Obstructive Sleep Apnea Modulates Clinical Outcomes Post-Acute Myocardial Infarction: A Large Longitudinal Veterans’ Dataset Report

Abstract

Background:

While the longer-term Obstructive Sleep apnea (OSA)-related intermittent hypoxia (IH) leads to various comorbidities, it has become increasingly evident that OSA confers protective advantages during and after acute myocardial infarction (AMI). We hypothesized in patients who were admitted with acute MI, the presence of OSA is associated with lower in-hospital mortality compared to those without a prior diagnosis of OSA.

Methods:

In this nationwide retrospective study utilizing Veterans Health Administration records, we included patients hospitalized for MI with a history of sleep disorders from 1999–2020. We divided patients into two cohorts: those with OSA and those without OSA. The primary outcome was in-hospital mortality during AMI hospitalization. We analyzed the data using logistic regression and calculated the odds ratio of in-hospital mortality.

Results:

Out of more than four million veterans with any sleep diagnosis, 76,359 patients were hospitalized with a diagnosis of AMI. We observed 30,116 with OSA (age, 64±10 years; BMI, 33±7 kg/m²) and 43,480 without OSA (age, 68±12 years; BMI, 29±6 kg/m²). The aOR of in-patient mortality (n = 333 (1.1%)) was lower in those with OSA (aOR, 0.43; 95% CI, 0.38 to 0.49) compared to without-OSA (n = 1,102, 2.5%). However, the OSA cohort had a higher proportion of the prolonged length of stay (28.1%).

Conclusions:

Presence of OSA is associated with lower in-hospital mortality among patients admitted for AMI, after adjusting for various demographic and co-morbidity factors. This study highlights the complex relationship between OSA and cardiovascular health and highlights the need for further research in this area.

Brief Summary

Current Knowledge/Study Rationale:

Obstructive sleep apnea often leads to intermittent hypoxia due to repetitive airway collapse. The study aimed to investigate the short-term impact of obstructive sleep apnea (OSA) on myocardial infarction (MI) related hospitalization.

Study impact:

Using a real-world longitudinal dataset analysis, the study found that patients with OSA had a lower in-hospital mortality rate compared to patients without OSA following an acute MI episode. This association remained significant even after adjusting for demographics and co-morbid conditions. However, patients with OSA had a prolonged length of stay in the hospital, defined as a stay over 6 days. The study provides valuable insights into the short-term impact of OSA on outcomes following MI, which has implications for healthcare resource allocation and planning.

Introduction

Obstructive sleep apnea (OSA) is a known risk factor for comorbidities such as coronary artery disease, cerebrovascular disease, and cognitive changes[1–5]. OSA-related intermittent hypoxia (IH) is a key component of the underlying pathophysiology. It is characterized by recurrent, intermittent, and short-duration episodes, often leading to severe desaturation during sleep[6, 7]. While longer-term IH is associated with various adverse outcomes, recent research suggests that it can also confer protective advantages during and after acute vital organ injuries, such as acute myocardial infarction (MI) and stroke [8, 9]. This phenomenon, known as ischemic or hypoxic preconditioning (HP), occurs when chronic IH is administered below the damage threshold, resulting in tolerance to severe ischemia exposure [10]. Ultimately this leads to a reduction in the size of myocardial infarction.

While animal studies have repeatedly demonstrated the HP phenomenon [11], the evidence related to HP is indirect and inferential in humans. While several studies have reported that OSA patients with acute MI may also have cardioprotective effects[12–14], others have offered contrary data, such as a larger infarct size in OSA patients[15]. Nonetheless, literature on OSA and ischemic cardioprotection remain quite limited. Moreover, previous studies were conducted on small sample sizes, therefore, additional research is necessary to delineate the complicated and varied effects of IH on the outcome of acute MI.

