Chapter 1 What is epidemiology? Key messages Epidemiology is a fundamental science of public health.

. Epidemiology has made major contributions to improving population health. Epidemiology is essential to the process of identifying and mapping emerging diseases. There is often a frustrating delay between acquiring epidemiological evidence and applying this evidence to health policy. The historical context Origins Epidemiology originates from Hippocrates observation more than 2000 years ago that environmental factors influence the occurrence of disease. However, it was not until the nineteenth century that the distribution of disease in specific human population groups was measured to any large extent. This work marked not only the formal beginnings of epidemiology but also some of its most spectacular achievements. 1 The finding by John Snow that the risk of cholera in London was related to the drinking of water supplied by a particular company provides a well-known example; the map (see Figure 4.1) highlights the clustering of cases. Snows epidemiological studies were one aspect of a wideranging series of investigations that examined related physical, chemical, biological, sociological and political processes. 2Comparing rates of disease in subgroups of the human population became common practice in the late nineteenth and early twentieth centuries. This approach was initially applied to the control of communicable diseases (see Chapter 7), but proved to be a useful way of linking environmental conditions or agents to specific diseases. In the second half of the twentieth century, these methods were applied to chronic noncommunicable diseases such as heart disease and cancer, especially in middleand high-income countries. Recent developments in epidemiology Epidemiology in its modern form is a relatively new discipline1 and uses quantitative methods to study diseases in human populations to inform prevention and control efforts. For example, Richard Doll and Andrew Hill studied the relationship between tobacco use and lung cancer, beginning in the 1950s.4 Their work was preceded by experimental studies on the carcinogenicity of tobacco tars and by clinical observations linking tobacco use and other possible factors to lung cancer. By using longterm cohort studies, they were able to establish the association between smoking and lung cancer (Figure 1.1). Definition, scope, and uses of epidemiology

Definition Epidemiology as defined by Last9 is the study of the distribution and determinants of health-related states or events in specified populations, and the application of this study to the prevention and control of health problems (see Box 1.2). Epidemiologists are concerned not only with death, illness and disability, but also with more positive health states and, most importantly, with the

means to improve health. The term disease encompasses all unfavourable health changes, including injuries and mental health. Definition of epidemiology Epidemiology is the study of disease in populations and of factors that determine its occurrence; the key word being populations. Veterinary epidemiology additionally includes investigation and assessment of other health-related events, notably productivity. All of these investigations involve observing animal populations and making inferences from the observations. l A literal translation of the word 'epidemiology', based on its Greek roots e - (epi-) = upon, o- (demo-) = people, and o'o- (logo-) = discoursing, is 'the study of that which is upon the people' or, in modern parlance, 'the study of disease in populations'. Traditionally, epidemiology' related to studies of human populations, and epizootiology', from the Greek wo- (zoo-) = , to the studies of animal (excluding human) populations (e.g., Karstad, 1962). Outbreaks of disease in human populations were called 'epidemics', in animal populations were called 'epizootics', and in avian popula tions were called 'epornitics', from the Greek opvt-(ornith-) = bird (e.g., Montgomery et al., 1979). Other derivatives, such as 'epidemein' ('to visit a community'), give hints of the early association between epidemiology and infections that periodically entered a community, in contrast to other diseases which were usually present in the population. The uses of epidemiology

There are five objectives of epidemiology: I . determination of the origin of a disease whose cause is known; 2. investigation and control of a disease whose cause is either unknown or poorly understood; 3. acquisition of information on the ecology and natural history of a disease; 4. planning and monitoring of disease control programmes; 5. assessment of the economic effects of a disease and analysis of the costs and economic benefits of alternative control programmes Determination of the origin of a disease whose cause is known Many diseases with a known cause can be diagnosed precisely by the signs exhibited by the affected animals, by appropriate laboratory tests and by other clinical procedures such as radiological investigation. For instance, the diagnosis of salmonellosis in a group of calves is relatively straightforward (the infection frequently produces distinct clinical signs). However, determining why an outbreak occurred and using the correct procedures to prevent recurrence can be difficult. For example, the outbreak may have been caused either by the purchase of infected animals or by contaminated food. Further investigations are required to identify the source of infection. When the food is suspected, the ration may consist of several components. Even if a sample of each component is still available, it would be expensive and

possibly uneconomic to submit all of the samples for laboratory examination. Consideration of the risk associated with the consumption of each component of the ration may narrow the field of investigation to only one or two items. There are many examples of the investigation of diseases with known causes that involve answering the questions `Why has an outbreak occurred?' or `Why has the number of cases increased?'. For instance, an increased number of actinobacillosis cases in a group of cattle might be associated with grazing a particular pasture of `burnt off' stubble. Such an occurrence could be associated with an increase in abrasions of the buccal mucosae which could increase the animals' susceptibility to infection with Actinobacillus lignieresi. An increased number of cases of bone defects in puppies might be due to local publicity given to the use of vitamin supplements, resulting in their administration to animals that were already fed a balanced diet, with consequent hypervitaminosis D and osteodystrophy (Jubb and Kennedy, 1971). An increase in the number of lamb carcasses with high ultimate pH values could be associated with excessive washing of the animals prior to slaughter (Petersen, 1983). These possible explanations can be verified only by epidemiological investigations. Investigation and control of a disease whose cause is either unknown or poorly understood There are many instances of disease control based on epidemiological observations before a cause was identified. Contagious bovine pleuropneumonia was eradicated from the US by an appreciation of the infectious nature of the disease before the causal agent, Mycoplasma mycoides, was isolated (Schwabe, 1984). Lancisi's slaughter policy to control rinderpest, mentioned in Chapter 1, was based on the assumption that the disease was infectious, even though the causal agent had not been discovered. Edward Jenner's classical observations on the protective effects of cowpox virus against human smallpox infection in the 18th century (Fisk, 1959), before viruses were isolated, laid the foundations for the global eradication of smallpox. More recently, epidemiological studies in the UK suggested that cattle develop bovine spongiform encephalopathy (BSE) following consumption of feedstuffs containing meat and bone meal contaminated with a scrapie-like agent (Wilesmith et al., 1988). This was sufficient to introduce legislation prohibiting the feeding of ruminant derived protein, although the causal agent had not been identified. Although the exact cause of `blood splashing' (eccymoses in muscle) in carcasses is still not known, observations have shown that there is a correlation between this defect and electrical stunning by a `head only' method (Blackmore, 1983); and the occurrence of this condition can be reduced by adopting a short 'stun-to-stick' interval, stunning animals with a captive bolt, or using a method of electrical stunning that causes concurrent cardiac dysfunction (Gracey, 1986). Similarly, there is a strong correlation between grass sickness and grazing, and the disease can be almost totally prevented by stabling horses continuously during spring and summer, although the cause of the disease is unknown (Gilmour, 1989). The cause of squamous cell carcinoma of the eye in Hereford cattle ('cancer eye') is not known. Epidemiological studies have shown that animals with unpigmented eyelids are much more likely to develop the condition than those with pigment (Anderson et al., 1957). This

