Retracted: The Impact of Treatment Description Format on Patient Preferences for Posttraumatic Stress Disorder Treatment
Abstract
enThe present study examined how the format in which treatment information is presented impacts individuals’ preferences for posttraumatic stress disorder (PTSD) treatments. Adults who screened positive for PTSD (N = 301) were randomized into groups to learn about five first-line treatments; participants either read sequential text descriptions or reviewed a comparison chart that presented side-by-side information. Participants rated treatment acceptability, rank ordered treatments from most to least preferred, and indicated their confidence in this ranking. Compared with participants in the text group, those in the chart group assigned more favorable acceptability ratings to prolonged exposure therapy (PE) and more moderate ratings to medications. Cognitive processing therapy was the most common first-choice treatment (43.6%). Forced-choice treatment rankings were similar across conditions, although participants in the chart group ranked PE more favorably than those in the text group, odds ratio (OR) = 0.54, 95% CI [0.35, 0.82], p = .004. Confidence in treatment rankings did not differ across conditions. The results suggest that perceptions of treatment acceptability can be influenced by the format in which treatment information is presented. In settings where the goal is to increase treatment acceptability, side-by-side formats may offer an advantage over sequential descriptions of each treatment.
Resumen
esSpanish Abstracts by Asociación Chilena de Estrés Traumático (ACET)
El impacto del formato de descripción del tratamiento en las preferencias del paciente para el tratamiento del trastorno de estrés postraumático
DESCRIPCION Y PREFERENCIA DEL TRATAMIENTO PARA TEPT
El presente estudio examinó cómo el formato en el que se presenta la información del tratamiento impacta las preferencias de los individuos en los tratamientos para el trastorno de estrés postraumático (TEPT). Los adultos que resultaron positivos para TEPT fueron asignados al azar en grupos para aprender acerca de cinco tratamientos de primera línea; los participantes leyeron descripciones de texto secuenciales o revisaron un cuadro comparativo que presentaba información en paralelo. Los participantes calificaron la aceptabilidad del tratamiento, clasificaron los tratamientos ordenados de mayor a menor preferencia e indicaron su confianza en esta clasificación. En comparación con los participantes del grupo de texto, los del grupo del cuadro comparativo asignaron calificaciones de aceptabilidad más favorables a la terapia de exposición prolongada (PE por sus siglas en ingles) y calificaciones más moderadas a los medicamentos. La terapia de procesamiento cognitivo fue el tratamiento de primera elección más frecuente (43,6%). Las clasificaciones de tratamiento de elección forzada fueron similares en todas las condiciones, aunque los participantes del grupo del cuadro comparativo clasifico la PE más favorablemente que los del grupo de texto, razón de probabilidades (OR)= 0.54, 95% IC [0.35, 0.82], p= .004. La confianza en las clasificaciones del tratamiento no difirió entre las condiciones. Los resultados sugieren que las percepciones de aceptabilidad al tratamiento pueden estar influenciadas por el formato en el cual se entrega la información. En lugares donde el objetivo es aumentar la aceptabilidad al tratamiento, los formatos en paralelo pueden ofrecer una ventaja sobre las descripciones secuenciales de cada tratamiento.
抽象
zhTraditional and Simplified Chinese Abstracts by the Asian Society for Traumatic Stress Studies (AsianSTSS)
簡體及繁體中文撮要由亞洲創傷心理研究學會翻譯
The Impact of Treatment Description Format on Patient Preferences for Posttraumatic Stress Disorder Treatment
Traditional Chinese
標題: 療程資訊呈現的方式對患者的創傷後壓力症治療取態的影響
撮要: 本研究檢視療程資訊呈現的方式如何影響患者對創傷後壓力症(PTSD)治療的取態。我們把被篩檢出有PTSD的成人(N = 301)隨機分組。樣本要認識五種首輪治療, 一組要閱讀順序的文字描述, 另一組則看一個把資訊並列的比較表。然後, 他們對療法可接受度給予評分, 把療法按最想選以至最不想選的方式排序, 亦為對療法的信心評分。相比閱讀文字描述的樣本, 閱讀比較表的樣本對延長暴露療法(PE)的可接受度評分較高, 對服藥的評分亦較為適度。認知整理治療是樣本中最普遍的首選療法(43.6%)。雖然, 閱讀比較表的樣本比閱讀文字描述的樣本, 對PE的評分較高)勝算比(OR) = 0.54, 95% CI [0.35, 0.82], p = .004(, 但強迫選擇治療在兩組所得的評分也差不多。兩組樣本對療法的信心評分並無分別。結果反映, 療法資訊呈現的方式能影響患者對療法可接受度的感知。假如希望提升療法對患者的可接受度的話, 使用比較表把療法資訊並列出來可能會比順序描述更為有效。
Simplified Chinese
标题: 疗程信息呈现的方式对患者的创伤后压力症治疗取态的影响
撮要: 本研究检视疗程信息呈现的方式如何影响患者对创伤后压力症(PTSD)治疗的取态。我们把被筛检出有PTSD的成人(N = 301)随机分组。样本要认识五种首轮治疗, 一组要阅读顺序的文字描述, 另一组则看一个把信息并列的比较表。然后, 他们对疗法可接受度给予评分, 把疗法按最想选以至最不想选的方式排序, 亦为对疗法的信心评分。相比阅读文字描述的样本, 阅读比较表的样本对延长暴露疗法(PE)的可接受度评分较高, 对服药的评分亦较为适度。认知整理治疗是样本中最普遍的首选疗法(43.6%)。虽然, 阅读比较表的样本比阅读文字描述的样本, 对PE的评分较高)胜算比(OR) = 0.54, 95% CI [0.35, 0.82], p = .004(, 但强迫选择治疗在两组所得的评分也差不多。两组样本对疗法的信心评分并无分别。结果反映, 疗法信息呈现的方式能影响患者对疗法可接受度的感知。假如希望提升疗法对患者的可接受度的话, 使用比较表把疗法信息并列出来可能会比顺序描述更为有效。
