Atrial fibrillation: a major contributor to stroke in the elderly. The Framingham Study

Chronic atrial fibrillation without valvular disease has been associated with increased stroke incidence. The impact of atrial fibrillation on the risk of stroke with increasing age was examined in 5184 men and women in the Framingham Heart Study. After 30 years of follow-up, chronic atrial fibrillation appeared in 303 persons. Age-specific incidence rates steadily increased from 0.2 per 1000 for ages 30 to 39 years to 39.0 per 1000 for ages 80 to 89 years. The proportion of strokes associated with this arrhythmia was 14.7%, 68 of the total 462 initial strokes, increasing steadily with age from 6.7% for ages 50 to 59 years to 36.2% for ages 80 to 89 years. In contrast to the impact of cardiac failure, coronary heart disease, and hypertension, which declined with age, atrial fibrillation was a significant contributor to stroke at all ages.

Postchallenge glucose concentration and coronary heart disease in men of Japanese ancestry. Honolulu Heart Program

Since 1965, the Honolulu Heart Program has followed 8006 men of Japanese ancestry, aged 45-70 yr at study entry, for the development of cardiovascular disease. To investigate the role of glucose concentration 1 h after a 50-g challenge on the risk of fatal coronary heart disease (CHD) and nonfatal myocardial infarction (MI), 6394 nondiabetic men were followed for 12 yr for the first development of CHD. The rate of fatal CHD increased linearly with amount of glucose. Men in the fourth quintile of postchallenge glucose (157-189 mg/dl) had twice the age-adjusted risk of fatal CHD of those in the lowest quintile (P less than .05). Relative risk increased to threefold among those in the top quintile and remained statistically significant after adjustment for other risk factors including body mass, total cholesterol, hypertension, left ventricular hypertrophy, and hematocrit (P less than .001). When glucose was considered as a linear term in the proportional hazards model, a highly significant relation was noted with fatal CHD alone and when combined with nonfatal MI (P less than .001). We conclude that a continuously increasing risk gradient exists between postchallenge glucose and subsequent CHD that is independent of other known risk factors.

Central obesity and coronary heart disease in men

The relation between central body fat distribution, determined by measurement of subscapular skinfold thickness (SSF), and the development of definite coronary heart disease (CHD) was examined after 12 years of follow-up in a cohort of 7692 men who participated in the Honolulu Heart Program. The risk of incident coronary events was directly related to SSF. Compared with men in the lowest tertile of SSF, those in the middle tertile experienced a 70% excess of definite CHD. For those in the highest tertile, the excess more than doubled. For a given level of body mass index (BMI), SSF remains a significant and independent predictor of definite CHD, even after adjustment for age, total cholesterol, glucose, triglycerides, hypertensive status, and cigarette smoking. In contrast, the independent effect of BMI was not significant after adjustment for SSF. Thus centrally obese individuals are at increased risk of CHD, independent of BMI.

Diabetes and the risk of stroke. The Honolulu Heart Program

From 1965 to 1968, the Honolulu Heart Program began following up a cohort of men in a prospective study of cardiovascular disease. For this report, we examined the 12-year risk of stroke in 690 diabetic and 6908 nondiabetic subjects free of coronary heart disease and a history of stroke at study entry. In 12 years of follow-up, 62.3 per 1000 diabetic men and 32.7 per 1000 nondiabetic men experienced a stroke. The relative risk of thromboembolic stroke for those with diabetes compared with those without diabetes was 2.0 (95% confidence limits, 1.4 to 3.0). Although diabetes was usually associated with an atherogenic risk profile, control of hypertension, complicating myocardial infarction, and other risk factors failed to diminish the effect of diabetes on stroke. Among those without diabetes, the relative risk of thromboembolic stroke for those at the 80th percentile of serum glucose level compared with those at the 20th percentile (199 vs 115 mg/dL [11.0 vs 6.4 mmol/L]) was 1.4 (95% confidence limits, 1.1 to 1.8). In the nondiabetic sample, the relative risk of thromboembolic stroke for those with glucosuria compared with those without glucosuria was 2.7 (95% confidence limits, 1.6 to 4.5). There was no association between diabetes, or measures of glucose intolerance, and hemorrhagic stroke. We conclude that diabetes, even in a possibly undiagnosed subset of hyperglycemic individuals, imparts an additional independent risk of stroke unexplained by clinically measured risk factors.

