Examining trauma cognition change trajectories among initial PTSD treatment non-optimal responders: a potential avenue to guide subsequent treatment selection

ABSTRACT

Background:

Despite their general effectiveness, 14-50% of individuals do not fully respond to evidence-based treatments for posttraumatic stress disorder (PTSD). Although changes in negative posttrauma cognitions (NPCs) are considered a likely PTSD treatment mechanism, less is known about how NPCs change among individuals who continue to be symptomatic following treatment (non-optimal responders).

Objective:

The objective of this study was to examine NPC change trajectories among individuals who were determined to be non-optimally responsive to intensive PTSD treatment.

Method:

Using a 3-week Cognitive Processing Therapy-based intensive PTSD treatment sample (ITP; N = 243), the present study examined the number of distinct NPC change trajectories among non-optimal responders via Group Based Trajectory Modeling and assessed predictors of non-optimal responders’ NPC change trajectory membership. Analyses were replicated in a separate 2-week ITP sample (N = 215).

Results:

In both non-optimal responder samples, two trajectories emerged; a no NPC change group which represented those with an overall lack of NPC change throughout treatment and an NPC change group which represented those with an overall reduction of NPCs occurring primarily later in treatment. Changes in PTSD symptom severity during treatment was the only consistent predictor of NPC change trajectory group membership among treatment non-optimal responders across ITPs.

Conclusions:

Findings suggest NPC change among non-optimal responders is nuanced and may inform subsequent intervention selection, resulting in testable hypotheses for future research.

HIGHLIGHTS

• Throughout intensive PTSD treatment, non-optimal responders exhibited two distinct negative posttrauma cognition change trajectories: (1) no change, and (2) slow change.

• Changes in PTSD symptom severity during treatment consistently predicted non-optimal responders’ trajectory of change in negative posttrauma cognitions.

• Additional research is needed to explore how nuances of negative posttrauma cognition change may inform subsequent treatment intervention in initial non-optimal responders.

Antecedentes: A pesar de su efectividad, entre el 14 y el 50% de las personas no responden completamente a los tratamientos basados en la evidencia para el trastorno de estrés postraumático (TEPT). Aunque los cambios en las cogniciones negativas postraumáticas (CPNs) se consideran un mecanismo probable de tratamiento del TEPT, se sabe menos sobre cómo cambian entre las personas que siguen sintomáticas tras el tratamiento (respondedores no óptimos).

Objetivo: El objetivo de este estudio fue examinar las trayectorias de cambio de las CNPs entre individuos que no respondían de manera óptima al tratamiento intensivo del TEPT.

Método: Utilizando una muestra de tratamiento intensivo para TEPT basado en la Terapia de Procesamiento Cognitivo de 3 semanas. (PTI; N = 243), el presente estudio examinó el número de trayectorias de cambio de CPN entre respondedores no óptimos a través del Modelado de Trayectoria Basado en Grupos y evaluó los predictores de la pertenencia a la trayectoria de cambio de CPN de los respondedores no óptimos. Los análisis se repitieron en una muestra de PTI separada de 2 semanas (N = 215).

Resultados: En ambas muestras de respondedores no óptimos aparecieron dos trayectorias: un grupo sin cambios en el CNP, que representaba a aquellos con una ausencia de cambio del CNP a lo largo del tratamiento y un grupo de cambio del CNP que representaba a aquellos con una reducción general de las CNP que se produjo principalmente al final del tratamiento. Los cambios en la gravedad de los síntomas de TEPT durante el tratamiento fueron el único predictor consistente de la pertenencia al grupo de trayectoria de cambio de CNP entre los respondedores no óptimos al tratamiento en todos los PTI.

Conclusiones: Los hallazgos sugieren que el cambio del CNP entre los respondedores no óptimos es matizado y puede informar la selección de la intervención posterior, dando lugar a hipótesis comprobables para futuras investigaciones.

