Diagnostic and treatment decision making in community nurses faced with a patient with possible venous leg ulceration: a signal detection analysis.

BACKGROUND Judgements and decisions about venous leg ulcer management are characterised by uncertainty. Good judgements and reduced variations in practice require nurses to identify relevant "signals" in clinical encounters. Nurses, even experienced ones, vary in their ability to separate these signals from surrounding noise. OBJECTIVES Examine specialist and generalist nurses' discrimination of clinical signals and noise when (i) diagnosing venous versus other causes leg ulceration, and (ii) starting multilayer compression therapy. DESIGN A signal detection analysis within a cross sectional survey. SETTINGS Four English NHS districts. PARTICIPANTS Tissue viability specialist (n = 18) and generalist (district and practice nurses, n = 18) sampled from networks of nurses caring for people with leg ulcers. Mean age was 46 years, 78% had more than 10 years nursing experience. They worked on average 32.5 h per week, of which 10 h were spent caring for people with leg ulcers. METHODS 110 clinical scenarios based on anonymous patient data from a large clinical trial of compression therapy for leg ulceration. The scenarios were classed as either signal (venous leg ulcer present and/or compression therapy warranted, n = 57) or no signal cases (other kind of ulcer and/or compression therapy contraindicated, n = 53) by four experts. Nurses made diagnostic and treatment judgements for each scenario. A signal detection analysis was undertaken for each nurse. Measures of signal detection (d prime or d') and judgement tendency or bias (C) were computed. Differences between specialist and generalist nurses were tested for using the Mann Whitney U test and graphically explored using Receiver Operating Curves (ROC). RESULTS Specialists identified more true positive cases than the generalist nurses: 75% vs. 59% for the diagnostic judgement (p < 0.01) and 70% vs. 60% for the treatment judgement. They were significantly more sensitive to the signals present (d' 1.68 vs. 1.08 for the diagnostic judgement and 1.62 vs. 1.11 for the treatment judgement). Specialists exhibited a significantly higher bias towards initiating treatment (C = .81 vs. .56, p < 0.01) but this did not extend to their diagnostic judgements. Specialists also varied slightly less in their signal detection abilities. CONCLUSIONS Nurse specialism was associated with better, but still variable, clinical diagnostic and treatment signal detection in simulated venous leg ulcer management.