A National Survey of Patient Completion of Cognitive Behavioral Therapy for Chronic Pain: The Role of Therapist Characteristics, Attempt Rates, and Modification

Abstract

The implementation of evidence-based psychotherapies, including patient-level measures such as penetration and rates of successfully completing a course of therapy, has received increasing attention. While much attention has been paid to the effect of patient-level factors on implementation, relatively little attention has been paid to therapist factors (e.g., professional training, experience).

Objective:

The current study explores therapists’ decisions to offer a particular evidence-based psychotherapy (cognitive behavioral therapy for chronic pain; CBT-CP), whether and how they modify CBT-CP, and the relationship between these decisions and patient completion rates.

Methods:

The study utilized survey responses from 141 Veterans Affairs therapists certified in CBT-CP.

Results:

Therapists reported attempting CBT-CP with a little less than one half of their patients with chronic pain (mean = 48.8%, s.d.=35.7). Therapist were generally split between reporting modifying CBT-CP for either very few or most of their patients. After controlling for therapist characteristics and modification, therapist-reported percentage of patients with attempted CBT-CP was positively associated with completion rates, t (111) = 4.57, p<.001.

Conclusions:

Therapists who attempt CBT-CP more frequently may experience better completion rates, perhaps due to practice effects or contextual factors that support both attempts and completion. Future research should examine this relationship using objective measures of attempt rates and completion.

Substantial attention has been paid to evidence-based psychotherapies (EBPs) in the treatment of chronic conditions and strategies to support their implementation in routine practice. Two important patient-level markers of implementation success include penetration (the number of eligible patients receiving the practice) and feasibility (e.g., percent of patients successfully completing a full course of the practice) (Proctor et al., 2011). These outcomes are dynamically affected by factors at various levels (Damschroder et al., 2015); however, one important source of influence is the EBP therapist. Therapists make decisions such as whether to attempt a course of the EBP with any given patient and whether to modify the practice from its original form. These decisions affect the number of patients receiving a full, pure “dose” of the EBP, or any at all.

The penetration of an EBP is defined as the degree of integration of the practice into a service setting; although there are several indicators of penetration, one important measure is the percentage of eligible patients who receive the EBP (Proctor et al., 2011). Prior research has indicated penetration is often low, even in specially trained EBP providers. For instance, work regarding EBPs for posttraumatic stress disorder (PTSD) both within the Veterans Health Administration (VHA) (Finley et al., 2015), and among community providers (Dondanville et al., 2018) demonstrated therapists often fail to provide or offer EBPs to patients. Importantly for the current study, attempting to offer or initiate a given EBP to a patient is a prerequisite for the patient to receive the EBP; therefore, the percentage of patients with whom the EBP is attempted is an important measure of penetration.

Once a course of an EBP is initiated, significant variability may occur in what follows. Led by seminal work by Stirman and colleagues (Stirman, Miller et al., 2013; Stirman, Baumann et al., 2019), research has led to a greater understanding of how practices may be modified in routine clinical care. Modification in practice is controversial. Some implementation models assume modification as a necessary process within implementation that maximizes fit within the clinical setting and patient population (Aarons et al., 2012). Evidence has consistently supported flexibility, particularly with manualized therapies (i.e., modifying the practice), to be associated with better outcomes (Kalichman et al., 1993; Kennedy et al., 2000; Marques et al., 2019). Interestingly, Marques and colleagues (2019) found only fidelity-consistent modifications to a PTSD EBP (e.g., tailoring language) to be associated with better outcomes. In contrast, other research found modification associated with poorer patient outcomes for a diverse set of presenting conditions (Kumpfer et al., 2002; Schulte & Eifert, 2002; Stanton et al., 2005). Some scholars have pointed to modification of practices in routine practice as a potential cause of the “voltage drop” in which EBPs show smaller effects in routine practice than those demonstrated in clinical trials (Chambers & Norton, 2016). In practice, modification may signal a perceived misfit of the EBP that may call for a change, including potentially discontinuing the EBP. Taken together, it is unclear whether and when modification of an EBP may be beneficial versus deleterious.

