The Relation of Prothrombin Times to Coronary Heart Disease Risk Factors among Men Aged Years

Transcription

1 American Journal of Epidemiology Vol 136,. 5 Copyright 1992 by The Johns Hopkins University School of Hygiene and Pubic Hearth Printed in U S.A. All rights reserved The Relation of Prothrombin Times to Coronary Heart Disease Risk Factors among Men Aged Years David S. Freedman, 1 Tim Byers, 1 Joseph J. Barboriak, 2 W. Dana Flanders, 3 Alexander Duncan, 4 Ray Yip, 1 and Elaine N. Meilahn 5 Although levels of coagulation factor VII and fibrinogen are predictive of cardiovascular disease, relatively little data describe hemostatic characteristics in healthy populations. The cross-sectional associations between the prothrombin time, a measure of the activity of the extrinsic and common pathways of coagulation, and traits associated with the risk of cardiovascular disease were therefore examined among 3,604 white and 514 black, male, US Army veterans aged years. The prothrombin time measurements, performed in 1985 and 1986, were precise, with an intraclass correlation of 0.98 (202 pairs). Overall, the mean prothrombin time was seconds (standard deviation, 0.4 seconds), and percent of the men had a value of less than 12 seconds. Many of the observed associations with the prothrombin time paralleled those that have been reported with clotting factor VII and fibrinogen. The mean prothrombin time was 0.15 seconds shorter among whites than among blacks and was 0.2 seconds shorter among current cigarette smokers than among men who had never smoked. Inverse associations were also seen with relative weight and with levels of total cholesterol and triglycerides (r = to -0.16). All associations were statistically significant at the 0.01 level, and the examined characteristics could jointly account for about 12 percent of the variability in prothrombin times. Additional data on characteristics related to coagulation may help elucidate the natural history of cardiovascular disease and aid in the design of dinical trials. Am J Epidemiol 1992;136: blacks; blood coagulation; cardiovascular diseases; lipids; prothrombin time; smoking Several characteristics are predictive of coronary heart disease, but about one-half of middle-aged men who develop clinical disease are not obese, do not smoke cigarettes, and do not have elevated levels of blood pressure or cholesterol (1). Although information on lipoproteins, such as levels of high density lipoprotein cholesterol (HDL Received for publication vember 29, 1991, and in final form April 15, 1992 Abbreviation: HDL cholesterol, high density lipoprotein cholesterol. ' Division of Nutrition, Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control, Atlanta, GA. 2 Department of Pharmacology and Toxicology, Medical College of Wisconsin, Milwaukee Veterans Affairs Medical Center, Milwaukee, Wl. 3 Department of Epidemiology, Emory University School of Public Health, Atlanta, GA 513 cholesterol) and various apolipoproteins, would improve disease prediction (2, 3), many persons will remain misclassified. With the recognition of the etiologic importance of thrombus formation in myocardial infarctions (4), increased attention has been given to hemostatic factors. Several cohort studies have shown that * Department of Pathology, Emory University Hospital, Atlanta, GA. 'Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA. Reprint requests to Dr David S. Freedman, Centers for Disease Control, K-26, 1600 Clifton Road, Atlanta GA The authors thank the leaders of veterans service organizations who provided important input and support for the study and Drew Baughman for his help with data management Data for this study were collected under an tnteragency agreement with the Veterans Administration.

