Table 4.

Comparing Rates of Discharge by Hospital Type, Location, and State^*

	Discharge Rate Ratio All Stays (LOS Analysis)		Discharge Rate Ratio Only Stays >3 days (CLOS Analysis)
Items for Comparison	Ratio	P-Value	Ratio	P-Value
Pennsylvania vs. New York:
Pittsburgh/Philadelphia children's   vs. NY City children's	1.20	<0.0001	1.05	0.30
Pittsburgh/Philadelphia general   vs. NY City general	1.26	<0.0001	1.00	0.98
Outside Pittsburgh/Philadelphia general   vs. outside NY general	1.14	<0.0001	1.04	0.38
General vs. Children's:
NY City general vs. NY City children's	0.99	0.47	1.14	0.63
Pittsburgh/Philadelphia general   vs. Pittsburgh/Philadelphia children's	1.05	0.23	1.08	0.33

^*On any given day, the discharge rate ratio compares the rate at which a given child would be sent home in two groups of hospitals. For instance, the ratio of 1.20 in the first row indicates that, on any given day, a child is 20 percent more likely to be discharged from a Pittsburgh or Philadelphia children's hospital than from a New York City children's hospital, so the length of stay will be longer in the New York City children's hospital. Importantly, this large difference for overall length of stay disappears after day three, where the rate ratio of 1.05 in the first row does not differ significantly from 1 (where a rate ratio of 1 means no difference). The table strongly suggests that the differences between New York and Pennsylvania are largely confined to the first few days. The tabulated values adjust for covariates using Cox's proportional hazards model.