1 INTRODUCTION

People affected by war, conflict, and humanitarian crises disproportionately live in low- and middle-income countries (LMICs), and they are at increased risk for psychological problems (Barbui & Tansella, 2013; Charlson et al., 2019). This is not surprising because these vulnerable populations can be exposed to significant adversity, including poverty, sexual violence, torture, war, and displacement. Despite the prevalence of mental health problems in LMICs, including people living in humanitarian contexts, it is estimated that at least 90% of adults with mental health needs do not receive minimum adequate care (Chisholm et al., 2016). Programs exist that can address mental health conditions in these typically poor resource settings (Morina et al., 2017a, 2017b), but they are often not implemented because (a) these interventions tend to only target a single diagnostic outcome (e.g., posttraumatic stress disorder (PTSD; (Neuner et al., 2008)), (b) are generally resource and time intensive (Bolton et al., 2007), and (c) often require mental health specialists who are lacking in LMICs (Bass et al., 2013). These factors preclude many LMICs and humanitarian responses from implementing much-needed mental health services that their populations require (Patel et al., 2018; Tol et al., 2012).

This situation has led to increased recognition that addressing the mental health and psychosocial needs of people in humanitarian crisis settings requires transferring a portion of mental health delivery to trained and supervised nonspecialist workers. This shift is appealing in low resource settings because it overcomes the limitation of the scarcity of mental health specialists, reduces costs, and can minimize stigma associated with specialist mental health care (Patel et al., 2018). Increasingly, programs have utilized these “task-shifting” approaches in which local providers who lack formal mental health qualifications are trained to deliver structured manualized interventions for common mental disorders. One meta-analysis found these interventions are moderately effective in LMICs (Singla et al., 2017).

The World Health Organization (WHO) has recently adopted this approach by developing a range of mental health interventions that are intended to address common mental disorders (i.e., anxiety and depression) in populations affected by adversity. Inherent in this approach is that these interventions are brief, affordable, can be easily trained to nonspecialists, and in this manner can be scaled up to achieve maximum reach in an LMIC and with limited resources. The first program of this type developed by the WHO was titled Problem Management Plus (PM+), a transdiagnostic intervention that involves five sessions that teach people skills that have been shown to be the most effective in reducing common mental disorders in controlled trials (Dawson et al., 2015). Specifically, PM+ teaches nonspecialist providers to train people with common mental disorders skills in arousal reduction, problem-solving, behavioral activation, and strengthening social supports. PM+ delivered to individuals has been shown to be effective in reducing psychological distress in adverse settings. The two large trials comprised women affected by gender-based violence in Kenya (Bryant et al., 2017) and men and women in in a conflict-affected region in Pakistan (Rahman et al., 2016). Additionally, PM+ has been delivered in a group format in Pakistan, which has also been shown to be effective relative to treatment as usual (Rahman et al., 2019).

Despite the overall efficacy of PM+ in reducing psychological distress and symptoms of common mental disorders, there is a need to better understand how it impacts specific psychological symptoms following adversity, such as the experiences of people who have lived through interpersonal trauma or humanitarian crises. One treatment study of female veterans with PTSD that employed trauma-focused cognitive behaviour therapy (TF-CBT) found that hyperarousal symptoms, and especially sleep problems and irritability, were most resistant to change with treatment (Schnurr & Lunney, 2019). This finding accords with a trial of female rape survivors with PTSD, and found that sleep problems were most resistant to TF-CBT (Larsen et al., 2019). The finding that many people continue to experience functional difficulties following remission of PTSD diagnosis (Bryant et al., 2016) points to the need to better understand psychological difficulties that persist after treatment. Although some inroads have been made in relation to trauma-focused psychotherapies, there is a need to understand how scalable interventions, such as PM+, address specific PTSD symptoms in humanitarian and development contexts where the prevalence of PTSD is elevated (Charlson et al., 2019). This is relevant for PM+ because although it is not a trauma-focused intervention, previous trials have shown that PM+ leads to significant reductions in PTSD severity, in adversity-affected LMICs and conflict settings (Rahman et al., 2016, 2019). In both situations, many people in these category do suffer from PTSD, or experience events that could lead to PTSD. In turn, understanding the residual symptoms of PTSD following PM+ would provide information on how the intervention reduces PTSD symptoms as well as inform the development of future scalable interventions.

