Does expectancy affect insomnia treatment response in cancer survivors receiving acupuncture and cognitive behavioral therapy?

Abstract

Purpose

Insomnia negatively affects quality of life in cancer survivors. Expectations of insomnia treatment efficacy may influence response to intervention. We sought to determine whether pre-treatment outcome expectancy predicts response to two non-pharmacological interventions for insomnia among cancer survivors.

Methods

We analyzed data from a randomized clinical trial that compared acupuncture versus cognitive behavioral therapy for insomnia (CBT-I) in cancer survivors. Patient expectancy was measured by the Mao Treatment Expectancy Scale (MTES) at baseline. Insomnia severity was assessed using the Insomnia Severity Index (ISI) at treatment completion (week 8). Multivariate linear regression was used to evaluate the associations between pre-treatment expectancy and ISI score at week, 8 adjusting for co-variates.

Results

Expectancy for acupuncture and CBT-I were similar at baseline (acupuncture: 13.3 ± 4.0; CBT-I: 13.2 ± 2.9, p = 0.17). Greater baseline expectancy scores were associated with a greater and statistically significant insomnia severity reduction at week 8 in the acupuncture group (beta coefficients [Coef.] = − 0.35, 95% confidence interval [CI] = − 0.6 to − 0.1, p = 0.016) adjusted for co-variates. Baseline expectancy was not statistically associated with insomnia severity reduction in the CBT-I group (Coef. = − 0.2, 95% CI = − 0.7 to 0.2, p = 0.31). High expectancy was significantly associated with greater proportion of treatment responders at week 8 in the acupuncture group (76% vs. 38%, p = 0.001) but not in the CBT-I group (83% vs. 70%, p = 0.21).

Conclusions

Higher pre-treatment outcome expectancy predicted significantly greater insomnia improvement in patients receiving acupuncture but not in those receiving CBT-I.

Implications for Cancer Survivors

Aligning treatment provision with expected outcomes may lead to personalized non-pharmacological insomnia management for cancer survivors.

Introduction

Insomnia affects up to 50% of cancer survivors [1–3] and negatively impacts their quality of life [2]. Although pharmacological therapy is only recommended for short-term use (less than 4 weeks) by the American College of Physicians (ACP) [4], the majority of patients (50.3%) use medications for more than 6 months [5]. Longer-term use of pharmacological sleep aids significantly increases the risk of drug tolerance and dependence [1]. Guidelines developed in the USA and Europe recommend cognitive behavioral therapy for insomnia (CBT-I) is as the first-line treatment for insomnia disorder in the general population [4, 6] and in cancer survivors [7]. Robust evidence indicates that CBT-I has comparable short-term and better long-term efficacy compared to pharmacotherapy [8, 9]. CBT-I also improves mood, fatigue, and overall quality of life in cancer survivors [10]. Acupuncture is another non-pharmacological option for the treatment of insomnia. Increasing evidence suggests that acupuncture may extend total sleep duration and improve sleep quality, daytime functioning, and quality of life; however, the findings are inconsistent [11–15]. When multiple treatment options for insomnia exist, consideration of personal beliefs is essential to guide patient-centered care and help cancer survivors make optimal treatment decisions.

Identifying how outcome expectancy—the belief that improvements will be achieved [16]—can predict patient response to a particular therapy and may play an important role in developing personalized approaches to insomnia research and management. A recent study focusing on non-specific effects of insomnia treatments illustrated that expectation of treatment effect rather than choice of treatment improved insomnia outcomes including insomnia severity, sleep quality, and insomnia-related fatigue [17]. In a comparative effectiveness study of CBT and exercise for pain, the proportion of participants reporting positive outcome was higher among those expecting improvement compared to those not expecting improvement in both groups [18]. These findings are consistent with a pooled analysis of four acupuncture randomized controlled trials (RCTs) with 864 participants, which also demonstrated a significant relationship between pre-treatment outcome expectancy and acupuncture outcomes in patients with chronic pain [19]. To date, there has been no research examining the role of expectancy in either acupuncture or CBT-I for the treatment of insomnia in cancer populations.

