alt="An illustration of extract from case series of bouncy castle injuries."/>
Figure 5.2 Extract from case series of bouncy castle injuries.
Source: From Singer and Freedman (1992), © 1992, BMJ Publishing Group Ltd.
Sometimes the author of a case series notices some common feature that the cases share and speculates that this factor might help to explain the condition. A famous example of such studies includes the early descriptions of the birth abnormalities that became linked with the drug thalidomide.
CROSS‐SECTIONAL STUDIES
Unlike the case series, which usually reports an unexpected clinical encounter, the researchers of a cross‐sectional study deliberately set out to assemble a group of study subjects – often patients in current or recent contact with part of the health service – and describe the presence and absence of various clinical features. For example, responding to the case series of bouncy castle injuries (Figure 5.2), another researcher extracted from a national survey of leisure accidents the clinical and circumstantial details of 105 bouncy castle injuries (Figure 5.3) – providing us with a more representative picture. This new, premeditated study of 105 cases is a fairly typical cross‐sectional study. It seems to show that, when the matter is examined using a suitably selected sample, there is no convincing evidence that bouncy castle injuries cause a specific fracture of the elbow region but, instead, lead to all manner of soft tissue, joint, and bony injuries – anywhere on the body.
Figure 5.3 Extract from cross‐sectional study about bouncy castle injuries.
Source: From Levene (1992), © 1992, BMJ Publishing Group Ltd.
A particular type of cross‐sectional, single‐group study is one in which incidence or prevalence of a condition is determined. A prevalence study determines how many cases of a condition or disease there are in a given population, or it might establish the frequency of a clinical finding in a study sample. For example, researchers might use hospital and pharmacy records to establish how many people have insulin‐receiving diabetes in a defined hospital catchment area. Then again, another study might estimate the proportion of older people in residential facilities whose ears are occluded by earwax (Figure 5.4).
Figure 5.4 Prevalence of a clinical feature, determined in a cross‐sectional study.
Source: From Culbertson et al. (2004), © 2004, Elsevier.
Incidence studies are rather similar but refine the above kind of study in two ways: the incidence of a condition is the number of new cases arising in a defined population over a defined time. Figure 5.5 describes such a study – to determine in a defined area of 22 adjoining electoral wards in South London, the number of first‐in‐a‐lifetime strokes, according to ethnic origin. The researchers established the incidence (sometimes called inception rate) of stroke in each of the ethnic groups under scrutiny – per 1000 population per year. Strictly speaking it is this incorporation of time, as well as the proportion of cases, that makes incidence a rate, while prevalence is merely a proportion.
Figure 5.5 Extract from a cross‐sectional (incidence) study about frequency of stroke.
Source: From Wolfe et al. (2002), © 2002, BMJ Publishing Group Ltd.
Figure 5.5 shows how the data were collected from hospitals and the community. Some incidence research data, as in this case, are derived from case‐registers – many of which have been set up with research and clinical service in mind, routinely recording data useful for both purposes. In other cross‐sectional studies the researchers undertake a survey of the study sample, where their survey – whether by interview, or by electronic or paper self‐report – has been set up specifically for the research project. For example, researchers might ask nurses who visit patients in their homes to describe, by filling in a paper questionnaire, how their patients are using prescribed medicines (Figure 5.6).
Figure 5.6 Use of a survey method in a cross‐sectional study.
Source: From Ellenbecker et al. (2004), © 2004, Elsevier.
LONGITUDINAL STUDIES
When researchers study a group of subjects over time in a longitudinal study, there is more research work to be done than in a cross‐sectional study; subjects must be followed up one or more times to determine their prognosis or outcome.
The kinds of observational studies we've seen above are among the simplest form of clinical research. In the next chapter, more complex observational studies are described – quantitative analytic studies. What they have in common (and in this way they differ from the above study types) is that they generally involve comparison of two or more groups of people and often attempt to infer something about cause of symptoms or conditions.
CHAPTER 6 Analytic Studies
Remember from the previous chapter that, compared with descriptive studies of a single group, analytic studies are more complex (and often more interesting). Analytic studies will usually involve some comparison and frequently aim to elucidate cause and effect in some way. Four kinds of observational analytic study will be described here:
Ecological studies
Cross‐sectional, two‐group studies
Case–control studies
Cohort analytic studies
ECOLOGICAL STUDIES
A neat way of tackling questions about the cause of disease or other health events is to sit in a library (or, more likely, at a computer), locate routinely collected data, and put population data about disease frequency (e.g. regional deaths from lung cancer) together with data about exposure to a risk (e.g. regional data on tobacco consumption). By so doing, you might find that
