We found an increase in adherence to published guidelines in the management of S. aureus bacteremia upon the initiation of an automated infectious disease consultation. However, there was no statistical improvement in patient outcomes.
There was a particularly high rate of baseline ID involvement when compared to other published studies. Jenkins et al. performed a similar before/after study in 2005 and had a baseline consultation rate of 53 % [14], whereas, in our study ID involvement was 54.2 and 88.4 % at our two hospital sites, respectively, prior to the intervention (overall rate 72.7 %). It is unclear why the rate of baseline ID involvement was significantly higher at one of our hospitals. It is possible that the practice pattern dictated this difference. As well, there are two distinct ID services at the hospital with higher baseline consultation rates (an immune-compromised and immune-competent service), which may increase accessibility to ID consultation. Despite the considerable rate of ID involvement at baseline, this study illustrated a significant reduction in time to ID consultation at the two sites combined (2 days in comparison to 1 day, p = 0.001). Overall, this highlights that the impact of a mandatory ID consult policy is dependent upon the practice pattern at the hospital where it is initiated.
The increase in adherence to guidelines between 2011 and 2012 was 16.4 % while the increase in ID consultation was 10.4 %. It is possible that the institution of a mandated consultation led to increased awareness around guideline-based management of s. aureus bacteremia both by the ID physicians as well as the most responsible physicians, which would have contributed further to the overall improvement in adherence to guidelines.
Following the intervention, it is unclear why the rate of ID involvement did not increase to one hundred percent, as would have been expected. Seven patients passed away within 72 h and five patients were palliated, which likely explains the lack of formal consultation. We cannot find an explanation for the remaining nine cases, however. It is possible that the staff physician reviewed the antibiotics and management with the care team upon being notified by the lab and never completed the formal consult, or planned to complete the formal consult however details were lost in handover.
Multiple previous studies have illustrated a mortality benefit from a similar intervention [2, 15, 16, 18]. The lack of difference in mortality in our study may be explained partially by our small sample size and potential confounding factors. Further, the overall high baseline rate of ID involvement and the relatively high level of adherence to guidelines at baseline may mask a potential effect. For example, in the study by Nagao et al. the baseline rate of TTE/TEE use prior to ID involvement was 37.1 % [15] and Jenkins et al. found a baseline rate of 57 % [14]. In our study the baseline use of echocardiogram prior to the intervention was 85.3 %. We found a very high rate of echocardiogram completion in community-acquired cases (93.8 %) with a lower rate (79.7 %) in hospital-acquired cases.
Our study differs from a number of previous studies published on this subject (Additional file 1: Table S1). Bai et al. and Fowler et al. had a similar study question, however ID consultations were offered rather than mandated thus could have led to a selection bias [4, 18]. Nagao et al. similarly developed a mandatory consultation process, however in Japan they note that ID consultation was rare for bloodstream infections and that the ID staff could not prescribe antibiotics themselves [15]. Jenkin et al. created a mandatory consultation, however, their results were underpowered and they note that there was no specific algorithm for the evaluation and management of SAB at the time of the study [14]. Most recently, Saunderson et al. published a study with mandatory ID consultations in SAB [17]. Their control arm prior to the mandatory ID consult had the benefit of phone advice from a microbiologist, which is not standard practice in many institutions.
Thus, our study offers a different perspective on a mandatory ID consultation policy. The main limitation of our study was the retrospective design with data collection from a chart review, which provides a potential for immeasurable confounding factors between the before and after patient population and relies on clear documentation from the healthcare team.