To this end, this study conducts an individual-patient data meta-analysis (IPD-MA) of the three large controlled trials that have been published to determine which symptoms of PTSD are most persistent after provision of PM+ (Bryant et al., 2017; Rahman et al., 2016, 2019). It further aims to inform the potential of PM+ to be a useful brief intervention across humanitarian and adversity settings where PTSD is considered one of the common mental disorders experienced.

2 METHODS 2.1 Materials and methods

This IPD analysis focused on prospectively registered randomized clinical trials (RCT) of individual PM+ in Pakistan ((Rahman et al., 2016), ACTRN12614001235695) and Kenya ((Bryant et al., 2017), ACTRN12616000032459), and group PM+ in Pakistan ((Rahman et al., 2019), ACTRN12616000037404). Details about the original studies have been published elsewhere (Chiumento et al., 2017; Sijbrandij et al., 2015, 2016). Initial studies of the PM+ manual required approval of their research protocols through the WHO Ethical Review Committee in addition to the relevant governing bodies in countries that the trials were conducted. All participants provided written informed consent before taking part in the trials.

2.2 Context and participants

Between November 2014 and August 2016 three RCTs were conducted to evaluate the effectiveness of PM+ in reducing the levels of psychological distress experienced by participants. A secondary outcome of each trial included PTSD symptoms. Data from these studies were combined into a single data set.

The study characteristics are presented in Table 1. In the Pakistan individual PM+ study, participants (N = 346) were recruited from three primary health care centres in the conflict-affected urban city of Peshawar (Rahman et al., 2016). Participants received clinical assessments from their physicians; participants who were deemed to be suffering from psychological distress were referred to the study. The Pakistan group-based PM+ study was conducted in two rural council districts of Swat (Rahman et al., 2019), and adult women (N = 612) were randomly screened in the community and invited into the study if they screened positively for psychological distress. The Kenya study recruited female participants (N = 522) through random sampling in peri-urban neighbourhoods of Nairobi, of whom 81% had experienced gender-based violence (Sijbrandij et al., 2016). In all three studies the initial screening procedure used the General Health Questionnaire-12 (GHQ-12; (Goldberg & Williams, 1988; Minhas & Mubbashar, 1996)) to identify psychological distress and the WHO-Disability Assessment Schedule 2.0 (WHODAS 2.0), which is a 12-item measure of general functioning (WHO, 2010). The common inclusion criteria across all three studies were (i) adults aged 18–60 years, (ii) score of ≥3 on the GHQ-12, and (iii) score of ≥17 on the WHODAS 2.0. The exclusion criteria were identical across all three studies: (i) imminent risk for suicide, (ii) severe mental illness (e.g., psychotic disorders, substance use disorders), and (iii) significant cognitive and neurological impairment (e.g., severe intellectual disability).

Table 1. Study characteristics Studies Study design Country Recruitment Inclusion criteria Intervention Control group Follow-up (weeks) N (Nint) Outcome Bryant et al. (2017) RCT Kenya Community screening Age = 18–60, GHQ ≥ 3; WHODAS ≥ 17 PM+ ETAU 6 (T1), 18 (T2) T0 = 522 (260) PCL-5 T1 = 419 (205) T2 = 389 (191) Rahman et al. (2016) RCT Pakistan Primary health care centres Age = 18–60, GHQ ≥ 3; WHODAS ≥ 17 PM+ ETAU 6 (T1), 18 (T2) T0 = 346 (172) PCL-5 T1 = 137 (60) T2 = 306 (146) Rahman et al. (2019) cRCT Pakistan Community screening Age = 18–60, GHQ ≥ 3; WHODAS ≥ 17 gPM+ ETAU 6 (T1), 18 (T2) T0 = 612 (306) PCL-5 T1 = 598 (298) T2 = 578 (288) Abbreviations: cRCT, clustered randomized controlled trial; ETAU, enhanced treatment as usual; GHQ, General Health Questionnaire; gPM+, Group Problem Management Plus; N, number of participants; Nint, number of participants randomized into intervention arm; PCL-5, PTSD checklist for DSM-5; PM+, problem management plus; RCT, randomized controlled trial; T0, Entry into the study (0 weeks); T1, 6-week follow up, T2, 18-week follow up; WHODAS, World Health Organization Disability Adjustment Scale 2.0. 2.3 Interventions 2.3.1 Problem Management Plus

