There is an established international body of literature investigating nursing care quality, a subset of which aims to explicate the association between nurse staffing levels and inpatient complication rates. Systematic reviews of the topic have reached inconsistent conclusions about this relationship, and some have proposed that overcoming certain methodological challenges may resolve incongruent findings. There is also some evidence that perceptions of the impact of nursing care on patient outcomes vary across nursing roles (e.g. patient care, education, research, or management). This study sought to explore this complex issue by applying data linkage analysis methods in a way which had not been previously examined in the nursing care quality research. Data linkage draws together information from varied administrative sources to provide a detailed longitudinal record of the current and previous health status of unique individuals. This approach enabled the following methodological challenges to be addressed: accurate attribution of inpatient complications to nursing care and not the patient’s underlying health condition; measurement of nurse staffing at the unit level per fortnight; measurement of specific inpatient exposure to nurse staffing levels based on the units patients were admitted to during each hospitalisation; and hospitalisation-specific adjustment for patient comorbidities and characteristics such as socioeconomic status.
This study addresses four aspects of nursing care quality: nurses’ perceptions of the impact of nursing care on clinical outcomes, identification of nursing sensitive inpatient complications, measurement of nurse staffing levels, and the association between nurse staffing levels and inpatient complication rates. An exploratory survey of nurses’ perceptions of the impact of nursing care on inpatient complications was conducted with a convenience sample of nurses attending a three day nursing practice conference.Investigation of nurse staffing and inpatient complications used a retrospective longitudinal hospitalisation-level cohort design comprising Western Australian Department of Health administrative data collections spanning 1 January 2004 to 31December 2008. Adult inpatient hospitalisation data and nurse staffing records for intensive care, high dependency, cardiac care, general medical, and general surgical units at three metropolitan acute care hospitals were analysed.
Existing algorithms which identify nursing sensitive inpatient complications in administrative data were modified to include look-back periods. Linked health data enabled the algorithms to include look-back periods which identified patient-specific disease information recorded during earlier hospital admissions when classifying whether an inpatient complication should be attributed to the patients underlying health condition or nursing care. Survival analysis was used to examine how look-back periods impacted on nursing sensitive inpatient complication rates.
Nursing payroll and inpatient unit movement records were combined to measure two aspects of unit-level nurse staffing per fortnight: nursing hours per patient day and nursing skill mix. Fortnightly nursing hours per patient day were validated against figures reported by the Western Australian Department of Health using linear mixed models, descriptive statistics compared the frequencies with which nursing hours per patient day met set targets, and the proportions of nursing care hours delivered by nurses of different skill levels were graphically presented.
Generalised estimating equations were used to test the impact of nurse staffing on inpatient complications. Nurse staffing was measured using average exposure to skill mix adjusted nursing hours per day in hospital, individually computed per hospitalisation. Nursing sensitive inpatient complications were identified with a two year look-back period included in the identification algorithms. T-tests and chi-squared tests were used to compare demographic characteristics of inpatients with and without complications.
Nurses with different educational backgrounds and work roles had different opinions regarding the extent to which nursing care impacts on thirteen clinical indicators but agreed on the indicators most affected by nursing care. Nurses who worked in roles delivering patient care or who had lower levels of educational attainment were less likely to indicate that nursing care impacted on inpatient complications than their non-clinical, postgraduate educated counterparts.
There was a reduction in estimated rates of all nursing sensitive inpatient complications when a look-back period was applied. The majority of relevant patient disease information was identified with a two year look-back period from the hospitalisation during which the nursing sensitive inpatient complication was identified.
There was no significant difference between mean nursing hours per patient day estimated using raw administrative data sources and those reported by the Western Australian Department of Health (p=0.788). Using raw administrative data to measure nursing skill mix and fortnightly (as opposed to annual) nursing hours represented a more detailed yet equally accurate alternative to publicly reported nurse staffing information. Hours of nursing care met set targets the majority of fortnights during the study period.
The directions of the associations between nurse staffing and each inpatient complication were not consistent; nor were they consistent across nurse skill mix groups, or hospitalisations with different unit movement patterns. Hospitalisations with complications had significantly different demographic characteristics compared to those without (all p<0.001).
This study has furthered our understanding of how nurses delivering patient care perceive the impact they have and how data linkage methods can strengthen nursing care quality research methodology. Using linked hospitalisation data to incorporate look-back periods improves the accuracy of estimating nursing sensitive inpatient complications in administrative data. Rates may be overestimated if look-back periods are not applied. Fortnightly, unit-level measures of both hours of nursing care and skill mix were derived from the administrative data sources, representing a more detailed alternative to publicly available staffing measures. Accessing raw data enabled analysis of nurse staffing skill mix which would not have been possible with publicly reported information. Despite addressing a number of methodological challenges cited in the literature, this study did not find a clear relationship between patient exposure to nurse staffing and nursing sensitive inpatient complication rates. Despite large sample sizes, detailed administrative data and increased methodological rigour, evidence suggests that there is a need for a wider scope to examine the nurse staffing patient complications paradigm.
|Qualification||Doctor of Philosophy|
|Publication status||Unpublished - 2015|