背景：尽管创伤后应激障碍 (PTSD) 的循证治疗总体有效，但 14-50% 的个体并未完全反应。尽管负性创伤后认知 (NPC) 的变化被认为是一种可能的 PTSD 治疗机制，对于在治疗后继续出现症状的个体（非最佳反应者）中 NPC 如何变化，人们知之甚少。

目的：本研究旨在考查那些被确定对强化 PTSD 治疗反应不佳个体的 NPC 变化轨迹。

方法：使用为期 3 周基于认知加工疗法的强化 PTSD 治疗样本（ITP；N = 243），本研究通过基于组别的轨迹模型考查了非最佳反应者中不同 NPC 变化轨迹的数量，并评估了非最佳反应者 NPC 变化轨迹组别的预测因素。在单独的 2 周 ITP 样本中重复分析 (N = 215)。

结果：在两个非最佳反应者样本中，出现了两条轨迹； 无 NPC 改变组代表在整个治疗过程中总体上没有 NPC 变化的组，以及 NPC改变组代表主要在治疗后期出现的 NPC 总体减少的组。 治疗期间 PTSD 症状严重程度的改变是 ITP 治疗非最佳反应者中 NPC 变化轨迹组别的唯一一致预测因素。

结论：研究结果表明，非最佳反应者的 NPC 变化是微妙的，可能会为后续的干预选择提供信息，从而为未来的研究提供可检验的假设。

KEYWORDS:

• Negative posttrauma cognitions
• non-response
• intensive treatment
• PTSD
• treatment guidelines

PALABRAS CLAVE:

• Cogniciones negativas postraumáticas
• No respuesta
• Tratamiento intensivo
• TEPT
• Pautas de tratamiento

关键词:

• 创伤后负性认知
• 无反应
• 强化治疗
• PTSD
• 治疗指南

Evidence-based treatments for posttraumatic stress disorder (PTSD), such as Cognitive Processing Therapy (CPT; Resick et al., 2017), have been shown to produce large PTSD symptom reductions in weekly (Asmundson et al., 2019; Lewis et al., 2020) and intensive treatment formats with daily sessions (Sciarrino et al., 2020). Despite strong effectiveness data, depending on the sample, between 14-50% of individuals receiving evidence-based treatments can generally be classified as only partially or minimally responsive to these first-line PTSD treatments (Held et al., 2021; Resick et al., 2017; Steenkamp et al., 2015, 2020). Moreover, research has shown that approximately two-thirds of individuals who underwent evidence-based treatments retained their PTSD diagnosis following treatment completion (Steenkamp et al., 2015). Currently, relatively little is known about the reasons for such non-optimal response. One hypothesis is that purported mechanisms may not be sufficiently engaged by the respective evidence-based treatments among non-optimal responders.

Change in negative posttrauma cognitions (NPCs) have been hypothesized to be an important PTSD treatment mechanism (Brown et al., 2019; Zalta, 2015). NPCs are a key treatment target in CPT (Resick et al., 2017). According to the theory of change in CPT, failure for PTSD to improve in treatment is often due to a lack of improvement in NPC’s, especially those specifically about the index trauma (Galovski et al., 2020). In support of this theory, it has been repeatedly shown that NPC reductions during treatment are associated with PTSD symptom severity reductions; less NPC change over the course of treatment is associated with less change in PTSD symptom severity (Brown et al., 2019; Foa & Rauch, 2004; Holliday et al., 2014; Schumm et al., 2015; Sripada et al., 2016; Zalta et al., 2014, 2018). Reductions in NPCs throughout traditional (Schumm et al., 2015; Zalta et al., 2014) and intensive PTSD treatment (Held et al., 2020; Zalta et al., 2018) has been shown to predict reductions in PTSD and depression symptoms with those individuals who reported greater changes in NPC’s most likely to maintain long-term gains. There were initial questions regarding whether intensive treatments would show similar patterns of NPC change as found with traditional timing of CPT delivery, especially given that there is less time for rehearsal of cognitive skills. However, studies of traditional (Scher et al., 2017) and intensive treatment (Held et al., 2020) demonstrate NPC improvements predict PTSD and depression reductions and are strongly associated with long-term maintenance of treatment effects. Consequently, a failure to engage this purported mechanism should increase the odds of being a PTSD treatment non-optimal responder.

Despite the repeatedly reported positive association between NPC change and PTSD change, it is still unclear whether NPCs change similarly for non-optimal responders to PTSD treatments, or whether there are nuances in whether and how NPCs change for these individuals. Non-optimal responders are likely a heterogenous group and non-optimal treatment responses may be due to multiple reasons (Galovski et al., 2020). As such, even the way in which NPCs may change (or not change) for this group may differ. For example, it has yet to be determined whether NPC change is identical for all non-optimal responders (e.g. a lack of NPC change) or whether NPC change follows different trajectories, as has been repeatedly demonstrated for PTSD symptom change (Dewar et al., 2020; Elliott et al., 2005; Held et al., 2021; Schumm et al., 2013; Stein et al., 2012).

