Date Published: September 27, 2019
Publisher: Public Library of Science
Author(s): Inger van Heijl, Valentijn A. Schweitzer, C. H. Edwin Boel, Jan Jelrik Oosterheert, Susanne M. Huijts, Wendelien Dorigo-Zetsma, Paul D. van der Linden, Marc J. M. Bonten, Cornelis H. van Werkhoven, Robert Ehrman.
Observational studies have demonstrated that de-escalation of antimicrobial therapy is independently associated with lower mortality. This most probably results from confounding by indication. Reaching clinical stability is associated with the decision to de-escalate and with survival. However, studies rarely adjust for this confounder. We quantified the potential confounding effect of clinical stability on the estimated impact of de-escalation on mortality in patients with community-acquired pneumonia. Data were used from the Community-Acquired Pneumonia immunization Trial in Adults (CAPiTA). The primary outcome was 30-day mortality. We performed Cox proportional-hazards regression with de-escalation as time-dependent variable and adjusted for baseline characteristics using propensity scores. The potential impact of unmeasured confounding was quantified through simulating a variable representing clinical stability on day three, using data on prevalence and associations with mortality from the literature. Of 1,536 included patients, 257 (16.7%) were de-escalated, 123 (8.0%) were escalated and in 1156 (75.3%) the antibiotic spectrum remained unchanged. Crude 30-day mortality was 3.5% (9/257) and 10.9% (107/986) in the de-escalation and continuation groups, respectively. The adjusted hazard ratio of de-escalation for 30-day mortality (compared to patients with unchanged coverage), without adjustment for clinical stability, was 0.39 (95%CI: 0.19–0.79). If 90% to 100% of de-escalated patients were clinically stable on day three, the fully adjusted hazard ratio would be 0.56 (95%CI: 0.27–1.12) to 1.04 (95%CI: 0.49–2.23), respectively. The simulated confounder was substantially stronger than any of the baseline confounders in our dataset. Quantification of effects of de-escalation on patient outcomes without proper adjustment for clinical stability results in strong negative bias. This study suggests the effect of de-escalation on mortality needs further well-designed prospective research to determine effect size more accurately.
The aim of antimicrobial stewardship is improving antibiotic use, without compromising clinical outcomes on the individual level . De-escalation of empirical antimicrobial therapy is highly recommended in antimicrobial stewardship programs. In a recent systematic review de-escalation of empirical antimicrobial therapy was associated with a 56% (95% CI 34%-70%) relative risk reduction in mortality . Although it seems a safe strategy, most studies evaluating de-escalation and reporting mortality were observational with a high risk of bias, high clinical heterogeneity and not sufficiently powered to demonstrate safety for mortality. To the best of our knowledge, there are two randomized trials evaluating de-escalation, and these trials did not show a survival benefit for de-escalation [3,4]. A possible physiological mechanism for decreased mortality due to de-escalation could be a result of a more effective strategy by narrow-spectrum antibiotics or in case of continuation of unnecessary broad-spectrum antibiotics due to more (severe) side-effects. However, it seems highly unlikely that this would lead to increased mortality in the population. Therefore, the association between de-escalation and improved survival in observational studies is most likely biased by unmeasured confounding by indication. Confounding by indication is present if the indication for the intervention (here: de-escalation of empirical antimicrobial therapy) is also a prognostic factor for the outcome (mortality). De-escalation is usually only performed when clinical stability is reached in the first days after starting antimicrobial therapy and this also is a strong prognostic factor for patient outcome. However, hardly any of the observational studies adjusts for clinical stability during admission. In the aforementioned systematic review  only one of nineteen observational studies corrected for this confounder . Potentially they did not consider this to be an important confounder, or they lacked data on clinical stability during admission. Not taking this into account causes a negative bias (towards a protective effect). However, the magnitude of this bias has never been established. The aim of the current study was to quantify the potential effect of unmeasured confounding by indication due to clinical stability in the association between de-escalation and patient outcome in patients with community-acquired pneumonia.
In this observational study of patients hospitalized with CAP, after adjustment for observed baseline confounders de-escalation of antimicrobial therapy was associated with a 61% lower hazard of day-30 mortality. However, our simulations have demonstrated that clinical stability at 72 hours, which was not measured in our study, could fully explain this effect under reasonable, literature based assumptions. Based on these findings we conclude that the effects of de-escalation on patient outcome cannot be reliably quantified without adjustment for clinical stability and that the true effect of de-escalation on mortality needs to be quantified by a well-designed prospective study.