Incidence of coronary heart disease and lipoprotein cholesterol levels. The Framingham Study

The first report from the Framingham Study that demonstrated an inverse relationship between high-density lipoprotein cholesterol (HDL-C) and the incidence of coronary heart disease (CHD) was based on four years of surveillance. These participants, aged 49 to 82 years, have now been followed up for 12 years, and this report shows that the relationship between the fasting HDL-C level and subsequent incidence of CHD does not diminish appreciably with time. Since a second measurement of HDL-C is available eight years after the initial determination, the relationship of HDL-C measurements on the same subjects at two points in time is examined. This second HDL-C measurement is also used in a multivariate model that includes cigarette smoking, relative weight, alcohol consumption, casual blood glucose, total cholesterol, and blood pressure. It is concluded that even after these adjustments, nonfasting HDL-C and total cholesterol levels are related to development of CHD in both men and women aged 49 years and older. Study participants at the 80th percentile of HDL-C were found to have half the risk of CHD developing when compared with subjects at the 20th percentile of HDL-C.

Risk of stroke in male cigarette smokers

From 1965 to 1968, the Honolulu Heart Program began following 8006 men of Japanese ancestry in a prospective study of cardiovascular disease. Of the subjects who had not had a stroke by the time of study entry, 3435 were cigarette smokers and 4437 were nonsmokers. In 12 years of follow-up, 171 smokers and 117 nonsmokers had a stroke. As compared with nonsmokers, cigarette smokers had two to three times the risk of thromboembolic or hemorrhagic stroke, after control for age, diastolic blood pressure, coronary heart disease, and other risk factors (P less than 0.001). Subjects who continued to smoke in the course of follow-up had the highest risk of stroke. When these subjects were compared with those who never smoked, their risk of hemorrhagic events was increased four- to six-fold (P less than 0.001). Subjects who were smokers at study entry but stopped smoking in the course of follow-up had a slight excess risk of stroke. When these subjects were compared with those who continued to smoke, however, their risk was reduced by more than half after adjustment for risk factors (P less than 0.05), indicating that stopping smoking had significant benefits.

Alcohol and hemorrhagic stroke. The Honolulu Heart Program

Since 1965, the Honolulu Heart Program has followed up 8,006 men in a prospective study of cardiovascular disease. Of those subjects free of stroke at the time of study entry, 2,916 were classified as nondrinkers of alcohol and 4,962 as drinkers. In 12 years of follow-up, 197 drinkers and 93 nondrinkers experienced a stroke. No significant relationships were noted between alcohol and thromboembolic stroke. When compared with nondrinkers, however, the risk of hemorrhagic stroke more than doubled for light drinkers and nearly tripled for those considered to be heavy drinkers. These findings are statistically significant and independent of hypertensive status and other risk factors. Results further indicate that alcohol has a greater effect on hemorrhagic strokes that are subarachnoid in origin, conferring a threefold to fourfold increased risk for moderate and heavy drinkers compared with nondrinkers.

Conditional regression models for transient state survival analysis

Survival models are important tools for the analysis of data when a disease event occurs in time and subjects are lost to follow-up. Many models, however, can also be adapted for use when an event is characterized by transitions through intermediate states of disease with increasing severity. In this presentation, such adaptations will be demonstrated for a class of conditional regression models for the analysis of transient state events occurring among grouped event times. The type of conditioning that will be described is useful in providing comparisons of specific disease states and an assessment of transition-dependent risk factor effects. An example will be given based on the Framingham Heart Study.

