Study population
NHANES III is a cross-sectional survey conducted between 1988 and 1994 by the National Center for Health Statistics (NCHS). NHANES III provides a nationally valid estimate of health and nutrition status among a representative group of civilian, noninstitutionalized US residents. Participants signed a consent form during the survey. NHANES Institutional Review Board (IRB) approval was obtained.
In order to provide accurate estimates, young children, the elderly, non-Hispanic blacks and Mexican Americans were oversampled. Sampling methods and design have been previously described [
16,
17]. Each participant was assigned a weight to allow for calculation of population-level estimates. NHANES III includes household interviews, physical exams in the Mobile Examination Center (MEC) and home examinations when participants could not reach the MEC.
Since COPD is uncommon in younger age groups, we limited our study population to participants ≥40 years old with good quality, reliable spirometry data based on the criteria of the American Thoracic Society [
18]. Exclusion criteria included individuals <40 years old, pregnant women, Asians (due to a limited number), home-assessed or unreliable spirometry, and a positive response to either of the following questions (“Have you had a cough, cold, or other acute illness in the past few days?” or “In the past three weeks have you had any respiratory infections, such as flu, pneumonia, bronchitis, or a severe cold?”). Our study population included 3,681 individuals who met these eligibility criteria.
Measurements and variables
In depth documentation of NHANES III laboratory and sample collection methods have been previously described and are briefly outlined below [
21].
Age, gender and race information were collected during the home interview. Race was self-classified as white, black, non-black Hispanic or other. The self-report of physical activity included a questionnaire that evaluated the frequency and intensity of leisure time physical activities including dancing, jogging, running, bicycling, swimming, aerobics, and other activities. Intensity of the physical activity was expressed as a weekly metabolic equivalent (MET). METs are defined by increasing energy expenditure with MET = 1 corresponding to the resting metabolic rate and MET = 10 representing running at 6 mph [
22].
Insurance status was determined during the home interview and includes coverage under Medicare, Medicaid, CHAMPUS, CHAMPVA, VA and other private or employer/union health plans. Marital status was measured as the percent of participants who were partnered, including those who were married or living as married. Education was measured as the percent of participants who did not complete high school (<12 years of schooling). Income was determined as the percent of participants with a total combined family income over the last year amounting to less than $20,000. Employment status was based on responses to questions asking if the participant had a job or business or if they worked within the past 2 weeks.
Blood pressure was reported as the average of five different measurements, taken by mercury sphygmomanometers (W. A. Baum Co., Copiague, NY) using standard technique [
23]. The designation of hypertension (HTN) was given to those who were taking anti-hypertensive medications or those with a positive response to the question “Were you told on 2 or more different visits that you had hypertension, also called high blood pressure?”
Diabetes status was defined by self-report or prior diabetic medication use. BMI was calculated as weight in kilograms divided by the square of height in meters with height rounded to the nearest 0.1 cm and weight to 0.01 kg.
Smoking status was divided into several categories including: current smokers, past smokers, never smokers and ever smokers. Current smokers were defined by answering “yes” to questions: “Do you smoke cigarettes now?” and “Have you smoked at least 100 cigarettes in your lifetime?” Past smokers included those who do not smoke cigarettes now but smoked over 100 cigarettes in their lifetime. Never smokers were those who smoked less than 100 cigarettes in their lifetime. Ever smokers were determined by combining past and current smokers. Number of cigarettes per day was also measured. A participant was considered exposed to environmental tobacco smoke (second-hand smoke) if he/she currently smoked or lived with someone who smoked in the home.
Alcohol consumption was assessed through questions from the household interview asking how many times over the past month participants drank beer, wine and hard liquor. Frequency across all three alcohol types was summated and expressed as number of drinks per week and dichotomized to “any alcohol” or none.
Clinical symptoms of cough and phlegm were determined by a positive response to questions which asked participants if they coughed or produced phlegm on most days for 3 consecutive months or more. Shortness of breath was assessed through interview as being “troubled by shortness of breath when hurrying on level ground or walking up a slight hill”.
Diagnosis of COPD was based on spirometry. Spirometry was performed in a designated room in the MEC devoted to spirometry using a customized Ohio Sensormed 827 dry rolling seal spirometer [
24]. Calibration was performed at the start of each session and participants were coached during the procedure so as to produce the best results. Participants were encouraged to blow for at least six seconds as required by ATS guidelines, thus measuring forced vital capacity (FVC). The session was complete when five acceptable trials were obtained, with a maximum of eight [
24]. The tests were reviewed by two senior technicians and classified for reliability. All unacceptable maneuvers were excluded before reproducibility calculations were performed [
24].
