Variations in Patient Portal Adoption in Four Primary Care Practices

Abstract

This case report reviews the patient portal adoption experiences of four primary care practices at Partners HealthCare during 2002 – 2009. Although each practice used the enterprise patient portal (Patient Gateway) and electronic health record, their patient enrollments varied substantially, as did their marketing approaches, new feature adoption, leadership approach, and staff involvement. Marketing limitations, leadership concerns, and limited staff engagement characterized the low-enrollment practices, but not the others. These factors, along with other practice characteristics such as location and patient demographics, should be explored in future research to identify best practices for successful adoption of a patient portal.

Introduction

Many reports anticipate that personal health records (PHRs) will help patients and providers alike to address healthcare priorities such as patient safety, management of chronic conditions, coordination of care, and seamless information sharing^1–5. A growing number of provider organizations, including the Veterans Administration, Kaiser, Group Health Cooperative in Seattle, Caregroup in Boston, and many others, offer secure internet portals (a type of PHR) used by patients to view shared chart information, access linked reference information, and communicate online with their doctor’s office and care team^3,6,7.

Partners HealthCare, a large integrated delivery network in Eastern Massachusetts, has offered secure online chart and communication services to patients through its enterprise patient portal, Patient Gateway (PG), since 2002. While overall industry adoption of PHRs has been slow, blamed on a myriad of technical barriers (e.g. non-interoperable systems, lack of data standards, fragmented policies, etc.) and consumer obstacles (e.g. computer competency, internet access, health literacy, etc.)⁵, Patient Gateway has been adopted in over 100 primary care, specialty care, and pediatric practices as of 12/31/2009.

The process of practice implementation of PG consists of staff and physician training, setting a “go live” date, configuring the software, marketing to patients, and daily use of the live system in response to patient requests or practice needs. Several steps are critical to how patients hear about and experience PG. Once aware of PG, the patient can request an account online or in person, and will receive a printed password letter to activate their account (note: starting in 2010, patients can get access immediately, without waiting for the mailed password letter, through knowledge-based authentication online).

Little is published about how the practice implementation of a patient portal may influence patient enrollment and use. This case study focuses on the experiences of four primary care practices to identify factors that may facilitate or slow adoption.

Methods

Enrollment data was reviewed for fifteen primary care practices at Brigham and Women’s Hospital and Massachusetts General Hospital that implemented Patient Gateway prior to September 2003. Following an initial post-launch phase, persistent differences in enrollment levels were noted among many practices.

Four practices were chosen for analysis in this case report to highlight different enrollment and patient usage patterns. Time periods were identified empirically based on enrollment data (Figure 1) showing consistent enrollments during October 2004 to September 2007 (Period 2), and greater enrollment fluctuations before (Period 1) and after (Period 3) that interval.

Figure 1.

Quarterly enrollments (per 1000 patients) in four primary care practices during three time periods, 2002–2009.

All practices using PG had already implemented the LMR (Longitudinal Medical Record) and were offered the same feature set when implementing Patient Gateway. Features initially available included secure messaging, requests (prescription, appointment, referral authorization), chart information (medications, allergies, appointments), and health library (licensed content from Healthwise®). Additional features were offered to all practices in subsequent years, such as lab results and physician results letters (2006), and radiology reports (2009). PG includes the ability for practices to send a ‘broadcast’ message (a single message) to all (or a defined set of) patients in a practice.

Patient Gateway routinely records usage data (patient enrollments, patients with sessions, patients with requests, and a marketing question – “How did you hear about Patient Gateway?”). Monthly and quarterly data was collected for each primary care practice, along with implementation notes based on unstructured discussions with practice leadership and staff.

Implementation plans documented the marketing strategies and the software configuration for each practice, including which features were activated (such as Lab Results, first available in 2006). Two implementation analysts in an operational (not research) role were responsible for the four practice implementations, and provided subjective judgments regarding the style of leadership and level of staff engagement in PG adoption at each practice to the author. Demographics for each practice (number of active patients, mean age, gender, race, and insurance status) were obtained from an enterprise data warehouse.

