Epidemiology in practice: an introduction
In future issues of the Journal there will be a series of articles on Epidemiology in Practice. These articles will give examples of some of the different types of epidemiological study that eye health workers can use, and explain how important data from epidemiological studies are in prevention of blindness. Some of the terms used in epidemiology will be defined and some of the concepts explained.
Definition of Epidemiology: The quantitative study of the distribution, determinants and control of diseases in populations.
In spite of this rather heavy definition, epidemiology provides information that enables rational planning of prevention of blindness activities. The information obtained can be used to:
1. Assess the need for health care services
For example, knowing how many people are blind from cataract in a population at a given point in time (i.e., the prevalence) will influence how many trained personnel and how much equipment is required to meet the current need. Knowing how many people in the same population are likely to become blind from cataract over a given number of years (i.e., the incidence) allows forward planning of service provision, and evaluation as to whether the number of cataracts being done is enough to prevent the cataract ‘backlog’ from increasing.
2. Identify high risk groups
Knowing who is most affected, or who is most at risk of a particular disease in a population, allows curative services or preventive measures to be targeted at those most in need. For example, epidemiological studies have shown that women are more likely to be blind from trachoma than men. It is girls and young women who need to be targeted for health education and preventive measures, and women who need to be encouraged to come forward for trichiasis surgery. Epidemiological studies have shown that young children are most affected by vitamin A deficiency, and it is this age group which needs to be targeted for health and nutrition education, or vitamin A supplementation.
3. Identify factors which, if removed, would prevent the disease or modify its course
Diseases can only be prevented once the underlying causes have been identified. For example, infants with vitamin A deficiency are less likely to have been breast fed than infants from the same area who do not have vitamin A deficiency. Not being breast fed is, therefore, a risk factor for vitamin A deficiency. Health education to promote breast feeding may help to reduce the number of children with vitamin A deficiency.
4. Select the most effective treatment
Clinical trials can be used to determine which methods give the best results with the lowest complication rates. Clinical trials can compare surgical treatments (e.g., intracapsular cataract extraction and aphakic glasses compared with extracapsular cataract extraction and intraocular lens implant, to determine which gives the best visual results, the best quality of life, and the lowest complication rate). They can also compare medical treatments (e.g., topical tetracycline eye ointment compared with systemic azithromycin for active trachoma).
5. Select the most efficient and effective way of delivering a curative or preventive measure
This allows the best use of scarce resources.
Table 1. The epidemiological approach to eye care
|Reason for Study
|Type of Study
|Finding out the magnitude and distribution of need:
|To measure how much blindness and serious eye disease there is in the community, and to find out who in the community is most affected (age group, sex, geographic area, etc)
|Population based cross sectional survey of prevalence and causes
|To determine how long it takes people in the community to develop blindness or serious eye disease
|Longitudinal studies of incidence
|Investigation of the causes of eye disease:
|To identify factors that increase the risk of eye disease and visual loss in the community
|Case control studies Longitudinal cohort studies
|To compare and evaluate two or more different preventive strategies or treatments, in terms of effectiveness and cost
|Randomised community based intervention trials
The following Questions and Answers may help to clarify how eye health workers can use epidemiological data, and when the types of study mentioned in Table 1 should be used.
Question A general doctor working in a remote District noticed that many blind patients came to the hospital, but the nearest ophthalmologist was 300 miles away. The doctor approached the Regional Minister of Health and asked that an ophthalmologist be appointed, and equipment provided. The Minister was not satisfied with being told that there were ‘a lot of blind people coming to the hospital’, and wanted to know more before committing limited resources. What information could the doctor provide relatively quickly and easily?
Answer The doctor could go through the hospital records for the previous year to find out how many blind people came to the hospital, and what were the causes of blindness.
Question The Minister agreed that there seemed to be a need for an ophthalmologist, but he wanted to know whether the kind of equipment requested was suitable to treat people who were blind in the District served by the hospital. What information would the Minister need to answer this question, and how could it be obtained?
Answer The doctor thought there must be many blind people in the community who did not come to the hospital, as some areas in the District were poor, and transport was difficult. The doctor realised the information he had given to the Minister from the hospital records would not truly reflect the causes of blindness in the district. In other words, the hospital data was biased. The only way to get an accurate picture of the seriousness of blindness in the community as well as reliable information on the commonest causes would be to go to the community and examine people there. A community based survey would be needed. The doctor realised that he could not do this study on his own, that it would need a lot of planning, resources and time. He asked an epidemiologist to assist him to design, plan and carry out the study, as well as to analyse the data.
To find out precisely how many people are blind in a population (in this case the term ‘population’ is used to describe everyone living in the District served by the hospital) it would be necessary to examine everyone. This is not possible, nor is it necessary, as an estimate of the number of blind can be determined by examining a carefully selected sample of people from the population of interest. It is vital that the sample represents the whole population with respect to a number of parameters, the main ones being age and sex. The sample also needs to be big enough to give an accurate estimate, and there are well established formulae which can be used to determine how big the sample should be. To reduce bias in the sample the best way of selecting which people should be included for examination is to use random sampling methods.
