Self-management intervention improves patient adherence to swallowing exercises during radiation for head and neck cancer

Abstract

Background:

While preventive swallowing exercises reduce the risk of radiation-associated dysphagia in head and neck cancer patients, strategies are needed to improve patient adherence.

Methods:

Before radiation, all participants were taught preventive swallowing exercises and randomized to either an adherence intervention or enhanced usual care. During radiation, all participants met twice with a speech pathologist for swallowing assessment and reinforcement of exercises. Intervention participants met weekly with a counselor in-person or by phone. At 6-week post-radiation follow-up, all participants completed a follow-up assessment of self-reported adherence, which was then corroborated with medical record documentation.

Results:

Newly diagnosed pharyngeal and laryngeal cancer patients without distant metastases were randomized (n=265; 135 to intervention, and 130 to usual care). Intervention participants were more likely to adhere to exercises during radiation compared to the control group (p<.0001).

Conclusion:

The weekly in-person adherence intervention program significantly increased patient’s adherence to preventive swallowing exercises during radiation.

1. BACKGROUND

It is estimated that nearly 32,000 Americans will develop laryngeal or pharyngeal cancers in 2024.¹ Five-year survival has risen dramatically from 35.4% for those diagnosed in 1975 to 74.7% for those diagnosed in 2010.² High-dose radiation treatment is often a component of management of laryngeal and pharyngeal cancers and can cause permanent sequelae, including fibrosis of swallowing muscles or dysphagia.³ Once radiation associated dysphagia is clinically detected, full restoration to pre-treatment function is rare.^4,5 Up to 63% of patients develop radiation-associated dysphagia within a 1–2 year period after treatment, the severity of which ranges from not being able to swallow foods without liquid assist, to being completely dependent on percutaneous gastromy (PEG) tube feeding.^6,7

Prevention of dysphagia is therefore important for head and neck cancer survivors. In addition to radiation strategies designed to spare the pharyngeal muscles, speech and language pathology strategies include preventive swallowing exercises, device-driven exercises (e.g., lingual resistance, expiratory strength trainers) and myofascial release and lymphedema therapies).^{8–10,11–19} In advanced oropharynx cancer patients, exercise therapy during radiation reduced the incidence of self-reported post-treatment dysphagia.^13,20–22

However, due to the significant toxicity profile of radiation and particularly chemoradiation,²³ patient adherence to preventive swallowing exercises is challenging: up to 80% of patients experience mucositis and over 50% experience moderate-to-severe levels of fatigue, dry mouth, radiation skin burn, and difficulty swallowing.^24,25 As symptom burden intensifies and side effect management becomes paramount, patient adherence progressively declines during the later weeks of radiation.^14,26 As such, overall rates of patient adherence to unsupervised in-home swallowing exercises range from 13²² to 70%,²⁷ depending on the timeframe relative to radiation treatment. ²⁸

Among the studies reporting reasons for patient nonadherence to swallowing exercises, having a limited sense of urgency in the absence of dysphagia, ²² forgetting exercises due to the competing demands of treatment,²⁹ oral pain and fatigue were other frequently cited reasons for patient nonadherence. ¹⁴ Other studies have also reported that poor patient understanding of the rationale for swallowing recommendations and prophylactic exercises is an important component to nonadherence.^26,29

The problem of patient nonadherence to medical regimens is not specific to cancer settings. Within the wider body of research examining modifiable factors associated with nonadherence to medical regimens, early meta-analyses found moderate effect sizes for depression (OR= 3.0)³⁰ and for lack of social support (OR=3.6).³¹ Given head and neck cancer patients’ suboptimal adherence to prophylactic swallowing exercises and considering the multifactorial nature of patient adherence, there is a need for effective adherence interventions that integrate evidence-based strategies within a systematic framework. These theoretical frameworks can not only explain why individual patients behave differently from each other, they can also be used to identify specific strategies for different subgroups of patients.

