gms | German Medical Science

Background and Purpose: While a number of case mix classification systems have been developed for nursing homes, still little is known about nursing resource use in hospitals. In the FiliP study a measure for case mix was needed, so a patient classification system was developed and compared to the groups of the Pflegepersonalregelung (PPR).

Methods: The sample consisted of N=196 patients from 3 wards (general medicine, respiratory and geriatric) out of 3 different hospitals. Every nurse on duty was accompanied by a rater that measured and assessed times for nursing effort during day shifts. Time measurement accounted for direct and indirect care. Additionally, assessment data was collected for each patient (64 dichotomous items related to characteristics of the patient). For PPR data only the group assignment was used, not the normative times of that classification. The statistical learning algorithms CART and evtree were used for modeling.

Results: The 12 groups PPR model explained R2=48.09% of the variance in the measures times. A variant (Filip-PPR), that collapses the PPR to 5 groups, explained R2=56.62%. A 7-groups model that uses dichotomous variables of the FiliP assessment (walking, showering, venous catheter, clothing upper body, changing position in bed and bowel continence) explained R2=56.52%.

Conclusions: The specific findings should not be generalized to other hospitals or even other wards of the hospitals that participated in the study. However, the results show that PPR groups are suitable to discriminate groups of patients with similar resource use.