Does Length of Stay and Institution Mix for AMI Patients Affect Subsequent Health and Costs?

All hospital care is expensive, and acute care is more expensive than non-acute care. Insurers respond by encouraging a shorter length of stay (LOS) overall and/or by discharging patients to post-acute care. Subsequently, over the last several years the length of stay for AMI patients has decreased dramatically. Many researchers have explored whether shorter LOS is detrimental to patients. However, LOS is most likely endogenous, because sicker patients stay in the hospital longer and thus have higher subsequent medical costs and lower survival rates. To address this issue, we used an instrumental variables (IV) approach to evaluate the effect of acute and non-acute LOS on 1-year survival, 1-year cardiovascular-event free survival, and total healthcare costs for 1 year after hospitalization for patients who have suffered an acute myocardial infarction (AMI).

Our cohort of Medicare patients (N=186,346) were hospitalized with AMI in 2007 or 2008. Covariates included demographics, co-morbidities, complications, procedures, insurance and prior medication use. Total LOS was calculated by measuring the number of days from admission to acute care until discharge home, and divided into acute care (mean 8.09 days) and non-acute, when applicable (mean 1.83 days).  Adjusted average acute and non-acute LOS in local areas around each patient’s residence were used as instruments; thus, results are conditional on unmeasured confounders not being correlated with average LOS variation across local areas.

Results, presented in the table below, reveal that when using ordinary least squares (OLS), a longer LOS in acute care is associated with a lower rate of survival and higher subsequent costs, and LOS in non-acute care is associated with a higher survival rate and lower rate of hospital readmission. However, those with longer LOS differed from those who did not: patients with longer LOS, for example, had a higher Charlson Comorbidity Score and were more likely to suffer a stroke, renal event, or sepsis. Using instruments, however (first stage F-statistics > 1900), these relationships are diminished. The IV results show a positive, but insignificant association with survival for all categories of LOS.  Positive associations with future costs remain. Also, longer acute LOS was associated with hospital readmission. In addition, increasing LOS in non-acute care was not associated with lower readmission rates nor survival, but it was associated with increased costs.

Our results suggest that LOS studies need to identify and address issues related to endogeneity. In the case of AMI patients, our IV results suggest that shorter LOS for acute and non-acute care were not associated with lower survival rates. However, we find that longer acute LOS was associated with higher readmission rates and higher future costs.  Although longer LOS may help many patients, decreasing LOS in high-LOS regions might not be detrimental.

 

OLS	1-year   survival	Healthcare   costs, 1 year post institutional stay	30-day   hospital readmission
Full Acute	-0.00355**	480.741**	0.00214
Non-Acute	0.000332**	33.486	-0.0012**
IV
Full Acute	0.00158	541.471**	0.0062**
Non-Acute	0.00083	1037.42**	-0.00033