In this study, we evaluated the largest integrated healthcare system in the United States, the Veterans Health Administration (VHA)[16], to gather more indirect evidence in a real-world setting if OSA is associated with lower mortality during hospitalization for MI. We further studied if there is a difference in mortality among patients with a prior history of MI compared to those with the first MI event. Finally, since MI management has progressively evolved over the years, we also analyzed stratified outcome differences by comparing cases managed prior to 2010 vs. those after 2010.

2. Methods

Baylor College of Medicine’s institutional review board (H-35366) and Michael E. DeBakey Veteran Affairs medical center research and development committee approved this research.

2.1. Study design and setting

This retrospective cohort study used the VHA Electronic Medical records (EMR). We included veterans with either in-patient or outpatient encounters from October 1st, 1999, to September 30^th, 2020, with any International Classification of Disease, 9th edition (ICD-9) or 10th edition (ICD-10) sleep disorders diagnosis (N = 4,237,444). In this patient group, we identified patients who had a diagnosis of acute myocardial infarction (MI) as a principal diagnosis in the in-patient discharge note (n = 73,399) (Figure 1). The index date was defined as the date of hospitalization with acute MI as the principal diagnosis. Among the eligible participants, two cohorts were defined: a) patients with OSA diagnosis within twelve months before or after the acute MI hospitalization (n = 30,001), and b) patients without OSA within twelve months before or after the acute MI hospitalization (n = 43,225). The OSA was considered a confirmed diagnosis based on two separate outpatient encounters with OSA as the primary diagnosis at least 30 days apart or one in-patient encounter[17]. We excluded the outliers that had a length of stay (LOS) of more than two months since we suspect those patients were in a post-acute care facility (n =166), or in cases where the data was clinically inconsistent and likely inaccurate (death was prior to admission date)(n = 7). All relevant ICD-9 and ICD-10 codes utilized for this analysis are listed in Supplementary Table 1. The sensitivity and specificity data for our OSA definition have been previously published [18].

Figure 1.

Strobe diagram of the patients who had myocardial infarction (MI) with obstructive sleep apnea (OSA) and without OSA in the cohort of veterans from 10/1999 to 10/2020.

2.2. Study variables

The primary outcome was in-hospital mortality during the index acute MI admission. In-hospital mortality was defined as any death between the admission and discharge interval. Mortality data was gathered from the CDW VHA Vital Status table (sensitivity 98.3% and specificity 99.8% relative to National Death Index)[19, 20]. The secondary outcome was the LOS in days during the acute MI hospitalization. We also calculated prolonged LOS, which we defined as all the LOS that were ≥ the 75^th percentile in the cohort (≥ 6 days).

We collected patients’ demographics such as age (categorized as <50, 50-<65, 65-<75, and ≥75 years), BMI (categorized as <18.5, 18.5–30, and ≥30 kg/m²), sex, race (White, Black, and others), ethnicity, and Charlson comorbidity index (CCI) [21] and relevant co-morbid conditions. The CCI was calculated with in-patient and outpatient co-morbid conditions within a year before the index date.

2.3. Statistical Analysis

The continuous variables are presented as mean with standard deviation, and categorical variables as numbers and percentages using SPSS, version 27 (SPSS Inc, Chicago, Illinois). A p value < 0.05 was considered significant.

Kaplan–Meier curves were used to demonstrate the relationship between those with/without OSA and in-hospital mortality. We used the R-packages (‘survival’, ‘survminer’, ‘ggplot2’). First, we used logistic regression to estimate the odds ratio (OR) for in-hospital mortality. We then estimated the adjusted OR (aOR) by initially adjusting for age, sex, race, ethnicity, and BMI and then by adding the Charlson Comorbidity index (CCI) to the previous list of factors. We present the complete list of variables utilized to estimate the adjusted OR in Supplemental Figure 2 and Supplemental table 2. We then performed additional analysis stratifying by history of prior MI; and given MI management has progressively evolved over the years, we stratified by occurrence of cases prior to 2010 vs. those 2010 and onwards. We also studied the association using Kaplan-Meier curves with 95% CIs (the survminer R package v0.4.9).

