Vaccines have become a centerpiece in public-health practice because they have helped to reduce the incidence and prevalence of many diseases and have been responsible for the eradication of smallpox from the world Citation[1]. It has been estimated that, currently, over 350 vaccine candidates against nearly 100 different infectious diseases are in the development pipeline Citation[2]. Currently available vaccines are the result of research in which vaccine trials play an essential role.
Recent conceptual and methodological advances in geographical analysis have greatly enhanced our ability to design and execute vaccine trials. Methods for incorporating a geographic dimension into field research on human disease and interventions against disease, through geographic information systems (GIS), have advanced greatly in recent years. GIS is a powerful automated system for the capture, storage, retrieval, analysis and display of spatial data that can be loaded on most computers Citation[3]. Within a GIS database, spatial data, which may include geopolitical boundaries, rivers, roads, health facilities, sources of risk and human settlements, are stored.
Perhaps the most obvious utility of GIS for vaccine trials is in the planning of such trials. A vaccine trial requires a study area with sufficient disease burden, adequate healthcare centers to detect the target disease, and a system for transportation of vaccines, frequently within a cold chain. Knowledge regarding the population, physical characteristics of the area and barriers to movement are essential prerequisites for selecting a trial site. Of the three important components of a vaccine trial, preparation, vaccination campaign and follow-up, the vaccination campaign perhaps requires the most intense effort. The participants to be vaccinated have to be distributed among teams in a fashion that allows the teams to achieve their assigned goals. The size of their catchment area, rather than the size of catchment population, is closely associated with the performance of staff Citation[4]. The use of geographic data assists in planning by facilitating the creation of geographically informed work schedules for the team.
If vaccines are administered in the community and not at a central site where vaccines are stored, the daily vaccine requirement has to be predicted accurately to allow shipping of the correct number of study vaccines within the cold chain. Shipping too few vaccine doses may result in disappointed volunteers and shipping too many vaccines requires the return of vaccine doses at risk of breaking the cold chain. The GIS may help to define the clusters to be vaccinated each day of vaccination and the target population residing within the clusters. Commonly, specimens such as serum are collected from study participants on the day of vaccination to compare the immunogenicity of the candidate vaccine with the control. The GIS may be used to estimate the distance between the specimen collection centers and the center for storage (vaccination control center) of the specimens.
The GIS can also help in monitoring the success of recruitment in a vaccine trial. Refusals and absentees are a common problem in vaccine trials. Identifying areas with low coverage during a vaccine campaign can help investigators to target these areas for sensitization prior to a next day/round of vaccination.
Distance to health facilities may play a vital role in healthcare-seeking behavior, particularly in low-income countries, where mobility is limited Citation[5]. In a vaccine trial, the target patients are usually enrolled in hospital-based passive-surveillance systems. If a person lives far away from the target health facilities, they will be less likely to report to these facilities, which could lead to detection bias. It is important to ensure that health facilities are within reach of the subjects of the trial area. Establishing health facilities requires knowledge of the spatial distribution of the population, existing health facilities, barriers to movements and spatial barriers to the establishment of health facilities. Using the GIS, suitable sites for health facilities can be established. In addition, by using the GIS, we are now able to determine spatial metrics (e.g., simple cost and network distances) of movements and to evaluate spatial factors and impediments for the uptake of healthcare. For instance, a well-developed road may accelerate and a water body may impede our movements, which can be evaluated through a GIS when determining accessibility of the target health facilities. When resource constraints do not allow the establishment of new health facilities, GIS can provide the spatial information needed to delineate and create trial areas and, subsequently, to analyze the trial data.
