Background
Sub-Saharan Africa (SSA), the world region with the highest burden from communicable diseases [
1], is currently experiencing a fertility transition [
2], contributing jointly with continued mortality decline and massive urbanisation, to an overall demographic transition (DT). However, an investigation of any potential effects on the epidemiology of infectious disease in areas of SSA (or indeed in the wider world) has hitherto been largely neglected.
Although demography is continually in transition, the conventional idealised picture of DT, as experienced by industrialised countries [
2‐
4] after 1750–1800, is of an initially stationary population characterised by high mortality and fertility rates — the
ancien or
Malthusian régime [
3] — which first experiences falling mortality rates accompanied by rapid population growth leading to a young age distribution. This mortality decline is later followed by a fertility decline with, in this idealised scenario, the eventual attainment of a
modern stationary regime at low levels of fertility and mortality. In classical demographic and demo-economic explanations, mortality decline has been the major trigger of fertility decline, which in turn resulted as the major engine of economic development of industrialised countries [
3]. The experience of industrialised countries after 1960 has shown however that replacement fertility is by no means a necessary outcome of this process, with the gradual spread of below replacement fertility, in some cases even to very low levels [
5], yielding rapid population ageing.
In the different regions of SSA, the joint action of fertility and mortality decline and rapid urbanisation — not to mention HIV/AIDS — are producing marked changes in population structures, primarily in the distribution of the population by age. These changes in the population age distributions are likely to affect the proportions of people of different age groups with whom individuals come into contact day by day. Such patterns of contact by age are key determinants [
6,
7] of the patterns of transmission of infectious diseases and, consequently, of the ensuing age distribution of immunity. Moreover, severity of disease can often be influenced by age at infection [
6], so that the overall population burden of disease also may change as the age distribution of the population changes. For infections that are vertically transmitted, time changes in age-specific fertility, combined with an evolving female age distribution and age distribution of female immunity, will also likely have an impact on the epidemiology of the infection [
6].
The aim of this work is to provide a general investigation of the overall effects that changes in population age structure associated with DT might have upon the burdens of infectious diseases with complex epidemiology, focusing on hepatitis B virus (HBV), and selecting the region of Senegal and The Gambia as our “laboratory” in SSA. Only a few studies have so far investigated the possible effects of demographic change on infection transmission dynamics and control. Moreover, most such studies have focused on common childhood infections such as measles and varicella in view of their epidemiological characteristics, primarily a single well-identified transmission route [
8‐
19]. The only work so far that has considered an infection with more complex epidemiology has focused on dengue [
20], and that study confirmed by statistical analyses effects similar to those predicted in the aforementioned literature.
HBV infection is a major example of an infection causing a substantial continuing burden of disease worldwide [
21‐
23], including cirrhosis and liver cancer; thus, it calls for a redoubling of efforts and prioritisation of actions aimed at elimination [
22]. HBV has a complex epidemiology in which such key processes as transmission, infection, and disease development are strongly age-related [
24‐
26]. More specifically, HBV has multiple transmission routes (see Additional file
1: Figure S3.1), with perinatal, sexual, and horizontal transmission by person-to-person contacts being most prominent in the general population. Each of these transmission routes is potentially affected by changes in the age distribution of the population. In addition, the probability, once infected, of developing persistent HBV infection is strongly age-related, being very high in infancy but declining steeply with age [
25]. Given these age dependencies, it is presently unclear how the ongoing DT might affect the overall transmission and resulting disease burden of HBV. Finally, the pre-vaccination landscape of HBV exhibited a dramatic epidemiologically significant variation in prevalence worldwide [
27]. In Medley et al. [
28] an explanation has been proposed for this variability by the nonlinear feedback between the HBV force of infection and the age-related probability of developing HBV carriage; moreover, it was conjectured that the demographic changes along the DT might have been a major determinant of this heterogeneity.
Mathematical modelling of infectious disease transmission dynamics is a technique which naturally lends itself to the investigation of the interplay between population and infectious disease dynamics. We used an age-structured model for the transmission dynamics of HBV with realistic population dynamics parameterised with demographic (vital rates) and epidemiological data (HBV prevalence) from Senegal and The Gambia, considered here as a unique demographic entity, in an attempt to cast light on the nature and extent of the potential influence of the course of the DT on HBV epidemiology and its possible implications for the variability of HBV prevalence worldwide.
Discussion
Demographic change can, according to circumstances, create great challenges and opportunities. Possibly the major example in history is represented by the interplay between DT and the industrial revolution, with the acknowledged role of mortality decline as the major trigger of fertility decline, via the switch from quantity to “quality” of children, which in turn resulted as a major engine of the sustained economic development of industrialised countries [
3]. Nonetheless, it has remained largely unclear where the balance may lie in terms of its effect on infectious disease epidemiology. As age distributions of infection and immunity are far from uniform, patterns of contact between different age groups are of fundamental importance for infection transmission, so that change in population age structure may strongly influence infectious disease epidemiology. It then becomes essential to attempt to understand how the two interact, especially for those world regions, such as SSA, where the burden from infectious diseases is still striking.
