Limited Additive Diagnostic Impact of Isolated Gastrointestinal Involvement for the Triage of Children with Suspected COVID-19

The strategy for the selection of patients with a suspected SARS-CoV-2 infection is relevant for the organization of a children’s hospital to provide optimal separation into COVID-19 and non-COVID-19 areas and pathways. We analyzed the proportion of children with COVID-19 presenting with gastrointestinal (GI) symptoms in 137 consecutive patients admitted between January 2020 and August 2021. GI symptoms were present as follows: diarrhea in 35 patients (26%), vomiting in 16 (12%), and both of them in five (3%); the combination of fever, respiratory symptoms, and diarrhea was observed in 16 patients (12%). Of the 676 adult patients with COVID-19 admitted to our hospital in the same time interval, 62 (9.2%) had diarrhea, 30 (4.4%) had vomiting, and 11 (1.6%) had nausea; only one patient, a 38-year-old male, presented with isolated GI symptoms at the diagnosis. Although diarrhea was observed in one quarter of cases, one-half of them had the complete triad of fever, respiratory syndrome, and diarrhea, and only five had isolated diarrhea, of which two were diagnosed with a Campylobacter infection. The occurrence of either respiratory symptoms or gastrointestinal symptoms in our patients was not related to the patient age, while younger children were more likely to have a fever. Of the 137 patients, 73 (53%) could be tested for their serum level of SARS-CoV-2 specific IgG antibodies. The observed titer ranged between 0 (n = 3) and 1729 BAU/mL (median, 425 BAU/mL). Of 137 consecutive patients with COVID-19 admitted to our referral children’s hospital, only three presented with an isolated GI manifestation. It is interesting to note that this finding turned out to be fully in keeping with what was observed on adult patients with COVID-19 in our hospital. The additive diagnostic impact of gastrointestinal involvement for the triage of children with suspected COVID-19 appears limited.

A Physician-in-the-Loop Approach by Means of Machine Learning for the Diagnosis of Lymphocytosis in the Clinical Laboratory

Context.— The goal of the lymphocytosis diagnosis approach is its classification into benign or neoplastic categories. Nevertheless, a nonnegligible percentage of laboratories fail in that classification. Objective.— To design and develop a machine learning model by using objective data from the DxH 800 analyzer, including cell population data, leukocyte and absolute lymphoid counts, hemoglobin concentration, and platelet counts, besides age and sex, with classification purposes for lymphocytosis diagnosis. Design.— A total of 1565 samples were included from 10 different lymphoid categories grouped into 4 diagnostic categories: normal controls (458), benign causes of lymphocytosis (567), neoplastic lymphocytosis (399), and spurious causes of lymphocytosis (141). The data set was distributed in a 60-20-20 scheme for training, testing, and validation stages. Six machine learning models were built and compared, and the selection of the final model was based on the minimum generalization error and 10-fold cross validation accuracy. Results.— The selected neural network classifier rendered a global 10-class classification validation accuracy corresponding to 89.9%, which, considering the aforementioned 4 diagnostic categories, presented a diagnostic impact accuracy corresponding to 95.8%. Finally, a prospective proof of concept was performed with 100 new cases with a global diagnostic accuracy corresponding to 91%. Conclusions.— The proposed machine learning model was feasible, with a high benefit-cost ratio, as the results were obtained within the complete blood count with differential. Finally, the diagnostic impact with high accuracies in both model validation and proof of concept encourages exploration of the model for real-world application on a daily basis.

Evaluation of diagnostic ultrasound use in a breast cancer detection strategy in Northern Peru

To evaluate the diagnostic impact of point-of-care breast ultrasound by trained primary care physicians (PCPs) as part of a breast cancer detection program using clinical breast exam in an underserved region of Peru. Medical records and breast ultrasound images of symptomatic women presenting to the Breast Cancer Detection Model (BCDM) in Trujillo, Peru were collected from 2017–2018. Performance was measured against final outcomes derived from regional cancer center medical records, fine needle aspiration results, patient follow-up (sensitivity, specificity, positive, and negative predictive values), and by percent agreement with the retrospective, blinded interpretation of images by a fellowship-trained breast radiologist, and a Peruvian breast surgeon. The diagnostic impact of ultrasound, compared to clinical breast exam (CBE), was calculated for actual practice and for potential impact of two alternative reporting systems. Of the 171 women presenting for breast ultrasound, 23 had breast cancer (13.5%). Breast ultrasound used as a triage test (current practice) detected all cancer cases (including four cancers missed on confirmatory CBE). PCPs showed strong agreement with radiologist and surgeon readings regarding the final management of masses (85.4% and 80.4%, respectively). While the triage system yielded a similar number of biopsies as CBE alone, using the condensed and full BI-RADS systems would have reduced biopsies by 60% while identifying 87% of cancers immediately and deferring 13% to six-month follow-up. Point-of-care ultrasound performed by trained PCPs improves diagnostic accuracy for managing symptomatic women over CBE alone and enhances access. Greater use of BI-RADS to guide management would reduce the diagnostic burden substantially.