Evaluating Clinical Implementation Approaches for Prostate Cancer Decision Support

Abstract

Introduction:

Shared decision making is widely promoted for counseling men with localized prostate cancer. Results of randomized trials suggest decision aid efficacy. However, few practices or institutions have implemented decision support as standard practice. In this study we evaluated various implementation strategies for the decision aid P3P (Personal Patient Profile-Prostate) and analyzed feedback from clinical site staff and providers.

Methods:

A hybrid type 1 effectiveness-implementation trial was conducted. Primary data were collected in 6 urology clinics of 3 geographically distinct health networks. During the implementation phase site specific strategies were codesigned with site leaders. Referral and access metrics for men with localized prostate cancer were monitored for up to 7 months. Clinical staff reports of barriers and facilitators of implementation were evaluated in professionally facilitated focus groups.

Results:

Of 495 men with localized prostate cancer seen in the clinics 252 (51%, 95% CI 46-55) were informed of the program and of those men 107 (43%, 95% CI 36-49) accessed it. The highest access rates were observed with patient care coordinator e-mail and telephone contact (82%) or verbal physician instruction followed by e-mail and telephone invitations (87%). During focus groups physicians appraised the summaries as useful. Staff identified barriers included creating new workflows within heavy workloads and staff misunderstanding of context and resources. Promoters of successful implementation included an identified clinical lead and physician engagement.

Conclusions:

Implementation success was realized when physicians engaged and staff provided followup contact. New practice changes to implement interventions require multimodal strategies for early success.

Development of treatment decision support interventions in the first decade of the 21st century was propelled by a rise in consumerism and patient engagement,¹ growing recognition of medical treatment decisions that had little evidence of efficacy between options² and increased research support from funding sources. Localized prostate cancer is a common disease lacking a definitive medical mandate for a specific treatment option. Men with low risk LPC often are counseled about active surveillance. Contemporary guidance for moderate to high risk LPC includes active therapy but does not dictate a particular modality (eg surgery vs radiation therapy). Upon receiving a positive biopsy result, men meet with providers and hear a range of options with associated adverse and survival outcomes and are at risk for uncertainty related to the care decision. Prostate cancer specialists engage in the time-consuming task of helping men, and often, men’s partners, navigate decision making and settle on a mutually acceptable option.

Few practices or institutions have implemented decision aids for men with LPC as standard practice and even fewer have evaluated such implementation.^3,4 In a white paper from the AUA (American Urological Association) Makarov et al posited that shared decision making requires health care providers to be properly trained in DA use and also willing to allow patients to talk after provider elicitation of patient preferences.⁵ The 2017 AUA Guidelines for localized prostate cancer begin with a strong recommendation for shared decision making support.⁶

The Personal Patient Profile-Prostate is a decision aid that elicits patient preferences in the context of a LPC diagnosis, and generates educational and coaching guidance tailored to the user’s values, preferences and characteristics.⁷ P3P does not recommend specific choices but rather prepares the user to make a decision in consultation with clinicians. Two randomized trials have demonstrated the efficacy of P3P, significantly reducing decisional conflict experienced by men considering options for LPC.^7,8 Our team next evaluated the implementation of P3P as routine practice in 6 urology and multidisciplinary clinics in 3 geographically distinct health networks. The purpose of the evaluation was to assess the various implementation strategies chosen by clinics by measuring referral and access rates to the DA and analyzing feedback from clinical staff and providers.

Materials and Methods

Design

This study was conducted as the implementation phase of a hybrid type 1 effectiveness-implementation trial in which a clinical intervention is tested while gathering information on its delivery and implementation in a real-world environment.⁹ P3P met suggested criteria of containing a low risk intervention that had face validity and evidence of efficacy in a different but related population. The Dana-Farber/Harvard Cancer Center institutional review board approved the study (IRB No. 12-363), followed by approval at each network site IRB. The 6 largest volume clinical settings in 3 health networks from the randomized, initial study phase were included in the second phase for implementation.

