Time trends in pediatric hospitalizations for hepatitis A in Greece (1999–2013): Assessment of the impact of universal infant immunization in 2008

ABSTRACT

Hepatitis A vaccine was introduced in the Greek National Immunization Program in 2008. To estimate possible impact of the universal vaccination implementation, time trends of hospitalizations for hepatitis A at the Infectious Diseases Unit of a Tertiary Pediatric Hospital in Athens during 1999–2013 were analyzed. Hepatitis A hospitalizations were recorded from the discharge database and were expressed as frequencies and rate of annual departmental hospitalizations. Time series analysis (ARIMA) was used to explore trends and the impact of the vaccination. Moreover, changes in patient age, population group distribution and the duration of hospitalization were also examined. Hepatitis A hospitalizations rate significantly decreased between pre-vaccination (1999–2008) and post-vaccination (2009–2013) era from 50.5 to 20.8/1000 hospitalizations (p = 0.005). A 3-year periodicity and a trend of reduction on hepatitis A hospitalizations rates across years were noted. Roma children had significant higher rates of hepatitis A hospitalization, followed by immigrant children. Importantly, possibly due to preceding vaccine availability with considerable uptake in private market and unvaccinated group/pockets of children (Roma), overall vaccination effect was less apparent when compared to data from other countries that implemented universal vaccination. No significant change in patient age, population group distribution, or duration of hospitalization was observed. High risk groups such as Roma children should be targeted for vaccination to reduce future outbreaks.

KEYWORDS:

• hepatitis A
• hospitalization
• Roma
• time trends
• universal vaccination

Introduction

Hepatitis A is a vaccine preventable disease still prevalent in many parts of the world.¹ Although the infection is usually asymptomatic in young children, it may cause severe morbidity and mortality in adolescents and adults.² Hepatitis A vaccine is safe and effective. It has been licensed for use and available in the private market in developed world for almost 2 decades. In countries where universal Hepatitis A vaccination of children has been implemented, a significant reduction of hepatitis A incidence has been observed among vaccinated children.^3-5 Importantly, decreased incidence among all other unvaccinated age groups was noted due to herd immunity.^3-6 In Greece, hepatitis A vaccine has been available in the market since 1999 and was introduced in the National Immunization Program in 2008.⁷ Between 1999 and 2008, children as well as adults could get immunized in private clinics but needed to pay out of their pockets to purchase the vaccine. Therefore, those children most at need of vaccination, namely immigrant and Roma children had lower immunization coverage.⁸ To date there had been no reports on the incidence of acute hepatitis A in Greece. Previously, a small seroprevalence study including children 5–15 y old reported that 98.3% of Roma children had anti-HAV antibodies.⁹ Additionally, an outbreak involving mainly unvaccinated Roma children occurred in 2013.¹⁰ Although morbidity is higher among adolescents and adults, we chose to describe hospital hospitalizations for hepatitis A at the Infectious Diseases unit of the “Agia Sophia” Children's Hospital, to focus on vaccinated group and delineate the trend of the disease in this specific group. Moreover, this is a tertiary Pediatric Hospital in Athens, Greece where all pediatric communicable cases are referred.¹¹ The aim of the study was to assess trends of hospitalization rates for hepatitis A post universal vaccination was implemented in order to assess the impact of the universal vaccination program.

Results

Among 10,866 hospitalizations at the Infectious Diseases Unit during the study period, 43.0/1000 patients (95% CI: 39.2–46.8) were diagnosed with acute hepatitis A. The hepatitis A admission rate decreased by almost 76% during the period 1999–2013 (77.3/1000 hospitalizations to 18.5/1000 hospitalizations) (Table 1). Hepatitis A hospitalization numbers and rates were reduced post universal immunization implementation (p < 0.05, Table 2). Although data on vaccination status was not systematically collected during this retrospective study, authors affirmatively report that no child with acute hepatitis A had evidence of receipt of any hepatitis A vaccine dose.

Table 1. Number and rates of hepatitis A hospitalization cases per year.