2 514 Freedman et al. the importance of levels of fibrinogen in predicting disease among middle-aged men rivals that of total cholesterol (5-10). Although not as extensively studied, an elevated level of coagulation factor VII is also related to subsequent coronary events (6, 10). Furthermore, it is possible that some of the traditional risk factors may act by precipitating a thrombotic event (): Age (12, 13), cigarette smoking (12-14), and relative weight (13) are associated with levels of fibrinogen and factor VII. High levels of factor VII have also been found in persons with high intakes of dietary fat (15), among hyperlipidemics (16), and among white men as compared with blacks (12, 17) and Japanese (18). In addition to these associations, the administration of oral anticoagulants, which suppress the activity of factor VII and other vitamin K-dependent enzymes, substantially reduces the rate of recurrent infarctions and cerebrovascular accidents (19). The prothrombin time, a widely used clinical test to monitor oral anticoagulant therapy, measures the overall activity of clotting factors V, VII, X, and II (prothrombin) and of fibrinogen. Our study describes the relation of prothrombin times to several characteristics associated with the risk of coronary heart disease in a biracial (black and white) sample of men aged years. MATERIALS AND METHODS Population Subjects were selected from men who were included in a historical cohort study comparing the health of veterans who served in Vietnam with that of other veterans. Sample selection and data collection, performed in 1985 and 1986 during telephone interviews and medical examinations, have been described (20, 21). Briefly, a random sample of 48,500 men who entered the US Army between 1965 and 1971 was generated. To increase the baseline comparability of Vietnam and non- Vietnam veterans, we limited participation to men who 1) served only one term, 2) had 16 or more weeks of active service, 3) earned an occupational specialty other than "trainee" or "duty soldier," and 4) had a pay grade at discharge no higher than that of sergeant. A total of 17,867 veterans were considered eligible, with 51 percent having served in Vietnam. Of the 15,288 (86 percent) men who were given a structured interview by telephone, 6,443 were randomly selected for the medical examination. Of those invited, 4,462 (69 percent) men participated in examinations conducted at the Lovelace Medical Foundation in Albuquerque, New Mexico. Possible differences between the 4,462 men who participated in the medical examination and the other 10,826 men were assessed using data collected during the telephone interview (22). Participation was not related to race, marital status, household income, cigarette smoking status, Quetelet index (calculated from self-reported weight and height), or alcohol consumption. However, men who participated in the medical examination tended to be slightly more educated than were men interviewed by telephone only. The current analyses are restricted to 3,604 white men and 514 black men who were fasting at the examination and who were between the ages of 31 and 45 years. Hispanics (n = 200), Asians (n = 34), and American Indians (n = 49) were excluded, as were men older than age 45 years (n = 5) and men who reported that they had not fasted (n = 49). We also excluded men who reported either warfarin use (n = 4) or a blood-clotting abnormality (n = 6). The prothrombin times of men using warfarin were all above the 99th percentile, with values ranging from 14.0 to 21.4 seconds. Of the six men who reported a blood-clotting abnormality, three had a prothrombin time below the 10th percentile (.7-.9 seconds), and one had a prothrombin time above the 90th percentile (13.0 seconds). Telephone interview and medical examination Quetelet index, a measure of relative weight, was calculated as weight (kg)/height (m) 2. Information on race, household in-

3 Correlates of Prothrombin Time 515 come, educational achievement, cigarette smoking, alcohol consumption, and medication use (including names of drugs) was obtained. At the medical examination, participants were also asked if they had a tendency to bleed or bruise easily, if they had had a nose bleed that they could not stop during the preceding year, and if they had been told by a physician that they had had a heart attack or had cirrhosis, liver damage due to alcohol, diabetes, angina, peripheral vascular disease, or a blood-clotting abnormality. Subjects were also asked if they had experienced chest pain when walking quickly or when walking up a hill. Because of the low prevalence of several of these characteristics, positive responses to several questions were combined in the analyses. Men were considered to have coronary heart disease if they reported being told either that they had angina (n = 26) or had had heart attack (n = 19). Liver damage was based on a positive response to questions concerning cirrhosis (n = 8) or liver damage due to alcohol (n = 51). Positive responses to questions concerning bleeding or bruising easily (n = 97) and nose bleed (n = 14) were also combined. Laboratory determinations Although some evidence suggests that exposure to 2,3,7,8-tetrachlorodibenzo-/?