If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Harmonising Individual Patient Level Cardiac Registry Data Across the Asia Pacific Region—A Feasibility Study of In-Hospital Outcomes of STEMI Patients From the Asia Pacific Evaluation of Cardiovascular Therapies (ASPECT) Network
Corresponding author at: Professor Christopher Reid, Curtin University, Kent Street, Bentley, WA 6102, Australia, Curtin University, GPO Box U1987, Perth, WA 6845, Australia
Affiliations
School of Population Health, Curtin University, Perth, WA, AustraliaSchool of Public Health and Preventive Medicine, Monash University, Melbourne, Vic, Australia
School of Public Health and Preventive Medicine, Monash University, Melbourne, Vic, AustraliaDepartment of Cardiology, Alfred Health, Melbourne, Vic, Australia
School of Public Health and Preventive Medicine, Monash University, Melbourne, Vic, AustraliaDepartment of Cardiology, The Royal Melbourne Hospital, Melbourne, Vic, Australia
Division of Cardiology, Department of Medicine and Therapeutics, Faculty of Medicine, The Chinese University of Hong Kong and Prince of Wales Hospital, Hong Kong SAR, China
The Asia-Pacific Evaluation of Cardiovascular Therapies (ASPECT) collaboration was established to inform on percutaneous coronary intervention (PCI) in the Asia-Pacific Region. Our aims were to (i) determine the operational requirements to assemble an international individual patient dataset and validate the processes of governance, data quality and data security, and subsequently (ii) describe the characteristics and outcomes for ST-elevation myocardial infarction (STEMI) patients undergoing PCI in the ASPECT registry.
Methods
Seven (7) ASPECT members were approached to provide a harmonised anonymised dataset from their local registry. Patient characteristics were summarised and associations between the characteristics and in-hospital outcomes for STEMI patients were analysed.
Results
Six (6) participating sites (86%) provided governance approvals for the collation of individual anonymised patient data from 2015 to 2017. Five (5) sites (83%) provided >90% of agreed data elements and 68% of the collated elements had <10% missingness. From the registry (n=12,620), 84% were male. The mean age was 59.2±12.3 years. The Malaysian cohort had a high prevalence of previous myocardial infarction (34%), almost twice that of any other sites (p<0.001). Adverse in-hospital outcomes were the lowest in Hong Kong whilst in-hospital mortality varied from 2.7% in Vietnam to 7.9% in Singapore.
Conclusions
Governance approvals for the collation of individual patient anonymised data was achieved with a high level of data alignment. Secure data transfer process and repository were established. Patient characteristics and presentation varied significantly across the Asia-Pacific region with this likely to be a major predictor of variations in the clinical outcomes observed across the region.
Cardiovascular disease (CVD), particularly coronary artery disease, has been the major cause of death and disability across the globe for the past two decades [
]. In the low- to middle- and high-income countries of the Asia-Pacific region, the CVD epidemic is in full swing as these countries see a rapid transition into modern industrial societies [
]. Not only are they experiencing changes to traditional diet and physical activity patterns that were cardio-protective, but Asia is also seeing an unprecedented change in population demographics with dramatic increases in the number of people entering their sixth, seventh and eighth decades of life in virtually all countries around the region [
]. As part of this rapid development, Asia has also seen major advances in treatments of acute coronary artery disease, and percutaneous coronary intervention (PCI) is now becoming more available across the Asia-Pacific region [
]. First conducted in the region in 1981, PCI was limited to specialist centres until the mid-to-late 1990s but the last 15–20 years has seen an explosive growth in annual numbers of PCI cases across the regions [
]. It was recently estimated that approximately one million PCI cases were performed in Asia in 2016, close to the North America and European experience, and this figure is likely to rapidly expand [
In recent years, we have seen a major recognition in the value of “real-world data” or data collected as part of the routine clinical practice of performing coronary interventions with endeavours to monitor PCI at national and international levels, and efforts in Europe and North America are world leading in this field [
Improved outcomes in patients with ST-elevation myocardial infarction during the last 20 years are related to implementation of evidence-based treatments: experiences from the SWEDEHEART registry 1995-2014.
