 Evaluate the effects of the patient-centered medical home (PCMH) model on medical costs and utilization in the nonpediatric population, particularly among high-risk patients.

 Longitudinal case-control design, comparing per member per month (PMPM) cost and utilization per 1000 patients for members enrolled in PCMH and non-PCMH practices from 2009 to 2011.

 Commercial health maintenance organization members in nonpediatric practices that adopted the PCMH model in 2009 were matched to patients in nonpediatric practices that did not adopt the model until 2011 or later. Propensity score matching was used to identify a pool of similar controls, and difference-in-differences regression analysis was used to compare PCMH and non-PCMH patients relative to baseline. Analysis was conducted using the complete pool of matched patients (N = 6940 cases and 6940 controls), then using the 10% of patients with the highest DxCG risk scores (N = 654 cases and 734 controls).

 There were no significant cost or utilization differences for the overall population. Total cost decreased significantly more for the PCMH group than for controls in the high-risk group in years 1 and 2 (reductions of $107 and $75 PMPM), driven by lower inpatient costs. The PCMH group experienced a significantly greater reduction in inpatient admissions in all 3 years (61, 48, and 94 hospitalizations per 1000).

 PCMH practices had significantly reduced costs and utilization for the highest risk patients, particularly with respect to inpatient care. As high-risk members represent a high-cost group, the most benefit can be gained by targeting these members.

High-risk patients enrolled in nonpediatric primary care practices which adopted the patient-centered medical home (PCMH) model had significantly lower per member per month medical costs and utilization per 1000 members compared with non- PCMH practices after adjusting for baseline characteristics.

		Costs and utilization did not significantly differ between PCMH and non-PCMH groups among all patients, suggesting that the benefits of the PCMH model are concentrated among high-risk, high-cost patients.

		PMCH cost reductions appear to have been driven by lower rates of hospitalization, and total costs fell even though utilization of specialist care saw significant increases in the first 2 years.

The patient-centered medical home (PCMH) has been advanced as a promising framework for transforming primary care. In 2007, the American Academy of Family Physicians, American Academy of Pediatrics, American College of Physicians, and American Osteopathic Association issued the “Joint Principles of the Patient-Centered Medical Home,” which outlined the PCMH model. The medical home model emphasizes a team-based approach to primary care, in which a physician-leader coordinates care by other providers across multiple sites and specialties. It encourages increased access, both in terms of expanding practice hours and opening new channels of communication with patients. Organizations such as the Patient-Centered Primary Care Collaborative have initiated numerous pilot programs aimed at studying the impact of PCMH adoption,

 and the PCMH model was written into the Patient Protection and Affordable Care Act of 2010 as an area for study.

A number of previous studies have shown early promise for the PCMH model as a vehicle for controlling costs and improving the quality of healthcare delivered by primary care practices,

 including for targeting subpopulations such as children with special health needs.

 However, reviews often point to the incomplete nature of this work, citing methodological concerns,

 insufficient time for practices to implement reforms, and inadequate policy support beyond the level of individual practices.

 This study aims to contribute to this literature by comparing the effects of adopting the PCMH model on the healthcare cost and utilization in the nonpediatric population, using propensity score matching in order to reduce variability in the PCMH and non- PCMH groups studied. Additionally, the analysis employs difference-indifferences regression analysis in order to further control for remaining differences in patients’ characteristics as well as cost and utilization at baseline.

This study aims to assess the impact of PCMH adoption on the patients identified as having the greatest health risks. While the Joint Principles envision the PCMH model as being applicable to all patients, other pilots have targeted only high-risk patients with complex needs.

 The high cost of care associated with relatively few individuals makes such targeting a potentially powerful mechanism: one study noted that virtually all of the growth in expenditures for Medicare over the period from 1987 to 2002 occurred among beneficiaries with 5 or more chronic illnesses

; another paper noted that only 10% of Medicare beneficiaries accounted for 70% of healthcare costs.

 In order to evaluate whether the benefits of medical homes are limited to such high-risk, high-cost subpopulations, this study reports cost and utilization comparisons 2 ways: first, for all matched patients; and subsequently, limited to the patients with risk scores in the 90th percentile or above for the study population.

The majority of the PCMH practices (15/17) included in this study were part of Pennsylvania’s Chronic Care Initiative, a multi-stakeholder effort to improve primary care, which launched a 3-year pilot project that started in May 2008 in the Philadelphia area. The largest commercial health plans— led by Independence Blue Cross, all 3 Medicaid managed care plans, and 32 practices caring for 250,000 members&mdash;were convened by the governor to establish the first of several regional PCMH programs. Practices received significant additional payments to participate in a learning collaborative which supported practice transformation and provided the care management and coordination services required by patients with chronic conditions. The learning collaborative was based on Wagner’s Chronic Care Model

 and led by the MacColl Institute, with Edward H. Wagner, MD, MPH, facilitating the first session.

Practices were required to send a physician and a practice support leader to 7 days of sessions. Practice coaches provided ongoing support and education on installing and using a registry to track and monitor patients and implementing a team-based approach involving care managers, health educators, and other nonphysician healthcare personnel to promote self-management skills and patient engagement.

 Additionally, clinical guidelines were used to create evidence-based standing orders to optimize patient care. Practices used their electronic systems to identify and stratify patients in need of care management, selfmanagement support, and goal monitoring. Patient engagement was promoted through the use of report cards, on which concrete data from each visit are logged for patients so they can monitor their progress.

To provide a definition of the PCMH model, the authors employed the recognition standards published by the National Committee for Quality Assurance (NCQA). These standards list 9 categories encompassing 30 possible practice improvements (including 10 “must pass” items), each of which has a point value. Three levels of certification are possible, with each defined by a minimum number of “must pass” items achieved and specified point totals. The full list of recognition criteria appears in 

 Practices qualified for PCMH status by achieving NCQA Level 1 recognition or higher during 2009, and the control group was limited to practices which did not receive recognition until 2011 or later.

This study employs a longitudinal, case-control design to compare members of commercial health maintenance organizations (HMOs) enrolled in nonpediatric PCMH and non-PCMH practices with respect to healthcare costs and utilization during the period from 2009 to 2011. All figures are reported first for the full matched cohort of cases and controls (N = 6940 cases and 6940 controls), and then using the 10% of patients with highest risk scores, drawn from the pooled population of cases and controls (N = 654 cases and 734 controls).

DxCG risk score refers to the concurrent medical risk score calculated using the commercial risk adjustment model developed by Verisk Health DxCG Risk Solutions, version 3.1 (Verisk Health, Cary, North Carolina).

 The DxCG models use linear, additive formulas obtained from ordinary least squares (OLS) regressions to combine expenses associated with clinical groupings (Condition Categories and Rx- Groups) and demographic factors (age, gender) to generate predictions. Hierarchies are imposed on Condition Categories before regression models are run to decrease their sensitivity to variations in coding and to strengthen the predictive power of the models across a wide range of benefit designs and pricing arrangements.

 The concurrent risk score in this analysis used each individual’s age, gender, and claims information from all medical encounters and enrollment data as inputs to predict annual total medical resource use for the patient.

