Centre for Infectious Disease Control, National Institute for Public Health and the Environment, PO Box 1, Bilthoven, 3720, BA, the Netherlands

Department of Public Health Medicine, School of Public Health, University of Bielefeld, Bielefeld, Germany

Julius Centre for Health Sciences & Primary Care, University Medical Centre Utrecht, Utrecht, The Netherlands

Abstract

Background

For accurate estimation of the future burden of communicable diseases, the dynamics of the population at risk – namely population growth and population ageing – need to be taken into account. Accurate burden estimates are necessary for informing policy-makers regarding the planning of vaccination and other control, intervention, and prevention measures. Our aim was to qualitatively explore the impact of population ageing on the estimated future burden of seasonal influenza and hepatitis B virus (HBV) infection in the Netherlands, in the period 2000–2030.

Methods

Population-level disease burden was quantified using the disability-adjusted life years (DALY) measure applied to all health outcomes following acute infection. We used national notification data, pre-defined disease progression models, and a simple model of demographic dynamics to investigate the impact of population ageing on the burden of seasonal influenza and HBV. Scenario analyses were conducted to explore the potential impact of intervention-associated changes in incidence rates.

Results

Including population dynamics resulted in increasing burden over the study period for influenza, whereas a relatively stable future burden was predicted for HBV. For influenza, the increase in DALYs was localised within YLL for the oldest age-groups (55 and older), and for HBV the effect of longer life expectancy in the future was offset by a reduction in incidence in the age-groups most at risk of infection. For both infections, the predicted disease burden was greater than if a static demography was assumed: 1.0 (in 2000) to 2.3-fold (in 2030) higher DALYs for influenza; 1.3 (in 2000) to 1.5-fold (in 2030) higher for HBV.

Conclusions

There are clear, but diverging effects of an ageing population on the estimated disease burden of influenza and HBV in the Netherlands. Replacing static assumptions with a dynamic demographic approach appears essential for deriving realistic burden estimates for informing health policy.

Background

Advances in medical science and improvements in nutrition, housing, sanitation and other hygenic conditions have contributed towards reducing mortality rates in the Netherlands and globally, with the result of prolonging the average length of life. Life expectancy (LE) in the Netherlands has increased over the period 1950–2010 from 70.3 to 78.8 years for men, and from 72.6 to 82.8 years for women

Demographic changes such as population ageing have an impact on estimates of the disease burden (i.e., mortality and morbidity) over time. The immune system becomes less efficient as it ages, with the consequence that for many infectious diseases symptoms can occur more frequently and with greater severity in elderly compared with younger individuals

We investigated the effects of two aspects of demographic change – population ageing and growth – on the future burden of two infectious diseases in the Netherlands. Hepatitis B (HBV) and seasonal influenza were selected as examples because their natural histories differ widely, especially with respect to time-scale. Severe, life-threatening sequelae of HBV develop slowly, often appearing many years following acute infection

In the current study, we integrated a simple dynamic model of demographic change with established methodology for the computation of disease burden, to assess the impact of population ageing and growth on the disease burden of seasonal influenza and HBV infection in the Netherlands over the period 2000–2030. Our goal was not a high degree of precision, but rather we aimed to show qualitatively the differences in estimated burden when demographic change is taken into account.

Methods

DALY methodology

We used the pathogen-based approach for estimating disease burden, in which the infection event serves as the starting point, and all (future) health outcomes – acute infection as well as short and long-term sequelae – causally related to infection with the pathogen are included in the total disease burden calculation

We calculated disease burden using the disability-adjusted life-years (DALY) measure, a concept developed by the World Health Organization

An advantage of the DALY methodology is that diseases (both communicable and non-communicable) can easily be compared to each other with respect to their impact on population health. DALY calculation for the pathogen-based disease burden approach requires a number of sources of data: (i) an outcome tree describing disease progression through various health outcomes with specified transition probabilities of developing each outcome, (ii) data on the incidence of infection, disability weights (on a scale from 0 to 1, where 0 represents perfect health and 1 represents death) and the average duration associated with each health outcome, and the rates of mortality from each outcome, and (iii) life expectancy for all ages (typically aggregated into 5-year age-groups).

