ARCHIVED - Canadian National Report on Immunization, 2006

Volume: 32S3 - November 2006

5. Vaccine Safety: Surveillance of Adverse Events Following Immunization

5.1 Introduction and historical background

Postmarketing surveillance of drugs, including vaccines, was started by the Bureau of Epidemiology of the Laboratory Centre for Disease Control (LCDC), following the thalidomide embryopathy epidemic in the early 1960s. In 1987, postmarketing surveillance of drugs was transferred to the Drugs Directorate (now Marketed Health Products Directorate), and surveillance of vaccine safety remained with LCDC, now the Centre for Infectious Disease Prevention and Control within PHAC. Spontaneous adverse event reporting (known as passive surveillance) by health care providers, manufacturers and consumers is the cornerstone of Canadian postmarketing surveillance of vaccine safety. In addition, 12 pediatric hospitals across Canada perform active surveillance for adverse events following immunization (AEFI) under IMPACT (Section 3.5). The Canadian adverse event reporting form is available on-line (http://www.phac-aspc.gc.ca/dird-dimr/pdf/hc4229e.pdf) , as well as in the Compendium of Pharmaceuticals and Specialties, all health centres, clinics and hospitals. Most reports are submitted first to P/T jurisdictions for public health action and follow-up, and then transferred to the national level, where all reports are aggregated and stored in a computerized,Web-enabled database. However, some reports are submitted directly to the national level (for a more complete description, see http://www.phac-aspc.gc.ca/im/vs-sv/caefiss-eng.php). An expert Advisory Committee on Causality Assessment (ACCA), operative since 1994, systematically reviews selected reports on a case-by-case basis to evaluate the likelihood that an event is causally related to a vaccine.

The primary objective of this report is to present 2004 vaccine safety surveillance data, but since it is the first summary published since 1998, temporal trends are also discussed.

5.2 Methodology

This report focuses on reported adverse events following vaccines with an immunization date from 1 January, 2004, through 31 December, 2004. Reports with no immunization date were excluded. Descriptive analyses were done, using SAS statistical programs, to characterize case reports in terms of age, sex, vaccines administered, medical attention sought, adverse events reported, time from immunization to adverse event onset, causality assigned by ACCA and reporting timeliness. The results are presented for all reports regardless of vaccine(s) given.

Where appropriate, the data for 2004 have been compared with those gathered for previous years in order to examine temporal trends in adverse event reporting frequency and rate, age distribution, vaccines administered and adverse event profile. For these analyses, published data were used for the years preceding 1997^43-48, and data were retrieved and analyzed from the existing Canadian Adverse Events Following Immunization (CAEFI) database for 1997 through 2004.

Reporting rates were calculated using total doses of distributed vaccines in each year from 1992 to 2004. The dose distribution information was not available for some vaccines given from 1999 through 2004, and therefore the corresponding adverse event reports were excluded from the rate calculation, which should then be considered an approximation of the actual rate.

Before 1997, no distinction was made for seniors as opposed to other adults, and the upper and lower limits for pre-school and school-age respectively were < 5 and ≥ 5 years, whereas they were < 7 and ≥ 7 years from 1997 onwards. Otherwise the age categories are the same across all reporting years.

Causality assessments carried out by ACCA for the serious cases with an immunization date from 1997 to 2004 were summarized. ACCA assigns causality using one of six terms derived from those originally described by WHO: very likely, probable, possible, unlikely, unrelated and unclassifiable⁴⁹. For the purposes of this report, the terms very likely/probable and unlikely/unrelated have been combined and presented as “probably related” and “unlikely related”, respectively.

Reporting timeliness was analyzed for the period 1997 to 2004 using the date of immunization, date of CAEFI form completion and the date the CAEFI report was received at the national level.

