PTSD in Urban Primary Care: High Prevalence and Low Physician Recognition
Posttraumatic stress disorder (PTSD) is associated with medical and psychological morbidity. The prevalence of PTSD in urban primary care has not been well described.
To measure the prevalence of PTSD in primary care patients overall and among those with selected conditions (chronic pain, depression, anxiety, heavy drinking, substance dependence (SD), irritable bowel syndrome (IBS), and immigrant status).
English-speaking patients aged 18–65 years old, awaiting primary care appointments in an urban academic medical center, were eligible for enrollment to determine PTSD prevalence (N = 509). Additional eligible participants (n = 98) with IBS or SD were subsequently enrolled.
PTSD (past year) and trauma exposure were measured with Composite International Diagnostic Interview. We calculated the prevalence of PTSD associated with depression, anxiety, heavy drinking, SD, IBS, and chronic pain. Only the analyses on heavy drinking, SD, and IBS used all 607 participants.
Among the 509 adults in primary care, 23% (95% CI, 19–26%) had PTSD, of whom 11% had it noted in the medical record. The prevalence of PTSD, adjusted for age, gender, race, and marital and socioeconomic statuses, was higher in participants with, compared to those without, the following conditions: chronic pain (23 vs 12%, p = .003), major depression (35 vs 11%, p < .0001), anxiety disorders (42 vs 14%, p < .0001), and IBS (34 vs 18%, p = .01) and lower in immigrants (13 vs 21%, p = .05).
The prevalence of PTSD in the urban primary care setting, and particularly among certain high-risk conditions, compels a critical examination of optimal approaches for screening, intervention, and referral to PTSD treatment.
KEY WORDSunderserved populations PTSD prevalence
Media coverage of the 2001 World Trade Center attacks increased American awareness of posttraumatic stress disorder (PTSD) as an important psychiatric diagnosis.1,2 Recent documentation of the burden of mental disorders, including PTSD, among US soldiers returning from Iraq3,4 and people affected by Hurricane Katrina5, 6, 7 further increased awareness of PTSD risk after exposure to trauma. Posttraumatic stress disorder is characterized by disabling symptoms of reexperiencing, avoidance, and numbing after a significant traumatic event.8 The World Health Organization and others have substantiated the major impact of posttraumatic mental disturbance on social and health functioning.9,10
In primary care settings, the prevalence of PTSD depends upon the features of the population studied and the methods for participant recruitment and assessment. In a nationally representative US sample, Kessler et al.11 found a lifetime PTSD prevalence of 6.8%; 61% of males and 51% of females reported trauma exposure.12 Breslau et al.13 reported 90% lifetime prevalence of trauma and 9% probability of developing PTSD after trauma in an urban sample in Detroit. Stein and colleagues14 found a 2% prevalence of current PTSD in a middle-class, university-affiliated primary care practice in San Diego. In a representative sample of national health clinics in Israel, 9% of patients had current PTSD.15 One-third of African-American primary care patients in an urban setting met PTSD diagnostic criteria.16 Prevalence at veterans affairs (VA) health facilities ranges from 11.5% (current) to 27% (lifetime) PTSD.17,18 Female gender, being divorced or widowed, other psychiatric or substance-use disorders, and exposure to certain traumas increase PTSD risk.12,13,19
However, these studies may not accurately estimate PTSD prevalence because of selective recruitment, nondiagnostic assessments, and low study participation.14,16,19, 20, 21 Only one of these studies was conducted in a primary care practice in an urban, impoverished, and under represented minority population, a group at risk for PTSD because of high trauma exposure and low levels of social support.22 PTSD is clinically under-recognized; of patients with PTSD, primary care providers recognized 2% in Israel and 47% in a VA setting.15,17 These findings are consistent with extensive literature on clinician under-recognition of other mental disorders.23
Accurate determination of PTSD prevalence in primary care settings serving patients at high risk is important for several reasons. First, PTSD is associated with increased health-care utilization.4,24, 25, 26 Second, PTSD and trauma exposure have been associated with conditions commonly treated by primary care clinicians: somatization functional impairment, depression, anxiety, chronic pain, irritable bowel syndrome (IBS) and substance-use disorders.15,18,20,27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42 Apart from depression,16 these associations have not been evaluated in primary care. Additionally, immigrant and refugee populations have high exposure to violence and merit further evaluation.43 If PTSD is sufficiently prevalent in urban primary care patients or in those with certain common conditions, then it fulfills a key criterion for a clinical screening recommendation.44 Determining if primary care patients with specific comorbid conditions have particularly high prevalence of current PTSD strengthens the case and may even suggest an approach for targeted screening in primary care.
