MAD vs. AHI: Sleep patterns, oxygenation among polysomnography diagnosed by OSA patients

ABSTRACT

Background and Objective:

Obstructive sleep apnea is a common sleep disorder characterized by repeated airway obstructive during sleep, leading to breating pauses, reduced sleep quality and decreased blood oxygen levels.

Materials and Methods:

Polysomnographic data in Obstructive Sleep Apnea patients were analysed to explore correlations between Apnea Hypopnea Index, Mean Apnea hypopnea Duration, oxygen desaturation and sleep stages.

Results:

Among 136 Obstructive Sleep Apnea patients, Mean Apnea hypopnea duration showed a mild correlation with Apnea Hypopnea Index but no association with age, weight and height. Patients with longer Mean Apnea hypopnea duration had association (P < 0.05) with poorer oxygenation, lower average saturation, loss of sleep, Mean oxygen desaturation, shorter sleep latency, less Non Rapid Eye Movement 3, more Non Rapid Eye Movement 1 and higher arousal/desaturation. After adjusting Apnea Hypopnea Index, Mean Apnea hypopnea duration remained significantly associated with Epworth Sleep Scale, sleep latency, Non Rapid Eye Movement 1/3, arousal index and all oxygen metrics. These associations were stronger in severe Obstructive Sleep Apnea (n = 90), while in mild-moderate cases (n = 46), only Non Rapid Eye Movement 3 duration correlated negatively with Mean Apnea hypopnea duration (r = –0.14, P = 0.02).

Conclusion:

Mean apnea hypopnea duration emerged as an independent marker of oxygen desaturation and sleep disruption in OSA, beyond the Apnea Hypopnea Index. Its critical relevance was more pronounced in severe cases suggesting added value in assessing disease burden. Incorporating event duration into routine evaluation may refine patient stratification and guide targeted management strategies.

BACKGROUND AND OBJECTIVES

Obstructive sleep apnea (OSA) is a heterogeneous disorder characterized by intermittent pharyngeal obstruction causing recurrent desaturation and arousals. Untreated OSA is associated with increased mortality. Cardiovascular diseases (CVD) and motor vehicle crashes but these outcomes can substantially vary between individuals.

Although the apnea-hypopnea index (AHI) is commonly used to quantify the severity of OSA, this frequency-based measurement provides an incomplete descriptor of physiological stressors and has variable association with symptom burden, outcomes or treatment adherence.[1,2] In recent study, Xiaojun Zhan had proposed the Mean Apnea-Hypopnea Duration (MAD) is a parameter that incorporates the severity of breathing events during sleep. Although the use of the MAD has been previously studied in the pathophysiology of OSA, the relationship between the clinical utility of the MAD compared with the AHI and with other sleep apnea risk factors and patients’ demographics remain unclear.[3,4]

Therefore, the aim of this study was to compare the use of the MAD with the AHI as indicator of clinical and demographic parameters, blood oxygenation and sleep parameters in patients diagnosed with OSA by polysomnography (PSG).

MATERIALS AND METHODS

Study design

This study is a retrospective observational study conducted using previously recorded polysomnography (PSG) data of patients diagnosed with OSA, which aims to compare the clinical relevance of the MAD and the AHI with respect to blood oxygenation parameters and sleep architecture patterns. Patient records will be reviewed from a sleep laboratory database over a defined period. Data including AHI, MAD, oxygen saturation levels and sleep stage distribution will be collected and statistically analysed to determine their strength and associations between these variables.

All sleep studies were scored and interpreted by certified sleep physicians according to the American academy of Sleep Medicine (AASM) scoring manual, version 2.6 (2020).[3] Adults range between 18 and 65 years [Table 1] diagnosed with obstructive sleep apnea (OSA) by overnight polysomnography (PSG), patients with apnea-hypopnea index (AHI). 5 events/hour and a good quality of PSG data available in complete sleep and oxygen parameters are said to be inclusion criteria. Patient with central sleep apnea or other primary sleep disorders (e.g., narcolepsy, restless leg syndrome, chronic lung disease, heart failure, neuromuscular diseases which affects breathing, on CPAP therapy, with incomplete or poor quality in PSG recordings, pregnant women, on medication with sleep architecture (e.g., sedative) are said to be exclusion criteria.

