- Email: [email protected]
Risk of febrile convulsions after MMRV vaccination in comparison to MMR or MMR+V vaccination
Vaccine 32 (2014) 645–650
Contents lists available at ScienceDirect
Vaccine journal homepage: www.elsevier.com/locate/vaccine
Risk of febrile convulsions after MMRV vaccination in comparison to MMR or MMR+V vaccination Tania Schink a,1 , Jakob Holstiege a,2 , Frank Kowalzik b,3 , Fred Zepp b,4 , Edeltraut Garbe a,∗ a b
Leibniz Institute for Prevention Research and Epidemiology – BIPS, Achterstraße 30, 28359 Bremen, Germany Center for Children and Adolescent Medicine of the Johannes Gutenberg-Universität, Langenbeckstraße 1, 55131 Mainz, Germany
a r t i c l e
i n f o
Article history: Received 13 March 2013 Received in revised form 30 October 2013 Accepted 10 December 2013 Available online 25 December 2013 Keywords: MMRV MMR Febrile convulsion Vaccination First dose
a b s t r a c t Background: In July 2006, Priorix-TetraTM , a combined measles-mumps-rubella-varicella (MMRV) vaccine, was licensed in Germany. Since a postlicensure study had shown a more than twofold elevated risk of febrile convulsions (FC) after first dose vaccination with the combined MMRV vaccine ProQuad® compared to separately administered MMR and V vaccines (MMR+V), the Paul-Ehrlich-Institute, the German regulatory agency for vaccine licensing and safety, requested a study investigating the risk of FC for Priorix-TetraTM . Methods: We performed a matched cohort study based on claims data of more than 17 million insurees in the German Pharmacoepidemiological Research Database. All children born between 01.01.2004 and 31.12.2008 who received a 1st dose of MMRV vaccine were matched to children vaccinated with MMR, MMR+V and MMR or MMR+V (combined group), respectively, by sex, age, month of vaccination and statutory health insurance. The primary outcome was defined as hospitalization with a diagnosis of FC without any alternative plausible cause of FC, e.g. an infection or neurological condition, coded as main discharge diagnosis. The secondary outcome excluded only neurological conditions to provide a more comparable outcome definition to the one used in the ProQuad® study. Numbers needed to harm (NNH), risk ratios and confounder adjusted odds ratios (ORs) with 95% CIs were estimated to compare the exposure groups. Results: In the main risk period 5–12 days after immunization, the adjusted ORs of the primary endpoint for immunization with MMRV vaccine relative to the comparator vaccine indicated in brackets were 4.1 [95% CI 1.3–12.7; MMR], 3.5 [0.7–19.0; MMR+V], and 4.1 [1.5–11.1; MMR and MMR+V]. The corresponding ORs for the secondary outcome were 2.3 [1.4–3.9; MMR], 1.5 [0.8–2.9; MMR+V] and 2.4 [1.5–3.9; MMR and MMR+V]. Conclusions: This study in children younger than 5 years, 90% of them between 11 and 23 months, shows a risk of FC similar in magnitude for Priorix-TetraTM as has previously been reported for ProQuad® suggesting a class effect for these quadrivalent vaccines. © 2013 Elsevier Ltd. All rights reserved.
1. Introduction In July 2006, the quadrivalent measles-mumps-rubella-varicella (MMRV) vaccine Priorix-TetraTM (GlaxoSmithKline) was licensed in Germany. Before, measles, mumps and rubella (MMR) vaccines
∗ Corresponding author. Tel.: +49 421 218 56862; fax: +49 421 218 56861. E-mail addresses: [email protected] (T. Schink), [email protected], [email protected] (J. Holstiege), [email protected] (F. Kowalzik), [email protected] (F. Zepp), [email protected] (E. Garbe). 1 Tel.: +49 421 218 56865; fax: +49 421 218 56941. 2 Present address: Institute of Research in Rehabilitation Medicine at Ulm University, Wuhrstraße 2/1, 88422 Bad Buchau, Germany. 3 Tel.: +49 6131 17 59 22; fax: +49 6131 17 34 82. 4 Tel.: +49 6131 17 73 26; fax: +49 6131 17 39 18. 0264-410X/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.vaccine.2013.12.011
were administered separately from varicella (V) vaccines or children were only vaccinated against MMR. The MMRV vaccine was developed to reduce the number of injections and to increase acceptance and coverage of the V vaccine. The German Standing Vaccination Committee (STIKO) recommends vaccination against measles, mumps, rubella, and varicella in all children at 11–14 months of age (1st dose) and revaccination at 15–23 months of age (2nd dose). Several months before Priorix-TetraTM was licensed in Germany, the first quadrivalent MMRV vaccine worldwide (ProQuad® ) was launched by Merck in the USA and was recommended for both the first and second dose over separately administered MMR and V vaccines (MMR+V) [1]. In 2009, an observational post-licensure study among 12–23 months old children found a more than 2fold significantly elevated relative risk (RR) of febrile convulsions (FC) in children in the time window 5–12 days after a first dose of
646
T. Schink et al. / Vaccine 32 (2014) 645–650
ProQuad® compared to separately administered MMR+V [2]. The timing of the peak in FC corresponded to the peak in fever rates which were shown to be 1.5 to 2-fold elevated after MMRV compared to MMR vaccination [3–5]. Another US study using data from the Vaccine Safety Datalink found a statistically elevated RR of 1.98 of FC in the time window 7–10 days after first dose immunization with ProQuad® in comparison to MMR+V [6]. Due to the similar composition of ProQuad® and Priorix-TetraTM and the similar pattern of post-vaccination fever [3,7], an elevated risk of FC was also of concern for Priorix-TetraTM and a study investigating this risk was requested by the Paul-Ehrlich-Institute in Germany. The objective of this study was to estimate the risk of FC after vaccination with Priorix-TetraTM compared to vaccination with MMR or MMR+V vaccines in the time-windows specified in the study by Jacobsen et al. [2]. 2. Methods 2.1. Source of data The study was carried out using data from the German Pharmacoepidemiological Research Database (GePaRD). This database consists of claims data from four German statutory health insurances (SHIs) and includes more than 17 million insurees covering all regions in Germany. It provides demographic information as well as information on hospital admissions, outpatient physician visits, and outpatient prescriptions. Hospital data include admission and discharge dates, information on in-hospital procedures and on four different types of hospital diagnoses: the main discharge diagnosis which codes the disease requiring the hospital stay, the admission diagnosis, which is a tentative diagnosis at hospital admission, diagnoses secondary to an admission or discharge diagnosis, and ancillary diagnoses (co-morbidities). Outpatient data include diagnoses, diagnostic procedures and non-drug treatments. Since outpatient physician visits are reimbursed on a quarterly basis, outpatient diagnoses can only be allocated to a quarter and not to an exact date. All diagnoses are coded according to the German modification of the 10th International Classification of Diseases (ICD-10 GM). Preliminary evaluations regarding the age and sex distribution, number of hospital admissions and drug use have shown that the database is representative for Germany [8–10]. In Germany, the utilization of health insurance data for scientific research is regulated by the Code of Social Law. The Federal Ministry of Health and the regulatory authority in Bremen approved the use of the data for this study. Informed consent was not required by law, since the necessary permissions were granted. 2.2. Study design A matched cohort study was conducted among insurees who received a first vaccination with one of the index vaccines MMRV, MMR, or MMR+V during the study period from January 1 2006 to December 31 2008. During the study period, the STIKO gave equal preference to the use of the MMRV vaccine and the separate administration of MMR and V vaccines. Eligible were insurees born between January 1 2004 and December 31 2008 for whom the insurance started no later than 180 days after the date of birth. Cohort entry was defined as the date of first immunization with one of the index vaccines. Cohort exit was defined as the first of the following dates: 91 days after cohort entry, interruption/end of insurance, death or December 31 2008. Children who received an immunization with MMRV vaccine were matched to each of the other vaccine exposure groups.
