Feasibility and Reach of Trial Recruitment
The eligibility list for the feasibility trial included 2,146 patients. Over 6 weeks, the clinical data manager sent names of 442 patients to 34 primary care clinicians for pre-review; clinicians asked that only 16 (3.6%) of these patients not be contacted. A single outreach Mychart message was sent to each of the remaining 426 patients, of whom 315 (73.9%) opened the message within 4 weeks, and 80 (18.8%) affirmed interest in the weight loss goal (Fig. 1).
<<Figure 1 about here>>
Demographic characteristics of the 80 participants (63% women; 29% non-Hispanic Black; 14% Hispanic/Latinx; 30% age ≥ 65) approximated the characteristics of the overall target population, with slight over-representation of women and those identifying as Hispanic/Latinx, non-Hispanic Black, or non-Hispanic Asian (Table 1). In the setting of relatively small sample sizes, trial patients randomized to CLS were more often than BLS patients to be older than 45, male, and Hispanic/Latinx. CLS patients were also more likely to have prediabetes and a higher Charlson comorbidity score.
Table 1. Baseline Characteristics of Study Patients
Baseline Characteristics*
|
Eligible, Approached
(N = 426)
|
Affirmed Weight Goal, Enrolled
(N = 80)
|
Basic Lifestyle Support (BLS)
(N = 38)
|
Customized
Lifestyle Support (CLS)
(N = 42)
|
Completed Interview†
(N = 15)
|
Age Group, N (%)
|
|
|
|
|
|
Age 18-44
|
61 (14.3%)
|
12 (15.0%)
|
8 (21.1%)
|
4 (9.5%)
|
0 (0.0%)
|
Age 45-64
|
230 (54.0%)
|
44 (55.0%)
|
18 (47.4%)
|
26 (61.9%)
|
12 (80.0%)
|
Age 65+
|
135 (31.7%)
|
24 (30.0%)
|
12 (31.6%)
|
12 (28.6%)
|
3 (20.0%)
|
|
|
|
|
|
|
Gender, N (%)
|
|
|
|
|
|
Female
|
234 (54.9%)
|
50 (62.5%)
|
26 (68.4%)
|
24 (57.1%)
|
10 (66.7%)
|
Male
|
192 (45.1%)
|
30 (37.5%)
|
12 (31.6%)
|
18 (42.9%)
|
5 (32.3%)
|
|
|
|
|
|
|
Race/Ethnicity, N (%)
|
|
|
|
|
|
Hispanic/Latinx
|
44 (10.3%)
|
11 (13.8%)
|
3 (7.9%)
|
8 (19.1%)
|
4 (26.7%)
|
Non-Hispanic Black
|
113 (26.5%)
|
23 (28.8%)
|
12 (31.6%)
|
11 (26.2%)
|
6 (40.0%)
|
Non-Hispanic Asian
|
14 (3.3%)
|
6 (7.5%)
|
2 (5.3%)
|
4 (9.5%)
|
0 (0.0%)
|
Non-Hispanic White
|
217 (50.9%)
|
34 (42.5%)
|
16 (42.1%)
|
18 (42.9%)
|
4 (26.7%)
|
Other
|
38 (8.9%)
|
6 (7.5%)
|
5 (13.2%)
|
1 (2.4%)
|
1 (6.7%)
|
|
|
|
|
|
|
Qualifying Diagnoses, N (%)
|
|
|
|
|
Prediabetes
|
128 (30.1%)
|
30 (37.5%)
|
19 (50.0%)
|
11 (26.2%)
|
6 (40.0%)
|
Diabetes
|
138 (32.4%)
|
19 (23.8%)
|
6 (15.8%)
|
13 (31.0%)
|
5 (33.3%)
|
High Blood Pressure
|
324 (76.1%)
|
62 (77.5%)
|
29 (76.3%)
|
33 (78.6%)
|
13 (86.7%)
|
Dyslipidemia
|
341 (80.1%)
|
61 (76.3%)
|
27 (71.1%)
|
34 (81.0%)
|
11 (73.3%)
|
|
|
|
|
|
|
Charlson Score‡, Mean (SD)
|
1.34 (1.79)
|
1.33 (1.97)
|
1.13 (1.70)
|
1.50 (2.19)
|
1.13 (1.46)
|
|
|
|
|
|
|
Baseline Risk Factor Levels, Mean (SD)b
|
|
|
Systèm International (SI) Units
|
|
|
Mean Weight, kg
|
95.0 (15.8)
|
94.3 (17.2)
|
95.1 (18.4)
|
93.6 (16.2)
|
95.6 (18.5)
|
Mean BMI, kg/m2
|
33.2 (5.6)
|
33.6 (6.1)
|
34.1 (7.1)
|
33.1 (4.9)
|
34.7 (8.5)
|
Hemoglobin A1c, mmol/mol
|
44.5 (11.5)
|
43.8 (12.6)
|
42.4 (9.8)
|
45.2 (14.8)
|
42.5 (6.7)
|
Systolic Blood Pressure, mmHg
|
131 (15)
|
135 (18)
|
137 (18)
|
133 (18)
|
135 (20)
|
Total Cholesterol, mmol/L
|
4.69 (1.15)
|
4.79 (1.17)
|
4.74 (1.06)
|
4.84 (1.27)
|
5.20 (1.18)
|
HDL Cholesterol, mmol/L
|
1.33 (0.36)
|
1.42 (0.35)
|
1.45 (0.40)
|
1.39 (0.31)
|
1.38 (0.30)
|