information can be utilized by cattle breeders to select animals with a low susceptibility to this neoplasm. Epidemiological studies are also used to identify causes of disease (many of which are multifactorial and initially poorly understood) so that the most appropriate disease control techniques can be applied. Thus, the identification of low levels of water intake as an important component of the cause of feline urolithiasis (Willeberg, 1981) facilitated control of this syndrome by dietary modification. Investigations can also be used to identify characteristics of animals that increase the risk of disease. For example, entire bitches with a history of oestrus irregularity and pseudopregnancy are particularly at risk of developing pyometra (Fidler et al., 1966); this information is of diagnostic value to the clinician, and is of assistance when advising owners on breeding policy. Acquisition of information on the ecology and natural history of a disease An animal that can become infected with an infectious agent is a host of that agent. Hosts and agents exist in communities that include other organisms, all of which live in particular environments. The aggregate of all facts relating to animals and plants is their natural history. Related communities and their environments are termed ecosystems. The study of ecosystems is ecology . A comprehensive understanding of the natural history of infectious agents is possible only when they are studied in the context of their hosts' ecosystems. Similarly, an improved knowledge of non-infectious diseases can be obtained by studying the ecosystems and the associated physical features with which affected animals are related. The geological structure of an ecosystem, for example, can affect the mineral content of plants and therefore can be an important factor in the occurrence of mineral deficiencies and excesses in animals. The environment of an ecosystem affects the survival rate of infectious agents and of their hosts. Thus, infection with the helminth Fasciola hepatica is a serious problem only in poorly drained areas, because the parasite spends part of its life-cycle in a snail which requires moist surroundings. Each of the 200 antigenic types (serovars) of Leptospira interrogans is maintained in one or more species of hosts. Serovar copenhageni, for instance, is maintained primarily in rats (Babudieri, 1958). Thus, if this serovar is associated with leptospirosis in either man or domestic stock, then part of a disease control programme must involve an ecological study of rat populations and control of infected rats. Similarly, in Africa, a herpesvirus that produces infections without signs in wildebeeste is responsible for malignant catarrhal fever of cattle (Plowright et al., 1960). Wildebeeste populations, therefore, must be investigated when attempting to control the disease in cattle. An ecosystem's climate also is important because it limits the geographical distribution of infectious agents that are transmitted by arthropods by limiting the distribution of the arthropods. For example, the tsetse fly, which transmits trypanosomiasis, is restricted to the humid parts of Sub-Saharan Africa (Ford, 1971). Infectious agents may extend beyond the ecosystems of their traditional hosts. This has occurred in bovine tuberculosis in the UK, where the badger population appears to be an alternative host for Mycobacterium tuberculosis (Little et al., 1982; Wilesmith et al., 1982). Similarly, in certain areas of New Zealand, wild opossums are infected with this bacterium and can therefore be a source of infection to cattle (Thorns and Morris, 1983). Purposeful routine observation of such infections provides valuable information on changes in the amount of disease and relevant ecological factors and may therefore indicate necessary changes in control

strategies. Infectious diseases that are transmitted by insects, ticks and other arthropods, and which may be maintained in wildlife, present complex ecological relationships and even more complex problems relating to their control. Comprehensive epidemiological studies of these diseases help to unravel their life -cycles, and can indicate suitable methods of control. Planning and monitoring of disease control programmes The institution of a programme to either control or eradicate a disease in an animal population must be based on a knowledge of the amount of the disease in that population, the factors associated with its occurrence, the facilities required to control the disease, and the costs and benefits involved. This information is equally important for a mastitis control programme on a single dairy farm and for a national brucellosis eradication scheme involving all the herds in a country. The epidemiological techniques that are employed include the routine collection of data on disease in populations (monitoring and surveillance) to decide if the various strategies are being successful. Surveillance is also required to determine whether the occurrence of a disease is being affected by new factors. For example, during the eradication scheme for bovine

18

The scope of epidemiology

tuberculosis in New Zealand, opossums became infected in certain areas. New strategies had to be introduced to control this problem (Julian, 1981). During the foot-andmouth disease epidemic in the UK in 1967 and 1968, surveillance programmes indicated the importance of wind-borne virus particles in the transmission of the disease (Smith and Hugh-Jones, 1969). This additional knowledge was relevant to the establishment of areas within which there was a restriction of animal movement, thus facilitating eradication of the disease. Assessing the economic effects of a disease and of its control The cost of the control of disease in the livestock industry must be balanced against the economic loss attributable to the disease. Economic analysis therefore is required. This is an essential part of most modern planned animal health programmes. Although it may be economic to reduce a high level of disease in a herd or flock, it may be uneconomic to reduce even further the level of a disease that is present at only a very low level. If 15 % of the cows in a herd were affected by mastitis, productivity would be severely affected and a control programme would be likely to reap financial benefit. On the other hand, if less than I % of the herd were affected, the cost of further reduction of the disease might not result in a sufficient increase in productivity to pay for the control programme.