3. Results

3.1. Patients:

Baseline characteristics of patients by presence or absence of OSA are shown in Table 1. Compared to the non-OSA cohort, the OSA cohort was younger (mean age 64.3 ± 10.3 vs. 67.9 ± 11.6 years, p <0.001) and had a higher proportion of obese patients (65.1% vs. 36.2%, p <0.001). Most patients were men (98%, p < 0.008), as seen in US Veteran datasets. CCI was similar in both groups. Among the co-morbid conditions, the presence of diabetes mellitus, either with (26.2% vs. 19.1%, p <0.001) or without (53.1% vs. 42.8%, p <0.001) complications, was higher in the OSA cohort.

Table 1.

Characteristics and demographics of patients who had myocardial infarction (MI) with obstructive sleep apnea (OSA) and without OSA

	Without OSA	With OSA
N	43,225	30,001
Gender-Male, N (%)	42,172(97.6)	29,360(97.9)
Race
White, N (%)	34,015(78.7)	22,931(76.4)
Black, N (%)	6,254(14.5)	5,012(16.7)
Others, N (%)	2,956(6.8)	2,058(6.9)
Ethnicity- Hispanic, N (%)*	2,771(6.4)	1,980(6.6)
Age	67.9(11.6)	64.3(10.3)
Age < 50, N (%)	1,964(4.5)	2,112(7.0)
Age 50 – <65, N (%)	15,822(36.6)	13,271(44.2)
Age 65 – <75, N (%)	12,977(30.0)	10,136(33.8)
Age ≥ 75, N (%)	12,462(28.8)	4,482(14.9)
BMI	28.6(5.9)	32.9(6.48)
BMI 18.5–30	26,585(61.5)	10,289(34.3)
BMI < 18.5	1,021(2.4)	169(0.6)
BMI ≥ 30	15,619(36.1)	19,543(65.1)
Comorbidities
CCI ≥ 2, N (%)	38,720(89.6)	27,138(90.5)
Myocardial Infarction, N (%)	24,583(56.9)	17,906(59.7)
Congestive heart failure, N (%)	11,457(26.5)	8,040(26.8)
Peripheral vascular disease, N (%)	7,937(18.4)	5,040(16.8)
Cerebrovascular disease, N (%)	7,022(16.2)	4,145(13.8)
Dementia, N (%)	1,431(3.3)	627(2.1)
Diabetes mellitus without complication, N (%)	18,486(42.8)	15,921(53.1)
Diabetes mellitus with complication, N (%)	8,236(19.1)	7,844(26.1)
Renal Disease, N (%)	8,462(19.6)	6,045(20.1)
Moderate to severe liver disease, N (%)	1,358(3.1)	1,075(3.6)

OSA = Obstructive Sleep Apnea, MI = myocardial, infarction, CCI = Charlson Comorbidity Index. M(SD) = mean and standard deviation.

^*Ethnicity is the dataset is reported as Hispanic and Non-Hispanic.

3.2. Outcomes:

The primary outcome of inpatient mortality differed significantly in the OSA cohort compared to the non-OSA cohort (1.1% vs. 2.5%). The unadjusted odds of in-hospital mortality were 57% lower in patients with OSA compared to those without OSA (OR, 0.43, 95%CI: 0.38, 0.49, p-value < 0.001). After adjusting for age, sex, race, BMI, and CCI, the odds of in-hospital mortality remained lower (aOR, 0.54, 95%CI: 0.47, 0.62) in the patients with OSA (Table 2) compared to without OSA. Kaplan Meier curves depicting in-hospital mortality (with 95% confidence intervals) for the first 21 days are shown in figure 2. OSA and non-OSA cohorts differed as early as day 2 and the curve plateaued around day 12.

Table 2.

Odds of in-hospital mortality when comparing patients with obstructive sleep apnea (OSA) versus those without OSA^*.

