The Effect of Chronic Altitude Exposure on Chronic Obstructive Pulmonary Disease Outcomes in the SPIROMICS Cohort: An Observational Cohort Study

Abstract

Rationale

Individuals with chronic obstructive pulmonary disease (COPD) have airflow obstruction and maldistribution of ventilation. For those living at high altitude, any gas exchange abnormality is compounded by reduced partial pressures of inspired oxygen.

Objectives

Does residence at higher altitude exposure affect COPD outcomes, including lung function, imaging characteristics, symptoms, health status, functional exercise capacity, exacerbations, and mortality?

Methods

From the SPIROMICS (Subpopulation and Intermediate Outcome Measures in COPD Study) cohort, we identified individuals with COPD living below 1,000 ft (305 m) elevation (n = 1,367) versus above 4,000 ft (1,219 m) elevation (n = 288). Multivariable regression models were used to evaluate associations of exposure to high altitude with COPD-related outcomes.

Measurements and Main Results

Living at higher altitude was associated with reduced functional exercise capacity as defined by 6-minute-walk distance (−32.3 m [95% confidence interval, −49.8 to −14.8 m]). There were no differences in patient-reported outcomes as defined by symptoms (COPD Assessment Test and modified Medical Research Council dyspnea scale), or health status (St. George’s Respiratory Questionnaire). Higher altitude was not associated with a different rate of FEV₁ decline. Higher altitude was associated with lower odds of severe exacerbations (incidence rate ratio, 0.65 [95% confidence interval, 0.46 to 0.90]). There were no differences in small airway disease, air trapping, or emphysema. In longitudinal analyses, higher altitude was associated with increased mortality (hazard ratio, 1.25 [95% confidence interval, 1.0 to 1.55]); however, this association was no longer significant when accounting for air pollution.

Conclusions

Long-term altitude exposure is associated with reduced functional exercise capacity in individuals with COPD, but this did not translate into differences in symptoms or health status. In addition, long-term high-altitude exposure did not affect progression of disease as defined by longitudinal changes in spirometry.

Clinical trial registered with www.clinicaltrials.gov (NCT 01969344).

At a Glance Commentary

Scientific Knowledge on the Subject

There is a paucity of data evaluating the long-term effect of chronic high-altitude exposure on individuals with chronic obstructive pulmonary disease. Prior population studies have shown an association of increased mortality for individuals with chronic obstructive pulmonary disease residing at high altitude but have not evaluated individual-level outcomes such as functional outcomes, symptom burden, exacerbations, imaging characteristics, and progression of disease.

What This Study Adds to the Field

Long-term high-altitude exposure was associated with worse functional outcome, but this did not translate to differences in self-reported symptom burden. Higher mortality was again demonstrated, but the association was attenuated after controlling for pollution.

Exposure to high altitude induces physiologic adaptations to lower partial pressure of inspired oxygen (Pi_O2) because of decreasing barometric pressure as one ascends (1). The lower Pi_O2 results in a cascade of reduced partial pressure of alveolar oxygen and ultimately reduced Pa_O2, which is known as hypobaric hypoxemia (2, 3). In the acute setting, these changes induce the hypoxic ventilatory response, resulting in hyperventilation (4). Hyperventilation reduces the partial pressure of alveolar CO₂, thus offsetting the reduction in partial pressure of alveolar oxygen because of their inverse relationship (2). With time, there are chronic physiologic adaptations in other organ systems. These include renal compensatory mechanisms to reduce HCO₃ reabsorption to compensate for the respiratory alkalosis, increased Hb through hemoconcentration and rise in erythropoietin, leftward shift of the oxyhemoglobin disassociation curve, and hypoxia-induced adjustments in stroke volume and cardiac output. The goal of these changes is to increase oxygen delivery to tissues in the presence of compromised arterial oxygen content (2). During prolonged high-altitude exposure, the acute physiologic response is attenuated as the system deploys chronic compensatory mechanisms. For example, the extent of hyperventilation with long-term altitude exposure is significantly less compared with acute altitude exposure. However, a degree of hyperventilation remains with long-term exposure to high altitude (5, 6).

