Abstract
Background:
Patients with vs. without posttraumatic stress disorder (PTSD) are more likely to have poor antidepressant medication (ADM) adherence but it is unclear if improved PTSD is associated with ADM adherence. We determined if clinically meaningful PTSD symptom reduction was associated with ADM adherence.
Methods:
Electronic health record data (2008-2015) was obtained from 742 Veterans Health Affairs (VHA) patients using PTSD specialty clinics with a PTSD diagnosis and PTSD checklist (PCL) score ≥50. The last PCL in the exposure year after the first PCL≥50 was used to identify patients with a clinically meaningful PCL decrease (≥20 point) versus those without (< 20 point). Patients had a depression diagnosis in the 12-months before the exposure year and received an ADM in the exposure year. Proportion of days covered ≥80% in exposure year defined adherence. Confounding was controlled using propensity scores and inverse probability of treatment weighting.
Results:
Patients were 42.2±13.1 years of age, 63.9% white and 18.9% had a clinically meaningful PCL decrease. After controlling for confounding variables, patients with vs. without a clinically meaningful PCL decrease were significantly more likely to be adherent (OR= 1.78; 95% CI: 1.16-2.73). However, adherence remained low in both patients with and without meaningful PCL decrease (53.5% vs. 39.3%).
Limitations.
The sample was limited to VHA patients. Patients may not have taken medication as prescribed.
Conclusions:
Large reductions in PTSD symptoms are associated with ADM adherence. Prior literature suggests ADM adherence improves depression symptoms. Thus, PTSD symptom reduction may lead to better depression outcomes.
Keywords: PTSD symptoms, depression, antidepressant medication adherence
INTRODUCTION
Patients with vs. without posttraumatic stress disorder (PTSD) are significantly more likely to report not taking medications as prescribed, forgetting to take medicines and skipping doses (Kronish et al., 2012). Patients with anxiety disorders are less adherent to antidepressant medications (ADM) (Stein et al., 2006) and a majority of patients with PTSD are non-adherent to ADMs in the 12-months following residential treatment (Lockwood et al., 2009). Yet, the evidence is inconsistent. A large retrospective cohort study of patients with depression discharged on ADMs found those with vs. without comorbid PTSD were more often ADM adherent (Zivin et al., 2009).
While the majority of existing studies support the conclusion that PTSD is associated with poor ADM adherence, to our knowledge, there is no literature on whether large PTSD improvement is associated with greater likelihood of ADM adherence. Several studies have found ADM adherence is a strong predictor of depression improvement (Akerblad et al., 2006; Ho et al., 2016; Stein et al., 2006). Among patients beginning ADM therapy, response to treatment by 24 weeks is markedly higher in adherent vs. non-adherent patients (82.5% vs. 55.8%) (Akerblad et al., 2006). If patients who have a clinically meaningful PTSD symptom decrease are more likely to be ADM adherent compared to patients with no or less than clinically meaningful improvement, the success of pharmacotherapy for depression may be enhanced by first targeting reductions in PTSD severity. Approximately 50% of people with PTSD also have a diagnosis of major depressive disorder (Flory and Yehuda, 2015), and PTSD and depressive symptoms change similarly over time such that comorbid depression improves in tandem with PTSD (Norman et al., 2011). The current study used a retrospective cohort of Veterans Health Affairs (VHA) patients with PTSD to determine if patients who experience a ≥20 point PTSD Checklist (PCL) score decrease compared to <20 point PCL score decrease are more likely to be ADM adherent.
METHODS
A retrospective cohort study was conducted with electronic medical record data from the VHA, which included type of clinic encounter, ICD-9-CM diagnostic codes, prescription fills, laboratory results, vital signs and demographic measures. Data included patients with encounters at a PTSD specialty health clinic at one of five VHA medical centers across the United States between fiscal years (FY) 2008 and 2012, with follow-up continuing through FY2015. The study procedures were reviewed and approved by the Saint Louis University and Harry S. Truman Memorial Veterans’ Hospital institutional review boards.
A random selection of 5,916 patients from a total of 17,476 patients 18-70 years of age, with ≥ 2 visits to PTSD specialty care in FY2008-2012, and a diagnosis of PTSD were selected. Demographic comparisons indicated that average age and distributions of gender, marital status, race, and insurance status were similar in the random and overall sample. A PTSD diagnosis was present if there were at least two outpatient visits in a 12-month period or one inpatient stay with ICD-9-CM code 309.81. This algorithm has good positive predictive value compared to a gold standard PTSD Checklist (PCL) score ≥50 (Gravely et al., 2011), and good agreement with lifetime diagnosis per the Structured Clinical Interview for DSM-IV (Holowka et al., 2014).
