ABGC Certification Examination (CGC credential) common mistakes
CGC · Understand the traps and how to avoid them.
Treating autosomal recessive recurrence risk as 25% for every future pregnancy without first confirming that both parents are confirmed carriers of the same pathogenic variant, then applying that number to a pregnancy conceived with a different partner.
The 25% figure is memorized as a reflex during prenatal and preconception questions, so candidates stop reading for partner status, consanguinity, or de novo status. Exam items often change one variable (new partner, unknown paternal carrier status, variant not detected in one parent) precisely to test whether the number is derived rather than recalled.
What to do instead
Before computing any recurrence risk, write out the genotypes of both parents explicitly, note whether each variant is confirmed by testing or inferred from ancestry or family history, and only then apply the Punnett square. If one parent's status is unknown, express the answer as a conditional risk (for example, carrier probability times 25%) rather than a flat number.
Reporting a variant as pathogenic because it is rare or because a patient has a matching phenotype, skipping the ACMG/AMP evidence codes and the distinction between pathogenic, likely pathogenic, and variant of uncertain significance.
Clinical urgency pushes candidates toward closure, and the five-tier classification vocabulary is familiar enough to feel mastered. Items test whether the candidate knows that a VUS must not be used to make reproductive or management decisions and that criteria combinations, not phenotype match alone, drive classification.
What to do instead
For every classification item, state the specific evidence categories being applied (population frequency, computational prediction, functional data, segregation, de novo occurrence) and the strength assigned to each. When evidence is incomplete, select VUS and describe what follow-up (segregation study, RNA analysis, reanalysis at a set interval) would upgrade it.
Calculating Bayesian risk using the wrong prior or forgetting to include the proband's affected status as conditioning information, producing a carrier probability that ignores an unaffected sibling or an unaffected parent.
Bayes problems are usually practiced in textbook form with clean pedigrees; real items add unaffected relatives whose existence changes the prior. Candidates also confuse the prior (population or family-history-based carrier rate) with the conditional probability from test results.
What to do instead
Set up a three-column table every time: hypothesis, prior, and conditional probability of the observed data under each hypothesis. List every piece of conditioning information in the pedigree, including unaffected siblings, before multiplying. Sanity-check that the posterior moves in the direction the new information should push it.
Offering testing to a patient without first addressing what the result could mean for relatives, insurance, or reproductive plans, and without documenting whether the patient wants that information shared.
Informed consent is taught as a form to sign, so candidates treat it as a checkbox rather than a process. Items describing a patient who agrees to testing but has not considered family implications are testing whether the counselor raised those issues before sample collection.
What to do instead
Structure consent around the specific test: what is being tested, possible results including VUS and incidental findings, implications for relatives, and who the patient authorizes to receive information. Document the discussion, not just the signature, and revisit consent when the test scope changes.
Giving direct advice ("you should terminate," "you should test") instead of presenting options neutrally and exploring the patient's own values before and after information is given.
Candidates want to be helpful and perceive neutrality as cold or evasive, especially in emotionally charged scenarios. Exam items reward nondirectiveness as a measurable behavior: eliciting values, reflecting, and supporting the patient's decision rather than substituting the counselor's judgment.
What to do instead
Practice a consistent sequence: elicit what the patient already knows and wants, provide information in small chunks, check understanding, then ask what matters most to them. When a patient asks "what would you do," reflect the question back and offer a structured decision framework rather than a recommendation.
Missing the counseling and psychosocial cues embedded in a vignette because the candidate jumps to the genetic or numeric answer, then selecting the option that addresses the medical facts while ignoring the patient's stated fear, grief, or family conflict.
The exam mixes quantitative and counseling items, and under time pressure candidates scan for numbers and diagnoses. Items frequently place the decisive information in an emotional statement or a family dynamic that must be addressed before or alongside the medical content.
What to do instead
Read each vignette twice: first for the clinical facts, second specifically for feeling words, relationship descriptions, and stated concerns. If the question asks what the counselor should do next, prefer the response that acknowledges and explores the patient's experience before providing more information.
Applying adult-onset predictive testing protocols to a minor, or arranging carrier testing for a child, without checking whether the condition is childhood-onset and treatable and whether the child assent and parental consent framework has been addressed.
Candidates generalize from adult cases and forget that predictive testing for adult-onset conditions is generally deferred until the person can consent. Items about adolescents requesting testing or parents requesting testing of a child hinge on this distinction.
What to do instead
Ask two questions before recommending testing in a minor: does the condition have childhood onset, and is there an intervention that changes outcome in childhood? If both answers are no, discuss deferral, document the rationale, and explore the family's concerns and the child's developing autonomy.
Quoting a detection rate or residual risk without stating the test's limitations, or presenting a negative result as eliminating risk entirely.
Screening and testing are conflated, and candidates quote sensitivity figures as if they were certainties. Items test whether the candidate communicates residual risk after a negative result and distinguishes screening from diagnostic testing.
What to do instead
For every test discussed, state what it detects, what it misses, and the residual risk after a negative result in plain language. Pair the number with the limitation in the same sentence so the patient hears both together, and confirm understanding with a teach-back question.
Failing to recognize a referral trigger such as a family history pattern, ancestry-based carrier risk, or a red-flag finding, and instead reassuring the patient that no action is needed.
Candidates focus on the presenting question and do not systematically screen the family history for patterns (early onset, multiple affected relatives, bilateral or multifocal disease, consanguinity). Reassurance feels supportive but bypasses the indication for referral.
What to do instead
Apply a fixed checklist to every family history: number of affected relatives, ages at diagnosis, lineage, consanguinity, and ancestry. When any threshold is met, state the indication explicitly and offer referral rather than reassurance, and document the reasoning.
Letting time pressure drive selection of the longest or most comprehensive-sounding option rather than the option that matches the specific task in the question stem.
Stems often ask for the next step, the most appropriate response, or the best explanation, and distractors are written to sound thorough. Candidates who do not identify the exact task choose by plausibility instead of by fit.
What to do instead
Underline the operative phrase in the stem before reading the options, eliminate options that answer a different question, and choose among the remainder by asking which action is appropriate at this point in the process. Flag uncertain items, return to them, and avoid changing answers unless a specific misreading is identified.