Skip to main content
Free AccessScientific Investigations

Primary care and sleep unit agreement in management decisions for sleep apnea: a prospective study in Spain

Published Online:https://doi.org/10.5664/jcsm.8492Cited by:6

ABSTRACT

Study Objectives:

Involvement of primary care teams in the care of patients with OSA is a focus of interest. The study objective was to compare diagnostic and therapeutic agreement between decisions taken by primary care professionals and sleep unit specialists.

Methods:

This was a prospective multicenter study conducted at primary care and specialized care centers in the urban area of Barcelona, Spain. Men and women aged 18–75 years who visited the participating primary care centers for any reason were recruited. Both primary care physicians and sleep specialists made a diagnostic and therapeutic decision with clinical data and results of a home sleep apnea test. All patients were finally assessed with respiratory polygraphy or polysomnography as a gold-standard test.

Results:

A total of 229 patients underwent a home sleep apnea test and were evaluated at the primary care centers and the sleep units. Diagnostic agreement using the same tools and excluding indeterminate decisions was 69.8% (Cohen’s kappa = 0.64; 95% confidence interval, 0.56–0.72). Agreement for therapeutic decisions (PAP vs conservative treatment) was obtained in 82.5% of patients (Cohen’s kappa = 0.62; 95% confidence interval, 0.51–0.73), increasing to 92.5% (Cohen’s kappa = 0.49, 95% confidence interval, 0.40–0.58) when indeterminate options were excluded. As compared with the final therapeutic decisions made at the sleep unit with respiratory polygraphy/polysomnography, primary care physicians agreed regarding 83.3% (Cohen’s kappa = 0.62; 95% confidence interval, 0.49–0.74) of patients.

Conclusions:

Primary care professionals may assume an important role in the management of OSA in coordination with sleep centers, identifying patients who require specific treatment and should be referred to specialized care.

Clinical Trial Registration:

Registry: ClinicalTrials.gov; Name: PASHOS Project: Advanced Platform for Sleep Apnea Syndrome Assessment; URL: https://clinicaltrials.gov/ct2/show/NCT02591979; Identifier: NCT02591979.

Citation:

Peñacoba P, Llauger MA, Fortuna AM, 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):1579–1589.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Involvement of primary care teams in the care of patients with OSA is a focus of interest. The study objective was to compare diagnostic and therapeutic agreement between decisions taken by primary care professionals and sleep unit specialists.

Study Impact: Diagnostic agreement between both levels of care was about 70% and therapeutic agreement was higher than 80%. Given the high prevalence and the burden associated with OSA, it would be desirable implementing a coordinated working approach to manage mild-moderate OSA by primary care physicians, and only referring severe disease to specialists for eventual treatment with positive airway pressure.

INTRODUCTION

OSA is a highly prevalent disease1,2 with important health consequences,313 some of which can be reverted with PAP treatment.6,1416 Rising OSA awareness by the medical community and the general population has been associated with an increase in patients seeking care in sleep units. Increasing demands for consultation and sleep studies have led to widespread delays in diagnosis,1719 probably contributing to underdiagnosis of OSA.20 Undiagnosed and untreated OSA has serious consequences, including increased cardiovascular disease, stroke, metabolic disease, excessive daytime sleepiness, workplace errors, traffic accidents, and death, which result in significant economic burdens.21 All of these issues have prompted the search for and implementation of new health care models.22 In the traditional care model, primary care providers refer all patients with suspected OSA to a sleep specialist who decides the optimal diagnostic test and the need for specific treatment. The new models of care integrate other professionals, such as respiratory therapists,23 specialized nurses,24 or primary care physicians2529 in a more cost-efficient diagnosis and treatment process. The main characteristic of studies focused on this new approach is the inclusion of highly symptomatic patients with moderate-severe OSA who are only representative of a limited clinical spectrum of the disease.30,31 The results obtained should be validated in a larger sample of patients with OSA with variable phenotypic expression of the disease. In addition, the feasibility of the implementation of some of these models of care in clinical practice has been questioned because they move the entire health care process, diagnosis, and treatment to a generally overburdened primary care level2529 and excessively simplify the management of patients who have a disease of increasing complexity.32

The Advanced Platform for Sleep Apnea Syndrome Assessment, known as the PASHOS project, was conceived with a different approach to the problem: the implementation of a network for tasks and coordination between primary care physicians and specialists from sleep units.33 Based on the work of the project, the current study includes a population of patients selected in primary care consultation, using a randomization program, with low and high clinical probabilities of OSA. In contrast to previous studies, patients with a low pretest clinical probability of OSA were intentionally included based on the hypothesis that a screening program in primary care may be effective for selecting patients with severe OSA who should refer for starting treatment and to appropriately exclude the presence of OSA, thus preventing unnecessary referrals to the sleep unit.

