We strongly disagree with the assertions made by Chandler et al. [1] and Martinez-Lavin [2] regarding our use of evidence [3] and with the validity of the alternative information presented by the authors.

Chandler et al. [1] contend that a hierarchy of evidence is outdated and that we use epidemiological evidence to “trump” the findings of case reports and case series. The evidence hierarchy used to structure our review is a globally accepted paradigm in modern clinical medicine and healthcare [4]. Modifications to this hierarchy routinely place case series and case reports as the lowest level of evidence [5], reflecting their high risk of bias. We note that both Chandler et al. [1] and Martinez-Lavin [2] cite multiple very small and largely observational studies with no controls, several of which have been criticized [6, 7]. This is despite Martinez-Lavin [2] questioning the validity of “small” clinical trials.

The vast majority of evidence-based reviews on any topic do not include case series or case reports at all because of their inherent biases. In our review, we included such reports and acknowledged that they can have a role in raising potential safety issues; we did not characterize them as “anecdotes” or “coincidence,” as suggested by Chandler et al. [1]. Case reports allow patients and physicians to raise concerns and may contribute to hypothesis generation [3]. This has been recognized for decades and has underpinned the development of specialist adverse events clinics and networks in some countries which not only provide assessment and support for those who have experienced an adverse event but also facilitate systematic gathering of data to investigate concerns [8,9,10,11]. In the case of human papillomavirus (HPV) vaccines, this has occurred. As described in our two reviews [3, 12], in addition to an extensive body of clinical trial evidence demonstrating the safety of HPV vaccines, dozens of robust well-designed studies to investigate specific concerns have been conducted.

Chandler et al. [1] discuss variability in immunological responses to vaccination and appear to contend that case reports present data on individuals of “unusual susceptibility” to adverse events too rare to detect in epidemiological studies [1]. Yet, the opposite is true. Well-designed epidemiologic studies have, at their center, carefully validated case definitions, such as those published for potential adverse events by multi-disciplinary experts from the Brighton Collaboration. Martinez-Lavin [2] cites publications, including Chandler et al. [13], that describe symptoms such as headache, fatigue, dizziness, and musculoskeletal pain at highly variable times post-vaccination. These are common concerns; their occurrence does not imply they are caused by vaccination. In contrast, well-designed population-based studies have been used to investigate signals for, and determine the post-vaccination risk of, clearly defined and validated adverse events, including Guillain-Barré syndrome following influenza vaccination [14], narcolepsy following adjuvanted pandemic influenza vaccine [15], thrombocytopenia following the measles, mumps, and rubella (MMR) vaccine [16], and anaphylaxis following numerous individual vaccines [17].

With regard to the new era of “predictive vaccinology” discussed at length by Chandler et al. [1], we agree future developments in this field will undoubtedly help us to better understand observed individual variation in immunogenicity, efficacy, and reactogenicity to vaccines. However, the inference, if intended, that further developments in this field give validity to unsupported assumptions of causal relationships between vaccination and adverse events based on temporal associations alone is dangerous. There is potential, over the coming decades, for immunogenomics and systems biology approaches to study vaccine effects and perhaps, in years to come, provide the ability to identify predictive biomarkers for different outcomes [18, 19]. Yet even then, practical applications would need careful consideration [18]. Vaccines routinely recommended at a population level are currently held to the highest possible standards with respect to safety and overall benefit:risk profile, underpinned by extensive high-quality evidence as detailed in our and others’ reviews of HPV vaccines [3].

Martinez-Lavin [2] also presents an analysis of clinical trial data from a previously published letter [20] that has been criticized by others [21]. The additional ad hoc analyses presented are flawed and do not account for a lack of temporal association [22], with no clear methodology provided for the calculation of number needed to vaccinate. Overwhelmingly, our and other quality reviews have demonstrated the safety of HPV vaccination and a positive benefit:risk profile. Regarding some of the other conditions cited in his letter, as stated in our paper [3], the European Medicines Agency (EMA) [23] and the World Health Organization’s Global Advisory Committee on Vaccine Safety [24] also concluded there was no evidence of an association between HPV vaccine and postural orthostatic tachycardia syndrome or complex regional pain syndrome; the EMA reviewed the work by Martinez-Lavin [21] for their report.

HPV vaccines are highly effective in reducing HPV infections, genital warts, and pre-cancerous lesions of the cervix [25, 26]; reductions in cervical cancers are expected to occur imminently in vaccinated populations. As evidenced in our review [3], and that of other independent groups [23, 24, 27, 28], robust scientific evidence from around the globe supports the safety of these vaccines.