A prognostic comparison of asymptomatic left ventricular hypertrophy and unrecognized myocardial infarction: the Framingham Study

In 30 years of follow-up in the Framingham study, routine biennial ECG examinations revealed 315 subjects with ECG-LVH and 164 with unrecognized ECG-MI without previous cardiac explanation. Among subjects initially free of clinically evident coronary heart disease and both ECG abnormalities, the incidence of ECG-LVH was about double that of ECG-MI. Both events exhibited a male predominance and hypertensive subjects were more vulnerable to each. In subjects with asymptomatic ECG-LVH and ECG-MI, the 10-year, age-adjusted incidence of clinical coronary heart disease was greater than the rate experienced by the general Framingham sample. Rates for ECG-LVH were almost as large as those for ECG-MI. Cardiac failure and stroke also occurred more frequently among subjects with either ECG abnormality, and rates for ECG-LVH exceeded those for ECG-MI. Death from coronary heart disease, and sudden death in particular, was also increased two- to fourfold with similar risks for ECG-LVH and ECG-MI. ECG-LVH carried a significantly greater risk than ECG-MI for cardiovascular deaths in women. These findings suggest that ECG-LVH and ECG-MI are similar subclinical events with respect to predisposing characteristics and prognosis for subsequent overt cardiovascular disease including clinical manifestations of coronary heart disease.

Characteristics and prognosis of lone atrial fibrillation. 30-year follow-up in the Framingham Study

In 30 years of follow-up of 5,209 participants in the Framingham Study, 193 men and 183 women developed atrial fibrillation (AF). Among this group, "lone" AF occurred in 32 men and 11 women free of coronary heart disease, congestive heart failure, rheumatic heart disease, and hypertensive cardiovascular disease. To determine the characteristics and prognosis of lone AF, each case was matched to controls in the remaining Framingham sample. Comparisons indicated that levels of several risk factors associated with coronary heart disease were similar between the two groups. Atrial fibrillation cases, however, had significantly higher rates of preexisting nonspecific T- or ST-wave abnormalities and intraventricular block as determined by electrocardiograms. Follow-up for new cardiovascular events indicated similar rates of coronary heart disease and congestive heart failure, but the rate of strokes was significantly greater in the lone AF group. Findings suggest that subjects with lone AF, despite similar cardiovascular risk profiles to normal controls, have a distinct preponderance of preexisting electrocardiographic abnormalities. Furthermore, contrary to general belief, lone AF is not a benign condition; it has a serious prognosis, indicating a greater need for detection and treatment.

On the use of the logistic model in prospective studies

We describe the usual statistical concepts and consequent appropriate simulations of a prospective study for the simple case of a single risk variable and an assumed logistic model. We examine the simulations of Lilienfeld and Pyne, and show that they are seriously flawed. Contrary to those authors' claims, the estimates of parameters by the Walker-Duncan technique are both accurate and reliable.

Logistic regression in survival analysis

Logistic regression has been applied to numerous investigations that examine the relationship between risk factors and various disease events. Recently, the ability to consider the time element of event occurrences by proportional hazards models has meant that logistic regression has played a less important role in the analysis of survival data. This paper, however, shows that when event times are grouped into intervals, logistic regression can be adapted to the analysis of such data by modeling the interval when an event occurs. Furthermore, it is shown that results from such an adaptation will often lead to parameter estimates close to those obtained by the proportional hazards model in the grouped event time setting. An illustration of the application of logistic regression to survival analysis is based on data from the Framingham Heart Study.

