The effect of an online support group on patients׳ treatment decisions for localized prostate cancer: An online survey

Urologic Oncology: Seminars and Original Investigations ] (2016) ∎∎∎–∎∎∎

Original article

The effect of an online support group on patients' treatment decisions for localized prostate cancer: An online survey Johannes Huber, M.D., Ph.D.a,*, Philipp Maatz, M.D.a, Tanja Muck, M.D.a, Bastian Keck, M.D.b, Hans-Christoph Friederich, M.D.c, Wolfgang Herzog, M.D.c, Andreas Ihrig, Ph.D.c a

Department of Urology, Medical Faculty Carl Gustav Carus, TU Dresden, Dresden, Germany b Department of Urology, University Hospital Erlangen, Erlangen, Germany c Department of General Internal and Psychosomatic Medicine, University of Heidelberg, Heidelberg, Germany Received 16 March 2016; received in revised form 14 July 2016; accepted 27 September 2016

Abstract Objective: To analyze the effect of an online support group (OSG) on the final treatment decision for localized prostate cancer. Methods: We performed a cross-sectional descriptive study of the largest German prostate cancer OSG between July and October 2013. The online survey comprised 127 questions covering sociodemographic and disease-related information, decision-making habits, healthrelated quality of life, distress, depression, and anxiety. The primary outcome was to measure the effect of an OSG on the final treatment decision. Results: We analyzed the completed questionnaires from 686 patients with prostate cancer, 200 (29.2%) of whom revised their initial treatment decision. After revising their decisions, these patients more frequently underwent external beam radiation therapy (44.5% vs. 36.4%, P ¼ 0.048) and active surveillance (10.5% vs. 3.7%, P o 0.001) and less frequently underwent radical prostatectomy (52.5% vs. 74.9%, P o 0.001). Engaging longer in the OSG, demanding a more active role in the decision-making process, and participating in a conventional support group were independently associated with revision of the initial treatment decision. Conclusions: Of all patients participating in the OSG, 29.2% revised their initial treatment decision. We estimate that this phenomenon may affect 17,000 patients with prostate cancer in the United States of America every year. This finding highlights the importance of OSGs for the health care system. The patient's desired degree of involvement in decision-making should be routinely clarified to adjust counseling accordingly. Trial registration: www.germanctr.de, number DRKS00005086 r 2016 Elsevier Inc. All rights reserved.

Keywords: Online support group; Decision-making; Peer-to-peer support; Bias; Prostate cancer

1. Introduction Patients with chronic and oncological diseases increasingly use the Internet as a source of patient information and Funding was provided by the Foundation of the Federal Bank of Baden-Wuerttemberg (Grant 2012030055). Moreover, we thank the umbrella organization of Germany's regional prostate cancer support groups (Bundesverband Prostatkrebs Selbsthilfe e.V.) and German Cancer Aid for supporting our project. * Corresponding author. Tel.: þ49-351-458-18954; fax: þ49-351458-4333. E-mail address: [email protected] (J. Huber). http://dx.doi.org/10.1016/j.urolonc.2016.09.010 1078-1439/r 2016 Elsevier Inc. All rights reserved.

support [1,2]. In addition to the general development of the “social web,” the demand for online support appears highest for stigmatizing and devastating diseases, including prostate cancer [3–5]. This evolving effect of modern media appears inevitable and offers a significant opportunity to improve the health care system and the physician-patient relationship by encouraging participatory decision-making [6,7]. Localized prostate cancer presents a good example for exploring the opportunities and risks of modern media for patient empowerment: it demands complex decision-making with a high level of patient involvement [8,9]. Several available treatment options might be equally effective while

2

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

causing different spectra of side effects [10]. Primary treatment options include radical prostatectomy (RP), external beam radiotherapy (EBRT), brachytherapy, active surveillance, and watchful waiting [10]. Economic factors and the involvement of urologists, radiation oncologists, and other specialists further complicate the decision-making process [11]. The process of treatment decision-making may be described as a complex interplay of 3 domains comprising physicians, patients, and contextual factors [12]. A comprehensive systematic review by Tariman et al. [12] describes these factors in the context of the most relevant theoretical models of decision-making. Modern media influence all 3 domains, particularly patient factors such as personal beliefs and values, decisional control preferences, previous health-related experience, social structure, and personal factors. Among these, social support plays a major role in decision-making for localized prostate cancer; moreover, the importance of online resources has become increasingly recognized [13]. The importance of the Internet was underscored recently by a study that randomized 494 men with localized prostate cancer to the use of a decision support system vs. usual care [14]. Only prior Internet use and perceived preparation for decision-making significantly predicted 6-month decision satisfaction [14]. Whether Internet use also affected the treatment modality that patients finally chose remained an open question. This finding exposes a general knowledge gap because the study endpoints were generally limited to descriptions of the surrounding conditions [2,15,16] and subjective perceptions [16,17] of decision-making. Online support groups (OSGs) represent one of the most widespread interactive Internet resources; they enable patients to discuss personal information anonymously and provide patients with information, advice, and emotional support [13,17]. Quantifying the effects of this peer-to-peer counseling is difficult, and well-designed comparative studies are lacking; specifically, no such data exist for patients with prostate cancer. For breast cancer, an uncontrolled longitudinal study demonstrated reduced depression and improved reactions to pain in 32 patients [18]. Beyond investigating patient-reported outcomes, determining the possible effect on final treatment decisions is imperative. Owing to the growing relevance of patient empowerment and shared decision-making, OSGs might directly affect health care delivery, health economics, and the physicianpatient relationship, adding significant epidemiologic and economic weight to understanding OSGs' effects on treatment decisions. The discourse within OSGs might stimulate patients to reconsider previous recommendations, and a share of patients might ultimately revise their choice. The size of this effect is unknown; however, at least modest evidence exists from a recent systematic review demonstrating that OSGs may be effective in changing participants' behavior [19]. To the best of our knowledge, no

