Understanding challenges of using routinely collected health data to address clinical care gaps: a case study in Alberta, Canada

High-quality data are fundamental to healthcare research, future applications of artificial intelligence and advancing healthcare delivery and outcomes through a learning health system. Although routinely collected administrative health and electronic medical record data are rich sources of information, they have significant limitations. Through four example projects from the Physician Learning Program in Edmonton, Alberta, Canada, we illustrate barriers to using routinely collected health data to conduct research and engage in clinical quality improvement. These include challenges with data availability for variables of clinical interest, data completeness within a clinical visit, missing and duplicate visits, and variability of data capture systems. We make four recommendations that highlight the need for increased clinical engagement to improve the collection and coding of routinely collected data. Advancing the quality and usability of health systems data will support the continuous quality improvement needed to achieve the quintuple aim.

Antibiotic Prescribing Trends in Belgian Out-of-Hours Primary Care during the COVID-19 Pandemic: Observational Study Using Routinely Collected Health Data

Antibiotic overprescribing is one of the main drivers of the global and growing problem of antibiotic resistance, especially in primary care and for respiratory tract infections (RTIs). RTIs are the most common reason for patients to consult out-of-hours (OOH) primary care. The COVID-19 pandemic has changed the way general practitioners (GPs) work, both during office hours and OOH. In Belgian OOH primary care, remote consultations with the possibility of issuing prescriptions and telephone triage were implemented. We aimed to describe the impact of COVID-19 on GPs’ antibiotic prescribing during OOH primary care. In an observational study, using routinely collected health data from GP cooperatives (GPCs) in Flanders, we analyzed GPs’ antibiotic prescriptions in 2019 (10 GPCs) and 2020 (20 GPCs) during OOH consultations (telephone and face-to-face). We used autoregressive integrated moving average (ARIMA) modeling to identify any changes after lockdowns were implemented. In total, 388,293 contacts and 268,430 prescriptions were analyzed in detail. The number of antibiotic prescriptions per weekend, per 100,000 population was 11.47 (95% CI: 9.08–13.87) or 42.9% lower after compared to before the implementation of lockdown among all contacts. For antibiotic prescribing per contact, we found a decrease of 12.2 percentage points (95% CI: 10.6–13.7) or 56.5% among all contacts and of 5.3 percentage points (95% CI: 3.7–6.9) or 23.2% for face-to-face contacts only. The decrease in the number of prescriptions was more pronounced for cases with respiratory symptoms that corresponded with symptoms of COVID-19 and for antibiotics that are frequently prescribed for RTIs, such as amoxicillin (a decrease of 64.9%) and amoxicillin/clavulanate (a decrease of 38.1%) but did not appear for others such as nitrofurantoin. The implementation of COVID-19 lockdown measures coincided with an unprecedented drop in the number of antibiotic prescriptions, which can be explained by a decrease in face-to-face patient contacts, as well as a lower number of antibiotics prescriptions per face-to-face patient contact. The decrease was seen for antibiotics used for RTIs but not for nitrofurantoin, the first-choice antibiotic for urinary tract infections.

Waving away waivers: an obligation to contribute to ‘herd knowledge’ for data linkage research?

In today’s online data-driven world, people constantly shed data and deposit digital footprints. When individuals access health services, governments and health providers collect and store large volumes of health information about people that can later be retrieved, linked and analysed for research purposes. This can lead to new discoveries in medicine and healthcare. In addition, when securely stored and de-identified, the privacy risks are minimal and manageable. In many jurisdictions, ethics committees routinely waive the requirement for researchers to obtain consent from data subjects before using and linking these datasets in an effort to balance respect for individuals with research efficiency. In this paper, I explore the ethical justification for using routinely collected health data for research without consent. I conclude that, not only is this morally justified but also that data subjects have a moral obligation to contribute their data to such research, which would obviate the need for ethics committees to consider consent waivers. In justifying this argument, I look to the duty of easy rescue, distributive justice and draw analogies with vaccination ethics.