PM+ integrates problem-solving and behavioural activation techniques that are amenable to low-intensity delivery and are evidence-based (Bennett-Levy & Farrand, 2010; Cuijpers et al., 2007; Farchione et al., 2012; van't Hof et al., 2011). The individual format is delivered over five weekly sessions of 90 min duration; the group format is delivered over five weekly sessions of 120 min duration. Clients are taught: (i) arousal reduction; (ii) problem management; (iii) behavioural activation, and (iv) skills to strengthen social support. The PM+ programme has been made available publicly by the WHO (WHO, 2016). Facilitators received 8 days of training in basic counselling skills and the delivery of the PM+ intervention. For the group modality, facilitators are additionally taught group facilitation skills. Following training, the PM+ providers are required to complete two practice cases whilst attending weekly group supervision sessions focused on fidelity to the intervention, challenges faced, and helpers self-care.

2.3.2 Enhanced treatment as usual (ETAU)

The ETAU arms of all three studies included psychoeducation and referral information for locally available resources for health, vocational training, and other relevant services. Participants were provided with individual, verbal feedback regarding the results of the assessments, brief psychoeducation and referral services to primary health care centres in both Kenya and the Swat trial, and participants met with a primary healthcare physician at least once where they received assessment feedback and a psychoeducation session in the Peshawar study.

2.4 Measure

PTSD symptoms were measured using the PTSD Checklist for DSM-5 (PCL-5) (Weathers et al., 2013). The PCL-5 is a 20-item checklist corresponding to the 20 symptoms for PTSD as defined in DSM-5. There are five items corresponding to intrusion symptoms, two to avoidance symptoms, seven to negative alterations in cognitions and moods, and six to alterations in arousal and reactivity. Items are rated on a 5-point Likert scale (0 = not at all, 4 = extremely) and total scores are obtained by summing the responses for individual items (range: 0–80). Across all three studies, the PCL-5 was adapted to ask for symptoms in the past week (instead of month) to enhance sensitivity to change. The PCL-5 measure was adapted in all three studies in accordance with gold-standard translation practices, with each item being translated and back-translated by accredited translators (Bontempo, 1993). In Kenya, items were translated into Kiswahili and in Pakistan, items were translated into Urdu. The comprehensibility of the adapted versions were piloted in feasibility trials which took place before the RCTs (Dawson et al., 2016; Rahman et al., 2016). The psychometric properties of the PCL-5 have been widely studied and has been shown to have high levels of diagnostic accuracy and internal consistency (Blevins et al., 2015; Bovin et al., 2016), including in LMIC (Mughal et al., 2020). Symptoms were rated as present if the respondent indicated that they were at least moderately bothered by them, as indicated by a score ≥2 on individual items (Weathers et al., 2013). Symptom retention was defined as the conditional probability of retaining a symptom at posttreatment and 3-month follow-up compared to before receiving treatment. A probable diagnosis of PTSD is made if the total score on the PCL-5 was ≥33 (Weathers et al., 2013).

2.5 Statistical analysis

Means and SD are reported for continuous variables while frequencies and percentages are reported for categorical variables. Missing outcome data at postintervention and follow-up assessments were estimated using multiple imputation (MI); data were assumed to be missing-at-random. As the PCL-5 results are scored on a Likert scale, MI was conducted by chained equations using ordered logistic regression (mi impute chained ologit in Stata). We analysed the effects of PM+ versus ETAU on the conditional probability that participants retained symptoms following intervention using a multilevel logistic regression, with random effects for each trial; separate models were run for each of the 20 PTSD symptoms. A complete-case sensitivity analysis was conducted to explore robustness of results from the MI using only participants who completed the post and follow-up assessments. Results are reported as the estimated odds ratio (OR) and the 95% confidence intervals (CI) for each symptom are presented. All analyses were conducted using Stata 13 (StataCorp., 2013).

