Article Text
Abstract
Objectives The selective vaccination programme against hepatitis B virus (HBV) was introduced in the Netherlands in 2002 targeting high-risk groups, including men who have sex with men (MSM). Despite the high average age of vaccination in MSM, the number of notifications of acute HBV recently declined. We investigate whether this can be attributed to the selective vaccination programme. We examine how vaccination strategies could be improved and the impact of universal infant vaccination introduced in 2011.
Methods We use a mathematical model for HBV transmission among MSM. The incidence of HBV was calculated from the model and from notification data of acute HBV.
Results A decline was observed in the incidence of HBV since 2006, as calculated from the model; this decline was smaller than that observed in data if all MSM were equally likely to be vaccinated. Assuming that high-risk MSM were more likely to be vaccinated than low-risk MSM resulted in a steeper decline in modelled incidence and better agreement with observed incidence. Vaccinating MSM at a younger age or doubling the vaccination rate would increase the impact of selective vaccination, but is less effective than vaccinating high-risk MSM.
Conclusions Selective HBV vaccination of MSM in the Netherlands has had a substantial impact in reducing HBV incidence. The reduction suggests that vaccination rates among high-risk MSM were higher than those among low-risk MSM. Countries that have not yet reached 35-year cohorts with universal childhood vaccination should actively implement or continue selective high-risk MSM vaccination.
- Hepatitis B
- Vaccination
- Mathematical Model
- Men
Statistics from Altmetric.com
Introduction
Until recently, universal childhood vaccination against hepatitis B virus (HBV) infection was not considered cost-effective in the Netherlands, mainly due to the low incidence of HBV in the general population.1 Therefore, in 2002, a selective vaccination programme was initiated targeting behavioural risk groups with high HBV incidence, including men who have sex with men (MSM), commercial sex workers, drug users and heterosexuals with high rates of partner change. New economic analyses suggested that universal vaccination of infants would be cost-effective to control hepatitis B in the Netherlands;2 since September 2011, all newborns are offered vaccination against HBV in the first year after birth. However, only infants are directly protected by universal vaccination in the beginning of its implementation; therefore, selective vaccination should be continued until the risk groups are protected against HBV.3
Intense efforts are invested in order to reach the individuals in the selected risk groups. It is therefore important to assess the impact of the programme and to explore adaptations that could further improve it. To do that, we used a mathematical model.4–6 Since sexual transmission among MSM is the most important transmission route in the Netherlands and the one with significant ongoing transmission,7–9 this study was focused only on transmission via sexual contacts among MSM.
An earlier modelling study among MSM in Amsterdam indicated that with the current vaccination rate, it may take rather long until substantial reductions in HBV incidence could be observed.10 However, since 2006, a considerable decline has been observed in the notifications of acute HBV infection among MSM,11 pointing to a reduction in HBV incidence. It is unclear whether this decline was a result of the vaccination programme or of other factors, such as a decline in sexual risk behaviour. Here we attempt to explain this decline by comparing the incidence of HBV infection as calculated from the notification data of acute HBV with the incidence calculated from the model. The impact of selective vaccination on reducing HBV incidence is evaluated. Special scenarios are investigated to assess how the current vaccination programme could be improved, for example, by changes in age at vaccination or changes in the vaccination coverage in groups with high sexual risk behaviour. Moreover, we study the impact of universal vaccination implemented in the Netherlands in 2011.
Methods
The model
The model was based on the model of Williams et al,4 as adapted by Kretzschmar et al (see online supplementary text and table A1)5 ,6 Briefly, the modelled population is structured by age and is divided into six sexual activity groups. For simplicity, these are called groups 1, 2, 3, 4, 5 and 6, with increasing sexual risk behaviour: group 1 has the lowest partner change rate and group 6 has the highest partner change rate (see online supplementary tables A2 and A3). Mixing between risk and age groups is proportionate, that is, random mixing weighted by sexual activity and size of the risk groups. For every risk group there is a different level of partner change rate and this varies also with age. After infection, there is a latent stage (not infectious), followed by acute HBV infection. A fraction of those with acute infection become carriers (chronic HBV) and the remaining recover and are immune to infection.12 The model is described by a set of partial differential equations in time and age, given in the online supplementary appendix. The equations were solved numerically without vaccination, until the model reached the endemic steady state. Then vaccination was introduced and the calculations were carried out for 50 years.