We here present data on diagnostic and therapeutic agreement between decisions made by primary care professionals and sleep unit specialists based on clinical findings and the results of screening using the home sleep apnea test (HSAT).

METHODS

Study design

We performed a prospective multicenter study, the objective of which was to assess, in an unselected population, primary care center and specialist sleep center agreement regarding diagnostic and management decisions in patients with OSA.

Setting

The study involved the participation of 10 medical institutions located in the urban area of Barcelona (Spain), including 4 sleep units from the hospital setting and 6 primary health care centers. Primary health care teams formed by a primary care physician (a specialist in family and community medicine) and a nurse attending the same population and working in a coordinated manner participated in the study. The study was approved by the clinical research ethics committee of each participating center, and written informed consent was obtained from all patients. The study was registered at ClinicalTrials.gov (identifier: NCT02591979).

Role of the funding source

This study was funded by the Spanish Society of Respiratory Medicine and Thoracic Surgery, the Catalan Society of Pneumology, and a grant from Fondo de Investigaciones Sanitarias (PI 14/01985). This was not an industry-backed study.

Participants

Men and women aged 18–75 years who visited a participating primary health care center for any reason were consecutively included according to a randomization schedule to reach the necessary sample size. Exclusion criteria were previous diagnosis of OSA; chronic insomnia based on a previous clinical diagnosis of insomnia, use of specific treatment for insomnia, or self-reported less than 5 hours of sleep per day; cognitive decline or psychosocial impairment preventing the completion of the study procedures; relevant respiratory comorbidity such as moderate or severe chronic obstructive pulmonary disease; neuromuscular disorder; unstable or acute cardiovascular or cerebrovascular disease; any other condition that could interfere with results of the study; and refusal to participate. Hypertension was diagnosed according to the criteria of the European Society of Hypertension/European Society of Cardiology.34

Training program

Before the study began, a training program was developed. A primary care physician and a nurse from each participating team received a 4-hour theoretical training covering the definition, epidemiology, clinical features, diagnosis, treatment, and prevention of OSA. Nurses received a 3-day training course on the management of home sleep diagnostic equipment and the analysis of recordings, which were finally supervised by the physicians, and basic training in more complex sleep studies. A specialized nurse in sleep medicine from the Spanish Sleep Society was the instructor for the 3-day training course. Nurses were not directly involved in the diagnostic and treatment process; the primary care physicians were fully responsible. Primary care physicians stayed for 3 days in a sleep unit learning about the indication and interpretation of different diagnostic sleep studies with an emphasis on ambulatory monitoring and outpatient clinical management. At the end of the training period, all participants analyzed 10 simulated patients with OSA of different levels of severity to check their knowledge acquisition.33 Details of characteristics and analysis of simulated cases were previously reported in the description of the study protocol.33 This training program was an overall evaluation (not quantified) of 10 simulated patients with OSA of different complexity and severity, including registers with technical errors. The degree of agreement between the different participants (primary health care physicians and sleep specialists) was analyzed, and finally therapeutic decisions for all simulated patients were reviewed in a joint meeting, with an emphasis on those with mild-to-moderate agreement.

Procedure

At the primary care consultation, the following variables were recorded: age, sex, anthropometric data (weight; height; body mass index; waist, hip, and neck circumferences), salient features of medical history, and cardiovascular risk factors. All patients completed a specific clinical questionnaire of suggestive symptoms of OSA, the Epworth Sleepiness Scale,35 and the Berlin questionnaire.36 Scoring of the Berlin questionnaire categorized patients into groups at high risk (positive on 2 or more categories) and low risk of having sleep apnea. All patients in the high-risk category and 1 out of 2 patients in the low-risk category randomly took the HSAT. The selection of 1 out of 2 patients with a low probability of OSA was established to balance the study sample with a similar final proportion of patients with high and low probability of OSA.

