Evidence-Based Practice

Genetics changes faster than most fields, and a counselor's knowledge decays accordingly: gene-disease relationships are reclassified, variant interpretations are revised, penetrance estimates shift as unselected cohorts are studied, and treatments arrive for conditions that had none. Evidence-based practice is the discipline of noticing that the thing you learned is no longer true.

• The core skill is appraisal, not accumulation. Reading more is less useful than being able to judge whether a given paper should change what you tell patients.
• Ascertainment bias inflates penetrance. Most early penetrance figures come from families identified because they were heavily affected. Estimates from unselected populations and biobanks are consistently lower, which is why several long-quoted penetrance numbers have fallen substantially. When quoting a figure, know which population it came from.
• Study design determines what a finding can support. Case reports generate hypotheses. Case-control studies establish association. Cohort studies with unselected ascertainment give the most usable risk figures for counseling an unaffected carrier.
• Gene-disease validity is itself assessed formally. ClinGen curates gene-disease relationships into categories from definitive to disputed and refuted, and genes have been refuted after years of clinical testing. A panel including a disputed gene can generate findings that mean nothing.
• Variant classification is time-stamped. An ACMG and AMP classification reflects the evidence on the date it was issued. Reanalysis and reclassification are expected, which is what makes recontact and periodic reanalysis a real clinical obligation rather than an optional courtesy.
• Guidelines differ in strength and in purpose. NCCN, ACMG, ACOG, and specialty statements vary in method, update frequency, and whether they grade evidence. Knowing when a guideline was last revised matters as much as what it says.
• Distinguish clinical validity from clinical utility. A test can accurately detect a variant, which is validity, while making no difference to outcome, which is utility. Payers, correctly, buy utility.
• Sources that update: ClinGen, ClinVar, GeneReviews, and OMIM, with the caveat that ClinVar aggregates submissions of varying quality and that conflicting interpretations are common and informative rather than a defect.
• Absence of evidence is a real answer. For rare conditions the honest statement is often that data are limited, and saying so is more useful than extrapolating from a handful of cases as though it were established.

• A counselor has quoted a penetrance figure for years, taken from an early study of highly affected families. A large unselected cohort reports a substantially lower figure. She updates what she tells unaffected carriers identified incidentally, while noting that risk in someone from a heavily affected family may be higher than the population estimate.
• A patient's variant of uncertain significance is reclassified as likely benign three years after testing. The counselor contacts her, corrects the record, and stops surveillance that was being done on the basis of the original report.
• A laboratory panel includes a gene ClinGen has classified as having limited evidence for the condition being tested. The counselor discusses that a variant in that gene would be difficult to interpret and might generate uncertainty without benefit, and considers a panel with better-curated content.
• A family asks about a treatment they read about in a single case report. The counselor explains what a case report can and cannot establish, without dismissing the family's hope, and describes what evidence would be needed and whether a trial exists.