Trends in Sepsis and Infection Sources in the United States. A Population-Based Study

To the Editor:

The recently published article by Walkey and colleagues, “Trends in Sepsis and Infection Sources in the United States,” reaffirmed the important epidemiological observation that the rate of diagnosis of sepsis is rising more rapidly than the rate of diagnosis of infection (1). The authors attempted to control for secular changes in definition by measuring incidence of hospitalizations with an infection in mechanically ventilated patients, a trend that increased with increasing sepsis diagnosis. The financial incentives that may influence clinical judgments inherent in diagnostic coding of severe sepsis are well described. Additional discussion centered on the possibility that the pathogenesis of sepsis is changing.

A possibility not addressed by the authors is that over the interval 2003–2009, the standard definition of sepsis became more sensitive. The Surviving Sepsis campaign does not prioritize use of arterial blood gas (ABG) over venous blood gas (VBG) lactate, nor does it require measurement of core body temperature to diagnose hypothermia.

A meta-analysis of 10 studies concluded “a venous lactate cannot be used as a proxy for arterial in patients with abnormal lactate values...as the venous lactate becomes abnormal, the true value of arterial is less predictable from the venous lactate” (2). A patient with chronic obstructive pulmonary disease with respiratory failure requiring intubation caused by focal pneumonia, and who has a normal ABG lactate and abnormal VBG lactate, would meet severe sepsis criteria if only the VBG lactate were measured.

In a comparison of tympanic membrane and core temperatures, Haugan and colleagues observed that “a tympanic membrane temperature is on average 0.85°C lower than the rectal temperature with commonly used equipment” (3). This discrepancy between tympanic membrane and core temperature could be greater in a low-output state produced by hypovolemia, or in the presence of pulmonary embolism or congestive heart failure.

Including venous lactate and not requiring core temperature in the diagnosis of severe sepsis will increase the sensitivity and reduce the specificity of the diagnosis of severe sepsis. The impact of more sensitive diagnostic criteria on incidence and mortality of pulmonary embolism has been well described (4). Similar effects are likely in patients diagnosed with a more sensitive definition of sepsis. As a community, we should recognize that “toxicities” are likely associated with increased sepsis diagnosis including increased antibiotic exposure for syndromes not caused by bacteria and anchoring bias that could limit pursuit of alternative diagnoses.