Non-HDL Cholesterol, mmol/L
|
3.36 (1.07)
|
3.37 (1.06)
|
3.29 (0.91)
|
3.45 (1.18)
|
3.82 (1.19)
|
|
|
|
|
|
|
Conventional Units
|
|
|
|
|
|
Mean Weight, lbs
|
209.0 (34.8)
|
207.5 (37.8)
|
209.2 (40.5)
|
205.9 (35.6)
|
210.4 (40.8)
|
Hemoglobin A1c, %
|
6.2 (1.1)
|
6.2 (1.2)
|
6.0 (0.9)
|
6.3 (1.4)
|
6.0 (0.6)
|
Total Cholesterol, mg/dL
|
181.3 (44.3)
|
185.4 (45.3)
|
183.2 (41.0)
|
187.3 (49.2)
|
201.1 (45.4)
|
HDL Cholesterol, mg/dL
|
51.3 (14.1)
|
54.9 (13.7)
|
56.1 (15.5)
|
53.8 (12.0)
|
53.4 (11.7)
|
Non-HDL Cholesterol, mg/dL
|
130.0 (41.4)
|
130.4 (40.9)
|
127.1 (35.2)
|
133.5 (45.6)
|
147.8 (45.9)
|
* Gender and race/ethnicity based upon self-reported values stored in each patient’s electronic chart; weight, blood pressure, hemoglobin A1c, and cholesterol values based upon last recorded value within the past 12 months in each patient’s electronic chart; BMI is based on the last recorded weight and height in each patient’s electronic chart
† 30 of the 80 trial participants were randomly selected and approached for interviews to recruit 15 respondents
‡ Charlson comorbidity score is a weighted score based on the number of comorbid diagnoses 29
b values are based on electronic health records; 100% of patients had a weight and systolic blood pressure value in the past 12 months; A1c values were available for 338, 67, 33, 34, & 13 and cholesterol values were available for 418, 78, 37, 41, and 14 patients in the eligible, enrolled, BLS, CLS, and interviewed groups, respectively
Engagement in Intervention Components
All but 3 (3.8%) patients were able to initialize their eScale and register a baseline weight. Over the 12-week MyChart messaging phase of the intervention, 11 (14%) of participants did not use their scales for self-weighing, 7 (9%) participated in weighing in some but not all weeks, and 62 (78%) engaged in self-weighing during all 12 weeks. Self-weighing daily (5 to 7 days per week) was performed by 26 (33%) of patients during all 12 weeks; more patients assigned to CLS (customized weekly messaging) performed daily self-weighing in all weeks (43% versus 21% of those offered BLS).
Overall, 36 (45%) patients (37% of the BLS arm and 52% of the CLS arm) enrolled in one of the two referral-based coaching platforms offered by the fitness industry partner; 22 (27.5%) elected the face-to-face option and 14 (17.5%) elected fully virtual access. Patients enrolled in these services between 15 and 77 days after receiving the initial outreach MyChart message. Access to fitness facilities (i.e., the face-to-face delivery option) was disrupted by a coronavirus pandemic stay-at-home order beginning the third week of March 2020. At that time, patients had been enrolled in the referral-based services for 19 to 75 days; the 22 patients in face-to-face support had completed an average of 3 (range 1 to 6) individual coaching sessions and attended a partnering fitness facility 0 to 24 times; the 14 patients receiving “virtual” support had completed an average of 4 (range 2 to 5) individual coaching sessions prior to the pandemic but were able to complete an average of 3 additional (range 0 to 5) coaching sessions after the pandemic began.
Among patients randomized to CLS, 41 (98%) received at least one step-up telephonic nurse coaching call; 19 (45%) received ≥ 5 nurse calls (maximum 11 calls received) over the 12-week intensive support phase.
