Canada introduced universal pertussis vaccination in the 1940s. The first vaccine used was a whole cell pertussis fluid vaccine produced by two manufacturers: the Institut Armand Frappier, which supplied the province of Quebec; and Connaught Laboratories, which supplied all other provinces. Fluid vaccines were replaced in the 1980s by an adsorbed whole cell pertussis vaccine. Because Institut Armand Frappier never developed an adsorbed pertussis vaccine, Connaught Laboratories became the only Canadian supplier of the adsorbed vaccine. In 1997 to 1998, a five component acellular pertussis vaccine also manufactured by Connaught Laboratories (now Aventis Pasteur) became the only pertussis vaccine in use in Canada.
Beginning in 1990 Canada experienced substantial resurgence of pertussis. Quebec, the second largest province (population, 7.5 million) has been most affected and reported 40% of cases while representing 24% of the Canadian population. This resurgence was not attributed to unvaccinated children as vaccine coverage with ≥3 doses of pertussis vaccine in children has been consistently found to be >95%. 1–4 Studies found a low effectiveness of the Connaught Laboratories adsorbed whole cell vaccine which was estimated to range from 48 to 69%. 1–4 There have been no similar efficacy studies of the fluid vaccine used before the adsorbed vaccine, and it was impossible to conclude that the resurgence was a result of a drop in protection introduced by the poorly efficacious adsorbed vaccine. On a theoretical basis, should there be a sudden change in protection in the children population, one should anticipate a cohort effect. A cohort effect is present when a risk factor is present only in a part of the population delimited by its year of birth. In that situation the age distribution of the disease will parallel that of the members of this cohort, initially affecting younger children and progressively moving toward older age groups as the cohort ages. This cohort effect would be superimposed over the epidemiologic situation previously created by the more protective vaccine. With a more protective vaccine the incidence in younger age groups would be low and would progressively increase toward adolescence with the waning of immunity. 5, 6 This is what was observed in Massachusetts where they found low incidence in children <10 years of age with a subsequent higher incidence in those 10 to 19 years old. 7 This was also found in France where they used a highly protective whole cell vaccine. 8, 9
To assess the hypothesis that the resurgence in Quebec was caused by a cohort effect resulting from a poorly protective adsorbed whole cell vaccine introduced in 1985, we assessed the trends of pertussis incidence and hospitalization in the Province of Quebec between 1983 and 1998.
The source of data on pertussis cases was from notifications (confirmed and clinical cases) to the Quebec Ministry of Health and Social Services. Confirmed cases were cases with a cough and a positive culture for Bordetella pertussis or clinical cases with an epidemiologic link with a culture positive case. 10 Clinical cases were cases with a cough lasting for ≥2 weeks with one or more of the following symptoms: paroxysm; posttussive vomiting; apnea; or whoop. Cases were individually investigated by the regional public health units to ascertain that they met the case definition before being reported to the Quebec Ministry of Health and Social Services.
Hospitalization data were extracted from Med-ECHO, the computerized administrative database that collects information on each hospitalization in Quebec (fully implemented in 1983). The records for each hospitalization include age, sex, date of hospitalization, duration of stay and the discharge diagnoses (1 principal diagnosis and as many as 15 secondary diagnoses). These diagnoses are coded according to the ninth edition of the International Classification of Diseases. We extracted from Med-ECHO patients who had 033.0, 033.1 or 033.9 as their principal diagnosis or within the first 3 secondary diagnoses. The other diagnoses of these patients were considered pertussis complications. There has been no validation of this database.
The incidence rate and the hospitalization rates were calculated dividing the number of incident cases or hospitalized patients by the population. Population data were obtained from the Institut de la Statistique du Québec. These rates were estimated by age groups. Proportions were compared by chi square analysis.
Between 1983 and 1998, 25 890 cases of pertussis were reported of which 98 had no data for age. For the period 1990 to 1998 for which data on confirmation were available, 66% were clinical cases, 21% were laboratory-confirmed and 12% had an epidemiologic link with a laboratory-confirmed case. The mean annual number of cases changed during the period with a nearly 10-fold increase between 1983 and 1989 and 1990 and 1998 (Table 1). The corresponding rates during these periods increased from 3.8 to 37.2 per 100 000. The highest number of cases occurred in 1993, 1994, 1995 and 1998. These years included 69% (17 814) of all cases and incidence ranged from 59 to 66 per 100 000.