Studying nuances of NPC change has the potential to inform and possibly guide future treatment selection for these individuals. For example, for non-optimal responders to a course of CPT who may not exhibit any NPC changes, it would likely be futile to continue the same treatment. These individuals may benefit from an intervention that engages the purported mechanism differently or that engages a different mechanism of change. Yet, for individuals who exhibit some changes in NPCs, it is plausible that extending the number of sessions may result in eventual PTSD symptom changes, as these individuals may require additional time for NPC change to translate to noticeable reductions in PTSD symptoms. The latter postulation has been partially supported through prior research which has demonstrated that extending CPT can be beneficial for a number of individuals who did not exhibit substantial PTSD severity reductions after the standard duration of treatment (Galovski et al., 2012; Resick et al., 2021).

Although there is a strong desire to personalize treatment approaches (Herzog & Kaiser, 2022), a substantially better understanding of the extent to which purported treatment mechanisms are engaged among non-optimal responders, and whether the rate and timing of mechanism engagement differs across individuals in this group, is required. The goal of the present study was to advance our understanding of NPC change among non-optimal responders. To do so, we examined two samples of veterans who completed Cognitive Processing Therapy-based intensive PTSD treatment programmes (ITPs) and who were classified as non-optimal responders based on not having achieved meaningful PTSD and depression symptom change over the course of the programme. In a sample of veterans who completed a 3-week ITP, we first examined how many distinct NPC change trajectories could be reliably identified in this group. We expected that individuals would report different rates and timing of NPC change and thus hypothesized identifying at least two trajectories, with one trajectory representing an overall lack of NPC change and another trajectory showing at least some NPC change, despite everyone being considered non-optimally responsive based on their lack of meaningful PTSD and depression symptom change. This hypothesis was based on cognitive theory and processing assimilated and overaccommodated beliefs (Resick et al., 2017). Theory and previous research have suggested that changes in overaccommodated beliefs are most effective following the successful processing of assimilated beliefs (Farmer et al., 2017; Resick et al., 2017). In the case of non-optimal responders, we assumed that the processing of overaccommodated beliefs could either be delayed or not occur at all, which captures our two hypothesized groups. We then assessed whether non-optimal responder NPC change trajectory membership could be predicted using demographic characteristics, baseline symptom severity, or PTSD and depression symptom change. Based on prior research, we expected that the identified NPC change trajectories would be associated with PTSD symptom change, with greater change in NPC trajectories being associated with greater PTSD symptom severity change. Finally, to reduce the likelihood that findings were due to chance, and to examine whether the findings were robust across varying ITP session timing, we tested these aims in a second, separate 2-week CPT-based intensive PTSD treatment programme with twice daily (rather than once daily) individual CPT sessions. In doing so, we intend to replicate study findings across multiple CPT-based ITP formats to better identify potential avenues for personalization for individuals who respond non-optimally to such treatments.

1. Method

1.1. Participants

Data were collected from veterans who completed either a 3-week (N = 243) or 2-week (N = 215) CPT-based intensive treatment programme (ITP) for PTSD at the Road Home Programme at Rush University Medical Center¹ and were determined to be non-optimal responders (see definition below). On average, veterans in the 3-week sample were 41.31 years old (SD = 9.34, range: 24–74 years), 64.61% identified as male, and 67.90% identified as White. On average, veterans in the 2-week sample were 43.45 years old (SD = 10.11, range: 21–72 years), 52.09% identified as male, and 63.26% identified as White. Additional demographic information can be found in Table 1.

Table 1. Demographic characteristics of veteran non-responders in 2- and 3-week Cognitive Processing Therapy-based intensive PTSD treatment programmes.