Less well appreciated is the role of GIS in enhancing our ability to extract critical information about vaccine protection. A Phase III vaccine efficacy trial is typically designed to measure vaccine protective efficacy, which reflects direct effects of vaccines Citation[6]. Individually randomized, placebo-controlled trials are considered ‘gold standard’ for evaluating such vaccine efficacy. With this trial design, the type of vaccine protection measured is thought to be ‘direct’ vaccine protection, which is the protection conferred to a vaccinated individual in isolation from any other vaccinated individuals. However, infectious diseases that are transmitted from person to person, either directly or indirectly, are considered dependent happenings, which means the rate of occurrence is dependent on the number of individuals already affected Citation[7]. Owing to this property, vaccines may protect individuals not only directly, but also via herd protective effects, which occur when vaccination of a group of individuals in proximity to one another reduces the intensity of transmission of the infection among members of the group Citation[8,9].
From a broader public-health perspective, the protective value of a vaccine should derive both from direct protection to vaccinees, owing to vaccine-induced protective immunity in these individuals, and from herd protective effects, which may protect nonvaccinated neighbors of vaccinees and may also enhance the protection of vaccinees who reside in the neighborhood of other vaccinees Citation[10–14]. Protection of nonvaccinees in this fashion is termed ‘indirect’ vaccine protection, whereas the enhanced protection of vaccinees is termed ‘total’ vaccine protection, both in distinction to direct vaccine protection. For instance, the introduction of conjugate Haemophilus influenzae type b vaccines in the USA, and later in The Gambia, demonstrated such indirect protection Citation[15]. The use of cluster-randomized trials has been proposed as a design to measure these herd protective effects Citation[16]. In this design, geographic clusters of individuals are randomized to the vaccine under study or the control agent. Contrasting the rates of the target infection among vaccinated individuals in the vaccinated clusters versus recipients of the control agent in the control clusters is thought to measure total vaccine protection, whereas contrasting the rates among nonvaccinated individuals in vaccinated clusters versus nonrecipients of the control agent in the control clusters measures indirect vaccine protection. However, the use of cluster-randomized trials to measure herd effects is dependent on several assumptions:
| • | The target infection must be transmitted from person to person (directly or indirectly); | ||||
| • | The clusters that are randomized must correspond to epidemiological units of transmission of the infection, in the sense the transmission must occur within clusters but be negligible between clusters; | ||||
| • | The clusters must be demographically stable. | ||||
The second and third assumptions require antecedent data on the geography of infection transmission and population movement, data that require earlier studies on geographically mapped populations.
While individually randomized trials of vaccines are thought to measure only direct vaccine protection, in practice, in any individually randomized trial there will be geographic differences in vaccine coverage of the target population owing to chance variations in randomized assignments and to different rates of eligibility and participation. In addition, if suitable geographic clusters can be identified and there is sufficient variation in vaccine coverage among these clusters, vaccine herd effects can be assessed by evaluating the correlation of disease incidence with levels of vaccine coverage in these clusters.
As an example, although the oral cholera vaccine was found to be efficacious from the data of an individually randomized, placebo-controlled trial in Bangladesh in 1985, until 2005 it was unknown that the vaccine conferred a significant level of herd protection to nonvaccinees, as well as enhanced protection to the vaccinees living in the high vaccine coverage areas Citation[17]. It was also unknown that, by vaccinating mothers and caregivers, the risk of cholera in infants and young children (<2 years of age) can be significantly reduced Citation[18]. By including ecological dimension through using GIS, it was possible to obtain such important findings of the trial. Moreover, use of the ecological dimension in the trial data analysis proved that with a coverage of 50%, cholera can theoretically be controlled in the endemic setting of Bangladesh Citation[19].
Beyond enabling assessment of herd effects within individually randomized trials, GIS methods can enhance the analysis of vaccine protection in these trials in other ways. Conventionally, vaccine efficacy is measured through global efficacy of the vaccines, which means the variations in vaccine efficacy in geographic subpopulations are not considered. We observed that vaccine coverage can be highly variable in space, even in an individually randomized trial Citation[17]. Vaccine herd effects can vary in areas of different levels of vaccine coverage. We have shown that these vaccine herd effects can have a perverse effect on conventional estimates of vaccine protective efficacy. For example, in the field trial of killed oral cholera vaccines in Bangladesh, areas with higher levels of vaccine coverage had lower estimates for vaccine protective efficacy Citation[17]. In this trial, therefore, herd effects tended to bias the overall estimates of vaccine protective efficacy in a downward direction. Only with the use of GIS analysis was it possible to estimate the direct vaccine protection in the areas of lower vaccine coverage.
