Effectiveness of abbreviated trauma-focused cognitive behavioural therapy for South African adolescents: a randomized controlled trial

ABSTRACT

Background: In low- and middle- income countries (LMICs) trauma exposure among youth is high, but mental health services are critically under-resourced. In such contexts, abbreviated trauma treatments are needed.

Objective: To evaluate the efficacy of an abbreviated eight-session version of Trauma-Focused Cognitive Behavioural Therapy (TF-CBT) for improving posttraumatic stress disorder (PTSD) and depression symptoms in a sample of South African adolescents.

Method: 75 trauma-exposed adolescents (21 males, 54 females; mean age = 14.92, range = 11–19) with posttraumatic stress disorder (PTSD) symptoms were randomly assigned to eight sessions of TF-CBT or to usual services. At baseline, post-treatment and three-month follow-up, participants completed the Child PTSD Symptom Scale for DSM 5 (CPSS-5) and the Beck Depression Inventory II (BDI-II). The trial is registered on the Pan African Trial Registry (PACTR202011506380839).6.

Results: 95% of TF-CBT participants completed treatment while only 47% of TAU participants accessed treatment. Intention-to-treat analyses found that the TF-CBT group had a significantly greater reduction in CPSS-5 PTSD symptom severity at post-treatment (Cohen’s d = 0. 60, p < .01) and three-month follow-up (Cohen’s d = 0.62, p < . 01), and a greater reduction in the proportion of participants meeting the CPSS-5 clinical cut-off for PTSD at both time points (p = .02 and p = .03, respectively). There was also a significantly greater reduction in depression symptom severity in the TF-CBT group at post-treatment (Cohen’s d = 0.51, p = .03) and three-month follow-up (Cohen’s d = 0.41, p = .05), and a greater reduction in the proportion of TF-CBT participants meeting the BDI clinical cut-off for depression at both time points (p = .02 and p = .03, respectively).

Conclusion: The findings provide preliminary evidence of the efficacy of an abbreviated eight-session version of TF-CBT for reducing PTSD and depression symptoms in a LMIC sample of adolescents with multiple trauma exposure.

HIGHLIGHTS

• Trauma-Focused Cognitive Behavioural Therapy (TF-CBT) is one of the leading evidence-based treatments for child and adolescent posttraumatic stress, but an abbreviated version has not been evaluated in low- and middle-income countries (LMICs).

• At post-treatment and three-month follow-up, eight sessions of TF-CBT were more effective than treatment-as-usual in reducing posttraumatic stress and depression symptoms in an LMIC sample of South African adolescents exposed to multiple traumas.

• Effect sizes were similar to those reported for standard length TF-CBT, indicating that abbreviated TF-CBT may be a suitable option for resource-constrained child and adolescent mental health services in LMICs.

Antecedentes: En los países de ingresos bajos y medios (LMIC, por sus siglas en inglés), la exposición al trauma entre los jóvenes es alta, pero los servicios de salud mental tienen una escasez crítica de recursos. En tales contextos, se necesitan tratamientos de trauma abreviados.

Objetivo: Evaluar la eficacia de una versión abreviada de ocho sesiones de la Terapia cognitiva conductual centrada en el trauma (TCC-CT) para mejorar los síntomas del trastorno de estrés postraumático (TEPT) y de depresión en una muestra de adolescentes sudafricanos.

Método: 75 adolescentes expuestos a trauma (21 hombres, 54 mujeres; edad media = 14,92, rango = 11–19) con síntomas de trastorno de estrés postraumático (TEPT) fueron asignados aleatoriamente a ocho sesiones de TCC-CT o a los servicios habituales (TAU por sus siglas en inglés). Al inicio del estudio, después del tratamiento y a los tres meses de seguimiento, los participantes completaron la Escala de síntomas de TEPT infantil para el DSM 5 (CPSS-5) y el Inventario de depresión de Beck II (BDI-II, por sus siglas en inglés). El ensayo está registrado en el Registro Panafricano de Ensayos (PACTR202011506380839).

