Changes in the epidemiology of hepatitis A outbreaks 13 years after the introduction of a mass vaccination program

Abstract

A hepatitis A+B vaccine vaccination program of 12-year-olds was introduced in Catalonia in 1998. The aim of this study was to investigate the evolution of hepatitis A outbreaks in Catalonia and estimate the preventable fraction of cases associated with outbreaks as a measure of the impact of the vaccination program. Hepatitis A outbreaks reported to the Health Department between 1991 and 2012 were analyzed. The incidence rates of outbreaks, outbreak-associated cases and hospitalizations were calculated. The preventable fraction (PF) and 95% confidence intervals (CI) were estimated for the whole study period (pre-vaccination and post-vaccination) and the post-vaccination period.

 

One-hundred-eight (108) outbreaks (rate of 2.21 per 10⁶ persons-year) were reported in the pre-vaccination period and 258 outbreaks (rate of 2.82 per 10⁶ persons-year) in the post-vaccination period. The rate of cases associated with outbreaks was 1.52 per 10⁵ persons-year in the pre-vaccination period and 1.28 per 10⁵ persons-year in the post-vaccination period. Hospitalization rates were 0.08 and 0.75 per 10⁶ persons-year, respectively. The number of person-to-person outbreaks whose index case was a school contact decreased in the post-vaccination period (aOR 2.72; 95%CI 1.35–5.48), but outbreaks whose index case was a man who has sex with men (MSM) or an immigrant increased. The PF of all outbreak-associated cases was 6.46% (95%CI 3.11–9.82) and the highest PF was in the 15–24 years age group (42.53%; 95%CI 29.30–55.75). In the 0–4 years age group, the PF was 18.35% (95%CI 9.59–27.11), suggesting a protective herd effect in unvaccinated subjects. Vaccination of immigrants traveling to endemic countries and MSM should be reinforced.

Keywords: :

• Hepatitis A
• outbreak
• mass immunization program
• preventable fraction
• incidence rate
• hospitalization

Introduction

Hepatitis A is an acute infection caused by the hepatitis A virus (HAV), an RNA virus of the Picornaviridae family. The virus can remain infectious in the environment for weeks and is transmitted person-to-person to non-immune individuals through the fecal-oral route and through ingestion of contaminated food or water. The disease has been seen as early as 15 d (days) and as late as 50 d after exposure; the mean incubation period is about a month, although this depends on the infectious dose.¹

HAV infection may produce a wide spectrum of outcomes, ranging from asymptomatic infection (without elevated serum aminotransferase levels) to subclinical forms (with elevated aminotransferase levels with no other clinical manifestation) or clinically evident disease.² The typical symptoms of HAV infection are indistinguishable from those caused by other forms of viral hepatitis. In older children and adults, HAV infection causes clinical disease with jaundice, but in young children jaundice is rare and most infections acquired are asymptomatic.³ A cholestatic form of hepatitis A has been reported in patients with persistent jaundice and itching. Other rare atypical clinical forms include immunologic, neurologic, hematologic, pancreatic and renal extrahepatic manifestations. The most severe form of the disease (fulminant hepatitis A) is rare and is associated with older age and underlying chronic liver disease.^1,2 In pregnant women with acute hepatitis A, premature rupture of membranes and placental separation have been reported.^4,5

In countries with low endemicity, most cases occur in specific risk groups, such as travelers to areas of high or intermediate endemicity, injection drug users and men who have sex with men (MSM).^6-9

In 1995, when inactivated HAV vaccines of proven immunogenicity and protective efficacy became available, a vaccination program of people belonging to risk groups was introduced in Catalonia, but the results showed that the impact of vaccination on the global incidence of the disease was small.¹⁰ Therefore, at the end of 1998, a universal program of vaccination of preadolescents aged 12 y with a combined hepatitis A+B vaccine containing 360 ELISA units of HAV antigen and 10 µg of hepatitis B virus surface antigen was initiated. This, together with improvements in hygienic, sanitary and socioeconomic circumstances has led to a situation of low endemicity in which infection is usually linked to outbreaks. The nature and number of outbreak-associated cases is an important public health burden, due to the need for case investigation, surveillance and control activities.¹¹ Therefore, it is of interest to investigate whether the vaccination program has involved changes in the epidemiology of outbreaks. According to the recommendations of the STROBE statement, the impact of the intervention should be estimated using absolute impact measures rather than only association measures.¹²

The aim of this study was to investigate the evolution of incidence rates of hepatitis A outbreaks, cases and hospitalizations for a period of 22 y, to compare the characteristics of the outbreaks in the pre- and post-vaccination periods and to estimate the preventable fraction (PF) of cases associated with outbreaks as a measure of the impact of the vaccination program.