Variable	2-Week ITP N = 215		3-Week ITP N = 243
		M [SD]		M [SD]
Age		43.44 [10.11]		41.31 [9.34]
	n	%	n	%
Sex		-		-
Female	103	47.91	86	35.39
Race		-		-
White	136	63.26	165	67.90
Black / African American	45	20.93	48	19.75
Asian	7	3.26	3	1.23
Hawaiian / Pacific Islander	5	2.33	0	0.00
American Indian or Alaska Native	1	0.47	6	2.47
Other	21	9.77	21	8.64
Ethnicity		-		-
Hispanic or Latino	41	19.07	51	20.99
Not Hispanic or Latino	174	80.93	192	79.01
Marital Status		-		-
Married	94	43.72	127	52.26
Divorced	50	23.26	44	18.11
Single	60	27.91	55	22.64
Legally Separated	6	2.79	15	6.17
Other	2	0.93	2	0.82
Trauma Type		-		-
Military Sexual Trauma	101	46.98	87	35.80
Combat	115	53.02	156	64.20
Deployed
Yes	120	55.56	184	75.72

Note. ITP = Intensive PTSD Treatment Programme.

1.2. Procedures

The inclusion criteria were identical for the two ITPs. Veterans who met the diagnostic criteria for PTSD based on the Clinician Administered PTSD Scale for DSM-5 (CAPS-5) were eligible. Exclusion criteria included active suicidality or homicidality (i.e. plan and/or attempt within the past 60 days), current significant non-suicidal self-harm behaviour, active mania or psychosis, active substance use that would interfere with ability to participate in the ITP or pose a risk of withdrawal symptoms, and/or active eating disorders that would require medical observation. Based on the nature of the index trauma they were planning to address throughout treatment, veterans were assigned to a combat trauma or military sexual trauma (MST) cohort.

The 3-week ITP was offered from April 2016 to March 2020. It included 14 daily 50-minute individual CPT sessions, 13 group CPT sessions, 13 mindfulness sessions, 12 yoga sessions, and additional services such as distress management education sessions and DBT (MST cohorts only). The 2-week ITP was offered from February 2021 to January 2023 and included 16 50-minute CPT sessions delivered twice per day, 9 mindfulness-based stress reduction sessions, 8 group yoga sessions, and additional services such as art therapy and Dialectical Behavioural Therapy (DBT). More information about the structure and clinical outcomes of the two treatment programmes can be found elsewhere (Held et al., 2022).

The Rush University Medical Center Institutional Review Board approved all study procedures and data collection. As all data collection occurred as a part of routine clinical care, a waiver of consent was obtained.

1.3. Measures

1.3.1. Demographic Characteristics

At intake, veterans reported age, sex, race, ethnicity, education, service branch, marital status, and service era (pre/post- September 11, 2001).

1.3.2. Posttraumatic Cognitions Inventory (PTCI)

The PTCI is a 33-item self-report measure of NPCs pertaining to self-blame, negative beliefs about the self, others, and the world that individuals are currently experiencing (Foa et al., 1999). PTCI total scores range from 33 to 231, with higher scores indicating greater NPC severity. In the 2-week ITP, the PTCI was administered at intake and treatment days 4 and 9. In the 3-week ITP, the PTCI was administered at intake and treatment days 4, 9, and 13. Internal reliability for the PTCI in the present samples ranged from .95 to .97.

1.3.3. PTSD Checklist for DSM-5 (PCL-5)

The PCL-5 is a 20-item self-report measure of PTSD symptom severity based on DSM-5 diagnostic criteria (Weathers et al., 2013). PCL-5 scores range from 0 to 80 and higher scores indicate greater PTSD symptom severity. A total score of ≥ 33 has been suggested to indicate ‘probable PTSD’ (Bovin et al., 2016). In the 2-week ITP, the PCL-5 was administered at baseline, treatment days 1, 3, 5, 6, 8, and at post-treatment. In the 3-week ITP, the PCL-5 was administered at intake, treatment days 1, 2, 3, 5, 6, 7, 11, 13, and at post-treatment. Aside from the intake assessments, which assessed past month PTSD symptom severity, all other timepoints assessed past week PTSD symptom severity. In the present samples, internal reliability for the PCL-5 ranged from .89 to .96.

1.3.4. Patient Health Questionnaire (PHQ-9)

The PHQ-9 is a 9-item self-report measure of depression symptom severity (Kroenke et al., 2001). The PHQ-9 scores range from 0 to 27 and higher scores indicate greater depression symptom severity. A total score of ≥ 10 has been suggested to indicate at least moderate depression. In both ITPs, the PHQ-9 was administered at baseline and post-treatment and assessed symptom severity in the past two weeks. In the present samples, internal reliability for the PHQ-9 ranged from .82 to .86.