Table 1.

Age, height and weight vs. 2 MAD groups

Parameter	Group 1 (short MAD)		Group 2 (long MAD)		P
	SD	Mean	SD	Mean
Age (years)	56.73±12.05 (n=77)		55.10±14.17 (n=59)		0.47
Height (cm)	162.05±9.52 (n=77)		163.48±8.88 (n=56)		0.38
Weight (kg)	90.63±16.73 (n=77)		84.74±15.67 (n=56)		0.04*

Data collection method

Data for this study have been collected in B7 Sleep Lab records. Ethical approval for this study was obtained from the Institutional Ethics Committee of Sri Ramachandra Institute Higher Education and Research (Approval No: CSP/22/FEB/105/31), timeframe from 2021-2022. Patients who underwent overnight diagnostic polysomnography (PSG) and were diagnosed with Obstructive Sleep Apnea (OSA) has been identified from the lab’s database. Relevant demographic information (age, sex, BMI), along with PSG parameters including AHI, MAD, oxygen saturation levels (mean SpO², minimum SpO², Oxygen Desaturation Index) and sleep architecture (total sleep time, sleep efficiency, distribution of NREM and REM stages) will be extracted.

The collected data have been entered into a structured data sheet and analysed using statistical software such as SPSS. Descriptive statistics have been used to as baseline characteristics. For correlation analyses, Pearson’s correlation coefficient was applied to normal distributed variables while Spearman’s rank correlation was used for non-normally distributed variables to assess the relationships between AHI, MAD, oxygenation parameters and sleep stages [Table 2]. Comparative analysis across OSA severity groups (mild, moderate, severe) was performed using one way ANOVA.

Table 2.

Correlation between N1 sleep duration, AI, ODI vs. 2 MAD groups

Parameter	Short MAD		Long MAD		P
	SD	Mean	SD	Mean
N1D (min)	50.85±32.40		49.49±27.78		0.8
ODI	31.98±27.36		30.22±22.99		0.7
AI	33.59±24.90		33.06±21.94		0.9

Ethical consideration

All relevant parameters were compared between both MAD groups (short and long MAD) using appropriate statistical tests including Pearson’s correlation and independent t-tests. Statistical significance was considered at P < 0.05.

Statistical analysis

All collected data are combined and analysed using statistical software such as SPSS (Statistical Package for the Social Sciences) has been used to summarize demographic variables and baseline polysomnographic parameters. Continuous variables (e.g., AHI, MAD, SpO² levels, sleep stage percentages) represent as mean ± standard deviation (for normally distributed data) or median with interquartile range (for non-normally distributed data) [Table 3]. To assess the relationship between AHI and MAD with blood oxygenation parameters (e.g., minimum SpO², oxygen desaturation index) and sleep architecture (e.g., REM sleep percentage, sleep efficiency) the following tests will be applied, Pearson’s correlation coefficient for normally distributed variables, Independent t-tests to compare mean values between two groups, One-way ANOVA test for comparison across multiple OSA severity groups (mild, moderate, severe). Multiple linear regression analysis may be conducted to evaluate the predictive value of AHI and MAD for outcomes such as oxygen desaturation and sleep disruption [Table 4]. A P value <0.05 will be considered statistically significant

Table 3.

Baseline characteristics of the study population (n=136)

Parameter	n (%)
Total number of patients	136
Age range (years)	18-75
Gender (Male: Female)	85 (62.5%): 51 (37.5%)
Subgroup with MAD data	136 patients
Mild OSA (AHI 5-14)	46 patients
Moderate OSA (AHI 15-29)	21 patients
Severe OSA (AHI ≥30)	69 patients
Mean MAD (Mean apnea hypopnea duration)	23.59 seconds
MAD distribution	Non-normal
MAD-AHI correlation	Mild, significant
MAD vs. age, weight, height	No significant association

Table 4.