- dioxin, a contaminant of Agent Orange, may result in hematologic abnormalities (23), previous analyses found no association between military service in Vietnam and hemoglobin levels, hematocrit, prothrombin time, or counts of red cells, platelets, leukocytes, and lymphocytes (21, 24). Venipuncture was performed between 8 a.m. and 10 a.m. after an overnight fast. Prothrombin times were assessed to the nearest 0.1 second in a one-stage test using an Electra 700 Automatic Coagulation Timer (Medical Laboratory Automation, Pleasantville, New York). Lyophilized rabbit-brain thromboplastin (Dade Division, American Hospital Supply Corporation, Miami, Florida) was added to the specimen, and using the second derivative of the optical density curve, the prothrombin time was automatically calculated as the interval between the addition of thromboplastin and the start of fibrin formation. Partial thromboplastin times and fibrinogen levels were not evaluated. Because various liver diseases can lead to decreased production of coagulation proteins, the relation of prothrombin time to several tests associated with liver function (e.g., alanine aminotransferase, aspartate aminotransferase, and bilirubin) was examined. Exploratory analyses indicated that the strongest associations were with levels of total and unconjugated bilirubin (r = ). Because levels of total bilirubin were not available for about 30 percent of all men, levels of unconjugated bilirubin were included in several analyses. Hepatitis B surface antigen was assayed using a mouse monoclonal antibody, while antibody to the core antigen was assayed using human hepatitis B core antigen (Abbott Laboratories Diagnostic Division, rth Chicago, Illinois) (25). Of the 30 men with the surface antigen, 28 also had antibodies to the core antigen and were considered to have a chronic hepatitis B infection. Levels of total cholesterol were assayed by an enzymatic method, and HDL cholesterol was measured after precipitating other lipoprotein fractions with dextran sulfate. The laboratory used an Ektachem 700 (Eastman Kodak, Rochester, New York), and in addition to having internal controls, the laboratory was standardized and monitored by the Centers for Disease Control (25). Levels of triglycerides were determined by an enzymatic method. All laboratory determinations were monitored by using bench and "blind" repeat quality control procedures consisting of a "low" and a "high" control; the coefficient of variation for prothrombin times ranged from 0.9 to 2.2 percent (25). Based on repeat determinations of prothrombin times from a 5 percent random sample (n = 202) of men included in the current study, the intraclass correlation coefficient was The mean absolute difference between pairs was 0.06 seconds; the two determinations were

4 516 Freedman et al. identical for 100 of the 202 men and differed by O.I seconds for 81 men and by 0.2 seconds for 21 men. Statistical methods The distribution of prothrombin times was positively skewed with a high kurtosis, and the use of logarithmic or square-root transformations did not improve the normality. (In contrast, a log transformation greatly improved the normality of the triglyceride distribution.) Furthermore, the arithmetic mean of prothrombin times across categories of several variables was very similar to both the median and geometric mean. Several analyses, therefore, examined not only the mean prothrombin time, but also 1) the overall distribution of prothrombin times by using histograms and cumulative frequency distributions, and 2) the proportion of men with a prothrombin time of less than 12 seconds (relatively short) or 13 or more seconds (relatively long), approximately the 10th and 90th percentiles, respectively. Variables with the largest number of missing values were the number of drinks per month (16 men) and the number of daily cigarettes (12 men); men with missing data were excluded only from analyses that made use of that information. We used t tests and x 2 tests to assess differences in prothrombin times between blacks and whites and across levels of other characteristics. Because of the low prevalence of certain characteristics (e.g., coronary heart disease), we used Fisher's exact test to assess the independence of race and several variables. Dose-dependent associations between prothrombin time and several continuous variables were assessed using stratification and correlation coefficients. Similar results were obtained using parametric or nonparametric tests. To control for the effects of race, we contrasted mean prothrombin times across categories of several characteristics using analysis of covariance. Possible differences in the proportion of men with relatively short or long prothrombin times were assessed using the Mantel-Haenszel x 2 test (26) or exact tests (27) to summarize data over racespecific 2x2 tables. We used multiple linear regression to assess the independent association of each examined characteristic with prothrombin times. Because of the apparent nonlinearity between prothrombin times and alcohol intake, we categorized current drinkers in some analyses as light (1-14 drinks per month), moderate ( drinks per month), or heavy (150 or more drinks per month) drinkers. RESULTS The mean prothrombin time was 0.15 seconds shorter ( vs seconds) among white men than among black men (table 1). Racial differences were also evident at the extremes of the distribution: A relatively short prothrombin time (less than 12 seconds) was seen among percent of whites versus 6 percent of blacks, whereas a relatively long prothrombin time (13 or more seconds) was seen more frequently among black men. Despite the narrow range of prothrombin times, a larger proportion of whites than blacks was seen in all prothrombin time categories less than 12.6 seconds (figure 1), while there was a proportionately greater number of blacks in all categories