reliable data exist to date on the influence of OSGs on patients' actual treatment decisions. Our cross-sectional descriptive study analyzed an OSG's effect on final treatment decisions for localized prostate cancer through an online survey within the largest German OSG to identify how many and which participants revised their initial treatment decision after consulting the OSG. 2. Methods 2.1. General conduct An online survey open from July to October 2013 was offered to participants in the largest German prostate cancer OSG present since June 2000. To widen the scope of our study, we did not apply a specific theory of decisionmaking a priori [12]. We followed the Checklist for Reporting Results of Internet E-Surveys [20]. 2.2. OSG structure At the time of data collection, 3,357 users were registered with more than 70,000 postings. The numbers of users and postings have nearly doubled within the past 5 years [13]. The OSG is freely accessible with use of the German language as the only prerequisite. This group is maintained by the umbrella organization of regional prostate cancer support groups in Germany (Bundesverband Prostatkrebs Selbsthilfe e.V.), a member of “Europa Uomo —The European Prostate Cancer Coalition” (http://www. europa-uomo.org). 2.3. Study design The survey contained 127 questions within a dynamic questionnaire (Appendix 1). The deciding question for group allocation was “Has your treatment decision changed through the use of the discussion board?” Three options were provided: “Yes,” “No,” and “Not specified.” We defined “Yes” to indicate a change of the initial treatment decision. Participants who chose “No” or “Not specified” were included in the comparison group without a change of the initial treatment decision. To compare the participants who revised their decisions and those who did not report a change, the survey contained 12 sociodemographic, 16 disease-related, and 64 psychological questions complemented by 35 questions on decision-making and information-seeking habits. Regarding the structure of the German education system, we defined education status according to the duration of schooling as low or medium (r11 years) vs. high (Z12 years). The psychological questions consisted of validated instruments for measuring quality of life (EORTC QLQ-C30 and the prostate module PR25) [21], depression and anxiety (PHQ-4 [22]), and the distress thermometer [23].

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

These well-established questionnaires provide good psychometric properties, and data analysis occurred according to the corresponding manuals. One question on decision-making habits consisted of a German version of the “Control Preferences Scale” (Appendix 1, item 21), which characterizes decisionmaking preferences within a range from paternalistic to autonomous using a 5-degree ordinal scale [24]. The variable “Discussion board activity” summarized whether participants actively posted messages (posters) or only received content passively (lurkers) as determined by the question “How many messages did you post in the BPS discussion board?”. Further variables were defined by the following items: “Registered discussion board user”, “Time spent within the OSG”, “Participation in a conventional support group”, and “Using the Internet for health information”. We asked about final treatment decisions and offered the following options: RP, EBRT, brachytherapy, highintensity focused ultrasound, active surveillance, watchful waiting, chemotherapy, and hormonal therapy. For all the treatment modalities indicated for nonmetastatic disease only [10], the provided statements retrospectively characterized the situation at initial diagnosis when posed to patients with metastatic disease at the time of the survey. Therefore, we did not apply information on metastatic disease when defining risk groups as in the recent European Association of Urology guidelines [10]: low risk (all T r 2, N0/X, prostate-specific antigen (PSA) o 10 ng/ml, Gleason r 6), intermediate risk (all T r 2, N0/X, PSA o 20 ng/ml, Gleason ¼ 7), and high risk (all other cases). To rule out a relevant bias because of including patients reporting metastatic disease at the time of data collection, we repeated all calculations after excluding patients with metastatic disease. This confirmed the same significant differences between both groups (data not shown). All the collected information represented self-reported data. Because of data protection requirements, it was not possible to re-contact participants or to compare their statements to medical reports on a systematic basis. The primary outcome of our study was to measure the effect of an OSG on the final treatment decision. Because no reliable data exist on a certain set of associated factors, we collected a broader range of variables within an explorative approach. Our goal was to describe differences between patients with revised and unchanged decisions. For example, one hypothesis is that patients who revise their initial treatment decision feel more autonomous and in control of their disease, which could benefit outcomes such as quality of life, depression, and anxiety. A total of 42 volunteers aged 40 to 75 years, who were not participants in the OSG, piloted the survey. Although no technical problems occurred, we modified some wording to improve understanding. The average time to complete the survey was 20 minutes.

