Desempeño del sistema de vigilancia colombiano durante la pandemia de COVID-19: evaluación rápida de los primeros 50 días

"Won't get fooled again": statistical fault detection in COVID-19 Latin American data

BACKGROUND

Claims of inconsistency in epidemiological data have emerged for both developed and developing countries during the COVID-19 pandemic.

METHODS

In this paper, we apply first-digit Newcomb-Benford Law (NBL) and Kullback-Leibler Divergence (KLD) to evaluate COVID-19 records reliability in all 20 Latin American countries. We replicate country-level aggregate information from Our World in Data.

RESULTS

We find that official reports do not follow NBL's theoretical expectations (n = 978; chi-square = 78.95; KS = 4.33, MD = 2.18; mantissa = .54; MAD = .02; DF = 12.75). KLD estimates indicate high divergence among countries, including some outliers.

CONCLUSIONS

This paper provides evidence that recorded COVID-19 cases in Latin America do not conform overall to NBL, which is a useful tool for detecting data manipulation. Our study suggests that further investigations should be made into surveillance systems that exhibit higher deviation from the theoretical distribution and divergence from other similar countries.

Desempeño del sistema de vigilancia COVID-19 en Honduras 2020: Análisis de la ley de Benford

El COVID-19 es una enfermedad infecciosa responsable hasta la fecha de 6,369,703 muertes a nivel mundial. Para poder controlarla es importante adaptar y reforzar los sistemas nacionales de salud y ampliar las capacidades en materia de vigilancia, siendo el reporte diario de nuevos casos y muertes por COVID-19, una de las principales herramientas para comprender y dar respuesta a la pandemia. El trabajo se centró en identificar inconsistencias en los datos reportados sobre casos y muertes confirmadas de COVID-19, por el sistema de salud hondureño durante el año 2020. Se analizaron los datos oficiales publicados entre el 11 de marzo y el 31 de diciembre de 2020, por el centro de recursos coronavirus de la Universidad Johns Hopkins, los datos fueron divididos en cuatro segmentos de análisis, el primero correspondió a los primeros 50 días de pandemia en el país, el segundo incluyó datos hasta el 24 junio, el tercero hasta el 1 diciembre y el cuarto incluyó todos los datos del 2020. Para el análisis se hizo uso de la ley de Benford, el cual es un fenómeno estadístico ampliamente usado que permite detectar datos anómalos en los sistemas de vigilancia de cada país. Hasta el 31 de diciembre el sistema de salud hondureño confirmó 121,827 casos positivos y 3,130 muertes por COVID-19 en Honduras. Respecto a la detección de datos anómalos el segundo segmento de análisis obtuvo un mejor resultado, sin embargo, se observo que el análisis acumulado correspondiente al 2020 no se ajustó a la Ley de Benford. En general, la vigilancia tuvo una mejoría en el desempeñó después de los primeros 50 días de pandemia, no obstante, los datos referentes al 2020 revelan una disconformidad con la ley de Benford, lo que se interpreta como un desempeño de notificación no adecuado. Palabras clave .COVID-19, distribuciones estadísticas, Honduras, monitoreo epidemiológico, notificación de enfermedades.

Opportunities for the use of routinely collected data for the generation of large randomized evidence in Colombia

Randomized clinical trials are the cornerstone design for the evaluation of the safety and efficacy of health interventions. Furthermore, morbidity and mortality rates could be reduced if evidence of better interventions is sought and used to inform medical practice. However, only small to moderate, yet worthwhile, effects can be expected from such interventions. Therefore, moderate random error and moderate biases must be avoided during the design, conduct and analysis of trials. Routinely collected data, such as vital statistics, hospital episode statistics and surveillance data, could be used to enhance recruitment and follow-up a large number of patients, reducing both random error and moderate biases. Here, we discuss the opportunities and challenges for the use of these data for clinical studies in Colombia.