of 12.6 or more seconds. The distribution of prothrombin times among Hispanics was virtually identical to that observed among whites (data not shown). Mean levels of several other characteristics also differed between blacks and whites (table 1). Black men had higher levels of HDL cholesterol, were more likely to smoke cigarettes, consumed more alcohol, and had a higher prevalence of chronic hepatitis B infection, bleeding or bruising easily, diabetes, and peripheral vascular disease. In contrast, white men had higher levels of triglycerides, and white smokers reported a larger number of daily cigarettes than did black men who smoked. Prevalent coronary heart disease, which included angina and myocardial infarction, was reported by 33 men, all of whom were white. As shown in figure 2, cigarette smoking was associated with a shorter prothrombin

5 * ; * ; Correlates of Prothrombin Time 517 TABLE 1. Mean levels of selected characteristics, by race, Vietnam Experience Study, Characteristic (n - 3,604) Age (years) Military service in Vietnam (%) Prothrombin time (seconds) <12(%) 2:13 (%) Quetelet index (kg/m*) Total cholesterol (mg/dl) TriglycerkJes (mg/dl) HDL cholesterol (mg/dl) Current smokers (%) Cigarettes per dayi Current drinkers (%) Drinks per montht Bllirubin, unconjugated (mg/dl) Chronic hepatitis B infection (%) Self-reported (%) Bleeding/bruising Liver damage Coronary heart disease Diabetes Peripheral vascular disease Chest pain ±2t 56* ±0.4* * 8* 26.8 ± ±41 7 (95) ±3* 44 ± 12* 44» 27 ± 12* (22) ± 59* 0.6 ± 0.3 1* 2* 1 1 1* 2* 6 ± ±0.5* 6* 15* 26.9 ± ±43 88 (75) ± 68* 51 ± 16* 54* 17 ±10* (31) ± 0* 0.6 ± 0.3 2* p < p values for continuous variables were assessed using a t test, p values for dtehotomous variables were assessed using a x* test or Fisher's exact test. t Values are mean ± standard deviation for continuous variables. t Percentages are shown for dichotomous variables. Numbers in parentheses, geometric means for trigiycerides and drinks per month. H HDL cholesterol, high-density lipoproteln cholesterol. 1 Restricted to current smokers or current drinkers > U c 0) 3 U" Q) 2 : 15" 10" 5- CO m at [-1 PI u> 1 hi OS n D m 4* 2 0 3* 4* 6 VI CO *- o «D CO o en to ^r o < > n CO o AJ Prothrombin Time (seconds) FIGURE 1. Histogram of prothrombin times among men aged years in the Vietnam Experience Study, Four white men had a prothrombin time of.3 seconds or less; the shortest time was 10.1 seconds. Ten men (four whites and six blacks) had a prothrombin time of 14.2 seconds or greater the longest time was 16.6 seconds.

6 518 Freedman et al. 100H > O c 0) 3 O" <D "5 E 3 o Current ; / Never n Current Prothrombin Time (seconds) FIGURE 2. Cumulative frequency polygons of prothrombin times among current smokers and men who had never smoked, the Vietnam Experience Study, Upper panel, distributions for white men; lower panel, distributions for black men. time. The mean prothrombin time among current smokers was 0. seconds (blacks) to 0.18 seconds (whites) lower than among men who had never smoked. (Ex-smokers had prothrombin times that were midway between those of current and never smokers.) Furthermore, the proportion of men with a prothrombin time of less than 12 seconds was about three times greater among current smokers (17 percent whites and 8 percent blacks) than among men who had never smoked (6 percent whites and 3 percent blacks). The intensity of cigarette smoking also tended to be inversely related to the prothrombin time. In regression analyses restricted to current smokers, each ad-

7 Correlates of Prothrombin Time 519 ditional pack of cigarettes was associated with a 0.04-second decrease in prothrombin time among whites {p = 0.02) and a second decrease among blacks (p = 0.39) (data not shown). The relation of prothrombin times to several dichotomous variables is shown in table 2. Men who consumed alcohol or reported having chest pains had an approximately 0.1-second shorter mean prothrombin time, as well as a higher proportion of relatively short prothrombin times, than did other men. Furthermore, the proportion of men with short prothrombin times was almost twice as common among men with peripheral vascular disease (19 percent) as among other men ( percent). In contrast, prothrombin times were prolonged among men with a chronic hepatitis B infection and among men who reported bleeding or bruising easily. Other examined characteristics were not significantly associated with prothrombin times, but several were reported by only a small number of men. For example, only 33 men reported having coronary heart disease. In analyses of prothrombin time and several continuous characteristics (table 3), no association was seen with either age or levels of HDL cholesterol, whereas the largest cor- TABLE 2. Relation of selected dichotomous characteristics to prothrombin times, Vietnam Experience Study, Characteristic and level Vietnam veteran Current drinker Chest pain Bleeding/bruising Liver damage Chronic hepatitis B infection Diabetes Peripheral vascular disease Phlebitis Coronary heart disease. 1,823 2,3% 1,021 3,197 3, , , , , , , , Meanf 12.5 ** 12.3* 12.5* ** Prothrombin time (seconds) <12$ (%) 10 9 * 10 17** * $ (%) * * * p < 0.05; * p < The null hypothesis Is no difference In mean levels or proportions across the two categories of the specified characteristic. t Mean levels are adjusted for the effects of age by using analysis of covariance. t p-values for proportions were obtained by summarizing data over race-specific 2x2 tables using the Mantet-Haenszel x' test or an exact test.