Patients evaluated were enrolled in 17 PCMH practices, 15 of which were participating in the Pennsylvania Chronic Care Initiative pilot (cases), or 1 of 103 non-PCMH comparison practices in PA (controls). The high-risk patients were slightly more concentrated, appearing in only 14 PCMH practices (12 of which were part of the pilot) and 69 non-PCMH practices. Costs were assessed as per member per month (PMPM) costs and utilization was measured as encounters per 1000 patients per year.

Criteria In order to qualify for inclusion, each member must have been enrolled with the same primary care physician for at least 3 months and have been associated consistently with a PCMH or non-PCMH practice for 6 months or more during each year of the study period, including baseline (2008). Only nonpediatric practices located in Pennsylvania were included in the analysis. Pediatric practices were excluded to reduce the variability in the age of members in the study and minimize the impact of any potential unobserved differences between pediatric practices and nonpediatric practices. Patients with end-stage renal disease or extremely high medical costs (

$100,000 per year) were excluded to limit the impact of extreme outliers on the cost analysis.

The pool of eligible controls was limited to later adopters of the PCMH model in order to reduce potential practice self-selection issues. Patients enrolled in control practices that received NCQA recognition in 2011 were included in analyses for the first 2 program years, but were removed from the final year’s analysis to avoid contaminating the non-PCMH group with late adopters, along with their matched cases. This reduced the numbers included in program year 3 to 1141 cases and 1141 controls in the overall analysis, and 109 cases and 131 controls in the highrisk subgroup. While the guidelines remained largely similar from 2008 to 2011, some changes were made to the NCQA requirements, and a crosswalk table detailing the differences is available from the NCQA.

 An attrition diagram detailing all exclusion steps appears in 

Propensity score matching was used to select a sample of controls which were similar to the PCMH cases with respect to both practice- and patient-level demographics and characteristics. The propensity score included practice size, practice location, age, gender, DxCG risk score, chronic conditions (including asthma, diabetes, congestive heart failure, chronic obstructive pulmonary disease, and coronary artery disease), and median income for member’s zip code of residence. Matching successfully controlled for the significant differences at baseline between the group of cases and pool of potential controls, both for the overall comparison and among the cohort of highest risk patients (

Differences in PMPM cost and utilization per 1000 patients between PCMH and non-PCMH practices for the 3 follow-up years were compared using regression analysis. This method provides an estimate of the differences between treatment and control groups before and after treatment. Since the members included in this study are the same in each time period (ie, panel data), the difference-in-differences model can be simplified and is more statistically powerful. The difference between baseline and each time period can be modeled as follows (separate model for each time period difference):

is the difference between the repeated outcome measure for each observation, δ is the effect of time on all units, X 

 is the treatment indicator, and β is the treatment effect.

 Cost and utilization regressions controlled for concurrent risk in the comparison year and having a chronic condition in the baseline year, factors that could influence both cost and utilization in the comparison year. For example, when comparing 2010 with baseline, the 2010 risk score was added to the model to adjust for any new diagnoses a member might have acquired in 2010. Presence of a chronic condition at baseline was used as a covariate because chronic members often continue to incur costs, or incur even higher costs, over time. In order to test for the treatment effect for just the high-risk members, a high-risk indicator was introduced as covariate (ie, as a main effect and interaction term with the treatment indicator) into the model using the whole population. Additionally, a random intercept term was used in the model to account for any practice-level effects.

When comparing utilization and PMPM cost for all matched cases and controls during the 3 follow-up years with respect to baseline characteristics, no difference-in-differences comparison reached significance at α = .1 (

). The regression results limited to the 10% of pooled cases and controls with the highest DxCG risk scores resulted in clear differences between PCMH and non-PCMH. PCMH practices achieved statistically significant decreases in utilization of inpatient medical services in all 3 program years, amounting to reductions of 61, 48, and 94 hospitalizations per 1000 patients. In the first 2 program years, these reductions were accompanied by statistically significant adjusted savings in inpatient costs ($115 PMPM in 2009 and $62 PMPM in 2010). The decline in utilization of inpatient hospital services contributed to lower total medical costs for patients enrolled in PCMH practices during the first 2 study years as well. Cases had an adjusted total savings of $107 PMPM in 2009, representing an 11.2% reduction from baseline. For 2010, the difference was $75 PMPM, corresponding to a 7.9% reduction.

While total medical costs decreased, there were areas in which high-risk patients in PCMH practices experienced increases in costs and utilization. This cohort saw a statistically significant increase in specialist visits in 2009, and corresponding increases in PMPM spending for specialist care in 2009 and 2010 (

This study offers particular methodological contributions to the PCMH literature. The approach of combining propensity score matching to create comparable case and control pools with difference-in-differences regression analysis to adjust for practice- and patient-level variation and baseline cost and utilization characteristics allows improved control for factors which might otherwise confound efforts to study the effects of PCMH adoption. Inclusion of a non-PCMH control group and assessing differences over time allows the analyses to control for trends which may have affected all practices, regardless of PCMH status. Along with controlling for practice size, excluding practices which transitioned to the PCMH model in 2010 helped to limit the impact of self-selection on results. Because adoption of the PCMH reforms does not occur instantaneously, some non-PCMH practices may have incorporated aspects of the medical home without having yet risen to the level of eligibility for NCQA recognition. Confining the control group to later adopters diminishes this issue.

The findings of this study add new texture to the existing PCMH literature. After controlling for baseline differences, no statistically significant differences between patients enrolled in PCMH and non-PCMH practices were observed. However, when looking at the patients with highest risk scores in the pool of matched patients and practices, PCMH model adoption was shown to lead to a significant relative reduction in total costs in years 1 and 2, and significantly lower numbers of inpatient admissions in all 3 years. This suggests that the average patient may not be the relevant unit of observation for evaluating the impact of PCMH adoption. Rather, high risk patients with multiple comorbidities are the most logical targets for interventions aimed at supporting self-management, conveying test results in a timely and clear fashion, and coordinating follow-up and specialist care. Researchers may miss cost and utilization improvements if they confine their analyses to the typical patient, since healthcare costs are primarily driven by relatively rare events concentrated in few individuals. For example, during the baseline year, all cases and controls had 73 and 78 admissions per 1000 patients, respectively; but among the high-risk pool, these numbers increased to 566 and 540.

The observation that relative cost reductions are driven by decrease in inpatient admissions among high-risk patients suggests that the PCMH model is having its intended effect— improvements to information technology permit physicians to better assess patient needs, and better coordination of care means that ongoing issues can be managed at lower levels of care intensity (self-management, primary care) rather than after manifesting as crises requiring hospitalization. Even the observation of increased spending for specialist care can be interpreted as being consistent with this conceptualization: as providers gain access to better quality information about patient needs through improved medical record-keeping and care is coordinated across multiple sites, the patients with the highest medical risk may be appropriately directed to more frequent contact with specialists. This could drive up the cost of 1 component of care, while the use of appropriate early interventions helps control costs overall.

Despite extensive controls through matching, regression analysis, and study design, some self-selection issues may persist. A number of the initial pool of control practices received NCQA recognition in 2011, meaning that there may have been some “wind up” time in the prior year for practices identified as controls. However, this should have had the effect of reducing the apparent difference between PCMH and non- PCMH practices for the 2010 analyses.

Additionally, other studies of the medical home model have focused on patient satisfaction and quality improvement measures, which are essential aspects of the PCMH model and central to providing exceptional primary care, but these attributes are not the focus of this study. Over a sufficient time frame, improved integration of care may lead to better outcomes for patients with lower levels of health risk as well, so ongoing assessments of PCMH adoption should be conducted with these considerations in mind.