We conducted a cross-sectional study to evaluate the prevalence of and risk factors for PTSD in primary care patients at an urban hospital-based outpatient department. We assessed the prevalence of PTSD among those with immigrant status, depression, generalized anxiety or panic disorder, chronic pain, IBS, heavy drinking, and substance dependence (SD), and we determined also whether PTSD was identified by physicians caring for these patients. We hypothesized that PTSD prevalence would be high, especially in patients with selected characteristics, and under-recognized by clinicians. We also hypothesized that clinicians might identify PTSD as depression, a more familiar mental illness.
We interviewed a consecutive sample of primary care patients at the outpatient department of an urban university-affiliated hospital to examine overall prevalence of PTSD and its prevalence among patients with seven selected conditions: being an immigrant to the United States, major or other depression, generalized anxiety or panic disorder, heavy drinking, SD, IBS, and chronic pain. We also reviewed medical records for documentation of PTSD and depression.
Patients visiting the Internal Medicine and Family Medicine clinics were eligible if they spoke English, were 18–65 years old, and had a scheduled appointment with a primary care clinician. Patients were excluded if they could not be interviewed alone or did not appear to understand study procedures. After a planned minimum study enrollment of 100 participants per selected condition was met for depression, anxiety, pain, and immigrant status with the first 509 enrolled participants (“consecutive sample”), enrollment was limited to eligible patients with heavy drinking, drug use or IBS, or none of the selected conditions (control group) to permit planned subgroup analyses (“oversampled group”, n = 98). Enrollment occurred from February 2003 to September 2004.
Recruitment and Enrollment
Interviewers approached consecutive patients upon arrival for appointments and screened them for eligibility. Eligible patients were asked to participate in an interview about stress and health. Each participant gave written informed consent. Study participants were compensated with $10, and they received safety referrals upon interview completion. Interviews not completed before clinician visit were completed immediately afterward. Boston University Medical Center’s Institutional Review and HIPAA Privacy Review Boards approved the study. A Certificate of Confidentiality was obtained from the National Institutes of Health.
The self-administered screening tool contained 19 questions about the following: demographics, symptoms of depression and anxiety (2 questions from the PRIME-MD screener),45 IBS (4 definition questions from the ROME II Integrative questionnaire),46 quantity and frequency of alcohol use in the past 30 days,47 use of heroin or cocaine in the past month, and chronic pain in the past 3 months.48
Interview included demographic questions, the Composite International Diagnostic Interview (CIDI) version 2.1 PTSD module,8,49 the Chronic Pain Definitional Questionnaire,48 the Patient Health Questionnaire (PHQ)45 modules measuring depression diagnoses (major and other depression in the past 2 weeks) and anxiety disorder (generalized anxiety and/or panic disorder in the past 4 weeks), and the CIDI-Short Form (CIDI-SF) substance-use disorder modules.49 The interview lasted for approximately 30 min.
Medical Record Review
Using standardized data-collection forms, researchers trained in chart abstraction reviewed patient electronic medical records (EMRs), including all outpatient and emergency department records and inpatient discharge summaries for the 12 months before and including the date of interview, to identify documentation of ICD-9 coded PTSD or depression in the problem list or in the visit assessment. All clinical encounters used EMR documentation.
Dependent Variables: Trauma Exposure, PTSD, Medical Record Documentation
Current (past 12 months) and lifetime PTSD were diagnosed using CIDI version 2.1.49 The CIDI includes 9 questions about specific traumatic events and 2 open-ended questions. The research team classified open-ended responses as traumatic or nontraumatic according to DSM-IV guidelines for qualifying trauma.8 We defined trauma exposure as self-reported experience of any category of qualifying trauma. Multiple experiences of a single category of trauma (e.g., rape or physical assault) were counted as one trauma exposure. Clinician recognition of PTSD and depression was defined as documentation of this diagnosis (PTSD or depression, respectively) in the active problem list or in the free text of a note in the medical record.
Independent Variables and Covariates
Independent variables included immigrant status, IBS, chronic pain, anxiety, depression, heavy drinking, and SD. Immigrants were participants born outside the US or Puerto Rico. Irritable bowel syndrome was present if participants endorsed diagnostic criteria from the Rome II Integrative Questionnaire.46 Chronic pain was present if participants endorsed pain for 3 months or more on the Chronic Pain Definitional Questionnaire.48 Current anxiety disorder was present if participants met diagnostic criteria for generalized anxiety and/or panic disorder in the past month according to the PHQ.45 Current depressive disorder was classified into major or other depression in the past 2 weeks according to the PHQ.45 Heavy drinking in the past month was defined using the National Institute of Alcohol Abuse and Alcoholism guidelines.47 Substance dependence included alcohol dependence and/or drug dependence in the past 6 months as determined by the CIDI-SF.50 Covariates were age, gender, income, employment, marital status, and race.