8 520 Freedman et al. TABLE 3. Relation of selected continuous characteristics to prothrombin times, Vietnam Experience Study, nhflrattfirtetir nrvi ntm. Age (years) Drinks per month Quetelet index (kg/m 2 ) Total cholesterol (mg/dl) Trigtycerides (mg/dl) HDL cholesterol (mg/dl) Bilirubin (mg/dl) < $ , Prothrombin time (seconds) , CormJntion * * -0.16* -0.15* -0.13* * 0.29* p<0.01. t Spearman correlation coefficient between prothrombin time and specified characteristic. i Values represent mean level of specified characteristic within category of prothrombin time; geometric means are shown for dnnks per month and trlgtyceride levels. Restricted to men who reported that they were currently consuming alcohol. HDL cholesterol, high density Ipoproteln cholesterol. relations (r s = ) were seen with bilirubin levels. Relative weight was inversely associated with the prothrombin time; over the five strata, the mean Quetelet index decreased by l.o kg/m 2 among whites and 1.8 kg/m 2 among blacks. Inverse associations were also observed with levels of total cholesterol and triglycerides, with Spearman correlations ranging from r 5 = -0. to Although the magnitude of most associations was similar between blacks and whites, an inverse association with alcohol consumption was apparent only among white men. Linear regression was then used to assess the independent relation of various characteristics to prothrombin times, and all variables shown in table 4 were associated with prothrombin times at the 0.10 level. Predicted changes in prothrombin time for a specified change in each characteristic are shown both in seconds and as a proportion of the standard deviation (0.41 seconds) of prothrombin time. After adjustment for other characteristics, the mean prothrombin time was estimated to be 0.17 seconds longer among blacks than among whites, equivalent to 41 percent (0.17/0.41) of the prothrombin time standard deviation. In addition, the prothrombin time decreased by 0.09 seconds with each pack of cigarettes, and compared with that for nondrinkers, the mean prothrombin time was lower (ranging from 0.03 seconds among light drinkers to 0.14 seconds among heavy drinkers) among men who consumed alcohol. In contrast, the predicted prothrombin time was prolonged among men with liver damage and among those with a chronic hepatitis B infection. We also found inverse associations between

9 Correlates of Prothrombin Time 521 TABLE 4. Linear regression analysis of changes in prothrombin time associated with changes in various characteristics, Vietnam Experience Study, Characteristic* Change In units Predicted change In prothrombin time Seconds Standard deviations! f valued Race Cigarette smoking (packs) Quetelet index (kg/m 2 ) Total cholesterol (mg/dl) Triglycertdes (mg/dl) HDL cholesterol (mg/dl) Alcohol consumption Liver damage Chronic hepatitis B infection Bilirubin (mg/dl) Age (years) White -»black > light» moderate» heavy yes ->yes n " 4,096; 22 men were excluded because of missing data for one or more characteristics. Model contains al variables that were related to the prothrombin time at the 0.10 level, with the exception of bteedlng/bnislng. Variables not associated with the prothrombin time at the 0.10 level: peripheral vascular disease, diabetes, phlebitis, chest pain, and mstary service in Vietnam. t Values represent the predicted change as a proportion of the standard deviation of prothrombin time (0.41 seconds). t A rvalue of 1.6 Is statistically significant at the 0.10 level, a t value of 2.6 at the 0.01 level, and that of 3.3 at the level. HDL cholesterol, high density lipoprotein cholesterol. Light drinking was defined as 1-14 drinks per month, moderate drinking as drinks per month, and heavy drinking as 150 drinks per month. A partial F test for the three alcohol consumption regression coefficients (H o. coefficients are all equal to 0) yielded a p value of <0.000i. the prothrombin time and levels of total cholesterol, triglycerides, and HDL cholesterol. For example, a man with a total cholesterol level of 250 mg/dl would, on average, have a 0.03 second shorter prothrombin time than would a man with a level of 200 mg/dl. As indicated by the multiple R 2 of the model, these characteristics could jointly account for about 12 percent of the variability in prothrombin time. Race-specific analyses indicated that the associations of most characteristics with prothrombin times were fairly similar between black and white men; the inverse association with alcohol consumption, however, was present among white men only. DISCUSSION Thrombosis plays an important role in the pathogenesis of cardiovascular disease: Levels of factor VII and fibrinogen are predictive of clinical disease among asymptomatic men (28), and anticoagulant therapy after a myocardial infarction reduces the risk of strokes and recurrent events (19). Relatively little data, however, describe characteristics associated with hemostatic variables in healthy populations. We found that the prothrombin