Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 https://doi.org/10.1186/s12890-022-01827-0 RESEARCH ARTICLE Open Access A new tool to screen patients with severe obstructive sleep apnea in the primary care setting: a prospective multicenter study Patricia Peñacoba1,2* , Maria Antònia Llauger2,3, Ana María Fortuna1,2, Xavier Flor2,4, Gabriel Sampol2,5,6, Anna Maria Pedro‑Pijoan7, Núria Grau6,8, Carme Santiveri9, Joan Juvanteny2,10, José Ignacio Aoiz11, Joan Bayó11, Patricia Lloberes5,6 and Mercè Mayos1,2,6 on behalf of the PASHOS Working Group Abstract Background: The coordination between different levels of care is essential for the management of obstructive sleep apnea (OSA). The objective of this multicenter project was to develop a screening model for OSA in the primary care setting. Methods: Anthropometric data, clinical history, and symptoms of OSA were recorded in randomly selected primary care patients, who also underwent a home sleep apnea test (HSAT). Respiratory polygraphy or polysomnography were performed at the sleep unit to establish definite indication for continuous positive airway pressure (CPAP). By means of cross-validation, a logistic regression model (CPAP yes/no) was designed, and with the clinical variables included in the model, a scoring system was established using the β coefficients (PASHOS Test). In a second stage, results of HSAT were added, and the final accuracy of the model was assessed. Results: 194 patients completed the study. The clinical test included the body mass index, neck circumference and observed apneas during sleep (AUC 0.824, 95% CI 0.763–0.886, P < 0.001). In a second stage, the oxygen desaturation index (ODI) of 3% (ODI3% ≥ 15%) from the HSAT was added (AUC 0.911, 95% CI 0.863–0.960, P < 0.001), with a sensitiv‑ ity of 85.5% (95% CI 74.7–92.1) and specificity of 67.8% (95% CI 55.1–78.3). Conclusions: The use of this model would prevent referral to the sleep unit for 55.1% of the patients. The two-stage PASHOS model is a useful and practical screening tool for OSA in primary care for detecting candidates for CPAP treatment. Clinical Trial Registration Registry: ClinicalTrials.gov; Name: PASHOS Project: Advanced Platform for Sleep Apnea Syn‑ drome Assessment; URL: https://​clini​caltr​ials.​gov/​ct2/​show/​NCT02​591979; Identifier: NCT02591979. Date of registra‑ tion: October 30, 2015. Keywords: Obstructive sleep apnea, Continuous positive airway pressure (CPAP), Primary healthcare, Sleep unit, Screening model, Home sleep apnea test Background Obstructive sleep apnea (OSA) is a breathing disorder with high and increasing prevalence [1] and important *Correspondence: ppenacoba@santpau.cat 1 deleterious effect on the patient’s health and quality of Sleep Unit, Department of Respiratory Diseases, Hospital de La Santa Creu I Sant Pau, Universitat Autònoma de Barcelona, Barcelona, Spain life [2–5]. There is a wide consensus about the need for Full list of author information is available at the end of the article involvement of all healthcare levels especially primary © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 2 of 10 care in the management of OSA [6, 7]. The participation methodology of the study has been previously described of primary care professionals may contribute to improve in detail [22]. The PASHOS project has been approved underdiagnosis of OSA [8], which is particularly rel- by the Clinical Research Ethics Committee of the 10 evant because of conclusive evidence of the beneficial participating centers, and written informed consent was effects of continuous positive airway pressure (CPAP) obtained from all patients. The study was registered at on the overall health and prognosis of the patients [9]. ClinicalTrials.gov (identifier NCT02591979). In this context, primary care professionals play a double Between May 2015 and May 2018, men and women role. On one hand, to identify patients with severe OSA aged ≥ 18 and ≤ 75 years who visited a participating pri- who should be referred to a specialized sleep unit for mary care center for any reason, were consecutively treatment and follow-up. On the other hand, to identify included according to a randomized schedule. Exclu- patients with mild-moderate OSA who can be managed sion criteria were as follows: previous diagnosis of OSA, in primary care, thus preventing unnecessary referrals chronic insomnia (< 5 h of sleep/day), cognitive impair- and workload for sleep units. Different strategies with ment or psychophysical inability to perform the HSAT, variable results have been proposed to meet this dou- acute or unstable cardiovascular or cerebrovascular ble objective, including the use of questionnaires alone disease, neuromuscular disease, moderate-to-severe [10–13] or combined with simplified models of home chronic obstructive pulmonary disease (COPD) ­(FEV1/ sleep apnea test (HSAT) [14–16], or strategies in which FVC < 0.7% and ­FEV1 < 50% predicted), and relevant res- primary care physicians assume a central role in the care piratory comorbidity that may interfere with arterial of patients with OSA [17–21]. The ideal model probably blood saturation measurements. includes a