Table 3:
Impact of Intervention, 180 days post discharge
| Control Mean (1) | Treatment Mean (2) | Unadjusted Difference (95% CI) (3) | Adjusted Difference (95% CI) (4) | |
|---|---|---|---|---|
| Analysis sample (N=782) | ||||
| Any readmission (%) | 61.70 | 62.34 | 0.64 (−6.17 to 7.46) | 0.82 (−5.97 to 7.61) |
| Number of readmissions | 1.54 | 1.52 | −0.02 (−0.29 to 0.26) | 0.01 (−0.25 to 0.27) |
| 2 or more readmissions (%) | 36.25 | 36.39 | 0.14 (−6.61 to 6.89) | 0.27 (−6.22 to 6.77) |
| Days in hospital | 9.95 | 9.36 | −0.59 (−2.49 to 1.31) | −0.32 (−2.17 to 1.53) |
| Hospital charges ($) | 114,768 | 116,422 | 1,654 (−25,523 to 28,831) | 3,722 (−23,438 to 30,882) |
| Hospital payments received ($) | 17,650 | 18,130 | 480 (−3,613 to 4,573) | 680 (−3,415 to 4,775) |
| Any readmission by subgroup (%) | ||||
| 2 admissions in prior year (N=336) | 52.12 | 52.63 | 0.51 (−10.2 to 11.22) | 0.78 (−10.35 to 11.91) |
| 3 + admissions in prior year (N=446) | 68.75 | 69.82 | 1.07 (−7.51 to 9.65) | 1.27 (−7.38 to 9.92) |
| English preferred language (N=638) | 63.11 | 62.61 | −0.49 (−8.01 to 7.02) | 0.1 (−7.42 to 7.61) |
| Other preferred language (N=144) | 56.25 | 60.94 | 4.69 (−11.58 to 20.96) | 8.49 (−9.08 to 26.06) |
Notes: Data consist of the analysis sample (N=782) and outcomes are measured in the hospital discharge data. Column (1) shows the mean value for each outcome in the control group. Column (2) displays the mean value for each outcome in the treatment group. Column (3) shows the coefficient (and 95% confidence interval in parentheses) on an indicator for treatment group from an ordinary least squares regression of the outcome with no other covariates. Column (4) shows the coefficient (and 95% confidence interval in parentheses) on an indicator for treatment group from an ordinary least squares regression of the outcome with pre-specified covariates. All confidence intervals are calculated using heteroskedasticity-robust standard errors. Pre-specified covariates include the number of admissions 0–6 months before the index admission, the number of admissions 7–12 months before the index admission, and indicators for age (in five-year bins), male, African-American non-Hispanic, Hispanic. The covariates are measured in the hospital discharge data except for race which is from the baseline survey. For three of the outcomes (days in hospital, hospital charges, and hospital payments received), we replace the number of admissions 0–6 months before index admission and 7–12 months before index admission with the values of the dependent variable over those two time periods. The p-value of the primary outcome (any readmission) for the adjusted difference is 0.81.