1.4. Definition of non-optimal response

There are no standard criteria for defining response or non-optimal response to PTSD treatment (Beck, 2020; Fonzo et al., 2020; Larsen et al., 2020; Varker et al., 2020). As one of the goals for the present study was to lay the foundation for future work examining subsequent treatment for non-optimal responders, we decided to use relatively strict criteria that were based on endpoint symptom severity. Specifically, individuals who reported PCL-5 ≥ 33 and PHQ-9 ≥ 10 were considered non-optimal responders. We decided against using symptom change as an indicator for non-optimal response because individuals who exhibit strong response to treatment based on symptom change alone may still be highly symptomatic at the end of treatment and would require additional care. As such, to establish the groundwork for personalizing subsequent treatment options for individuals who remain highly symptomatic following an initial course of treatment, non-optimal response intends to refer even to individuals who exhibited symptom change but who still meet clinical cut-off scores and could benefit from additional intervention (Bovin et al., 2016; Kroenke et al., 2001).

1.5. Statistical analysis

To identify and define distinct groups based on trajectory of NPC change among non-optimal responders, we used Group Based Trajectory Modelling (GBTM). This finite mixture modelling-based approach probabilistically identifies latent groups with similar trajectories during the treatment programme using maximum likelihood based on a general quasi-Newton procedure (Jones & Nagin, 2012; Nagin, 2005). Determination of the number of trajectories was determined based on apriori theory, Nagin’s (2005) guidelines of trajectory adequacy, information criteria (Akaike’s Information Criterion; AIC, and Bayesian Information Criterion; BIC), and group composition and adequate representation within individual trajectories. Relative trajectories, based on amount of PTCI change that occurred from baseline to each measurement timepoint, were the focus of this trajectory analysis.

The GBTM approach also allows researchers to assess whether the NPC change trajectories were associated with covariates, which consists of multinomial logistic regression predicting trajectory membership with predictors of interest. Baseline covariates, including depression and PTSD² severity, age, sex, race, ethnicity, and cohort type, were explored as predictors of trajectory membership among non-optimal responders. Clinical predictors of interest included baseline depression and PTSD severity as well as amount of depression and PTSD severity change during the programme. All the analyses described were conducted in both the 3- and 2-week ITP. All analyses were conducted in Stata version 15.

2. Results

For NPC change among non-optimal responders, a two-trajectory model was selected based on prior hypotheses, similarity to expected group composition, and information criteria preferences among models with adequate representation in each trajectory group. Information criteria indicated general preference for the two-trajectory solution (AIC = −4507.00, BIC = −4520.97) over a one-trajectory solution (BIC_Δ = 86.91, AIC_Δ. =  103.64) or three-trajectory solution (BIC_Δ = 13.73, AIC_Δ = 5.00). The AIC and BIC analyses also clearly supported quadratic curvature for both trajectories. Average posterior probabilities of group membership in each programme exceeded .88 and odds of correct classification exceeded 5, which supports model adequacy (Nagin, 2005). One NPC change trajectory group represented non-optimal responders who exhibited an initial increase in NPCs and a lack of overall change from pre- to post-treatment (n = 143 (58.8%) in the 3-week programme and n = 135 (62.6%) in the 2-week programme; hereafter referred to as the no NPC change group). The other NPC change trajectory group included non-optimal responders who exhibited an overall reduction in NPCs, with a slightly steeper reduction in later phases of treatment (n = 100 (41.2%) in the 3-week programme and n = 80 (37.4%) in the 2-week programme; hereafter referred to as the NPC change group). Despite the different number of during-programme PTCI measurement time points in the 3- and 2-week programmes the same general trajectory groups emerged. Figures 1 and 2 illustrate PTCI trajectories and confidence intervals among non-optimal responders in the 3- and 2-week ITPs, respectively.

Figure 1. Negative posttrauma cognitions trajectories among non-optimal responders in the 3-week intensive PTSD treatment programme.

Note: Thin dotted lines represent 95% confidence intervals.

Figure 2. Negative posttrauma cognitions trajectories among non-optimal responders in the 2-week intensive PTSD treatment programme.

Note: Thin dotted lines represent 95% confidence intervals.

