10.2 Drug Literature Evaluation & Evidence-Based Application

Drug Literature Evaluation & Evidence-Based Application

Evaluating drug literature goes beyond reading the abstract or the authors' conclusions. Clinical pharmacists must dissect the methodology to determine if the findings are valid, clinically significant, and applicable to their specific patient population. This requires a strong grasp of validity, bias, and critical appraisal techniques.

Validity: Internal vs. External

When evaluating a trial, one must assess both its internal and external validity.

Internal Validity

Internal validity refers to the degree to which the study accurately answers the question it was designed to investigate, free from bias and confounding.

• High Internal Validity: Achieved through rigorous methodology, such as robust randomization, blinding (single, double, or triple), and controlling for confounding variables.
• Threats: Selection bias, attrition bias (loss to follow-up), measurement bias, and confounding. If a study lacks internal validity, its results cannot be trusted, rendering external validity irrelevant.

External Validity (Generalizability)

External validity is the extent to which the study's results can be generalized to clinical practice and other populations.

• High External Validity: The study population closely resembles patients seen in real-world practice, and the intervention is feasible in a clinical setting.
• Threats: Highly restrictive inclusion/exclusion criteria can create a homogenous study population that does not reflect real-world diversity. For example, a trial on a heart failure medication that excludes all patients with chronic kidney disease has low external validity for a general cardiology clinic.

Bias and Confounding

• Bias: A systematic error in study design or execution that leads to an incorrect estimate of the true treatment effect.
  • Selection Bias: Occurs when the groups being compared are not comparable at baseline. Randomization is the primary defense against selection bias.
  • Observation/Information Bias: Errors in measuring or collecting data. Blinding minimizes this bias.
  • Attrition Bias: Unequal loss of participants from different groups during the study.
• Confounding: Occurs when a third variable is associated with both the exposure and the outcome, distorting the apparent relationship between them. For instance, in a study assessing alcohol use and lung cancer, smoking status is a major confounder. Randomization helps distribute known and unknown confounders evenly between groups.

Critical Appraisal of RCTs

Randomization and Blinding

• Randomization: Ensures each participant has an equal chance of being assigned to any treatment group. Stratified randomization can be used to ensure balance for key prognostic factors (e.g., age, disease severity).
• Blinding: Prevents participants, investigators, and outcome assessors from knowing the treatment assignment. Double-blind (neither investigator nor patient knows) is standard for drug trials to prevent placebo effects and observer bias.

Intention-to-Treat (ITT) vs. Per-Protocol (PP) Analysis

• Intention-to-Treat (ITT): Analyzes patients in the groups to which they were originally randomized, regardless of whether they received the treatment, adhered to the protocol, or withdrew. ITT preserves the benefits of randomization and provides a conservative, real-world estimate of the treatment effect. It is the preferred method for superiority trials.
• Per-Protocol (PP): Analyzes only those patients who completed the study exactly according to the protocol. This method disrupts randomization and is highly susceptible to selection bias (since patients who drop out often differ systematically from those who complete the study). PP analysis can exaggerate the treatment effect. It is often used in conjunction with ITT, especially in non-inferiority trials.

Composite Endpoints

Many modern cardiovascular and diabetes trials utilize composite endpoints (e.g., MACE: Major Adverse Cardiovascular Events, defined as cardiovascular death, non-fatal MI, or non-fatal stroke). When interpreting composite endpoints, it is critical to evaluate the individual components. A positive composite endpoint might be driven entirely by the least clinically significant component (e.g., a reduction in hospitalizations rather than a reduction in mortality).

Interpreting Clinical Practice Guidelines

Clinical guidelines synthesize primary literature into actionable recommendations. Understanding the grading systems used is crucial for applying these recommendations.

ACC/AHA Guideline Methodology

The American College of Cardiology and American Heart Association use a standardized system:

• Class of Recommendation (COR): Indicates the strength of the recommendation (Risk vs. Benefit).
  • Class I: Strong (Benefit >>> Risk). Is recommended; should be performed.
  • Class IIa: Moderate (Benefit >> Risk). Is reasonable; can be useful.
  • Class IIb: Weak (Benefit ≥ Risk). Might be considered.
  • Class III: No Benefit or Harm (Benefit ≤ Risk). Should not be performed.
• Level of Evidence (LOE): Indicates the quality of evidence supporting the recommendation.
  • Level A: High-quality evidence from more than one RCT or meta-analyses of RCTs.
  • Level B-R: Moderate-quality evidence from 1 or more RCTs.
  • Level B-NR: Moderate-quality evidence from nonrandomized, observational studies.
  • Level C-LD: Limited data (observational studies, registries).
  • Level C-EO: Expert opinion.

Drug Information Resources

Efficiently locating and evaluating drug information requires knowing the hierarchy of literature.

• Tertiary Literature: Summarizes and synthesizes existing knowledge. Examples: Lexicomp, Micromedex, Clinical Pharmacology, textbooks, clinical practice guidelines, Review articles. They are easy to use but may be outdated.
• Secondary Literature: Indexing and abstracting services that point to primary literature. Examples: PubMed/MEDLINE, Embase, Cochrane Library, Google Scholar.
• Primary Literature: Original research published in peer-reviewed journals. Examples: Randomized controlled trials, cohort studies, case reports published in NEJM, JAMA, Pharmacotherapy. Provides the most current data but requires critical appraisal skills to interpret correctly.