Patients were then required to have a PCL ≥ 50, above the threshold for probable PTSD (Monson et al., 2008; National Center for PTSD, 2012), in FY2009 -2014. This allowed for an ‘exposure year’ to measure PCL change after the first PCL ≥ 50, and a year prior to measure prevalent depression. Prevalent depression (ICD-9-CM: 296.2x, 296.3x, 311) was defined as a single inpatient stay or at least two outpatient codes in a 12-month period where the second outpatient code must have occurred in the year prior to PCL ≥ 50. This algorithm has excellent agreement with patient reported depression and manual chart abstraction (Frayne et al., 2010;Solberg et al., 2006). Patients were further required to have a PCL at least 8 weeks after in the ‘exposure year’ and depression in the year prior to the first PCL ≥ 50. Finally, patients must have had at least one fill for an antidepressant medication (ADM) in the ‘exposure year’, leaving a final analytic sample of 742. See e-Figure 1 for sample selection.
Variable Definitions
Exposure: PCL Change
The PCL measures PTSD symptoms on a 17-item self-report measure based on DSM-IV criteria (Weathers et al., 1991). Participants rate how much they have been bothered by each symptom over the last month on a scale from 1=Not at all to 5=Extremely. Total symptom score ranges from 17 to 85. We did not distinguish whether scores were from the PCL-Military, PCL-Civilian, and PCL-Specific versions. The PCL has been shown to have good internal consistency, test-retest reliability, and convergent validity (Blanchard et al., 1996; Ruggiero et al., 2003).
PCL scores were obtained from administrative medical record data and supplemented with scores from manual chart abstraction because VHA administrative data do not capture all PCL scores in a discrete field. Abt Associates (www.abtassociates.com) conducted chart abstraction over a 6-month period on the 5,916 PTSD patients, obtaining 22,287 valid PCL scores from FY2008-2015. Administrative medical record scores were added and after removing duplicate scores, there were 4,441 patients with 26,631 PCL scores.
A meaningful reduction in PCL score was classified as a ≥20 point decrease versus a < 20 point decrease (including an increase) from the first PCL ≥ 50 to the last PCL in the ‘exposure year’. This ≥20 decrease is consistent with a large, clinically meaningful PTSD improvement (Monson et al., 2008). The average number of days between first PCL ≥ 50 and the last PCL in the exposure year was 217.9 (±96.8) and median days was 213.5 (IQR=126 to 308).
Outcome: ADM adherence
ADM’s included monoamine oxidase inhibitors (MAOIs), selective serotonin reuptake inhibitors (SSRIs), serotonin-norepinephrine reuptake inhibitors (SNRIs), tricyclics (TCAs), and non-classified ADMs (see e-table 1 for full list of drugs). ADM adherence was defined using proportion of days covered (PDC) from PCL ≥ 50 date to 12-months after (Cramer et al., 2008; Peterson et al., 2007). PDC is an adherence measure calculated as the number of days where a particular drug is available divided by the number of total days in a specified time frame. PDC ranges from 0 to 1 because unlike the medication possession ratio where the numerator is the sum of days supply for each fill in a time period, PDC does not double count covered days. Days in the exposure year when an ADM was available were identified by using fill start and end dates relative to the PCL ≥ 50 date. Overlapping days were counted as a single covered day. The PDC was the total number of ADM days covered divided by days in the exposure year. Standard thresholds were used to dichotomize PDC into adherent (≥80%) and non-adherent (< 80%) (Cramer et al., 2008; Peterson et al., 2007).
A secondary outcome indicating longest duration of continuous ADM use was created. All periods of continuous use in the exposure year were examined, where a period of continuous use stopped if there was a gap of > 30 days between fills or at the end of the exposure year.
Covariates
Detailed definitions for all variables are shown in e-table 1. Sociodemographic variables included age, race, gender, marital status and access to non-VHA health insurance. A ‘missing’ category was included for demographic variables to retain all cases. Age, gender, marital status, and access to non-VHA insurance were missing in 0% of cases while race was missing in 2.8% cases. Comorbid conditions were measured from one year prior to one year post PCL ≥ 50 and included anxiety disorders, sleep disorders, alcohol and drug abuse/dependence, smoking, and the Charlson-Romano comorbidity index (Romano et al., 1993). Adequate PTSD psychotherapy, defined as ≥ 9 visits in any 15-week period (Seal et al., 2010), was measured in the exposure year. Finally, an indicator of severe PTSD based on the first PCL was defined as ≥70 versus 50-69.