	Raw OR (95%CI)	Adjusted OR (95%CI)†	Adjusted OR (95%CI)‡
Non stratified analysis	0.43(0.38, 0.49)	0.54(0.47, 0.62)	0.54(0.47, 0.62)
Stratified analysis
Without history of MI	0.44(0.35, 0.54)	0.54(0.43, 0.67)	0.49(0.39, 0.62)
With history of MI	0.42(0.36, 0.49)	0.53(0.45, 0.62)	0.49(0.42, 0.58)
Prior to 2010	0.42 (0.37,0.49)		0.48(0.41, 0.56)
2010 onwards	0.34 (0.26,0.44)		0.54(0.41, 0.72)

^*Non-OSA cohort as a reference.

In all the cases p <0.001

OR (95%CI) = odds ratio and 95 percentage confidence interval.

^†Adjusted odds ratio with sex, race, ethnicity, BMI, and age.

^‡Adjusted odds ratio with sex, race, ethnicity, BMI, age, and Charlson Comorbidity Index.

OSA = obstructive sleep apnea, BMI = Body Mass Index

Figure 2:

Kaplan Meier curves depicting in-hospital mortality (with 95% confidence intervals) for the first 21 days

We observed the no significant difference in mean of LOS between OSA (7.51±9.37 days) and non-OSA (7.56±8.31) groups. Prolonged LOS (defined as ≥75^th percentile LOS, ≥ 6 days) in non-OSA group was 21.7% whereas it was 28.1% in OSA group.

We observed that patients with a BMI <18.5 kg/m2 compared to a BMI of 18.5–30 kg/m2 as a reference had higher odds of mortality (OR, 1.77, 95%CI: 1.35, 2.32). The odds of in-hospital mortality progressively increased with age. They were highest in the oldest veterans of age ≥ 75 years (OR, 5.94, 95%CI: 3.47, 10.17). We observed a similar trend as the CCI increased. Furthermore, the Hispanic ethnicity was more vulnerable to in-hospital acute MI mortality (OR, 1.2795% CI: 1.05, 1.54) when compared to the White race as a reference.

3.2. Stratified analysis:

With stratification based on prior history of MI, we did not find a significant difference in mortality whether a patient had a history of previous MI. However, we observed that OSA patients without a history of MI had lower adjusted odds of mortality (aOR, 0.49, 95%CI:0.39, 0.62). Similarly, among patients with a history of MI there was a 51% lower adjusted odds of mortality (aOR, 0.49, 95% CI:0.42, 0.58) for the OSA cohort when compared to the non-OSA cohort.

Temporal changes did not impact the mortality odds either. When we evaluated the period prior to 2010, the estimated odds of in-hospital mortality was 58% lower (OR, 0.42, 95%CI: 0.37, 0.49) in patients with OSA compared to those without OSA. Similarly, from 2010 onwards, inpatients with OSA had improved adjusted odds of in-hospital mortality (aOR, 0.34 (95%CI:0.26, 0.44) when compared to those without OSA

4. Discussion

We report one of the largest nationwide, retrospective administrative dataset analyses spanning over two decades in elderly veterans. Withing the constraints inherent to a dataset study, our finding suggests that patients with OSA died less during acute MI hospitalization than those without OSA. Moreover, the association remained significant after adjusting for demographics, co-morbid conditions, and prior MI history. In contrast, a higher proportion of patients with OSA stayed longer than six days in the hospital although the overall length of stay was not different.

Our results offer real-world healthcare system-based confirmation of prior studies showing the cardioprotective association of OSA during acute MI hospitalization. The acquired pre-conditioning from nightly intermittent hypoxic episodes in patients with OSA presumably results in increased collateral circulation in the myocardium [12, 22] and consequently reduced myocardial infarct size as estimated using serial cardiac troponin I levels [23] or by peak troponin I levels [24]. OSA cohort also had a survival advantage with lower overall in-hospital and intensive care unit mortality[25, 26]. Furthermore, we and others have reported that patients with OSA have superior/non-inferior postoperative survival [27, 28]. Co-morbid OSA offers protection during that early period. OSA likely offers an independent mortality advantage since the effect remained significant after adjusting for patient demographics and comorbid conditions.

Various mechanisms have been proposed to explain the reduced mortality rate among acute MI patients with co-morbid OSA. Studies have demonstrated that IH and oxidative stress can trigger the protective mechanism of endothelial cell-colony forming units, which is strongly linked to vascular function and protects vascular health[29]. Rapid IH in patients with OSA has been suggested to increase hypoxia-inducible factor (HIF)-1 concentration, which promotes hypoxia adaptation and cellular survival. This HIF stability results in a transcriptional mechanism that increases extracellular synthesis of adenosine, activating adenosine receptors, which has been linked to cardioprotection[30–32]. Additionally, the hypoxic preconditioning related to myocardial protection displays a biphasic pattern, with the resulting hypoxic preconditioning from OSA’s recurrent and nightly nature likely conferring further protection via both the first and second windows[11]. These potential mechanisms may provide insights into the relationship between OSA and the observed reduced mortality rate among acute MI patients.

Reassuringly our data was similar to another non-veteran nationwide database study involving 1,850,625 patients in approximately 1000 US hospitals by Mohananey et al. This study showed that ST-elevation MI patients with a history of recognized OSA had significantly lower in-hospital mortality than patients without OSA [33]. After controlling for baseline patient characteristics and hospital characteristics, in this cohort, OSA patients had lower in-hospital mortality (aOR, 0.83 [95% CI, 0.81–0.84]; P<0.001). Our study provides additive knowledge to this study. In contrast to Mohananey et al. study, the US Veterans’ dataset had a significantly higher proportion of older patients (64.3 years vs 59.3 years). Despite being older and possibly at a higher risk for mortality, the veterans that had OSA died less during the acute hospitalization. We further extend knowledge where we evaluated the mortality over a longer duration of time (> 2 decades) where practice patterns, guidelines, and therapeutic options will have changed significantly. We found that even with changing practice patterns over the decades, the short-term cardioprotective impact remained significant.

Patients with OSA had a significantly higher proportion of prolonged acute MI hospitalization when compared to non-OSA counterparts. While MI patients with OSA have lower in-hospital mortality, our findings suggest that they may have a complicated hospital course leading to prolonged hospitalization. Prior studies have shown that OSA is independently associated with hospital readmissions and outpatient visits[34, 35]. Thus OSA’s protective benefits appear to be most pronounced during the acute phase; however, in the sub-acute and chronic phases, OSA remains a condition that is associated with poorer outcomes.

This study’s findings of the cardioprotective benefit of OSA during acute MI are thought-provoking and may serve as hypothesis-generating. While the study adds to the literature in meaningful ways, it also has several limitations. Inherent to studies using dataset, a diagnosis made by ICD coding is often less accurate than a clinical chart review and tends to underestimate OSA[36, 37]. Furthermore, granular details such as the role of duration and adherence to PAP therapy pre and post-index date, the severity of sleep apnea, history of smoking, and ejection fraction were not determined. Additional lack of information of CPAP compliance or use of PAP therapy prior to, during or after MI adds to the limitation. Similarly, for MI, finer details, such as ST elevation vs. non-ST elevation MI, troponin levels, infarct size, the role of medications, and coronary interventions, such as angioplasty, are not included in this study. Unmeasured confounders such as time of symptoms to hospitals, ability of facility to revascularize also potentially played a role and were not measured in this study. The male predominant veteran dataset also limits generalizability to the population. Nevertheless, the large size of the data and the real-world scenario are major strengths of this study. Further studies with advanced data science techniques, such as natural language processing, may alleviate some of these concerns.

Conclusions

Co-morbid OSA is associated with lower in-hospital mortality among patients admitted for an acute MI. This association had an independent mortality advantage during an acute MI episode after adjusting for demographics and co-morbid conditions. Additionally, prior acute MI did not alter these outcomes. Further studies will explore type of MI, extent of myocardial damage, and extent of vascular lesions in relation to effect of OSA on MI mortality.