3

After a field test, SurveyMonkey (http://www.surveymon key.com), a web-based survey solution, was used to host our survey for 3 months. This service automatically captures responses and guarantees submissions' privacy. The OSG's Webmaster sent an e-mail including the survey link and an invitation to participate to all registered users. A thread and hyperlink were posted on the front page, reading “Please participate: online survey to investigate the potential and limitations of self-help.” No incentives were offered. The e-mail explained our project's aim and guaranteed anonymity. Furthermore, we established that refusing to participate would have no negative consequences. Written informed consent was obtained online from each participant whereby every potential participant was provided with the complete study information including the declaration of consent. The general inclusion criteria were informed consent and age 414 years. We chose this lower age limit to include relatives and friends who searched the OSG by proxy, for example, a patient's grandchild. However, we analyzed only patients with prostate cancer for this study. This subgroup was defined as persons searching for information for themselves and having been diagnosed with prostate cancer. 2.4. Analyses To analyze the effects of OSG participation on final treatment decisions, we compared the participants who revised their decision and those who did not report a change. The data from both groups were examined for differences using explorative univariate statistics. We applied the chi-square test for categorical variables, the Mann-Whitney U-test for ordinal-scaled variables or not normally distributed raw data, and Student's t test for mean differences of normally distributed variables. Multivariate logistic regression including variables that described patients' decision-making behavior was performed to identify factors for a change of the final treatment decision. For all tests, P o 0.05 denoted statistical significance. Andreas Ihrig performed all calculations using SPSS 21.0 (IBM Corp., Armonk, NY). 2.5. Trial registration and ethics committee approval The present study was part of a larger project comparing conventional self-help groups and OSGs for prostate cancer. The Ethics Committee of the University of Dresden approved the study protocol (EK 75032013). We registered the entire project within the German Clinical Trials Register (http://www.germanctr.de), number DRKS00005086. 3. Results We received 1,426 responses; of these, 814 were from registered users. The questionnaire return rate, which could

4

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

Fig. 1. Flowchart of study participants, group allocation, and data workup. (Color version of figure is available online.)

be estimated only for registered users, was 814/3,357 (24.2%). Fig. 1 provides a synopsis of the data workup: by comparing the IP addresses of the clients' computers and their voluntarily provided e-mail addresses, we identified 182 potential duplicates. Additionally, 116 responses contained no answers, and 30 users declined to participate. Of the 1,098 participants, 139 (12.7%) were patients' relatives or friends, 69 (6.3%) were men without a prostate cancer diagnosis, and 890 (81.1%) were patients with prostate cancer. For this study, we only analyzed the patients with prostate cancer (n ¼ 890) and excluded 204 because of incomplete data. We analyzed 686 completed questionnaires from the patients with prostate cancer in whom current metastatic disease was reported by 105 (15.3%). The average initial PSA (n ¼ 366) was 20.6 ⫾ 75.6 ng/ml (median ¼ 7.5). The median time from diagnosis was 5.8 ⫾ 4.1 years. Following OSG participation, 200 (29.2%) patients revised their initial treatment decision. Comparing patients with revised (n ¼ 200) and unchanged (n ¼ 486) decisions, no major differences were observed in sociodemographic, disease-related, psychometric, or quality-oflife measures (Table 1); moreover, revisions were not associated with prostate cancer risk groups. Decision-making and information-seeking habits differed considerably when comparing the patients with revised (n ¼ 200) and unchanged (n ¼ 486) decisions (Table 2): the patients who revised their decisions wanted a more active role in the physician-patient relationship (P o 0.001), used the Internet for health information more frequently, spent more time in the OSG (410 h: 71.0% vs. 37.2%, P o 0.001), were more likely to actively post messages (69.5% vs. 51.2%, P o 0.001), and reported a higher participation rate in conventional support groups (50.5% vs. 33.3%, P o 0.001).

Comparing the final treatment of patients with revised (n ¼ 200) and unchanged (n ¼ 486) initial decisions (Fig. 2), those who shifted their decision chose EBRT (44.5% (89/200) vs. 36.4% (177/486), P ¼ 0.048) and active surveillance (10.5% [21/200] vs. 3.7% [18/468], P o 0.001) more frequently and opted for RP less frequently (52.5% [105/200] vs. 74.9% [364/468], P o 0.001). Particularly regarding intermediate- and high-risk prostate cancer cases (n ¼ 414), active surveillance was more frequent in patients who changed their initial treatment decision than in those who reported no change (6.4% [8/125] vs. 1.4% [4/289], P ¼ 0.005). Three patient characteristics were independently associated with revision of an initial treatment decision in the multivariate logistic regression (Table 3): longer OSG engagement (odds ratio [OR] ¼ 2.4), demanding a more active role in the decision-making process (OR ¼ 1.6), and participating in a conventional support group (OR ¼ 1.3). To control for attrition bias, 21 people who decided not to participate after reviewing the study information were included in a dropout analysis after answering 5 brief questions. This group did not differ from the participants in age (t test: P ¼ 0.78) or discussion board activity (U-test: P ¼ 0.07). Data protection concerns comprised the most commonly stated reason for not participating. Additionally, the patients with prostate cancer were included in a missing-value analysis [25] comparing the 686 complete and 204 incomplete responses (terminated early). The groups did not differ in any sociodemographic or disease-related parameters. However, the participants who terminated the questionnaire early reported less discussion board activity. Because we asked about possible revisions of the initial treatment decision in the questionnaire's final section, only 16 of the 204 (7.8%) incomplete respondents provided an answer. Of these 16 incomplete respondents 2 reported a change in their treatment decision.