Posttraumatic stress disorder (PTSD) can be treated effectively with a variety of interventions. Recommended treatments for PTSD include trauma-focused psychotherapies, such as prolonged exposure (PE), cognitive processing therapy (CPT), and eye movement desensitization and reprocessing (EMDR), as well as medications from two antidepressant classes: selective serotonin reuptake inhibitors (SSRIs) and serotonin norepinephrine reuptake inhibitors (SNRIs; Department of Veterans Affairs/Department of Defense [VA/DoD], 2017). With few exceptions, head to head comparisons suggest that these interventions are similarly effective at reducing PTSD symptoms (Jonas et al., 2013), and there is little empirical evidence to guide the selection of one first-line treatment over another for specific patients. However, these first-line treatments differ with respect to how they work, what they require of the patient, and their risk and benefit profiles. Individual patients, therefore, may have unique perspectives on which treatments are most and least desirable. Although there is a growing body of literature assessing patient preferences for various PTSD treatments, several knowledge gaps remain.
A systematic review of 41 PTSD treatment preference studies found that, overall, most participants preferred psychotherapy over medication (Simiola, Neilson, Thompson, & Cook, 2015). This finding was consistent across studies conducted within analog samples asked to choose which treatment they would hypothetically prefer (Becker et al., 2009) and studies of treatment-seeking participants with diagnosed PTSD (Kehle-Forbes, Polusny, Erbes, & Gerould, 2014; Shalev, 2012). Little is known, however, about preferences between first-line psychotherapies. Most studies have examined preferences for one psychotherapy versus one medication (e.g., Feeny, Zoellner, & Kahana, 2009; Kehle-Forbes et al., 2014; Zoellner et al., 2018) or compared one first-line psychotherapy with non–first-line options (e.g., Najavits, Kivlahan, & Kosten, 2011). Among the preference studies that have included more than one first-line psychotherapy option (Becker et al., 2009; Mott, Stanley, Street, Grady, & Teng, 2014; Schumm, Walter, Bartone, & Chard, 2015; Shalev, 2012; Tarrier, Liversidge, & Gregg, 2006), the only two studies that included all of the most highly recommended psychotherapy options (i.e., CPT, PE, and EMDR; VA/DoD, 2017) used analogue samples in which most participants did not present with PTSD symptoms (Becker et al., 2009; Tarrier et al., 2006).
An additional limitation of the existing PTSD preference literature is that most studies have asked patients to rank order preferred treatments or to identify a single preferred treatment without assessing preferences separately for each intervention; Tarrier et al. (2006) is an exception. When a participant ranks treatments or selects a single top choice, the strength of the preference for one treatment over another is not known. The top-ranked treatment could be much preferred over alternatives or simply seen as the slightly better choice among several similarly appealing, or unappealing, options. In a recent methodological review of patient preferences research in mental health, the authors recommended assessing both absolute and relative treatment preferences (Mott, Koucky, & Teng, 2015).
It is interesting to note that some PTSD preference studies have provided no information about how patients were educated about potential treatment options (e.g., Shalev, 2012). This is problematic because, in these studies, descriptions of the treatments may have varied. For example, patients may have received more comprehensive information about one treatment versus another treatment, or different patients may have received different descriptions of the same treatment or treatments. Among those PTSD preference studies that have described how patients were educated, most asked participants to read sequential, standardized text descriptions or, in some cases, to view sequential video descriptions (Angelo, Miller, Zoellner, & Feeny, 2008) of treatment options. Although sequential treatment descriptions allow patients to give their undivided attention to each treatment option individually, this presentation method may not be ideal for facilitating comparison across treatments. After reading several descriptions, patients’ recall for details of earlier treatment descriptions may be limited. In addition, sequential presentation of treatment information is particularly vulnerable to ordering effects, as users tend to perceive that information presented first is more important (Feldman-Stewart, Chammas, Hayter, Pater, & Mackillop, 1996).
