Schackman, Bruce R PhD*; Leff, Jared A MS*; Botsko, Michael MSW, MPhil†; Fiellin, David A MD‡; Altice, Fredrick L MD‡; Korthuis, P Todd MD, MPH§; Sohler, Nancy PhD, MPH‖; Weiss, Linda PhD¶; Egan, James E MPH¶; Netherland, Julie MSW, MPhil¶; Gass, Jonathan MPH¶; Finkelstein, Ruth ScD¶; for the BHIVES Collaborative
Many HIV-infected patients have concomitant opioid dependence and addiction. Addressing these health conditions by integrating buprenorphine/naloxone (bup/nx) treatment into HIV care requires additional provider resources at the HIV clinic and new medication and monitoring costs.1,2 Patients need to invest time and resources in addition to what they invest for their HIV care alone, although perhaps less than for HIV care and substance abuse treatment provided at separate locations. In addition, as patients successfully address their opioid dependence, they may engage in more frequent care for other medical conditions that can change health system costs.3-5 Potential economic benefits outside of the health care system of successful opioid dependence treatment include reductions in criminal justice system involvement and unemployment.6,7
Treatments for substance abuse and HIV care historically have been reimbursed through separate funding streams. Provider training and reimbursement policies are frequently quite different for substance abuse treatment and HIV care, even when reimbursement comes from the same payer. We previously provided a framework for enumerating costs associated with integrated HIV and bup/nx treatment to identify costs that may or may not be covered by different payers.8
The primary purpose of this study was to identify costs associated with integrated HIV and bup/nx care, as part of a cross-site evaluation of integrated care programs using a variety of integration approaches.9,10 We identified costs for integrated care incurred at the HIV care sites (clinic perspective), bup/nx medication costs (payer perspective), and costs to patients receiving this care at these sites (patient perspective). For comparative purposes, we also report the costs incurred at these sites for patients enrolled in the evaluation who did not initially receive integrated care, recognizing that at most sites patients were not randomly assigned to the comparison group. Finally, we explore potential cost offsets for integrated care elsewhere.
The overall evaluation of integrated HIV care and bup/nx treatment funded by the Health Resources and Services Administration HIV/AIDS Bureau's Special Projects of National Significance Program has been described elsewhere.9 Ten clinical sites were participants in the cost portion of the evaluation, including 1 evaluation site that provided care in 2 locations with different staffing models that were considered 2 separate sites for the cost evaluation. These sites represent a total of 352 patients with HIV clinical encounters recorded at the care site and 3171 patient months. Six are located on the East Coast, 1 is in the Midwest, 1 is in the Southwest, and 2 are on the West Coast. Integrated care models differed as to whether sites provided integration at the clinic level (services provided by different clinicians at a single site), at the individual level (HIV and bup/nx services provided by the same clinician), or either option depending on the clinician. All models were consistent with national guidelines for bup/nx treatment.11 Only 2 sites had extensive prior experience providing integrated care. All sites employed a designated bup/nx coordinator, who was a counselor (5 sites), registered nurse (3 sites), nurse practitioner (1 site), or pharmacist (1 site).
All but 1 site included a comparison group of opioid-dependent HIV patients not receiving integrated care but receiving substance abuse treatment at a different location than the site of HIV care; assignment to the comparison group was randomized at 4 sites. Initial substance abuse treatment for these comparison patients represented referral for standard of care at the site and included a range of modalities that varied by site: methadone (9 sites), bup/nx (7 sites), and nonpharmacological treatments (5 sites). At 7 sites, comparison patients were allowed to subsequently request and receive integrated bup/nx during the follow-up period.
Cross-site evaluation data collection was coordinated by the New York Academy of Medicine. The Institutional Review Boards at The New York Academy of Medicine and participating sites approved the relevant data collection protocols.
Site staff recruited all individuals meeting cross-site eligibility criteria described elsewhere9 and each site's program-specific criteria. Patients were recruited between July 2005 and December 2007 and followed for up to 1 year. In addition, to be included in this cost analysis, patients had to have at least 1 clinical encounter with a provider at the HIV care site recorded in the study database (details below); their monthly costs were observed until the last month in which a clinical encounter was documented. All patients who received bup/nx treatment at the HIV care site within 44 days of baseline were considered to be in integrated care. The remaining patients were considered to have been assigned to a comparison group and are labeled “all other.”