Leventhal et al.’s theory of Self-Management may be a reasonable model to understand how an individual’s psychosocial factors, disease characteristics and symptom burden interact to prompt nonadherence to swallowing exercises. The model assumes that control lies within the patient, and that an individual’s perceptions of his or her illness will shape his or her coping response to illness demands.³² Specifically, there are five dimensions of illness perceptions that regulate coping: identity, cause, consequences, timing and perceived control. We designed a 10-session weekly psychosocial intervention called PREPARE to target these perceptions, and measured these illness perceptions at the beginning and end of the PREPARE intervention to determine whether they mediated patient adherence to swallowing exercises.

The intervention was delivered by a trained counselor who met with the patient once a week for ten weeks, starting with the first week of radiation. The intervention targeted illness perceptions with education about radiation side effects and the rationale for exercises, practical side effect management tips and communication skills training. Patients were encouraged to perceive their swallowing exercises as not only achievable, but also necessary for eventual return to normalcy. First-person narratives from former patients, practical tips and encouraging quotes were selected to foster positive emotional coping with the rigors of head and neck cancer treatment.

The main purpose of this study was to test the efficacy of the intervention in improving adherence to preventive swallowing exercises during radiation. Its secondary purpose was to identify potential mechanisms for intervention effects by conducting both mediation analyses on patients’ illness perceptions, coping, and appraisal of coping and moderator analyses for depression, pain, fatigue, and social support (Figure 1).

Figure 1. Planned Mediation and moderation analyses.

Planned mediation and moderation analyses. Mediation models were run as unconditional and conditional models. For the conditional models, baseline age, sex, race, education level, marital status, income, chemotherapy status, diseases site, and tumor stage were controlled. Four separate moderation models (one for each moderator) were run by using multiple ordinary least squares regression model including independent variable, moderator, and their interaction (product) term.

2. METHODS

2.1. Participants.

Following Institutional Review Board approval at M. D. Anderson Cancer Center, informed consent was obtained for each patient prior to enrollment onto study. Patients were eligible if they had received radiation treatment at M. D. Anderson for oropharyngeal, laryngeal, hypopharyngeal, nasopharyngeal, or an unknown primary cancer with cervical metastases, were stage II-IVb (AJCC 7^th ed.) for oropharyngeal and laryngeal cancer or stage I-IVB for hypopharyngeal and nasopharyngeal cancer, were at least 18 years old, and spoke English well enough to complete the questionnaires. Patients were excluded if they had distant metastases, recurrent or progressive disease, other cancer diagnoses (except non-melanoma skin cancer), or previous head and neck surgery (excluding tonsillectomy, previous biopsy, or partial laryngectomy).

2.2. Design

After obtaining informed consent, participants were randomized 1:1 to either the adherence intervention plus 3 speech pathology visits during radiation or to 3 speech pathology visits alone (enhanced usual care). An adaptive randomization algorithm was used to balance for tumor stage greater than III, age over 70, gender, concurrent chemotherapy, and disease site. Participants completed psychosocial and cancer symptom questionnaires at study entry prior to radiation start and again after the completion of the 10-week PREPARE intervention. If questionnaires were not received within 2 weeks, questionnaires were either mailed again with postage-paid return envelopes or completed in clinic during appointments.

2.2.1. Speech pathology services

Prior to radiation start, all participants were referred to a study speech pathologist for baseline evaluation and swallowing exercise training who was blinded to treatment status. At weeks 3 and 5 of radiation, all participants were scheduled for follow-up speech pathology appointments to monitor swallowing function and assess adherence to swallowing exercises. Study staff called and rescheduled missed appointments to minimize patient no-show rates.

The following eight exercises were part of the standard regimen prescribed to head and neck cancer patients about to start radiation: the Shaker (isometric and isokinetic), Mendelsohn, jaw stretching, supraglottic swallow, pitch glide, Masako, and effortful swallow. Each of these exercises were to be repeated 3–5 times per set, with 4 sets of exercises to be performed every day, for a total of 15–20 minutes. Some exercises, such as the Shaker, were withheld at the discretion of the treating speech pathologist when anatomical abnormality or dysfunction prevented safe performance of the exercise.