time, an overall measure of the activity of the intrinsic and extrinsic pathways of coagulation, was longer among blacks than among whites and was prolonged among men with a chronic hepatitis B infection, self-reported liver disease, and a tendency to bleed or bruise easily. In addition, prothrombin times were shorter among cigarette smokers and were inversely related to relative weight, alcohol consumption, and levels of lipids and lipoproteins. Although not statistically significant, age tended to be inversely associated with prothrombin times in multivariable analyses. Whereas a prolonged prothrombin time is indicative of a deficiency of factor V, VII, X, or II (prothrombin) or of fibrinogen, a shortened prothrombin time may be associated with an increased risk of thrombosis. Levels of factor VII (6, 10) and fibrinogen (5-10) are predictive of cardiovascular disease, with the magnitude of the associations comparable with those observed for total cholesterol. (Factor VII levels are typically

10 522 Freedman et al. assessed through a modified prothrombin time assay using factor Vll-deficient plasma.) During a 10-year follow-up of 1,5 men, for example, the incidence of fatal coronary events increased by percent with each standard deviation change in levels of factor VII and fibrinogen (6). Relatively high levels of factor VII have also been found in relatives of patients with coronary heart disease (29). Other characteristics related to coagulation, such as the combined activity of factors II, VII, and X (5) or heparin-neutralizing activity (30), may also be related to cardiovascular disease. Because of the nonlinear association between prothrombin time and clotting-factor activity, even small increases in the prothrombin time may indicate a substantial reduction in the activity of several clotting factors. (These deficiencies can be congenital or can arise from liver damage or a lack of vitamin K.) For example, in factor VIIdeficient samples, each 1-second increase in the prothrombin time, assessed with rabbitbrain thromboplastin, was related to a 50 percent reduction in factor VII levels (31). Although a substantial proportion of mildly prolonged prothrombin times may be due to a factor VII deficiency (32, 33), the current findings cannot be related to the concentration of a specific coagulation factor. In addition to its association with factor VII, prothrombin times also show a strong inverse association with levels of factors V and X and with fibrinogen and prothrombin (34-36). Although few population-based studies have examined correlates of the prothrombin time, various diseases of the liver (the site of synthesis of most coagulation factors) can prolong the prothrombin time. We found that the men who had a chronic hepatitis B infection and those who reported having liver damage due to alcohol or cirrhosis had relatively long prothrombin times; furthermore, of all the examined characteristics, bilirubin levels were the most strongly associated with the prothrombin time. (It is possible that the lack of association between the prothrombin time and prevalent coronary heart disease was due to the influence of various medications on liver function.) In contrast, alcohol consumption was inversely associated with the prothrombin time, possibly due to an alcoholstimulated increase in the hepatic synthesis of various clotting factors. This mechanism may parallel the increased hepatic synthesis of apolipoprotein A-I seen with moderate alcohol consumption (37) and would be consistent with the inverse association between HDL cholesterol levels and prothrombin times. A positive association between alcohol consumption and levels of factor VII has been reported (12). The prothrombin time was also related to several other characteristics that influence the risk of coronary heart disease, and the inverse association with smoking is of particular interest as the increased risk of disease among cigarette smokers may be due to a thrombotic tendency. For example, cigarette smokers do not consistently have extensive coronary atherosclerosis (), and smokers remain at increased risk for myocardial infarction even after accounting for the extent of coronary artery occlusion (39, 40). Other investigators have reported that smoking is associated positively with levels of fibrinogen (12-14, 18, 41), possibly due to smoking-induced endothelial damage (13). We found that the mean prothrombin time was also longer among black men than among white men, consistent with reports of racial differences in levels of factor VII (12, 17). It is possible that decreased coagulation activity among black men may offset their higher prevalence of hypertension, resulting in coronary heart disease rates that are no higher than those seen among white men in the United States. Prothrombin times were also inversely related to levels of triglycerides, total cholesterol, and HDL cholesterol. Previous investigations have found that levels of factor VII are positively associated with levels of total cholesterol and triglycerides (16, 18, 42-44), and although not as extensively studied, a positive association has also been reported with levels of HDL cholesterol and apolivania, unpublished manuscript). (Since the activity of prothrombin and factors VII and X is present in the plasma fraction that is free of lipoproteins, the increased clotting