Propensity Score Methods
Propensity scores (PS) and inverse probability of exposure weighting (IPEW) were used to balance the distribution of potential confounders between patients with and without a clinically meaningful PCL decrease. Potential confounders are variables in Table 1, with the exception of mean first and last PCL scores which were presented for descriptive information. The PS is the probability of a PCL decrease ≥ 20 points, given covariates, that is calculated using a binary logistic regression model (Rosenbaum and Rubin, 1983). The PS is used to compute a stabilized weight (Austin and Stuart, 2015) by multiplying initial weights by the marginal probability of PCL decrease ≥20 or < 20 in the overall sample, in each respective exposure group. Stabilizing weights helps reduce bias associated with extreme weights due to increased variance and retains original sample size for analysis preserving Type I error rate. Stabilized weights should have a mean close to one and a maximum < 10, thus, weights ≥10 were trimmed (Sturmer et al., 2014). After applying IPEW, a pseudo-population is created where confounding is controlled for when variables balance between PCL decrease ≥ 20 versus < 20, indicated by a standardized mean difference (SMD%) < 10% (Austin and Stuart, 2015).
Table 1.
Sample characteristics overall and by PCL decrease for PTSD cases, veterans age 18-70 years with treated depression (n=742)
| Variable, n(%) or mean (±sd) | Overall (n=742) |
PCL dec < 20 (n=602) |
PCL dec ≥ 20 (n=140) |
p- value |
|---|---|---|---|---|
| Age (years), mean (±sd) | 42.2 (±13.1) | 41.6 (±13.1) | 44.6 (±12.6) | .015 |
| Male gender, n(%) | 597 (80.5) | 491 (81.6) | 106 (75.7) | .116 |
| Race, n(%) | ||||
| White | 474 (63.9) | 377 (62.6) | 97 (69.3) | |
| Black | 203 (27.4) | 171 (28.4) | 32 (22.9) | .174 |
| Other | 44 (5.9) | 39 (6.5) | 5 (3.6) | |
| Missing | 21 (2.8) | 15 (2.5) | 6 (4.3) | |
| Married, n(%) | 321 (43.3) | 266 (44.2) | 55 (39.3) | .292 |
| VHA only insurance, n(%) | 484 (65.2) | 398 (66.1) | 86 (61.4) | .294 |
| High primary HCU, n(%) | 186 (25.1) | 160 (26.6) | 26 (28.6) | .049 |
| First PCL severe (≥70), n(%) | 252 (34.0) | 198 (32.9) | 54 (38.6) | .301 |
| First PCL, mean (±sd) | 65.1 (±9.2) | 64.9 (±9.2) | 65.9 (±9.1) | .229 |
| Last PCL, mean (±sd) | 57.5 (±15.7) | 62.7 (±11.6) | 35.2 (±10.3) | <.0001 |
| Comorbidities and treatments a | ||||
| Other anxietyb, n(%) | 216 (29.1) | 167 (27.7) | 49 (35.0) | .089 |
| Sleep disorder, n(%) | 364 (49.1) | 297 (49.3) | 67 (47.9) | .753 |
| Alcohol abuse/dependence, n(%) | 323 (43.5) | 254 (42.2) | 69 (49.3) | .127 |
| Drug abuse/dependence, n(%) | 239 (32.2) | 185 (30.7) | 54 (38.6) | .074 |
| Smokingc, n(%) | 387 (52.2) | 306 (50.8) | 81 (57.9) | .134 |
| Adequate PTSD psychotherapy, n(%)d | 335 (45.2) | 251 (41.7) | 84 (60.0) | <.0001 |
| Comorbidity index, n(%) | ||||
| 0 | 461 (62.1) | 378 (62.8) | 83 (59.3) | |
| 1 – 2 | 167 (22.5) | 135 (22.4) | 32 (22.9) | .627 |
| ≥ 3 | 114 (15.4) | 89 (4.8) | 25 (17.9) |
PTSD=posttraumatic stress disorder; PCL=PTSD checklist (range: 17-85); FY=fiscal year; DEC=decrease; HCU=healthcare utilization
Comorbidities occur from one year prior to one year post first PCL ≥50
Composite of panic disorder, obsessive compulsive disorder, social phobia, generalized anxiety disorder, anxiety not otherwise specified.