Pre-Implementation Procedures

Deployment planning sessions were convened at each site. The PI attended at least 1 session with key personnel at each of the 3 networks. The physician staff were provided a brief presentation of the program of research efforts to date plus copies of publications. The nature of face-to-face meetings with nurses and staff varied in that some attendees were not available for more than 15 minutes and others met for up to an hour. During these sessions attendees described current practice processes and considered various approaches to P3P deployment, recommending the process in each individual setting that was believed to be possible and potentially successful. Web trainings also were conducted with the clinic staff, including patient care coordinators and nurses, who would refer patients to P3P and/or receive P3P reports sent by patients. The staff were provided with instruction sheets for using the site and flyers for P3P referrals.

Using the Personal Patient Profile-Prostate

Methods of informing patients that P3P was available varied among sites per clinical providers and staff preferences. All methods were intended to make P3P available after the disclosure of a positive biopsy but before the full options review visit. The supplementary figure (http://urologypracticejournal.com/) illustrates the basics of the study periods. Patients were invited to use P3P but no patient was told that use was mandatory. Once a patient had received the link, remote access to P3P was possible or, for those men without personal Internet access, iPads were provided at each site so the user could access before the options review clinic visit. The components of P3P have been described previously.^7,8,10 In brief, users answer a series of questions focused on personal characteristics, concerns and values, current symptoms and where they are with regard to making a decision. A tailored educational and communication coaching intervention in the form of text, graphs and short video clips on medical facts and personally relevant factors for the prostate cancer decision is delivered. Patients were given the option of sharing a report of their P3P responses with providers, either by printing and bringing the report to clinic, or when clinics permitted, sending an e-mail link to the report to the clinic. A unique staff member was identified at each clinic to accept reports electronically. This person downloaded and/or printed a copy of the report for the clinic encounter.

Analytic Considerations

We planned a multimethod evaluation, described by Creswell as “explanatory sequential mixed methods.”¹¹ Specifically we evaluated the feasibility and implementation processes of P3P in the health network settings regarding patient awareness and access rates, followed by qualitative assessment of clinician utility and promoters and barriers to physician/clinical staff implementation.

Research coordinators at each site retrospectively screened the appointment schedules monthly for men who should have been referred to P3P, ie those who had planned consult visits occurring after a diagnosis of prostate cancer. The clinic staff who had been designated to offer P3P to patients reported the number of patients they had referred. We calculated the referral rate, that is the ratio of men referred to P3P relative to the total number of scheduled consults for options discussion. The P3P server logged patient access rates with a plug-in that matched users to clinic sites geographically in cases in which users did not identify the clinics. We calculated the number of referred men who accessed P3P and provided 95% CIs of the estimates. Basic demographics were queried, with optional additional questions about education level and previous computer use.¹²

Approximately 6 months after beginning implementation, site specific research coordinators sent e-mail invitations for focus groups to all clinic personnel who may have had any contact with P3P. Semi-structured focus groups were convened at each site to explore the successes, promoters and barriers to P3P implementation, and future plans and suggestions. Experienced facilitators from outside the study team were engaged to lead the focus groups, allowing participants the freedom to critically appraise the process and program. The study team held webinars with the facilitators to review the implementation process, focus group guide and goals of the focus groups. Site specific P3P informed and access rates were provided to the participants at each group. Audio recorded focus group data were de-identified before analysis by a central office research coordinator and then transcribed by a contracted, independent company.

Analyses

Engagement Rates.

Descriptive statistics were used to summarize continuous and categorical data. The informed and access rates were reported with point estimates and 95% CIs.

Qualitative Analysis.

Focus group narratives were entered in NVivo 11 (QSR International Pty Ltd). Using a conventional content analytic approach,¹³ the PI (DB) and a research assistant coded each transcript and came to consensus on the initial coding structure. The PI finalized the categories describing the experiences of participants relevant to the implementation period.

Results

During the planning sessions with the PI and site leadership and staff, the 2 sites where a consistent approach to patient education already existed chose to add P3P to an educational resource packet. Other sites provided P3P as the primary method for decision support. Each site created a workflow for implementation, inserting P3P referrals into the work of various clinic personnel. The strategies selected and resources provided are listed for each clinic site in table 1.

Table 1.

