Study design

This study analyses data from a RCT, with two study arms. It compares the effects of an eHealth support, known as Patient Activation in Treatment at Home (ePATH), which provides additional assistance for patients to adhere to rehabilitation recommendations in the intervention group, against standard care alone in the control group. The study design has been described in detail in the study protocol [25] and in a previous report on primary outcomes [26]. The study is registered, as an International Standard Randomized Controlled Trial (18055968), International Registered Report Identifier (10.2196/11625) and follows CONSORT guidelines (S1 Checklist).

Ethics approval and consent to participate

The study was performed in accordance with principles in the Declaration of Helsinki and had ethical approval from the Regional Ethics Committee (Ref No 2016/484-31; 2017/512-32; 2018/147-32) in Sweden. All patients gave written informed consent to participate and were informed that data would be handled confidentially, and they could withdraw their consent at any time, without giving a reason.

Inclusion criteria and sample size

The inclusion criteria were: >18 years, elective radical prostatectomy, having a smartphone-based e-identification, and being able to communicate in Swedish. All men who met the inclusion criteria and accepted participation at any of three county hospitals in southern Sweden were included. An exclusion criterion was having a cognitive impairment (assessed by the cancer nurse specialist).

An independent statistician performed the sample size estimation based on the “The sample size and suggested end point” table of the primary outcome measure for the RCT, the Expanded Prostate Index Composite (EPIC). To show clinical importance, the effect size was set to 0.5, as recommended. A maximum of 4 domains of EPIC, with a two-tailed independent samples t-test per domain, a 5% overall significance level and a power of 80%, 91 patients per group had to be included. Considering a dropout of 25%, 114 patients per group was needed [26–28]. Sample size was not calculated for the secondary outcome measure, the Patient Activation Measure (PAM), on which the present study reports [25].

Setting and standard care

Three county hospitals in southern Sweden included patients planned for radical prostatectomy. The care organizations differed somewhat between the sites [25], but all clinics applied the national standardized care trajectory [9]. Men with prostate cancer are allocated a cancer nurse specialist, for easy access to healthcare and support. Generic recommendations for rehabilitation were provided verbally and in writing to both intervention and control groups. The cancer nurse specialists contacted the patients in conjunction with regular check-ups after surgery (usually at 3- or 6-month intervals) and provided individual consultation on rehabilitation and further treatment. Other contacts were on each patient’s initiative [9].

The ePATH intervention

ePATH consisted of a web-based and mobile application, to be used as additional support in cancer rehabilitation. The intervention is based on the self-determination theory, to satisfy the three basic psychological needs for motivation: competence, autonomy, and relatedness [15]. Competence was supported as patients could read/re-read individualized information, explanations, and instruction for self-care at any time. Patient autonomy was respected as they had the option to customize ePATH, and independently choose actions and when to perform them. Patients could register self-care activities performed, rate their symptoms, get notifications, set goals for themselves and see graphs of improvements over time. Relatedness was pursued through the simplified interaction with the cancer nurse specialist through ePATH’s messaging function. Links to patient associations were also provided [29].

Data collection and procedures

Measures

The characteristics of participants obtained at baseline were age, marital status, household income, and education. Use of ePATH was based on data metrics after data collection was finalized. Self-rated general health was measured using a single item from the Short Form Health Survey (SF-36) (in general, how would you describe your overall health?) rated on a 5-point scale from 1 = excellent to 5 = poor [30, 31].

Motivation was measured using the short form of the Needs Satisfaction and Frustration scale [32], with three subscales, a patient’s perceived autonomy, social belonging, and own competence. The scale comprises 6 items (2 per subscale), with 7 response options ranging from very often (1) to very seldom (7). One of two items per subscale should be reversed before summarizing and divided by two, for the final score. A Swedish version of the scale has been validated in a general Swedish population [33].

The Patient Health Questionnaire 9 (PHQ-9) comprises 9 questions, measuring the severity of depression (as defined by the Diagnostic and Statistical Manual of Mental Disorders 4^th edition), with a total score ranging from 0 to 27, and a cut-off score at 10. Higher scores indicate a higher degree of depression, with scores categorized on a five-level scale: minimal depression 1–4, mild depression 5–9, moderate depression 10–14, moderately severe depression 15–19, and severe depression 20–27. One item assesses suicide risk [34]. The PHQ-9 fulfils the criteria for sensitivity and specificity of instruments used to diagnose and grade depression severity [35]. The Swedish version has demonstrated good measurement properties in a primary care context and in a sample of psychiatric patients [36].

