Serum albumin, death Richard

F. Gillum,

coronary

MD, and Diane M. Makuc,

heart disease,

DrPH.

A marked decrease in mortality from all causes, cardiovascular disease, and cancer, with increasing concentrations of serum albumin, a measure of nutritional status, was an unexpected finding in a follow-up study of middle-aged British men.l Given the lack of a prior hypothesis, confirmatory reports from other studies, and causal mechanisms, replication of this result is important to exclude its having been a chance finding. Data from the National Health and Nutrition Examination Survey Epidemiologic Follow-up Study (NHEFS) provided an opportunity to confirm the mortality finding in white men and to extend the findings to the incidence of coronary heart disease (CHD) and to white women and black men and women. METHODOLOGY

The first National Health and Nutrition Examination Survey (NHANES I) collected data from a nationwide, multistage, probability sample of the civilian noninstitutionalized population, aged 1 to 74 years, of the United States, excluding Alaska, Hawaii, and reservation lands of American Indians.2j 3Details of the plan, complex sample design, response, and operation were published previously, as were procedures used to obtain informed consent and maintain

From

the National

Center

for Health

Statistics.

The NHANES I Epidemiologic Follow-up Study was developed and funded by the following agencies: National Center for Health Statistics; National Institute on Aging; National Cancer Institute; National Institute of Child Health and Human Development; National Heart, Lung, and Blood Institute; National Institute of Mental Health; National Institute of Diabetes and Digestive and Kidney Diseases; National Institute of Arthritis and Musculoskeletal and Skin Diseases; National Institute of Allergy and Infectious Diseases; National Institute of Neurological and Communicative Disorders and Stroke; Centers for Disease Control; and United States Department of Agriculture. The field work was conducted by Westat, Inc. (contract numbers 233-80-2049 and 282-84-2111). Received

for publication

Reprint requests: National Center Hyattsville, MD 4/l/34094

June

19, 1991;

accepted

Aug.

R. F. Gillum, MD, Office of Analysis for Health Statistics, 6525 Belcrest 20782.

2, 1991. and Epidemiology, Rd., Room 1000,

Hyattsville,

and

Add.

confidentiality of obtained information.2, 3 The elderly, women of childbearing age, and persons residing in poverty areas were oversampled. Serum albumin levels were measured only in the nutrition component, examination locations 1 to 65 from 1971 to 1974. The NHANES I Epidemiologic Follow-up Study (NHEFS) is a longitudinal study of participants in NHANES I, who were 25 to 74 years of age at the time of the survey from 1971 to 197L4, 5 Personal interviews and results of physical and laboratory examinations from NHANES I provided the baseline data for the NHEFS. This analysis was based on three waves of follow-up data collection during 1982 to 1984, 1986, and 1987. Data collected consisted of three interview surveys, medical records from health care facilities for the period between baseline and last follow-up examination, and death certificates for all decedents. Information concerning place and date of hospitalizations was obtained from the follow-up interview, the death certificate, or both; hence without one of these data sources hospital records were usually unavailable. Hospital records may have been missing because of subject or proxy refusals, hospital refusals, or inability to locate specific hospitals or hospital records. The completeness of data collection has been reported. 5t6 Persons who were institutionalized after the baseline examination were included in the NHEFS. Age, race, and sex-specific mortality among the NHEFS cohort have been quite similar to what was expected for the United States population.7 This analysis included white and black persons, aged 45 to 74 years at baseline, examined at locations 1 to 65 in NHANES I. Of the 6299 white and black persons aged 45 to 74 years at baseline, 103 (1.6%) were lost to follow-up. Excluded from all analyses were 431 persons whose baseline serum albumin, systolic blood pressure, and serum cholesterol values and level of education were unknown. Also excluded from CHD analyses were 92 persons who had no follow-up interview or death certificate or were missing baseline data on history of heart disease and 804 persons who had a positive history of heart diseaseat 507

and Makuc

February 1992 Heart Journal

508

Gillum

Table

1.Number of incident casesof CHD, deaths, and personsat risk of death by sex, ageat baseline,and race