While no differences were observed when assessing patients from the full distribution of risk scores, this study demonstrated that adoption of the PCMH model was associated with significantly reduced costs and utilization for those members at highest risk, particularly with respect to inpatient care. High-risk members are the most costly to health plans, so it logically follows that the most benefit can be gained by targeting these members with programs such as the medical home. When evaluating PCMH programs, it is important to focus both interventions and evaluations on relevant populations

 Independence Blue Cross, Philadelphia, PA (SH, RC, CC, RS, SN).

This study was funded by Independence Blue Cross, which is an independent licensee of the Blue Cross and Blue Shield Association. All of the authors were employed by the Independence Blue Cross during the course of the study.

Articles

WEB EXCLUSIVE

A longitudinal case-control design was used to evaluate the effects of the patient-centered medical home model on medical costs and utilization among high-risk patients.

Medical Homes and Cost and Utilization Among High-Risk Patients

 We examined the cost-effectiveness of an intervention to reduce coronary heart disease (CHD) risk and blood pressure in African Americans.

 Stochastic cost-effectiveness analysis alongside a clinical trial, augmented by a Markov model of lifetime cost-effectiveness.

 In 2 urban academic primary care practices, we randomized African American patients with uncontrolled hypertension to a 6-month intervention of office practice and peer coach behavioral support (N = 136) or informational brochures about CHD risk factors (N = 144). Costs were estimated from the perspective of the provider. Outcomes included estimated CHD events avoided over 6 months and reduction in systolic blood pressure (SBP) (mm Hg). Subgroup analysis was performed for compliers who received an “effective” dose of the peer coach and office staff visits. Long-term cost-effectiveness was estimated by applying the clinical trial cost and effectiveness into a Markov model of CHD risk.

The average cost for the behavioral support intervention group was $435.36 compared with $74.39 for the brochure control group. The incremental cost-effectiveness ratio (ICER) was $47 per mm Hg reduction in SBP and $453,419 per CHD event avoided in 6 months. Modeled over a 10-year horizon, the intervention had an ICER only as high as $3998 per incremental quality-adjusted life-year.

 A community-primary care practice behavioral intervention to reduce hypertension in African Americans with sustained uncontrolled hypertension does not appear to be cost-effective in the first 6 months. If intervention results are sustained over the long term, the program may be cost-effective over the patient’s lifetime.

A community and primary care practice behavioral intervention may be cost-effective to reduce hypertension in African Americans with sustained uncontrolled hypertension. From the payer perspective, the intervention appears to be:

		Not likely to be cost-effective in reducing coronary heart disease in the first 6 months.

		Likely to be cost-effective in lowering blood pressure in the first 6 months.

		Cost-effective in terms of incremental cost per quality-adjusted life-year over the lifetime of the patient if the intervention results are sustained.

Innovative models of primary care offer the potential to reduce disparities in health outcomes for vulnerable populations with chronic diseases.

 Cost-effectiveness studies have reported that future healthcare utilization and costs can be reduced by implementing a diabetes patient registry along with clinical meetings.

 Supplementing a registry with education for patients with diabetes about self-management and improving professional quality of care also meets standard guidelines of increasing quality-adjusted life-years (QALYs).

However, few cost analyses have been conducted of interventions based on Wagner’s chronic care model that focus on reducing coronary heart disease (CHD) risk and improving blood pressure (BP) in those from racial/ethnic minority backgrounds.

We report a cost-effectiveness analysis of a randomized, controlled trial of a 6-month community- and staff-based intervention of behavioral support and education for African Americans with sustained, uncontrolled hypertension based on a practice-based registry. All subjects received educational brochures and usual physician care while intervention subjects received 3 community support phone calls from trained peers from the same practices alternating with personal counseling by trained mid-level staff at 2 practice visits on alternate months. The study outcomes were 6-month changes in systolic blood pressure (SBP) and CHD risk. These results add to evidence of the potential cost-effectiveness from a provider standpoint of adopting features of the chronic care model to empower patients to reduce CHD risk due to poorly controlled risk factors.

Study subjects were recruited from July 2007 through November 2009 in 2 urban academic general internal medicine practices largely serving low-income patients. Subjects were identified from a registry of all 9135 African American patients aged 40 to 75 years receiving longitudinal care (2+ visits in 2 years). We identified those patients with treated but uncontrolled hypertension per the 7th Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7) targets (ie, SBP 130 mm Hg or higher, or diastolic blood pressure [DBP] 80 mm Hg or higher, if chronic kidney disease or diabetes; otherwise, SBP 140 mm Hg or higher, or DBP 90 mm Hg or higher) and at least 1 value 10 mm Hg above goal.

 Of 1057 subjects with uncontrolled hypertension, recent lipid levels, and at least moderate adherence to keeping primary care visits, 810 were randomly selected to ask approval from the provider for the trial; 574 were approved and sent a recruitment letter. Of 440 patients who were contacted, 280 subjects were randomized.

The Healthy Heart trial randomly assigned subjects to 1 telephone-based peer counseling session and office-based visits with trained mid-level providers over 6 time frames, or to a control condition of usual physician care. Practice providers nominated 20 patients to serve as peer coaches from lists of African Americans with well-controlled hypertension aged 50 to 75 years, because they were perceived to be “good communicators.” Of these, 11 completed training and 5 continued to the end of the study while 3 replacements were recruited and trained to assist with peer support. Training involved viewing and discussing illustrated slide shows created by the study team about CHD in the community and risks and barriers to control. Peer coaches were taught elements of motivational interviewing and practiced phone calls before being assigned patients. Peer coaches contacted intervention patients every other month for 6 months (minimum of 3 calls).

For practice-based counseling visits, we trained 3 African American staff members (eg, medical assistants) with the same slide shows used to train peers, and also trained them to use a personalized, computer-based, 4-year CHD risk calculator as a teaching tool. On alternate months from peer calls, patients made two 15- to 30-minute visits with trained practice staff to review personal CHD risk factors. All study subjects received culturally appropriate educational brochures and healthy recipes from the American Heart Association. Participants received $50 in gift certificates for participation ($20 at enrollment and $30 at end point visit). Peer coaches received $20 per completed phone call and followed a mean of 8 patients at once (other costs in 

). The protocol and procedures were reviewed and approved by the University of Pennsylvania Institutional Review Board.

Direct intervention costs were estimated from the perspective of the provider/ health system and measured in 2010 US dollars. Because the trial was based on outcomes monitored over 6 months, we did not apply any discounting. Specific resources used for the intervention included cost of training peer coaches, labor cost of peer coach telephone calls, cost of training office-based health educators, cost of clinic visits and laboratory tests, incentives for patients, cost of brochures, and cost of materials such as transportation, postage, and office supplies (

). We also included costs for the overall administration of the trial, but not solely research-related costs. For the brochure control group, resources included the cost of clinic and laboratory visits and the cost of the brochures. Patients in the intervention group incurred all of these costs. Costs that were not relevant to the provider perspective, such as indirect costs for patients, were not considered.

