Use of an ultrasound picture archiving and communication system to answer research questions: Description of data cleaning methods.

Introduction/Purpose: Ultrasound picture archiving and communication system (PACS) databases are useful for quality improvement and clinical research but frequently contain free text that is not easily readable. Here, we present a method to extract and clean a semi-structured echocardiography (cardiac ultrasound) PACS database. Methods: Echocardiography studies between 1 January 2010 and 31 December 2018 were extracted using a data mining tool. Numeric variables were recoded with extreme values excluded. Analysis of free text, including descriptions of the heart valves and right and left ventricular size and function, was performed using a rule-based system. Different levels of free text variables were initially identified using commonly used phrases and then iteratively developed. Randomly selected sets of 100 studies were compared to the electronic health record to validate the data cleaning process. Results: The data validation step was performed three times in total, with Cohen's kappa ranging between 0.88 and 1.00 for the final set of data validation across all measures. Conclusion: Free text cleaning of semi-structured PACS databases is possible using freely available open-source software. The accuracy of this method is high, and the resulting dataset can be linked to administrative data to answer research questions. We present a method that could be used to answer clinical questions or to develop quality improvement initiatives.

Outcomes of patients with early calcific aortic valve disease detected by clinically indicated echocardiography.

AIMS: Previous studies have demonstrated relatively slow rates of progression of early calcific aortic valve disease (CAVD), which encompasses aortic sclerosis (ASc) and mild aortic stenosis (AS). The potential evolution to clinically significant AS is unclear, and we therefore examined the long-term outcomes of patients with ASc and mild AS detected at the time of clinically indicated echocardiography. METHODS AND RESULTS: Data from initial clinically indicated echocardiograms performed between 2010 and 2018 in patients aged ≥18 years were extracted and linked to nationally collected outcome data. Those with impaired right or left systolic ventricular function or other significant left-sided valve disease were excluded. A time to first event analysis was performed with a composite primary outcome of cardiovascular death and aortic valve intervention (AVI). Of the 13 313 patients, 8973 had no CAVD, 3436 had ASc, and 455 had mild AS. The remainder had moderate or worse stenosis. Over a median follow-up period of 4.2 (interquartile range 1.8-6.7) years (and after adjustment for age and sex), those with ASc were at greater risk of the primary outcome [hazard ratio (HR) 2.9, 95% confidence interval (CI) 2.1-4.0] and need for AVI (HR 26.8, 95% CI 9.1-79.1) compared with those with no CAVD. Clinical event rates accelerated after ∼5 years in those with mild AS. CONCLUSION: Patients with ASc are >25 times more likely to require AVI than those with no CAVD, and follow-up echocardiography should be considered within 3-4 years in those with mild AS.

Image quality and technical limitations in emergency department cardiac point-of-care ultrasound: A retrospective cohort study.

OBJECTIVE: To assess the image quality and common technical limitations seen on cardiac point-of-care ultrasound (POCUS) performed and archived in a single New Zealand ED. METHODS: A retrospective cohort study of clinically indicated cardiac POCUS, archived from 1 October 2019 to 20 May 2020. Archived examinations were retrospectively reviewed by an ED POCUS expert, and an expert cardiac sonographer to determine diagnostic image quality, technical limitations present and opportunities for image quality improvement. Image quality of credentialed examinations was compared to uncredentialed examinations and examinations that were undocumented in the medical record. RESULTS: A total of 211 cardiac POCUS examinations were included. The impact of image quality on diagnostic interpretation was only documented in <2% of examinations. There was no difference in median global image quality scores for uncredentialed and credentialed examinations (8.5 vs 9, P = 0.55) and median score for undocumented examinations (5.5) was lower than credentialed examinations (P < 0.01). Common technical limitations identified were off-axis imaging and artefacts limiting image quality. CONCLUSION: In the present study of clinically indicated cardiac POCUS, low image quality was common but the impact of image quality on diagnostic interpretation was very rarely documented in the medical record. Local quality assurance and training should be directed at credentialed and uncredentialed clinicians including strategies to improve off-axis imaging and managing artefacts where possible. Standardised documentation of image quality that may impact diagnostic accuracy should be encouraged.

Ethnicity Is Associated With Differing Presentation and Outcomes of Patients Undergoing Aortic Valve Replacement for Calcific Aortic Stenosis in Aotearoa New Zealand.