Understanding expectancy in the context of patient beliefs may increase the overall treatment effect, which is essential to provide evidence-based medicine and patient-centered care [20]. In our recently published CHOICE study, both CBT-I and acupuncture produced clinically meaningful and durable reductions in insomnia severity with minimal side effects among survivors of diverse cancer types [21]. The current study is a pre-specified secondary analysis of the parent trial to evaluate the association between pre-treatment expectancy and reduction of insomnia severity among cancer survivors who received acupuncture or CBT-I.

Materials and methods

Study design, participants, and procedures

The details of the original trial have been published elsewhere [21, 22] and are summarized here. The study was a dual-center, parallel randomized controlled trial to compare the effectiveness of acupuncture and CBT-I for insomnia in cancer survivors. Participants received CBT-I or acupuncture over the course of 8 weeks, and outcomes were assessed at week 8. The study was conducted between March 2015 and July 2017. This study was approved by the institutional review boards at the University of Pennsylvania and Memorial Sloan Kettering Cancer Center. The protocol was registered at ClinicalTrials.gov (identifier: NCT02356575).

English-speaking adult patients diagnosed with cancer of all types and stages were considered eligible. They also were required to meet the criteria for insomnia disorder defined by the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) and score eight or greater on the Insomnia Severity Index (ISI). Active treatment such as surgery, chemotherapy, and/or radiotherapy needed to be completed at least 1 month prior to study initiation. Patients receiving continued hormone treatment or maintenance targeted therapies, or on stable doses of psychotropic medication, or hypnotics or sedatives, also remained eligible.

Patients were not eligible if they had a sleep disorder other than insomnia, such as delayed/advanced sleep phase syndrome, or restless legs syndrome, that were not adequately treated; had previous experience with CBT or acupuncture to treat insomnia; had other psychiatric disorders that were not in remission or adequately treated; were employed in a job requiring shift work that would impair the ability to establish a regular sleep schedule; or received additional medication, such as corticosteroids, which may induce insomnia. Patients using psychotropic medication remained eligible provided that the dose had been stable for the past 6 weeks. Considering the high use of benzodiazepines such as sedatives and hypnotics within cancer survivors, patients using the two medicines were also considered eligible for study participation. We tracked the use of sleep medication during the entire study period. Trained research study assistants and staff conducted screening and diagnostic interviews to confirm patient eligibility.

Eligible patients provided informed consent and completed the baseline assessment, which included measurement of treatment expectancy, demographic characteristics (age, gender, race, and education), and clinical factors (cancer type, time since cancer diagnosis, insomnia severity, and time since insomnia). Pre-treatment expectancy for acupuncture and CBT-I was measured separately for each patient. Then, patients were sequentially randomized and stratified by study site using permuted block randomization. The study biostatistician generated the randomization sequence prior to participant recruitment. Randomization information was concealed in a letter inside a sealed envelope, which the participant opened after completing the baseline assessment. The study investigators including statistician were blinded to treatment assignment. Patients, research staff, and therapists were not blinded. Treatment expectancy was assessed only at baseline. Insomnia severity was evaluated at both baseline and week 8 (end of treatment). The study protocol has been published with details [23].

Interventions

CBT-I is a multicomponent intervention that addresses the behaviors that contribute to the development and maintenance of insomnia, while also identifying and restructuring thoughts and beliefs that may contribute to, or reinforce, behaviors that produce pre-sleep arousal and/or performance anxiety. Over time, CBT-I works to reassociate sleep-related stimuli to elicit sleep as opposed to wakefulness. The intervention consists of five components: sleep restriction, stimulus control, cognitive restructuring, relaxation training, and education. Four licensed therapists and five psychology trainees delivered the intervention. The first treatment session was 60 min, and remaining sessions were 30 min each. A total of seven CBT-I sessions were delivered over 8 weeks, once weekly for the first 5 weeks and then twice weekly for the remaining 3 weeks.

