Page 714 - ISC PROCEEDINGS 21.4

classroom settings where full random group allocation is not feasible [14]. Internal validity
was protected through three mechanisms: (1) parallel-form questionnaires constructed
with controlled difficulty levels; (2) a two-week interval between T0 and T1 to reduce
rote- learning effects; and (3) subgroup analysis by AI tool to control for the intervention
variable.
Parallel-form questionnaire construction: Two versions (T0 and T1) were
independently designed by the research team and content-validated by two
pharmacoepidemiology experts with over five years of experience. Each version
comprised 15 items (10 MCQs + 5 short-answer questions) measuring the same seven
competency domains, applied to different articles of the same study design type.
Difficulty control was achieved through:
(a) balanced item count per indicator group; (b) equivalent computational
complexity (all NNT/ARR items provided two event rates requiring calculation and
interpretation); and
(c) pilot administration to five students outside the study sample to estimate
relative difficulty. The marked difference in task completion time between T0 and T1 is
therefore unlikely to be explained solely by differences in questionnaire difficulty.
2.2. Study participants
All 37 fourth-year pharmacy students (regular five-year program) enrolled in the
Pharmacoepidemiology module at Thanh Do University during Semester I of the 2025-
2026 academic year were included using census sampling.
Inclusion criteria: (1) fourth-year pharmacy students in the regular program at
Thanh Do University; (2) currently enrolled in Pharmacoepidemiology in Semester I/2025-
2026; (3) voluntary participation with signed informed consent; (4) personal device with
stable internet access. No students were excluded. The completion rate was 100% (37/37).
2.3. Practice materials
The intervention used three Open Access original articles selected according to
three criteria: (1) relevance to proton pump inhibitors (PPIs), a topic of high clinical
pharmacological value; (2) representation of three distinct research designs; and (3)
coverage of all pharmacoepidemiological indicator types required for learning.
Article 1 - SCCS: Maret-Ouda J et al. “Proton pump inhibitor use and risk of
pneumonia: a self-controlled case series study.” J Gastroenterol. 2023;58(8):734-
740. DOI: 10.1007/s00535-023-02007-5 (PMID: 37314495; Open Access PMC).
Practice indicators: Incidence Rate Ratio (IRR), 95% CI, person-time units.
Article 2 - Meta-analysis: Wu Z et al. “The impact of proton pump inhibitor exposure
on pneumonia: an updated meta-analysis based on randomized controlled trials.” Front
Pharmacol. 2025;16:1713256. DOI: 10.3389/fphar.2025.1713256 (Open Access). Practice
indicators: pooled RR =
1.10 (95% CI: 0.99-1.21; p = 0.07), low I², forest plot. Pedagogical note: this meta-
analysis of 20 RCTs (29,100 participants) did not reach statistical significance (p
= 0.07) with low I² a paradigmatic illustration of the distinction between statistical
significance and large sample size.
Article 3 - RCT: Bhatt DL et al. (COGENT Investigators). “Clopidogrel with or without
omeprazole in coronary artery disease.” N Engl J Med. 2010;363(20):1909-1917. DOI:
10.1056/NEJMoa1007964. Practice data: GI events omeprazole 1.1% vs. placebo 2.9%; HR
= 0.34 (95% CI: 0.18-0.63); ARR = 1.8%; NNT = 56.

713