2.3 Research Methods & Evidence Analysis

Evidence-based practice is a defining skill of the registered dietitian. The RD exam expects you to recognize study designs, judge their strength, interpret basic statistics, and apply graded evidence through the Nutrition Care Process. You do not need to be a statistician, but you must read a study stem and decide how much to trust it.

The Evidence Hierarchy

Not all evidence is equal. From strongest to weakest:

1. Systematic reviews and meta-analyses — synthesize many studies; a meta-analysis statistically pools their results.
2. Randomized controlled trials (RCTs) — randomization controls confounding; the gold standard for cause and effect.
3. Cohort studies — follow groups forward over time (prospective or retrospective).
4. Case-control studies — start with an outcome and look backward for exposures.
5. Cross-sectional studies — a snapshot of exposure and outcome at one point in time.
6. Case reports, case series, expert opinion, anecdote — weakest, hypothesis-generating only.

The higher a design sits, the better it controls bias and confounding — but feasibility and ethics fall as you climb, which is why much nutrition evidence is observational.

Study Design Comparison

Key principle: observational designs (cohort, case-control, cross-sectional) demonstrate association, not causation, because they cannot rule out confounding variables — a third factor linked to both exposure and outcome. Only a well-conducted RCT supports causal claims, because randomization distributes both measured and unmeasured confounders evenly across arms.

Two design refinements worth knowing: a double-blind trial blinds both subjects and investigators to reduce placebo and observer bias, and an intention-to-treat analysis keeps participants in their assigned group regardless of adherence, preserving the benefits of randomization.

Statistics You Must Interpret

• p-value: the probability the observed result (or one more extreme) occurred by chance if the null hypothesis were true. p < 0.05 is the conventional threshold for statistical significance.
• Confidence interval (CI): the range likely to contain the true value; a 95% CI for a risk ratio or odds ratio that crosses 1.0 indicates a non-significant result, and a CI for a mean difference that crosses 0 is non-significant.
• Effect size / clinical significance: a finding can be statistically significant yet too small to matter clinically — a 0.4-lb weight difference in 5,000 people may reach p < 0.05 but change no practice.

Diagnostic accuracy and association:

• Sensitivity: correctly identifies those with the condition (true-positive rate); high sensitivity rules a condition out when negative ("SnNout").
• Specificity: correctly identifies those without the condition (true-negative rate); high specificity rules a condition in when positive ("SpPin").
• Relative risk (RR) is used in cohorts/RCTs; odds ratio (OR) in case-control studies; values near 1.0 suggest no effect, above 1.0 suggest increased risk.

Mean, median, and mode describe central tendency; standard deviation describes spread. A skewed dataset (e.g., income, length of stay) is better summarized by the median, because the mean is pulled toward the tail.

The Evidence Analysis Library and EBP

The Academy of Nutrition and Dietetics Evidence Analysis Library (EAL) is the profession's system for systematically reviewing nutrition research and assigning conclusion grades (Grade I = good/strong evidence, down through Grade V = expert opinion only / insufficient). It produces Evidence-Based Nutrition Practice Guidelines that dietitians apply at the point of care, and it feeds the Academy's Evidence Analysis Manual methodology.

Evidence-based practice (EBP) integrates three elements — none sufficient alone:

1. The best available research evidence (the EAL supplies this).
2. The clinician's expertise and clinical judgment.
3. The patient's values, preferences, and circumstances.

A dietitian who follows a guideline but ignores a patient's cultural food practices, budget, or goals is not practicing EBP — and neither is one who acts on personal habit while ignoring strong evidence. On the exam, the "best" answer to an intervention question typically reflects all three legs of the stool. The EAL connects directly to the Nutrition Care Process introduced later in this domain: graded evidence informs the standardized assessment, diagnosis, intervention, and monitoring steps every RDN uses.

Recognizing Bias and Validity

Beyond design and statistics, the exam asks you to spot threats to a study's trustworthiness. Internal validity is whether the study measured what it claims within its own sample; external validity (generalizability) is whether the result applies to your patient population. Common threats to recognize in a stem:

Apply this to nutrition research specifically: dietary studies lean heavily on self-reported intake (24-hour recalls, food-frequency questionnaires, food diaries), which systematically underestimates calories and is hard to blind. That measurement limitation is one reason much nutrition evidence sits at the observational level rather than the RCT level, and it explains why so many population diet-disease associations never translate into proven causation.