Donate Help Contact The AHA Sign In Home
American Heart Association
Circulation
Search: search_blue_button Advanced Search
« Previous Article | Table of Contents | Next Article »
Circulation. 1997;96:1408-1415

This Article
Right arrow Abstract Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fagot-Campagna, A.
Right arrow Articles by Knowler, W. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fagot-Campagna, A.
Right arrow Articles by Knowler, W. C.

(Circulation. 1997;96:1408-1415.)
© 1997 American Heart Association, Inc.

Articles

Serum Cholesterol and Mortality Rates in a Native American Population With Low Cholesterol Concentrations

A U-Shaped Association

Anne Fagot-Campagna, MD, MPH; Robert L. Hanson, MD, MPH; K. M. Venkat Narayan, MD, MSc; Maurice L. Sievers, MD; David J. Pettitt, MD; Robert G. Nelson, MD, MPH; ; William C. Knowler, MD, DrPH

From the National Institute of Diabetes and Digestive and Kidney Diseases, Phoenix, Ariz.

Correspondence to Anne Fagot-Campagna, National Institute of Diabetes and Digestive and Kidney Diseases, 1550 E Indian School Rd, Phoenix, AZ 85014. E-mail fax{at}cu.nih.gov


*    Abstract
up arrowTop
 *Abstract
down arrowIntroduction
 down arrowMethods
 down arrowResults
 down arrowDiscussion
 down arrowReferences
 
Background Low serum cholesterol concentrations are associated with high death rates from cancer, trauma, and infectious diseases, but the meaning of these associations remains controversial. The present report evaluates whether low cholesterol is likely to be a causal factor for mortality from all causes or from specific causes.

Methods and Results Among 4553 Pima Indians >=20 years old, a population with low serum cholesterol (median, 4.50 mmol/L), 1077 deaths occurred during a mean follow-up of 12.8 years. Trauma was the most common cause. The relationship between serum cholesterol measured at 2-year intervals and age- and sex-standardized mortality rates was U-shaped. Cholesterol was related positively to mortality from cardiovascular diseases and diabetes (including nephropathy) and negatively to mortality from cancer and alcohol-related diseases. The relationship was U-shaped for mortality from infectious diseases, and cholesterol was not related to mortality from trauma. Change in cholesterol from one examination to the next was positively related to mortality from diabetes. In proportional-hazards models adjusted for potential confounders, the relationship between baseline cholesterol and mortality was U-shaped for all causes and diabetes and positive for cardiovascular diseases. Other relationships were nonsignificant. Among 3358 subjects followed >=5 years, the relationship was significant and positive only for mortality from cardiovascular diseases.

Conclusions Despite a high exposure risk for Pima Indians, if low cholesterol level is a causal factor, the relationships between low serum cholesterol and high mortality rates probably result from diseases lowering cholesterol rather than from a low cholesterol causing the diseases.


Key Words: cholesterol • risk factors • mortality


*    Introduction
up arrowTop
 up arrowAbstract
 *Introduction
down arrowMethods
 down arrowResults
 down arrowDiscussion
 down arrowReferences
 
Lowering serum cholesterol concentration reduces the incidence of coronary heart disease.1 However, low cholesterol levels are also associated with higher death rates from cancer, accidents, suicides, homicides, and infectious diseases, and serum cholesterol has a U-shaped relationship with mortality in men and no significant relationship in women.2 Although low serum cholesterol concentration may be a risk factor for certain diseases, it may also reflect the presence of underlying disease, such as undiagnosed cancer or alcoholism, that is associated with early mortality.2 3 4 Previous studies that excluded the first years of follow-up supported the latter conclusion because they showed nonsignificant5 6 7 or weaker8 9 10 11 12 13 associations of mortality rates with low cholesterol concentration, and excess deaths have been reported to occur in older and less healthy subjects.10

Pima Indians have, on average, lower cholesterol levels14 15 and lower death rates from cardiovascular diseases than the general US population,16 but they have high mortality rates from diabetes, accidents, homicides, suicides, and cirrhosis.16 Therefore, the association of low cholesterol with mortality may be highlighted in this population, even though cancer is not a leading cause of death. The present report evaluates the relationship between serum cholesterol levels and total and cause-specific mortality in Pima Indians.


*    Methods
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 *Methods
down arrowResults
 down arrowDiscussion
 down arrowReferences
 
Since 1965, NIDDK has conducted a longitudinal health survey in the Gila River Indian community in Arizona,17 where the inhabitants are primarily Pima, Maricopa, or Tohono O'odham Indians. All community residents >=5 years old are invited to participate in biennial research examinations, and {approx}88% of the population has participated.18 At each examination, the subject is asked about smoking habits, and height and weight are measured with the subject wearing light clothing and no shoes. BMI is calculated as weight (kg) divided by the square of height (m2). Blood pressures are measured to the nearest 2 mm Hg with the subject in the supine position. Hypertension is defined by a systolic blood pressure >=140 mm Hg or a diastolic pressure >=90 mm Hg and/or by current use of antihypertensive medicines. Venous blood samples are collected after an overnight fast. Accuracy of the cholesterol assay19 was monitored and verified by the Centers for Disease Control and Prevention Laboratory Program Office or the College of American Pathologists Survey Program. A 75-g oral glucose tolerance test is administered, and type II diabetes is diagnosed if the 2-hour glucose concentration is >=11.1 mmol/L (200 mg/dL)20 or if diabetes is documented in the course of routine medical care.18

All Pima Indians examined at least once after the age of 20 years were included in the present analysis. Subjects were followed from this first examination until death or December 31, 1990. The study was approved by the institutional review board of the NIDDK and by the Tribal Council of the Gila River Indian Community.