Every in-person encounter with a service provider at the HIV care site was documented on a standard form completed by the site's research staff and submitted on a quarterly basis. On-site encounters were identified on the form by (1) the type of provider who delivered care; (2) the type of service(s) delivered at the encounter including HIV care with and without bup/nx care, bup/nx induction (when patients receive initial doses of bup/nx, often under observation), other bup/nx care (stabilization/maintenance), methadone-related (methadone was dispensed off-site), other drug treatment related, mental health, and other; (3) date of the encounter; and (4) intervention delivery mode (individual or group). Because we were unable to consistently obtain data at this level of detail for substance abuse treatment provided off-site, we did not conduct a societal-level analysis that would include costs incurred at all locations. Additional cost-related data collected quarterly included receipt of bup/nx and dose, dates of urine toxicology analyses, and receipt of antiretroviral therapy.
We conducted in-person interviews and telephone surveys with investigators, providers, finance personnel, and evaluation staff, with telephone and e-mail follow-up, to collect site-level data relevant to evaluating costs including the following: (1) estimated average duration of time spent providing clinical care at each type of encounter; (2) estimated average duration of time spent completing administrative duties, such as charting, for each type of encounter; (3) administrative time spent scheduling appointments; (4) estimated travel and wait time and cost of travel for patients; (5) number of providers trained to administer bup/nx and duration of provider training; (6) local wage and fringe benefit rates for providers and staff; and (7) clinic overhead costs.
Site-Level Data Analysis
Personnel time and associated overhead was summarized for each site as total start-up training costs for the site, average cost of each bup/nx induction, and average of the remaining personnel and overhead cost per patient per month. The labor cost for each clinical encounter was determined as follows. First, the total number of encounters provided by each provider professional category was multiplied by the site's estimated duration of time for that encounter type and provider type, including administrative time by providers and scheduling time by administrative staff. Encounters that were recorded as occurring in a group setting were adjusted for the estimated size of the group.
The total duration of time for each provider professional category was summed and multiplied by the compensation rate for that category. We report results using national labor and benefit (43.4%) rates reported by the US Bureau of Labor Statistics,12 so that sites are represented on a comparable basis; readers can compare these costs with their locality by applying a ratio of their own labor costs to these national costs. We also report results applying each site's own labor and fringe benefit rate for each provider professional category. In the few cases where local labor rates were not available, we substituted regional labor rates in this analysis.12 Start-up training time costs were calculated using the same approach for the 48 physicians and 26 other providers who completed the 8-hour training required by Federal law for most prescribing physicians. We also assigned a cost of $150 per person for training materials based on the cost of on-line training;13 although some training was conducted in-person at the time of the study, most training is currently conducted on-line.
We did not use overhead rates applied to grant funding, which reflect the institutional costs related to research and evaluation. Overhead rates were calculated for each site based on interviews and reviews of site financial records; one site that was not able to separate research and clinical overhead provided us with a proxy rate from a comparable nonresearch site. For the analysis using national labor rates, we used the median overhead rate across sites (28.6%). The total cost of urine toxicology analyses was determined by multiplying the number of analyses by the cost for the drugs most frequently tested at the site. These costs varied depending on whether the site was using kits on-site, sending samples to a commercial laboratory, or sending samples to a hospital laboratory. The median cost per analysis across sites was used in the national analysis.
The cost of bup/nx was the average wholesale price less the discount provided by manufacturers for all medications purchased by state Medicaid programs, using state-specific Medicaid discounts and dispensing fees to determine site-specific bup/nx costs.14 We used the median discount (10%) and dispensing fee ($4.07) in the national analysis and for 1 site where an applicable Medicaid discount was unavailable. If the same dose was reported in the subsequent quarter, we assumed this dose was dispensed throughout the quarter; if the dose was changed or no bup/nx use was reported in the subsequent quarter, we assumed the dose was changed midway through the quarter.
