We confirm an increase in IA cases among hospitalized patients in the decade from 2004 to 2013. The magnitude of growth in the age-adjusted annual volume of IA hospitalizations in the current study (4.4%) echoes that noted within the National Inpatient Sample (NIS) by Vallabhaneni and co-authors (2.9%) [3]. To build on those authors’ work, we examined in detail the changes in IA hospitalizations over time. We stratified the data by such important factors as age and geography, and these stratifications clarify the nuances of the fluctuations in the IA incidence over the time frame of the study, which are smoothed out when looking at the overall population [3]. Despite these fluctuations, however, there was general growth in volume, which neared 50% over the decade examined. Normalizing the annual incidence to all US hospitalizations adds further granularity to understanding the impact of IA over time at the hospital level. Finally, our study adds to the understanding of the hospital resource utilization burden associated with IA.
We further observed that the proportion of IA hospitalizations with IA as the principal diagnosis dropped from 14 to 9% during the study period. Both of these estimates are far below a prior analysis of the NIS, a part of the HCUP, in 1996, in which over one quarter of all IA admissions carried it as the principal diagnosis [4]. Although it is reasonable to assume that most IA requires inpatient treatment, the precipitous and continuing drop in IA as the primary reason for hospitalization we note implies that other maladies are taking precedence over IA in necessitating admissions. In other words, IA may no longer represent the primary driver for hospitalization.
Our finding of a rise in crude mortality from 10 to 12% among patients with IA as the principal diagnosis, if not a cause for concern, at the very least requires further exploration. This is particularly necessary since analyses of the same source have consistently reported the death rate in this subgroup of 14 to 15% [4, 20]. It is likely that the explanation for the increase in mortality that we observed is due to the parallel increase in the severity of underlying illness among groups of patients whose diseases are now treated more aggressively and routinely, resulting in heightened susceptibility to IA. However, this requires confirmation.
We also found that while the IA-associated LOS decreased, the inflation-adjusted charges attributed to it rose by over $20/admission between 2004 and 2013. However, despite this increase in charges, the costs remained stable at approximately $32,000. This disparity between charges and costs may be indicative of a widening gap between expenditures and reimbursement incurred by hospitals over time. Of equal importance, the national bill for hospitalizations with IA as the principal diagnosis nearly doubled from a little over $300 million in 2004 to nearly $600 million in 2013. Not unexpectedly, these charges far exceed those noted in 1996 ($112.5 million) [4]. Notably, in Dasbach and colleagues’ study, patients with IA as the primary diagnosis (27% of all IA patients) contributed approximately 18% of the total aggregate US costs associated with all IA hospitalizations [4]. This implies that, given the 9% prevalence of IA as the principal diagnosis relative to all hospitalizations involving an IA code, and the total national bill for IA as principal diagnosis of approximately $600 million, the aggregate annual US charges for all IA-related hospitalizations in 2013 are likely exceeded $6 billion.
Our findings point to year-to-year and geographic variations in IA hospitalization incidence. Although our data do not allow for examinations of causality, these results point to potential directions for future research to shed light on whether there are modifiable risk factors and preventive strategies that may lessen the burden of IA hospitalizations.
Our study has a number of limitations. We may have overestimated the number of IA cases when all ICD-9-CM codes were added together, as any overlaps in these diagnoses within a single hospitalization cannot be identified in the HCUPnet. Though unavailable to us in the current analysis, in a separate study in this population, we reported the crude 30-day readmission rate to be 18% among survivors of a prior IA hospitalization [20]. Similarly, the annual incidence of IA may also be overestimated, as all cases were counted as first occurrences, when in fact a certain proportion of them may have represented readmissions. Additionally, the HCUPnet data lack the granularity necessary to examine important clinical, demographic and hospital factors that may explain some of our findings. The strength of the current study, however, lies in its vast numbers and broad generalizability to all US institutions.