5.3 Results

As of 1 March, 2006, a total of 3,625 AEFI reports were received for vaccines given in 2004. On the basis of analyses of reporting timeliness over the last 5 years it is expected that > 95% of AEFI reports for vaccines administered during a given calendar year will be captured in an analysis done 1 year after the end of the calendar year.

A total of 8,409 adverse events were identified in the 3,625 reports for 2004. Table 5 shows the total number reported, by specific adverse event, and the respective proportion of all case reports. The three most commonly reported adverse events were local reactions (32.4% of 3,625), allergic reactions including rash (31.7%) and fever (23%).

Trends in the reported occurrence of adverse events meeting the case definition for ORS (see footnote, Table 5) following influenza vaccine have been analyzed and previously described in detail for the 1997-1998 through 2003-2004 influenza seasons⁵⁰. For the 2004-2005 season, the ORS reporting rate was 0.4 per 100,000 doses of distributed influenza vaccine.

Table 5. Profile of reported adverse events following immunization, 2004

An interval from immunization to onset was specified for 7,160 adverse events (85.2%). Cumulatively, 49% of reported events occurred within 1 day, 89.4% within 1 week and 99.3% within 1 month of immunization.

Figure 10 shows the total number of adverse event reports received and the associated reporting rates per 100,000 doses of distributed vaccines by year from 1992 to 2004.

Age could be calculated for 3,584 (95%) of vaccinees with a reported adverse event following vaccine(s) given in 2004. The age distribution was as follows: < 1 year, 12.3% (n = 447); 1 to < 2 years, 17.2% (n = 623); 2 to < 7 years, 14.5%(n = 524); 7 to < 20 years, 12.5% (n = 453); 20 to < 65 years, 31.5% (n = 1,146); and > 65 years, 6.8% (n = 248). The mean and median age, respectively, were 23 and 10 years (range, birth to 101 years). Figure 11 shows the trends in age distribution from 1989 and 1992 to 2004, and includes the proportion of reports with missing data for age.

Figure 10. Number of AEFI reports and reporting rates per 100,000 doses of distributed* vaccines, 1992 to 2004

* Net number of doses distributed (doses distributed minus doses returned)

Figure 11. Age distribution of reported AEFI, 1989, 1992 to 2004*

*Numbers in the graph columns indicate the total reports received for each age group.

Before 1997, 5 years was the limit for pre-school/school-age children rather than 7 years; 65+ years was included with the 20 to < 65 year age group.

Published data on age groups were not available for 1990 and 1991.

Overall, 60% of the 3,625 reports involved females, although distribution by sex varied with age. Among children aged < 7 years there was a predominance of males (54% male), whereas females predominated after age 7, with specific proportions by age group as follows: 7 to < 20 years, 55%female; 20 to < 65 years, 81% female; 65+ years, 76% female.

For 2004, a total of 4,905 separate vaccine events were named in the 3,625 adverse event reports following immunization. Figure 12 shows trends in vaccine types in AEFI reports between 1992 and 2004.

Figure 12. Vaccines types in AEFI reports by year of immunization, 1992 to 2004

Figure 13 shows the temporal trend, from 1992 to 2004, for relatively frequent adverse events, specifically those that accounted for ≥ 5% of all events reported for at least one of the years shown.

Medical attention associated with the AEFI reports for 2004 was as follows: hospital admission for 5% (n = 173), emergency department assessment for 7% (277), non-urgent outpatient visit for 31% (1,113), no medical attention for 28% (1,021) and not indicated for 28% (1,020).

Figure 13. Distribution of relatively frequent* AEFI, 1992 to 2004

*Accounted for ≥ 5% of all events reported for at least one of the years shown.

Outcome data were specified in 2,947 (81%) of the reports, of which 2,488 (84%) had full recovery, 272 (9%) were recovering and 179 (6%) had residual effects at the time of reporting. There were eight reports of deaths (0.3%).

Table 6 shows the results of causality assessment by ACCA for 502 of the serious events reported from 1997 to 2004. Deaths assigned as probably related to immunization were First Nations infants who received BCG and died of disseminated BCG due to underlying, but undiagnosed, severe combined immune deficiency syndrome⁵¹.

