Search results

ABSTRACTObjectivesThere have been a number of key changes in the clinical definition and diagnostic threshold of acute coronary syndromes in the last 10 years. We have characterised temporal and geographic changes in the incidence and outcomes following Acute Coronary Syndrome (ACS: Unstable Angina (UA), Non ST-segment elevation myocardial infarction (NSTEMI) and ST-segment elevation myocardial infarction (STEMI)) between 2009 and 2013.
Approach65,137 hospitals stays were identified involving ACS (ICD-10: I20.0, I21 and I22) relating to 55,369 individuals identified through secondary care primary diagnosis records during 2009-2013. All prior and subsequent secondary care diagnoses from 1981-2014 were sourced for these patients and records were deterministically matched on a pseudo patient identifier to obtain the cause and date of death for purposes of follow-up. An incident ACS case was defined as such if the patient had not suffered an ACS in the five years prior to the hospital admission and all co-morbidities were derived from hospital diagnostic codes accompanying the ACS codes.
ResultsFor the entire cohort, patients with an incident ACS were predominantly male (61.5%) with mean age 68 (SD=13.7 years). Co-morbidities included: 65.5% Other Ischaemic Heart Disease; 5.2% Stroke; 7.5% Peripheral Artery Disease; 14.8% Atrial Fibrillation; 42.0% Hypertension; 18.0% Diabetes Mellitus and 8.4% Chronic Kidney Disease.
The overall incidence of ACS in 2009 was 204/100,000 and fell by 8.1% to 188/100,000 in 2013. Subtypes of ACS comprised 9.4% UA, 50.9% NSTEMI, 29.0% STEMI and 10.8% MI unspecified in 2013.
In-hospital mortality following an incident ACS was 9.7% (95% CI: 9.2-10.3%) in 2009 and varied from 7.9 to 19.0% across the NHS boards. In 2013, in-hospital mortality was 8.5% (95% CI: 7.9-9.0%) ranging from 4.5 to 10.5% across the NHS boards. One-year mortality following an incident ACS in 2009 was 18.6% (95% CI: 17.9-19.4%) falling to 16.8% (95% CI: 16.1-17.5%) in 2013. Stratified by NHS board, the one-year mortality rate in 2009 varied from 16.9 to 28.0% and in 2013 ranged from 11.9 to 20.0% across the NHS boards.
ConclusionThese findings highlight the importance of a cohort based record linkage approach to routine healthcare datasets. While there appears to be changes in incidence of ACS and its subtypes and changes in mortality over time, these findings reflect significant changes in clinical practice with respect to definition and diagnosis. Cautious interpretation is needed combined with further research to fully understand the epidemiological implications of our findings.

ABSTRACT
ObjectivesThe use of "real-time" data to support individual patient management and outcome assessment requires the development of risk assessment models. This could be delivered through a learning health system by the building robust statistical analysis tools onto the existing linked data held by NHS Scotland's Infection Intelligence Platform (IIP) and developed within the Scottish Healthcare Associated Infection Prevention Institute (SHAIPI).
This project will create prediction models for the risk of acquiring a healthcare associated infection (HAI), and particular outcomes, at the point of GP consultation/ hospital admission which could aid clinical decision making.
ApproachWe demonstrate the capability using the HAI Clostridium difficile (CDI) from 2010-2013. Using linked national individual level data on community prescribing, hospitalisations, infections and death records we extracted all cases of CDI and by comparing to matched population-based controls, examined the impact of prior hospital admissions, care home residence, comorbidities, exposure to gastric acid suppressive drugs and antibiotic exposure, defined as both cumulative (total defined daily dose (DDD)) and temporal antimicrobial exposure in the previous 6 months, to the risk of CDI acquisition. Antimicrobial exposure was considered for all drugs and the higher risk broad spectrum antibiotics (4Cs). Associations are assessed using conditional logistic regression. Using cross-validation we assess the ability of the model to accurately predict CDI infection. Risk scores for acquisition of CDI are estimated by combining these predictions with age and gender population incidence.
ResultsIn the period 2010-2013 there were 1446 cases of CDI with matched 7964 controls. A significant dose-response relationship for exposure to any antimicrobial (1-7 DDDs OR=2.3 rising to OR=4.4 for 29+ DDDs) and, with elevated risk, to the 4C group (1-7 DDDs OR=3.8 rising to OR=17.9 for 29+ DDDs). Exposure elevates CDI risk most in the month after prescription but for 4C antimicrobials the elevated risk remains 6 months later (4C OR=12.4 within 1 month, OR=2.6 4-6 months later). The risk of CDI was also increased with more co-morbidities, previous hospitalisations, care home residency, increased number of prescriptions, and gastric acid suppression.
ConclusionDespite limitations to current application in practice,(paucity of patient level in-hospital prescribing data and constraints of the timeliness of the data), when fully developed this system will enable risk classification to identify patients most at risk of HAI and adverse outcomes to aid clinical decision making.

ABSTRACT
ObjectivesUrinary tract infections (UTIs) are amongst the most common infections treated in community and hospital settings. Initial antibiotic treatment of UTI is usually empirical, that is, where the prescriber has no definitive information on the organism or its antibiotic sensitivity.. Overall the prevalence of antimicrobial resistance is increasing, and specifically so for antibiotics commonly used for UTI.
By linking NHS surveillance data to routinely collected administrative health data this study aims to investigate risk factors for antibiotic resistance in urine samples
ApproachAll positive urine samples included in the "Surveillance of Antimicrobial Resistance in Urinary Isolates in Scotland" dataset in the period from January 2012 to June 2015 (all NHS Health Boards in Scotland submit susceptibility data for up to eleven antibiotics on 400 positive urinary samples per quarter) were analysed. Cases were assigned a resistance status of Sensitive, Resistant or Multi-drug resistant based on the antibiotic susceptibility data recorded.
Using the NHS Scotland Infection Intelligence Platform all cases were linked to national coverage data: (i) hospital discharge data to create the Charlson score for comorbidity and (ii) patient-level community prescribing data to measure cumulative antibiotic exposure (number of defined daily doses) in the 3 months prior to infection. Risk factors associated with the infection susceptibility to antibiotics were assessed using multivariable multinomial logistic regression.
Results40,984 positive urine samples were examined. Overall 29.0% were sensitive, 48.1% resistant and 22.9% multi drug resistant. Around a third of the cases (33.9%) had no antibiotic prescribing in the 3 months prior to infection. Age, care home residence and increasing comorbidity were both found to be associated with resistance and multidrug resistance. Cumulative antibiotic exposure had a clear dose-response effect. Those with 1-7DDDs were 1.2 times (95% CI: 1.11-1.29) more likely to have a multidrug resistant infection (compared to a sensitive infection) rising to 7.45 times (95% CI: 95% CI 6.45-8.6) for 29+ DDDs. Similar dose response held for resistant infection but at a lesser scale (1-7DDDs OR=1.36 (95% CI: 1.2-1.5) rising to OR=3.04 (95% CI: 2.38-3.89).
ConclusionA clear effect of cumulative antibiotic exposure in the community and multidrug resistance in UTI cases has been demonstrated. Such quantification is key to ensuring and supporting robust antimicrobial stewardship policy and will form the evidence base for development of prescribing decision support tools for more patient centred treatment of UTI.

