P-108 - ARE MAJOR LOWER EXTREMITY AMPUTATIONS WELL RECORDED IN PRIMARY CARE - INSIGHTS FROM ELECTRONIC HEALTH DATA IN ENGLAND

TOPIC:
Other
AUTHORS:
Meffen A. (Department of Health Sciences, University of Leicester ~ Leicester ~ United Kingdom) , Sayers R. (Department of Cardiovascular Sciences, University of Leicester ~ Leicester ~ United Kingdom) , Gillies C. (Diabetes Research Centre, University of Leicester ~ Leicester ~ United Kingdom) , Khunti K. (Diabetes Research Centre, University of Leicester ~ Leicester ~ United Kingdom) , Gray L. (Department of Health Sciences, University of Leicester ~ Leicester ~ United Kingdom)
Introduction:
Major lower extremity amputations (MLEA) are expected to be well recorded in secondary care (hospital) databases in England as hospitals receive payment for procedures performed based on events recorded in these databases. Primary care databases in England work differently with information on secondary care based events being manually entered into the database on the receipt of a letter informing them of an event from a hospital consultant. Methods of recording vary between primary care practices. A number of studies have ascertained cases of MLEA in secondary care data, however, very few have ascertained cases in primary care data therefore it is unclear how well recorded MLEA are in primary care data[1, 2]. This study aims to compare the recording of MLEA events between the secondary care database Hospital Episode Statistics (HES) and the primary care database Clinical Practice Research Datalink (CPRD) in England[3, 4].
Methods:
MLEA events in England were ascertained in CPRD and in the CPRD available HES linkage between 01/01/2009 and 31/12/2019. Full CPRD patient history and all available linked HES data were obtained for these patients. The number of MLEA events and the number of patients with at least one MLEA in each database was recorded and compared. Individual events were matched between the two databases using varying event date matching windows. The numbers of matched events from each database were recorded and compared. Additionally, the level of MLEA (below/at/above knee etc.) was compared for those events that matched.
Results:
A total of 23,312 people were found to have at least one MLEA event recorded within the study period. 39.3% (n = 9,153) of which had an MLEA record in both the primary and secondary care databases (Figure 1). 37.4% (n = 8,716) only had MLEA recording in HES and 23.4% (n = 5,443) only had MLEA recording in CPRD. When matching individual events between the two databases using patient ID and exact event date, only 7.1% (n = 3,840) of events in CPRD found a matching event in HES and only 17.7% (n = 3,751) of events in HES found a matching event in CPRD (Figure 2). The databases were in agreement in MLEA level for 89.9% (n = 3,371) of events matched on exact date. As the date matching window increased to a matching window that ranges between the hospital admission date and 28 days post-event, the match percentage also increased to 18.6% (n = 10,125) of CPRD events finding a match in HES and 37.6% (n = 7,874) of events in HES finding a match in CPRD. However, the percentage of agreement in MLEA level between the databases then decreased to 86.7% (n = 6,880).
Conclusion:
MLEAs are not well recorded in primary care records. Reasons for MLEA not being recorded are unlikely due to patient demographic and are predominantly due to a number of factors including recording type (coded/free text/scanned), amputations performed abroad and re-recordings of the same event multiple times including events prior to the study date. It is also possible that some events in HES are recorded using separate location and procedure codes instead of the MLEA specific codes. Only the MLEA specific codes were used in this study. Future research could benefit from the added patient data provided by the primary care databases, however, it would not be recommended to ascertain cases of MLEA solely using primary care data. Ideally, recordings of MLEA cases should include triangulation of data from primary and secondary care databases. Accuracy of recordings in primary care databases could be improved if methods and standards of electronic health recordings were homogenised between primary care practices.
References:
1. Gunn LH, Vamos EP, Majeed A, Normahani P, Jaffer U, Molina G, et al. Associations between attainment of incentivized primary care indicators and incident lower limb amputation among those with type 2 diabetes: a population-based historical cohort study. BMJ open diabetes research & care. 2021;9(1). 2. Meffen A, Houghton JSM, Nickinson ATO, Pepper CJ, Sayers RD, Gray LJ. Understanding variations in reported epidemiology of major lower extremity amputation in the UK: a systematic review. BMJ Open. 2021;11(10):e053599. 3. Wolf A, Dedman D, Campbell J, Booth H, Lunn D, Chapman J, et al. Data resource profile: Clinical Practice Research Datalink (CPRD) Aurum. International Journal of Epidemiology. 2019;48(6):1740-g. 4. NHS-Digital. Hospital Episode Statistics (HES). 2019. [updated 20/03/2019]. Available from: https://digital.nhs.uk/data-and-information/data-tools-and-services/data-services/hospital-episode-statistics.
ATTACHMENTS: