You may think so, with this title: “Oral mucositis as a pathway for fatal outcome among critically ill patients exposed to chlorhexidine (CHX)”, with the conclusion that the “data points to oral mucositis as the main pathway for the association between CHX exposure and enhanced in-hospital mortality.” The research letter is a post hoc analysis of a randomized clinical trial. Time to stop using CHX mouthwash? Or time to stop building strong stories on weak data?
CHX mouthwash is widely used for mouth hygiene, usually in a 0,12% concentration. In higher concentrations it has – in mechanically ventilated patients – been associated with prevention of pneumonia. A meta-analysis back in 2011 of randomized studies actually provided evidence of a dose-dependent preventive effect. The highest concentrations tested (CHX 2%) was associated with the highest preventive effect: a Relative Risk of 0.53 (95% CI 0.31-0.91) for pneumonia.
Based on this finding we included CHX 2% as one of the interventions in an international cluster-randomized study, that was executed between 2013 and 2017. CHX 2% mouthwash was to be implemented as routine practice during 6 months in 13 ICUs, but after its use in the first 2 centers it was discontinued because of unexpected side effects in the mouth in 29 of 295 patients (9.8 %), see. For the remainder of the study we used an oral gel with CHX 1%, without further notifications of side effects.
In 2014 2 meta-analyses suggested – again – that CHX mouthwash was associated with prevention of pneumonia (albeit non-significant), but also with a higher risk of mortality (albeit non-significant in one), see & see. This was unexpected and unexplained: how can a measure that prevents pneumonia increase mortality? The fact that 2 meta-analyses found this association should not be interpreted as cumulative evidence, but rather as confirmation that a meta-analysis is reproducible wen using more or less the same studies as input.
A biological explanation for this increased mortality was not apparent. Yet, intratracheal installation of CHX leads to acute pulmonary toxic effects and fatal respiratory distress syndrome. In rats.
Back to the research letter, in which the mechanism of mortality was “explained”. Data came from a randomized clinical trial (performed between 2011-2013) evaluating a dental care intervention aimed to prevent respiratory infections in ICU patients. “The experimental group received dental care provided by a dentist plus routine oral care, while the control group had access only to routine oral care provided by the nursing staff. Both groups used CHX 0.12% oral solution, if fully conscious, or CHX 2% oral gel, if unconscious, three times a day throughout their ICU stay.” 18 of 254 patients had CHX-induced mucositis and this was exclusively reported in patients exposed to CHX 2%; 12 of the 18 patients died in ICU. The odds ratio (adjusted for age, gender and APACHE II) for CHX-induced mucositis and ICU-death was 6.14 (95% CI 1.98–19.08).
Yet, patients developing CHX-induced mucositis also had a longer stay in ICU; 14 vs 7 days, with no information on when CHX-mucositis was diagnosed. The conclusion that CHX-mucositis contributed to death in these patients can only be drawn if these patients had – at the day that CHX-mucositis developed – the same risk of dying in ICU as patients that did not develop CHX-mucositis. The APACHE II score that was used for adjustment was derived at the time of ICU-admission. It may adjust for some of the differences of mortality risk at that time point, but everything that happened to patients from then until the day that CHX-mucositis occurred (or not) is not adjusted for. It is, therefore, not (at all) excluded that these patients were more sick, and that the underlying disease and, for instance, the lack of response to treatment lead to death. The not-so-sick may have been discharged and the even-more-sick may have died before they actually could develop mucositis. Competing events that should not be ignored when quantifying the risks.
Nevertheless, the post-hoc analysis provides further evidence that CHX 2% should not be recommended for mouth care in critically ill patients. But does that kill patients? Not sure, and we should be careful in following the conclusions from meta-analyses! In the largest evaluation of the effects of CHX (2% and mostly 1%) 2108 patients with CHX 2%/1% were compared to 2251 control patients (most receiving CHX 0.12%). For patient outcome there was (1) no evidence that carriage with Gram-negative bacteria in the respiratory tract (as a proxy for pneumonia risk) was lower than during standard care and (2) no evidence that CHX2%/1% mouth care was associated with higher mortality (Hazard ratios between 0.97 and 1.03 for death at day 28, in ICU or in hospital). So far, for the predictive value of meta-analyses, but stop using CHX2%.