American

Deaths Sex, age at baseline, and race

CHD

Total Men 45-64 yr White Black 65-74 yr White Black Women 45-64 yr White Black 65-74 yr White Black *A total of 804 persons death only.

cases* 1,352

with

a history

All causes

CHD*

2,399

513

Cardiovascular disease

Noncardiovascular disease

1,268

No. at risk of death*

1,266

5,765

225 27

287 72

72 12

145 30

142 42

1,055 182

401 71

833 188

180 31

451 92

382 96

1,236 255

138 28

155 58

30 11

60 30

95 28

1,154 221

394 68

651 155

150 27

374 86

277 204

of heart

disease

at baseline

and 92 persons

baseline, defined as those who had ever been told by a doctor that they had had a heart attack or heart failure or who had used any medicine, drugs, or pills for a weak heart during the 6 months before the baseline interview. After all exclusions 5765 persons remained for mortality analyses (4834 white and 931 black persons) and 4869 remained for CHD analyses. The length of follow-up in the NHEFS for mortality analyses ranged from 9 to 16 years (median 15 years). Coronary heart disease criteria. Incident cases of CHD met at least one of the following criteria: (1) a death certificate with underlying or nonunderlying cause of death coded 410-414 according to the International Classification of Diseases Ninth Revision (ICD-9) or (2) one or more hospital stays during the follow-up period with any discharge diagnosis coded 410-414 according to the Clinical Modification of ICD-9 (ICD-9 CM). The date of incidence was estimated as one of the following: (1) date of first hospital admission with a diagnosis of CHD, (2) date of death for CHD deaths without any CHD hospital records. Cause of death for mortality analyses was defined by the underlying cause of death: CHD, ICD-9, 410-414 or cardiovascular disease, ICD-9 390-448. Baseline variables. Blood samples were obtained and frozen serum was sent to the Centers for Disease Control for determination of serum albumin and serum total cholesterol levels8, g For persons examined at locations 1 to 65 from 1971 to 1974, serum albumin was measured by a modification of the automated

with

missing

data

were excluded

from

analyses

1,389

273 of CHD

incidence

and CHD

bromcresol green method of Doumas et al.g The Centers for Disease Control method for measuring serum albumin is similar to the Technicon AutoAnalyzer II (Technicon Diagnostics, Tarrytown, N.Y.) and SMA AutoAnalyzer (Technicon Diagnostics) method. Serum albumin values were missing for less than 3% of persons. Categories of serum albumin used in the analysis were approximate tertiles for the entire study population. The baseline medical history questionnaire included questions about selected conditions (e.g., diabetes) diagnosed by a doctor and about medications used for selected conditions during the preceding 6 months.2 At the beginning of the baseline physical examination, the physician measured the blood pressure with the examinee seated.lO Smoking status was obtained at baseline for the subsample that received the detailed examination. For the remaining persons smoking status at baseline was derived from questions on the follow-up interview on lifetime smoking history or imputed.ll, I2 The validity of this approach has been documented.ll, I2 Other baseline variables were measured as described elsewhere.2 Statistical methods. Estimates of the risk of CHD and death for persons with higher serum albumin concentrations relative to those with lower concentrations derive from Cox proportional hazards regression models computed with the PHGLM procedure in the Statistical Analysis System.i3 Separate analyses have been carried out for six groups based on age at baseline (45 to 64 and 65 to 74 years), race