During the entire study period the highest incidence was in infants (Table 1;Fig. 1). The increase in incidence between 1983 and 1989 and 1990 and 1998 differed according to age (Table 1). Infants had the smallest increase when compared with children between 1 and 19 years and adults. The proportion of cases among adults increased steadily during the study period (Fig. 2). In contrast the proportion of cases age 0 to 4 years constantly decreased (Fig. 2).
Because case by case data for incidence were avail able only from 1990 to 1998, we compared the evolution of the age distribution of cases and that of the cohort vaccinated with Connaught adsorbed whole cell vaccine. The median age of cases increased from 4.4 to 7.8 years compared with 2.5 to 6.5 years for the cohort members. For that same period, plotting the percentage of all cases by year of age, a slow but steady shift to older age was observed (Fig. 3). In 1990 to 1993 the most affected age was <1 year, whereas it was age 3 and 4 years in 1994 to 1997 and ages 5, 6 and 7 years in 1998. In 1990 to 1993, there was a steady decline with age, children 0 to 4 and 5 to 9 years represented 53 and 29% of all cases. The proportion of cases was highest in children 6 years and younger with a rapid drop between age 7 and 11 years. Almost 5 years later, in 1998, the age of highest proportion of case spread from age 3 years to age 10 years with a decline now extending between 11 and 15 years. In 1998 children 5 to 9 years were the most affected groups with 35% of cases, infants <1 year represented only one-half their 1990 to 1993 proportion (6%vs. 12%) and children 0 to 4 years then represented 37% of all cases. The years 1994 to 1997 were intermediate years in the changes of epidemiologic trends between 1990 and 1993 and 1998. Between 1990 and 1998 the proportion of all cases doubled in adolescents (10 to 17 years) and adults (Fig. 3).
Because school attendance may increase transmission, we looked whether seasonality differed by age group and found small differences. Infants had no seasonal change in incidence, whereas in children age 1 to 4 years October was the month with the highest proportion of cases. The peak month for older children, adolescents and adults was November.
Overall 4655 patients were hospitalized, and 66% of hospitalizations occurred in infants (Table 1). Hospitalization rates between 1983 and 1989 were low and increased between 1990 and 1998 (Fig. 1, bottom right). The mean annual number of hospitalizations was 183 between 1983 and 1989 as compared with 375 between 1990 and 1998 for estimated rates of 2.7 and 5.2 per 100 000, respectively.
Among infants the hospitalization rate was highest in those age 3 to 5 months and lowest in those 9 to 11 months old. Although overall, hospitalization rates in infants age 6 to 8 months were similar to those in the 0- to 2-month age group (185 vs. 186 per 100 000), the rates in the older infants were generally higher in epidemic years with the reverse observed during “low” years (1987, 1989, 1991, 1996).
A complication occurred in 14% of hospitalized patients (Table 2). Otitis media represented one-half of all complications followed in decreasing frequency by pneumonia, sinusitis, severe vomiting and convulsion. One death was reported between 1990 and 1998 in an unvaccinated 3-month-old infant. Hospitalizations were slightly more frequent in girls than in boys (1.2:1), but not complications (13%vs. 15%).
Infants contributed to 77% of all hospitalization days. The median duration of stay was longer in infants than in older children (Table 2). The mean annual number of days of hospitalization increased by 41% between 1989 and 1989 and 1990 and 1998. Although the number of hospitalizations increased during the latter period, the median duration of stay decreased from 6 to 4 days (Table 2).
Comparison of incidence and hospitalization
Hospitalizations paralleled incidence for each age group (Fig. 1). However, in Fig. 1 the scales for incidence and hospitalization rates change with age: in infants the scales are identical; whereas the scale ratio (incidence/hospitalization) is 10:1 in children 1 to 4 years old, 40:1 in those 5 to 9 years old, and 50:1 in those older than 10 year of age. For the entire study period the number of reported cases per hospitalized case varied with age and was lowest in infants and highest in 10- to 19-year-olds (Table 1). Between 1983 and 1989 and 1990 and 1998, the hospitalization rates doubled as compared with a 10-fold increase in incidence rate. This happened because the increase in pertussis cases occurred across all ages whereas hospitalizations were mainly limited to infants <12 months of age (Table 1). The incidence:hospitalization ratio for each age group increased in all age groups with the greatest change in children 1 to 19 years old (Table 1).