2.2.2. Enhanced Usual Care Arm.

In addition to ensuring that speech pathology services were scheduled, rescheduled and received as described above (representing an enhancement to usual care), all patients in the control group received an instructional brochure describing the swallowing exercises and their rationale. They were also encouraged to log completion of their daily exercises but logs were not collected by the study team.

2.2.3. Self-Management Adherence Intervention

Participants randomized to the self-management arm received intervention sessions for 10 weeks, starting with the first week of radiation. Ten weekly psychoeducational sessions which were delivered by a master’s level counselor, both in person and by telephone. During each 20–40 minute session, a weekly newsletter presenting patient narratives to normalize loss of taste and set realistic expectations, practical coping strategies for radiation side effects, and psychological skills training were given to the patient and reviewed with a trained counselor. The counselor also briefly reviewed the participant’s daily exercise logs for completion and discussed any potential barriers to adherence for that week. The intervention sessions were manualized and recorded weekly reviews with the principal investigator for treatment fidelity.

2.3. Measures

2.3.1. Main outcome:

Adherence to swallowing exercises was measured multimodally with a) daily activity logs completed by the patient, which were reviewed for completion by the counselor at each weekly session. While the counselor did ask at each session whether the patient had done any swallowing exercises that week, she did not train or re-train the patient, b) self-report in the form of a questionnaire completed 6 weeks after the end of radiation treatment. Patients were asked to indicate which exercises had been assigned, whether the exercises had been performed at home during each week of radiation, and the number of daily sets that had been performed for each exercise during that specific week, and c) verification with independent observation. During radiation, all participants met three times with the study’s medical speech pathologist, who was blinded to randomization status. At each visit, the speech pathologist asked whether the patient was adherent to the assigned swallowing exercises and asked patients to demonstrate the exercises without instructional cues. If patients were unable to perform the exercises independently, they were retrained and this was noted in the patient’s record. The speech pathologist’s documentation of inability to perform the exercises with or without paper instruction cues was counted as nonadherence for that week of radiation. If the medical record for that week did not agree with the patient’s self-report for that week, the speech pathology record superseded the patient’s self-report for that week only. Records also verified which exercises had been withheld, and the number of weeks in which the exercise was withheld. Medical record abstraction was conducted by assessors blinded to the patient’s study status. Adherence was a continuous variable ranging from 0 to 1 and was defined as the number of times each exercise was performed during the 6 week radiation period divided by the total number of exercises that should have been performed.

2.3.2. Mediating variables

Coping was measured with the 60-item full-scale COPE, which measures 6 styles of problem-focused coping (active, planning, suppression of competing activities, restraint, instrumental social support) and 5 aspects of emotion-focused coping (emotional support, positive reinterpretation, acceptance, denial, and religion) Internal reliability of subscales ranged from 0.65 to 0.90. The COPE has strong convergent and discriminant validity with conceptually related measures of optimism, hardiness and trait anxiety.³³

The Illness Perception Questionnaire-Revised

The Illness Perception Questionnaire-Revised has over 80 five-point Likert scale items assessing five cognitive illness perception domains (Identity, Timeline, Consequences, Control, and Cause). The widely used scale demonstrated validity and internal consistency.³⁴

Appraisal of Coping

A 7-item, 4-point Likert scale was developed based on Aspinwall & Taylor’s (1997) theoretical paper describing feedback appraisal loops within the Self-Regulation model.³⁵ The items assessed three domains of appraisal: a) whether the illness event improved, b) whether coping efforts were effective, and c) whether new understanding of barriers to coping were gained.