Volume

123

Number

2

Serum albumin and coronary disease 509

(white or black), and sex. All models included age at baseline in single years as a covariate. Persons who died of other causes in cause-specific analyses were censored at the date of death. Survivors were censored at the date of the follow-up interview. To assess the effect of preexisting morbidity, analyses were repeated for groups with and without a history of any of a list of 26 chronic diseases at baseline; further analyses were repeated excluding the first 5 years of follow-up. To assess the effect of the sample weights on the results, person-months logistic regression analyses were done weighted and unweighted by means of the LOGIST procedure in the Statistical Analysis System. 13rl4 The results were consistent with minimal effects of weighting on the main conclusions derived from the unweighted estimates, which have smaller variances. Therefore only the data from the unweighted Cox regression models are presented. Means, unpaired t tests, and Pearson correlation analyses were performed with the use of standard methods. OBSERVATIONS Serum albumin

Serum albumin concentrations ranged from 2.9 to 5.5 gm/dl (mean 4.34) in white men, 2.8 to 5.5 gm/dl (mean 4.31) in white women, 3.0 to 5.6 gm/dl in black men (mean 4.27), and 2.7 to 5.4 gm/dl in black women (mean 4.25) among persons in the mortality analyses. Serum albumin values were negatively correlated with age (r = -0.22 in white men and -0.12 in white women, p = 0.0001). Thus other correlations were examined separately within two age groups, 45 to 64 years and 65 to 74 years. Other statistically significant associations (p < 0.01) within age groups in white men and women included positive correlations with serum cholesterol levels and blood pressure. In white men serum albumin levels were only slightly lower in smokers than in nonsmokers (e.g., means at age 45 to 64 years for smokers, 4.38 gm/dl; for nonsmokers, 4.42 gm/dl; p = 0.03). In white men and women years of education and family income were positively associated with serum albumin values only in men aged 45 to 64 years (p = 0.001). White men with one or more of a list of chronic diseases at baseline generally had lower serum albumin levels than those with no disease (e.g., means for white men aged 45 to 64 years with no disease, 4.44 gm/dl; with one or more diseases, 4.37 gm/dl; p = 0.0004). However, differences for other sex-race groups were generally not statistically significant. Incidence of coronary heart disease. Table I shows the number of incident cases of CHD and the number of deaths by sex, race, and age at baseline examand cardiovascular

risk factors.

Table II. Relative risks for CHD incidence associated with

serum albumin concentration in white men and women Sex, age at baseline, and serum albumin (gmldl) Men 45-64 yr 4.4 65-74 yr 4.4 Women 45-64 yr 4.4 65-74 yr 4.4 45-74 yi4.4

Age-adjusted RR (95% CL)

Risk-adjusted* RR (95% CL)

1.00 0.69 (0.50-0.96)

1.00 0.68 (0.48-0.95) 0.51 (0.36-0.73)

0.55 (0.39-0.78)

1.00 1.07 (0.84-1.38) 1.27 (0.98-1.64)

1.00

1.00 1.02 (0.79-1.31) 1.15 (0.89-1.50)

0.95 (0.63-1.44) 0.70 (0.44-1.11)

1.00 0.93 (0.61-1.41) 0.64 (0.40-1.03)

1.00 0.82 (0.66-1.03) 0.75 (0.57-0.97)

1.00 0.81 (0.64-1.02) 0.73 (0.56-0.96)

1.00 0.85 (0.70-1.04) 0.73 (0.58-0.92)

1.00 0.83 (0.68-1.01) 0.70 (0.55-0.88)

RR, Relative risk; 95% CL, 95% confidence limits. *Adjusted for baseline age, smoking, systolic blood pressure, cholesterol, history of diabetes, and level of education.