coordinated network between different levels of care, adapted to the health care characteristics of the Study procedures and data collection region and workload of the different settings, and able to The two-stage study design included a clinical assessment be applicable to the broad spectrum of OSA phenotypes. at the primary care center and a HSAT. At the patient’s The PASHOS project (PASHOS is the Spanish acronym visit in the primary care center, the following data were of Advanced Platform for Sleep Apnea Assessment) is a recorded: sociodemographic (age and sex), anthropomet- multicenter study conceived for the implementation of ric (weight, height, body mass index [BMI], and neck, an inter-level working model, coordinated between pri- waist and hip circumference), history of relevant diseases mary care and sleep units. Previous development of the and cardiovascular risk factors, forced spirometry, clini- PASHOS project included specific training of primary cal history directed to assessment of sleep breathing dis- care professionals (physicians and registered nurses), orders, daytime sleepiness using the Epworth sleepiness establishment of inter-level network tools [22] and analy- scale [24], the Berlin questionnaire [25], OSA50 ques- sis of primary care and sleep unit agreement in manage- tionnaire [14] and the STOP-Bang sleep apnea question- ment decisions for OSA patients [23]. A moderate level naire [11]. of concordance on diagnostic classification between pri- The clinical probability of OSA based on scores of the mary care physicians and sleep specialists was found, Berlin questionnaire [25] was used to assess the preva- there was substantial agreement in patients with severe lence of patients with low and high risk of OSA. To ensure OSA who were candidates for CPAP therapy [23]. How- a balanced sample between patients with low and high ever, the overall sensitivity for detecting candidates for risk, according to the method proposed by Chai-Coetzer CPAP treatment by primary care physicians was only et al. [14] and considering an expected prevalence of 62.2% [23]. Therefore, the development of a screen- OSA of 25–30% in primary care [25], all patients in the ing tool that would help in the clinical assessment and high-risk category and, randomly, 1 out of 2 patients in improve the ability to identify patients with OSA in the the low-risk underwent a HSAT. In all cases, sleep stud- & Go® polygraph device (Bitmed, Sibelgroup, Barcelona, primary care setting seems justified. The objective of this ies were performed using a self-applied Sibelmed Screen study was to develop a two-stage screening model based on a clinical questionnaire and a HSAT, to detect patients Spain), with a 2-channel monitor: nasal cannula for air- with severe OSA who are candidates for treatment with flow measure and oximetry. The device also provided CPAP. body position. The primary care nurses assessed the quality of the HSAT and removed the periods of poor Methods signaling (artifacts or lost signal). The minimum valid Design and participants recording time was 5 h, and the sleep study was repeated This was a prospective, multicenter study with the par- if a poor signal acquisition was detected or the record- ticipation of six primary health care centers and four ing time was less than 5 h. Variables recorded included sleep units from the urban area of Barcelona (Spain). The peripheral oxygen saturation (­SpO2), falls in S­ pO2 ≥ 3% Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 3 of 10 (oxygen desaturation index—ODI3%) and ≥ 4% (ODI4%) specificity, positive and negative predictive values, over- per hour of recording, and time spent in the supine posi- all accuracy, likelihood ratio, odds ratio, and the post-test tion. A hypopnea was defined as an airflow reduction probability with the 95% CI were calculated. The cut-off of ≥ 30% but < 90% lasting at least 10 s, with a ≥ 3% drop point of maximum sensitivity and moderate specificity in oximetry, and an apnea was defined as an absence of was chosen. The same data was calculated for the results airflow or ≥ 90% reduction for ≥ 10 s. Cut-off points for of the Epworth sleepiness scale, the Berlin question- AHI were 5–15 for mild OSA, > 15 and < 30 for moderate naire, the STOP-Bang sleep apnea questionnaire, and OSA, and ≥ 30 for severe OSA. the OSA50 questionnaire. In a second analysis, the ROC All patients independent of the clinical probability of curve and the AUC were recalculated after including OSA and within 3 months after HSAT, were referred to variables collected in the HSAT, such as apnea–hypopnea the sleep unit and underwent a respiratory polygraphy index (AHI), ODI3% and ODI4%. Statistical significance or conventional polysomnography according to stand- was set at P < 0.05. Statistical analyses were performed ard practice [26, 27]. Specialists at the sleep unit with all using the statistical software IBM-SPSS version 26.0 for clinical documentation available established a final diag- Windows. nosis and therapeutic decision regarding whether or not patients were candidates for CPAP therapy. The diagno- Results sis of OSA and indication of CPAP