Provimi i Informatizuar i Mjekësisë common mistakes
Provimi i Informatizuar i Mjekësisë · Understand the traps and how to avoid them.
When a question describes a patient with fever, dyspnea and hypotension, the candidate jumps straight to the diagnosis (e.g., 'septic shock') without first stating the physiologic problem (hypoperfusion) and the immediate stabilizing step, so they pick a diagnostic test or an antibiotic instead of the correct answer about airway/fluids/vasopressors.
Exam pressure and pattern recognition push candidates toward the most familiar-sounding disease label; they read the stem as a 'name the diagnosis' item when it is actually a 'next best step' item.
What to do instead
Before answering, classify the question type in one word: diagnose, next step, mechanism, or drug. For 'next step' items, rank options by immediacy (ABC first) and eliminate any option that is diagnostic before lifesaving.
The candidate computes a sensitivity, specificity, positive predictive value, or number needed to treat by plugging numbers into a memorized formula without checking which marginal totals the stem gives, so they swap the PPV denominator (all positives) with the sensitivity denominator (all diseased).
Formulas are memorized as symbols rather than as sentences about populations, and the stem's wording ('of those who tested positive...') is skimmed.
What to do instead
Rewrite each 2x2 cell in words before calculating, and state the denominator aloud: sensitivity = diseased who test positive; PPV = positives who are diseased. Practice reconstructing the table from any three given values.
On an ECG or rhythm strip described in text, the candidate treats any wide-complex tachycardia as ventricular tachycardia and any narrow-complex tachycardia as sinus, ignoring rate regularity, P waves, and the patient's stability, and then chooses a drug rather than cardioversion for an unstable patient.
Pattern matching on a single feature ('wide = VT') replaces a systematic reading; stability is treated as background rather than as the decision node.
What to do instead
Use a fixed sequence every time: rate, rhythm/regularity, QRS width, P waves, then stability. If unstable, the answer is synchronized cardioversion (or defibrillation if pulseless) regardless of the exact rhythm.
The candidate answers an ethics or medical-law item by choosing the option that feels most compassionate (e.g., disclose everything to the family, or treat a minor without consent) instead of the legally correct principle (confidentiality, assent plus parental consent, or emergency exception).
Clinical empathy is over-weighted relative to the formal hierarchy of autonomy, capacity, and legal exceptions, and Albanian and EU legal frameworks are not actively recalled.
What to do instead
Build a one-page decision tree: is the patient competent? is there an emergency? is there a legal duty to report? Apply autonomy first, then check the specific exception (minors, public-health reporting, court order) before choosing.
In pharmacology items, the candidate selects a drug based on indication alone and ignores contraindications, pregnancy category, renal/hepatic dosing, or the interaction named in the stem, so they choose a correct-for-disease but unsafe-for-this-patient option.
Study is organized by drug class and indication, so contraindications and interactions are learned as isolated lists rather than as filters applied after indication.
What to do instead
For every drug question, run a three-filter checklist in order: indication, contraindication in this patient, then interaction with the patient's current medications. If two options pass the first filter, the stem usually contains the discriminator.
The candidate interprets a lab value or vital sign against adult reference ranges in a pediatric or obstetric stem, missing age-specific norms (neonatal bilirubin, pediatric heart and respiratory rates, pregnancy-adjusted hemoglobin and blood pressure).
Adult medicine dominates study time, and reference ranges are memorized as single numbers rather than as age- and pregnancy-stratified values.
What to do instead
Before interpreting any value, write the patient's age and pregnancy status at the top of your scratch space, then recall the stratified range. Drill the five most-tested exceptions: neonatal jaundice, pediatric vitals, pregnancy physiology, elderly renal function, and postmenopausal ranges.
On surgery items, the candidate picks a definitive operation for a patient who first needs resuscitation or source control, or picks conservative management for a patient with peritonitis, strangulation signs, or free air, misjudging the timing of intervention.
The stem's operative findings are read as the answer key rather than as triggers for a decision about urgency; 'when to operate' is studied less than 'what operation'.
What to do instead
Classify every surgical stem into one of three buckets: resuscitate first, operate now (peritonitis, ischemia, obstruction with strangulation), or investigate further. Answer only after the bucket is chosen.
In biostatistics and epidemiology items, the candidate confuses study-design implications (cohort vs case-control vs cross-sectional) with the measure of association, or mistakes a p-value for effect size, and selects an answer that misstates what the design can prove.
Design and analysis are learned as separate chapters, so the link between design and the valid measure (risk ratio, odds ratio, prevalence ratio) is not automatic.
What to do instead
For every study-design question, state the design, then the only valid measure it yields, then whether incidence or prevalence was sampled. Reject any option that claims causation from a cross-sectional or case-control design.
The candidate answers pathology and pathophysiology items by naming the disease instead of the mechanism the question asks for, or reverses cause and consequence (e.g., calls hyperkalemia the cause of renal failure rather than its result), producing a plausible but directionally wrong option.
Disease labels are more strongly encoded than mechanistic chains, and stems using 'because', 'due to', or 'results in' are read as identification questions.
What to do instead
Circle the causal word in the stem and draw a two-arrow chain (cause to mechanism to consequence) before reading options. Eliminate any option that reverses an arrow, even if the terms are correct.
In medical informatics and health-management items, the candidate answers from general IT intuition (e.g., 'more data is better', 'the newest system is best') rather than from health-system concepts such as interoperability standards, data protection, clinical decision support limits, and cost-effectiveness, and so picks the option that sounds modern but is not the standard.
These domains receive the least study time and are treated as common sense, so technical-sounding distractors win over the actual governance and evaluation principles.
What to do instead
Learn a short list of anchor concepts: HL7/FHIR interoperability, GDPR and patient-data confidentiality, alert fatigue and CDS specificity, and the difference between efficacy and cost-effectiveness. Match the stem to one anchor before choosing.