Welcome to the 9th EAPS Congress Programme Scheduling

Background and Aims

Although severe acute COVID-19 is rare in children, the disease is temporally associated with the novel post-infectious condition multisystem inflammatory syndrome in children (MIS-C). The aim of the study was to assess risk factors for MIS-C with the aim to identify vulnerable children.

Methods

A register-based cohort study including all children and adolescents <19 years born in Sweden between March, 2001- December, 2020 was performed. Data on sociodemographic risk factors and comorbidities were retrieved from national health and population registers. The outcome was MIS-C diagnosis according to the Swedish Pediatric Rheumatology Quality Register during March 1, 2020 – December 8, 2021. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated using Cox regression analysis. Incidence rates per 100 000 person-years were calculated assuming a Poisson distribution.

Results

Among 2 117 443 children included in the study, 253 children developed MIS-C, corresponding to an incidence rate of 6.8 (95% CI: 6.0-7.6) per 100 000 person-years. Male sex (HR 1.65, 95% CI: 1.28-2.14), age 5-11 years (adjusted HR 1.44, 95% CI: 1.06-1.95 using children 0-4 years as reference), non-Swedish origin of parents (HR 2.53, 95% CI: 1.93-3.34), asthma (aHR 1.49, 95% CI: 1.00-2.20), obesity (aHR 2.15, 95% CI: 1.09-4.25) and life-limiting conditions (aHR 3.10, 95% CI: 1.80-5.33) were associated with MIS-C. Children 16-18 years had a reduced risk for MIS-C (aHR 0.45, 95% CI: 0.24-0.85).

Conclusions

Knowing these risk populations might facilitate identification of children with MIS-C and potentially guide targeted public health interventions. Nevertheless, the absolute risks for MIS-C were very low.

Background and Aims

Administration of corticosteroid to decrease extubation failure during the neonatal period in extremely preterm (EPT) infants (GA<28+0) is still controversial. Dexamethasone has been associated with impaired cognitive development and cerebral palsy. Betamethasone is used as a rescue treatment according to Swedish guidelines. Long-term adverse effects of betamethasone have not been evaluated in previous studies.

Methods

A prospective cohort study, including all EPT infants (22+0-26+6) in Sweden born 2004-2007 (the EXPRESS trial). In total 441 children completed 6,5 years follow up, of those 115 children were treated with betamethasone and 314 children were not treated. Children treated with other corticosteroids (hydrocortisone and prednisolone, n=12) were excluded. The primary outcome was neurodevelopment impairment (NDI) at 6,5 years of age, defined as a composite of cerebral palsy, and impairment in cognition, neuromotor function, hearing and vision. Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were estimated using Logistic regression.

Results

Children treated with betamethasone were more likely to have NDI at 6.5 years of age (49% of children treated compared with 26% of children not treated, p<0.001). Treatment with betamethasone was associated with 1.9-fold higher odds of NDI at 6.5 years of age (aOR=1.9, 95% ci 1.1-3.2). The results remained similar after matching with propensity scores on background factors with near identical distribution of confounders (40% vs. 24%, p=0.004).

Conclusions

Among extremely preterm infants’ postnatal treatment with betamethasone is associated with increased risk of neurodevelopment impairment at 6,5 years of age.

Background and Aims

Preterm infants are at higher risk of developing iron deficiency. Thus, this study aims to investigate iron intake during complementary feeding (CF) in very low birth weight (VLBW) infants.

Methods

This is a secondary outcome analysis of a randomized, intervention trial in VLBW infants. Infants were randomized to an early or late CF group (introduction between 10^th-12^th or 16^th-18^th week corrected for term) and, in addition to formula or breastfeeding, fed a standardized CF concept. Iron intake was assessed using monthly 3-day dietary records from 3 until 12 months (M3-M12) corrected for term. Infants received 2-3 mg/kg/d iron supplementation until meat was fed on a regular basis. Iron intake was compared with mixed-effects models accounting for possible correlations between siblings and other covariates (e.g. sex).

Results

Dietary records could be assessed in 80% (71/89) of infants in the early and 72% (63/88) in the late group. There was no difference in mean dietary iron intake between groups. However, breastfed infants had significantly lower mean dietary iron intake in M3-M8. Iron supplementation compensated this effect, however there was still a trend towards lower total iron intake in breastfed infants. ESPGHAN iron intake recommendations (2-3 mg/kg/d) were not met in M8-M12 regardless of feeding type even though meat was already introduced.

Conclusions

Dietary iron intake was significantly different between breastfed and formula-fed infants and iron reference values were not met from M8 onwards. Thus, prolonged iron supplementation regardless of feeding type and higher iron supplementation in breastfed infants should be considered.

Background and Aims

Background

Down Syndrome is associated with high mortality in India, due to non-diagnosis/ late-diagnosis caused by unavailability of qualified doctors and/ or lack of access to expensive medical/ diagnostic facilities – especially in rural India. Using AI/ML graphical pattern recognition tools, relevant facial points can be extracted from children’s photographs, facial anomalies can be identified, and probability of Down Syndrome affliction can be predicted.

Objective

The objectives of this research are to assess the suitability of AI/ ML models for first-level screening of Down Syndrome and to assess the accuracy of Race-specific AI/ML models v/s a Unified AI/ML model (encompassing all Races).

Methods

Trained Google’s Cloud Vision AutoML Image Classification model with ~2000 photographs of Down Syndrome positive children and ~3000 photographs of Down Syndrome negative children.

Used a subset of 300 images, 100 each of Asian, Caucasian and Other-Race children to train and test three Race Specific Models. Compared these results against a Unified Model trained and tested with the same 300 images.

Results

The CloudML model trained with ~5,000 images initially achieved: Sensitivity - 94.6%, Specificity - 96.9%, Accuracy - 96.0%. Upon optimizing confidence threshold to 0.1, model maximized Sensitivity at 99.6%, Specificity dropped to 93.8%, Accuracy maintained at 96.0%.

Each of the Race specific models trained with 100 images each, after optimization, yielded perfect scores on Sensitivity, Specificity and Accuracy of 100% each. Against this, the Unified model with 300 images yielded overall Accuracy of 98% (100% Sensitivity, 83% Specificity for Caucasian children and 100% Sensitivity, 100% Specificity for Asian/ Other children).

Conclusions

Post optimization, this model can be used as an effective post-natal screening tool for Down Syndrome detection. Preliminary results indicate that Race specific models can achieve even higher Accuracy, Sensitivity and Specificity.