This introduction to the uses of epidemiology indicates that the subject is relevant to many areas of veterinary science. The general agricultural practitioner is becoming increasingly concerned with herd health. The companion animal practitioner is faced with chronic refractory diseases, such as the idiopathic dermatoses, which may be understood better by an investigation of the factors that are common to all cases. The state veterinarian cannot perform his routine duties without reference to disease in the national animal population. The diagnostic pathologist investigates the associations between causes and effects (i.e., lesions); this approach is epidemiological when inferences are made from groups of animals. The veterinarian in abattoirs and meat-processing plants attempts to reduce the occurrence of defects and contamination by understanding and eliminating their causes. Similarly, industrial veterinarians, concerned with the design of clinical trials, compare disease rates and response to treatment in groups of animals to which different prophylactic and therapeutic compounds are administered. Types of epidemiological investigation

There are four approaches to epidemiological investigation that traditionally have been called `types' of epidemiology. These types are descriptive, analytical, experimental and theoretical epidemiology.

18

The scope of epidemiology

Descriptive epidemiology Descriptive epidemiology involves observing and recording diseases and possible causal factors. It is usually the first part of an investigation. The observations are sometimes partially subjective, but, in common with observations in other scientific disciplines, may generate hypotheses that can be tested more rigorously later. Darwin's theory of evolution, for example, was derived mainly from subjective observations, but with slight modification it has withstood rigorous testing by plant and animal scientists. Analytical epidemiology Analytical epidemiology is the analysis of observations using suitable diagnostic and statistical tests. Experimental epidemiology The experimental epidemiologist observes and analyses data from groups of animals from which he can select, and in which he can alter, the factors associated with the groups. An important component of the experimental approach is the control of the groups. Experimental epidemiology developed in the 1920s and 1930s, and utilized laboratory animals whose short lifespans enabled events to be observed more rapidly than in humans (see Chapter 18). A notable example is the work of Topley (1942) who infected colonies of mice with ectromelia virus and Pasteurella spp. The effects of varying the rate of exposure of mice maintained in groups of various sizes provided insights into the behaviour of human epidemic diseases such as measles, scarlet fever, whooping cough and diptheria which followed similar patterns to the experimental infections (MRC, 1938). This work demonstrated the importance of the proportion of susceptible individuals in the population in determining the progress of epidemics (see Chapter 8); hitherto, changes in the virulence of a microorganism were thought to be the most important factor affecting epidemic patterns. Rarely, a `natural' experiment can be conducted when the naturally occurring disease or other fortuitous circumstance approximates closely to the ideally designed experiment. For instance, when BSE occurred in the UK, outbreaks of the disease on the Channel Islands (Jersey

Types of epidemiological investigation

19

and Guernsey), which maintain isolated populations of cattle, provided an ideal situation in which to study the disease, uncomplicated by the possibility of transmission by contact with infected animals (Wilesmith, 1993). This added credence to the hypothesis that the disease was transmitted in contaminated feedstuffs. Theoretical epidemiology Theoretical epidemiology consists of the representation of disease using mathematical 'models' that attempt to simulate natural patterns of disease occurrence. Epidemiological subdisciplines Various epidemiological subdisciplines are now recognized. These reflect different areas of interest, rather than fundamentally different techniques. They all apply the four types of epidemiology described above, and can overlap, but their separate identities are considered by some to be justifiable. Clinical epidemiology Clinical epidemiology is the use of epidemiological principles, methods and findings in the care of individuals, with particular reference to diagnosis and prognosis (Last, 1988), and therefore brings a numerate approach to traditional clinical medicine, which has tended to be anecdotal and subjective (Grufferman and Kimm, 1984). It is concerned with the frequency and cause of disease, the factors that affect prognosis, the validity of diagnostic tests, and the effectiveness of therapeutic and preventive techniques (Fletcher et al., 1988). Computational epidemiology Computational epidemiology involves the application of computer science to epidemiological studies (Habtemariam et al., 1988). This includes the representation of disease by mathematical models (see 'Quantitative investigations', below) and the use of expert systems. These systems are commonly applied to disease diagnosis where they incorporate a set of rules for solving problems, details of clinical signs, lesions, laboratory results, and the opinions of experts; examples are the identification of the cause of coughing in dogs (Roudebush, 1984), and the diagnosis of bovine mastitis (Hogeveen et a!., 1993). Expert systems are also employed in formulating disease control strategies (e.g., for East coast fever: Gettinby and Byrom, 1989), predicting animal productivity (e.g., reproductive performance in dairy herds: McKay e t al., 1988), and supporting management decisions (e.g., decisions on replacing sows: Huirne et al., 1991). Genetic epidemiology Genetic epidemiology is the study of the cause, distribution and control of disease in related individuals, and of inherited defects in populations (Morton, 1982; Roberts, 1985). It indicates that the disciplinary boundary between genetics and epidemiology is blurred. Many diseases involve both genetic and non-genetic factors (see Chapter 5), and genes are increasingly incriminated in diseases of all organ systems ( F i g u r e 1.3). Thus, the geneticist and epidemiologist are both concerned with interactions between genetic and non-genetic factors - only the frequently indistinct time of interaction may be used to classify an investigation as genetic or epidemiological. Molecular epidemiology New biochemical techniques now enable microbiologists and molecular biologists to study small genetic and antigenic differences between viruses and other microorganisms at a higher level of discrimination than has been

possible using conventional serological techniques. The methods include peptide mapping, n ucleic acid 'fingerprinting' and hybridization (Keller and Manak, 1989; Kricka, 1992), restriction enzyme analysis, monoclonal antibodies (Oxford, 1985; Goldspink and Gerlach, 1990; Goldspink, 1993) and the polymerase chain reaction (Belak and Ballagi-Pordany, 1993). For example, nucleotide sequencing of European foot-and-mouth disease virus has indicated that recent outbreaks of the disease involved vaccinal strains, suggesting that improper inactivation or escape of virus from vaccine production plants may have been responsible for the outbreaks (Beck and Strohmaier, 1987). Similarly, infections that hitherto have been difficult to identify are now readily distinguished using these new molecular techniques; examples are infection with Mycobacterium paratuberculosis (the cause of Johne's disease) (Murray et al., 1989) and latent infection with Aujeszky's disease virus (Belak et al., 1989). The application of these new diagnostic techniques constitutes molecular epidemiology. A general description of the methods is given by Persing et al. (1993). Molecular epidemiology is part of the wider use of biological markers (Hulka et al., 1990). These are cellular, biochemical or molecular alterations that are measurable in biological media such as tissues, cells or fluids. They may indicate susceptibility to a causal factor, or a biological response, suggesting a sequence of events from exposure to disease (Perera and Weinstein, 1982). Some have been used by veterinarians for many years, for