serum total

ination. In white persons age-adjusted relative risk (RR) for CHD during the follow-up period was decreased for younger men with serum albumin levels of 4.2 to 4.4 gm/dl at baseline and decreased still further for those with levels of 4.5 gm/dl or more compared to men with levels of 4.1 gm/dl or less (Table II). Relative risk was also decreased for women aged 45 to 74 years with levels of 4.5 gm/dl or more compared to 4.1 gm/dl or less. Table II shows the relative risk for CHD incidence after adjusting for baseline smoking, systolic blood pressure, serum total cholesterol, history of diabetes, and level of education in addition to age in white men and women. Relative risk for CHD changed little and remained reduced for men aged 45 to 64 years and women aged 45 to 74 years with albumin levels of 4.5 gm/dl or more. To be noted was the lack of association in men aged 65 to 74 years, which contrasted with the association in women. A statistically significant independent reduction in the incidence of CHD with serum albumin levels of 4.5 gm/dl or more was detected in black men aged 45 to 74 years (men: RR = 0.54, 95% confidence limits

5 10

Gillum and Makuc

American

III. Relative risks for death from all causesassociated with serum albumin concentration in white men and women Table

Sex, age at baseline, and serum albumin (gmldl) Men 45-64 yr 4.4 65-74 yr 4.4 45-74 yr 4.4 Women 45-64 yr 4.4 65-74 yr 4.4 45-74 yr 4.4

Age-adjusted RR (95% CL)

Risk-adjusted* RR (95% CL)

1.00 0.64 (0.48-0.86) 0.61 (0.46-0.82)

1.00 0.65 (0.48-0.87) 0.62 (0.45-0.84)

1.00 0.89 (0.75-1.04) 0.79 (0.66-0.94)

1.00 0.86 (0.73-1.01) 0.77 (0.64-0.92)

1.00 0.82 (0.72-0.95) 0.74 (0.64-0.87)

1.00 0.81 (0.70-0.93) 0.73 (0.62-0.85)

1.00 0.71 (0.48-1.04) 0.70 (0.47-1.05)

1.00 0.74 (0.50-1.09) 0.71 (0.47-1.07)

1.00 0.84 (0.70-1.00) 0.74 (0.60-0.91)

1.00 0.81 (0.68-0.97) 0.71 (0.57-0.87)

1.00 0.82 (0.70-0.96) 0.74 (0.61-0.88)

1.00 0.79 (0.67-0.93) 0.71 (0.59-0.85)

RR, Relative risk, 95% CL, 95O1 confidence limits. *Adjusted for baseline age, smoking, systolic blood pressure, cholesterol, history of diabetes, and level of education.

[CL]: = 0.32,0.90;womenRR 1.49). However, the number Death in white

serum total

= 0.98,95% CL = 0.65, of cases was small.

men and women

All causes. Table I shows the number

of deaths and population at risk by age, sex, and race. The age-adjusted risk of death from all causes was reduced in younger and older white men and women with serum albumin levels of 4.5 gm/dl or more relative to those with concentrations of 4.1 gm/dl or less (Table III). After adjusting for baseline risk factors, relative risk was essentially unchanged (Table III). Death from coronary heart disease. Age-adjusted risk of death from CHD in persons with no heart disease at baseline examination was statistically significantly reduced only in white men, aged 45 to 64 years, with baseline serum albumin values of 4.2 to 4.4 (RR = 0.45; 95% CL = 0.25, 0.80) and 4.5 gm/dl or more (RR = 0.45; 95% CL = 0.25, 0.79). The reduction in risk in middle-aged white men remained unchanged and statistically significant after adjust-