There is some evidence that displaying treatment information in a chart format, whereby information on multiple treatments can be viewed simultaneously, helps users draw direct comparisons and may facilitate decision making (Feldman-Stewart & Brundage, 2004). Participants in decision-making studies have demonstrated a preference for strategies that require the least amount of effort and allow for the simultaneous consideration of options (Wiggins & Bollwork, 2006). A review found that decision support tools that included side-by-side information displays were rated by respondents as being more “balanced” than those that did not include side-by-side presentations (Abhyankar et al., 2013). Although this prior work suggests that information format may impact decision making, we are unaware of any study to date that has evaluated how different methods of presenting treatment information may impact PTSD treatment preferences.
In the current study, we compared two methods of presentation for standardized information about first-line PTSD treatments. We examined whether these delivery formats differentially impacted three aspects of participants’ treatment preferences: acceptability, forced-choice ranking, and confidence in their ranking. We addressed the limitations in the existing literature by assessing preference for each treatment individually in addition to asking patients to rank order-preferred interventions.
Method
This study used data collected as part of a national online survey examining multiple aspects of preferences for PTSD treatment decision making. The Dartmouth College Committee for the Protection of Human Subjects determined that this study was exempt from the requirement for approval. This study was approved by the White River Junction VA Medical Center Research and Development Committee.
Participants
Participants were members of an online research panel maintained by the GfK Group (formerly Knowledge Networks), a professional survey firm. The panel included a nationally representative sample of approximately 55,000 adult members recruited through probability-based sampling. The GfK Group obtained consent from all panel members during an initial panel recruitment interview. A randomly selected subset of panel members (n = 2,655) was recruited via email to participate in the online survey. A total of 1,944 panel members (73.2% participation rate) accessed the link to the survey and completed the Primary Care PTSD Screen (PC-PTSD) for the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; Prins et al., 2004). Of these individuals, 355 (18.3%) were eligible for participation on the basis of a positive PC-PTSD screen (i.e., a score of 3–4) suggesting probable PTSD. Each of these 355 participants completed and submitted the survey. By design, we aimed to fulfill the following four quota groups: 35% male veterans, 15% female veterans, 35% male nonveterans, and 15% female nonveterans. Quota groups were determined based on our interest in including both veterans and nonveterans. Given that the veteran population is predominately male, we purposefully sampled 30% women to ensure enough female respondents to examine potential gender effects.
Of the 355 completed cases, 54 were removed due to quota control (i.e., the respondent's data were removed by the survey firm because their respective quota group was already full), yielding a final sample of 301 participants (see Figure 1). Participants received $10 (USD) for completing the survey.
Participant characteristics are displayed in Table 1. The final participant sample included 192 men (63.8%) and 109 women (36.2%). Participants’ mean age was 49.11 years (SD = 16.21). By design, half the sample comprised United States veterans (n = 149, 49.5%). All participants reported exposure to at least one traumatic event and identified their index trauma as interpersonal assault (24.6%), witnessing sudden or violent death (13.3%), life-threatening injury or illness (13.3%), accident (13.0%), combat (10.3%), natural disaster (6.0%), fire explosion (4.0%), or other (15.5%). The majority of participants (79.0%) reported that they had never been diagnosed with PTSD and most (83.7%) denied a history of PTSD treatment.