3 RESULTS

Descriptive information about study participants is presented in Table 2. In total, 1480 participants were randomised across the three trials, with 738 receiving the PM+ intervention and 742 receiving ETAU. There were 1154 (78.0%) participants who completed the postintervention assessment and 1,273 (86.0%) who completed the 3-month follow-up assessments. The average age of the study participants was 35.31 years (SD = 11.84). The studies comprised 1407 women (95%).

Table 2. Demographic and clinical characteristics of the study population (n = 1480) Kenya (N = 522) Pakistan (N = 346) SWAT (N = 612) Combined (N = 1480) Randomization PM+, N 260 172 306 738 ETAU, N 262 174 306 742 Age, mean (SD) 35.95 (13.66) 33.08 (11.80) 36.26 (9.88) 35.41 (11.84) Education (years), mean (SD) 8.50 (3.98) - - - Education - No schooling, N (%) - 203 (58.7%) 503 (82.2%) - Primary (6 years), N (%) - 42 (12.1%) 51 (8.3%) - Middle (8 years), N (%) - 26 (7.5%) 22 (3.6%) - Matriculate (10 years), N (%) - 29 (8.4%) 18 (2.9%) - Intermediate, college and university (11–16 years), N (%) - 46 (13.3%) 15 (2.5%) - Gender, N (%) Female 522 (100) 273 (78.90) 612 (100) 1407 (95.07) Male 0 (0) 73 (21.10) 0 (0) 73 (4.93) PCL-5, mean (SD) PM+ ETAU PM+  ETAU PM+ ETAU PM+ ETAU Baseline 32.86 (20.10) 30.67 (19.17) 31.81 (14.34) 30.30 (14.22) 17.26 (11.04) 19.75 (11.70) 26.14 (17.21) 26.06 (16.14) Postintervention 9.62 (12.54) 13.36 (15.74) 12.86 (10.69) 18.73 (11.11) 8.66 (9.17) 12.41 (8.29) 9.75 (10.76) 13.73 (12.03) 3-month 7.56 (11.42) 7.41 (10.40) 11.90 (9.18) 17.72 (9.35) 8.48 (6.47) 10.92 (7.96) 9.00 (9.01) 11.52 (9.88) PTSD diagnosis Baseline 126 (48.65) 115 (44.40) 77 (44.77) 77 (44.25) 27 (8.82) 41 (17.60) 230 (31.21) 233 (31.53) Postintervention 13 (6.40) 31 (14.49) 9 (7.89) 12 (12.63) 10 (3.33) 8 (2.68) 32 (5.19) 51 (8.39) 3-month 9 (4.71) 6 (3.02) 6 (4.11) 12 (7.50) 2 (0.69) 7 (2.41) 17 (2.72) 25 (3.85) PTSD diagnosis retention Postintervention, N (%) 12 (12.24) 28 (28.00) 5 (9.80) 7 (15.56) 4 (14.81) 2 (5.00) 21 (11.93) 37 (20.00) 3-month, N (%) 7 (7.14) 5 (5.95) 4 (6.35) 7 (10.61) 0 (0.00) 4 (11.11) 11 (5.95) 16 (8.60) Abbreviations: ETAU, enhanced treatment as usual; PM+, problem management plus.

The average PCL-5 score at baseline was 26.1 (SD = 16.76) while 463 (31.3%) participants scored ≥33, indicating that it was probable they meet diagnostic criteria for PTSD. There were 230 (31.21%) and 233 (31.53%) persons who scored ≥33 on the PCL-5 at baseline randomized in the PM+ and ETAU arms, respectively. The Kenya and individual PM+ Pakistan trials had higher baseline PCL-5 scores (31.76 and 31.05, respectively) and numbers of people with probable PTSD (46% and 45%, respectively) compared to the baseline PTSD levels in the Swat study (M = 18.51, 11% > 33). At postintervention assessment, 5.19% of people who received PM+ and 8.39% of people who received ETAU reported a score ≥33 and at 3-month follow-up, 2.72% (PM+) and 3.85% (ETAU) had a score ≥33. For those who scored ≥33 at baseline, 11.93% and 20% (postintervention), and 5.95% and 8.60% (3-month follow-up) maintained a score of ≥33 in the PM+ and ETAU arms, respectively.