Matching was conducted one-to-one on sex, age at vaccination in months (±1 month), SHI and calendar month of vaccination (±1 month) to children who received an immunization with MMR vaccine (matched MMR cohort), MMR+V vaccine (matched MMR+V cohort) or to children who received either MMR or MMR+V vaccine (matched MMR/MMR+V cohort). 2.3. Exposure assessment Vaccinations were identified by outpatient codes used for reimbursement of administration of vaccines. For MMR and V vaccines these codes cover all brands available in Germany from different manufacturers. Vaccine dispensations in the pharmacy could not be considered, as physicians generally use vaccines kept in their own medical practices. 2.4. Outcome definition Cases were defined as hospitalizations with a diagnosis of FC, i.e. an ICD-10-GM code R56.0 in any of the hospital diagnoses. Two outcome definitions were used based on their presumed different sensitivity and specificity. The primary outcome “FC narrow” was defined as hospitalization where no alternative plausible cause of FC, e.g. an infection or a neurological condition, was coded as main discharge diagnosis. In detail, this endpoint included: (i) all hospitalizations with FC as main discharge diagnosis; (ii) all hospitalizations with FC as main admission diagnosis and without a main discharge diagnosis of an infectious disease (except measles, mumps, rubella, or chickenpox) or a neurological condition; and (iii) all hospitalizations with FC as secondary or ancillary diagnosis and a main discharge diagnosis coded as complication following immunization (ICD-10-GM code “T88.0 infection following immunization” or “T88.1 other complications following immunization, not elsewhere classified”). Due to exclusion of alternative causes of FC in this outcome definition, it was assumed that it would have higher specificity, but lower sensitivity. The secondary outcome “FC Jacobsen” was defined as closely as possible to the outcome-criteria specified by Jacobsen et al. [2]. That is, only hospitalizations for FC with a neurological condition coded as main discharge diagnosis were excluded. In consequence, “FC Jacobsen” included (i) all hospitalizations with FC as main discharge diagnosis; (ii) all hospitalizations with FC as main admission diagnosis and without a main discharge diagnosis of a neurological condition; and (iii) all hospitalizations with FC as secondary or ancillary diagnosis and with a main discharge diagnosis coded as complication following immunization. Due to inclusion of cases with an infection coded as main discharge diagnosis in this outcome definition, it was assumed to have lower specificity, but higher sensitivity. By definition, “FC narrow” cases are a subset of “FC Jacobsen” cases. To assess whether our assumptions regarding the specificity of these two outcome definitions based on coding guidelines [11] in Germany were empirically confirmed, an additional investigation was conducted to ascertain the time window in which the endpoints “FC narrow” and “FC Jacobsen” occurred after vaccine administration. 2.5. Assessment of potential confounders Potential confounders included age, sex, a prior FC, hospitalization for an infectious disease 15 days before until 30 days after vaccination, administration of other vaccines 30 days prior to 30 days after immunization with MMRV, MMR or MMR+V vaccine, and calendar month of vaccination to take into account the seasonality of infectious diseases. For FC cases, confounder assessment regarding hospitalizations for an infectious disease
T. Schink et al. / Vaccine 32 (2014) 645–650
oradministrations of other vaccines was for each risk interval limited to the time period until the onset of FC.
647
of a prior feasibility study) were entered in the model; no backward or forward selection was performed. Additionally, several other potential confounders, e.g. district, SHI, and month of vaccination, were included in separate sensitivity analyses. As these yielded no differences in the estimates for the exposure only the parsimonious model was presented. All logistic regression models were programmed using the PROC LOGISTIC procedure in SAS 9.2. The number needed to harm (NNH) was calculated for the risk interval 5–12 days after vaccination as the inverse of the unrounded RD between immunization with MMRV vaccine and immunization with MMR or MMR+V vaccine or both vaccines, respectively. Due to its higher sensitivity, this calculation was based on the outcome “FC Jacobsen”.
2.6. Statistical analysis The occurrence of the primary and secondary outcomes was compared between the different exposure groups in the following pre-specified risk intervals: 0–4 days after immunization, 5–12 days after immunization (main risk interval), 13–30 days after immunization, and the entire risk period, that is 0–30 days after immunization. Cumulative incidences (=risks) of the primary and secondary outcomes with 95% CIs were calculated for all exposure groups within each of the pre-specified risk intervals. Risk ratios (RRs) and risk differences (RDs) of the primary and secondary outcomes for the comparison of exposure groups were calculated with 95% CIs. Confounder adjusted odds ratios (ORs) with corresponding 95% CIs were estimated to compare the MMRV exposure group with each of the comparison exposure groups using. a separate binary logistic regression model for every risk interval. Accounting for the matching to avoid bias is not necessary in matched cohort studies [12], thus an unconditional logistic model was used. All variables considered as possible confounding factors (based on the results
3. Results 3.1. Cohort characteristics During the study period, 584,745 newborns were identified in the database. Of these, 477,662 (82%) fulfilled the inclusion criterion of start of insurance no later than 180 days after birth. About 51% of these children (n = 226,267) received an immunization with one of the index vaccines during the study period (2006–2008)
Table 1 Demographic characteristics and distribution of potential confounders of children in the three matched cohorts.
Sex Male (%) Age at cohort entry (months) Mean ± standard deviation Median (interquartile range)
MMRVa vs. MMRb matched cohort N = 74,734
MMRV vs. MMR+Vc matched cohort N = 32,180
MMRV vs. MMR/MMR+V matched cohort N = 82,561
MMRV
MMR
MMRV
MMR+V
MMRV
MMR or MMR+V
51.0
51.0
51.4
51.4
51.1
51.1
13.7 ± 4.70
13.8 ± 4.62
13.0 ± 3.44
12.9 ± 3.38
13.4 ± 4.53
13.5 ± 4.46
12 (11; 14)
12 (11; 14)
12 (11; 13)
12 (11; 13)
12 (11; 14)
12 (11; 14)
0–57 Minimum–maximum Between 11 and 23 months (n, %) History of FC (%) History of Epilepsy (%) Hospitalization for an infectious disease before immunization (%) Administration of other vaccines 30 days to 1 day before immunization (%) from cohort entry to 30 days after immunization (%) Month of Immunization January (%) February (%) March (%) April (%) May (%) June (%) July (%) August (%) September (%) October (%) November (%) December (%) a b c
1–58
2–50
2–51
0–57
1–58
89.88
91.41
92.53
93.97
89.96
91.95
0.77 0.76
0.76 0.76
0.69 0.76
0.58 0.82
0.73 0.75
0.72 0.77
0.17
0.15
0.20
0.16
0.17
0.15
4.80
6.08
5.09
4.15
4.58
5.43
22.32
21.86
24.33
14.10
22.37
19.88
8.02 6.21 6.21 9.63 8.66 8.93 9.95 10.06 9.67 9.47 9.99 3.20
7.71 6.05 7.56 8.47 10.34 10.88 10.70 10.13 8.96 7.83 7.92 3.45
10.99 7.12 7.52 8.71 8.89 8.78 7.96 9.06 8.78 8.97 9.07 4.15
10.00 7.41 8.19 8.67 9.65 9.94 9.88 9.17 8.98 7.09 7.03 4.01
7.26 5.62 5.63 8.73 7.93 8.39 9.66 10.40 10.88 10.88 11.48 3.15
6.76 5.84 6.52 7.57 8.94 9.78 10.83 11.13 10.51 9.13 8.98 4.02
MMRV: children vaccinated with the quadrivalent measles-mumps-rubella-varicella vaccine. MMR: children vaccinated with MMR vaccine. MMR+V: children vaccinated with MMR and varicella vaccine separately on the same day.
648
T. Schink et al. / Vaccine 32 (2014) 645–650
Table 2 Number of (a) cases, (b) risk ratios, (c) risk differences and numbers needed to harma per 10,000 of febrile convulsions based on the endpoints FC narrow and FC Jacobsen for the comparison of exposure groups with corresponding 95% CIs for the risk period 0–4 days, 5–12 days (main risk period), 13–30 days, and the entire risk period. MMRVb vs. MMRc matched cohort N = 74,734
MMRV vs. MMR+Vd matched cohort N = 32,180
MMRV vs. MMR/MMR+V matched cohort N = 82,561
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
0–4 5–12 13–30 0–30
4 vs. 5 14 vs. 3 4 vs. 9 22 vs. 17
7 vs. 13 45 vs. 19 35 vs. 31 87 vs. 63
2 vs. 0 5 vs. 1 2 vs. 1 9 vs. 2
5 vs. 4 21 vs. 14 18 vs. 12 44 vs. 30
4 vs. 4 18 vs. 4 4 vs. 8 26 vs. 16
8 vs. 15 51 vs. 21 40 vs. 31 99 vs. 67
0–4 5–12 13–30 0–30
0.8 (0.2–3.0) 4.7 (1.3–16.2) 0.4 (0.1–1.4) 1.3 (0.7–2.4)
0.5 (0.2–1.4) 2.4 (1.4–4.1) 1.1 (0.7–1.8) 1.4 (1.0–1.9)
–e 5.0 (0.6–42.8) 2.0 (0.2–22.1) 4.5 (1.0–20.8)
1.3 (0.3–4.7) 1.5 (0.8–3.0) 1.5 (0.7–3.1) 1.5 (0.9–2.3)
1.0 (0.3–4.0) 4.5 (1.5–13.3) 0.5 (0.2–1.7) 1.6 (0.9–3.0)
0.5 (0.2–1.3) 2.4 (1.5–4.0) 1.3 (0.8–2.1) 1.5 (1.1–2.0)
0–4 5–12 13–30 0–30
−0.1 (−1.6–1.4) 1.5 (−1.6–3.2) 6803 −0.7 (−2.3–1.5) 0.7 (−4.1–4.7)
−0.8 (−3.0–2.2) 3.5 (−6.0–8.2) 2874 0.54 (−6.8–7.2) 3.2 (−14.3–15.8)
0.62 (−0.2–2.3) 1.2 (−1.9–3.6) 8065 0.3 (−1.8–2.3) 2.2 (−2.7–5.3)
0.3 (−3.3–3.7) 2.2 (−8.5–10.3) 4587 1.9 (−7.4–9.1) 4.4 (−16.8–19.5)
0.0 (−1.3–1.3) 1.7 (−1.9–3.5) 5882 −0.5 (−1.9–1.3) 1.2 (−3.8–4.8)
−0.9 (−3.0–2.2) 3.6 (−6.1–8.3) 8264 1.1 (−6.4–7.1) 3.9 (−14.3–15.9)
Days after immunization
(a)
(b)
(c)
a b c d e
In the main risk period based on unrounded risk differences with three decimal places. MMRV: children vaccinated with the quadrivalent measles-mumps-rubella-varicella vaccine. MMR: children vaccinated with MMR vaccine. MMR+V: children vaccinated with MMR and varicella vaccine separately on the same day. No FC cases in the MMR+V matched cohort.