Resultados: El 95 % de los participantes de TCC-CT completó el tratamiento, mientras que solo el 47 % de los participantes de TAU accedió al tratamiento. Los análisis por intención de tratamiento encontraron que el grupo TCC-CT tuvo una reducción significativamente mayor en la gravedad de los síntomas de TEPT CPSS-5 después del tratamiento (d de Cohen = 0,60, p < 0,01) y a los tres meses de seguimiento (d de Cohen = 0,62, p < 0,01), y una mayor reducción en la proporción de participantes que cumplieron con el límite clínico CPSS-5 para TEPT en ambos puntos temporales (p = 0,02 y p = 0,03, respectivamente). También hubo una reducción significativamente mayor en la gravedad de los síntomas de depresión en el grupo de TCC-CT después del tratamiento (d de Cohen = 0,51, p = 0,03) y a los tres meses de seguimiento (d de Cohen = 0,41, p = 0,05), y una mayor reducción en la proporción de participantes de TCC-CT que alcanzaron el límite clínico BDI para la depresión en ambos puntos temporales (p = 0,02 y p = 0,03, respectivamente).

Conclusión: Los hallazgos brindan evidencia preliminar de la eficacia de una versión abreviada de ocho sesiones de TCC-CT para reducir el TEPT y los síntomas de depresión en una muestra de adolescentes de LMIC con exposición a múltiples traumas.

背景：在低收入和中等收入国家 (LMIC)，年轻人创伤暴露程度很高，但心理健康服务资源严重不足。 在这种情况下，需要简短的创伤治疗。

目的：在南非青少年样本中评估简化版八疗程聚焦创伤的认知行为疗法 (TF-CBT)对改善创伤后应激障碍 (PTSD) 和抑郁症状的有效性。

方法：75 名有创伤后应激障碍 (PTSD) 症状的创伤暴露青少年（21 名男性，54 名女性；平均年龄 = 14.92，范围 = 11-19）被随机分配到八疗程TF-CBT 或常规服务。 在基线、治疗后和三个月的随访中，参与者完成了 DSM 5 儿童 PTSD 症状量表 (CPSS-5) 和贝克抑郁量表 II (BDI-II)。 该试验已在泛非试验注册中心注册(PACTR202011506380839)。

结果：95% 的 TF-CBT 参与者完成了治疗，而只有 47% 的 TAU 参与者接受了治疗。治疗意向分析发现，TF-CBT 组在治疗后 (Cohen's d = 0. 60, p < .01) 和三个月随访时 CPSS-5 PTSD 症状严重程度显著降低 ( Cohen 的 d = 0.62， p <.01)，并且在这两个时间点满足 CPSS-5 PTSD 临床临界值的参与者比例都大幅下降（分别为 p = .02 和 p = .03）。TF-CBT 组的抑郁症状严重程度在治疗后（Cohen's d = 0.51， p = .03）和三个月的随访（Cohen's d = 0.41， p = .05）时也有显著降低，并且在两个时间点达到抑郁症 BDI 临床临界值的 TF-CBT 参与者比例大幅降低（分别为 p = 0.02 和 p = 0.03）。

结论：研究结果为简化版八疗程TF-CBT 在 LMIC多重创伤暴露青少年中减少 PTSD 和抑郁症状的有效性提供了初步证据，。

KEYWORDS:

• Trauma-focused cognitive behavioral therapy
• abbreviated
• adolescents
• South Africa
• posttraumatic stress disorder
• depression

PALABRAS CLAVE:

• Terapia cognitiva conductual centrada en el trauma
• abreviada
• adolescentes
• Sudáfrica
• trastorno de estrés postraumático
• depresión

关键词:

• 聚焦创伤的认知行为疗法
• 简化版
• 青少年
• 南非
• 创伤后应激障碍
• 抑郁

Children and adolescents living in low- and middle-income countries (LMICs) experience higher rates of multiple trauma (Le et al., 2016) and posttraumatic stress disorder (PTSD) (Yatham et al., 2018) than youth in high income countries (HICs) but are far less likely to receive treatment (Patel et al., 2013). Untreated childhood trauma yields a substantial mental health and economic burden in HICs (Fang et al., 2012). This is likely to be exacerbated in LMICs, where resources are already scarce. Therefore, offering effective treatments to trauma-affected children and adolescents in LMICs is an important global mental health priority. However, the evidence base for youth trauma interventions in these settings remains vastly under-represented compared to HICs (Uppendahl et al., 2020). Further, standard-length traumatic stress interventions developed in higher income settings may not be easily transferable to LMICs, where child and adolescent mental health services (CAMHS) are often critically under-resourced (Lu et al., 2018; Morris et al., 2011). In the constrained health systems of LMICs, it is crucial to identify brief interventions for trauma-affected youth that can balance effectiveness and efficiency (Yatham et al., 2018).