2.3.3. Moderating side effects

Pain was measured with the Brief Pain Inventory (BPI), a 9-item self-report measure designed to assess pain severity and pain interference.³⁶ Internal consistency reliability for pain intensity ranges from .78 to .96 and pain interference from .83 to .95.³⁷ Fatigue was measured with the Brief Fatigue Inventory (BFI), a 9-item self-report measure similar in format to the BPI. The BFI’s internal consistency is 0.96 and correlates well with physiological markers of fatigue.³⁸

Depressive symptomatology were measured using the Physicians Health Questionnaire (PHQ-9), which ranges from 0 to 27. ³⁹ A cutoff score of 10 has a high positive predictive value for major depressive episode. ⁴⁰

Social support was measured with the Medical Outcomes Study Social Support Scale (MOS Social Support), a 20-item, 5-point Likert scale assessing four dimensions of social support (emotional/informational, tangible, affectionate, and positive social interaction). Internal consistency for all subscales is greater than .91 ⁴¹

2.3.4. Demographic and medical information

Demographic and medical information were collected at the beginning of radiation therapy. Age, gender, race, educational level and marital status were collected by self-report questionnaire. Disease site, stage of disease, induction and/or concurrent chemotherapy status were abstracted from the medical record.

2.4. Analysis

Adjusted ordinary least squares (OLS) regression was applied to examine whether adherence differed between the PREPARE intervention and enhanced usual care group, adjusting for age, race, education, marital status, chemotherapy regimen (induction vs. concurrent vs none), disease site, and cancer stage.

Coping, illness perception constructs and appraisal of coping were tested as potential mediators between treatment group status and adherence total score first as separate single-mediator models and then in multiple-mediator models. For the single-mediator mediation models, each subscale in the variable of interest (COPE, IPQ) was run separately. For appraisal of coping, each of its 7 items was tested separately. For the multiple-mediator models, coping, illness perception and appraisal of coping were run as three separate models, but with the subscales for each scale of interest included in one model (the 7 items of the appraisal analysis were tested in one model). Both single- and multiple-mediator models were run with and without controlling for baseline characteristics (age, gender, race, education level, marital status, income, chemotherapy status, disease site, and tumor stage).

The indirect effect for each potential mediator was obtained by first obtaining path coefficients for regression of the mediator on treatment status and for regression of adherence on both treatment status and mediator. The indirect effect was obtained by using the product of the two path coefficients (Figure 1). The indirect effect was also standardized to aid in its interpretability relative to the predictor, mediator, and outcome variables, which all had different numerical ranges of scores.⁴² The standard error and 95% confidence interval of indirect effect were calculated by the bootstrap method using 1000 simulations. The extent of mediation was also quantified using proportion mediated effect (PME).

For the moderation analysis, treatment group was the independent variable, adherence score was the outcome variable, and the proposed moderator variables were baseline depression, pain, fatigue, and social support. Four separate moderation models (one for each moderator) were run by using multiple ordinary least squares regression model including independent variable, moderator, and their interaction (product) term.

2.4. Power analysis

Based on the performance of our swallowing adherence measure in a previous unpublished study, we expected a correlation of 0.2 among repeated measures. Under these assumptions, we calculated that a total sample size of 200 subjects (100 in each study arm) provided 80% power to detect a difference of 0.27 standard deviations between the Self Regulation intervention group and the control group using a 2-tailed test, and alpha=.05.

3. RESULTS

After obtaining informed consent, 402 patients were enrolled onto the study and randomized to enhanced usual care or intervention, 326 completed baseline assessments, and 265 participants completed the post-radiation assessment and were included in the analysis. The accrual rate was 77.0%. Of the 265 participants, 130 were randomly assigned to enhanced usual care, and 135 were assigned to intervention (Figure 2). Of the 135 intervention participants, 134 completed the 10-week program. The median number of completed sessions was 7 (std. dev = 1.9) and the range of completed sessions was 3–10 (data not shown).

Figure 2. CONSORT.

The average age of the sample was 57.8 years, 84.2% were male, 78.1% were married or living with a partner, 83% of the sample was non-Hispanic white, 49.8% had obtained a college degree or higher and 43.4% reported an annual household income of $75,001 or higher. Demographic and clinical characteristics were not significantly different between the two study groups (Table 1).

Table 1.