Both penetration and modification may be driven by perceived feasibility – the degree to which an EBP can be used in a service setting (Proctor et al., 2011). While variably assessed, feasibility is often evaluated retrospectively based on the percentage of patients successfully completing the full course of the practice. Dropout from psychotherapy is recognized as common and problematic. A seminal meta-analysis estimated that, across a variety of clinical contexts and practices, 47% of patients drop out before clinically indicated (Wierzbicki & Pekarik, 1993); however, a more recent meta-analysis estimated rates of about 20% (Swift & Greenberg, 2012). Premature dropout is important as patients may not benefit from abbreviated courses of therapy. Furthermore, missed appointments due to dropout can represent lost opportunities for patients to receive effective care and for healthcare systems to deliver this care. Scholars generally recognize dropout rates are affected by numerous factors, with research on patient-related factors (e.g., patient socio-economic status) receiving disproportionate attention relative to system- and therapist-related factors (Roos & Werbart, 2013).

The current study explores therapists’ decisions to offer a particular EBP (cognitive behavioral therapy for chronic pain; CBT-CP), whether and how they modify the CBT-CP, and the relationship between these decisions and patient completion rates. CBT-CP is an approximately 12-session EBP that has been shown in numerous clinical trials to reduce the burden of chronic musculoskeletal pain, including pain catastrophizing and pain interference in functioning (Ehde et al., 2014; Murphy, in press; Stewart, 2015). CBT-CP is the target of an ongoing national VHA EBP rollout that involves extensive therapist training, evaluation of session recordings, and weekly consultation calls (Stewart et al., 2015). CBT-CP was chosen as a focus of the current study because of the importance of chronic pain in the VHA and US population as well as its focus on a physical health condition which contrasts and complements prior research on modification and completion of EBPs focused on psychiatric conditions. Of note, while diagnosis was associated with differential dropout rates in Swift and Greenburg’s (2012) meta-analysis, chronic pain was not specifically examined; therefore, it is unclear how EBPs for chronic pain may differ in regards to completion and dropout. The current study examines data from a national survey of VHA therapists who had completed the CBT-CP training and consultation process regarding their CBT-CP patient penetration, practice modification, and case completion rates.

Methods

Sampling and Procedures

The lead author (who did not conduct VHA CBT-CP trainings and is not officially affiliated with the office that provides the trainings) sent invitations to all VHA therapists nationwide who had completed the formal VHA CBT-CP training and consultation process between 6/12/2012 and 2/15/2018 (n = 404). Invitations were sent via email which included a survey link. To promote the survey completion, we used strategies from the Dillman total design survey method (Smyth et al., 2009). This method involves frequent follow-up; thus, a reminder email was sent to therapists who had not completed the survey one week after survey distribution. Following the initial reminder contact, periodic e-mail reminders were sent (up to four) until the survey was closed. Participants reviewed an informed consent document and indicated their agreement electronically. This study was approved by the Indiana University-Purdue University Indianapolis Institutional Review Board (IRB), the medical center Research and Development Committee, the Organizational Assessment Sub-Committee of the Human Resources Committee of the Veterans Affairs National Leadership Board and reviewed by an American Federation of Government Employees union representative.

Measures

All participants completed an online survey. Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools. REDCap is a secure, web-based software platform designed to support data capture for research studies (Harris et al., 2009). Adapted from a CBT interview protocol utilized by Stirman and colleagues (Stirman et al., 2015), the survey included 28 closed-ended quantitative items and 10 open-ended qualitative items regarding therapist characteristics, CBT-CP modification, and completion and dropout rates. Regarding therapist characteristics, therapists reported on their clinical setting, geographic location, and professional background (Table 1). Therapists were asked about 1) penetration (the percentage of patients with chronic pain to whom they attempt to deliver CBT-CP; 2) how many patients they have seen for individual and group CBT-CP in the past three months; 3) modification of CBT-CP in terms of the percentage of patients for whom they modify CBT-CP (0–20%; 21–40%;41–60%;61–80%; 81–100%) and the extent to which they typically modify CBT-CP (not at all, a little, some, quite a bit, a lot); 4) and course of CBT-CP including the percentage of patients starting CBT-CP who drop out and the percentage completing CBT-CP.

Table 1.