Acupuncture is a component of Traditional Chinese medicine (TCM) that involves inserting thin needles into targeted areas of the body. It has gained popularity in the USA in recent decades due to its safety and efficacy [24]. In this study, the acupuncture regimen used a manualized protocol, which was developed based on a Traditional Chinese Medicine Acupuncture Textbook [25] and in consultation with local and China-based acupuncturists. Acupoint selection was semifixed and included standardized points to address insomnia and additional points to treat comorbid symptoms, such as anxiety, pain, or fatigue, as needed. Four licensed acupuncturists with 11–14 years of experience delivered the intervention using sterile, single-use, metallic needles to achieve “De Qi,” a needle sensation including feelings of soreness, numbness, fullness, burning, heaviness, and/or aching [26]. Needle retention for every treatment was 30 min. A total of ten treatments were performed over 8 weeks, twice weekly for the first 2 weeks and then weekly for an additional 6 weeks.

Primary exposure: expectancy

The Mao Treatment Expectancy Scale (MTES) was developed by the senior author (JM) and has been validated in various socio-cultural contexts to evaluate treatment expectancy in a wide range of populations, including cancer, according to patient age and health conditions [27–30]. This four-item instrument asks patients to rate on a five-point scale (where 1 is total disagreement with a statement and 5 is total agreement) their expected effect of treatment on insomnia. The items of this instrument include insomnia improvement, the ability to cope with insomnia, symptoms related to insomnia, and energy level. The scale has demonstrated reliability with a Cronbach’s α of 0.82 [31]. We modified the MTES for CBT-I and acupuncture by changing the word “treatment” to “cognitive behavioral therapy” or “acupuncture.” The scores range between 4 and 20, with a higher score indicating greater expectancy.

Primary outcome: insomnia severity

The Insomnia Severity Index (ISI) is a well-validated measure of perceived insomnia severity and impact on daytime functions [32]. This seven-item instrument targets the subjective symptoms and consequences of insomnia as well as the degree of insomnia-related distress [32]. Each item is rated on a 0 to 4 scale (0, no problem; 4, very severe problem), and the total score ranges from 0 to 28 with a higher score suggesting more severe insomnia. A reduction of eight points is considered clinically significant improvement [33]. As such, patients with reduction of 8 or greater on the ISI score from baseline to post-treatment (week 8) were considered treatment responders.

Statistical analyses

The sample size of 158 participants was predetermined by the parent study [21]. The primary statistical approach was linear regression to model post-treatment ISI scores at week 8. The primary covariate of interest was the pre-treatment expectancy (MTES score). Two separate regressions were fitted, for the acupuncture and the CBT-I groups, respectively. We wanted to first examine, with each group, whether the association between expectancy and ISI scores was reliable. We first established that, within the acupuncture group, there was a statistically reliable association between expectancy and ISI scores, adjusting for baseline ISI score. Next, to account for putative contributors of insomnia, we included additional four covariates (age, gender, education, and race). All results reported included these covariate adjustments.

Additionally, we dichotomized ISI treatment response into responder vs. non-responder as per ≥ 8 point reduction rule [33] so that it facilitates the clinical interpretation of the data. Dichotomization on MTES was also done for the same purpose (MTES ≥ 14 as “high pre-treatment expectancy” vs. “low pre-treatment expectancy”). Chi-square test was used to evaluate if baseline expectancy type was associated with treatment responder type.

Data analyses followed the intention-to-treat principle (analyzed as randomized). Missing assessment scores were excluded in the analyses. A sensitivity analysis was done, using last value carried forward, to evaluate the influence of missing questionnaire responses, despite that only 7% of respondents had missing data. Inferential statistics were based on a two-sided p value of 0.05 or lower for statistical significance. Statistical analyses were conducted using STATA (version 15.0; STATA Corporation, College Station, TX) and SAS (version 9.2; SAS Institute, Inc, Cary, NC).