instance, serum magnesium levels as indicators of susceptibility to clinical hypocalcaemia (Whitaker and Kelly, 1982; Van de Braak et al., 1987), serum transaminase levels as markers for liver disease, and antibodies as indicators of exposure to infectious agents (see Chapter 17). Other subdisciplines Several other epidemiological subdisciplines have also been defined. Chronic disease epidemiology is involved with diseases of long duration (e.g., cancers), many of which are noninfectious. Environmental epidemiology is concerned with the relationship between disease and environmental factors such as industrial pollution and, in human medicine, occupational hazards. Domestic animals can act as monitors of environmental hazards and can provide early warning of disease in man (see Chapter 18). Micro-epidemiology is the study of disease in a small group of individuals with respect to factors that influence its occurrence in larger segments of the population. For example, studies of feline acquired immunodeficiency syndrome (FAIDS) in groups of kittens have provided insights into the widespread human disease, AIDS (Torres-Anjel and Tshikuka, 1988). Micro-epidemiology, which frequently uses animal biological models of disease, therefore is closely related to comparative epidemiology (Chapter18). In contrast, macroepidemiology is the study of national patterns of disease, and the social, economic and political factors that influence them (Hueston and Walker, 1993). Other subdisciplines, such as nutritional epidemiology (Willett, 1990) and subclinical epidemiology (Evans, 1987), can also be identified to reflect particular areas of interest. The components of epidemiology are summarized in Figure 2.1. The first stage in any investigation is the collection of relevant data. The main sources of information are outlined in Chapter 10. Methods of storing, retrieving and disseminating information are discussed in Chapter 11. Investigations can be either qualitative or quantitative or a combination of these two approaches. Components of epidemiology Qualitative investigations The natural history of disease

The ecology of diseases, including the distribution, mode of transmission and maintenance of infectious diseases, is investigated by field observation. Ecological principles are outlined in Chapter 7. Methods of transmission and maintenance are described in Chapter 6, and patterns of disease occurrence are described in Chapter 8. Field observations also may reveal information about factors that may directly or indirectly cause disease. The various factors that act to produce disease are described in Chapter 5.

Causal hypothesis testing If field observations suggest that certain factors may be causally associated with a disease, then the association must be assessed by formulating a causal hypothesis. Causality (the relating of causes to effects) and hypothesis formulation are described in Chapter 3. Qualitative investigations were the mainstay of epidemiologists before the Second World War. These epidemiologists were concerned largely with the identification of unknown causes of infectious disease and sources of infection. Some interesting examples of the epidemiologist acting as a medical `detective' are described by Roueche (1991). Quantitative investigations Quantitative investigations involve measurement (e.g., the number of cases of disease), and therefore expression and analysis of numerical values. Basic methods of expressing these values are outlined in Chapters 4 and 12. The types of measurement that are encountered in veterinary medicine are described in Chapter 9. Quantitative investigations include surveys, monitoring and surveillance, studies, modelling, and the biological and economic evaluation of disease control. Surveys A survey is an examination of an aggregate of units (Kendall and Buckland, 1982). A group of animals is an example of an aggregate. The examination usually involves counting members of the aggregate and characteristics of the members. In epidemiological surveys, characteristics might include the presence of particular diseases, weight, and milk yield. Surveys can be undertaken on a sample of the population. Less commonly, a census, which examines the total animal population, can be undertaken (e.g., tuberculin testing). A cross-sectional survey records events occurring at a particular point in time. A longitudinal survey records events over a period of time. These latter events may be recorded prospectively from the present into the future; or may be a retrospective record of past event

Scope A focus of an epidemiological study is the population defined in geographical or other terms; for example, a specific group of hospital patients or factory workers could be the unit of study. A common population used in epidemiology is one selected from a specific area or country at a specific time. This forms the base for defining subgroups with respect to sex, age group or ethnicity. The structures of populations vary between geographical areas and time periods. Epidemiological analyses must take such variation into account. The scope of epidemiology

A particular type of diagnostic survey is screening. This is the identification of undiagnosed cases of disease using rapid tests or examinations. The aim is to separate individuals that probably have a disease from those that probably do not. Screening tests are not intended to be definitive; individuals with positive test results (i.e., that are classified as diseased by the screening test) require further investigation for definite diagnosis. Diagnostic tests, including serological surveys and screening, are considered in Chapter 17. The design of surveys in general is described in Chapter 13. Monitoring and surveillance Monitoring is the making of routine observations on health, productivity and environmental factors and the recording and transmission of these observations. Thus, the regular recording of milk yields is monitoring, as is the routine recording of meat inspection findings at abattoirs. The identity of individual diseased animals usually is not recorded. Surveillance is a more intensive form of data recording than monitoring. Originally, surveillance was used to describe the tracing and observation of people who were in contact with cases of infectious disease. It is now used in a much wider sense (Langmuir, 1965) to include all types of disease infectious and non-infectious - and involves the collation and interpretation of data collected during monitoring programmes, usually with the recording of the identity of diseased individuals, with a view to detecting changes in a population's health. It is normally part of control programmes for specific diseases. The recording of tuberculosis lesions at an abattoir, followed by tracing of infected animals from the abattoir back to their farms of origin, is an example of surveillance. The terms `monitoring' and `surveillance' have previously been used synonymously, but the distinction between them is now generally accepted. The national and international aspects of surveillance are reviewed by Blajan (1979), Davies (1980, 1993), Ellis (1980) and Blajan and Welte (1988), and some animal disease information systems are described in Chapter 11. Monitoring and surveillance can include all of the national herd. Alternatively, a few farms, abattoirs, veterinary practices or laboratories may be selected; these are then referred to as `sentinel' units, because they are designed to `keep watch' on a disease. Similarly, horses can be used as sentinels for Venezeulan equine encephalitis virus infection (Dickerman and Scherer, 1983), and stray dogs as sentinels for canine parvovirus infection (Gordon and Angrick, 1985), the infections being identified serologically. Other species of animals that also