Regression analysis of PSG parameters and clinical variables

Predictor	β coefficient	SE	P
AHI	–0.21	0.09	0.04*
MAD	–0.29	0.11	0.01*
Age	–0.08	0.07	0.26
BMI	–0.05	0.06	0.41

RESULTS

Baseline characteristics of patients

Baseline characteristics of patients The patients ranged from age group 18 to 75 years, with a predominant male population considering of 85 men(62.5%) and 51 women (37.5%). The study population was further categorized based on OSA severity: mild (n = 46), moderate (n = 21), and severe (n = 69), for a subgroup analysis of 136 patients for which Mean Apnea-Hypopnea Duration (MAD) data were available. There were no significant differences in age, body weight or height between the MAD subgroups (<23.59 sec and >23.59 sec). The median MAD in the population was 23.59 seconds, which has been used as the cutoff to stratify patients into short and long MAD groups for comparative analysis [Tables 1-3].

Regression analysis

In multiple regression [Tables 4-7] models including AHI, MAD, age, sex and BMI, MAD remained a significant independent predictor of Lowest Oxygen Desaturation (LOD) (β = –0.29, P = 0.01) even after adjusting for AHI (β = –0.21, P = 0.04). The overall model explained 32% of the variance in LOD (adjusted R² = 0.32). This suggests that longer apnea duration contributes uniquely to oxygen desaturation severity independent of apnea hypopnea frequency. These parameters were compared between both MAD groups and showed no statistically significant difference.[5,6]

Table 7.

Correlation betweem 2MAD groups and all other parameters

Parameter	Short MAD (mean)	Long MAD (mean)	P
MAD	20.02 sec	28.26 sec	0.00*
LOD (Lowest oxygen desaturation)	77.23%	75.69%	0.01*
N2S (% of stage N2 sleep)	20.82%	13.71%	0.03*

DISCUSSION

Evaluation of mean apnea hypopnea duration and its association with blood oxygenation and sleep architecture in obstructive sleep apnea patients

This retrospective study to evaluate the utility of the MEAN MAD as a complementary metric to the traditional AHI in assessing the physiological impact of OSA. While AHI is the gold standard for diagnosing OSA by quantifying the frequency of respiratory events, it inherently fails to capture the duration of these events. Our findings suggest that MAD provides crucial insights into the severity of hypoxemia and sleep disruption, offering a more understanding of the disease burden.

Demographic factors and MAD

Our analysis of demographic factors revealed that only body weight had a significant association with MAD, while age height did not. This finding is consistent with the established pathophysiology of OSA. Higher body weight particularly increases Body Mass Index (BMI) contributes to greater adipose tissue deposition around the pharynx, leading to increased airway collapsibility and reduce upper airway muscle tone. This anatomical predisposition can logically lead to more prolonged and severe airway obstructions during sleep thereby increasing the MAD. This aligns with previous research such as that by Hazhar and Hadi,[7] which demonstrated a significant correlation between BMI and AHI, indicating that OSA severity worsens with higher BMI values.[8]

Blood oxygen parameters and sleep architecture

A critical finding of our study was the specific relationship between MAD and blood oxygenation. When comparing [Table 5] Average Oxygen Desaturation (AOD), Mean Oxygen Desaturation (MOD) and Lowest Oxygen Desaturation (LOD) only LOD was significantly associated with MAD. This highlights a key physiological mechanism; longer cessations of breathing provide a greater window for blood oxygen levels to fall resulting in a more profound of oxygen saturation. This suggests that MAD is a potent indicator of the severity of intermittent hypoxemia a primary driver of OSA related cardiovascular and neurological complications.[9] This supports reports cardiovascular and neurological complications. This supports reports from Hadice Selimoglu Sen[10] and Teki,[11] who found that apnea hypopnea duration is a useful indicator of blood oxygenation independent of AHI.[12]

Table 5.

