“Sepsis to the Fore: Taking the Lead in Critical Care Admissions”

The current demographic trends in the US are not a secret: baby boomers are reaching their golden years, Americans are living longer and the disease landscape continues to change. From 2000–2010, we saw the US population ≥ 65 years-old increase 15.1% while the rest of the population grew 8.9% (1). With concern for the increased demands for critical care among older adults, US hospitals have responded with a 15.1% growth of critical care capacity over this same time period, from 67,579 beds to 77,809 beds (2). Professional societies and governing bodies have expressed concern over workforce shortages, with a 2006 US Department of Health and Human Services report predicting the population trend alone would increase the demand for adult intensivists from 2000 to 2020 by 38%, outpacing predicted supply (3, 4). In addition to this shaping force, the advances in cardiac and cancer care have improved mortality rates from these most common diseases affecting Americans ≥ 65 years-old. From 2000–2010 the age-adjusted mortality from major cardiovascular diseases in this age group decreased from 2,279 to 1,530 per 100,000 (32.9%) and mortality from malignant neoplasms decreased from 1,123 to 987 per 100,000 (12.1%) (5). How are these population shifts and disease compositions changing the face of healthcare, and specifically, critical care? The original article by Sjoding et al. provides important guidance by examining the epidemiologic trends in critical care among older Americans (6).

To briefly summarize, the article examines the 27.8 million hospital claims submitted from 1996–2010 to the Centers for Medicaid & Medicare Services (CMS) for Medicare beneficiaries ≥ 65 years-old that contained a bill for an intensive care or coronary care unit. Given the population growth in this American subpopulation over the time period, there was a surprising 29.2% decrease in the annual number of critical care admissions, from 2.16 million in 1996 to 1.53 million in 2010. And while the major cardiovascular disease group collectively remained the most common reason for hospital admissions with a critical care stay, there was a decline in the proportion of admissions with this group listed as the primary diagnosis from 50.8% to 33.3%. Conversely, the proportion of admissions with an infectious disease listed as the primary diagnosis saw an increase from 8.8% to 17.2%. Narrowing the focus from diagnostic categories to prevalence rankings for individually listed primary diagnoses, sepsis moved from the eleventh position (1.9% of primary diagnoses) to first (10.2% of primary diagnoses), supplanting coronary atherosclerosis (which fell from first to third, decreasing from 14.6% to 5.3% of primary diagnoses). In addition to these findings, there were increases in coding for chronic organ dysfunction, acute organ dysfunction, mechanical ventilation and renal replacement therapy. There was a slight increase in in-hospital mortality in the study cohort, from 11.3% to 12.0% despite a nearly 50-fold relative increase in discharges to hospice (0.1% to 4.6%).

These are certainly provocative findings. Despite the millennial panic over the aging population’s potential demand for critical care, there has been a decrease in critical care admissions (although we would need data on ICU length of stay to more fully describe critical care utilization) and a concomitant increase in hospice services for older Americans. Furthermore, the results paint an alarming picture of sepsis as an increasingly common reason for critical care admission. Despite the weight of evidence that such a massive, inclusive study population carries, all studies relying on billing claims must clear the hurdle of skepticism regarding data sources and coding practices. The data sources for the current study are the annual Medicare Provider Analysis and Review (MEDPAR) datasets, which contain all (not a selected sample) of the hospital discharge fee-for-service billing claims made to CMS in the United States. In addition, MEDPAR includes ICD-9-CM coded information on up to 6 procedures, up to 10 diagnoses, dates of service, discharge disposition and beneficiary demographics. Notably, up until 2008 this dataset did not capture the claims of Medicare beneficiaries that were submitted to any managed care organizations, and therefore from 1996–2008 this study may have underestimated the total number of admissions among Medicare beneficiaries ≥ 65 (7). Given that 12.6%–22.6% of US Medicare beneficiaries were also enrolled in optional managed care over the time period, this may be a significant underestimation and the downward trend in admissions with a critical care stay may actually be more pronounced (8). In regards to the increasing proportion of critical care hospital admissions with a primary diagnosis of sepsis, changes in coding practices certainly introduce the need for critical appraisal of this finding. A recent surge in investigation into this topic has revealed trends in the increasing use of sepsis and accompanying organ dysfunction codes without clinically documented bacteremia, evidence for organ dysfunction or necessary concomitant codes for infection (9–13). Skeptics may see these trends as financially driven, pointing to the 1999 Office of the Inspector General report that found that “a small number of hospitals have atypically high Medicare billings for Diagnosis Related Group 416 [i.e. septicemia]” often improperly ‘upcoding’ a urinary tract infection that would otherwise receive a lower reimbursement (14). On the contrary, others may argue that physicians have little personal financial incentive to ‘upcode’ and that the coding trend reflects a true increasing clinical recognition of sepsis and the application of the term to infections not limited to the bloodstream. Finally, in understanding the current study’s results in regards to national trends in sepsis, it is important to remember that its purpose is not to provide a complete description of the incidence of sepsis but rather the proportion of discharges that list sepsis as a primary diagnosis. Other studies have examined the national incidence of sepsis with more inclusive case definition that include the presence of a sepsis code among any of the 10 codes listed on the discharge claim rather than the only as the primary diagnosis.

Overall, while there is rational plausibility for the observed increases in sepsis based on biological immunosenescence and the reductions in common competing causes of morbidity and mortality among older Americans, the science of epidemiological trends remains unavoidably obscured by coding practices (15). Despite the presence of these scientific impurities, we believe that the work of Sjoding et al. still clearly demonstrates that: 1) the anticipated increase in the demand for critical care required by the aging US population needs to be re-examined and 2) sepsis is a leading national priority for the health of older Americans. The next steps in critical care utilization research require co-examination of population growth, critical care bed growth, trends in length of intensive care unit stay and critical care bed occupancy. In regards to the epidemiology of sepsis, we hope the increasing recognition of the scope of the problem and the debate over clinical and surveillance definitions are creating a conflagration to clear the forest for a wave of science to rationally develop and validate new definitions that facilitate the advancement of clinical and scientific knowledge of severe infections for the improvement of our public’s health.