Ryan Nahapetian1, Graciela E Silva2, Kimberly D Vana3, Sairam Parthasarathy4, Stuart F Quan4,5. 1. Arizona Respiratory Center and Department of Medicine, Division of Pulmonary, Critical Care, Allergy, and Sleep Medicine, University of Arizona, 1501 North Campbell Avenue, Tucson, AZ, 85724, USA. rnahapetian@deptofmed.arizona.edu. 2. College of Nursing, University of Arizona, Tucson, AZ, USA. 3. College of Nursing and Health Innovation, Arizona State University, Phoenix, AZ, USA. 4. Arizona Respiratory Center and Department of Medicine, Division of Pulmonary, Critical Care, Allergy, and Sleep Medicine, University of Arizona, 1501 North Campbell Avenue, Tucson, AZ, 85724, USA. 5. Division of Sleep Medicine, Harvard Medical School, Boston, MA, USA.
Abstract
BACKGROUND: STOP-Bang is a tool for predicting the likelihood for sleep-disordered breathing (SDB). In the conventional score, all variables are dichotomous. Our aim was to identify whether modifying the STOP-Bang scoring tool by weighting the variables could improve test characteristics. METHODS: Subjects who participated in the Sleep Heart Health Study (SHHS) were included in this analysis using a derivation dataset (n = 1667) and a validation dataset (n = 4774). In the derivation dataset, each STOP-Bang variable was evaluated using linear regression against the presence of SDB (AHI > 15/h) in order to determine the coefficients that would allow variable weighting. In other models, BMI, age, and neck circumference were entered as continuous variables. The sum of the weighted dichotomous variables yielded a weighted STOP-Bang (wSTOP-Bang). The sum of the weighted-continuous variables yielded a continuous STOP-Bang (cSTOP-Bang). The wSTOP-Bang, cSTOP-Bang, and the conventional STOP-Bang scores were then applied to the validation dataset, and receiver operating characteristic (ROC) curves were constructed. RESULTS: The area under the curve (AUC) for cSTOP-Bang (0.738) was greater than the AUC for conventional STOP-Bang (0.706) and wSTOP-Bang (0.69). The sensitivities for cSTOP-Bang, STOP-Bang, and wSTOP-Bang were similar at 93.2, 93.2, and 93.3 %, respectively. The cSTOP-Bang had a higher specificity (31.8 %) than both STOP-Bang (23.2 %) and wSTOP-Bang (23.6 %). The cSTOP-Bang had a higher likelihood ratio of a positive test (1.36) than both STOP-Bang (1.21) and wSTOP-Bang (1.22). CONCLUSIONS: Modifying the STOP-Bang score by weighting the variables and using continuous variables for BMI, age, and neck circumference can maintain sensitivity while improving specificity, positive likelihood ratio, and area under the receiver operating characteristic curve.
BACKGROUND: STOP-Bang is a tool for predicting the likelihood for sleep-disordered breathing (SDB). In the conventional score, all variables are dichotomous. Our aim was to identify whether modifying the STOP-Bang scoring tool by weighting the variables could improve test characteristics. METHODS: Subjects who participated in the Sleep Heart Health Study (SHHS) were included in this analysis using a derivation dataset (n = 1667) and a validation dataset (n = 4774). In the derivation dataset, each STOP-Bang variable was evaluated using linear regression against the presence of SDB (AHI > 15/h) in order to determine the coefficients that would allow variable weighting. In other models, BMI, age, and neck circumference were entered as continuous variables. The sum of the weighted dichotomous variables yielded a weighted STOP-Bang (wSTOP-Bang). The sum of the weighted-continuous variables yielded a continuous STOP-Bang (cSTOP-Bang). The wSTOP-Bang, cSTOP-Bang, and the conventional STOP-Bang scores were then applied to the validation dataset, and receiver operating characteristic (ROC) curves were constructed. RESULTS: The area under the curve (AUC) for cSTOP-Bang (0.738) was greater than the AUC for conventional STOP-Bang (0.706) and wSTOP-Bang (0.69). The sensitivities for cSTOP-Bang, STOP-Bang, and wSTOP-Bang were similar at 93.2, 93.2, and 93.3 %, respectively. The cSTOP-Bang had a higher specificity (31.8 %) than both STOP-Bang (23.2 %) and wSTOP-Bang (23.6 %). The cSTOP-Bang had a higher likelihood ratio of a positive test (1.36) than both STOP-Bang (1.21) and wSTOP-Bang (1.22). CONCLUSIONS: Modifying the STOP-Bang score by weighting the variables and using continuous variables for BMI, age, and neck circumference can maintain sensitivity while improving specificity, positive likelihood ratio, and area under the receiver operating characteristic curve.
Authors: S Redline; M H Sanders; B K Lind; S F Quan; C Iber; D J Gottlieb; W H Bonekat; D M Rapoport; P L Smith; J P Kiley Journal: Sleep Date: 1998-11-01 Impact factor: 5.849
Authors: Robert J Farney; Brandon S Walker; Robert M Farney; Gregory L Snow; James M Walker Journal: J Clin Sleep Med Date: 2011-10-15 Impact factor: 4.062
Authors: Lucas M Donovan; Laura C Feemster; Edmunds M Udris; Matthew F Griffith; Laura J Spece; Brian N Palen; Ken He; Sairam Parthasarathy; Kingman P Strohl; Vishesh K Kapur; David H Au Journal: J Clin Sleep Med Date: 2019-01-15 Impact factor: 4.062
Authors: Jin Mou; Bethann M Pflugeisen; Brian A Crick; Paul J Amoroso; Kirk T Harmon; Stephen F Tarnoczy; S Shirley Ho; Kimberly A Mebust Journal: Sleep Breath Date: 2018-04-24 Impact factor: 2.816