Blood oxygen (average oxygen desaturation, lowest oxygen desaturation and mean oxygen desaturation) vs. 2 MAD groups

Parameter	Group 1 (short MAD)		Group 2 (long MAD)		P
	SD	Mean	SD	Mean
AOD (average O2 desaturation)	92.66±4.87		93.31±4.52		0.43
MOD (minimum O2 desaturation)	77.23±11.66		75.69±13.19		0.47
LOD (level of desaturation)	5.12±2.21		6.47±3.28		0.01*

Furthermore, we exposed the link between MAD and sleep stages (N1, N2, N3, REM) [Table 6]. The analysis revealed that prolonged MAD was significantly correlated the majority of total sleep time prolonged and severe respiratory events during this stage can have an outsized impact on the patient’s overall oxygen burden and sleep fragmentation.[13,14] this suggests that the physiological instability during N2 sleep may be a key factor in determining the severity of oxygen desaturation in patients with long duration events, a finding that complements S. Hussains report on the variation of event duration between sleep stages.[15]

Table 6.

Correlation between sleep latency vs. 2 MAD groups

Variable	Group 1 (short MAD)		Group 2 (long MAD)		P
	SD	Mean	SD	Mean
Non rapid eye movement stage 1 (%)	17.50±27.88		22.06±40.82		0.44
Non rapid eye movement stage 2 (%)	20.82±20.73		13.71±15.17		0.03*
Non rapid eye movement stage 3 (%)	46.91±52.90		34.73±47.87		0.16
Rapid eye movement	86.18±110.41		92.08±126.42		0.78

Clinical implications

The integration of MAD into clinical assessments has significantly potential implications.

1. Improved Risk Stratification: Clinicians could use MAD alongside AHI to better identify high risk patients. An individual with a moderate AHI but a long MAD may be at greater risk for comorbidities than a patient with a severe AHI composed of many short events.

2. Personalized Treatment: This metric could guide more personalized therapy. Patients with a long MAD might be candidates for more aggressive treatment such as CPAP and require closer follow-up.

This study demonstrates that MAD is not just a redundant metric but a valuable tool that captures the durational aspect of respiratory events. Its significant association with body weight profound oxygen desaturation (LOD) and disruptions during the critical N2 sleep stage indicates its utility in providing a deeper understanding of the physiological severity of OSA. While MAD may not replace AHI its use as a complementary metric could lead to more accurate risk assessment and more effective patient management.

CONCLUSION

This study highlights the potential value of Mean Apnea Hypopnea Duration (MAD) as an adjunct parameter to the Apnea Hypopnea Index (AHI) in the evaluation of Obstructive Sleep Apnea (OSA). Our findings indicate that MAD appears to be independently associated with specific markers of the disease severity indicating higher body weight the degree of oxygen desaturation (LD) and respiratory events occurring during NREM stage 2 sleep [Table 7].

However, the lack of an overall statistically significant difference between the short and long MAD groups suggests that MAD as a standalone metric may not be sufficient to independently predict overall clinical severity. Instead, its strength seems to lie in providing a more nuanced view of the physiological burden particularly the durational aspect of hypoxemia which is not captured by the frequency-based AHI.

In summary while MAD may complement AHI in assessing the disease burden by offering additional insight into the patient’s physiological status larger prospective studies are nodded to validate these preliminary findings before its clinical adoption can be recommended. Future research should focus on determining how MAD correlates with long term clinical outcomes such as cardiovascular events to fully establish its role in routine OSA management.

Limitations

• Retrospective design, single centre sample reduce generalization.

• PSG scoring, variation in MAD may affect accuracy.

• Limited oxygenation metrics AOD, MOD, LOD. No CO₂ data.

• No link to clinical outcomes such as symptoms, cardiovascular events or treatment response.

Recommendations

• Conduct larger, prospective and multicentre studies.

• Standardize MAD definitions and include stage and position specific analysis like NREM, REM, Supine/Lateral.

• Tests MAD prognosis value for symptoms, cardiovasular risk and therapy outcomes.

Conflict of interest

The author declares no potential conflicts of interest with respect to the research, authorship or publication of this thesis. No financial support was received from commercial entities related to mandibular advancement devices, polysomnography systems or oxygen monitoring technologies.

Funding Statement

Nil.