Outcomes
The vital status of all subjects was ascertained as of the close of the study. Records of deaths for community residents are maintained on a continuing basis. Copies of death certificates are obtained from the Arizona Department of Vital Statistics or from agencies in other states for deaths that occur outside of Arizona. Cause of death was determined from the underlying cause listed on the death certificate and, in 686 cases occurring between 1975 and 1990, was recoded according to a review of clinical and autopsy records.16 Deaths occurring through 1974 were classified according to the eighth revision of the ICD (ICD-8), and those from 1975 on were classified according to ICD-9. Deaths were divided into traumatic causes (ICD codes E 800 to 999, including accidents, suicides, and homicides) and nontraumatic causes (ICD codes 000 to 799). Nontraumatic causes were further classified as cardiovascular diseases, including ischemic heart diseases, hypertension, and stroke (ICD codes 400 to 459); cancer (ICD codes 140 to 208); chronic liver disease or alcoholism (ICD-8 codes 291, 303, 571; ICD-9 codes 291, 303, 571 to 572); diabetes or nephropathy (ICD codes 250, 580 to 587); infectious diseases (ICD-8 and -9 codes 000 to 139, 320 to 326, 460 to 466, 480 to 487, 540 to 543, 572, 590, 599, 680 to 686; ICD-8 codes 470 to 474, 720, 732; ICD-9 codes 728, 729, 730); and miscellaneous causes not specified above. For 22 deaths (2%), death certificates could not be obtained, and these cases were classified as miscellaneous causes of death.

Statistical Methods
Characteristics at the baseline examination were compared over quartiles of serum cholesterol with a {chi}2 test for categorical variables and ANOVA for continuous variables. Three variables were used to assess the effect of cholesterol levels: interval cholesterol, baseline cholesterol, and change in cholesterol (Table 1Down). Mortality rates by strata of age, sex, diabetes, and quartiles of interval cholesterol or change in cholesterol were calculated as deaths per 1000 person-years and standardized by the direct method for age and sex to the 1980 US population. The significance of the difference in mortality rates between quartiles was determined by a {chi}2 test for stratified incidence data.21 The Mantel-Haenszel procedure was used to control for age and sex.22 The hypothesis of a difference between the strata was evaluated by the general association test (P), and the hypothesis of a linear trend over strata was evaluated with the Mantel extension correlation test (Ptrend).23 Mortality rate ratios by quartiles of baseline cholesterol compared with the lowest quartile were calculated by use of Cox's proportional-hazards regression models.24 For most of the analyses, the covariates age, 2-hour glucose, and systolic blood pressure violated the proportionality assumption.25 Proportional-hazards models, therefore, were stratified by age groups (<30, 30 to 40, 40 to 50, 50 to 60, and >=60 years) and by quartiles of 2-hour glucose and systolic blood pressure. The continuous variable and the quadratic term for BMI were both introduced into the models because of a U-shaped relationship with mortality. Linear and nonlinear relationships were tested by the likelihood ratio test,26 with cholesterol as a continuous variable (Ptrend) and its quadratic term (P). The interaction between baseline cholesterol and follow-up time was analyzed by use of baseline cholesterol, the follow-up time as continuous, and the interaction term as time dependent.


View this table:
[in this window]
[in a new window]
  		Table 1. Variables Used to Assess the Effect of Cholesterol Levels on Mortality


*    Results
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 up arrowMethods
 *Results
down arrowDiscussion
 down arrowReferences
 
During a mean follow-up of 12.8 years (range, 0.04 to 25.8 years), 1077 deaths occurred among 4553 subjects (24%). Among the categories used in this analysis, trauma was the most and cancer the least common cause of death (Table 2Down). At baseline, age, male sex, hypertension, and diabetes were positively associated with serum cholesterol levels and BMI was negatively associated with cholesterol (Table 3Down).


View this table:
[in this window]
[in a new window]
  		Table 2. Causes of Death by Sex, and Mean±SD for Age at Death and Follow-up


View this table:
[in this window]
[in a new window]
  		Table 3. Characteristics of 4553 Subjects by Cholesterol Quartiles at the First Examination

Age- and Sex-Standardized Mortality Rates
The age- and sex-standardized all-cause mortality was significantly associated with interval cholesterol (P<.001), and the relationship was U-shaped, with higher mortality rates in subjects in the lowest and highest cholesterol quartiles (Fig 1Down). When analyzed according to specific causes of death (Fig 2Down), age- and sex-standardized mortality rates from cardiovascular diseases and diabetes or nephropathy were positively related to cholesterol (Ptrend=.001 for each), whereas mortality rates from cancer and alcohol-related diseases were negatively related (Ptrend<=.05 for each). Note that "P" refers to a test of general association with three degrees of freedom (for four quartiles of cholesterol) and "Ptrend" refers to a one-degree-of-freedom test of linear trend (see "Methods"). The relationship with mortality rates from infectious diseases showed a U shape (P=.002), with the highest death rates among subjects in the lowest cholesterol quartile. No significant relationship was found for trauma (P=.15). Results were similar when controlled for the presence of diabetes (data not shown).



