Table 6.
Decomposition of Gains in Survival Over Time
| Condition | (1) | (2) | (3) | (4) | (5) | (6) |
|---|---|---|---|---|---|---|
| Contributions in Pctage
Points |
Contributions as Share of
Total |
|||||
| AMI | HF | Pneu | AMI | HF | Pneu | |
| Total Change in Wtd Survival | 0.0389 | 0.0092 | 0.0316 | 1.00 | 1.00 | 1.00 |
| Within | 0.0298 | 0.0076 | 0.0297 | 0.77 | 0.82 | 0.94 |
| Between | 0.0000 | −0.0004 | −0.0004 | 0.00 | −0.05 | −0.01 |
| Cross | 0.0062 | 0.0015 | 0.0015 | 0.16 | 0.16 | 0.05 |
| Entry | 0.0021 | 0.0006 | 0.0010 | 0.05 | 0.07 | 0.03 |
| Exit | −0.0007 | 0.0000 | 0.0002 | −0.02 | 0.01 | 0.01 |
This table decomposes the gains in risk-adjusted survival for the emergent conditions over our full sample window (between 1996 and 2008) using the decomposition shown in equation (7). Columns 1-3 show the contribution of each component to gains in survival in percentage points (e.g. a value of 0.1 is 10 percentage points). Columns 4-6 show the share of total gains that can be attributed to each component. The exit component enters negatively, so a negative value indicates that exit accounts for a gain in survival.
The decomposition is performed for each market, then averaged together weighted by the market's size in 1996. Risk-adjusted survival is calculated from a regression of survival on hospital fixed effects and patient risk-adjusters. A separate regression is run for each of the year groups 1994-1996 (yielding the 1996 survival rates) and 2006-2008 (yielding the 2008 survival rates).