6 Evidence Appraisal: Research 111
Prevalence and incidence studies are descriptive designs frequently used by epi-
demiological researchers (Polit & Beck, 2017). The aim of prevalence studies is
to determine the proportion of a population that has a particular condition at a
specific point in time (known as prevalence or point prevalence). This provides
researchers with a useful metric to better understand the burden of a specific
disease in the community. Incidence studies seek to determine the frequency of
new cases (or incidence rate) and are useful in understanding risk for disease
development.
Example: Nonexperimental Epidemiological Descriptive Designs
An interdisciplinary research team (Dybitz, Thompson, Molotsky, & Stuart, 2011)
studied the prevalence of diabetes and the burden of comorbid conditions
in elderly nursing home residents. Prevalence of diabetes was determined
by laboratory values recorded in the medical record over a 12-month period,
documented medical-chart diagnosis of diabetes, and evidence of medications
prescribed for diabetes in a prescription claims database. They found a diabetes
prevalence of 32.8% of residents from a national sample of 250 skilled nursing
facilities and characterized the disease burden of diabetes in these settings.
Roberts (2010) conducted electronic chart reviews of hospitalized children in
a midsize urban hospital over a 2-week period to determine the incidence of
parental/guardian absence in the previous 24 hours. The researchers were
interested in understanding the risk for unaccompanied pediatric patients in a
culture that promotes patient-family centered care.
Predictive Designs
Predictive designs seek to predict relationships. Two basic questions are asked in
predictive research: “If phenomenon X occurs, will phenomenon Y follow? If we
introduce an intervention, will a particular outcome follow?” Predictive designs
range from simple predictive correlational studies that look at whether a single
variable predicts a particular outcome to more complex predictive designs that
use multiple or logistic regression to examine whether several variables predict a
particular outcome.