Although a healthy person can adapt to high altitude via multisystem compensatory mechanisms, for individuals with chronic obstructive pulmonary disease (COPD) with airway obstruction and impaired gas exchange, the question arises as to whether the respiratory system can adequately compensate and whether prolonged altitude exposure affects COPD-related outcomes. The airway limitations and parenchymal changes in COPD may affect one’s ability to appropriately meet the increased ventilatory needs. Although the effects of acute high-altitude exposure in COPD have been studied, there is a paucity of data investigating the effects of long-term high-altitude exposure on clinical outcomes in people with COPD (3, 4). Prior population-based studies have demonstrated an association of increased mortality for individuals with COPD residing at higher altitude (7, 8). These studies have hypothesized that hypoxia-induced cor pulmonale is an underlying mechanism given the increased prevalence of this condition in patients with COPD at higher altitude (7, 9). Nevertheless, the latter studies date back to the mid-1960s, before supplemental home oxygen therapy, long-acting bronchodilators, and smoking cessation were included in treatment algorithms.

To our knowledge, no prior studies have evaluated the long-term effect of high-altitude exposure on COPD functional outcomes, symptom burden, exacerbations, thoracic imaging characteristics, and progression of disease. We hypothesized that residing at high altitude would be associated with reduced functional status and increased symptom burden, exacerbations, and progression of disease for individuals with COPD.

Methods

Study Design and Variable Specification

SPIROMICS (Subpopulation and Intermediate Outcome Measures in COPD Study) is an ongoing prospective cohort study across 12 sites in the United States that enrolled individuals 40–80 years of age who had never smoked, or who had smoked at least 20 pack-years over their lives, with and without COPD at the time of enrollment. Participants were followed longitudinally via in-person and telephone visits to complete physiologic testing, questionnaires, and exacerbation reporting (10).

In the present analysis we evaluated a subset of these participants, defined as individuals with postbronchodilator FEV₁:FVC ratios <0.70 during the baseline examination and whose primary residence at the time of enrollment was situated at an altitude of either <1,000 ft (305 m) or >4,000 ft (1,219 m). Individuals residing between 1,000 and 4,000 ft were excluded from this analysis. Historically, high altitude has been inconsistently defined, although several references use a cutoff of 1,500 m (4,921 ft) to define the altitude grading, ranging from moderate to high and very high altitude. Prior literature has suggested significant differences in cardiac and orthopedic outcomes above 4,000 ft (1,219 m) elevation (11, 12). For the purpose of this analysis, we elected to use a minimum of 4,000 ft to define a high-altitude living condition on the basis of the average altitude of Salt Lake City, Utah, and Denver, Colorado, the two high-altitude SPIROMICS sites.

Individuals enrolled in SPIROMICS I (2011–2016) had a baseline visit followed by up to three annual in-person visits. All these visits included questionnaires, measurements of spirometry, and 6-minute-walk distance (6MWD). We recorded exacerbations and vital status using quarterly phone calls (10). Additional vital status data for the enrolled participants were obtained from National Death Index data, with the last query dated December 31, 2021, which allowed the assessment of vital status even for participants who were unable to complete follow-up. High-resolution computed tomography images were obtained at TLC and residual volume at the baseline visit (13).