Weight loss feasibility
Despite lower than usual office visit rates during the early phase of the coronavirus pandemic, 44 (55%) of feasibility trial patients had an office weight measure recorded in the EHR at 6 ± 3 months from their trial enrollment. Among the remaining 36 patients, 26 (72%) had a weight recorded at 12 ± 3 months, and of the remaining 10 patients, 7 (70%) had an eScale weight received at 6 ± 3 months. Thus, our pragmatic data capture strategy enabled estimation of a 6-month weight change for 77 (96%) of all trial participants. The average time between trial enrollment and the weight used for 6-month data analysis was 184.5 days. Overall, 62% of participants had lower weight at follow up than at baseline; 15.0% exhibited a weight loss of ≥ 5% of baseline weight, with no statistically significant difference between CLS or BLS arms (p = 0.85).
Using eScale data and assuming that the 14% of patients who did not self-weigh also did not lose weight, 28 (35%) of all participants achieved ≥ 10 lbs of weight loss (i.e., the initial goal recommended by the MyChart message) and 24 (30%) achieved ≥ 5% of weight loss at some point during the 12-week automated messaging phase. Thus, half of patients who achieved a 5% weight loss during the 12 weeks of active support maintained the loss at 6 months.
Intervention Delivery Costs
Direct medical costs associated with offering each intervention component are summarized in Additional File Table 3. Mean health system costs to offer the intervention as delivered during the feasibility trial were estimated at $335 per person over the first 6 months ($284 for BLS patients; $382 for CLS patients) and $150 per person over the second 6 months ($124 for BLS; $174 for CLS patients).
Patient Perceptions about Acceptability of Practice Components
Among the 80 feasibility trial participants, 30 were randomly selected and approached before 15 agreed to complete the telephone interview; 11 were unreachable, 5 declined participation, 2 asked to be re-contacted at a future time. Respondents included 5 men and 10 women, with demographic and clinical characteristics generally representative of the overall target population (Table 1). Patient perceptions are summarized below; representative quotes and design implications are organized in Additional File Table 2.
Most patients described the Mychart messages and website resources as generally easy to understand and navigate. However, some reported feeling frustrated or less engaged by not having a clear understanding upfront for how the different intervention components were coordinated or should be utilized. Trust in the person or organization providing lifestyle support services emerged as an important theme. Importantly, some patients expressed being distrustful when offered resources for “free” from a commercial partner, unless it was clear that their doctor or the health system were recommending and coordinating the activities. Others indicated that they had not initially realized they could access dietary and physical activity coaching at the partnering fitness organization and were unsure how to engage with the practice nurse if they had problems.
Nearly all patients reported that an initial weight loss goal of about 10 pounds in 10 weeks was appropriate and achievable. Some expressed interest in larger weight loss goals and access to services beyond 10 to 12 weeks. Many patients reported that their weight loss goal was motivated by other health or social goals, such as reducing the need for more pills to treat high blood pressure.
Several patients highlighted the critical importance of supportive accountability in successfully changing behavior. They emphasized the practice nurse, registered dietitian, and physical trainers as key sources of supportive accountability, more so than the automated messages. Patients expressed strengths and limitations of the automated messages, highlighting that the messages were a helpful reminder or source of encouragement when making progress but were insufficient or even demoralizing when not making progress.
Many patients viewed self-weighing as an additional source of accountability. Some expressed interest in more elaborate electronic scales or integration with other technologies, such as smartphone apps that provide immediate and customized support. Other patients indicated that they preferred support coming from a member of the healthcare or fitness center team. Generally, patients expressed frustration when they viewed eScales (and self-weighing) as a direct source of support, rather than as a means for healthcare providers and coaches to track and customize their support. Importantly, some viewed daily weighing as a source of stress and felt that no matter how closely they followed diet and physical activity plans, their daily weight remained unpredictable. Several patients highlighted failure expectations stemming from negative experiences with prior weight loss attempts. Those patients recommended a greater focus on recognizing and addressing failure expectations, as well as de-emphasizing a focus on the daily weight readings as a sign of success or failure, while offering more immediate support beyond automated messaging. Relatedly, some patients highlighted how emotions have a larger influence on dietary behaviors than reminders to eat healthy. Those patients recommended greater focus on acknowledging and addressing emotions both as drivers and consequences of attempts to lose weight and change diet.
Finally, all patients indicated that the abrupt start of the coronavirus pandemic presented a profound barrier to healthy behaviors and weight loss. Many indicated frustration with becoming more sedentary, adopting ‘stress eating’ behaviors, and gaining weight. The pandemic restrictions eliminated access to fitness facilities, which added to frustration.