Incidence data is typically retrieved from national notification or hospitalisation databases, and where necessary is adjusted using multiplication factors (MFs) to correct for under-estimation. Under-estimation can be divided into under-ascertainment (cases at the community level who do not seek healthcare, possibly because infection is mild or even asymptomatic) and under-reporting (cases who have attended healthcare services, but who are not reported to any notification or surveillance system, possibly because notification is not mandatory, notification failure, misdiagnosis or misclassification). YLD and YLL are computed as summations over

where _{
i,a
} is the number of cases developing health outcome _{
i,a
} and _{
i,a
} are the disability duration and disability weight for outcome _{
i,a
} is the number of deaths occurring in health outcome _{
a
} is the life expectancy for age-group _{
i,a
}.

DALYs were estimated for acute infections occurring during the period 2000–2030, and 95% confidence intervals were constructed around DALYs using Latin hypercube sampling methods

Modelling demographic dynamics

Dynamic modelling of the Netherlands population was undertaken also with the year 2000 as a starting point, using demographic information publicly available from CBS for that year (population size according to sex and 1-year age-group (from 0 to 85+ years), age- and sex-specific mortality rates

Modelling disease progression

Hepatitis B

The outcome tree for HBV disease progression was adopted from that developed as part of the Burden of Communicable Diseases in Europe (BCoDE) project

**Outcome trees and parameter values for hepatitis B and influenza.**

Click here for file

Acute HBV cases in the Netherlands have been notifiable since 1976 and are openly available; data on age, sex, transmission route, and country of birth are included. In the period 2000–2010 for example, the majority of acute HBV notifications occurred in males (76.3%), and in the 35–39 years age group (15.1%; mean of 38 notified cases per year). Notification counts were first adjusted to account for the under-estimation of the incidence of acute infection when using notification data (range: 2.4–3.0)

Disease burden, in DALYs per year, was then computed (Eqs. 1a and 1b above) taking into consideration the evolving population age-structure and life expectancy and dynamically calculated incidence. To isolate the effect of demographic change on future disease burden, DALYs were also computed using a ‘static’ demographic model in which the age distribution, population size, and life expectancy were assumed to remain constant throughout the simulation period.

Scenario analyses examined the effect of decreasing incidence rates over the period 2012–2030 (crudely simulating the effects of increasing coverage of HBV vaccination in all age-groups) by 2% per year, and the effect of decreasing incidence rates in the youngest age-groups only (<15 years) of 5% per year (simulating decreases in transmission attributable to age-targetted vaccination

Seasonal influenza

The outcome tree for seasonal influenza and most parameter values were also taken from the BCoDE project and earlier work

We calculated age-specific incidence rates from Dutch sentinel GP consultation rates (data were obtained from the Netherlands Institute for Health Services Research (NIVEL)) for ILI averaged over the period 2000–2010

The rate of GP consultations due to influenza in the UK (period 1991–1996) has been reported to be 9.9/1000 person-years in the 65+ years age group, 16.4/1000 in the 15–64 years age group, and 12.2/1000 in those aged <15 years

Over 90% of influenza-related mortality occurs in adults aged 65 years or older

Influenza vaccine uptake in persons aged 60 years or older has been approximately 75% since 2008

Results

Population dynamics

Figure

Simulated change in the age distribution of the Netherlands population over the period 2000 to 2030 (left panel) and the corresponding change in the size of each age-group over the same period (right panel)

**Simulated change in the age distribution of the Netherlands population over the period 2000 to 2030 (left panel) and the corresponding change in the size of each age-group over the same period (right panel).**

Hepatitis B

The annual disease burden of HBV in the Netherlands (in DALYs, split into YLD and YLL per year) predicted using the dynamic model is shown in Figure

The modelled annual disease burden of HBV in the Netherlands (split into YLD and YLL per year), for acute infections in the period 2000–2030

**The modelled annual disease burden of HBV in the Netherlands (split into YLD and YLL per year), for acute infections in the period 2000–2030.** Vertical lines indicate 95% confidence intervals.

Age-group specific estimated burden of HBV, comparing acute infections occurring in the years 2000 and 2030, plotted separately for males (upper) and females (lower)

**Age-group specific estimated burden of HBV, comparing acute infections occurring in the years 2000 and 2030, plotted separately for males (upper) and females (lower).** Vertical lines indicate 95% confidence intervals.

