Guilherme Gonçalves ¹, A. de Araújo ², M. L. Monteiro Cardoso ^2
1 Instituto Nacional de Saúde do Dr. Ricardo Jorge, Delegação no Porto (National Institute of Health), Portugal
² Delegação de Saúde Concelhia do Porto (Oporto Health Authority), Portugal

Vaccination against mumps in Portugal began in 1987, with the introduction of the combined measles, mumps, and rubella vaccine (MMR) in the national vaccination programme (Programa Nacional de Vacinação: PNV) for both sexes at 15 months. In November 1990, a second dose of MMR, at 11 to 13 years of age, was added to the PNV. In October 1992, adverse reactions attributed to the mumps strain (Urabe Am9) led the Portuguese Health Ministry to replace the MMR used throughout the country with a MMR with the mumps vaccine strain Rubini. The first supply of the ‘new’ vaccine arrived in Oporto and was distributed to district health centres in October 1992.

After vaccination against mumps began in 1987, the numbers of cases reported in Portugal fell to a low of 627 in 1993. Thereafter, the numbers of reported cases increased each year until 30 000 cases were reported in a national epidemic in 1996-97. This epidemic in the presence of high vaccine coverage, suggested that the Rubini vaccine might be associated with widespread vaccine failures (1,2).

Mumps outbreaks in highly vaccinated populations have been described elsewhere as a result of poor vaccine efficacy (3). Outbreak studies that followed a resurgence of mumps in Switzerland suggested that vaccines incorporating the Urabe strain had greater efficacy than Rubini vaccines (4,5).

Only 13 cases of mumps were reported among residents of Oporto (approximately 276 000 inhabitants) in 1995. Early in 1996, however, routine reports suggested that a mumps epidemic was approaching the town from neighbouring regions. An incident case control study was conducted to test the hypothesis that the Urabe and Rubini vaccine strains had different protective efficacies during the anticipated epidemic of mumps in the town of Oporto.

Methods

Only children born after 1979, aged 15 months or more when they developed mumps, were included as cases. This was done to prevent bias against the vaccine because children under 15 months of age and those born before 1980 would not have been vaccinated (6). Cases that arose in 1995 or 1996 were selected from the notification files of the health authority. Notification forms included the diagnosis, date of onset, and whether the patient was admitted to hospital, but no details of signs and symptoms. Individual vaccination records were traced and reviewed in the health centres where the children were registered. Two consecutive vaccination records, corresponding to children of the same sex as the case and born in the same month and year, were selected as controls, whether or not they had already had mumps. This sampling scheme for controls was used so that the odds ratio for the exposure would yield an estimate of the relative risk (7,8).

Unfortunately, the batch of MMR given was seldom recorded in the vaccination files. For our study, it was assumed that those who were vaccinated before the first of November 1992 had received the Urabe strain while those vaccinated subsequently had received the Rubini strain. Very few children enrolled in the study were vaccinated close to that date, so there was little doubt about which strain had been received. As controls were matched with cases by sex and age, conditional logistic regression was used to calculate odds ratios and 95% confidence intervals (CI). Vaccine efficacy (VE) was calculated as: VE = (1 – odds ratio) x 100.

Thirteen cases of mumps were reported in 1995 and 267 in 1996. Most cases were reported by general practitioners (GPs) (46%) and hospital doctors (35%). Two hundred and thirty-eight reported cases met the selection criteria and written vaccination records could be traced for 189. The age and sex distributions of the 280 reported mumps cases and the 189 included in the study were similar (table 1). Six cases were confirmed serologically. Eleven cases were admitted to hospital, seven of them with meningitis; no deaths were reported.

Age Group

All reported cases

Cases selected for the case control study

Male

Female

Male

Female

<1

8

2

0

0

01 – 04

73

46

52

37

05 – 09

51

32

42

25

10 – 19

26

15

22

11

20 – 39

7

13

0

0

40 – 59

1

3

0

0

60 +

1

0

0

0

Total

167

111

116

73

Ninety-one per cent of cases and 93% of controls had been vaccinated, confirming the high coverage of MMR vaccine reported by the Oporto District Health Department (table 2).

The vaccine efficacy of the Rubini vaccine strain was estimated to be 1% (95% CI – 108%-53%) while the vaccine efficacy of the Urabe vaccine strain was estimated to be 70% (95% CI 25%-88%) (table 3).

