The differential impact of respiratory event scoring criteria on CPAP eligibility in women and men

Abstract

Study Objectives:

Obstructive sleep apnea is more prevalent and severe in men than women. The American Academy of Sleep Medicine offers 2 definitions for scoring hypopneas: “acceptable” = associated with a ≥ 4% oxygen desaturation, adopted by Center for Medicare and Medicaid Services (CMS), and “recommended” = associated with a ≥ 3% oxygen desaturation and/or an arousal. We hypothesized that CMS vs American Academy of Sleep Medicine scoring criteria would differentially impact continuous positive airway pressure eligibility in women and men.

Methods:

We conducted a retrospective review of adult diagnostic in-lab polysomnography at an urban academic institution. All polysomnographies were scored by both CMS and American Academy of Sleep Medicine scoring criteria, and an analysis by sex was performed that considered demographics and other polysomnography variables.

Results:

Of 969 polysomnographies reviewed, 674 (69.6%) were in women. Women were younger (51.5 vs 53.3 years old) and had a higher body mass index (38.6 kg/m² vs 33.8 kg/m²) but had similar Epworth Sleepiness Scale scores compared to men. The odds of an American Academy of Sleep Medicine apnea-hypopnea index > 5 events/h being missed by CMS scoring in women was 1.89 (95% confidence interval: 1.40–2.53; P < .001) compared to men and increased to 6.87 among women 40–60 years of age with a body mass index ≥ 40 kg/m². After controlling for age, body mass index, % rapid eye movement sleep, and mean oxygen saturation, the sex effect remained significant (odds ratio 1.87; 95% confidence interval: 1.36–2.58; P < .001).

Conclusions:

CMS scoring criteria imparts a sex bias toward women, potentially resulting in denial of therapy to symptomatic women with obstructive sleep apnea. Larger, prospective cohort studies are needed to confirm these findings.

Citation:

Khalid F, Ayache M, Auckley D. The differential impact of respiratory event scoring criteria on CPAP eligibility in women and men. J Clin Sleep Med. 2021;17(12):2409–2414.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Obstructive sleep apnea is more prevalent in men compared to women despite a similar symptom burden. Although reasons for this are likely multifactorial, the underdiagnosis of obstructive sleep apnea in women resulting from Centers for Medicare and Medicaid Services scoring criteria may be playing a role.

Study Impact: The odds of missing clinically significant obstructive sleep apnea are higher in women than in men when using Centers for Medicare and Medicaid Services scoring criteria compared to American Academy of Sleep Medicine scoring criteria to define obstructive sleep apnea, and this is most notable in obese middle-aged women.

INTRODUCTION

Obstructive sleep apnea (OSA) is characterized by repeated episodes of sleep-related upper airway obstructions that are characterized as complete (apneas) or partial (hypopneas). The apnea-hypopnea index (AHI), the sum of apneas plus hypopneas per hour of sleep, is the conventional metric used to both diagnose OSA and to grade its severity.¹ According to the American Academy of Sleep Medicine’s (AASM) scoring manual², apneas are to be scored when there is an absence of airflow, while hypopneas can be scored by 1 of 2 definitions: a 30% reduction in airflow associated with a ≥ 4% oxygen desaturation (“acceptable” criteria) or a 30% reduction in airflow associated with a ≥ 3% oxygen desaturation and/or arousal (“recommended”’ criteria). The prevalence and severity of OSA vary considerably depending on which of these scoring criteria are used for hypopneas.^3,4 The “acceptable” hypopnea definition has been adopted by the Center for Medicare and Medicaid Services (CMS), as well as other health care insurers, for scoring respiratory events to determine the AHI that qualifies patients for therapy.⁵ As a result, eligibility for continuous positive airway pressure (CPAP) therapy may be significantly influenced by the scoring criteria used by insurers to define OSA (as determined by the AHI) and thus reimburse for treatment.⁴

OSA prevalence has been reported to have a female-to-male ratio in the range of 1:3 and 1:5 in the general population.^6,7 In addition, both OSA severity and its distribution across sleep stages differ between women and men, with women having less severe OSA,⁸ particularly in nonrapid eye movement sleep.^8,9 This may be related in part to women having shorter apneic events and less severe oxygen desaturation than men as seen during polysomnography (PSG).¹⁰