and were included in the cohort. Of these children, 82,656 (37%) received an immunization with MMRV vaccine, 111,241 (49%) received an immunization with MMR vaccine and 32,370 (14%) received an immunization with MMR+V vaccine. 3.2. Matched cohorts About 90% (n = 74,734) of the members in the MMRV exposure group could be matched to members of the MMR exposure group, while matching to the MMR+V exposure group was only possible for 39% (n = 32,180) members in the MMRV exposure group. Nearly all (n = 82,561) children who received MMRV vaccine could be matched to members of the combined exposure group (MMR or MMR+V vaccine). The characteristics of the MMRV-MMR, MMRVMMR+V and MMRV-MMR/MMR+V matched cohorts are displayed in Table 1. Besides age and sex as matching factors, other potential confounders were also similarly distributed in the matched cohorts. 3.3. Primary and secondary outcomes The unadjusted RRs and RDs of the primary outcome “FC narrow” and the secondary outcome “FC Jacobsen” are presented in Table 2. The adjusted OR regarding the primary outcome “FC narrow” in the main risk period 5–12 days after immunization was 4.1 [1.3–12.7] for immunization with MMRV vaccine compared to MMR alone, 3.5 (0.7–19.0) for MMRV compared to MMR+V and 4.1 [1.5–11.1] for MMRV compared to MMR or MMR+V (see Table 3). The corresponding ORs for the secondary outcome “FC Jacobsen” to the respective other immunizations (in brackets) were 2.3 [1.4–3.9; MMR], 1.5 [0.8–2.9; MMR+V] and 2.4 [1.5–3.9; MMR and MMR+V] (Table 3). The NNH for MMRV immunization in comparison to MMR (MMR+V and MMR/MMR+V) in the main risk period 5–12 days was 2874 (4587 and 2747) vaccinations per 1 excess case of FC for the secondary outcome “FC Jacobsen”. Occurrence of “FC Jacobsen” cases peaked between day 6–10 with 18 out of 31 cases at day 9 (see Fig. 1). A similar pattern was seen for “FC narrow”.
Fig. 1. Time of onset for (a) FC narrow cases and (b) FC Jacobsen cases in the measles-mumps-rubella-varicella vaccinated (MMRV) group (n = 82,564, dotted line) compared to the combined group of children either vaccinated with measlesmumps-rubella (MMR) alone or MMR and varicella separately on the same day (MMR+V), indicated by the solid line (n = 82,564).
T. Schink et al. / Vaccine 32 (2014) 645–650
649
Table 3 Confounder adjusteda odds ratios of febrile convulsions based on the endpoints FC narrow and FC Jacobsen for the comparison of exposure groups with corresponding 95% CIs in the risk periods 0–4 days, 5–12 days (main risk period), 13–30 days, and the entire risk period. Days after immunization
0–4 5–12 13–30 0–30
MMRVb vs. MMRc matched cohort N = 74,734
MMRV vs. MMR+Vd matched cohort N = 32,180
MMRV vs. MMR/MMR+V matched cohort N = 82,561
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
0.8 (0.3–2.5) 4.1 (1.3–12.7) 0.5 (0.2–1.4) 1.3 (0.7–2.4)
0.5 (0.2–1.3) 2.3 (1.4–3.9) 1.1 (0.7–1.8) 1.4 (1.0–1.9)
5.3 (0.4–70.0) 3.5 (0.76–19.0) 1.5 (0.3–8.7) 3.9 (1.0–14.5)
1.1 (0.3–3.5) 1.5 (0.8–2.9) 1.6 (0.8–3.2) 1.5 (1.0–2.4)
1.0 (0.3–3.3) 4.1 (1.5–11.1) 0.5 (0.2–1.6) 1.6 (0.9–3.0)
0.5 (0.2–1.2) 2.4 (1.5–3.9) 1.3 (0.8–2.0) 1.5 (1.1–2.0)
a Cohorts were matched on sex, age at vaccination in months, statutory health insurance provider (SHI) and calendar month of vaccination. Adjustments were made for the potential confounders: history of febrile convulsions, hospitalization for infectious diseases and administration of other vaccines. b MMRV: children vaccinated with the quadrivalent measles-mumps-rubella-varicella vaccine. c MMR: children vaccinated with MMR vaccine. d MMR+V: children vaccinated with MMR and varicella vaccine separately on the same day.
4. Discussion In this study, the risk of FC in children younger than 5 years, 90% of them between 11 and 23 months, was 3- to 4-fold increased 5–12 days following first dose vaccination with MMRV as compared to MMR or MMR+V based on the outcome definition “FC narrow”. Employing the broader outcome definition “FC Jacobsen” which was defined as closely as possible to the outcome definition utilized by Jacobsen et al. [2], the risk of FC was 1.5- to 2.4-fold increased. The latter risk estimate is similar in magnitude to the risk estimate observed for ProQuad® in the Jacobsen study where a more than twofold (RR: 2.20, 95% CI: 1.04–4.65) increased risk was calculated for first dose vaccination with ProQuad® in the risk interval 5–12 days in comparison to MMR+V. A similar risk estimate for ProQuad® was also found in another U.S. study using data from the Vaccine Safety Datalink [6] which reported a RR of 1.98 [1.43–2.37] of FC in the time-window 7–10 days after first dose immunization with ProQuad® in comparison to MMR+V. Our results are also compatible with those of a meta-analysis of randomized controlled trials (RCTs) conducted on behalf of GlaxoSmithKline to further explore the risk of FC associated with Priorix-TetraTM [13]. This meta-analysis included 10 RCTs including 7317 subjects who had been vaccinated with Priorix-TetraTM compared to 4455 subjects who had been vaccinated with MMR or MMR+V as a combined comparison group. The study showed a non-statistically significant elevated RR of 3.96 (0.78–38.87) of FC in the time-window 5–12 days following immunization comparing Priorix-TetraTM to MMR or MMR+V. Due to its higher sensitivity, calculation of NNH was based on the outcome “FC Jacobsen”, resulting in 1 additional FC for every 4587 doses of MMRV instead of MMR+V, 1 additional FC for every 2874 doses instead of MMR and 1 additional FC for every 2747 doses instead of MMR/MMR+V. These numbers are in line with those observed in the two U.S. studies with ProQuad® : Klein et al. [6] calculated an NNH of about 2300 and Jacobsen et al. [2] an NNH of about 2600. “FC narrow” was defined as specific as possible to avoid dilution of the outcome and to assess FC as potential vaccine complication. In Germany, a child admitted with FC to hospital would receive an admission diagnosis of FC, but if the cause of the fever and thus the FC was determined, for example, as nephritis, this should be coded as main discharge diagnosis. Thus, “FC narrow” was defined as hospitalization where no alternative plausible cause of FC, e.g. an infection or a neurological condition, was coded as main discharge diagnosis. As the appropriateness of “FC narrow” as the primary outcome was based on the compliance of physicians with coding guidelines in Germany [11], an additional analysis was conducted to explore whether “FC Jacobsen” would show the assumed lower specificity
than “FC narrow”. In this analysis, the time to the onset of FC after vaccination was compared between both endpoints. If specificity of “FC Jacobsen” was low, its cases would be rather evenly distributed across all risk windows, whereas for “FC narrow” a peak would be observed during the main risk interval 5–12 days after immunization. Interestingly, also “FC Jacobsen” showed a prominent peak in this risk window, leading to the conclusion that “FC Jacobsen” is indeed also rather specific. Due to the higher sensitivity of “FC Jacobsen”, we think that analyses based on “FC Jacobsen” give the more meaningful results and thus that “FC Jacobsen” should be regarded as the more important endpoint. Our findings of an increased risk of FC for Priorix-Tetra® as compared to MMR/MMR+V are in line with studies showing an increased risk of fever after administration of the quadrivalent vaccines Priorix-Tetra or ProQuad® . Both quadrivalent MMRV vaccines have been shown to be associated with higher fever rates than MMR+V: 21.5% vs. 14.9% for ProQuad® Dose 1 compared to M-MR® +Varivax® [7] and 11.2% vs. 7.5% for Priorix-Tetra® compared to Priorix® +Varilrix® [3]. The higher rates of fever and febrile convulsions in MMRV vaccinated children might be explained by the significantly higher geometric mean titer response to the measles component of the MMRV vaccine compared to the MMR vaccine [14]. There are some limitations of our study. Due to strict data protection regulations in Germany, confirmation of the FC diagnosis via chart review was not possible. Plausibility and completeness of FC diagnoses in the database were therefore assessed in a prior feasibility study which showed that the age distribution and characteristics of hospitalizations for FC were as expected, but had somewhat lower incidence rates compared to the literature [15]. FC could only be ascertained in hospital diagnoses, since outpatient diagnoses are related to a quarter and not to a distinct date. In Germany, hospitalization of first FCs is recommended to exclude meningitis [16]. However, the hospitalization rate for second FCs and first FCs in older children could be lower and may result in an underestimation of the occurrence of FC. Comparison of the cumulative incidences of FC in our study with those from the two U.S. studies [2,6] indicates that underascertainment may have occurred. Since underascertainment of FCs will likely be the same in all cohorts (i.e. non-differential), this should have little impact on the RRs and ORs [17]. A comparison of the occurrence of outpatient diagnoses of FCs in the quarters overlapping the risk windows revealed no relevant differences between the exposure groups, supporting our assumption of non-differential underascertainment of FCs between the exposure groups. Not all potential confounders could be assessed in the desired detail in the database, e.g. information on overall health status of the child and family history of FC was unavailable. Outpatient diagnoses of infectious diseases could not be taken into account due to lack of an exact date, however, we believe that this underascertainment is also likely to be
