In-Person and Video-Based Post-Traumatic Stress Disorder Treatment for Veterans: A Location-Allocation Model

ABSTRACT Post-traumatic stress disorder (PTSD) is associated with poor health but there is a gap between need and receipt of care. It is useful to understand where to optimally locate in-person care and where video-based PTSD care would be most useful to minimize access to care barriers, care outside the Veterans Affairs system, and total costs. We developed a service location systems engineering model based on 2010 to 2020 projected care needs for veterans across New England to help determine where to best locate and use in-person and video-based care. This analysis determined specific locations and capacities of each type of PTSD care relative to patient home locations to help inform allocation of mental health resources. Not surprisingly Massachusetts, Connecticut, and Rhode Island are well suited for in-person care, whereas some rural areas of Maine, Vermont, and New Hampshire where in-patient services are infeasible could be better served by video-based care than external care, if the latter is even available. Results in New England alone suggest a potential $3,655,387 reduction in average annual total costs by shifting 9.73% of care to video-based treatment, with an average 12.6 miles travel distance for the remaining in-person care.

INTRODUCTION

Post-traumatic stress disorder (PTSD) is a serious mental health disorder that results in significant effects on physical and social well-being.1,2 PTSD is common among Americans with a prevalence of 3.7% in the general population, yet only 7% of those affected seek treatment in their first year, with a lifetime treatment-seeking rate of just over 50%.3,4 Given the vast medical and social costs of untreated or undertreated PTSD, there is a critical need to improve access to the most effective treatments.5-7 However, access to treat- ment, particularly for those residing in rural areas, is a prac- tical issue. This is even more the case for both active duty and veteran populations, with recent estimates from the wars in Afghanistan and Iraq indicating 13.8% of those who served developed PTSD.8

Innovative treatment modalities, such as telehealth or video-based services, hold great promise to increase access to important effective mental health treatment services, par- ticularly in rural areas where access to in-person mental health treatment is limited.9 Large integrated health systems such as the Veterans Affairs' (VA) 21 regional service net- works across the U.S. (Veterans Integrated Service Networks [VISNs]) can benefit from systems engineering tools to help plan the most appropriate ways to meet mental health needs of patient populations. Our aim was to apply longitudinal systems engineering care location models to help determine the optimal geographic locations and capacities for in-person care for veterans with PTSD across the New England VA network (VISN-1) and where instead video-based care would be advantageous over either expensive within-system capac- ity expansion or fee-based external care, if even feasible.

METHODS

Study Location

Because receipt of a PTSD diagnosis requires access to care, we used estimates of overall PTSD prevalence among VA users rather than diagnostic codes in VISN-1 to estimate PTSD prevalence. The most contemporary figures available on PTSD prevalence among VA users are derived from the current cohort of returning veterans. The VA Environmental Epidemiology Service found that 741,954 of Iraq and Afghanistan veterans had used the VA by the end of the 2011 fiscal year and that 207,954 (27.9%) of these had been diagnosed with PTSD.10,11 The rate of PTSD diagnosis also has been increasing over time.12-14 Therefore, we lowered this figure to 25% since we were interested in VISN-1 users as of 2010. We multiplied the number of New England veterans who use VA health services by 0.25 to estimate the number of PTSD cases in VISN-1, producing 36,469 in 2010 and with the majority residing in Massachusetts (11,521) and Connecticut (8,992). Within the VA, veterans with PTSD are treated in specialized PTSD clinics, general mental health clinics, or primary care clinics. Specialized PTSD clinics currently are located only at some of the 9 medical centers across New England. General mental health services are pro- vided at medical centers and large community-based out- patient clinics, whereas smaller community-based outpatient clinics typically provide PTSD care either by referral to their affiliated medical center or through the use of video-based telemental health services.15 The New England VA network consists of nine comprehensive hospital facilities (medical centers) and 39 community-based outpatient clinics. These hospital and clinics are geographically distributed throughout the New England region (Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, and Connecticut). Six of nine VA medical centers (Boston, Massachusetts; Brockton, Massachusetts; West Haven, Connecticut; Providence, Rhode Island; Togus, Maine; and White River Junction, Vermont) have specialized PTSD care teams. In 2010, 35.15% of patients who received a primary diagnosis of PTSD used specialized PTSD services and an additional 56.35% of veterans with a primary PTSD diagnosis received care in general mental health services.