Sample characteristics

Characteristic	n	% or Mean (s.d.)
Physically located in VA Medical Center	140	70.0
Psychologist (yes)a	140	67.9
Psychologists	95	67.9
Social Workers	42	30
Other	3	2.1
Tenure in field (years)	141	14.5 (s.d.=8.6)
Pain clinic (yes)b	119	18.5
Provide individual sessions	119	83.2
Provide group sessions	120	30.0
Proportion patients CBT-CP attempted	120	47.9 (s.d=35.9)
Proportion modified:	114
0–20
21–40		35.1
41–60		9.7
61–80		8.8
81–100		14.0
		32.5
Extent modified:	115
Not at all		13.9
A little		38.3
Some		22.6
Quite a bit		17.4
A lot		7.8

Note: CBT-CP=cognitive-behavioral therapy for chronic pain

^a= no psychologist = 0; psychologist = 1

^b= not pain clinic = 0; pain clinic = 1

Data Analysis

Descriptive statistics were calculated for therapist characteristics and therapist reported CBT-CP process variables (percentage of patients offered CBT-CP, percentage of patients completing CBT-CP, the percentage of patients for whom CBT-CP is modified, and the extent of modification). To examine the relationship between therapist characteristics and penetration and modification (percentage and extent), we conducted Pearson correlations, point biserial correlations, polychoric or polyserial correlations, as necessitated by variable type. To examine CBT-CP course completion and dropout, two separate models were developed. Therapist-reported percentage of completers and dropouts were estimated using beta distribution, a model that is tailored to represent proportional outcomes. A customization known as the “zero-one” inflated beta regression was employed to account for outcomes with observed scores that were exactly 0 or 1 (Ferrari & Cribari-Neto, 2004; Ospina & Ferrari, 2010; Swearingen et al., 2012). Specifically, the proportion of cases completing the full CBT-CP course was reported as 1.0 by 10 therapists and as 0 by 8 therapists; and, proportion of cases discontinuing CBT-CP was reported as 0 by 17 therapists (no 1’s were reported for rate of discontinuing). Initial beta regression models estimated the effect of predictors (therapist characteristics and modification) on the proportion of cases (an expected value, referred to as μ); precision parameters incorporating variance components (referred to as ɸ); and, parameters linked to the chances that the proportion is at one extreme or the other (referred to as π parameters). In the final models, all three CBT-CP practice variables were retained in final models; however, only therapist characteristics significant at p<.20 were retained in final models. Analyses were conducted using SAS software 9.4. Copyright © 2014 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA.

Qualitative analyses of open-ended questions were conducted using a conventional content analysis (Hsieh & Shannon, 2005). First, three coders read the responses and independently identified themes using an inductive approach (Crabtree & Miller, 1999). Coders then met and discussed emerging themes in the data and resolved discrepancies. During the ongoing coding process, the coders wrote memos and revised codes until a final a set of focused codes were developed. Focused coding was then used to code all of the responses. Analyses were conducted using Microsoft Excel version 2002.

Results

Four hundred and four CBT-CP therapists were invited to participate; 21 were unavailable because they were no longer with the VHA or on extended leave, leaving a total available sample of 383. One hundred and fifty four (40.2%) of the currently available therapists responded to the survey. Finally, two therapists who responded to the survey indicated they no longer provided CBT-CP, leaving a final total sample of 141 (36.8% of therapists still with the VHA). The majority of therapists were located in VA Medical Centers (versus community-based outpatient clinics, residential programs, etc.), trained as psychologists, and provided individual CBT-CP sessions (see Table 1).

CBT-CP Penetration

On average, therapists reported attempting CBT-CP with a little less than one half of their patients with chronic pain (mean = 48.8%, s.d.=35.7) in the last three months. While the modal response was that therapists attempted CBT-CP with 50% of patients with pain, the extremes (attempting with 0–20% or with 100%) were also common (see Figure 1). Therapists based in a VA Medical Center and who provided individual CBT-CP reported providing CBT-CP to a higher percentage of their patients (Table 2). Therapists who provided individual therapy (n = 99, 70.2%) had served about 9 patients (mean = 8.8, s.d. = 14.1) over the past three months, while those who provided group therapy (n = 97, 68.8%%) had served about 66 (mean = 65.6, s.d. = 26.7) during this same period.