View larger version (22K):
[in this window]
[in a new window]
  		Figure 1. Age-standardized all-cause mortality rates by sex and cholesterol quartile measured at 2-year intervals. PY indicates person-years.



View larger version (26K):
[in this window]
[in a new window]
  		Figure 2. Age- and sex-standardized mortality rates by cause of death and cholesterol quartile measured at 2-year intervals. PY indicates person-years.

There was no significant relationship between change in cholesterol between examinations and age- and sex- standardized all-cause mortality (P=.30) (Fig 3Down). When analyzed according to specific causes of death, the relationship between changes in cholesterol and mortality rates was significant for diabetes or nephropathy (Ptrend=.03), with the highest rates among subjects with the greatest increase in cholesterol (Fig 4Down). There were no significant associations between changes in cholesterol and mortality rates from cardiovascular diseases (P=.21), cancer (P=.68), alcohol-related diseases (P=.44), infectious diseases (P=.88), or trauma (P=.72).



View larger version (20K):
[in this window]
[in a new window]
  		Figure 3. Age- and sex-standardized all-cause mortality rates by sex and quartile of cholesterol change between interval examinations. PY indicates person-years.



View larger version (24K):
[in this window]
[in a new window]
  		Figure 4. Age- and sex-standardized mortality rates by cause of death and quartile of cholesterol change between biennial examinations. PY indicates person-years.

In subjects who had a cholesterol increase, the age- and sex-standardized all-cause mortality was significantly associated with interval cholesterol (P=.003) and the relationship was U-shaped, and in those who had a cholesterol decrease there was a negative association with mortality (Ptrend=.04). The age- and sex-standardized mortality rates from cardiovascular diseases and diabetes or nephropathy were positively related to cholesterol in subjects who had a cholesterol increase (Ptrend<=.007 for each) but not in those with a cholesterol decrease, whereas mortality from cancer was negatively related only in subjects who had a cholesterol decrease (Ptrend=.002). The relationship with mortality rates from alcohol was negative in subjects who had a cholesterol increase (Ptrend=.004). No significant relationship was found for infectious diseases and trauma.

Mortality Rate Ratios
Mortality rate ratios from Cox's proportional-hazards regression models in 4257 subjects with complete covariate data at baseline are shown in Table 4Down by cholesterol quartiles at baseline, controlled for the potential confounding factors age, sex, BMI, 2-hour glucose, and systolic blood pressure. Baseline cholesterol had a significant U-shaped relationship with all-cause mortality and diabetes or nephropathy, and the relationship was positive with mortality from cardiovascular diseases (including stroke) and with stroke alone. No significant relationship was found for infectious diseases, cancer, alcohol-related diseases, and trauma. Smoking did not predict any cause of death in Pima Indians, and controlling for smoking did not modify the results (data not shown). The association between low cholesterol and high mortality rate ratios from cancer was reduced with longer follow-up, whereas the association between low cholesterol and deaths from infectious diseases became strongest: the interaction between follow-up duration and baseline cholesterol was negative for deaths from infectious diseases (P=.02) and positive for deaths from cancer (P=.02).


View this table:
[in this window]
[in a new window]
  		Table 4. Mortality Rate Ratios for Three Baseline Cholesterol Quartiles vs Lowest Quartile, by Cause of Death, Controlled for Age, Sex, BMI, 2-Hour Glucose, and Systolic Blood Pressure: Cox Proportional Regression Hazards Models in 4257 Subjects

Analyses were performed separately in 982 subjects with diabetes (432 deaths) and 3275 subjects without diabetes (530 deaths) at baseline. Results were similar, except for infectious diseases and cardiovascular diseases. The relationship between baseline cholesterol and cardiovascular diseases was significant and positive in diabetic subjects (Ptrend=.03) but not in nondiabetic subjects (P=.37). The relationship between baseline cholesterol and infectious diseases was positive in diabetic subjects (Ptrend=.002) and the interaction between follow-up duration and baseline cholesterol was significantly negative (P=.03), suggesting that the association between high cholesterol and high mortality rate ratios from infectious diseases was weaker among diabetic subjects with longer follow-up. The relationship was U-shaped in nondiabetic subjects (P=.02), with no significant interaction with follow-up duration.

To analyze the potential effect of underlying disease on baseline cholesterol level, subjects were divided into two groups according to duration of follow-up. Among the 4257 subjects with covariate data and any length of follow-up, 255 deaths occurred within 5 years of baseline. Among the 3358 subjects followed >=5 years, there were an additional 707 deaths. The relationships between baseline cholesterol and all-cause mortality (P=.28 versus P=.002), diabetes or nephropathy (P=.49 versus Ptrend=.0003), infectious diseases (P=.83 versus Ptrend=.05), and cancer (P=.56 versus Ptrend=.01) were weaker in subjects followed >=5 years than in subjects followed <5 years (Fig 5Down). No significant relationship was found in subjects followed >=5 or <5 years for alcohol-related diseases and trauma. Baseline cholesterol had a stronger and positive relationship with cardiovascular diseases (for all subjects, Ptrend=.003 versus P=.63) and stroke (Ptrend=.01 versus P=.58) in subjects followed >=5 than <5 years, but was significant only after 5 years among diabetic subjects (Ptrend=.04 for cardiovascular diseases). No significant relationship was found among diabetic (P=.26) and nondiabetic (P=.06) subjects for infectious diseases after 5 years of follow-up.



