Chronic Obstructive Pulmonary Disease Mortality

Chronic Obstructive Pulmonary Disease Mortality

Chronic Obstructive Pulmonary Disease Mortality* A Role for Altitude TImothy R. Cote, M.D.; Donna F. Stroup, Ph.D.; Diane M. Dwyer, M.D.; john M. Hora...

605KB Sizes 1 Downloads 67 Views

Chronic Obstructive Pulmonary Disease Mortality* A Role for Altitude TImothy R. Cote, M.D.; Donna F. Stroup, Ph.D.; Diane M. Dwyer, M.D.; john M. Horan, M.D.; and Dan E. Peterson, M.D.

A map of US COPD mortality rates by state suggested that the relative hypoxia of increased altitude may be independently associated with COPD mortality. This was investigated using linear regression analysis of 1986 state-specific data on COPD mortality rates, history of cigarette consumption, and altitude. County seat altitudes and county populations were used to calculate the median altitude of state residents. We found independent signi6cant associations between COPD and both smoking and altitude. State COPD mortality rose by 1/1()1 for every 5.4 increase in mean

packs consumed per capita per year or for each 95-m increase in resident altitude. There was DO association between altitude and smoking. If increased altitude does contribute to COPD mortality, persons with this disease may bene6t from down-migration. (Chat 1993; 103:1194-97)

the United States in 1986, 71,099 persons died of I nchronic obstructive pulmonary disease (COPO,


ICO-9-CM codes 491 [chronic bronchitis], 492 [emphysema], and 496 [chronic airway obstruction, not otherwise specifiedj).' COPO mortality often occurs by terminal hypoxemia." Although cigarette smoking is recognized as the single most important risk factor for COPO, occupational and air pollution exposures are now receiving more attention, mainly as contributors to the effects of cigarette smoking. 3 After studying a map of state-specific US COPO mortality rates, we noted that rates are generally higher for mountainous states-" (Fig 1). A prospective Veterans Administration study> of men with COPD showed higher COPO mortality at moderate altitudes in Utah and New Mexico compared with those at sea level. Sauer' found a positive correlation with mortality from all chronic respiratory diseases with altitude. This correlation was also found among Colorado counties" but was rejected in analysis of COPO mortality in New Mexico." The New Mexico study's statistical power to find a meaningful association is neither directly stated nor computable from data in the report. None of these studies controlled for smoking or other risk factors. We undertook this investigation to control for the effect of smoking, as measured by state-specific cigarette consumption data, in assessing the effect of altitude on COPO mortality rates.

*From the Maryland Department of Health and Mental Hygiene, Baltimore (Drs. Cote and Dwyer); the Centers for Disease Control, Epidemiology Program Office, Atlanta (Drs. Cote, Stroup, Horan, and Peterson); and the Wisconsin Department of Health and Social Services, Madison (Dr. Peterson). Manuscript received May 11; revision accepted August 3. Reprint requests: Dr. Cote, NClIVES, 6130 Executive Blvd, EPN434, Roc1roUle, Maryland 20852


ALT=median altitude of state residents; COPDMR=chronic obstructive pulmonary disease mortality rate; ICD-9-CM "lratemt.dional C~ fJlDiNaa, Nifalla BeciIion, CUnietJl Modification; PACICS = median pacb of cigarettes purchased per state resident per year


We used age-adjusted state COPD mortality rates that were computed from 1986 US death certificate dataS for ICD-9-CM codes 491 (chronic bronchitis), 492 (emphysema), and 496 (chronic airway obstruction, not otherwise specified).' State cigarette consumption was calculated from US Commerce Department data for the mean packs purchased per state resident per year from 1965 to 1984.8 The most readily available measure of state altitude is a topographic average of surface altitudesr'" however, since residence of the population is not uniformly distributed within states, that measure was unsuitable for this study of human disease. Therefore, to estimate the median altitude of residence of state residents, we ranked each slate's counties by the altitude of the county seat" and used the 1980 US census population of each county to determine each state's median altitude of residence. There are no county seats in Alaska. Its median altitude of state residents was thus indeterminable and Alaska was excluded from analysis. The median altitude of state residents for the remaining 49 states and the District of Columbia is shown in Table 1. Commerce Department records" showed that more than 10 percent of cigarettes sold in four other states (New Hampshire, Vermont, North Carolina, and Kentucky) were sold to non-residents; these states were excluded from analysis involvingcigarette consumption. We used simple and multiple linear regression to determine if COPD mortality is associated with cigarette consumption and/or altitude independent of cigarette consumption. RESULTS

