‘P’ is for population

The population included individuals aged ≥16 years experiencing PTSD symptoms. Studies were eligible if at least 70% of participants met PTSD diagnostic criteria, assessed through clinical interviews. No exclusions were based on the type of trauma experienced, symptom severity or duration, time since the traumatic event or comorbidities.

‘I’ is for intervention

Interventions included online or mobile-delivered cognitive therapy, behavioural therapy or combined cognitive–behavioural therapy (CBT) programmes. Eye movement desensitisation and reprocessing and mindfulness-based therapies were excluded, except for mindfulness-based I-C/BT. Therapist guidance was limited to 5 h delivered across a range of modalities (e.g. messaging, telephone, face-to-face contact), with no restrictions on frequency. Treatment durations were unrestricted but averaged approximately 9 weeks. Of the 13 studies, 4 used <8 sessions and 3 used unstandardised, ad hoc sessions, despite the National Institute for Health and Care Excellence (NICE) recommendation of 8–12 sessions for PTSD treatment (NICE 2018).

‘C’ is for comparison

All 13 included studies were RCTs, a design typically regarded as providing high-quality evidence. Comparator interventions varied and included face-to-face CBT, face-to-face non-CBT, psychological therapies, waiting list/TAU, internet-delivered psychoeducation and internet-delivered non-CBT interventions. The present commentary will focus on the majority of studies, which used waiting-list/TAU comparators (10/13).

‘O’ is for outcome

Primary outcomes included PTSD symptom severity, assessed via standardised scales (10 out of 13 studies) and clinician interviews (Box 1), and drop-out rates. Secondary outcomes included the proportion of participants meeting PTSD diagnostic thresholds, depression and anxiety symptom severity, cost-effectiveness, adverse events, treatment acceptability and quality of life. Outcomes were reported chronologically: immediately post-treatment (the primary interest) and <6 months, between 6 and 12 months and over 12 months post-treatment.

BOX 1

Post-traumatic stress disorder (PTSD) outcomes: tools at a glance

Randomised controlled trials included in the Cochrane Review employed a range of standardised measures to assess PTSD outcomes, exhibiting different strengths and limitations. Understanding their properties can help clinicians interpret findings and choose appropriate instruments in practice

Most of these tools were developed under DSM-4 criteria and adapted for DSM-5; some instruments do not fully capture DSM-5 cognitive and mood symptoms. The self-report tools (IES-R, PCL-C, PDS-5) are user-friendly and sensitive to change, but rely on accurate patient recall and may misestimate symptom severity. In contrast, the CAPS-5 and PSS-I provide clinical depth and diagnostic clarity, at the cost of administration time and clinician training.

Search results

To address their objective, Simon et al conducted a comprehensive search of studies published between March 2018 and June 2020 across seven databases, in line with Cochrane recommendations (Higgins Reference Higgins, Thomas and Chandler2019). The search strategy included grey literature (unpublished studies, dissertations and theses), reference lists and ongoing work. This yielded 13 RCTs, accounting for 808 participants. Only three new studies were added to an earlier version of the review (Lewis Reference Lewis, Roberts and Bethell2018). More RCTs on the topic have become available since 2020, perhaps reflecting the pandemic-driven demand for remote interventions not yet captured in this search. Experimental group samples ranged from 3 to 79 participants. The higher prevalence of PTSD in females was accurately reflected, as women constituted 18.75–100% of samples.

The review thoughtfully evaluated the characteristics and methodological quality of the included studies. Certainty of evidence was assessed using the GRADE approach (Box 2), reflecting a clear effort to account for factors that may affect confidence in the meta-analytic findings.

BOX 2

Certainty of evidence in meta-analyses

Meta-analyses routinely report gradings of the certainty of evidence in ‘Results’ and a ‘Summary of findings table’. Essentially, this can be considered as the trustworthiness of a study’s observations. Assessing the quality of evidence allows researchers to make informed judgements on whether a study’s effect sizes reflect treatment effectiveness accurately, or whether other factors, such as study design, may have reduced precision. This is especially important when evaluating recommendations produced from these estimates of effect, which can result in large consequences for healthcare if incorrect.

