Epidemiology of Equine Infectious Disease

Table 65-2 A 2 × 2 Table Constructed for Chi-Square Test (as Described in Text)



Table 65-3 Example of Chi-Square Table (as Described in Text)



CHAPTER 65 Epidemiology of Equine Infectious Disease




BASICS



Definition


Epidemiology is the study of the occurrence of disease in populations1 and the application of this knowledge to control or prevent disease. The underlying tenet of epidemiology is that disease does not occur randomly in a population. This means that there are always reasons why some horses become sick and others stay healthy, even if we do not always understand those reasons. This has tremendous implications for veterinarians: they can identify causes and risk factors for disease and take actions to prevent or decrease the impact of a disease. In this sense, it is critical to consider more than just disease agents and hosts; it is also critical to consider the environment and management factors that impact interactions among agents and hosts. Because of this broader implication, epidemiology is also sometimes defined as “medical ecology,” as discussed below.


The mare reproductive loss syndrome outbreak in Kentucky in 2001 was an excellent example of epidemiology in action.2 Even before veterinarians and producers understood the etiology of this disease, veterinarians were able to use epidemiology to identify risk factors for disease (exposure to eastern tent caterpillars2 or pasture3), which allowed farm managers to implement control measures and prevent some abortions that would have otherwise occurred.



Epidemiologic Approach


A key epidemiologic approach to understanding and controlling disease involves looking for patterns of disease in the population of interest. Which horses are sick, and what do they have in common? Which horses are healthy, and what do they have in common? Which groups have been most affected? Great insight can be gained into causal mechanisms and control points that can be exploited in disease prevention efforts by (1) describing a population and identifying patterns, (2) making comparisons among different groups within a population, (3) comparing different populations, and (4) comparing the same population at different time points.


It is useful to consider the “five Ws” when trying to understand disease occurrence in a population: who, what, where, when, and why. Who is affected (and unaffected)? Include age, breed, gender, housing, water source, and vaccination status, as well as any other variables that may be relevant. What are the circumstances related to disease occurrence, and has anything changed? Where are the affected and unaffected animals located? Use a map of the barn or farm with food, water, and ventilation sources marked, and spatially locate the ill animals. When did each ill animal develop disease? Use this information to identify groups most affected (e.g., age groups, barns, breeds) using the tools described later in this chapter. All this should be interpreted with a focus on ultimately identifying “why.” Why did these animals develop disease, and why were others not affected? Understanding why disease occurs allows identification of ways that disease can be prevented.



Disease Ecology


When trying to understand reasons why particular animals become diseased, it is clearly important to consider more than just an individual host and a particular agent as causes for a specific occurrence. The population to which an individual belongs must also be considered, in addition to the patterns of interactions and the environment that influences these interactions and impacts the likelihood of contagious transmission. Because of the importance of these broader considerations, epidemiology is sometimes referred to as medical ecology, or the interactions of all organisms and their environment as these pertain to health.


Mare reproductive loss syndrome (MRLS), which was initially reported among broodmares in central Kentucky in 2001, is one example of disease arising from a combination of host, agent, and environmental factors. An epidemic of equine abortion, endophthalmitis, and pericarditis began in late April 2001 and lasted until June 2001;46 fetal losses occurred both early and late in pregnancy3,7 and affected more than 60% of mares on some farms.7 Multiple bacterial species were identified8 in tissue of aborted fetuses. The syndrome was subsequently found to be associated with ingestion of the eastern tent caterpillar,4 and it has been proposed that bacterially contaminated barbed caterpillar hairs migrated out of the alimentary tract, spread hematogenously, and were directly responsible for the observable signs of MRLS.9 Eastern tent caterpillars are ubiquitous in the eastern United States but were particularly abundant in Kentucky that spring because a rapid temperature increase in early spring was superimposed on an unusually dry winter and spring.10 These climatic conditions caused an explosion of biologic activity, including growth of black cherry trees on which eastern tent caterpillar eggs are laid and larvae develop.11 During that spring with its unusual climactic conditions, grazing on pasture4 with black cherry trees12 exposed horses to disease; fetuses were particularly vulnerable. The sensitivity of the fetus to disease, the environmental conditions that led to the overgrowth of caterpillars, the bacteria themselves, and the management of the broodmares all contributed to the occurrence of MRLS.



