The development and testing of a contextual model for healthcare quality improvement using Lean and the Model for Understanding Success in Quality (MUSIQ) : a thesis presented in partial fulfilment of the requirements for the degree of Master of Quality Systems at Massey University, Palmerston North, New Zealand

Abstract

This study developed a new theoretical model of quality improvement (QI) contextual factors, for QI activity undertaken at the healthcare microsystem level. The Model for Understanding Success in Quality (MUSIQ) (Kaplan, Provost, Froehle, & Margolis, 2012), was aligned with Lean improvement activity using the Toyota Way framework. The aim of the research was to improve the effectiveness of healthcare quality improvement initiatives by providing more understanding of the associations, relative importance and precise functioning of critical contextual factors. A new survey instrument, based on the literature, was developed to collect data and the hypothesised theoretical relationships were tested using the partial least squares path modelling (PLSPM) technique. QI practitioners at a large New Zealand District Health Board were surveyed on a range of contextual factors hypothesised to influence improvement outcomes. All survey participants had recently completed a small-scale improvement project using Lean, or were participants in training programmes that introduced them to Lean thinking and methods. Some participants worked autonomously on improvements of their own selection; others were part of a wider training programme derived from the National Health Service’s (UK) ‘productive ward’ programme. In the healthcare organisational context, the majority of these improvement initiatives were carried out at the microsystem level – initiated and delivered by the teams responsible for the work processes being modified. Survey responses were first analysed via principal components analysis (to examine the dimensionality of the scales) and then PLSPM. The defined contextual factors for ‘Teamwork’, ‘Respect for People’, ‘Lean Actions’ and the influence of negatively motivating factors all reached significance. Defined contextual factors for ‘Previous Experience’ and the influence of positive motivating factors did not reach significance at 5% level. The final model showed a statistically significant, moderate predictive strength, with an overall adjusted R2 of 0.58. This result was an encouraging validation of the microsystem-level layer of the MUSIQ model using Lean as the QI method (context). The relative influence of ‘Teamwork’, ‘Respect for People’, ‘Motivation’, and a mediating mechanism for making process changes (in this instance, Lean) were measured and found to be consistent with the MUSIQ model. Identifying more detailed causal mechanisms (the present model was intentionally parsimonious due to the time frame allowed and the resources available for the research), refining the operational definitions, and developing and testing predictive models for the defined contextual factors are the proposed next steps in the research.