Effects of an Internet Support System to Assist Cancer Patients in Reducing Symptom Distress

A Randomized Controlled Trial

Cornelia M. Ruland, PhD, RN; Trine Andersen, MSc, RN; Annette Jeneson, MSc; Shirley Moore, PhD, RN; Gro H. Grimsbø, MSc, RN; Elin Børøsund, MSc, RN; Misoo C. Ellison, PhD

Disclosures

Cancer Nurs. 2013;36(1):6-17. 

In This Article

Results

Four hundred forty-five persons who called met the inclusion criteria and expressed continued interest to participate and therefore received baseline questionnaires and consent forms. Three hundred twenty-five of those (73%) returned completed consent forms and questionnaires and were enrolled into the study; 162 (96 breast cancer and 66 prostate cancer patients) were randomized into the experimental group and 163 participants (93 breast cancer and 70 prostate cancer patients) into the control group (Figure 1).

Figure 1.

Flowchart of enrollment, randomization, and follow-up of study participants.

System Use

One hundred twenty-five of the study participants assigned to the experimental group (77%) logged onto WebChoice at least once. Twenty-three percent never logged on. The only difference between users and nonusers on any demographic or other variables was that users had slightly more previous computer experience, although this difference was only borderline significant (P = .07). Those of the 103 study participants (64%) who used WebChoice more than once used it on average 60 times over the 1-year study period; however, there were large individual variations (range, 2–892). The e-forum and e-communication with expert nurses were most frequently used. Sixty-two patients wrote personal messages to the nurse (total, 385; range, 1–49; average, 6.2), and 50 patients posted messages to the forum (total, 506; range, 1–58; average, 10.15). However, patients visited the forum and messaging service many times more to read information without posting messages. On average, nurses spent 15 minutes to answer a message. More details on usage patterns are provided elsewhere.[48]

Baseline Data

To ensure that the 2 study groups did not differ at baseline, we compared baseline characteristics of age, time since diagnosis, socioeconomic status, treatment, gender, stage of disease, and comorbidity. As seen in Table 1, there were no statisticall significant group differences at baseline on these variables. However, because time since diagnosis showed close-to-significant group differences (P = .07), and there were large variations among patients, we included this variable as a covariate in the analyses. Baseline data are displayed in Table 1.

Primary Outcome: Symptom Distress

To test the hypothesis that patients in the experimental group would have significantly less symptom distress over time, we compared within- and between-group differences on the MSAS-SF total score and subscales in trends over time and controlled for time since diagnosis. Within- and between-group variances and slopes were compared between the 2 groups using the linear mixed-effects model.

Results are displayed in Table 2. Figure 2 depicts the findings graphically, displaying the slopes for changes over time in symptom distress scores from the baseline data collection point over the 12-month study period for the intervention group compared with the control group.

Figure 2.

Group differences over time on symptom distress, including the F statistic, degrees of freedom (df ), and P values for global symptom distress, physical symptoms, psychological symptoms, and the total MSAS-SF score, controlled for time since diagnosis.

Between-group differences were statistically significant for the GDI only (slope estimate, −0.052; 95% confidence interval [CI], −0.101 to −0.004; t = 4.42; P = .037). There were no significant within- or between-group differences on the other MSAS-SF subscales or the total score. Therefore, the hypothesis that WebChoice would reduce the primary outcome of symptom distress was only partially supported. However, Figure 2 shows a downward trend, although not significant, toward less symptom distress in the WebChoice group on all subscales and the MSAS-SF total score, whereas the control group showed a trend in the opposite direction toward increased symptom distress.

Secondary Outcomes: Depression, Self-efficacy, HRQoL, and Social Support

To test the hypothesis that the experimental group would have significantly better scores over time on depression, self-efficacy, HRQoL, and social support, we again compared within- and between-group variances and slopes on the Center for Epidemiological Studies–Depression Scale, Cancer Behavior Inventory, 15D, and Medical Outcomes Study Social Support Survey, using the linear mixed-effects model.

As seen in Table 2, there were no significant between-group differences on any secondary outcomes. So the hypothesis that WebChoice would improve secondary outcomes was not supported. However, participants in the experimental group showed significant within-group improvements in depression (slope estimate, −0.41; 95% CI, −0.71 to −0.11; t = −2.71; P = .007) during the study period that were not observed in the control group. Furthermore, the control group worsened their within-group self-efficacy (slope estimate, −3.77; 95% CI, −6.38 to −1.15; t = −2.82; P = .005) and HRQoL scores significantly (slope estimate, −0.01; 95% CI, −0.01 to −0.00; t = −2.77; P = .006), but the experimental group did not. There were no within- or between-group differences in social support (Figure 3).

Figure 3.

Group differences over time on secondary outcomes, including the t statistics, degrees of freedom (df ), and P values for depression, self-efficacy, health-related quality of life and social support, controlled for time since diagnosis.

Additional Exploratory Analyses

Because there were large variations in participants' time since diagnosis, a variable that could potentially influence symptoms as well as patients' need for support, we wanted to explore if the data could tell us something about whether WebChoice may work differently for patients at different stages of their illness trajectory. We therefore analyzed primary and secondary outcomes separately for patients who at the time of enrollment had been diagnosed within the last 12 months and for patients who had been diagnosed for more than 2 years. Because the sample size for the 2 groups did not leave us with enough power, our additional analyses are purely exploratory.

There was a statistically significant difference between patients newly diagnosed (<12 months) and those who were diagnosed for more than 24 months in terms of cancer recurrence and metastases (P < .001) but not in frequency of other illnesses (P = .24). However, when we performed the separate subgroup analyses on group differences for patients newly diagnosed compared with those diagnosed for more than 2 years, these differences were no longer statistically significant, meaning that the group differences described below were not confounded by stage of disease.

When patients were diagnosed within 1 year (n = 174), there were significant within-group reductions in MSAS-SF total scores (slope estimate, −0.11; 95% CI, −0.20 to −0.01; t = −2.27; P = .024), the global symptom distress (slope estimate, −0.13; 95% CI, −0.22 to −0.05; t = −3.07; P = .003), the physical symptoms subscale (slope estimate, −0.12; 95% CI, −0.22 to −0.03; t = −2.51; P = .013), and close to significant reductions in the psychological symptoms subscale (slope estimate, −0.10; 95% CI, −0.20 to 0.00; t = −1.9; P = .06) in the experimental group, but not in the control group. Also, the WebChoice group had significant within-group improvements in self-efficacy (slope estimate, 5.97; t = 2.36; P = .02), not observed in the control group.

For patients who were diagnosed for more than 2 years (n = 74), there were no significant within- or between-group differences on any subscales measuring symptoms or global symptom distress. However, the control group significantly worsened their self-efficacy scores over the study period (slope estimate, −3.46; 95% CI, −6.84 to −0.08; t = −2.01; P = .04), and there was a downward trend in HRQoL (slope estimate, −0.01; 95% CI, −0.01 to 0.00; t = −1.76; P = .08) and depression (slope estimate, 0.37; 95% CI, −0.06 to 0.79; t = 1.7; P = .09), but not in the WebChoice group.

processing....