Disease Agent


Characteristics of the disease agent, including infectivity, contagiousness, pathogenicity and virulence, immunogenicity, host range, life cycle, and antimicrobial susceptibility, influence the speed and scope of disease spread. Infectiousness (infectivity) refers to the ease with which an agent infects susceptible hosts, which is sometimes quantified in relation to the amount of agent required to reliably infect an individual. Contagiousness relates to the likelihood that an agent will move between infected and susceptible hosts; it is sometimes quantified by the number of new infections that will likely result from exposure to an infected animal or as the speed with which a disease agent is transmitted through a susceptible population. Equine influenza virus and equine herpesvirus are both highly infectious, but influenza virus is more contagious. Although equine protozoal myeloencephalitis (EPM) is an infectious disease, it is not a contagious disease because the etiologic agent is not transmitted directly between horses. Pathogenicity describes the likelihood that an infected horse will develop clinical disease, and virulence describes the likelihood that disease will be severe. West Nile virus (WNV) is highly virulent in horses; more than 30% of horses with clinical disease die.13 In contrast, EPM is not highly pathogenic; most equids exposed to the disease agent do not develop clinical disease.1416


Characteristics of the disease agent that enable it to survive and spread without detection are particularly important to consider when instituting preventive or control measures. Agents that can persist in the environment, such as Clostridium difficile17 or Streptococcus equi subsp. equi, require different control measures than does equine influenza, which does not persist well outside the host. Some diseases spread undetected through infected horses without clinical signs of illness. Subclinically, persistently, and latently infected animals are often important reservoirs and sources of exposure for susceptible animals in a population because they go unnoticed or undiagnosed. Animals often are infected with a potentially pathogenic organism without showing clinical signs, and this can even be the predominant presentation depending on the pathogenicity of the agent. The term subclinical is also used to describe animals during the induction or incubation period for infectious diseases. Animals that remain infected for extended periods are sometimes described as being “persistently infected,” especially if infections continue after clinical signs of disease resolve. Persistent infection and long-term shedding of S. equi subsp. equi are common1820 and important to the spread of disease among populations.20,21 In contrast, latency describes a state of dormant viral infection in which shedding stops and the virus cannot be detected until later, when the infection reactivates or recrudesces. This is a common feature among alpha herpesviruses, such as equine herpesvirus (EHV) types 1 and 4.2229




Environment


A horse’s environment includes its location, climate, and the local surroundings and interactions created by its management.30 Characteristics of a horse’s environment affect which diseases and vectors a horse is exposed to, the magnitude of that exposure, and the likelihood of developing disease if exposed. Horses that have been vaccinated with efficacious vaccines or immunized by natural exposure are more resistant to a particular disease than naive horses. Horses that are stressed for any reason, including poor diet, concurrent disease, weaning, transport, or mixing, are more likely to develop a disease than their unstressed counterparts. The risk of disease is not equal for similar horses when managed differently or housed in different environments.




Population.


In addition to characteristics of individuals that affect their disease risk, the aggregate characteristics of the population to which the individual belongs affects the disease risk for that individual. This aggregate of the population’s susceptibility to disease is often called herd immunity, described as immunity of an individual that is conferred by the population to which it belongs, or the ability of a population of animals to withstand exposure without succumbing to disease because the immunity of a population is more than the sum of its parts.31 Herd immunity is created when the likelihood is small that an infected horse shedding a disease agent will encounter a susceptible horse (Fig. 65-1). If most horses are immune or if contact among horses is heavily restricted, it is unlikely that the few susceptible horses will have contact with the infected horse sufficient to allow transmission. For example, consider a barn in which 90% of horses are immune to influenza virus, and each horse in the barn contacts four other horses per day. If a newly introduced horse happens to be infected with influenza virus, and conditions are adequate for transmission of virus to in-contact horses, the probability that any other horse in the barn will become infected is about 35%, and the probability that more than one horse will become infected is about 12% (Fig. 65-1). For herd immunity to be effective, the disease agent (e.g., influenza) must only reside in horses, must not have an environmental reservoir, must be transmitted directly from horse to horse, and must have a short infectious period.32




Disease Causation


Many veterinarians are accustomed to thinking that infectious diseases have a single cause: the disease agent. Epidemiologists think of cause in a more general sense. Any “exposure” that leads to new cases of disease can be considered a “cause” of that disease. By removing that exposure, therefore, some cases of disease can be prevented. Causation has multiple levels. Again, consider MRLS. What “exposures” are associated with MRLS? The bacteria on the caterpillar hairs,8 the caterpillars themselves,33 exposure to cherry trees,2 pasture grazing,3,7 and the convergence of climatic factors resulting in caterpillar overgrowth have all been implicated in the epidemic occurrence of MRLS. Epidemiologically, all these factors are causes. Cases of disease can be reduced by removing bacteria from the caterpillars (an obviously impractical approach), minimizing exposure to caterpillars, reducing exposure to environments shared with the caterpillars (cherry trees or pasture), or returning to a more typical climate, as happened in subsequent years.