are susceptible to an infectious agent can be used as sentinels for the infection in the main animal population. For example, wild birds can be used to monitor the activity of St Louis encephalitis virus, providing early information on the activity of the virus at a time when avian infection rates are still too low to pose an immediate threat to man (Lord et al., 1974). Domestic animals can also be used as sentinels of human environmental health hazards such as carcinogens and insecticides; this topic is discussed in detail in Chapter 18. Studies `Study' is a general term that refers to any type of investigation. However, in epidemiology, a study usually involves comparison of groups of animals, for example, a comparison of the weights of animals that are fed different diets. Thus, although a survey generally could be classified as a study, it is excluded from epidemiological studies because it involves only description rather than comparison and the analysis that the comparison requires. There are four main types of epidemiological study: 1. experimental studies; 2. cross-sectional studies; 3. case-control studies; 4. cohort studies. In an experimental study the investigator has the ability to allocate animals to various groups, according to factors which the investigator can randomly assign to animals (e.g., treatment regimen, preventive technique); such studies are therefore part of experimental epidemiology. An important example is the clinical trial. In a clinical trial, the investigator assigns animals either to a group to which a prophylactic or therapeutic procedure is applied, or to a control group. It is then possible to evaluate the efficacy of the procedure by comparing the two groups. Clinical trials are discussed in Chapter 16. The other types of study - cross-sectional, case-control and cohort - are observational. An observational study is similar to an experimental study: animals are allocated to groups with respect to certain characteristics that they possess (trait, disease, etc.). However, in observational studies, it is not possible to assign animals to groups randomly because the investigator has little control over the factors that are being studied; the characteristics are inherent (e.g., sex, weight or normal diet). A cross-sectional study investigates relationships between disease (or other health-related factors) and

hypothesized causal factors in a specified population. Animals are categorized according to presence and absence of disease and hypothesized causal factors; inferences can then be made about associations between disease and the hypothesized causal factors, for example, between heart valve incompetence (the disease) and breed (the hypothesized causal factor). A case-control study compares a group of diseased animals with a group of healthy animals with respect to exposure to hypothesized causal factors. For example, a group of cats with urolithiasis (the disease) can be compared with a group of cats without urolithiasis with respect to consumption of dry cat food (the factor) to determine whether that type of food has an effect on the pathogenesis of the disease. In a cohort study, a group exposed to factors is compared with a group not exposed to the factors with respect to the development of a disease. It is then possible to calculate a level of risk of developing the disease in relation to exposure to the hypothesized causal factors. Case-control and cohort studies have often been applied in human medicine in which experimental investigations of cause are usually unethical. For example, it would not be possible to investigate the suspected toxicity of a drug by intentionally administering the drug to a group of people in order to study possible sideeffects. However, if symptoms of toxicity have occurred, then a case-control study could be used to evaluate the association between the symptoms and the drug suspected of causing the toxicity. There are fewer ethical restraints on experimental investigation in veterinary medicine than in human medicine and so experimental investigation of serious conditions is more tenable. However, observational studies have a role in veterinary epidemiology; for example, when investigating diseases in farm and companion animal populations. The increasing concern for animal welfare is making these techniques even more useful than previously. Basic methods of assessing association between disease and hypothesized causal factors in observational studies are described in Chapters 14 and 15. Observational studies form the majority of epidemiological studies. Observational and experimental science have their own strengths and weaknesses which are discussed in detail by Trotter (1930). A major advantage of an observational investigation is that it studies the natural occurrence of disease. Experimentation may separate factors associated with disease from other factors that may have important interactions with them in natural outbreaks. Modelling Disease dynamics and the effects of different control strategies can be represented using mathematical equations. This representation is `modelling'. Many modern methods rely heavily on computers. Another type of modelling is biological simulation using experimental animals (frequently laboratory animals) to simulate the pathogenesis of diseases that occur naturally in animals and man. Additionally, the spontaneous occurrence of disease in animals can be studied in the field (e.g., using observational studies) to increase understanding of human diseases. Mathematical modelling is outlined in Chapter 19, and spontaneous disease models are described in Chapter 18. Disease control The goal of epidemiology is to improve the veterinarian's knowledge so that diseases can be controlled effectively, and productivity thereby optimized. This can be fulfilled by treatment, prevention or eradication. The economic evaluation of disease and its control is discussed in Chapter 20. Health schemes are described in Chapter 21. Finally, the principles of disease control are outlined in Chapter 22.

The different components of epidemiology apply the four epidemiological approaches to varying degrees. Surveys and studies, for example, consist of a descriptive and an analytical part. Modelling additionally may include a theoretical approach

