August 5, 2009

Association of Torture and Other Potentially Traumatic Events With Mental Health Outcomes Among Populations Exposed to Mass Conflict and DisplacementA Systematic Review and Meta-analysis

Author Affiliations

Author Affiliations: Psychiatry Research and Teaching Unit, School of Psychiatry (Drs Steel and Silove and Mss Chey and Marnane), and School of Psychology (Dr Bryant), University New South Wales, Liverpool New South Wales, Australia; and Department of Mental Health and Substance Abuse, World Health Organization, Geneva, Switzerland (Dr van Ommeren).

JAMA. 2009;302(5):537-549. doi:10.1001/jama.2009.1132

Context Uncertainties continue about the roles that methodological factors and key risk factors, particularly torture and other potentially traumatic events (PTEs), play in the variation of reported prevalence rates of posttraumatic stress disorder (PTSD) and depression across epidemiologic surveys among postconflict populations worldwide.

Objective To undertake a systematic review and meta-regression of the prevalence rates of PTSD and depression in the refugee and postconflict mental health field.

Data Sources An initial pool of 5904 articles, identified through MEDLINE, PsycINFO and PILOTS, of surveys involving refugee, conflict-affected populations, or both, published in English-language journals between 1980 and May 2009.

Study Selection Surveys were limited to those of adult populations (n ≥ 50) reporting PTSD prevalence, depression prevalence, or both. Excluded surveys comprised patients, war veterans, and civilian populations (nonrefugees/asylum seekers) from high-income countries exposed to terrorist attacks or involved in distal conflicts (≥25 years).

Data Extraction Methodological factors (response rate, sample size and design, diagnostic method) and substantive factors (sociodemographics, place of survey, torture and other PTEs, Political Terror Scale score, residency status, time since conflict).

Data Synthesis A total of 161 articles reporting results of 181 surveys comprising 81 866 refugees and other conflict-affected persons from 40 countries were identified. Rates of reported PTSD and depression showed large intersurvey variability (0%-99% and 3%-85.5%, respectively). The unadjusted weighted prevalence rate reported across all surveys for PTSD was 30.6% (95% CI, 26.3%-35.2%) and for depression was 30.8% (95% CI, 26.3%-35.6%). Methodological factors accounted for 12.9% and 27.7% PTSD and depression, respectively. Nonrandom sampling, small sample sizes, and self-report questionnaires were associated with higher rates of mental disorder. Adjusting for methodological factors, reported torture (Δ total R2 between base methodological model and base model + substantive factor [ΔR2] = 23.6%; OR, 2.01; 95% CI, 1.52-2.65) emerged as the strongest factor associated with PTSD, followed by cumulative exposure to PTEs (ΔR2 = 10.8%; OR, 1.52; 95% CI, 1.21-1.91), time since conflict (ΔR2 = 10%; OR, 0.77; 95% CI, 0.66-0.91), and assessed level of political terror (ΔR2 = 3.5%; OR, 1.60; 95% CI, 1.03-2.50). For depression, significant factors were number of PTEs (ΔR2 = 22.0%; OR, 1.64; 95% CI, 1.39-1.93), time since conflict (ΔR2 = 21.9%; OR, 0.80; 95% CI, 0.69-0.93), reported torture (ΔR2 = 11.4%; OR, 1.48; 95% CI, 1.07-2.04), and residency status (ΔR2 = 5.0%; OR, 1.30; 95% CI, 1.07-1.57).

Conclusion Methodological factors and substantive population risk factors, such as exposure to torture and other PTEs, after adjusting for methodological factors account for higher rates of reported prevalence of PTSD and depression.