]. Countries in Asia are rapidly catching up in terms of establishing registries across the region, led initially by efforts in Korea and Japan, which have well established national registries [
Impact of institutional and operator volume on short-term outcomes of percutaneous coronary intervention: a report from the Japanese Nationwide Registry.
The current status of percutaneous coronary intervention in Korea: based on year 2014 cohort of Korean Percutaneous Coronary Intervention (K-PCI) Registry.
Using the LACE index to predict 30-day all-cause unplanned readmission and mortality in acute myocardial infarction patients: insights from the CADOSA registry.
Effect of age on clinical outcomes in elderly patients (>80 years) undergoing percutaneous coronary intervention: insights from a multi-centre australian PCI registry.
]. Some countries, like Malaysia, have developed national registries, and major advances in technology have led to Big Data approaches in countries like Singapore to gather key information on the management of cardiac disease, including PCI [
]. However, the extent of the development of individual registries varies across the region and the Asia Pacific Evaluation of Cardiovascular Therapies (ASPECT) initiative was established and endorsed through the Asia Pacific Society of Cardiology, with the aim to foster collaboration on PCI registry-based activities across the region.
A feasibility project was designed to validate the methods, identify governance and analysis issues for a merged data registry analysis. The focus was to describe the regional characteristics and outcomes for patients presenting with ST-elevation myocardial infarction (STEMI) undergoing PCI. This paper describes the findings from the feasibility project and presents the initial individual patient data analysis from the ASPECT registry.
Materials and Methods
Study Setting
At the 2018 European Society of Cardiology Scientific Sessions representatives from PCI clinical quality registry initiatives across the region met and agreed to progress with a demonstration project regarding harmonisation of data for pooled individual data analysis. All volunteer registries were accepted as participants irrespective of level of maturity and representativeness from a country. Registries included single-hospital registries in Singapore, Hong Kong and Vietnam, state-based registries from Australia (Victoria and South Australia), and a national registry from Malaysia [
The Asia-Pacific Evaluation of Cardiovascular Therapies (ASPECT) Collaboration—improving the quality of cardiovascular care in the Asia Pacific Region.
]. Briefly, data from Singapore were based on a single public tertiary care hospital that contributes to a national registry, whereas data from Australia’s Melbourne Interventional Group were from six public tertiary care hospitals. Whilst the collaboration is opportunistic and with the exception of Malaysia does not reflect populations data, these registries assist to identify trends in variation in patient and procedural details and monitor the quality of activity and changes in clinical practice [
]. The participating clinical quality registries collect data from all patients admitted to the hospital between 2015 and 2017 with the exception of Vietnam, for which the data collection was restricted to all patients over a few typical days according to a pre-specified protocol to manage workload and feasibility [
Protocol was developed and submitted to the Alfred Hospital in Melbourne as the lead Ethics Committee (Ethics Committee No. 586/18). Following approval, each participating registry satisfies the ethical review requirements for the collection of confidential data within the local jurisdiction. All STEMI patients who had undergone PCI at the participating registries during the period were included in the analyses. All participating registries then extracted the agreed set of data elements (Supplementary file) between 2015 and 2017 and prepared anonymised individual patient data sets for merging.
Data Dictionary
The collaborators established definitions and a data dictionary for relevant fields to minimise misclassification bias. For instance, STEMI was defined according the Third Universal Definition of Myocardial Infarction (MI) [
]: Was a peri-/postoperative myocardial infarction (MI) diagnosed by finding at least two of the following criteria:
a.
Enzyme level elevation either:
i.
CK-MB >30 units; OR
ii.
Troponin > 20.0 micrograms/L; OR
iii.
Troponin level equivalent documented at your instruction, provided operation does not involve myocardial incision
b.
New wall motion abnormalities
c.
Serial ECG (at least two) showing Q waves, duration ≥0.03 ms in two contiguous leads. Major bleeding was defined according to the Bleeding Academic Research Consortium (BARC) consensus [
], specifically, BARC 3 and 5 types of bleeding. New renal impairment was defined as a Yes/No response by asking; “Was there acute post-operative renal insufficiency” characterised by one of the following:
i.
Increased serum creatinine to >0.2 mmol/L (>200 μmol/L) AND a doubling or greater increase in creatinine over the baseline preoperative value AND the patient did not require preoperative dialysis/haemofiltration, OR
ii.
A new postoperative requirement for dialysis/haemofiltration (when the patient did not require this preoperatively).
The definitions and a data dictionary as well as a data extract schema defining the attributes for each field required were provided to each registry. Whilst the process of data collection varied at each site, data were checked and reviewed locally at each site prior to entry in the registry. No formal data validation procedures were undertaken as part of the registry data collection.
Data Management
Each registry dataset was transferred by Secure Unified File Exchange (SUFEX) provided by the Centre for Clinical Research and Education (CCRE) at Curtin University. The data were initially checked for completeness and queries were generated and sent back to the individual registry for confirmation to clarify whether any data marked as “missing” meant that the data were not supplied, and not due to some oversight or conversion error. Clean data were then resubmitted to the CCRE for compilation into an integrated ASPECT STEMI dataset for analysis. The ASPECT STEMI dataset was housed in a SafeHaven at Curtin University using secure ISO27001 compliant technology deployed via a mutual collaboration with Swansea Universities Secure Anonymous Information Linkage (SAIL) Gateway in the United Kingdom [
A case study of the Secure Anonymous Information Linkage (SAIL) Gateway: a privacy-protecting remote access system for health-related research and evaluation.
Feasibility was determined through the following measures:
-
The number of ASPECT members able to gain governance approvals within the allocated time frame to provide data. The allocated time frame was within 6 months of project ethical approval;
-
The number of sites able to provide >90% of agreed data elements; and
-
The number of sites with missing data <10% for all variables included.
Statistical Analyses of Patient Characteristics and In-Hospital Outcomes
Demographics, medical conditions, presentation details, procedural characteristics and in-hospital clinical outcomes of patients admitted with STEMI at each of the PCI registries (the relatively small South Australia registry was combined with the Victoria registry) were described in mean and standard deviation (or median and interquartile range) or number of observations and percentage. Comparison of these variables by sites were performed using one-way analysis of variance, Kruskal-Wallis, or Chi-squared tests, where appropriate. The in-hospital clinical outcomes included new renal impairment, major bleeding, peri-procedural recurrent myocardial infarction, stent thrombosis, emergency coronary artery bypass graft (CABG), target vessel revascularisation (TVR), cerebrovascular events (CVE)/stroke, death and any major adverse cardiovascular and cerebrovascular events (MACCE) in hospital. In-hospital MACCE was defined as having experienced peri-procedural recurrent myocardial infarction, stent thrombosis, emergency CABG, CVE/stroke, or death.
Predictors of in-hospital MACCE on univariate logistic regression models were included in the corresponding multiple logistic regression models with registry sites as an additional covariate. Stepwise approach was then applied until the final models contained only the predictors associated significantly (p<0.05) with in-hospital MACCE. The adjusted odds ratios (OR) and 95% confidence intervals (CI) were reported. It is acknowledged that residual confounding is unavoidable in this analysis due to lack of data, such as data on marker of use of statin and marker of frailty. Complete case analyses were performed as the missing data were likely to be missing completely at random. Significance was set as <0.05. All statistical analyses were performed using Stata MP version 16 (StataCorp, College Station, TX, USA).