Estimated HBV burden as a function of time since infection, for acute infections occurring in the year 2000 only

**Estimated HBV burden as a function of time since infection, for acute infections occurring in the year 2000 only.**

Compared with results using a ‘static’ demographic model, in which the population size, age distribution, and life expectancy were held constant, the dynamic model predicted an overall increased burden, ranging from 1.34-fold higher burden in 2000 through a 1.50-fold higher burden in 2030. This greater burden for the dynamic model was largely represented by persons aged 15–39 years at infection (Figure

Comparison of dynamic with static modelling assumptions with respect to population growth and ageing on the annual burden of HBV in the Netherlands (in DALYs per year), for acute infections occurring in the year 2000 and aggregated across sex

**Comparison of dynamic with static modelling assumptions with respect to population growth and ageing on the annual burden of HBV in the Netherlands (in DALYs per year), for acute infections occurring in the year 2000 and aggregated across sex.**

The scenario analysis indicated a decrease of 32% in the expected DALYs in 2030 if the incidence rate decreased by 2% per year in all age groups from 2012, compared with the constant incidence rate assumption (Figure

HBV burden estimates in three scenarios: age-specific incidence rates are constant over time; decreasing incidence rates of 2% per year from 2012 (corresponding to increasing vaccination rates in all age groups); and decreased incidence rates of 5% per year in the under-15 years age-groups only (reflecting the effect of age-targetted vaccination)

**HBV burden estimates in three scenarios: age-specific incidence rates are constant over time; decreasing incidence rates of 2% per year from 2012 (corresponding to increasing vaccination rates in all age groups); and decreased incidence rates of 5% per year in the under-15 years age-groups only (reflecting the effect of age-targetted vaccination).**

Influenza

Figure

The modelled annual disease burden of seasonal influenza in the Netherlands (split into YLD and YLL per year), for acute infections occurring in the period 2000–2030

**The modelled annual disease burden of seasonal influenza in the Netherlands (split into YLD and YLL per year), for acute infections occurring in the period 2000–2030.** Vertical lines indicate 95% confidence intervals.

Age-group specific estimated burden of seasonal influenza, comparing acute infections occurring in the years 2000 and 2030, plotted separately for males (upper) and females (lower)

**Age-group specific estimated burden of seasonal influenza, comparing acute infections occurring in the years 2000 and 2030, plotted separately for males (upper) and females (lower).** Vertical lines indicate 95% confidence intervals.

Compared with a ‘static’ population model (with constant population size, age distribution, and life expectancy across the simulation period), the model incorporating demographic dynamics resulted in a total disease burden varying between 1.00-fold (in 2000) and 2.27-fold (in 2030) higher.

The scenario analysis investigating the consequences of decreasing trends in incidence rates for the 60+ years age group from 2012 indicated substantial reductions in the overall disease burden, mainly in YLL, were achieved with moderate reductions in the incidence of infection (Figure

Influenza burden estimates in four scenarios for trends in incidence rates from 2012, for the 60+ years age group only (upper) and for all ages combined (lower)

**Influenza burden estimates in four scenarios for trends in incidence rates from 2012, for the 60+ years age group only (upper) and for all ages combined (lower).** In each scenario, the affected incidence rates for the 60+ age group were (i) constant, (ii) decreased by 2% per year from 2012, (iii) decreased by 5% per year, and (iv) increased by 2% per year.

Discussion

By simulating the ageing and growth of the Dutch population over the period 2000 through 2030, we found diverging trends in the future disease burden associated with infection with acute HBV and seasonal influenza. For both infections considered, incorporation of a simple model of population growth and ageing predicted a higher overall disease burden for both infections than if population size, age distribution, and life expectancy were assumed to remain constant over time (the steady-state assumption). For HBV, this greater disease burden is mainly attributable to the increased life expectancy at the time sequelae develop; note that this effect is larger for HBV (total DALYs for persons infected in 2000 were 1.3 times higher; Figure

For HBV, population ageing also meant that the relative size of the population in the age group most at risk for acute infection diminished over the simulation period, giving rise to fewer incident infections overall and a lower projected future burden. At the same time, the increase in life expectancy over the simulation period predicts an increase in the opportunity to develop long-term sequelae, and for those persons who die from HBV-related severe sequelae, the YLL component of the disease burden will also increase (Figure

In contrast, for seasonal influenza our results indicated an overall increasing burden over the simulation period. The effect of population ageing is mostly attributable to the growing size of the elderly population. This effect is localised in YLL, due to increased life expectancy over time and the highest mortality rates occurring in the oldest age groups. The effect of population ageing on disease burden was more pronounced for the female than the male elderly, due to longer female life expectancy.