Likewise, consider encephalitis associated with WNV infection. This agent is propagated in a mosquito-bird-mosquito life cycle.34 West Nile virus can replicate in multiple mosquito species,35 although its primary vectors are Culex mosquitoes36 The primary hosts are birds,37 which develop transient viremia followed by long-term (lifelong) immunity. Ticks may also play a role in maintenance of WNV.35,38,39 New cases of equine disease can therefore be reduced by minimizing exposure to mosquitoes and ticks (e.g., controlling vector populations, using animal-safe repellents, housing horses inside at dusk and dawn) or by increasing immunity to the virus, either in birds or in horses. In birds the natural immunity that develops after initial exposure reduces the total amount of virus circulating in the mosquito population, which in turn reduces equine exposures. The increase in WNV immunity among birds is one likely reason that the number of reported equine cases of WNV-associated disease in Colorado was 378 and 426 in 2002 and 2003, respectively, and then decreased to 33 in 2004.40


One useful model used to understand complex causal relationships is to classify causes as component causes, necessary causes, or sufficient causes.31 In this model, a component cause is anything that contributes to new cases of disease. Component causes can be characteristics of the host (e.g., age, vaccination status), the agent (e.g., subtype), or the environment (e.g., presence or absence of caterpillars). A sufficient cause is any set of components that, when present together, is capable of causing disease; once a sufficient cause is present or complete, disease will occur. A necessary cause is a component cause that must be present for disease to occur; without the necessary cause, disease cannot occur. For infectious diseases, exposure or infection with the infectious agent is never sufficient by itself, but it is necessary for disease to occur.


Using this model, we can see that there may be multiple sufficient causes, and that disease can “flow” through any of these paths. Thus, removing exposure to one component cause will only mitigate disease through the sufficient causal paths that include that particular component cause. When a particular component cause is part of a high proportion of sufficient pathways, then exposure to this particular component cause will be strongly associated with disease occurrence. The extreme of this example is when a component cause is included in all sufficient causes, in which case the component cause can also be called a “necessary cause.” Necessary causes are rare among all component causes, and there are always multiple sufficient causal sets. Thus, by removing exposure to some of the component causes, we only expect to prevent some disease occurrence and not all occurrences. The objective is to maximize efficiency of disease prevention efforts by targeting component causes that are strongly associated with disease occurrence.



IDENTIFYING CAUSAL FACTORS


In epidemiologic studies the main objective is to determine the factors (risk factors) associated with occurrence of disease so that they can be targeted in control and prevention programs. In general, we identify risk factors for a disease by comparing measures of disease frequency between different populations or groups. More specifically, this is accomplished by summarizing the occurrence of disease in the population, measuring disease frequency, and then comparing the risk of disease among horses with different exposures. By identifying differences in disease risk for groups with different exposures, we determine which exposures are associated with disease. Multiple studies are required to label exposures confidently as risk factors that are truly causal, which can then be targeted for minimizing exposure and thereby reducing the occurrence of new cases.


For example, in a study to identify the risk factors associated with equine protozoal myeloencephalitis (EPM), horses affected with EPM and nonaffected horses were compared using a case-control study design.41 In that study, presence of opossums on the premises, lack of feed security, and recent occurrences of major health events, among other factors, were associated with an increased likelihood of disease and thus were identified as potential risk factors for the disease.41



Measuring Disease


The frequency of disease occurrence is measured for different purposes, including determining and comparing the health status of populations, monitoring changes in disease occurrence over time, and establishing the risks associated with certain events in the population. For example, the practitioner might be interested in the number or proportion of diseased horses in a herd, the increase or decrease in the number of disease cases over time, or the risk of disease introduction associated with new horses introduced to the herd. Common measures of disease frequency include prevalence and incidence, as well as related measures such as attack, case-fatality, and mortality risks.


Disease can be measured in whole populations or in specific subgroups. Measuring disease in the whole population (sometimes called a “crude” measure) tells you about the overall scope of the problem. Measuring disease in subgroups (called “specific” measures) enables you to compare those groups, which is essential when attempting to identify factors affecting the occurrence of disease. For example, if you had 10 cases of neurologic disease on a farm of 100 horses, you could report that 10% of horses on the farm are affected, as a crude prevalence estimate. In contrast, you could also report that 8 of 25 (32%) horses grazing in Pasture A were affected and 2 of 75 (3%) horses in Pasture B were affected. These pasture-specific attack rates suggest that something associated with housing in Pasture A may be the problem.




