Vaccine Safety Signal detection

As per CIOMS, signal is “Information that arises from one or multiple sources (including observations and experiments) which suggests a new potentially causal association, or a new aspect of a known association, between an intervention and an event or set of related events, either adverse or beneficial, that is judged to be of sufficient likelihood to justify verificatory action”.

Signals of possible unexpected adverse reactions or changes in severity, characteristics or frequency of expected adverse reactions may arise from any sources including preclinical, clinical and post-marketing data (e.g. spontaneous reports from Healthcare Professionals or Consumers; epidemiological studies; clinical trials), published scientific and lay literature.

Vaccine clinical trials are substantially larger than those for drugs and consequently are powered to detect rarer adverse events. In the field of vaccines, a signal may also relate to evidence of reduced efficacy or effectiveness, vaccine failures and quality deviations with potential impact on safety, efficacy or effectiveness (which may be batch-specific).

In signal detection for vaccines aside from a qualitative analysis of spontaneous case reports or case series, quantitative methods such as disproportionality analyses and observed vs. expected (O/E) analyses are routinely used.

Disproportionality analyses

Estimates of disproportionality should be calculated based on comparisons of groups that have a similar likelihood of receiving similar vaccines and experiencing similar adverse events, to prevent false disproportionality from occurring. The choice of the comparator group will depend of the objectives of the analysis and the information available in the database. A comparison with all medicinal products may result in the detection of reactions specifically related to vaccines, but may also identify a high number of false signals (e.g. SIDS in infants) or already known mild and expected reactions (e.g. local reactions). On the other hand, using all vaccine-related reports available in the database may result in signals of age- related reactions (e.g. cardiac disorders if the vaccine of interest is used in the elderly).

The new Stratified disproportionality ratio (SDR) that potentially represent genuine safety signals. The most inclusive and sensitive signal detection method would be the combination of a crude and subgroup-based data mining approach, based on the ratio between the proportion of vaccines within the ADR of interest and within all other ADRs. SDRs decreases the total number of highlighted vaccine-event pairs, it is likely to increase efficiency and therefore might reduce workload.

While detecting signals for vaccines differences between vaccines and other medicinal products should be considered, for example frequent reporting of unrelated adverse events in the target population (e.g. Sudden Infant Death Syndrome (SIDS) and childhood vaccination, myocardial infarction and influenza vaccines). Furthermore, the safety profile of a vaccine may differ substantially among the target population (e.g. higher risks in younger vaccinees).

A single report of a serious adverse event occurring in temporal association with the vaccination, especially if the event is unexpected or fatal, could have a detrimental impact on vaccination programmes due to perception of unsubstantiated risks or risk amplification. A single report of a serious adverse event should be processed as a signal only if there is a possible causal association to the vaccine.

In a first step, it may therefore be appropriate to examine results of statistical methods using both comparator groups, or to use reports for other vaccines as the comparator group with a stratification made at least by age. Given the large differences in reporting rates between regions and countries, stratification by geographical region may also be considered. Stratification by co-morbidity or co-medication is desirable, but may be difficult to achieve. If Consumer/Patient reports of suspected adverse reactions are included in the database, signal detection could also be stratified by source (Healthcare Professionals, Consumers/Patients). Stratification between study reports and spontaneous reports may be appropriate. Seasonality of vaccine administration may be relevant for some vaccines and needs consideration.

The MedDRA hierarchy needs to be considered before commencing a database search. Grouping of medically related Preferred Terms may also be considered.

Observed-to-expected analyses:

When there is little time to validate signals, it is essential to make best use of suspected adverse reaction reports. Observed vs. expected (O/E) analyses based on good-quality data can optimise the utility of passive surveillance data, allowing determination of the strength of a signal for prioritisation and further evaluation, and can help in communication of these data (particularly when serious, rare reported events are well within an expected range). O/E analyses are particularly useful during mass vaccination programmes where there is little time to review individual cases and prompt decision- making about a safety concern is required. Although such analyses cannot exclude risks or determine causality, they can help put suspected adverse reaction reports into context and should be used as a routine tool for real-time surveillance. They can also be useful in signal validation and, in the absence of robust epidemiological data, in preliminary signal evaluation.

In this method a signal is to compare the number of cases observed in temporal relationship to a suspected exposure during a period of time (O) to the number of natural incidences of the disease estimated to occur in the same period of time (E), assuming no relationship to the suspected exposure. Observed means usually reported via spontaneous reporting. O/E analyses are the first level of evaluation of safety signals. A classical approach is to calculate the O/E ratio and determine if this ratio is significantly different from one. Less conservative but more complex approaches have been developed recently. These approaches focus on E rather than on O/E and accounts for an age effect on E. In this analysis E is not a fixed number and O/E must be interpreted as a point estimate with variability around them.

Sequential methods:

The group of sequential statistical testing methods aims to test sequentially (e.g. on a monthly basis) the null hypothesis – the event rate is higher among exposed patients compared to unexposed – on prospective cohort data, as it could be done in a routine signal-detection activity. Each new analysis takes into account the number of new patients exposed and unexposed to the vaccine of interest since the last analysis, and the increment in exposure time for patients already included in the previous analysis. A signal is raised if the test statistic exceeds a predefined critical value, which is chosen so that the overall type I error is maintained at α = 0.05 across the multiple tests to reduce the generation of false positives.

References:

Vaccine Safety Signal detection

Comments

Leave a ReplyCancel reply