Year	Total hepA	Departmental admissions	Rate per 1000 admissions	95% CI
1999	61	789	77.3	58.7–95.9
2000	44	853	51.6	36.7–66.4
2001	70	1043	67.1	51.9–82.3
2002	62	831	74.6	56.7–92.5
2003	13	623	20.9	9.6–32.1
2004	21	718	29.2	16.9–41.6
2005	44	797	55.2	39.4–71.1
2006	25	796	31.4	19.3–43.5
2007	22	757	29.1	17.1–41.0
2008	28	566	49.5	31.6–67.3
2009	20	665	30.1	17.1–43.1
2010	13	625	20.8	9.6–32.0
2011	21	584	36	20.9–51.1
2012	11	572	19.2	8.0–30.5
2013	12	647	18.5	8.2–28.9
Total	467	10866	43	39.2–46.8

Table 2. Comparison of hepatitis A hospital admissions between pre- and post- universal vaccination era.

		median (IQR) or % (95% CI)
		Total	Pre-vaccination 1999–2008	Post-vaccination 2009–2013	p-value
Hospitalizations		22.0 (13.0–44.0)	36.0 (22.0–61.0)	13.0 (12.0–20.0)	0.005^*
Hospitalization rate		31.4 (20.9–55.2)	50.5 (29.2–67.1)	20.8 (19.2–30.1)	0.028^*
Hospitalization duration in days		5.0 (3.0–6.0)	5.0 (3.0–6.0)	4.0 (3.0–6.0)	0.079^*
Age distribution of HepA cases	<6 y old	56.6	58.0	54.5	0.635^**
	≥ 6 y old	43.4	42.0	45.5
Population group distribution of HepA cases	Greek	33.6	34.8	27.3	0.355^**
	Immigrant	29.6	28.5	35.1
	Roma	36.8	36.7	37.6

* Mann-Whitney test

** Chi-square test

IQR: interquartile range

By considering the different population groups of children hospitalized for hepatitis A during the study period, a decrease of hepatitis A hospitalization rates of 52%, 69.2% and 76.7% is observed among Greek, immigrant and Roma children respectively (Table 3). No significant changes in the population group distribution of our cohort were observed between pre- and post- vaccination era (Table 2). However, when hospitalization rates within different population group were assessed, significant higher rates among Roma children were noted (p < 0.001, Fig. 1).

Figure 1. Hepatitis A hospitalization rates by population group.

Table 3. Hepatitis A hospitalization rates per 1000 admissions by nationality/race: changes between pre- and post- universal vaccination period.

Population group of hospitalized children	Pre-vaccination 1999–2008	Post-vaccination 2009–2013	p-value ^*
	median (IQR)	median (IQR)
Greek	25.6 (22.2–28.3)	12.3 (10.2–13.2)	0.099
Immigrant	57.1 (36.9–85.4)	17.6 (15.5–27.0)	0.019
Roma	233.1 (93.8–410.3)	54.3 (47.1–100.0)	0.080

* Mann-Whitney test

Roma children had the shortest duration of hospitalization (mean of 5 days, p = 0.011). There was no significant change in the hospitalization duration after the implementation of universal vaccination (Table 2). There was no change in the age distribution of hepatitis A cases (Table 2).

ARIMA analysis reveals a periodicity every 3 y (p < 0.001) and a trend of reduction on Hepatitis A rates across years (p = 0.01). Interrupted time series analysis reveals no significant effect of introduction of hepatitis A vaccination in 2008 on admission rates (p = 0.222). Additionally, although a statistically significant negative relationship between GDP per capita and Hepatitis A admissions rate (p=0.029) was found, when intervention was also included in the analysis, the GDP per capita marginally was not found to be statistically significant associated with the admissions rate (p = 0.070).