In SSA, although mortality decline possibly had its beginnings prior to 1900 [
3], much of the process has taken place since 1950 and is still ongoing. On the other hand, the fertility transition was unexpectedly delayed and has been proceeding at a slower rate compared with other regions such as Asia and Latin America [
2]. Indeed, fertility decline began, in regions such as SG, as recently as the 1990s [
2] and is predicted to continue to display its effects over a span of a century or so. The resulting massive changes in the age distribution of the populations are likely to have greatest impact on those infections in which processes of transmission and disease are strongly age-related and infection may be long-lasting. HBV is one such infection and one which gives rise to a substantial burden of disease. Global mortality associated with viral hepatitis ranks equally with that from HIV/AIDS and tuberculosis, and in 2016 the World Health Assembly adopted a strategy for its global elimination [
21], with HBV being responsible for the majority of deaths associated with viral hepatitis [
22]. Nonetheless, the great heterogeneity in the global distribution of the burden of disease arising from HBV [
27,
28] might hinder steps towards its elimination. For these reasons a mathematical modelling approach may provide a key to disentangling the mutual interplay of demographic changes and HBV epidemiology during the different phases of DT, and, by depicting an entire range of dynamic regimes, possibly offer some clues towards explaining the puzzle of the heterogeneity in HBV burden.
Using a mathematical model, we assessed the possible extent of the impact of the entire course of the DT on the transmission and burden of HBV in a region of SSA. Our results predict that, departing from the pre-transitional demo-epidemiologic equilibrium prevailing at some time prior to 1900 and later destabilised by the onset of the mortality transition, HBV burden may have followed a complex “epidemiological transition” pattern. This pattern is initially characterised by a long epoch of expansion until ~ 2000, corresponding to the epoch of maximal population rejuvenation, and it is predicted to be subsequently followed by a dramatic “retreat” corresponding to the major epoch of fertility decline observed from 1990 onward and continuing for decades after the completion of fertility decline. This demographically driven pattern of expansion-retreat of HBV is mostly explained by the underlying dramatic changes in horizontal transmission due to the changes in the age distribution of the population. During the HBV expansion phase, this change in transmission is the consequence of the period of rejuvenation following mortality decline in young age groups while fertility remained persistently high during the first stages of the transition. During the phase of HBV retreat, the change in transmission is mostly the consequence of the decline in fertility associated with the onset and completion of the fertility transition in SG. Thereafter, between 2000 and 2150 (when below replacement fertility prevails), HBV burden is predicted to decline dramatically by around 70%.
These results seem to be of importance from a number of standpoints. First, they shed light on the nature, extent, and determinants of the potential influence of demographic change on the burden of HBV, and quite possibly other infections with a similarly complex epidemiology having in common age dependency in key epidemiological processes. Second, they supply a contribution to the interpretation of the long-term natural history of HBV epidemiology. Third, they provide considerable evidence in support of a conjecture formulated in [
28], namely that the great variability in pre-vaccination HBV prevalence worldwide might largely be a consequence of the DT. More precisely, the present work has shown that the greatest part of this variability might result from the differences in the timing of onset and pace at which the DT was experienced in different regions of the world [
2,
3]. Fourth, they indicate a precise causative role of the DT in triggering “epidemiological transitions”, broadly interpreted as patterns of retreat of the burdens of disease which are a key component of Omran’s classical definition [
37]. Last, the onset of a demographically driven decline in HBV prevalence in the model coincided with a real-world expansion of HBV vaccination, suggesting a synergy potentially boosting effectiveness of control in the decades leading up to elimination. This appears to be of fundamental importance in helping to achieve the aim of HBV global elimination [
21]. Indeed, the predicted initiation of the HBV retreat phase in the last 15 years suggests that a “window of opportunity” has presented itself where fertility decline will synergistically concur with immunisation programmes to hasten the global target of HBV elimination.
Finally, although due to paucity of data our analyses focused on our characterisation of a single region of SSA, our model representations of the DT and of HBV epidemiology are fairly general. We therefore feel that a major strength of this work is that it provides a pointer for interpreting past and future long-term trends of HBV in SSA and also at the global scale.
The major limitations in studies of the impact of population change on infection transmission typically lie in the lack of availability of sufficiently long-term time series of reliable epidemiological data allowing robust estimation of the impact of demographic processes on age-related epidemiology. As here, most available studies [
12,
15,
16,
18] have in common the same drawback: they have to postulate a direct influence of age-structured demographic change on transmission and typically rely on a single age-structured infection datum (e.g. a single serological profile) to estimate transmission. However, as argued in [
11], the formulation of horizontal transmission adopted here, though simple, has the advantage of potentially capturing a whole range of processes triggered by fertility decline, namely the ageing of population in the groups most involved with horizontal transmission and also the ensuing decline of family size as an engine of the decline in intra-family horizontal transmission. In this study we modelled horizontal transmission by a WAIFW matrix as previously available for the study setting [
30], being obliged to do so given that direct data on social contact patterns in SSA are so far available only for Eastern Africa [
38], where socio-cultural patterns are believed to be markedly different from those of Western Africa and the setting considered. The lack of direct contact data in most of Africa remains a serious limitation. Another limitation lies in the lack of precise predictions about mortality resulting from chronic HBV infection, a choice motivated by a lack of appropriate data.
To sum up, while the focus of this work has been on isolating and highlighting the impact of changing demography on the epidemiology of HBV, it is important to note that patterns of evolution of social and sexual behaviour over time will add a further layer of complexity to the resulting pattern of HBV epidemiology. More generally, it is reasonable to conclude that where processes of infection and disease or relevant behaviours display age dependencies, potential changes in age distribution associated with fertility, mortality, or migration may have a significant bearing on observed patterns of infection and should be taken into account when forward planning interventions.