New enrollments during each period are reported based on practice size in Table 1 and Figure 1 (e.g., “New per 1000 pts/yr”). Similarly, unique patients with sessions or requests are reported based on the level of patient adoption (e.g., “per 1000 enrolled pts”) in Figures 2 and 3.

Table 1.

Patient demographics in October 2007 (top section); Adoption choices and enrollment totals for each time period (middle section); and Total enrollment (bottom section) for four primary care practices. Marketing key: P=Phone, R=participation in major Research study, S=Staff communication with patients, V=posters and postcards available during an office Visit, L=mailed Letters to practice patients, Q=Quality incentive, $=staff contest.

Label Practice site	D Downtown	S Suburban	C Campus	H Health Ctr
Patients (P)	7721	13306	17138	9842
Mean age	50.2y	53.3y	51.8y	35.9y
Female	49%	67%	73%	60%
Non-white	17%	28%	53%	36%
Comm. Insur.	79%	85%	68%	49%
	Period 1 (6/11/02 to 9/30/04)
PG go live	Jun-02	Apr-03	Jun-03	Sep-03
Period 1 mktg	PRSVL	RSVL	VL	RVL
New per 1000 pts/yr	133	64	49	36
Clin. Msg feature	Yes	Yes	No	No
	Period 2 (10/1/04 to 9/30/07)
Period 2 mktg	PRSV	RSV	V	RV
New per 1000 pts/yr	84	28	8	4
Lab results (2006)	Yes	Yes	No	No
Leadership focus	High	High	Low	Low
Staff engagement	High	High	Low	Low
	How did you hear about PG?
Visit	32%	56%	20%,	54%
Phone	49%	2%	14%	3%
Mail	7%	24%	27%	14%
Web. friend, other	12%	18%	39%	23%
	Period 3 (10/1/07 to 12/31/09)
Period 3 mktg	PSVBQ	SVBQ$	VQ$	VQ
New per 1000 pts/yr	68	126	29	26
	All Periods through 12/31/2009
PG Accounts (A)	5559	5951	2510	1038
New per 1000 pts/yr	96	67	23	17
Penetration (A/P)	72%	45%	15%	11%

Figure 2.

Period 2 monthly count of unique patients who signed in (a “session”) to Patient Gateway per 1000 enrolled patients, October 2004 – September 2007.

Figure 3.

Period 3 monthly count of unique patients making a request via Patient Gateway per 1000 enrolled patients, October 2007 – December 2009.

Results

Practice data and observations are grouped into three sequential time periods corresponding to initial implementation (Period 1: go-live until 9/30/04), production use (Period 2: 10/1/04 to 9/30/07), and new growth (Period 3: 10/1/07 to 12/31/09). Marketing and implementation observations for each practice, including enrollment totals, are reported for each time period in lower sections of Table 1.

Overall PG enrollment penetration (# PG enrollees / # Patients active in the practice) showed 6-fold variation from practice D (72%) to practice H (11%) across all periods (Table 1, lower section).

Patient demographics differed among the four practices, as shown in the upper section of Table 1. Health Center (H) patients were youngest (35.9y vs. 50–53y) and the least likely to have commercial insurance (49% vs 68–85%). Together, the Health Center and Campus practices had the most non-white patients (36% & 53%) and fewest commercially insured patients (49% and 68%) compared to the Downtown and Suburban practices.

During Period 1, rates of patient enrollment (per 1000 practice patients per year) differed 4-fold from practice H (with fewest enrollments, 36) to practice D (with the most, 133). Practice D utilized 5 marketing strategies, compared to only 2 or 3 in the low-enrollment practices (C and H). Those practices also disabled the clinical messaging feature from use by patients whereas practices D and S allowed its use.

During Period 2, the rates of patient enrollment (per 1000 practice patients per year) were lower for each practice compared with Period 1, and dropped to negligible rates for practices C (8) and H (4). Practice D differed from the other practices in using telephone-based marketing, and a large number of their patients reported learning about PG via the telephone (49%), followed by the visit (32%). Practice S had moderate rates of new enrollment, with most patients hearing about PG through a visit (56%) or by mail (24%). Practice S providers and staff often dropped a postcard into patient correspondence, such as a lab letter, informing the patient about Patient Gateway.