Discussion

The retrospective analysis suggests that hepatitis A hospitalization rate among children 0–14 y old in Athens, Greece has been significantly reduced during the study period (1999–2013). Moreover, when comparing the hospitalization frequency and rates during pre- and post- universal vaccination era, a significant impact of the introduction of universal hepatitis A vaccination in 2008 could be identified. The year of introduction of hepatitis A in the National Immunization Program (2008) is included since vaccination of children had been initiated through the private market since 1999. Importantly, although a prevailing temporal trend of reduction of hepatitis A hospital admissions post universal vaccination was observed, it mainly affected immigrant and Roma children (Table 3, Fig. 1). This in contrast to the epidemiologic changes observed in other countries where universal hepatitis A vaccination implementation resulted in up to 98% incidence reduction and significant prevention of outbreak associated cases.^4-5,12,13 This finding could be explained in various ways. Firstly, a significant periodicity of the hepatitis A associated hospitalizations was observed impeding the time series analysis to reveal a significant effect post vaccination.¹⁴ Hospitalization rate due to hepatitis A showed a periodicity, mainly apparent among Roma children (Fig. 1). Almost 2 out of 10 Roma children hospitalized during the pre-universal vaccination period (1999–2008) were admitted because of acute hepatitis A. This was further investigated and was mainly due to an outbreak in 2001–2002 during which almost 50% of hospitalized Roma children were acute hepatitis A cases. Furthermore, possibly due to the lower vaccination coverage rates among immigrant children before the implementation of universal vaccination, transmission during outbreaks was enabled and therefore a periodicity among this group was also observed.⁸

Most importantly, universal vaccination was implemented when already at least a third of children on a national level had been vaccinated according to a study including children visiting outpatient department of public hospitals.⁸ Evidently, this most likely reduced the effect of the national program. Of note, the vaccine uptake was even greater (> 50%) in Athens metropolitan area where this study took place.⁸ One may postulate that the vaccination coverage in the former study was underestimated, since children visiting public hospitals for their medical care are less likely to be vaccinated with a vaccine without reimbursement. This is supported by the findings of a national vaccination coverage study in 2012 (4 y post incorporation of hepatitis A vaccine in our NIP) that indeed has indicated that 69% of hepatitis A vaccinations have been administered by private pediatricians.¹⁵ According to the same study, vaccination coverage with HAV vaccine is quite high among children 6 y old, with 88% and 82% vaccine receipt of 1 and with 2 doses respectively.¹⁵ Greece does not have a national vaccination registry and therefore information on vaccine coverage and timely administration is gathered from published studies. Vaccine coverage studies in children 1–3 y old, have indicated that although hepatitis A vaccine is well accepted by both pediatricians and parents, it is often administered just before entering kindergarten rather then at 12 months of age as recommended by our National Immunization Program.^{16, 17} This is possibly due to the overcrowded vaccination schedule during the first 24 months of life.

It should be stressed however, that recent epidemiologic data from Greece indicate that more than 5 y post implementation of universal vaccination against hepatitis A, there are still outbreaks mainly affecting Roma population.¹⁰ More importantly, in this overview of the recent epidemiology of hepatitis A in Greece, it was noted that after the implementation of vaccination in 2008 an increase of the mean age of hepatitis A cases was observed (from 21.2 y to 30.6 y).¹⁰ This data questions the cost effectiveness of universal immunization program in Greece and this is most important in view of the severe economic crisis. It should be noted that in 2008, when universal vaccination against hepatitis A was initiated, Greece did not fulfill the WHO criteria.¹⁸ WHO proposes that universal vaccination is introduced in countries where a change in the endemicity from high to intermediate is observed and with the consideration of cost-effectiveness.¹⁸ However, according to national data reported by ECDC, the disease incidence was lower than the average of EU (1.07/100,000 habitants) indicating that Greece was a country of low endemicity.¹⁹

In support of the benefit of immunization to protect individuals from hepatitis A however, one may note that in the 2013 outbreak among Roma children, cases have been geographically restricted and extended community outbreaks have been avoided. This is possibly due to the increased vaccine coverage in the general pediatric community.¹⁵ This is in contrast to data from 2004–2012 where a geographical correlation of the occurrence of cases among Roma with the occurrence of community cases in the general population was identified.¹⁰ The role of Roma in the transmission of vaccine-preventable diseases in the community has been documented in the context of recent outbreaks in Greece and in other countries ^20,21.