Nevertheless, women tend to be more symptomatic at lower AHI values compared with men. Women with an AHI of 2–5 events/h will report symptom severity similar to that of men with an AHI of ≥ 15 events/h.¹¹ While the etiology for this is not entirely clear, a number of pathophysiologic mechanisms have been proposed. Women tend to have a clustering of respiratory events in rapid eye movement (REM) sleep,¹² and REM-related OSA has been associated with daytime sleepiness. Episodes of upper airway resistance, respiratory events that do not meet criteria to be scored as hypopneas, are more commonly seen in women and can lead to sleep fragmentation.^13,14 Recent work has found that despite women having a lower loop gain and less airway collapsibility in non-rapid eye movement sleep compared to men, women have a lower arousal threshold.⁹

As a consequence of these differences between women and men in the manifestations of OSA, symptomatic women may be less likely to meet a ≥ 4% oxygen desaturation hypopnea definition (“CMS criteria”) for the diagnosis of OSA compared to a ≥ 3% oxygen desaturation and/or arousal hypopnea definition (“AASM criteria”). In addition, the severity of OSA may be significantly underestimated in women compared to men when scoring by CMS criteria.

We therefore hypothesized that women with symptomatic OSA would be less likely than men to meet eligibility criteria for treatment with CPAP when utilizing CMS scoring criteria compared to AASM scoring criteria. To test this hypothesis, we examined the impact of CMS scoring criteria vs AASM scoring criteria on CPAP eligibility for patients who underwent PSGs in our sleep laboratory.

METHODS

Study overview

This study was a retrospective review of adult diagnostic in-lab PSGs completed during a 2-year period (2017–2018) at an urban academic institution. Split-night PSGs were excluded. All PSGs were scored using both CMS criteria and AASM criteria (as defined below). Patient characteristics (sex, age, body mass index [BMI]) and comorbidities were recorded at the time of the PSG and subsequently extracted from the medical record. An analysis was then performed comparing scoring of the PSGs by both CMS and AASM criteria between women and men. Institutional Review Board approval was obtained prior to the initiation of the study (IRB# 18-00852).

PSG scoring

CMS¹⁵ and AASM scoring criteria were obtained from the current version of The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications (2020, version 2.6).² In both CMS and AAMS scoring guidelines, an apnea is scored when peak signal excursions drop by ≥ 90% of pre-event baseline for at least 10 seconds using an oronasal thermal sensor. Hypopneas were scored by CMS guidelines criteria (“acceptable criteria”) for scoring a hypopnea that required 1) the peak signal excursion to drop by > 30% of pre-event baseline using nasal pressure, 2) the duration of the 30% drop in signal excursion is ≥ 10 seconds, and 3) a ≥ 4% oxygen desaturation from pre-event baseline. Hypopneas were also scored by AASM guidelines criteria (“recommended criteria”) for scoring a hypopnea that required CMS criteria numbers 1 and 2 be fulfilled, but that the reduction in airflow is associated with either a ≥ 3% oxygen desaturation from pre-event baseline or an arousal (as defined in the AASM scoring manual). All scoring was performed by registered PSG technicians enrolled in the AASM Interscorer Reliability program.

A total of 984 PSGs were reviewed. Fifteen PSGs were excluded due to incomplete data. PSG parameters were extracted from each study and included the AHI by CMS and AASM criteria, % REM sleep time, and mean O₂ saturation.

Statistical analysis

To compare the impact of scoring by the different criteria, mean CMS-to-AASM AHI ratios were calculated in women and men and then stratified by age and BMI. These ratios were analyzed using unpaired t-test (level of significance P value < .05). Then, a CMSmiss dummy variable was created to represent the diagnosis of OSA by AASM criteria but not by CMS criteria, ie, CMS AHI < 5 events/h and AASM AHI ≥ 5 events/h. Univariate logistic regression was used to calculate the odds of CMSmiss in women compared to men, stratified by age and BMI. Multivariate logistic regression was subsequently used to calculate the effect of sex on the odds of a CMSmiss while taking into consideration demographic and PSG independent variables (% REM sleep, mean oxygen saturation) that could influence the AHI. STATA 12.0 was used for statistical analysis.

RESULTS

The total number of individual studies analyzed was 969, of which 674 (69.6%) were performed on women and 295 (30.4%) on men. Overall and sex-specific descriptive patient characteristics are outlined in Table 1. Men were slightly older (average age 53.3 years old vs 51.5 years old for women), while women had a higher average BMI (average BMI 38.6 kg/m² for women vs 33.8 kg/m² for men). Rates of significant comorbidities were similar, other than men had slightly more heart failure and atrial fibrillation and women were more frequently diagnosed with asthma. Median Epworth Sleepiness Scale scores were not significantly different, and men had statistically significant higher AHIs by both CMS and AASM scoring criteria (Table 1).

Table 1.

Patient characteristics.

