Evaluación de la precisión diagnóstica del análisis automático de la poligrafía frente al análisis manual en la apnea obstructiva del sueño

Introduction: Obstructive sleep apnea (OSA) is a chronic pathology that affects more than 20% of the adult population. It is one of the main sleep disorders with great clinical, economic and social repercussions. To evaluate the impact and severity of OSA, the number of apneas and hypopneas per hour (AHI) were counted. To define that a person has OSA, they must have an AHI ≥15/h, predominantly obstructive, in a sleep study, or the presence of an AHI ≥5/h accompanied by symptoms. The diagnosis of certainty or exclusion of OSA, as well as the severity, is established with a sleep study. Polysomnography (PSG) continues to be the reference standard for the diagnosis of OSA, it covers the recording of cardiorespiratory and neurophysiological variables, which allows the analysis of the time and structure of sleep, the presence of different respiratory episodes and their repercussions. Respiratory polygraphy (RP) includes the recording of a flow sensor, respiratory effort, oxygen saturation, heart rate and also position but not EEG. There are several studies that have explored the diagnostic agreement of PR versus PSG, being a validated, useful and necessary test for the diagnosis of OSA in different clinical situations. Being more economical and accessible. When we talk about the diagnosis of OSA, it refers to establishing whether or not there is, the severity and the therapeutic decision that will greatly affect the quality of life, prognosis and day-to-day life of the patient, since it is a chronic disease. It must be taken into account that the majority of studies are carried out in a field specialized in dream interpretation, so caution must be taken when interpreting results in another field. PR equipment incorporates increasingly better developed software that allows automatic analysis of records, but the technology and algorithms used vary depending on the device, and until now the AASM continues to recommend manual analysis based on the evidence that exists today. Several studies have analyzed the agreement between automatic and manual analysis of PR recording or between automatic analysis of PR and PSG. It seems that this agreement is reached above all at higher AHIs, above 25-30, which may limit its use in clinical practice. The PR studies previously carried out were not standardized, they were applied to a non-OSA population and using different softwares. Therefore, it is important to develop a study with a large number of patients to achieve statistical significance and strong conclusions that would support normal clinical practice, and to disable a study that does not meet the scientific requirements when it comes to interpretation and reading. Objetives: - Establish the diagnostic performance of AHI measured by automatic analysis with respect to manual analysis in polygraphy and depending on the severity of OSA. - Assess the degree of agreement between the automatic analysis with respect to the manual analysis in the calculation of the AHI, average SatO2 and CT90%. - Descriptive analysis of the demographic variables of patients who undergo polygraphy. Methodology: Descriptive, observational and retrospective study in which polygraphs performed at HUSP during the years 2014 to March 2020 are recruited with the sample size is 3144 subjects First, an automatic analysis of the polygraphy was performed to obtain the variables: AHI, supine AHI, non-supine AHI, average SatO2 and Ct90 and subsequently, according to usual clinical practice, the manual analysis was performed. Polygraphy analyzes are performed with the emletta MPR 3 model. Results: The polygraphies performed progressively increased during the study period: 169 in 2014, 695 in 2015, 502 in 2016, 417 in 2017, 530 in 2018, 723 in 2019 and 114 in 2020 (until March). The most prevalent type of patient who attended the consultation and underwent the test was male (67.6%) and with an average age of 56.63+/-14.6 years. However, according to sex, with similar ages in both groups, it is slightly higher in the group of women. This age difference was statistically significant. From the automatic analysis of the PR, a mean AHI of 17.7 +/- 14.6 (moderate OSA) was obtained, while in the manual analysis the mean AHI was 31.6 +/-36.24 (severe OSA). . When the degree of OSA was analyzed with the automatic analysis, OSA was ruled out in 29.2% of the subjects, 29.4% were diagnosed with mild OSA, 20.1% moderate, 13.5% severe, and 7.8% very serious. With manual analysis the results were different. OSA was ruled out in 13.9% of subjects, and mild OSA was diagnosed in 22.7%, moderate in 22.5%, severe in 18.1%, and very severe in 19.2%. Therefore, although the automatic analysis ruled out false positives, it undervalued and consequently undertreated patients with OSA. 86.1% of the polygraphs performed by manual analysis were diagnosed with OSA, so only 1 in 8 of the polygraphs performed was not diagnostic of OSA, of which 40.0% were serious or very serious. The diagnostic frequency of OSA analyzed by sex was 87.8% in men (42.4% severe or very severe). On the other hand, it was 81.8% in women (28.4% is serious or very serious). Therefore, in women the severity of OSA was less than in men and more frequent of a moderate nature. The linear correlation coefficient between both types of analysis was high (p<0.001) when studying the AHI (supine and non-supine). The highest correlation was obtained with SatO2 (0.969) and to a lesser extent with CT90 (0.77). The agreement in the severity of OSA between both PR evaluations was 44.52% and the kappa index was 0.3068, indicating low/moderate agreement. When the ability to rule out or not rule out OSA and to diagnose its degree was analyzed with the automatic analysis compared to the manual analysis, the automatic analysis was sensitive (0.94), but not very specific (0.51). The positive predictive value was also low (0.46), and on the other hand, the negative predictive value was high (0.95). The sensitivity to diagnose OSA was moderate (0.78) and with very low specificity (0.46). The positive predictive value was high (0.98), and on the other hand, the negative predictive value was very low (0.49). The sensitivity to diagnose mild OSA was very low (0.31) with a specificity of 0.82, which although higher, was not adequate (0.82) either. The positive predictive value was also very low (0.55) and the negative predictive value is high (0.86). The sensitivity for diagnosing moderate OSA was very low (0.29), with a specificity of 0.88. The positive predictive value was very low (0.36, and on the other hand the negative predictive value was higher (0.84). The sensitivity to diagnose severe OSA is very low (0.29) with a specificity of 0.89. The positive predictive value was very low (0.37) and the negative predictive value was, on the other hand, higher (0.84).And finally, the sensitivity to diagnose very severe OSA was very low (0.39) with a specificity of 0.99, while the positive predictive value was very high (0.97) and the negative predictive value was also high, although to a lesser extent (0.87). Conclusions: Compared the automatic PR analysis with the manual analysis, the automatic analysis is not valid in the diagnosis of OSA, it only has diagnostic capacity in the case of no OSA or in the case of very severe OSA, but not in intermediates OSA diagnoses. Automatic analysis of PR is not recommended in the definitive diagnosis of OSA. It is only recommended as a screening method to later demonstrate the diagnosis with a polygraph analysis with the manual method performed by healthcare personnel expert in sleep.