Roma population accounts for about 1.5% of total population residing in Greece. During the 2012 national vaccination coverage study, a sub-study on the Roma population was included, indicated that Roma children remain significantly under-vaccinated with only 25% and 13% of them having received 1 and 2 doses of hepatitis A vaccine respectively.¹⁵ According to Puglia study, at least 45% of vaccination coverage is needed in order to interrupt person-to-person transmission of hepatitis A infection.²² Roma children are a difficult group to target due to travel and parental beliefs. Ongoing outbreaks in 2013 were mainly attributed to unvaccinated Roma children living under low socioeconomic conditions and inadequate sewage systems.¹⁰ Although targeting a high risk population alone, such as Roma children, may be cost effective, it is well known that reaching these populations may be quite challenging.

Conversely, one may argue that the observed decline of hospitalization rates is not only due to the implementation of universal vaccination but also attributed to the improved sanitation and living conditions as indicated by the increased GDP per capita. This would indeed apply to Roma children who have low vaccination coverage. However, it should be noted that since 2009 the GDP per capita is decreasing in Greece.

The most important limitation of this study is that data presented refer to a single tertiary pediatric center. Therefore data presented might not be representative of the general population. Also, data on social demographics which are important when discussing hepatitis A epidemiology, were not collected. As aforementioned, although vaccination data of cases was not available, authors affirmatively confirm that all children hospitalized with acute hepatitis A were unvaccinated. Most importantly it should be noted that assessing the impact of universal vaccination on hepatitis A epidemiology using the hospitalization rates among children is possibly misleading since it is well documented that children are often asymptomatic. Alternatively, using national laboratory data and examining the incidence of anti-HAV-IgM detection could provide stronger evidence. Data from general practitioners would be extremely difficult to gather in Greece since clinical practitioners have not been accustomed to report. These important limitations indicate that since the results of this study imply that additional targeted measures need to be undertaken, a national study using laboratory data might be necessary.

Conclusions

In conclusion, the retrospective analysis of hepatitis A attributed hospitalization rate indicated a temporal trend of a reduction due to the vaccine implementation. However, vaccination of Greek children was initiated by private pediatricians as vaccines became available in 1999, before its incorporation in the NIP. Therefore, universal infant hepatitis A vaccination mainly had an impact on immigrant and Roma children. In view of the economic crisis and need of prioritization of resourses allocated to public health, the need of additional measures is concerning. Targeting high risk groups such as Roma children is challenging, but education and improving living conditions may decrease hepatitis A outbreaks among this population.

Methods

A retrospective study was performed at the aforementioned Infectious Diseases Unit between 1999–2013. All hospitalized children 0–14 y of age were included. According to the census 2011, the hospital covers about 650,000 children aged 0–14 y, which is about 41% of the entire Greek population (0–14 y old).⁷ The total number of hospitalizations at the infectious diseases unit was obtained. Pediatric hepatitis A hospitalizations were noted from the discharge log and were expressed as frequencies and rate of annual departmental hospital hospitalizations (per 1000 hospitalizations). Hepatitis A diagnosis was made on the basis of consistent clinical presentation and serologic confirmation by the detection of anti-HAV IgM antibodies (ICD-10 B15.9). Age of children admitted for acute hepatitis A as well as duration of hospitalization (in days) were noted. Importantly, to be able to describe potential changes in the distribution of different population groups during study period, population group of total hospitalizations and hepatitis A cases were noted. The Gross Domestic Product (GDP) per capita for the years 1999–2013 was collected from the Hellenic Statistical Authority. Quantitative data are presented either as median with interquartile range (IQR) or as mean with standard deviation. Qualitative data are presented as frequencies with percentages or with 95% confidence intervals (95% CI). Interrupted time series analysis [ARIMA (0, 0, 0) modeling procedure] was used to explore the trends and the impact of the vaccination and GDP per capita on the yearly hepatitis A hospitalizations rate. Non-parametric test, Mann-Whitney test, for non-normal distributed data, was used to assess possible vaccination impact comparing hospitalization rates, hospitalization duration, age and nationality distribution between pre- (1999–2008) and post- universal vaccination era (2009–2013). Chi-square test was used to explore associations between qualitative variables. Analyses were undertaken with the SAS 9.0, using a p< 0.05 as the criterion of significance.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

Special thanks to the Wellcome Trust for sponsoring the participation of Dr. Zoi Alexopoulou.