The Field Guide › Paper
Develops and validates the Safety Organizing Scale (SOS), a 9-item self-report measure of the concrete behaviours that constitute a unit-level safety culture, addressing a real gap in the prior literature: existing safety-culture measures mostly captured the surrounding context that promotes safety (supportive leadership, procedures, whether errors get reported) rather than the behaviours themselves. Built directly on Weick and Sutcliffe's high-reliability-organising tradition, mapping each SOS item onto one of the five component processes of 'collective mindfulness': preoccupation with failure (a chronic, proactive wariness about what could go wrong, e.g. 'we spend time identifying activities we do not want to go wrong'), reluctance to simplify interpretations (deliberately questioning assumptions about routine work rather than accepting received wisdom), sensitivity to operations (an ongoing, mutual 'map' of colleagues' skills and knowledge), commitment to resilience (discussing errors and how they could have been prevented, so the unit gets better at catching and containing problems rather than just avoiding them), and deference to expertise (decision authority migrating to whoever actually has the relevant expertise during a crisis, regardless of formal rank, e.g. 'when a patient crisis occurs, we rapidly pool our collective expertise'). Validated with 1,685 registered nurses across 125 nursing units in 13 Catholic hospitals spanning six US states and a wide range of sizes and settings (rural, suburban, urban). Reliability was strong (Cronbach's alpha 0.88), confirmatory factor analysis supported a single underlying factor, and the SOS was shown to be statistically distinct from two theoretically related constructs, organisational commitment and trust in manager, via nested model comparisons. Aggregating individual responses up to the unit level was explicitly and rigorously justified (median within-unit agreement of 0.98, meaningful between-unit variance), confirming the SOS captures something genuinely shared across a unit rather than just individual attitudes. The paper's most consequential finding is prospective, not merely concurrent: units scoring higher on the SOS had significantly fewer reported medication errors and patient falls over the following six months, while trust in manager, organisational commitment, and a lower patient-to-nurse ratio all predicted higher SOS scores, as theorised. Addresses a real threat to this kind of validity directly: since safety-culture measures could plausibly just capture a unit's willingness to report rather than its actual safety, the authors show that units with more reported errors and falls were independently rated lower on quality of care by their own nurse managers, evidence that high reported-error counts in this sample reflect genuinely less safe units rather than merely more transparent ones. Limitations are candidly flagged: the sample is drawn from a single Catholic hospital system (albeit one spanning a wide range of sizes and locations), the measure was validated using nurses only even though safety is shaped by a much broader set of care providers, and the outcome data rely on incident reports rather than independently observed or audited events, a gap the authors explicitly flag for future work using direct observation or prospective clinical surveillance.