Olfson M, Marcus SC, Druss B, Elinson L, Tanielian T, Pincus HA. National trends in the outpatient treatment of depression. Journal of the American Medical Association. 2002;287(2):203-209. 17. Olfson M, Gameroff MJ, Marcus SC, Waslic k BD. Outpatient treatment of child and adolescent depression in the United States. Archives of General Psychiatry . 2003;60(12):1236-1242. 18. Stromberg C, Loeb L, Thomsen S, et al. State initiatives in health care reform. The Psychologist222s Legal Update. 1996;8:1-16. 19. O222Malley S, Jaffe A, Chang G, et al. Naltrexone and copi ng skills therapy for alcohol dependence: A controlled study. Archives of General Psychiatry . 1992;49:881-887. 20. Volpicelli J, Alterman A, Hayasguda M, et al. Naltrexone in the treatment of alcohol dependence. Archives of General Psychiatry . 1992;49(11):876-880. 21. National Health Policy Forum (NHPF). Depressi on: A Decade of Progress, More to Do. No. 786; 2002. 22. Anderson N, Estee S. Medical cost offsets a ssociated with mental health care: A brief review. Research and Data Analysis Division Report Number 3.28. Washington Department of Social and Health Services. 2002. 23. Holder HD, Lennox RD, Blose JO. The economic benefits of alcoholism treatment: A summary of twenty years of research. Journal of Employee Assistance Research . 1992;1(1):63-82. Table 1 Distribution of MH/SA service window lengtha, b (in days) by MH/SA category, service agency,c and state over the period 19962261998, ages 0 to 64 Delaware Oklahoma Washington Service Window Percentiles by MH/SA Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency A l l s e r v i c e u s e r s N 35,009 14,011 15,618 5,380 195,513 47,305 124,887 23,321 325,608 62,059 146,258 117,291 25% 8 1 23 264 2 1 1 199 5 1 3 100 50% (median) 139 45 135 629 66 82 36 537 103 1 58 344 75% 507 325 394 1,021 361 394 207 975 352 126 215 765 M H - O n l y N 19,367 11,987 4,682 2,698 130,177 43,049 71,246 15,882 227,987 52,191 95,725 80,071 25% 1 1 23 193 1 1 1 174 1 1 1 77 50% (median) 98 41 141 562 59 78 24 493 72 1 33 297 75% 498 307 687 1,019 337 378 157 954 316 122 168 728 (continued) 25 Table 1 Distribution of MH/SA service window lengtha, b (in days) by MH/SA category, service agency,c and state over the period 19962261998, ages 0 to 64 (continued) Delaware Oklahoma Washington Service Window Percentiles by MH/SA Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency S A - o n l y N 9,879 881 8,550 448 31,719 1,008 30,605 106 64,695 7,011 45,950 11,734 25% 11 1 18 248 1 1 1 39 22 1 25 101 50% (median) 113 8 117 502 14 3 15 204 107 1 102 271 75% 366 89 366 896 86 48 86 417 314 56 288 559 M H + S A N 5,763 1,143 2,386 2,234 33,617 3,248 23,036 7,333 32,926 2,857 4,583 25,486 25% 116 57 45 366 70 44 49 276 179 17 99 241 50% (median) 432 301 255 723 302 265 220 643 465 162 274 557 75% 856 623 687 1,039 743 702 615 1,009 836 444 540 903 aA service window is defined as the number of days between an individual222s first and last observed service encounter date over t he 3-year study period. bBecause the IDB spans a 3-year period (1996-1998), the maximum service window length MH/SA clients may have is 1,096 days. cService agency refers to the agency (Medicaid and/or State Agency) from which each IDB record is obtained. 26 Table 2 Distribution of MH/SA se rvice encounter datesa by MH/SA category, service agency,b and state over the period 19962261998, ages 0 to 64 Delaware Oklahoma Washington Service Date Percentiles by MH/SA Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency A l l s e r v i c e u s e r s N 35,009 14,011 15,618 5,380 195,513 47,305 124,887 23,321 325,608 62,059 146,258 117,291 25% 2 1 6 17 2 1 1 15 2 1 2 7 50% (median) 12 3 22 62 7 5 6 40 8 1 7 24 75% 56 11 70 263 29 22 22 99 36 3 29 77 MH-Only N 19,367 11,987 4,682 2,698 130,177 43,049 71,246 15,882 227,987 52,191 95,725 80,071 25% 1 1 3 11 1 1 1 12 1 1 1 5 50% (median) 5 2 12 34 5 4 4 35 5 1 4 16 75% 27 10 40 155 23 21 14 88 18 3 13 53 (continued) 27 Table 2 Distribution of MH/SA se rvice encounter datesa by MH/SA category, service agency,b and state over the period 19962261998, ages 0 to 64 (continued) Delaware Oklahoma Washington Service Date Percentiles by MH/SA Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency SA-only N 9,879 881 8,550 44 8 31,719 1,008 30,605 10 6 64,695 7,011 45,950 11,734 25% 6 1 8 31 1 1 1 7 6 1 7 17 50% (median) 26 2 29 119 5 2 5 15 26 1 28 45 75% 83 9 84 383 19 6 20 35 67 6 67 109 MH+SA N 5,763 1,143 2,386 2,234 33,617 3,248 23,036 7,333 32,926 2,857 4,583 25,486 25% 8 3 6 32 8 5 7 22 13 2 14 18 50% (median) 37 9 29 99 25 15 20 54 46 3 45 56 75% 143 28 94 344 65 52 51 124 123 8 106 140 aService encounter dates are the number of unique dates of service over the 3-year period. bService agency refers to the agency (Medicaid and/or State Agency) from which each IDB record is obtained. 