Results
Feasibility—Governance Approval
Six (6) of seven ASPECT registries (86%) gained governance approvals for the project and were able to provide data in the time frame for analysis; South Australia, Melbourne (combined into one “Australia” due to the relatively small number from South Australia), Vietnam, Hong Kong, Malaysia and Singapore. Governance requirements from a single non-participating registry (Indonesia) were not in place for inclusion in this analysis.
Feasibility—Data Quality
Across the six registries, five (83%) were able to provide >90% of the agreed data elements (Tables 1 and 2). About 68% (21/31) of the collated elements had <10% of missingness. The patient characteristic variables with missingness over 40% were family history of coronary artery disease (CAD) and lung disease (none from Malaysia and Vietnam), estimated ejection fraction data (none from Hong Kong, 68% missing from Malaysia) and door to balloon time (none from Vietnam, 79% missing from Hong Kong and Malaysia). None of the key variables (age, gender, treatments given prior to procedure, major bleeding and deceased status) had any missingness.
Table 1Characteristics and presentation details of ASPECT STEMI patients, and treatment provided, by sites (n=12,620)
The combined ASPECT STEMI dataset comprised a total of 12,620 cases (Table 1) with the number varying from each participating registry due to a differing number and size of contributing sites. Most of the patients were from Malaysia (39%), followed by Hong Kong (27%), Melbourne and South Australia combined (24%), Singapore (9%) and Vietnam (1%).
There were more male patients admitted for STEMI across all sites (p<0.001; Table 1). On average, the Vietnamese patients were the oldest across all sites (p<0.001). Forty-four per cent (44%) of the Malaysian patients were current smokers while 59% of the Singaporean patients never smoked. Whilst 29% of the Australian patients had family history of CAD, the majority of patients from Hong Kong and Singapore had no family history of CAD. Diabetes was more prevalent in Singapore (30%), Hong Kong (34%) and Malaysia (38%) than in Australia (18%) and Vietnam (14%). Of note, 98% and 88% of the patients from Hong Kong had dyslipidaemia and hypertension respectively. Most of the patients across all registries did not have a history of congestive heart failure, cerebrovascular disease, peripheral vascular disease, previous PCI or previous CABG. About 34% of the Malaysian patients had previous myocardial infarction (MI), which was at least two times the rate observed in the other registries (p<0.001).
Presentation Details and Treatment
The percutaneous entry location was primarily radial or brachial across all sites whilst Singapore reported similar percentage for radial or brachial (49%) and femoral (51%) (Table 1). The median door to balloon time was the longest in Malaysia (101 mins) and shortest in Vietnam (65 mins). Treatments provided were similar across the region with high use of aspirin and any P2Y12 inhibitors. Majority of stents used were drug eluting (87% of cases). Bare-metal stent use ranged from 19.4% in Australia to no bare metal stents used in Vietnam.
The majority of lesions (95%) were de novo. The lesion type, as defined by the American College of Cardiology/American Heart Association (ACC/AHA) guidelines [
2013 ACCF/AHA key data elements and definitions for measuring the clinical management and outcomes of patients with acute coronary syndromes and coronary artery disease: a report of the American College of Cardiology Foundation/American Heart Association Task Force on clinical data standards.
], differed across the registries. With respect to Type C lesions, Vietnam reported a rate of 87% whereas Malaysia reported 43% and Hong Kong reported just 12% (Figure 1). Data related to lesion types, lesion length >20 mm, lesion success and complications are presented in Figure 1. Lesion success was high across all sites. The lesion complication rate ranged from 0.3 to 8.2%.
Figure 1STEMI lesion details across the ASPECT registry.
Table 2 illustrates in-hospital outcomes, with low rates of major bleeding, peri-procedural recurrent MI, stent thrombosis, emergency CABG, TVR or stroke. New renal impairment was reported in 6% of the patients in Vietnam, 5% in Australia and 3% in Singapore (p<0.001). The in-hospital mortality ranged from 3% in Vietnam to 8% in Singapore (p=0.012) whilst the prevalence of MACCE ranged from 3% in Vietnam to 10% in Singapore (p<0.001).