February 1992 Heart Journal

ing for baseline risk factors. There were trends toward reduced risk with higher serum albumin levels in white women but not in older white men. Cardiovascular death. Age-adjusted risk of death from cardiovascular diseases was reduced in white men, aged 45 to 74 years, with serum albumin levels of 4.5 gm/dl or more (Table IV). At age 45 to 64 years relative risk for serum albumin levels of 4.5 gm/dl or more was 0.61(95 % CL = 0.40,0.91) in men (data not shown). After adjusting for baseline risk factors, relative risk for death from cardiovascular diseases with serum albumin levels of 4.5 gm/dl or more was statistically significantly reduced for white men and women, aged 45 to 74 years (Table IV). At age 45 to 64 years, adjusted relative risk was 0.57 (95:; CL = 0.37, 0.87) for men (data not shown). Noncardiovascular death. Age-adjusted risk of death from noncardiovascular diseases was reduced in white men and white women, aged 45 to 74 years, with serum albumin levels of 4.5 gm/dl or more (Table IV). After adjusting for baseline risk factors, relative risk for death from noncardiovascular diseases was similar and still reduced for white men and women, aged 45 to 74 years (Table IV). Effects were similar at ages 45 to 64 and 65 to 74 years in women and men. Preexisting disease at baseline. To exclude effects of preexisting disease at baseline, the association between serum albumin concentrations and death from all causes was examined in persons with (about 70% ) and without (about 30 % ) a history of a doctor’s diagnosis of any of a list of 26 chronic diseases and five major types of prescription medications ascertained by questionnaire at baseline. In white persons higher serum albumin levels were associated with reduced age-adjusted mortality in men aged 45 to 74 years in the groups with and without prior disease and in women aged 45 to 74 years in the group with prior disease at baseline. In women aged 45 to 74 years, with no prior disease at baseline, higher serum albumin levels had no protective effect (age-adjusted RR = 1.23; 95% CL = 0.79, 1.93). The relative risk changed little after adjusting for baseline risk factors. Similar patterns were seen for death from cardiovascular and noncardiovascular causes, although the reduced risk with increasing albumin concentrations was not statistically significant in the smaller group with no history of disease. In further analyses of the entire cohort, the independent protective effect of higher serum albumin levels persisted for white men (RR = 0.72; 95% CL = 0.60, 0.87) and women (RR = 0.78; 95%) CL = 0.63,0.76) when deaths from all causes occurring in the first 5 years of follow-up were excluded. This was

Volume Number

123 2

Serum albumin

IV. Relative risks for death from selected causes associated with serum albumin concentration in white men and women, aged 45 to 74 years at baseline

Table

Sex, Cause of death, and serum albumin (gmldl)

Age-adjusted RR (95% CL)

V. Relative risks for death from selected causes associated with serum albumin concentration in black men and women, aged 45 to 74 years at baseline

Table

Sex, cause of death, and serum albumin

Risk-adjusted * RR (95% CL)

(gmJdl) Men CVD 4.4 Non-CVD 4.4 Women CVD 4.4 Non-CVD 4.4

1.00 0.83 (0.68-1.01) 0.77 (0.62-0.95)

1.00 0.78 (0.64-0.94) 0.69 (0.55-0.85)

1.00 0.82 (0.67-1.01) 0.71 (0.57-0.89)

1.00 0.84 (0.68-1.03) 0.77 (0.61-0.97)

1.00 0.83 (0.66-1.03) 0.81 (0.63-1.03)

1.00 0.78 (0.62-0.97) 0.74 (0.58-0.96)

1.00 0.81 (0.64-1.02) 0.66 (0.50-0.87)

1.00 0.80 (0.63-1.01) 0.66 (0.50-0.88)

RR, Relative risk; 95% CL, 95% confidence limits; disease; non-CVD, noncardiovascular disease.

CVD,

Men All causes ~4.2 4.2-4.4 >4.4 CVD 4.4 Non-CVD 4.4 Women All causes 4.4 CVD 4.4 Non-CVD 4.4

cardiovascular

*Adjusted for baseline age, smoking, systolic blood pressure, cholesterol, history of diabetes, and level of education.

serum total

also true for cardiovascular deaths in white men and noncardiovascular deaths in white women, Death in black men and women. Table V shows that higher serum albumin levels were related to reduced risk of death from all causes in black men and women Bged 45 to 74 years. Higher serum albumin levels were also associated with reduced risk of death from cardiovascular disease in black men and women (Table V). This association was significant after controlling for baseline risk factors. Higher serum albumin levels were associated with reductions in risk of noncardiovascular disease among black women that were not statistically significant after controlling for baseline risk factors, perhaps because of the small number of deaths. COMMENTS