	Overall	Women	Men	P*
	(n = 969)	(n = 674)	(n = 295)
Mean age (± SD) years	52.1 (14.9)	51.5 (15.1)	53.3 (14.3)	.0831*
Mean BMI (± SD) kg/m2	37.2 (9.8)	38.6 (9.8)	33.8 (8.9)	<.001*
Comorbidities, n (%)
CHF	63 (6.5)	34 (5.1)	29 (9.8)	.007†
Atrial fibrillation	47 (4.9)	21 (3.1)	26 (8.8)	<.001†
COPD	96 (9.9)	63 (9.4)	33 (11.2)	.413†
Asthma	160 (16.6)	135 (20.0)	25 (8.1)	<.001†
Anxiety	158(16.3)	109 (16.1)	49 (16.7)	.851†
Depression	274 (28.3)	194 (28.9)	80 (27.3)	.642†
Hypothyroidism	50 (5.2)	36(5.3)	14 (4.8)	.752†
ESS median (IQR)	8.0 (9.0)	9.0 (9.0)	8.0 (7.75)	.288‡
PSG data (AHI events/h)
CMS AHI median (IQR)	6.2 (12.7)	5.0 (10.7)	8.65 (18.4)	<.001‡
AASM AHI median (IQR)	20.2 (23.8)	18.5 (20.4)	25.8 (28.05)	<.001‡

*P value for women vs men calculated using unpaired t-test. †P value for women vs men calculated using Fisher’s exact test. ‡P value for women vs men calculated using Mann-Whitney test. AASM AHI = American Academy of Sleep Medicine recommended scoring criteria for AHI, AHI = apnea-hypopnea index, BMI = body mass index, CHF = congestive heart failure, CMS AHI = Centers for Medicare and Medicaid Services scoring criteria for AHI, COPD = chronic obstructive pulmonary disease, ESS = Epworth Sleepiness scale, IQR = interquartile range, PSG = polysomnogram, SD = standard deviation.

The overall mean AHI ratio (CMS to AASM) in women was 0.33 (95% confidence interval [CI]: 0.32–0.35) compared to 0.41 (95% CI: 0.38–0.44) in men (difference in mean ratio of 0.08, 95%CI: 0.04–0.11; P < .001). This is consistent with a greater difference in the AASM-scored AHI vs the CMS-scored AHI in women than in men. Stratified by age and BMI, the size effect of sex on lower CMS-to-AASM AHI ratio was greatest in the middle-aged and morbidly obese subgroup (age 40–60 years; BMI ≥ 40 kg/m²), with a difference in the mean AHI ratio of 0.25 (95% CI: 0.17–0.33; P < .001) between the sexes (Table 2). A statistically significant difference in this ratio was not found for those over > 60 years old, regardless of BMI, or in those with a BMI < 30 kg/m².

Table 2.

Mean CMS-to-AASM AHI ratios in women and men stratified by age and BMI (95% confidence intervals in parenthesis).

Age (y)	BMI < 30 kg/m2			BMI 30–39 kg/m2			BMI ≥ 40 kg/m2
	W	M	Diff	W	M	Diff	W	M	Diff
<40	0.20	0.31	0.11	0.21	0.34	0.13*	0.32	0.53	0.20*
			(−0.07 to 0.29)			(0.05 to 0.22)			(0.08 to 0.33)
40–60	0.23	0.27	0.03	0.35	0.42	0.07*	0.40	0.65	0.25*
			(−0.05 to 0.11)			(0.01 to 0.13)			(0.17 to 0.33)
>60	0.33	0.36	0.04	0.34	0.42	0.09	0.39	0.49	0.11
			(−0.05 to 0.13)			(−0.01 to 0.17)			(−0.08 to 0.29)

*Statistically significant difference in mean apnea-hypopnea index (AHI) ratio between women and men in the same subgroup; P < .05. AASM = American Academy of Sleep Medicine, BMI = body mass index, CMS = Centers for Medicare and Medicaid Services, Diff = difference in mean CMS to AASM ratios between men and women, M = men, W = women.

OSA diagnosis was missed by using CMS criteria compared to AASM criteria in 48.9% of women and in 28.5% of men. In a univariate regression model, the odds of OSA being missed by CMS scoring in women was 1.89 times the odds in men (95% CI: 1.40–2.53; P < .001). Table 3 shows the odds ratio of a CMSmiss in women compared to men as stratified by age and BMI. The association between CMSmiss and sex was statistically significant in the middle age group (40–60 years old) but not in the older age group (>60 years old). The strength of the association appeared greatest for the morbidly obese middle-age group (age 40–60 years old, BMI ≥ 40 kg/m²), with an odds ratio of 6.87 (95% CI: 1.57– 30.13; P = .011). In the younger age group (age <40 years old), morbidly obese (BMI ≥ 40 kg/m²) women were more likely to be a CMSmiss than men, but the finding did not reach statistical significance at an odds ratio of 3.58 (95% CI: 0.92–13.92; P = .066).