Parameter values
Data from the first 7 years (November 2002–2009) of the National HBV Vaccination Programme for MSM in the Netherlands were used to estimate the current vaccination rate (see online supplementary tables A4 and A7). Estimates of sexual partner change rates were obtained from data from Rutgers World Population Foundation (WPF)13 and were adjusted to calibrate the model to estimates of incidence and the age distribution of incident infections from the national data on notifications of acute HBV infection14 (see online supplementary appendix for details). The other parameters are described in detail in the online supplementary appendix.
Calculation of incidence from notification data
In order to calculate the incidence of HBV infection from the number of notifications of acute HBV infection, it was assumed that a third of the cases is clinical15 and that there is 50% under-reporting.16 This means that the number of notifications of acute HBV should be multiplied by a factor of 6 and then divided by the total size of the MSM population to provide the estimate of incidence:
The total number of MSM in the Netherlands was estimated to be in the range 278 000–392 000.13 Using this interval, the incidence from the notification data in a specific year was estimated to be between 6*notifications/392 000 and 6*notifications/278 000. This interval is shown in figure 1 with the grey shaded area.
Model scenarios
In the Results section, different scenarios are presented for the current situation in the Netherlands and how that could be improved. These scenarios were implemented as follows.
Current vaccination rates equally distributed among risk groups: The annual vaccination rate was calculated from data from the National HBV Vaccination Programme (see online supplementary table A4). The number of vaccinations administered was assumed to be equally distributed among all age and all sexual risk groups in the MSM population.
Only high-risk MSM are vaccinated: For these scenarios, it was assumed that the vaccination rates are higher among high-risk MSM. This could be true because high-risk MSM are aware of their higher risk to get infected and that could make them more interested in receiving vaccination, but also because the programme recruited participants at outreach locations, where high-risk behaviour is frequently seen. The total number of vaccinations administered was as calculated from the National HBV Vaccination Programme, but it was not equally divided among the sexual risk groups: only men from the four highest risk groups 6, 5, 4 and 3 (19.6% of the population) were vaccinated; or only men from the three highest risk groups 6, 5 and 4 (7.1% of the population) were vaccinated; or only men from risk groups 6, 5 and half of group 4 (4.1% of the population) were vaccinated (see online supplementary table A6 for the vaccination rates in these scenarios).
Decline in sexual risk behaviour: To examine whether the observed decline in reported acute HBV cases since 2006 could be explained by reductions in sexual risk behaviour, a specific scenario was examined with a reduction in the partner change rates of 20% or 50% taking place in 2005.
Getting vaccinated 5 years younger: In 2009, the average age of MSM at vaccination was 35 years (data from the National HBV Vaccination Programme), while the average age at infection was 42 years.7 To evaluate the impact of vaccinating MSM at a younger age, we recalculated the current vaccination rates assuming that the same numbers of vaccinations were administered to men 5 years younger than what was reported at the National HBV Vaccination Programme (see online supplementary table A6).
Double vaccinations: To assess the impact of scaling up the vaccination programme, a hypothetical scenario was calculated with twice as many vaccinations administered as those reported at the National HBV Vaccination Programme.
Universal vaccination since 2011: In this scenario, universal vaccination was introduced in 2011; all infants are vaccinated in the first 2 years of life.
The last four scenarios were examined with equal vaccination rates among all risk groups and with higher vaccination rates among high-risk men (only groups 6, 5 and 4 vaccinated).
Results
Prevaccination
From the model, the basic reproduction number was calculated at R0=1.6 and the incidence without vaccination was 190 new HBV infections per 100 000 per year. In 25% of new HBV infections, the partner transmitting HBV had acute HBV, while in 75% he had chronic HBV. The prevalence of immunes was 12% and the prevalence of carriers was 1% among MSM aged 15–60 years.
What caused the observed decline in reports of acute HBV infection?
The incidence from the notification data was considerably reduced in 2006 and continued to decline thereafter (figure 1, grey shaded area). To determine the factors that could explain this decline, we compared the incidence from the data with the incidence of HBV obtained from the model under different scenarios. In the model, assuming that all MSM in the population were equally likely to be vaccinated resulted in a moderate decline in incidence, much smaller than the decline observed in the notification data. The scenarios with only high-risk MSM being vaccinated were in better agreement with the data, in particular the scenarios where only the three highest risk groups were vaccinated (figure 1A). Figure 1B shows that a moderate decline of 20% in 2005 in sexual risk behaviour could also explain the sudden decline in the notifications of acute HBV after 2005; however, a sharp reduction in risk behaviour is not expected to be the underlying factor, as that would have resulted in a much steeper decline in incidence. Similarly, assuming that only high-risk MSM were vaccinated and that sexual risk behaviour was reduced resulted in a steeper decline than the decline observed in the notification data (figure 1B).