| Total(N = 301) | Chart(n = 157) | Text(n = 144) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Characteristic | n | % | n | % | n | % | χ2(N = 301) | df | p |
| Female sex | 109 | 36.2 | 60 | 38.2 | 49 | 34.0 | 0.57 | 1 | .450 |
| Race | 5.92 | 3 | .205 | ||||||
| White | 181 | 60.1 | 100 | 63.7 | 81 | 56.3 | |||
| Black | 34 | 11.3 | 18 | 11.5 | 16 | 11.1 | |||
| Hispanic | 53 | 17.6 | 28 | 17.8 | 25 | 17.4 | |||
| Other | 32 | 10.6 | 11 | 7.0 | 21 | 15.3 | |||
| U.S. veteran | 149 | 49.5 | 75 | 47.8 | 74 | 51.4 | 0.39 | 1 | .531 |
| Education | 5.97 | 3 | .113 | ||||||
| Less than high school | 22 | 7.3 | 12 | 7.4 | 10 | 6.8 | |||
| High school | 75 | 24.9 | 34 | 20.9 | 41 | 27.7 | |||
| Some college | 121 | 40.2 | 71 | 43.6 | 50 | 33.8 | |||
| College degree or higher | 83 | 27.6 | 36 | 22.1 | 47 | 31.8 | |||
| PC-PTSD score | 0.56 | 1 | .454 | ||||||
| 3 | 125 | 41.5 | 62 | 39.5 | 63 | 43.8 | |||
| 4 | 176 | 58.5 | 95 | 60.5 | 81 | 56.3 | |||
| Index trauma type | 0.76 | 3 | .859 | ||||||
| Physical assault | 39 | 13.0 | 21 | 13.9 | 18 | 13.1 | |||
| Sexual trauma | 35 | 11.6 | 19 | 12.6 | 16 | 11.7 | |||
| Combat | 31 | 10.3 | 14 | 9.3 | 17 | 12.4 | |||
| Other | 183 | 60.8 | 97 | 64.2 | 86 | 62.8 | |||
| Prior PTSD therapy | 43 | 14.3 | 19 | 12.1 | 24 | 16.7 | 1.27 | 1 | .258 |
| Prior PTSD medication | 40 | 13.3 | 18 | 11.5 | 22 | 15.3 | 0.95 | 1 | .330 |
- Note. Differences between groups were nonsignificant for all demographic variables. PTSD = posttraumatic stress disorder; PC-PTSD = Primary Care PTSD Screen.
Procedure
Study participants (N = 301) responded to online survey items that assessed demographics, treatment history, and preferences for treatment characteristics; these data have been published previously (Harik, Hundt, Bernardy, Norman, & Hamblen, 2016; Harik, Matteo, Hermann, & Hamblen, 2017). Using a computer-generated 1:1 randomization procedure, participants were then randomly assigned to review descriptions of five first-line PTSD interventions (PE, CPT, EMDR, SSRIs/SNRIs, and stress inoculation training [SIT]) either by reading sequential text descriptions of each treatment or by reviewing a comparison chart that included side-by-side information about each intervention. To minimize the potential impact of ordering effects, the sequence in which the treatments were presented in the sequential text condition and the order in which the treatments were listed in the comparison chart was randomized for each participant. Immediately after reviewing treatment options, participants responded to items that assessed treatment preferences.
The five treatments featured in the present study represent those that received the highest recommendation in the 2010 VA/DoD (2010) guideline for the management of PTSD. Of note, the updated VA/DoD guideline (2017), which was published after data collection for the current study was completed, no longer gives the highest recommendation to SIT; additionally, the update expands its first-line recommendations to include additional trauma-focused therapies beyond CPT, PE, and EMDR, such as written exposure therapy, narrative exposure therapy, and brief eclectic psychotherapy.
Treatment descriptions (see Supplementary Materials) were drafted by the first and third authors and revised according to feedback from the team of experts at the National Center for PTSD responsible for the development of educational materials on PTSD and PTSD treatment. The same information was conveyed in both formats, and in both conditions, treatment descriptions were written at or below an eighth-grade reading level, according to the Flesch-Kincaid readability formula. Within each condition, descriptions for each treatment were matched as closely as possible with respect to organization, sentence structure, wording, and length.
The sequential treatment descriptions featured separate summaries of each featured treatment. Descriptions began with a brief overview paragraph, and additional information was organized under the following three headings: How does it work?, What will I do in this treatment?, and What are the risks? With the exception of the occasional use of bulleting to organize lists of information, the descriptions were written in full sentences and organized into paragraphs. On average, descriptions were 267 words in length (range: 225–299 words).
The comparison chart featured side-by-side information about each of the five featured interventions. Each treatment represented a column in the chart, and information about the treatment was organized within rows with the following labels: What type of treatment is this?, How does the treatment work?, Is the treatment effective?, What are the risks?, Is this a group or individual treatment?, Will I talk in detail about my trauma?, Will I need to practice anything between sessions?, and How long does treatment last? The comparison chart included a brief answer to each question for each of the five interventions. Many of the answers were not complete sentences. For example, in the row labeled “Is this a group or individual treatment?”, the answer for PE was simply “individual.” As such, the mean word count of 147 words (range: 125–166 words), inclusive of row labels, for comparison chart descriptions was shorter than that of the sequential text descriptions even though they contained similar factual information.
