Introduction | Part 1: Hierarchy of Evidence | Part 2: Case Reports | Part 3: Cross-Sectional Studies | Part 4: Case Control Studies | Part 5: Cohort Studies | Part 6: Randomized Controlled Studies | Part 7: Meta-Analysis | Part 8: Systematic Reviews
Cross-sectional studies describe the presence and absence of disease within a slice of a population at a specific time by describing a facet shared by that population during the same time. Although they involve no controls, comparisons can be made within the sampled population, such as breed, sex, comorbidity, environment, etc. These surveys assess the magnitude and can generate a theory about causation. The relationship between exposure and disease is extrapolated from data obtained in the survey.
An important distinction here is the difference between prevalence vs. incidence which can be misunderstoods . For prevalence, a specific point in time will include patients with the same disease who may have been diagnosed recently or who have had the condition for years. It is the probability of the disease — whether it is peracute or chronic — in the population at a specific time with onset, temporally, outside the bounds of the period of interest. Since incidence assesses newly diagnosed disease, only those conditions diagnosed within the time period are included. If the time interval is one year, patients with chronic illness diagnosed during the first week are included along with those suffering acute or self-limiting illness during the last week; as long as the diagnosis was made within the specified period of time.
Cross-sectional studies assess the prevalence of disease. Prevalence (expressed as a probability) is the likelihood of a disease in a specific population at a specific point in time. Incidence is the probability of newly diagnosed disease disease within a population within an explicit period of time. Surveys tell us about the present or how things are now.
Also, be aware that some studies may use survey to indicate the research method but the term is occasionally misused to mean what is actually a cohort study used to understand incidence.
From 2000-2010, the leading journals publishing cross-sectional studies (in descending order) were JAVMA (165), Preventative Veterinary Medicine (104), Veterinary Parasitology (72), Tropical Animal Health Production (54), Veterinary Record (35), Australian Veterinary Journal (24), Veterinary Microbiology (23), Journal Dairy Science (19), New Zealand Veterinary Journal (16) and the Journal of Veterinary Internal Medicine (16).
The leading MeSH headings during this period were cattle, prevalence, risk factors, cattle diseases, serioepidemiologic studies, dogs, animal husbandry, dog diseases and enzyme-linked immunosorbent assay respectively. Not surprisingly, common cross-sectional study themes in the veterinary literature are those relating to epidemiology and animal husbandry with prominent representation of cattle, dogs and sheep as a species and investigations regarding parasitology and microbiology, especially studies about antibodies (e.g. protozoan, bacterial and viral).
Johnston AR, Gillespie TR, Rwego IB, McLachlan TL, Kent AD, Goldberg TL. | Center for Zoonoses and Infectious Disease Research and Department of Pathobiology, University of Illinois, Urbana, Illinois, United States of America. | PLoS Negl Trop Dis. 2010 May 11;4(5):e683.
A study wishes to assess the prevalence multiple organ failure (MOF) in patients with acute pancreatitis. Comparisons can also be made regarding the presence or absence of sepsis among those with or without MOF as well. This emphasizes that although the survey may not have a separate control or comparison group, subgroups within the sample may be compared with one another (e.g. those patients with or without the presence of sepsis in the multiple organ failure group).
As mentioned these studies are ideal to understand prevalence in a population or comparisons between populations. Because they require no longitudinal commitments (i.e. follow-up) they are less demanding of resources and inherently less costly to perform.
As previously mentioned, associations between patient attributes or exposures cannot be reliably used for causation. While case reports can reveal rare or unusual observations, there is no guarantee that a sampling of a population during a temporally finite period of time will reveal exceptional cases.
An introduction of left censorship results when patients (or clients) in a survey either drop-out, die or perhaps move away from the study area and are lost to follow-up. Although not a usually problem in veterinary medicine, when cross-sectional studies evaluate prevalence (point in time) in large populations (e.g. government health surveys), certain stereotypical patient populations are often under-represented. Patients with acute, high risk illness with high mortality and low morbidity due to another co-morbid illness or genetic influences may appear to have a lower prevalence.
This series has been loosely organized from a set of lectures given by the author within graduate courses in Biomedical Informatics beginning in 2002. Content is being edited to improve organization, depth, correct inaccuracies as well as updates with new information during Winter/Spring 2012. Feedback is greatly appreciated. © 2011, Stuart Turner.