treatment were made Characteristics of participants according to clinical practice guidelines [26, 27]. A total of 1036 patients were assessed by primary care physicians, 466 (45%) of which were eligible and agreed Statistical analysis to take part in the study. A total of 249 (53.4%) patients The sample size was estimated considering a minimum underwent HSAT. There were no significant differences prevalence of OSA of 25% in the population attended between patients with low clinical probability of OSA to in the primary care setting [25]. Assuming a loss of based on scoring of the Berlin questionnaire that were 15% at follow-up and an alpha error of 5%, a total sam- randomized to or not to perform a HSAT (Table S1, Sup- ple of 162 patients was needed to detect a sensitivity of plementary material). Finally, 194 (41.6%) patients com- 90% in the validation process. Categorical variables are pleted the second stage with a respiratory polygraphy or expressed as frequencies and percentages, and continu- conventional polysomnography at the sleep unit, and the ous variables as mean, standard deviation (SD) and 95% definitive diagnosis and therapeutic decision was estab- confidence interval (CI). Bivariate analysis included the lished. Of the total 194 patients who completed the study, chi-square test (χ2) test or the Fisher’s exact probability 126 (64.9%) were not considered candidates for CPAP test for categorical data, and the Student’s t test or the therapy, whereas the remaining 68 (35.0%) were candi- Mann–Whitney U test for continuous data according to dates for CPAP treatment. The flow chart of the study the conditions of application. population is shown in Fig. 1. The clinical characteris- Development of the screening predictive model was tics of patients in the overall study population and in the made according to a cross-validation procedure, with groups of candidates and non-candidates for CPAP ther- the whole sample available for estimation of the model apy are shown in Table 1. and further validation. Thus, all clinical variables were included in a multivariate logistic regression analysis, in Model based on clinical variables (PASHOS test) which indication of CPAP treatment (yes/no) made by After logistic regression analysis (candidate for CPAP sleep specialists was the dependent variable. Categorical treatment, yes or no) and taking into account all clinical variables of clinical questionnaires were dichotomously data, the final model included the following three vari- recodified. In order to simplify the predictive clinical ables: BMI, neck circumference, and the question “Have screening model and facilitate its applicability in daily you been told that you repeatedly stop breathing when practice, a system of stratification and scoring according asleep?”. The collinearity between these three variables to β coefficients of the logistic regression model [13] was was excluded. The AUC of the ROC curve was 0.824 developed. The variance inflation factor (VIF) was used (95% CI 0.763–0.886). Results of the logistic regression to assess multicollinearity of the model, with VIF < 5 indi- analysis and scores assigned to each variable are shown cating a complete absence of collinearity [28]. Diagnos- in Table 2. The final screening questionnaire, named the tic accuracy of clinical variables and HSAT was assessed PASHOS Test (Table S2, Supplementary material), was with the area under the curve (AUC) of the receiver oper- scored between 0 and 11 (higher scores indicated higher ating characteristic (ROC) curve. probability of CPAP treatment for OSA). The AUC of A first analysis was performed including only the clini- the ROC curve for discriminating patients who were cal variables. For each of the final scores, sensitivity, candidates or not for CPAP treatment was 0.824 (95% Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 4 of 10 Primary Care Patients evaluated (N = 1036) Patients Exclusion criteria: excluded - Age (N = 299) (N = 570) - Refused participation Patients included (N = 198) (N = 466) Withdrawals - Known OSA (N = 18) (N = 37) - Other reasons (N = 55) “Low-risk” of OSA by “High-risk” of OSA by Berlin questionnaire Berlin questionnaire (N = 279) (N = 150) “Low-risk”: no HSAT Withdrawals (N = 149) (N = 31) Patients studied with Repeated HSAT HSAT (N = 39) (N = 249) Withdrawals (n=20) Patients evaluated Sleep Unit (N = 229) Withdrawals (N = 34) Patients studied with RP Patients studied with PSG (N = 149) (N = 46) Withdrawals (N = 1) Initial “Low-risk” of OSA by Initial “High-risk” of OSA by Berlin questionnaire Berlin questionnaire (N = 85) (N = 109) Fig. 1 Flow chart of the study population (OSA: obstructive sleep apnea; HSAT: home sleep apnea test; RP: respiratory polygraphy; PSG; polysomnography) CI 0.764–0.884), almost identical to that of the regres- treatment with a sensitivity of 92.6% (95% CI 83.9–96.8) sion model. The diagnostic accuracy of the test according and a specificity of 53.2% (95% CI 44.5–61.7), so that with to different cut-off values is shown in Table 3. A cut-off a test score < 5, only 5 patients were excluded who were value of ≥ 5 classified a patient as candidate for CPAP finally candidates for CPAP treatment. Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 5 of 10 Table 1 Characteristics of patients and results of sleep studies according indication of CPAP treatment by the sleep specialist Variables All patients (n = 194) Non-candidates for CPAP Candidates for CPAP P value (n = 126) (n = 68) Men:women 106:88 56:70 50:18 < 0.001 Age, years, mean (SD) 55.9 (12) 55.1 (12.3) 57.4 (11.4) 0.202 Anthropometric variables, mean (SD) Body mass index, kg/m2 29.2 (5.1) 27.7 (4.5) 32 (5.2) < 0.001 Neck circumference, cm 38.2 (4.6) 36.7 (4.2) 41.1 (4) < 0.001 Waist circumference, cm 98.7 (13.7) 94 (12.6) 107.4 (11.2) < 0.001 Hip circumference, cm 106.6 (10.5) 103.8 (10) 111.8 (9.4) < 0.001 Waist-hip ratio 0.92 (0.11) 0.9 (0.1) 0.94 (0.12) 0.008 Comorbidities Hypertension 68 (35.1) 36 (28.6) 32 (47.1) 0.010 Diabetes mellitus 35 (18) 23 (18.3) 12 (17.6) 0.008 Dyslipidemia 70 (36.1) 44 (34.9) 26 (38.2) 0.656 Depression 18 (9.3) 10 (7.9) 8 (11.8) 0.381 Anxiety 35 (18) 23 (18.3) 12 (17.6) 0.916 Stroke 2 (1) 2 (1.6) 0 (0) 0.542 Arrhythmia 7 (3.6) 2 (1.6) 5 (7.4) 0.053 Peripheral artery disease 8 (4.1) 4 (3.2) 4 (5.9) 0.365 Hypothyroidism 12 (6.2) 7 (5.6) 5 (7.4) 0.620 Heart disease 13 (6.7) 6 (4.8) 7 (10.3) 0.141 Active smoking 44 (22.7) 27 (21.4) 17 (25) 0.847 Alcohol consumption 69 (35.9) 38 (30.4) 31 (46.3) 0.029 Spirometry, mean (SD)  ­FEV1, % 90.6 (13.8) 91.5 (14) 89 (13.5) 0.236 FVC, % 92 (14.9) 92.3 (15.1) 91.5 (14.6) 0.730 Epworth sleepiness scale, mean (SD) 7.8 (5) 7.2 (4.6) 8.8 (5.5) 0.031 Patients with low/high risk of OSA Berlin questionnaire, n = 194 85/109 65/61 20/48 STOP-Bang, n = 193 57/136 49/76 8/60 OSA50, n = 185 44/141 39/80 5/61 Epworth scale score ≥ 11, n = 193 142/51 101/24 41/27 HSAT, mean (SD) Recording time, min 414 (84.8) 418.3 (87.4) 406 (79.9) 0.334 AHI 13.7 (15.4) 7.7 (7.5) 24.9 (19.5) < 0.001 AHI supine 24.2 (23.2) 15 (16) 41.3 (24.8) < 0.001 Oxygen saturation ­(SpO2) 92.5 (3.2) 93.2 (2.2) 91.3 (4.1) < 0.001 CT90% 11.3 (19.5) 7.6 (18) 18.3 (20.6) < 0.001 ODI3% 18.5 (16.3) 10.9 (8.3) 32.7 (17.9) < 0.001 ODI4% 12.9 (14.7) 6.1 (6.1) 25.8 (17.4) < 0.001 RP or conventional PSG, mean (SD) Recording time, min 463 (58.2) 462.3 (50.8) 464.2 (70.2) 0.829 AHI 21.3 (19.7) 10.1 (7.4) 42.1 (18.5) < 0.001 AHI supine 24.3 (21.6) 12.9 (10.6) 47.2 (20) < 0.001 Oxygen saturation ­(SpO2) 93.5 (2) 94.1 (1.7) 92.2 (2) < 0.001 CT90% 7.1 (13.8) 2.9 (9.9) 16.4 (0.1) < 0.001 ODI3% 18.9 (18.2) 8.6 (6.5) 34.9 (19.1) < 0.001 ODI4% 13.3 (15.7) 5.2 (5.4) 27.6 (17.7) < 0.001 Data expressed as frequencies and percentages in parenthesis unless otherwise stated. CPAP: continuous positive airway pressure; HSAT: Home sleep apnea test; SD: standard deviation; ­FEV1: forced expiratory value in one second; FVC: forced vital capacity; AHI: apnea–hypopnea index; CT90%: cumulative percentages of time at Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 6 of 10 Table 1 (continued) saturations below 90%; ODI3%: oxygen desaturation index of 3%; ODI4%: oxygen desaturation index of 4%; RP: respiratory polygraphy; PSG: polysomnography Table 2 Results of multivariate analysis with stratification of variables and scores assigned to each variable in the PASHOS Test Variables Coefficient (β) Standard error Wald (χ2) P value Odds ratio (95% CI) Scores (PASHOS Test) Neck circumference, cm ≤ 35 Reference 0 > 35 and ≤ 41 1.472 0.545 7.300 0.007 4.359 (1.5–12.68) 3 > 41 2.466 0.605 16.621 < 0.001 11.771 (3.6–38.47) 5 BMI, kg/m2 ≤ 26 Reference 0 > 26 and ≤ 30 1.215 0.585 4.316 0.038 3.369 (1.07–10.6) 2 > 30 and ≤ 33 1.587 0.610 6.763 0.009 4.887 (1.48–16.15) 3 > 33 1.949 0.632 9.498 0.002 7.021 (2.03–24.24) 4 Breathing pauses during sleep No Reference 0 Yes 0.942 0.379 6.171 < 0.001 2.566 (1.22–5.4) 2 BMI: body mass index; CI: confidence interval Table 3 Diagnostic accuracy of the PASHOS Test according to different cut-off values PASHOS Sensitivity Specificity Positive Negative Overall Positive Negative Odds ratio Post-test Test cut-off % (95% CI) % (95% CI) predictive predictive accuracy % likelihood likelihood (95% CI) probability % value value % value % (95% CI) ratio (95% ratio (95% (95% CI) (95% CI) (95% CI) CI) CI) ≥2 98.5 11.9 37.6 93.8 42.2 1.11 0.12 9.05 31.2 (35.5– (92.1–99.7) (7.3–18.7) (30.9–44.9) (71.1–98.9) (35.5–49.3) (1.04–1.2) (0.02–0.92) (1.17–70.1) 38.8) ≥3 98.5 27.0 42.1 97.1 52.1 1.35 0.05 24.76 41.7 (39.0– (91.1–99.7) (20–35.3) (34.7–49.9) (85.5–99.5) (45.1–59.0) (1.21–1.51) (0.01–0.39) (3.3–185.4) 44.4) ≥4 95.6 46.0 48.9 95.1 63.4 1.77 (1.5–2.1) 0.1 18.48 48.4 (44.2– (87.8–98.5) (37.6–54.7) (40.5–57.3) (86.5–98.3) (56.4–69.9) (0.03–0.29) (5.8–58.2) 52.6) ≥5 92.6 53.2 51.6 93.1 67.0 1.98 0.14 13.31 51.1 (46.2– (83.9–96.8) (44.5–61.7) (42.9–60.3) (84.8–97.0) (60.1–73.2) (1.62–2.41) (0.06–0.33) (5.54–37) 56.1) ≥6 79.4 69.8 58.7 86.3 73.2 2.63 0.29 8.93 58.2 (51.0– (68.4–87.3) (61.3–77.2) (48.5–68.2) (78.3–91.6) (66.6–78.9) (1.97–3.53) (0.18–0.48) (4.46–17.9) 65.1) ≥7 70.6 77.0 62.3 82.9 74.7 3.07 0.38 8.02 61.9 (53.2– (58.9–80.1) (68.9–83.5) (51.2–73.3) (75.1–88.7) (68.2–80.3) (2.15–4.37) (0.26–0.56) (4.14–15.6) 69.8) ≥8 52.9 85.7 66.7 77.1 74.2 3.71 0.55 6.75 66.2 (54.7– (41.2–64.3) (78.5–90.8) (53.4–77.8) (69.5–83.3) (67.6–79.9) (2.29–6.01) (0.42–0.71) (3.4–13.4) 76.1) ≥9 44.1 91.3 73.2 75.2 74.7 5.05 0.61 8.25 72.8 (58.9– (32.9–55.9) (85.0–95.1) (58.1–84.3) (67.8–81.3) (68.2–80.3) (2.71–9.44) (0.49–0.76) (3.81–17.8) 83.3) ≥ 10 20.6 98.4 87.5 69.7 71.1 13.0 0.80 16.1 87.3 (61.6– (12.7–31.6) (94.4–99.6) (64.0–96.5) (62.6–75.9) (66.4–77.1) (3.04–55.4) (0.71–0.91) (3.5–73.2) 96.7) 11 14.7 99.2 90.9 68.3 69.6 18.5 0.86 21.6 90.7 (56.2– (8.2–25.0) (95.6–99,9) (62.3–98.4) (61.2–74.6) (62.8–75.6) (2.4–141.7) (0.78–0.95) (2.7–172.4) 98.7) CI: confidence interval The diagnostic accuracy of the PASHOS Test as com- questionnaire is shown in Fig. 2. The AUC of the ROC pared with results of the Epworth sleepiness scale, Berlin curve was more favorable for the PASHOS Test, although questionnaire, STOP-Bang sleep apnea test, and OSA50 differences were only statistically significant with the Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 7 of 10 Fig. 2 Diagnostic accuracy of the PASHOS Test for a cut-off value of ≥ 5 and different study questionnaires (left panel) as well as variables obtained with the home sleep apnea test (right panel) (AUC: area under the ROC curve; CI: confidence interval; AHI: apnea–hypopnea index; ODI3%: oxygen desaturation index of 3%; ODI4%: oxygen desaturation index of 4%) Epworth sleepiness scale. The sensitivity, specificity, posi- Discussion tive and negative predictive values, and overall diagnostic This study proposes a two-stage screening model for OSA accuracy for the PASHOS Test as compared to the study developed in non-selected patients attended to in the pri- questionnaires are shown in Table S3 (Supplementary mary care setting. This model includes a questionnaire material). In this comparisons, the PASHOS Test showed (PASHOS Test) based on three clinical variables (BMI, the highest sensitivity (92.6%, 95% CI 83.9–96.8) and neck circumference, breathing pauses during sleep) that negative predictive value (93.1%, 95% CI 84.8–97.0). in case of positivity (cut-off ≥ 5 points) is complemented by a HSAT. This model, easy to use in the clinical set- Model based on home sleep apnea test variables ting, has been able to identify patients at high risk of The diagnostic accuracy of variables obtained in the severe OSA who are candidates for CPAP treatment and HSAT were assessed and included in the logistic regres- sion model. As shown in Fig. 2, the AUC of the ROC curve for ODI3% (0.911, 95% CI 0.863–0.960) was Table 4 Classification of patients according to results of the slightly higher than those corresponding to the AHI and PAHOS Test alone and combined with ODI3% ODI4%. The diagnostic accuracy of the PASHOS model Variable Final indication of CPAP Total patients in relation to cut-off values of the ODI3% between ≥ 10 treatment by sleep and ≥ 30 is shown in Table S3 (Supplementary material). specialists The sensitivity of the PASHOS model was 85.5% (95% CI No Yes 74.7–92.1) and the specificity 67.8% (95% CI 55.1–78.3) with an ODI3% of ≥ 15. PASHOS Test ≥ 5 When the dependent variable in the logistic regression CPAP treatment model was an AHI ≥ 30 obtained in the final respiratory No 67 (34.5%) 5 (2.6%) 72 polygraphy or conventional polysomnography, it was Yes 59 (30.4%) 63 (32.5%) 122 shown that the model included the same predictive varia- Total patients 126 68 194 bles and very similar values of the AUC of the ROC curve PASHOS model (Test ≥ 5 and ODI3% ≥ 15) (0.819, 95% CI 0.753–0.885). CPAP treatment Based on the final decision taken by sleep specialists, No 40 (33.1%) 10 (7.4%) 50 the PASHOS Test with a cut-off value of ≥ 5 classified cor- Yes 19 (15.7%) 53 (43.8%) 72 rectly 67 patients (34.5%) who were non-candidates for Total patients 59 63 122 CPAP therapy, with a further 40 patients (20.6%) when an ODI3% of ≥ 15 was added (Table 4). CPAP: continuous positive airway pressure; ODI3%: oxygen desaturation index of 3% Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 8 of 10 should be referred to a sleep unit with a diagnostic accu- recommended for OSA screening, such as the Berlin racy higher than 80%. On the other hand, the use of the questionnaire [10], OSA50 [14], and STOP-Bang [11], model would prevent referral to the sleep unit for 55.1% and clearly higher than the Epworth sleepiness scale, of patients with low to moderate risk of OSA. These which shows a high specificity but a low sensitivity for patients could continue to be attended to in primary care, detecting OSA [12]. as a similar management model to other chronic diseases The HSAT was performed using a 2-channel monitor such as asthma or COPD. with a nasal cannula for airflow measure and oximetry. Patients were recruited in the primary care setting In agreement with data reported in the study of Chai- independent of the reason for consultation and suspicion Coetzer et al. [14], the ODI3% was the parameter with of OSA. In a second stage, the HSAT was performed in the highest diagnostic reliability. The primary care nurse all patients in the OSA high-risk category of Berlin ques- assessed the quality of sleep studies. The studies were tionnaire and, randomly, 1 