Statistical Analysis

Altitude exposure was dichotomized as <1,000 and >4,000 ft. Baseline characteristics, including demographic variables, patient history, functional capacity, and participant-reported symptom burden, were compared between those residing at high versus low altitude using chi-square or rank sum tests as appropriate. Regression modeling was used to evaluate the association of high-altitude exposure to the outcome. We used a linear regression model for continuous variables, including 6MWD, handgrip strength, resting oxygen saturation, St. George’s Respiratory Questionnaire (SGRQ) score (14), and COPD Assessment Test score (15). We measured total lung volume, airway wall thickening, air trapping (Hounsfield units < −856), and percentage emphysema by lung density (Hounsfield units ⩽ −950) on high-resolution computed tomography performed at TLC. We also measured parametric response mapping (PRM) measures of emphysema (PRM^emphysema) and functional small airway disease (PRM^fSAD) (16, 17). The resting oxygen saturation test was measured while breathing room air, and the 6-minute-walk test (6MWT) was performed with supplemental oxygen if previously prescribed. A logistic regression model was used for binary outcomes, including measurements of supplemental oxygen use, oxygen desaturation ⩾ 4% (oxygen saturation as measured by pulse oximetry [Sp_O2]) during the 6MWT, and modified Medical Research Council dyspnea scale score dichotomized as <2 and ⩾2 (18).

Exacerbations were defined as any participant-reported acute worsening of symptoms requiring corticosteroids and/or antibiotics, and severe exacerbations were defined as needing an emergency department visit or hospitalization during the study period (19). A zero-inflated negative binomial model was used to evaluate the association of altitude with exacerbations, with the total follow-up days as an offset. Complete case analysis was used for all models.

Additional longitudinal analyses were conducted using a series of mixed-effects linear regression models allowing repeated measurements at unequal intervals with subject-specific random intercepts to evaluate the association between the baseline altitude and the rate of progression of postbronchodilator FEV₁, 6MWD, and SGRQ score across the duration of the study; the two-way interaction between baseline altitude and time was the parameter of interest. For analysis of the association of long-term altitude dwelling with mortality, we used a Cox proportional-hazards model.

Covariate Selection

Covariates in these models included age, sex, neighborhood area deprivation index (ADI) national rank as determined by geocoded addresses, postbronchodilator percentage predicted FEV₁, smoking status, cumulative tobacco exposure (pack-years), body mass index (BMI), and history of asthma and cardiovascular disease. The ADI is a national ranking of neighborhoods by socioeconomic factors, including income, education, employment, and housing quality (20, 21). Of the total 1,655 participants, 1% had missing ADI data, which we imputed using the site-specific median. Regression models for exacerbations also controlled for history of exacerbations 12 months before enrollment and use of inhaled corticosteroids at baseline. In addition, we separately adjusted for ambient air pollution concentrations obtained from the SPIROMICS AIR (SPIROMICS Air Pollution Study) ancillary study (22). This included average concentrations of both particulate matter with an aerodynamic diameter ⩽2.5 μm (PM_2.5) and NO₂ during the year before enrollment as potential confounders, on the basis of known independent associations of each of these pollutants with worse COPD-related outcomes (23–25). We assessed multicollinearity using variance inflation factors to determine the appropriateness of including covariates.

Stata version 15 (StataCorp) was used for all analyses. SPIROMICS was approved by the institutional review board at each site, and all participants provided written informed consent (see Table E1 in the online supplement).

Results

Baseline Characteristics

Among 1,687 participants with spirometry-confirmed COPD and eligibility on the basis of altitude of residence, 1,655 participants (high altitude, n = 288; low altitude, n = 1,367) had complete covariate data for our analyses (Table 1). Almost all had lived at their respective altitudes for at least 10 years (91.3% at high altitude vs. 98.3% at low altitude). The high-altitude participants resided at a median elevation of 4,524 ft (1,379 m) (interquartile range [IQR], 4,314–5,325 ft [1,314–1,623 m]), compared with the low-altitude group at 381 ft (116 m) (IQR, 98–761 ft [30–241 m]). At 4,524 ft, the barometric pressure is 650 mm Hg (86.7 kPa), resulting in a Pi_O2 of 127 mm Hg (16.9 kPa), compared with a Pi_O2 of 143 mm Hg (19.1 kPa) at 381 ft (26).

Table 1.