Strategies for informing patients about P3P, percent informed and access at 6 clinics in 3 networks

Urology Clinic Location	Strategy	In-Clinic Access*	No. Men with LPC Seen in Clinic	No. Men Informed about P3P (%)	No. Informed Men who Accessed P3P (%)
Southwest #1	URL on after visit summary, nurse reviewed summary with pts	None	104	51 (49)	7 (14)
Southwest #2	P3P flyer added to information packet along with clinic developed booklet and video, nurse reviewed packet with pts	None	40	39 (98)	0 (0)
Southeast #1	E-mail contact by nurse navigator	Tablets	137	73 (53)	23 (32)
Southeast #2	Phone + e-mail contact by pt care coordinator	Education room	35	17 (49)	14 (82)
Southeast #3	In person, in clinic only	Tablets	34	5 (15)	5 (100)
Northeast	Physician instruction, phone + e-mail contact by pt care coordinator	Tablets	145	67 (46)	58 (87)

^*Access and equipment provided to the clinic for men without Internet access.

A total of 495 men with LPC were seen at the 6 clinics during the approximate 6-month implementation period. Table 1 lists the percentages of these men who were informed of P3P availability using strategies selected at each clinic. Of all men with LPC seen in the clinics 252 (51%, range 15 to 98 per clinic, 95% CI 46-55) were informed of P3P and 107 (43%, range 0 to 100 per clinic, 95% CI 36-49) of informed participants accessed P3P. Among the 107 men who accessed P3P the majority were white, 60 years old or older and married (table 2). Clinics in the Southeast engaged a higher proportion of black men and the Northeast clinic engaged an older, more educated group. Of the 81 (76%) men who were willing to provide optional demographics the majority had college or higher education and reported frequent computer use at home. Most men (63%) preferred shared decision making, 49% were thinking about treatment options and 35% were close to or already made a decision.

Table 2.

Baseline demographics of 107 patients

	No. Clinical Sites (%)
	Northeast		Southeast		Southwest		No. All Sites (%)
Age:*
Missing	0	(0)	1	(2.4)	1	(14.3)	2	(1.9)
Less than 60	18	(31.0)	21	(50.0)	2	(28.6)	41	(38.3)
60 or Older	40	(69.0)	20	(47.6)	4	(57.1)	64	(59.8)
Marital status:*
Missing	2	(3.4)	0	(0)	1	(14.3)	3	(2.8)
Married/partnered	43	(74.1)	34	(81.0)	4	(57.1)	81	(75.7)
Not married/Partnered	13	(22.4)	8	(19.0)	2	(28.6)	23	(21.5)
Race/Ethnicity:*
Missing	2	(3.4)	1	(2.4)	1	(14.3)	4	(3.7)
Black	5	(8.6)	13	(31.0)	0	(0)	18	(16.8)
Others	1	(1.7)	0	(0)	0	(0)	1	(0.9)
White-Hispanic	1	(1.7)	1	(2.4)	1	(14.3)	3	(2.8)
White-nonHispanic	49	(84.5)	27	(64.3)	5	(71.4)	81	(75.7)
Education:†
Missing	14	(24.1)	12	(28.6)	2	(28.6)	28	(26.2)
College or higher	40	(69.0)	25	(59.5)	4	(57.1)	69	(64.5)
No college	4	(6.9)	5	(11.9)	1	(14.3)	10	(9.3)
Computer use at home:†
Missing	13	(22.4)	11	(26.2)	3	(42.9)	27	(25.2)
Never/rarely/sometimes	5	(8.6)	4	(9.6)	0	(0)	9	(8.4)
Often/very often	40	(69.0)	27	(64.3)	4	(57.1)	71	(66.3)
Use website resources:†
Missing	13	(22.4)	10	(23.8)	3	(42.9)	26	(24.3)
Never/rarely/sometimes	17	(29.3)	14	(34.3)	0	(0)	31	(28.9)
Often/very often	28	(48.3)	18	(42.9)	4	(57.2)	50	(46.8)

^*Asked for the purpose of delivering the tailored intervention. Missing indicates the men who skipped the item.

^†Asked for sample description purposes, required additional online permission. Missing indicates those who did not give permission and were not offered the item.

The highest observed rates of P3P access, outside of clinic and before the consult visit, were observed with patient care coordinator e-mail and telephone contact (82%) or verbal physician instruction followed by e-mail and telephone invitations (87%). Clinic sites that chose only to provide written material with suggestions to access P3P had the lowest access rates (range 0% to 14%). Staff at the 1 clinic (Southeast #3) that chose in-clinic only administration on tablets reported that all 5 invited patients were able to use P3P once the staff assisted with creating a user name and password, but did not complete the full program before being called in to see a provider.

