Symposium 23 Finding Haystacks and Needles - Using routinely collected data for nursing workforce research in the National Nursing Research Unit large data sets project

Symposium lead:

Peter Griffiths, Director, Nursing Reseach Unit, Nursing Research Unit, King's College London, London, United Kingdom peter.griffiths@kcl.ac.uk

Symposium chair:
Peter Griffiths (as above) OR Anne Marie Rafferty, Dean, Florence Nightingale School of Nursing, King's College London

Symposia focus:

In common with many elements of healthcare provision, the contribution of ‘nursing’ within the complex structures and processes of healthcare remains elusive. Optimal skills, education, numbers and configuration of nursing services remain unclear. Meanwhile significant changes are made to the composition and organisation of nursing work and the nursing workforce. These changes are predicated on assumed benefit or assumption that the workforce is infinitely plastic and there are no necessary consequences for client or staff well being. Identifying the marginal benefits (or otherwise) of more subtle effects of workforce configuration presents a major challenge because of the dominating effects of a patient’s condition and the availability of successful treatments in determining outcomes and so massive samples are required. Large datasets of routinely collected data are increasingly available for both workforce and patient outcomes.. Currently these resources are under utilised for research purposes in general and represent a largely untapped resource for assessing the impact of changes in workforce composition and organisation. These datasets can provide longitudinal measures of change and provide the opportunity to utilise naturally occurring variation across localities to address a range of questions. In these papers we address and discuss issues arising from the use of these data from the National Nursing Research Unit’s programme to identify, assess and link data sources that can support the nursing workforce studies. We will explore issues arising from this work including a critical review of existing research on nurse staffing and patient outcomes, the importance of ‘nurse sensitive’ outcomes, the invisibility of ‘nursing’ within NHS workforce data, the limitations inherent in conducting research at this level and the opportunities created by bringing a participative approach to this mode of research.

Source of Funding: Department of Health Policy Research Programme

Level of Funding: Core funding from Department of Health

Abstract 1: The Needle in the Haystack: the evidence base on nurse staffing and patient outcomes

Jill Maben, Senior Research Fellow National Nursing Research Unit, King's College London Peter Griffiths, Director National Nursing Research Unit, King's College London

In this paper we review existing research that has utilised large data sets to explore the impact of nursing. In relation to nursing and care work there are relatively few examples of studies which have utilised large data sets. We present the results of a review of the literature which seeks studies where routinely collected discharge data sets or equivalents have been used to determine the link between nursing workforce configuration and characteristics and patient outcomes. The field has been dominated by studies examining the link between nurse patient staffing ratios and patient outcomes in acute (generally surgical) hospital settings in the USA (1.2). This research has limitations both in terms of limited generalisation, potential confounding and misattribution of outcomes through differences in the level of measurement of staff and patient data. Further, where complex statistical models are deployed there may be contested assumptions and alternative perspectives which warrant further analysis. Opportunities for replication or reanalysis are limited. The ability to assess the impact of ongoing change in the wider context on these isolated findings is also limited (3). There is essentially no research that examines nursing work in other settings or specialities and in particular little research to support current developments which envisage a greater role for nurses in chronic and acute disease management in settings outside of acute hospital.

References:

  • Aiken, LH, Clarke, SP, Sloane, DM, Sochalski, J, Silber, JH, 2002. Hospital Nurse Staffing and Patient Mortality, Nurse Burnout, and Job Dissatisfaction. JAMA 288 (16), 1987-1993
  • Needleman J, Buerhaus P, Mattke S, Stewart M and Zelevinsky K (2002). Nurse-Staffing Levels and the Quality of Care in Hospitals. New England Journal of Medicine, 346, 1715-1722, Massachusetts Medical Society, USA
  • West E, Maben J, Rafferty AM (2006) Nursing and patient outcomes: How can employers provide the right environment for nurses to deliver high quality care. Harvard Health Policy Review 7 (1), Spring 2006: 64-84

Abstract 2: Is it a needle? The challenges of linking nurse sensitive indicators and to workforce data

Trevor Murrells, Statistician, National Nursing Research Unit, King's College London

In this paper we discus potential data sources for nurse sensitive outcomes and patient safety indicators (PSIs) and make a preliminary assessment of their utility when linked to staffing data. PSIs and other nurse sensitive outcome indicators need good face and construct validity, amenability to precise measurement, minimum bias and insulation from perverse incentives to improve performance. There must be evidence of sensitivity to variations in the quantity and/or quality of nursing inputs (1). PSIs that are recognised as nurse sensitive include decubitus ulcer, failure to rescue, postoperative respiratory failure and postoperative pulmonary embolus or deep vein thrombosis (1). However, these provide limited opportunities to demonstrate positive impacts of wider nursing practice (e.g. prevention, treatment, risk factor reduction, avoiding additional complications). Some of the criteria may apply to other routinely collected outcome indicators in other settings. We will examine the evidence and discuss our ongoing work to identify other indicators.