Measures
PTSD screen
The PC-PTSD (Prins et al., 2003) is a four-item, nondiagnostic screening tool for PTSD. An introductory sentence instructs participants to consider if they have experienced a traumatic event. Four yes or no items then assess the presence of four types of past-month symptoms related to the event; items reflect DSM symptoms clusters and relate to reexperiencing, avoidance, numbing, and hyperarousal symptoms. The screen is considered positive if a respondent answers yes to any three items (i.e., a total score of 3 or higher), indicating probable PTSD. In a primary care sample (N = 188), the PC-PTSD has demonstrated strong psychometric properties for a cutoff score of 3 (sensitivity = .78, specificity = .87; Prins et al., 2003) when compared with the Clinician-Administered PTSD Scale (Blake et al., 1995), which is considered to be the gold standard in PTSD diagnostic assessment. Although a newer version of this measure, which corresponds to the fifth edition of the DSM, exists (i.e., the PC-PTSD-5; Prins et al., 2016), it was not available at the time of data collection.
Treatment ranking and preference
To assess treatment acceptability, participants were instructed to use a 4-point Likert-type scale ranging from 1 (definitely) to 4 (definitely not) to indicate whether they would consider participating in each of the featured treatments. Participants were then instructed to rank the five treatments from 1 to 5 according to their personal preference, with 1 being the most preferred. Participants also indicated their confidence in their ranking (“How confident are you in your ranking?”) on a 4-point Likert-type scale ranging from 1 (very confident) to 4 (not confident).
Data Analysis
We assessed whether PTSD treatment acceptability, rankings, and confidence in these rankings varied by the format of treatment information (sequential text or comparison chart). To examine differences in treatment acceptability ratings by randomized condition, we conducted a separate chi-square analysis for each of the five treatments (i.e., CPT, PE, SIT, EMDR SSRI/SNRI). To account for multiple comparisons, we performed a Bonferroni correction whereby we divided the critical value by the number of comparisons (.05/5); an alpha of .01 was used as a cutoff for statistical significance.
To examine the impact of treatment format on forced-choice rankings, we constructed a series of cumulative odds ordinal logistic regression models with proportional odds. For ranked data, ordinal regression offers an advantage over other approaches given that it accounts for the ordinal nature of the outcome (Ananth & Kleinbaum, 1997). In each model, condition was entered as a predictor variable, and the ranking for each treatment served as the dependent variable in its respective model. Additional candidate predictor variables included gender, race (White vs. non-White), education (high school diploma or less vs. post–high school education), veteran status (veteran vs. nonveteran), trauma type (physical, sexual, combat, other), age, any prior therapy for PTSD (yes vs. no), and any prior medication for PTSD (yes vs. no). Candidate predictor variables that were significant at a p value of .10 in a bivariate ordinal regression model were entered simultaneously as independent variables in the multivariable model. The odds ratio (OR) for each predictor variable represents the odds that the group ranked the treatment rank level lower (i.e., more favorably) relative to participants in the other condition. For each model, we assessed goodness of fit using chi-square tests and the proportional odds assumption using a test for parallel lines. Finally, we used chi-square tests to assess for differences between conditions with respect to confidence in assigned treatment rankings and again performed a Bonferroni correction to account for multiple comparisons (α = .01). For all analyses, we included all available data. Given the small amount of missing data (range: 0%–4.7% missingness among included variables), we did not attempt to impute or otherwise substitute missing values. All analyses were conducted with SPSS (Version 25).
Results
Treatment Acceptability
Frequency data for the treatment acceptability ratings are presented in Table 2. Across the full sample, 22.6% of participants reported that they would definitely (i.e., score of 1) consider CPT, and 18.6%, 15.9%, 15.3%, and 9.3% said that they would definitely consider SIT, SSRI/SNRI, PE, and EMDR, respectively. The proportion of participants who indicated that they would definitely not (i.e., score of 4) consider the treatment ranged from 10.3% for CPT to 28.6% for SSRI/SNRI. For all treatments, the modal response from the full sample was possibly (i.e., score of 3).
| CPT | PE | SIT | EMDR | SSRI/SNRI | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Response | n | % | n | % | n | % | n | % | n | % |
| Definitely | 68 | 22.6 | 46 | 15.3 | 56 | 18.6 | 28 | 9.3 | 48 | 15.9 |
| Probably | 97 | 32.2 | 87 | 28.9 | 83 | 27.6 | 74 | 24.6 | 71 | 23.6 |
| Possibly | 98 | 32.5 | 113 | 37.5 | 125 | 41.5 | 123 | 40.9 | 92 | 30.6 |
| Definitely not | 31 | 10.3 | 50 | 16.6 | 34 | 11.3 | 72 | 23.9 | 86 | 28.6 |
- Note. CPT = cognitive processing therapy; PE = prolonged exposure; SIT = stress inoculation training; EMDR = eye movement desensitization and reprocessing; SSRI/SNRI = selective serotonin reuptake inhibitors/serotonin norepinephrine reuptake inhibitors.
