Health equity assessment of machine learning performance (HEAL): a framework and dermatology AI model case study

Background

Artificial intelligence (AI) has repeatedly been shown to encode historical inequities in healthcare. We aimed to develop a framework to quantitatively assess the performance equity of health AI technologies and to illustrate its utility via a case study.

Methods

Here, we propose a methodology to assess whether health AI technologies prioritise performance for patient populations experiencing worse outcomes, that is complementary to existing fairness metrics. We developed the Health Equity Assessment of machine Learning performance (HEAL) framework designed to quantitatively assess the performance equity of health AI technologies via a four-step interdisciplinary process to understand and quantify domain-specific criteria, and the resulting HEAL metric. As an illustrative case study (analysis conducted between October 2022 and January 2023), we applied the HEAL framework to a dermatology AI model. A set of 5420 teledermatology cases (store-and-forward cases from patients of 20 years or older, submitted from primary care providers in the USA and skin cancer clinics in Australia), enriched for diversity in age, sex and race/ethnicity, was used to retrospectively evaluate the AI model's HEAL metric, defined as the likelihood that the AI model performs better for subpopulations with worse average health outcomes as compared to others. The likelihood that AI performance was anticorrelated to pre-existing health outcomes was estimated using bootstrap methods as the probability that the negated Spearman's rank correlation coefficient (i.e., "R") was greater than zero. Positive values of R suggest that subpopulations with poorer health outcomes have better AI model performance. Thus, the HEAL metric, defined as p (R >0), measures how likely the AI technology is to prioritise performance for subpopulations with worse average health outcomes as compared to others (presented as a percentage below). Health outcomes were quantified as disability-adjusted life years (DALYs) when grouping by sex and age, and years of life lost (YLLs) when grouping by race/ethnicity. AI performance was measured as top-3 agreement with the reference diagnosis from a panel of 3 dermatologists per case.

Findings

Across all dermatologic conditions, the HEAL metric was 80.5% for prioritizing AI performance of racial/ethnic subpopulations based on YLLs, and 92.1% and 0.0% respectively for prioritizing AI performance of sex and age subpopulations based on DALYs. Certain dermatologic conditions were significantly associated with greater AI model performance compared to a reference category of less common conditions. For skin cancer conditions, the HEAL metric was 73.8% for prioritizing AI performance of age subpopulations based on DALYs.

Interpretation

Analysis using the proposed HEAL framework showed that the dermatology AI model prioritised performance for race/ethnicity, sex (all conditions) and age (cancer conditions) subpopulations with respect to pre-existing health disparities. More work is needed to investigate ways of promoting equitable AI performance across age for non-cancer conditions and to better understand how AI models can contribute towards improving equity in health outcomes.

Funding

Google LLC.

Risk Stratification for Diabetic Retinopathy Screening Order Using Deep Learning: A Multicenter Prospective Study

Purpose

Real-world evaluation of a deep learning model that prioritizes patients based on risk of progression to moderate or worse (MOD+) diabetic retinopathy (DR).

Methods

This nonrandomized, single-arm, prospective, interventional study included patients attending DR screening at four centers across Thailand from September 2019 to January 2020, with mild or no DR. Fundus photographs were input into the model, and patients were scheduled for their subsequent screening from September 2020 to January 2021 in order of predicted risk. Evaluation focused on model sensitivity, defined as correctly ranking patients that developed MOD+ within the first 50% of subsequent screens.

Results

We analyzed 1,757 patients, of which 52 (3.0%) developed MOD+. Using the model-proposed order, the model's sensitivity was 90.4%. Both the model-proposed order and mild/no DR plus HbA1c had significantly higher sensitivity than the random order (P < 0.001). Excluding one major (rural) site that had practical implementation challenges, the remaining sites included 567 patients and 15 (2.6%) developed MOD+. Here, the model-proposed order achieved 86.7% versus 73.3% for the ranking that used DR grade and hemoglobin A1c.

Conclusions

The model can help prioritize follow-up visits for the largest subgroups of DR patients (those with no or mild DR). Further research is needed to evaluate the impact on clinical management and outcomes.

Translational Relevance

Deep learning demonstrated potential for risk stratification in DR screening. However, real-world practicalities must be resolved to fully realize the benefit.

Predicting the risk of developing diabetic retinopathy using deep learning

BACKGROUND

Diabetic retinopathy screening is instrumental to preventing blindness, but scaling up screening is challenging because of the increasing number of patients with all forms of diabetes. We aimed to create a deep-learning system to predict the risk of patients with diabetes developing diabetic retinopathy within 2 years.

