Skip to main content
. 2017 Mar 13;53(2):1110–1136. doi: 10.1111/1475-6773.12670

Table 2.

Unadjusted Frequency of Social Factors and Association with Readmission by Data Source Used in Identification (N = 93,606 Admissions)

Characteristic ICD‐9 Diagnoses ICD‐9 Plus Structured EHR Data Physician Notes All Sources (ICD‐9 + EHR data + Physician Notes)
No. of Patients (%) Readmission Rate, % Odds Ratio (95% CI)§ No. of Patients (%) Readmission Rate, % Odds Ratio (95% CI)§ No. of Patients (%) Readmission Rate, % Odds Ratio (95% CI)§ No. of Patients (%) Readmission Rate, % Odds Ratio (95% CI)§
Tobacco Use 11,359 (12) 20.8 1.37* (1.32, 1.41) 13,791 (15) 20.3 1.33* (1.27, 1.39) 23,727 (25) 21.9 1.466* (1.41,1.52) 27,763 (30) 21.0 1.39* (1.34,1.44)
Alcohol Abuse 6,899 (7) 20.4 1.34* (1.26, 1.42) 7,461 (8) 20.2 1.317* (1.24, 1.40) 8,790 (9) 20.9 1.377* (1.31, 1.45) 11,070 (12) 20.0 1.30* (1.24 1.37)
Drug Abuse 5,952 (6) 25.3 1.77* (1.67, 1.88) 6,042 (6) 24.3 1.67* (1.67, 1.88) 8,959 (10) 25.4 1.78* (1.69, 1.87) 11,272 (12) 25.6 1.79* (1.71, 1.87)
Depression 18,615 (20) 24.0 1.65* (1.58, 1.71) 20,469 (22) 23.6 1.61* (1.55, 1.67) 20,358 (22) 24.5 1.69* (1.62, 1.76) 26,904 (29) 23.9 1.64* (1.59, 1.70)
Housing Instability 1,298 (1) 29.9 2.22* (1.97, 2.49) 1,309 (1) 29.9 2.22* (1.98, 2.50) 763 (1) 34.0 2.69* (2.32, 3.12) 1,551 (2) 29.4 2.17* (1.95, 2.42)
Fall Risk 2,209 (2) 18.3 1.17** (1.06, 1.30) 2,302 (2) 18.7 1.20* (1.08, 1.32) 23,737 (25) 21.8 1.45* (1.40, 1.50) 24,261 (26) 21.6 1.44* (1.39, 1.50)
Poor Social Support 342 (0) 14.2 0.86 (0.64, 1.14) 342 (0) 14.0 0.85 (0.64, 1.14) 14,872 (16) 21.3 1.41* (1.36, 1.48) 15,004 (16) 21.3 1.41* (1.35, 1.47)

*p < .001, **p < .01.

EHR refers specifically to problem lists (diagnoses) and health maintenance registries in the EHR.

All sources include patients identified through ICD‐9 diagnoses, problem lists and health maintenance registries in the EHR, and text analysis of physician discharge summary notes.

§Unadjusted OR and 95% confidence intervals (CIs) were computed from series of logistic regressions of a readmission indicator on an indicator variable for a patient social factor that was determined based on a data source. Each social factor and data source represents one unadjusted logistic model.