4. Discussion 4.1. Summary of findings This study explored an OSG's effect on the final treatment decision of patients with localized prostate cancer. Specifically, an online survey was open to users of the largest German OSG for a 3-month period, and 686 questionnaires were ultimately analyzed. The results showed that 29.2% of the respondents revised their treatment decision owing to OSG participation. The patients who changed their decisions chose EBRT or active surveillance rather than RP; additionally, such patients spent more time in the OSG, demanded a more active role in the physician-patient relationship, and had participated in a conventional support group more often. Although this study focused on an OSG targeting people affected by prostate cancer, the results have potentially significant implications

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

5

Table 1 Univariate analysis of sociodemographic and disease-related data (means ⫾ standard deviations or absolute [relative] frequencies; P o 0.05 indicates statistical significance). Variable

Treatment decision unchanged n ¼ 486

Treatment decision changed n ¼ 200

P

Mean age (y) City (410,000 inhabitants) Income (€ per mo) o1000 1000–3000 43000

65.0 ⫾ 8.3 336 (69.1%)

66.2 ⫾ 8.6 123 (61.5%)

0.08a 0.05b 0.83c

32 (6.6%) 291 (59.9%) 163 (33.5%)

9 (4.5%) 125 (62.5%) 66 (33.0%)

Language skills Native speaker Nonnative speaker

454 (93.4%) 32 (6.6%)

187 (93.5%) 13 (6.5%)

Family status With a partner Single

439 (90.3%) 47 (9.7%)

174 (87.0%) 26 (13.0%)

Children Yes No

418 (86.0%) 68 (14.0%)

169 (84.5%) 31 (15.5%)

Insurance status Public health insurance Private health insurance

309 (64.6%) 169 (35.4%)

132 (66.3%) 67 (33. 7%)

Education High (Z12 years) Low or medium (r11 years)

235 (48.4%) 251 (51.6%)

90 (45.0%) 110 (55.0%)

0.97b

0.20b

0.61b

0.67b

0.42b

Initial PSA value, ng/ml (n ¼ 235/131)d Time from diagnosis, y (n ¼ 481/198) Biopsy Gleason score (n ¼ 405/176)d 5 6 7 8 9 10

22.2 ⫾ 89.9 5.7 ⫾ 4.0

17.9 ⫾ 38.7 6.2 ⫾ 4.2

20 (4.9%) 136 (33.6%) 169 (41.7%) 39 (9.6%) 30 (7.4%) 11 (2.7%)

17 (9.7%) 47 (26.7%) 69 (39.2%) 22 (12.5%) 19 (10.8%) 2 (1.1%)

T stage (initial) (n ¼ 353/166)d T1 T2 T3 T4

58 (16.4%) 153 (43.2%) 126 (35.7%) 16 (4.5%)

31 (18.7%) 66 (39.8%) 65 (39.2%) 4 (2.4%)

Nþ (n ¼ 440/175)d Mþ (n ¼ 444/180)d

86 (19.5%) 70 (15.8%)

34 (19.4%) 35 (19.4%)

Prostate cancer risk group (n ¼ 391/170)d Low Intermediate High

102 (26.1%) 83 (21.2%) 206 (52.7%)

45 (26.5%) 33 (19.4%) 92 (54.1%)

Distress thermometer EORTC Global Health PHQ-4 sum score (n ¼ 479/196)d

3.9 ⫾ 2.7 68.3 ⫾ 21.9 2.4 ⫾ 2.6

3.7 ⫾ 2.6 68.2 ⫾ 22.9 2.1 ⫾ 2.5

0.76e 0.09e 0.63c

0.82c

0.97b 0.27b 0.86c

0.61a 0.95a 0.25a

EORTC ¼ European Organisation for Research and Treatment of Cancer; PHQ-4 ¼ Patient Health Questionnaire (short version). a Student t test was used to examine differences of means between both groups. b Chi-square test was used to examine differences of relative frequencies between both groups. c Mann-Whitney U-test was used to examine differences of ordinal-scaled variables between both groups. d Corrected sample size in both groups because of incomplete data (numbers are provided per group); the percentages refer to the effective proportion of answers. e Mann-Whitney U-test was used to examine differences of not normally distributed raw data between both groups.

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

6

for health communication and the role of online health in patient decision-making.

and gastrointestinal toxicity may indeed outweigh the risk of cancer progression and death in individual decisionmaking [9,26].

4.2. Changes in final treatment decisions 4.3. Strengths and limitations One important question is whether the changes in final treatment decisions that occurred after consulting the OSG were appropriate. Because no strong evidence exists for the efficacy difference of RP and EBRT [10], active surveillance was the only reliable parameter for assessing the quality of the final treatment decisions. Active surveillance is indicated in low-risk prostate cancer only [10]; however, regarding intermediate- and high-risk cases, it was chosen more frequently in patients who revised their decision (6.4% vs. 1.4%, P ¼ 0.01). This finding might indicate that the treatment decision quality in terms of an accurate medical indication is impaired in some cases. However, the possibility exists that some patients may decide against active treatment such as RP and EBRT in intermediate- or high-risk constellations despite awareness about the increased likelihood of cancer progression. Potential side effects such as urinary incontinence, erectile dysfunction,