Costs from the patient perspective include time and transportation costs. Patient time was estimated by summing the total time spent with providers, waiting in the clinic, traveling to-and-from the site, and time spent making appointments. In the analysis using national labor rates, we applied the US minimum wage because approximately 75% of patients were unemployed and 44% did not have a high school diploma. In the analysis using local labor rates, we applied state minimum wages. To determine the cost of patient travel, we multiplied the total number of visits by the local cost of a round trip on public transportation. The median cost ($3.00) was used in the national analysis.
We calculated average resource utilization and costs per patient month for each site. With site as the unit of analysis, findings are reported as the median and range of these individual site averages. The medians and ranges are reported separately for patients receiving integrated bup/nx care (10 sites) and all other patients (9 sites). The site-level analyses were conducted in Excel 2003 (Microsoft Corporation, Redmond, WA), and costs are reported in 2008 US dollars.
Patient-Level Data Analysis
The type and duration of encounters by each type of provider and the number of urine toxicology analyses may have been affected by case mix. We used data collected at the patient level to address this issue. Patients completed assessments at baseline and at the end of each subsequent quarter for up to 1 year. We incorporated these data in multivariate models using generalized estimating equations, with separate models for number of physician encounters, total number of provider encounters, and number of monthly toxicology analyses. In these models, we tested for the significance of site as an independent variable controlling for baseline patient demographics (age, sex, and race/ethnicity), drug use (Addiction Severity Index15), HIV symptoms (HIV Symptom Index16), and use of antiretroviral therapy.
We also explored the potential impact of initiating integrated care on other cost offsets over time in generalized estimating equations models that adjusted for clustering by site. The dependent variables were self-reported hospitalizations and emergency room visits in the last 3 months (any event and number of events); currently on parole or probation; currently awaiting trial, charges or sentencing; in jail or prison for 3 days or more in the last 3 months; currently working for pay; and received unemployment insurance, public assistance, or disability insurance in the past 3 months.
There were a total of 2034 patient months of observation among patients assigned to integrated care, and the median (range) average months of observation per patient site was 9.5 (6.1-11.1). For all other patients, there were a total of 1137 patient months of observation, and the median (range) average months of observation per patient was 9.8 (5.0-13.5). The median (range) average monthly number of patient encounters with any provider was 3.2 (1.5-13.3) for integrated care and 1.7 (1.1-4.2) for all other patients (Table 1). The median number of monthly encounters was higher in integrated care for encounters with nurses, social workers, and other providers than in nonintegrated care. The median number of monthly encounters was lower in integrated care for physicians and for nurse practitioners or physician assistants. Median duration of these encounters (including charting and follow-up time) was similar between the 2 groups for most providers except physicians, where the median duration was 45.0 minutes for integrated care patients versus 37.8 minutes for all other patients (Table 1).
There was substantial variation among sites in the number of monthly patient encounters in integrated care (Fig. 1). This did not appear to be associated with differences among sites in how long patients remained in care. In multivariate analyses, site remained a statistically significant predictor of number of physician encounters and all encounters (P < 0.001) after controlling for differences in demographic characteristics, HIV symptoms, use of antiretroviral therapy, and addiction severity.
The median (range) labor and overhead costs of providing services for integrated care was $136 ($67-$677) per patient month using national labor rates and $113 ($67-$550) using local labor rates (Table 2). Physician and registered nurse costs were the largest contributors and also varied substantially among sites, reflecting differences in level of use. The mean total labor and overhead costs were higher for integrated care than for all other patients at 7 sites (Fig. 2); at the remaining 2 sites with comparison groups, these costs were lower for integrated care due to fewer encounters with physicians. The median (range) difference was $14 (−$65-$73) per patient month.