Figure 1.

Percentage of Veterans with Chronic Pain Therapists Report Attempting CBT-CP

Table 2.

Association between therapist characteristics, CBT-CP attempts, proportion modified, and extent of modifications

Characteristic	Correlation coefficient (p-value)
	Percent patients CBT-CP attempted	Proportion of patients CBT-CP is modified	Extent of modification to CBT-CP
Physically located in VA Medical Center	.22 *	.06	.02
Psychologist (yes)a	.11	.28 *	.27*
Tenure in field (years)	−.08	.02	−.01
Pain clinic (yes)b	.06	.18	.18
Provide individual sessions	.30 ***	−.14	−.28 *
Provide group sessions	.18 *	.32 *	.03
Proportion patients CBT-CP attempted	1.0	-	-
Proportion modified:	−.05	1.0	-
Extent modified:	−.40 ***	.78 ***	1.0

Note: CBT-CP=cognitive-behavioral therapy for chronic pain

^*= p <.05

^**= p <.01

^***= p <.001

^a= no psychologist = 0; psychologist = 1

^b= not pain clinic = 0; pain clinic = 1

When asked “For those patients with chronic pain you do not attempt CBT-CP, what are the most common reasons you do not?”, therapists reported several reasons for not providing CBT-CP. The most commonly reported reasons were attributed to patients’ attitudes toward the practice (n = 54; 38.3%). For instance, several therapists reported patients do not like the home practice (e.g., “lack of interest or inability to comply with homework requirements”), the format (e.g., “Veteran not interested in that many psychotherapy sessions”), or that patients would prefer another form of treatment (e.g., “don’t believe ‘psychological interventions’ will help” and “they decline because they are only interested in medications”).

Many therapists reported not offering CBT-CP for reasons other than patient preferences. Some (n = 37, 26.2%) reported not offering CBT-CP due to clinical comorbidities that they believed make CBT-CP not the ideal option. For instance, “They have other mental illnesses that are the current focus of therapy” and “they have character/personality variables that would make a straightforward CBT approach not feasible.” Therapists also commonly reported offering an alternate treatment (irrespective of patient preferences; n = 34, 24.1%). Examples include “Referral to Pain School or Chronic Pain Program” and “hypnosis or ACT-CP [Acceptance-Commitment Therapy for Chronic Pain].” Finally, a number of therapists deem CBT-CP inappropriate for some patients: “not ready for EBP protocol,” “they need to process their pain journey/story,” “not the best fit for the patients,” and “I find CBT to be a bit dry and clients tend to be more needy and yet jaded at the same time and want other services.”

CBT-CP Modification

Most therapists reported modifying CBT-CP for either very few patients (0–20% of patients seen; 35.1% of therapists) or most of patients seen (81–100%; 32.5%). In contrast, the extent to which they modified was more evenly distributed, with “a little” being the modal response (Table 1). In terms of correlates of modification, psychologists reported more modification (extent and proportion of patients) than master’s level therapists (e.g., social workers, licensed mental health counselors, etc.). Therapists who provide CBT-CP in group format reported modifying for a greater proportion of patients, but extent of modification did not differ based on whether the therapist provided group CBT-CP.

The most common reasons therapists modified therapy pertained to patient clinical needs or preferences (n = 53; 41.1%). For example, one therapist stated, “most of the patients I see are dealing with multiple issues, so the CBT skills, pacing, etc. have relevance to other issues, and they lack the focus/patience to put aside the other material and focus exclusively upon pain.” Several therapists reported modifying to respond to clinical needs in the context of CBT-CP groups. For instance, “I see most of my pain patients in group. Consequently, I want to be aware of group needs, i.e. make room for them to speak and share, process and digest the material.” Some therapists (n = 28; 21.7%) indicated that they made modifications because, in their views, their changes made CBT-CP more effective. For example, “I have found offering alternative ways to manage unhelpful thoughts such as cognitive defusion…I also find it beneficial to discuss values when talking about functioning and quality of life.” Another therapist justified his or her modifications, stating, “there is not significant evidence supporting adherence to manualized models over work conducted by an experienced clinician in a topic area.” Other therapists reported modifying so that CBT-CP would better fit their clinic structure (n = 26; 20.2%). For example, several therapists noted that CBT-CP fits poorly in primary care settings (e.g., “not conducive to twelve 60-minute sessions” and “[primary care- mental health integration]” model only allows 3–4 sessions of therapy”). Others reported modifications to fit a group format (e.g., “session 1 [is modified] for obvious reasons because I cannot assess a group”) or “due to scheduling availability” (e.g., less frequent sessions because the therapist does not have weekly availability). Finally, some therapists reported modifying CBT-CP in response to patient logistical needs (n = 14; 10.9%) such as “case by case basis, sometimes it’s Veteran’s work schedule, sometimes [internet] connectivity issues” and if a “topic can be addressed in one session, [it] saves patient time and travel.”