Use of routinely collected health data in randomised clinical trials: comparison of trial-specific death data in the BOSS trial with NHS Digital data

Background A promising approach to reduce the increasing costs of clinical trials is the use of routinely collected health data as participant data. However, the quality of this data could limit its usability as trial participant data. Methods The BOSS trial is a randomised controlled trial comparing regular endoscopies versus endoscopies at need in patients with Barrett’s oesophagus with primary endpoint death. Data on death and cancer collected every 2 years after randomisation (trial-specific data) were compared to data received annually (all patients on one date) from the routinely collected health data source National Health Service (NHS) Digital. We investigated completeness, agreement and timeliness and looked at the implications for the primary trial outcome. Completeness and agreement were assessed by evaluating the number of reported and missing cases and any disparities between reported dates. Timeliness was considered by graphing the year a death was first reported in the trial-specific data against that for NHS Digital data. Implications on the primary trial outcome, overall survival, of using one of the data sources alone were investigated using Kaplan-Meier graphs. To assess the utility of cause of death and cancer diagnoses, oesophageal cancer cases were compared. Results NHS Digital datasets included more deaths and often reported them sooner than the trial-specific data. The number reported as being from oesophageal cancer was similar in both datasets. Due to time lag in reporting and missing cases, the event rate appeared higher using the NHS Digital data. Conclusion NHS Digital death data is useful for calculating overall survival where trial-specific follow-up is only every 2 years from randomisation and the follow-up requires patient response. The cancer data was not a large enough sample to assess usability. We suggest that this assessment of registry data is done for more phase III RCTs and for more registry data to get a more complete picture of when RCHD would be useful in phase III RCT. Trial registration ISRCTN54190466 (BOSS) 1 Oct 2009.

Patterns of care for people presenting to Australian general practice with musculoskeletal complaints based on routinely collected data: protocol for an observational cohort study using the Population Level Analysis and Reporting (POLAR) database

IntroductionGeneral practice is integral to the Australian healthcare system. Outcome Health’s POpulation Level Analysis and Reporting (POLAR) database uses de-identified electronic health records to analyse general practice data in Australia. Previous studies using routinely collected health data for research have not consistently reported the codes and algorithms used to describe the population, exposures, interventions and outcomes in sufficient detail to allow replication. This paper reports a study protocol investigating patterns of care for people presenting with musculoskeletal conditions to general practice in Victoria, Australia. Its focus is on the systematic approach used to classify and select eligible records from the POLAR database to facilitate replication. This will be useful for other researchers using routinely collected health data for research.Methods and analysisThis is a retrospective cohort study. Patient-related data will be obtained through electronic health records from a subset of general practices across three primary health networks (PHN) in southeastern Victoria. Data for patients with a low back, neck, shoulder and/or knee condition and who received at least one general practitioner (GP) face-to-face consultation between 1 January 2014 and 31 December 2018 will be included. Data quality checks will be conducted to exclude patients with poor data recording and/or non-continuous follow-up. Relational data files with eligible and valid records will be merged to select the study cohort and the GP care received (consultations, imaging requests, prescriptions and referrals) between diagnosis and 31 December 2018. Number and characteristics of patients and GPs, and number, type and timing of imaging requests, prescriptions for pain relief and referrals to other health providers will be investigated.Ethics and disseminationEthics approval was obtained from the Cabrini and Monash University Human Research Ethics Committees (Reference Numbers 02-21-01-19 and 16975, respectively). Study findings will be reported to Outcome Health, participating PHNs, disseminated in academic journals and presented in conferences.

437Comparison of stillbirth trends in Wales and Western Australia using pooled routinely collected health data