Location of Patients

We determined the 3-digit zip code geographic location of each veteran in New England using administrative data from the New England VA Data Warehouse. Since relying on a diagno- sis of PTSD requires access to services, we assigned a PTSD diagnosis to 25% of the patients in each zip code. Figure 1 depicts the heterogeneous geographic distribution of veterans with PTSD in raw counts (Fig. 1A) and per square mile (Fig. 1B) across VISN-1, with the locations of VA facilities.

Location of Clinic Sites

We mapped the location of all current VA care sites in New England. Each site was characterized by its location, staffing, and clinical volume.16 We estimated the service requirements to deliver both initial evidence-based PTSD care and ongoing treatment for the portion unlikely to remit. Taking into account non-PTSD mental services workload based on diag- nostic information for all VISN-1 veterans, we used engi- neering optimization models to determine the optimal total amount of mental health manpower to serve each location of veterans. We determined for each facility the optimal amount of each type of PTSD care service to provide and for each PTSD patient whether service could be economically pro- vided within a specified travel distance to any existing PTSD treatment facility. The remainder was considered candidates for video-based services as a cost-effective alternative to external care, which may not even be available within any reasonable distance.

Modeling Clinical Needs for Each Location

We used our mathematical model to assign each PTSD patient to the best close VA clinic site. Location optimization models of this type typically are used by systems engineers in many industries to determine the location of facilities relative to geographic and longitudinal demand, including some applications in health care.17-19 This approach also can produce a more patient-centered balance between use of in-person and video-based services. Assignments of care types to sites were made regardless of current services at them, along with assignment of PTSD patients to these sites to determine who ideally should be seen where. VA goals for maximum acceptable travel distances (currently 30 miles for PTSD care) were used to determine these service locations, patient destinations, and if a particular patient could receive within-network in-person care. Those that could not receive in-person care within the VISN-1 network were considered candidates for video-based care rather than outside the VA system in a manner so as to minimize total care and delivery costs. Although the 30 mile travel threshold is defined by the VA's central office, we also repeated our analysis for other maximum distances to investigate how results might change for other thresholds. We additionally computed the clinical care and travel costs associated with each patient, aggregated by clinic site and for the region as a whole.

Other model inputs, constraints, and assumptions included:

(1) PTSD treatment preferences among VA users were assumed to be (a) 60% want evidence-based psycho- therapy only, (b) 25% want medication only, and (c) 15% want both.20

(2) Assumptions of evidence-based psychotherapy included: (a) Conducted by psychotherapists (i.e., with at mini- mum a master's degree); (b) consisted of one evalua- tion, 12 treatment sessions, and three phone calls; and (c) a total of 13.75 clinical hours needed per patient per year. Clinical hours were allotted 1 hour for evalu- ation, 1 hour for each treatment session, and 15 minutes for each phone call, based on typical session lengths. Evidence-based psychotherapy was defined by the VA/DoD Clinical Practice Guideline for the Manage- ment of Post-Traumatic Stress Disorder21 and Uni- form Mental Health Services in VA Medical Centers and Clinics.22

(3) Medication assumptions included the following: (a) Prescribed by psychiatrists; (b) consisted of one evaluation session, six follow-up visits, and three phone calls; and (c) a total of 4.75 clinical hours needed per patient per year. For psychiatrist clinical hours, we allotted 1 hour for evaluation session, 1 hour for each follow-up visit, and 15 minutes for each phone call.

(4) Mental health staff not located in specialized PTSD clinics spent 75% of their time addressing disorders other than PTSD.

The primary goal of the model (the objective function, in systems engineering language) was to minimize VISN-1's total cost of PTSD care while still satisfying the above travel distance requirements, treatment preferences, and logistical constraints. The following variables are calculated by our model: treatment and travel costs for patients who receive services within the VA and care costs for patients not able to obtain care at any VA clinic. VA care costs also included travel costs for distances more than 30 miles, based on the VA's travel reimbursement policy. An overview of our math- ematical model is given in Table I.

RESULTS

Data and Parameters

The location of VA users with PTSD in New England was determined using 3-digit zip codes. Distances between each 3-digit zip code and all VA clinics then were determined. Future PTSD care demand for the next 10 years was esti- mated via the VA's veteran population VET DATA projec- tions (http://www.va.gov/VETDATA/Veteran_Population.asp), which estimates significant decreases over the next 20 years because of aging veteran populations and smaller modern mil- itary sizes. Total need estimates for each future year were geographically distributed among all 59 VISN-1 zip codes based on current relative proportions. Since locations of care services, capacities, and patient routings are interdependent, our model then simultaneously and optimally located each type of PTSD care to each candidate facility, determined the optimal capacities for these services at each facility, and determined the optimal facilities at which PTSD patients at each 3-digit zip code should receive their care if possible within 30 miles.