Greater PCL-5 change reductions were associated with greater odds of being in the NPC change group in both the 3- and 2-week programmes (see Table 2). Results indicate that for each point reduction in PCL-5 during the programme, the odds of a treatment non-optimal responder being in the NPC change group increased between 6% and 11% across the two programmes. Change in PHQ-9 was also a significant predictor of NPC trajectory in the 3-week programme, but not the 2-week programme. Neither baseline PTSD and depression severity nor demographic variables were associated with greater likelihood of being in either NPC change group (Table 2).

Table 2. Predictors of PTCI trajectory membership among non-optimal responders.

	3-week programme			2-week programme
Variable	b^1	95% CI	p	b	95% CI	p
PCL-5 change	1.06	1.01, 1.13	.020	1.11	1.06, 1.16	< .001
PHQ-9 change	1.16	1.01, 1.34	.031	1.06	0.94, 1.20	.335
Baseline PCL-5	1.03	0.97, 1.09	.264	1.00	0.95, 1.05	.905
Baseline PHQ-9	0.93	0.84, 1.12	.330	1.10	0.98, 1.24	.115
Age	1.00	0.96, 1.04	.806	1.00	0.97, 1.04	.892
Sex^2	1.60	0.55, 4.62	.381	0.72	0.21, 2.41	.591
Race	1.21	0.62, 2.36	.582	0.61	0.29, 1.30	.200
Ethnicity	1.00	0.42, 2.34	.992	0.57	0.24, 1.36	.207
Cohort Type	0.66	0.23, 1.92	.448	0.72	0.51, 5.80	.383

Note. ¹Slopes and corresponding confidence intervals represent the increase in odds of being in trajectory 2 for each unit increase of the variable. ²Reference categories were male, white, non-Hispanic, and MST cohort.

3. Discussion

The present study adds to the extensive literature examining the role of NPC changes in PTSD treatment (Brown et al., 2019). Until now, studies have not examined the association between NPCs and PTSD symptoms specifically in non-optimal responders. This has limited our ability to understand whether NPCs change at the same rate for all individuals in this group or whether there may be individual nuances, which would provide empirical support for the notion that non-optimal responders are different with regard to their ability to achieve NPC change. Understanding whether different NPC change trajectories may exist in this particular group of individuals is important as it may help tailor interventions to more effectively engage NPC change in these individuals and thus facilitate improved outcomes. Further, replicating these associations across two distinct CPT-based treatment formats provides a stronger foundation for understanding and examining implications of NPC change trajectories on treatment personalization especially during intensive treatment delivery.

The results from the present study demonstrated that, NPC changes during treatment could be grouped into two distinct trajectories among non-optimal treatment responders. Individuals could be grouped into a no NPC change and a slow NPC change group. Thus, being non-optimally responsive to treatment does not necessarily mean that NPCs are not changing at all. Instead, there appear to be important nuances in how NPCs change (or do not change) over the course of treatment in this group.

Our ability to predict NPC change trajectory group membership among treatment non-optimal responders was relatively limited, with changes in PTSD symptom severity during treatment being the only consistent predictor. Neither demographic variables nor baseline PTSD and depression symptom severity consistently predicted whether non-optimal responders would be assigned to the slow NPC change or the no NPC change group across both samples. In general, predicting how individuals’ NPCs will change at the onset of treatment appears to be extremely challenging, similar to what has previously been suggested for predicting PTSD symptom severity changes using only demographic and baseline clinical characteristics (Held et al., 2022). In line with prior research which has demonstrated that treatment response predictions become increasingly accurate as treatment progresses (Nixon et al., 2021; Smith & Held, 2022), the present study suggested that PTSD symptom severity change over the course of treatment was a much better predictor of NPC change group membership than baseline PTSD symptom severity (Brown et al., 2019). Across both ITP samples of non-optimal responders in the NPC change groups, NPC change accelerated around mid-treatment, though among those in the no NPC change trajectory group this acceleration only brought them back to baseline NPC levels. Understanding individuals’ NPC change trajectories can be important when evaluating treatment options for non-optimal responders (e.g. extending CPT or switching treatment approaches). It may therefore be prudent to wait to make such decisions until after mid-treatment when NPC change trajectories become clearer.

