BMJ: The prophylactic human papillomavirus (HPV) vaccines are remarkable both for their efficacy against HPV infection and related diseases,1 and for their potential to prevent cervical cancer. Cervical cancer, which is caused by persistent infection with oncogenic HPV types, remains a cause of premature death in women around the world, most of whom have no access to secondary prevention through organised cervical screening programmes.2 The linked study by Arnheim-Dahlström and colleagues (doi:10.1136/bmj.f5906) provides a timely and important contribution to the evidence base on the safety of the quadrivalent HPV vaccine,3 which prevents HPV infection and disease due to the oncogenic types HPV-16 and HPV-18 and types HPV-6 and HPV-11, which cause genital warts.
This population based cohort analysis provides strong evidence that autoimmune conditions, neurological diseases, and thromboembolic disease are not triggered by quadrivalent HPV vaccination. Serious sudden onset conditions such as these, which are largely of undetermined cause, are sometimes falsely attributed to vaccination when population based vaccination programmes are implemented.4 It is crucial that surveillance systems can rule out false associations and identify rare but real adverse effects in the post-vaccine licensure period.5
Although vaccines are subject to rigorous pre-licensure evaluations, with many thousands of subjects in clinical trials, these trials are never powerful enough to detect rare adverse events, such as those that occur in less than one in several thousand vaccinated people. Systematic whole of population analyses, such as the current one, overcome the inherent limitations of passive surveillance systems for assessing adverse events after immunisation. In a passive surveillance system, it is up to individuals or their healthcare providers to report details of adverse health events that occur in the period after vaccination. Reporting can be haphazard and evaluable information is not guaranteed. Such reporting forms the basis for evaluating cases with similar diagnoses and assessing whether there is any indication that such cases are occurring at higher than background rates. Formal epidemiological assessment then follows using a rigorous systematic approach.
Instead, Arnheim-Dahlström and colleagues took a systematic approach from the outset, which removes reporting bias (both under-reporting and over-reporting) from the initial assessment of risk. They used the enviable registry systems available in Sweden and Denmark to identify girls aged 10-17 years who were vaccinated in both countries between 2006 and 2010 (nearly 700 000 doses). Once vaccination status and timing were defined for each girl, the authors calculated the rates of 53 outcomes of interest over the 180 days after vaccination using international classification of diseases codes in hospital records (inpatient and outpatient).
Among the 29 outcomes of interest with five or more vaccinated cases, only three conditions had a significantly increased relative risk—Behçet’s syndrome, type 1 diabetes, and Raynaud’s disease. However, they each fulfilled only one of three criteria examining the strength, consistency, and reliability of the association. Particularly reassuring is the lack of any consistency in the timing of disease onset after vaccination, as visually summarised in the accompanying figures. It is reasonable to assume that these associations were chance findings, given the multiple comparisons made and lack of consistency with risk estimates for these conditions from previous studies.6
The paper lays to rest earlier concerns about an association between vaccination and venous thromboembolism that emerged from a previous analysis of US data from passive surveillance.7 With more than 116 million doses distributed globally, our experience with HPV vaccines is now considerable. The World Health Organization global advisory committee on vaccine safety has reviewed HPV vaccines four times, most recently in June 2013, and each time agreed that the available data suggest these vaccines are safe.8 Although there are published case reports of rare illnesses coincident with HPV vaccination,9 this is the weakest form of evidence, and it is only through more sophisticated studies that such hypotheses can be tested.
The major strength of Arnheim-Dahlström and colleagues’ study is its size and the assessment of a whole population. Its main weakness is that vaccination coverage in the countries under study was still relatively low during the study period and early adopters may be different from non-vaccinees. The authors adjusted for potential confounders, such as socioeconomic status and ethnicity, to control for these differences as far as possible. But ongoing monitoring remains important to provide cumulative evidence of safety, particularly as vaccines are rolled out to new population groups such as boys.
The type of evidence reported by Arnheim-Dahlström and colleagues will help secure continuing confidence in both HPV vaccination and immunisation programmes more generally. The global immunisation environment is now one in which anti-vaccination misinformation about coincidental temporal associations can readily be promulgated through websites and social media. Clear and honest communication about vaccine safety, and rapid high level response and reassurance when the safety of a vaccine comes into question, are the best ways to prevent the erosion of public confidence in immunisation. This new evidence about the quadrivalent HPV vaccine firmly indicates that concern about vaccine related adverse events is not a rational reason to forgo this potentially lifesaving vaccine. This reassurance is particularly important as the vaccine is rapidly being made available to those girls in developing countries who need it most.1
Original article found here: http://www.bmj.com/content/347/bmj.f5631?etoc=