During Periods 2 and 3, practices C and H had disabled the Lab Results feature and were nervous about engaging office staff in marketing PG to patients by telephone or using other methods. In contrast, practices D and S maintained high leadership focus, high staff engagement, and activated the new Lab Results feature as soon as it became available.

During Period 3, staff contests were introduced into practices S and C, and all practices were (for the first time) given enrollment targets by primary care management. Practice D was given a very small incremental target because of its high enrollment by that time. During this incentive period, enrollments rose dramatically in three practices, rising three-fold in practice C (8→29), six-fold in practice H (4→26), and four-fold in practice S (28→126).

Usage data

Unique patients per month (per 1000 enrolled patients) with sessions (i.e., who signed in to PG) during Period 2 are shown in Figure 2. Average unique patients with sessions measured for the starting 2-months vs. the last 2-months of this 3-year period showed differing growth. The moderate/high enrollment practices grew in patients-with-sessions by 96% (Practice S: 72→142) and 55% (Practice D: 107→165), whereas the other practices showed 9% (Practice H: 82→89) and −25% (Practice C: 90→67) growth. Early and late “spikes” appear in practices D, S, and H in 2005 and 2007, when broadcast messages (▴) were sent via PG secure mail to all PG patients in those 3 practices as part of a research study. In addition, several new feature broadcast announcements (‡) were sent to patients in practices D and S during June–September 2006, with observed spikes in the number of patients having sessions.

The number of patients making “requests” was also recorded. A “request” is a message sent to the practice by a patient asking for service (a medication renewal, an appointment, a referral, a registration change, or a clinical question). Patients per month making service requests was typically about 35–40% of the number of patients having a session, indicating that “self-service” use of PG (e.g. browsing lab results) without making a request was very common.

During Period 3, October 2007 to December 2009, enrollment rates rose substantially in practices S, H, and C (Figure 1). Patient service requests for this period are shown in Figure 3. The average unique patients per 1000 enrollees making requests to the practice during the first 2 months vs. the last 2 months of Period 3 grew considerably in three practices – by 307% in Practice H (37→151), 155% in Practice S (125→320), and 78% in Practice C (35→62). Practice D patients-with-requests was highest (of the four practices) at the start of Period 3 (144), with −5% growth (137) by the end of the period.

Request rates in late 2007 vs. late 2009 show a rise in overall patients-with-requests across all practices, and about four-fold practice variation from highest to lowest rate in late 2007 (D=144 vs. C=35) and in late 2009 (S=320 vs. C=62). Some practices (D,C) showed consistent rates during this period, while others (S,H) showed growth.

Discussion

In this case report, no attempt is made to quantify the relative contribution of factors such as patient demographics as enablers or barriers to patient adoption or use. They likely play a role. In addition, the fact that enrollment rates were observed to rise substantially during Period 3 (with the use of incentives) suggests that practice-related factors were also important during Periods 1 and 2. Several implementation issues deserve special mention as potential influencers of patient adoption and use of a patient portal.

Leadership concerns

Practice C leadership repeatedly expressed concerns that PG use would lead to a high volume of unwanted or burdensome messages to physicians, though measurement of PG request volume suggested low message volumes among practices using PG, and many staff who handled electronic requests viewed them as more efficient than similar telephone-based requests (personal communications). It is possible that practice C, a resident clinic, had specific workflow or staffing issues contributing to this concern.

Adoption activities at practice C seemed consistent with leadership concerns. Staff did not feel encouraged to offer PG to patients, practice-wide marketing methods were avoided, and some physicians did not market PG to their patients at all. In addition, as new PG features important to patients became available (such as Lab Results), they were deferred. The overall effect was to reduce new patient enrollments, reduce patient use, and diminish the value of PG to patients and providers alike.

Low levels of adoption in practice C compared to other practices persisted through all three periods. This raises the question whether early assessment of leadership and staff to fully understand and address any preconceptions or concerns that were raised might have improved adoption. It is possible that delay of practice adoption until concerns were addressed might be a better strategy than putting practice staff and physicians into a “bind” – having to adopt something they believe might be burdensome to the practice. It is also possible that only with direct experience in offering a patient portal would practice leaders learn what would really happen to staff workload, and workflow with regular patient portal use.