Toxicology analysis use varied by site, ranging from 0.21 to 2.93 average urine toxicology screens per patient per month, with a median (range) monthly cost of $8 ($2-$23). In the multivariate analysis site was a significant predictor of use (P < 0.001) after controlling for patient-level factors. The median labor and overhead costs for a bup/nx induction was $160 using national labor rates and $150 using local labor rates. One site was an outlier, with an average induction cost of approximately $2200 because the induction protocol required patients to be seen daily for 4 days; at all other sites initial inductions occurred over 1 or 2 days and the range of costs for these sites was $80-$500 using national labor rates and $70-$480 using local labor rates. The median cost per site for training fees would have been approximately $1050 assuming all sites had used on-line training, and the median (range) cost of provider time for on-line training would have been $8350 ($460-$13,070) using national labor rates and $6750 ($450-$12,710) using local labor rates if training occurred during paid work hours; these costs varied depending on the number of providers trained.
The median (range) monthly cost of bup/nx from the payer perspective for patients assigned to integrated care was $209 ($165-$272) using the median Medicaid discount and $212 ($164-$266) applying state-specific discounts to each site. There was little variation among sites in bup/nx dose; 44% of all doses were 16mg, 17% were 24 mg, and 9% were 8 mg.
The median (range) total monthly cost from the patient perspective was approximately $11 ($1-$54) higher for integrated care than for all other patients, primarily reflecting 0.8 (−0.1 to 3.5) additional trips to the HIV clinic each month. This does not reflect patient time and transportation costs for substance abuse treatment at other locations, for example, methadone clinics that often require daily visits.
In exploratory analyses of potential cost offsets, initiating integrated care was not associated with differences in hospitalizations, emergency room visits, criminal justice involvement, or employment (P > 0.100).
In this cross-site evaluation of integrated HIV care and bup/nx treatment, we found that median monthly clinic costs were $136 per patient for labor and overhead and $8 per patient for toxicology analyses. This represents an incremental cost to the clinic of only about $22 per month compared with treating HIV-infected patients with an opioid dependence diagnosis who were not assigned to integrated care. In integrated care, however, there were fewer encounters with physicians and more encounters with nonphysician providers, whose services are less expensive but also less likely to be billable to third party payers. This finding reinforces the critical role played by bup/nx coordinators in integrated care models described elsewhere17 and the need to obtain funding for their services.
All patients in integrated care do not remain on bup/nx continuously. In this study, we found the monthly cost of bup/nx was about $209 per patient at current prices, somewhat higher than a recent national estimate of $166 per prescription in 2007.18 The bup/nx patent recently expired,19 so future costs may be lower if generic bup/nx becomes available. If incremental HIV clinic costs are $22 and bup/nx prices remain unchanged, the total additional cost of $229 per patient month still compares favorably with monthly per patient costs reported elsewhere for office-based ($238) and methadone clinic-based ($159-$486) methadone maintenance treatment adjusted to 2008 US dollars.20,21
Paradoxically, with fewer billable physician encounters, HIV clinics may actually receive lower reimbursement from third parties using these models of integrated care. However, there were wide variations among sites in the number of encounters, reflecting differences in program design and implementation. The funding for this project allowed sites to use integration approaches without regard to third-party reimbursement and different approaches might be designed to maximize reimbursement. Up-front expenditures for clinics were relatively small, consisting primarily of provider time spent on training. Ongoing training support may also be required and is available without charge through a national mentoring program.22
Integrated care also requires additional patient visits to the clinic, especially during the early phases of bup/nx treatment. For some patients, these requirements may be a barrier to participating in integrated care, even though they can incur similar or higher costs for off-site outpatient substance abuse treatment.23 Programs should consider providing financial and nonfinancial incentives to address these barriers, such as transportation reimbursement, meals, or other incentive programs.24,25
A limitation of this study is that we only calculated costs incurred at HIV care sites because we were unable to collect utilization data for substance abuse care provided elsewhere for most sites. This means that we are unable to determine potential savings from integrated care to third-party payers and state agencies currently paying for substance abuse care in other settings. Using patient self-reports, we were also unable to identify offsetting cost savings from integrated care elsewhere in the health care system. We might have been able to identify more cost savings if we had access to clinical data systems. For example, Barnett26 found total costs of care were lower for bup/nx versus methadone in the first 6 months of care in the Veterans Health Administration, with no significant differences in costs in subsequent months. This result was attributable to significantly lower use of ambulatory care services (defined as substance abuse/mental health and other ambulatory care) and fewer inpatient days. In addition, potential long-term health system savings from better HIV care outcomes 27,28 and better screening and treatment of comorbid medical conditions3-5 could not be measured in this study. Although we were also unable to identify cost offsets in the criminal justice system or due to increased employment, these savings have been found in other studies of opioid dependence treatment.29
There are several other limitations to this study. The increase in labor and overhead costs associated with integrated care may have been understated because our analysis was based on care initially received. Some patients not initially receiving integrated care subsequently received this care during the follow-up period. Moreover, initial group assignment was not randomized at the majority of sites, so patients with lower costs may have been assigned to integrated care. These factors may also have limited our ability to detect cost offsets from patient-reported data.