Prediction of CBT-CP Completion and Dropout

The predictive model for the proportion of patients who completed their CBT-CP course is presented in Table 3. In the final model, only two variables remained significant or marginally significant: penetration and clinic location. Therapists who treat a larger percentage of their caseload with CBT-CP reported higher completion rates. For example, based on the model, for a therapist who employs CBT-CP with 25% of patients, the expected proportion completing CBT-CP course is 49%; whereas, at 75% attempted, the expected proportion completing CBT-CP course is 61%. Also, location in a pain clinic had a noticeably negative impact on CBT-CP course completion. Course completion is estimated at 54% in a pain clinic and 66% for other clinic types, keeping other variables constant. Neither proportion of patients for whom CBT-CP is modified nor extent of modification were related to proportion of patients completing.

Table 3.

Combined Model Predicting Therapist-Reported Rates of CBT-CP Completion (n=111)

Parameter	Variable	Estimate	Standard Error	t-value	Pr>|t|
μ	Intercept	−0.22	0.44	−0.51	.61
	Pain clinic	−0.49	0.28	−1.72	.09
	Individual sessions	0.55	0.38	1.44	.15
	Proportion Patients CBT-CP attempted	0.01	0.003	4.57	<.001
	Proportion modified	−0.05	0.07	−0.77	.44
	Extent modified	−0.09	0.11	−0.83	.41
Parameter	Variable	Estimate	Standard Error	t-value	Pr>|t|
ϕ	Intercept	0.02	0.64	0.03	.98
	Pain clinic	−0.77	0.37	−2.11	.04
	Individual sessions	0.64	0.42	1.52	.13
	Proportion Patients CBT-CP attempted	0.02	0.004	4.15	<.001
	Proportion modified	−0.21	0.11	−1.91	.06
	Extent modified	0.29	0.20	1.48	.14
Parameter	Variable	Estimate	Standard Error	t-value	Pr>|t|
π0	Intercept	−2.46	0.37	−6.66	<.001
π1	Intercept	−2.22	0.33	−6.70	<.001

Note: CBT-CP=cognitive-behavioral therapy for chronic pain, μ=mean of the response variable, ϕ=precision parameter, π₀, π₁=model for extreme outcome

The predictive model for the proportion of patients for whom therapists report drop out is presented in Table 4. Only a proportion of patients undergoing CBT-CP attempts was significant and therapist training as a psychologist was marginally related to lower reported dropout rate. As the proportion of attempts increases, the expected reported dropout rate decreases. For example, for a therapist who employs CBT-CP with 25% of patients, the expected proportion dropping out is 21%; whereas, at 75% attempted, the expected proportion of dropout is 14%. Also, doctoral-level psychologists (versus master’s level counselors, social workers, etc.) reported less dropout; dropout rate is estimated at 17% for psychologists and 23% for other types of therapists. Neither proportion of patients for whom CBT-CP is modified nor extent of modification were related to proportion of patients who dropout.

Table 4.