Summary statistics of participants’ characteristics

	Total (N=330)	Control (N=151)	Treatment (N=179)
Variables	Mean (SD)	Mean (SD)	Mean (SD)	Two-sample t test p-value
Age (continuous)	57.83 (9.90)	57.96 (9.68)	57.73 (10.11)	0.837
	N (%)	N (%)	N (%)	Chi-squared test p-value
Age (binary)
< = 70	298 (90.3%)	133 (88.1%)	165 (92.2%)	0.107
> 70	30 (9.1%)	18 (11.9%)	12 (6.7%)
Missing	2 (0.6%)	0	2 (1.1%)
Race
White	274 (83.0%)	126 (83.4%)	148 (82.7%)	0.987
African American	14 (4.2%)	7 (4.6%)	7 (3.9%)
Hispanic	26 (7.9%)	12 (7.9%)	14 (7.8%)
Asian	10 (3.0%)	5 (3.3%)	5 (2.8%)
Other	3 (0.9%)	1 (0.7%)	2 (1.1%)
Missing	3 (0.9%)	0	3 (1.7%)
Education
No high school	20 (6.1%)	8 (5.3%)	12 (6.7%)	0.182
High school or GED	44 (13.3%)	19 (12.6%)	25 (14.0%)
Technical/Vocational degree	20 (6.1%)	5 (3.3%)	15 (8.4%)
Some college	84 (25.5%)	39 (25.8%)	45 (25.1%)
Bachelor’s degree	91 (27.6%)	43 (28.5%)	48 (26.8%)
Master’s or higher	38 (11.5%)	24 (15.9%)	14 (7.8%)
M.D., Ph.D., or other doctorate degree	27 (8.2%)	11 (7.3%)	16 (8.9%)
Missing	6 (1.8%)	2 (1.3%)	4 (2.2%)
Marital Status
Single, not living with significant other	20 (6.1%)	8 (5.3%)	12 (6.7%)	0.830
Single, living with significant other	16 (4.8%)	5 (3.3%)	11 (6.1%)
Married, living with partner	237 (71.8%)	110 (72.8%)	127 (70.9%)
Married, living apart	9 (2.7%)	4 (2.6%)	5 (2.8%)
Separated	4 (1.2%)	1 (0.7%)	3 (1.7%)
Divorced	33 (10.0%)	17 (11.3%)	16 (8.9%)
Widow	4 (1.2%)	2 (1.3%)	2 (1.1%)
Missing	7 (2.1%)	4 (2.6%)	3 (1.7%)
Chemotherapy
No	64 (19.4%)	28 (18.5%)	36 (20.1%)	0.762
Yes	266 (80.6%)	123 (81.5%)	143 (79.9%)
Disease Site
Oropharynx	207 (62.7%)	96 (63.6%)	111 (62.0%)	0.935
Larynx	43 (13.0%)	21 (13.9%)	22 (12.3%)
Hypopharynx	9 (2.7%)	5 (3.3%)	4 (2.2%)
Nasopharynx	31 (9.4%)	13 (8.6%)	18 (10.1%)
Unknown Primary	26 (7.9%)	11 (7.3%)	15 (8.4%)
Missing	14 (4.2%)	5 (3.3%)	9 (5.0%)
Tumor Stage
Stage I	4 (1.2%)	2 (1.3%)	2 (1.1%)	0.495
Stage II	21 (6.4%)	8 (5.3%)	13 (7.3%)
Stage III	49 (14.9)	28 (18.5%)	21 (11.7%)
Stage IV	237 (71.8%)	105 (69.5%)	132 (73.7%)
Missing	19 (5.8)	8 (5.3%)	11 (6.2)

The average adherence total score for the entire sample was 0.39 and the standard deviation was 0.26. The enhanced usual care group had a mean adherence total score of 0.33 with standard deviation of 0.23; the intervention group had a mean adherence total score of 0.44 with standard deviation of 0.27 (p = 0.0003; supplemental figure 1).