Definition of epidemiology9 The word epidemiology is derived from the Greek words: epi upon, demos people and logos study. This broad definition of epidemiology can be further elaborated as follows: Term Explanation Study includes : surveillance, observation, hypothesis testing, analytic research and experiments. Distribution : refers to analysis of: times, persons, places and classes of people affected. Determinants : include factors that influence health: biological, chemical, physical, social, cultural, economic, genetic and behavioural. Health-related states and events : refer to: diseases, causes of death, behaviours such as use of tobacco,positive health states, reactions to preventive regimes and provision anduse of health services. Specified populations : include those with identifiable characteristics, such as occupational groups. Application to prevention and control : the aims of public healthto promote, protect, and restore health Epidemiology and public health Public health, broadly speaking, refers to collective actions to improve population health.1 Epidemiology, one of the tools for improving public health, is used in several ways (Figures 1.31.6). Early studies in epidemiology were concerned with the causes(etiology) of communicable diseases, and such work continues to be essential sinceit can lead to the identification of preventive methods. In this sense, epidemiology isa basic medical science with the goal of improving the health of populations, andespecially the health of the disadvantaged. Causation of disease Although some diseases are caused solely by genetic factors, most result from an interaction between genetic and environmental factors. Diabetes, for example, has both genetic and environmental components. We define environment broadly to include any biological, chemical, physical, psychological, economic or cultural factors that can affect health (see Chapter 9). Personal behaviours affect this interplay, and epidemiology is used to study their influence and the effects of preventive interventions through health promotion (Figure 1.3). Morbidity Death rates are particularly useful for investigating diseases with a high case-fatality. However, many diseases have low case-fatality, for example, most mental illnesses, musculoskeletal diseases, rheumatoid arthritis, chickenpox and mumps. In this situation, data on morbidity (illness) are more useful than mortality rates. Morbidity data are often helpful in clarifying the reasons for particular trends in mortality. Changes in death rates could be due to changes in morbidity rates or in case-fatality. For example, the recent decline in cardiovascular disease mortality rates in many developed countries could be due to a fall in either incidence (suggesting improvements in primary prevention) or in case-fatality (suggesting improvements in treatment). Because population age structures change with time, time-trend analyses should be based on age-standardized morbidity and mortality rates. Other sources of morbidity data include:

 hospital admissions and discharges outpatient and primary health care consultations specialist services (such as injury treatment) registers of disease events (such as cancer and congenital malformations). To be useful for epidemiological studies, the data must be relevant and easily accessible. In some countries, the confidential nature of patient medical records may make hospital data inaccessible for epidemiological studies. A recording system focusing on administrative or financial data, rather than on diagnostic and individual characteristics may diminish the epidemiological value of routine health service data. Hospital admission rates are influenced by factors other than the morbidity of the population,such as the availability of beds, hospital admission policies and social factors. Because of the numerous limitations of routinely recorded morbidity data, many epidemiological studies of morbidity rely on the collection of new data using specially designed questionnaires and screening methods. This enables investigators to have more confidence in the data and the rates calculated from them. Mortality Epidemiologists often investigate the health status of a population by starting with information that is routinely collected. In many high-income countries the fact and cause of death are recorded on a standard death certificate, which also carries information on age, sex, and place of residence. The International Statistical Classification of Diseases and Related Health Problems (ICD) provides guidelines on classifying deaths.14 The procedures are revised periodically to account for new diseases and changes in case-definitions, and are used for coding causes of death (see Box 2.2). The International Classification of Diseases is now in its 10th revision, so it is called the ICD-10. Limitations of death certificates Data derived from death statistics are prone to various sources of error but, from an epidemiological perspective, often provide invaluable information on trends in a populations health status. The usefulness of the data depends on many factors, including the completeness of records and the accuracy in assigning the underlying causes of deathespecially in elderly people for whom autopsy rates are often low. Epidemiologists rely heavily on death statistics for assessing the burden of disease, as well as for tracking changes in diseases over time. However, in many countries basic mortality statistics are not available, usually because of a lack of resources to establish routine vital registration systems. The provision of accurate cause-of-death information is a priority for health services.15 Death rates The death rate (or crude mortality rate) for all deaths or a specific cause of death is calculated as follows: Crude mortal it y rate = Number of deaths during a specified period ( 10n )

Number of persons at risk of dying during the same period The main disadvantage of the crude mortality rate is that it does not take into account the fact that the chance of dying varies according to age, sex, race, socioeconomic class and other factors. It is not usually appropriate to use it for comparing different time periods or geographical areas. For example, patterns of death in newly occupied urban developments with many young families are likely to be very different from those in seaside resorts, where retired people may choose to live. Comparisons of mortality rates between groups of diverse age structure are usually based on agestandardized rates. Age-specific death rates Death rates can be expressed for specific groups in a population which are definedby age, race, sex, occupation or geographical location, or for specific causes of death. For example, an age- and sex-specific death rate is defined as: Total number of deaths occurring in a specific age and sex group of the population in a defined area during a specified period Estimated total population of the same age and sex group of the population in the same area during the same period (10n ) Proportionate mortality Occasionally the mortality in a population is described by using proportionate mortality, which is actually a ratio: the number of deaths from a given cause per 100 or 1000 total deaths in the same period. Proportionate mortality does not express the risk of members of a population contracting or dying from a disease. Comparisons of proportionate rates between groups may show interesting differences. However, unless the crude or age-group-specific mortality rates are known, it may not be clear whether a difference between groups relates to variations in the numerators or the denominators. For example, proportionate mortality rates for cancer would be much greater in high-income countries with many old people than in low- and middle-income countries with few old people, even if the actual lifetime risk of cancer is the same. Infant mortality The infant mortality rate is commonly used as an indicator of the level of health in a community. It measures the rate of death in children during the first year of life, the denominator being the number of live births in the same year. The infant mortality rate is calculated as follows: Infant mortality rate = Number of deaths in a year of children less than 1 year of age Number of live births in the same year 1000 The use of infant mortality rates as a measure of overall health status for a given population is based on the assumption that it is particularly sensitive to socioeconomic changes and to health care interventions. Infant mortality has declined in all