Patient activation was measured using the Patient Activation Measure (PAM-13), which aims to assess patient’s self-reported knowledge, skill, and confidence regarding management of own health and illness [12]. Patient activation is assessed by 13 items, that gauges a person’s self-concept as a manager of their own health and care. Each question has 5 response categories, ranging from strongly disagree (1) to strongly agree (4), with 0 corresponding to ‘not applicable’. A scoring sheet provided by the developer, Insignia Health, indicates the PAM scores (0–100) and correlating levels (1–4) of patient activation. Level 1: disengaged and overwhelmed; level 2: becoming aware, but still struggling; level 3: taking action; level 4: maintaining behaviours and pushing further. Each point change in the score is meaningful and a change of 3–4 points may be associated with the difference between being engaged and not. However, no cut-offs are available. Most individuals have activation scores between 30 and 90 [37]. The PAM-13 has shown good measurement properties regarding validity and reliability [12, 38] and the Swedish version has been validated in discharged medical patients, geriatric patients, and surgical patients [39].

Statistical analysis

Data were analysed using SPSS software (version 29; IBM Corp) for Windows [40]. Of the 170 patients included, five declined participation after randomization (remaining: 83 intervention and 82 control group). Patients without any data (n = 20) were considered lost to follow-up and excluded from final analyses, resulting in 145 patients (71 intervention and 74 control group) included in a modified intention-to-treat analysis (Fig 1) [26]. For an intention-to-treat analysis to be valid, its assumptions need to be valid; not all patients must be included in the analysis [41]. A missing data analysis showed randomly missing data at the individual level for approximately 15% of outcome measures, no outliers were removed. We therefore imputed data, using multiple imputations with predictive means matching [42]. Five imputation rounds were performed, and the pooled values of these imputations have been presented [43]. The pooled values and imputations were further compared with the original data for validity (S1 Table). All statistical tests employed a two-sided significance level of < .05. Descriptive statistics were used to describe participant characteristics and baseline data. Means and standard deviations were used for the normally distributed continuous variables (S2 Table). To identify differences between groups, Pearson’s chi-squared test was used for nominal variables, the Mann-Whitney U-test for ordinal variables, and the independent Student’s t-test for normally distributed continuous variables. Some response options for self-rated general health were merged– 1 (poor) and 2 (fair), and 4 (very good) and 5 (excellent), respectively–due to a small number of responses at the extremes, resulting in a 3-point scale. To investigate the mean PAM score differences over time between groups, a linear mixed model with repeated measures and an unstructured correlation structure was used. An interaction term between time and group was employed to identify differences in the pattern of change. The assumptions of the model were checked based on residuals and Akaike Information Criterion (AIC). The model was controlled for the effects of baseline PAM scores, PHQ scores, and age. To illuminate the time-varying covariance of the PHQ, we used mean PHQ scores to distinguish from changes in time.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. CONSORT flow diagram.

170 men were included and both groups allocated 85 men. Five men withdrew participation. Men with no answers (19) were considered lost to follow-up.

https://doi.org/10.1371/journal.pone.0308555.g001

Participant characteristics, baseline data, and use

The mean age of participants in both groups at inclusion was 64 years (SD 6.2) and age ranged from 48 to 78 years. The participant characteristics did not statistically significantly differ between the groups as regards marital status (p = .82), household income (p = .51), or level of education (p = .74). In the intervention group 43% (36/83) were married/had a partner compared to 54% (44/82) in the control group. The largest proportion of participants had a monthly household income >45,000 Swedish crowns (intervention group 40%, 33/83; control group 39%, 32/82), and university education (intervention group 27%, 22/83; control group 29%, 24/82).