out of 2 patients in the low- allowed to be repeated in case of poor signaling acquisi- risk category. However, results are unlikely to be affected tion or even discarded when invalid recording. by this randomization, since differences between patients The logistic regression model was calculated accord- randomized to or not to the sleep study were not found. ing to decision of treatment with CPAP (yes/no). How- Our approach is similar to the two-stage model devel- ever, this decision may include a subjective component so oped by Chai-Coetzer et al. [14] based on 157 patients the AUC of the ROC curve was also calculated taking the attending their primary care physician, for any reason, at AHI ≥ 30 as severe OSA obtained in the gold-standard six primary care clinics in rural and metropolitan regions sleep studies. The AUC obtained in this model was very of South Australia. The screening questionnaire in this similar, which supports the usefulness of the two-stage study included snoring, waist circumference, witnessed model for selecting candidates for CPAP treatment and apneas and age, and was followed by a HSAT. This two- patients with severe OSA according to AHI. stage diagnostic model showed a sensitivity of 97% and Strengths of the study include the multicenter design specificity of 87% [14]. In this study, as in other previ- and the fact that the inclusion of patients was not limited ous studies carried out in selected populations [15, 16], to those with a pretest high clinical probability of OSA, so the predictive capacity of the model was higher, which that the results obtained can be applicable to a large phe- may be explained by the fact that they included a HSAT notype spectrum of patients, but considering the limita- in all patients, independently of the clinical variables. In tions of the study. The limitations include the eligibility our opinion, the indication of HSAT without any previ- criterion of an upper age range of 75 years, which excludes ous clinical filter may impose a work overload for pri- a substantial percentage of patients attended to in pri- mary care, transferring the problem of waiting lists to mary care. Patients with a previous clinical diagnosis of this setting and limiting its applicability. Other models insomnia are also excluded, which may have contributed of integrated management of patients with high clinical to the underdiagnosis of patients with OSA primarily with probability of OSA in primary care have been validated, complaints of sleep disruption, and it is already known with non-inferior mid-term results to conventional man- the existence of a clinical cluster of OSA related to a high agement in specialized sleep units [17–21]. These results, prevalence of hypertension, diabetes, and cardiovascular however, although they offer evidence that non-sleep disease [30]. Also, in contrast to other studies [14–19, 21], specialists are capable of providing care to patients with the definite diagnosis was not made using polysomnogra- OSA, have been questioned for difficulties of implemen- phy in all patients [22]. Although respiratory polygraphy tation in real-world clinical practice given the work over- may underestimate the diagnosis of OSA, our model was load in primary care [29]. adapted to routine clinical practice and recommendations The two-stage screening model proposed in the pre- of current clinical practice guidelines [27]. sent study, includes a short, simple and easy to adminis- ter questionnaire (PASHOS Test) with a high sensitivity Conclusion (93%) and negative predictive value of 93% that, in prac- This two-stage screening model that includes a short clin- tice, correctly identified subjects at low risk of OSA ical questionnaire (PASHOS Test) and results obtained by in 34.5% of the study sample. The association of the a HSAT, is a useful approach in which primary care pro- PASHOS Test with ODI3% in a second stage increases fessionals have an important role in the management of the specificity and positive predictive value of the model, OSA. With a previous adequate training program and in avoiding referral to sleep specialists to further 21% of coordination with sleep specialists, unnecessary referrals patients. The PASHOS Test includes only three variables of patients who are non-candidates for CPAP with mild- but shows sensitivity and negative predictive values simi- moderate OSA could be prevented. lar or slightly higher than clinical questionnaires usually Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 9 of 10 Abbreviations Author details 1 AHI: Apnea–hypopnea index; AUC​: Area under the curve; BMI: Body mass Sleep Unit, Department of Respiratory Diseases, Hospital de La Santa Creu I index; CI: Confidence interval; COPD: Chronic obstructive pulmonary disease; Sant Pau, Universitat Autònoma de Barcelona, Barcelona, Spain. 2 Department CPAP: Continuous positive airway pressure; CT90%: Percentage of time with of Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain. 3 Equip oxygen saturation below 90%; FEV1: Forced expiratory volume in 1 s; FVC: d’Atenció Primària Encants, Centre d’Atenció Primària (CAP) Maragall, Institut Forced vital capacity; HSAT: Home sleep apnea test; ODI3%: Oxygen desatura‑ Català de La Salut, Barcelona, Spain. 