In 1986 the age-adjusted US COPO mortality rate (COPOMR) was 29.51100,000 and ranged from 16.61 100,000 in Hawaii to 49.l/100,OOO in Wyoming. Average annual cigarette consumption (PACKS) ranged from 71.5 packs per capita Utah to 185.2 packs per capita in Nevada. Average state residence altitude (ALT) ranged from 3 m in Connecticut and Rhode Island to 1,627 m in Colorado. The simple regression of (PACKS) on COPOMR is shown in Figure 2 (slope=O.151 [PACKS], Y-intercept= 12.2, and r2=O.18). Thus, COPOMR increased COPO Mortality (CoMet 81)




19 .0-25.6


() 30.7-34.9 ~

26.5-29.9 ~ 50


COPO Mortality per 100 ,000

FI GURE 1. Age-adjusted chronic 0bstruetive pulmonary disease mortality rates per 100,000 population, by quartile - United States, 1986.

rl - - - - - - - - - - - - - - - - - - - - - , , - - - - - - - - : - : - : - : - - - - ;


• IN

45 ,~ NM


FIGURE 2. The 1986 state chronic 0bstructive pulmonary disease mortality rates per 100,000 as a function of cigarette consumption per state resident per


COPO Mortality per 100,000 50 r - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

45~ 40

• Wy

• NV

• MT



I ~



35 1 ' I •

• UT

25 20





Altitude in Meters


CHEST I 103 I 4 I APRIL. 1993


by one per 100,000 for every 6.6 average packs smoked per state resident per year and only 18 percent of the variation in COPD mortality was explained by the smoking measure. The simple regression of median altitude of state residents on COPDMR is shown in Figure 3 (slope = 0.0094 ALT, Y-intercept = 28.0, and r=O.33). Thus, COPD mortality increased by one per lOS for every l06-m rise in the median altitude of state residents and 33 percent of the variation in COPDMR was explained by the effect of altitude. The simple regression of smoking and altitude showed no association between these variables (r = 0.01). Multiple linear regression yielded the equation: COPDMR =4.68 + O.OI05(ALT) + O.I86(PACKS) (p
Significance values for each coefficient estimate are given in parentheses. There was no significant interaction between the altitude and smoking variables. This equation predicts that state COPD mortality rises by 1 per 100,000 for every increase in 5.4 average packs consumed per state resident per year (with no change in altitude) or for a 95-m increase in resident altitude (with no change in smoking). Sixty percent of the variability in COPD mortality was explained by combined effects of smoking and altitude. DISCUSSION

This ecologic analysis shows that cigarette consumption and altitude are independently related to a state's COPD mortality rate; and the combination of these two explains over halfthe variation in COPD mortality rates. Interpreting these results cannot be done by simple comparison of correlation coefficients. Although the correlation coefficient is higher for altitude, both altitude and cigarettes are highly significant predictors ofCOPD mortality. Furthermore, the effect of smoking (as estimated by the product of the smoking coefficient and the number of packs) is several times greater than the effect of altitude. In addition to the main study goals, we estimated the median altitude at which state residents live (Table 1). This index may be useful to investigators interested in other possible health effects of altitude such as cor pulmonale or low birth weight. Because in most states the median altitude of residents is relatively low, the results may have been influenced by a few outlying, high-altitude states. We tested the robustness of the results by sequentially removing from the data set the six states with altitudes higher than Nevada. Only after all six states were removed did the relationship become nonsignificant. Even if those points do "drive the regression;' the effect is important because they represent thousands of deaths. This association is biologically plausible. Air inspired in Miami contains approximately 20 percent more oxygen than the same volume of air inspired in 1111

Table I-Median Altitude o[SItIte Baidenta*, Meten Abooe SetJ Ucel, United StGta, 1980 State ~a

Arkansas Arizona California Colorado Connecticut District of Columbia Delaware Florida Georgia

Altitude, m 1~

84 285 35

1627 3 9 20


Hawaii Iowa

231 10 275



Illinois Indiana Kansas Kentucky wub~

Massachusetts Maryland Maine Mi~

Minnesota Mbsouri Mb~p~


North Carolina North Dakota Nebraska New Hampshire New Jersey New Mexico Nevada New York Oklahoma Oregon Pennsylvania Rhode Island ~~Carolina

South Dakota

Tennessee Texas U~

Virginia Vermont Washington Wisconsin \\est Virginia



223 343 198

10 12 30 38


248 236 ~

972 200 469 361 51


1491 619 22 306 48 146 3



192 141



153 8 227 215


*MedUm altitude of state residents was computed by ranking each state's counties by the altitude of the county seat (source: US Geological Survey, National Mapping Division, OfBce of Ge0graphic and Cartographic Research, Geographic Names Information System, 1990), listing the population of each county (source: 1980 US Census), and interpolating the altitude below which 50 percent of the state's population resided.