Baseline data (Table 1) on the motivational subscales ‘autonomy’ and ‘relatedness’ and depression were similar between groups. However, the intervention group had higher PAM scores and higher ratings on the motivational subscale ‘competence’ than those in the control group at baseline. The largest proportion of participants (in both groups) had minimal depression at baseline. Use of ePATH in the intervention group was defined as using ePATH more than once, and 64% of the intervention group fell into that category.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline data (n = 165).

https://doi.org/10.1371/journal.pone.0308555.t001

Changes in depression scores (PHQ-9)

The depression scores (PHQ-9 mean value) increased from 6 months to 12 months from minimal depression (mean 2.72 (SD 3.94)) to moderate depression (mean 12.92 (SD 4.11)) in both the intervention and control group. No statistically significant difference between groups was detected (6 months p = .69; 12 months p = .68).

Effects of ePATH on patient activation (Patient Activation Measure, PAM-13)

The linear mixed model did not detect any statistically significant differences in PAM score between the intervention and control groups at 6 or 12 months (Table 2). No statistically significant interaction between time and group was found. Higher PAM score at baseline (higher patient activation) was statistically significantly correlated with increased PAM score at 6 and 12 months. Higher PHQ score (higher levels of depression) was statistically significantly correlated with decreased PAM score (less patient activation) at 6 and 12 months. Age did not statistically significantly correlate with PAM score.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Changes in patient activation over time (6 and 12 months), linear mixed model.

https://doi.org/10.1371/journal.pone.0308555.t002

Both intervention and control groups had improved PAM scores between 6 months and 12 months (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Mean values (PAM) at 6 and 12 months (linear mixed model).

https://doi.org/10.1371/journal.pone.0308555.t003

Strengths and limitations

The study’s power was calculated based on primary outcome measures in another study [26]. Recruitment of patients was terminated after two years, of concern of the involved clinics (workload) and for standard care to remain the same for all included participants. The current study therefore included 170 patients. Dropout can be a threat to validity, as it can introduce selection bias. An intention-to-treat analysis, where dropouts are assumed to have a neutral effect, is a way to avoid bias. However, such a procedure can diminish the power of a study [57]. eHealth interventions generally struggle with high dropout rates and one of the fundamental methodological issues in eHealth trials is that a substantial proportion of the intervention group will use the intervention sparingly or not at all [58]. The study was performed in routine clinical practice, to ensure minimal burden on the valuable time of the cancer nurse specialists and to show how this kind of intervention could work in contemporary clinical scenarios. This approach stems from the prevalent demand for clinical RCTs [22]. The frequency of use of ePATH [26] in the intervention group might have affected the results, as not all men used ePATH, even though they had gone through recruitment and given informed consent, thus agreeing to adopt the intervention [57]. Participants might have experienced dissonance between the intervention and their expectations, affecting attrition [59] or not perceived as creating benefits [57]. The procedure at inclusion might have influenced the use as well; incorporation of a hands-on introduction or choosing another time point might have increased use. Multiple factors affect how the findings can be generalized and eHealth support do not reach the total population. Further studies should focus on investigating which patients would gain from eHealth support for patient activation.

Applicability and validity should be considered in the evaluation. The study had a homogeneous sample of men after radical prostatectomy, which increases the generalizability of the findings. The sample had high PAM scores in general, which reflects the physiological capacity of this population. Further, that fact raises questions of selection bias and ceiling effects, causing difficulties in identifying statistical differences between groups. The use of linear mixed models with two assessment points improved the representability of the sample, as individuals without complete data could be included [60]. However, the use of imputations should be considered. An accepted method for imputation was used, which increases validity [61]. Valid results were seen when the imputed dataset was compared with the original dataset.

Patient activation is a complex outcome with heterogeneous variables and processes involved. The lack of conceptual clarity concerning patient empowerment, engagement, and activation, makes it challenging to operationalize interventions and measure all the aspects involved, influencing the eHealth context [62]. Although it appears that eHealth interventions can be effective, the interventions show a great heterogeneity of technologies, variables, and outcomes assessed [20]. The PAM offers an adequate measure to monitor patient activation [12] and is viewed as a successful way to measure patient activation [63], but can be arbitrary in nature [64]. In the present study, patients with elevated depression scores correspondingly showed diminished patient activation scores. This underscores the importance of illuminating the complex dimension of patient activation within the context of cancer rehabilitation. Moreover, it emphasises the need for a comprehensive approach where the evaluation and support of patients encompass both depression and patient activation in parallel throughout the rehabilitation process.