We studied 2 measures of effectiveness for the withintrial stochastic cost-effectiveness analysis: predicted 6-month CHD risk avoided and 6-month improvements in SBP (per mm Hg). The 6-month CHD risk measure was derived from D’Agostino’s risk equations for primary events and secondary events in Framingham data (eg, myocardial infarction, angina), using original (rather than the calibrated) versions of these equations.

 Separate risk equations were used for men and women. Our primary CHD event end point combined predicted 6-month risk of primary and secondary CHD events for patients. All outcomes were based on an intention-to-treat approach. In the long-term costeffectiveness analysis, we studied 2 measures of effectiveness: years of life saved (YLS) and QALYs, both over a 10-year lifetime horizon.

Two cost-effectiveness analyses were performed: a withintrial stochastic cost-effectiveness analysis that focused on cost-effectiveness during the 6-month trial period, and longterm cost-effectiveness based on a Markov model that modeled the longer-term benefits of BP reduction.

 For the within-trial stochastic cost-effectiveness analysis, we estimated 2 incremental cost-effectiveness ratios (ICERs): incremental cost per predicted CHD event avoided within 6 months, and incremental cost per mm Hg in SBP reduced. To characterize the uncertainty of the within-trial cost-effectiveness results, we used bootstrapping to estimate a 95% confidence ellipse around the ICER.

 The bootstrap method resampled the data 10,000 times with replacement, and computed the ICER for each replicate. From the bootstrap samples, we estimated the probability that one treatment was cost-effective compared with the other for a given willingness to pay (WTP). In addition, we computed the cost-effectiveness acceptability curve (CEAC) and plotted the probability that the behavioral health intervention was cost-effective over a reasonable range of levels of WTP.

We performed a subgroup analysis for patients who were compliant with the intervention. We defined an effective “dose” as at least 2 peer coach calls and 1 practice visit. We then estimated the ICER for the intervention in this intervention subgroup relative to the control group. All stochastic cost-effectiveness analyses were performed using R statistical software (version 2.10.1, http://www.r-project.org).

To estimate the long-term costs and benefits of BP reduction observed in the trial, data on costs and effectiveness from the clinical trial were entered into a Markov model of CHD risk in order to extrapolate trial results to a 10-year lifetime horizon. The Markov model was designed to study the impact of antihypertensive medications and was adapted to this setting. The model has a 10-year time horizon with yearly cycles. All costs begin in year 2010 US dollars and were discounted at a rate of 3%. Utility values for health states were drawn from Sullivan et al and Currie et al.

 As effectiveness measures, the model estimates YLS and QALYs. Additional details of the model are provided in Baker et al.

 Because the trial only lasted 6 months, it was necessary to make assumptions about how the intervention would be provided over the 10-year time horizon of the Markov model. We assumed that yearly reinforcements of the intervention would be required in order to sustain improvements.

The 280 intervention (N = 136) and control (N = 144) subjects were well balanced on demographics and clinical conditions (

). Complete data to estimate CHD risk were available for 212 (94 intervention and 118 control)subjects and complete end point data were available for 247 (116 intervention and 131 control) subjects. Sixty-eight percent of intervention subjects (N = 79) were compliant with an effective “dose” intervention (ie, at least 2 peer coach calls and 1 practice visit) and included in the compliance subgroup analysis.

) show that the average cost over 6 months of the intervention was $435 for intervention subjects and $74 for control subjects. The intervention was successful in reducing both SBP and CHD risk. SBP fell by 7.2 mm Hg among intervention subjects, compared with only 0.77 mm Hg for control subjects (

 = .0011). The average difference in CHD risk among intervention subjects fell by 0.046% but rose by 0.034% among control subjects (

 = .07). The ICERs were $453,419 per predicted CHD event avoided over 6 months and $55 per mm Hg reduced in 6 months.

The uncertainty analysis for these ICERs (

) shows a high probability that the intervention would rensult in a reduction in systolic blood pressure. The probability that the intervention is cost-effective is 25%, 50%, and 75% if the decision maker is willing to pay $45.20, $55.40, and $70.80, respectively, to reduce SBP by 1 mm Hg for at least 6 months (Figure 1B). For CHD risk, the CEAC suggests that the probability that the intervention is cost-effective for reducing CHD risk is 25%, 50%, and 75% if the decision maker is willing to pay $324,000, $449,000, and $674,000, respectively, to avoid 1 CHD event over 6 months (

Similar results were observed for the subgroup of patients who were compliant with the intervention (Table 3). The average cost of the intervention over 6 months was $442 for compliant intervention subjects and $74 for control subjects. SBP fell 8.5 mm Hg among intervention subjects, compared with only 0.77 mm Hg for control subjects. The average difference in CHD risk among compliant subjects decreased by 0.07% and increased by 0.03% among control subjects. The ICERs were $48 per mm Hg reduced in 6 months and $350,134 per predicted CHD event avoided over 6 months. The CEACs for the compliant subgroup, presented in Figure 1 (

), suggest that the probability that the intervention is cost-effective in reducing SBP is 25%, 50%, and 75% if the decision maker is willing to pay $34, $40, and $48, respectively, to reduce SBP by 1 mm Hg for at least 6 months. The probability that the intervention is cost-effective in reducing CHD risk is 25%, 50%, and 75% if the decision maker is willing to pay $274,000, $351,000, and $476,000, respectively, to avoid 1 additional CHD event in 6 months (

Long-term cost-effectiveness results are presented in 

. Assuming that the intervention was given every year over the patient’s lifetime (10-year horizon), the intervention was more costly ($7324 vs $5584), but also more effective, both in terms of YLS (8 vs 7) and QALYs (6.3 vs 6.2). This yields ICERs for the intervention of $12,373 per incremental YLS and $10,866 per incremental QALY saved.

This study finds that a combined community and office staff behavioral health intervention to reduce hypertension and risk of CHD among African American primary care patients with uncontrolled hypertension and other CHD risks is potentially cost-effective at reducing SBP in the short term, and in terms of cost per YLS and QALY in the long term, but not cost-effective for reducing CHD risk in the short term (6 months). There are 2 explanations for the relatively high within-trial ICER to prevent CHD events. First, substantial costs are expended for the initial training of peer coaches and practice team members. With a more mature program, these may be reduced, yielding lower cost-effectiveness ratios. We conducted qualitative interviews with our peer coaches and identified key characteristics that motivated sustained participation and may be used to find a cohort of committed, long-term peer coaches.

 Second, the number of predicted CHD events avoided was low, as would be expected for a short (6-month) time frame. Extrapolating for 4 years, we found that predicted CHD risk would be reduced by 12% (0.73% absolute reduction from 6.1% baseline risk). Of course, additional costs would need to be incurred as well.

Another practical concern in regard to the high ICER for preventing 1 CHD event is that our within-trial analysis takes the provider or healthcare system’s perspective, so the provider must cover the cost of the additional services to prevent this event. At this point in the United States, peer coaches are not routinely covered by payers and behavioral health visits with mid-level staff would not be reimbursed unless payment for care were changed from fee-for-service to another bundled scheme, such as one under the medical home model. Indeed, providers are not rewarded directly for avoiding CHD events. Only in a setting such as a national healthcare system would this expense be clearly beneficial from the provider perspective. On the other hand, pay-forperformance measures in the United States do evaluate BP control for persons with hypertension, so a $47 expense per mm Hg reduction in BP in persons whose control continues to be inadequate despite treatment might presently be attractive to a provider.

Our within-trial cost-effectiveness results regarding the role of lay counselors in CHD risk reduction are much higher than those from other countries. Barton et al studied the costeffectiveness of lay health trainers in a randomized trial conducted in Liverpool, where the intervention involved offering information, advice, and support aimed at changing beliefs and behaviors in order to reduce risk of cardiovascular disease.

 The intervention had an estimated ICER of £14,480 (approximately $22,709 per QALY) over 6 months. However, there was only a 39.5% chance that it would be cost-effective at a WTP threshold of £20,000, and at no WTP threshold was it cost-effective with a probability greater than 50%. A costeffectiveness analysis of an Australian cluster-randomized trial of telephone counseling addressing diet and physician activity in 434 adult participants with type 2 diabetes mellitus or hypertension reported that the cost per QALY gained in 2008 Australian dollars was $29,375.

 Although these studies use different metrics for cost-effectiveness assessment, collectively they paint a much more favorable picture for the use of lay counselors to reduce cardiovascular risk than our withintrial analysis. Extrapolating the trial results using the Markov model suggested that our intervention was cost-effective, even under the conservative assumption that the intervention would have to be repeated every year.

This study targeted a highly vulnerable population— African Americans&mdash;who have a greatly increased CHD risk in the United States.

 CHD risk factor screening and counseling interventions that target low-income older persons in the community who are uninsured, such as the WISEWOMAN study, have found that, in the best-case analysis, it costs US $4400 per discounted life-year gained, but a sensitivity analysis revealed substantial uncertainty around this estimate.

 Nonetheless, these data suggest that patients in healthcare delivery settings and persons in the community with poor access to care may benefit from interventions to reduce CHD risks that exact a great toll on minority populations. Of course, the financial resources required for these interventions may not be insignificant.

METHODS

A cost-effectiveness analysis of a peer and practice staff support intervention to reduce coronary heart disease risk and blood pressure in African Americans.

Cost-Effectiveness of a Peer and Practice Staff Support Intervention

 Despite the role of mobile clinics in delivering care to the full spectrum of at-risk populations, the collective impact of mobile clinics has never been assessed. This study characterizes the scope of the mobile clinic sector and its impact on access, costs, and quality. It explores the role of mobile clinics in the era of delivery reform and expanded insurance coverage.

 A synthesis of observational data collected through Mobile Health Map and published literature related to mobile clinics.

Analysis of data from the Mobile Health Map Project, an online platform that aggregates data on mobile health clinics in the United States, supplemented by a comprehensive literature review.

 Mobile clinics represent an integral component of the healthcare system that serves vulnerable populations and promotes high-quality care at low cost. There are an estimated 1500 mobile clinics receiving 5 million visits nationwide per year. Mobile clinics improve access for vulnerable populations, bolster prevention and chronic disease management, and reduce costs. Expanded coverage and delivery reform increase opportunities for mobile clinics to partner with hospitals, health systems, and insurers to improve care and lower costs.

 Mobile clinics have a critical role to play in providing high-quality, low-cost care to vulnerable populations. The postreform environment, with increasing accountability for population health management and expanded access among historically underserved populations, should strengthen the ability for mobile clinics to partner with hospitals, health systems, and payers to improve care and lower costs.

Mobile clinics are an integral component of the healthcare system, serving vulnerable populations and promoting high-quality care at low cost.

		There are an estimated 1500 mobile clinics, receiving 5 million or more annual visits nationwide.

		Mobile clinics improve access for vulnerable populations, bolster prevention and chronic disease management, and reduce costs.

		Expanded coverage and delivery reform increase opportunities for mobile clinics to partner with hospitals, health systems, and insurers to improve care and lower costs.

The Affordable Care Act (ACA) has expanded insurance coverage to millions, but other barriers to access remain. Mobile clinics are customized vehicles that travel to the heart of communities, both urban and rural, and provide prevention and healthcare services where people work, live, and play. They overcome barriers of time, money, and trust, and provide community-tailored care to vulnerable populations.

Despite their scope and importance in delivering care to the full spectrum of at-risk populations, the collective impact of mobile clinics has never been assessed. Mobile clinics have historically lacked resources for evaluation, advocacy, and dissemination of best practices, which has reduced their visibility in the healthcare landscape. Recent collaborations and data sharing initiatives within the mobile clinic community provide the first opportunity to gather data and characterize the mobile clinic sector in the United States.

The Mobile Health Map project was launched to enable the first opportunity for the mobile clinic community to gather data about itself. Mobile Health Map is an online platform where the community of mobile clinics aggregates its data for the purpose of documenting the scope, geographic reach, and value of the services provided. In this article, we use data collected through this project as well as the small body of published research to provide the first comprehensive description of the mobile clinic sector and its role in the evolving care delivery environment. Specifically, we provide a description of the demographics of the mobile clinic sector and document its impact on access, quality, and costs. In the discussion, we use these results to explore the role of mobile clinics in the era of delivery form and expanded access.

Analysis in the study is based on data retrieved from the Mobile Health Map project as of March 11, 2013. As of this date, 644 clinics were registered on Mobile Health Map. The number of mobile clinics providing data as well as the type of data provided was as follows: 528 on service type, 282 on average annual number of visitors, 65 on visitor insurance status, 69 on visitor gender, 38 on visitor race, 72 on Hispanic status, 69 on visitor age, and 104 on rural/urban status. The data from the Mobile Health Map project were supplemented by a comprehensive literature review on mobile health clinics and their impact on access, quality, and costs.

Scale and Scope of the Mobile Clinic Sector

Scale. Mobile clinics operate in every state across the country, and in Washington, DC, and Puerto Rico (Figure). Epidemiological modeling estimates that 1500 to 2000 mobile clinics exist nationwide, based on the 644 mobile clinics currently registered on the Mobile Health Map project.

 Each mobile clinic serves an average of 3100 visitors per year, totaling approximately 5 to 6 million visits annually for the sector as a whole.

. Mobile clinics serve communities that have the poorest access to health services in the United States. Target populations include those of racial and ethnic minority backgrounds, the homeless, displaced populations, recent immigrants, migrant workers, people lacking insurance, and children. The clinics typically serve uninsured and lower-income individuals: 57% of visits are made by uninsured persons, and 35% by publically insured persons. Men utilize mobile clinics almost as much as women (46% of visits are made by men, 54% by women). People from racial and ethnic minority backgrounds are a key demographic for mobile clinics: 37% of visits are made by individuals that identify as non-white, while 43% are made by individuals identifying as Hispanic or Latino. Mobile clinics reach individuals of all ages: 42% of visits are by those younger than age 18, 50% by those aged 18 to 65, and 8% by those older than age 65. Mobile clinics serve both rural and urban populations: 14% of clinics serve rural areas, 41% serve urban areas, and 44% serve both.

. Mobile clinics provide a wide range of services tailored to community needs. Some function as comprehensive patient-centered medical homes, while others focus on specific diseases, such as pediatric asthma. Overall, 44% of mobile clinics provide primary care, 42% provide prevention services, and 31% provide dental care. Many also provide mammography, mental health, and a variety of specialty services.

. Mobile clinics have a diverse spectrum of organizational and funding arrangements. They can exist as independent entities (45%) or be operated by hospitals (25%), community health centers (16%), or academic medical centers (15%). Funding sources are similarly varied. Mobile clinics are primarily funded through philanthropy (38%), but state and federal programs (17%) as well as independent companies and private insurance providers (10%) are common sources of financial support.

Impact of Mobile Clinics on Access, Quality, and Costs 

Expanding access for vulnerable populations

. Mobile clinics are particularly successful in reaching vulnerable populations that have poorer health and less-than-optimal access to healthcare (

). By traveling to these communities and offering affordable, or, oftentimes, free services, mobile clinics remove logistical constraints such as transportation issues, difficulties making appointments, long wait times, complex administrative processes, and financial barriers such as health insurance requirements and copayments.

Mobile clinics also overcome subtler barrier challenges in maintaining trusting relationships between healthcare providers and community members. Qualitative research has found that the mobile clinic staff’s marriage of professional and personal discourses fosters trusting relationships, allowing them to stay engaged with clients and to support them through behavior changes and navigation of the healthcare system.

Promoting prevention. There has been considerable national focus on the need for safety-net programs to provide communitybased prevention, screening, and chronic disease management, particularly for low-income, minority, and rural communities. Several studies have found that mobile clinics are successful at improving screening and identifying high rates of chronic and infectious disease among underserved populations.

Improved screening allows mobile clinics to deploy interventions that improve treatment and prognosis. For example, studies found that underserved women utilizing the van services for prenatal care initiated care earlier than women using other services.

. Many mobile clinics provide ongoing support for chronic disease management to vulnerable populations. Patients who presented at the Family Van clinic in Boston with high blood pressure experienced reductions in their systolic and diastolic blood pressure of 10.7 and 6.2 mm Hg, respectively, during followup visits. This correlated with a 32% reduction in the relative risk for heart attack and a 45% reduction in the relative risk for stroke.

 Similarly, a school-based mobile asthma-care clinic for urban underprivileged children was associated with increased daily anti-inflammatory medication usage, decreased emergency department (ED) visits, hospitalizations, and school absenteeism compared with before enrollment.

. Mobile clinics can produce significant cost savings as a result of their ability to provide community-tailored care in high-risk areas. The principal source of savings is a reduction in avoidable hospital and ED visits. An asthma-oriented mobile clinic generated savings of $3500 per child due to reduced ED visits and hospitalizations resulting from improved asthma control.

 A study that aggregated data from 10 larger, more comprehensive mobile clinics estimated cost savings of $6.8 million from avoidable ED visits over a 1-year period.

 Significant cost savings are also possible from improved disease management and increased use of preventive services. The improved hypertension control in the Family Van mobile clinic was associated with a lower-bound return on investment estimate of 1.3 when combined with avoided ED visits.

When the long-term benefit of prevention is considered, the return on investment is more substantial. A study of the Family Van in Boston used estimates of the value of quality life-years saved from preventive services and the cost of avoidable ED visits to estimate a staggering return on investment of $30 for every $1 invested in the program.

Mobile clinics represent an integral component of the healthcare system that delivers care to populations that are hard to reach by the traditional system, improving access and supporting prevention and chronic disease management. Mobile clinics are able to leverage their ability to overcome barriers to access and build trusting relationships to reduce disparities, improve health, and reduce costs.

Despite this potential, many mobile clinics often struggle to forge financially sustainable arrangements with care delivery organizations and payers. Recent reforms to the delivery system and health insurance coverage have the potential to improve the integration of mobile clinics into existing care delivery structures.

Since the passage of the ACA in 2010, federal and private payers are increasingly negotiating at-risk accountable care contracts with hospitals and health systems. The expectation of these arrangements is that not only will they improve the health and experience of care for individuals, but also that they will improve the health of populations while reducing the rate of cost growth. But hospitals and health systems often lack access to, or expertise in, the levers necessary to control health at the community level. Instead of struggling to develop these capacities, care delivery organizations can partner with mobile clinics that have experience in addressing the health needs of communities. Mobile clinics are uniquely qualified to provide high-quality care to underserved populations, and they can do so with considerable cost savings. St. Joseph Health, a $4.4-billion nonprofit Catholic health system with 14 hospitals serving California, Texas, and New Mexico, is a prime example. St Joseph Health invests $5 million annually in mobile health clinics, 11% of which is offset by reimbursements. These mobile clinics provide care through their more than 32,000 patient encounters annually, and offer services such as primary care and dental and vision care to people in need.

Historically, one of the major challenges for mobile clinics has been demonstrating return to investors. Delivery reform focused on population health management has the potential to help mobile clinics secure stable funding streams. There is evidence that mobile clinics can generate cost savings and substantial return on investment. Traditionally, however, these savings did not accrue to the delivery organization. The expansion of contracts that tie hospital compensation to performance on quality and costs will bolster the business case for investments in mobile clinics; savings generated by mobile clinics will accrue directly to the hospital.

A closely related recent development is the expansion of health insurance coverage. Mobile clinics serve mainly the publically insured and uninsured. This mix will shift as previously uninsured individuals enroll in Medicaid programs during the implementation of ACA coverage reforms. It is notable that the Family Van clinic in Massachusetts did not see a decline in visitors after healthcare reform took effect in 2006—92% of individuals visiting the Van have health insurance.

 Even with expanded coverage, there are still barriers to primary care services, such as waiting times, copayments, complexities of navigating the system, and feelings of intimidation.

A growing Medicaid population should facilitate partnerships with care-delivery organizations and strengthen funding streams. State Medicaid programs are turning to contracting mechanisms that place providers at financial risk for population health management and cost control. These reforms should promote partnerships between mobile clinics and health centers that care for large Medicaid populations under accountable care contracts. Similarly, as Medicaid programs look to improve value amid increasingly tight state budgets, policy makers should consider increasing funding to mobile clinics. Sick, disadvantaged patients with complicated social circumstances account for a majority of Medicaid spending. The ability of mobile clinics to reach vulnerable populations and promote improved disease management suggests that they will be effective in addressing the needs of this population.

 Harvard Medical School, Boston, MA (CFH, BWP, SHJ, JB, NEO); Boston-VA Medical Center, Boston, MA (SHJ); Merck and Co, Inc, Boston, MA (SHJ); Callen-Lorde Community Health Center, New York, NY (AV); Beth Israel Deaconess Medical Center, Boston, MA (NEO).

 The Mobile Health Map project has been supported by the Boeing Company, the President & Fellows of Harvard College, Provost Fund, Ronald McDonald House Charities, and the US Department of Health and Human Services’ Office for Minority Health through the Hispanic Association of Colleges and Universities Award # 6PCMP111058-01-02.

 Dr Vavasis reports being the board chair of the Mobile Health Clinics Association, which stands to benefit financially from increased membership, a possible result of this article. Ms Hill reports doing paid work for the Family Van mobile health program and applying for and receiving grants to support mobile healthcare programs. The other authors (BWP, SHJ, JB, NEO) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (CFH, BWP, AV, NEO); acquisition of data (CFH, NEO); analysis and interpretation of data (CFH, BWP, NEO); drafting of the manuscript (CFH, BWP, AV, NEO); critical revision of the manuscript for important intellectual content (CFH, BWP, JB, AV, NEO); statistical analysis (CFH); obtaining funding (JB, NEO); administrative, technical, or logistic support (CFH); supervision (JB, NEO).