As shown in Figure 2, chi-square analyses revealed significant differences between the sequential text and comparison chart conditions regarding treatment acceptability ratings for two of the five featured treatments. Relative to participants in the sequential text group, those in the comparison chart group gave more favorable acceptability ratings to PE, χ2(3, N = 296) = 11.50, p = .009; and more moderate rankings (i.e., probably or possibly as opposed to definitely or definitely not) for SSRIs/SNRIs, χ2(3, N = 298), 22.97 p < .001. The ratings for CPT, χ2(3, N = 296) = 0.67, p = .880; EMDR, χ2(3, N = 297) = 7.92, p = .048; and SIT, χ2(3, N = 297) = 9.18, p = .020, were similar across the two conditions.
Treatment ranking
Frequency data for treatment rankings are presented in Table 3. When instructed to identify their most preferred treatments, 43.6 % of the overall sample selected CPT as their first-choice intervention, 22.3% selected SIT, 19.5% selected SSRIs/SNRIs, 11.8% selected PE, and only 2.8% selected EMDR. Bivariate analyses revealed three significant (i.e., p <.10) candidate predictors for CPT (age, prior medication for PTSD, and prior therapy for PTSD) and two significant candidate predictors for SSRI/SNRIs (age and prior medication for PTSD); these variables were entered as predictors in ordinal logistic models predicting CPT ranking and SSRI/SNRI ranking, respectively. There were no significant candidate predictors for PE, SIT, or EMDR.
| CPT | PE | SIT | EMDR | SSRI/SNRI | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Chart | Text | Chart | Text | Chart | Text | Chart | Text | Chart | Text | |||||||||||
| Rank | n | % | n | % | n | % | n | % | n | % | n | % | n | % | n | % | n | % | n | % |
| 1 | 64 | 42.7 | 61 | 44.5 | 26 | 17.3 | 8 | 5.8 | 36 | 24.0 | 28 | 20.4 | 3 | 2.0 | 5 | 3.6 | 21 | 14.0 | 35 | 25.5 |
| 2 | 43 | 28.7 | 31 | 22.6 | 28 | 18.7 | 23 | 16.8 | 29 | 19.3 | 36 | 26.3 | 18 | 12.0 | 21 | 15.3 | 32 | 21.3 | 26 | 19.0 |
| 3 | 23 | 15.3 | 20 | 15.0 | 47 | 31.3 | 43 | 31.4 | 32 | 21.3 | 35 | 25.5 | 27 | 18.0 | 27 | 19.7 | 21 | 14.0 | 12 | 8.8 |
| 4 | 13 | 8.7 | 18 | 13.1 | 28 | 18.7 | 36 | 26.2 | 32 | 21.3 | 29 | 21.2 | 66 | 44.0 | 38 | 27.7 | 11 | 7.3 | 16 | 11.7 |
| 5 | 7 | 4.6 | 7 | 5.1 | 21 | 14.0 | 27 | 19.7 | 21 | 14.0 | 9 | 6.6 | 36 | 24.0 | 46 | 33.6 | 65 | 43.3 | 48 | 4.3 |
| M | SD | M | SD | M | SD | M | SD | M | SD | M | SD | M | SD | M | SD | M | SD | M | SD | |
| Averge | 2.04 | 1.16 | 2.12 | 1.25 | 2.93 | 1.27 | 3.37 | 1.15 | 2.82 | 1.38 | 2.67 | 1.21 | 3.76 | 1.02 | 3.72 | 1.19 | 3.45 | 1.54 | 3.12 | 1.65 |
- Note. CPT = cognitive processing therapy; PE = prolonged exposure; SIT = stress inoculation training; EMDR = eye movement desensitization and reprocessing; SSRI/SNRI = selective serotonin reuptake inhibitors/serotonin norepinephrine reuptake inhibitors.
The results of the ordinal logistic regression models are presented in Table 4. In all ordinal logistic regression models, the proportional odds assumption was not violated as indicated by a p value greater than .01: CPT model, p = .948; SSRI/SNRI model, p = .404; PE model, p = .345; SIT model, p = .130; and EMDR model, p = .029. Randomized condition (i.e., sequential text vs. comparison chart) significantly predicted PE ranking such that chart participants ranked PE lower (i.e., more favorably) than text participants, OR = 0.54 95% CI [0.35, 0.82], p = .004. Prior use of medication significantly predicted CPT ranking such that participants who had not previously used medication for PTSD assigned a more favorable ranking, OR = 0.14, 95% CI [0.04, 0.46], p = .001. Older participants, OR = 0.99 CI [0.97, 1.00], p < .001, and those with a history of medication use for PTSD, OR = 3.75, CI [1.99, 7.04], p < .001, assigned more favorable rankings to SSRIs/SNRIs compared to younger participants and those who had not previously used medication for PTSD. All other associations were nonsignificant. The final regression models for CPT, χ2(4, N = 296) = 17.92 p = .001; PE, χ2(1, N = 296) = 8.55, p = .003; and SSRI/SNRIs, χ2(3, N = 298) = 26.35 p = .001, improved upon the intercept-only model, which did not control for predictor variables, whereas the models for EMDR and SIT did not improve upon the intercept only model, ps = .393–.921.