Children and adolescents in South Africa experience high levels of trauma exposure in both their homes and communities. A national survey with South African adolescents found prevalence rates of 12% for sexual abuse, 18% for physical abuse and 25% for family violence exposure (Ward et al., 2018). Further, most adolescents have been exposed to violence in their communities (Kaminer et al., 2013; Otwombe et al., 2015). The prevalence of poly-victimisation (exposure to multiple forms of violence) among young South Africans is over 90% (Collings et al., 2014; Herrero Romero et al., 2021), far exceeding rates in both HICs (Aho et al., 2016; Mossige & Huang, 2017) and other LMICs (Le et al., 2016). Like elsewhere, cumulative trauma exposure in South African adolescents is associated with an increased risk of both PTSD and depression (Collings et al., 2014; Stansfeld et al., 2017). Yet CAMHS in South Africa are inadequate and severely under-resourced: South Africans under 18 years old represent only 6% of all mental health outpatient visits and have a high unmet need for mental health services (Docrat et al., 2019).

There is, therefore, an urgent need to develop trauma interventions for South African youth that are brief enough to be implemented in an over-burdened CAMHS, but robust enough to effectively address traumatic stress in a context of multiple trauma exposure and ongoing socioeconomic precarity. Unfortunately, the evidence base for trauma treatments for South African youth is poor. Only one randomised controlled trial (RCT) has been conducted in the country to date, which found 7–14 sessions of prolonged exposure therapy to be more effective than supportive counselling in addressing PTSD in adolescents (Rossouw et al., 2018). Expanding this local evidence base is critical for meeting the needs of a highly trauma-exposed youth population. We aimed to identify a trauma intervention for youth that has been culturally adapted and shown to be effective in LMIC settings and adapt it for the South African context.

Trauma-Focused Cognitive Behavioural Therapy (TF-CBT) was developed to treat PTSD, depression and anxiety symptoms in children and adolescents who have been exposed to traumatic events (Cohen et al., 2017). It is currently considered to be one of the leading trauma treatments for youth, based on its extensive evidence base (Bennett et al., 2020; Morina et al., 2016). In TF-CBT, a stabilisation phase focused on psychoeducation, affect regulation skills and cognitive coping skills is followed by narration and re-processing of traumatic memories through graduated exposure and processing of dysfunctional trauma-related beliefs (Cohen et al., 2017). There is also a focus on safety planning to reduce risk of re-exposure to trauma, which is essential in a context of ongoing violence exposure like South Africa. A particular strength of TF-CBT is that it includes caregivers throughout the intervention, with a focus on strengthening attachment relationships between youth and caregivers. Although the evidence base for TF-CBT is largely based in HICs (Thomas et al., 2020), several RCTs have found TF-CBT to be effective in a range of LMIC settings (Dorsey et al., 2020; McMullen et al., 2013; Murray et al., 2015; O'Callaghan et al., 2013), making it a strong choice for the South African context. Other trauma treatments previously used with youth in high-resource settings, such as Prolonged Exposure and Eye Movement Desensitisation and Reprocessing, have seldom been evaluated in LMIC settings (Morina et al., 2016).

Standard TF-CBT is between 12−15 sessions (Cohen et al., 2017). This could be challenging to implement at scale in South Africa’s over-burdened CAMHS, where limited resources result in high workloads for treatment providers (Mokitimi et al., 2019) and transport costs to reach mental health services are a burden to many service users (Mkabile & Swartz, 2020). However, two previous studies in the United States found an abbreviated eight-session TF-CBT format to be effective in reducing traumatic stress symptoms in children exposed to sexual abuse (Deblinger et al., 2011) and intimate partner violence (Cohen et al., 2011). To date, an eight session TF-CBT model has not been evaluated in LMIC settings.

Previously, we developed an eight-session version of TF-CBT for the South African context and piloted it to assess its acceptability to local participants and to establish fidelity of treatment delivery (Kaminer et al., 2022). The aim of the present study was to evaluate the efficacy of this eight-session version of TF-CBT compared with treatment as usual (TAU) in reducing symptoms of posttraumatic stress and depression in South African adolescents. We hypothesise that abbreviated TF-CBT is more effective than TAU in reducing PTSD and depression symptom severity and reducing the proportion of participants at clinical risk of being diagnosed with PTSD and depression in South Africa.