View larger version (31K):
[in this window]
[in a new window]
  		Figure 5. Mortality rate ratios controlled for age, sex, BMI, 2-hour glucose, and systolic blood pressure by follow-up duration and cause of death.


*    Discussion
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 up arrowMethods
 up arrowResults
 *Discussion
down arrowReferences
 
The present results, summarized in Table 5Down, suggest that low serum cholesterol is not a causal factor for mortality from any of the causes, as described below.


View this table:
[in this window]
[in a new window]
  		Table 5. Relationship Between Cholesterol Measurements and Total and Specific Mortality: Summary

Mortality From Cardiovascular Diseases
The Pima Indians have lower cholesterol levels14 15 and mortality rates from cardiovascular diseases16 than other US populations. However, higher cholesterol levels measured at baseline or at 2-year intervals were significant risk factors for cardiovascular mortality when controlled for potential confounders, such as age, sex, BMI, 2-hour glucose, systolic blood pressure, and smoking. This effect was stronger in those followed for >=5 years, and a high cholesterol level was a risk factor only in subjects whose cholesterol increased. The relationship was weaker in nondiabetic subjects, perhaps because of a low number of cardiovascular deaths. Two other studies in populations with low cholesterol concentrations have reported disparate results: a strong positive relationship between baseline cholesterol and cardiovascular diseases in one27 and a nonsignificant relationship due to an inverse association between serum cholesterol and cerebral hemorrhage in the other.28 In the present study, a positive relationship was found between cholesterol and stroke from all causes that was stronger after 5 years of follow-up. The consistency and strength of the present results demonstrated that high and increasing cholesterol is a risk factor for cardiovascular deaths, at least among diabetic subjects, and that low cholesterol (<3.93 mmol/L) is not associated with increased death rates from cardiovascular diseases in Pima Indians.

Mortality From Diabetes or Nephropathy
Because diabetes is common in Pima Indians17 and is responsible for almost all deaths attributed to nephropathy in this population,29 deaths from diabetes and nephropathy were pooled. Higher mortality rate ratios and hazard rate ratios were observed in the highest cholesterol quartile, measured at baseline or at 2-year intervals, as was found in the Multiple Risk Factor Intervention study for deaths attributed to diabetes alone.30 However, the relationship between baseline cholesterol and mortality from diabetes/nephropathy was weaker when deaths occurring during the first 5 years of follow-up were excluded. Increasing cholesterol between 2-year examinations was related to higher mortality rates, whereas a high interval cholesterol was not a risk factor among persons whose cholesterol decreased. Thus, mortality from diabetes/nephropathy was not related to the cholesterol level per se but only to increasing cholesterol.

A high cholesterol concentration has been suggested to be a risk or an exacerbating factor for diabetic nephropathy,31 and a relationship between serum cholesterol and incidence of elevated urinary albumin excretion has been reported previously in Pima Indians with longer duration of diabetes.32 However, an increase in total cholesterol is reported when nephropathy (eg, nephrotic syndrome) occurs, although it is rather a decrease in HDL cholesterol that occurs during the development of diabetes.33 The weak relationships when baseline cholesterol is used or the first 5 years are excluded, the effect of cholesterol change, and the positive association between serum cholesterol and diabetes/nephropathy mortality may be explained by an increase in serum cholesterol due to worsening nephropathy rather than an effect of cholesterol on the development of nephropathy.

Mortality From Cancer
Cancer is not a leading cause of death in Pima Indians, although they have low cholesterol levels compared with the US population.14 15 However, the present findings are limited because of the small number of deaths from cancer. Mortality from cancer was negatively associated with interval cholesterol but was not significantly associated with baseline cholesterol after potential confounders were controlled for or the first 5 years of follow-up were excluded, both of which minimize the effect of an underlying disease on baseline cholesterol. The positive interaction between follow-up duration and cholesterol implies that the associations between low cholesterol and high mortality rate ratios from cancer are reduced in subjects with longer follow-up. Results from other studies are conflicting as to the nature of the relationship between cholesterol and cancer mortality.6 7 8 9 11 12 13 28 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 In two other populations with low cholesterol levels, the inverse association between baseline cholesterol and mortality was not significant27 or was significant only in men after adjustments for confounders.28 Change in serum cholesterol, which may emphasize the effect of underlying disease on cholesterol level, was not related to cancer mortality in our study, in contrast to the findings of two other studies,11 41 but a low interval cholesterol was a strong risk factor for cancer mortality only in subjects with decreasing cholesterol.

Although physiological hypotheses have been suggested to explain the association between a low cholesterol level and cancer,2 alcohol use, smoking, or dietary patterns may be responsible for both low cholesterol and cancer, as described in some studies,9 46 47 48 and a low cholesterol level may also reflect the influence of preclinical cancer on serum cholesterol. Because of the weak time-dependent effect of low cholesterol level on mortality rates and the strong effect in subjects with cholesterol decrease, our results favored the latter hypothesis.