28 Table 3 Distribution of service use by MH/SA diagnosis category, service agency,a and state over the period 19962261998, ages 0 to 64 Delaware Oklahoma Washington Service Use Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All service u sers N 35,009 14,011 15,618 5,380 195,513 47,305 124,887 23,321 325,608 62,059 146,258 117,291 Single encounter utilizers 17.6% 35.7% 7.2% 0.6% 24.5% 27.1% 27.9% 1.2% 21.9% 51.6% 22.8% 5.1% Clients with 3 months in the system (short-term utilizers) 26.0% 21.8% 34.9% 10.9% 29.8% 24.4% 35.1% 12.6% 26.3% 20.1% 35.0% 18.7% Clients with >3 months in the system but <10 encounter dates (occasional utilizers) 10.3% 18.7% 3.7% 7.4% 9.8% 16.3% 7.7% 7.9% 11.3% 20.2% 7.5% 11.3% Clients with > 3 months in the system and >10 encounter dates (frequent utilizers) 46.2% 23.8% 54.2% 81.2% 35.9% 32.2% 29.3% 78.3% 40.5% 8.1% 34.6% 64.9% (continued) 29 Table 3 Distribution of service use by MH/SA diagnosis category, service agency,a and state over the period 19962261998, ages 0 to 64 (continued) Delaware Oklahoma Washington Service Use Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency MH- only N 19,367 11,987 4,682 2,698 130,177 43,049 71,246 15,882 227,987 52,191 95,725 80,071 Single encounter utilizers 26.7% 37.8% 13.0% 0.9% 27.5% 28.2% 32.9% 1.6% 28.5% 53.0% 32.9% 7.3% Clients with 3 months in the system (short-term utilizers) 22.3% 20.9% 29.7% 15.9% 28.2% 24.0% 33.9% 13.9% 24.9% 19.2% 31.6% 20.5% Clients with >3 months in the system but <10 encounter dates (occasional utilizers) 14.7% 18.7% 7.2% 9.8% 10.9% 16.5% 8.1% 8.6% 13.6% 21.1% 9.7% 13.4% Clients with > 3 months in the system and >10 encounter dates (frequent utilizers) 36.3% 22.6% 50.0% 73.4% 33.3% 31.4% 25.0% 75.9% 33.0% 6.7% 25.8% 58.8% (continued) 30 Table 3 Distribution of service use by MH/SA diagnosis category, service agency,a and state over the period 19962261998, ages 0 to 64 (continued) Delaware Oklahoma Washington Service Use Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency SA- only N 9,879 881 8,550 448 31,719 1,008 30,605 106 64,695 7,011 45,950 11,734 Single encounter utilizers 6.8% 40.6% 3.7% 0.0% 34.5% 44.9% 34.2% 0.9% 9.0% 55.8% 3.9% 0.9% Clients with 3 months in the system (short-term utilizers) 38.3% 35.0% 40.4% 4.5% 41.5% 37.3% 41.6% 37.7% 37.5% 25.0% 43.5% 21.9% Clients with >3 months in the system but <10 encounter dates (occasional utilizers) 2.3% 9.9% 1.3% 7.6% 4.7% 7.1% 4.6% 12.3% 3.7% 5.9% 3.1% 4.6% Clients with > 3 months in the system and >10 encounter dates (frequent utilizers) 52.5% 14.5% 54.6% 87.9% 19.3% 10.6% 19.5% 49.1% 49.8% 13.3% 49.5% 72.7% (continued) 31 Table 3 Distribution of service use by MH/SA diagnosis category, service agency,a and state over the period 19962261998, ages 0 to 64 (continued) Delaware Oklahoma Washington Service Use Category All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency All Service Users Medicaid Only State MH/SA Agency Only Medicaid + Agency MH +SA N 5,763 1,143 2,386 2,234 33,617 3,248 23,036 7,333 32,926 2,857 4,583 25,486 Single encounter utilizers 5.4% 9.7% 8.0% 0.3% 3.3% 6.7% 3.8% 0.4% 1.8% 15.1% 1.5% 0.3% Clients with 3 months in the system (short-term utilizers) 17.0% 20.7% 25.3% 6.1% 25.2% 26.4% 30.1% 9.2% 14.2% 25.2% 21.8% 11.6% Clients with >3 months in the system but <10 encounter dates (occasional utilizers) 9.0% 25.5% 5.3% 4.4% 10.1% 16.7% 10.3% 6.4% 10.3% 40.4% 6.6% 7.6% Clients with > 3 months in the system and >10 encounter dates (frequent utilizers) 68.7% 44.0% 61.3% 89.1% 61.4% 50.1% 55.7% 84.0% 73.7% 19.3% 70.0% 80.5% aService agency refers to the agency (Medicaid and/or State Agency) from which each IDB record is obtained. 32