Results

Participant characteristics

As previously reported [21], we screened 604 patients for eligibility between February 2015 and March 2017. Of those, 444 patients were excluded due to ineligibility, no interest, scheduling difficulty, and location issues. The remaining 160 patients were randomly assigned to acupuncture or CBT-I groups with 80 patients in each group. One hundred fifty-eight participants completed the expectancy assessment of two treatments at baseline and 147 patients (75 in the acupuncture group and 72 in the CBT-I group) provided the ISI evaluation at week 8 (Fig. 1). The observed adherence rates were high in both groups (97.3% in acupuncture and 86.1% in CBT-I).

Fig. 1.

Consolidated Standards of Reporting Trials (CONSORT) diagram describing participant flow through the study

The pre-treatment characteristics of the participants can be seen in Table 1. The mean age was 61.3 (SD, 11.6 years), 90 (57.0%) were women, 46 (29.1%) self-identified as a race other than white, and 113 (71.5%) had a college or above education level. The most common cancer types were breast (31.7%) and prostate (22.2%). The mean time since the cancer diagnosis was 6.1 (SD, 5.4 years) and time since insomnia onset was 9.2 (SD, 9.1 years). The pre-treatment ISI score was 18.1 (SD, 4.3) ranging between 9 and 28.

Table 1.

Sociodemographic and clinical characteristics

	Total		Acupuncture		CBT-I
Characteristics	No	%	No	%	No	%
	158	100	80	50.6	78	49.4
Mean age (SD), y	61.3 (11.6)		62.3 (11.4)		60.3 (11.9)
Gender
Male	68	43.0	37	46.3	31	39.7
Female	90	57.0	43	53.8	47	60.3
Race
White	112	70.9	61	76.3	51	65.4
Nonwhite*	46	29.1	19	23.8	27	34.6
Education
Under college	45	28.5	24	30	21	26.9
College or above	113	71.5	56	70	57	73.1
Cancer type
Breast	50	31.7	24	30	26	33.3
Prostate	35	22.2	19	23.8	16	20.5
Other†	73	46.2	37	46.3	36	46.2
Years since cancer diagnosis, mean (SD)	6.1 (5.4)		6.4 (5.1)		5.7 (5.6)
ISI score, mean (SD)	18.1 (4.3)		17.6 (4.1)		18.6 (4.5)
Years since insomnia onset, mean (SD)	9.2 (9.1)		9.7 (9.6)		8.5 (8.6)
Pre-treatment expectancy, mean (SD)
Expectancy for acupuncture treatment	13.3 (3.6)		13.3 (4.0)		13.3 (3.2)
Expectancy for CBT-I treatment	13.0 (3.2)		12.8 (3.5)		13.2 (2.9)

CBT-I, cognitive behavioral therapy for insomnia; ISI, Insomnia Severity Index

^*Other includes Asian and more than one race

^†Other cancer types included colorectal, head and neck, hematologic, gynecologic, skin, lung, other gastrointestinal, and > 1 cancer type

Among all 158 patients, the pre-treatment expectancy score for acupuncture and CBT-I was similar (13.3 ± 3.6 vs. 13.0 ± 3.2), and it also did not differ between treatment groups (acupuncture: p = 0.94, CBT-I: p = 0.40) (Table 1). Higher pre-treatment expectancy for CBT-I was moderately correlated with higher pre-treatment expectancy for acupuncture (correlation coef. = 0.6, p < 0.001).