AIM: Surgical aortic valve replacement (SAVR) has been the gold standard for treatment of severe symptomatic aortic stenosis (AS) for decades. We examined whether ethnic differences exist in the presentation and outcomes of patients undergoing aortic valve replacement (AVR) for AS in New Zealand. METHODS: Patients of New Zealand European, Maori, and Pacific Island ethnicities undergoing SAVR with or without other procedures in New Zealand public hospitals from 2017 to 2019 were included. Major postoperative outcomes were compared between ethnic groups, with 30-day mortality being the primary outcome. RESULTS: A total of 1,175 patients were included: 1,085 European, 50 Maori, and 40 Pacific. The mean age was 71.1±9.4 years, and men accounted for more than half of all patients (69.9%). Maori (64.7±9.4 years) and Pacific (65.4±10.1 years) patients were younger when undergoing SAVR compared with European patients (71.7±9.2; analysis of variance p<0.001). Maori and Pacific patients had a higher prevalence of diabetes, poorer renal function, and worse left ventricular function; 30-day mortality was higher in Maori and Pacific compared with European patients (6% and 10% vs 2.4%, respectively; Fisher's exact test p=0.011), with odds ratio of 3.06 (95% confidence interval [CI] 0.88-10.66) for Maori patients after adjustment for EuroSCORE II and odds ratio of 5.23 (95% CI 1.79-16.07) for Pacific patients. CONCLUSIONS: There are significant differences in presentation and outcomes of patients undergoing AVR in New Zealand. Maori and Pacific patients undergo SAVR at a younger age, have more preoperative comorbidities, and have higher rates of 30-day mortality than European patients.

Survey of echocardiography practice across five continents.

BACKGROUND AND AIM: The term echocardiography refers to a diverse range of cardiovascular ultrasound imaging methods, both inside and outside specialist cardiology practice. While guidelines exist, we hypothesized that there are significant worldwide differences in the way echocardiography is practiced. We surveyed echocardiography practitioners around the world to characterize the workforce and their practice. METHOD: Social media and word of mouth were used in an explosive sampling approach to recruit echo users, who then completed an online survey that included personal demographics and questions about their practice, their resources, and daily use of echocardiography. RESULTS: In total, 594 participants completed the survey: 54.9% sonographers; 30% cardiologists, with the remainder other physicians or trainees. Significant variation in the number of echoes performed and the time allocated to scanning was observed. There were also differences in the gathering of adjunct measures such as blood pressure and body size. CONCLUSION: There is wide variation in echocardiography practices across the world. Differences are likely to be both clinician- and healthcare system-driven. Guidelines for practice developed in well-resourced western countries and intended for use in cardiology-based echocardiography laboratories may not be applicable to other countries or indeed to new echo users.

Abbreviated Echocardiographic Screening for Rheumatic Heart Disease by Nonexperts with and without Offsite Expert Review: A Diagnostic Accuracy Study.

BACKGROUND: Early detection of rheumatic heart disease (RHD) through echocardiographic screening can facilitate early access to effective treatment, which reduces the risk for progression. Accurate, feasible approaches to echocardiographic screening that can be incorporated into routine health services are needed. The authors hypothesized that offsite expert review could improve the diagnostic accuracy of nonexpert-obtained echocardiographic images. METHODS: This prospective cross-sectional study was performed to evaluate the diagnostic accuracy of health worker-conducted single parasternal long-axis view with a sweep of the heart using hand-carried ultrasound for the detection of RHD in high-risk populations in Timor-Leste and Australia. In the primary analysis, the presence of any mitral or aortic regurgitation met the criteria for a positive screening result. Sensitivity and specificity were calculated for a screen-and-refer approach based on nonexpert practitioner assessment (approach 1) and for an approach using offsite expert review of nonexpert practitioner-obtained images to decide onward referral (approach 2). Each participant had a reference test performed by an expert echocardiographer on the same day as the index test. Diagnosis of RHD was determined by a panel of three experts, using 2012 World Heart Federation criteria. RESULTS: The prevalence of borderline or definite RHD among 3,329 participants was 4.0% (95% CI, 3.4%-4.7%). The sensitivity of approach 1 for borderline or definite RHD was 86.5% (95% CI, 79.5%-91.8%), and the specificity was 61.4% (95% CI, 59.7%-63.1%). Approach 2 achieved similar sensitivity (88.4%; 95% CI, 81.5%-93.3%) and improved specificity (77.1%; 95% CI, 75.6%-78.6%). CONCLUSION: Nonexpert practitioner-obtained single parasternal long-axis view with a sweep of the heart images, reviewed by an offsite expert, can detect borderline and definite RHD on screening with reasonable sensitivity and specificity. Brief training of nonexpert practitioners with ongoing support could be used as an effective strategy for scaling up echocardiographic screening for RHD in high-risk settings.

Adjunct Methods for the Detection of Patent Foramen Ovale: The Contribution of Transcranial Doppler and the Valsalva Manoeuvre.

A patent foramen ovale (PFO) is present in 25% of the population. In some patients, especially those without traditional stroke risk factors and with no immediately apparent cause, a cryptogenic stroke may be caused by an embolus passing through the PFO to the systemic circulation. The identification, or indeed exclusion, of a PFO is sought in these patients, most commonly using contrast-enhanced transthoracic or transoesophageal echocardiography. Another method for detecting a PFO is transcranial Doppler, which allows the detection of PFO possibly without the need for an echo laboratory, and with arguably improved sensitivity. This review will focus on transcranial Doppler detection of PFO, with a brief summary of echocardiographic techniques and the use of ultrasound contrast agents, and the role of provocations to increase diagnostic accuracy, specifically the Valsalva manoeuvre. We discuss the phases alongside the direct and indirect signs of an adequate Valsalva manoeuvre.