Baseline serum albumin concentrations in the upper third of the population distribution compared with the lower third were associated with a decreased risk of CHD incidence and death from cardiovascular and noncardiovascular diseases independent of other risk factors in white men and women, aged 45 to 74 years, with the exception that no association between CHD incidence and CHD death was seen in white men aged 65 to 74 at baseline. Among black

and coronary disease 5 11

Age-adjusted

RR (95%

CL)

Risk-adjusted* RR 195%

CL)

1.00 0.81 (0.61-1.06) 0.64 (0.46-0.89)

1.00 0.83 (0.63-1.11) 0.67 (0.48-0.94)

1.00 0.82 (0.55-1.22) 0.60 (0.36-0.97)

1.00 0.76 (0.51-1.15 0.54 (0.33-0.90)

1.00 0.80 (0.54-1.17) 0.68 (0.44-1.06)

1.00 0.89 (0.60-1.33) 0.81 (0.51-1.28)

1.00 0.64 (0.47-0.87) 0.56 (0.39-0.80)

1.00 0.68 (0.50-0.93) 0.56 (0.39-0.80)

1.00 0.63 (0.41-0.95) 0.53 (0.33-0.87)

1.00 0.68 (0.44-1.04) 0.53 (0.32-0.86)

1.00 0.66 (0.42-1.05) 0.59 (0.35-0.99)

1.00 0.69 (0.43-1.09) 0.60 (0.36-1.02)

Relative risk; 95% CL, 95 “i) confidence limits: CVD, cardiovascular disease; non-CVD, noncardiovascular disease. *Adjusted for baseline age, smoking, systolic blood pressure, serum total cholesterol, history of diabetes, and level of education.

RR,

persons risks of CHD incidence in men and deaths from all causes and from cardiovascular diseases in men and women were reduced with higher serum albumin levels. Serum albumin has not previously been identified as a coronary risk factor.t5 Protein deficiency has been listed as a possible initiator of atherosclerosis.15 Low intake of animal protein has been suggested as a risk factor for stroke in Japan based on ecologic analyses.16 However, in unpublished analyses of NHANES I data, the serum albumin concentration was not significantly correlated with absolute or percentage of recommended dietary protein, fat, or calorie intake from 24-hour dietary recall in any age, sex, or race group except 65 to 74year-old white men (r = 0.11 for protein intake) (National Center for Health Statistics: unpublished data). Patterns of correlation of serum albumin concentrations with multiple baseline variables in NHANES I were sim-

February

5 12

Gillum

and Makuc

ilar to those reported above for the NHEFS sample with complete follow-up data in published and unpublished analyses. l7 Serum albumin concentration has been reported to be positively correlated with high-density lipoprotein (HDL) cholesterol concentration in one sample of healthy men.18 However, in preliminary unpublished analyses of data from the second NHANES, serum albumin levels were not significantly correlated with HDL cholesterol levels at ages 45 to 64 or 65 to 74 years. This suggests that confounding by HDL cholesterol probably does not explain the association of albumin with cardiovascular disease in NHEFS, even if a low concentration of HDL cholesterol is a risk factor for CHD. Further, the significant univariate association of HDL cholesterol with CHD in the British study was greatly diminished and no longer statistically significant (p = 0.07) in multiple logistic models.lg Presumably because of this, HDL cholesterol was not controlled in the British study of albumin1 The lack of a plausible mechanism for an effect of serum albumin on the incidence of CHD suggests that it may be an indicator of some factor influencing some stage of the atherosclerotic process from initiating lesions to precipitating events. Effects of albumin concentration on blood viscosity and free fatty acid transport might be considered. The lack of effect on risk of CHD in older men is puzzling. However, serum total cholesterol has also been reported to lose its predictive power in older persons in this study and others.20 Similarly puzzling is the lack of a protective effect of higher serum albumin levels in white women with no prior disease. Mechanisms for effects on noncardiovascular mortality are also unknown. However, replication of the British findings in American white men and women virtually eliminates chance as an explanation for the findings. In preliminary analyses not shown, associations between albumin levels and the incidence of CHD at the 1984 follow-up were similar to those seen with additional follow-up through 1987. Associations with CHD death in white men aged 45 to 64 years were slightly stronger and associations with cardiovascular death in white women aged 65 to 74 years were also stronger. Limitations. Limitations of the study include possible bias arising from loss to follow-up and, for incidence, missing hospital data at follow-up. Among white men and women, a small number of persons (66 white and 26 black) reported hospitalization for heart attack, angina, or “heart trouble” without a matching hospital record to establish the diagnosis. A further source of possible bias was misclassification resulting from inaccuracy of diagnoses coded on death