Association between expectancy and insomnia severity

A greater pre-treatment expectancy was significantly associated with a greater reduction in post-treatment ISI scores at week 8 in the acupuncture group. A one point increase in the baseline MTES score was associated with a 0.4-point decrease in week 8 ISI outcome (lower ISI means less insomnia severity) (Coef. = − 0.4, 95% CI = − 0.7 to − 0.1, p = 0.003). After adjusting for demographic characteristics (age, gender, race, and education), expectancy remained significantly associated with the ISI score at week 8 in the acupuncture group (Coef. = − 0.35, 95% CI = − 0.6 to − 0.1, p = 0.016) (Table 2). None of the demographic characteristics showed significant association with week 8 ISI score in the multivariate model. No statistically significant association was found between pre-treatment expectancy score and ISI score at week 8 in the CBT-I group (Coef. = − 0.2, 95% CI = − 0.7 to 0.2, p = 0.31). Sensitivity analyses using last value carried forward did not change our primary findings.

Table 2.

Predictors of Insomnia Severity Index (ISI) at week 8—multivariate regression analysis in the acupuncture group

	Coefficient	95% CI*	P
MTES † for acupuncture	− 0.4	− 0.6 to − 0.1	0.016
ISI‡ score	0.2	− 0.1 to 0.5	0.11
Age	0.0	− 0.1 to 0.1	0.47
Gender	0.1	− 1.9 to 2.2	0.91
Race	− 1.2	− 3.7 to 1.3	0.35
Education	1.7	− 0.7 to 4.1	0.16
Constants	6.6	− 3.2 to 16.4	0.18

^*CI, confidence interval

^†The Mao Treatment Expectancy Scale

^‡Insomnia Severity Index

Figure 2 shows the calculated insomnia severity change over time by high and low pre-treatment expectancy status in each treatment group. As illustrated, in the acupuncture group, patients with high treatment expectancy scores exhibited more insomnia severity reduction than those with low treatment expectancy scores (2.5 points) at week 8 (p = 0.02). However, in the CBT-I group, the difference in insomnia severity between patients with high and low treatment expectancy scores at week 8 was not significant (1.6 points; p = 0.28).

Fig. 2.

Insomnia Severity Index (ISI) change score over time by pre-treatment expectancy

Treatment response based on high vs. low expectancy

To further explore the association between pre-treatment expectancy and treatment response, we conducted additional analysis using dichotomized expectancy and treatment responder status. Figure 3 shows that in the acupuncture group, patients with high expectancy are twice as likely as those with low expectancy (76% vs. 38% p = 0.001) to be treatment responders. However, in the CBT-I group, treatment response was not associated with pre-treatment expectancy (83% vs. 70%, p = 0.21).

Fig. 3.

Treatment response by high versus low expectancy

Discussion

Insomnia is a prevalent and persistent symptom that negatively affects quality of life among cancer survivors and requires individualized management. Using data from a comparative effectiveness trial of diverse cancer survivors with insomnia, we demonstrated an association between pre-treatment outcome expectancy and insomnia severity in the acupuncture group but not in the CBT-I group. In the acupuncture group, patients with high expectancy were twice as likely to achieve a clinically significant response to treatment than those with low expectancy. This result suggests that pre-treatment expectancy can be used to personalize insomnia management in cancer survivors.

To our knowledge, this trial is the first study evaluating the role of pre-treatment expectancy on insomnia outcome for individual CBT-I in cancer survivors. Current evidence shows that higher pre-treatment expectancy is related to better outcomes for CBT in patients with anxiety [34], obsessive compulsive disorder [35], depression [36, 37], panic disorder, and agoraphobia [38]. Although we were not able to identify literature on the role of expectancy on treatment outcome of CBT-I in cancer survivors, one case series study (N = 86) was conducted in group CBT-I among non-cancer population with insomnia that used the ISI as an outcome [39]. Consistent with our study, treatment expectancy was not associated with a reduction in insomnia severity. However, the researchers measured during-treatment expectancy (following the second treatment) rather than pre-treatment expectancy. They did, however, find a modest interaction between patient expectancy and therapeutic alliance and the change in total wake time. The most essential components of CBT-I focus on behavioral changes and vary according to specific health conditions [40–42]. Therefore, when evaluating the effect of outcome expectancy on a particular group of therapies, consideration needs to focus on specific components of the intervention. Pre-treatment expectancy may be relevant for some but not for others.