3.1. OLS Regression Results: Efficacy of the Proposed Intervention on Patient-Reported Adherence

The adjusted OLS regression results showed that the treatment group’s adherence rate was on average 13% higher than the control group after controlling for the seven covariates, and this treatment benefit was statistically significant (p <0.001; Table 3). Regarding the covariates, patients with higher tumor stage (i.e., III or IV) had higher adherence than those patients with lower tumor stage (i.e., I or II). No other significant differences between treatment groups were observed with regard to demographic and clinical characteristics.

Table 3:

Single-mediator mediation analyses results (N=265)

Mediator	Indirect Effect	Bootstrap S.E.	P-value	Lower CI	Upper CI	PME
Appraisal of Coping
I have performed my exercises effectively	0.043	0.038	0.263	−0.032	0.117	53.9%
My swallowing exercises have been effective in keeping my swallowing problem from getting worse	0.007	0.013	0.594	−0.019	0.033	12.7%
In order to get my exercises done, I have learned how to handle my emotions more effectively	0.010	0.013	0.413	−0.014	0.035	6.0%
In order to get my exercises done, I have learned how to handle practical problems more effectively	0.013	0.012	0.285	−0.011	0.037	8.0%
My skills and strategies help me get my swallowing exercises done every day	0.034	0.020	0.092	−0.006	0.073	20.6%
My emotional skills help me get my swallowing exercises done every day	0.039	0.018	0.033*	0.003	0.075	24.1%
Trying new ways to cope with my cancer is often not worth the effort	0.004	0.009	0.671	−0.013	0.021	2.0%
COPE Subscales
Emotional	0.005	0.010	0.582	−0.014	0.024	2.4%
Objective	0.005	0.009	0.596	−0.012	0.021	2.1%
IPQR Subscales
Timeline (acute/chronic)	0.000	0.008	0.984	−0.013	0.013	0.1%
Timeline cyclical	0.007	0.012	0.514	−0.014	0.029	3.3%
Consequences	0.007	0.010	0.497	−0.013	0.027	3.2%
Personal control	−0.003	0.010	0.725	−0.021	0.015	1.5%
Treatment control	−0.006	0.010	0.539	−0.027	0.014	2.9%
Illness coherence	−0.001	0.008	0.946	−0.017	0.015	0.2%
Emotional representation	−0.001	0.009	0.944	−0.017	0.016	0.3%

^*p < .05.

3.2. Mediation Analysis Results

Bootstrap simulation results indicated that one of the appraisals of coping items, “My emotional skills help me get my swallowing exercises done every day” mediated 24% of the intervention’s effect on adherence, with an indirect effect of 0.039 and a p-value of 0.03. However, this mediation effect became nonsignificant (p=.08) when controlling for baseline characteristics (results not shown in the table). None of the other proposed single-mediator models showed statistically significant effects. For the multiple-mediator models controlling for baseline characteristics (not shown in the results), neither appraisal of coping (p of the 7 items ranged from .34 to .92), coping (subscales p =.78, p =.65) nor the illness perception subscales (p ranged from .29 to .88) had significant indirect effects.

3.3. Moderation Analysis Results

No statistically significant moderation effects were found for any of the four moderator scales tested, indicating that there is no evidence that the treatment effect on the outcome changes across the levels of baseline depression (p-value 0.607), pain (p-value 0.855), fatigue (p-value 0.689), or social support (p-value 0.539).

4. DISCUSSION

The main finding of this study was that participants randomized to the PREPARE adherence intervention plus speech pathology had significantly better adherence to swallowing exercises during radiation compared to the participants randomized to enhanced usual care with a corresponding moderate effect size. To our knowledge, this is the second prospective randomized trial testing the effect of an intervention on adherence to swallowing exercises during radiation and the first to report improved adherence. While several studies have shown that preventive swallowing exercises improved self-report and clinician-rated swallowing scales post-treatment, especially when adherence is reinforced with ¹⁶ or weekly^14,19 visits by speech pathologists, these studies did not randomly assign adherence interventions nor focus on adherence as the main outcome. The only other study of which we are aware that tested adherence is Baudelet et al.’s PRESTO trial, which compared three strategies for increasing adherence to exercises: educational brochure plus exercise log, gamified mobile app with embedded instructional video and daily in-person delivery of speech pathologist training during the first four weeks of radiation. This large trial of 148 head and neck cancer patients found nonsignificant differences of adherence among the delivery modes.²⁸ In contrast, our study developed and tested an intensive 10-session adherence intervention delivered by a masters-level counselor plus three speech pathology visits compared to three speech pathology visits alone.