Results

A total of 376 outbreaks [312 (83%) transmitted person to person, 36 (9.6%) transmitted through a common source, and 28 (7.5%) of unknown transmission mechanism] involving 1955 associated cases were reported between 1991 and 2012. The evolution of the incidence rates of outbreaks, associated cases and hospitalized cases is shown in Figure 1. Only hospitalization rate showed a significant increase (r = 0.66; P = 0.001).

Figure 1. Incidence rate of reported hepatitis A outbreaks, associated cases, and hospitalizations. Catalonia, 1991–2012

A total of 108 outbreaks (rate of 2.21 per 10⁶ persons-year; 95%CI 1.81–2.67) were reported in the pre-vaccination period and 258 (2.82 per 10⁶ persons-year; 95%CI 2.49–3.19) in the post-vaccination period. The number of cases associated with outbreaks in the pre-vaccination and post-vaccination periods were 746 (rate of 1.52 per 10⁵ persons-year; 95%CI 1.42–1.64) and 1165 (1.28 per 10⁵ persons-year; 95%CI 1.20–1.35), respectively, with the differences being statistically significant (P < 0.001).

The distribution of the number of person-to-person outbreaks according to the risk factor of the index case is shown in Table 1. In the bivariate analysis, the risk factors that fell significantly in the post-vaccination period were being a child attending a day care center (OR 3.07; 95%CI 1.20–7.84) and being a school contact of a case (OR 4.34; 95%CI 2.18–8.63), whereas outbreaks whose index case was an immigrant (OR 0.17; 95%CI 0.07–0.44) or a MSM (OR 0.04; 95%CI 0.002–0.74) increased. In the multivariate analysis, immigrant status (aOR 0.19; 95%CI 0.07–0.49), having a school contact with a case (aOR 2.72; 95%CI 1.35–5.48) and being a MSM (aOR 0.04; 95%CI 0.002–0.67) were the only statistically-significant risk factors.

Table 1. Person-to-person hepatitis A outbreaks according to the risk factor of the index case in the pre-vaccination (1991–98) and in the post-vaccination (2000–12) periods

	Number of outbreaks		OR* (95% CI)	P	aOR (95% CI)	P
	Period 1: 1991–98	Period 2: 2000–12
Day care (child)	10	9	3.07 (1.20–7.84)	0.019
Day care (adult)	1	7	0.36 (0.04–2.94)	0.338
Immigrant	5	58	0.17 (0.07–0.44)	< 0.001	0.19 (0.07–0.49)	0.005
Household contact	29	60	1.35 (0.79–2.31)	0.275
School contact	23	17	4.34 (2.18–8.63)	< 0.001	2.72 (1.35–5.48)	< 0.001
Travel	1	5	0.51 (0.06–4.37)	1
MSM	0	26	0.04 (0.002–0.74)	0.003	0.04 (0.002–0.67)	0.026
Others** and unknown***	17	37	1.21 (0.64–2.29)	0.555
All outbreaks	86	219	–

MSM, men who have sex with men; OR, Odds ratio; aOR, adjusted odds ratio; CI, confidence interval; *comparing each category of risk factor with all the others; **others, adopted children; ***unknown, outbreaks with unknown risk factors

The estimated incidence rates of outbreak-associated cases and the PF of all outbreak-associated cases in the periods considered are shown in Table 2. The PF of all cases was 6.46% (95%CI 3.11–9.82) and the highest value was in the 15–24 y age group (42.53%; 95%CI 29.30–55.75), but PF values higher than 14% were observed also in children aged < 15 y. The PF for women was 10.32% (95%CI 3.95–16.69), but no reduction was observed in men.