What is the impact of the current vaccination programme?
Figure 2 shows the impact of HBV vaccination for the first 50 years after its introduction. The impact is measured by the percentage decline in HBV incidence compared with the endemic state without vaccination. Although the decline in incidence is relatively small in the first years, vaccination can result in considerable reductions in HBV incidence, even with the relatively low vaccination rates that have been achieved thus far in the Netherlands. Assuming that vaccinations are equally distributed among all risk groups, the incidence of HBV will be reduced by 24% 20 years after the start of vaccination and by 30% 50 years after the start of vaccination. HBV incidence will be reduced by 57%–60% 20 years after the start of the vaccination programme if only high-risk men are vaccinated and by 44% if sexual risk behaviour was reduced by 20% in 2005. With universal vaccination the reduction in HBV incidence is by far the greatest among the scenarios examined here. However, this decline occurs at least 20 years after the introduction of universal vaccination, after the infants currently vaccinated under this programme become sexually active and the inflow of young men into the sexually active population consists of individuals who are (almost all) vaccinated against HBV.
Can we make the HBV vaccination programme more effective?
To investigate how the vaccination programme could become more effective, we investigated two additional hypothetical scenarios where the number of vaccinations administered is doubled or the same number of vaccinations is administered to men 5 years younger than what is currently the practice (figure 3). By vaccinating MSM at a younger age or by vaccinating more MSM, the impact of selective vaccination will be greater: HBV incidence is reduced more in these two scenarios than in the scenario with the current vaccination rates (with both equal and unequal vaccination rates among risk groups). It is also interesting to note that if high-risk MSM are mostly vaccinated (figure 3B), the scenarios with vaccination at a younger age or double vaccinations bring the impact of selective vaccination closer to that of universal vaccination. However, this is a combined effect of vaccinating high-risk MSM and vaccinating them at a younger age: comparing the two plots in figure 3 shows that with a specific number of vaccinations administered mostly to the high-risk men of the population, the reduction in HBV incidence (figure 3B) is much higher than the reduction that would be achieved with twice as many vaccinations equally distributed among all men in the population (figure 3A). Looking at the long-term impact of these programmes (after 2035), universal vaccination is the best strategy achieving the highest decrease in HBV incidence.
Discussion
We show that the decline in the notifications of acute HBV infection in the Netherlands in the last 6 years can be most likely attributed to the selective vaccination of MSM, which has had a substantial impact in reducing HBV incidence within this population. By comparing our model results with data, it seems that this decline was caused by higher vaccination rates among high-risk MSM than the rates among low-risk MSM. With the universal vaccination of all infants introduced in 2011, the incidence of HBV infection will be further reduced considerably. However, this will happen at least 20 years after the introduction of universal vaccination when a considerable fraction of the sexually active population is protected against HBV infection.
In the Netherlands, it has been proposed that the impact of selective vaccination could be increased by vaccinating MSM at a younger age. Preliminary data from the National HBV Vaccination Programme indicate that this has not been achieved despite considerable efforts. The difficulty to vaccinate MSM at a younger age has been noticed in other countries as well.17–20 Internet sites and social media provide nowadays new potentials for targeting specific population groups, especially among young adults. However, based on this modelling study, we suggest that the vaccination programme should focus on vaccinating at a younger age, and on increasing even more the coverage of the programme among the more sexually active MSM. However, it has to be mentioned that the programme recruits men at outreach locations that high-risk MSM frequently visit. Therefore, reaching even more high-risk MSM may be difficult to implement. Also, it is important to notice that identification of high-risk MSM may not be that straightforward. Recent studies point to higher sexual risk behaviour among younger MSM and among those finding their partners via internet or in darkrooms.21 However, there is no clear demographic grouping of MSM into high- and low-risk MSM.