1. Methods

1.1. Study design

The study used a two-arm, single-blind RCT design with participants allocated to eight sessions of TF-CBT or to TAU. The primary outcome was mean change in PTSD symptom severity assessed with the Child PTSD Symptom Scale for DSM 5 (CPSS-5; Foa et al., 2018). As a secondary outcome we also assessed change in the proportion of cases meeting the clinical cut-off for PTSD on the CPSS-5. As major depressive disorder (MDD) is commonly comorbid with PTSD (Flory & Yehuda, 2015), we assessed change in depression symptom severity on the Beck Depression Inventory II (Beck et al., 1996), and change in the proportion of cases meeting the BDI II clinical cut-off for depression, as secondary outcomes. All participants completed a post-treatment assessment after eight sessions of TF-CBT or eight weeks in the TAU condition, and again at a three-month follow-up. The same assessor, a six-year university qualified clinical psychologist, conducted all the assessments for all participants across all three time points and was blinded to treatment condition. The trial was registered during study implementation on the Pan African Trial Registry (PACTR202011506380839).

1.2. Participants

An a priori power analysis indicated that a sample size of 36 participants per group would provide 80% power for a two-tailed alpha of .05, based on the average effect size reported in previous meta-analyses of TF-CBT studies (.67; Cary & Curtis McMillen, 2012). A statistician-generated block randomisation schedule was generated prior to recruitment to ensure a balanced allocation of participants across groups.

Inclusion criteria were: (1) 11–19 years of age, (2) exposure to at least one traumatic event on the trauma screening list of the Kiddie Schedule for Affective Disorders and Schizophrenia-Present and Lifetime Version K-SADS-PL PTSD module (Kaufman et al., 1997, 2016), which assesses exposure to 11 types of trauma, and (3) active post-traumatic stress symptoms as indicated by having at least moderate PTSD symptom severity (a score of 21 or higher) on the Child PTSD Symptom Scale for DSM 5 (CPSS-5; Foa et al., 2018) or meeting a DSM 5 diagnosis of PTSD on the K-SADS-PL (Kaufman et al., 1997, 2016). Exclusion criteria included the presence of autism, psychotic disorders or substance use disorders on the K-SADS-PL (Kaufman et al., 1997, 2016) or collateral evidence of intellectual disability.

Participants were recruited through an existing university-based trauma research clinic in Cape Town, South Africa. This clinic conducts assessments and research (including treatment research) with trauma-exposed adolescents (aged 11–19 years) referred from local schools, hospitals, the legal system, and non-governmental organisations (NGOs). Flyers and informational talks about the TF-CBT study were provided to these referral sources during the recruitment phase. The majority of participants referred to the clinic do not have the financial means to access private mental health care.

Of 96 adolescents screened, 75 were eligible (21 males, 54 females; ages 11–19 years, M [SD] 14.92 [1.84]) and were randomised to either TF-CBT (n = 37) or TAU (n = 38).

1.3. Outcome measures

PTSD symptom severity was assessed with the self-report version of the CPSS-5 (Foa et al., 2018), administered verbally to participants by the assessor. The CPSS-5 is a 20-item scale assessing past-month severity of DSM 5 symptoms of PTSD in youth. The total severity score, a sum of all 20 items, ranges from 0 to 80. A clinical cut-off score of 31 is used to identify respondents at high risk of a PTSD diagnosis (Foa et al., 2018). The CPSS-5 self-report version has high convergence with the interviewer-administered version and has previously demonstrated excellent internal consistency, good test-retest reliability, and discriminant validity in United States samples (Foa et al., 2018). In our sample, the CPSS-5 had a Cronbach’s alpha of .89.

Depression symptom severity was assessed with the Beck Depression Inventory II (BDI-II; Beck et al., 1996), a 21-item scale with scores ranging from 0 to 63. A clinical cut-off of 23 is recommended to identify adolescents at high risk of MDD (Dolle et al., 2012). In studies with adolescents in the United States, the BDI-II has excellent internal consistency (Cronbach’s alpha = .93) as well as convergent validity (Rausch et al., 2017). The reliability and validity of the BDI-II has been demonstrated with young adults in South Africa (Makhubela & Mashegoane, 2016) and excellent internal consistency has been reported for adolescent samples (Schwartz et al., 2019). In our sample, the BDI-II had a Cronbach’s alpha of .91.

1.4. Randomisation procedures

After the baseline assessment, a research coordinator working off-site allocated participants to either the TF-CBT treatment group or the TAU control group, using the randomisation schedule. The research coordinator informed the caregivers directly about the treatment condition to which the adolescent had been assigned, without communicating with the assessor. The assessor was therefore blinded to participants’ treatment allocation.