650
T. Schink et al. / Vaccine 32 (2014) 645–650
non-differential between the exposure groups. Misclassification of the outcome might have occurred if FCs were coded as convulsion instead of FC, however a sensitivity analysis in which we extended the definition of FC to all convulsions or seizures yielded similar, yet smaller associations, probably due to dilution of the effect by a more unspecific case definition. Strengths of our study include its sample size and the representativeness of the data. Results of a prior feasibility study we conducted showed that there was no strong indication for important misclassification of exposure or outcome and that the coverage was in accordance with the literature available for Germany [18,19]. Coverage estimates for MMR vaccination before the age of 2 vary between 75% and 93% based on the type of data collection and region in Germany. Our study results were also robust to sensitivity analyses such as exclusion of regions with low coverage of vaccination (data not shown). In conclusion, this study suggests a similar risk of FC after a first dose of Priorix-TetraTM as has been observed for a first dose of ProQuad® pointing to a class effect of these quadrivalent vaccines. The elevated risk of FC observed for the quadrivalent vaccines has to be weighed against the advantage of only one injection for the child and the potential benefit of an increased varicella immunization coverage.
Acknowledgments The authors would like to thank all SHIs that provided data for this study, inter alia the Allgemeine Ortskrankenkasse (AOK) Bremen/Bremerhaven and the DAK-Gesundheit (DAK). The authors thank the members of GlaxoSmithKline Biologicals SA for reviewing the manuscript. Disclosure statement: (1) EG is running and TS is working for a department that occasionally performs studies for pharmaceutical industries. These companies include Bayer, Celgene, GlaxoSmithKline, Mundipharma, Novartis, Sanofi-Aventis, Sanofi Pasteur MDS, and STADA. (2) EG has been a consultant to BayerSchering, Nycomed, GlaxoSmithKline, Teva and Novartis. (3) EG is a member of the German Standing Vaccination Committee (Ständige Impfkommission, STIKO). Role of the funding source: Funding for this study was provided by GlaxoSmithKline Biologicals SA. The authors had complete autonomy for the process of establishing the protocol, carrying out the analyses and interpreting the results. This also includes the full right to publish the results without limitation. GlaxoSmithKline Biologicals SA was provided the opportunity to review a preliminary version of this manuscript for factual accuracy but the authors are solely responsible for final content and interpretation. The authors received no financial support or other form of compensation related to the development of the manuscript. Contribution disclosure: TS and EG were responsible for the conception and the design of the study; TS and JH analyzed the data; all authors contributed to the interpretation of data; TS and EG drafted
the paper; JH, FK and FZ revised it critically and all authors approved the final version. References [1] CDC. Notice to readers: licensure of a combined live attenuated measles, mumps, rubella, and varicella vaccine. MMWR 2005;54(47):1212–4. [2] Jacobsen SJ, Ackerson BK, Sy LS, Tran TN, Jones TL, Yao JF, et al. Observational safety study of febrile convulsion following first dose MMRV vaccination in a managed care setting. Vaccine 2009;27(34):4656–61. [3] Czajka H, Schuster V, Zepp F, Esposito S, Douha M, Willems P. A combined measles, mumps, rubella and varicella vaccine (Priorix-Tetra): immunogenicity and safety profile. Vaccine 2009;27(47):6504–11. [4] Rumke HC, Loch HP, Hoppenbrouwers K, Vandermeulen C, Malfroot A, Helm K, et al. Immunogenicity and safety of a measles-mumps-rubella-varicella vaccine following a 4-week or a 12-month interval between two doses. Vaccine 2011;29(22):3842–9. [5] Shinefield H, Black S, Digilio L, Reisinger K, Blatter M, Gress JO, et al. Evaluation of a quadrivalent measles, mumps, rubella and varicella vaccine in healthy children. Pediatr Infect Dis J 2005;24(8):665–9. [6] Klein NP, Fireman B, Yih WK, Lewis E, Kulldorff M, Ray P, et al. Measles-mumpsrubella-varicella combination vaccine and the risk of febrile seizures. Pediatrics 2010;126(1):e1–8. [7] Kuter BJ, Brown ML, Hartzel J, Williams WR, Eves KA, Black S, et al. Safety and immunogenicity of a combination measles, mumps, rubella and varicella vaccine (ProQuad). Hum Vaccine 2006;2(5):205–14. [8] Pigeot I, Ahrens W. Establishment of a pharmacoepidemiological database in Germany: methodological potential, scientific value and practical limitations. Pharmacoepidemiol Drug Saf 2008;17(3):215–23. [9] Schink T, Garbe E. Representativity of dispensations of non-steroidal antiinflammatory drugs (NSAIDs) in the German Pharmacoepidemiological Research Database. Pharmacoepidemiol Drug Saf 2010;19:S294. [10] Schink T, Garbe E. Assessment of the representativity of in-patient hospital diagnoses in the German Pharmacoepidemiological Research Database. Pharmacoepidemiol Drug Saf 2010;19:S178–9. [11] InEK gGmbH. Deutsche Kodierrichtlinien 2007. Allgemeine und spezielle Kodierrichtlinien für die Verschlüsselung von Krankheiten und Prozeduren [German Coding Guidelines 2007. General and specific coding guidelines for coding of diseases and procedures]. Köln: Deutscher Ärzte-Verlag; 2006. p. 163. [12] Rothman KJ, Greenland S, Lash TL. Design strategies to improve study accuracy. Modern epidemiology. 3rd ed. Philadelphia: Lippincott, Williams & Wilkins; 2008. p. 168–82. [13] GlaxoSmithKline. Integrated Summary Output (ISO) on febrile convulsions; 2011. Internet 01.04.11 [cited 24.09.11]; available from: URL: http://www.gskclinicalstudyregister.com [14] Nolan T, McIntyre P, Roberton D, Descamps D. Reactogenicity and immunogenicity of a live attenuated tetravalent measles-mumps-rubella-varicella (MMRV) vaccine. Vaccine 2002;21(3–4):281–9. [15] Verburgh ME, Bruijnzeels MA, van der Wouden JC, van Suijlekom-Smit LW, van der valden J, Hoes AW, et al. Incidence of febrile seizures in The Netherlands. Neuroepidemiology 1992;11(4–6):169–72. [16] Muschenborn-Koglin S, Weiss M, Reinhardt D. [Infants and young children with fever of unknown origin. Systematic procedure based on a diagnosis-therapy diagram]. MMW Fortschr Med 1999;141(37):26–9. [17] Rothman KJ, Greenland S, Lash TL. Validity in epidemiologic studies. Modern epidemiology. 3rd ed. Philadelphia: Lippincott, Williams & Wilkins; 2008. p. 128–47. [18] Poethko-Muller C, Kuhnert R, Schlaud M. Durchimpfung und Determinanten des Impfstatus in Deutschland. Ergebnisse des Kinder- und Jugendgesundheitssurveys (KiGGS) [[Vaccination coverage and predictors for vaccination level. Results of the German Health Interview and Examination Survey for Children and Adolescents (KiGGS)]]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2007;50(5–6):851–62. [19] Poggensee G, Reuss A, Reiter S, Siedler A. Überblick und Bewertung der verfügbaren Datenquellen zur Inzidenz impfpräventabler Krankheiten, zum Durchimpfungsgrad und zum Immunstatus [[Overview and assessment of available data sources to determine incidence of vaccine preventable diseases, vaccination coverage, and immune status in Germany]]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2009;52(11):1019–28.
