Table 2.
Regression Results of the Association Between Health Information Technology, Accountable Care Organization, Social Vulnerability Index, Alzheimer's Disease and Related Dementia, and Total Medicare Payment
| MODEL 1: HIT, SVI, ADRD | COEFFICIENT | 95% CI | p | |
|---|---|---|---|---|
| Telehealth-post discharge | 0.069 | 0.066 | 0.072 | <0.001 |
| ACO | 0.112 | 0.109 | 0.114 | <0.001 |
| ADRD | 0.150 | 0.147 | 0.152 | <0.001 |
| SVI Q1: 0–28.7% | Ref. | |||
| SVI Q2: 28.7–52.5% | −0.005 | −0.008 | −0.001 | 0.006 |
| SVI Q3: 52.5–69.2% | 0.048 | 0.044 | 0.051 | <0.001 |
| SVI Q4: >69.2% (the most vulnerable area) | 0.076 | 0.072 | 0.080 | <0.001 |
| MODEL 2: HIT × SVI × ADRD | COEFFICIENT | 95% CI | p | |
|---|---|---|---|---|
| Telehealth-post discharge |
0.048 |
0.044 |
0.051 |
<0.001 |
| SVI Q4 (the most vulnerable area) vs. the rest |
0.024 |
0.020 |
0.028 |
<0.001 |
| ADRD |
0.143 |
0.140 |
0.146 |
<0.001 |
| Telehealth-post discharge × SVI Q4 |
0.086 |
0.080 |
0.092 |
<0.001 |
| Telehealth-post discharge × ADRD |
−0.026 |
−0.032 |
−0.020 |
<0.001 |
| SVI Q4 × ADRD |
0.044 |
0.039 |
0.050 |
<0.001 |
| Telehealth-post discharge × SVI Q4 × ADRD | −0.043 | −0.054 | −0.032 | <0.001 |
| MODEL 3: ACO × SVI × ADRD | COEFFICIENT | 95% CI | p | |
|---|---|---|---|---|
| ACO affiliation |
0.132 |
0.129 |
0.135 |
<0.001 |
| SVI Q4 (the most vulnerable area) vs. the rest |
0.064 |
0.060 |
0.068 |
<0.001 |
| ADRD |
0.163 |
0.159 |
0.166 |
<0.001 |
| ACO × SVI Q4 |
−0.020 |
−0.025 |
−0.014 |
<0.001 |
| ACO × ADRD |
−0.064 |
−0.070 |
−0.058 |
<0.001 |
| SVI Q4 × ADRD |
0.030 |
0.024 |
0.036 |
<0.001 |
| ACO × SVI Q4 × ADRD | −0.012 | −0.022 | −0.002 | 0.022 |
| MODEL 4: HIT AND ACO × SVI × ADRD | COEFFICIENT | 95% CI | p | |
|---|---|---|---|---|
| Telehealth-post discharge and ACO |
0.131 |
0.126 |
0.135 |
<0.001 |
| SVI Q4 (the most vulnerable area) vs. the rest |
0.051 |
0.047 |
0.054 |
<0.001 |
| Telehealth-post discharge and ACO × SVI Q4 |
0.144 |
0.141 |
0.147 |
<0.001 |
| ADRD |
0.061 |
0.051 |
0.072 |
<0.001 |
| Telehealth-post discharge and ACO × ADRD |
−0.063 |
−0.072 |
−0.054 |
<0.001 |
| SVI Q4 × ADRD |
0.032 |
0.027 |
0.037 |
<0.001 |
| Telehealth-post discharge and ACO × SVI Q4 × ADRD | −0.043 | −0.060 | −0.025 | <0.001 |
Notes: We used the generalized linear model with log link and gamma variance distribution with state fixed effect. All other covariates included race, age, sex, and health indicators. Health indicators included common comorbidities of ADRD, such as heart disease, diabetes, hyperlipidemia, hypertension, and asthma. Covariates at the hospital level included teaching status, type of controls (for-profit, not-for-profit, and government), and bed size. Area-level variables included the rurality index using the Core-Based Statistical Areas, the U.S. Office of Management and Budget. Two-way interaction effects were tested and computed post estimation to test the significance and robustness of the findings. Particularly, the effect of the interacted variables was calculated through the partial derivative and the first difference. The ginteff command was used. ∂ Telehealth-Post Discharge × ∂ SVI Q4 × ∂ ADRD = −$1,431.5 (95% CI: −$1,905.4 to −$957.6); ∂ ACO × ∂ SVI Q4 × ∂ ADRD = −$685.6 (95% CI: −$1,107.3 to −$263.8); and ∂ Telehealth-Post Discharge and ACO × ∂ SVI Q4 × ∂ ADRD = −$1,766.2 (95% CI: −$2,576.4 to −$976).
CI, confidence interval; HIT, Health Information Technology.