Report From Bolsonaro's Brazil: The Consequences of Ignoring Science

Currently, the fast spread of COVID-19 is the cause of a sanitary emergency in Brazil. This situation is largely due to President Bolsonaro's denial and the uncoordinated actions between the federal and local governments. In addition, the Brazilian government has reported that it would change its method of sharing information about the pandemic. On June 6, 2020, the presentation of accumulated cases and deaths was stopped, and the Supreme Court of Brazil determined that the federal government should continue to consolidate and disseminate the accumulated figures of cases and deaths. However, doubt about the transparency of the data remained. We used data reported by the government from Situation Reports 38-209 of the World Health Organization to assess the Benford's law fulfillment as an indicator of data quality. This rapid evaluation of data quality during the ongoing COVID-19 pandemic in Brazil suggests that the Brazilian public health surveillance system had an acceptable performance at the beginning of the epidemic. Since the end of June, the quality of cumulative death data began to decrease and remains in that condition as of August 2020. A similar situation has existed since August, with the data of accumulated new cases.

Multimorbidity patterns among COVID-19 deaths: proposal for the construction of etiological models

OBJECTIVES

To describe patterns of multimorbidity among fatal cases of COVID-19, and to propose a classification of patients based on age and multimorbidity patterns to begin the construction of etiological models.

METHODS

Data of Colombian confirmed deaths of COVID-19 until June 11, 2020, were included in this analysis (n=1488 deaths). Relationships between COVID-19, combinations of health conditions and age were explored using locally weighted polynomial regressions.

RESULTS

The most frequent health conditions were high blood pressure, respiratory disease, diabetes, cardiovascular disease, and kidney disease. Dyads more frequents were high blood pressure with diabetes, cardiovascular disease or respiratory disease. Some multimorbidity patterns increase probability of death among older individuals, whereas other patterns are not age-related, or decrease the probability of death among older people. Not all multimorbidity increases with age, as is commonly thought. Obesity, alone or with other diseases, was associated with a higher risk of severity among young people, while the risk of the high blood pressure/diabetes dyad tends to have an inverted U distribution in relation with age.

CONCLUSIONS

Classification of individuals according to multimorbidity in the medical management of COVID-19 patients is important to determine the possible etiological models and to define patient triage for hospitalization. Moreover, identification of non-infected individuals with high-risk ages and multimorbidity patterns serves to define possible interventions of selective confinement or special management.

Performance of rapid IgM-IgG combined antibody tests in the occupational surveillance of COVID-19 in Colombian enterprises

Introduction: Rapid IgM-IgG combined antibody tests can play an important role in the COVID-19 surveillance by supporting the diagnosis of infection, assessing the immune response, and verifying the progress towards herd immunity. Objective: To evaluate the performance of rapid IgM-IgG combined antibody tests in COVID-19 occupational surveillance in a group of Colombian enterprises. Materials and methods: We used the occupational surveillance data from companies that had performed periodic serological tests on all personnel from the end of April to the beginning of July, 2020. Workers were organized in small groups (“social bubbles”) to prevent outbreaks and optimize surveillance. The sensitivity was estimated as if the sampling had a prospective design. We describe here the changes in serological testing through periodic rounds. Results: Data were obtained from 4,740 workers, of whom only 23 were symptomatic showing changes from IgM(-)/IgG(-) to IgM(+) and then to IgM(+)/IgG(+) and IgG(+). The sensitivity was 40.94% for IgM(+) and 47.95% for IgM(+)/IgG(+). This implies that a little less than half of the cases can be detected. Conclusion: Antibody rapid tests have a role in the diagnostic process of infection and they must be evaluated taking into account the moment of the epidemic, the type of test purchased, and the populations at risk since their results depend on the number of infections and cases. In the context of a health crisis, they can be optimized by organizing workers into “social bubbles”

Benchmarking of public health surveillance of COVID-19 in Colombia: First semester

Introduction: Public health surveillance together with good sanitary decisions is essential for the proper management of the SARS-CoV-2 pandemic. Objective: To compare the performance of Colombian departments based on the quality of the data and to build the national ranking. Materials and methods: We analyzed the accumulated cases published between March 6 and September 1, 2020, by the Instituto Nacional de Salud. To achieve comparability, the analyses considered the day the first case was diagnosed as the first analysis date for each department. The fulfillment of Benford’s law was assessed with p-values in the log-likelihood ratio or chi-square tests. The analysis was completed with the lethality observed in each department and then the performance ranking was established. Results: Bogotá and Valle del Cauca had optimal public health surveillance performance all along. The data suggest that Antioquia, Nariño, and Tolima had good containment and adequate public health surveillance after the economic opening beginning on June 1, 2020. Conclusion: We obtained the ranking of the departments regarding the quality of public health surveillance data. The best five departments can be case studies to identify the elements associated with good performance.