| Model | Outcome | Predictor | OR | 95% CI |
|---|---|---|---|---|
| 1 | CPT rank | |||
| Comparison chart | 0.98 | [0.64, 1.50] | ||
| No prior PTSD medications | 0.14*** | [0.04, 0.46] | ||
| No prior PTSD therapy | 3.13 | [0.97, 10.21] | ||
| 2 | PE rank | Comparison chart | 0.54** | [0.35, 0.82] |
| 3 | SIT rank | Comparison chart | 1.20 | [0.80, 1.81] |
| 4 | EMDR rank | Comparison chart | 0.98 | [0.64, 1.48] |
| 5 | SSRI/SNRI rank | |||
| Comparison chart | 1.52 | [0.99, 2.32] | ||
| No prior PTSD medications | 3.75*** | [1.99, 7.04] | ||
| Participant age | 0.99* | [0.97,1.00] |
- Note. OR = odds ratio; CPT = cognitive processing therapy; PE = prolonged exposure; SIT = stress inoculation training; EMDR = eye movement desensitization and reprocessing; SSRI/SNRI = selective serotonin reuptake inhibitors/serotonin norepinephrine reuptake inhibitors; PTSD = posttraumatic stress disorder.
- *p < .05. **p < .01. ***p < .001.
Confidence in Ranking
One-quarter of the full sample (25.8%) reported that they were very confident in their ranking, 36.9% were fairly confident, 26.8% were somewhat confident, and 10.8% were not confident. There were no significant differences in confidence ratings in sequential text and comparison chart participants, χ2(3, N = 298) = 6.92, p = .075.
Discussion
The present study assessed PTSD treatment preferences in a sample of adults with PTSD symptoms and examined whether treatment information format (i.e., sequential text vs. comparison chart) affected these preferences. Among the full study sample, CPT was the most preferred treatment in terms of both acceptability and forced-choice ranking. A preference for CPT relative to other treatments has been observed in several other studies. For example, among 183 veterans attending an outpatient VA PTSD treatment orientation group, veterans had a stronger preference for CPT over other psychotherapies (Schumm et al., 2015). Similarly, data from a small PTSD shared decision-making pilot trial demonstrated that veterans differentially preferred CPT over other evidence-based and non–evidence-based treatments (Mott et al., 2014). Given that some research suggests that patients’ PTSD treatment decisions are often heavily influenced by their perceptions of the treatment mechanism, there may be some aspects of the CPT mechanism that patients find particularly appealing (Angelo et al., 2008; Zoellner, Feeny, Cochran, & Pruitt, 2003).
In general, medications were seen as less acceptable than psychotherapies; most participants selected a psychotherapy as their first-choice treatment, and the SSRI/SNRI treatment option had the lowest overall acceptability ratings of any treatment. This finding is consistent with a growing number of studies showing that people prefer psychotherapy for PTSD over medications (Kehle-Forbes et al., 2014; Schumm et al., 2015; Simiola et al., 2015). Although some treatments were ranted as more preferable than others, a relatively small portion of the sample said they would definitely not be willing to try a specific treatment, with the percentage of participants who chose this rating option ranging from 10.3% for CPT to 28.6% for SSRI/SNRI. Most participants, therefore, were willing to consider each of the first-line treatments for PTSD. This finding highlights the importance of assessing acceptability for each treatment in addition to considering rankings or relative preferences. For example, although only 2.6% of the sample selected EMDR as a first-choice treatment, more than 30% said that they would definitely or probably consider this treatment.
Overall, we observed that the format of treatment information had a larger impact on the acceptability of some treatments compared to others. Whereas format had no bearing on the acceptability of CPT, SIT, and EMDR, participants in the chart group viewed PE as more acceptable than those in the text group. Prior research suggests that information displayed side by side in a chart format facilitates both comparisons between the treatments and a deeper understanding of each individual treatment (Feldman-Stewart & Brundage, 2004). It may be that the acceptability of PE increases as an individual gains a more in-depth understanding of this approach. Indeed, prior research suggests that providers view exposure therapy more positively as their understanding of the protocol and rationale increases (Ruzek et al., 2016). For SSRI/SNRIs, we observed a slightly different pattern such that participants in the chart group endorsed more moderate acceptability ratings (i.e., were less likely to indicate that they would definitely or definitely not consider medication) than participants in the text group. Given that SSRI/SNRIs were the only featured medication option, our data cannot determine whether this pattern is unique to SSRIs/SNRIs or would be observed with other medications.