Types of Data


The types of data available largely determine which methods will be most appropriate for measuring disease frequency or comparing disease risk. Most data can be described as interval (measurement) or categorical data. Interval data quantify a characteristic such as temperature, age, or weight, which can be measured as an infinite number of possible values. For example, in a group of five horses, you might take temperature measurements of 100°, 100°, 101°, 101.5°, and 102° F. Average or median values are often used to summarize interval data, and for this example the average temperature for these five horses is 100.9° F. Interval data are often compared by subtracting one average or median from another, and differences in interval measurements among groups are often statistically tested using z-tests, t-tests, and analyses of variance (ANOVA).


Categorical data divide groups into mutually exclusive categories (e.g., young horses vs. older horses, Quarter Horses vs. Thoroughbreds), and counts are used to characterize horses fitting into each category. Categorical data can be further characterized as “ordinal” if categories have an inherent order to them (e.g., young or old, light or heavy) or “nominal” if categories cannot be numerically ordered (e.g., categories for gender or breed). Interval measurements can be converted to ordinal measurements by dividing your range of values into categories. For example, if you wanted to describe temperature ordinally, as <101.5° F vs. temperature ≥101.5° F for these five horses, you would report that three horses had temperatures <101.5° F, and two had temperatures ≥101.5° F. Categories for nominal or ordinal data can be dichotomous (only two values are possible; e.g., live/dead, yes/no, sick/well) or can have more than two possible values (e.g., breed, age group). Ordinal and nominal data are summarized using ratios, proportions, and rates. These summary measurements can then be compared using relative risks, odds ratios, and attributable risks. These comparisons are often tested statistically using different types of chi-square tests.


Generally, all characterizations of disease occurrence in populations include some type of categorical assessment of presence or absence of disease signs using a specific case definition. In summarizing these measurements, the data are standardized to account for population sizes using the number of affected animals as a numerator, and some context measurement of the “opportunity” for disease to have occurred in the population (e.g., the population at risk). The denominator that we choose greatly affects the conclusions we can draw from these measurements of disease frequency. In general, these measures of disease frequency take the form of ratios, proportions, or rates.




Using Categorical Data



Ratios, Proportions, and Rates.


When using ratios, proportions, and rates to summarize disease occurrence, a count of affected animals meeting a specific case definition is used as the numerator. Do 10 affected horses represent a significant number of cases? The answer depends on the type of disease and the size of the population in which these observations were made. Are we referring to 10 sick horses in a barn of 15 horses, or 10 sick horses at an entire racetrack facility with 3500 horses? The denominator provides context (e.g., is the PAR 15 or 3500 horses) and improves standardization and the ability to extrapolate or make comparisons. The type of denominator we choose affects the conclusions we can draw. The ratios, proportions, and rates used as epidemiologic measures principally differ in how the denominator is calculated (e.g., which animals are included, is time considered).






Epidemiologic Measures of Disease Frequency




Cumulative Incidence.

The cumulative incidence (CI) is the proportion of new cases of disease occurring in a population during a specific time period43 and is calculated as follows:



image



Cumulative incidence is used to assess the progression of disease in the population during a specific time period and can be used to predict disease occurrence. The cumulative incidence measures the risk or probability of becoming diseased in a population during a defined time period (Box 65-2).



The cumulative incidence is an appropriate measure of disease incidence when the population is relatively “closed” (i.e., minimal movement of animals in and out of the population). When there is substantial movement of animals (“open” population), the cumulative incidence might underestimate or overestimate (bias) disease incidence,43 and the incidence rate, also called incidence density, is a more appropriate measure of disease incidence.


Other common measures that could be described as specific types of cumulative incidence include the attack rate, mortality rate, and the case-fatality rate. The attack rate is simply a different name attributed to the cumulative incidence in an outbreak situation and is calculated exactly as the cumulative incidence. The mortality rate is the proportion of all deaths (“crude” mortality rate) or deaths attributable to a specific disease (“cause-specific” mortality rate) over the total PAR of death at the beginning of the time period (Box 65-3). Note that these measures are called “rates,” but in reality they are proportions because they do not include time measurements in their denominator.



Mortality can be calculated as a proportion, as just noted, or as an incidence using one of the methods described next. The term “mortality rate” is commonly used to describe mortality in the population whether it is a proportion or an actual rate.


The case-fatality rate is the proportion of deaths attributable to the disease of interest during a specific period of time (Box 65-4).43 The case-fatality rate is calculated as follows:




image



This measure of disease occurrence is often used to characterize the severity of disease and the effectiveness of treatment. Therefore the specific case-fatality rates for treated and untreated animals are often cited and compared.

Stay updated, free articles. Join our Telegram channel

Jun 8, 2016 | Posted by in EQUINE MEDICINE | Comments Off on Epidemiology of Equine Infectious Disease

Full access? Get Clinical Tree

Get Clinical Tree app for offline access