Other issues arise when attempting to utilise indicators to study nursing related variables. Available data is often aggregated at levels which pay scant regard to aspects of nursing care. Hospital stays are reported by consultant episode but not by nursing unit or specialty (i.e. hospital ward). Very little information exists on nurses working at the GP practice level and again data is aggregated at a high level. To analyse outcomes from a workforce perspective, data from different sources may need to be combined. We will illustrate the challenges with an analysis linking data from the NHS Workforce Benchmarking database and the Quality Outcome Framework (QOF) indicators collected from general practices in England in order to determine if nurse staffing impacts upon general practice performance. Limitations will be highlighted and new or emerging sources of outcome/workforce data will be identified. (289) (1) Department of Health and Human Services Agency for Healthcare Research and Quality, 2007b. Guide to Patient Safety Indicators, Version 3.1 Other issues arise when attempting to utilise indicators to study nursing related variables. Available data is often aggregated at levels which pay scant regard to aspects of nursing care. Hospital stays are reported by consultant episode but not by nursing unit or specialty (i.e. hospital ward). Very little information exists on nurses working at the GP practice level and again data is aggregated at a high level. To analyse outcomes from a workforce perspective, data from different sources may need to be combined. We will illustrate the challenges with an analysis linking data from the NHS Workforce Benchmarking database and the Quality Outcome Framework (QOF) indicators collected from general practices in England in order to determine if nurse staffing impacts upon general practice performance. Limitations will be highlighted and new or emerging sources of outcome/workforce data will be identified.

References:

  • Department of Health and Human Services Agency for Healthcare Research and Quality, 2007b. Guide to Patient Safety Indicators, Version 3.1

Abstract 3: Where’s the nurse where’s the patient? (or are we looking in the right haystack?)

Elizabeth Smith, Research Associate, National Nursing Research Unit, King’s College London Sally Brearley , Honorary Research Fellow, National nursing Research Unit, King’s College London

The use of large data sets offers considerable potential to advance our knowledge of the role of nursing within health services. It may also tell us something worthwhile about the experience of patients. However, such studies raise as many questions as they answer. For example, the aggregation of data on a large scale masks considerable variation amongst both individuals and groups. This applies to nurses and patients. It is also unlikely to explain why the observed results occur, or how to move forward. Nurse-specific data relates almost exclusively to acute settings: in the data sets currently generated in primary care it is difficult, if not impossible, to disentangle the nursing component of care, as highlighted in paper 2.

This paper will argue that participative approaches to research can enhance studies of large data sets. Such approaches have the power to explain the reasons for, and the relative importance of trends from the perspectives of policy makers, managers, healthcare staff, patients, carers and the public and they can help to ensure that research is relevant to the groups that research intends to serve (1). If linked to strong theoretical propositions participative approaches can reveal possibilities for how to move forward in the context of wider social or political challenges. Our discussion is informed by an examination of trends in user involvement in research (2), and recent findings on the role of community nursing, particularly in relation to long-term conditions (3). We discus the implications of these for the future generation of large data sets: How might they develop in order to better capture what nurses do and what patients value?

References:

  • Department of Health (2004) Patient and public involvement in health: The evidence for policy implementation, London: Department of Health
  • Smith, E., Ross, F., Donovan, S., Manthorpe, J., Brearley, S., Sitzia, J., Beresford, P., (2005). User involvement in the design and undertaking of nursing, midwifery and health visiting research. London: Nursing Research Unit, King’s College London. Full report (Ref: SDO/69/2003) and annotated bibliography available from http://www.sdo.lshtm.ac.uk
  • Sargent, P; Pickard, S; Sheaff, R and Boaden, R. (2007) Patient and carer perceptions of case management for long-term conditions. Health & Social Care in the Community (OnlineEarly Articles) doi:10.1111/j.1365-2524.2007.00708.x