 Nancy E. Oriol, MD, Dean of Students, Harvard Medical School, Ste 244, 260 Longwood Ave, Boston, MA 02115. Email: Nancy_Oriol@hms.harvard.edu.1. US Department of Health and Human Services, Office of Minority Health. Mobile health clinics in the United States: reducing disparities: improving care, improving health, controlling costs. http://www.mobile healthmap.org/documents/Mobile_Health_Clinics_in_the_United_States_ March_2013.pdf. Published March 2013. Accessed April 11, 2013.

2. Fleurant M, Hill C, Bennett J, Simon SR, Oriol NE, Hicks L. The use of health information technology among mobile health clinics in the United States in 2010. Mobile Health Clinics Network Conference 2011. http://www.mobilehealthmap.org/roi.php. Accessed April 11, 2013.

3. Harris DE, Hamel L, Aboueissa AM, Johnson D. A cardiovascular disease risk factor screening program designed to reach rural residents of Maine, USA. Rural Remote Health. 2011;11(3):1-15.

4. Campos M, Olmstead-Rose L. Mobile health clinics: increasing access to care in central and eastern Contra Costa County: final report. http://www.johnmuirhealth.com/content/dam/jmh/Documents/Community/ Mobile_Health_Clinics-Increasing_Access_to_Care.pdf. Published April 2012, Accessed June 10, 2013.

5. Hill C, Zurakowski D, Bennet J, et al. Knowledgeable Neighbors: a mobile clinics model for disease prevention and screening in underserved communities. Am J Public Health. 2012;102(3):406-410.

6. US Department of Health and Human Services, Agency for Healthcare Research and Quality 2011. National healthcare disparities report. http://www.ahrq.gov/research/findings/nhqrdr/nhdr11/. Published April 2012. Accessed April 23, 2013.

7. Altice FL, Bruce RD, Walton MR, Buitrago MI. Adherence to hepatitis B virus vaccination at syringe exchange sites. J Urban Health. 2005;82(1):151-161.

8. Kahn RH, Moseley KE, Thilges JN, Johnson G, Farley TA. Community- based screening and treatment for STDs: results from a mobile clinics initiative. Sex Transm Dis. 2003;30(8):654-658.

9. Carmack HJ. “What happens on the van, stays on the van”: the (re) structuring of privacy and disclosure scripts on an Appalachian mobile health clinic. Qual Health Res. 2010;20(10):1393-1405.

10. Rodriguez KL, Appelt CJ, Young AJ, Fox AR. African American veterans’ experiences with mobile geriatric care. J Health Care Poor Underserved. 2007;18(1):44-53.

11. Guruge S, Hunter J, Barker K, McNally MJ, Magalhães L. Immigrant women’s experiences of receiving care in a mobile health clinic. J Adv Nurs. 2010;66(2):350-359.

12. Edgerley LP, El-Sayed YY, Druzin ML, Kiernan M, Daniels KI. Use of a community mobile health van to increase early access to prenatal care. Matern Child Health J. 2007;11(3):235-239.

13. Ellen JM, Bonu S, Arruda JS, Ward MA, Vogel R. Comparison of clients of a mobile health van and a traditional STD clinic. J Acquir Immune Defic Syndr. 2003;32(4):388-393.

TRENDS FROM THE FIELD

This article reviews the mobile clinic sector's impact on access, quality, and costs, and explores postreform opportunities for leveraging them nationally.

Mobile Health Clinics in the Era of Reform

 We measure the impact of Medicaid formulary restrictions (FRs) on the rate at which patients who previously failed a drug therapy for schizophrenia are returned to that therapy.

 We collect drug-level information on FRs in state Medicaid programs and examine claims of noninstitutionalized Medicaid enrollees with schizophrenia.

A difference-in-differences technique is used to compute the change in the probability of adverse outcomes before and after a state adopts an FR. This change is compared with the change in failure probabilities in states with no FRs.

 Regardless of FRs, patients tend to resume the same drug after an adverse medical event. In 2005, 69% of inpatient mental health—related admissions resulted in patients resuming the same therapy within 6 months of the event, and 63% of patients resumed the same drug after a mental health–related emergency department admission. In states where FRs limit access to all atypicals, the likelihood of a patient resuming the same atypical after having ceased treatment for at least 30 days increases by 20.1% relative to patients in states without restrictions. Additionally, patients in states that impose FRs on all atypicals are 11.6% more likely to discontinue all treatments.

 FR may increase the likelihood that patients will return to failed treatments or cease treatment altogether. Although formularies are designed to reduce drug spending, an unintended consequence may be an increase in the use of other services needed to treat patients with schizophrenia.

We measure the impact of Medicaid formulary restrictions (FRs) on the rate at which patients who previously failed on a particular drug therapy for schizophrenia are returned to that same therapy. Our findings suggest that FRs increase the likelihood that patients will return to failed treatments, or totally cease treatment.

		In 2005, 69% of all schizophrenia patients’ mental health—related admissions resulted in patients resuming the same schizophrenia treatment within 6 months

		Patients in states that impose FRs on all atypical antipsychotics are 20.1% more likely to resume use of the same atypical after having ceased treatment for at least 30 days and 11.6% more likely to discontinue all treatments relative to patients in states without restrictions.

Between 1995 and 2005, Medicaid spending on drugs increased an average of 13% per year—an increase due in part to physicians favoring newer, more expensive drugs.

 To attempt to control costs, many state Medicaid programs responded by imposing policies that encourage the use of lower-cost drugs.

Prior authorization, for example, is designed to reduce pharmacy spending by requiring physicians to seek approval from the state’s Medicaid program before prescribing certain high-cost drugs to patients. These controls are designed to reduce the demand for expensive drugs when lowercost alternatives exist, which ultimately results in reduced Medicaid spending. But these policies may affect healthcare spending and utilization through other channels, especially if patients do not respond well to the preferred drugs. Physicians match patients to the drug and dosage that is optimal for each patient based on various practice guidelines, recommendations, and formulary practices.

On average, several treatments might be equally effective at the population level, but heterogeneous patients will respond differently to each. Removing a drug from coverage could adversely affect those who would have most benefited from it. If the use of these drugs leads to deteriorating health status, the result may be higher nonpharmaceutical medical spending.

An important example of this trade-off is the second generation of drugs for the treatment of schizophrenia, called atypical antipsychotics, which account for almost 15% of Medicaid spending.

 For this reason, they are increasingly popular targets of state Medicaid formulary restrictions.

 Because a number of patients may respond only to 1 drug, formulary restrictions can place patients with idiosyncratic treatment responses at greater risk of uncontrolled symptoms.

 Uncontrolled schizophrenia has dramatic costs for patients (about 10% of whom commit suicide), and for society in terms of increased social service use and crime.

 Although lower-cost alternatives to atypicals do exist, they are problematic. The first generation of schizophrenia drugs resulted in improvement of symptoms but with potentially debilitating side effects.

 Atypical antipsychotics were introduced in the 1990s and have largely replaced first-generation treatments because atypicals control schizophrenia better and have fewer negative side effects. For example, about 30% to 40% of patients relapse with first-generation drugs (relative to 80% without treatment), but with second-generation drugs, relapse rates fall to about 25%.

Formulary restrictions appear to alter doctors’ prescribing behavior. Previous studies have examined the impact of utilization management programs and found that formulary restrictions reduced the use of atypicals by about 5%.