Model Results

Figure 2 shows the trade-off between maximum acceptable driving distance, total minimal cost, and access for the esti- mated 2010 PTSD care demand. Also shown is the current situation for comparison, estimated based on current patient allocations, which has an 11.08% ($10.9 million) higher cost and 56.26% lower access than the optimized case even with no telehealth service, because of current suboptimal system design. Using telehealth services where demand cannot be covered by the VA within the acceptable distance of 30 miles, rather than providing care at non-VA facilities, would result in further savings for 2010 of $4,001,664 (4.04%) given the same access coverage, or alternately a roughly 50% reduction in maximum driving distance to 30 miles given the same cost. Repeating this analysis for 2010 to 2020 produces a potential average estimated savings across all 10 years of $9.8 million/year by optimizing the cur- rent care system and an additional $3,655,387 per year by using telehealth. Note that all demand can be covered by the VA if a maximum acceptable driving distance of more than 60 miles is allowed. This, however, causes a 45% increase in average driving distance per patient that exceeds the VA's maximum travel distance policy (30 miles for PTSD care).

Table II summarizes the resulting optimal locations of in-person PTSD services and the number of psychotherapist and psychiatrist FTEs required at each site to provide this care. As shown, rural pockets of Maine, Vermont, and New Hampshire would be better served with video-based mental health services since demand in these areas cannot be covered by any of the VA facilities. In Massachusetts, Connecticut, and Rhode Island, there is sufficient need and access for in-person PTSD services, with the possible exception of the medical center located in Jamaica Plain, Massachusetts, which might be treated in the next closest location instead since only an estimated 17 patients here will need care during the year. Figure 3 illustrates how optimal within-network and video-based care locations change if some other maximum travel distances (15, 30, 45, 60 miles) were allowed. Finally, Figure 4 summarizes future projected needs for PTSD treatment by therapists and psychiatrists, along with the estimated number of veterans needing PTSD care. As expected, the number of providers required decreases over the next 20 years as a result of shrinking veteran population projections, whereas zip codes that would be better served with video-based services remain the same.

DISCUSSION

This study illustrates the potential value of the use of systems engineering models to help design an overall system that provides appropriate access and considers patient needs and preferences, at minimal cost. A few limitations include logic, data, and preference assumptions in our model. For example, we assume providers trained in treating PTSD are available or could be relocated to each clinic site, which may not be possible immediately, in which case results can represent an upper bound on maximal potential savings for benchmark comparison; additional or temporary reliance on telemental health services also may be possible. The few trials that compared the effectiveness of in-person versus remote PTSD treatments did not detect differences but also were not powered to determine noninferiority.23 -25 Assumptions regarding patient treatment preference also were based on a pilot sample20 and were assumed to be equally distributed across the patient population. Rural patients instead, for example, may have higher preference for medication treat- ment. It also is not clear that all VA users have a preference for or an understanding of evidence-based psychotherapy. Although the VA follows guidelines that promote evidence- based treatment,21,22,26 the definition of evidence-based psy- chotherapy is not universal and instead for example could be defined as listed in the National Registry of Evidence-based Programs and Practices.27,28 The recent Re-Engineering Systems of Primary Care Treatment (RESPECT)-PTSD trial14 also showed the difficulty of implementing evidence- based care for PTSD by primary care providers. In addition to psychiatrists, medications also can be prescribed by nurse practitioners, physician assistants, or primary care physicians and therefore our medication-prescribing estimate may be conservative. Finally, our results have based only on the New England VA network, although the general value of this approach seems generalizable to all VISNs.

CONCLUSION

This study highlights the need and value of appropriate allo- cation of PTSD services for veterans. Results can be useful as input for informing PTSD resource planning within the U.S. Department of Veteran Affairs. Access to care is a particular concern for veterans who are unable to readily access PTSD treatment, such as those living in rural environments. In such cases, new treatment modalities such as telehealth (i.e., via video conference) can be considered. Such treatment actually may be preferred by some veterans given it increases conve- nience and reduces privacy concerns.