Baseline Characteristics of Participants Residing at Higher versus Lower Altitude

	Total n	Low Altitude (<1,000 ft [305 m]) (n = 1,367)	High Altitude (>4,000 ft [1,219 m]) (n = 288)	P Value
Elevation, ft	1,655	381 (98–761)	4,524 (4,314–5,325)	—
Age, yr	1,655	66 (59–71)	65 (59–70)	0.28
Female sex, n (%)	1,655	597 (44)	107 (37)	0.04
Race, n (%)	1,655
White	—	1,062 (78)	267 (93)	<0.01
Black or African American	—	254 (19)	7 (2)	—
Other*	—	51 (4)	14 (5)	—
Postbronchodilator FEV1 (percentage predicted)	1,655	61 (42–77)	65 (46–80)	0.20
Smoking	1,655
Active smoking status, n (%)	—	473 (35)	105 (36)	0.55
Cumulative exposure, pack-years	—	46 (35–62)	45 (35–60)	0.14
NO2, ppb	1,655	9.5 (6.1–14.7)	14.6 (11.5–17.0)	<0.01
PM2.5, μg/m3	1,655	9.2 (8.2–10.3)	7.8 (6.7–8.7)	<0.01
ADI, national percentile rank	1,655	33 (11–61)	39 (24–54)	<0.01
BMI, kg/m2	1,655	27 (23–31)	26 (23–30)	0.12
Asthma history, n (%)	1,655	352 (26)	50 (17)	<0.01
Any cardiovascular history, n (%)	1,637	901 (67)	191 (67)	0.95
Exacerbation in prior 12 mo, n (%)	1,640	438 (32)	81 (28)	0.16
Baseline ICS use, n (%)	1,640	631 (47)	137 (48)	0.64
6-min-walk distance, m	1,568	393 (314–457)	360 (280–419)	<0.01
Handgrip strength, kg	697	28 (22–36)	23 (16–34)	<0.01
Resting SpO2, %	1,578	95 (93–97)	94 (92–95)	<0.01
Home oxygen, n (%)	1,647	269 (20)	125 (44)	<0.01
Hb, mg/dl	1,640	14.2 (13.2–15.1)	15.2 (14.0–16.3)	<0.01
CAT score	1,583	15 (9–21)	16 (9–22)	0.65
mMRC score ⩾ 2, n (%)	1,645	437 (32)	92 (32)	1.00
SGRQ score (total)	1,495	37 (23–53)	38 (23–51)	0.61
Emphysema (<−950 HU at TLC), %	1,645	7.2 (2.3–16.3)	7.3 (2.6–17.3)	0.55
Air trapping (<−856 HU at RV), %	1,638	30.0 (15.5–50.8)	31.8 (15.8–52.5)	0.44
PRMemphysema, %	1,467	4.4 (0.8–14.2)	4.9 (1.0–16.0)	0.27
PRMfSAD, %	1,467	26.4 (15.5–37.1)	28.1 (15.5–38.8)	0.38
Total lung volume at TLC, L	1,645	6.0 (5.0–7.1)	6.6 (5.6–7.5)	<0.01
Pi10, mm	1,645	3.7 (3.7-3.8)	3.8 (3.7-3.8)	<0.01

Definition of abbreviations: ADI = area deprivation index; BMI = body mass index; CAT = COPD Assessment Test; fSAD = functional small airway disease; HU = Hounsfield units; ICS = inhaled corticosteroid; mMRC = modified Medical Research Council dyspnea scale; Pi10 = average airway wall thickness normalized to 10-mm inner perimeter; PM_2.5 = particulate matter with an aerodynamic diameter ⩽2.5 μm; PRM = parametric response mapping; RV = residual volume; SGRQ = St. George’s Respiratory Questionnaire; Sp_O2 = oxygen saturation as measured by pulse oximetry.

Data are descriptive statistics with median (interquartile range) (rank sum test) for continuous variables and total number (percentage) (chi-square test) for categorical variables.