On the other hand, bup/nx integrated care was supported with dedicated funding and may have been more intensive and expensive than would occur elsewhere, especially because site funding levels were fixed and enrollment at most sites was lower than anticipated. Costs may be lower in subsequent years if providers require less time with more patient experience; in addition, patients may require less additional time as they stabilize on bup/nx. Costs of inductions may also be overstated because at some sites re-inductions that required fewer resources were not reported separately from regular visits; had they been included the average induction costs would have been lower.
The wide variation among sites in provider utilization and costs, confirmed in multivariate analyses, indicates that the costs we report may not be broadly generalizable. HIV clinics will need to forecast their own costs based on the approach described here and then determine which of these costs would be reimbursed to project additional funding requirements. Our analysis does suggest that clinics may have the opportunity to approach third-party payers for assistance, in addition to traditional funders such as Ryan White CARE Act programs.
In summary, integrated HIV care and bup/nx treatment requires additional resources at the clinic level, including for nonphysicians and other providers acting as bup/nx coordinators whose costs are usually not third-party reimbursed. Even if overall costs do not increase substantially, implementing integrated care will require additional funding at the clinic level in addition to appropriate reimbursement for bup/nx drug costs. By forecasting service utilization and costs of integrated care, HIV clinics can determine the budget impact of implementing different treatment models and identify potential sources of funding.
We gratefully acknowledge the assistance of the staff at each site that participated in this cost study and the patient participants.
1. Altice FL, Sullivan LE, Smith-Rohrberg D, et al. The potential role of buprenorphine in the treatment of opioid dependence in HIV-infected individuals and in HIV infection prevention. Clin Infect Dis. 2006;43(Suppl 4):S178-S183.
2. Sullivan LE, Barry D, Moore BA, et al. A trial of integrated buprenorphine/naloxone and HIV clinical care. Clin Infect Dis. 2006;43(Suppl 4):S184-S190.
3. Parthasarathy S, Weisner CM. Five-year trajectories of health care utilization and cost in a drug and alcohol treatment sample. Drug Alcohol Depend. 2005;80:231-240.
4. Weisner C, Mertens J, Parthasarathy S, et al. Integrating primary medical care with addiction treatment: a randomized controlled trial. JAMA. 2001;286:1715-1723.
5. Basu S, Smith-Rohrberg D, Bruce RD, et al. Models for integrating buprenorphine therapy into the primary HIV care setting. Clin Infect Dis. 2006;42:716-721.
6. Ettner SL, Huang D, Evans E, et al. Benefit-cost in the California treatment outcome project: does substance abuse treatment “pay for itself”? Health Serv Res. 2006;41:192-213.
7. Sindelar JL, Fiellin DA. Innovations in treatment for drug abuse: solutions to a public health problem. Annu Rev Public Health. 2001;22:249-272.
8. Schackman BR, Merrill JO, McCarty D, et al. Overcoming policy and financing barriers to integrated buprenorphine and HIV primary care. Clin Infect Dis. 2006;43(Suppl 4):S247-S253.
9. Weiss L, Egan JE, Botsko M, et al. A multi-site evaluation of integrated buprenorphine/naloxone and HIV treatment: overview of the BHIVES collaborative. J Acquir Immune Defic Syndr. 2011;56(Suppl 1):S1-S2.