1.5. Intervention

1.5.1. TF-CBT

This group received an abbreviated eight session version of Cohen and colleagues’ (2017) original 12–15 session TF-CBT. All the components of the original TF-CBT model were retained but compressed into fewer sessions (see Table 1). Sessions were 90 min in length. In each session, except the conjoint session, the counsellor spent 45 min with the adolescent and then 45 min with the caregiver. In most cases, one biological parent attended the caregiver sessions; due to practical and logistical reasons (such as employment commitments), it was often difficult for both parents to attend. In a minority of cases, where both biological parents could not attend or no longer resided with the child, another primary caregiver of the child (such as a grandmother who resided with the child) attended the caregiver sessions. All skills development with adolescents and caregivers involved in-session role play with the counsellor, a homework plan for between-session practice, problem-solving of any potential barriers to between-session practice (for example, how to find time or a quiet place to practice), and a review of between-session practice at the next session. The process of adapting and piloting the intervention for use with South African adolescents has been described in a previous publication (Kaminer et al., 2022).

Table 1. Example of average components received by participants in eight-session TF-CBT.

Session No.	Components with adolescent (45 min)	Components with caregiver (45 min)
1	Psychoeducation and therapeutic engagement	Psychoeducation, therapeutic engagement and introduction to parenting skills*
2	Relaxation skills and affective modulation	Relaxation skills, affective modulation and parenting skills
3	Cognitive coping skills	Cognitive coping skills and parenting skills
4	Trauma narrative (eliciting details of traumatic event)	Share trauma narrative with caregiver (if adolescent has consented) and continue parenting skills
5	Continue trauma narrative (eliciting worst moment and difficult thoughts and feelings) and begin cognitive processing	Continue sharing of trauma narrative with caregiver, begin cognitive processing and continue parenting skills
6	Continue cognitive processing and prepare for conjoint session	Continue cognitive processing, prepare caregiver for conjoint session and continue parenting skills
7	Conjoint session (combined 90 min)
8	Enhancing safety	Enhancing safety

*Parenting skills included the following: Praise, rewards and special thanks, positive one-on-one time, giving effective requests, and consequences. The order in which skills were delivered was determined by the most pressing parenting needs in each case (based on the counsellor’s assessment in the first session and ongoing tracking during treatment).

Treatment was delivered by six registered mental health professionals (three registered psychological counsellors and three clinical psychologists), who received training in both the full and the abbreviated versions of TF-CBT by a certified TF-CBT trainer, followed by weekly supervision from a TF-CBT certified trainer who had co-developed the abbreviated TF-CBT treatment. Before each session, counsellors agreed on steps for the session with the clinical supervisor, based on the treatment manual. After each session, counsellors documented which treatment components they had delivered and how. The clinical supervisor read the recorded session notes and then elicited further details in weekly supervision. If a counsellor missed a treatment component or step that should have been implemented, the supervisor requested completion in the next session.

1.5.2. TAU

The research coordinator provided all participants assigned to the TAU condition with contact details of existing CAMHS in, or near to, their residential area. This included their local government community health clinic, as well as non-government agencies providing trauma services to youth. Treatment decisions were then determined by the clinic or treatment agency accessed by participants. TAU participants who had CPSS scores in the clinically at-risk range at the three-month follow-up assessment were offered TF-CBT.

1.6. Safety protocol

Informed consent for participation in the research was obtained from all participants. A safety protocol was developed and included in the assessment procedures for both groups. The protocol included assessment for risk, procedures for notification, and possible referral routes. The study protocol was approved by the Health Research Ethics Committee at Stellenbosch University (N15/05/047) and the Faculty of Humanities at the University of Cape Town (PSY2014-020). With regard to adverse events, three hospital admissions were recorded in the TAU group: one for suicidality after a family conflict during the post-treatment follow-up period, one for medication assessment and adjustment, and one for what turned out to be an isolated panic attack. Two psychiatric hospital admissions were reported in the TF-CBT group during the treatment phase, but upon assessment neither were judged to be the result of the intervention: one participant experienced suicidal ideation activated by a weekend family conflict and the other experienced a panic attack following the death of a family member. One TF-CBT participant was briefly admitted to hospital due to suicidal ideation two months after treatment had ended; again, there was no apparent link to TF-CBT treatment.

1.7. Data analysis

We conducted an intent to treat analysis. To assess whether the TF-CBT and TAU groups were balanced through randomisation, the groups were compared on baseline characteristics using independent samples t-tests for continuous variables and χ² tests for categorical data. To examine the effectiveness of TF-CBT, linear mixed models (LMM) were conducted on the CPSS-5 and BDI-II symptom severity scores. These models included participants as a random effect and group and time as fixed effects. For post hoc testing, Fisher least significant difference (LSD) tests were used. Results were augmented by reporting Cohen’s d effect sizes. Normality was checked by inspecting normal probability plots and in all cases was found to be acceptable. To evaluate differences between the two groups over time in the proportion of participants meeting the clinical cut-offs on the CPSS-5 and BDI-II, we used generalised estimating equations (GEE) with binomial as the underlying distribution. As there were no significant differences in baseline sociodemographic characteristics or trauma exposure between the two groups, no covariates were included in the analyses of primary and secondary outcomes.