^*Asian, American Indian, Alaska Native, Native Hawaiian, other Pacific Islander, mixed, and missing.

Although some participants classified as high-altitude residents come from various SPIROMICS sites, the two major cities recruiting most of these individuals were Denver and Salt Lake City. Compared with participants residing at low altitude, those living at high altitude were more commonly male and White and less frequently had histories of asthma (Table 1). Air quality and pollution weight differed significantly between cities (see Table E2), but overall, individuals at high altitude were exposed to lower average annual fine particulate matter and more NO₂ (Table 1). As anticipated, high-altitude participants had lower average resting Sp_O2, more frequently used supplemental oxygen, and had higher Hb values. No differences between groups were noted regarding age, FEV₁ percentage predicted, current smoking status, exacerbations in the year before enrollment, BMI, or history of cardiovascular disease.

Functional Capacity and Physical Performance at High Altitude

In a multivariable analysis (Table 2), residence at high altitude was associated with reduced exercise capacity. High-altitude individuals demonstrated shorter 6MWD by a mean of 32.3 m (P < 0.01) and increased odds of desaturation (defined as a ⩾4% decrease in Sp_O2) during the 6MWT. They also had weaker mean handgrip strength by 4.1 kg (P < 0.01). In these adjusted models, living at high altitude was still associated with poorer outcomes related to oxygenation; individuals experienced lower resting Sp_O2 and had higher odds of needing supplemental oxygen (Table 2).

Table 2.

Adjusted Differences in Clinical Outcomes Comparing Participants Residing at Higher versus Lower Altitude

Outcome	Unadjusted for Air Pollution		Adjusted for Air Pollution
	Effect Size (95% CI)	P Value	Effect Size (95% CI)	P Value
6MWD, m (MD)*	−42.2 (−55.7 to −28.6)	<0.01	−32.3 (−49.8 to −14.8)	<0.01
Handgrip strength, kg (MD)*	−4.5 (−6.4 to −2.7)	<0.01	−4.3 (−6.6 to −2.0)	<0.01
Use of oxygen therapy (OR)†	7.0 (4.9 to 10.1)	<0.01	7.5 (4.6 to 12.2)	<0.01
Resting oxygen saturation, % (MD)*	−1.9 (−2.3 to −1.4)	<0.01	−2.1 (−2.6 to −1.5)	<0.01
Desaturation with 6MWD (OR)†	1.8 (1.3 to 2.4)	<0.01	1.6 (1.1 to 2.4)	0.01
CAT score (MD)*	0.5 (−0.4 to 1.4)	0.25	0.0 (−1.1 to 1.2)	0.94
mMRC (OR)†	1.2 (0.9 to 1.7)	0.2	1.0 (0.7 to 1.5)	0.98
SGRQ (total score) (MD)*	0.3 (−1.8 to 2.5)	0.77	−0.8 (−3.6 to 2.0)	0.58
SGRQ (activity score) (MD)*	−0.2 (−2.7 to 2.4)	0.90	−2.0 (−5.4 to 1.4)	0.24
SGRQ (impact score) (MD)*	0.6 (−1.5 to 2.7)	0.56	0.4 (−3.0 to 2.3)	0.80
SGRQ (symptom score) (MD)*	0.8 (−2.0 to 3.6)	0.58	0.1 (−3.5 to 3.6)	0.98
Total exacerbations (IRR)‡	0.88 (0.74 to 1.05)	0.14	0.85 (0.67 to 1.07)	0.17
Severe exacerbations (IRR)‡	0.74 (0.58 to 0.95)	0.02	0.65 (0.46 to 0.90)	<0.01

Definition of abbreviations: 6MWD = 6-min-walk distance; ADI = area deprivation index; BMI = body mass index; CAT = COPD Assessment Test; CI = confidence interval; IRR = incidence rate ratio; MD = mean difference; mMRC = modified Medical Research Council dyspnea scale; OR = odds ratio; PM_2.5 = particulate matter with an aerodynamic diameter ⩽2.5 μm; SGRQ = St. George’s Respiratory Questionnaire.

^*Linear regression model adjusted for FEV₁ (percentage predicted), smoking pack-years, smoking status, age, sex, ADI, asthma, BMI, cardiovascular disease, and presence or absence of pollution (PM_2.5 and NO₂).

^† Logit model adjusted for FEV₁ (percentage predicted), smoking pack-years, smoking status, age, sex, ADI, asthma, BMI, cardiovascular disease, and presence or absence of pollution (PM_2.5 and NO₂).

^‡ Zero-inflation negative binomial model adjusted for FEV₁ (percentage predicted), smoking pack-years, age, sex, ADI, asthma history, BMI, presence or absence of pollution (PM_2.5 and NO₂) and inflation adjusted for FEV₁ (percentage predicted), current smoking status, exacerbation in the 12 months before enrollment, and baseline inhaled corticosteroid use.

Despite the differences in the functional exercise capacity between the residents at high versus low altitude, there were no altitude-related differences in self-reported symptoms or health status, as defined by modified Medical Research Council dyspnea scale score, COPD Assessment Test score, and SGRQ score assessed at baseline (Table 2).

Prospective Exacerbations

When controlling for relevant confounders, residing at high altitude versus low altitude was associated with an annual incidence rate ratio of 0.65 (P < 0.01) for severe exacerbations after enrollment but not for total exacerbations (incidence rate ratio, 0.85; P = 0.17) (Table 2).

Computed Tomography Imaging Characteristics Associated with Living at High Altitude

Individuals at high altitude showed significant increased air trapping (+2.6%; P < 0.01), greater PRM^fSAD (+1.6%; P = 0.04), and a trend toward increased PRM^emphysema (+1.1%; P = 0.08) when controlling for FEV₁ percentage predicted, smoking status, cumulative smoking history, age, sex, ADI, and asthma history. These differences were no longer present when also controlling for average annual pollution exposure (Tables 3 and E3), but associations of high altitude with increased total lung volume and airway wall thickness persisted.

Table 3.

Adjusted Differences in Imaging Characteristics Comparing Participants Residing at Higher versus Lower Altitude

	Unadjusted for Air Pollution		Adjusted for Air Pollution
	Mean Difference (95% CI)	P Value	Mean Difference (95% CI)	P Value
Emphysema (<−950 HU at TLC), %*	0.8 (−0.3 to 2.0)	0.16	0.4 (−1.1 to 1.9)	0.57
PRMemphysema, %*	1.1 (−0.1 to 2.4)	0.08	0.6 (−1.0 to 2.3)	0.45
PRMfSAD, %*	1.6 (1.0 to 3.1)	0.04	0.4 (−1.6 to 2.4)	0.71
Air trapping (<−856 HU at RV), %*	2.6 (0.7 to 4.4)	<0.01	1.3 (−1.2 to 3.7)	0.32
Total lung volume at TLC, ml*	411 (266 to 556)	<0.01	384 (197 to 570)	<0.01
Pi10 (whole lung), mm*	0.07 (0.06 to 0.08)	<0.01	0.08 (0.07 to 0.09)	<0.01

Definition of abbreviations: CI = confidence interval; fSAD = functional small airway disease; HU = Hounsfield units, Pi10 = average airway wall thickness normalized to 10-mm inner perimeter; PRM = parametric response mapping, RV = residual volume.

^*Linear regression model controlling for FEV₁ (percentage predicted), smoking status, smoking pack-years, age, sex, area deprivation index, asthma, and presence or absence of pollution (particulate matter with an aerodynamic diameter ⩽2.5 μm and NO₂).

