Incidence, Microbiology, and Outcomes in Patients Hospitalized With Infective Endocarditis

Supplemental Digital Content is available in the text.


Supplementary figure II: Sensitivity analysis restricting to diagnostic coding in position one (primary diagnosis). Estimated incidence rate per
8 Odds ratio (95% CI)

Scottish hospitalization record
Scottish hospitalizations from infective endocarditis were defined from the Scottish morbidity record 01 (SMR01) -General/Acute Inpatient & Day Case. SMR01 is an episode-based patient record relating to all inpatients and day cases discharged from non-obstetric and nonpsychiatric specialties. A record is generated when a patient completes an episode of inpatient or day case care. Data collected include patient identifiable and demographic details, episode management details and general clinical information. Currently diagnoses are recorded using the ICD-10 classification and operations are recorded using the OPCS-4 classification. Further information on the national dataset and variables contained is available at https://www.ndc.scot.nhs.uk/Data-Dictionary/SMR-Datasets//Episode-Management/SMR-Record-Type/ Past history, operation codes and demographics were also derived from the Scottish Morbidity Record 01 and linked to incident cases of infective endocarditis. Subsequent hospitalizations based on ICD codes are further linked to incident cases of infective endocarditis identified.

National Records of Scotland (NRS)
The NRS covers all deaths in Scotland with approximately 55,000 deaths registered annually.

Prescribing Information System (PIS)
The Prescribing Information System (PIS) is the definitive data source for all prescribing relating to all medicines and their costs that are prescribed and dispensed in the community in Scotland. The information is supplied by Practitioner & Counter Fraud Services Division (P&CFS) who is responsible for the processing and pricing of all prescriptions dispensed in Scotland. Primary care physicians write the vast majority of these prescriptions, with the remainder written by other authorised prescribers such as nurses and dentists. Also included in the dataset are prescriptions written in hospitals that are dispensed in the community. Note that prescriptions dispensed within hospitals are not included.
Both the diabetes and chronic lung disease status in our cohort were based on community prescribing data as per Supplementary table 5. Incident cases of endocarditis prescribed either anti-diabetic drugs or drugs for chronic lung disease with one year prior to hospitalization were defined as having the condition. Complete prescribing data was available from 2009. Further information on the Prescribing Information System operational in Scotland is available at https://www.ndc.scot.nhs.uk/National-Datasets/data.asp?SubID=9

National microbiology register (Electronic Communication of Surveillance in Scotland [ECOSS])
The Scottish microbiology surveillance registry, or 'Electronic Communication of Surveillance in Scotland' (ECOSS) as it is termed by NHS National Services Scotland, was used in the present study to provide individual patient-level data on positive blood culture results (from diagnostic microbiology laboratories within NHS Scotland health boards and national reference laboratories) related to incident cases of infective endocarditis identified from SMR01 between 2008 and 2014. ECOSS is part of NHS Scotland's Infection Intelligence Platform (IIP),46,47 which was set-up in response to the UK's antimicrobial resistance (AMR) strategy (2013-2018) with the aim of providing "better access to and use of surveillance data".48 Data were first collected and recorded within ECOSS in 2007. The dataset is maintained by NHS National Services Scotland on behalf of Health Protection Scotland. ECOSS is updated monthly and, as of 2017, it contained approximately 29 million records of positive microbiology laboratory specimens from across Scotland.1 It provides data for numerous national clinical and research activities, audit projects and Scottish Government reports, including: the identification of cases of severe infectious disease, infectious disease outbreaks and the evaluation of longer term trends in the incidence of laboratory-reported infections; surveillance of episodes of Clostridium difficile infections, Escherichia coli bacteremia, Staphylococcus aureus bacteremia and surgical site infections.2 NHS National Services Scotland monitors the completeness and accuracy of ECOSS data through its 'Data Monitoring and Support Service'.1 Further, NHS National Services Scotland routinely informs data users of any problems affecting the accuracy or assurance of these data.
In the present study, causative organisms were defined as those identified within 90 days on either side of the index admission date. Using this timeframe, 950 cases of infective endocarditis were associated with positive microbiological results.
Polymicrobial status was defined specifically is more than one causative organism was identified on the same culture date. If more than one causative organism was identified on differing dates, then the organism identified closest to the index admission date was assigned as the causative organism.
Near complete blood culture microbiology data were available from 2008. Three small Scottish laboratories did not provide complete data. These laboratories were as follows: Overall these laboratories served <1.5% of the Scottish Population and would therefore have a negligible effect on the rates of non-positive blood cultures observed.

Approvals for use of data
Access to the data was approved by the NHS Scotland Public Benefit and Privacy Panel and in accordance with the Declaration of Helsinki. As the analysis used routinely collected and anonymized data, individual patient consent was not sought.

Supplementary text IV: Description of 5 year lookback period and calculation of persontime Lookback
The schematic below demonstrates an example of how the 5-year lookback period was employed in the period between 2000 and 2015 (inclusive) to identify incident events of infective endocarditis in three exemplar patients (patients A, B and C). The total incident count for each year is shown in the final column. Where a patient has been admitted with an episode of infective endocarditis, a '1' appears in the 'Admission' column. If no infective endocarditis event has occurred in the 5-years prior (i.e. the 'lookback' period, as indicated by the light grey shading), then the event is considered an incident event and a '1' will also appear in the 'Incident' column.

Person-time calculation
The schematic below demonstrates an example across three patients on how the person-time was calculated. 20

Chapter and section in the British National Formulary
Chronic respiratory disease Respiratory system: all drugs Chapter 3 Diabetes mellitus Endocrine system: drugs used in diabetes Chapter 6, section 6.1

Supplementary text VI: Details of interrupted time series analysis
To evaluate any change in the incidence of infective endocarditis before and after introduction of guidelines on antibiotic prophylaxis published by the National Institute of Health and Care Excellence, an interrupted time series analysis model was created.
We chose a step-change model. A Poisson model was used as we were predicting count data. We modelled the count data directly (rather than the rate which doesn't follow a Poisson distribution), using the person-time (log transformed) as an offset variable in order to transform back to rates. In order to account for overdispersion we used a quasipoisson model (allowing the variance to be proportional rather than equal to the mean).
The following model was created with corresponding output:

Supplementary table I: Baseline characteristics and short-and long-term outcomes stratified by single calendar years from 2008 to 2014
Single