Denver (159 mm Hg vs 134 mm Hg). Although these differences are small, the relatively hypoxic exposure is continuous and the need for oxygen is a biologic imperative. It is reasonable to expect that persons with smoking-induced pulmonary impairment might COPD Mort8Ity (CoM et eI)

experience increased mortality at higher altitudes. While the findings were biologically plausible, this was an ecologic study and consequently subject to important limitations of interpretation. Some third factor such as an occupational exposure (eg, mining) or lack of access to health care may be causal and associated with both altitude and COPD mortality As death certificate coding of occupation continues to improve, future analyses that control for occupational exposures willbe valuable. Our cigarette consumption data were limited to 1965 to 1986; because cigarettes smoked before those years were unmeasured, we probably underestimated the effect of smoking. While attribution of mortality to state of residence at death may have caused misclassification of persons who migrated, people with pulmonary problems have also been noted to bemore likely to migrate to rural areas" and to migrate downward." COPD is a significant cause of mortality; in 1986 in the United States, COPD was the cause of 128,590 years of potential life lost before age 65 years. 1 Because smoking remains the single best documented cause of COPD, efforts to prevent COPD mortality should continue to focus on smoking cessation. This study has provided some evidence for an independent effect of altitude in COPD mortality Should future studies establish that altitude is truly contributory, downmigration may be recommended for persons with this disease. ACKNOWLEDGMENTS:The writers thank Dr. RogerL. Payne of the US Department of the Interior, Geological Survey, _for altitude

data and Drs. Thomas Novotny and Roy Baron for useful comments on the manuscript.


1 Centers for Disease Control. Chronic disease reports: chronic obstructive pulmonary disease mortality- United States, 1986.

MMWR 1989; 38:549

2 Ingram RG. Chronic bronchitis, emphysema, and airways 0bstruction. In: Wilson JD, Braunwald E, Isselbacber KJ, eds. 3




7 8

Harrison's principles of internal medicine. New York: McGraw Hm, 1991; 1074-82 Speizer FE. Environmental lung disease. In: Wilson JD, Braunwald E, Isselbacher KJ, eds. Harrisons principles of internal medicine. New York: McGraw Hill, 1991; 1056-63 Sauer HI. 1980 Geographic patterns in the risk of dying and associated factors, ages 35-74 years. United States, 1968-1972, DHHS publication No. (PHS) 80-1402, Vital and Health Statistics; Series 3, No.18 Renzetti AD, McClement JH, Litt BD. The Veterans Administration cooperative study of pulmonary function: III. Mortality in relation to respiratory function in chronic obstructive pulmonary disease. Am J Med 1966; 41:115-29 Emphysema Moore LG, Bohr AL, Maisenbach JK, Reeves mortality is increased in Colorado residents at high altitude. Am Rev Respir Dis 1982; 126:225-28 Coultas DB, Samet JM, Wiggins CL. Altitude and mortality from chronic obstructive lung disease in New Mexico. Arch Environ Health 1984; 39:355-59 National Centers for Health Statistics. Vital statistics mortality data, multiple cause of death detail, 1986 (machine-readable public-use data tape). Hyattsville, MD: US Department of Health and Human Services, Public Health Service, 1988 (ICD-


9-CM 491-2, 496) 9 Tobacco Institute. The tax burden on tobacco-historical compilation (volume 23, 1988). Washington, DC: Tobacco Institute, 1989 10 US Geological Survey. Elevations and Distances in the United States, 1980 11 US Geological Survey National Mapping Division, Office of Geographic and Cartographic Research, Geographic Names Information System, 1990 12 Advisory Committee on Intergovernmental Relations. Cigarette tax evasion: a second look. Washington,· DC: Advisory Committee on Intergovernmental Relations, 1985 13 Kelsey JL, Mood E~ Acheson RM. Population mobility and epidemiology of chronic bronchitis in Connecticut. Arch Environ Health 1968; 16:853-61 14 Regensteiner JG, Moore LG. Migration of the elderly from high altitudes in Colorado. JAMA 1985; 253:3124-28

CHEST I 103 I 4 I APRIL, 1993