Table 2.
Characteristics of the study population at each study site
| JHHS |
KPMAS |
KPSC |
||||
|---|---|---|---|---|---|---|
| Total | Residential instability | Total | Residential instability | Total | Residential instability | |
| Study population, n (%) | ||||||
| 47 440 (100)a | 1786 (3.8) | 25 727 (100)b | 2905 (11.3) | 300c | 138 | |
| Age, n (col%) | ||||||
| 18–34 | 4592 (9.7) | 283 (15.8) | 3612 (14.0) | 505 (17.4) | 97 (32.3) | 39 (28.3) |
| 35–44 | 3303 (7.0) | 290 (16.2) | 1858 (7.2) | 342 (11.8) | 57 (19.0) | 34 (24.6) |
| 45–64 | 12 526 (26.4) | 747 (41.8) | 8404 (32.7) | 1217 (41.9) | 95 (31.7) | 52 (37.8) |
| 65–74 | 8732 (18.4) | 186 (10.4%) | 5634 (21.9) | 467 (16.1) | 27 (9.0) | 11 (8.0) |
| 75 | 12 996 (27.4) | 109 (6.1%) | 6219 (24.2) | 374 (12.8) | 24 (8.0) | 2 (1.5) |
| Gender, n (col%) | ||||||
| Male | 25 425 (53.6) | 667 (37.3) | 10 204 (39.7) | 1053 (36.3) | 156 (52.0) | 88 (63.8) |
| Female | 22 013 (46.4) | 1119 (62.6) | 15 523 (60.3) | 1852 (63.7) | 144 (48.0) | 50 (36.2) |
| Race/Ethnicity, n (col%) | ||||||
| Hispanic | 6 (0.01) | 1 (0.05) | 2010 (7.8) | 222 (7.7) | 112 (37.3) | 45 (32.6) |
| Non-Hispanic Black | 15 102 (31.8) | 844 (49.5) | 12 345 (48.0) | 1641 (56.5) | 42 (14.0) | 24 (17.4) |
| Non-Hispanic White | 26 899 (56.7) | 801 (44.8) | 8547 (33.2) | 820 (28.2) | 96 (32.0) | 47 (34.1) |
| Asian/Pacific-Islander | 1802 (3.8) | 12 (0.7) | 2244 (8.7) | 193 (6.6) | 21 (7.0) | 2 (1.5) |
| Other/unknown | 3483 (7.3) | 115 (6.4) | 581 (2.3) | 29 (1.0) | 29 (9.7) | 20 (14.5) |
| Insurance, n (col%) | ||||||
| Medicaid | 31 (0.1) | 3 (0.2) | 4391 (17.1) | 877 (30.2) | 21 (7.0) | 8 (5.8) |
| Medicare | 0 (0) | 0 (0) | 13 154 (51.1) | 1,095 (37.7) | 45 (15.0) | 8 (5.8) |
| Deductible | 702 (1.5) | 12 (0.7) | 2114 (8.2) | 236 (8.1) | 19 (6.3) | 3 (2.2) |
| Standard HMO | 236 (0.5%) | 11 (0.6) | 5756 (22.4) | 640 (22.0) | 89 (29.7) | 16 (11.6) |
| Other | 41 180 (86.8)d | 1589 (88.9) | 312 (1.2) | 57 (2.0) | 126 (42.0)e | 103 (74.5)f |
The JHHS site assessed over 30 EHR questionnaires and flowsheets addressing residential instability and identified 5 relevant ones. Between July 2016 and June 2020 we identified, 1786 patients with positive response and 45 654 patients with negative response to residential instability questions.
The KPMAS site extracted the YCLS survey data from the EHR. Between March 2017 and June 2020 we identified a total of 40 372 YCLS survey responses completed by 25 727 KPMAS adult members, 2905 of whom indicated residential instability. We used YCLS survey responses to generate a binary label of a patient’s needs related to residential instability. We assigned a positive label (1) to patients with a survey response indicating an unmet housing need and a negative label (0) to patients with a survey response indicating no current housing need.
KPSC site randomly selected 150 hospitalized or ED patients between January 2016 and December 2019 with residential instability diagnosis or homeless checklist and 150 without residential instability diagnosis or homeless checklist. Positive cases (138 patients) were those identified as residentially unstable during the chart review process.
Including self-pay.
124 of 126 are non-KP members.
102 of 103 are non-KP member.
EHR: Electronic Health Record, HMO: Health Maintenance Organization, JHHS: Johns Hopkins Health System, KPMAS: Kaiser Permanente Mid-Atlantic States, KPSC: KP Southern California, YCLS: Your Current Life Situation.