METHODS

We created and validated two versions of a deep-learning system to predict the development of diabetic retinopathy in patients with diabetes who had had teleretinal diabetic retinopathy screening in a primary care setting. The input for the two versions was either a set of three-field or one-field colour fundus photographs. Of the 575 431 eyes in the development set 28 899 had known outcomes, with the remaining 546 532 eyes used to augment the training process via multitask learning. Validation was done on one eye (selected at random) per patient from two datasets: an internal validation (from EyePACS, a teleretinal screening service in the USA) set of 3678 eyes with known outcomes and an external validation (from Thailand) set of 2345 eyes with known outcomes.

FINDINGS

The three-field deep-learning system had an area under the receiver operating characteristic curve (AUC) of 0·79 (95% CI 0·77-0·81) in the internal validation set. Assessment of the external validation set-which contained only one-field colour fundus photographs-with the one-field deep-learning system gave an AUC of 0·70 (0·67-0·74). In the internal validation set, the AUC of available risk factors was 0·72 (0·68-0·76), which improved to 0·81 (0·77-0·84) after combining the deep-learning system with these risk factors (p<0·0001). In the external validation set, the corresponding AUC improved from 0·62 (0·58-0·66) to 0·71 (0·68-0·75; p<0·0001) following the addition of the deep-learning system to available risk factors.

INTERPRETATION

The deep-learning systems predicted diabetic retinopathy development using colour fundus photographs, and the systems were independent of and more informative than available risk factors. Such a risk stratification tool might help to optimise screening intervals to reduce costs while improving vision-related outcomes.

FUNDING

Google.

Fundus photograph-based deep learning algorithms in detecting diabetic retinopathy

Remarkable advances in biomedical research have led to the generation of large amounts of data. Using artificial intelligence, it has become possible to extract meaningful information from large volumes of data, in a shorter frame of time, with very less human interference. In effect, convolutional neural networks (a deep learning method) have been taught to recognize pathological lesions from images. Diabetes has high morbidity, with millions of people who need to be screened for diabetic retinopathy (DR). Deep neural networks offer a great advantage of screening for DR from retinal images, in improved identification of DR lesions and risk factors for diseases, with high accuracy and reliability. This review aims to compare the current evidences on various deep learning models for diagnosis of diabetic retinopathy (DR).

Electronic colon-cleansing for CT colonography: diagnostic performance

BACKGROUND

The purpose of this study was to evaluate whether an electronic-colonic-cleansing (ECC) algorithm is beneficial for the diagnostic performance compared to a CT colonography (CTC) evaluation without electronic cleansing in tagged datasets.

METHODS

Two blinded readers evaluated CTC datasets from 79 patients with 153 colorectal polyps confirmed by optical colonoscopy. Cases were read in a randomized order with and without the use of electronic colon-cleansing software. Per-polyp sensitivity, per-polyp/per-patient specificity and reading times (with and without ECC) have been calculated and reported.

RESULTS

Per-polyp sensitivity for polyps >6 mm without using ECC was 60.4% (Reader 1: 59.7%, Reader 2: 61.1%), while polyps >10 mm were detected with a sensitivity of 58.3% (Reader 1: 66.7%, Reader 2: 50%). On electronically cleansed datasets, the sensitivity was 73.6% (Reader 1: 76.4%; Reader 2: 70.8%) for polyps >6 mm and 83.3% (Reader 1: 83.3%; Reader 2: 83.3%), respectively. Per-patient specificity was 75% without using cleansing (Reader 1: 68%, Reader 2: 82%) and 81.5% using ECC (Reader 1: 86%, Reader 2: 77%).

CONCLUSION

Reading CTC cases using ECC software improves sensitivity in detecting clinically relevant colorectal polyps.

Age- and anatomy-related values of blood-brain barrier permeability measured by perfusion-CT in non-stroke patients

BACKGROUND AND PURPOSE

The goal of this study was to determine blood-brain barrier permeability (BBBP) values extracted from perfusion-CT (PCT) using the Patlak model and possible variations related to age, gender, race, vascular risk factors and their treatment and anatomy in non-stroke patients.