Because there was no great change in hospitalization practices for pertussis during the study period, a change in the gap between hospitalization and incidence rates in Figure 1 indicates a change in either in diagnosis or reporting of incident cases or both. Underreporting in infants is demonstrated by the hospitalizations rates that exceeded the reported incidence in infants between 1983 and 1988 and became almost identical afterwards (Fig. 1, top left). Considering that not all infants affected by pertussis are hospitalized, underreporting continued after 1989 but to a lesser extent than before. In contrast, in children 1 to 4 years of age (Fig. 1, top right) and those 10 to 19 years of age (Fig. 1, middle right) and in adults, the changing gap between hospitalization rates and incidence rates is likely to be caused also by an improved diagnosis. Interestingly children 5 to 9 years old were those with the smallest change of the gap between these rates.
This study shows four major features of the epidemiology of pertussis in Quebec. (1) The incidence of pertussis increased substantially between 1990 and 1998 as compared with the previous period. (2) This increase did not affect all groups similarly. It progressed from younger children to older ones after a yearly pace corresponding to a cohort effect. (3) Although the incidence increased disproportionately in older children, adolescents and adults, hospitalization still occurred primarily in infants throughout period. (4) It is likely that there were diagnosis or surveillance (reporting) improvements as evidenced by the increasing ratio of incident cases/hospitalized cases observed in infants, in children 1 to 4 years of age and those 10 to 19 years of age.
The large increase in reported incidence of pertussis in Quebec is not likely to be an artifact when considering the magnitude of the change between the two periods, the occurrence of the same trends in other Canadian provinces and the increase in hospitalization in infants who are more likely to be properly investigated and diagnosed. Throughout the study period children <10 years remained the group with the highest incidence. This cannot be attributed to a low vaccine coverage as studies consistently found vaccine coverage with ≥3 doses to be >95%. 2–4, 11
The important resurgence of pertussis occurred 5 years after the replacement of the Institut Armand Frappier fluid vaccine by Connaught Laboratories adsorbed vaccine in 1985. There has been no direct comparison of the efficacy of the two vaccines and no efficacy studies of the Institut Armand Frappier fluid vaccine alone, but the Connaught Laboratories adsorbed vaccine was repetitively found to have a low effectiveness (48 to 69%). 1–4 Despite high vaccine coverage the continuous use of that vaccine left a large proportion of children born since 1985 vulnerable to pertussis. As the cohort vaccinated with this vaccine aged, the average and median age of cases became older, and the proportion of cases among older children increased, matching the theoretical epidemiologic features of a cohort effect. In 1998, 13 years after the introduction of the poorly protective vaccine, the plateau of higher incidence extended to 10 years of age and declined between 11 and 14 years of age (Fig. 3). The lower incidence in children between 11 and 13 years of age who were also vaccinated with Connaught adsorbed vaccine was likely caused by a reduction of their vulnerability compared with younger children caused by the greater cumulative exposure to pertussis since 1985. In British Columbia where Connaught Laboratories adsorbed vaccine was introduced in 1982, the epidemiology shows similar features. 12 The average and median age of cases increased during the 1990s, pertussis affecting progressively older and older children. In 2000 the age of peak incidence was 11 years and declined thereafter to level off at age 17 year years, corresponding to the children who first received this vaccine.
Hospitalizations were concentrated in younger children. Even if we did not performed a validation of the Med-ECHO database, it is likely to properly reflect the situation of the hospitalization morbidity attributable to pertussis given that most cases were found in infants and young children. Pneumonia and convulsions occurred in 3.4 and 0.5% of hospitalized patients, respectively. These rates were lower than the 9 and 2.3%, respectively, observed in Canadian pediatric tertiary care centers. 13 The sensitivity and specificity of the diagnosis in infants and young children are likely to be better than in adolescents and adults, because the laboratory confirmation by culture is generally easier in the former than in the latter. Thus it is likely that some adolescents and adults hospitalized for pertussis were never diagnosed. In spite of that and although there has been a 5- to 10-fold higher hospitalization rates in adolescents and adults after 1990 compared with the period before that year, they still contribute relatively little into the total hospitalization rates.