Failure to consider the geographic dimension in vaccine trials may lead to inadequate planning, conduct and analysis of the trials. The methods available in GIS offer public-health practitioners a valuable new toolbox to incorporate ecological aspects in vaccine trials Citation[20]. Consideration of ecological aspects through GIS in vaccine trials may help investigators to improve their study design, management, analysis and interpretation of data to enhance the scientific quality of vaccine trials.
Financial & competing interests disclosure
The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.
References
- Keja K, Chan C, Hayden G, Henderson RH. Expanded programme on immunization. World Health Stat. Q.41(2), 59–63 (1988).
- NIAID. National Institute of Allergy and Infectious Disease. Accelerated Development of Vaccines: the Jordan Report (2000).
- Clarke KC. Analytical and Computer Cartography (2nd Edition). Prentice-Hall, Englewood Cliffs, NJ, USA (1995).
- Ali M, de Francisco A, Khan MM, Chakraborty J, Myaux J. Factors affecting the performance of family planning workers: importance of geographical information system in empirical analysis. Int. J. Popul. Geogr.5, 19–29 (1999).
- Wilkinson D, Tanser F. GIS/GPS to document increased access to community-based treatment for tuberculosis in Africa. Geographic information system/global positioning system. Lancet354(9176), 394–395 (1999).
- Longini IM Jr, Halloran ME, Nizam A. Model-based estimation of vaccine effects from community vaccine trials. Stat. Med.21(4), 481–495 (2000).
- Ross R. An application of the theory of probabilities to the study of a priori pathometry, part 1. Proc. R. Soc. Lond. Ser. A92, 204–230 (1916).
- Halloran ME, Longini IM Jr, Struchiner CJ. Design and interpretation of vaccine field studies. Epidemiol. Rev.21(1), 73–88 (1999).
- Struchiner CJ, Halloran ME, Robins JM, Spielman A. The behaviour of common measures of association used to assess a vaccination programme under complex disease transmission patterns – a computer simulation study of malaria vaccines. Int. J. Epidemiol.19(1), 187–196 (1990).
- Anderson RM, May RM. Immunisation and herd immunity. Lancet335(8690), 641–645 (1990).
- Fine PE. Herd immunity: history, theory, practice. Epidemiol. Rev.15(2), 265–302 (1993).
- Fox JP, Elveback L, Scott W, Gatewood L, Ackerman E. Herd immunity: basic concept and relevance to public health immunization practices.1971. Am. J. Epidemiol.141(3), 187–197 (1995).
- Greenland S, Frerichs RR. On measures and models for the effectiveness of vaccines and vaccination programmes. Int. J. Epidemiol.17(2), 456–463 (1988).
- Halloran ME, Struchiner CJ. Study designs for dependent happenings. Epidemiology2(5), 331–338 (1991).
- Robbins JB, Schneerson R, Anderson P, Smith DH. The 1996 Albert Lasker Medical Research Awards. Prevention of systemic infections, especially meningitis, caused by Haemophilus influenzae type b. Impact on public health and implications for other polysaccharide-based vaccines. JAMA276(14), 1181–1185 (1996).
- Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. Arnold, London, UK (2000).
- Ali M, Emch M, von Seidlein L et al. Herd immunity conferred by killed oral cholera vaccines in Bangladesh: a reanalysis. Lancet366(9479), 44–49 (2005).
- Ali M, Emch M, Yunus M et al. Vaccination of adult women against cholera protects young children in rural Bangladesh. Pediatr. Infect. Dis. (In press) (2008).
- Longini IM, Azhar N, Ali M, Yunus M, Shenvi N, Clemens J. Controlling endemic cholera with oral vaccine. PLoS Med.4(11), e336 (2007).
- Krieger N. Place, space, and health: GIS and epidemiology. Epidemiology14(4), 384–385 (2003).