Taking the heat or taking the temperature? A qualitative study of a large-scale exercise in seeking to measure for improvement, not blame

Natalie Armstrong, Liz Brewster, Carolyn Tarrant, Ruth Dixon, Janet Willars, Maxine Power, Mary Dixon-Woods

Measurement of quality and safety has an important role in improving healthcare, but is susceptible to unintended consequences. One frequently made argument is that optimising the benefits from measurement requires controlling the risks of blame, but whether it is possible to do this remains unclear. We examined responses to a programme known as the NHS Safety Thermometer (NHS-ST). Measuring four common patient harms in diverse care settings with the goal of supporting local improvement, the programme explicitly eschews a role for blame. The study design was ethnographic. We conducted 115 hours of observation across 19 care organisations and conducted 126 interviews with frontline staff, senior national leaders, experts in the four harms, and the NHS-ST programme leadership and development team. We also collected and analysed relevant documents. The programme theory of the NHS-ST was based in a logic of measurement for improvement: the designers of the programme sought to avoid the appropriation of the data for any purpose other than supporting improvement. However, organisational participants - both at frontline and senior levels - were concerned that the NHS-ST functioned latently as a blame allocation device. These perceptions were influenced, first, by field-level logics of accountability and managerialism and, second, by specific features of the programme, including public reporting, financial incentives, and ambiguities about definitions that amplified the concerns. In consequence, organisational participants, while they identified some merits of the programme, tended to identify and categorise it as another example of performance management, rich in potential for blame. These findings indicate that the search to optimise the benefits of measurement by controlling the risks of blame remains challenging. They further suggest that a well-intentioned programme theory, while necessary, may not be sufficient for achieving goals for improvement in healthcare systems dominated by institutional logics that run counter to the programme theory.