Longitudinal Outcomes

We evaluated longitudinal differences between the high-altitude and low-altitude groups in the annualized rate of change of postbronchodilator FEV₁, 6MWD, and SGRQ score from the initial baseline visit to the final visit. The median follow-up time was 37 months (IQR, 13–70 mo). There was no observed altitude-related difference in the annualized mean rate of change for FEV₁ (P = 0.55) or 6MWD (P = 0.37) (Figures 1A and 1B). In contrast, individuals at high altitude demonstrated numerically greater worsening of health status, reflected in the increase of SGRQ total score with time (P < 0.01) (Figure 1C). Although differences in the rate of progression were not statistically significant, this appears to have been driven by a divergence in the impact score and symptom score (see Figure E1).

Figure 1.

Mixed models–based means over time using regression-based marginal predicted values, adjusted for FEV₁ (percentage predicted), smoking status, smoking pack-years, age, sex, area deprivation index, asthma, and pollution (particulate matter with an aerodynamic diameter ⩽2.5 μm and NO₂). (A) Postbronchodilator FEV₁ (n = 1,655). (B) Six-minute-walk distance (n = 1,605). (C) SGRQ score (n = 1,635). SGRQ =  St. George’s Respiratory Questionnaire.

Survival analysis, controlling for percentage predicted FEV₁, age, sex, ADI, smoking status, cumulative smoking history, BMI, and cardiovascular disease, demonstrated a hazard ratio of 1.25 (95% confidence interval, 1.0–1.55; P = 0.04) for dwelling at higher altitude (Figure 2). However, when also controlling for air pollution, the effect of altitude on mortality was no longer found (hazard ratio, 1.03 [95% confidence interval, 0.77–1.37]; P = 0.85).

Figure 2.

(A) Cox proportional-hazards regression model–based survival estimates adjusted for FEV₁ (percentage predicted), smoking status, smoking pack-years, age, sex, area deprivation index, body mass index, and history of cardiovascular disease and (B) Cox proportional-hazards regression model also adjusted for pollution (particulate matter with an aerodynamic diameter ⩽2.5 μm and NO₂).

Discussion

Based on a well-profiled, longitudinal cohort of participants with COPD who could be distinguished as residing at high versus low altitudes for ⩾10 years before enrollment, our results offer novel insights into the effect of higher altitude and the associated lower partial pressure of oxygen on the natural history of COPD. Our findings differ from those of prior studies by focusing on long-term residence at high altitude.

We report several significant findings. For individuals with spirometry-confirmed COPD, after adjusting for relevant covariates, long-term high-altitude exposure was associated with poorer physical performance during the 6MWT, weaker handgrip strength, development of oxygen desaturation during exercise, and increased supplemental oxygen use. Although some of these findings might be expected, interestingly, there was no difference in patient-reported outcomes, which may suggest adaptation of perceived symptom burden. Alternatively, this may reflect discordance between objective measures and symptoms that are important to individuals with COPD. This disparity between objective and patient-reported measures raises the concern that well-validated tools for the assessment of patient symptoms and health status may fail to reflect differences in physical performance related to residence at different altitudes. Indeed, the validation of these questionnaires identified trends in selected populations toward different sensitivities in ranking the scores (27, 28). For example, in the analysis of the IBERPOC (Epidemiological Study of Chronic Obstructive Pulmonary Disease in Spain) study, Ferrer and colleagues (29) found that symptom scores on the SGRQ were higher in men, but activity scores were higher in women. The latter study also showed that groups with lower education and social class had significantly higher (worse) scores on all scales. Our findings may suggest that more sensitive tools to capture early functional status decline are needed when monitoring individuals with COPD at high altitude, as generally used questionnaires may not fully capture the reduced functional status. This disparity is consistent with a previous systematic review, which demonstrated only a weak correlation between 6MWD and SGRQ score (30). Similarly, Punekar and colleagues (30) argued for the need to incorporate objective measures in addition to patient-reported outcomes.

Living at high altitude, despite worse physical performance and similar symptom burden compared with low-altitude dwellers, was associated with fewer severe exacerbations. We find this observation notable, as it is consistent with prior analyses in individuals with asthma, which showed that living at high altitude was associated with fewer exacerbations, reduced oral steroid use, and improvement in health status (6, 31). In asthma, this has been attributed to decreased allergen exposure at high altitude, but our analysis interestingly demonstrates persistent significant findings after controlling for PM_2.5 and NO₂ (6). Prior literature has suggested improved airflow dynamics at low air density, which may contribute to reduced exacerbations (32). Additional unmeasured confounders such as access to health care may contribute to this finding.

Consistent with published data (7, 8), we demonstrated that long-term high-altitude residence was associated with an increased risk of death in individuals with COPD. Importantly, we extend those data by showing this effect size was attenuated by controlling for ambient pollution, highlighting air pollution as an additional contributing factor on mortality in COPD. Previously, an increased prevalence of cor pulmonale at high altitude had been suggested as a cause of higher mortality (7, 8). Our study did not have available echocardiographic or heart catheterization data to evaluate for this difference. Unlike prior literature suggesting that long-term exposure to hypoxia may worsen airflow obstruction (6), we found no differences according to altitude in the annualized rate of FEV₁ decline or exacerbation frequency, despite an association with worse functional status.

We identified an important contribution of air pollution to the association of long-term altitude exposure with imaging characteristics. Specifically, adjusting for air pollution (PM_2.5 and NO₂) abolished the association of high-altitude residence with increased small airway disease and greater air trapping but not with higher total lung volume and greater airway wall thickness. On the basis of our analysis, the impact of air pollution at the SPIROMICS sites on outcomes in COPD may be more significant than high altitude itself, and future studies analyzing outcomes in COPD in people living at high altitude need to incorporate the pollution effect when analyzing clinical data. The attenuating effect of pollution within this analysis is consistent with prior research that identified associations of increased air pollution with higher percentage emphysema, COPD exacerbations, morbidity, and mortality (23–25, 33).

Our study has several limitations. The cohort study design limits the ability for causal inferences. Although our multivariable regression modeling controlled for baseline differences, the study design is subject to influences from unmeasured confounders. A majority of the high-altitude participants were from two centers, which could have introduced unmeasured site-specific confounders. Longitudinal analysis was unable to control for time varying covariates such as changes in altitude exposure after enrollment, smoking habit, and air pollution. In addition, certain outcome measures, such as exacerbation history, that were collected during interval follow-up phone and in-person visits are subject to recall bias. Those residing at high versus low altitude had differing characteristics, and although accounting for select confounders allowed better conclusions to be drawn, additional possible confounders may be missed. Finally, our study included relatively few individuals with COPD residing at high altitude, so that absence of significant findings because of type 2 statistical error is possible.

Despite its limitations, this study has some notable strengths. The findings are based on a large multicenter cohort study that followed participants who reported lifetime histories of ⩾20 pack-years of combusted tobacco exposure with and without COPD over a >5-year postenrollment period. The SPIROMICS cohort is well profiled and is suitable for such relatively comprehensive analyses because of its reliable physiologic and imaging characterization of the study participants and geocoding for residential air pollution. A large majority of participants reported long-term high- or low-altitude exposure, thereby permitting a comparatively robust evaluation of the association of long-term residence at high altitude with COPD outcomes. The collection of baseline information and detailed prospective outcome measurements allowed modeling to evaluate the association of long-term altitude exposure with COPD outcomes.

Conclusions

Long-term high-altitude exposure in individuals with COPD was associated with multiple markers of worse functional performance status relative to residence near sea level, including 6MWD, handgrip strength, oxygen desaturation during exercise, and supplemental oxygen use. Interestingly, these objective impairments did not result in differences in patient-reported symptom burden. Individuals at high altitude experienced higher mortality, with an effect size attenuated after accounting for air pollution. In addition, living at high altitude was not associated with worse progression of COPD, as defined by longitudinal changes in postbronchodilator FEV₁.