How We Review Clinical Content
Every Clinical Drills problem is generated from written rules, not by AI at run time. Each answer is checked by automated tests and an independently re-solved reference set, every rule cites a published source, and a physician signs off each content pack before it can be deployed. If you find an error, tell us.
1. Problems are generated from rules
Each question type is a template: a set of rules that picks realistic values (real product strengths, survivable lab values, realistic infusion rates) and computes the answer with exact decimal arithmetic, never floating-point maths. The same template, version, seed and settings always produce the same problem, so any reported problem can be reproduced exactly. No AI is used at run time to write questions, grade answers or do the maths.
2. Answers are checked more than one way
- Automated tests. Every template has unit tests and property-based tests run across thousands of random seeds per template (10,000 per template for the ECG, Labs, ABG and pharmacy drills in the pre-release run; 1,000 to 3,000 for most dosage templates, and 10,000 for oral liquids), checking that answers are correct and problems stay within realistic limits.
- Reference ("gold") problems. Each template is tested against a fixed set of worked problems whose answers were solved separately.
- Independent re-solve. Before release, samples of generated problems were re-solved independently. For the newer drills this covered about 3,480 ECG, 7,200 labs, 3,600 ABG and 1,500 pharmacy problems, and no wrong answer keys were found.
3. Every rule cites its source
Every problem template and case records who reviewed it, when, and the sources it rests on. Reference ranges and rhythm definitions carry their own source references and are checked by automated tests against a hand-verified reference set. Typical sources by drill:
| Drill | Main sources | Signed off |
|---|---|---|
| Dosage Drills | FDA drug labels (DailyMed, FDA NDC directory), ISMP error-prone abbreviations, a published heparin nomogram (Raschke 1993) | |
| Pharm Calc Drills | FDA labels and NDC directory for strengths, pack sizes and reconstitution volumes; NIST Handbook 44 for unit conversions | |
| ECG Rhythm Drills | ACC/AHA/HRS guidelines (bradycardia, SVT, atrial fibrillation, ventricular arrhythmias), AHA 2025 adult ALS, LITFL ECG Library, StatPearls | |
| Labs Drills | MedlinePlus lab test pages, Mayo Clinic Laboratories critical values, ADA Standards of Care, drug labels for therapeutic levels | |
| ABG Drills | Berend et al., NEJM 2014; Winters 1967; Kraut and Madias 2007; Figge 1998; Berlin ARDS definition 2012; StatPearls |
Where sources disagree (for example the heart-rate cut-off for sinus bradycardia, or the normal bicarbonate range), the drill either keeps generated problems inside every cited range or makes the convention a setting so it matches your course.
4. Who reviews it
Clinical content is reviewed by Dr. Rafiq Muhammad, MBBS, PhD, the physician who founded Clinical Drills. To be transparent about the process: drafts are first put through a structured AI-assisted adversarial review (source checks, reading at least 50 generated problems per template, independent re-solving). Issues found are fixed or queued as decisions, and the founder then reviews and signs off. That sign-off is recorded in the content itself, and the deploy check refuses to publish any template or case without a recorded reviewer, date and sources. Some items are flagged for further specialist review (for example a pharmacist or telemetry reviewer) and are tracked until done.
5. Review dates
The dates in the table above are when each content pack was last signed off. Guide pages show their own "Last reviewed" date under the title. When a rule changes, it is re-reviewed before release.
6. Reporting an error
If an answer, explanation or value looks wrong, please contact us and choose "Clinical content or an answer". Include the drill and, if you can, the problem text. We investigate every report; a confirmed error is fixed and added to the reference problem set so it cannot come back.
What this does not mean
Educational practice only. Clinical Drills is for education and exam practice. It is not a clinical calculator or decision-support tool, and reviewed content is not a substitute for your facility's policies, current product labeling or your supervising clinician. Lab reference ranges vary between laboratories. See also the glossary and terms.