Reporting timeliness was calculated for all reports received from 1997 through 2004. The median interval from vaccine administration date to completion of an adverse event report was15 days (range 0 days to 7 years). The median interval from completion of the adverse event report to receipt at the national level was 2.2 months (range 0 days to 8 years).

Table 6. ACCA causality assessment for serious AEFI, 1997 to 2004 (n = 502)

5.4 Discussion

It must be remembered that the results reported here reflect surveillance, not research data. The type of information requested and the coding conventions have changed over time and will continue to do so. The vast majority of reports in the CAEFI system are submitted on a voluntary basis, and there is marked variability in both the quantity and quality of information provided. The total number of reports received fluctuates from year to year for a variety of reasons, related not only to changes in immunization programs (e.g. changes in the type and number of vaccines provided through publicly funded programs; mass immunization programs in response to an outbreak; catch-up programs) but also to adverse event reporting/data entry practices and capacity (e.g. changes in personnel at the level of federal, P/T and local health departments; changes in emphasis on the types of reports sent for inclusion in the national database; changes in computer systems leading to a backlog in forwarding reports that could last 4 to 6 years). It is also important to remember that a primary purpose of the voluntary reporting system is to detect signals of concern, for which capture of all events is not essential. Despite these limitations a number of interesting trends are apparent.

The annual total number of adverse event reports has varied by a factor of approximately 2-fold, from a low of 3,022 to a high of 5,992, with no consistent upward or downward trend over time. The reporting rate per 100,000 distributed vaccine doses has also varied by about 2-fold (range 16 to 40) with no consistent pattern, but clearly the two variables are independent of each other. This is a good example of why it is dangerous to rely on the absolute number of reports alone when considering vaccine safety surveillance data.

There have been some dramatic and consistent age-specific changes in AEFI reporting patterns for infants aged < 1 year, school-aged children and adults (Figure 11). Much concern has been voiced about the increasing number and complexity of vaccines given to infants over the last decade. Yet it is clear that there has been a steady decrease in both the absolute and relative number of AEFI reports submitted for infants aged < 1 year. Since vaccine coverage in this group has not changed dramatically, it is likely that the change is explainable largely by the shift from whole cell to acellular pertussis vaccines in infant vaccination programs in all jurisdictions during 1997-1998. The fact that both absolute and relative numbers of reports for toddlers and pre-school children have not changed as much over the same time frame suggests that the infant trend is not due simply to variation in reporting practice or the degree to which reports are forwarded to the national level.

Among school-aged children and adolescents (5 to 7 years through < 20 years), there has been a nearly 10-fold variation in total number of annual reports (range 232 in 1989 to 2,268 in 1996). Similarly, the proportion of all reports involving this age group has varied from 10% to 40%. Several factors may account for this variation, including meningococcal vaccine campaigns focused on school-aged children in several jurisdictions; introduction of universal hepatitis B immunization in schools across Canada during the mid to late 1990s; and adoption of 2-dose measles vaccine programs with catch-up campaigns involving school-aged children in 1996. Since 2002, the total number of reports involving school-aged children and adolescents has levelled off to approximately 400 per year, which represents just over 10% of all submitted AEFI reports.

Among adults there has been a marked increase in report frequency starting in 2000, primarily as a result of ORS associated with the influenza vaccine used in 2000 and subsequent increased reporting through a combination of heightened public and provider awareness and enhanced P/T surveillance. Ontario's adoption of universal influenza immunization may also have played a role. While the proportion of AEFI reports involving adults has remained high, at just under 40% of all reports received, the total number of reports has decreased dramatically over the last 3 years, again largely related to the drop in ORS case reporting. There have been few changes in age distribution for AEFIs over the last 2 years, and the total number of reports has remained under 4,000 despite the fact that a steadily increasing number of jurisdictions have added universal infant/toddler programs against varicella, S. pneumoniae and N. meningitidis. This trend has to be interpreted with caution since some jurisdictions temporarily stopped sending reports for technical reasons.

