Erratum

This is the correct, full version of the letter published in the 8 April 2016 issue of the BDJ (2016; 220 : 324). We apologise to Dr Phil Alderson and Professor Mark Baker for our administrative error.

Sir, the article by Thornhill and colleagues1 compared the differing guidelines from NICE2 and the European Society of Cardiology3 on antibiotic prophylaxis for infective endocarditis (IE).

Unfortunately, they are not accurate in their representation of the NICE process and uncritically re-present data from their own analyses previously published in the Lancet,4 recommending that practitioners should present a summary of the Lancet data as an aid to decision making for patients.

The NICE process carefully considered both evidence of the effectiveness of policies of antibiotic prophylaxis and the evidence for the logical case underpinning such a policy. This is discussed in detail in the guideline report. This was then subjected to public consultation and the comments and responses are documented on NICE's website. Thornhill and colleagues repeat comments made in consultation, but do not acknowledge the responses that NICE have already made. They repeat a claim that NICE has stated it will not update the guidelines without a randomised controlled trial: this is not true, and readers will note that this update was triggered by the need to consider evidence from an observational interrupted time series.

Thornhill et al.'s proposal relies very heavily on the data from the work he and colleagues published in the Lancet. This paper was very carefully considered by the NICE committee and a review of the methods requested from a recognised independent expert in the field. There is clearly an increasing incidence of IE (unadjusted for age or other demographic factors) that has been present for many years. The Lancet paper examined a hypothesis that there was a single time point at which the slope of this line changed, shortly after the publication of the NICE guideline. They did not examine other hypotheses, or check whether a single change in slope was the best fit for the data rather than two or more changes in slope. Having fitted two straight lines to the data, joining in 2008, they then attribute all the difference in these slopes to the introduction of the NICE guideline, effectively dismissing any other potential explanation for the increasing incidence of IE. Thornhill et al. then propose that the Lancet summary data are presented directly to patients as the consequence of the NICE recommendation, with almost no acknowledgment of the uncertainty surrounding their estimate. In their paper in the BDJ they do not acknowledge or address any criticism of the analysis or their interpretation. This may lead to inappropriate clinical practice and we are sure readers of the BDJ will recognise that this perspective should not be presented uncritically to patients.

What is beyond dispute is that there is an increasing incidence of IE, which is not properly understood and continues around the world despite various antibiotic prophylaxis policies. As recommended in the guideline in 2008 and again in 2015, the research community needs to design better epidemiological research to understand the causes of this phenomenon and thus suggest better preventive strategies.