Due to loss to follow-up, there were some cases with missing data at post-treatment (7 of 75; 9.33%) and follow-up (2 of 75; 2.67%). There were no significant differences in demographic characteristics or trauma exposure between those who were lost to follow-up and those who completed all assessments. LMM and GEE are suitable for handling missing data because all measurements done at each time point are included in the model.

To assess whether group differences in time between assessments affected the results, mixed model ANOVAs were conducted where the number of days from baseline was entered into the model as a continuous covariate. This resulted in homogeneity-of-slopes analyses.

Mixed models were done using the R package ‘lmerTest’ version 3.1-0, and GEE analyses done using the R package ‘geepack’ version 1.2-1.

2. Results

2.1. Baseline characteristics

Baseline characteristics of gender, age, ethnicity, government grant status (in South Africa receiving a government social assistance grant is an indicator of low socioeconomic status), trauma exposure and scores on the CPSS-5 and BDI-II for the two groups are reported in Table 2. There were no significant differences between the groups on any of these variables. Both groups reported multiple trauma exposures, with an average of six different incidents and three different types of trauma exposure. Frequencies of exposure to different trauma types for the total sample are reported in Figure 1. Traumas involving violence or abuse were far more common than non-interpersonal traumas.

Figure 1. Frequency of Lifetime Trauma Exposure and Index Traumas Among Participants (n = 75).

Table 2. Baseline sociodemographic, trauma exposure and clinical characteristics of study participants.

Baseline characteristic	TF-CBT group (n = 37)	TAU group (n = 38)
Gender n (%)
Male	9 (24.32)	12 (31.58)
Female	28 (75.68)	26 (68.42)
Age M (SD)	14.78 (1.81)	15.05 (1.89)
Ethnicity n (%)
Coloured/mixed-race*	24 (64.86)	31 (81.58)
Black African	10 (27.03)	3 (7.89)
White	3 (8.11)	4 (10.53)
Receiving a government grant n (%)	18 (48.65)	17 (44.74)
Total number of trauma exposure incidents M (SD)	6.92 (4.72)	6.79 (6.83)
Different trauma types M (SD)	3.59 (1.21)	3.35 (1.44)
CPSS-5 M (SD)	47.84 (14.78)	47.03 (17.24)
BDI-II M (SD)	26.97 (14.96)	26.79 (13.27)
Met CPSS-5 clinical cut-off n (%)	32 (86.49)	29 (76.32)
Met BDI-II clinical cut-off n (%)	23 (62.16)	23 (60.53)

*Citizens of mixed-race are frequently termed ‘coloured’ in the South African context, based on the legacy of racial categories used during apartheid.

Note: There were no significant differences between groups on any of the baseline variables.

2.2. Time between assessments

In both groups, scheduled appointments were cancelled by some participants due to conflicting commitments (for example, school exams or sports events), illness, or transport difficulties, and had to be re-scheduled. The average number of days (M, SD) between baseline and post-assessment was significantly longer for the TF-CBT group (123.09, 28.86) than the TAU group (99.18, 7.03) at the p < .01 level, as was the total number of days between post-assessment and three-month follow up (TF-CBT 96.59, 10.11; TAU 92.66, 16.69; p = .04).

2.3. Treatment completion and retention

Participant retention is summarised in Figure 2. Of the 37 participants allocated to the TF-CBT condition, two dropped out of treatment. One had to relocate to another province after six sessions but was able to attend the three-month assessment. The other terminated treatment after five sessions due to transport difficulties (they lived in a semi-rural area outside Cape Town) but was able to attend both follow-up assessments. A further five TF-CBT participants completed treatment but did not attend their post-treatment assessment. Four of these attended the three-month follow-up assessment and one did not. One other participant, who had completed the post-treatment assessment, did not attend the three-month follow assessment. Of the 38 participants assigned to TAU, 17 (44.73%) reported that they had accessed the service they had been referred to. The remainder either did not attempt to access services or attempted but were unsuccessful due to various barriers typical of CAMHS in South Africa (for example, long waiting lists or difficulty contacting the provider). Four TAU participants did not attend the post-treatment assessment but only one missed the three-month assessment.

Figure 2. Flowchart of Study Participants.