Vaccine status

Cases

Controls

Vaccine efficacy*

95% CI

N

%

N

%

(%)

Urabe

56

29.7

142

37.6

70

25 - 88

Rubini

116

61.4

209

55.3

1

-108-53

Not vaccinated

17

8.9

27

7.1

Reference

TOTAL

189

100

378

100

Type of study

Vaccine strain

VE % (95%)

Case-control 1996, Oporto, Portugal

Urabe

70 ( 25 ; 88)

01 (-108 ; 53)

Retrospective 1995, Leiria, Portugal (9)

Urabe (a)

46 ( 24 ; 62)

Retrospective 1996, Switzerland (5)

Urabe

76 (36 ; 91)

12 (-102 ; 62)

65 (11 ; 86)

Case-control 1995, Switzerland (4)

Rubini

22 (-10 ; 45)

The point estimate of efficacy for the Rubini strain vaccine observed in our study was lower than that reported in Swiss studies (4,5), but all three studies had very wide 95% CIs (table 3). Our study confirms the difference in efficacy between vaccines containing different strains of mumps virus (4,9) (table 3). These results are consistent with poor seroconversion rates induced by the Rubini strain vaccine in Portugal in the 1990s (10).

We could not trace vaccination records for all cases, and it could be argued that selection bias was possible – for example, if study cases were more likely to have been vaccinated. If this was the case, vaccine efficacy (VE) would have been underestimated, a problem reported elsewhere (6). Our inability to trace all records resulted from the right of Portuguese citizens to access any health service wherever they reside. It is common for parents to have their children vaccinated outside their area of residence, but records of their vaccination would not be traced in their residence area.

Incompleteness of notification is not a problem for efficacy studies, but notifications of misclassified cases would lead to an underestimate of vaccine efficacy (6). We did not have data on the clinical characteristics of the reported cases, but mumps is generally an easy diagnosis to make clinically. Also, during a mumps epidemic, the positive predictive value of a diagnosis (and report) of mumps is likely to be higher than during an interepidemic period. We therefore believe that misclassification of mumps was unlikely to bias estimates of vaccine efficacy.

The vaccination records did not specify the MMR batch given but the dates of arrival of the batches with Rubini to Oporto were well documented. Thus, misclassification of the mumps vaccine strain is unlikely to have occurred.

The outbreak in Oporto was not due to insufficient vaccination coverage but to the low (or absent) protection induced by the Rubini vaccine strain. This and other studies, both from Portugal (1,2,9,10) and elsewhere (4,5), support the decision taken in June 1997 by the Portuguese Health Ministry to replace the MMR vaccine with the Rubini strain by another one with the Jeryl Lynn strain, which has been shown to be better (5). By January 1998, all Portuguese health centres had been given the new MMR vaccine.

Dr Áurea Almeida, Clotilde Moutinho, Delfina Antunes, Eduarda Couceiro, Hamilton Coutinho, Isabel Moita, Nilza Magalhães, and Teresa Saraiva, deputy health authorities of the municipality of Oporto, cooperated in this study.

We are also grateful to Dr Clarisse Teixeira and Mrs Gabriela from the Oporto health region.

1. Neto MC. Parotidite & VASPR: uma questão de vigilância epidemiológica. Saúde em Números 1997; 12: 20-4.

2. Dias JA, Cordeiro M, Afzal MA, Freitas MG, Morgado MR, Silva J, et al. Mumps epidemic in Portugal despite high vaccine coverage - preliminary report. Eurosurveillance 1996; 1: 25-8.

3. Briss PA, Fehrs LJ, Parker RA, Wright PF, Sannella EC, Hutcheson RH, et al. Sustained transmission of mumps in a highly vaccinated population: assessment of primary vaccine failure and wainning vaccine-induced immunity. J Infect Dis 1994; 169: 77-82.

4. Paccaud MF, Hazeghi P, Bourquin M, Maurer AM, Steiner CA, Seiler AJ, et al. Rückblick auf zwei Mumpsausbrüche. Soz Praventivmed 1995; 40: 72-9.

5. Toscani L, Batou M, Bouvier P, Schlaepfer A. Comparaison de l’efficacité de différentes souches de vaccin ourlien: une enquête en milieu scolaire. Soz Praventivmed 1996; 41: 341-7.

6. Clarkson JA, Fine PEM. An assessment of methods for routine local monitoring of vaccine efficacy, with particular reference to measles and pertussis. Epidem Inf 1987; 99: 485-99.

7. Smith PG, Rodrigues LC, Fine PEM. Assessment of the protective efficacy of vaccines against common diseases using case-control and cohort studies. Int J Epidemiol 1984; 13: 87-93.

8. Rodrigues LC, Kirkwood BR. Case-control designs in the study of common diseases: updates on the demise of the rare disease assumption and the choice of sampling scheme for controls. Int J Epidemiol 1990; 19: 205-13.

9. Passadouro R. Silva A. Mendes O, Lopes C. Investigação epidemiológica de um surto de parotidite epidémica no Concelho de Leiria. Saúde em Números 1995; 10: 25-8.

10. Gonçalves G, Tavares F, Andrade HR. Estudo de seroconversão pelo componente parotidite epidémica (Rubini) da VASPR, em adolescentes do Concelho de Sabrosa. Saúde em Números 1997; 12: 12-4.