Table 2. Incidence rate and preventable fraction of hepatitis A cases associated with all reported outbreaks

Variable	Incidence rate* Period 1: 1991–98 (95%CI)	Incidence rate* Period 2: 2000–12 (95%CI)	Incidence rate* Period 1 + period 2 (95%CI)	Preventable fraction** (95%CI)
Age group (in years)
0–4	5.320 (4.403–6.371)	3.119 (2.637–3.664)	3.820 (3.375–4.308)	18.35 (9.59–27.11)
5–14	4.189 (3.666–4.766)	2.893 (2.543–3.278)	3.403 (3.104–3.723)	14.99 (7.59–22.38)
15–24	1.287 (1.049–1.562)	0.477 (0.353–0.630)	0.830 (0.702–0.973)	42.53 (29.30–55.75)
25–44	0.958 (0.805–1.131)	1.825 (1.676–1.982)	1.547 (1.434–1.666)
≥ 45	0.027 (0.009–0.062)	0.075 (0.050–0.109)	0.059 (0.041–0.083)
All	1.528 (1.420–1.641)	1.275 (1.202–1.350)	1.363 (1.302–1.425)	6.46 (3.11–9.82)
Sex
Men	1.503 (1.351–1.667)	1.551 (1.438–1.670)	1.534 (1.443–1.630)
Women	0.914 (0.799–1.041)	0.688 (0.615–0.768)	0.767 (0.704–0.834)	10.32 (3.95–16.69)

CI, confidence interval; *per 100,000 persons-year; **in percentages. Only calculated if a decrease was observed.

Table 3 shows the incidence rates and PF of outbreak-associated cases in the periods considered according to the risk factor of the index case. Being a school contact of a case was the risk factor with the highest PF (34.79%; 95%CI 11.29–59.30).

Table 3. Incidence rate and preventable fraction of hepatitis A cases associated with reported person-to person outbreaks according to risk factor of the index case

Risk factor	Incidence rate* Period 1: 1991–98 (95%CI)	Incidence rate* Period 2: 2000–12 (95%CI)	Incidence rate* gPeriod 1+Period 2 (95%CI)	Preventable fraction** (95%CI)
Household contact	0.059 (0.040–0.085)	0.066 (0.050–0.084)	0.063 (0.051–0.078)	-
Day-care contact	0.023 (0.011–0.040)	0.018 (0.010–0.028)	0.019 (0.013–0.028)	9.08 (< 0–37.52)
School contact	0.047 (0.030–0.071)	0.019 (0.011–0.030)	0.029 (0.020–0.039)	34.79 (11.29–59.30)
Immigrant	0.010 (0.003–0.024)	0.063 (0.048–0.082)	0.045 (0.035–0.057)
Travel	0.002 (0.000–0.011)	0.005 (0.002–0.013)	0.004 (0.002–0.009)
MSM	-	0.028 (0.019–0.042)	0.019 (0.012–0.027)

MSM, men who have sex with men; CI, confidence interval; *per 100,000 persons-year; **in percentages. Only calculated if a decrease was observed

Discussion

Our results show a significant increase in hospitalization rates of cases of HAV infection throughout the whole study period, as found by Arteaga Rodriguez et al.¹³ in a Spanish study using the minimum basic data set from the years 2005–2008. Studies in other countries have also found higher hospitalization rates in older people.^14,15 The increased number of cases adults, especially those aged >45 y, due to a shift to older ages after the introduction of the vaccination program, may explain these results.⁶

A study in the United States by Craig et al.¹¹ found a reduction in the incidence of outbreaks after the introduction of a mass vaccination program in states with high rates of HAV infection. In contrast, we found that although the incidence rates of outbreaks did not significantly change in the two periods considered, the incidence rate of cases associated with outbreaks fell in the post-vaccination period. This may reflect the fact that sensitivity to the reporting of outbreaks has increased (small size of reported outbreaks) and also that the mass vaccination of preadolescents is associated with a limited spread when outbreaks occur. Decreases in the incidence of HAV infection in the whole population (vaccinated and unvaccinated) have been reported by other authors after a mass vaccination program. Chironna et al.¹⁶ reported a reduction in the incidence rate of HAV infection from 14.8 per 100 000 in 1998 to 0.8 per 100 000 in 2009, after mass vaccination in Puglia (Italy). Wasley et al.¹⁷ assessed the impact of a routine vaccination strategy in some territories of the United States and found that the incidence rate fell from 12 per 100 000 in 1995 to 2.6 per 100 000 in 2003, with the reduction being observed not only in vaccinated age groups but in the whole population. In Argentina, following the introduction of a hepatitis A immunization program in 2005, a reduction of 88% with respect to the average incidence rate in the pre-vaccination period was observed.¹⁸

Only person-to-person outbreaks in which the risk factor of index case was being a school contact of a case showed a reduction in associated cases in the post-vaccination period (aOR 2.72; 95%CI 1.35–5.48). The most logical explanation is that the hepatitis A+B vaccination was performed in preadolescents in schools. Other authors have also reported greater reductions in vaccinated than in unvaccinated age groups.

