OBJECTIVE: To test a prediction model for sleep apnea based on clinical and sociodemographic variables in a population suspected of having sleep disorders and submitted to polysomnography. METHODS: We included 323 consecutive patients submitted to polysomnography because of the clinical suspicion of having sleep disorders. We used a questionnaire with sociodemographic questions and the Epworth sleepiness scale. Blood pressure, weight, height, and SpO2 were measured. Multiple linear regression was used in order to create a prediction model for sleep apnea, the apnea-hypopnea index (AHI) being the dependent variable. Multinomial logistic regression was used in order to identify factors independently associated with the severity of apnea (mild, moderate, or severe) in comparison with the absence of apnea. RESULTS: The prevalence of sleep apnea in the study population was 71.2%. Sleep apnea was more prevalent in men than in women (81.2% vs. 56.8%; p < 0.001). The multiple linear regression model, using log AHI as the dependent variable, was composed of the following independent variables: neck circumference, witnessed apnea, age, BMI, and allergic rhinitis. The best-fit linear regression model explained 39% of the AHI variation. In the multinomial logistic regression, mild apnea was associated with BMI and neck circumference, whereas severe apnea was associated with age, BMI, neck circumference, and witnessed apnea. CONCLUSIONS: Although the use of clinical prediction models for sleep apnea does not replace polysomnography as a tool for its diagnosis, they can optimize the process of deciding when polysomnography is indicated and increase the chance of obtaining positive polysomnography findings.
OBJECTIVE: To test a prediction model for sleep apnea based on clinical and sociodemographic variables in a population suspected of having sleep disorders and submitted to polysomnography. METHODS: We included 323 consecutive patients submitted to polysomnography because of the clinical suspicion of having sleep disorders. We used a questionnaire with sociodemographic questions and the Epworth sleepiness scale. Blood pressure, weight, height, and SpO2 were measured. Multiple linear regression was used in order to create a prediction model for sleep apnea, the apnea-hypopnea index (AHI) being the dependent variable. Multinomial logistic regression was used in order to identify factors independently associated with the severity of apnea (mild, moderate, or severe) in comparison with the absence of apnea. RESULTS: The prevalence of sleep apnea in the study population was 71.2%. Sleep apnea was more prevalent in men than in women (81.2% vs. 56.8%; p < 0.001). The multiple linear regression model, using log AHI as the dependent variable, was composed of the following independent variables: neck circumference, witnessed apnea, age, BMI, and allergic rhinitis. The best-fit linear regression model explained 39% of the AHI variation. In the multinomial logistic regression, mild apnea was associated with BMI and neck circumference, whereas severe apnea was associated with age, BMI, neck circumference, and witnessed apnea. CONCLUSIONS: Although the use of clinical prediction models for sleep apnea does not replace polysomnography as a tool for its diagnosis, they can optimize the process of deciding when polysomnography is indicated and increase the chance of obtaining positive polysomnography findings.
Authors: Paulo Henrique Godoy; Ana Paula Cassetta Dos Santos Nucera; Andressa de Paiva Colcher; Jéssica Escorcio de Andrade; Davi da Silveira Barroso Alves Journal: Sleep Sci Date: 2022 Apr-Jun
Authors: Patricia A Espinoza López; Kelly Jéssica Fernández Landeo; Rodrigo Ricardo Pérez Silva Mercado; Jesús José Quiñones Ardela; Rodrigo M Carrillo-Larco Journal: Wellcome Open Res Date: 2021-01-26
Authors: Mukoso N Ozieh; Kinfe G Bishu; Rebekah J Walker; Jennifer A Campbell; Leonard E Egede Journal: BMC Health Serv Res Date: 2016-10-18 Impact factor: 2.655