MATERIALS AND METHODS

We retrospectively identified 96 non-stroke patients who underwent a PCT study using a prolonged acquisition time up to 3 minutes. Patients' charts were reviewed for demographic data, vascular risk factors and their treatment. The Patlak model was applied to calculate BBBP values in regions of interest drawn within the basal ganglia and the gray and white matter of the different cerebral lobes. Differences in BBBP values were analyzed using a multivariate analysis considering clinical variables and anatomy.

RESULTS

Mean absolute BBBP values were 1.2 ml 100 g(-1) min(-1) and relative BBBP/CBF values were 3.5%. Statistical differences between gray and white matter were not clinically relevant. BBBP values were influenced by age, history of diabetes and/or hypertension and aspirin intake.

CONCLUSION

This study reports ranges of BBBP values in non-stroke patients calculated from delayed phase PCT data using the Patlak model. These ranges will be useful to detect abnormal BBBP values when assessing patients with cerebral infarction for the risk of hemorrhagic transformation.

Accuracy and anatomical coverage of perfusion CT assessment of the blood-brain barrier permeability: one bolus versus two boluses

PURPOSE

To assess whether blood-brain barrier permeability (BBBP) values, extracted with the Patlak model from the second perfusion CT (PCT) contrast bolus, are significantly lower than the values extracted from the first bolus in the same patient.

MATERIALS AND METHODS

125 consecutive patients (29 with acute hemispheric stroke and 96 without stroke) who underwent a PCT study using a prolonged acquisition time up to 3 min were retrospectively identified. The Patlak model was applied to calculate the rate of contrast leakage out of the vascular compartment. Patlak plots were created from the arterial and parenchymal time enhancement curves obtained in multiple regions of interest drawn in ischemic brain tissue and in nonischemic brain tissue. The slope of a regression line fit to the Patlak plot was used as an indicator of BBBP. Square roots of the mean squared errors and correlation coefficients were used to describe the quality of the linear regression model. This was performed separately for the first and the second PCT bolus. Results from the first and the second bolus were compared in terms of BBBP values and the quality of the linear model fitted to the Patlak plot, using generalized estimating equations with robust variance estimation.

RESULTS

BBBP values from the second bolus were not lower than BBBP values from the first bolus in either nonischemic brain tissue [estimated mean with 95% confidence interval: 1.42 (1.10-1.82) ml x 100 g(-1) x min(-1) for the first bolus versus 1.64 (1.31-2.05) ml x 100 g(-1) x min(-1) for the second bolus, p = 1.00] or in ischemic tissue [1.04 (0.97-1.12) ml x 100 g(-1) x min(-1) for the first bolus versus 1.19 (1.11-1.28) ml x 100 g(-1)min(-1) for the second bolus, p = 0.79]. Compared to regression models from the first bolus, the Patlak regression models obtained from the second bolus were of similar or slightly better quality. This was true both in nonischemic and ischemic brain tissue.

CONCLUSION

The contrast material from the first bolus of contrast for PCT does not negatively influence measurements of BBBP values from the second bolus. The second bolus can thus be used to increase anatomical coverage of BBBP assessment using PCT.

Dynamic perfusion CT assessment of the blood-brain barrier permeability: first pass versus delayed acquisition

BACKGROUND AND PURPOSE

The Patlak model has been applied to first-pass perfusion CT (PCT) data to extract information on blood-brain barrier permeability (BBBP) to predict hemorrhagic transformation in patients with acute stroke. However, the Patlak model was originally described for the delayed steady-state phase of contrast circulation. The goal of this study was to assess whether the first pass or the delayed phase of a contrast bolus injection better respects the assumptions of the Patlak model for the assessment of BBBP in patients with acute stroke by using PCT.

MATERIALS AND METHODS

We retrospectively identified 125 consecutive patients (29 with acute hemispheric stroke and 96 without) who underwent a PCT study by using a prolonged acquisition time up to 3 minutes. The Patlak model was applied to calculate BBBP in ischemic and nonischemic brain tissue. Linear regression of the Patlak plot was performed separately for the first pass and for the delayed phase of the contrast bolus injection. Patlak linear regression models for the first pass and the delayed phase were compared in terms of their respective square root mean squared errors (square root MSE) and correlation coefficients (R) by using generalized estimating equations with robust variance estimation.