In contrast, person-to person outbreaks in which the risk factor of the index case was immigrant status increased in the post-vaccination period (aOR 0.19; 95% CI 0.07–0.40), in agreement with other studies that show the importance of children of immigrants returning from visiting relatives and friends in their countries of origin as a cause of outbreaks.^19-23 Likewise, outbreaks in which the index case was a MSM increased (aOR 0.04; 95%CI 0.002–0.67) in the post-vaccination period. A large outbreak in this group, involving 189 cases, was reported by Tortajada et al.²⁴ between September 2008 and June 2009 in Barcelona. Hepatitis A vaccination has been recommended and offered free-of-charge to MSM in Catalonia since 1995, but unfortunately this strategy has not been successful, as has also occurred in other countries with low disease incidence.^25-27 Weerakon et al.²⁸ suggested that the level of immunization required to prevent outbreaks in MSM is about 50%, a coverage not reached in MSM in Catalonia due to the difficulty of contacting these persons and actively offering vaccination.²⁹ Our results support the need to reinforce vaccination in MSM.

Although the use of measures of absolute risk has been recommended in order to determine the potential impact of preventive measures,^12,30 not all reports use these measures to show the impact of factors associated with hepatitis A incidence rates.³¹ In our study, the PF of cases associated with outbreaks was estimated as >40% in people aged 15–24 y, but the PF was ≥15% in younger age groups. This suggests, in agreement with the results of other studies,^32,33 that the mass vaccination program has produced a protective herd effect, meaning that the disease incidence has decreased in an unimmunized segment of the population due to immunization of a proportion of the population,³⁴ such as the vaccination program of preadolescents in Catalonia. However, the results of studies in the United States,³³ Israel,³⁵ and Argentina¹⁸ with universal immunization programs in the second year of life suggest that the benefits for the whole population could be still higher in Catalonia if the age of universal vaccination were advanced to the second year of life, as infections in infants are frequently asymptomatic, and unvaccinated infants can act as a source of infection of their parents, caregivers and relatives.³⁶ The previously-mentioned increase in cases associated with MSM may explain why we observed a PF > 10% in women but no reduction in men.

According to the risk factor of the index case, the highest PF (34.79%; 95%CI 11.29–59.30) was found in outbreaks in which the index case was a school contact of a case. This seems logical because vaccination was administered in schools, and the coverage was >90%.³⁷ Some benefit was also observed in the prevention of cases associated with outbreaks whose index case was a day-care contact, reinforcing the previously-mentioned protective herd effect, although the lower CI of the PF was <0.

Our work has strengths and limitations. A main strength is that the study was based-population, with all reported outbreaks of hepatitis A in Catalonia during the study period being included, and the same definitions of cases and outbreaks was used. Another strength is the large number of years analyzed (8 y in the pre-vaccination period and 13 y in the post-vaccination period), which minimized the possible influence of disease cycles.

The limitations of the study include the fact that it is difficult to determine whether the decrease in incidence rates in the post-vaccination period was due only to the vaccination program or whether there are other environmental and hygienic factors that could contribute to this reduction. Therefore, the possibility that factors other than vaccination could explain some of the differences observed in the post vaccination period cannot be ruled out. Another possible limitation is underreporting during the study period. However, as the incidence rates of associated cases show, the size of outbreaks fell during the last years of the study and this suggests greater sensitivity in detecting hepatitis A cases and outbreaks in the post-vaccination period. Consequently, the decreases observed and the estimated PF are probably lower than the true figures.