Contents lists available at ScienceDirect
Vaccine journal homepage: www.elsevier.com/locate/vaccine
Risk of febrile convulsions after MMRV vaccination in comparison to MMR or MMR+V vaccination Tania Schink a,1 , Jakob Holstiege a,2 , Frank Kowalzik b,3 , Fred Zepp b,4 , Edeltraut Garbe a,∗ a b
Leibniz Institute for Prevention Research and Epidemiology – BIPS, Achterstraße 30, 28359 Bremen, Germany Center for Children and Adolescent Medicine of the Johannes Gutenberg-Universität, Langenbeckstraße 1, 55131 Mainz, Germany
a r t i c l e
i n f o
Article history: Received 13 March 2013 Received in revised form 30 October 2013 Accepted 10 December 2013 Available online 25 December 2013 Keywords: MMRV MMR Febrile convulsion Vaccination First dose
a b s t r a c t Background: In July 2006, Priorix-TetraTM , a combined measles-mumps-rubella-varicella (MMRV) vaccine, was licensed in Germany. Since a postlicensure study had shown a more than twofold elevated risk of febrile convulsions (FC) after first dose vaccination with the combined MMRV vaccine ProQuad® compared to separately administered MMR and V vaccines (MMR+V), the Paul-Ehrlich-Institute, the German regulatory agency for vaccine licensing and safety, requested a study investigating the risk of FC for Priorix-TetraTM . Methods: We performed a matched cohort study based on claims data of more than 17 million insurees in the German Pharmacoepidemiological Research Database. All children born between 01.01.2004 and 31.12.2008 who received a 1st dose of MMRV vaccine were matched to children vaccinated with MMR, MMR+V and MMR or MMR+V (combined group), respectively, by sex, age, month of vaccination and statutory health insurance. The primary outcome was defined as hospitalization with a diagnosis of FC without any alternative plausible cause of FC, e.g. an infection or neurological condition, coded as main discharge diagnosis. The secondary outcome excluded only neurological conditions to provide a more comparable outcome definition to the one used in the ProQuad® study. Numbers needed to harm (NNH), risk ratios and confounder adjusted odds ratios (ORs) with 95% CIs were estimated to compare the exposure groups. Results: In the main risk period 5–12 days after immunization, the adjusted ORs of the primary endpoint for immunization with MMRV vaccine relative to the comparator vaccine indicated in brackets were 4.1 [95% CI 1.3–12.7; MMR], 3.5 [0.7–19.0; MMR+V], and 4.1 [1.5–11.1; MMR and MMR+V]. The corresponding ORs for the secondary outcome were 2.3 [1.4–3.9; MMR], 1.5 [0.8–2.9; MMR+V] and 2.4 [1.5–3.9; MMR and MMR+V]. Conclusions: This study in children younger than 5 years, 90% of them between 11 and 23 months, shows a risk of FC similar in magnitude for Priorix-TetraTM as has previously been reported for ProQuad® suggesting a class effect for these quadrivalent vaccines. © 2013 Elsevier Ltd. All rights reserved.
1. Introduction In July 2006, the quadrivalent measles-mumps-rubella-varicella (MMRV) vaccine Priorix-TetraTM (GlaxoSmithKline) was licensed in Germany. Before, measles, mumps and rubella (MMR) vaccines
∗ Corresponding author. Tel.: +49 421 218 56862; fax: +49 421 218 56861. E-mail addresses: [email protected] (T. Schink), [email protected], [email protected] (J. Holstiege), [email protected] (F. Kowalzik), [email protected] (F. Zepp), [email protected] (E. Garbe). 1 Tel.: +49 421 218 56865; fax: +49 421 218 56941. 2 Present address: Institute of Research in Rehabilitation Medicine at Ulm University, Wuhrstraße 2/1, 88422 Bad Buchau, Germany. 3 Tel.: +49 6131 17 59 22; fax: +49 6131 17 34 82. 4 Tel.: +49 6131 17 73 26; fax: +49 6131 17 39 18. 0264-410X/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.vaccine.2013.12.011
were administered separately from varicella (V) vaccines or children were only vaccinated against MMR. The MMRV vaccine was developed to reduce the number of injections and to increase acceptance and coverage of the V vaccine. The German Standing Vaccination Committee (STIKO) recommends vaccination against measles, mumps, rubella, and varicella in all children at 11–14 months of age (1st dose) and revaccination at 15–23 months of age (2nd dose). Several months before Priorix-TetraTM was licensed in Germany, the first quadrivalent MMRV vaccine worldwide (ProQuad® ) was launched by Merck in the USA and was recommended for both the first and second dose over separately administered MMR and V vaccines (MMR+V) [1]. In 2009, an observational post-licensure study among 12–23 months old children found a more than 2fold significantly elevated relative risk (RR) of febrile convulsions (FC) in children in the time window 5–12 days after a first dose of
646
T. Schink et al. / Vaccine 32 (2014) 645–650
ProQuad® compared to separately administered MMR+V [2]. The timing of the peak in FC corresponded to the peak in fever rates which were shown to be 1.5 to 2-fold elevated after MMRV compared to MMR vaccination [3–5]. Another US study using data from the Vaccine Safety Datalink found a statistically elevated RR of 1.98 of FC in the time window 7–10 days after first dose immunization with ProQuad® in comparison to MMR+V [6]. Due to the similar composition of ProQuad® and Priorix-TetraTM and the similar pattern of post-vaccination fever [3,7], an elevated risk of FC was also of concern for Priorix-TetraTM and a study investigating this risk was requested by the Paul-Ehrlich-Institute in Germany. The objective of this study was to estimate the risk of FC after vaccination with Priorix-TetraTM compared to vaccination with MMR or MMR+V vaccines in the time-windows specified in the study by Jacobsen et al. [2]. 2. Methods 2.1. Source of data The study was carried out using data from the German Pharmacoepidemiological Research Database (GePaRD). This database consists of claims data from four German statutory health insurances (SHIs) and includes more than 17 million insurees covering all regions in Germany. It provides demographic information as well as information on hospital admissions, outpatient physician visits, and outpatient prescriptions. Hospital data include admission and discharge dates, information on in-hospital procedures and on four different types of hospital diagnoses: the main discharge diagnosis which codes the disease requiring the hospital stay, the admission diagnosis, which is a tentative diagnosis at hospital admission, diagnoses secondary to an admission or discharge diagnosis, and ancillary diagnoses (co-morbidities). Outpatient data include diagnoses, diagnostic procedures and non-drug treatments. Since outpatient physician visits are reimbursed on a quarterly basis, outpatient diagnoses can only be allocated to a quarter and not to an exact date. All diagnoses are coded according to the German modification of the 10th International Classification of Diseases (ICD-10 GM). Preliminary evaluations regarding the age and sex distribution, number of hospital admissions and drug use have shown that the database is representative for Germany [8–10]. In Germany, the utilization of health insurance data for scientific research is regulated by the Code of Social Law. The Federal Ministry of Health and the regulatory authority in Bremen approved the use of the data for this study. Informed consent was not required by law, since the necessary permissions were granted. 2.2. Study design A matched cohort study was conducted among insurees who received a first vaccination with one of the index vaccines MMRV, MMR, or MMR+V during the study period from January 1 2006 to December 31 2008. During the study period, the STIKO gave equal preference to the use of the MMRV vaccine and the separate administration of MMR and V vaccines. Eligible were insurees born between January 1 2004 and December 31 2008 for whom the insurance started no later than 180 days after the date of birth. Cohort entry was defined as the date of first immunization with one of the index vaccines. Cohort exit was defined as the first of the following dates: 91 days after cohort entry, interruption/end of insurance, death or December 31 2008. Children who received an immunization with MMRV vaccine were matched to each of the other vaccine exposure groups.