The treatment information format influenced forced-choice treatment rankings for PE but not for the other four featured treatments (i.e., CPT, EMDR, SIT, SSRI/SNRI). Participants who viewed the comparison chart ranked PE more favorably than those who viewed the sequential text descriptions. It is not clear why format had a unique effect on PE. However, one distinctive aspect of PE is that it uses exposure techniques (i.e., patients are asked to purposefully and repeatedly recall the details of their trauma), which are often initially perceived as counterintuitive to PTSD patients. Patients are sometimes reluctant to initiate PE due to fear of symptom exacerbation (Cook et al., 2013; Hundt et al., 2015), though education about the mechanism and effectiveness of exposure is seen as key to increasing patients’ comfort with these techniques. It is possible that the comparison chart format facilitated understanding of the treatment mechanism and/or effectiveness, translating into a more favorable ranking.
We examined the influence of a number of demographic and clinical factors on forced-choice rank order and observed that only two of these variables—a history of medication use and age—were predictive of treatment ranking. Specifically, participants who had previously used medications for PTSD ranked medication more favorably, whereas those who had never used medication ranked CPT more favorably. One plausible explanation is that participants with a prior positive experience with medication are more accepting of future medication use. However, prior research has been inconsistent, with some studies suggesting that a history of medication use is related to a preference for medications (Kehle-Forbes et al., 2014) and others showing no association (Feeny et al., 2009). Older age was also predictive of a more favorable ranking for medication. Given that older adults experience more concurrent illnesses and are prescribed more medications than younger people, older adults may have more experience with medications in general and, therefore, see medication as a more acceptable option (Gallagher, Ryan, Kennedy, & O'Mahony, 2008). Although little is known about the PTSD treatment preferences of older adults, research suggests that older adults with depression prefer counseling or psychotherapy over medication (Gum et al., 2006; Unützer et al., 2003). Confidence in treatment decision, which is often evaluated in the context of treatment decision making as one index of high-quality decision making (Burton, Blendell, Jones, Fraser, & Elwyn, 2010; Diefenbach et al., 2012), did not differ across the two randomized groups. Thus, participants were similarly confident in their ranking regardless of whether they viewed the comparison chart or sequential text descriptions.
The results of the present study offer insight into the influence of information format on three dimensions of treatment preference: acceptability, forced-choice ranking, and confidence. However, several limitations warrant mention. All participants screened positive for PTSD but were not required to have a formal diagnosis of PTSD nor did they represent a treatment-seeking sample. Participants were only educated about first-line interventions (VA/DoD, 2010); preferences may differ according to the number and type of alternative options presented. Although we collected general information about participants’ prior receipt of medication and psychotherapy for PTSD, we did not measure prior exposure to each of the featured treatments, and it is unknown how participants’ baseline knowledge of these protocols may have shaped their preferences. Our goal was to provide equivalent information in the two formats (i.e., comparison chart and sequential text), but the information was not identical. The formats differed with respect to length and content, and they used different key questions to organize the information. It is possible that these subtle differences contributed to observed differences between the two randomized conditions. Given that participants were randomized to a single format of treatment information, we could not evaluate participant preferences for one format versus the other. We also did not compare the two conditions with respect to time spent reviewing treatment options. Given that time pressure is known to influence participants’ preferred information format (Morrison, Wiggins, & Porter, 2010), future studies may wish to consider how certain presentations of information interact with different time constraints. Similarly, it may be of interest to test whether a comparison chart format is preferred by individuals with PTSD who may be experiencing memory deficits; specifically, the comparison chart may help individuals overcome working memory limitations by allowing simultaneous review and comparison of various treatments. Future studies are also needed to investigate how other format variations—such as variations in length (short vs. long) and delivery format (online vs. print, graphics vs. no graphics)—may impact opinions about care. Similarly, in addition to evaluating the impact of format on patient preferences, it will be important to examine whether other downstream outcomes, such as knowledge about treatment engagement, initiation, completion, and response, are impacted by format.
In summary, the present results demonstrated that participants who viewed a treatment comparison chart rated several specific psychotherapies as more acceptable than participants who viewed sequential text descriptions of the same treatments. Treatment information format had little influence on forced-choice rankings of treatment and no influence on confidence in the ranking. These results suggest that the format in which information is provided impacts patients’ perceived acceptability of alternate treatments but may not impact their first-choice selection. In settings where the goal is to increase treatment acceptability to multiple treatments, side-by-side formats, such as comparison charts, may offer an advantage over individual sequential descriptions of each treatment.