RESULTS

BBBP values calculated from the first pass were significantly higher than those from the delayed phase, both in nonischemic brain tissue (2.81 mL x 100 g(-1) x min(-1) for the first pass versus 1.05 mL x 100 g(-1) x min(-1) for the delayed phase, P < .001) and in ischemic tissue (7.63 mL x 100 g(-1) x min(-1) for the first pass versus 1.31 mL x 100 g(-1) x min(-1) for the delayed phase, P < .001). Compared with regression models from the first pass, Patlak regression models obtained from the delayed data were of better quality, showing significantly lower square root MSE and higher R.

CONCLUSION

Only the delayed phase of PCT acquisition respects the assumptions of linearity of the Patlak model in patients with and without stroke.

Molecular weight of Hydroxyethyl Starch Molecules influences Coagulation Profile measured by thrombelastography

Background

The consensus about the ideal intravenous fluid in trauma patients remains open. However, hypertonic saline and hydroxyethyl starch (HES) seems to have advantages in terms of immuno-modulatory and haemodynamic effects. Nevertheless clotting abnormalities are frequently reported in association with the use of HES. We investigated the influence of light, medium and heavy molecular weight (MW) hydroxyethyl starch (HES) on coagulation in 29 healthy subjects.

Methods

Ringer’s lactate (RL) served as a control solution. Thrombelastography using Haemoscope’s Thrombelastograph® (TEG®) hemostasis system was used to assess the effect of HES polymers and RL. TEG analysis was performed using recalcified native whole blood both with and without the addition of platelet activating factor IV (PAF IV) before and immediately after infusion of one of the solutions.

Results

Infusion of RL or one of the three HES solutions exerts an anticoagulant effect as demonstrated by a increase in clot formation time (R) and a decrease in maximum amplitude (MA), and the angle. The addition of PAF IV reversed these changes.

Conclusions

This data indicate clear evidence of platelet activity per se or platelet interaction with the plasmatic coagulation system. Key words: Coagulation, thrombelastography, platelets, hydroxyethyl-starch

Molecular weight of hydroxyethyl starch molecules influences coagulation profile measured by thrombelastography

BACKGROUND

The consensus about the ideal intravenous fluid in trauma patients remains open. However, hypertonic saline and hydroxyethyl starch (HES) seems to have advantages in terms of immuno-modulatory and haemodynamic effects. Nevertheless clotting abnormalities are frequently reported in association with the use of HES. We investigated the influence of light, medium and heavy molecular weight (MW) hydroxyethyl starch (HES) on coagulation in 29 healthy subjects.

METHODS

Ringer's lactate (RL) served as a control solution. Thrombelastography using Haemoscope's Thrombelastograph (TEG) hemostasis system was used to assess the effect of HES polymers and RL. TEG analysis was performed using recalcified native whole blood both with and without the addition of platelet activating factor IV (PAF IV) before and immediately after infusion of one of the solutions.

RESULTS

Infusion of RL or one of the three HES solutions exerts an anticoagulant effect as demonstrated by a increase in clot formation time (R) and a decrease in maximum amplitude (MA), and the angle. The addition of PAF IV reversed these changes.

CONCLUSIONS

This data indicate clear evidence of platelet activity per se or platelet interaction with the plasmatic coagulation system.

Spin chains and channels with memory

In most studies of quantum channels, it is assumed that the errors in each use of the channel are independent. However, recent investigations of the effect of memory or correlations in error have led to speculation that nonanalytic behavior may occur in the capacity. Motivated by these observations, we connect the study of channels with correlated error to the study of many-body systems. This enables us to use many-body theory to solve some interesting models of correlated error. These models can display nonanalyticities analogous to quantum phase transitions.

Length of insertion for pulmonary artery catheters to locate different cardiac chambers in patients undergoing cardiac surgery

BACKGROUND

Although, guidelines related to length of insertion of a pulmonary artery catheter to reach a particular cardiac chamber are available, these are not backed by clinical studies. We measured the length of insertion of pulmonary artery catheters to locate the right ventricle, pulmonary artery and pulmonary capillary wedge positions in 300 adult patients undergoing elective cardiac surgery.

METHODS

The pulmonary artery catheters were inserted using a standard technique through the right internal jugular vein. The right ventricle, pulmonary artery and wedge position of the catheter were confirmed by the characteristic waveforms, and the length of insertion to these points was measured.