In conclusion, our results show that the incidence rate of cases of HAV infection associated with outbreaks has decreased in the post-vaccination period in Catalonia, mainly due to a reduction in person-to-person outbreaks whose index case was a school contact. The preventable fractions attributed to the vaccination program show that the highest value was in the 15–24-y-age-group, which includes the cohorts vaccinated in schools, but a protective herd effect was also observed in children under this age. The benefit of universal vaccination would probably be greater if the age of administration of the vaccine was advanced to the second year of life. In addition to universal vaccination, our results show the need to reinforce the vaccination of immigrants who travel to endemic areas and of men who have sex with men.

Material and Methods

The study was performed in Catalonia, a region with 7.5 million inhabitants situated in the northeast of Spain.

Cases of hepatitis A must be statutorily reported to the Health Department in Catalonia. A clinical case of hepatitis A infection was defined as an acute illness with discrete onset of symptoms (malaise, nausea, anorexia, fever, malaise, or abdominal pain) and jaundice, dark urine or elevated aminotransferase levels. A hepatitis A outbreak was defined as ≥2 epidemiologically-linked cases with at least one laboratory-confirmed case (detection of immunoglobulin antibody to HAV). Outbreaks were classified as due to a common source or transmitted person to person.

For each reported hepatitis A outbreak to the Health Department between 1991 and 2012, the following variables were collected: year, number of cases associated with outbreaks, number of hospitalizations (cases associated with outbreaks that required hospitalization), and risk factor of the index case identified as the origin of the outbreak. Risk factors were defined as: living in the household of a case, being a school contact of a case, day-care attendance, day-care worker, immigrant status, travel to endemic areas, and being a MSM. When the index case had travelled to their or their parents’ native country and was also an immigrant, immigrant status was the risk factor considered. An immigrant was defined as a person who moved to Spain from another country and children born in Spain to non-Spanish nationals. The rate of outbreaks (per 10⁶ inhabitants), the rate of cases associated with outbreaks (per 10⁵ inhabitants) and the rate of hospitalizations (per 10⁶ inhabitants) and their 95% confidence intervals (CI) were calculated. The yearly population of Catalonia estimated by the National Institute of Statistics was used to calculate incidence rates.³⁸

Spearman’s correlation coefficients were calculated to evaluate the trend of rates. Statistical significance was established assuming an α error of 0.05. Crude odds ratios (OR), and their 95%CI were calculated by bivariate analysis comparing each category of risk factor with all the others. Adjusted odds ratios (aOR) were calculated by multiple logistic regression. We excluded the year 1999 from the analysis, as this was the year when preadolescents completed vaccination. The statistical analyses were made using the Statistical Package for Social Sciences (SPSS 19.0 for Windows) and R 3.0.3 (R Development Core Team 2014). To avoid the problem of quasi-complete separation, the method proposed by Firth³⁹ was used and implemented in R.⁴⁰ In order to determine the proportion of the disease that would be prevented if the vaccination program was implemented during the whole period and not only since 1999, we calculated the PF, which was estimated using the formula (Ip-Iv)/Ip,³⁰ where Ip was the incidence rate in the whole period considered (before the introduction of the vaccination program [1991–1998] and after the vaccination program [2000–2012]), and Iv the incidence in the period after the introduction of the vaccination program (2000–2012); the 95% CI of the PF were estimated using the method proposed by Szklo and Nieto.⁴¹

Abbreviations:
aOR	=	adjusted odds ratio
CI	=	confidence interval
d	=	days
HAV	=	hepatitis A virus
ELISA	=	enzyme-linked immunosorbent assay
MSM	=	men who have sex with men
OR	=	odds ratio
PF	=	preventable fraction
y	=	years

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgements

This work was partially funded by CIBER Epidemiología y Salud Pública (CIBERESP), Spain and by AGAUR (expedient number 2009/SGR 42).

We thank the reporting Physicians and the technicians of the Epidemiological Surveillance Units of the Public Health Agency of Catalonia and of the Epidemiology Service of the Public Health Agency of Barcelona.

The other members of the Working Group for the Study of Hepatitis A in Catalonia

M Alsedà, C Arias, I Barrabeig, N Camps, M Carol, M Company, J Ferràs, G Ferrús, I Parrón, A Rovira, R Torras (Public Health Agency of Catalonia); P Olalla, and J Caylà (Public Health Agency of Barcelona and CIBER Epidemiología y Salud Pública).