2.4. Treatment implementation

Of the 35 TF-CBT participants who completed treatment, 34 completed in eight sessions, as per the treatment manual, while one completed in nine sessions because of very high levels of avoidance during the trauma narrative component. Due to sessions that were postponed due to illness, school-related commitments, or transport difficulties, the average time between first to last TF-CBT session was 10.36 weeks (SD = 2.80); treatment took place over exactly eight consecutive weeks for only 25.7% of TF-CBT participants.

2.5. Treatment effects

As there were no significant differences in baseline sociodemographic characteristics or trauma exposure between the two groups, no covariates were included in the analyses of primary and secondary outcomes. Results for treatment effects for all outcomes are reported in Table 3.

Table 3. Post-treatment and three-month Outcomes for TF-CBT (n = 37) and TAU (n = 38) groups.

	Baseline	Post-treatment	Three-month follow-up	% Reduction from baseline – post treatment	% Reduction from baseline – three-month follow-up	Difference between groups at post-treatment	Difference between groups at three-month follow-up
Primary outcome
CPSS-5 score						d = 0.60	d = 0.62
M (SD)						p < .01	p < .01
						95% CI	95% CI
TF-CBT	47.84 (14.78)	22.85 (19.22)	20.00 (17.40)	52.24	58.19	(0.11-1.08)	(0.15-1.09)
TAU	47.03 (17.24)	35.29 (22.30)	31.73 (20.48)	24.96	32.53
Secondary outcomes
Proportion meeting CPSS-5 clinical cut-off % (n)						p = .02	p = .03
TF-CBT	86.49 (32)	27.03 (10)	21.62 (8)	68.75	75.00
TAU	76.32 (29)	52.63 (20)	44.74 (17)	31.04	41.38
BDI-II score						d = 0.51	d = 0.42
M (SD)						p = .03 95% CI	p = .05 95% CI
TF-CBT	26.97 (14.96)	15.44 (14.16)	12.80 (13.46)	42.75	52.54	(0.03-0.99)	(−0.05-0.89)
TAU	26.79 (13.27)	23.44 (17.00)	18.95 (15.80)	12.50	29.26
Proportion meeting BDI-II clinical cut-off % (n)						p = .02	p = .03
TF-CBT	62.16 (23)	24.32 (9)	18.92 (7)	60.88	22.20
TAU	60.53 (23)	47.37 (18)	42.11 (16)	21.74	11.10

For the primary outcome, at both time points there was a significantly greater reduction in CPSS-5 PTSD symptom severity in the TF-CBT group compared with the TAU group (p < . 01 at both time points), with effect sizes of 0.60 at post-treatment and 0.62 at three-month follow-up.

For the secondary outcomes, there was a significantly greater reduction in the proportion of participants meeting the CPSS-5 clinical cut-off for PTSD in the TF-CBT group, compared with the TAU group, at both time points (p = .02 at post-treatment and p = .03 at three-month follow-up), the TF-CBT group had a significantly greater reduction in BDI-II symptom severity at both time points, with an effect size of 0.51 (p = .03) at post-treatment and 0.42 (p = .05) at three-month follow-up, and there was a significantly greater reduction in the proportion of participants meeting the BDI-II clinical cut-off for depression in the TF-CBT group, compared with the TAU group, at both post-treatment (p = .02) and three-month follow-up (p = .03).

Group differences in time between assessments did not influence the outcome for PTSD (p < .01) or depression (p = .05) symptom severity. The results were consistent with the findings of time as a fixed effect.

3. Discussion

We aimed to identify an intervention that is effective for treating trauma-exposed youth in culturally diverse LMIC settings and adapt it for use in South Africa, where the evidence-base for youth trauma interventions is currently poor. TF-CBT is one of the leading evidence-based treatments for traumatic stress among youth in HICs and has also been found to be effective in several LMICs. To our knowledge this is the first time that an abbreviated TF-CBT protocol has been systematically evaluated in a LMIC setting or with youth exposed to multiple traumas. This RCT contributes to the limited findings on treating traumatic stress among youth in LMICs, where trauma exposure rates are high and CAMHS services are poorly resourced.

Our findings support the efficacy of abbreviated eight-session TF-CBT in reducing PTSD symptom severity in South African adolescents who have experienced multiple traumas. Our abbreviated TF-CBT protocol produced a medium effect size for PTSD symptoms at post-treatment (.60) and three-month follow-up (.62), comparable to the average effect sizes reported in previous meta-analyses of TF-CBT (.67; Cary & Curtis McMillen, 2012; 0.66 in comparison to active treatments, Morina et al., 2016). This suggests that abbreviating TF-CBT to align with the needs of under-resourced CAMHS in LMICs is feasible and produces comparable outcomes to full-length TF-CBT, with no loss of treatment effect up to three months after treatment. Our effect size is lower than that reported in Murray et al.’s (2015) TF-CBT study in Zambia and some sub-samples in Dorsey et al.’s (2020) study in Tanzania and Kenya, but this is to be expected given the shorter duration of our TF-CBT treatment protocol. Further, in the context of violent communities, such as the ones in which many of our participants reside, some level of ongoing vigilance, avoidance and arousal may be an adaptive survival response (Eagle & Kaminer, 2013; Gaylord-Harden et al., 2018).

The TF-CBT group also experienced a significantly greater reduction in clinically at-risk levels of PTSD symptoms, compared with the TAU group. This indicates that PTSD symptom reductions were of clinical importance, with more participants in the TF-CBT group moving out of the clinically at-risk symptom range compared with the TAU group.

At post-treatment and three-month follow-up, participants who received TF-CBT had significantly lower depression scores and showed a greater reduction in clinically at-risk levels of depression symptoms, compared with TAU participants, indicating that abbreviated TF-CBT can improve comorbid depressive symptoms also. However, it is not possible to ascertain from the current study whether this is an indirect response to the reduction in PTSD symptoms or whether TF-CBT acts directly on depression symptoms, for example by developing affect regulation and cognitive coping skills or through other specific treatment components.

Our findings suggest that in under-resourced settings such as South Africa it may be optimal to adopt a flexible-length approach to treatment delivery (Galovski et al., 2012), using the abbreviated version of TF-CBT when there are practical barriers to full-length treatment and the full-length protocol when this is feasible for the setting and the client. This will allow an optimal balance between access and effectiveness. However, our study also raised some implementation issues. Although the abbreviated TF-CBT protocol was eight sessions in length, the average treatment time was closer to 10 weeks, due to the postponement of some sessions by participants due to illness, transport difficulties, or after-school commitments. Treatment took place over exactly eight consecutive weeks for only a quarter of TF-CBT participants. Given barriers to regular treatment attendance, it may be practical to deliver double sessions at each appointment, to reduce the number of visits required to complete treatment. Such an approach would need to be formally evaluated to assess whether it allows sufficient opportunity for incubation of skills and trauma processing. Further, the barriers that many TAU participants faced when accessing existing CAMHS highlight the need to explore the feasibility and effectiveness of alternative treatment delivery sites for abbreviated TF-CBT in South Africa. For example, there is some evidence that trauma interventions can be delivered using a task-sharing approach in local schools (Rossouw et al., 2018).

This study had several limitations. It is possible that PTSD and depression symptoms may be either over- or under-endorsed on self-report scales, compared to structured psychiatric interviews that include clinician judgement. Further, caregiver reports of changes in adolescent participants’ PTSD and depression symptoms were not obtained; these would have provided an additional source of information on clinical outcomes. Over half of participants in the TAU condition did not access services, due in part to existing access barriers for usual CAMHS in South Africa. For those who did access TAU, details of services received (type of intervention, number and duration of sessions, provider qualifications etc.) could not be reliably established. As a result, our control condition, while reflecting the realities of current services, was not a fully active one. It therefore cannot be conclusively demonstrated that TF-CBT, rather than access to regular mental health treatment, is responsible for the outcomes and abbreviated TF-CBT still needs to be evaluated against other active short-term trauma interventions. Our sample size did not allow for investigation of moderators of treatment response to this abbreviated TF-CBT model, which will be important to explore in future research to optimise efficiency and effectiveness. The absence of a standardised and independent measure of treatment fidelity is also a limitation. Finally, testing multiple secondary outcomes raises the possibility of Type I error, so these findings should be viewed with some caution until they can be replicated in future studies of abbreviated TF-CBT in South Africa.

In conclusion, our findings provide preliminary evidence that eight session TF-CBT is effective in reducing PTSD and depression symptoms in a sample of South African adolescents with multiple trauma exposure. These effects were maintained at three-month follow up, despite participants residing in communities characterised by ongoing violence and multiple adversity. Further research with active control conditions and the inclusion of clinician- and caregiver-rated outcome measures is needed to refine the evidence base for abbreviated TF-CBT in South Africa.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

For ethical reasons, primary data cannot be published online, as we had not included this in our informed consent and assent for participants. The dataset analysed for the current study is available from the corresponding author after complying with data protection agreements.

Additional information

Funding

This work was supported by National Research Foundation South Africa [grant number 93568].