RESULTS

The right ventricle was reached at 24.6 (3) cm (95% CI 24.2-24.9 cm), pulmonary artery at 36 (4) cm (95% CI 35.6-36.5 cm) and wedge position at 42.8 (5.7) cm (95% CI 42.2-43.5 cm). The length of catheter to reach the right ventricle, pulmonary artery and wedge position was significantly more in patients undergoing valve surgery as compared with those undergoing coronary artery bypass grafting [26 (3.8) and 24 (2.5) cm; 38.5 (4.6) and 35 (3.2) cm; and 47.8 (6.9) and 41.2 (4.1) cm, respectively, P<0.001]. The length of insertion to reach pulmonary artery and pulmonary capillary wedge position was directly related to height of the patient (Pearson's correlation 0.157 and 0.15, respectively).

CONCLUSIONS

We have provided the norms related to length of insertion of pulmonary artery catheter, which should be useful in accurate placement of the catheter and minimize complications related to coiling of the catheter.

Bounds on multipartite entangled orthogonal state discrimination using local operations and classical communication

We show that entanglement guarantees difficulty in the discrimination of orthogonal multipartite states locally. The number of pure states that can be discriminated by local operations and classical communication is bounded by the total dimension over the average entanglement. A similar, general condition is also shown for pure and mixed states. These results offer a rare operational interpretation for three abstractly defined distancelike measures of multipartite entanglement.

Asymptotic relative entropy of entanglement

We present an analytical formula for the asymptotic relative entropy of entanglement with respect to positive partial transpose states for Werner states of arbitrary dimension. We then demonstrate its validity using methods from convex optimization. This is the first case in which the asymptotic value of a subadditive entanglement measure has been calculated.

Comparison of the effects of a cell saver and low-dose aprotinin on blood loss and homologous blood usage in patients undergoing valve surgery

OBJECTIVE

To compare 2 important techniques of blood conservation, use of a cell saver and low-dose aprotinin, in terms of blood loss and homologous blood usage in patients undergoing cardiac valve surgery.

DESIGN

Prospective, randomized.

SETTING

Tertiary care hospital.

PARTICIPANTS

Sixty adult patients undergoing elective valve surgery.

INTERVENTIONS

The patients were divided into 3 groups of 20 each. In group 1, aprotinin in the dose of 30,000 KIU/kg was added to the pump prime, with a further dose of 15,000 KIU/kg added at the end of each hour of cardiopulmonary bypass. In group 2, a cell-saver system was used to collect all blood at the operation site for processing in preparation for subsequent reinfusion. Group 3 patients acted as a control group and underwent routine management, which included collection of autologous blood during the pre-cardiopulmonary bypass period. A hemoglobin of <8 g/dL was considered as an indication for bank blood transfusion in the postoperative period.

MEASUREMENTS AND MAIN RESULTS

The chest tube drainage was significantly less in group 1 compared with groups 2 and 3, with total drainage (median [interquartile range]) amounting to 250 mL [105 to 325 mL] vs. 700 mL [525 to 910 mL] in group 2 and 800 mL [650 to 880 mL] in group 3 (p < 0.001). The patients in groups 1 and 2 required significantly less bank blood (median [interquartile range]) as compared with group 3 (350 mL [0 to 525 mL], 350 mL [0 to 350 mL], and 1050 mL [875 to 1050 mL]; p < 0.001), respectively. Cell saver provided 410 +/- 130 mL of hemoconcentrated blood in group 2. The average preoperative hemoglobin concentration was 11.3 g/dL, and it was around 9 g/dL on the 7th postoperative day. The hemoglobin concentration at various stages during hospitalization in all 3 groups was similar.

CONCLUSIONS

Low-dose aprotinin and a cell saver are effective and comparable methods of blood conservation. Aprotinin helps by decreasing the postoperative drainage, and a cell saver helps by making the patient's own blood available for transfusion.

Pharmacologic support of circulation in patients undergoing cardiac surgery

Maintenance of adequate oxygen balance to all tissues is one of the primary objectives when dealing with patients undergoing cardiac surgery. Cardiac output is one of the major components of oxygen delivery so that its maintenance is an important consideration. Due to pre-operative cardiac lesion and myocardial dysfunction secondary to the events related to cardiac surgery and cardiopulmonary by-pass, circulatory support by pharmacological or mechanical means is frequently required after surgery. Therefore, inotropes and vasodilators are used to improve the myocardial performance after cardiac surgery. Epinephrine, dopamine and dobutamine are commonly used inotropes. Dopexamine and phosphodiesterase inhibitors such as amrinone, milrinone and enoximone are some of the newer agents that have been introduced in clinical practice. Amongst the vasodilators, sodium nitroprusside and nitroglycerin are commonly used. Alpha adrenergic blockers such as phentolamine and phenoxybenzamine and calcium channel blockers such as diltiazem are some other vasodilators that can be used. Many units still regard epinephrine as an inotrope of choice and use its predominant beta agonist effect in the dose range of 0.02 to 0.04 mg/kg/minute. Some prefer dobutamine and others a combination of inotrope and vasodilator or an inodilator. Phosphodiesterase inhibitors can be useful in certain situations such as pre-existing ventricular dysfunction or when stunning of the myocardium is suspected with down regulation of beta receptors. Dopamine is useful in the renal vasodilating dose to improve renal perfusion and improve output. There is no ideal inotrope at present and each one has its own drawbacks. The clinician must learn to use the inotropes (especially the newer ones) based on his own clinical experience.