Figure 12 is a clear reflection of the marked changes in Canada's immunization programs over the last 17 years. The increasing number and proportion of reports involving influenza vaccine reflect not only the occurrence of ORS in 2000 but also the marked increase in vaccine coverage over the last 5 years. The variation in reports associated with measles containing vaccines, peaking in 1996, reflects the move from one- to two-dose measles and catch-up campaigns initiated in that year. Aside from that, the total number of reports involving MMR vaccines has remained very similar throughout the period shown. The figure also charts, since 1997, the cessation of whole cell pertussis and oral polio vaccine use and introduction of new vaccines in Canada: first, pediatric acellular pertussis vaccine combinations followed by live attenuated varicella zoster vaccine, conjugate pneumococcal and meningococcal vaccines and, most recently, adult formulations of acellular pertussis vaccine with Td toxoids. It is clearly important to critically examine any changes in AEFI reporting for several years after the introduction of new vaccines. Aside from the sheer number of new vaccines added to the schedule over the last few years, the fact that many are given in combination adds to the challenge of monitoring vaccine safety. Vaccine-specific AEFI data analyses are currently under way (starting with hepatitis B, MMR and influenza vaccines) and will be published on the Web in 2006 and in CCDR in the future.

In terms of specific adverse events, the most dramatic temporal trend observed for children has been the marked drop in the frequency of reports involving fever ≥ 39°C and screaming/persistent crying since the 1997 introduction of acellular pertussis vaccines. Despite the introduction of several new vaccines for universal use in Canada over the last 5 years, with accelerated adoption of these vaccines by P/T jurisdictions during 2003 and 2004, the distribution of relatively commonly reported AEFI, as shown in Figure 13, has changed little since 2000, with the single exception of ORS. While the 2004 data do not reveal any worrying trends regarding AEFI following newly introduced vaccines, it is much too early to draw any firm conclusions other than the need for continued surveillance.

The degree of medical attention associated with AEFI reports has not been consistently provided in previously published reports, and thus temporal trends in medical attention were not calculated for this report. The figure of 5% for hospitalization observed for AEFIs related to immunizations given in 2004 is similar to what was reported for both 1989 (5.2%) and 1992 (3.0%)^45,48.

An innovation for this vaccine safety report is an analysis of regional and national AEFI reporting timeliness. The vast majority of reported AEFI occur within 30 days of immunization. How quickly an AEFI is identified and reported, however, depends on many factors, including the degree of severity, the vaccinee's own or his/her caretaker's concern regarding the event and recognition by a health care professional that an event may be related to a prior vaccine. As a rule, written AEFI reports are first sent to the local health unit, then to the central P/T health departments and then onto PHAC for entry into the CAEFI database. Thus, there are many stages at which delays in reporting can occur, often for good reason, for example, delays caused by investigation into an accurate medical diagnosis or a search for etiologies other than the vaccine that may have caused the AEFI. Nevertheless, receiving reports in a timely fashion is clearly important for the rapid detection of any safety signals of concern. Collecting and analyzing AEFI reports at the national level increases the likelihood of earlier detection of rare or unusual AEFI, since it is possible to examine trends across multiple jurisdictions that might not be noticeable in a single region. It is not uncommon for P/T jurisdictions to contact the national office regarding a possible new AEFI or increased frequency of AEFI. The validity of national data analyses to address such concerns clearly depends on the timely forwarding of reports by all jurisdictions. Thus we have chosen to examine year-to-year trends in reporting timelines as one measure of the quality and utility of vaccine safety surveillance in Canada. Over the coming year, efforts to improve the timeliness of national reporting of serious adverse events will be made in collaboration with our P/T partners and the IMPACT network.