Impact of a Wellness Portal on the Delivery of Patient-Centered Preventive Care

Development and Pilot Testing of the Wellness Portal

We first assembled a Portal Advisory Committee, which included 3 clinicians, 2 office staff, and 6 patients. Based on input from 3 committee meetings, a prototype personal Wellness Portal website was developed over a 6-month period. The prototype design incorporated management of preventive services history, tracking of personal risk factors and preferences, and generation of a tailored wellness plan. The Advisory Committee suggested a set of additional capabilities that included tracking and charting vital signs and laboratory test results, history of medical encounters, a symptom diary, management of medication lists, an immunization record for children, secure messaging with practices, and generation of an interoperable personal health record in the continuity of care record format.

We incorporated and tested these features over a 6-month period in a convenience sample of two practices in the Oklahoma Physicians Resource/Research Network where elements of a preventive care delivery system had already been implemented. Survey data on satisfaction with the Portal were analyzed to describe portal utility on four-point Likert scales. Open-ended questions were also analyzed to describe general impressions of the Wellness Portal use. Information regarding features that facilitated or hindered the improvement of preventive care and recommendations for further Portal enhancements were compiled.

Cluster Randomized Controlled Trial

Following pilot testing, eight Oklahoma Physicians Resource/Research Network clinicians practices were selected, recruited, and enrolled into a cluster randomized controlled trial (c-RCT). Pairs of clinician practices were matched on location and practice type (urban, suburban, or rural and solo, small, or midsize) and then randomized within pairs to intervention and control arms, as proposed by Imai et al.⁸ Selection criteria included clinician experience with the existing patient registry (PSRS), inclusion of both young children (age <6) and adults (age >50) in the patient panel, and no previous experience with the Wellness Portal.

Two of the intervention sites were small, rural physician practices in western and central Oklahoma (each included a male physician and a female nurse practitioner), whereas the other two represented a typical, medium-size, suburban family practice environment including several male physicians, a female nurse practitioner, and a physician assistant. Correspondingly, the control group consisted of a matched sample of small and medium-size rural and suburban practices in northeast, southwest and central Oklahoma, serving comparable patient populations in terms of demographics, socioeconomic status, and payer mix.

After practices were randomized, patients in each practice were notified about the availability of the study through handouts, flyers, and verbal communication. Interested patients were enrolled by a research assistant (RA) in the waiting room of the clinicians' offices until 70 patients from each practice were recruited (N = 560). At the baseline visit, the RA asked patients to read and sign a patient consent form and complete baseline surveys. Patients were included in the study if they had been seen at least twice by the enrolled clinician in the last 12 months, were either children (≤6 years older) or between 40 and 75 years old (women) or 50 and 75 years old (men), could understand and respond in English, and had a basic level of computer skills that included the ability to navigate a standard, consumer-oriented web site, use the keyboard and the mouse to interact with a browser, and understand/respond to web content phrased at 6th grade level. A sample size of approximately 400 participants (50 patients per practice) gave us 80% power to detect differences between intervention and control groups. The CONSORT diagram of the Wellness Portal trial is shown in Figure 2.

• Download figure
• Open in new tab

Figure 2.

CONSORT diagram of the Wellness Portal randomized controlled trial. *Participant did not complete or send the survey back to investigators and/or was unreachable. ^†Participant signed up as a spouse of another participant and this became evident to investigators only later.

Implementation Support By Practice Enhancement Assistants

Portal implementation was facilitated by practice enhancement assistants (PEAs; trained practice facilitators) in intervention practices during the entire study period. PEAs helped practices identify study participants, map the existing preventive care workflow, tailor portal implementation in the intervention arm, recruit patients, and support portal users with basic technical assistance. Typically, portal users received a link to the portal registration page via printed study materials or over e-mail. They could sign up for the portal from home, work, or another Internet-enabled location, including portal kiosks made available in the office. At the first log on, the portal would offer the following menu of options and resources: maintain demographics profile, update preventive services history, assess personal risk factors, review personalized wellness plan (recommendations), track and chart vitals and lab results, work on symptom diary, update medical encounters history, manage medication and problem list, and download a personal health record. Patients were also encouraged to print their wellness plan and discuss the plan with their doctor at their next office visit. Clinicians then worked with PEAs to implement systems in their practices to respond to preventive care recommendations (eg, improve their referral approach, incorporate health behavior counseling, and optimize coding for preventive services).

Patient Surveys

Patients were surveyed in their doctor's office via a self-administered paper questionnaire at two time points: at baseline and 12 months later. Surveys included items describing patient experience with preventive services, satisfaction with the system of care, perception of patient-centeredness of care, patient activation, and patient empowerment. Patients in both arms of the study completed the surveys that were identical except that patients in the portal arm were asked to assess their satisfaction with the portal tool after the intervention. Patients in intervention practices were asked at enrollment to review and update their demographics, risk factor information, allergy and contraindication profile, and personal preferences through the Wellness Portal before each visit to their primary care clinician. Clinicians and their staff were asked to either use the linked practice portal (PSRS) to generate care recommendations or to ask patients to print a list of recommendations from their portal account before their next visit. Patients in control practices were not given access to the portal and they did not receive personalized recommendations or a wellness plan.