Combined Model Predicting Therapist-Reported Rates of CBT-CP Drop-out (n=111)

Parameter	Variable	Estimate	Standard Error	t-value	Pr>|t|
μ	Intercept	−1.00	0.36	−2.74	.007
	Psychologist	−0.35	0.18	−1.91	.06
	Proportion Patients CBT-CP attempted	−0.01	0.003	−2.79	.006
	Proportion modified	0.10	0.07	1.54	.13
	Extent modified	0.17	0.12	1.46	.15
Parameter	Variable	Estimate	Standard Error	t-value	Pr>|t|
ϕ	Intercept	1.52	0.59	2.58	.01
	Psychologist	−0.08	0.30	−0.25	.80
	Proportion Patients CBT-CP attempted	0.01	0.005	2.64	.009
	Proportion modified	−0.10	0.12	−0.86	.39
	Extent modified	−0.04	0.20	−0.19	.85
Parameter	Variable	Estimate	Standard Error	t-value	Pr>|t|
π0	Intercept	−1.71	0.26	−6.49	<.001

Note: CBT-CP=cognitive-behavioral therapy for chronic pain, μ=mean of the response variable, ϕ=precision parameter, π₀=model for extreme outcome

Discussion

The current study sought to explore the relationship between therapist factors, modification, and penetration rates on CBT-CP completion and dropout. After controlling for therapist factors such as background and clinical setting, as well as reported levels of modification, therapist penetration (reported percentage of eligible patients with whom CBT-CP is offered) was positively associated with reported CBT-CP completion rates. This finding is important; completion of a course of an EBP maximizes the chances that patients will realize the benefits demonstrated in clinical trials. The association between completion and penetration rates may have several explanations. Therapists reporting higher rates of both may have an overall more positive attitude toward CBT-CP. This explanation would be consistent with prior work showing that provider attitudes towards EBPs are predictive of EBP implementation, including penetration (Garcia, Mignogna, et al., 2019). Additionally, therapists may benefit from a practice effect in which attempting CBT-CP more frequently may hone clinical skills leading to greater success. Consistent with this idea, patients have identified therapist confidence in a given EBP as key to their own willingness to engage in these approaches (Hundt et al., 2015). Irrespective of the explanation, the positive relationship between therapist penetration and completion rates adds to literature indicating mutable therapist behavior may affect patient EBP completion.

Given the potential benefits of higher attempt rates, this variable bears additional examination. Overall rates of penetration of CBT-CP were high in comparison to rates found for other EBPs. It is difficult to compare directly to other findings since penetration was measured differently. CBT-CP therapists in our study averaged 9 individual CBT-CP cases in the past three months (the approximate length of time needed to complete a standard course of CBT-CP). Reports for PTSD EBPs have found therapists in specialty clinics provided about four hours per week (Finley et al., 2015); if therapists were following the recommended weekly sessions for about 12 weeks, that would equate to approximately 4 patients over three months. Other studies have found as few as two cases at a time (Rosen et al., 2016). While therapists reported higher penetration rates for CBT-CP in the current study, there is ample room for improvement, as on average therapists attempted CBT-CP with only half of patients with chronic pain. Moreover, it is possible that therapists with higher levels of CBT-CP use were more likely to respond to the survey and thus these findings may not be generalizable.

The reasons provided for not attempting CBT-CP are consistent with prior research on other EBPs. Patient factors such as patients’ attitudes, readiness, and clinical complexity are consistently cited by therapists as the key barriers to EBP implementation (Hamblen et al., 2015; Osei-Bonsu et al., 2017; Zubkoff et al., 2016). However, several findings provide reasons for optimism that CBT-CP (and other EBPs) may be successfully attempted with more patients. First, in our study some therapists reported very high levels of penetration, indicating that many therapists have overcome these barriers. Second, some evidence indicates that putative patient-level barriers may be driven at least in part by therapist related factors, rather than immutable, objective truths about patients. For instance, Garcia and colleagues (Garcia, DeBeer, et al., 2019) found therapists’ with certain theoretical orientations were more likely to assume patients would not prefer an EBP for PTSD; such therapists attitudes likely suppress even offering an EBP. Consistent with this idea, therapist characteristics such as attitudes toward EBPs have been linked to their utilization of these approaches (Cook et al., 2014). Given that therapists sometimes make treatment decisions based on their own understanding of patients’ preferences, it is particularly unfortunate that some therapists’ views of clinical contraindications for EBPs are inconsistent with available scientific evidence (Rosen, Matthieu et al., 2016; Finley, Garcia et al., 2015). Taken together, the barriers that therapists see may be strongly influenced by their own backgrounds, circumstances, knowledge of chronic pain, and subjective evaluations of patients, rather than immovable, objective characteristics of the patients they serve. Fortunately, several studies have shown that efforts to better prepare patients for participation in EBPs and better inform patients about EBPs through decision support aids have led to greater patient uptake (Feeny et al., 2009; Mott et al., 2014; Schumm et al., 2015). Importantly, a growing body of literature emphasizes the benefits of responsiveness to patient preferences in terms of not only dropout rates but clinical outcomes (Swift & Callahan, 2009; Swift et al., 2018).

