Statins have been found to be associated with peripheral neuropathy and muscle adverse reactions such as weakness, increased fatigability, myopathy, and rhabdomyolysis [1]. In addition, sporadic cases of amyotrophic lateral sclerosis (ALS) with statins have been also reported and disproportionality for the ALS and statins has been observed in a recent analysis of the FAERS database [1,2,3].

We searched in Vigibase for individual case reports (ICSRs) in which the MedDRA preferred term “Amyotrophic lateral sclerosis (ALS)” was reported for HMG-CoA reductase inhibitors according to the anatomical therapeutic chemical (ATC) classification. A de-duplicated dataset version of Vigibase with 16.619.231 ICSRs was used to minimize the risk of identifying duplicate reports. Disproportional reporting was investigated through frequentist and Bayesian approaches by the calculation of the reporting odds ratio and the information component (IC) respectively. Cumulative RORs by year were also calculated.

Following our search, 903 ICSRs of ALS were found. From these, 220 ISCRs reported a statin and ALS. The median age of the patients in the reports was 63 years old (range 32–84). The reports involved 140 (63.6%) men and 67 (30.5%) women; sex was not specified in 13 (5.9%) reports. Of the 220 ISCRs, 107 (49%) were reported by health professionals and 48 (22%) by consumers or lawyers; reporting source was not specified in 59 (27%) reports. By regions, the reports originated in America 188 (85.5%), Europe 28 (12.7%), Asia 3 (1.4%), and Africa 1 (0.5%). From the 15 different countries in where the reports were originated, the USA contributed the most with 183 ISCRS (83.2%). For 77 (35.0%) ICSRs, there was available information on the outcome at the time of reporting: 13 died, 41 did not recover, 8 seemed to have made some recovery though with sequelae, 8 were recovering, and 7 recovered.

A statistically significant disproportionality was found for most individual statins (Table 1). The signal of ALS and statins has been strengthened for the drug-class over time. Nevertheless, the cumulative analysis of the RORs (Fig. 1) suggest a potential notoriety bias in 2008 (and 2009) consistent with the concerns and discussions from the public about a possible connection between statins and ALS published in previous data mining studies [2, 3].

Table 1 Information component (IC) and reporting odds ratio (ROR) values for statins and ALS
Fig. 1
figure 1

Cumulative RORs of statins and ALS by year. a Analysis for all the ICSRs included in Vigibase (on top). b Analysis excluding the ICSRs from the USA

Causal associations cannot (and should not) be made using data from spontaneous reports. Signals coming from spontaneous report warrant additional epidemiological and clinical investigations and a further confirmation in controlled studies are usually required [3]. However, the evidence available from controlled clinical studies has yielded conflicting results. In 2008, a pooled analysis from data provided by the drug manufacturers to the FDA (not publicly available) found an overall incidence rate of ALS of 4.2 per 100,000 person–years and 5.0 per 100,000 person–years for statin and placebo respectively (rate ratio 0.85, 95% CI 0.34–2.08) [4]. On the other hand, a population-based case–control study in the Netherlands including 722 sporadic ALS found a protective effect of statins (odds ratio 0.45, 95% CI 0.35–0.59) [5]. The pooled analysis of three observational studies (128.405 patients) included in a systematic review was inconclusive for the comparison of ALS risk for statin versus non-statin groups (rate ratio 0.89, 95% CI 0.55–1.42).

The analyses carried out to date could be statistically underpowered to detect risk differences for ALS whose worldwide annual incidence has been estimated to be 1.9 per 100,000 [6]. Despite big population samples and longer follow-up could be available in the so-called statinization era [7], large pharmacoepidemiological studies based on electronic health care databases analyzing the risk of ALS associated to statins are lacking. Moreover, there is much evidence already existent from large RCTs that has not yet be considered and critically assessed [8].