This is the first study to demonstrate the degree to which final treatment decisions might be affected by participation in a prostate cancer OSG. Categorically, we have described an association because a cross-sectional study cannot determine causation. Our findings relied on a sufficient sample size and a methodically sound research design [27]. We chose the largest German OSG, which represents nearly complete coverage of existing OSGs for prostate cancer in the German language. Consequently, this focus limited our results' generalizability because the effects might differ for other diseases, countries, and age groups, as well as in women. Additionally, our findings may not be representative owing to selection bias. The questionnaire's return rate could be estimated only for registered users (814/3,357; 24.2%). This number may be inaccurate because 381

Table 2 Univariate analysis of decision-making and Internet usage habits (absolute [relative] frequencies; P o 0.05 indicates statistical significance). Variable

Treatment decision unchanged n ¼ 486

Control preferences scale I prefer to make the final choice about which treatment I receive myself. 27 (5.6%) I prefer to make the final choice about my treatment after seriously considering my doctor's 233 (47.9%) opinion. I prefer that my doctor and I share responsibility for deciding which treatment is best for me. 214 (44.0%) I prefer that my doctor makes the final decision about which treatment would be used but 10 (2.1%) seriously considers my opinion. I prefer to leave all decisions regarding my treatment to my doctor. 2 (0.4%) Registered discussion board user Discussion board activity Active (posters) Passive (lurkers)

Treatment decision changed n ¼ 200

o0.001a 20 (10.0%) 116 (58.0%) 63 (31.5%) 1 (0.5%) 0 (0%)

313 (64.4%)

160 (80.0%)

249 (51.2%) 237 (48.8%)

139 (69.5%) 61 (30.5%)

82 (16.9%) 223 (45.9%) 181 (37.2%)

12 (6.0%) 46 (23.0%) 142 (71.0%)

Participation in a conventional support group Never 1 or 2 times More than 2 times

324 (66.7%) 50 (10.3%) 112 (23.0%)

99 (49.5%) 35 (17.5%) 66 (33.0%)

a

o0.001a

o0.001a 56 131 152 147

(11.5%) (27.0%) (31.3%) (30.2%)

Mann-Whitney U-test was used to examine differences of ordinal-scaled variables between both groups. Chi-square test was used to examine differences of relative frequencies between both groups.

b

o0.001b o0.001b

o0.001a

Time spent within the OSG o1 h 1–10 h 410 h

Using the Internet for health information Every day Once per week Once per month Less frequent

P

25 77 69 29

(12.5%) (38.5%) (34.5%) (14.5%)

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

7

Fig. 2. Final treatment decisions among participants with an unchanged or changed initial decision (multiple answers possible). HIFU ¼ high-intensity focused ultrasound; P o 0.05 indicates statistical significance. (Color version of figure is available online.)

questionnaires lacked user status information. A return rate for unregistered users cannot be obtained. However, a return rate of one-fourth is a satisfactory number considering that most eHealth users are inactive or discontinue usage after a brief period of time [28]. Although we attempted to control for the volunteer effect [15] with a dropout analysis, only 21 replies were received. However, a missing-value analysis showed a lower OSG effect on treatment decisions among the participants who terminated the questionnaire early. Only 2/16 (13%) incomplete respondents reported a change compared with 200/686 (29.2%) in the study sample. Despite the low explanatory power of this small number of incomplete respondents, the patients who benefitted from the OSG were understandably more likely to complete the questionnaire. Therefore, the percentage of treatment changes (29.2%) might have been slightly overestimated. Regardless, no strong systematic bias was apparent, and the study sample was representative of the OSG participants. Undeniably, the sociodemographic parameters indicate that the sample was not representative of the general population: participants reported a higher education level (high education level: 47.4% vs. 19.3%

among those aged 50þ years in 2012; German Federal Statistical Office; http://www.destatis.de) and a higher share of private health insurance (34.9% vs. 10.9% in the general population in 2012; Private Health Insurance Association; http://www.pkv.de). Another possible weakness could result from group allocation because we included 62 participants who chose “Not specified” in the unchanged treatment decision group (Appendix 1, item 39). We assumed that patients who had changed their treatment decision would actively choose “Yes.” Additionally, the patients who had not changed their treatment decision likely would have considered this question irrelevant (“Not specified”). To exclude a major bias, we also calculated a comparison between patients with revised (n ¼ 200) and unchanged (n ¼ 424) initial decisions after excluding these 62 participants. There were no relevant changes in the results. Owing to the reduced numbers, however, the result of the chi-square test for comparing EBRT changed from significant (P ¼ 0.048) to nonsignificant (P ¼ 0.066), although the percentages remained nearly the same (36.4% vs. 36.8% in the group with an unchanged treatment decision).

Table 3 Multivariate logistic regression analysis identifying factors associated with a change in the initial treatment decision (N ¼ 686; P o 0.05 indicates statistical significance). Variable (levels)

Estimated regression coefficient (B) Standard error Wald test Exp (B): odds ratio (95% CI) P

Time spent within the OSG (3) Control preferences scale (5) Participation in a conventional support group (3) Discussion board activity (2) Registered discussion board user (2) Use of the Internet to obtain health information (4)

0.89 0.49 0.29 0.24 0.23 0.08

0.16 0.14 0.10 0.21 0.23 0.10

29.31 11.82 7.98 1.31 1.00 0.59

2.4 1.6 1.3 1.3 1.3 1.1

(1.8–3.3) (1.2–2.2) (1.1–1.6) (0.8–1.9) (0.8–2.0) (0.9–1.3)

o0.001 0.001 0.005 0.25 0.32 0.44

8

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

Further limitations include the uncertain accuracy of patient-reported clinical data and the retrospective judgment on decision-making. Because the mean time from diagnosis was 5.8 years, some statements might have been imprecise, and a few men might have been posttreatment when participating in the OSG. Finally, this long interval since diagnosis could introduce a recall bias that might vary after different treatments. This could either lead to overestimate (personal empowerment and change of initial decision) or underestimate (compliance with initial decision) treatment changes. Nevertheless, as these statements are of little emotional importance, we expect a minor effect on our study outcomes.

