The objectives of the reported research were: (1) to further develop a variance model for nurse staffing which includes stochastic aspects of patient care demand for nursing services; and (2) to quantify the expected values and variances of patient care demand time due to errors in patient classification.

A variance model is proposed because of the need to be able to specify the probability of having sufficient staff to care for the patients. The model incorporates the patient care demand time for nursing services for inpatients and the variance at the time of classification, in addition to the expected patient care demand time and variance for admissions, discharges, transfers, leaves, and patient classification errors. The model is constructed to determine the patient care needs by shift and day of the week for the next three shifts as well as the present shift. Methodologies are proposed to collect the necessary data to implement the model.

One of the components of the model, patient classification error was selected to quantify the magnitude and sources of errors. A number of two observer experiments were run. The major error sources were: lack of up to date kardex information, lack of understanding about classification definitions, and classification by nursing personnel not personally involved with assessing a particular patient's needs. An overall estimate of the mean difference and variance in patient care demand requirements was determined. The major conclusion was that the errors, even with careful implementation, are of sufficient magnitude that they must be explicitly considered as a model parameter.