regions of the world, but wide differences persist between and within countries Child mortality rate The child mortality rate (under-5 mortality rate) is based on deaths of children aged 14 years, and is frequently used as a basic health indicator. Injuries, malnutrition and infectious diseases are common causes of death in this age group. The under-5 mortality rate describes the probability (expressed per 1000 live births) of a child dying before reaching 5 years of age. Table 2.5 shows the mortality rates for countries representing a range of income categories. The areas of uncertainty around the estimates for middle-income and low-income countries are shown in parentheses. Data in Table 2.5 have been calculated so that the information can be compared between countries. Mortality rates per 1000 live births vary from as low as 4 for Japan (based on precise data) to 297 for males in Sierra Leone (with a wide range ofuncertainty: between 250 and 340 per 1000 live births).23 Gathering accurate data is not easy and alternative approaches have been developed Maternal mortality rate The maternal mortality rate refers to the risk of mothers dying from causes associated with delivering babies, complications of pregnancy or childbirth. This important statistic is often neglected because it is difficult to calculate accurately. The maternal mortality rate is given by: Maternal mortal it y rate= Number of maternal deaths in a given geographic area in a given year Number of live births that occurred among the population of the given geographic area during the same year (10n ) The maternal mortality rate ranges from about 3 per 100 000 live births in high-incomecountries to over 1500 per 100 00live births in low-income countries.23 However,even this comparison does not adequately reflect the much greater lifetime risk of dying from pregnancy-related causes in poorer countries. Adult mortality rate The adult mortality rate is defined as the probability ofdying between the ages of 15 and 60 years per 1000 population. The adult mortality rate offers a way to analysehealth gaps between countries in the main working agegroups.24 The probability of dying in adulthood is greaterfor men than for women in almost all countries, but thevariation between countries is very large. In Japan, lessthan 1 in 10 men (and 1 in 20 women) die in these productiveage groups, compared with almost 2 in 3 men (and1 in 2 women) in Angola

Life expectancy Life expectancy is another summary measure of the health status of a population. It is defined as the average number of years an individual of a given age is expected to live if current mortality rates continue. It is not always easy to interpret the reasons for the differences in life expectancy between countries; different patterns may emerge according to the measures that are used. For the world as a whole, life expectancy at birth has

increased from 46.5 years during the period 19501955 to 65.0 years during the period 19952000 (see Figure 2.5). Reversals in life expectancy have occurred in some sub-Saharan countries largely due to AIDS. Similar reversals in life expectancy have also occurred in middle-aged men in the former Soviet Union, where almost 1 in 2 men die between the ages of 15 and 60 years, largely due to changes in the use of alcohol and tobacco.26 Life expectancy at birth, as an overall measure of health status, attaches greater importance to deaths in infancy than to deaths later in life. Table 2.7 gives data for selected countries. As the data are based on existing age-specific death rates, additional calculation is necessary to allow comparability between countries; the uncertainty of the estimates are shown in parentheses. Confidence intervals can be quite large as in Zimbabwe but quite precise in countries like Japan which has complete vital registration. These data show the large variations in life expectancies between countries. For example, a girl born in Japan in 2004 can expect to live 86 years, whereas a girl born in Zimbabwe at the same time will live between 30 and 38 years. In almost all countries, women live longer than men.27 Causation of disease Although some diseases are caused solely by genetic factors, most result from an interaction between genetic and environmental factors. Diabetes, for example, has both genetic and environmental components. We define environment broadly to include any biological, chemical, physical, psychological, economic or cultural factors that can affect health (see Chapter 9). Personal behaviours affect this interplay, and epidemiology is used to study their influence and the effects of preventive interventions through health promotion

Causation of disease Genetic factors

Good health

Ill health

Environmental factors (including behaviours)

Screening Screening people for disease or risk factors which predict disease is motivated by the potential benefits of secondary prevention through early detection and treatment. Definition Screening is the process of using tests on a large scale to identify the presence of disease in apparently healthy people. Screening tests do not usually establish a diagnosis, but rather the presence or absence of an identified risk factor, and thus require individual follow-up and treatment. As the recipients of screening are usually people who have no illness it is important that the screening test itself is very unlikely to cause harm.26 Screening can also be used to identify high exposure to risk factors. For instance, childrens blood samples can be screened for lead in areas of high use of lead in paint. Types of screening There are different types of screening, each with specific aims: mass screening aims to screen the whole population (or subset); multiple or multiphasic screening uses several screening tests at the same time; targeted screening of groups with specific exposures, e.g. workers in lead battery factories, is often used in environmental and occupational health (Box 6.5); and case-finding or opportunistic screening is aimed at patients who consult a health practitioner for some other purpose. Criteria for screening Table 6.4 lists the main criteria for establishing a screening programme.27 These relate to the characteristics of the disorder or disease, its treatment and the screening test. Table 6.4. Requirements for instituting a medical screening programme Disorder Well-defined Prevalence Known Natural history Long period between first signs and overt disease; medically important disorder for which there is an effective remedy Test choice Simple and safe Test performance Distributions of test values in affected and unaffected individuals known Financial Cost-effective Facilities Available or easily provided Acceptability Procedures following a positive result are generally agreed upon and acceptable to both the screening authorities and to those screened. Equity Equity of access to screening services; effective, acceptable and safe treatment available

In addition, several issues need to be addressed before establishing a screening programme. Costs The costs of a screening programme must be balanced against the number of cases

detected and the consequences of not screening. Generally, the prevalence of the preclinical stage of the disease should be high in the population screened, but occasionally it may be worthwhile to screen even for diseases of low prevalence which have serious consequences, such as phenylketonuria. If children with phenylketonuria are identified at birth, they can be given a special diet that will allow them to develop normally. If they are not given the diet, they become mentally retarded and require special care throughout life. In spite of the low incidence of this metabolic disease (24 per 100 000 births), secondary prevention screening programmes are highly cost-effective. Lead time The disease must have a reasonably long lead time; that is, the interval between the time when the disease can be first diagnosed by screening and when it is usually diagnosed in patients presenting with symptoms. Noise-induced hearing loss has a very long lead time; pancreatic cancer usually has a short one. A short lead time implies a rapidly progressing disease, and treatment initiated after screening is unlikely to be more effective than that begun after the more usual diagnostic procedures. Length bias Early treatment should be more effective in reducing mortality or morbidity than treatment begun after the development of overt disease, as, for example, in the treatment of cervical cancer in situ. A treatment must be effective and acceptable to people who are asymptomatic. If treatment is ineffective, earlier diagnosis only increases the Table 6.4. Requirements for instituting a medical screening programme Disorder Well-defined Prevalence Known Natural history Long period between first signs and overt disease; medically important disorder for which there is an effective remedy Test choice Simple and safe Test performance Distributions of test values in affected and unaffected individuals known Financial Cost-effective Facilities Available or easily provided Acceptability Procedures following a positive result are generally agreed upon and acceptable to both the screening authorities and to those screened. Equity Equity of access to screening services; effective, acceptable and safe treatment available Epidemiology and prevention: chronic noncommunicable diseases 111 time period during which the participant is aware of the disease; this effect is known as length bias or length/time bias. Screening test The screening test itself must be cheap, easy to apply, acceptable to the public, reliable and valid. A test is reliable if it provides consistent results, and valid if it correctly categorizes people into groups with and without disease, as measured by its sensitivity and specificity. Sensitivity is the proportion of people with the disease in the screened population who are identified as ill by the screening test. (When the disease is