4.4. Comparison with existing evidence The present study confirms the results of our descriptive analysis of 82 threads from the same OSG over a 32-month period [13]. Men newly diagnosed with localized prostate cancer started all these conversations and asked for advice concerning their treatment decision. Two independent investigators characterized every thread following a standardized protocol. Of all men with an initial therapeutic preference, 75% were finally confirmed herein; 25% of this sample changed their treatment preference, which corroborates the present percentage of 29.2%. Moreover, the frequency of advice favoring RP was lower, that is, 67% compared with 82% for EBRT, corresponding to the current finding that a changed treatment preference resulted in less RP and more EBRT (52.5% vs. 74.9%). Imbalanced user activity may explain this tendency because only 5% of users wrote 70% of all postings [13]. Therefore, a small user group might be opinion-forming [28]. Physicians also presume that this effect occurs in conventional support groups [29]. Studies of face-to-face support groups have shown that patients with prostate cancer consider informational support useful for treatment decision-making [30]. However, the actual effect on emerging treatment decisions remains unclear. The present study is the first to examine this issue in an OSG. Thus far, the literature on OSGs has focused mainly on the empowerment resulting from increased participation [17,31–33]. Herein, empowerment is defined as an educational process that enables patients to take responsibility for their treatment decision. This educational process may result from gathered information, acquired skills, and increased selfawareness [34]. Congruently, all mentioned studies concluded that OSGs strengthen the empowerment process. In contrast, an older meta-analysis showed no effects for OSG participation [15]. However, a decade ago, the Internet was less established for acquiring health-related information. Recent data show that after consultations with physicians, the Internet has become the number-one source of health-related information [35].

4.5. Population-based effect To judge the relevance of our findings, the populationbased effect of OSGs must be estimated. According to data from the German Federal Statistical Office (http://www. destatis.de), 24% of people aged 50 to 70 years use online social networks and search for health-related information on the Internet. Every year, therefore, 7% of the 66,000 patients with newly diagnosed prostate cancer might be affected, and OSGs may determine 4,800 subsequent treatment decisions in Germany. Because relevant cultural differences may exist between German-speaking patients with prostate cancer and others, we may carefully extrapolate these numbers considering this limitation. In the United States of America, for example, 88% of those aged 50 to 64 years and 57% aged 65þ years have Internet access (2014 Internet User Demographics; http://www. pewresearch.org). Of all Internet users, 34% report having used peer-to-peer support, including OSGs [1]. With a mean age at initial diagnosis of approximately 65 years [10], an estimated 25% of all patients with prostate cancer in the United States of America might consult an OSG. This estimate is modest because people facing a chronic disease or a cancer diagnosis are significantly more likely than the average Internet user to search for peer support online [1]. Thus, at least 7% of all affected patients might change their initial treatment decision after the participation in OSG. Given the 2013 USA incidence of prostate cancer (238,600 cases), we estimate that approximately 17,000 treatment decisions therein may be influenced by an OSG, illustrating the potentially huge effect of this phenomenon on the health care system. Importantly, the percentage of Internet users is growing, particularly among people aged Z65 years [1]. 4.6. Consequences for patient counseling Physicians are confronted with a significant number of patients who gather information from the Internet and OSGs [35]. There are similar offers of peer-to-peer support for prostate cancer in many western countries. “Prostate Cancer UK” hosts an OSG with more than 1,500 threads (www.prostatecanceruk.org), in the United States of America “Us TOO prostate cancer” hosts an OSG with 9,000 members (www.ustoo.org), and the “Canadian Prostate Cancer Network” runs a Twitter-network with 11,500 followers (www.prostatecancer.ca). As an exception the “Australian Prostate Cancer Foundation” hosts more than 150 regional face-to-face support groups, but does not run an OSG (www.prostate.org.au). To provide good counseling and patient care, identifying the subgroup of patients prone to acting against their true values and beliefs [16,36,37] is essential. Although sociodemographic and health-related characteristics did not offer a useful discriminator (Table 1), decision-making habits and Internet usage amounts differed largely between the groups (Table 2). Based on our experience, few patients diagnosed