While therapist factors may account for some variability in implementation, organizational factors should not be ignored. In the current study, respondents from specialty pain clinics reported lower completion rates than their colleagues in other settings; however, clinic location was not associated with the proportion of patients for whom CBT-CP is attempted (i.e., penetration). These findings could be attributable to factors associated with the inner setting of specialty clinics (e.g., culture, workflow, etc.) (Damschroder et al., 2015), the patient mix served by these clinics, or both. Contrary to our findings, prior work in PTSD EBPs found higher penetration to be associated with less competing demands (Chard et al., 2012; Rosen et al., 2016).Therapists in specialty pain clinics likely have fewer roles outside of providing CBT-CP than therapists in other settings (e.g., general outpatient mental health or primary care clinics). However, providers in specialty clinics also have more clinical expertise and tools and therefore may be more likely to change to another modality. Regarding the patient population, though, indicators of clinical severity such as inpatient hospitalization rates (Rosen et al., 2016) have been associated with lower penetration (Sripada et al., 2018) and higher dropout rates. Patients referred to pain clinics may be disproportionately severe, which may explain the lower reported completion rates. Future work should examine differences in pain specialty clinic settings and patient mixes and how therapists in these settings may be better supported to serve those patients.

Contrary to the notion that practice modification may increase fit and therefore acceptability of an EBP, this study found no association between modification and completion or dropout rates. The findings regarding CBT-CP modification were only partially consistent with extant literature. Therapists in the current study varied significantly, with many reporting modifying CBT-CP for nearly all patients, while others reported modifying for very few. This contrasts with Meza and colleagues’ (2019) findings, in which therapists reported modifying an EBP for PTSD for less than half of their patients; however, consistent with Meza as well as Stirman and colleagues (2013) in that most therapists in this study reported modifying the EBP at least a little. As with prior studies, the extent and frequency of modification varied substantially from therapist to therapist. While overall levels of modification are not associated with success, a more nuanced examination of modifications exploring type (e.g., fidelity-consistent modifications) (Marques et al., 2019) and processes (e.g., planned versus ad hoc) (Mackie et al., 2020) may provide additional insight. For instance, modifications made on unsubstantiated clinical instinct or in contrast with core elements of the model may be less successful than planned modification of EBPs that use alternative forms to fulfil essential functions of the EBP (e.g., changing maladaptive thinking) (Jolles et al., 2019; McGuire et al., 2020).

Taken together, the current study contributes to the extant literature in several ways. First, it contributes additional knowledge to the relatively neglected question of the role of therapist factors and setting in dropout and completion of EBPs. While prior research as well as therapist attitudes focus disproportionately on patient barriers, our study adds to a growing body of literature suggesting therapists and setting are at least equally associated. Secondly, while prior work has focused on therapists’ attitudes and background, the current study examines the role of a specific therapist behavior- attempting an EBP- on penetration. Behaviors may be more easily targeted, changeable, and potentially influence subsequent attitudes (Aronson, 1992). Finally, the current study expands the literature regarding EBP modification and penetration into a novel presenting condition- chronic pain. Patients with chronic pain and associated clinical interventions may differ systematically from more traditional psychological conditions; therefore, complementary results from this population addresses the generalizability of prior work.