11 Correlates of Prothrombin Time 523 activity is not due to nonspecific effects on the assay (16).) Phospholipids from large lipoprotein particles may activate various clotting factors (45), and it has been suggested that dietary fat may predispose to coronary heart disease through its effects on the coagulation system (15). Various limitations of this cross-sectional study, based on a sample of army veterans aged years, should be considered in interpreting our results. Because the comparison of prothrombin times across laboratories is complicated by the use of various thromboplastins and methods to assess fibrin formation (31), the prothrombin time is often reported as a ratio (prothrombin time of test plasma/prothrombin time of control plasma). Although we reported prothrombin times in seconds, all measurements were made in a single laboratory using one instrument and one type of thromboplastin. It has also been suggested that relatively large random errors are associated with the measurement of several coagulation factors (46). An analysis of blind duplicates in this study, however, indicated that the prothrombin time was measured precisely: The mean, absolute difference between 202 pairs was 0.06 seconds, and the largest difference was ±0.2 seconds. Hereditary deficiencies of one or more clotting factors do not prevent thrombosis or myocardial infarction (47), but several coagulation factors are predictive of cardiovascular disease in the general population. Furthermore, because oral anticoagulants are effective in the secondary prevention of cardiovascular disease (19), a primary prevention trial assessing low-dose warfarin has been initiated (48). Additional data on correlates of hemostatic variables may clarify the mechanisms by which various clotting factors influence the risk of cardiovascular disease and may aid in the design of future studies aimed at modifying this risk. REFERENCES 1. Shaper AG, Pocock SJ, Phillips AN, et al. Identifying men at high risk of heart attacks: strategy for use in general practice. Br Med J (Clin Res Ed) 1986;293: Salonen JT, Salonen R, Seppanen K, et al. HDL, HDL 2, and HDL 3 subfractions, and theriskof acute myocardial infarction. A prospective population study in eastern Finnish men. Circulation 1991; 84: Maciejko JJ, Holmes DR, Kottke BA, et al. Apolipoprotein A-I as a marker of angiographically assessed coronary-artery disease. N Engl J Med 1983,309: DeWood MA, Spores J, tske R, et al. Prevalence of total coronary occlusion during the early hours of transmural myocardial infarction. N Engl J Med 1980,303: Wilhelmsen L, Svardsudd K, Korsan-Bengtsen K, et al. Fibrinogen as a risk factor for stroke and myocardial infarction. N Engl J Med 1984;3: Meade TW, Mellows S, Brozovic M, et al. Haemostatic function and ischaemic heart disease: principal results of the rthwick Park Heart Study. Lancet 1986,2: Kannel WB, Wolf PA, Castelli WP, et al. Fibrinogen and risk of cardiovascular disease. The Framingham Study. JAMA 1987^258: Yarnell JWG, Baker IA, Sweetnam PM, et al. Fibrinogen, viscosity, and white blood cell count are major risk factors for ischemic heart disease. The Caerphilly and Speedwell collaborative heart disease studies. Circulation 1991-,83: Stone MC, Thorp JM. Plasma fibrinogen a major coronary risk factor. J R Coll Gen Pract 1985;35: Balleisen L, Schulte H, Assmann G, et al. Coagulation factors and the progress of coronary heart disease. (Letter). Lancet 1987;2:461.. Pearson TA. Coronary arteriography in the study of the epidemiology of coronary artery disease. Epidemiol Rev 1984;6: Meade TW, rth WRS, Chakrabarti R, et al. Population-based distributions of haemostatic variables. Br Med Bull 1977,33: Balleisen L, Bailey J, Epping PH, et al. Epidemiological study on factor VII, factor VIII and fibrinogen in an industrial population. I. Baseline data on the relation to age, gender, body-weight, smoking, alcohol, pill-using, and menopause. Thromb Haemost 1985;54: Yarnell JWG, Sweetnam PM, Rogers S, et al. Some long term effects of smoking on the haemostatic system: a report from the Caerphilly and Speedwell Collaborative Surveys. J Clin Pathol 1987;40: Miller GJ, Martin JC, Webster J, et al. Association between dietary fat intake and plasma factor VII coagulant activity a predictor of cardiovascular mortality. Atherosclerosis 1986;60: Constantino M, Merskey C, Kudzma DJ, et al. Increased activity of vitamin K-dependent clotting factors in human hyperlipoproteinaemia association with cholesterol and triglyceride levels. Thromb Haeraost 1977;: Mead TW, Stirling Y, Thompson SG, et al. An international and interregional comparison of haemostatic variables in the study of ischemic heart disease. Report of a working group. Int J Epidemiol 1986;15: Iso H, Folsom AR, Wu KK, et al. Hemostatic variables in Japanese and Caucasian men: plasma fibrinogen, factor VIIc, factor VIIIc, and von Willebrand factor and their relations to cardiovas-