Patient Experience and Patient-Centeredness of Care

To measure patient-centeredness of care, we adapted the Consumer Assessment of Health Care Providers and Systems (CAHPS) survey developed and validated by the Agency for Health Care Research and Quality ambulatory CAHPS Initiative.⁹ We created a composite score from CAHPS survey questions 1 to 8, plus 10 and 11 to measure the patient-centeredness of care as it relates to prevention. The 10 questions were coded as binary (yes/no) values describing patient–clinician interactions as they relate to preventive care: assisting patients to make changes in their health habits, helping patients to make changes, encouraging patients to address health concerns, healthy diet, physical activity, worry and stress, depression, pros and cons of care options, and clinician guidance for appropriate care choices. In addition to the patient-centeredness composite score, an item measuring patient perception of provider knowledge of the patient's medical history was also included. Differences in composite patient-centeredness scores before and after intervention were calculated for each patient and the differences were compared between control and intervention groups.

Patient Activation

Patient activation was assessed using an adapted Patient Activation Measure (PAM) questionnaire.¹⁰ The shorter form of the PAM instrument included 13 items in 3 subscales, measuring patient knowledge, confidence, and skills that demonstrate self-efficacy.¹¹ These measures allowed us to determine the distribution of patients within the four stages of activation identified by the PAM instrument.¹¹

Patient Demographics and Health Status

Patient age, sex, race/ethnicity, level of education, marital status, insurance coverage information, and employment status were collected via medical record reviews and baseline surveys. Patients' baseline health status was determined by documenting chronic health conditions and the number and type of office visits during a 12-month period before and during the study. We used the patients' comorbid conditions to calculate the Charlson comorbidity index¹² as a measure of the severity of illness.

Provision of Preventive Services

Medical records of patients (paper and electronic) were reviewed in the practice to determine the number and type of selected preventive services received before and during the 12-month study period. We also documented the number and type of risk factors, contraindications, and personal care preferences that patients entered into the portal database. We used medical history to populate the PSRS risk engine that generated personalized preventive services recommendations based on the US Preventive Services Task Force, Advisory Committee on Immunization Practices, and American Academy of Family Physicians guidelines.

Preventive Care Delivery System

We administered an adapted version of the Assessment of Chronic Illness Care survey¹³ in each participating clinician's office before and after the study to evaluate the level and nature of practice system improvements, specific to the delivery of preventive services, made by clinicians and their staff. The Assessment of Chronic Illness Care addresses several chronic care model domains that include community linkages, organization of care delivery, self-management support, decision support, clinical information systems, activated patients, and prepared, proactive practice teams.

Delivery of Appropriate Services At the Right Time

We examined the delivery of preventive services under various conditions where recommendations differ based on individual patient risk factors. For example, pneumococcal vaccination is recommended for patients younger than 65 years old (2–64 years) if they have chronic cardiovascular disease, chronic pulmonary disease, or diabetes mellitus. In the absence of these chronic conditions, patients are usually eligible for pneumococcal vaccination at 65 years of age or older. Review of medical records and portal records allowed us to assess the individualization of recommendations as a result of patient input via the Wellness Portal and documentation of the patient's risk profile. We created a binary variable to describe appropriate delivery of services. If the appropriate service was delivered at the right time, the variable was assigned the value of 1, whereas if a service should have been delivered and was not delivered, the variable was assigned the value of 0.

Patient Utilization of the Wellness Portal

We created two variables associated with portal use during the 12-month period: (1) a dichotomous variable where the value of “use” was assigned as 1 if the patient accessed the portal and reviewed his/her recommendations at least once; and (2) a continuous variable documenting the number of times patients have accessed the portal. Categorical variables were created to describe the type of information patients provided to clinicians through the portal that contributed to the individualization of the wellness plan. Utilization of the portal was tracked via electronic security logs that were automatically created during each user session and contained detailed information about user behavior and information exchange between users and the portal database.

Analytic Strategy

We analyzed quantitative data with the help of SAS 9.1.3 software (SAS Institute, Cary, NC). All quantitative analyses were conducted in an intention-to-treat framework, adjusting for the clustering of patients within clinician practices. We conducted difference-in-differences analyses¹⁴ to determine the impact of the portal on the patient-centeredness of care. These analyses compared the magnitude of change in patient-centeredness scores between groups and time points. We also examined missed opportunities for care to gauge the impact of the portal on the delivery of preventive services. In this analysis, we compared the number of preventive services received per person between groups before and after the study using t tests, considering patient clustering.