Matching was conducted one-to-one on sex, age at vaccination in months (±1 month), SHI and calendar month of vaccination (±1 month) to children who received an immunization with MMR vaccine (matched MMR cohort), MMR+V vaccine (matched MMR+V cohort) or to children who received either MMR or MMR+V vaccine (matched MMR/MMR+V cohort). 2.3. Exposure assessment Vaccinations were identified by outpatient codes used for reimbursement of administration of vaccines. For MMR and V vaccines these codes cover all brands available in Germany from different manufacturers. Vaccine dispensations in the pharmacy could not be considered, as physicians generally use vaccines kept in their own medical practices. 2.4. Outcome definition Cases were defined as hospitalizations with a diagnosis of FC, i.e. an ICD-10-GM code R56.0 in any of the hospital diagnoses. Two outcome definitions were used based on their presumed different sensitivity and specificity. The primary outcome “FC narrow” was defined as hospitalization where no alternative plausible cause of FC, e.g. an infection or a neurological condition, was coded as main discharge diagnosis. In detail, this endpoint included: (i) all hospitalizations with FC as main discharge diagnosis; (ii) all hospitalizations with FC as main admission diagnosis and without a main discharge diagnosis of an infectious disease (except measles, mumps, rubella, or chickenpox) or a neurological condition; and (iii) all hospitalizations with FC as secondary or ancillary diagnosis and a main discharge diagnosis coded as complication following immunization (ICD-10-GM code “T88.0 infection following immunization” or “T88.1 other complications following immunization, not elsewhere classified”). Due to exclusion of alternative causes of FC in this outcome definition, it was assumed that it would have higher specificity, but lower sensitivity. The secondary outcome “FC Jacobsen” was defined as closely as possible to the outcome-criteria specified by Jacobsen et al. [2]. That is, only hospitalizations for FC with a neurological condition coded as main discharge diagnosis were excluded. In consequence, “FC Jacobsen” included (i) all hospitalizations with FC as main discharge diagnosis; (ii) all hospitalizations with FC as main admission diagnosis and without a main discharge diagnosis of a neurological condition; and (iii) all hospitalizations with FC as secondary or ancillary diagnosis and with a main discharge diagnosis coded as complication following immunization. Due to inclusion of cases with an infection coded as main discharge diagnosis in this outcome definition, it was assumed to have lower specificity, but higher sensitivity. By definition, “FC narrow” cases are a subset of “FC Jacobsen” cases. To assess whether our assumptions regarding the specificity of these two outcome definitions based on coding guidelines [11] in Germany were empirically confirmed, an additional investigation was conducted to ascertain the time window in which the endpoints “FC narrow” and “FC Jacobsen” occurred after vaccine administration. 2.5. Assessment of potential confounders Potential confounders included age, sex, a prior FC, hospitalization for an infectious disease 15 days before until 30 days after vaccination, administration of other vaccines 30 days prior to 30 days after immunization with MMRV, MMR or MMR+V vaccine, and calendar month of vaccination to take into account the seasonality of infectious diseases. For FC cases, confounder assessment regarding hospitalizations for an infectious disease
T. Schink et al. / Vaccine 32 (2014) 645–650
oradministrations of other vaccines was for each risk interval limited to the time period until the onset of FC.
647
of a prior feasibility study) were entered in the model; no backward or forward selection was performed. Additionally, several other potential confounders, e.g. district, SHI, and month of vaccination, were included in separate sensitivity analyses. As these yielded no differences in the estimates for the exposure only the parsimonious model was presented. All logistic regression models were programmed using the PROC LOGISTIC procedure in SAS 9.2. The number needed to harm (NNH) was calculated for the risk interval 5–12 days after vaccination as the inverse of the unrounded RD between immunization with MMRV vaccine and immunization with MMR or MMR+V vaccine or both vaccines, respectively. Due to its higher sensitivity, this calculation was based on the outcome “FC Jacobsen”.
2.6. Statistical analysis The occurrence of the primary and secondary outcomes was compared between the different exposure groups in the following pre-specified risk intervals: 0–4 days after immunization, 5–12 days after immunization (main risk interval), 13–30 days after immunization, and the entire risk period, that is 0–30 days after immunization. Cumulative incidences (=risks) of the primary and secondary outcomes with 95% CIs were calculated for all exposure groups within each of the pre-specified risk intervals. Risk ratios (RRs) and risk differences (RDs) of the primary and secondary outcomes for the comparison of exposure groups were calculated with 95% CIs. Confounder adjusted odds ratios (ORs) with corresponding 95% CIs were estimated to compare the MMRV exposure group with each of the comparison exposure groups using. a separate binary logistic regression model for every risk interval. Accounting for the matching to avoid bias is not necessary in matched cohort studies [12], thus an unconditional logistic model was used. All variables considered as possible confounding factors (based on the results
3. Results 3.1. Cohort characteristics During the study period, 584,745 newborns were identified in the database. Of these, 477,662 (82%) fulfilled the inclusion criterion of start of insurance no later than 180 days after birth. About 51% of these children (n = 226,267) received an immunization with one of the index vaccines during the study period (2006–2008)
Table 1 Demographic characteristics and distribution of potential confounders of children in the three matched cohorts.
Sex Male (%) Age at cohort entry (months) Mean ± standard deviation Median (interquartile range)
MMRVa vs. MMRb matched cohort N = 74,734
MMRV vs. MMR+Vc matched cohort N = 32,180
MMRV vs. MMR/MMR+V matched cohort N = 82,561
MMRV
MMR
MMRV
MMR+V
MMRV
MMR or MMR+V
51.0
51.0
51.4
51.4
51.1
51.1
13.7 ± 4.70
13.8 ± 4.62
13.0 ± 3.44
12.9 ± 3.38
13.4 ± 4.53
13.5 ± 4.46
12 (11; 14)
12 (11; 14)
12 (11; 13)
12 (11; 13)
12 (11; 14)
12 (11; 14)
0–57 Minimum–maximum Between 11 and 23 months (n, %) History of FC (%) History of Epilepsy (%) Hospitalization for an infectious disease before immunization (%) Administration of other vaccines 30 days to 1 day before immunization (%) from cohort entry to 30 days after immunization (%) Month of Immunization January (%) February (%) March (%) April (%) May (%) June (%) July (%) August (%) September (%) October (%) November (%) December (%) a b c
1–58
2–50
2–51
0–57
1–58
89.88
91.41
92.53
93.97
89.96
91.95
0.77 0.76
0.76 0.76
0.69 0.76
0.58 0.82
0.73 0.75
0.72 0.77
0.17
0.15
0.20
0.16
0.17
0.15
4.80
6.08
5.09
4.15
4.58
5.43
22.32
21.86
24.33
14.10
22.37
19.88
8.02 6.21 6.21 9.63 8.66 8.93 9.95 10.06 9.67 9.47 9.99 3.20
7.71 6.05 7.56 8.47 10.34 10.88 10.70 10.13 8.96 7.83 7.92 3.45
10.99 7.12 7.52 8.71 8.89 8.78 7.96 9.06 8.78 8.97 9.07 4.15
10.00 7.41 8.19 8.67 9.65 9.94 9.88 9.17 8.98 7.09 7.03 4.01
7.26 5.62 5.63 8.73 7.93 8.39 9.66 10.40 10.88 10.88 11.48 3.15
6.76 5.84 6.52 7.57 8.94 9.78 10.83 11.13 10.51 9.13 8.98 4.02
MMRV: children vaccinated with the quadrivalent measles-mumps-rubella-varicella vaccine. MMR: children vaccinated with MMR vaccine. MMR+V: children vaccinated with MMR and varicella vaccine separately on the same day.
648
T. Schink et al. / Vaccine 32 (2014) 645–650
Table 2 Number of (a) cases, (b) risk ratios, (c) risk differences and numbers needed to harma per 10,000 of febrile convulsions based on the endpoints FC narrow and FC Jacobsen for the comparison of exposure groups with corresponding 95% CIs for the risk period 0–4 days, 5–12 days (main risk period), 13–30 days, and the entire risk period. MMRVb vs. MMRc matched cohort N = 74,734
MMRV vs. MMR+Vd matched cohort N = 32,180
MMRV vs. MMR/MMR+V matched cohort N = 82,561
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
0–4 5–12 13–30 0–30
4 vs. 5 14 vs. 3 4 vs. 9 22 vs. 17
7 vs. 13 45 vs. 19 35 vs. 31 87 vs. 63
2 vs. 0 5 vs. 1 2 vs. 1 9 vs. 2
5 vs. 4 21 vs. 14 18 vs. 12 44 vs. 30
4 vs. 4 18 vs. 4 4 vs. 8 26 vs. 16
8 vs. 15 51 vs. 21 40 vs. 31 99 vs. 67
0–4 5–12 13–30 0–30
0.8 (0.2–3.0) 4.7 (1.3–16.2) 0.4 (0.1–1.4) 1.3 (0.7–2.4)
0.5 (0.2–1.4) 2.4 (1.4–4.1) 1.1 (0.7–1.8) 1.4 (1.0–1.9)
–e 5.0 (0.6–42.8) 2.0 (0.2–22.1) 4.5 (1.0–20.8)
1.3 (0.3–4.7) 1.5 (0.8–3.0) 1.5 (0.7–3.1) 1.5 (0.9–2.3)
1.0 (0.3–4.0) 4.5 (1.5–13.3) 0.5 (0.2–1.7) 1.6 (0.9–3.0)
0.5 (0.2–1.3) 2.4 (1.5–4.0) 1.3 (0.8–2.1) 1.5 (1.1–2.0)
0–4 5–12 13–30 0–30
−0.1 (−1.6–1.4) 1.5 (−1.6–3.2) 6803 −0.7 (−2.3–1.5) 0.7 (−4.1–4.7)
−0.8 (−3.0–2.2) 3.5 (−6.0–8.2) 2874 0.54 (−6.8–7.2) 3.2 (−14.3–15.8)
0.62 (−0.2–2.3) 1.2 (−1.9–3.6) 8065 0.3 (−1.8–2.3) 2.2 (−2.7–5.3)
0.3 (−3.3–3.7) 2.2 (−8.5–10.3) 4587 1.9 (−7.4–9.1) 4.4 (−16.8–19.5)
0.0 (−1.3–1.3) 1.7 (−1.9–3.5) 5882 −0.5 (−1.9–1.3) 1.2 (−3.8–4.8)
−0.9 (−3.0–2.2) 3.6 (−6.1–8.3) 8264 1.1 (−6.4–7.1) 3.9 (−14.3–15.9)
Days after immunization
(a)
(b)
(c)
a b c d e
In the main risk period based on unrounded risk differences with three decimal places. MMRV: children vaccinated with the quadrivalent measles-mumps-rubella-varicella vaccine. MMR: children vaccinated with MMR vaccine. MMR+V: children vaccinated with MMR and varicella vaccine separately on the same day. No FC cases in the MMR+V matched cohort.
