Suitability of indexing terms in the MEDLINE bibliographic database on drug-related problems

For years, there has been controversy about the meaning of medication-related problems (MRPs). This has led several authors to attempt to redefine and classify this term with the aim of using it correctly in the healthcare setting. So far without achieving the desired objective, resulting in erroneous results in the sources of information and thus in malpractice in the sector.

To describe and analyze the appropriateness of the existing indexing of scientific publications in the MEDLINE bibliographical database with respect to drug-related problems (DRPs) and to determine whether the descriptors used fulfilled the function of suitably representing this concept.

A descriptive study was conducted, using the following search terms: Medication Errors; Drug Interactions; Drug Overdose; Drug-Related Side Effects and Adverse Reactions; Contraindications, Drug. The sample size was calculated by estimating population parameters in an infinite population (expected value = 0.05; precision of interval = 0.05; level of confidence = 0.95) and the selection method was simple random sampling without replacement, taking the total number of bibliographical references in the database as the basis. The agreement of the indexing with DRPs was evaluated with the coefficient of determination (R²), and the Cohen kappa coefficient was used for the association between the definition of the descriptors and the objective of the article.

The 1930 records analyzed showed a total of 2888 different major topics. These major topics were present, with at least one of the five descriptors studied, in 482 (25.0%; 95% CI 23.0-27.0) documentary files, with statistically significant differences between the two phases analyzed (χ2 = 183.8; degrees of freedom (df) = 1; p < 0.001): 1st phase, 295 (13.3%; 95% CI 13.7-16.9) and 2nd phase, 187 (9.7%; 95% CI 8.4-11.0). Overall scientific output with the five descriptors showed a coefficient of determination (R²) of 0.9 (p < 0.001) and the relationship between the objective of the study and the definitions of the five descriptors was 0.9 (p < 0.001).

There was a very good direct exponential trend of the overall scientific output retrieved with the terms associated with DRPs, although the progression of the five descriptors separately did not show a growth model conforming to expectations. There was a moderate agreement between the objective of the study and the definition of each of the five descriptors used and a low relationship between the objective of the study and the concept of DRPs used for this investigation. It is essential to have a descriptor that unifies the terminological diffusion that has existed up till now, since process (causes) and effects (outcomes) have been mixed together under the various definitions and classifications of DRPs found in the studies.