We used descriptive analyses to determine the prevalence of current and lifetime PTSD and trauma exposure. Bivariate analyses compared the prevalence of PTSD and trauma exposure in people with and without 7 selected conditions. We employed multivariable logistic regression to adjust for covariates. Adjusted proportions were calculated using the mean values of background covariates; P values comparing adjusted proportions are based on the regression coefficients for the group variable. Bivariate analyses compared PTSD or depression EMR documentation with research interview diagnoses. We used the consecutive sample (N = 509) for most analyses. Analyses examining heavy drinking, SD, and IBS utilized the enhanced sample (the consecutive sample plus the oversampled group, n = 98) of 607 participants.
Participant Demographic and Clinical Characteristics
Demographic and Clinical Characteristics of Participants Recruited in Primary Care (Consecutive Sample)
Total (N = 509)
Male (n = 251)
Female (n = 258)
Mean age in years (SD)
USA- or Puerto Rico-born
Married/Living with partner
Never married/not living with a partner
Unemployed or disabled
Irritable bowel syndrome
Other depressive disorder*
PTSD and Trauma Prevalence
Thirty-four percent (95% CI, 30–38%) of the consecutive sample met the diagnostic criteria for lifetime PTSD and 23% (CI 19–26%) for current PTSD. Median duration of PTSD symptoms was 6 years (range 1 month–44 years). Twenty-six percent (CI 21–32%) of women and 19% (CI 14–24%) of men met the criteria for current PTSD (p = .04). Trauma exposure was very common (79%), with 65% reporting exposure to more than one trauma category. See Appendix for details of trauma exposure.
Current PTSD was more common in participants with incomes <$20,000 (30 vs 14%, p < .0001), unemployed or disabled (30 vs 16%, p = .0001), or separated/divorced (31 vs 15% for married, p = .009; or 31 vs 22% for never married, p = .009).
PTSD was significantly more common among participants with at least one comorbid condition (depression, pain, IBS, anxiety, and/or SD) than among those with none of these conditions (27 vs 8%, p < .0001). Ninety-one percent of participants with current PTSD had at least one comorbid condition compared to 71% of those without PTSD. Of note, 75% of the entire consecutive sample (including those with PTSD) met the diagnostic criteria for at least one comorbid condition. In those with a comorbid condition, 28% had PTSD, whereas in those without a comorbid condition, 91% did not have PTSD.
Relationship of Comorbid Conditions with PTSD Prevalence: Adjusted Analyses
PTSD Prevalence Among Participants with Selected Conditions (Consecutive Sample, N = 509)
Unadjusted prevalence (%)
Adjusted prevalence‡ (95% CI)
Unadjusted prevalence (%)
Adjusted prevalence‡ (95% CI)
PTSD Prevalence Among Participants with Comorbid Conditions (Enhanced Sample, N = 607)
Unadjusted prevalence (%)
Adjusted prevalence‡ (95% CI)
Unadjusted prevalence (%)
Adjusted prevalence‡ (95% CI)
Clinician Recognition: Medical Record Documentation
Based on EMRs of the participants in the consecutive sample, 11% of patients with CIDI-determined current PTSD had a PTSD diagnosis documented in the medical record. Of participants with current PTSD, 51% had depression noted in the medical record. Depression was noted in the EMR of 55% of those with both PTSD and depression (major and other), 43% of those with PTSD alone, 32% of those with depression but not PTSD, and 18% of those with neither PTSD nor depression (p < .0001 for differences between groups).
Among the patients attending an urban hospital-based primary care practice, PTSD is an exceptionally common diagnosis.51 Almost one-quarter of patients met the criteria for current PTSD and one-third met the criteria for lifetime PTSD. Importantly, documentation of this diagnosis was exceedingly uncommon. Only 11% of those with current PTSD were correctly identified in the medical record. In patients with certain conditions (chronic pain, IBS, depression, anxiety disorder, and SD), PTSD prevalence was two to three times as high; patients with these conditions accounted for over 90% of all cases of PTSD. While over a quarter of those with a co-occurring condition had PTSD, not having a co-occurring condition made the diagnosis of PTSD very unlikely.