and were included in the cohort. Of these children, 82,656 (37%) received an immunization with MMRV vaccine, 111,241 (49%) received an immunization with MMR vaccine and 32,370 (14%) received an immunization with MMR+V vaccine. 3.2. Matched cohorts About 90% (n = 74,734) of the members in the MMRV exposure group could be matched to members of the MMR exposure group, while matching to the MMR+V exposure group was only possible for 39% (n = 32,180) members in the MMRV exposure group. Nearly all (n = 82,561) children who received MMRV vaccine could be matched to members of the combined exposure group (MMR or MMR+V vaccine). The characteristics of the MMRV-MMR, MMRVMMR+V and MMRV-MMR/MMR+V matched cohorts are displayed in Table 1. Besides age and sex as matching factors, other potential confounders were also similarly distributed in the matched cohorts. 3.3. Primary and secondary outcomes The unadjusted RRs and RDs of the primary outcome “FC narrow” and the secondary outcome “FC Jacobsen” are presented in Table 2. The adjusted OR regarding the primary outcome “FC narrow” in the main risk period 5–12 days after immunization was 4.1 [1.3–12.7] for immunization with MMRV vaccine compared to MMR alone, 3.5 (0.7–19.0) for MMRV compared to MMR+V and 4.1 [1.5–11.1] for MMRV compared to MMR or MMR+V (see Table 3). The corresponding ORs for the secondary outcome “FC Jacobsen” to the respective other immunizations (in brackets) were 2.3 [1.4–3.9; MMR], 1.5 [0.8–2.9; MMR+V] and 2.4 [1.5–3.9; MMR and MMR+V] (Table 3). The NNH for MMRV immunization in comparison to MMR (MMR+V and MMR/MMR+V) in the main risk period 5–12 days was 2874 (4587 and 2747) vaccinations per 1 excess case of FC for the secondary outcome “FC Jacobsen”. Occurrence of “FC Jacobsen” cases peaked between day 6–10 with 18 out of 31 cases at day 9 (see Fig. 1). A similar pattern was seen for “FC narrow”.
Fig. 1. Time of onset for (a) FC narrow cases and (b) FC Jacobsen cases in the measles-mumps-rubella-varicella vaccinated (MMRV) group (n = 82,564, dotted line) compared to the combined group of children either vaccinated with measlesmumps-rubella (MMR) alone or MMR and varicella separately on the same day (MMR+V), indicated by the solid line (n = 82,564).
T. Schink et al. / Vaccine 32 (2014) 645–650
649
Table 3 Confounder adjusteda odds ratios of febrile convulsions based on the endpoints FC narrow and FC Jacobsen for the comparison of exposure groups with corresponding 95% CIs in the risk periods 0–4 days, 5–12 days (main risk period), 13–30 days, and the entire risk period. Days after immunization
0–4 5–12 13–30 0–30
MMRVb vs. MMRc matched cohort N = 74,734
MMRV vs. MMR+Vd matched cohort N = 32,180
MMRV vs. MMR/MMR+V matched cohort N = 82,561
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
FC narrow
FC Jacobsen
0.8 (0.3–2.5) 4.1 (1.3–12.7) 0.5 (0.2–1.4) 1.3 (0.7–2.4)
0.5 (0.2–1.3) 2.3 (1.4–3.9) 1.1 (0.7–1.8) 1.4 (1.0–1.9)
5.3 (0.4–70.0) 3.5 (0.76–19.0) 1.5 (0.3–8.7) 3.9 (1.0–14.5)
1.1 (0.3–3.5) 1.5 (0.8–2.9) 1.6 (0.8–3.2) 1.5 (1.0–2.4)
1.0 (0.3–3.3) 4.1 (1.5–11.1) 0.5 (0.2–1.6) 1.6 (0.9–3.0)
0.5 (0.2–1.2) 2.4 (1.5–3.9) 1.3 (0.8–2.0) 1.5 (1.1–2.0)
a Cohorts were matched on sex, age at vaccination in months, statutory health insurance provider (SHI) and calendar month of vaccination. Adjustments were made for the potential confounders: history of febrile convulsions, hospitalization for infectious diseases and administration of other vaccines. b MMRV: children vaccinated with the quadrivalent measles-mumps-rubella-varicella vaccine. c MMR: children vaccinated with MMR vaccine. d MMR+V: children vaccinated with MMR and varicella vaccine separately on the same day.
4. Discussion In this study, the risk of FC in children younger than 5 years, 90% of them between 11 and 23 months, was 3- to 4-fold increased 5–12 days following first dose vaccination with MMRV as compared to MMR or MMR+V based on the outcome definition “FC narrow”. Employing the broader outcome definition “FC Jacobsen” which was defined as closely as possible to the outcome definition utilized by Jacobsen et al. [2], the risk of FC was 1.5- to 2.4-fold increased. The latter risk estimate is similar in magnitude to the risk estimate observed for ProQuad® in the Jacobsen study where a more than twofold (RR: 2.20, 95% CI: 1.04–4.65) increased risk was calculated for first dose vaccination with ProQuad® in the risk interval 5–12 days in comparison to MMR+V. A similar risk estimate for ProQuad® was also found in another U.S. study using data from the Vaccine Safety Datalink [6] which reported a RR of 1.98 [1.43–2.37] of FC in the time-window 7–10 days after first dose immunization with ProQuad® in comparison to MMR+V. Our results are also compatible with those of a meta-analysis of randomized controlled trials (RCTs) conducted on behalf of GlaxoSmithKline to further explore the risk of FC associated with Priorix-TetraTM [13]. This meta-analysis included 10 RCTs including 7317 subjects who had been vaccinated with Priorix-TetraTM compared to 4455 subjects who had been vaccinated with MMR or MMR+V as a combined comparison group. The study showed a non-statistically significant elevated RR of 3.96 (0.78–38.87) of FC in the time-window 5–12 days following immunization comparing Priorix-TetraTM to MMR or MMR+V. Due to its higher sensitivity, calculation of NNH was based on the outcome “FC Jacobsen”, resulting in 1 additional FC for every 4587 doses of MMRV instead of MMR+V, 1 additional FC for every 2874 doses instead of MMR and 1 additional FC for every 2747 doses instead of MMR/MMR+V. These numbers are in line with those observed in the two U.S. studies with ProQuad® : Klein et al. [6] calculated an NNH of about 2300 and Jacobsen et al. [2] an NNH of about 2600. “FC narrow” was defined as specific as possible to avoid dilution of the outcome and to assess FC as potential vaccine complication. In Germany, a child admitted with FC to hospital would receive an admission diagnosis of FC, but if the cause of the fever and thus the FC was determined, for example, as nephritis, this should be coded as main discharge diagnosis. Thus, “FC narrow” was defined as hospitalization where no alternative plausible cause of FC, e.g. an infection or a neurological condition, was coded as main discharge diagnosis. As the appropriateness of “FC narrow” as the primary outcome was based on the compliance of physicians with coding guidelines in Germany [11], an additional analysis was conducted to explore whether “FC Jacobsen” would show the assumed lower specificity
than “FC narrow”. In this analysis, the time to the onset of FC after vaccination was compared between both endpoints. If specificity of “FC Jacobsen” was low, its cases would be rather evenly distributed across all risk windows, whereas for “FC narrow” a peak would be observed during the main risk interval 5–12 days after immunization. Interestingly, also “FC Jacobsen” showed a prominent peak in this risk window, leading to the conclusion that “FC Jacobsen” is indeed also rather specific. Due to the higher sensitivity of “FC Jacobsen”, we think that analyses based on “FC Jacobsen” give the more meaningful results and thus that “FC Jacobsen” should be regarded as the more important endpoint. Our findings of an increased risk of FC for Priorix-Tetra® as compared to MMR/MMR+V are in line with studies showing an increased risk of fever after administration of the quadrivalent vaccines Priorix-Tetra or ProQuad® . Both quadrivalent MMRV vaccines have been shown to be associated with higher fever rates than MMR+V: 21.5% vs. 14.9% for ProQuad® Dose 1 compared to M-MR® +Varivax® [7] and 11.2% vs. 7.5% for Priorix-Tetra® compared to Priorix® +Varilrix® [3]. The higher rates of fever and febrile convulsions in MMRV vaccinated children might be explained by the significantly higher geometric mean titer response to the measles component of the MMRV vaccine compared to the MMR vaccine [14]. There are some limitations of our study. Due to strict data protection regulations in Germany, confirmation of the FC diagnosis via chart review was not possible. Plausibility and completeness of FC diagnoses in the database were therefore assessed in a prior feasibility study which showed that the age distribution and characteristics of hospitalizations for FC were as expected, but had somewhat lower incidence rates compared to the literature [15]. FC could only be ascertained in hospital diagnoses, since outpatient diagnoses are related to a quarter and not to a distinct date. In Germany, hospitalization of first FCs is recommended to exclude meningitis [16]. However, the hospitalization rate for second FCs and first FCs in older children could be lower and may result in an underestimation of the occurrence of FC. Comparison of the cumulative incidences of FC in our study with those from the two U.S. studies [2,6] indicates that underascertainment may have occurred. Since underascertainment of FCs will likely be the same in all cohorts (i.e. non-differential), this should have little impact on the RRs and ORs [17]. A comparison of the occurrence of outpatient diagnoses of FCs in the quarters overlapping the risk windows revealed no relevant differences between the exposure groups, supporting our assumption of non-differential underascertainment of FCs between the exposure groups. Not all potential confounders could be assessed in the desired detail in the database, e.g. information on overall health status of the child and family history of FC was unavailable. Outpatient diagnoses of infectious diseases could not be taken into account due to lack of an exact date, however, we believe that this underascertainment is also likely to be
650
T. Schink et al. / Vaccine 32 (2014) 645–650
non-differential between the exposure groups. Misclassification of the outcome might have occurred if FCs were coded as convulsion instead of FC, however a sensitivity analysis in which we extended the definition of FC to all convulsions or seizures yielded similar, yet smaller associations, probably due to dilution of the effect by a more unspecific case definition. Strengths of our study include its sample size and the representativeness of the data. Results of a prior feasibility study we conducted showed that there was no strong indication for important misclassification of exposure or outcome and that the coverage was in accordance with the literature available for Germany [18,19]. Coverage estimates for MMR vaccination before the age of 2 vary between 75% and 93% based on the type of data collection and region in Germany. Our study results were also robust to sensitivity analyses such as exclusion of regions with low coverage of vaccination (data not shown). In conclusion, this study suggests a similar risk of FC after a first dose of Priorix-TetraTM as has been observed for a first dose of ProQuad® pointing to a class effect of these quadrivalent vaccines. The elevated risk of FC observed for the quadrivalent vaccines has to be weighed against the advantage of only one injection for the child and the potential benefit of an increased varicella immunization coverage.