Potential Predictors of In-Hospital MACCE and Mortality
Predictors associated with odds of any MACCE in hospital varied while adjusted for sites (Figure 2). Having had cardiogenic shock at PCI, experienced complications with lesion and had bare-metal stent were associated with higher odds of in-hospital MACCE. However, the estimation for cardiogenic shock is imprecise (95% CI of aOR 6.5–15.3). Lower odds of in-hospital MACCE were associated with history of dyslipidaemia, having eGFR of >30, having ejection fraction of >35% (Figure 2).
Figure 2In-hospital major adverse cardiovascular event (MACCE).
The ASPECT collaboration has successfully merged individual anonymised patient data from six independent registries across the Asia-Pacific region. The registries have been independently established and collected data for varying time periods, however a high degree of alignment of data definitions was observed with the standard definitions of variables being used and there was no issue with translation being documented because English is commonly spoken across the six sites. Nevertheless, missingness was observed. Majority of the high level of missingness (>40%) was contributed by missing data from the entire registries and hence routine data check at individual registry site and where possible, multiple imputation are recommended to further improve data harmonising. It is worth noting that some of the variables with high level of missingness (such as history of lung disease and target vessel revascularisation) are not influential of the main clinical outcomes (death or MACE) whilst the main predictors (age, gender, treatments given prior to procedure, major bleeding) had no missingness. To our knowledge, our study is the first demonstration of the prospects of pooling of anonymised individual patient data from independent registries across the Asia-Pacific region.
Risk factors, in particular smoking, diabetes and hypertension varied widely across the region which could impact on management decisions and clinical outcomes. High rates of diabetes in Hong Kong, Malaysia and Singapore in comparison to the Australian and the Vietnamese registries is consistent with other population-based cohort studies of cardiovascular risk factors across the region [
]. The relatively low rate of current smoking amongst patients from Vietnam is unusual, given the high population rates of smoking, particularly in males. This could be due to following medical advice to quit smoking. The quality of self-reported family and personal medical history also varied across the registries, however did reflect previously reported higher rates of family history of CAD in predominantly Caucasian populations and lower rates amongst Chinese [
Family history of coronary heart disease and the incidence and progression of coronary artery calcification: Multi-Ethnic Study of Atherosclerosis (MESA).
Multi-vessel coronary disease (MVD) was highly prevalent across the region with Vietnam being the exception (range 16.2% to 68.7%). It is unclear why the rate of MVD reported was so low in Vietnam. This may reflect the level of background cardiovascular risk factors including diabetes and hyperlipidaemia, as Vietnam might be considered as the least ‘westernised society’ of the countries involved in the collaboration. Collecting data over a longer period instead of a few typical days from registries may be useful to explain the differences.
Previous research into comparative studies of clinical outcomes in patients with STEMI undergoing interventional procedures across the region have used a meta-analytical approach from individual registry reporting [
Comparative overview of ST-elevation myocardial infarction epidemiology, demographics, management, and outcomes in five Asia-Pacific countries: a meta-analysis.
]. The current analysis sees Singapore and Malaysian in-hospital mortality rates similar to those reported in Australia, Hong Kong and Vietnam, with the overall mortality approximately 5%. This compares favourably with that reported in the UK, Europe and the United States [
Predictors of in-hospital mortality of ST-segment elevation myocardial infarction patients undergoing interventional treatment. An analysis of data from the RO-STEMI registry.
Whilst this study demonstrated the prospects of pooling anonymised individual patient data from independent registries across the Asia-Pacific region, and the potential to identify predictors of the outcomes using the current state of pooled data, there is room for improvements in relation to the feasibility of data quality. For instance, formal data validation routine (especially of variables that are prone to missing) may need to be integrated in registry data collection in order to harmonise the data for robust statistical modelling. Additional variables such as use of statin or marker for statin use, marker of frailty could also be included in the data elements to minimise the impact of residual confounding. The predictors identified from this preliminary feasibility study need to be interpreted with caution.
Whilst the co-authors across different countries noted that the findings were comparable to observations at their clinical practices, this study has some limitations. For instance, lack of data to minimise impact of missingness and residual confounding in the modelling. In addition, with the exception of Malaysia, which collects data from all PCI hospitals in the country through the National Cardiac Databank, the data presented represent one or more PCI capable centres in the region and may not provide a true representation of all patients undergoing PCI in the country or region. We learnt that it may also be useful to collect data over a longer period instead of across a few typical days. In addition, a number of the registries across the region link the respective National Death data sets and incorporate 12-month follow-up into their registry outcomes ascertainment. Future studies may aim to look at harmonising longer-term outcomes data for robust statistical modelling in order to identify influential factors on quality of care and outcomes for patients across the Asia-Pacific region.
Conclusions
It is possible to identify variation across the region in terms of patient characteristics and procedural details, and to monitor the quality of activity in terms of time for treatment, contemporary clinical practice and patient outcomes at the regional level using the ASPECT registry data. The process of individual patient data merging has been validated and has the potential to provide a platform for research, education and training on the delivery of cardiac interventions at a national and international level. As appropriate ethics and governance procedures have been established, and more registries are established through ASPECT, broader participation from sites across the Asia-Pacific region is possible to facilitate further pooled data analyses addressing questions related to the quality of care and outcomes for patients across the Asia-Pacific region. The collaboration will inform clinicians of patient outcomes from cardiac interventions in the region and provide opportunity to compare longitudinal data with the North American and European data to predict (long-term) patient outcomes.
Funding
The ASPECT Collaboration activities are supported by an Australian National Health and Medical Research Council (NHMRC) Program Grant [grant number: GNT 546272]. The first author is supported by a NHMRC Principal Research Fellowship [grant number: GNT 1136372].
Conflict of Interest
All authors declare no competing interest.
Acknowledgement
We acknowledge the contribution of all data managers and clinicians participating in country specific registries.
Improved outcomes in patients with ST-elevation myocardial infarction during the last 20 years are related to implementation of evidence-based treatments: experiences from the SWEDEHEART registry 1995-2014.
Impact of institutional and operator volume on short-term outcomes of percutaneous coronary intervention: a report from the Japanese Nationwide Registry.
The current status of percutaneous coronary intervention in Korea: based on year 2014 cohort of Korean Percutaneous Coronary Intervention (K-PCI) Registry.
Using the LACE index to predict 30-day all-cause unplanned readmission and mortality in acute myocardial infarction patients: insights from the CADOSA registry.
Effect of age on clinical outcomes in elderly patients (>80 years) undergoing percutaneous coronary intervention: insights from a multi-centre australian PCI registry.
The Asia-Pacific Evaluation of Cardiovascular Therapies (ASPECT) Collaboration—improving the quality of cardiovascular care in the Asia Pacific Region.
A case study of the Secure Anonymous Information Linkage (SAIL) Gateway: a privacy-protecting remote access system for health-related research and evaluation.
2013 ACCF/AHA key data elements and definitions for measuring the clinical management and outcomes of patients with acute coronary syndromes and coronary artery disease: a report of the American College of Cardiology Foundation/American Heart Association Task Force on clinical data standards.
Family history of coronary heart disease and the incidence and progression of coronary artery calcification: Multi-Ethnic Study of Atherosclerosis (MESA).
Comparative overview of ST-elevation myocardial infarction epidemiology, demographics, management, and outcomes in five Asia-Pacific countries: a meta-analysis.
Predictors of in-hospital mortality of ST-segment elevation myocardial infarction patients undergoing interventional treatment. An analysis of data from the RO-STEMI registry.
01084-8&id=gr1.jpg){kind=link}
01084-8&id=gr2.jpg){kind=link}