Changes in Psychotherapy Utilization Among Veterans With Depression, Anxiety, and PTSD

This study focused on patients who received a new diagnosis of depression, anxiety, or PTSD in VHA outpatient facilities during the specified years. Patients were categorized, using ICD-9-CM codes, into three diagnostic groups: depression (codes 293.83, 296.20–296.36, 300.4, and 311), anxiety (codes 293.84, 293.89, 300.00–300.02, 300.09, 300.20–300.23, 300.29, and 300.3), and PTSD (codes 308 and 309.81). These diagnostic categories were selected because they are highly prevalent and disabling within the VHA patient population (26), and substantial evidence supports the use of psychotherapy as a primary or secondary treatment for these diagnoses (27). PTSD was examined separately because of its high prevalence and impact in the veteran population and because of the focus on enhancing access to PTSD-specific treatments within the VHA during the study time frame (including nationwide rollouts of two evidence-based psychotherapies).

As detailed in previous work (15), restricting the sample to patients with new-onset diagnoses allowed the assessment of time between diagnosis and receipt of psychotherapy and provided a fair test of whether those newly identified by the system received indicated care. A new-onset condition was defined as six months without a related diagnosis before the index date (the date of first diagnosis during the study period). To avoid overlapping ICD-9-CM coding, conditions occurring within diagnostic categories (such as major depression and dysthymia) were considered one condition. Patients with multiple new-onset conditions were classified into more than one diagnostic category (for example, patients with new-onset depression and new-onset anxiety were classified into both categories). Frequency statistics reported for the full fiscal year (FY) cohort (with diagnostic groups combined) are reported for unique patients. For diagnostic subgroup analyses, patients who were classified into multiple categories were represented in each of these diagnostic groups. We restricted the sample to patients with adequate opportunities to use outpatient psychotherapy services by excluding those with 60 or more inpatient days in the 180 days after diagnosis.

Patient use of mental health services was assessed during the 12 months after each patient’s index date. Psychotherapy use was determined on the basis of Current Procedural Terminology (CPT) codes and classified by treatment modality into individual psychotherapy (CPT codes 90804, 90806, 90808, 90810, 90812, 90814, 90845, 90875, 90876, and 96152) and group therapy (90849, 90853, 90857, and 96153). Psychotherapy exposure, defined as the number of patients receiving at least one session (any exposure to psychotherapy), four sessions (approximate length of briefer forms of psychotherapy [28–30]) or eight or more sessions (the minimum length of traditional, evidence-based psychotherapies for depressive and anxiety disorders [19,31]) was also assessed. Timeliness of psychotherapy was assessed by calculating the number of days between the patient’s index date and first psychotherapy visit.

We examined patients’ sociodemographic characteristics, including age, gender, and marital status. Patients’ residential zip codes were used to estimate income (using the average adjusted gross income for each zip code, based on 2008 Internal Revenue Service data) and distance in miles to the nearest VHA facility. Patient race was populated inconsistently in the NPCD and was therefore not reported. Because of the increased access to care provided to veterans with service-related disabilities, patients were categorized into three groups by service-related disability percentage, including 0%, 1%−49%, and ≥50%. Illness burden was assessed with a diagnosis-based, risk-adjustment methodology (32,33), which provides an odds ratio representing the predicted health care cost of each veteran compared with the average predicted cost (that is, the relative risk score).

All analyses were descriptive. The proportion of patients receiving at least one, four, or eight psychotherapy sessions during the 12 months after their index date and the delay between diagnosis and psychotherapy were calculated for unique patients in each FY cohort and for specific diagnostic categories within each cohort. Subsequent analyses described treatment modality patterns (individual versus group psychotherapy) across the study time points. Effect size was calculated with odds ratios. Analyses were conducted with SAS version 9.2.

Table 1 lists the demographic and clinical characteristics of the cohorts from FY 2004 (N=424,428), FY 2007 (N=494,318), and FY 2010 (N=583,733); the steady growth in cohort size represents increases in the number of patients who received new-onset diagnoses of depression, anxiety, or PTSD. A minority of the sample met criteria for two diagnostic categories (FY 2004, 7.7%; FY 2007, 9.3%; and FY 2010, 10.9%). Notable changes in patient characteristics between FY 2004 and FY 2010 included a decrease in age, increased numbers of female veterans, greater medical illness burden, increased VA service-connected disability, and increases in co-occurring substance use diagnoses.

The raw number of patients receiving any psychotherapy grew by 72% between FY 2004 (N=90,095) and FY 2010 (N=155,191) (Figure 1). The effect size, as measured by odds ratio (OR), was 1.34, which had a 95% confidence interval (CI) of 1.33–1.36. Examination of psychotherapy use from FY 2007 to FY 2010 (OR=1.28, CI=1.27–1.29) indicates greater increases in psychotherapy use during this period relative to between FYs 2004 and 2007 (OR=1.05, CI=1.04–1.06).