Current smoker in health factors or ICD-9-CM code
Measured from first PCL≥50 to one year post. Presence of ≥9 psychotherapy visits in any 15 week period.
Analytic approach
All analyses were performed using SAS v9.4 (SAS institute, Cary, NC) at an alpha level of 0.05. Bivariate analyses in crude (unweighted) and weighted data assessed the relationship of PCL decrease ≥ 20 versus < 20 and potential confounders using chi-square tests for categorical variables and independent samples t-tests for continuous variables as well as SMD%. Unweighted and weighted binary logistic regression and linear regression models were used to test the association of meaningful PCL decrease with ADM adherence ≥ 80% and longest ADM duration, respectively. Robust, sandwich-type variance estimators were used to calculate confidence intervals and p-values in weighted models.
RESULTS
Overall, patients were 42.2 (±13.1) years old, 63.9% white, 43.3% married, and 18.9% had a clinically meaningful decrease in PCL score. Average first PCL was 65.1±9.2 and 34.0% had a severe initial score. Older age and high primary healthcare utilization were positively associated with PCL decrease ≥ 20 (p<0.05). Adequate PTSD psychotherapy was about 1.5 times more prevalent in the PCL decrease ≥ 20 versus < 20 groups (60.0% vs. 41.7%; p<0.0001). Average initial PCL was similar between groups but the last PCL in the exposure year was higher for PCL decrease < 20 (mean=62.7) than ≥ 20 (mean=35.2) (p<0.0001). No other significant differences were found. See Table 1.
Although a majority of confounders were not significantly related to PCL decrease in crude analyses, all variables with the exception of insurance, sleep disorder, and comorbidity index showed significant imbalance between groups (SMD%≥10%). IPEW balanced all confounders (SMD<10%) between groups. See e-Figure 2. Stabilized weights ranged from 0.31 to 3.87 with a mean=1.00 (±0.30) and median=0.95 (iqr=0.90-1.05). No weights were trimmed.
Descriptive outcome results and regression models are shown in Table 2. In crude analyses, 55.7% of PCL decrease ≥ 20 compared to 39.4% of PCL decrease < 20 were ADM adherent. Average longest ADM duration was 239.3 (±108.9) and 204.3 (±114.5) days in PCL decrease ≥20 versus < 20, respectively. In weighted models, there was a 78% increased likelihood of ADM adherence for PCL ≥20 versus < 20 decrease (OR=1.78; 95% CI: 1.16-2.73). Those with a clinically meaningful decrease versus not also had about 30 more days of longest continuous ADM use (B=30.22; 95% CI: 6.39, 54.04).
Table 2.
Crude and weighted by inverse probability of exposure comparisons of ADM treatment adherence and PCL decrease among veterans with PTSD age 18-70 years with PTSD and treated depression. Linear and logistic regression models (n=742)
| PDC ≥ 80%1 | Longest ADM duration (days)2 | ||||
|---|---|---|---|---|---|
| n | % | OR (95% CI) | mean (±sd) | B (95% CI) | |
| Crude | |||||
| PCL decrease < 20 | 237 | 39.4% | reference | 204.3 (±114.5) | reference |
| PCL decrease ≥ 20 | 78 | 55.7% | 1.94 (1.34-2.81) | 239.3 (±108.9) | 34.97 (14.13, 55.81) |
| Weighted | |||||
| PCL decrease < 20 | -- | 39.3% | reference | 204.5 (±114.7) | reference |
| PCL decrease ≥ 20 | -- | 53.5% | 1.78 (1.16-2.73) | 234.7 (±112.4) | 30.22 (6.39, 54.04) |
PCL=PTSD checklist; PTSD=posttraumatic stress disorder; ADM = antidepressant; OR=odds ratio; B=unstandardized regression coefficient; CI=confidence interval; PDC=proportion of days covered
Logistic regression
Linear regression
DISCUSSION
Patients treated in VHA PTSD specialty clinics who did vs. did not experience a clinically meaningful PCL decrease were nearly twice as likely to be adherent ADM users. These results remained largely unchanged after control for confounding. Although adherence was better in the patients with large PTSD improvement, the percent of adherent patients remained low. After controlling for confounding, only 53.5% of patients with and 39.3% without clinically meaningful PCL decrease were adherent, which is similar to the 47% of patients with comorbid depression and anxiety disorder adherent to ADMs, revealed from analysis of a private sector medical claims data (Stein et al., 2006). The percent of ADM adherent patients in those without clinically meaningful PCL reductions is similar to a previous study of patients discharged from PTSD residential treatment, of whom 34% were ADM adherent during 12-month follow-up (Lockwood et al., 2009).