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

with localized prostate cancer attend a face-to-face support group before making their primary treatment decision [35]. Therefore, this factor appears less helpful initially. The 2 strongest discriminators according to our multivariate model are spending 410 hours in an OSG (sensitivity ¼ 44% and specificity ¼ 84%) and demanding a more active role in the decision-making process (sensitivity ¼ 38% and specificity ¼ 78%). The desired degree of patient involvement should be routinely clarified to adjust counseling accordingly. For this the “Control Preferences Scale” is a straightforward single-item instrument to obtain this information in a standardized manner [24]. Knowing whether a patient prefers a paternalistic or autonomous decisionmaking process may increase satisfaction [38] and save time and energy for patients and physicians. Physicians should invest special care in identifying the true values of those patients prone to bias by Internet sources. Additional neutral information could also enable patients to more critically judge acquired information [7,39,40]. 4.7. Future research directions The OSG's substantial effect on final treatment decisions for localized prostate cancer highlights the major importance of understanding this phenomenon. To affirm generalizability of our results, analogous studies should explore OSG effects in other benign and malignant diseases, in different countries and age groups, and in both sexes. Furthermore, a population-based sample might be used to construct a representative picture of the main information sources for health care treatment decisions and to quantify effects of OSG more accurately. 4.8. Conclusion OSGs play an important role in the decision-making process for prostate cancer treatment. Here, 29.2% of patients participating in an OSG revised their initial treatment decision. They tended to choose against RP, favoring EBRT or active surveillance. This phenomenon may affect nearly 17,000 yearly treatment decisions for prostate cancer in the United States of America. Therefore, OSGs are not a neutral source of information and clearly affect patient care. The time spent in an OSG and decision-making preferences might help to identify patients who are prone to this bias. The desired degree of patient involvement in decision-making should be routinely clarified to adjust counseling accordingly. Further research should explore the reasons for patients to revise their initial treatment decisions after gathering information on the Internet. Author contributions Huber and Ihrig had full access to all the study data and are responsible for the integrity of the data and the accuracy of the data analysis.

9

Study concept and design: Huber, Maatz, and Ihrig. Acquisition of data: Huber, Maatz, Muck, and Ihrig. Analysis and interpretation of data: All authors. Drafting of the article: Huber, Maatz, and Ihrig. Critical revision of the article for important intellectual content: All authors. Statistical analysis: Huber and Ihrig. Administrative, technical, or material support: Maatz, Muck, Keck, and Friederich. Study supervision: Huber, Herzog, and Ihrig. Conflict of interest All authors have completed the ICMJE Form for Disclosure of Potential Conflicts of Interest. All authors reported no disclosures. Role of the sponsor The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the article. Acknowledgments We gratefully thank Ralf-Rainer Damm and Holger Juenemann for their assistance in the execution of our study. American Journal Experts provided language-editing services. Portions of this work were presented in abstract form at the Annual Meeting of the German Urological Association in 2014 and at the Annual Meeting of the European Association of Urology in 2015. Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/ j.urolonc.2016.09.010.

References [1] Susannah F. The social life of health information. Washington, DC: Pew Internet & American Life Project, 2011;2011. Available at: pewinternet.org/Reports/2011/Social-Life-of-Health-Info.aspx: [accessed 30.04.14]. [2] McHugh SM, Corrigan M, Morney N, Sheikh A, Lehane E, Hill AD. A quantitative assessment of changing trends in internet usage for cancer information. World J Surg 2011;35:253–7. [3] Davison KP, Pennebaker JW, Dickerson SS. Who talks? The social psychology of illness support groups. Am Psychol 2000;55:205–17. [4] Fang F, Keating NL, Mucci LA, Adami HO, Stampfer MJ, Valdimarsdottir U, et al. Immediate risk of suicide and cardiovascular death after a prostate cancer diagnosis: cohort study in the United States. J Natl Cancer Inst 2010;102:307–14.

10

J. Huber et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–10

[5] Gray RE, Fitch M, Phillips C, Labrecque M, Fergus K. To tell or not to tell: patterns of disclosure among men with prostate cancer. Psychooncology 2000;9:273–82. [6] Wald HS, Dube CE, Anthony DC. Untangling the Web—the impact of Internet use on health care and the physician-patient relationship. Patient Educ Couns 2007;68:218–24. [7] Huber J, Ihrig A, Yass M, Bruckner T, Peters T, Huber CG, et al. Multimedia support for improving preoperative patient education: a randomized controlled trial using the example of radical prostatectomy. Ann Surg Oncol 2013;20:15–23. [8] Gwede CK, Pow-Sang J, Seigne J, Heysek R, Helal M, Shade K, et al. Treatment decision-making strategies and influences in patients with localized prostate carcinoma. Cancer 2005;104:1381–90. [9] O'Callaghan C, Dryden T, Hyatt A, Brooker J, Burney S, Wootten AC, et al. ‘What is this active surveillance thing?’ Men's and partners' reactions to treatment decision making for prostate cancer when active surveillance is the recommended treatment option. Psychooncology 2014;23:1391–8. [10] Heidenreich A, Bastian PJ, Bellmunt J, Bolla M, Joniau S, van der Kwast T, et al. EAU guidelines on prostate cancer. Part 1: screening, diagnosis, and local treatment with curative intent—update 2013. Eur Urol 2014;65:124–37. [11] Bekelman JE, Suneja G, Guzzo T, Pollack CE, Armstrong K, Epstein AJ. Effect of practice integration between urologists and radiation oncologists on prostate cancer treatment patterns. J Urol 2013;190:97–101. [12] Tariman JD, Berry DL, Cochrane B, Doorenbos A, Schepp KG. Physician, patient, and contextual factors affecting treatment decisions in older adults with cancer and models of decision making: a literature review. Oncol Nurs Forum 2011;39:E70–83. [13] Huber J, Ihrig A, Peters T, Huber CG, Kessler A, Hadaschik B, et al. Decision-making in localized prostate cancer: lessons learned from an online support group. BJU Int 2011;107:1570–5. [14] Berry DL, Wang Q, Halpenny B, Hong F. Decision preparation, satisfaction and regret in a multi-center sample of men with newly diagnosed localized prostate cancer. Patient Educ Couns 2012;88:262–7. [15] Eysenbach G. Health related virtual communities and electronic support groups: systematic review of the effects of online peer to peer interactions. Br Med J 2004;328:1166–70. [16] Bosco JLF, Halpenny B, Berry DL. Personal preferences and discordant prostate cancer treatment choice in an intervention trial of men newly diagnosed with localized prostate cancer. Health Qual Life Outcomes 2012;10:123. [17] van Uden-Kraan CF, Drossaert CH, Taal E, Shaw BR, Seydel ER, van de Laar MA. Empowering processes and outcomes of participation in online support groups for patients with breast cancer, arthritis, or fibromyalgia. Qual Health Res 2008;18:405–17. [18] Lieberman MA, Golant M, Giese-Davis J, Winzlenberg A, Benjamin H, Humphreys K, et al. Electronic support groups for breast carcinoma: a clinical trial of effectiveness. Cancer 2003;97:920–5. [19] Maher CA, Lewis LK, Ferrar K, Marshall S, De Bourdeaudhuij I, Vandelanotte C. Are health behavior change interventions that use online social networks effective? A systematic review. J Med Internet Res 2014;16:e40. [20] Eysenbach G. Improving the quality of Web surveys: the Checklist for Reporting Results of Internet E-Surveys (CHERRIES). J Med Internet Res 2004;6:e34. [21] Sprangers MA, Cull A, Groenvold M, Bjordal K, Blazeby J, Aaronson NK. The European Organization for Research and