Mortality From Alcohol-Related Diseases
Alcohol-related disease is a leading cause of death among Pima Indians, who have higher rates of heavy and binge drinking than reported in national surveys.15 There was a negative relationship between alcohol-related mortality and cholesterol measured at 2-year intervals but not measured at baseline or when controlled for other confounders. No effect of cholesterol change was found in our study, but a low cholesterol level was a risk factor only in subjects whose cholesterol increased. Other studies showed no association34 or a negative association9 27 30 between baseline cholesterol and liver cirrhosis, which in one study was weaker when the first years of follow-up were excluded.42

Liver production of cholesterol decreases in cirrhosis and serum cholesterol tends to decline.11 Moderate alcohol intake increases HDL cholesterol, decreases LDL cholesterol level, and may or may not increase total cholesterol level.49 Nevertheless, common nutritional deficiencies in heavy drinkers may lower serum cholesterol.50 Thus, the association between low cholesterol measured at 2-year intervals, which emphasizes a short-term effect of cholesterol, and high alcohol-related mortality could reflect the cholesterol-lowering effect of cirrhosis or the nutritional deficiencies associated with heavy drinking.

Mortality From Infectious Diseases
The U-shaped or nonsignificant association between cholesterol measured at 2-year intervals or at baseline when controlled for confounders and mortality from infectious diseases was due to a positive relationship in diabetic subjects and to a U-shaped relationship in nondiabetic subjects. Change in cholesterol was not associated with mortality. In diabetic subjects, the effect of baseline cholesterol was weaker when the first 5 years were excluded to decrease the potential effects of an underlying disease, and the negative interaction between cholesterol and follow-up duration suggests also that the association between high cholesterol and high mortality rate ratios from infectious diseases was weaker with longer follow-up. However, in nondiabetic subjects, excluding the first 5 years of follow-up slightly weakened the relationship, and the interaction with follow-up duration was not significant. In other studies, a negative relationship has been observed between respiratory diseases and cholesterol levels5 34 when the first 10 years were excluded, but this relationship was confounded by health and social status.46

Hypocholesterolemia may be due to poor nutrition and to increased plasma levels of the inflammatory cytokine interleukin-6 during infectious diseases.51 Poor glucose control may be responsible for a high cholesterol level and for some infectious diseases, but mainly during a short follow-up. Thus, in diabetic subjects, the positive association between high serum cholesterol and infectious diseases may be explained by changes in cholesterol due to an underlying disease that increases cholesterol (eg, diabetes or nephropathy) rather than by cholesterol being a risk factor for infectious diseases. In nondiabetic subjects, misclassifications with diseases or chronic infections associated with low cholesterol (eg, tuberculosis, alcohol-related diseases, cancer) may be responsible for the negative association between a low cholesterol and mortality from infectious diseases.

Mortality From Trauma
Depression predisposes to suicide and may lower serum cholesterol through poor diet, weight loss, or an effect of the cytokine interleukin-2, possibly mediated by melatonin.52 However, hypothetical explanations for a low serum cholesterol leading to violent death and mental disorders have also implicated modifications of serotonin metabolism.53 Because Pima Indians have low levels of cholesterol14 and trauma is the most frequent cause of death, one might have expected a negative relationship between cholesterol and mortality from trauma, but the relationship was not significant, even when cholesterol was measured at 2-year intervals or when cholesterol change, which emphasizes the short-term effect of cholesterol on mortality, was used. Other studies have reported a nonsignificant34 or a negative association.30 54 55 56 Two studies supported the hypothesis that the relationship is confounded by alcohol consumption.57 58 In fact, even the nonsignificant trend for a negative relationship between cholesterol and deaths from trauma in Pima Indians may be related to an excessive long-term alcohol intake. Thus, our data do not confirm a hypothetical link between low cholesterol and traumatic deaths.

Implications
Lowering cholesterol concentrations in populations as a whole is currently combined with a "high-risk" approach to decrease cardiovascular diseases. Nevertheless, cholesterol levels and cardiovascular diseases vary widely between populations, and the most desirable serum cholesterol concentration remains unknown. We studied a Native American Community with, on average, a low concentration of cholesterol14 and low mortality rates from cardiovascular diseases.16 The relationship between serum cholesterol level and mortality rate was U-shaped as a result of a U-shaped association with mortality from infectious diseases, a negative association with mortality from cancer and alcohol-related diseases, and a positive association with mortality from cardiovascular diseases and diabetes/nephropathy. No significant association was found with trauma, despite high death rates from this cause. The negative or U-shaped association of serum cholesterol with mortality from some diseases probably results from these diseases lowering cholesterol rather than from a low cholesterol level causing the diseases. This is suggested by weak relationships with infectious diseases and cancer, weaker relationships with baseline cholesterol or when the first 5 years of follow-up were excluded to decrease the potential effects of an underlying disease, a significant interaction with duration of follow-up and a stronger relationship with cancer in subjects with a decrease in cholesterol, and a negative relationship with alcohol-related diseases. Similar results have been reported in other populations with low cholesterol, such as Chinese27 and Japanese,28 in whom high cholesterol was still a risk factor for cardiovascular mortality. Perhaps more important than a low cholesterol level, a spontaneous decrease in serum cholesterol may suggest the presence of other chronic diseases or adverse behaviors, which themselves result in increased mortality, and it may therefore be a marker of a poor prognosis.