Our results are limited in several ways. We do not know if patients took their medication once it was dispensed. Actual or perceived side effects of ADMs and inability to tolerate side effects are associated with early discontinuation of medication treatment (Hung, 2014). We assume poor adherence due to side-effects would be randomly distributed in patients with and without large PCL decrease and thus, unlikely to bias our analyses. We only have depression symptom scores on a small subset of patients and we are unable to draw conclusions about the effect of depression severity or magnitude of depression symptom change in those who are and are not adherent. While we controlled for a large number of confounding variables, it is possible that residual confounding exists. Last it is not known if these results generalize to non-VHA patients with PTSD.
Although we were unable to measure depression symptom change, we do estimate that about 44% of patients with clinically meaningful PCL decrease would experience depression improvement compared to 32% of patients without meaningful PCL decrease, based on prior evidence suggesting that 82.5% of ADM adherent patients have improved depression symptoms (Akerblad et al., 2006). Targeting improvements in PTSD may be an effective way to improve depression in patients with this psychiatric comorbidity (Norman et al., 2011). Improvement in both PTSD and depression may reduce the risk of some health consequences associated with these conditions, like type 2 diabetes (Scherrer et al., 2019). Additional research is needed to determine if PTSD symptom reduction is associated with greater likelihood of adherence to non-psychotropics, such as anti-hypertensives and anti-diabetic medications.
Supplementary Material
Highlights.
PTSD is associated with poor antidepressant medication (ADM) adherence.
PTSD and depressive symptoms change similarly over time.
Significant PTSD symptom improvement is associated with greater ADM adherence.
PTSD symptom reduction may improve depression treatment outcomes.
Acknowledgments
Funding: This work was supported by the National Heart, Lung, and Blood Institute [NHLBI grant number: R01HL125424]. This material is the result of work supported with resources and the use of facilities at the Harry S. Truman Memorial Veterans’ Hospital.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest
Jeffrey F. Scherrer receives compensation as Editor of Oxford Press’s Family Practice.
Declarations of interest for all other authors: none.
Access to Data and Data Analysis:
Jeffrey F. Scherrer and Joanne Salas had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Joanne Salas conducted the analysis.
Role of Funding Sources: Funding sources had no role in study design; in the collection, analysis and interpretation of data; in the writing of the manuscript; or in the decision to submit for publication.
REFERENCES
- Akerblad AC, Bengtsson F, von Knorring L, Ekselius L, 2006. Response, remission and relapse in relation to adherence in primary care treatment of depression: a 2-year outcome study. Int Clin Psychopharmacol 21, 117–124. [DOI] [PubMed] [Google Scholar]
- Austin PC, Stuart EA, 2015. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Statistics in medicine 34, 3661–3679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Blanchard EB, Jones-Alexander J, Buckley TC, Forneris CA, 1996. Psychometric properties of the PTSD Checklist (PCL). Behaviour research and therapy 34, 669–673. [DOI] [PubMed] [Google Scholar]
- Cramer JA, Roy A, Burrell A, Fairchild CJ, Fuldeore MJ, Ollendorf DA, Wong PK, 2008. Medication compliance and persistence: terminology and definitions. Value Health 11, 44–47. [DOI] [PubMed] [Google Scholar]
- Flory JD, Yehuda R, 2015. Comorbidity between post-traumatic stress disorder and major depressive disorder: alternative explanations and treatment considerations. Dialogues Clin Neurosci 17, 141–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Frayne SM, Miller DR, Sharkansky EJ, Jackson VW, Wang F, Halanych JH, Berlowitz DR, Kader B, Rosen CS, Keane TM, 2010. Using administrative data to identify mental illness: what approach is best? Am J Med Qual 25, 42–50. [DOI] [PubMed] [Google Scholar]
- Gravely AA, Cutting A, Nugent S, Grill J, Carlson K, Spoont M, 2011. Validity of PTSD diagnoses in VA administrative data: comparison of VA administrative PTSD diagnoses to self-reported PTSD Checklist scores. Journal of rehabilitation research and development 48, 21–30. [DOI] [PubMed] [Google Scholar]
- Ho SC, Chong HY, Chaiyakunapruk N, Tangiisuran B, Jacob SA, 2016. Clinical and economic impact of non-adherence to antidepressants in major depressive disorder: A systematic review. J Affect Disord 193, 1–10. [DOI] [PubMed] [Google Scholar]
- Holowka DW, Marx BP, Gates MA, Litman HJ, Ranganathan G, Rosen RC, Keane TM, 2014. PTSD diagnostic validity in Veterans Affairs electronic records of Iraq and Afghanistan veterans. Journal of consulting and clinical psychology 82, 569–579. [DOI] [PubMed] [Google Scholar]
- Hung CI, 2014. Factors predicting adherence to antidepressant treatment. Curr Opin Psychiatry 27, 344–349. [DOI] [PubMed] [Google Scholar]
- Kronish IM, Edmondson D, Li Y, Cohen BE, 2012. Post-traumatic stress disorder and medication adherence: results from the Mind Your Heart study. Journal of psychiatric research 46, 1595–1599. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lockwood A, Steinke DT, Botts SR, 2009. Medication adherence and its effect on relapse among patients discharged from a Veterans Affairs posttraumatic stress disorder treatment program. The Annals of pharmacotherapy 43, 1227–1232. [DOI] [PubMed] [Google Scholar]
- Monson CM, Gradus JL, Young-Xu Y, Schnurr PP, Price JL, Schumm JA, 2008. Change in posttraumatic stress disorder symptoms: do clinicians and patients agree? Psychol Assess 20, 131–138. [DOI] [PubMed] [Google Scholar]
- National Center for PTSD, 2012. Using the PTSD Checklist for DSM-IV (PCL). National Center for PTSD. [Google Scholar]
- Norman SB, Trim RS, Goldsmith AA, Dimsdale JE, Hoyt DB, Norman GJ, Stein MB, 2011. Role of risk factors proximate to time of trauma in the course of PTSD and MDD symptoms following traumatic injury. Journal of traumatic stress 24, 390–398. [DOI] [PubMed] [Google Scholar]
- Peterson AM, Nau DP, Cramer JA, Benner J, Gwadry-Sridhar F, Nichol M, 2007. A checklist for medication compliance and persistence studies using retrospective databases. Value Health 10, 3–12. [DOI] [PubMed] [Google Scholar]
- Romano PS, Roos LL, Jollis JG, 1993. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol 46, 1075–1079; discussion 1081-1090. [DOI] [PubMed] [Google Scholar]
- Rosenbaum PR, Rubin DB, 1983. The central role of the propensity score in observational studies for causal effects. Biometrika Trust 70, 41–55. [Google Scholar]
- Ruggiero KJ, Del Ben K, Scotti JR, Rabalais AE, 2003. Psychometric properties of the PTSD Checklist-Civilian Version. Journal of traumatic stress 16, 495–502. [DOI] [PubMed] [Google Scholar]
- Scherrer JF, Salas J, Norman SB, Schnurr PP, Chard KM, Tuerk P, Schneider FD, van den Berk-Clark C, Cohen BE, Friedman MJ, Lustman PJ, 2019. Reductions in risk of type 2 diabetes following clinically meaningful PTSD improvement. Manuscript submitted for publication.
- Seal KH, Maguen S, Cohen B, Gima KS, Metzler TJ, Ren L, Bertenthal D, Marmar CR, 2010. VA mental health services utilization in Iraq and Afghanistan veterans in the first year of receiving new mental health diagnoses. Journal of traumatic stress 23, 5–16. [DOI] [PubMed] [Google Scholar]
- Solberg LI, Engebretson KI, Sperl-Hillen JM, Hroscikoski MC, O'Connor PJ, 2006. Are claims data accurate enough to identify patients for performance measures or quality improvement? The case of diabetes, heart disease, and depression. Am J Med Qual 21,238–245. [DOI] [PubMed] [Google Scholar]
- Stein MB, Cantrell CR, Sokol MC, Eaddy MT, Shah MB, 2006. Antidepressant adherence and medical resource use among managed care patients with anxiety disorders. Psychiatric services 57, 673–680. [DOI] [PubMed] [Google Scholar]
- Sturmer T, Wyss R, Glynn RJ, Brookhart MA, 2014. Propensity scores for confounder adjustment when assessing the effects of medical interventions using nonexperimental study designs. J Intern Med 275, 570–580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weathers FW, Huska JA, Keene TM, 1991. PCL-M for DSM-IV. National Center for PTSD-Behavioral Division, Boston, MA. [Google Scholar]
- Zivin K, Ganoczy D, Pfeiffer PN, Miller EM, Valenstein M, 2009. Antidepressant adherence after psychiatric hospitalization among VA patients with depression. Administration and policy in mental health 36, 406–415. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.