Acetazolamide for the Prevention of Acute Mountain Sickness: Time to Move On

Kayser and colleagues have assessed the benefits and harms of acetazolamide for the prevention of acute mountain sickness in their interesting meta-analysis (Kayser et al., 2012). Compared with placebo, acetazolamide reduced the risk of acute mountain sickness with 49% when all doses were considered, and with 45%, 50%, and 55% when daily doses of 250 mg, 500 mg, and 750 mg acetazolamide were considered (Kayser et al., 2012). However, repetitive meta-analyses are at risk of producing random errors because of sparse data, and due to multiple looks on accumulating evidence when new trials emerge (Wetterslev et al., 2008; Brok et al., 2009). The question arises whether we have sufficient evidence for the efficacy of acetazolamide? Are we convinced and are further trials which compare acetazolamide with placebo superfluous?

To minimize random errors, trial sequential analysis can be used to calculate the required information size (i.e., the number of participants needed in a meta-analysis to detect or reject a certain intervention effect, Wetterslev et al., 2008). The underlying assumption of trial sequential analysis is that significance testing may be performed each time a new trial is added to the meta-analysis. On the basis of the risk for type I (5%) and type II (20%) errors, the proportion with the outcome in the control group, the chosen risk reduction of the intervention, and the observed diversity, the diversity-adjusted required information size is calculated and the trial sequential alpha-spending and beta-spending monitoring boundaries are constructed (Wetterslev et al., 2008; Wetterslev et al., 2009; Thorlund et al., 2011).

We applied trial sequential analysis on the meta-analysis by Kayser et al., choosing a relative risk reduction (RRR) of 40%. For the comparison of acetazolamide 250 mg versus placebo, we estimated a required information size of 365 patients (currently 603 patients have been studied). For the comparison of acetazolamide 500 mg vs. placebo, we estimated a required information size of 521 patients (currently 995 patients have been studied). For the comparison of acetazolamide 750 mg vs. placebo, we estimated a required information size of 231 patients (currently 272 patients have been studied). For the comparison of all doses of acetazolamide vs. placebo, we estimated a required information size of 347 patients (currently 1870 patients have been studied). Hence, the required information size to accept a 40% RRR was reached for all doses. Even more, the alpha-spending monitoring boundaries were crossed for each individual dose, and for all doses combined before the required information sizes were reached. This means that we may exclude undue risk of random error regarding the evidence for a 40% RRR (being the lower 95% confidence limit of the intervention effect found in the traditional meta-analysis) of acetazolamide for the prevention of acute mountain sickness.

Not all of the trials included in the meta-analysis (Kayser et al., 2012) are with low risk of bias (proper randomization, blinding, etc.), so even though random errors can be excluded, the meta-analysis may still be at risk of systematic errors. However, when only including trials with low risk of systematic errors (i.e., trials with low risk of bias according to the Cochrane tool for assessment of risk of bias (Basnyat 2003; Gertsch 2004; Chow 2005; Basnyat 2006; Basnyat 2008; Gertsch 2010; Higgins 2011), acetazolamide was still better than placebo in preventing acute mountain sickness (all doses vs. placebo: RR 0.46; 95% CI 0.37–0.58, and acetazolamide 250 mg (being the dose with the lowest number of adverse effects) vs. placebo: RR 0.54; 95% CI 0.38–0.76). Furthermore, when applying trial sequential analysis on trials with low risk of bias choosing a 40% RRR, the required information size was also reached for all doses (365 patients required, and currently 1165 patients studied) as well as for a daily dose of 250 mg (351 patients required, and currently 428 patients studied). Furthermore, the trial sequential alpha-spending monitoring boundaries for trials with low risk of bias, dose of acetazolamide 250 mg daily, and the anticipation of a 25% RRR (being the lower 95% confidence limit of the intervention effect found in traditional meta-analysis) were crossed.

In short, we may exclude both random errors and systematic errors regarding a realistic efficacy of acetazolamide compared with placebo. It is unlikely that new trials will change this evidence. Hence, acetazolamide is useful for the prevention of acute mountain sickness, and it is time to move on beyond doing trials comparing acetazolamide against placebo. Questions that remain to be addressed in future trials include treatment initiation, treatment duration related to mode and speed of ascent, and combinations of acetazolamide with other drugs.