present, how often does the test detect it?) Specificity is the proportion of disease-free people who are so identified by the screening test. (When the disease is absent, how often does the test provide a negative result?) The methods for calculating these measures and the positive and negative predictive values are given in Table 6.5. Although a screening test ideally is both highly sensitive and highly specific, we need to strike a balance between these characteristics, because most tests cannot do both. We determine this balance by an arbitrary cut-off point between normal and abnormal. If we want to increase sensitivity and to include all true positives, we are obliged to increase the number of false positives, which means decreasing specificity. Reducing the strictness of the criteria for a positive test can increase sensitivity, but by doing this the tests specificity is reduced. Likewise, increasing the strictness of the criteria increases specificity but decreases sensitivity. We also need to account for predictive value when interpreting the results of screening tests Surveillance and response Definition Health surveillance is the ongoing systematic collection, analysis and interpretation of health data essential for planning, implementing and evaluating public health activities. Surveillance needs to be linked to timely dissemination of the data, so that effective action can be taken to prevent disease. Surveillance mechanisms include compulsory notification regarding specific diseases, specific disease registries (population-based or hospital-based), continuous or repeated population surveys and aggregate data that show trends of consumption patterns and economic activity. Box 7.6. Uses of surveillance Surveillance is an essential feature of epidemiologic practice and may be used to: recognize isolated or clustered cases; assess the public health impact of events and assess trends; measure the causal factors of disease; monitor effectiveness and evaluate the impact of prevention and control measures, intervention strategies and health policy changes; and plan and provide care. In addition to estimating the magnitude of an epidemic and monitoring its trends, data can also be used to: strengthen commitment, mobilize communities, and advocate for sufficient resources.19 128 Ch The scope of surveillance The scope of surveillance is broad, from early warning systems for rapid response in the case of communicable diseases, to planned response in the case of chronic diseases which generally have a longer lag time between exposure and disease. Most countries have regulations for mandatory reporting of a list of diseases. These lists of notifiable diseases often include vaccine-preventable diseases such as polio,

measles, tetanus and diphtheria as well as other communicable diseases such as tuberculosis, hepatitis, meningitis and leprosy. Reporting may be required also for Box 7.5. Immunization: key to prevention and control Immunization is a powerful tool in the management and control of infectious diseases. Systematic immunization programmes can be very effective. For example, by the late 1980s, most countries in South and Latin America had incorporated measles vaccination into routine immunization programs and many had done follow-up immunization campaigns to reach all children and interrupt measles transmission.17 non-communicable conditions, such as maternal deaths, injuries and occupational and environmental diseases such as pesticide poisoning Mandatory reporting of specific conditions is a subset of surveillance. There are many other uses of surveillance Principles of surveillance A key principle is to include only conditions for which surveillance can effectively lead to prevention. Another important principle is that surveillance systems should reflect the overall disease burden of the community. Other criteria for selecting diseases include: incidence and prevalence indices of severity (case-fatality ratio) mortality rate and premature mortality an index of lost productivity (bed-disability days) medical costs preventability epidemic potential information gaps on new diseases. Sources of data Sources of data may be general or disease-specific, and include the following: mortality and morbidity reports hospital records laboratory diagnoses outbreak reports vaccine utilization sickness absence records biological changes in agent, vectors, or reservoirs blood banks. Surveillance can collect data on any element of the causal chain of disease behavioural risk factors, preventive actions, cases and program or treatment costs. The scope of a surveillance system is constrained by human and financial resources. Surveillance in practice Surveillance relies upon a routine system of reporting suspected cases within the health system, followed by validation and confirmation. Active and appropriate responses ranging from local containment measures to investigation and containment by a highly specialized team, are then put in place. Surveillance requires continuing scrutiny of all aspects of the occurrence and spread of disease, generally using methods distinguished by their practicability, uniformity and, frequently, their rapidity, rather than by complete accuracy. The analysis

of data from a surveillance system indicates whether there has been a significant increase in the reported number of cases. In many countries, unfortunately, surveillance systems are inadequate, particularly if they depend on voluntary notification. A wide range of networks, including nongovernmental organizations, electronic discussion groups, search engines on the World Wide Web, and laboratory and training networks, offer powerful ways of obtaining information that leads to a coordinated international response. Sentinel health information systems, in which a limited number of general practitioners report on a defined list of carefully chosen topics that may be changed from time to time, are increasingly used to provide supplementary information for the surveillance of both communicable and chronic diseases. Surveillance of chronic disease risk factors is discussed in Chapter 2. A sentinel network keeps a watchful eye on a sample of the population by supplying regular, standardized reports on specific diseases and procedures in primary health care. Regular feedback of information occurs and the participants usually have a permanent link with researchers. Analysis and interpretation of surveillance data Surveillance is not only a matter of collecting data, as the analysis, dissemination and use of the data for prevention and control are equally important. Many public health programs have far more data than they can presently analyse Table 7.3 outlines Millennium Development Goal 6, which focuses on HIV/AIDS, malaria and other diseases, which are largely interpreted as communicable diseases. Non-communicable diseases which account for the bulk of death and disability in most countries have been omitted. The indicators, operational definitions and overall objectives to be met for tuberculosis (target 8) are also shown in Table 7.3; all require detailed surveillance.