Correlates of Utilization of PTSD Specialty Treatment Among Recently Diagnosed Veterans at the VA

Veterans presenting for care at U.S. Department of Veterans Affairs (VA) facilities have a higher prevalence of posttraumatic stress disorder (PTSD) than is found in general community samples. The lifetime prevalence of PTSD in the community is about 7% (1). A reanalysis of data from the National Vietnam Veterans' Readjustment Study estimated that the lifetime prevalence of PTSD for Vietnam veterans was 19% (2).

Recent statistics suggest that the wars in Iraq and Afghanistan are also associated with an increase in prevalence of PTSD among veterans. Before the current Iraq conflict, PTSD prevalence among all veterans seen in primary care clinics at four VA medical centers was 12% (3). Among veterans of Operation Enduring Freedom (OEF), in Afghanistan, or Operation Iraqi Freedom (OIF), in Iraq, who began receiving VA care between April 1, 2002, and March 31, 2008, 36% received new mental health diagnoses. Of these, 58% received new PTSD diagnoses (4).

Most veterans who are newly diagnosed as having PTSD receive some mental health care, but often they do not complete a comprehensive course of treatment. The VA specifically supports the use of two evidence-based PTSD psychotherapies, cognitive processing therapy and prolonged exposure therapy. Typically, these consist of nine or more treatment sessions (4–7). Among all veterans with recent PTSD diagnoses from VA providers, about two-thirds received any VA mental health care in the six months after diagnosis, and about half of those who received any mental health care received either a four months' supply of medication or at least eight sessions of counseling (5). Among a national sample of OEF-OIF veterans, 80% of those with new PTSD diagnoses had at least one VA mental health follow-up visit in the following year, but only 27% attended nine or more mental health visits of any type within the year (4). Predictors of having had nine or more mental health visits within 15 weeks of receiving a new PTSD diagnosis included having received the diagnosis in a mental health clinic, having had any comorbid mental health diagnosis, female gender, older age, living closer to the clinic, and having been seen primarily at a VA community clinic (4).

Little is known about the extent to which veterans recently diagnosed as having PTSD receive specialty PTSD treatment, and one previous study of characteristics of patients (N=87) associated with receipt of PTSD specialty care did not find any significant predictors of the number of visits (8). However, specialized PTSD clinics may be more likely than general mental health clinics to provide efficacious and adequate treatment for PTSD because of the providers' specialized training and expertise. Receiving a PTSD diagnosis in a PTSD clinic, compared with in a general mental health or general medical clinic, has been shown to predict receiving an adequate course of treatment (5). Among VA mental health treatment settings, a follow-up visit to a PTSD clinic is much more likely than a visit to a general mental health clinic or primary care clinic to include psychotherapy (5).

Andersen's (9) behavioral model of service utilization characterizes predictors of health services utilization as predisposing, enabling, and need factors. It has been used in mental health services utilization research that concerns veterans with PTSD (10), veterans filing VA disability claims for PTSD (11), and veterans with new mental health diagnoses (4). The primary objectives of our study were to describe PTSD specialty treatment received and, using Andersen's model, to identify predictors of receiving minimally adequate specialty treatment, defined as nine or more visits in 12 months, among veterans newly diagnosed as having PTSD who attended at least one visit to a specialized PTSD clinic. A secondary objective was to examine differences in and predictors of receipt of minimally adequate specialty treatment among OEF-OIF veterans and non-OEF-OIF veterans.

Methods

Data collection

Data related to demographic characteristics, diagnosis, screening, and visits were collected from the records of the VA Northwest Health Network, Veterans Integrated Service Network (VISN) 20 (12). The information was located in the VISN 20 Data Warehouse, a collection of databases extracted from the electronic patient medical records at each regional facility and updated monthly. OEF-OIF status was determined by linking the VISN 20 Data Warehouse to the national VA OEF-OIF roster, which uses files from the Department of Defense Manpower Data Center and is considered the gold standard for identifying veterans of the OEF-OIF conflicts. The local VA institutional review board approved the study and granted a waiver of informed consent. All data were deidentified before analysis.

Setting and participants

The study included the four sites in the Pacific Northwest region of the VA health care system, or VISN 20, with specialized PTSD clinics (Portland, Puget Sound, Spokane, and Boise).