Cuspal fracture resistance and microleakage of glass ionomer cements in primary molars

The purpose of this study was to comparatively evaluate the fracture strength and microleakage among two new glass ionomer cements. One hundred and ten primary first and second molars were divided into two main groups, one group comprising of fifty teeth for microleakage evaluation and second group of sixty for fracture strength. Both the groups were subdivided equally into four based on the materials used, consisting of the same number of teeth respectively, in each. Two controls were used for cuspal fracture strength and an additional trial group was added for microleakage evaluation. Ideal, non retentive class II were prepared and restored with amalgam, Fuji IX, Fuji II LC and Vitremer. The statistical analysis revealed highly significant differences among all the subgroups except between subgroup 3 vs. 4 and 1 for cuspal fracture strength. Similarly microleakage values were significantly different statistically among sub groups 4 and 5 vs 1 and 2. The present study revealed that Fuji IX was the best in terms of cuspal fracture strength and Fuji IX with an additional application of light cure resin sealant gave the lowest degree of microleakage.

Familial coarctation of the aortic arch with bilateral ptosis: a new syndrome?

A Newfoundland family with the apparently unique syndrome of complex coarctation of the aortic arch, bilateral stenoses of the subclavian arteries, bilateral ptosis, sensorineural deafness, and bronchial asthma is reported. This syndrome appears to have affected at least four generations, and has the characteristics of autosomal dominant inheritance.

Echocardiography of intracardiac filling defects in infants and children

Intracardiac masses are rare in infants and children. Early detection is essential to their successful management. We present seven patients in whom echocardiography established the diagnosis and was crucial in the management. Three of the masses were primary cardiac tumors and four were thrombi. Patient 1: an infant with a calcified left ventricular fibroma. Patient 2: a neonate who presented with cyanosis due to obstruction of the right ventricular inflow tract by a fibroblastic tumor. Patient 3: an infant with a right atrial myxoma presenting as sepsis. Patient 4: a child who had a pulmonary embolus after a pulmonary valvotomy and was found to have a right ventricular thrombus. Patient 5: a child with a right atrial thrombus following a Fontan procedure for univentricular atrioventricular connection. Patient 6: a child with a left ventricular thrombus due to a dilated cardiomyopathy in association with epidermolysis bullosa. Patient 7: An infant with bilateral lobar emphysema, an aorticopulmonary window with left ventricular fibroelastosis, who developed a left ventricular thrombus.

Successful one stage repair of coarctation of the aorta and aneurysm of the ascending aorta in a child

A case of successful one stage repair of aortic coarctation and ascending aortic aneurysm in a 10 year old boy with cystic medial necrosis and congenital bicuspid aortic valve is presented. The patient underwent correction of both lesions at the same operation through two separate incisions. The coarctation was repaired first. The aortic valve was found to be hemodynamically normal and was not replaced. The patient leads a normal life at five and half years after operation and is playing ice hockey with no limitations. Close and long-term follow-up is considered essential in view of the potential complications.

Left-axis deviation in otherwise healthy children

Eighteen children with left-axis deviation on the electrocardiogram but no other detected abnormalities are reported. The implications of this finding are discussed. No untoward cardiovascular events occurred during the period of follow-up and the electrocardiographic findings remained unchanged. This, therefore, appears to be a benign entity in the short term but longer follow-up is required before the ultimate prognosis can be determined.

A simplified parameter of LV function in children with congenital heart disease

Myocardial state in children with various congenital anomalies was assessed using pressured-derived parameters; Vmax and Vpm. A simple approach, based on the relationship between peak value of the first derivative of the left ventricular isovolumic pressure and its associated total pressure, for identifying patients with impaired myocardial function is described.