Figure 1 illustrates the proportions of participants that were randomised to receive PM+ and who scored positive for symptom presence at baseline, postintervention, and 3-month follow-up. The most prevalent symptoms at baseline were irritability/aggression, difficulty concentrating, and sleeping (i.e., arousal symptoms). The least prevalent symptoms were inability to recall the trauma, exaggerated blame, and risky/destructive behavior. Following treatment, at both postintervention and 3-month assessment, arousal symptoms remained the most prevalent, although the proportions of individuals who were positive for these arousal symptoms decreased markedly. Mood symptoms were among the lowest reported following PM+ at both postintervention time points, although risky behavior (an arousal symptom) was the least prevalent at all three time points.

Problem management plus—symptom presence for each posttraumatic stress disorder symptom at baseline, postintervention assessment, and 3-month follow up

Figure 2 illustrates the prevalence of PTSD symptoms for those who received ETAU. The baseline rates were similar to participants who received PM+. Even though this group showed marked reductions of symptoms of PTSD, the prevalence of all symptoms were higher than the PM+ participants; particularly mood and arousal symptoms. This was also true at 3-month follow-up.

Enhanced treatment as usual—symptom presence for each posttraumatic stress disorder symptom at baseline, postintervention assessment, and 3-month follow up

Tables 3 and 4 present the proportions of individuals who retained symptoms following intervention (Table 3) and at 3 months (Table 4); they also present the ORs and corresponding 95% CI. Across all 20 symptoms, there were higher rates of amelioration of symptoms for those who received PM+ compared to those who received ETAU, with the largest effects seen in mood and arousal symptoms. The symptoms with the greatest absolute remission following the PM+ intervention were avoidance of thoughts and behaviours. When comparing the conditional probabilities of symptom retention immediately following interventions and at 3-month follow-up, the ORs point estimates indicate that participants receiving the ETAU intervention had higher odds of symptom retention that those who received PM+, when compared to baseline. Specifically, the odds of symptoms amelioration were significantly higher for those who received PM+ compared to ETAU in: physical reactivity, avoidance of thoughts, avoidance of people/places, negative affect, decreased interest, difficulty with positive affect, irritability/aggression, heightened startle reaction, difficulty concentrating, and difficulty sleeping. At 3-month follow-up, significant reductions in symptom rates were maintained for: physical reactivity, avoidance of people/places, decreased interest, irritability/aggression, heightened startle reactions, difficulty concentrating, and difficulty sleeping. In addition, at 3-month follow-up, the following symptoms had significantly higher odds of being reduced in those who received PM+ compared to ETAU: negative thoughts/assumptions and hypervigilance. There were no changes in the results of the sensitivity analysis; this indicates robustness of the primary analysis strategy.

Table 3. Percentage of participants retaining each posttraumatic stress disorder symptom after treatment by treatment type—postassessment PM+ ETAU Multiple imputation Complete case Item % N % N OR 95% CI OR 95% CI B–1 Intrusive memories 15.25 34 21.61 51 1.44 0.92–2.26 1.53 0.95–2.49 B–2 Nightmares 15.03 26 22.83 42 1.53 0.90–2.62 1.66 0.96–2.87 B–3 Flashbacks 14.37 37 20.56 37 1.42 0.82–2.45 1.54 0.88–2.69 B–4 Emotional distress 23.50 51 27.87 68 1.27 0.85–1.90 1.28 0.83–1.96 B–5** Physical reactivity 21.78 49 29.66 70 1.57 1.05–2.35 1.68 1.09–2.60 C–1** Avoidance of thoughts 12.89 25 22.39 45 1.89 1.13–3.17 2.04 1.18–3.52 C–2** Avoidance of people/places 11.48 21 20.99 38 2.01 1.10–3.66 2.13 1.17–3.88 D–1 Dissociative amnesia 8.39 13 15.15 25 1.75 0.93–3.31 1.96 0.96–3.99 D–2 Negative thoughts/assumptions 18.58 34 21.80 46 1.23 0.76–1.97 1.26 0.76–2.08 D–3 Exaggerated blame 17.76 27 23.35 39 1.22 0.72–2.09 1.41 0.81–2.44 D–4** Negative affect 18.58 32