After selecting the sample the next step is to examine everyone in the sample using exactly the same methods. This is usually done by field workers who have been rigorously trained and assessed to make sure they are competent and reliable. To determine the proportion of people in the sample who are blind the visual acuity of everyone in the sample needs to be measured. Those who are found to be blind are then examined (usually by an ophthalmologist or ophthalmically trained paramedic) to determine the cause of blindness. All the information for each individual (age, sex, place of residence, visual acuity, results of eye examination, etc.) is entered onto a carefully designed form, and the data entered into a computer for analysis. The information collected allows the proportion of the sample who were blind to be calculated. This proportion, which is usually expressed as a percentage, is called the prevalence of blindness. As long as the sample size is big enough, and enough of the sample have been examined, this information can be used to estimate the number of people who are blind in the whole population from which the sample was selected (in this case the District served by the hospital). Information can also be obtained on the most important causes of blindness.
Question The result of this study showed that one cause of blindness occurred only in one geographical area, and another condition only affected children. Why is it important to know this information?
Answer It is important to know this so that specific groups in the District can be targeted with specific interventions. Knowing who in a population is most affected is called the distribution.
Question The Minister wanted to know whether the contract for the ophthalmologist should be long term, or whether a short term contract would meet the need in the District. He also wanted to know whether the need was likely to increase with time, and whether he should be budgeting for a second ophthalmologist. What kind of study would need to be done to supply this information?
Answer The Minister was wanting to know the incidence of blindness in the district, which is a measure of the number of new cases occurring over time. To provide this information a longitudinal study would be needed in which a sample of people who were not blind are selected, and then re-examined after a period of time to find out the proportion who had become blind.
Question The ophthalmologist was appointed, but was soon asking for more equipment. Before supplying this the Minister wanted to know the underlying causes of the blinding conditions, and whether anything could be done to prevent blindness. What kind of studies would need to be done to address these issues, and why is it important to know this?
Answer It is very important to know what might be causing a disease, as this provides the opportunity for prevention, or for modifying the course of the disease. There are several different types of epidemiological study which address issues of causation, i.e., case control studies, cohort studies and ecological studies. These studies are complicated to design, and require sophisticated analysis of data. For this reason these types of study will not be discussed in detail in this series of articles.
Question The ophthalmologist kept up to date with the medical literature, and read of a new treatment for a condition which seemed to be effective in patients in the USA. He wanted to know whether the new treatment would also be effective in the patients he was seeing in the hospital. What kind of study could he do?
Answer Clinical trials are the best method of comparing treatments, and the design which gives the most reliable results is a ‘double masked, randomised controlled trial’. In these studies patients with the same condition are randomly allocated to one treatment, or to another treatment (which may be ‘no treatment’). In some studies it is possible for the doctor doing the study as well as the patient not to know who is receiving which treatment (‘double masked’). After the period of follow up the findings in the two treatment groups are analysed and compared to see which treatment gives the best results.
Question The doctor noticed that there were patients coming to the hospital with advanced eye diseases which were in fact preventable. The preventive measure entailed going into the community to treat people before the disease was advanced, but the hospital only had one vehicle and limited staff. The ophthalmologist knew that there were Community Health Workers (CHWs), and that the Immunisation Programme visited the villages on a regular basis. He wondered whether treatment would be as effective if given by the CHWs as by the Immunisation Programme staff, and which approach would be the cheapest. What kind of research is needed to answer these questions?
Answer Operational research, also known as health services research. Operational research is concerned with the planning, management, logistics and delivery of health care. It deals with the use of human resources, organisation, the utilisation of facilities, the quality of health care, cost, and the relationship between need and demand.
Question The ophthalmologist had been working in the hospital for 10 years. He had trained nurses in the Primary Health Clinics in Primary Eye Care; had started a community based programme of eyelid surgery for the scarring effects of trachoma; and had persuaded the Immunisation Programme teams to supplement 9 month old babies with vitamin A 100,000 IU at the time of measles immunisation. After this period of hard work he wanted to know whether there had been any impact on blindness in the District. How could this be determined?
Answer A repeat survey could be done, and the prevalence of blindness in the repeat survey compared to the prevalence in the initial survey. Care needs to be taken when interpreting the findings of repeat surveys, however, because if the prevalence is lower in the second survey it may be due to other factors – not the provision of improved eye health care.
Epidemiological studies describe the health status of populations, provide data on the causes of disease, and identify the most effective and efficient way of delivering health care or preventive measures.
In the next issues of the Journal different types of epidemiological study will be described in more detail, using studies that have actually been undertaken as practical examples.