Acknowledgments The authors would like to thank all SHIs that provided data for this study, inter alia the Allgemeine Ortskrankenkasse (AOK) Bremen/Bremerhaven and the DAK-Gesundheit (DAK). The authors thank the members of GlaxoSmithKline Biologicals SA for reviewing the manuscript. Disclosure statement: (1) EG is running and TS is working for a department that occasionally performs studies for pharmaceutical industries. These companies include Bayer, Celgene, GlaxoSmithKline, Mundipharma, Novartis, Sanofi-Aventis, Sanofi Pasteur MDS, and STADA. (2) EG has been a consultant to BayerSchering, Nycomed, GlaxoSmithKline, Teva and Novartis. (3) EG is a member of the German Standing Vaccination Committee (Ständige Impfkommission, STIKO). Role of the funding source: Funding for this study was provided by GlaxoSmithKline Biologicals SA. The authors had complete autonomy for the process of establishing the protocol, carrying out the analyses and interpreting the results. This also includes the full right to publish the results without limitation. GlaxoSmithKline Biologicals SA was provided the opportunity to review a preliminary version of this manuscript for factual accuracy but the authors are solely responsible for final content and interpretation. The authors received no financial support or other form of compensation related to the development of the manuscript. Contribution disclosure: TS and EG were responsible for the conception and the design of the study; TS and JH analyzed the data; all authors contributed to the interpretation of data; TS and EG drafted
the paper; JH, FK and FZ revised it critically and all authors approved the final version. References [1] CDC. Notice to readers: licensure of a combined live attenuated measles, mumps, rubella, and varicella vaccine. MMWR 2005;54(47):1212–4. [2] Jacobsen SJ, Ackerson BK, Sy LS, Tran TN, Jones TL, Yao JF, et al. Observational safety study of febrile convulsion following first dose MMRV vaccination in a managed care setting. Vaccine 2009;27(34):4656–61. [3] Czajka H, Schuster V, Zepp F, Esposito S, Douha M, Willems P. A combined measles, mumps, rubella and varicella vaccine (Priorix-Tetra): immunogenicity and safety profile. Vaccine 2009;27(47):6504–11. [4] Rumke HC, Loch HP, Hoppenbrouwers K, Vandermeulen C, Malfroot A, Helm K, et al. Immunogenicity and safety of a measles-mumps-rubella-varicella vaccine following a 4-week or a 12-month interval between two doses. Vaccine 2011;29(22):3842–9. [5] Shinefield H, Black S, Digilio L, Reisinger K, Blatter M, Gress JO, et al. Evaluation of a quadrivalent measles, mumps, rubella and varicella vaccine in healthy children. Pediatr Infect Dis J 2005;24(8):665–9. [6] Klein NP, Fireman B, Yih WK, Lewis E, Kulldorff M, Ray P, et al. Measles-mumpsrubella-varicella combination vaccine and the risk of febrile seizures. Pediatrics 2010;126(1):e1–8. [7] Kuter BJ, Brown ML, Hartzel J, Williams WR, Eves KA, Black S, et al. Safety and immunogenicity of a combination measles, mumps, rubella and varicella vaccine (ProQuad). Hum Vaccine 2006;2(5):205–14. [8] Pigeot I, Ahrens W. Establishment of a pharmacoepidemiological database in Germany: methodological potential, scientific value and practical limitations. Pharmacoepidemiol Drug Saf 2008;17(3):215–23. [9] Schink T, Garbe E. Representativity of dispensations of non-steroidal antiinflammatory drugs (NSAIDs) in the German Pharmacoepidemiological Research Database. Pharmacoepidemiol Drug Saf 2010;19:S294. [10] Schink T, Garbe E. Assessment of the representativity of in-patient hospital diagnoses in the German Pharmacoepidemiological Research Database. Pharmacoepidemiol Drug Saf 2010;19:S178–9. [11] InEK gGmbH. Deutsche Kodierrichtlinien 2007. Allgemeine und spezielle Kodierrichtlinien für die Verschlüsselung von Krankheiten und Prozeduren [German Coding Guidelines 2007. General and specific coding guidelines for coding of diseases and procedures]. Köln: Deutscher Ärzte-Verlag; 2006. p. 163. [12] Rothman KJ, Greenland S, Lash TL. Design strategies to improve study accuracy. Modern epidemiology. 3rd ed. Philadelphia: Lippincott, Williams & Wilkins; 2008. p. 168–82. [13] GlaxoSmithKline. Integrated Summary Output (ISO) on febrile convulsions; 2011. Internet 01.04.11 [cited 24.09.11]; available from: URL: http://www.gskclinicalstudyregister.com [14] Nolan T, McIntyre P, Roberton D, Descamps D. Reactogenicity and immunogenicity of a live attenuated tetravalent measles-mumps-rubella-varicella (MMRV) vaccine. Vaccine 2002;21(3–4):281–9. [15] Verburgh ME, Bruijnzeels MA, van der Wouden JC, van Suijlekom-Smit LW, van der valden J, Hoes AW, et al. Incidence of febrile seizures in The Netherlands. Neuroepidemiology 1992;11(4–6):169–72. [16] Muschenborn-Koglin S, Weiss M, Reinhardt D. [Infants and young children with fever of unknown origin. Systematic procedure based on a diagnosis-therapy diagram]. MMW Fortschr Med 1999;141(37):26–9. [17] Rothman KJ, Greenland S, Lash TL. Validity in epidemiologic studies. Modern epidemiology. 3rd ed. Philadelphia: Lippincott, Williams & Wilkins; 2008. p. 128–47. [18] Poethko-Muller C, Kuhnert R, Schlaud M. Durchimpfung und Determinanten des Impfstatus in Deutschland. Ergebnisse des Kinder- und Jugendgesundheitssurveys (KiGGS) [[Vaccination coverage and predictors for vaccination level. Results of the German Health Interview and Examination Survey for Children and Adolescents (KiGGS)]]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2007;50(5–6):851–62. [19] Poggensee G, Reuss A, Reiter S, Siedler A. Überblick und Bewertung der verfügbaren Datenquellen zur Inzidenz impfpräventabler Krankheiten, zum Durchimpfungsgrad und zum Immunstatus [[Overview and assessment of available data sources to determine incidence of vaccine preventable diseases, vaccination coverage, and immune status in Germany]]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2009;52(11):1019–28.