More recent discussions among implementation staff and practice leadership, based on the experiences at many practices using PG, have led to a consensus that offering patients all features of PG creates the most positive and consistent user experience and offers the greatest value for all users.

Marketing methods

This case study suggests that automated telephone system scripts informing patients about PG services (each time they phone the practice) can be highly effective. This marketing method is advantageous because staff time is not consumed to market PG, and patients learn about PG at a moment of need – when they are phoning to speak with the practice. Now (in 2010) that patients can get their PG initial password immediately via online authentication, they can make use of practice online services as soon as they have a need.

Another finding suggested by this study is that marketing collateral (posters, postcards, etc.) available during a visit is marginally effective without enthusiastic staff and providers presenting it. It also appears likely that “selective marketing” to some patients can be difficult or impossible to implement. In other words, when all physicians in a practice support a tool like PG, staff can enthusiastically mention it to patients without worry that it does not apply. Conversely, if some physicians are not enthusiastic about recommending the patient portal, staff may have to avoid suggesting PG to patients (in general) to avoid mixed messages to the patients.

Staff and physician knowledge and enthusiasm about PG when communicating with patients seemed to be important for patient adoption regardless of the practice. A number of staff reported that having their own PG account helped improve their understanding of the tool and its potential value to patients.

Recently, practices planning to go live with PG in 2010 have begun to collect email addresses from their patients in advance of the go live date. Once that date arrives, an email message is sent out (with a link) to remind patients to sign up. In discussing the Period 1 enrollment spike with practice D, it became apparent that physicians already using email to communicate with their patients began to inform them that they would be using PG instead of email for security reasons. This explains the “double spike” (one from emails to patients at go live, and one from a mailing done later) in Figure 1 for practice D.

Service orientation

Patient experience seems to be affected by many factors, including the service culture of the practice staff – especially those who triage and respond to patient messages. Leadership and staff at practices D and S shared their view that the patient portal was an extension of their service offering to patients, and should provide the same high quality experience and fast response times whether patients interact online, by telephone message, or directly with staff.

Incentives and Innovation

The dramatic increase in enrollments during Period 3, following three years of “flat” enrollments in practices C and H, shows the power of incentives in adoption. A side benefit of using practice incentives was the interest generated among staff in monitoring progress and frequently reviewing metrics of their success. Reports similar to the ones in this paper were made available to all practice staff and physicians, so monitoring of enrollments, sessions, and requests was easily accessible.

The introduction of new features can have several benefits for adoption. First, they increase the value of the patient portal, attracting more users. Second, the opportunity to “announce” the new feature to existing users creates additional marketing among patients who might have become inactive, helping them gain more experience using the system.

The increased industry focus on personal health records and patient portals during 2002 to 2009 has probably helped the adoption of PG in the case study practices. Partners “internal” practice incentives are likely to intensify patient portal adoption at Partners in 2010, and federal Meaningful Use requirements are creating even stronger provider incentives for patient portal use.

Study Limitations

The case study approach is more useful for identifying factors influencing patient portal adoption than determining the relative contribution of each factor. Physicians and staff engagement were not assessed in a standardized way, and patient engagement was not directly measured. The effects of trends over time are also harder to isolate, such as general growth in the use of the internet. More rigorous studies are needed to help identify when, for example, marketing and leadership strategies are more effective, and when patient demographics are the most limiting factor.

Conclusion

Variations in practice adoption of a patient portal were observed in this case study of four primary care practices at Partners HealthCare observed over an eight-year period. Differences between patient characteristics, practice leadership focus, staff engagement, feature activation, marketing practices, and incentives were identified as potential factors in adoption and use among patients, providers, and staff.

Practices with the highest levels of adoption among their patients and enthusiasm among their staff and physicians appear to conceptualize the patient portal as a critical component of high quality service. Best practices such as offering patients a fully featured system, engaging all physicians, staff, and patients in marketing activities, using incentives to create focus, and beginning the marketing process prior to “go live” appear to be important factors distinguishing high enrollment practices from others.