Our study also offers novel insight on how pre-treatment expectancy predicts insomnia outcomes as related to acupuncture treatment in cancer survivors. Prior research has evaluated the impact of expectation in relation to acupuncture were conducted with patients who have pain; however, no studies have been conducted with cancer survivors who have insomnia [43, 44]. One systematic review has shown expectancy is relevant to acupuncture response in five out of seven studies conducted in musculoskeletal pain, including two with a large sample size (> 500 participants) [44]. Our results differ from the only other study conducted on evaluating the role of expectancy on insomnia in 116 non-cancer participants [45]. This study utilized the terminology expectancy but evaluated the during-treatment credibility measured by the Credibility of Treatment Rating Scale (confidence in effectiveness, recommendation to others, perceived logic, etc.), which was fundamentally different from the baseline outcome expectancy in our trial [46]. Previous studies have shown that expectancy seemed to predict outcome more often than credibility, suggesting that these two terminologies are separate constructs [16].

In the context of patient-centered care, assessment of patient beliefs, including expectancy about treatment interventions, is an essential element of evidence-based medicine [47, 48]. As suggested in our study, the different predicting roles of pre-treatment expectancy on outcome using CBT-I and acupuncture could be key in tailoring treatments to manage insomnia among cancer survivors. Our results suggest that pre-treatment expectancy does not affect outcome for CBT-I but it does for acupuncture. For patients with high expectancy for acupuncture, both acupuncture and CBT-I provide an almost identical reduction in insomnia; therefore, the final treatment decision rests on availability of providers as well as patient preference. However, for patients with low expectancy for acupuncture, they should be directed to CBT-I, as CBT-I is twice as likely to improve their treatment response for insomnia than acupuncture. Furthermore, future research should evaluate what modifiable factors that may increase expectancy for acupuncture. Understanding of these factors may help inform targeted intervention to enhance the therapeutic effect of acupuncture for insomnia and other outcomes.

Findings of the current study should be considered within the context of several limitations. First, the parent study was a randomized trial; thus, patients that had extremely high or low expectancy for only one treatment may not have been willing to participate. Therefore, the range of observed expectancy values may be more narrow than what may be observed in a clinical setting. Second, patients in this trial were not blinded because CBT-I and acupuncture both require active participation. Third, additional treatment process variables, such as therapeutic alliance or the interaction between expectancy and other clinical or treatment-related factors, were not assessed. Fourth, this study did not measure the change of expectancy during the whole treatment process as the study aimed only to evaluate the predictive effect of pre-treatment expectancy towards outcome; however, some studies have shown that expectancy may change throughout the therapeutic course [49], which may also have an effect on the final outcome. Further, our study is an RCT with good adherence in both groups. While both expectancy and adherence in the real world setting may be different from an RCT, their relationship needs to be further examined in real world clinical settings.

Despite these limitations, the current study has a number of strengths including using data from a rigorous randomized trial that included a diverse population. In addition, a validated instrument was used to evaluate outcome expectancy. The measure of insomnia severity is also validated in cancer survivors with well-established minimally clinically important change scores. Our results show baseline outcome expectancy may predict response to acupuncture but not to CBT-I among the cancer population. This finding provides evidence of personalized insomnia management based on expectancy to improve sleep for cancer survivors.

Abbreviations

CBT-I

Cognitive behavioral therapy for insomnia

MTES

Mao Treatment Expectancy Scale

ISI

Insomnia Severity Index

ACP

American College of Physicians

RCTs

Randomized controlled trials

DSM-5

Diagnostic and Statistical Manual of Mental Disorders, 5th Edition

TCM

Traditional Chinese medicine