The results presented here are in accordance with results from other studies. Among MSM 15–60-years-old, the prevalence of immunes was 12% and the prevalence of carriers was around 1% from the model. These model estimates agree with the respective estimates from data: among those receiving the first dose of the HBV vaccine in 2002 (the first year of the vaccination programme), 16% were anti-HBc positive and 1.3% were HBsAg positive (data from the National HBV Vaccination Programme). Previous studies have also shown the great impact of universal infant vaccination6 and the increased impact of selective vaccination when targeted to high-risk MSM.10
Certain limitations of our study should be mentioned. In the model, we did not account for vertical transmission, for heterosexual transmission or for transmission via sharing injecting equipment. These routes of transmission are extremely rare among MSM in the Netherlands and, therefore, we expect that the dynamics of HBV and the impact of vaccination are qualitatively well approximated by our model. Another limitation is that HBV transmission via household contacts was modelled only indirectly by assuming an exposure of young ages proportional to the prevalence of chronic hepatitis B. Modelling transmission within households would require an individual based model or a description of within and among household contacts.6 ,22 Furthermore, as with most modelling studies, a shortcoming is the uncertainty in model parameters, such as the transmission probabilities or the percentage of those infected becoming carriers. Based on sensitivity analyses carried out in previous work,5 ,6 ,23 we believe that this uncertainty affects our results only quantitatively, but not qualitatively, such that the conclusions are robust.
Despite the criticism against selective vaccination of MSM, our results indicate that a substantial decline in HBV incidence can be attributed to this programme. This decline would have been indeed rather small, if the vaccinations administered had been equally distributed among all MSM and no other changes had occurred in the population. However, a much higher reduction in HBV incidence has been observed since 2006, indicating that high-risk MSM are the ones mostly being vaccinated or some other conditions (important for the transmission of HBV) have changed. The most reasonable explanation would be a reduction in sexual risk behaviour around the year 2005, such that HBV incidence would start declining after 2006. Our model results indicate that this could also be a contributing factor. However, there has been no epidemiological evidence indicating a decline in sexual risk behaviour around this time.24 Furthermore, reporting of other sexually transmitted infections among MSM remained at the same level or increased in this period.11 Hence, it seems unlikely that the decline in reported HBV cases resulted solely from a reduction in sexual risk behaviour. Also, we expect that the reduction in notifications does not merely reflect deterioration in surveillance or reporting due to the introduction of a new law in 2008 that requires laboratories to report cases. Recent data from one region in the Netherlands indicate actually that the completeness of reporting has improved between 2005 and 2010. Therefore, it is highly likely that the impact of selective vaccination was enhanced among MSM because the men who were vaccinated were the ones who were the most at risk to get infected.
Comparing selective vaccination with universal infant vaccination leads to the same conclusion as in previous work:6 in terms of the reduction in HBV incidence, universal vaccination is by far the best strategy, 20 years after its introduction. Until then, however, it is imperative to continue the selective vaccination of MSM especially targeting high-risk MSM. This implies that for countries with a high transmission of HBV among MSM, if universal infant vaccination is not yet introduced or does not cover over 35 years age cohorts, selective high-risk MSM vaccination should actively be implemented and continued until elderly high-risk MSM are also protected against HBV infection by universal vaccination. On the other hand, for countries where universal vaccination was introduced long ago and where more than 35 years age cohorts are protected or where the HBV epidemic is not concentrated on specific risk groups, selective vaccination will not be beneficial.
Key messages
-
In the Netherlands, selective vaccination has had a substantial impact in reducing hepatitis B virus (HBV) incidence among men who have sex with men (MSM).
-
Our data suggest that the decline in HBV incidence is partly attributable to the fact that mostly high-risk MSM have been vaccinated.
-
Universal infant vaccination can considerably reduce HBV incidence, but only several decades after its introduction.
-
Countries with universal childhood vaccination and ongoing transmission among MSM should actively implement and continue selective high-risk MSM vaccination until elderly MSM are protected.
Acknowledgments
The authors would like to thank Rutgers WPF for the data on sexual risk behaviour and the National HBV Vaccination Programme for data on HBV vaccinations. The suggestions of two anonymous referees and of the editorial committee are also acknowledged.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online appendix
Footnotes
-
Handling editor Jackie A Cassell.
-
Contributors MX contributed to the design of the study, carried out data analyses for the parameters and calibration of the model, and designed the intervention scenarios. MK developed the model and carried out the numerical calculations. RvH contributed to the study set up and the processing of data on sexual behaviour and vaccination coverage. RC and SH contributed to the set up and design of the study. SH and JvS assisted with data from the National HBV Vaccination Programme and with the interpretation of data for vaccination strategies in other countries. All authors contributed to the interpretation of results and to the writing of the paper.
-
Funding The work was partially supported by the Netherlands Organisation for Health Research and Development (ZonMW), grant number 125010004.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.