Figure 1 presents percentages of patients who received at least one, four, or eight psychotherapy sessions. The number and proportion of patients who received each level of exposure increased across study time points. The most notable increase was in the number of patients who received at least one psychotherapy session, which grew by more than 65,000 between FYs 2004 and 2010. The number of newly diagnosed patients who received at least four sessions nearly doubled between FY 2004 (N=32,581) and FY 2010 (N=64,459), as did the number who received eight or more psychotherapy sessions (FY 2004, N=17,249; FY 2010, N=33,938). Across all levels of exposure, modest increases in psychotherapy use were observed between FY 2004 and FY 2007, with more substantial increases occurring between FY 2007 and FY 2010.

Examination of psychotherapy exposure according to diagnostic group revealed unique trends for patients with PTSD (Table 2). Across all FY cohorts, PTSD patients demonstrated the highest rate of psychotherapy initiation and number of psychotherapy sessions. Although the proportion of PTSD patients who received psychotherapy remained relatively consistent from FY 2004 to FY 2010 (OR=.95, CI=.93–.97), rates of psychotherapy use increased among patients with depression (OR=1.40, CI=1.38–1.41) and among patients with anxiety (OR=1.59, CI=1.56–1.62), reducing disparities by diagnostic group in receipt of psychotherapy.

Descriptive analyses also revealed differences in utilization rates for group versus individual psychotherapy (Table 3). For each cohort, patients were at least three times more likely to initiate individual psychotherapy than group psychotherapy. The proportion of each cohort that received at least one session of group psychotherapy consistently ranged from 6% to 7%. The proportion of patients receiving individual psychotherapy showed little change between FY 2004 (18%) and FY 2007 (19%) (OR=1.05, CI=1.04–1.06), with a more substantial increase between FY 2007 and FY 2010 (26%) (OR=1.33, CI=1.33–1.34). Data from all three cohorts revealed that group psychotherapy patients were more likely to receive more psychotherapy sessions. Across the full study sample, 34% of those who initiated group therapy received eight or more sessions (FY 2004, 35%; FY 2007, 32%; and FY 2010, 34%), compared with 12% of those who initiated individual psychotherapy (FY 2004, 10%; FY 2007, 11%; and FY 2010, 14%).

Treatment delays decreased across study time points. The median time between index diagnosis and psychotherapy dropped from 56 days (FY 2004) to 47 days (FY 2010), with the majority of change occurring from FY 2007 to FY 2010 (Table 4). Although patients with PTSD consistently had shorter delays than patients with anxiety or depression, diagnostic disparities in time until treatment grew smaller across the study time points.

Improvements in the reach, amount, and timeliness of psychotherapy occurred during a period of dramatic increases in patient load, which likely reflects the success of several recent programmatic efforts within the VHA. Required screenings and evidence-based guidelines for assessment of mental disorders have allowed increased detection of psychiatric problems, a critical step in access to care.

Among the diagnostic categories examined in this study, rates of PTSD grew most substantially, increasing by more than twofold between 2004 and 2010. This is likely because of the VHA’s particular focus on identification and treatment of PTSD, given its substantial psychological and economic burden among veterans and the clear link between military trauma exposure and subsequent development of PTSD. Specific policy changes that may have contributed to increases in PTSD diagnosis include mandatory screening for PTSD symptoms and military sexual trauma in primary care (40). Veterans with PTSD also had the shortest delays between diagnosis and psychotherapy and were more likely to initiate psychotherapy and receive adequate exposure than those in other diagnostic groups.

Although the VHA provided psychotherapy to many more veterans with new diagnoses of PTSD in 2010 than 2004, a majority continued to receive no exposure to VHA psychotherapy, despite numerous, ongoing PTSD-specific initiatives (such as evidence-based psychotherapy rollouts). There is a great need within the VHA for additional efforts to engage veterans in psychotherapy services.

Our study period also corresponded to a period of increased funding for VHA mental health services, expansion of mental health staff, evidence-based psychotherapy training initiatives, and the integration of psychotherapy into primary care. More substantial increases in psychotherapy reach, amount, and timeliness occurred in 2007−2010 relative to 2004–2007, which aligns with the implementation of several of these initiatives. Other recent policy changes that may have contributed to increases in both the number of newly diagnosed patients and the use of psychotherapy include cost-free VHA health care for Iraq-Afghanistan combat veterans for five years postdischarge and expanded efforts to engage rural veterans through community-based outpatient clinics and telepsychotherapy services (41).

This study used CPT codes to assess mental health service use. Although widely used in studies examining psychotherapy utilization, they are associated with several limitations: CPT codes do not capture the type of psychotherapy provided, and, although entered by the provider (and therefore reflective of the provider’s perspective of the care provided), the fidelity of these codes is not independently verified. The validity of provider-recorded diagnoses is similarly unknown. We defined psychotherapy amount as the number of sessions received and did not extract codes that reflect the length of each session, because these codes often represent ranges of time rather than a precise duration. We also did not examine use of non-VHA psychotherapy, and, given that many veterans are dual users of VHA and non-VHA services (42), results likely underestimated the true rates of psychotherapy exposure within the study sample. Despite this limitation, these data are valuable in that they present a descriptive account of trends in VHA psychotherapy and reflect improvements in care that have likely resulted from efforts to transform VHA mental health services. Finally, this study did not examine medication use; evaluation of psychopharmacology for these diagnostic groups over this period would provide an even more comprehensive picture of how veterans’ mental health needs are being addressed.