 After prior authorization of atypicals became a requirement in Maine, the risk of discontinuation rose by 29%.

 Studies in Georgia and New Hampshire found that prior authorization reduced the use of targeted drugs and increased the use of other health services.

 Previous studies of the impact of restrictions on treatment have focused on individual states. This study, to the best of our knowledge, is the first nationally representative study of formulary restrictions’ impact on treatment. In this study, we examine how formulary restrictions of atypicals affect the likelihood that a patient who has been diagnosed with schizophrenia, and who has had some marker of an unsuccessful response to a particular drug therapy, is continued on that same drug therapy, changes the type of drug, or discontinues all treatment.

Policy information on the restrictions placed on atypical antipsychotics from 1999 to 2005 was collected from 24 state Medicaid programs via a mailed survey and personal communications with policy directors. States reported whether prior authorization, quantity limits, step therapy (fail-first), and other policies applied to a list of drugs identified by US brand name. To supplement this information, additional formulary information was collected using publicly available documents. Regulation data were collected at the individual drug level. The survey included 10 atypicals. All major atypical antipsychotic agents are addressed in the survey, except for paliperidone (brand: Invega) and olanzapine/fluoxetine combined therapy (brand: Symbax), both of which were not yet commercially available during the study period. States reported if each drug was placed under prior authorization, quantity limits, step therapy, dose restriction, or duplicate therapy restrictions in each year.

Medicaid claims for 24 states (12 states that adopted formulary restrictions and 12 states that did not) for a 5-year period (2001-2005) were used to measure medical and pharmacy use at the individual level. We selected the states shown in 

 based on the size of the Medicaid population; the fraction of enrollees in fee-for-service Medicaid (with complete claims); and whether the state Medicaid program had imposed restrictions on selected classes of medications over the study period. The years included in this study (2001-2005) captured the widespread introduction and use of prior authorization requirements in many state Medicaid programs, particularly for atypical antipsychotics.

The sample consisted of Medicaid enrollees 18 years and older who have schizophrenia in each of the study states. These patients were identified by (1) the presence of an 

International Classification of Diseases, 9th Revision, Clinical Modification

) code beginning with 295 in any diagnosis position for inpatient, outpatient, or long-term care claims, and (2) a confirmatory schizophrenia diagnosis at least 30 days after the initial diagnosis. Thus, we eliminated individuals who were diagnosed only once without a subsequent diagnosis. This eliminated individuals with a socalled “rule-out” diagnosis: a physician may make an initial diagnosis of suspected schizophrenia that is later ruled out. The requirement of a subsequent “confirmatory” diagnosis excluded “ruled out” cases from the sample. This criterion also imposed a 30-day minimum window of observation for inclusion in the study. In addition, to be included, a patient must have filled at least 1 prescription for one of the drugs included in our sample. In practice, this restriction eliminated very few patients because those with a subsequent confirmatory diagnosis are almost always given a prescription.

presents a consort diagram for the sample construction: 944,609 patients over the age of 18 years who have had at least 1 schizophrenia diagnosis. Of these, about 77% had a confirmatory diagnosis after 30 days. Of the remaining population, 63%, or 460,991 patients, were prescribed an atypical.

To estimate the impact of prior authorization on treatment failure and “replication of failure,” a difference-in-differences method (DD) was used. In the DD method, 2 groups—patients in states with restrictions and patients in states without restrictions&mdash; were observed in 2 different time periods: 2001 and 2005. Patients living in states that adopted formulary restrictions after 2001 were, in the terminology of DD models, exposed to “treatment” in 2005, while those in nonformulary states were not exposed in either period. This method first calculates the change in the average likelihood of the outcome variable in the treated group, and then calculates the corresponding change in the nontreated group. The excess change in the treated group, above and beyond the change in the nontreated group, represents the estimated DD effect. For example, in states with formulary restrictions, we calculated the change in the likelihood of a patient resuming the same medication after a 30-day gap before and after the implementation of the restriction. We then compared this with the corresponding difference in likelihood within states that never adopted restrictions. The approach removes the effects of trends due, for example, to general changes in atypical use between 2001 and 2005. Because formulary restrictions were measured at the state level, we clustered the standard errors at the state level. We utilized an alpha of 0.10, although the tables included alternative alpha values.

To examine the replication of failure, all patients currently with schizophrenia who are taking an atypical antipsychotic for their condition were identified. Then, patients were classified as failing on a particular therapy if they either were admitted to a hospital for their disease or had a medication gap greater than 30 days (in a baseline year). Patients who failed treatment were followed over time, and their claims were examined for resumption of treatment with the same drug after at least a 30-day gap. As a robustness check, we examined longer gaps, but faced serious sample size limitations due to the limited number of people observed restarting after longer gaps. Finally, we also examined whether individuals on an atypical discontinued drug therapy altogether, switched from an atypical to a typical (first generation) antipsychotic, or switced from one atypical to another.

We included an indicator variable equal to 1 in the year 2005 and an indicator for states that adopt formulary restrictions. The interaction between the 2 indicators, year 2005 and formulary restrictions, is the difference-in-differences estimate discussed above (labeled DD in the tables).

 shows summary statistics measuring atypical antipsychotic usage along with emergency department (ED) and inpatient admissions related to mental health causes, stratified by the number of atypicals under a formulary restriction in a given state-year. The observed population with schizophrenia increased slightly over time, from just fewer than 500,000 in 2001 to about 541,000 in 2005. The number of patients with schizophrenia receiving atypical antipsychotics increased from nearly 195,000 in 2001 to over 320,000 in 2004, and then fell to about 273,000 in 2005.

Each patient, on average, experienced 0.22 inpatient admissions related to mental health issues and 0.06 ED visits per year. Approximately 2.8% of the sample was admitted to the ED and 7.9% was admitted on an inpatient basis for issues related to mental health.

 summarizes the replication of failure by year, and the restrictiveness of Medicaid formulary in that year. The variable of interest is the number of times a patient goes off a particular medication for 30 days and then resumes that same medication. For example, a value of 1 indicates that the average patient had one 30-day gap in their medication. Patients in states with formulary restrictions tend to have slightly more replication of failure, measured as stoppages and restarts. For example, in 2005, individuals in states with all atypicals under restriction experienced an average of 1.04 stoppages and subsequent restarts of the same atypical in the same year, while those in states with no restrictions experienced an average of 0.99 stoppages and restarts per year.

This study measures the impact of Medicaid formulary restrictions on the rate at which patients who failed on a drug therapy for schizophrenia return to the same therapy.

March 2014

Medical Homes and Cost and Utilization Among High-Risk Patients

Cost-Effectiveness of a Peer and Practice Staff Support Intervention

Mobile Health Clinics in the Era of Reform

Do Strict Formularies Replicate Failure for Patients With Schizophrenia?

Beyond Black and White: Race/Ethnicity and Health Status Among Older Adults

The Economics of Resistant Pathogens and Antibiotic Innovation

STABLE Results: Warfarin Home Monitoring Achieves Excellent INR Control

Collaborative DTM Reduces Hospitalization and Healthcare Costs in Patients With Diabetes Treated With Polypharmacy

Documentation of the 5 As for Smoking Cessation by PCPs Across Distinct Health Systems

Assessing the Chiral Switch: Approval and Use of Single-Enantiomer Drugs, 2001 to 2011