10. Cheever LW, Kresina TF, Cajina A, et al. A model federal collaborative to increase patient access to buprenorphine treatment in HIV primary care. J Acquir Immune Defic Syndr. 2011;56(Suppl 1):S3-S6.
11. Center for Substance Abuse Treatment. Clinical Guidelines for the Use of Buprenorphine in the Treatment of Opioid Addiction. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2004.
12. US Department of Labor Bureau of Labor Statistics. National compensation survey. Available at: http://data.bls.gov
. Accessed August 14, 2009.
14. Murray L, ed. Red Book. Montvale, NJ: Thomson PDR; 2008.
15. Cacciola JS, Alterman AI, McLellan AT, et al. Initial evidence for the reliability and validity of a “Lite” version of the Addiction Severity Index. Drug Alcohol Depend. 2007;87:297-302.
16. Justice AC, Holmes W, Gifford AL, et al. Development and validation of a self-completed HIV symptom index. J Clin Epidemiol. 2001;54(Suppl 1):S77-S90.
17. Weiss L, Netherland J, Egan JE, et al. Integration of buprenorphine/naloxone treatment into HIV clinical care: lessons from the BHIVES collaborative. J Acquir Immune Defic Syndr. 2011;56(Suppl 1):S68-S75.
18. Mark TL, Kassed CA, Vandivort-Warren R, et al. Alcohol and opioid dependence medications: prescription trends, overall and by physician specialty. Drug Alcohol Depend. 2009;99:345-349.
20. Jones ES, Moore BA, Sindelar JL, et al. Cost analysis of clinic and office-based treatment of opioid dependence: results with methadone and buprenorphine in clinically stable patients. Drug Alcohol Depend. 2009;99:132-140.
21. Zarkin GA, Dunlap LJ, Homsi G. The substance abuse services cost analysis program (SASCAP): a new method for estimating drug treatment services costs. Evaluation and Program Planning. 2004;27:35-43.
23. McCollister KE, French MT, Pyne JM, et al. The cost of treating addiction from the client's perspective: results from a multi-modality application of the Client DATCAP. Drug Alcohol Depend. 2009;104:241-248.
24. Olmstead TA, Sindelar JL, Petry NM. Cost-effectiveness of prize-based incentives for stimulant abusers in outpatient psychosocial treatment programs. Drug Alcohol Depend. 2007;87:175-182.
25. Sindelar JL, Olmstead TA, Peirce JM. Cost-effectiveness of prize-based contingency management in methadone maintenance treatment programs. Addiction. 2007;102:1463-1471.
26. Barnett PG. Comparison of costs and utilization among buprenorphine and methadone patients. Addiction. 2009;104:982-992.
27. Altice FL, Bruce RD, Lucas GM, et al. HIV treatment outcomes among HIV-infected, opioid-dependent patients receiving buprenorphine/naloxone treatment within HIV clinical care settings: results from a multisite study. J Acquir Immune Defic Syndr. 2011;56(Suppl 1):S22-S32.
28. Buchacz K, Baker RK, Moorman AC, et al. Rates of hospitalizations and associated diagnoses in a large multisite cohort of HIV patients in the United States, 1994-2005. AIDS. 2008;22:1345-1354.
29. Zarkin GA, Dunlap LJ, Hicks KA, et al. Benefits and costs of methadone treatment: results from a lifetime simulation model. Health Econ. 2005;14:1133-1150.
APPENDIX I: BHIVES COLLABORATIVE
The CORE Center (Chicago, IL), El Rio Santa Cruz Neighborhood Health Center (Tucson, AZ), Johns Hopkins University (Baltimore, MD), Miriam Hospital (Providence, RI), Montefiore Medical Center (Bronx, NY), OASIS (Oakland, CA), Oregon Health Sciences University (Portland, OR), University of California San Francisco Positive Health Program at San Francisco General Hospital (San Francisco, CA), University of Miami Medical School (Miami, FL), Yale University School of Medicine (New Haven, CT) and The New York Academy of Medicine (New York, NY). Cited Here...
© 2011 Lippincott Williams & Wilkins, Inc.