After assessing the clinical probability of OSA according to history, clinical findings, and results of the HSAT, the primary care physician chose 1 of the following diagnostic-therapeutic scenarios: (1) low suspicion of OSA and no therapeutic measures, (2) suspicion of mild-moderate OSA and conservative treatment, (3) suspicion of OSA candidate for PAP therapy, and (4) indeterminate. The indeterminate scenario included patients with difficulties in the interpretation of the HSAT (carried out up to 2 times maximum) because of the presence of artifacts in the signals or a discrepancy between the symptomatology and the HSAT results (eg, low pretest clinical probability but high AHI or high pretest clinical probability with low AHI). The clinical decision to start PAP treatment was based on the recommendations of the clinical practice guidelines in patients with AHI >30 events/h or >5 events/h in the presence of excessive daytime sleepiness and/or relevant comorbidity37 (the criteria for defining AHI are detailed in the supplemental material). All data from the patients’ medical records were considered—not only the results of the Berlin questionnaire, which was used only for the selection of a balanced sample. Because treatment with PAP is covered by the Spanish health care system, economic limitations did not influence treatment decisions.

All patients who took the HSAT were referred to the sleep unit, where specialists with all clinical documentation available made a diagnostic and therapeutic decision for each patient with the same tools. These specialists were blinded to the clinical decisions made by the primary care physicians. Finally, patients underwent a complete respiratory polygraphy or conventional polysomnography to establish a final diagnosis and therapeutic decision.

HSAT

In all cases, screening sleep testing was performed at the patients’ homes with a self-applied Sibelmed Screen&Go device (Bitmed, Sibelgroup, Barcelona, Spain), with a 2-channel monitor and a nasal cannula for airflow measure and oximetry. The device also provides data on body position. The minimum valid recording time was 5 hours. The primary care nurses assessed the quality of the sleep studies and removed the periods of poor signal (artifacts or lost signal) but did not analyze the respiratory events manually. Clinical data and results of specific questionnaires together with results of HSAT recordings were telematically sent to the referral sleep centers, where sleep specialists analyzed HSAT data manually.

In both automatic and manual analyses, the respiratory events were defined as follows: hypopnea was set as an airflow reduction of ≥30% and <90% lasting at least 10 seconds, with a ≥3% drop in oximetry, and apnea was defined as an absence of airflow or ≥90% reduction for ≥10 seconds.

Respiratory polygraphy and polysomnography

Conventional polysomnography recordings included: electroencephalography, electro-oculography, chin electromyography, oximetry, oronasal airflow by thermistor and nasal cannula, thoracoabdominal movements by inductance plethysmography, electrocardiography, anterior-tibial electromyography, and body position. Respiratory polygraphy included every channel except electroencephalography, electro-oculography, electromyography, and electrocardiography.

Respiratory polygraphy and polysomnography recordings were analyzed manually by sleep specialists according to the Spanish Society of Respiratory Medicine and Thoracic Surgery manual procedures37 and the American Academy of Sleep Medicine 2012 recommendations38 (details of these recommendations are included in the supplemental material).

Statistical analyses

The sample size was calculated considering a minimal OSA prevalence of 25% in the population attended in the primary care setting. Assuming a loss of 15% at follow-up, an alpha error of 5%, a noninferiority limit of 10%, and an 80% statistical power to reject the null hypothesis with an asymptotic noninferiority test, a total sample of 156 patients was needed. Categorical variables were expressed as frequencies and percentages and quantitative variables were expressed as mean and standard deviation, median and interquartile range (25th–75th percentile), and 95% confidence interval (CI). Bivariate analysis included the χ2 test or the Fisher exact probability test for categorical data and the Student t test or Mann-Whitney U test for continuous data according to the conditions of application. Interrater agreement for diagnostic and treatment decisions between primary care physicians and sleep specialists were analyzed with Cohen’s kappa coefficient (κ) and the linearly weighted kappa when needed. Statistical significance was set at P <.05. Statistical analyses were performed using the statistical software SPSS version 15.0 (SPSS Inc., Chicago) for Windows.

RESULTS

Of 1,036 patients evaluated at primary care consultations, 466 (44.9%) were eligible and agreed to participate in the study. A total of 279 were classified as low-risk for OSA and 150 as high-risk according to the results of the Berlin questionnaire.36 Of those classified as low-risk, 149 (53.4%) did not undergo HSAT by randomization. Thus, 249 patients underwent HSAT and 229 (92%) were evaluated at the sleep unit. Finally, 194 completed the study with respiratory polygraphy or polysomnography. The flowchart of the study population is shown in Figure 1.