The increased incidence in older adolescents and adults is likely explained by factors other than the general increase in younger individuals. In the United States where the efficacy of the adsorbed vaccines was estimated to be >80%, 14 and the incidence in children younger than 5 years remained stable between 1990 and 1996, whereas it nearly doubled in 10- to 19- (1.7 to 3.4 per 100 000) and 20+-year-old (0.3 to 0.5) individuals. 15 In Massachusetts between 1989 and 1998, the incidence also remained unchanged in children younger than 10 years but increased almost 9-fold in those 10 to 19 years old (13 to 121 per 100 000) and 15-fold in adults (0.4 to 6 per 100 000). 7 In Quebec in 1998 the incidence for these age groups were very similar to the latter figures at 113 and 12.7 per 100 000, respectively. This suggests that the epidemiologic changes in Quebec in individuals too old to have received the Connaught Laboratories adsorbed vaccine are not explained by the increased transmission in younger age groups but mostly by the waning of whole cell vaccine immunity.
We thank Louis Rochette and Dr. Scott Halperin for their comments and support.
This study was supported by an unrestricted grant from Aventis Pasteur.
1. Halperin SA, Bortolussi R, MacLean D, Chisholm N. Persistence of pertussis in an immunized population: results of the Nova Scotia enhanced pertussis surveillance program. J Pediatr 1989; 115: 686–93.
2. De Serres G, Boulianne N, Duval B. Field effectiveness of erythromycin prophylaxis to prevent pertussis within families. Pediatr Infect Dis J 1995; 14: 969–75.
3. De Serres G, Boulianne N, Duval B, et al. Effectiveness of a whole cell pertussis vaccine
in child-care centers and schools. Pediatr Infect Dis J 1996; 15: 519–24.
4. Bentsi-Enchill AD, Halperin SA, Scott J, MacIsaac K, Duclos P. Estimates of the effectiveness of a whole-cell pertussis vaccine
from an outbreak in an immunized population. Vaccine 1997; 15: 301–6.
5. Jenkinson D. Duration of effectiveness of pertussis vaccine
: evidence from a 10 year community study. Br Med J 1988; 296: 612–14.
6. Lambert HP. Epidemiology of a small pertussis outbreak. Public Health Rep 1965; 80: 365–9.
7. Yih WK, Lett SM, des Vignes FN, et al. The increasing incidence of pertussis in Massachusetts adolescents and adults, 1989–1998. J Infect Dis 2000; 182: 1409–16.
8. Baron S, Njamkepo E, Grimprel E, et al. Epidemiology of pertussis in French hospitals in 1993 and 1994: thirty years after a routine use of vaccination. Pediatr Infect Dis J 1998; 17: 412–18.
9. Baron S, Grimprel E, Daurat G, et al. Estimation épidémiologique de l’efficacité de la vaccination anticoquelucheuse au cours d’épidémies en collectivité. Arch Pediatr 1997; 4: 744–50.
10. Statement on management of persons exposed to pertussis and pertussis outbreak control. Can Commun Dis Rep 1994; 20: 193–9.
11. Bentsi-Enchill A. National estimates of vaccination coverage at two years of age: Canada. Division of Immunization, Bureau of Infectious Diseases, Laboratory Centre for Disease Control. Ottawa: Health Canada, 1997.
12. Skowronski DM, De Serres G, MacDonald D, et al. The changing age and seasonal profile of pertussis. J Infect Dis 2002; 185: 1448–53.
13. Halperin SA, Wang EEL, Law B, et al. Epidemiological features of pertussis in hospitalized patients in Canada, 1991–1997: report of the immunization monitoring program-active (IMPACT). Clin Infect Dis 1999; 28: 1238–43.
14. Onorato IM, Wassilak SG, Meade B. Efficacy of whole-cell pertussis vaccine
in preschool children in the United States. JAMA 1992; 267: 2745–9.
15. Guris D, Strebel PM, Bardenheir B, et al. Changing epidemiology of pertussis in the United States: increasing reported incidence among adolescents and adults 1990–1996. Clin Infect Dis 1999; 28: 1230–7.