Caring for the patient, caring for the record: An ethnographic study of 'back office' work in upholding quality of care in general practice

© 2015 Swinglehurst and Greenhalgh; licensee BioMed Central. Background: The quality of information recorded about patient care is considered key to improving the overall quality, safety and efficiency of patient care. Assigning codes to patients' records is an important aspect of this documentation. Current interest in large datasets in which individual patient data are collated (e.g. proposed NHS care.data project) pays little attention to the details of how 'data' get onto the record. This paper explores the work of summarising and coding records, focusing on 'back office' practices, identifying contributors and barriers to quality of care. Methods: Ethnographic observation (187 hours) of clinical, management and administrative staff in two UK general practices with contrasting organisational characteristics. This involved observation of working practices, including shadowing, recording detailed field notes, naturalistic interviews and analysis of key documents relating to summarising and coding. Ethnographic analysis drew on key sensitizing concepts to build a 'thick description' of coding practices, drawing these together in a narrative synthesis. Results: Coding and summarising electronic patient records is complex work. It depends crucially on nuanced judgements made by administrators who combine their understanding of: clinical diagnostics; classification systems; how healthcare is organised; particular working practices of individual colleagues; current health policy. Working with imperfect classification systems, diagnostic uncertainty and a range of local practical constraints, they manage a moral tension between their idealised aspiration of a 'gold standard' record and a pragmatic recognition that this is rarely achievable in practice. Adopting a range of practical workarounds, administrators position themselves as both formally accountable to their employers (general practitioners), and informally accountability to individual patients, in a coding process which is shaped not only by the 'facts' of the case, but by ongoing working relationships which are co-constructed alongside the patient's summary. Conclusion: Data coding is usually conceptualised as either a technical task, or as mundane, routine work, and usually remains invisible. This study offers a characterisation of coding as a socially complex site of moral work through which new lines of accountability are enacted in the workplace, and casts new light on the meaning of coded data as conceptualised in the 'quality of care' discourse.