Of the 39 physicians, 17 nurses and 15 clinic staff invited to attend the focus groups 13 (33%), 2 (12%) and 10 (67%) attended, respectively. Fifty percent of participants were men, and of those who reported 22 of 25 (88%) were non-Hispanic, 13 of 22 were white (59%) and 13 of 24 (54%) had worked in the clinic for 0 to 3 years. The focus group results included categories of issues discussed relevant to implementation (see Appendix). Barriers common to most settings included creating new workflows on top of heavy workloads, and staff and administrator misunderstanding of P3P context and resources. Staff inability to identify men with new LPC (vs those with advanced disease) hampered access. Promoters of successful implementation included an identified clinical lead, physician engagement and telephone combined with e-mail contact. The physicians who recalled receiving the clinician summary reported positively on utility and format.

Discussion

In an implementation study with a multimethod evaluation we found wide variation among 6 clinical sites with regard to selected implementation strategies, rates of informing men of decision support and program access rates. Identified barriers to implementation and recommended strategies for success were more consistent among sites.

Despite numerous clinical trials of decision support for men with LPC, only 1 published planned protocol has addressed decision aid implementation as routine care.¹⁴ Is our 51% access rate low or high? With no published results of LPC decision aid implementation as usual care, it is not possible to compare rates in the urology setting. Belkora et al reported successful implementation of a breast cancer decision aid but implementation depended on unpaid interns and students, not the real-world setting of our evaluation.¹⁵

Compared to our randomized trial enrollment, a lower percentage of minority men accessed P3P during the implementation evaluation.⁷ This could be an artifact of the digital divide and/or the fact that during the trial, research coordinators were readily available with tablets for men without home Internet access. However, we do know that certain implementation strategies, notably the combination of in-person verbal instruction followed by telephone/e-mail modes of contact, were most successful and these relative rates are instructive for planning future DA interventions. Clearly, adding a flyer to many pages in a packet of information was not successful at all.

The barriers reported in the focus groups were not unlike barriers to many new practice changes. Elwyn et al reviewed barriers and facilitators to implementation of decision support for a variety of clinical conditions.³ Although not reported for the LPC setting, barriers were consistent with our focus group findings. DAs were perceived as competing with existing patient education materials, referring to DAs was not perceived to be part of the clinician and staff role and competing demands precluded referral and supportive activities. As in our findings, systematic identification of eligible users facilitated uptake as well as remote use before arrival at the clinic.

The usefulness of the clinical summary was not unexpected. In our previous work clinicians did use and appreciated having a pre-visit summary of patient reported outcomes, values and preferences.

Our findings are limited by several circumstances. We had no method to document the language used when clinicians and staff referred patients to use P3P. Variability certainly existed among staff members. Because reporting education level and experience with computers required online consent, we had fewer data on these variables. The majority of users who did consent reported some college education, a higher percentage than in our randomized trial at the same sites.⁷ It is possible that staff selectively referred computer users. Alternately, the men who accessed P3P more often and answered the additional demographic questions may have used computers more frequently. Either possibility may be related to the clinics’ decisions that providing point of service access was not feasible or could seldom be done, limiting users generally to those with remote Internet access. Further, clinic staff did not track reasons for nonaccess if they had been aware. Finally, only about a third of clinicians and staff who had been identified as having knowledge of P3P were able and willing to attend the focus groups. This self-selection may have impacted the feedback provided. The feedback from the clinicians as to the usefulness of the summaries was solely spontaneous self-report.

Conclusions

Implementation of an efficacious decision aid for men newly diagnosed with prostate cancer varied depending on strategies chosen by clinical staff. Higher access rates were realized at clinics that combined in-person, verbal and e-mail modes of communication to inform men of P3P availability. Barriers included staff inability to identify appropriate users, creating new workflows on top of heavy workloads, and misunderstanding of P3P context and resources by staff and administrators. Suggested facilitators included having a physician active during referral and an identified clinical staff lead. Future clinical implementation efforts will need to address these issues. P3P is available at https://www.p3p4me.org.

Abbreviations and Acronyms

DA

decision aid

LPC

localized prostate cancer

P3P

Personal Patient Profile-Prostate

PI

principal investigator