Posttraumatic Stress Disorder and Risk for CHD
Posttraumatic Stress Disorder and Risk for CHD
This research was performed in accordance with MOOSE guidelines for meta-analyses of observational studies. Potentially relevant articles were identified by a trained medical librarian (L.F.) who searched publicly available computerized databases. Potentially relevant articles were identified by searching the electronic databases Ovid MEDLINE, PsycINFO, Scopus, PILOTS, and the Cochrane Review. Dates searched were from inception to July 2013. The searches were conducted on August 15, 2012, and July 1, 2013. All relevant subject headings and free-text terms were used to represent PTSD and incident CHD/cardiac-specific mortality, and the sets of terms were combined with AND. Terms for MEDLINE included the following: exp Stress Disorders, Traumatic/OR ptsd.tw. OR (post-traumatic OR (post adj traumatic)).tw OR posttraumatic.tw. OR acute stress disorder$.tw. OR asd.tw, exp Acute Coronary Syndrome/OR acute coronary.tw. OR acs.tw OR exp Myocardial Infarction/OR myocardial infarct$.tw. OR (mi OR ami).tw. OR (heart adj attack$).tw. OR (stemi OR nstemi).tw. OR ((preinfarction OR unstable) adj angina$).tw. These terms were adapted for the other databases. Additional records were identified by scanning the reference lists of relevant studies and reviews and by using the Related Articles feature in PubMed and the Cited Reference Search in ISI Web of Science.
Eligible study designs included all studies that measured PTSD caused by any traumatic event in initially healthy participants and estimated its association with subsequent CHD or cardiac-specific mortality. Two investigators (D.E., M.M.B.) independently reviewed all citations identified through the literature search, using a predefined protocol to assess study eligibility. Articles that clearly did not meet the inclusion criteria were excluded at the title and abstract level. The remaining articles were selected for full-text review; articles that did not meet the inclusion criteria were excluded. Disagreements regarding the selection of articles were resolved through discussion between the 2 reviewers, and full consensus was achieved at each stage of review.
Two investigators (D.E. and M.M.B.) abstracted information about the dates and cohort sizes of the studies, characteristics of participants enrolled, methods for assessing PTSD, covariates including depression, and the strength of the association between PTSD and incident CHD and cardiac-specific mortality. One study reported estimates for men and women separately, and we thus extracted data separately from this study. Similarly, one study reported estimates for participants with and without depression separately, and we extracted data from this study separately, as well.
Four studies reported the association of PTSD diagnosis or of a positive screen on a validated PTSD self-report screening questionnaire with incident CHD or cardiac mortality. The other 2 studies reported the risk of incident CHD associated with a 1-SD increase in PTSD symptoms on a self-report PTSD screen questionnaire. For both of those studies, we transformed the findings into a single risk estimate, corresponding to 2 SDs above each study's mean as an approximation of a positive PTSD screen—because the sample scores at that level best approximated published cutoffs for the scales used. Furthermore, although half of the studies reported associations in hazard ratios (HRs), one study reported a relative risk and 2 reported an odds ratio. For this meta-analysis, we calculated an aggregate point estimate as an HR, using the following equations to convert the relative risk and odds ratio to HRs:
and
, where RR is the relative risk, OR is the odds ratio, HR is the hazards ratio, and r is the event rate for the reference group (ie, no PTSD).
Comprehensive MetaAnalysis (version 2; BioStat Software, Engelwood, NJ) was used for completing all statistical tests and associated graphic results. Estimates of the overall risk of PTSD with incident CHD were pooled using a random-effects model. A secondary pooled analysis was performed by restricting to the 5 studies that adjusted for depression as a covariate for the association of PTSD with incident CHD. Heterogeneity assessments using Cochrane Q statistic and I preceded all meta-analytic tests. There was a statistically significant heterogeneity, justifying the use of a random-effects model to estimate and test effects. Log-transformed HRs and 95% CIs were calculated for each study using the reported effect size and estimates of the SE of each effect drawn from data reported in the article. When articles reported multiple models, we selected the model with the highest level of covariate adjustment. There were too few studies to test for moderator effects; however, we provide descriptive information concerning potential moderators below. To address the issue of publication bias, we calculated Orwin's fail-safe N and created a funnel plot of SEs by log HRs.
This work was supported by Grants HL-088117, HL-084438, and CA-156709 from the National Institutes of Health (NIH), Bethesda, MD. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of the National Center for Research Resources or the NIH. This work was supported, in part, by Columbia University's Clinical and Translational Science Awards (CTSA) Grant No. UL1RR024156 from National Center for Advancing Translational Sciences-National Center for Research Resources (NCATs-NCRR)/NIH. Dr Kronish is supported by Grant K23-HL098359 from the National Heart, Lung, and Blood Institute. Dr Shaffer is supported by Grant 12CRP8870004 from the American Heart Association and by Grant K23HL112850 from the NIH, Bethesda, MD. No funding bodies had any role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents.