[22]

[23] [24]

[25]

[26]

[27] [28] [29]

[30] [31]

[32]

[33]

[34] [35]

[36]

[37]

[38]

[39]

[40]

Treatment of Cancer approach to developing questionnaire modules: an update and overview. EORTC Quality of Life Study Group. Qual Life Res. 1998;7:291–300. Kroenke K, Spitzer RL, Williams JB, Lowe B. An ultra-brief screening scale for anxiety and depression: the PHQ-4. Psychosomatics 2009;50:613–21. National Comprehensive Cancer Network. Distress management. Clinical practice guidelines. J Natl Compr Canc Netw 2003;1:344–74. Degner LF, Sloan JA. Decision making during serious illness: what role do patients really want to play? J Clin Epidemiol 1992;45: 941–50. Little RJ, D'Agostino R, Cohen ML, Dickersin K, Emerson SS, Farrar JT, et al. The prevention and treatment of missing data in clinical trials. N Engl J Med 2012;367:1355–60. Hayes JH, Ollendorf DA, Pearson SD, Barry MJ, Kantoff PW, Stewart ST, et al. Active surveillance compared with initial treatment for men with low-risk prostate cancer: a decision analysis. J Am Med Assoc 2010;304:2373–80. Wyatt JC. When to use web-based surveys. J Am Med Inform Assoc 2000;7:426–9. van Mierlo T. The 1% rule in four digital health social networks: an observational study. J Med Internet Res 2014;16:e33. Steginga SK, Smith DP, Pinnock C, Metcalfe R, Gardiner RA, Dunn J. Clinicians' attitudes to prostate cancer peer-support groups. BJU Int 2007;99:68–71. Coreil J, Behal R. Man to man prostate cancer support groups. Cancer Pract 1999;7:122–9. Mo PKH, Coulson NS. Empowering processes in online support groups among people living with HIV/AIDS: a comparative analysis of ‘lurkers’ and ‘posters’. Comput Hum Behav 2010;26:1183–93. Setoyama Y, Yamazaki Y, Nakayama K. Comparing support to breast cancer patients from online communities and face-to-face support groups. Patient Educ Couns 2011;85:e95–e100. Broom A. Virtually he@lthy: the impact of internet use on disease experience and the doctor-patient relationship. Qual Health Res 2005;15:325–45. Feste C, Anderson RM. Empowerment: from philosophy to practice. Patient Educ Couns 1995;26:139–44. Huber J, Ihrig A, Huber CG, Hadaschik B, Pahernik S, Hohenfellner M. Patient centeredness and decision-making in localised prostate cancer: possible fields of health services research in urology. Urologe 2011;50:691–6. Owen JE, Bantum EO, Golant M. Benefits and challenges experienced by professional facilitators of online support groups for cancer survivors. Psychooncology 2009;18:144–55. Stephen JE, Christie G, Flood K, Golant M, Rahn M, Rennie H, et al. Facilitating online support groups for cancer patients: the learning experience of psycho-oncology clinicians. Psychooncology 2011;20: 832–40. Shabason JE, Mao JJ, Frankel ES, Vapiwala N. Shared decisionmaking and patient control in radiation oncology: implications for patient satisfaction. Cancer 2014;120:1863–70. Violette PD, Agoritsas T, Alexander P, Riikonen J, Santti H, Agarwal A, et al. Decision aids for localized prostate cancer treatment choice: Systematic review and meta-analysis. CA Cancer J Clin 2015; 65:239–51. Groeben C, Streuli JC, Krones T, Keck B, Wirth MP, Huber J. Treatment of nonmetastatic prostate cancer. Urologe 2014;53: 854–864.