ABSTRACTObjectivesPrior use of antimicrobials, in particular broad-spectrum antimicrobials, is associated with the development of Clostridium difficile infection (CDI). Our previous work has demonstrated increased risk with cumulative exposure but there is limited evidence on specific patterns of cumulative antimicrobial prescribing prior to infection. Understanding this pattern will help inform antimicrobial stewardship. This study aims to investigate the prescribing patterns for CDI cases with more than 4 weeks cumulative antimicrobial exposure during the 6 months prior to their CDI date.ApproachWe linked three Scottish patient-level data sets: laboratory confirmed CDI (ECOSS), prescriptions for antimicrobials in primary care (PIS) and all hospital admissions (SMR01). From ECOSS all cases of CDI reported in the period August 2010 to July 2013 were identified. Each CDI case was linked to SMR01 to allocate case type (either hospital or community associated) and to PIS for previous antimicrobial prescriptions. Visual representation of temporal exposure indicating both duration and drug type of each antimicrobial dispensed during the 6 month period before the infection was produced to clearly understand the pattern of prescribing.ResultsIn the study period, there were a total of 1557 community acquired CDI cases without recent (prior 3 months) hospitalisation. Among them, 287 (18%) cases had more than 4 weeks prior cumulative antimicrobial exposure accumulating a total of 1311 dispensed prescriptions. Cases had an average 4 (Q1:3,Q3:6) prescriptions and 2 (Q1:2,Q3:3) different types of antimicrobials dispensed. The timeline plot shows that repeated short duration prescriptions contributed more to cumulative exposure than long duration prophylaxis. The most common antimicrobials prescribed were amoxicillin (38% of cases), trimethoprim (30%) and flucloxacillin (27%). The median duration per prescription for each of these was less than 2 weeks. The antimicrobials that had a median duration of around a month per prescription were oxytetracycline (3%) and clindamycin (6%).
ConclusionsThis study of a national linked patient level data set contained sufficient numbers of cases to enable investigation of the prescribing pattern in individuals with highest risk cumulative exposure. This study uses NHS Scotland's developing Infection Intelligence Platform which will place Scotland as a world leader in the use of health informatics to support infection control and antimicrobial stewardship.

ABSTRACTObjectivesBiologics are a class of medicines recommended by UK and international guidelines for treatment of patients with rheumatoid arthritis (RA) who are unresponsive, intolerant or who have contraindications to conventional therapies. Biologics are often supplied in the UK to patients in their home by external healthcare providers commissioned by the National Health Service (NHS). Patient adherence and persistence to biologics has not been assessed within NHS Scotland.The aim of this study was to evaluate adherence and persistence patterns of subcutaneous biologics delivered by homecare services to patients with rheumatoid arthritis in a Scottish NHS Board.
ApproachA population-based cohort study was conducted using records from patients receiving medication through healthcare delivery companies in the NHS Greater Glasgow and Clyde health board. RA patients with deliveries between 2012- 2015 were linked via their Community Health Index number to their prescriptions in primary care, hospital admissions and death records. For this study those with at least two deliveries of biologics and who did not switch biologic treatment were included. Length of therapy was determined from first delivery to last day treatment was covered. Adherence rate referring to the supply received for the whole study period was evaluated through Medication Refill Adherence (MRA). Persistence from first delivery to discontinuation was estimated using survival analysis and a drug–free interval of 90 days.
ResultsA total of 434 patients with RA received a subcutaneous biologic. The mean age at first delivery was 54.7 (SD 13.4) years and 78% were female. From this cohort, 322 had at least 2 deliveries of the same biologic. The overall median length of treatment was 387.5 days (IQR 544.5); the longest for certolizumab pegol users with a median of 511.5 days (IQR 629.2) and the shortest for tocilizumab (median of 187 days, IQR 90.5). The adherence throughout the study had a mean MRA 79.73% (SD 29.95) and was similar across the different biologics. Persistence at follow-up decreased with time from 81% of the patients receiving a biologic at 6 months, to 67% after 1 year and 65% after two years. Adalimumab, certolizumab pegol, etanercept and golimumab users had higher persistence rates.
ConclusionLength of treatment and persistence, but not adherence, are influenced by the type of subcutaneous biologic received by the RA patients in this cohort.Future research will gain a better insight into the factors influencing adherence and persistence to these medicines.