4 CAP Chafarinas, Institut Català de La tion index of 3%; ODI4%: Oxygen desaturation index of 4%; OSA: Obstructive Salut, Barcelona, Spain. 5 Sleep Unit, Service of Pneumology, Hospital Universi‑ sleep apnea; PSG: Polysomnography; ROC: Receiver operating characteristic; tari Vall d’Hebron, Universitat Autònoma de Barcelona, Barcelona, Spain. 6 Cen‑ RP: Respiratory polygraphy; SD: Standard deviation; SpO2: Peripheral oxygen tro de Investigación Biomédica en Red Enfermedades Respiratorias (CIBERES), saturation; VIF: Variance inflation factor. Instituto de Investigación Carlos III, Madrid, Spain. 7 Àrea Bàsica de Salut Gaudí, Consorci Sanitari Integral, Barcelona, Spain. 8 Sleep Unit, Department of Res‑ piratory Medicine, Parc de Salut Mar-IMIM, Hospital del Mar, Barcelona, Spain. Supplementary Information 9 Service of Pneumology, Hospital Dos de Maig, Consorci Sanitari Integral, The online version contains supplementary material available at https://​doi.​ Barcelona, Spain. 10 CAP Trinitat Vella, Institut Català de La Salut, Barcelona, org/​10.​1186/​s12890-​022-​01827-0. Spain. 11 CAP El Clot, Institut Català de La Salut, Barcelona, Spain. Received: 11 October 2021 Accepted: 16 December 2021 Additional file 1: Table S1. Characteristics of patients at low risk of OSA based on scores of the Berlin questionnaire randomized or not randomized to home sleep apnea test (HSAT). Table S2. PASHOS Test. Table S3. Diagnostic accuracy of the PASHOS Test as compared with the study questionnaires and with the addition of ODI3%. References 1. Benjafield AV, Ayas NT, Eastwood PR, Heinzer R, Ip MSM, Morrell MJ, et al. Estimation of the global prevalence and burden of obstructive sleep Acknowledgements apnoea: a literature-based analysis. Lancet Respir Med. 2019;7(8):687–98. The authors thank Marta Pulido, MD, for editing the manuscript and editorial 2. Tietjens JR, Claman D, Kezirian EJ, de Marco T, Mirzayan A, Sadroonri B, assistance. et al. Obstructive sleep apnea in cardiovascular disease: a review of the PASHOS Working Group: Leandra Domínguez Olivera (Equip d’Atenció literature and proposed multidisciplinary clinical management strategy. J Primària Encants, Centre d’Atenció Primària (CAP) Maragall, Institut Català de la Am Heart Assoc. 2019;8(1):e010440. Salut); Pepi Valverde Trillo (Àrea Bàsica de Salut (ABS) Gaudí, Consorci Sanitari 3. Punjabi NM, Caffo BS, Goodwin JL, Gottlied DJ, Newman AB, O’Connor Integral); Mª Ángeles Santos Santos and Mª del Mar Farga Martínez (CAP GT, et al. Sleep-disordered breathing and mortality: prospective cohort Chafarinas, Institut Català de la Salut); Montserrat Reverté Simó (CAP Trinitat study. PLoS Med. 2009;6(8):e1000132. Vella, Institut Català de la Salut); and Núria Argemí Saburit, Casimira Medrano 4. Campos-Rodriguez F, Martinez-Garcia MA, Martinez M, Duran-Cantolla J, Medrano, and Marineus Puig García (CAP El Clot, Institut Català de la Salut), De La Penã M, Masdeu MJ, et al. Association between obstructive sleep Barcelona, Spain. apnea and cancer incidence in a large multicenter spanish cohort. Am J Respir Crit Care Med. 2013;187:99–105. Authors’ contributions 5. Lajoie AC, Lafontaine A-L, Kimoff RJ, Kaminska M. Obstructive sleep apnea Design: P.P., M.A.Ll., A.M.P.P., C.S., X.F., J.J., G.S., P. Ll., J.I.A., J.B., N.G., A.M.F., and in neurodegenerative disorders: current evidence in support of benefit M.M. Acquisition of data: P.P., M.A.Ll., A.M.P.P., C.S., X.F., J.J., G.S., P. Ll., J.I.A., J.B., from sleep apnea treatment. J Clin Med. 2020;9:297. N.G., A.M.F., and M.M. Analysis of data: P.P. and M.M.; Interpretation of data: P.P., 6. Suárez M, Osorio J, Torres M, Montserrat JM. Should the diagno‑ M.A.Ll., A.M.P.P., C.S., X.F., J.J., G.S., P. Ll., J.I.A., J.B., N.G., A.M.F., and M.M. Writing sis and management of OSA move into general practice? Breathe. the manuscript: P.P. and M.M. Approval of the version to be published and 2016;12(3):243–7. agreement to be accountable for all aspects of the work: all authors. 7. Donovan LM, Shah A, Chai-Coetzer CHL, Barbé F, Ayas NT, Kapur VK. Rede‑ signing care for OSA. Chest. 2020;157(4):966–76. Funding 8. Rotenberg BW, George CF, Sullivan KM, Wong E. Wait times for sleep The Spanish Society of Pneumology and Thoracic Surgery (SEPAR), the Catalan Apnea care in Ontario: a multidisciplinary assessment. Can Respir J. Society of Pneumology (SOCAP) and the Fondo de Investigaciones Sanitarias 2010;17(4):170–4. (FIS) (PI 14/01985) funded this study to acquire the necessary devices, to 9. Patil SP, Ayappa IA, Caples SM, John Kimoff R, Patel SR, Harrod CG. Treat‑ design and maintain the database, for statistical advice and data analysis. This ment of adult obstructive sleep apnea with positive airway pressure: an was not an industry-backed study. American academy of sleep medicine clinical practice guideline. J Clin Sleep Med. 2019;15(2):335–43. Availability of data and materials 10. Netzer NC, Stoohs RA, Netzer CM, Clark K, Strohl KP. Using the Berlin Study data are available from the corresponding author upon request. Questionnaire to identify patients at risk for the sleep apnea syndrome. Ann Intern Med. 1999;131:485–91. Declarations 11. Chung F, Abdullah HR, Liao P. Stop-bang questionnaire: a practical approach to screen for obstructive sleep apnea. Chest. 2016;149(3):631–8. Ethics approval and consent to participate 12. Miller JN, Berger AM. Screening and assessment for obstructive sleep This project was approved by the Clinical Research Ethics Committee of apnea in primary care. Sleep Med Rev 2016:41–51. the Hospital de la Santa Creu i Sant Pau. Committee’s reference number: 13. Duarte RLM, Rabahi MF, Magalhães-da-Silveira FJ, De Oliveira-E-Sá TS, IIBSP-PAS-2013–151. Written informed consent to participate in the study was Mello FCQ, Gozal D. Simplifying the screening of obstructive sleep apnea obtained from all patients. with a 2-item model, no-apnea: a cross-sectional study. J Clin Sleep Med. 2018;14(7):1097–107. Consent for publication 14. Chai-Coetzer CL, Antic NA, Rowland LS, Catcheside PG, Esterman A, Written informed consent was obtained from all patients. In this manuscript Reed RL, et al. A simplified model of screening questionnaire and there is no personal or clinical data of the participants that compromises home monitoring for obstructive sleep apnoea in primary care. Thorax. anonymity. 2011;66(3):213–9. 15. Gurubhagavatula I, Maislin G, Pack AI. An algorithm to stratify sleep apnea Competing interests risk in a sleep disorders clinic population. Am J Respir Crit Care Med. The first author and none of the other authors have any affiliations or financial 2001;164(101):1904–9. involvement that conflict with the material presented in this study. Peñacoba et al. BMC Pulmonary Medicine (2022) 22:38 Page 10 of 10 16. Mulgrew AT, Fox N, Ayas NT, Ryan CF. Diagnosis and initial management of obstructive sleep apnea without polysomnography: A randomized validation study. Ann Intern Med. 2007;146(3):157–66. 17. Antic NA, Buchan C, Esterman A, Hensley M, Naughton MT, Rowland S, et al. A randomized controlled trial of nurse-led care for symptomatic moderate-severe obstructive sleep apnea. Am J Respir Crit Care Med. 2009;179(6):501–8. 18. Sánchez-De-La-Torre M, Nadal N, Cortijo A, Masa JF, Duran-Cantolla J, Valls J, et al. Role of primary care in the follow-up of patients with obstructive sleep apnoea undergoing CPAP treatment: a randomised controlled trial. Thorax. 2015;70(4):346–52. 19. Chai-Coetzer CL, Antic NA, Rowland LS, Reed RL, Esterman A, Catcheside PG, et al. Primary care vs specialist sleep center management of obstruc‑ tive sleep apnea and daytime sleepiness and quality of life: a randomized trial. JAMA. 2013;309(10):997–1004. 20. Tarraubella N, Sánchez-De-La-Torre M, Nadal N, De Batlle J, Benítez I, Cortijo A, et al. Management of obstructive sleep apnoea in a pri‑ mary care vs sleep unit setting: a randomised controlled trial. Thorax. 2018;73(12):1152–60. 21. Ángeles Sánchez-Quiroga M, Corral J, Gómez-de-Terreros FJ, Carmona- Bernal C, Isabel Asensio-Cruz M, Cabello M, et al. Primary care physi‑ cians can comprehensively manage patients with sleep apnea a noninferiority randomized controlled trial. Am J Respir Crit Care Med. 2018;198(5):648–56. 22. Mayos M, Peñacoba P, Pijoan AMP, Santiveri C, Flor X, Juvanteny J, et al. Coordinated program between primary care and sleep unit for the management of obstructive sleep apnea. Npj Prim Care Respir Med. 2019;29(1):39. 23. Peñacoba P, Antònia Llauger M, Fortuna AM, Flor X, Sampol G, Pijoan AMP, et al. Primary care and sleep unit agreement in management decisions for sleep apnea: a prospective study in Spain. J Clin Sleep Med. 2020;16(9):1579–89. 24. Walker NA, Sunderram J, Zhang P, Lu S-E, Scharf MT. Clinical utility of the Epworth sleepiness scale. Sleep Breath. 2020;24(4):1759–65. 25. Netzer NC, Hoegel JJ, Loube D, Netzer CM, Hay B, Alvarez-Sala R, et al. Prevalence of symptoms and risk of sleep apnea in primary care. Chest. 2003;124(4):1406–14. 26. Lloberes P, Duran-Cantolla J, Martinez-Garcia MA, Marin JM, Ferrer A, Cor‑ ral J, et al. Diagnosis and treatment of sleep apnea-hypopnea syndrome. Spanish Society of Pulmonology and Thoracic Surgery. Arch Bronconeu‑ mol. 2011;47:143–56. 27. Kapur VK, Auckley DH, Chowdhuri S, Kuhlmann DC, Mehra R, Ramar K, et al. Clinical practice guideline for diagnostic testing for adult obstruc‑ tive sleep apnea: an American academy of sleep medicine clinical practice guideline. J Clin Sleep Med. 2017;13:479–504. 28. Yoo W, Mayberry R, Bae S, Singh K, Peter He Q, Lillard JW. A study of effects of multicollinearity in the multivariable analysis. Int J Appl Sci Technol. 2014;4:9–19. 29. Kuna ST. Diagnosis and management of patients with obstructive sleep apnea in primary care. Ready or not? Am J Respir Crit Care Med. 2018;198:557–8. 30. Keenan BT, Kim J, Singh B, et al. Recognizable clinical subtypes of obstruc‑ tive sleep apnea across international sleep centers: a cluster analysis. Sleep. 2018;41(3):zsx214. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ Ready to submit your research ? Choose BMC and benefit from: lished maps and institutional affiliations. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions