2024/25

Aims: To highlight the limitations and challenges of combining primary and secondary care medicines datasets to understand medicines use across an Integrated Care System (ICS) footprint1

Background: To optimise the use of medicines within an ICS, analysis of multiple datasets is required. In primary care, medicines data is available through ePACT2 as dispensed data (NHS BSA), with data fields that include the number of patients and basic demographics. However, secondary care medicines data (SCMD) (excluding outpatient prescriptions) is only available as a national dataset in the form of stock control medicines data, either through the Secondary Care Medicines Dataset or Define/Refine (Rx-info)2. Stock control datasets do not contain the diversity of data fields available for primary care medicines.

Patients access hospital treatment beyond their ICS of residence, this cross border activity will be captured through contractual reporting, however it is not reflected in secondary care medicines usage data, so it cannot be assumed that this relates to the ICS population for which the secondary care provider is located.

Methodology: The proportions of cross ICS border hospital admissions & attendances, including treatment speciality, have been analysed for the North of England (NoE) ICSs. The preferred dataset for SCMD use would be ‘Electronic Prescribing and Medicines Administration’ data, but this is not available nationally. Data (FY 2021/22) obtained from the Hospital Episode Statistics (HES) data warehouse held by NHS Digital3 was obtained and amalgamated to present the residence of the patient, the place and speciality of hospital admission or attendance. Analysis of the patterns of hospital attendance in relation to the geographical residence of the patient, informed the limitations of combined medicines usage. This was extended to an analysis of carbon foot printing of anaesthetic gases across primary and secondary care in North-East North Cumbria (NENC) ICS.

Results: Of the total number of attendances and admissions in South Yorkshire ICS, 12.3% and 12.5% respectively are resident outside the ICS, the highest proportion across the NoE. In contrast, the NENC ICS has minimal out of area activity; 2.8% of attendances and 3.2% of admissions. Humber and North Yorkshire (HNY) ICS has the greatest disparity between the proportion of attendances and admissions (6.6% and 9.1%). Conclusions: The lower the cross-border activity, the more accurate the SCMD will be e.g. utilising the specialities information for NENC ICS, the percentage of out of area admissions and attendances that were either surgical or obstetrics (where anaesthetic gases are likely to be used), was 1.3% and 1.2% respectively. It is therefore a reasonable assumption that the corresponding stock utilisation in secondary care is nearly entirely due to the NENC population

Download abstract here

Download the poster here

Abstract

Aim: to understand the potential impact of electronic repeat dispensing (eRD) on the volume of medicines utilised, and hospital admissions.

Method: All GP practices in England were analysed to determine the % eRD implementation (items) in both 2017/18 and 2022/23. Two cohorts of GP practices were then defined; <1% eRD (control group) and >50% eRD (intervention group), in 2022/23. The ‘defined daily dose (DDD) per identified patient prescribed a specific drug group’ was then calculated using primary care dispensed medicines data1 for the most common BNF paragraphs prescribed by eRD: lipid-regulating drugs (LRDs), proton pump inhibitors (PPIs), renin-angiotensin system (RAS), beta-adrenoceptor blockers (BB), calcium channel blockers (CCBs), antidiabetic (AD) and antiplatelet (AP) drugs. The statistical significance between the DDDs/patient in the control and intervention groups were analysed at both time points, using a 2 tailed T-test. Hospital admissions for gastro-intestinal (GI) bleeds, stroke and myocardial infarction (MI) were obtained from the Hospital Episode Statistics data warehouse2, for the two time periods for both cohorts. The admissions were weighted by the cohort list size and scaled to admissions per 100,000 population.

Results: There were ~1300 GP practices in the control group (<1% eRD) and ~200 GP practices in the intervention group (>50% eRD). In 2017/18, there was no significant difference in the average volume of drug a patient received (DDD/identified patient) between the control and intervention groups for all the BNF groups analysed, with the exception of PPIs. At the outset, the intervention group had a significantly lower volume of PPI prescribing per patient and higher admissions for GI bleeds (474 versus 434 per 100,000 people). In 2022/23, the weighted volume of prescribing of PPIs in the control group remained unchanged, whereas there was a significant increase in the intervention group to 215 DDD/patient (control group 229 DDD/patient). In 2022/23, the number of hospital admissions for GI bleeds had reduced by 62 and 17 admissions per 100,000 list size, for the intervention and control groups respectively, relative to the 2017/18 baseline. For the cardiovascular drug groups; AP, CCBs and LRDs, there was a significant increase in volume per patient over time in the intervention cohort only. Hospital admissions for stroke were unchanged, however, admissions for MI reduced in both cohorts to 49 and 67 admissions per 100,000 list size for the intervention and control group respectively. The volume of BBs decreased in both groups over time and there was no significant change in RAS drug volume.

Conclusion: The data suggests that in GP practices with higher levels of eRD implementation, the volume of drug per patient dispensed may increase for specific drug groups. This was associated with a reduction in hospital admissions for GI bleeds and MIs. Further analysis using linked patient data is suggested, to confirm the improved outcomes are attributable to the changes in prescribing volume.

Download abstract here