CPCD common mistakes
CPCD · Understand the traps and how to avoid them.
Treating 21 CFR Part 11 as interchangeable with ICH E6 GCP when answering questions about electronic records, electronic signatures, and audit trails, instead of distinguishing the legally binding regulation from the guideline.
Both are taught together in study start-up training, so candidates memorise them as one 'compliance' topic and cannot separate the US regulatory requirement from the international guideline when a question tests one specifically.
What to do instead
Build a two-column comparison of Part 11 (scope, predicate rules, e-signature components, audit trail requirements, record retention) versus E6 (sponsor/investigator responsibilities, essential documents, monitoring). Before answering, identify which framework the stem invokes.
Selecting 'the monitor reviews the CRFs' or 'the sponsor audits the site' as the control that prevents erroneous data entering the database, rather than source data verification and edit-check/programmed validation at the point of entry.
Monitoring and auditing are highly visible activities, so candidates default to them for any data quality question without distinguishing real-time/preventive controls from retrospective detection controls.
What to do instead
Classify each control by timing (preventive at entry vs. detective after entry) and by who performs it. Practise sorting controls into a grid before choosing an answer.
Confusing a query with an edit check, and confusing a data clarification with a protocol deviation, when describing discrepancy workflows.
The terms appear in the same SOP sections and the candidate's site experience often involved only one of them, so the definitions blur.
What to do instead
Write one-sentence definitions with an example of each: automated edit check (system rule at entry), manual query (reviewer-generated), data clarification form (site response), deviation (protocol non-compliance). Match the stem's actor and trigger to the term.
Assuming database lock can occur once the last patient completes, without listing the prerequisite steps (all queries closed, SAE reconciliation complete, coding complete, external data transfers received and reconciled, blind review/unblinding decision, signed lock authorisation).
Lock is memorable as a single milestone, and candidates underestimate the checklist because they have not sequenced the close-out activities.
What to do instead
Memorise a linear lock-readiness checklist and practise ordering tasks. If an answer option skips reconciliation or coding, it is usually wrong.
Mishandling MedDRA and WHO Drug Dictionary questions by mixing up the hierarchy levels (SOC, HLGT, HLT, PT, LLT) or treating MedDRA as the product/drug dictionary.
Both dictionaries are 'coding' topics studied together, and the multi-level MedDRA structure is easy to memorise in the wrong order or to conflate with WHO Drug's ATC structure.
What to do instead
Draw both hierarchies from memory: MedDRA top-down (SOC to LLT) and WHO Drug (ATC levels plus preferred name). Rehearse which dictionary codes adverse events versus concomitant medications.
Answering safety reconciliation questions by describing only the site-to-sponsor SAE flow, omitting the comparison of the safety database against the clinical database (and against external sources such as the IB or coding dictionary).
Candidates recall the reporting pathway but not that reconciliation is a two-way comparison exercise with documented discrepancy resolution.
What to do instead
Frame every reconciliation answer as 'compare source A to source B, log discrepancies, resolve, document, re-verify.' Apply that template to SAE, labs, and external vendor data.
Designing or critiquing a CRF without linking each field to the protocol objective and analysis plan, producing fields that are collected but never analysed or analyses with no supporting field.
Candidates focus on layout and formatting because those are visible, and skip the traceability step from protocol endpoint to CRF item to database variable.
What to do instead
For any CRF design question, mentally walk protocol schedule of assessments to CRF page to database field. Flag orphan fields and missing endpoints as the defect the question is testing.
Describing data integrity only in terms of 'accurate data' and omitting ALCOA+ attributes (attributable, legible, contemporaneous, original, accurate, plus complete, consistent, enduring, available) and the audit trail that supports them.
'Data integrity' sounds self-explanatory, so candidates answer from intuition rather than applying the formal attribute framework the exam expects.
What to do instead
Memorise ALCOA+ as an acronym and practise naming which attribute a scenario violates (e.g., back-dated entries violate contemporaneous; unsigned corrections violate attributable).
Treating archiving and close-out as 'store the files' and missing requirements for retention period, readability/restorability of electronic records, transfer of investigator obligations, and reconciliation of returned drug and essential documents.
Close-out is studied last and often skimmed, and candidates conflate physical storage with regulatory record retention obligations.
What to do instead
Create a close-out checklist covering essential document finalisation, retention timelines, e-record accessibility, and accountability reconciliation; use it to eliminate incomplete answer options.
Skipping the data management plan's role when answering study start-up questions, instead naming only the CRF and database as start-up deliverables.
The DMP is a planning document rather than a tangible artefact, so candidates underweight it and forget it governs validation, query, coding, and lock procedures.
What to do instead
Memorise the DMP's core sections (data flow, CRF/database specs, validation and query rules, coding conventions, SAE reconciliation, external data, lock criteria) and treat 'documented plan' as the default correct answer for process questions.