Exploring the construct validity of the Context Assessment Index (CAI) using the Rasch model of item response theory with data collected in a Western Australian tertiary hospital

Abstract

The adoption and integration of new knowledge into nursing practice is inconsistent. This is despite the active promotion of evidence-based practice by health care organisations and the nursing profession. The nursing profession is aware that there is a problem transferring empirical findings into accepted practices; however, it is not armed with the tools to determine what variables within a given clinical setting have the most influence on whether a change in practice is adopted. Nursing is not alone. There is growing recognition within health that context, i.e., the surroundings, circumstances, environment or settings in which care is delivered, is a critical, but poorly understood, determinate of success for any desired change in clinical behaviours across a range of professional groups.A structured quantitative approach for assessing the relative influence of specific factors within each context is therefore advocated. It is believed that the most appropriate selection of strategies to overcome the identified contextual barriers can then occur. To achieve this goal, measurement instruments that explore the role of context within health settings need to be created and tested in multiple settings. Ideally such tools evolve from within sound theoretical frameworks and are then extensively tested in the real world. Such a tool was recently created in the Republic of Ireland.The Context Assessment Index (CAI) was developed to contextually assess the readiness of different health care settings to implement procedural changes that reflect evidence-based practice. The CAI was derived from the Promoting Action on Research Implementation in Health Services (PARiHS) conceptual framework, which recognises the complexities of implementing change within health care organisations.The current research contributes to the testing of the CAI, by assessing the validity of the measurement instrument for potential use in Western Australian hospitals. This validation study is necessary because of differences in health care systems and the potential influence of Ireland‟s proximity to Europe and Australia‟s proximity to Asia on the recruitment of migrant workers. It is likely that the cultural mix of health care workers in these two settings is different. Consequently it is important to establish that the items within the CAI are measuring the same variables in Western Australia as in Ireland.Study participants were recruited over a 2 week period from the medical and surgical wards of a metropolitan tertiary hospital in Western Australia. Participants completed the CAI and a demographic questionnaire to assist in identifying cultural difference. A total of 210 research instruments were returned for a response rate of 61%.The Rasch measurement model of item response theory was used to statistically analyse the construct validity of the CAI in this Western Australian hospital. As nursing has historically relied on classical test theory to test measurement tools, this thesis discusses the relative contributions of different statistical approaches to validate the CAI. While preliminary Rasch analysis of the sample data has confirmed that the CAI is a unidimensional instrument, it has also identified that one item may not be contributing to the overall measurement of context. Further study using a larger sample is recommended to confirm these findings and to validate the absence of differential item functioning based on the cultural backgrounds of the respondents.