[Methodology of a scientific study: from the formulation of the problem to the interpretation of the results--applications to radiology].
Sante, 1996/5-1996/6;6(3):139-44.
Bertrand P[1]
Affiliations
PMID: 8764446
Abstract
A scientific article should convey valid information to the reader. To reach this goal, the article must meet several criteria. First, the problem must be clearly stated. For example with medical imaging, the two types of studies are: -a descriptive study depicting a particular situation; -a comparative analytical study, either evaluating a diagnostic method or searching for prognostic factors. The process of defining the sample population can be distinguished: -Experimental studies can compare two groups of people who, after drawing lots, received differential treatment. -Alternatively, observational studies can investigate either representative samples, exposed versus nonexposed cases, or control cases. Characteristics of the variables are very important. Statistical methods facilitate the analysis of three types of variables: qualitative, quantitative and survival variables. Statistical analysis is used to extrapolate the results of the sample population of the study to the target population. Then, presentation, interpretation and extrapolation of the results must be done using three methods: First, the descriptive study should accurately depict the disease. It must be representative of the pathology relative to the clinical manifestations and avoid any bias. Then, the comparative analytical study for revealing the prognostic factors. These types of studies can be divided into three steps: -by performing the suitable statistical tests, a significant correlation between the prognostic factors and the outcome of the disease can or cannot be demonstrated; -any bias must be eliminated; -this correlation can be quantified by measuring it as a risk; Finally, the evaluation of the diagnostic methods are mostly used in medical imaging to determine which factors need to be measured. Five criteria can be followed for these types of studies: -to obtain good technical quality; -to interpret the results consistently; -to give valid information (requiring proper classification of the clinical signs of the patients according to the definition of disease); -to be helpful in determining the treatment; -to be efficient for the patient and the population. The results of a study on the diagnostic value of a medical test must be expressed as an index of sensitivity, or the rate of true positives versus negatives. This facilitates the calculation of the positive predictive value, i.e. the probability for a patient to become ill if the medical test is positive, versus the negative predictive value, i.e. the probability for a patient to remain healthy if the test is negative. The rigorous methodology of scientific work requires: -clear formulation of the problem; -good definition of the situation and the variables to be measured; -the proper method to collect the data; -the description of the study protocol.
MeSH terms
Bias; Data Collection; Diagnosis; Diagnostic Imaging; Documentation; Humans; Predictive Value of Tests; Prognosis; Radiology; Random Allocation; Reproducibility of Results; Research Design; Risk Factors; Science; Sensitivity and Specificity; Statistics as Topic; Survival Analysis; Therapeutics; Treatment Outcome
More resources
EndNote: Download