American

Heart

1992 Journal

certificates and hospital records. Inaccuracy of baseline history of heart disease could result in casesbeing identified during follow-up that were actually diagnosed before baseline. Further, for CHD incidence the year of onset was taken as the year of first hospitalization, but this could be in error if an earlier hospitalization for CHD was not reported or the records found. Errors in measurement of serum albumin concentrations and other baseline variables could also result in bias, but this seemsunlikely given the standardized nature of the NHANES I examination. The effects of use of sample weights on logistic regression coefficients described previously generally supported the validity of using unweighted data. Confounding by variables not measured cannot be excluded; however, most important risk factors were available with the exception of HDL cholesterol. Statistical power was relatively low for younger white women and for black men and women. Additional research is needed on serum albumin as a risk factor for CHD and mortality in white women, the elderly, and black men and women. Study of larger cohorts of black men and women would permit examination of associations between albumin and CHD and cause-specific mortality. Mechanisms for these associations in white and black persons might be further elucidated by including measures of HDL cholesterol, left ventricular mass, serum insulin, prior disease, and functional status, as well as multiple measurements of serum albumin levels during the follow-up period. SUMMARY

To confirm a reported association between elevated serum albumin concentrations and reduced risk of death in middle-aged white men and to determine whether such associations exist for CHD incidence in white men and CHD and death in white women and black men and women, data w,ere examined from the NHANES I Epidemiologic Follow-up Study. Over a follow-up period of 9 to 16 years, serum albumin concentrations of 4.5 gm/dl or more were associated with reduced risk of CHD incidence in white men aged 45 to 64 years (RR = 0.51; 95% CL = 0.36, 0.73) and in white women aged 45 to 74 years (RR = 0.70; 95 % CL = 0.55,0.88), independent of baseline risk factors. Independent reductions in risk of death from all causes,cardiovascular diseases, and noncardiovascular diseases were also seen in white men and women. Relative risk of death from all causesat ages 45 to 74 years in the white population was 0.73 (95% CL = 0.62, 0.85) for men and 0.71 (95% CL = 0.59,0.85) for women. Similar reductions in risk of death from all causes and cardiovascular

Volume Number

123 2

diseases were seen in black men and women, despite the small numbers. Further studies are needed to confirm these findings for women and black persons and to elucidate mechanisms for the effect of serum albumin. REFERENCES