We note that our estimated influenza burden is much higher than previously reported for seasonal influenza

The evolution of the age distribution in the population is entirely driven by the demographic processes of mortality, fertility, and migration. We have not attempted to closely fit population-level data on temporal trends in these rates; our goal was to show the impact on disease burden from a typical ageing population that is the consequence of such processes; which does not require accurate model fit.

Previous research has used models of transmission dynamics to predict the effects of demographic change (population decline and ageing) on the incidence of childhood disease

One limitation of the current study is that we have assumed that the parameters underlying transmission dynamics do not change over time; incidence rates reflect the force of infection (determined by contact patterns between age/risk groups and the prevalence of acute infection in the population), and interventions such as vaccination will influence future incidence. In addition, demographic changes such as population growth and ageing can influence the transmission of infection through shifts in the size of the subpopulations that are susceptible to infection. Our findings thus illustrate the simple scenario in which transmission dynamics remain constant, localising any trends in disease burden to population dynamics only.

A second limitation concerns data quality, namely the assumptions underlying the estimation of annual incidence. For HBV, a single MF was applied over the entire simulation period to account for under-estimation inherent in the notification data, namely under-reporting and under-ascertainment of cases. For influenza, the MF range was based on two studies only (conducted in Hong Kong in 2008 and Scotland in 1993–1994); thus, generalisability to the Netherlands setting is an issue. The accuracy of the calculated DALYs depends strongly on the accuracy of the reconstructed incidence of acute infection, and thus also on the MFs.

For HBV we have not considered the influence of acute infection in first-generation migrants (FGMs) from endemic countries, or secondary transmission from FGMs. Based on serological survey data from 2007, it is estimated that 1.4% of FGMs in the Netherlands are chronically infected, representing 51% of the infected population

Furthermore, for HBV we have simplified certain parameters; e.g., we assumed the disability durations for compensated cirrhosis and HCC to be equivalent for males and females and constant for all ages. For influenza, age-specific disability weights for the long-term sequela of otitis media were simplified to a single value.

Finally, the pathogen-based approach attributes all future burden arising from initial infection to the year of infection. Figure

Conclusions

The ageing of the population is expected to have a clear impact on the future disease burden of HBV and influenza in the Netherlands; computing disease burden using steady-state assumptions is not realistic. We emphasise that our results should not be interpreted as forecasts, but rather as a qualitative investigation into the effects of population change. As such, results are generalisable to other countries with similar demographic dynamics and age-specific patterns in incidence. Future work is required to refine these findings; for instance, by incorporating the time-dependent effects of potential public health interventions, changes in contact patterns, risk factors, or, in the case of HBV, the number of chronic carriers entering the population

Abbreviations

BCoDE: Burden of Communicable Diseases in Europe; CI: Confidence interval; CBS: Statistics Netherlands; DALY: Disability-adjusted life-years; HBV: Hepatitis B virus; MF: Multiplication factor; ILI: Influenza-like illness; YLD: Years of life lost due to disability; YLL: Years of life lost.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

SM conceptualised the study, carried out the simulations and burden computations, and drafted the manuscript. MK conceptualised the study, advised on modelling, interpreted the results, and edited the manuscript. AvL and DP interpreted the results and edited the manuscript. All authors read and approved the final manuscript.

Acknowledgements

This research was partly funded by the European Centre for Disease Prevention and Control (Specific agreement No 1 to Framework Partnership Agreement GRANT/2008/003). We would like to thank Jan Gravestein and Gé Donker at NIVEL, Femke Koedijk and Kees van den Wijngaard at RIVM, and Alessandro Cassini, Edoardo Colzani, and Piotr Kramarz of the European Centre for Disease Prevention and Control and the members of the BCoDE consortium.

Pre-publication history

The pre-publication history for this paper can be accessed here: