A new multiple test approach for nursing care administration of deep vein thrombosis patients

In nursing care administration, it is critical to evaluate the risk assessment ability of the nurses for patients at different risk status. For an investigation carried out in the Jobst Vascular Institute at The Toledo Hospital, the nurse risk evaluation necessitates the analysis of risk assessment data for prophylaxis of deep venous thrombosis when comparing nurse risk assessment scores with master scores simultaneously at different risk categories. While conventional statistical methods fail to make any conclusion from the data, we construct a new stepwise confidence procedure that strongly controls the family-wise error rate and successfully detects the difference between the nurse score and the Master score. Compared with existing statistical methods, the new bivariate method is more powerful than the Bonferroni procedure and the Holm s step-down algorithm for this data set. It is also more robust than the Hochberg s step-up approach (which relies on an un-checkable assumption of positive dependence among test statistics to strongly control the family-wise error rate). In the data analysis of patients with deep vein thrombosis, the new method successfully detects the difference between the master risk assessment score and the nurse score, while the conventional statistical methods are unable to make any conclusive statement. The new statistical method is applicable to other fields of administration research simultaneously comparing management performances of two different groups under different scenarios.