*    Selected Abbreviations and Acronyms
 
BMI = body mass index
ICD = International Classification of Diseases
NIDDK = National Institute of Diabetes and Digestive and Kidney Diseases


*    Acknowledgments
 
The authors would like to thank the members of the Gila River Community who participated in the study, Dr Peter Bennett for support and assistance, the staff of the Diabetes and Arthritis and Epidemiology Section, NIDDK, for their assistance, and Dr Barbara V. Howard for advice and reviewing the manuscript.

Received December 11, 1996; revision received March 18, 1997; accepted March 26, 1997.


*    References
up arrowTop
 up arrowAbstract
 up arrowIntroduction
 up arrowMethods
 up arrowResults
 up arrowDiscussion
 *References
 

  1. Gould AL, Rossouw JE, Santanello NC, Heyse JF, Furberg CD. Cholesterol reduction yields clinical benefit: a new look at old data. Circulation. 1995;91:2274-2282.[Abstract/Free Full Text]
  2. Jacobs D, Blackburn H, Higgins M, Reed Dwayne, Iso H, McMillan G, Neaton J, Nelson J, Potter J, Rifkind B, Rossouw J, Shekelle R, Yusuf S. Report of the conference on low blood cholesterol: mortality associations. Circulation. 1992;86:1046-1060.[Abstract/Free Full Text]
  3. Lewis B, Tikkanen MJ. Low blood total cholesterol and mortality: causality, consequence and confounders. Am J Cardiol. 1994;73:80-85.[Medline] [Order article via Infotrieve]
  4. Meilahn EN, Ferrell RE. `Naturally occurring' low blood cholesterol and excess mortality. Coron Artery Dis. 1993;4:843-853.[Medline] [Order article via Infotrieve]
  5. Davey Smith G, Shipley MJ, Marmot MG, Rose G. Plasma cholesterol concentration and mortality: the Whitehall Study. JAMA. 1992;267:70-76.[Abstract]
  6. Dyer AR, Stamler J, Paul O, Shekelle RB, Schoenberger JA, Berkson DM, Lepper M, Colette P, Shekelle S, Lindberg HA. Serum cholesterol and risk of death from cancer and other causes in three Chicago epidemiological studies. J Chronic Dis. 1981;34:249-260.[Medline] [Order article via Infotrieve]
  7. Kromhout D, Bosschieter EB, Drijver M, de Lezenne Coulander C. Serum cholesterol and 25-year incidence of and mortality from myocardial infarction and cancer: the Zuphten Study. Arch Intern Med. 1988;148:1051-1055.[Abstract]
  8. Pekkanen J, Nissinen A, Punsar S, Karvonen MJ. Short- and long-term association of serum cholesterol with mortality: the 25-year follow-up of the Finnish Cohorts of the Seven Countries Study. Am J Epidemiol. 1992;135:1251-1258.[Abstract/Free Full Text]
  9. Wannamethee G, Shaper AG, Whincup PH, Walker M. Low serum total cholesterol concentrations and mortality in middle aged British men. BMJ. 1995;311:409-413.[Abstract/Free Full Text]
  10. Harris T, Feldman JJ, Kleinman JC, Ettinger WH, Makuc DM, Schatzkin AG. The low cholesterol-mortality association in a national cohort. J Clin Epidemiol. 1992;45:595-601.[Medline] [Order article via Infotrieve]
  11. Sherwin RW, Wentworth DN, Cutler JA, Hulley SB, Kuller LH, Stamler J. Serum cholesterol levels and cancer mortality in 361,662 men screened for the Multiple Risk Factor Intervention Trial. JAMA. 1987;257:943-948.[Abstract]
  12. Rose G, Shipley MJ. Plasma lipids and mortality: a source of error. Lancet. 1980;8:523-526.
  13. Stemmermann GN, Chyou PH, Kagan A, Nomura AMY, Yano K. Serum cholesterol and mortality among Japanese-American men: the Honolulu (Hawaii) Heart Program. Arch Intern Med. 1991;151:969-972.[Abstract]
  14. Howard BV, Davis MP, Pettitt DJ, Knowler WC, Bennett PH. Plasma and lipoprotein cholesterol and triglyceride concentrations in the Pima Indians: distributions differing from those of Caucasians. Circulation. 1983;68:714-724.[Abstract/Free Full Text]
  15. Welty TK, Lee ET, Yeh J, Cowan LD, Go O, Fabsitz RR, Le NA, Oopik AJ, Robbins DC, Howard BV. Cardiovascular disease risk factors among American Indians: the Strong Heart Study. Am J Epidemiol. 1995;142:269-287.[Abstract/Free Full Text]
  16. Sievers ML, Nelson RG, Bennett PH. Adverse mortality experience of a southwestern American Indian community: overall death rates and underlying causes of death in Pima Indians. J Clin Epidemiol. 1990;43:1231-1242.[Medline] [Order article via Infotrieve]
  17. Bennett PH, Burch TA, Miller M. Diabetes mellitus in American (Pima) Indians. Lancet. 1971;2:125-128.[Medline] [Order article via Infotrieve]