Grantor dc:publisher

Universidad de La Rioja (España)

Year dc:date

2024

Author dc:creator

• Roncero Lázaro, Alejandra

Contributors dc:contributor

• García Pichel, José Manuel (Universidad de La Rioja)
• Ruiz Martínez, Carlos (Universidad de La Rioja)

Statement dc:rights

• LICENCIA DE USO: Los documentos a texto completo incluidos en Dialnet son de acceso libre y propiedad de sus autores y/o editores. Por tanto, cualquier acto de reproducción, distribución, comunicación pública y/o transformación total o parcial requiere el consentimiento expreso y escrito de aquéllos. Cualquier enlace al texto completo de estos documentos deberá hacerse a través de la URL oficial de éstos en Dialnet. Más información: https://dialnet.unirioja.es/info/derechosOAI | INTELLECTUAL PROPERTY RIGHTS STATEMENT: Full text documents hosted by Dialnet are protected by copyright and/or related rights. This digital object is accessible without charge, but its use is subject to the licensing conditions set by its authors or editors. Unless expressly stated otherwise in the licensing conditions, you are free to linking, browsing, printing and making a copy for your own personal purposes. All other acts of reproduction and communication to the public are subject to the licensing conditions expressed by editors and authors and require consent from them. Any link to this document should be made using its official URL in Dialnet. More info: https://dialnet.unirioja.es/info/derechosOAI

Language dc:language

spa

Repository record dc:identifier

https://dialnet.unirioja.es/servlet/oaites?codigo=342207

OAI identifier oai:identifier

oai:dialnet.unirioja.es:TES0000023151