A number of swallowing exercise intervention studies have investigated other determinants of patient adherence to swallowing exercises. Hadju et al found nonsignificant effects for HPV status, partner status, concurrent chemotherapy, disease site or SLP visit attendance on adherence to exercises in 45 oral cavity, larynx and pharyngeal cancer patients during active treatment.⁴³ We also found no association between disease site or marital status on adherence. However, we found that patients with disease stage ≥ III were more likely to adhere to their exercises compared to patients with lower stage disease. One possible explanation is that patients with more advanced disease were more concerned about dysfunction and thus more motivated to follow therapy recommendations.

Patients in the PREPARE intervention group had an adherence score of 0.44 out of a range of 0–1, meaning that, as a group, they were perfectly adherent to four daily sets of all assigned exercises for each week of radiation 44% of the time. Within the extant literature, adherence tends to be reported differently, making direct comparison with our results difficult. Cnossen conducted a prospective cohort trial with 60 oral cavity, larynx and pharynx cancer patients during radiation and for six additional weeks thereafter. All patients received an initial 15-min training session with speech pathologists supplemented with 12 weekly 10-min booster phone sessions and either online video or paper-based photo instructions. At 6 weeks, 40% of patients were completing all exercises at most once a day, 34% between 1 and 2 times a day and 26% at least twice a day.²⁷ After meeting with an occupational therapist 3 times a week during radiation, Hadju et al. reported that 47% of 45 intervention patients self-reported completing at least 80% of their assigned exercises for an average duration of 5 weeks of radiation.⁴³ Within the PRESTO trial of late-stage oropharynx cancer, 27% of the total sample (n = 115) performed at least 75% of all assigned repetitions of swallowing exercises during the first 4 weeks of radiation, regardless of speech pathology service delivery mode. This rate improved to 35% for the patients receiving the mobile app, and was as low as 22% for patients who received the instructional brochure.²⁸ Kotz et al. noted that 69% of 13 patients randomized to weekly in-person speech pathologist visits were able to continue an unspecified number of exercises until the 4th week of radiation, although only 31% were able to continue past the 5th week of radiation.¹⁴ Jansen et al. randomized 92 cancer survivors who had been treated within the past 5 years to either a self-help exercise intervention or self-care control group. Within the exercise group, 59% of the 46 survivors performed eight range of motion exercises at least once a day for 12 weeks.⁴⁴ Given that no study has thus far reported100% adherence to recommended exercises, these results underscore the enormous difficulty of daily exercises throughout the course of radiation with curative intent.

Illness perceptions, emotion-focused coping, social support and depression did not moderate PREPARE’s effect on adherence, suggesting that the treatment may have uniform clinical benefit over all subgroups of the enrolled participants. One item on the appraisal of coping scale was related to significantly increased adherence in the treatment group. Patients’ positive endorsement of the item “My emotional skills help me get my swallowing exercises done every day,” explained 24% of the variance of the intervention program’s effect on adherence. This suggests that the intervention’s therapeutic benefit was driven by increased patients’ confidence in their ability to emotionally cope with radiation side effects. No other statistically significant indirect effects were observed. However it should be noted that our study was not powered for these secondary analyses.