The Grades of Recommendation Assessment, Development and Evaluation (GRADE) working group (www.gradeworkinggroup.org/) has a standardised framework within which certainty of evidence can be assessed. Randomised controlled trials and non-randomised studies are initially (before assessment) considered to have ‘high’ and ‘low’ certainty respectively.

Authors then consider each outcome with regard to the following domains, which can lead to a decrease in the certainty of evidence:

• • risk of bias or limitations in the design and implementation: whether there are concerns about the design and conduct of the studies (e.g. poor masking (‘blinding’) or randomisation, incomplete outcome data);

• • inconsistency: whether there is unexplained variability in results across different studies (e.g. in the direction of results or magnitude of effect sizes);

• • indirectness: whether the evidence directly addresses the research question;

• • imprecision: whether the results are imprecise (e.g. large confidence intervals), which could be due to issues such as too small a sample size;

• • publication bias: whether studies reporting a statistically significant effect are more likely to be published than those that fail to reach significance, resulting in an overestimation of effects when considering standardised mean differences.

Usually, certainty ratings drop by one level for each factor, or two in the case of particularly serious issues.

Each observation is also assessed with the following domains, which can increase certainty of evidence:

• • large magnitude of effect: large effect size reported;

• • dose–response gradient: a reported relationship between the dose of an intervention and the strength of its response, suggesting causality;

• • opposing plausible residual confounding or bias: when all plausible confounds and/or biases would reduce the observed effect, despite it remaining significant.

When these factors are considered, the body of evidence can be categorised into the following levels of certainty with their respective implications:

Is I-C/BT better than waiting list/usual care?

Overall, results suggest that I-C/BT was more effective than waiting list/usual care at reducing PTSD, depression and anxiety symptoms immediately post-treatment (Box 3).

For the primary outcome of PTSD symptom severity immediately after treatment, I-C/BT showed a moderate benefit over waiting list/TAU (s.m.d. = −0.61, 95% CI −0.93 to 0.29; 10 studies, 608 participants). However, this benefit did not persist up to 6 months post-treatment (s.m.d. = −0.45, 95% CI −1.29 to 0.39; 4 studies, 154 participants). Regarding drop-out rates, no difference was found between I-C/BT and waiting-list/TAU groups (RR = 1.25, 95% CI 0.97–1.60; I ² = 34%; 9 studies, 634 participants).

Regarding secondary outcomes, I-C/BT did not significantly reduce the risk of meeting PTSD diagnostic criteria post-treatment compared with waiting list (RR = 0.53, 95% CI 0.28 to 1.00; 1 study, 62 participants). However, I-C/BT was more effective than waiting list at reducing depressive (s.m.d. = −0.51, 95% CI −0.97 to −0.06; 7 studies, 473 participants) and anxiety symptoms (s.m.d. = −0.61, 95% CI −0.89 to −0.33; 5 studies, 345 participants). One study found that the anxiolytic effects persisted 6 months post-treatment (m.d. = −12.59, 95% CI −20.74 to −4.44; 42 participants).

All findings were rated as low or very low certainty, primarily because of risk of bias, indirectness of evidence, serious inconsistency, imprecision of effect estimates and possible publication bias. There was consistently high risk of performance bias across all included studies, as neither participants nor therapists could be masked to treatment. However, eight out of the ten studies reported adequate masking of outcome assessors, partly mitigating this limitation.

Does therapist involvement or trauma focus matter?

Findings suggest that therapist guidance may meaningfully enhance I-C/BT effectiveness. Interventions incorporating therapist support were associated with greater reductions in PTSD symptoms (s.m.d. = −0.78, 95% CI −1.09 to −0.47; 8 studies, 439 participants), whereas unguided interventions showed no meaningful benefit over waiting list (s.m.d. = −0.09, 95% CI −0.39 to 0.22; 2 studies, 169 participants).

Similarly, trauma-focused I-C/BT appeared more effective than non-trauma-focused approaches. Trauma-focused interventions showed greater symptom improvement (s.m.d. = −0.94, 95% CI −1.24 to −0.65; 6 studies, 342 participants), whereas non-trauma-focused treatments showed no difference compared with waiting-list controls (s.m.d. = −0.16, 95% CI −0.41 to 0.08; 4 studies, 266 participants).