Figure 1: Flowchart of the study population.

HSAT = home sleep apnea test, PSG = polysomnography, RP = respiratory polygraphy.

Baseline characteristics of the 429 patients who completed the Berlin questionnaire are shown in Table 1. Patients in the high-risk group for OSA as compared with the low-risk group showed significantly higher values of obesity-related variables, score on the Epworth Sleepiness Scale, comorbidities (hypertension, diabetes, depression, anxiety), and reduced forced vital capacity. There were no statistically significant differences among patients in the low-risk group for OSA according to whether or not they had been scheduled for the HSAT by randomization (Table S1 in the supplemental material).

Table 1 Baseline characteristics of 429 patients evaluated in primary care.

VariablesFull Sample (n = 429)Berlin QuestionnaireP
Low-Risk (n = 279)High-Risk (n = 150)
Gender (men:women)211:218135:14476:74.652
Age, y, mean (SD)54.4 (13)54.3 (13.6)54.5 (11.9).870
BMI, kg/m2, mean (SD)28.4 (5.2)26.9 (4.5)31.2 (5.3)<.001
Neck circumference, cm, mean (SD)37.3 (5.3)36.5 (5.5)38.8 (4.6)<.001
Waist circumference, cm, mean (SD)95.9 (14.2)92.2 (13.3)102.8 (13.2)<.001
Hip circumference, cm, mean (SD)105 (10.6)102 (9.5)110.6 (10.4)<.001
Waist-hip ratio, mean (SD)0.91 (0.10)0.90 (0.10)0.92 (0.11).198
Comorbidities
 Obesity (BMI >30 kg/m2)140 (32.6)47 (16.8)93 (62)<.001
 Hypertension146 (34.8)72 (26.5)74 (50)<.001
 Diabetes mellitus60 (14)32 (11.5)28 (18.7).040
 Dyslipidemia129 (30.1)77 (27.6)52 (34.7).128
 Depression44 (10.3)22 (7.9)22 (14.7).027
 Anxiety84 (19.6)44 (15.8)40 (26.7).007
 Stroke6 (1.4)5 (1.8)1 (0.7).670
 Arrhythmia16 (3.7)11 (3.9)5 (3.3).751
 Peripheral artery disease13 (3)8 (2.9)5 (3.3).788
 Hypothyroidism28 (6.5)15 (5.4)13 (8.7).188
 Heart disease21 (4.9)11 (3.9)q10 (6.7).212
 Active smoking98 (22.8)57 (20.4)41 (27.3).251
 Alcohol consumption155 (36.4)102 (37)53 (35.3).739
 Profession of risk36 (10.3)20 (9.1)16 (12.5).314
FVC%, mean (SD)92.1 (14.4)93.4 (14.6)89.6 (13.9).01
FEV1%, mean (SD)93 (15.2)93.8 (15.8)91.4 (13.8).111
Epworth Sleepiness Scale score, mean (SD)6.7 (4.8)5.6 (4.1)8.9 (5.3)<.001

Data expressed as frequencies and percentages in parentheses unless otherwise stated. BMI = body mass index, FEV1 = forced expiratory volume in 1 second, FVC = forced vital capacity, SD = standard deviation.

Baseline data for the 229 patients who continued the assessment at the sleep units are shown in Table 2. Statistically significant differences between the high-risk and low-risk groups for OSA were similar to those observed in the full sample of patients who completed the Berlin questionnaire.

Table 2 Baseline characteristics of 229 patients evaluated in the sleep unit.