  18. Knowler WC, Bennett PH, Hamman RF, Miller M. Diabetes incidence and prevalence in Pima Indians: a 19-fold greater incidence than in Rochester, Minnesota. Am J Epidemiol. 1978;108:497-505.[Abstract/Free Full Text]
  19. Levine J, Zak B. Total serum cholesterol in serum. Clin Chim Acta. 1964;10:381-384.[Medline] [Order article via Infotrieve]
  20. World Health Organization Study Group on Diabetes Mellitus. Diabetes Mellitus. Geneva, Switzerland: World Health Organization; WHO Technical Reports No. 127:1985.
  21. Rothman KJ, Boice JC. Follow-up (cohort) studies. In: Epidemiologic Analysis With a Programmable Calculator. Washington, DC: National Institutes of Health (NIH publication 79-1649) 1979;11-17.
  22. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst. 1959;22:719-748.
  23. Mantel N. Chi-square tests with one degree of freedom: extensions of the Mantel-Haenszel procedure. J Am Stat Assoc. 1963;58:690-700.
  24. Cox DR. Regression models and life tables. J R Stat Soc B. 1972;34:187-220.
  25. Kalbfleisch JD, Prentice RL. Likelihood construction and further results on the proportional hazards model. In: The Statistical Analysis of Failure Time Data. New York, NY: John Wiley & Sons; 1980:119-142.
  26. Breslow NE, Day NE. Unconditional logistic regression for large strata. In: Statistical Methods in Cancer Research Vol I: The Analysis of Case-Control Studies. Lyon, France: International Agency for Research on Cancer, IARC publication 32; 1980:191-246.
  27. Chen Z, Peto R, Collins R, MacMahon S, Lu J, Li W. Serum cholesterol concentration and coronary heart disease in population with low cholesterol concentrations. BMJ. 1991;303:276-282.
  28. Iso H, Naito Y, Kitamura A, Sato S, Kiyama M, Takayama Y, Iida M, Shimamoto T, Sankai T, Komachi Y. Serum total cholesterol and mortality in a Japanese population. J Clin Epidemiol. 1994;47:961-969.[Medline] [Order article via Infotrieve]
  29. Nelson RG, Pettitt DJ, Carraher MJ, Baird HR, Knowler WC. Effect of proteinuria on mortality in NIDDM. Diabetes. 1988;37:1499-1504.[Abstract]
  30. Neaton JD, Blackburn H, Jacobs D, Kuller L, Lee DJ, Sherwin R, Shih J, Stamler J, Wentworth D. Serum cholesterol level and mortality findings for men screened in the Multiple Risk Factor Intervention Trial. Arch Intern Med. 1992;152:1490-1500.[Abstract]
  31. Diamond JR. The role of cholesterol in glomerular injury. Contemporary Issues in Nephrology. 2nd ed. 1991:109-126.
  32. Nelson RG, Knowler WC, Pettitt DJ, Hanson RL, Bennett PH. Incidence and determinants of elevated urinary albumin excretion in Pima Indians with NIDDM. Diabetes Care. 1995;18:182-187.[Abstract]
  33. Reaven GM. Lipid disorders in patients with NIDDM. Symposium. IDF Bull. 1995;35:11-14.
  34. Kozarevic D, McGee D, Vojvodic N, Gordon N, Racic Z, Zukel W, Dawber T. Serum cholesterol and mortality: the Yugoslavia Cardiovascular Disease Study. Am J Epidemiol. 1981;114:21-28.[Abstract/Free Full Text]
  35. Salonen JT. Risk of cancer and death in relation to serum cholesterol: a longitudinal study in an eastern Finnish population with high overall cholesterol level. Am J Epidemiol. 1982;116:622-630.[Abstract/Free Full Text]
  36. Yaari S, Goldbourt U, Even-Zohar S, Neufeld HN. Associations of serum high density lipoprotein and total cholesterol with total, cardiovascular, and cancer mortality in a 7-year prospective study of 10 000 men. Lancet. 1981;30:1011-1015.
  37. Cowan LD, O'Connell DL, Criqui MH, Barrett-Connor E, Bush TL, Wallace RB. Cancer mortality and lipid and lipoprotein levels: the lipid research clinics program mortality follow-up study. Am J Epidemiol. 1990;131:468-482.[Abstract/Free Full Text]
  38. Kark JD, Smith AH, Hames CG. The relationship of serum cholesterol to the incidence of cancer in Evans County, Georgia. J Chronic Dis. 1980;33:311-322.[Medline] [Order article via Infotrieve]
  39. Törnberg SA, Holm LE, Carstensen JM, Eklund GA. Risks of cancer of the colon and rectum in relation to serum cholesterol and beta-lipoprotein. N Engl J Med. 1986;315:1629-1633.[Abstract]
  40. Kagan A, McGee DL, Yano K, Rhoads GG, Nomura A. Serum cholesterol and mortality in a Japanese-American population: the Honolulu Heart Program. Am J Epidemiol. 1981;114:11-20.[Abstract/Free Full Text]
  41. Cambien F, Ducimetière P, Richard J. Total serum cholesterol and cancer mortality in a middle-aged male population. Am J Epidemiol. 1980;112:388-394.[Abstract/Free Full Text]
  42. Frank JW, Reed DM, Grove JS, Benfante R. Will lowering population levels of serum cholesterol affect total mortality? Expectations from the Honolulu Heart Program. J Clin Epidemiol. 1992;4:333-346.
  43. Schuit AJ, Van Dijk CEMJ, Dekker JM, Schouten EG, Kok FJ. Inverse association between serum total cholesterol and cancer mortality in Dutch civil servants. Am J Epidemiol. 1993;137:966-976.[Abstract/Free Full Text]
  44. Isles CG, Hole DJ, Gillis CR, Hawthorne VM, Lever AF. Plasma cholesterol, coronary heart disease, and cancer in the Renfrew and Paisley survey. BMJ. 1989;298:920-924.
  45. Beaglehole R, Foulkes MA, Prior IAM, Eyles EF. Cholesterol and mortality in New Zealand Maoris. BMJ. 1980;280:285-287.
  46. D'Agostino RB, Belanger AJ, Kannel WB, Higgins M. Role of smoking in the U-shaped relation of cholesterol to mortality in men: the Framingham Study. Am J Epidemiol. 1995;141:822-827.[Abstract/Free Full Text]
  47. Iribarren C, Reed DM, Burchfiel CM, Dwyer JH. Serum total cholesterol and mortality: confounding factors and risk modification in Japanese-American men. JAMA. 1995;273:1926-1932.[Abstract]
  48. Garcia-Palmieri MR, Sorlie PD, Costas R, Havlik RJ. An apparent inverse relationship between serum cholesterol and cancer mortality in Puerto Rico. Am J Epidemiol. 1981;114:29-40.[Abstract/Free Full Text]
  49. Castelli WP, Doyle JT, Gordon T, Hames CG, Hjortland MC, Hulley SB, Kagan A, Zukel WJ. Alcohol and blood lipids: the Co-operative Lipoprotein Phenotyping Study. Lancet. 1977;2:153-155.[Medline] [Order article via Infotrieve]
  50. Muldoon MF, Rossouw JE, Manuck SB, Glueck CJ, Kaplan JR, Kaufmann PG. Low and lowered cholesterol and risk of death from suicide and trauma. Metabolism. 1993;42:45-56.[Medline] [Order article via Infotrieve]
  51. Alvarez C, Ramos A. Lipids, lipoproteins, and apoproteins in serum during infection. Clin Chem. 1986;32:142-145.[Abstract/Free Full Text]
  52. Penttinen J. Hypothesis: low serum cholesterol, suicide, and interleukine-2. Am J Epidemiol. 1995;141:716-718.[Abstract/Free Full Text]
  53. Engelberg H. Low serum cholesterol and suicide. Lancet. 1992;339:727-729.[Medline] [Order article via Infotrieve]
  54. Schuit AJ, Dekker JM, Schouten EG, Kok FJ. Low serum cholesterol and death due to accidents, violence, or suicide. Lancet. 1993;341:827.
  55. Lindberg G, Rastam L, Gullberg B, Eklund GA. Low serum cholesterol concentration and short term mortality from injuries in men and women. BMJ. 1992;305:277-279.
  56. Zureik M, Courbon D, Ducimetière P. Serum cholesterol concentration and death from suicide in men: Paris Prospective Study I. BMJ. 1996;313:649-651.[Abstract/Free Full Text]
  57. Iribarren C, Reed DM, Wergowske G, Burchfiel CM, Dwyer JH. Serum cholesterol level and mortality due to suicide and trauma in the Honolulu Heart Program. Arch Intern Med. 1995;155:695-700.[Abstract]
  58. Vartiainen E, Puska P, Pekkanen J, Tuomiletho J, Lonnqvist J, Ehnholm C. Serum cholesterol concentration and mortality from accidents, suicide, and other violent causes. BMJ. 1994;309:445-447.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 J. Gerontol. A Biol. Sci. Med. Sci.Home page
G. Zuliani, A. Cherubini, A. R. Atti, A. Ble, C. Vavalle, F. Di Todaro, C. Benedetti, S. Volpato, M. G. Marinescu, U. Senin, et al.
Low Cholesterol Levels Are Associated With Short-Term Mortality in Older Patients With Ischemic Stroke
J. Gerontol. A Biol. Sci. Med. Sci., March 1, 2004; 59(3): M293 - M297.
[Abstract] [Full Text] [PDF]

Home page
 Arterioscler. Thromb. Vasc. Bio.Home page
B. V. Howard, D. C. Robbins, M. L. Sievers, E. T. Lee, D. Rhoades, R. B. Devereux, L. D. Cowan, R. S. Gray, T. K. Welty, O. T. Go, et al.
LDL Cholesterol as a Strong Predictor of Coronary Heart Disease in Diabetic Individuals With Insulin Resistance and Low LDL : The Strong Heart Study
Arterioscler. Thromb. Vasc. Biol., March 1, 2000; 20(3): 830 - 835.
[Abstract] [Full Text] [PDF]

Home page
 CirculationHome page
M. L. Sievers, P. H. Bennett, J. Roumain, and R. G. Nelson
Effect of Hypertension on Mortality in Pima Indians
Circulation, July 6, 1999; 100(1): 33 - 40.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fagot-Campagna, A.
Right arrow Articles by Knowler, W. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fagot-Campagna, A.
Right arrow Articles by Knowler, W. C.

Circulation Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search
Copyright © 1997 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited.