Evaluating an Abstract's Causal Claim
A nursing journal abstract reports: 'In a convenience sample of 120 patients discharged from a 30-bed unit, 62% of patients who received a new handoff checklist (n=60) correctly stated three discharge instructions, compared with 48% of patients who received usual care (n=60). The difference was statistically significant (p=0.03).' The authors conclude: 'The handoff checklist proves that structured handoffs cause better patient comprehension and should be mandated hospital-wide.' The checklist was implemented on the day shift only, and nurses were not blinded. The unit's patients speak primarily English, and the checklist was piloted for two weeks. The study did not randomly assign patients or nurses to groups. The authors report no conflicts of interest.
1. Identify two words or phrases in the conclusion that overstate what the study can support, and explain why each is a problem.
Read the worked answer
'Proves' is the first overstatement. A single study, especially one with a convenience sample and no randomization, cannot prove a causal relationship; at most it provides evidence consistent with one. 'Cause' is the second. The study compared patients who received the checklist with those who did not, but it did not randomly assign them. Differences in patient acuity, language, education, nurse assignment, or day-shift staffing could explain the 62% versus 48% gap. The phrase 'hospital-wide' is a third overstatement: the checklist was tested on one 30-bed unit, on day shift, for two weeks. A finding in that setting does not automatically generalize to all units, shifts, or patient populations. The conclusion should be limited to an association in the studied unit.
2. Rewrite the conclusion in one sentence that accurately reflects the findings, including one limitation.
Read the worked answer
'In one 30-bed unit, patients who received a handoff checklist were more likely to state three discharge instructions correctly than patients who received usual care, but the non-randomized, single-unit design means the result is an association, not proof of causation.' This sentence keeps the direction of the finding (checklist associated with higher comprehension), avoids 'proves' and 'cause,' and names the design limitation. A second limitation could be added: the study was conducted on day shift only, so the result may not apply to night shift.
3. What additional information about the 120 patients would help you judge whether the comparison is fair?
Read the worked answer
You would want baseline characteristics of the two groups: age, primary language, hearing or vision impairment, education level, diagnosis, and acuity. If the checklist group had more English-speaking patients or less acutely ill patients, the 62% versus 48% difference might reflect those differences rather than the checklist. You would also want to know who collected the outcome and whether they knew which patients received the checklist; unblinded assessment can bias results. Finally, nurse staffing and shift matter: if the checklist was used only on day shift when more nurses were available, the comparison mixes the checklist with staffing. A fair comparison would randomly assign patients or units and blind outcome assessors.