VariablesFull Sample (n = 229)Berlin QuestionnaireP
Low-Risk (n = 102)High-Risk (n = 127)
Gender (men:women)123:10655:4768:59.955
Age, y, mean (SD)54.5 (12.6)54.5 (13.2)54.5 (12.2).956
BMI, kg/m2, mean (SD)29 (5.4)26.5 (4.5)31 (5.2)<.001
Neck circumference, cm, mean (SD)38.2 (4.7)37.2 (4.3)39 (4.8).004
Waist circumference, cm, mean (SD)98.5 (14.2)93.2 (13.4)102.7 (13.4)<.001
Hip circumference, cm, mean (SD)106.6 (10.4)102.2 (9.2)110.1 (10)<.001
Waist-hip ratio, mean (SD)0.92 (0.11)0.91 (0.09)0.92 (0.12).596
Comorbidities
 Hypertension85 (37.9)24 (24.5)61 (48.4)<.001
 Diabetes mellitus39 (17)15 (14.7)24 (18.9).402
 Dyslipidemia76 (33.2)30 (29.4)46 (36.2).277
 Depression22 (9.6)5 (4.9)17 (13.4).030
 Anxiety42 (18.3)11 (10.8)31 (24.4).008
 Stroke2 (0.9)1 (1)1 (0.8).000
 Arrhythmia7 (3.1)2 (2)5 (3.9).466
 Peripheral artery disease8 (3.5)5 (4.9)3 (2.4).472
 Hypothyroidism17 (7.4)4 (3.9)13 (10.2).070
 Heart disease14 (6.1)5 (4.9)9 (7.1).493
 Active smoking59 (25.8)25 (24.5)34 (26.8).919
 Alcohol consumption87 (38.3)40 (40)47 (37).645
 Profession of risk24 (12.3)8 (9)16 (15.1).196
FVC%, mean (SD)90.9 (14.1)92.7 (14.7)89.5 (13.5).090
FEV1%, mean (SD)92.3 (14.8)93.4 (15.9)91.5 (13.9).343
Epworth Sleepiness Scale score, mean (SD)7.9 (5.1)6.3 (4.2)9.2 (5.4)<.001

Data expressed as frequencies and percentages in parentheses unless otherwise stated. BMI = body mass index, FVC = forced vital capacity, FEV1 = forced expiratory volume in 1 second, SD = standard deviation.

Baseline characteristics of the 194 patients who completed the study are shown in Table S2. In relation to these patients, patients who withdrew (n = 86) were younger, with a lower number of working professions of risk (eg, professional drivers), a higher prevalence of smoking and depression, and a lower score on the Epworth Sleepiness Scale (Table S3).

Diagnostic and therapeutic agreement according to HSAT

Diagnostic agreement between the primary and specialized care providers was obtained in 62.9% of patients (κ = 0.49, 95% CI, 0.40–0.58). The level of agreement according to each decision is shown in Figure 2A. Minimal disagreement was observed in suspicions of severe OSA. In 17% of patients, low suspicion of OSA was established by the primary care physician, whereas the sleep specialist considered suspicion of mild-moderate OSA. The diagnosis of indeterminate OSA was made in 30 patients, 16 in the primary care setting and 19 patients in the sleep unit, with a coincident diagnosis in 5 patients. Discrepancy between symptoms and HSAT occurred in 13 patients, a low quality of study signal occurred with 11 patients, and a disproportionate nocturnal oxygen saturation occurred in 6 patients. When indeterminate cases were excluded from the analysis, diagnostic agreement between both levels of care was 69.8% (κ = 0.64; 95% CI, 0.56–0.72).

Figure 2: Agreement for diagnostic decisions.

In light blue, concordant patients; in red, no concordant patients. The size of the circle represents the proportion of patients. (A) Agreement between primary care and sleep units based on HSAT. (B) Agreement between sleep units based on RP/PSG and primary care based on HSAT. (C) Agreement at sleep unit based on RP/PSG and HSAT. HSAT = home sleep apnea test, PSG = polysomnography, RP = respiratory polygraphy.

Agreement for therapeutic decisions (PAP vs conservative treatment) between primary and specialized care physicians was obtained in 82.5% of patients (κ = 0.62; 95% CI, 0.51–0.73) (Figure 3A). In addition, when indeterminate decisions were excluded, therapeutic agreement between providers at both levels of care increased to 92.5% (κ = 0.49; 95% CI, 0.40–0.58).

Figure 3: Agreement for therapeutic decisions.

In light blue, concordant patients; in red, no concordant patients. The size of the circle represents the proportion of patients. (A) Agreement between primary care and sleep units based on HSAT. (B) Agreement between sleep units based on RP/PSG and primary care based on HSAT. (C) Agreement at sleep unit based on RP/PSG and HSAT. HSAT = home sleep apnea test, PSG = polysomnography, RP = respiratory polygraphy.

Diagnostic and therapeutic agreement according to respiratory polygraphy or polysomnography

Of the 194 patients who completed the study and underwent complete respiratory polygraphy or polysomnography, 14 patients (7.2%) diagnosed as indeterminate based on the HSAT were excluded from the analysis. As compared with the final diagnosis made at the sleep unit, diagnoses made by primary care physicians agreed in 62.2% of patients (κ = 0.44; 95% CI, 0.33–0.54) (Figure 2B). Agreement for therapeutic decisions was 83.3% (κ = 0.62; 95% CI, 0.49–0.74) (Figure 3B).