The current findings should be viewed considering the study’s limitations. Most importantly, all data was self-reported by therapists, and is thus subject to certain biases. For instance, prior work has demonstrated that most therapists view themselves as “above average” and may therefore underestimate patient dropout (Kruger, 1999; Tracey et al., 2014). Additional research utilizing more objective measures of penetration, modification, and completion is needed, such as objective rating of recorded or directly observed sessions. Moreover, it is important to ultimately examine the association between modifications and patient outcomes. It should be noted that completion of a course of an EBP does not guarantee clinically significant improvement. The response rate for the current survey was modest and respondents were all VHA therapists; therefore, results may not be representative of all VHA therapists or community providers. The current study provides important insights into the extent, reasons, and interconnection of EBP penetration, modification, and completion rates which should be complemented with future research with more objective measures and additional samples. Taken together, though, there is reason to believe that additional penetration of CBT-CP is warranted, especially in light of the pervasive need for evidence-based treatments for chronic pain. Programs and providers may benefit from patient shared decision support tools that provide information on all available options and their support, thus supporting both greater EBP penetration and patient-centered care.

Public Significance Statement:

Trained therapists reported attempting an evidence-based psychotherapy for chronic pain with less than half of their eligible patients. Therapists outside specialty pain clinics and those with higher reported percentages of patients with whom they attempt the therapy reported lower dropout rates. These findings indicate trained therapists should offer information about evidence-based psychotherapies more frequently and engage patients in treatment decisions.

Biography

ALAN MCGUIRE received his PhD in clinical psychology from Indiana University-Purdue University at Indianapolis. He is currently a Clinical Research Psychologist at the Richard L. Roudebush VA Medical Center in the HSR&D Center for Health Information and Communication and an Associate Research Professor in the Department of Psychology at Indiana University-Purdue University at Indianapolis. His professional interests include dissemination and implementation of psychosocial interventions for chronic conditions.

MARIANNE MATTHIAS holds a PhD in health communication from Purdue University. She is a Research Scientist at the Roudebush VA Medical Center and Regenstrief Institute in Indianapolis, IN, and Associate Research Professor in the Department of General Internal Medicine and Geriatrics at the Indiana University School of Medicine. Additionally, she serves as Associate Editor of the Journal of General Internal Medicine and Pain Medicine. She is a health services researcher who focuses on communication in healthcare contexts, with particular interest in treatment decision-making in chronic pain care.

MARINA KUKLA received her PhD in clinical psychology from Indiana University-Purdue University at Indianapolis. She is currently a Research Scientist and Clinical Psychologist at the Richard L. Roudebush VA Medical Center in the HSR&D Center for Health Information and Communication and an Associate Research Professor in the Department of Psychology at Indiana University-Purdue University at Indianapolis. Her areas of professional interest include improving work functioning and other community outcomes in persons with mental illness.

NANCY HENRY received her BA in psychology from Indiana University. She is a project manager for the Richard L. Roudebush VA Medical Center in the HSR&D Center for Health Information and Communication and the Department of Psychology at Indiana University-Purdue University at Indianapolis. Her professional interests include mental health issues affecting those with serious persistent mental illness, recovery-oriented care, shared decision making, mental health provider burnout, implementation of evidence-based psychotherapies.

JESSICA CARTER received her BS in psychology from Indiana University. She is a research assistant for the Richard L. Roudebush VA Medical Center in the HSR&D Center for Health Information and Communication. Her professional interests are implementation and dissemination of evidence-based practices and health disparities.

MINDY FLANAGAN received her PhD in social psychology with a statistics minor from the University of Missouri-Columbia. She is currently a Senior Research Scientist at Parkview Health and Research Consultant at Richard L. Roudebush VA HSR&D Center for Health Information and Communication. Her professional interests include mixed methods research and applied data analysis.

MATTHEW BAIR received his MD from the Medical College of Wisconsin and MS in clinical research from Indiana University-Purdue University at Indianapolis. He is a Research Scientist for the Richard L. Roudebush VA Health Services Research and Development Center for Health Information and Communication, Staff Physician at Richard L. Roudebush VA Medical Center, and Regenstrief Research Scientist. His professional interests include chronic pain and psychological comorbidity and developing strategies to improve pain management in the primary care setting.

JENNIFER MURPHY received her PhD in clinical psychology from Adelphi University. She serves as the Director of Behavioral Pain Medicine for the Veterans Health Administration and is Master Trainer for VA’s Cognitive Behavioral Therapy for Chronic Pain (CBT-CP). Dr. Murphy is Associate Professor in the University of South Florida’s Morsani College of Medicine and serves on the editorial board of Pain Medicine. Her professional interests include developing and implementing nonpharmacological treatments for chronic pain across healthcare systems.