Table 1.
Participant baseline characteristics for those included versus excluded from the analytical cohort
| Included | Excluded | P value | |
|---|---|---|---|
| Variable | Median (IQR) or N (%) | ||
| Subjects (n) | 6,042 | 208,164 | |
| Age | 56.69 (41.45–67.62) | 56.91 (40.96–68.24) | 0.373 |
| Race | <0.001 | ||
| Black | 336 (5.6) | 45,661 (21.9) | |
| Other | 309 (5.1) | 11,112 (5.3) | |
| White | 5,072 (83.9) | 108,141 (51.9) | |
| Sex at birth | <0.001 | ||
| Female | 4,379 (72.5) | 126,159 (60.6) | |
| Male | 1,579 (26.1) | 77,969 (37.5) | |
| Ethnicity | <0.001 | ||
| Hispanic or Latino | 376 (6.2) | 41,638 (20.0) | |
| Not Hispanic or Latino | 5,590 (92.5) | 160,368 (77.0) | |
| Education | <0.001 | ||
| College degree | 4,317 (71.4) | 82,407 (39.6) | |
| Some college | 1,346 (22.3) | 53,973 (25.9) | |
| No college | 356 (5.9) | 66,925 (32.2) | |
| BMI | 28.10 (24.32–32.85) | 28.80 (24.70–34.10) | <0.001 |
| Baseline conditions | |||
| CAD | 170 (2.8) | 14,684 (7.1) | <0.001 |
| Cancer | 1,429 (23.7) | 58,050 (27.9) | <0.001 |
| Smoking (100 cigarettes) | |||
| >100 cigarettes | 1,932 (32.0) | 84,466 (40.6) | <0.001 |
| Alcohol | |||
| ≥1 drink | 5,846 (96.8) | 177,735 (85.4) | <0.001 |
| Fitbit variables | |||
| Duration (years) | 3.99 (2.15–5.58) | ||
| Average daily steps | 7,731.30 (5,866.84–9,826.85) | ||
Data are median (IQR) for continuous variables and frequency (percentage) for categorical variables.
Participants were excluded because of not having EHR and Fitbit data or valid Fitbit data for at least 6 months. For included participants, missingness for each variable was as follows: Race, 325 (5.4%); Sex, 84 (1.4%); Ethnicity, 76 (1.3%); BMI, 2,644 (43.8%); Smoking, 108 (1.8%); Education, 23 (0.4%); Alcohol, 31 (0.5%). For excluded participants, missingness for each variable was as follows: Race, 43,250 (20.8%); Sex, 4,036 (1.9%); Ethnicity, 6,158 (3.0%); BMI, 51,496 (24.7%); Smoking, 6,094 (2.9%); Education, 4,859 (2.3%); Alcohol, 6,477 (3.1%). For included participants, BMI had a high amount of missingness because the measurement must have occurred before the Fitbit monitoring period. Mann–Whitney U and chi-squared tests for continuous and categorical variables, respectively, were used to compare these clinical characteristics for the participants who were excluded versus included in the analytical dataset for this study.