Our findings are consistent with the published studies of people with PTSD reporting more physical symptoms.52, 53, 54, 55, 56, 57, 58 Multiple factors may contribute to the association between PTSD and physical symptoms including organic illnesses,31,57,59, 60, 61 altered physiology,62, 63, 64 and psychological mechanisms.30
In this study, lifetime PTSD was present in half of patients with current heavy drinking and SD. These findings are similar to those from substance-abuse treatment settings and primary care.38,39,65 Surprisingly, current SD and heavy drinking were strongly associated with lifetime PTSD but only weakly with current PTSD. One explanation may be that substance use dampens PTSD symptoms, an assertion supported by reports of PTSD patient self-medication in other studies.66, 67, 68 Another explanation may be that patterns of substance use are initiated in a period of current PTSD but persist beyond its resolution.
The higher PTSD prevalence in those with a diagnosis of depression and anxiety was not unexpected based on the literature detailing the psychiatric conditions comorbid with PTSD.12,69 While primary care clinicians have become relatively more comfortable with identifying and managing depression in the past 10 years, more work remains to optimize screening and treatment for other psychiatric disorders.70, 71, 72, 73 More than half of our participants with EMR documentation of depression met the diagnostic criteria for PTSD, suggesting that clinicians may have labeled psychological distress as the more familiar diagnosis of depression.
Unlike depression, for which medication alone can be effective, evidence-based care for PTSD includes psychotherapy with or without medication.74 Trauma-focused psychotherapies can result in significant improvement of symptoms in more than half of patients.75 Thus, treatment by primary care clinicians may require new strategies beyond psychoactive medication prescription. In particular, the close relationship between physical problems and PTSD suggests an avenue to develop new interventions, better coordinating the care of general medicine and mental-health clinicians. Such coordination of care to address mental-health conditions in the primary care setting has been encouraged by a recent Institute of Medicine report.76 Based on this study’s findings, clinicians should consider evaluating patients who present with more than 3 months of physical pain or IBS for PTSD and referring those with PTSD for care. In addition, effective PTSD treatment may improve pain. Extensive literature suggests that treating depressive symptoms can improve pain and disability.77, 78, 79, 80, 81
An unexpected finding was the lower PTSD prevalence among immigrants. The study eligibility requirement to speak English likely selected acculturated immigrants, producing a “healthy immigrant effect”.82 Immigrants in this study had more social support as well. Other studies suggest lower psychiatric morbidity among some immigrant groups, depending on home country, socioeconomic status and subsequent adjustment to their adopted land.83 Additionally, it is likely that immigrants with PTSD would have had substantial barriers to learning English fluently enough to participate in this study.84 Future studies should examine both English and non-English speaking immigrants for PTSD to better understand its prevalence.
The prevalence of PTSD in these urban primary care patients was markedly higher than that found in most other studies of primary care settings, including those involving veterans, where PTSD prevalence is expected to be high.14,15,18,19,21,31,37,85 Possible explanations for this population’s high PTSD prevalence are rooted in its socioeconomic profile with high exposure to trauma and low levels of social support. In a recent study of high school students in Boston, 71% witnessed violence and 44% were directly victimized in the prior year.86 Social support is an important source of protection from the development of PTSD after trauma exposure,87, 88, 89, 90 and several demographic characteristics can serve as proxies for social support in this study. For example, half of our participants earned less than $20,000 per year. Likewise, half of the participants, all of whom were of working age, were unemployed or disabled. In addition, only 27% were married or living with a partner.91,92
There were limitations to this study. As a diagnostic instrument, the CIDI PTSD module only assessed for symptoms after one trauma selected as most stressful by the participant, whereas other diagnostic interviews assessed for symptoms after multiple traumas, offering more opportunities for PTSD diagnosis. However, the instrument used in this study is the same or very similar to those instruments used in national PTSD studies.12,13 Our sample was drawn from a single hospital care system whose mission is to serve vulnerable populations and may not represent settings with a broader range of patients. However, our sample is similar to other urban medical settings serving low-income populations in demographic characteristics as well as prevalence of: depression, trauma exposure, substance-use disorders, and co-occurring chronic conditions.16,93, 94, 95, 96, 97 Further research in primary care settings serving other populations is warranted. Lastly, the prevalence would only apply to patients aged 18–65 years old presenting for care, not to a population of all primary care patients.
Despite these limitations, the study had numerous strengths. We enrolled a large primary care sample of patients, oversampled for less prevalent comorbid conditions of interest, utilized a well-regarded structured diagnostic interview for PTSD, and analyzed data with multivariable approaches. These methodological features address some of the concerns with prior studies of PTSD prevalence in primary care.
Our results underscore the need to focus on the identification and treatment of PTSD in urban primary care settings as well as to explore the relationship between PTSD and certain conditions (chronic pain, IBS, SD, heavy drinking, depression, anxiety, and immigrant status) in varied settings with lower baseline PTSD prevalence. Before recommending broad screening for PTSD, screening tools and interventions in the primary care setting need to be developed and tested. In the meantime, it is appropriate for primary care clinicians treating patients in high-risk areas, particularly those with the conditions identified in this study, to assess for PTSD and refer those identified to effective care.
This work was supported by a Generalist Physician Faculty Scholar Award from the Robert Wood Johnson Foundation, Princeton, New Jersey (RWJF #045452) and by a career development award from the National Institute on Drug Abuse, National Institutes of Health (K23 DA016665).
We thank Jessica Geier, Minga Claggett-Borne, Lauren Kelly, Michael Rosas, Mary Reyes, Pavan Sekhar, Jen Tran, Eric Holder, and Mary Benitta Schickel for their aid in data collection, and Joann Elmore (MD MPH), Roger Weiss (MD), and Larry Culpepper (MD) for their comments on research design and analysis.
Conflict of Interest Summary
- 2.Goode E. A nation challenged: psychological trauma. The New York Times. September 18, 2001;B:1.Google Scholar
- 3.Mishra R. Mental toll on troops detailed. The Boston Globe. July 1, 2004:A1.Google Scholar
- 6.Centers for Disease Control and Prevention. Health hazard evaluation of police officers and firefighters after Hurricane Katrina, New Orleans, Louisiana, October 17–28 and November 30–December 5, 2005: Centers for Disease Control and Prevention; 2006.Google Scholar
- 8.American Psychiatric Association, ed. Diagnostic and Statistical Manual of Mental Disorders Fourth Edition. Washington, DC: American Psychiatric Association; 1994.Google Scholar
- 10.Murray CJ, Lopez AD, eds. The global burden of disease: a comprehensive assessment of mortality and disability from diseases, injuries and risk factors in 1990 and projected to 2020. Cambridge, MA: Harvard University Press; 1996; No. 1.Google Scholar
- 44.Force UPST. Guide to Clinical Preventive Services, 2nd Edition. Bethesda, MD; 1996.Google Scholar
- 46.Drossman DA, Corazziari E, Talley NJ, Thompson WG, Whitehead WE. Rome II. The Functional Gastrointestinal Disorders. Diagnosis, Pathophysiology and Treatment: A Multinational Consensus. McLean, VA: Degnon Associates; 2000.Google Scholar
- 47.NIAAA. Helping Patients Who Drink Too Much: A Clinician’s Guide. http://pubs.niaaa.nih.gov/publications/Practitioner/CliniciansGuide2005/clinicians_guide.htm; 2006.
- 48.Purves AM, Penny KI, Munro C, et al. Defining chronic pain for epidemiological research assessing a subjective definition. Pain Clinic. 1998;10(3):139–47.Google Scholar
- 49.World Health Organization. Composite International Diagnostic Interview (CIDI): Version 2.1. Geneva: World Health Organization; 1997.Google Scholar
- 51.Health, United States, 2004 With Chartbook on Trends in the Health of Americans. Hyattsville, MD: National Center for Health Statistics; 2004.Google Scholar
- 58.Beckham J, Crawford A, Feldman ME, et al. Chronic posttraumatic stress disorder and chronic pain in Vietnam combat veterans. J Psychiatr Res. 1997;43(4):379–89.Google Scholar
- 69.Bruce SE, Weisberg RB, Dolan RT, et al. Trauma and posttraumatic stress disorder in primary care patients. Prim Care Companion J Clin Psychiat. 2001;3(5):211–7 (Oct).Google Scholar
- 74.Foa EB, Davidson JRT, Frances A. The expert consensus guideline series: treatment of post-traumatic stress disorder. J Clin Psychiatry. 1999;60(suppl 16):1–76.Google Scholar
- 76.Institute of Medicine, ed. Improving the Quality of Health Care for Mental and Substance Use Conditions: Quality Chasm Series. Washington, DC: The National Academies Press; 2006.Google Scholar
- 86.Hemenway D, Prothrow-Stith D, Browne A. Report of the 2004 Boston Youth Survey. Boston: City of Boston; 2005 (Aug).Google Scholar