New Echocardiography Reference Ranges for Aotearoa (NewERA) Study: the application of international echocardiographic reference values to linear measurements of the hearts of healthy, young Maori and Pacific adults may not detect cardiac enlargement.

AIMS: To develop ethnic-specific echocardiography reference ranges for Aotearoa, and to investigate the impact of indexation to body surface area (BSA). Current reference international ranges are derived from people of mostly NZ European ethnicity and may not be appropriate for Maori and New Zealanders of Pacific ethnicity, who both experience high rates of cardiovascular disease. METHODS: Echocardiography was performed in a cross-sectional study of 263 healthy adults (18-50 years): Maori (N=71, 43 female), Pacific (N=53, 28 female), European (N=139, 74 female). Linear measurements of the left heart are reported and indexed to BSA. The upper/lower limit of normal (ULN/LLN) by ethnicity and sex were derived (quantile regression). Ethnic- and sex-specific differences were examined using ANOVA. RESULTS: The ULN was higher for all un-indexed dimensions in men compared to women, and for most indices the ULN was smallest in NZ Europeans and largest in Maori and Pacific peoples. Indexation reversed these relationships: NZ Europeans had higher ULN for many measurements. CONCLUSIONS: Indexing to BSA introduced bias that preferences the NZ European ethnicity by creating an upper limit reference threshold that far exceeds this sample's upper range. As a result, this may lead to under-recognition of cardiac enlargement in Maori and Pacific patients, and in particular for women. Unique reference ranges for all ethnic groups and sexes are required to optimally detect and manage cardiovascular diseases (CVD) in Aotearoa.

Estimating heart mass from heart volume as measured from post-mortem computed tomography.

Heart mass can be predicted from heart volume as measured from post-mortem computed tomography (PMCT), but with limited accuracy. Although related to heart mass, age, sex, and body dimensions have not been included in previous studies using heart volume to estimate heart mass. This study aimed to determine whether heart mass estimation can be improved when age, sex, and body dimensions are used as well as heart volume. Eighty-seven (24 female) adult post-mortem cases were investigated. Univariable predictors of heart mass were determined by Spearman correlation and simple linear regression. Stepwise linear regression was used to generate heart mass prediction equations. Heart mass estimate performance was tested using median mass comparison, linear regression, and Bland-Altman plots. Median heart mass (P = 0.0008) and heart volume (P = 0.008) were significantly greater in male relative to female cases. Alongside female sex and body surface area (BSA), heart mass was univariably associated with heart volume in all cases (R2 = 0.72) and in male (R2 = 0.70) and female cases (R2 = 0.64) when segregated. In multivariable regression, heart mass was independently associated with age and BSA (R2 adjusted = 0.46-0.54). Addition of heart volume improved multivariable heart mass prediction in the total cohort (R2 adjusted = 0.78), and in male (R2 adjusted = 0.74) and female (R2 adjusted = 0.74) cases. Heart mass estimated from multivariable models incorporating heart volume, age, sex, and BSA was more predictive of actual heart mass (R2 = 0.75-0.79) than models incorporating either age, sex, and BSA only (R2 = 0.48-0.57) or heart volume only (R2 = 0.64-0.73). Heart mass can be more accurately predicted from heart volume measured from PMCT when combined with the classical predictors, age, sex, and BSA.

Composite Echocardiographic Score to Predict Long-Term Survival Following Myocardial Infarction.

BACKGROUND: Whilst the left ventricular ejection fraction (LVEF) remains the primary echocardiographic measure widely utilised for risk stratification following myocardial infarction (MI), it has a number of well recognised limitations. The aim of this study was to compare the prognostic utility of a composite echocardiographic score (EchoScore) composed of prognostically validated measures of left-ventricular (LV) size, geometry and function, to the utility of LVEF alone, for predicting survival following MI. METHODS: Retrospective data on 394 consecutive patients with a first-ever MI were included. Comprehensive echocardiography was performed within 24 hours of admission for all patients. EchoScore consisted of LVEF<50%, left atrial volume index>34 mL/m2, average E/e >14, E/A ratio>2, abnormal LV mass index, and abnormal LV end-systolic volume index. A single point was allocated for each measure to derive a score out of 6. The primary outcome measure was all-cause mortality. RESULTS: At a median follow-up of 24 months there were 33 deaths. On Kaplan-Meier analysis, a high EchoScore (>3) displayed significant association with all-cause mortality (log-rank χ2=74.48 p<0.001), and was a better predictor than LVEF<35% (log-rank χ2=17.01 p<0.001). On Cox proportional-hazards multivariate analysis incorporating significant clinical and echocardiographic predictors, a high EchoScore was the strongest independent predictor of all-cause mortality (HR 6.44 95%CI 2.94-14.01 p<0.001), and the addition of EchoScore resulted in greater increment in model power compared to addition of LVEF (model χ2 56.29 vs 44.71 p<0.001, Harrell's C values 0.83 vs 0.79). CONCLUSIONS: A composite echocardiographic score composed of prognostically validated measures of LV size, geometry, and function is superior to LVEF alone for predicting survival following MI.