Patient and primary care provider experience using a family health history collection, risk stratification, and clinical decision support tool: a type 2 hybrid controlled implementation-effectiveness trial

MeTree^© is a patient-facing FHH computerized collection tool with embedded clinical decision support (CDS) for patients and providers on actionable prevention strategies and education support for collecting FHH. Patients collect their FHH, then enter it into MeTree^© along with other relevant personal history needed to run the integrated risk calculators. MeTree^© then risk-stratifies patients for five diseases (breast, ovarian and colorectal cancer, thrombosis, and hereditary cancer syndromes) and recommends risk-guided prevention strategies endorsed by evidence-based guidelines [15‐24]. CDS is provided in the form of a pedigree and tabular FHH along with individualized reports- one for patients with general comments about their disease risk and points to discuss with their providers, and another for providers outlining personalized action-oriented evidenced-based prevention strategies along with details of the FHH triggers and resources for additional information. Details regarding MeTree^©’s development and validation, including the evidence-based guidelines, risk calculators, programming and validity have been published [25].

We performed a controlled hybrid type 2 implementation-effectiveness clinical trial in three community-based primary care practices in Cone Health System, Greensboro, NC. Hybrid studies are an emerging research tool for combining studies with effectiveness and implementation outcomes; primary outcomes in type 2 studies are effectiveness and secondary outcomes address both effectiveness and implementation [26, 27]. The details of the study design are reported in a previously published protocol paper [28]. The study was IRB approved by all 3 institutions and the DoD. Written and informed consent was obtained from all participants in the study.

MeTree^© was integrated into the work flow of two community-based primary care clinics, while the 3^rd served as a concurrent control for comparison of contemporaneous screening and referral rates. Clinic practices were compensated with a small sum of money for their participation. Integration included educating patients about the importance of FHH and how to collect it, providing a FHH worksheet to facilitate collection, completing MeTree^© at a dedicated clinic kiosk prior to their appointment, and generating CDS output. Patient reports were given to patients immediately and provider reports were integrated into the medical record for use at the patient visit. It was left up to the patient and the provider to choose whether to act, or not, on MeTree^©’s recommendations. A study coordinator was available for assistance.

Invitation letters were mailed to 11,177 patients with upcoming primary care well visits at the two intervention practices between October 15^th 2009 and April 14^th 2012. Children, adoptees, and non-English speakers were excluded. Patients were required to enter their FHH at the kiosk in the clinic and since only one kiosk was available at each clinic, only one person per hour time slot could enroll. See study flow diagram (Figure 1); of the 5,971 patients that were contacted by phone about the study, 4,277 (72%) agreed to participate. Of the 5,971 contacted, 1,694 (28%) declined to participate, 288 (5%) were unable to come one hour prior to their PCP appointment to complete MeTree^©, 2,805 (47%) could not be scheduled due to only one clinic kiosk being available per one hour time slot, and 1,184 patients (20%) completed MeTree^©.

All intervention clinic patients completed a baseline survey derived from the Health Information National Trends Survey [29] at the time of study enrollment. They were surveyed on disease risk perceptions, lifestyle, and their knowledge of health, cancer and genetics. An exit survey, completed after their primary care appointment, assessed their experience with MeTree^© and their discussions with their provider. At 3 months post- MeTree^©, patients re-took the baseline survey and answered questions about MeTree^©’s impact on their health perceptions, cancer screening practices, and discussions with family members. Physicians were interviewed prior to implementation regarding perceived barriers to integration and surveyed at 3 months post-integration regarding their experience with MeTree^©. All surveys were paper-based and self-administered. Data were entered into RedCap by study personnel [30].

Data was analyzed using R statistical software and all hypotheses tests were assessed at a significance level of p < 0.05 [31]. Since the intervention was allocated at the level of the clinic, to assess for the presence of data clustering, we calculated the design effect. The result, 1, indicated the absence of clustering and permitted the use of standard models without adjustment [32]. We used Pearson's chi-square test to analyze the independence of relationships for two categorical variables and ANOVA F-test for one categorical and one numeric variable. Fisher's least significant difference test assessed differences in pairs of levels. Linear regression models and t-tests analyzed the relationship of two numeric variables. When appropriate, numerical outcome variables were analyzed using multivariate analysis with standard linear regression, and categorical variables with logistic linear regression. Acceptability was evaluated based on: age, gender, ethnicity, education level, family size, and percent of family with cancer. Odds ratios for continuous variables, when used, were presented in the following units: age, per one year increase; family size, per one person increase; percent of family with cancer, per one percent increase. For multivariate analyses related to patients’ experience using MeTree^©, a covariate regarding whether patients talked with relatives prior to using MeTree^© was also included. There were variable amounts of missing data. For any particular analysis individuals with missing data were dropped from that analysis. For subjects that were lost to follow-up, reminders to complete the surveys were sent three times.

Characteristics of the 1,184 enrolled patients compared to the general clinic population are presented in Table 1. They entered information on 27,406 relatives.

The user experience is presented in the following areas: ease of use, time, satisfaction, and preparedness (Table 2).

Of the 1184 study participants, only 370/1184 (31.2%) answered the survey question regarding discussions with their provider (Table 3). With the exception of age, these participants are reflective of the population as a whole, with no difference in gender, ethnicity, or likelihood of having received a non-routine recommendation from MeTree^©. Those who were younger were more likely to answer this question (mean age 57.1 vs. 59.6, p-value = 0.001), though the difference in the ages is not clinically significant. Discussions related to each CDS condition are described below.

At 3 months (N = 454, 38.3%) follow-up, patients still had a strongly positive view of their experience, feeling that it was helpful to them and their doctor, made them more aware of their personal risk and family health risk, and changed how they think about their health (Table 4). In addition, they would “recommend MeTree^© to others”. In multivariate models positive impressions were not associated with demographic factors or whether patients received a non-routine recommendation from MeTree^©. Regarding what they “wished they knew before using MeTree^©, 55/214 (25.7%) wished they had a greater knowledge of their FHH. Only 10 had questions regarding MeTree^©’s CDS recommendations or their provider discussion.

There were 14 providers at the intervention sites, 9 physicians and one nurse practitioner at one and 4 physicians at the other. Demographics are age range 29–65 and 50% female. During pre-implementation interviews, providers indicated skepticism about the potential benefit of MeTree^© and concern that it would impact workflow. In particular they reported that: 1) they felt they were already collecting high quality FHHs; 2) integration of MeTree^© would not lead to clinically important changes in their patients’ health care plans; 3) patients would redirect discussions during the appointment towards MeTree^© recommendations and away from the providers’ high priority topics; and 4) MeTree^© would negatively impact work flow either when addressing questions or implementing recommendations.

On the 3 month post-integration survey providers were overwhelmingly supportive of MeTree^©. In particular, they felt MeTree^©: raised awareness about the importance of FHH and risk stratification, improved the way they practiced medicine, and made their daily routine easier; and they reported actively recommending MeTree^© to their peers (Table 5). In addition, MeTree^© made them more aware of the importance of genetic counselors and genetic counseling (81% wanted to establish relationships with genetic counselors).