Red X iconGreen tick iconYellow tick icon

Research Team

Principal Investigator

  • Jackie Cumming [Health Services Research Centre, School of Government, Victoria U]


  • Dr Patrick Graham [Public Health and General Practice, UOC]
  • Antony Raymont [Health Services Research Centre, School of Government Victoria U]
  • Associate Professor Phil Hider [Public Health and General Practice, UOC]
  • Amohia Boulton [Te Pumanawa Hauora, Massey U]
  • Gregor Coster [Dept General Practice, School of Population Health, U of Auckland]
  • Mary Finlayson [School of Nursing, U of Auckland].

Host Institution Victoria University of Wellington, Health Services Research Centre


This project will adapt and apply new outcome measures and modern statistical approaches to evaluating hospital performance in New Zealand. As well as providing information on variability and trends in in-patient outcomes and hospital-level predictors of these outcomes, this project will provide the statistical basis for a sophisticated system for analysis and monitoring of hospital outcomes. A particular focus of this project is the identification of aspects of hospital organisation and working environment which may be related to patient outcomes.


In New Zealand, public hospitals account for between 40 to 50% of Vote: Health and form a focal point for public interest in the health system. In view of the substantial public investment in the hospital sector, the performance of the sector as whole and of individual providers is an important issue for hospital managers, users, staff and policy-makers. However, there has been relatively little research on comparative performance of New Zealand public hospitals, and New Zealand research on hospital-level determinants of patient outcomes is particularly scarce. This project is concerned with variation in hospital performance, as measured by patient outcomes, and seeks to identify potentially modifiable hospital-level predictors of patient outcomes. In addition the proposed research will also address issues relating to equity of outcomes between patient groups, particularly for Māori and non-Māori.


The specific aims of the project are:

  1. To quantify variability between New Zealand hospitals in a range of mortality and patient safety outcome indicators;
  2. To examine if differences in outcomes exists for Māori and non-Maori admitted to New Zealand public hospitals;
  3. To examine if differences in outcomes exist for rural and urban dwelling hospital patients and whether deprivation level of area of domicile is a predictor of outcomes;
  4. To analyse the effect of features of hospital organisation and working environment on patient outcomes;
  5. To adapt recent developments in the statistical methodology for monitoring hospital performance to the New Zealand context and make these developments available to the New Zealand health sector.

Research Methods

Patient outcomes, together with demographic and diagnostic information will be accessed via the National Minimum dataset (NMDS), for the period 1996-2004. Two broad sets of outcome variables will be emphasised: (1) Mortality indicators for specific conditions for which there is evidence that higher mortality may be directly associated with poorer processes of care; (2) Patient safety indicators including failure-to-rescue and specific, potentially preventable, hospital-acquired complications. Case-mix adjustment algorithms proposed in the international literature to maximise co-morbidity information extracted from administrative databases will be adapted for New Zealand use. Information on hospital organisation and the implementation of organisational structures intended to promote patient safety, as well as other patient safety initiatives, will be obtained from reviews of official documents, particularly DHB and CHE annual reports. This information will be supplemented by key-informant interviews with DHB quality managers and chairs of clinical boards. The financial position of hospitals over the study period will be ascertained from financial information reported in CHE and DHB annual reports and from financial reports of the Ministry of Health, Treasury and the Crown Financing Agency. Other hospital characteristics such as, teaching status and patient volumes will be obtained from official documents or derived from the NMDS. Indicators of hospital working environments will be developed from existing survey data, particularly the revised Nursing Work Index, which has been administered in national nursing surveys conducted in 2000 and 2004. Statistical analysis will employ hierarchical Bayesian generalised linear models, which markedly improve on reporting of crude hospital-specific estimates and non-hierarchical modelling by improving precision and correctly accommodating predictors operating at different levels of aggregation, such as patient and hospital-level characteristics.

Significance and relevance

This project is ultimately concerned with improving patient outcomes by identifying aspects of hospital organisation and work environment which promote positive patient outcomes. The methods developed in this study provide the statistical basis for a sophisticated system of performance feedback for health managers. Through the provision of timely information, it is to be hoped that hospital managers and the Ministry of Health will identify potential issues early, with the result that patient lives can be saved, health related quality-of-life improved, and resources not wasted through, for example, preventable readmissions.

Funded by Health Research Council

Back to top