The analysis of agreement at the sleep unit was 74.4% (κ = 0.66; 95% CI, 0.58–0.75) for diagnosis of OSA (Figure 2C) and 87.2% (κ = 0.70; 95% CI, 0.59–0.81) for therapeutic decisions (Figure 3C). There were 19 patients with severe OSA who had been rejected for PAP treatment because of the HSAT results. These patients, as compared with candidates for PAP treatment, presented with more atypical OSA with significantly lower BMI and lower prevalence of hypertension and daytime somnolence (Table 3). They also presented with a greater positional component in the HSAT, with AHI events/h in the supine position that were more than double those of the global AHI. Unfortunately, we do not have the results of the AHI events/h in the nonsupine position or the time in each position.

Table 3 Characteristics of patient candidates for PAP treatment by sleep specialists based on result of respiratory polygraphy or polysomnography.

VariablesFull Sample (n = 63)Sleep Specialist Decision Based on HSATP
No PAP Candidates (n = 19)PAP Candidates (n = 44)
Gender (men:women)47:1613:634:10.459
Age, y, mean (SD)57.8 (11.5)57.4 (9)58 (12.5).837
BMI, kg/m2, mean (SD)32 (5.2)30 (4.6)32.8 (5.3).043
Neck circumference, cm, mean (SD)41.2 (3.9)40.2 (5)41.6 (3.2).262
Waist circumference, cm, mean (SD)107.3 (11)103.4 (12.1)109 (10.1).060
Hypertension34 (54.8)3 (15.8)31 (72.1)<.001
Diabetes mellitus16 (25.4)3 (15.8)13 (29.5).250
Dyslipidemia24 (38.1)7 (36.8)17 (38.6).893
Profession of risk8 (14.3)3 (21.4)5 (11.9).378
Epworth Sleepiness Scale score, mean (SD)9 (5.6)6.3 (5.3)10.1 (5.3).013
HSAT, mean (SD)
 Recording time, minutes401 (79.8)416.8 (86.3)394.2 (76.9).305
 AHI26.1 (19.5)13.1 (7.6)31.7 (20.5)<.001
 AHI supine43.2 (24.4)27.2 (20.6)50.1 (22.8)<.001
 Average O2 saturation91.7 (2.4)92.8 (1.7)91.2 (2.6).016
 CT90%18.7 (21.2)9.2 (9.9)23.1 (23.5).002
 ODI4%27 (17.3)10.7 (5.7)34.3 (15.7)<.001
RP/PSG, mean (SD)
 Recording time, min462.1 (71.8)447.9 (78.7)468.3 (68.6).303
 AHI42.5 (18.2)38.1 (11.3)44.4 (20.3).017
 AHI supine47.7 (19.8)47 (15.3)48 (21.5).871
 Average O2 saturation92.2 (2)92.7 (1.9)91.9 (2).176
 Minimum O2 saturation78.3 (7.9)81.6 (5.2)76.9 (8.5).010
 CT90%14.8 (16.3)10.6 (12.8)16.6 (17.5).183
 CT80%1.3 (5.3)0.2 (0.5)1.8 (6.2).280
 ODI4%27.8 (17)20 (12.8)30.7 (17.6).036

Data expressed as frequencies and percentages in parentheses unless otherwise stated. BMI = body mass index, CT80% = percentage of time with oxygen saturation below 80%, CT90% = percentage of time with oxygen saturation below 90%, HSAT = home sleep apnea test, ODI4% = oxygen desaturation index of 4%, RP = respiratory polygraphy, PSG = polysomnography, SD = standard deviation.

DISCUSSION

This study assessed the concordance between primary care physicians and sleep specialists regarding patients with OSA, specifically the level of agreement in diagnosis and therapeutic decisions using clinical data and the HSAT as screening tools. Main findings of the study were a moderate agreement on diagnostic classification and a good level of concordance in deciding whether to begin treatment with PAP. It seems clear that clinical data and a simple HSAT allow primary care physicians to identify patients who may require specific treatment and should be referred to specialized care, avoiding referral in a high percentage of patients, with a subsequent reduction in workload for sleep units. These results support the feasibility of coordinated work between primary care physicians and sleep specialists in the framework of routine daily practice.