For the VA population, most clinic sites have sufficient volume to justify in-person staff for face-to-face PTSD treat- ment. There are a small number of community-based clinics, however, whose estimated demand is insufficient to justify in-person staff. For these sites, our analysis suggests that PTSD treatment needs would be better met by video-based services. In general, these sites were in rural or highly rural regions of New England, primarily in central Vermont, south- western New Hampshire, and much of northern Maine. All other areas have sufficient need to support in-person services, assuming availability. Additional research could examine if teletreatment modalities affect treatment retention among veterans or the active duty population. More generally, use of operations research and systems engineering methods appears useful to help inform these types of macro system design and policy issues. Similar analyses for the location of other evidence-based mental health services therefore also may be similarly beneficial.

REFERENCES

1. Hermann BA, Shiner B, Friedman MJ: Epidemiology and prevention of combat-related post-traumatic stress in OEF/OIF/OND service members. Mil Med 2012; 177(8 Suppl): 1- 6.

2. Shiner B, Drake RE, Watts BV, Desai RA, Schnurr PP: Access to VA services for returning veterans with PTSD. Mil Med 2012; 177(7): 814-22.

3. Kessler RC, Chiu WT, Demler O, Merikangas KR, Walters EE: Preva- lence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005; 62(6): 617-27.

4. Wang PS, Berglund P, Olfson M, Pincus HA, Wells KB, Kessler RC: Failure and delay in initial treatment contact after first onset of mental disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005; 62(6): 603-13.

5. Ivanova JI, Birnbaum HG, Chen L, Duhig AM, Dayoub EJ, Kantor ES, Schiller MB, Phillips GA: Cost of post-traumatic stress disorder vs. major depressive disorder among patients covered by Medicaid or private insurance. Am J Manag Care 2011; 17(8): e314 -23.

6. Cohen BE, Gima K, Bertenthal D, Kim S, Marmar CR, Seal KH: Mental health diagnoses and utilization of VA non-mental health medical services among returning Iraq and Afghanistan veterans. J Gen Intern Med 2010; 25(1): 18-24.

7. Murdoch M, van Ryn M, Hodges J, Cowper D: Mitigating effect of Department of Veterans Affairs disability benefits for post-traumatic stress disorder on low income. Mil Med 2005; 170(2): 137- 40.

8. Schell TL, Marshall GN: Survey of individuals previously deployed for OEF/OIF. In: Invisible Wounds of War: Psychological and Cogni- tive Injuries, Their Consequences, and Services to Assist Recovery, pp 87-115. Edited by Tanielian TL, Jaycox LH. Santa Monica, CA, RAND Corporation, 2008.

9. Barnwell SV, Juretic MA, Hoerster KD, Van de Plasch R, Felker BL: VA Puget Sound telemental health service to rural veterans: a growing program. Psychol Serv 2012; 9(2): 209 -11.

10. VA Office of Public Health: Analysis of VA Health Care Utilization among Operation Enduring Freedom (OEF), Operation Iraqi Freedom (OIF), and Operation New Dawn (OND) Veterans: Cumulative from 1st Qtr FY 2002 through 4th Qtr FY 2011 (October 1, 2001-September 30, 2011). Washington, DC, Department of Veterans Affairs, 2012. Available at http://www.publichealth.va.gov/epidemiology/; accessed November 28, 2012.

11. VA Office of Public Health: Report on VA Facility Specific Operation Enduring Freedom (OEF), Operation Iraqi Freedom (OIF), and Operation New Dawn (OND) Veterans Coded with Potential PTSD: Cumulative from 1st Qtr FY 2002 through 4th Qtr FY 2011 (October 1, 2001- September 30, 2011). Washington, DC, Department of Veterans Affairs, 2012. Available at http://www.publichealth.va.gov/epidemiology/; accessed November 28, 2012.

12. Rosenheck RA, Fontana AF: Recent trends in VA treatment of post- traumatic stress disorder and other mental disorders. Health Aff (Millwood) 2007; 26(6): 1720 -7.

13. Hermes ED, Rosenheck RA, Desai R, Fontana AF: Recent trends in the treatment of posttraumatic stress disorder and other mental disorders in the VHA. Psychiatr Serv 2012; 63(5): 471- 6.

14. Schnurr PP, Friedman MJ, Oxman TE, et al: RESPECT-PTSD: re-engineering systems for the primary care treatment for PTSD, a randomized clinical trial. J Gen Intern Med 2013; 28(1): 32- 40.

15. Kussman MJ: VHA Handbook 1160.01: Uniform Mental Health Services in VA Medical Centers and Clinics. Washington, DC, Veterans Health Administration, 2008. Available at http://www1.va.gov/vhapublications/ ViewPublication.asp?pubID=1762; accessed March 3, 2010.

16. Wennberg J, Gittelsohn A: Small area variations in health care delivery: a population-based health information system can guide planning and regulatory decision-making. Science 1973; 182: 1102- 8.

17. Benneyan JC, Musdal H, Ceyhan ME, Shiner B, Watts BV: Specialty care single and multi-period service location-allocation models within the Veterans Health Administration. Socioecon Plann Sci 2012; 46(2): 136-48.

18. Watts BV, Shiner B, Ceyhan ME, Musdal H, Sinangil S, Benneyan J: Health systems engineering as an improvement strategy: a case exam- ple using location-allocation modeling. J Healthc Qual 2013; 35(3): 35-40.

19. Leira EC, Fairchild G, Segre AM, Rushton G, Froehler MT, Polgren PM: Primary stroke centers should be located using maximal coverage models for optimal access. Stroke 2012; 43(9): 2417- 22.

20. Watts BV, Schnurr PP, Zayed MH, Sinnott PL: Effects of a patient decision aid for veterans with post-traumatic stress disorder. 2012 VA Health Services Research and Development Annual Meeting. Avail- able at http://www.hsrd.research.va.gov/meetings/2012/abstracts.cfm; accessed November 28, 2012.

21. VA/DoD Clinical Practice Guideline for the Management of Post- Traumatic Stress. Available at http://www.healthquality.va.gov/ptsd/ cpg_PTSD-FULL-201011612.pdf; accessed February 4, 2013.

22. VHA: Uniform Mental Health Services in VA Medical Centers and Clinics. Washington, DC, Veterans Health Administration, 2008. Avail- able at http://www.mirecc.va.gov/VISN16/docs/UMHS_Handbook_1160 .pdf; accessed February 4, 2013.

23. Litz BT, Engel CC, Bryant RA, Papa A: A randomized, controlled proof-of-concept trial of an Internet-based, therapist-assisted self- management treatment for posttraumatic stress disorder. Am J Psychiatry 2007; 164(11): 1676-83.

24. Frueh BC, Monnier J, Yim E, Grubaugh AL, Hamner MB, Knapp RG: A randomized trial of telepsychiatry for post-traumatic stress disorder. J Telemed Telecare 2007; 13(3): 142-7.

25. Spence J, Titov N, Dear BF, et al: Randomized controlled trial of Internet-delivered cognitive behavioral therapy for posttraumatic stress disorder. Depress Anxiety 2011: 28(7): 541-50.

26. Cloitre M, Courtois CA, Ford JD, et al: The ISTSS Expert Consensus Treatment Guidelines for Complex PTSD in Adults. Deerfield, IL, International Society for Traumatic Stress Studies, 2012. Available at http://www.istss.org/AM/Template.cfm?Section=ISTSS_Complex_ PTSD_Treatment_Guidelines&Template=/CM/ContentDisplay.cfm& ContentID =5185; accessed November 30, 2012.

27. Sorensen JL: From Cat's Cradle to Beat the Reaper: getting evidence- based treatments into practice in spite of ourselves. Addict Behav 2011; 36(6): 597- 600.

28. Miller WR, Sorensen JL, Selzer JL, Brigham GS: Disseminating evidence- based practices in substance abuse treatment: a review with suggestions. J Subst Abuse Treat 2006; 31(1): 25-39.

Hande Musdal, MSE*; Brian Shiner, MD, MPH[dagger][double dagger]§?; TeChieh Chen, BA[dagger]; Mehmet E. Ceyhan, PhD;

Bradley V. Watts, MD, MPH[dagger][double dagger]§?; James Benneyan, PhD*[dagger]

*Healthcare Systems Engineering Institute, Northeastern University, 334 Snell Engineering Center, Boston, MA 02115.

[dagger]New England Veterans Engineering Resource Center, WRJ VAMC, 215 North Main Street, White River Junction, VT 05009.

[double dagger]White River Junction VA Medical Center, WRJ VAMC, 215 North Main Street, White River Junction, VT 05009.

§VA National Center for Patient Safety Field Office, WRJ VAMC, 215 North Main Street, White River Junction, VT 05009.

||Geisel School of Medicine at Dartmouth, 1 Rope Ferry Road, Hanover, NH 03755.

Lahey Clinic, Operations Management, 41 Mall Road, Burlington, MA 01805.

doi: 10.7205/MILMED-D-13-00177