12 524 Freedman et al. cular disease risk factors. Am J Epidemiol 1989; 130: Smith P, Arnesen H, Holme I. The effect of warfarin on mortality and reinfarction after myocardial infarction. N Engl J Med 1990;323: Centers for Disease Control. Health status of Vietnam veterans. I. Psychosocial characteristics. Centers for Disease Control Vietnam Experience Study. JAMA 1988,259: Centers for Disease Control. Health status of Vietnam veterans II. Physical health. Centers for Disease Control Vietnam Experience Study. JAMA 1988^259: Freedman DS, Strogatz DS, Eaker E, et al. Differences between black and white men in correlates of high density lipoprotein cholesterol. Am J Epidemiol 1990; 132: Hoffman RE, Stehr-Green PA, Webb KB, et al. Health effects of long-term exposure to 2,3,7,8- tetrachlorodibenzo-p-dioxin. JAMA 1986;255: Centers for Disease Control. Health status of Vietnam veterans. Volume III. Medical examination. Centers for Disease Control Vietnam Experience Study. Atlanta, GA: Centers for Disease Control, Centers for Disease Control. Health status of Vietnam veterans. Supplement A. Laboratory methods and quality control. Centers for Disease Control Vietnam Experience Study. Atlanta, GA: Centers for Disease Control, Selvin S. Statistical analysis of epidemiologic data. New York, NY: Oxford University Press, 1991: Mehta CR, Patel NR. StatXact: statistical software for exact nonparametric inference (version 1.0). User manual. Cambridge, MA: Cytel Software Corporation, 1989: Hultin MB. Fibrinogen and factor VII as risk factors in vascular disease. Prog Hemost Thromb 1991;IO: Hoffman CJ, Miller RH, Lawson WE, et al. Elevation of factor VII activity and mass in young adults at risk of ischemic heart disease. J Am Coll Cardiol 1989;14: Yarnell JWG, Sweetnam PM, Elwood PC, et al. Haemostatic factors and ischaemic heart disease. The Caerphilly study. Br Heart J 1985;53: The International Committee on Thrombosis and Haemostasis, The International Committee for Standardization in Hematology. Prothrombin time standardization: report of the expert panel on oral anticoagulant control. Thromb Haemost 1979; 42: Green G, Poller L, Thomson JM, et al. Factor VII as a marker of hepatocellular synthetic function in liver disease. J Clin Pathol 1976;29: Mazzucconi MG, Mariani G, Chistolini A, et al. Evaluation of the nature of mildly prolonged prothrombin times. Am J Hematol 1987;24: Doellgast GJ, Triscott MX, Buss DH, et al. Extrinsic-pathway enzyme-linked coagulation assay (EP- ELCA). A clot-based alternative to prothrombin time for measurement of extrinsic pathway factors in plasma. Clin Chem 1988;34: Kato H, Uchida K. Automated fluorogenic method for the evaluation of the extrinsic coagulation reactions in human plasma. Thromb Res 1988;5O: Poller L. Laboratory control of anticoagulant therapy. Semin Thromb Hemost 1986; 12: Savolainen MJ. How does alcohol raise HDLcholesterol concentration? (Editorial). Ann Med 1990,22: Solberg LA, Strong JP. Risk factors and atherosclerotic lesions. A review of autopsy studies. Arteriosclerosis 1983;3: McKenna WJ, Chew CYC, Oakley. Myocardial infarction with normal coronary angiogram. Possible mechanism of smoking risk in coronary artery disease. Br Heart J 1980;43: Freedman DS, Gruchow HW, Walker JA, et al. Cigarette smoking and non-fatal myocardial infarction in women: is the relation independent of coronary artery disease? Br Heart J 1989,62: Korsan-Bengtsen K, Wilhelmsen L, Tibblin G. Blood coagulation and fibrinolysis in a random sample of 788 men 54 years old. II. Relations of the variables to "risk factors" for myocardial infarction. Thromb Diath Haemorrh 1972;29: Balleisen L, Assmann G, Bailey J, et al. Epidemiological study on factor VII, factor VII, factor VIII and fibrinogen in an industrial population. II. Baseline data on the relation to blood pressure, blood glucose, uric acid, and lipid fractions. Thromb Haemost 1985;54: Heller RF, Meade TW, Haines AP, et al. Interrelationships between factor VII, serum testosterone and plasma lipoproteins. Thromb Res 1982; 28: Lowe GDO, Wood DA, Douglas JT, et al. Relationships of plasma viscosity, coagulation and fibrinolysis to coronary risk factors and angina. Thromb Haemost 1991 ;65: Hunt BJ. The relation between abnormal hemostatic function and the progression of coronary disease. Curr Opin Cardiol 1990;5: Thompson SG, Martin JC, Meade TW. Sources of variability in coagulation factor assays. Thromb Haemost 1987;58: Ratnoff OD. Thrombosis and the hypercoagulable state. Circulation 1984;70(5, pt 2):III Meade TW, Wilkes HC, Stirling Y, et al. Randomized controlled trial of low dose warfarin in the primary prevention of ischaemic heart disease in men at high risk: design and pilot study. Eur Heart J 1988,9: