Statistical Analysis
The statistical software package SAS 9.4 (SAS Institute Inc., Cary,
North Carolina) was utilized for the analyses, which accounted for the
complex survey design and clustering. Since NIS represents a 20%
stratified random sample of US hospitals, analyses were performed using
hospital-level discharge weights provided by the NIS to obtain national
estimates of arrhythmia associated with HIV hospitalizations. For
categorical variables such as annual change in HIV hospitalization rate
and in-hospital mortality, the modified chi-squared test of trend for
proportions (Cochrane Armitage test) was used21. For
continuous variable such as LOS and cost of care, simple linear
regression was used. Multivariable models for predictors of arrhythmia
were performed with independent variables which were either clinically
significant or statistically significant in univariate model.
Multivariate model included patient-level variables such as age and
gender; year of admission, admission type (elective vs non-elective),day
of admission (weekdays vs weekend), primary payer (private insurance
[including Health Maintenance Organization] and self-pay versus
Medicaid/Medicare); co-morbidities such as diabetes mellitus, chronic
lung disease, peripheral vascular disease etc.; hospital course
requiring ICD implantation, use of vasopressors, cardiac
catheterization, endotracheal intubation and CPR; A two-sided p-value
< 0.05 was considered statistically significant.