We included all veterans at these sites who had a positive screen result on the Primary Care PTSD (PC-PTSD) screen between November 7, 2006, and September 30, 2008, excluding any who died during the year after the screen. These dates were chosen because they represented the most sustained period in the recent past during which reliable data for PC-PTSD screening results were available. Data on PTSD clinic visits were collected for one year from the screen date.

The PC-PTSD, a four-question screen assessing the presence of PTSD symptoms, is routinely used across the VA system for clinical care and has been validated in the VA primary care patient population (13). During the study period, a positive PC-PTSD screen was defined as three “yes” responses. Among male and female primary care patients combined, a cutoff of 3 has a sensitivity of .78, a specificity of .87, a positive predictive value of .65, and a negative predictive value of .92 (13) when using the Clinician-Administered PTSD Scale (CAPS) as the reference standard (14).

Because we wished to focus on veterans newly screened and diagnosed as having PTSD, we excluded veterans who had any mental health visits, PTSD clinic visits, or a previous diagnosis of PTSD, other mental illness, or substance use disorder (ICD-9 codes 295–316.9) in the five years before the date of the index positive screen. [A chart explaining participants' inclusion is available in an online appendix to this report at ps.psychiatryonline.org.] After these exclusions, 4,720 of the 20,414 veterans who screened positive for PTSD remained, and of those, 2,032 received a new PTSD diagnosis in the year following the positive screen. A total of 869 of these veterans attended at least one PTSD clinic visit and were included in our descriptive analysis.

Data about race were missing for 21% of veterans (N=185), but they were classified as veterans of unknown race and retained in the regression analysis. Seventeen veterans with missing data on marital status were excluded from the regression analysis. The sample for the regression analysis included 852 veterans, of whom 395 had served in either OEF or OIF and 457 had not served in either operation.

Measures

Outcome variables.

The primary outcome variable, designed to approximate receipt of minimally adequate specialty treatment in a PTSD clinic, was defined as having had nine or more PTSD clinic visits within 365 days of the index positive PTSD screen. A secondary outcome was the number of PTSD clinic visits attended within 365 days of the index positive PTSD screen. PTSD visits were defined by the VA clinic stop codes 519, 525, 540, 542, 561, and 580. Clinics with stop codes 571 or 572 were also included if they were PTSD specialty clinics. Clinic status was confirmed by contacting sites when necessary.

Predictor variables.

Independent variables were selected to be consistent with Anderson's behavioral model (9). Predisposing variables included demographic characteristics including age, marital status, gender, and race. Age was divided into quartiles in order to achieve homogeneity of variance. Enabling variables included residence in a rural or urban area and whether the first PTSD clinic visit occurred within 30 days of the index positive screen. Need variables included major comorbidities associated with PTSD, namely depressive disorders, alcohol abuse or dependence, traumatic brain injury, and anxiety disorders. An aggregate measure of medical comorbidity, the Charlson Comorbidity Index (CCI) score, was also included (15).

Rural versus urban residence was determined by matching zip codes with Rural-Urban Commuting Area Codes, which utilize the standard U.S. Census Bureau Urbanized Area and Urban Cluster definitions in combination with work-commuting information to characterize all U.S census tracts as rural or urban (16,17). An ICD-9-CM code for depression, alcohol use disorders, traumatic brain injury, or anxiety disorders associated with any VA visit during the year-long follow-up period was considered an indication of the presence of a comorbid diagnosis. The CCI score (15) uses ICD-9-CM codes for 17 health conditions to assess medical comorbidity (18,19); possible scores range from 0 to 37, with higher scores representing more severe illness. ICD-9-CM inpatient and outpatient diagnoses during the follow-up period were used to calculate CCI scores, which were categorized as 0, 1, or ≥2, because most scores were ≤2.

Analysis

Statistical analyses were conducted using SPSS statistical software, version 17.0. We used a logistic regression model for the dichotomous outcome of the presence of minimally adequate specialty treatment and a negative binomial regression model for the number of clinic visits (20). Generalized estimating equations were used to account for intrasite correlation. The logistic regression analyses were stratified by OEF-OIF status. Independent variables included the predisposing, enabling, and need variables described above and were entered simultaneously into each model.

Results

Sociodemographic and clinical characteristics

Compared with non-OEF-OIF veterans, OEF-OIF veterans were significantly younger; were less likely to be married, divorced, separated, or widowed; were more likely to reside in an urban area; were more likely to have attended a PTSD clinic visit within 30 days of the index positive PTSD screen; had lower CCI scores, indicating less overall medical comorbidity; and were more likely to be diagnosed as having alcohol use disorders or traumatic brain injury (Table 1).