We used a 2-channel respiratory polygraph, a level IV study (American Sleep Disorders Association classification), with simple settings and easy recording analysis. Portable respiratory recording devices provide high diagnostic yield and facilitate detection of OSA in the ambulatory setting. However, most of this equipment has been validated in sleep units3941 in patients that were specifically referred for suspected OSA and the clinical decisions were made by sleep specialists. Therefore, validation studies in the primary care setting are necessary.

We found a moderate level of diagnostic concordance between primary care physicians and sleep specialists regarding patients with intermediate suspicion of OSA, but there was substantial agreement in extreme cases, especially in patients diagnosed with severe OSA and candidates for PAP therapy. These results may be attributed to primary care physicians with less experience in the management of OSA and, particularly, to the well-known limitations of screening studies. Automatic analysis of respiratory polygraph recordings was used in primary care, with only a rapid revision of artifacts or deficient signal periods, whereas sleep specialists analyzed recordings manually. In the 2014 Masa et al study,39 in which the diagnostic efficacy of both scorings (automated and manual) of a home single-channel nasal pressure device was evaluated, manual scoring showed a better diagnostic accuracy than automatic scoring for patients with low AHI, but automatic and manual scoring had a similar accuracy for intermediate and high AHI levels. In our opinion, even considering the limitations, the automatic analysis option seems preferable in the primary care setting, where time constraints and the level of providers’ experience in the analysis of sleep studies play a crucial role in daily practice.

In relation to the treatment of OSA, we limited the therapeutic decision whether to initiate PAP without considering other approaches.42 For the design of the study, we relied on current clinical guidelines that basically recommend treatment with PAP in patients with severe OSA, those who are highly symptomatic, or those with relevant comorbidity.37,43 However, OSA is a heterogeneous disease with a recent focus on personalized medicine that approaches the treatment of OSA based on its phenotypes.44 New learning about OSA would require adapting our approach to treatment, which may be facilitated by establishing a framework of coordinated tasks between clinicians at different levels of care.

Other strategies for managing OSA in the primary care scenario have been recently evaluated in randomized clinical trials. Chai-Coetzer and colleagues26 conducted a randomized trial in 2013 in which primary care physicians identified 155 patients with moderate to severe OSA using a previous validated 2-step diagnostic strategy that included a screening questionnaire and home oximetry. Patients who had at least moderate daytime sleepiness (Epworth Sleepiness Scale ≥ 8) or resistant hypertension were randomly assigned to either ambulatory management by the general practitioner and community-based nurse or to usual laboratory-based care by a sleep specialist. After 6 months, management in the primary care setting was found to be non-inferior to specialist sleep center management in terms of improvement in daytime somnolence, quality of life, OSA symptoms, PAP adherence, and overall patient satisfaction. Per-patient costs were 38% less in the primary care arm of the study.

Recently, Sánchez-Quiroga and colleagues,28 in another non-inferiority randomized controlled trial, reported that in patients with intermediate to high OSA probability, diagnosis and management by general practitioners had similar outcomes as in patients receiving care by sleep specialists in a sleep laboratory. The main contribution of this study is the use of a prescribed clinical algorithm in primary care practices that required minimal training of doctors and nurses. In addition, the cost-effectiveness relationship favored the primary care practice arm, largely because of the higher cost of polysomnography. Finally, Tarraubella et al27 also found that in patients with high clinical suspicion of OSA or resistant hypertension, the primary care model including primary care physicians and nurses did not result in worsening the Epworth Sleepiness Scale score or the Health Utilities Index (measure of general health status and health-related quality of life) at 6 months of PAP treatment and generated savings in terms of management costs.

However, in none of these studies were indirect costs or the long-term economic implications of this simplified strategy in primary care assessed.

In contrast to these previously published studies, our patient population was selected according to a randomization program, so it was more representative of patients usually cared for in primary care centers. The fact that the patients included in the study did not specifically attend primary care for OSA symptoms would be associated with a lower pretest clinical risk of OSA and, therefore, a more complex diagnosis. Despite these circumstances, there was substantial agreement, particularly for therapeutic decisions, made by professionals from the primary care and specialized care centers that we attribute to the coordinated network model and the training program, which was brief but systematized. It is important to emphasize that for the assessment, all the clinical history data were considered, not just the Berlin questionnaire that was used only for the selection of a balanced sample.