Table 3.

Odds ratio of diagnosing OSA by AASM but not CMS (CMSmiss) scoring criteria in women compared to men stratified by age and BMI (95% confidence intervals in parenthesis).

Age (y)	BMI < 30 kg/m2	BMI 30–39 kg/m2	BMI ≥ 40 kg/m2
<40	2.44 (0.55–10.83)	1.79 (0.69–4.64)	3.58 (0.92–13.92)**
40–60	2.55 (1.10–5.91)*	2.40 (1.27–4.53)*	6.87 (1.57–30.14)*
>60	1.64 (0.68–3.96)	1.44 (0.57–3.66)	1.28 (0.30–5.47)

**P = .066. *P < .05. AASM = American Academy of Sleep Medicine, BMI = body mass index, CMS = Centers for Medicare and Medicaid Services, OSA = obstructive sleep apnea.

In a multiple regression model that included age, BMI, and sex, both older age and higher BMI were associated with reduced odds of missing an OSA diagnosis by CMS criteria, while the odds of a CMSmiss in women was 2.18 (95% CI: 1.59– 2.97; P < .001) compared to men. After adding the percentage of REM sleep time and mean oxygen saturation to the model, the sex effect on missing an OSA diagnosis by CMS criteria remained statistically significant (OR 1.87; 95% CI: 1.36–2.58; P < .001) as shown in Table 4.

Table 4.

Odds ratio of diagnosing OSA by AASM but not CMS scoring criteria (CMSmiss) in a multiple regression model including sex, age and BMI, mean saturation and percentage REM sleep.

	Odds Ratio	95% CI	P
Sex*	1.87	1.36–2.58	<.001
Age**	0.98	0.97–0.99	.001
Body mass index***	0.97	0.96–0.99	<.001
Mean saturation†	1.20	1.13–1.27	<.001
REM sleep percentage‡	0.99	0.97–1.01	.038

*Women compared to men. **Scaled in 1-year units. ***Scaled in 1 kg/m² units. ^†Mean saturation is mean oxygen saturation during sleep and is scaled in 1% units. ^‡REM sleep percentage is percentage of sleep time spent in rapid eye movement sleep and is scaled in 1% units. AASM = American Academy of Sleep Medicine, BMI = body mass index, CI = confidence interval, CMS = Centers for Medicare and Medicaid Services, OSA = obstructive sleep apnea, REM = rapid eye movement.

DISCUSSION

Utilizing different hypopnea scoring criteria to define OSA has been shown to significantly impact the prevalence and severity classification of OSA and even eligibility for CPAP therapy.^3,4,15 The use of the more strict CMS criteria to score hypopneas, compared to AASM criteria, results in fewer patients being eligible for treatment of OSA in both women and men.^3,4 However, the differential impact of varying hypopnea scoring criteria on the clinical care of women vs men has not been fully explored.^12,15,16 This study not only examined the role that different hypopnea definitions have on CPAP eligibility for women vs men, but uniquely investigated how weight and age may influence this relationship.

Not unexpected, using CMS scoring criteria (vs AASM criteria) doubled the odds of a missed diagnosis of OSA in women compared to men with a similar symptom burden as measured by daytime sleepiness. This finding did not significantly change even after accounting for age, BMI, and polysomnographic variables (percent REM sleep time and mean oxygen saturation). Subgroup analysis after stratification based on age and BMI showed that the odds of a missed diagnosis was 6-fold higher in middle-aged (40–60 years) women with obesity (BMI >40 kg/m²) than in comparable men. It is interesting to note that while weight gain appeared to lessen the discrepancy between AHIs determined by CMS and AASM scoring criteria, sex was a considerably more robust determining factor and superseded the relationship between BMI and differences in AHI by scoring criteria. Indeed, the women in this study had an average BMI nearly 5 kg/m² higher than the men. As all patients undergoing PSGs in this study had symptoms concerning for OSA, the clinical implication of these findings is that half as many women as men would qualify for CPAP therapy using CMS scoring criteria, and this is particularly notable in the middle-aged population with obesity. This suggests a significant sex bias in eligibility for OSA treatment directly attributable to CMS scoring guidelines.

The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications recommends that the definition used for hypopnea scoring needs to be specified in the PSG report and acknowledges that the hypopnea criteria may impact the patient’s ability to qualify and be reimbursed for OSA treatment.² Previous work has shown that utilizing CMS scoring criteria leads to a lower overall AHI and thus decreases eligibility for CPAP in up to 17% of patients, although this effect was mostly driven by those younger than 65 years old (20% decrease in eligibility).⁴ Similarly, our study found that age was an independent factor influencing CPAP eligibility based on differing scoring criteria, with the likelihood of a CMSmiss being higher in younger patients, particularly those < 60 years old (Table 2 and Table 3). While sex was a much stronger predictor of a CMSmiss than age, the combination of the 2 factors further increased the sex bias of CMS scoring criteria (Table 2 and Table 3). This age–sex interaction is not surprising as it has been recognized for some time that OSA incidence rates in women and men equalize around the sixth decade of life (50s), felt to be the result of women transitioning through menopause.^17,18 So while utilizing CMS scoring criteria vs AASM scoring criteria might be less of a concern in the traditional Medicare population (> 65 years of age), insurer’s following CMS scoring criteria for CPAP eligibility may differentially deprive younger women of appropriate therapy. Multiple potential etiologies (ie, sociocultural factors, health care professional bias, etc.) are suspected to contribute to the underdiagnosis and undertreatment of women with OSA,^11,12,19 and the addition of a biased scoring system will further compound this problem.

Untreated OSA has been associated with several unfavorable health outcomes, including impaired quality of life and adverse cardiovascular consequences. Intermittent hypoxia, sleep disruption, and sympathetic nervous system activation may all play a role in promoting a proinflammatory state and increasing oxidative stress, resulting in neurocognitive and cardiovascular morbidities.^20,21 Treatment of OSA with CPAP therapy improves alertness and quality of life in many patients,^22,23 although controversy exists regarding the impact of CPAP therapy on long-term cardiovascular outcomes.^24,25 Furthermore, treatment of mild OSA (AHI < 15 events/h) in the absence of comorbid conditions has been a subject of debate. There is some evidence supporting the treatment of mild OSA in patients with OSA-related symptoms,^25,26 although indications for therapy to improve other outcomes has been inconsistent.²⁶ In a multicenter parallel randomized control trial, treatment of mild OSA compared to standard care led to improvement in quality of life.²⁷ In our patient population, the mean CMS and AASM AHIs for women were 10.1 and 23.9 events/h, respectively, which would place them in the mild-to-moderate OSA category. As women with OSA tend to report lower functional status, more self-reported daytime sleepiness, more mood disturbances, and worse neurobehavioral performance compared to men of comparable age, BMIs, and AHIs,²⁸ it appears likely that they would stand to benefit as much, if not more so, than men from treatment of mild-to-moderate OSA.

Our study has some strengths and limitations. The strengths included that that all the PSGs were full-night diagnostic studies and were scored by both standardized scoring criteria at the time of the studies and that a number of important factors that could influence scoring outcomes were accounted for in the multiple regression analysis. The higher rates of PSG in women than men during the period of the study review is likely accounted for by the facts that our Sleep Lab receives a higher number of referrals for women than men (in part related to our Bariatric Surgery Center) and that men had a higher number of split studies than women (183 vs 138), which were not included in the analysis. Limitations to the study include that the cohort of patients was obtained from a single urban academic center, which may limit generalizability. Additionally, the study was conducted retrospectively and as a result we could not control for all variables that may influence the associations found (ie, time spent in the supine position). Interscorer variability may also have had some impact on the findings, although the sleep lab is fully AASM accredited and all technicians participate actively and successfully in the AASM’s Interscorer Reliability program. And finally, the study was based on single-night PSG studies, and significant night-to-night variability in the AHI, and thus OSA severity, can occur in a nontrivial minority of patients.²⁹ However, this effect is not influenced by sex²⁹ and thus unlikely to play a significant role in the differential impact of scoring criteria on treatment eligibility.

The findings from this study suggest that CMS scoring criteria imparts a sex bias that is unfavorable toward women and will result in symptomatic women with OSA being denied appropriate therapy. Larger, prospective cohort studies are needed to confirm these findings as this result may have significant policy implications.

DISCLOSURE STATEMENT

All authors have seen and approved this manuscript. The authors report no conflicts of interest.

ABBREVIATIONS

AASM,

American Academy of Sleep Medicine

AHI,

apnea-hypopnea index

BMI,

body mass index

CI,

confidence interval

CMS,

Centers for Medicare and Medicaid Services

CPAP,

continuous positive airway pressure

OSA,

obstructive sleep apnea

PSG,

polysomnogram

REM,

rapid eye movement