We used an FEV1/FVC ratio (forced expiratory volume in one second/forced vital capacity) of <0.70 to define airflow obstruction. Mild, moderate, severe, and very severe airflow obstruction were defined as respectively: stage 1 -FEV1 ≥ 80% of predicted; Stage 2 -50% ≤ FEV1 < 80% predicted; Stage 3 - 30% ≤ FEV1 < 50% predicted; and Stage 4 -FEV1 < 30% predicted based on the GOLD COPD Guidelines [
25] and the predicted FEV1 based on the equation developed by Hankinson et al. [
26].
CHD diagnosis was based on an affirmative response to the interview question: “Has your doctor ever told you that you had a heart attack?” Participants were classified into CHD risk groups based on the number of CHD risk factors or risk equivalents they had [
27].
Laboratory analyses: fasting blood samples were obtained in the MEC or during the home interview. All blood samples were frozen and shipped on dry ice to laboratories for analysis. Upon arrival at the laboratories, frozen specimens were initially stored at -20°C, refrigerated samples were stored at 4-8°C, and frozen specimens with delayed analysis were stored at -70°C or lower [
21].
Triglycerides (TG), HDL cholesterol (HDL-c), and total cholesterol (TC) analyses were performed by the Hitachi 704 Analyzer (Boehringer Mannheim Diagnostics, Indianapolis, IN). LDL cholesterol (LDL-c) was not calculated from these measurements because a certain percentage of the population did not fast prior to blood draw. Non-HDL was calculated as the difference between total cholesterol and HDL, which may be a superior predictor of cardiovascular events compared to LDL-c [
28].
White blood cell (WBC) count was determined by the Coulter Counter Model S-PLUS JR with Coulter histogram differential, an automated hematology analyzer. Homocysteine (Hcy) level was determined by reverse-phase high-performance liquid chromatography (HPLC) and fluorescence detection [
29]. Due to NHANES limitations, Hcy testing was performed only on participants ≥12 years of age in phase II (1991–1994).
Fibrinogen was measured by the Coagamate XC plus automated coagulation analyzer (Organon Teknika, Durham, NC). Ferritin was measured using the Bio-Rad Laboratories “Quantimune Ferritin IRMA” 1000 kit (Hercules, CA). Hemoglobin A1c was measured by the Bio-Rad DIAMAT glycosylated hemoglobin analyzer system, using HPLC principles (Bio-Rad Laboratories, Hercules, CA).
Serum C-Reactive Protein (CRP) is a marker of systemic inflammation and a validated independent risk factor for CHD [
30]. It was quantified via latex-enhanced nephelometry using the Behring Nephelometer Analyzer System (Behring Diagnostics Inc., Somerville, NJ). Since most participants had CRP values below the lowest detectable level (20.95 nmol/L), CRP was dichotomized into low (<95.24 nmol/L) and high (≥95.24 nmol/L) levels which are consistent with previous NHANES III categorization [
31,
32]. Low levels were used to define the non-inflammatory COPD group and high levels were used to define the inflammatory COPD group.
Diagnostic groups used for analyses. Participants were divided into five exposure groups based upon their COPD status from spirometry testing, their CRP level, and their self-reported CHD status. Participants whose spirometry results indicated no COPD were placed into Group 1- no COPD and no CHD
(“Healthy”individuals free of hypertension, diabetes, and arthritis); participants who had COPD and had no inflammation (CRP <95.24 nmol/L) [
31,
32] were included in Group 2
(“non-Inflammatory COPD”); participants who had COPD and had inflammation (CRP ≥95.24 nmol/L) [
31,
32] were included in Group 3
(“inflammatory COPD”); participants who reported “yes” to the question “Has a doctor ever told you that you had a heart attack?” and did not have COPD based on spirometry results were placed in Group 4 (the
“CHD” Group); and participants who reported “yes” to the question “Has a doctor ever told you that you had a heart attack?” and had COPD based on spirometry results were placed in Group 5 (the
“CHD + COPD” Group).
Statistical analysis
Descriptive statistics were used to examine demographics, CHD risk factors and inflammatory biomarkers of the five study groups. Analysis of variance and chi-square tests were used to evaluate differences in prevalence of continuous and categorical risk factors, respectively, among the five study groups while adjusting for the complex sampling design in NHANES III.
Cox proportional hazards models were used to examine the time to CHD mortality. In addition, we explored the relation between COPD subgroups and CVD as well as COPD subgroups and all-cause mortality after adjusting for the confounding effect of CHD risk factors and inflammatory biomarkers including age, gender, race/ethnicity, family history of CHD, diabetes, BMI, physical activity level, cholesterol ratio, current smoking, household income, education, and blood pressure medications. Tests for violations in the proportional hazards assumptions were also conducted and found to be insignificant.
The adjusted hazards ratios and their 95% confidence intervals were used to measure the association between COPD subgroups and (CHD, CVD and all-cause) mortality. Statistical analyses were performed using SAS-callable (version 9.2, SAS Institute Inc., Cary, NC) SUDAAN version 11.0.1 (Research Triangle Institute, Research Triangle Park, NC).