An interesting aspect of the study was the comparison of diagnostic and therapeutic decisions made based on the HSAT with the results of sleep studies in the sleep unit (respiratory polygraphy/polysomnography). We found that a nonnegligible percentage of patients was misclassified even in the sleep centers where the HSAT recordings were evaluated. Of 63 patients who would have been treated with PAP based on the gold-standard study, 19 (30.2%) would not have been treated if a decision had been made according to the results of the HSAT. These patients were characterized by reduced daytime sleepiness, low cardiovascular load (absence of obesity and hypertension), and predominant positional OSA in the HSAT but not in the respiratory polygraphy/polysomnography. The overestimation of severity by complete studies in patients with positional OSA is already known.45 These results illustrate the obvious limitations of simplified techniques already reported41 and reveal that body position signal is a key feature in ambulatory screening sleep studies.

Our study is closely related to clinical practice and was dependent on the characteristics of the national health care system, which limits the external validity of the study. The degree of knowledge about sleep medicine may differ among primary care professionals in different countries according to the inclusion of specific training topics in sleep disorders in their curricular training systems. We believe that the key to a successful network system is a close interrelationship between professionals and good communication channels. This interrelationship allowed us to design a training program tailored to primary care physicians on the basis of their lack of specific training in sleep disorders and the primary reliance of their training on clinical practice. Unfortunately, in this study we do not have data on objective feedback of the training program. However, for the training program’s implementation in clinical practice we would recommend including a quantifiable evaluation of the objectives of the program. In this respect, the implementation of the model in other geographical areas and health care systems should always consider this interrelationship that allows designing the specific training programs to be adapted to the characteristics of professionals in each particular setting.

Other limitations of the study include the eligibility criterion of an upper age range of 75 years, which is not fully representative of the adult population attended in primary care, and the exclusion of patients with a previous clinical diagnosis of insomnia, which may have contributed to the underdiagnosis of patients with OSA primarily with complaints of sleep disruption, a clinical cluster of OSA related to the highest rates of hypertension, diabetes, and cardiovascular disease.31 Moreover, unlike other studies, conventional polysomnography was not used only as the gold standard. Although respiratory polygraphy can underestimate the diagnosis of OSA, we adapted our model to routine clinical practice and to the recommendations of guidelines that suggest the use of respiratory polygraphy only in selected patients.46 Finally, the study was focused on the management of patients with simple OSA, so that the present results are only applicable to this population, excluding patients who may have other respiratory and nonrespiratory sleep disorders. Strengths of the study include its multicenter design and that the inclusion of patients was not restricted to a certain a priori clinical probability of OSA, so the results may be applicable to a wide phenotypic spectrum of patients.

In summary, with appropriate training and simplified management tools, clinical data, and a level IV study, primary care physicians and nurses may assume an important role in the management of OSA in coordination with sleep centers, selecting patients who must be referred to specialized care to begin specific treatment and preventing unnecessary referral of a high percentage of patients with mild-moderate OSA. The proposed model does not move the entire health care process to a usually overburdened primary care level, and it favors the coordinated work and the necessary flexibility to adapt the model to new challenges and perspectives of OSA.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. Work for this study was performed at Barcelona. This study was funded by the Spanish Society of Respiratory Medicine and Thoracic Surgery, the Catalan Society of Pneumology, and a grant from Fondo de Investigaciones Sanitarias (FIS) (PI 14/01985). This was not an industry-backed study. The fees of medical editing were paid by RESPIRA Fundación Española del Pulmón, the Spanish Society of Pneumology and Thoracic Surgery, Barcelona, Spain. Anna Maria Pedro Pijoan reports grants from Teva, grants from Chiesi, grants from Boehringer, and grants from GSK outside the submitted work. Gabriel Sampol reports nonfinancial support from Respironics and personal fees from Resmed outside the submitted work. The authors report no conflicts of interest.

ABBREVIATIONS

BMI

body mass index

CI

confidence interval

COPD

chronic obstructive pulmonary disease

CT80%

percentage of time with oxygen saturation below 80%

CT90%

percentage of time with oxygen saturation below 90%

ECG

electrocardiography

EEG

electroencephalography

EMG

electromyography

EOG

electro-oculography

FEV1

forced expiratory volume in 1 second

FVC

forced vital capacity

HSAT

home sleep apnea test

ODI4%

oxygen desaturation index of 4%

PSG

polysomnography

RP

respiratory polygraphy

SD

standard deviation

REFERENCES