Self-management adherence interventions have been conducted in a wide range of patient populations. However, none of the adherence interventions of which we are aware have been applied in head and neck cancer populations. Aujla et al.’s meta-analysis of 21 studies examining illness perceptions’ effect on adherence outcomes showed that several domains of illness perception (identity, acute vs. chronic timeline, consequences, personal control) had significant but small effects on adherence (weighted r ranged from .04 to .13).⁴⁵ In another review of 12 self-management adherence intervention studies for at-home physiotherapy regimens in a range of patient populations (e.g., post-stroke, post-injury, or osteoarthritic patients), Peek et al. found that interventions incorporating utilizing multiple sessions over an extended span of time with sports psychologists increased long-term adherence, which was consistent with our finding of increased adherence if 8 or more sessions were completed.⁴⁶

Several of the self-management variables in our study can reasonably be seen as facets of social cognitive constructs. For example, the concept of self-appraisal of emotional and objective coping is consistent with self-efficacy, or the positive belief in one’s capability to enact a goal.⁴⁷ If self-appraisal of coping is interpreted as a proxy for self-efficacy, our study is consistent with others’ who have found self-efficacy to be a mediator for increased adherence to prescribed exercise. For example, both Harnirattisai et al.’s and Rejeski et al.’s studies found that increased physical activity in the intervention group was related to positive changes in self-efficacy.^{48 49} However, other studies directly targeting self-efficacy have not shown increases in adherence compared to controls.⁵⁰

4.1. Study Limitations

Our study had several limitations. Study retention was impacted by the burden of our baseline questionnaire, which was in excess of 20 pages, of which 74 participants did not complete (18.4%). Furthermore, the median number of weeks from end of radiation to completion of the follow-up questionnaire was 8.4 (SD=15.5); 80% of our sample completed the questionnaire within 12.8 weeks after completing treatment. While this is somewhat longer than the planned 6-week timeframe from the end of treatment, positive recall bias was likely mitigated with cross-verification of speech pathology notes of inability to demonstrate swallowing exercises without instructional cues. While both groups were asked to complete daily exercise logs, we did not collect or analyze the data within the logs as the patients found them cumbersome and we did not wish to introduce the logs as an intervention strategy within the control arm. While we were able retain 80.7 % of the participants who completed the baseline, our study was not powered to detect potential moderating and mediating effects. Finally, our mediation models should be interpreted with caution since the outcome and mediator variables were obtained at the same time point.

5. Conclusion

PREPARE increases patient adherence to a challenging swallowing exercise protocol during radiation for head and neck cancer. Our clinical trial had several strengths: well-defined eligibility criteria, high accrual and completion rates, adherence measurements verified with speech pathologist documentation, and high fidelity to a manualized intervention. While the weekly counseling intervention is not a substitute for speech pathology exercises in the prevention of radiation-associated dysphagia, the PREPARE program presents a more scalable model for increasing adherence. The program builds upon earlier studies, which used daily or weekly speech pathology visits to reinforce patient adherence. Future directions should explore the translation of the adherence program to more mobile applications to increase accessibility and ease patient burden.

Table 2.

Adjusted ordinary least squares (OLS) regression result

	Adherence Score (N=257)
Variables	Coefficient (S.E.)	95% CI	p-value
Constant	0.03 (0.10)	(−0.17, 0.23)	0.765
Treatment	0.13 (0.03)	(0.07, 0.19)	<0.001
Age	0.07 (0.05)	(−0.04, 0.17)	0.222
Race	−0.04 (0.05)	(−0.13, 0.05)	0.394
Education	0.02 (0.04)	(−0.05, 0.09)	0.606
Marital status	0.05 (0.04)	(−0.03, 0.13)	0.235
Chemo	−0.03 (0.04)	(−0.11, 0.05)	0.473
Disease site	−.003 (0.03)	(−0.07, 0.07)	0.924
Stage	0.17 (0.06)	(0.05, 0.30)	0.007

Treatment: (Reference: control group)

Age: (Reference: 70 or younger)

Race: (Reference: non-white)

Education: (Reference: less than college)

Marital status: (Reference: not living with partner)

Chemo: (Reference: no chemotherapy)

Disease site: (Reference: not oropharynx)

Stage: (Reference: Stage I-II)

• Note: Treatment: (Reference: control group). Age: (Reference: 70 or younger). Race: (Reference: non-white). Education: (Reference: less than college). Marital status: (Reference: not living with partner). Chemo: (Reference: no chemotherapy). Disease site: (Reference: not oropharynx). Stage: (Reference: Stage I-II).