Methods
Data Sources and Searching
This research was performed in accordance with MOOSE guidelines for meta-analyses of observational studies. Potentially relevant articles were identified by a trained medical librarian (L.F.) who searched publicly available computerized databases. Potentially relevant articles were identified by searching the electronic databases Ovid MEDLINE, PsycINFO, Scopus, PILOTS, and the Cochrane Review. Dates searched were from inception to July 2013. The searches were conducted on August 15, 2012, and July 1, 2013. All relevant subject headings and free-text terms were used to represent PTSD and incident CHD/cardiac-specific mortality, and the sets of terms were combined with AND. Terms for MEDLINE included the following: exp Stress Disorders, Traumatic/OR ptsd.tw. OR (post-traumatic OR (post adj traumatic)).tw OR posttraumatic.tw. OR acute stress disorder$.tw. OR asd.tw, exp Acute Coronary Syndrome/OR acute coronary.tw. OR acs.tw OR exp Myocardial Infarction/OR myocardial infarct$.tw. OR (mi OR ami).tw. OR (heart adj attack$).tw. OR (stemi OR nstemi).tw. OR ((preinfarction OR unstable) adj angina$).tw. These terms were adapted for the other databases. Additional records were identified by scanning the reference lists of relevant studies and reviews and by using the Related Articles feature in PubMed and the Cited Reference Search in ISI Web of Science.
Study Selection
Eligible study designs included all studies that measured PTSD caused by any traumatic event in initially healthy participants and estimated its association with subsequent CHD or cardiac-specific mortality. Two investigators (D.E., M.M.B.) independently reviewed all citations identified through the literature search, using a predefined protocol to assess study eligibility. Articles that clearly did not meet the inclusion criteria were excluded at the title and abstract level. The remaining articles were selected for full-text review; articles that did not meet the inclusion criteria were excluded. Disagreements regarding the selection of articles were resolved through discussion between the 2 reviewers, and full consensus was achieved at each stage of review.
Data Extraction
Two investigators (D.E. and M.M.B.) abstracted information about the dates and cohort sizes of the studies, characteristics of participants enrolled, methods for assessing PTSD, covariates including depression, and the strength of the association between PTSD and incident CHD and cardiac-specific mortality. One study reported estimates for men and women separately, and we thus extracted data separately from this study. Similarly, one study reported estimates for participants with and without depression separately, and we extracted data from this study separately, as well.
Data Synthesis and Analysis
Four studies reported the association of PTSD diagnosis or of a positive screen on a validated PTSD self-report screening questionnaire with incident CHD or cardiac mortality. The other 2 studies reported the risk of incident CHD associated with a 1-SD increase in PTSD symptoms on a self-report PTSD screen questionnaire. For both of those studies, we transformed the findings into a single risk estimate, corresponding to 2 SDs above each study's mean as an approximation of a positive PTSD screen—because the sample scores at that level best approximated published cutoffs for the scales used. Furthermore, although half of the studies reported associations in hazard ratios (HRs), one study reported a relative risk and 2 reported an odds ratio. For this meta-analysis, we calculated an aggregate point estimate as an HR, using the following equations to convert the relative risk and odds ratio to HRs:
and
, where RR is the relative risk, OR is the odds ratio, HR is the hazards ratio, and r is the event rate for the reference group (ie, no PTSD).
Comprehensive MetaAnalysis (version 2; BioStat Software, Engelwood, NJ) was used for completing all statistical tests and associated graphic results. Estimates of the overall risk of PTSD with incident CHD were pooled using a random-effects model. A secondary pooled analysis was performed by restricting to the 5 studies that adjusted for depression as a covariate for the association of PTSD with incident CHD. Heterogeneity assessments using Cochrane Q statistic and I preceded all meta-analytic tests. There was a statistically significant heterogeneity, justifying the use of a random-effects model to estimate and test effects. Log-transformed HRs and 95% CIs were calculated for each study using the reported effect size and estimates of the SE of each effect drawn from data reported in the article. When articles reported multiple models, we selected the model with the highest level of covariate adjustment. There were too few studies to test for moderator effects; however, we provide descriptive information concerning potential moderators below. To address the issue of publication bias, we calculated Orwin's fail-safe N and created a funnel plot of SEs by log HRs.
This work was supported by Grants HL-088117, HL-084438, and CA-156709 from the National Institutes of Health (NIH), Bethesda, MD. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of the National Center for Research Resources or the NIH. This work was supported, in part, by Columbia University's Clinical and Translational Science Awards (CTSA) Grant No. UL1RR024156 from National Center for Advancing Translational Sciences-National Center for Research Resources (NCATs-NCRR)/NIH. Dr Kronish is supported by Grant K23-HL098359 from the National Heart, Lung, and Blood Institute. Dr Shaffer is supported by Grant 12CRP8870004 from the American Heart Association and by Grant K23HL112850 from the NIH, Bethesda, MD. No funding bodies had any role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents.
Source...