1. Phillips A, Shaper AG, Whincup PH. Association between serum albumin and mortality from cardiovascular disease, cancer. and other causes. Lancet 1989;2:1434-6. 2. Miller HW. Plan and operation of.the Health and Nutrition Examination Survey, United States, 1971-73. Hyattsville, Maryland: National Center for Health Statistics, 1973; DHEW publication no. (HSM) 73-1310. (Vital and health statistics; series 1, no. lOa, lob). 3. Engel A, Murphy RS, Maurer K, Collins E. Plan and operation of the HANES I Augmentation Survey of Adults 25-74 Years, United States, 1974-75. Hyattsville, Maryland: National Center for Health Statistics, 1978; DHEW publication no. (PHS) 78-1314. (Vital and health statistics; series 1; no. 14). 4. Cohen BB, Barbano HE, Cox CS, et al. Plan and operation of the NHANES I Epidemiologic Follow-up Study: 198284. Hyattsville, Maryland: National Center for Health Statistics, 1987; DHHS publication no. (PHS) 87-1324. (Vital and health statistics; series 1, no. 22). 5. Finucane FF, Freid VM, Madans JH, et al. Plan and operation of the NHANES I Epidemiologic Follow-up Study, 1986. Hyattsville, Maryland: National Center for Health Statistics, 1990; DHHS publication no. (PHS) 90-1307. (Vital and health statistics; series 1; no. 25). 6. Madans JH, Kleinman JC, Cox CS, et al. Ten years after NHANES I: report of initial follow-up, 1982-84. Public Health Rep 1986;101:465-73. 7. Madans JH, Cox CS, Kleinman JC, et al. Ten years after NHANES I: mortality experience at initial follow-up, 198284. Public Health Rep 1986;101:474-81. 8. Fulwood R, Abraham S, Johnson CL. Serum cholesterol levels of persons 4-74 years of age, by socioeconomic characteristics: United States 1971-74. Hyattsville, Maryland: National Center for Health Statistics. 1980: DHEW publication no. (PHS) 80-1667. (Vital and health statistics; series 11; no. 217).

Serum albumin and coronary disease

513

9. HANESI hematology and clinical chemistry procedures developed or utilized by the Centers for Disease Control, Bureau of Laboratories, 1971-1975. Hyattsville, Maryland: National Center for Health Statistics, 1979; DHEW (PHS). 10. Roberts J, Maurer K. Blood pressure levels of persons 6-74 years, United States, 1971-74. Hyattsville, Maryland National Center for Health Statistics, 1977; DHEW publication no. (HRA) 78-1648. (Vital and health statistics; series 11; no. 203). 11. McLaughlin JK, Dietz MS, Mehl ES, Blot WJ. Reliability of surrogate information on cigarette smoking by type of informant. Am J Epidemiol 1987;126:144-6. 12. Machlin SR, Kleinman JC, Madans JH. Validity of mortality analysis based on retrospective smoking information. Stat Med 1989;8:997-1009. 13. Hastings RP, ed. SUGI supplemental library user’s guide. Version 5 ed. Cary, NC: SAS Institute, Inc., 1986:269-93,43774. 14. Ingram DD, Kleinman JC. Empirical comparisons of proportional hazards and logistic regression models. Stat Med 1989;8:525-38. 15. Hopkins PN, Williams RR. Identification and relative weight of cardiovascular risk factors. Cardiol Clin 1986;4:3-31. 16. Omura T, Hisamatsu S, Takizawa Y, Minowa M, Yanagawa H, Shigematsu I. Geographical distribution of cerebrovascular disease mortality and food intakes in Japan. Sot Sci Med 1987;24:401-7. 17. Micozzi MS, Albanes D, Stevens RG. Relation of body size and composition to clinical, biochemical and hematologic indices in US men and women. Am J Clin Nutr 1989,50:1276-81. 18. Nanji AA, Reddy S. Use of total cholesterol/albumin ratio as an alternative to high density lipoprotein cholesterol measurement. J Clin Path01 1983;36:716-8. 19. Shaper AG, Pocock SJ, Walker M, Phillips AN, Whitehead TP, Macfarlane PW. Risk factors for ischaemic heart disease: the prospective phase of the British Regional Heart Study. 3 Epidemiol Community Health 1985;39:197-209. 20. Leaverton PE, Havlik RJ, Ingster-Moore LM, Lacroix AZ, Cornoni-Huntley JC. Coronary heart disease and hypertension. In: Cornoni-Huntley JC, Huntley RR, Feldman JJ, eds. Health status and well-being of the elderly: National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study. New York: Oxford University Press, 1990:53-70.