PTSD treatment utilization

Among the 869 veterans who attended at least one PTSD clinic visit, the mean±SD number of PTSD clinic visits was 9.1±12.6 (range 1–124). A total of 286 veterans (33%) completed at least nine visits, our standard for minimally adequate specialty treatment; 570 veterans (66%) completed at least three visits, and 91 of the 286 veterans who completed at least nine visits completed more than 20 visits (11% of veterans who completed at least one visit) (Figure 1).

Significantly fewer OEF-OIF veterans completed minimally adequate specialty treatment, compared with non-OEF-OIF veterans (29% versus 36%; p=.021, two-sided chi square test). OEF-OIF veterans also attended significantly fewer visits than non-OEF-OIF veterans (8.2±11.4 versus 9.9±13.5, p=.045, two-sided t test). A total of 252 (62%) OEF-OIF veterans and 318 (69%) non-OEF-OIF veterans completed three clinic visits; 39 (10%) OEF-OIF veterans and 52 (11%) non-OEF-OIF veterans completed more than 20 visits.

Predictors of PTSD treatment utilization

In regression models, among OEF-OIF veterans, significant predictors of utilization of minimally adequate specialty treatment were race other than white, urban residence, an initial PTSD visit within 30 days of the positive PTSD screen, CCI score of 1 (versus scores of 1 or ≥2), and comorbid depressive disorder, alcohol abuse or dependence, traumatic brain injury, or other anxiety disorder (Table 2). Predictors significantly associated with the number of visits included older age (35 years and older), urban residence, an initial PTSD visit within 30 days of PTSD screen, and comorbid depressive disorder, alcohol abuse or dependence, traumatic brain injury, or other anxiety disorder.

In the sample of veterans who did not serve in OEF or OIF, significant predictors of utilization of minimally adequate specialty treatment were race other than white, being married (versus having never married), urban residence, an initial PTSD visit within 30 days of PTSD screen, and comorbid depressive disorder or anxiety disorder (Table 3). Predictors positively associated with the number of PTSD clinic visits included female gender, race other than white, older age (59 years and older), being married (versus having never married or being divorced, separated, or widowed), urban residence, an initial PTSD visit within 30 days of PTSD screen, CCI score≥2, and comorbid depressive disorder or anxiety disorder.

Discussion

To our knowledge, this study is the first to examine correlates of minimally adequate specialty treatment in VA PTSD clinics and to describe PTSD specialty treatment utilization by veterans newly diagnosed as having PTSD after a positive screen. Only about one-third of veterans who began PTSD specialty treatment received minimally adequate specialty treatment. Several factors were associated with receipt of minimally adequate specialty treatment, including attending initial PTSD clinic visits within 30 days of positive screens, living in urban locations, and having psychiatric comorbidities.

We believe that although the data do not allow us to differentiate between PTSD clinic visits for psychotherapy or medication management, they may provide an initial estimate of receipt of specialized psychotherapy for PTSD. The proportion of veterans who received minimally adequate specialty treatment (33%) was comparable to the finding by Spoont and others (5) that 28% of veterans who received PTSD diagnoses in PTSD clinics attended at least eight counseling sessions. That study, which also excluded veterans with any recent history of VA mental health care, found that about 69% of follow-up visits in PTSD clinics involved psychotherapy (5). The rate of receipt of minimally adequate specialty treatment found in this study is also comparable to the rate of receipt of adequate psychotherapy for depression and anxiety among patients in community samples (21–23). Further study is necessary to determine what proportion of patients, among those receiving minimally adequate specialty treatment, utilize evidence-based psychotherapies.

The clinical significance of decreased receipt of minimally adequate specialty treatment and attendance of fewer visits by OEF-OIF veterans is unclear, although several factors may contribute. For example, Vietnam veterans, who are older, may respond differently to treatment than OEF-OIF veterans (24). An older cohort might be more likely to persist in treatment because of their age, time since trauma, PTSD chronicity, or other factors (24). Despite attending fewer visits and being less likely to complete minimally adequate specialty treatment, OEF-OIF veterans were more likely to receive an initial PTSD visit within 30 days of a positive screen. This finding may suggest decreased access to services for Vietnam veterans because of limited treatment resources. Alternatively, older veterans with more chronic PTSD who did not serve in OEF or OIF may feel less urgency to initiate treatment but have more motivation to continue receiving treatment after initial engagement, compared with the OEF-OIF veterans.