Background As there are variations in stillbirth rates and trends, even among high income countries, international comparisons can provide insights into how reductions in stillbirths can be achieved. We compared stillbirth rates and trends over time in Wales and Western Australia (WA). Methods We pooled population-based data of all births of at least 24 weeks’ gestation occurring between 1993-2015 in Wales and WA, divided into 6 time-periods. The stillbirth rate per 1,000 births was estimated for each cohort in each time-period. Multivariable Poisson regression analyses, were performed to evaluate the interaction between cohort and time-period. relative risk (RRs) and 95% Confidence Intervals (CIs) for each time-period and cohort were calculated. Results The overall stillbirth rate declined by 15.9% in Wales and 40.4% in WA. Using WA and 1993-1996 as the reference group, the adjusted RRs for stillbirths at 39-41 weeks’ gestation in the most recent study period (2013-15) were 0.85 (95% CI 0.64 to 1.13) in Wales and 0.51 (95% CI 0.36 to 0.73) in WA. Conclusions The stillbirth rate disparities between Wales and WA have widened in the last two decades (especially among term births). Some of these differences may be partially explained by maternal lifestyle behaviours such as smoking, but we had insufficient population-level data to investigate their contribution. Key messages The stillbirth rate was persistently higher in Wales than WA from 1993 to 2015, with widening disparities after adjustment for important risk factors.

Why are feasibility studies accessing routinely collected health data? A systematic review

Background: Feasibility trials are often undertaken to determine whether a larger randomised controlled trial (RCT) is achievable. In a recent review, 15 feasibility trials accessed routinely collected health data (RCHD) from UK national databases and registries. This paper looks at attributes of these trials and the reasons why they accessed RCHD. Methods: We extracted data from all publicly available sources for the 15 feasibility studies found in a previous review of trials successfully accessing RCHD in the UK between 2013–2018 for the purpose of informing or supplementing participant data. We extracted trial characteristics, the registry accessed, and the way the RCHD was used. Results: The 15 feasibility RCTs were conducted in a variety of disease areas, and were generally small (median sample size 100, range 41–4061) and individually randomised (60%, 9/15). The primary trial outcome was predominantly administrative (non-clinical) (80%, 12/15) such as feasibility of patient recruitment. They were more likely to recruit from secondary care (67%, 10/15) settings than primary (33%, 5/15). NHS Digital was the most commonly accessed registry (33% (5/15)) with SAIL databank (20% (3/15)), electronic Data Research and Innovation Service (eDRIS) and Paediatric Intensive Care Audit Network (PICANET) (each 13% 2/15) also being accessed. Where the information was clear, the trials used RCHD for data collection during the trial (47%, 7/15), follow-up after the trial (27%, 4/15) and recruitment (13%, 2/15). Conclusions: Between 2013 and 2018, 15 feasibility trials successfully accessed UK RCHD. Feasibility trials would benefit, as with other trials, from guidance on reporting the use of RCHD in protocols and publications.

Days alive and at home after hip fracture: a cross-sectional validation of a patient-centred outcome measure using routinely collected data

BackgroundDays alive and at home (DAH) is a patient centered outcome measureable in routinely collected health data. The validity and minimally important difference (MID) in hip fracture have not been evaluated.ObjectiveWe assessed construct and predictive validity and estimated a MID for the patient-centred outcome of DAH after hip fracture admission.MethodsThis is a cross-sectional observational study using linked health administrative data in Ontario, Canada. DAH was calculated as the number of days alive within 90 days of admission minus the number of days hospitalised or institutionalised. All hospital admissions (2012–2018) for hip fracture in adults aged >50 years were included. Construct validity analyses used Bayesian quantile regression to estimate the associations of postulated patient, admission and process-related variables with DAH. The predictive validity assessed was the correlation of DAH in 90 days with the value from 91 to 365 days; and the association and discrimination of DAH in 90 days predicting subsequent mortality. MID was estimated by averaging distribution-based and clinical anchor-based estimates.ResultsWe identified 63 778 patients with hip fracture. The median number of DAH was 43 (range 0–87). In the 90 days after admission, 8050 (12.6%) people died; a further 6366 (10.0%) died from days 91 to 365. Associations between patient-level and admission-level factors with the median DAH (lower with greater age, frailty and comorbidity, lower if admitted to intensive care or having had a complication) supported construct validity. DAH in 90 days after admission was strongly correlated with DAH in 365 days after admission (r=0.922). An 11-day MID was estimated.ConclusionDAH has face, construct and predictive validity as a patient-centred outcome in patients with hip fracture, with an estimated MID of 11 days. Future research is required to include direct patient perspectives in confirming MID.