A few limitations need to be considered when evaluating the findings from this study. Although the present study utilized two large intensive PTSD treatment samples to replicate the results, findings may not translate as well to evidence-based treatments that are delivered in non-intensive treatment formats. Similarly, study findings may not be generalizable to populations engaged in standalone interventions, as there is potential for the ITPs’ adjunctive services to impact symptom change. It is also important to note that the study data came from clinical programmes rather than highly controlled clinical trials and it was impossible to evaluate the quality of CPT that was provided. It is therefore possible that NPC change trajectories may differ in settings where fidelity to the CPT model is verified. Another limitation is the relatively small number of repeated NPC assessments, especially compared to other studies that have evaluated PTSD severity trajectories, which may have limited the number of NPC change trajectories that could be identified. Since PTSD symptom changes were not examined longitudinally, it was not possible to assess timing or potential mechanistic roles, although there are several published studies that have addressed the temporality of changes in NPC and PTSD symptoms (Held et al., 2022; Lee et al., 2021). Finally, NPCs were measured in the present moment, whereas PTSD and depression symptoms were rated over the past one and two weeks, respectively. In other words, because of these differing assessment timeframes PTSD and depression symptoms may ‘lag’ the NPC assessments. It is therefore possible that individuals rated as non-optimal responders may have been classified otherwise had PTSD and depression symptom assessments been conducted one or two weeks after individuals completed the ITP.

4. Conclusions

As the first of its kind, the present study examined NPC change trajectories among non-optimal responders to CPT. This study marks a potentially important step toward understanding NPC changes among individuals who may not respond to an initial course of care, as findings from a 3-week ITP were replicated in a separate 2-week ITP. Among non-optimal responders, NPC change followed two distinct patterns. It will be critical to further study these nuances and evaluate whether they may provide clues about the optimal subsequent course of care for individuals who are initially non-optimally responsive to treatment. For example, individuals who reported a slow change in NPCs may simply need additional time for treatment to demonstrate effects. This notion aligns with previous research on CPT which has demonstrated that extending treatment can be helpful for some individuals who have not responded during the initial course of care (Galovski et al., 2012; Resick et al., 2021). Importantly, in these previous studies not all individuals responded even after receiving extended courses of CPT and it is plausible that these individuals may have been similar to those classified in the no NPC change group in the present study. In the referenced clinical trial examining the benefit of extending CPT, 8% of initial non-optimal responders remained non-responsive even after receiving additional CPT sessions (Galovski et al., 2012). It is possible that this subgroup of individuals who do not exhibit NPC change during an initial course of CPT may not benefit from extending the same kind of treatment. Future research should examine whether there are other mechanisms of change that further differentiation treatment trajectories as these may help inform treatment matching for non-optimal responders. Along with the aforementioned hypotheses, future research also needs to replicate the presented results in samples where CPT is delivered as a standalone intervention rather than as part of comprehensive ITPs in order to determine the extent to which these findings generalize. It will also be important to further evaluate non-optimal response, given the large number of potential definitions. The present study based non-optimal response categories solely on endpoint symptom severity, but it will also be important to evaluate the amount of symptom change individuals who experienced meaningful symptom improvement throughout treatment but still remain above a clinical threshold at endpoint and those who did not respond experience meaningful symptom improvement. In general, additional research on non-optimal responders to initial PTSD treatment is critically needed as this group of individuals is currently understudied and heterogenous; without a more nuanced understanding of non-optimal response, it will be extremely difficult to identify or design optimal, personalized subsequent treatment options.

Acknowledgments

The National Center for PTSD, VA Palo Alto Health Care System. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Veterans Administration, Wounded Warrior Project, the Patient-Centered Outcomes Research Institute (PCORI), Board of Governors, or Methodology Committee or that of any other funding agency. All other authors declare that they have no competing interests.

Data availability statement

The datasets used in the current study are not publicly available. Datasets can be obtained from the corresponding author upon reasonable request.

Disclosure statement

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

Additional information

Funding

This work was supported by Agency for Healthcare Research and Quality: [grant no R21 HS028511]; National Institute on Alcohol Abuse and Alcoholism: [grant no R01-AA027499]; Wounded Warrior Project: [grant no. N/A]; Patient-Centered Outcomes Research Institute: [grant no PTSD-2019C1-15636]; US Department of Defense: [grant no. W81XWH-22-1-0739].

Notes

1 A total of 500 veterans started and completed the 3-week ITP and 330 veterans started and completed 2-week ITP.

2 Due to the overlap between PTCI and PCL-5 items 9 and 10, we explored trajectory models with both the full PCL-5 and a version without these two items. Since results were robust to this change, we have only reported the full PCL-5 scores here.