Our findings confirm the usefulness of the behavioral model in considering predisposing, enabling, and need-related factors that may influence PTSD specialty treatment utilization. Regarding predisposing variables, marriage was predictive of receiving minimally adequate specialty treatment among veterans who did not serve in OEF or OIF, possibly because it confers increased social support for getting treatment or because veterans who are able to maintain marital relationships are also more likely to continue in treatment relationships.

Our findings associating race other than white with receipt of minimally adequate specialty treatment are consistent with some (25) but not all (26) previous studies of veterans. Psychotherapy for PTSD may be more accessible to nonwhite patients in the VA system than it is in the community, where white patients are more likely to seek PTSD treatment (27). Enabling variables, including geography and timing of the first PTSD visit, are considered the most amenable to interventions to change health service distribution (9). Living further from the VA has previously been found to correspond with decreased receipt of adequate mental health treatment for PTSD (4). Our findings regarding the need variable of psychiatric comorbidities, particularly depression and anxiety disorders, support previous findings that psychiatric comorbidities predict both PTSD treatment initiation and receipt of adequate mental health treatment for PTSD (4,10,28).

The study's limitations, including two characteristics of the sample, should be noted in interpreting our findings. We included only veterans new to VA mental health care (5), a decision that may have caused us to observe lower rates of receipt of minimally adequate specialty treatment. This sample characteristic may preclude generalizability to all veterans receiving care at the VA but also provides a valuable perspective on individuals in the early stages of treatment. We included only veterans who received at least one PTSD clinic visit, in order to minimize variance due to unmeasured factors related to VA providers, facilities, and systems that have an impact on initial access to PTSD clinic treatment. Such structural components of resource organization are the most difficult to define and to relate to utilization patterns (10). However, limiting the sample to those who attended PTSD specialty clinic visits also may have selected for veterans who were motivated and who perhaps had more severe PTSD than veterans who were not referred for PTSD specialty care.

We sought to measure PTSD treatment plausibly related to the index screen; thus a veteran who began treatment but did not complete nine visits in the 12 months following the screen was not considered to have received minimally adequate specialty treatment, even if he or she ultimately completed treatment. It is not known whether delays in treatment receipt affected clinical outcomes. In post hoc analyses, extending the follow-up period to 365 days after the first PTSD clinic visit slightly increased the rate of completion of minimally adequate specialty treatment and of number of visits in both cohorts; it also slightly widened the gap between veterans who had or had not participated in OEF or OIF in the number of visits received and in the proportion receiving minimally adequate specialty treatment.

We did not include a measure of pharmacotherapy, because medication for PTSD is prescribed by a variety of VA providers. About half of all veterans in our sample received at least 120 days' supply of antidepressant medication (5). Combining this measure of pharmacotherapy with minimally adequate specialty treatment would yield an overall rate of adequate treatment comparable to or greater than the rates that have been found among U.S. patients with depression (29) or various psychiatric diagnoses (30).

Several other limitations apply to interpretation of our data. We could not confirm the accuracy of PTSD diagnoses made by clinicians. We could not assess the impact of VA disability ratings for PTSD on treatment completion rates because the data did not contain information on the temporal relationship between the assignment of disability ratings and treatment attendance. Any implications about our findings related to race are limited by the significant proportion of missing data (Table 1).

Further research would be needed to evaluate whether rural veterans are receiving less PTSD treatment, different PTSD treatment, or both; VA facilities may use contracts or referrals to rural community providers for veterans who cannot travel to other VA facilities for PTSD specialty care (31). The effect of receiving the first PTSD clinic visit within 30 days of a positive screen result may reflect patient motivation, treatment access factors, and PTSD symptom severity. We cannot interpret the impact of having psychiatric comorbidities as causal because of the possible confound between treatment utilization and receiving additional diagnoses.

Conclusions

Psychiatric comorbidities, rural versus urban residence, and early access to specialty treatment may play important roles in the early receipt of comprehensive specialty care for PTSD. Although the majority of veterans who began PTSD specialty treatment did not receive minimally adequate specialty treatment, perhaps not all veterans with PTSD need the same amount of treatment. Further research could examine the relationship between PTSD severity and treatment seeking, veterans' subjective experiences of treatment, and specific interventions received. Related clinical interventions could focus on evaluation of treatment readiness or need and on enhancement of motivation to complete treatment.

Figures and Tables