Incidence and prognosis of unrecognized myocardial infarction. An update on the Framingham study

Of 708 myocardial infarctions among 5127 participants in the Framingham Study, more than 25 per cent were discovered only through the appearance of new diagnostic evidence during routine biennial electrocardiographic examinations. Of these unrecognized infarctions almost half were "silent," and the others caused atypical symptoms. The proportion of all infarcts that were unrecognized was higher in women and in older men. Such infarcts were uncommon in persons with angina. Unrecognized infarctions were as likely as recognized ones to cause death, heart failure, or strokes. Recurrent infarctions were more common in women with recognized than with unrecognized infarcts, but this difference was not present in men. Recurrent infarctions were more likely to be recognized than were first infarctions. We conclude that unrecognized infarctions are common and have as serious a prognosis as recognized infarctions.

Interpreting multiple logistic regression coefficients in prospective observational studies

Multiple logistic models are frequently used in observational studies to assess the contribution of a risk factor to disease while controlling for one or more covariates. Often, the covariates are correlated with the risk factor, resulting in multiple logistic coefficients that are difficult to interpret. This paper highlights the problem of assessing the magnitude of a multiple logistic coefficient and proposes a supplemental procedure to the usual logistic analysis for describing the relationship between a risk factor and disease. An example is given, along with results that are not apparent when the multiple logistic coefficient is considered alone. Conclusions that are presented are important in biologic studies if describing the effect of a risk factor is influenced by correlation with a covariate.

Factors associated with lipoprotein cholesterol levels. The Framingham study

Lipoprotein cholesterol determinations in 6328 individuals aged 20 to 79 years in the Framingham Heart Study Cohort and Offspring were related by multiple regression analysis to reported cigarette smoking, alcohol intake, and Quetelet Index (wt/ht2). Cigarette smoking was found to be strongly associated with "atherogenic" lipoprotein cholesterol profiles in young adults, and particularly in women. The associations for alcohol intake were mostly uniform across age groups and lipoprotein cholesterol fractions, while coefficients for Quetelet Index varied considerably. Blood chemistry associations were studied in participants under 50 years, after controlling for smoking, Quetelet Index, and alcohol intake. Significant associations for high density lipoprotein cholesterol (HDL-C) were seen for both sexes with alkaline phosphatase, serum calcium, serum u ric acid, and leucocyte count. Low density lipoprotein cholesterol (LDL-C) associations were observed with calcium, hematocrit, lactate dehydrogenase, and leucocyte count in men and women, while very low density lipoprotein cholesterol (VLDL-C) associations occurred with alkaline phosphatase, glucose, uric acid, and leucocyte count in both sexes. A 1 mg/dl higher calcium corresponded to an HDL-C approximately 4 mg/dl greater and a LDL-C typically 6 mg/dl greater after controlling for 12 other variables. A 1000/ml increase in leucocyte count was typically associated with a decrease in HDL-C by 1 mg/dl and an increase in LDL-C and VLDL-C of 1 mg/dl each.

Association of hyperestrogenemia and coronary heart disease in men in the Framingham cohort

The serum levels of estradiol and testosterone as well as established risk factors for coronary heart disease were estimated in 61 men (mean age 70.0 +/- 6.4 [SD] years) with coronary heart disease and in 61 matched control subjects enrolled in the Framingham Heart Study. The mean serum estradiol level was significantly higher in the subjects with coronary disease (p = 0.011). This difference in estradiol level increased with the exclusion of subjects older than 75 years (p less than 0.001). The mean serum testosterone level was not significantly different. None of the established risk factors for coronary heart disease was different between subjects with coronary disease and control subjects except blood glucose level, which was higher in the subjects with coronary disease (p = 0.025). We conclude that hyperestrogenemia is an important correlate of coronary heart disease in men.

Joint distribution of lipoprotein cholesterol classes. The Framingham study

During the period 1971 to 1975, the cholesterol content of lipoprotein density classes were measured in 2415 male and female Framingham Heart Study participants and 4342 of their offspring and offspring spouses who were free of coronary heart disease. In this first of three papers, the univariate and bivariate distributions of high density lipoprotein cholesterol, low density lipoprotein cholesterol, and very low density lipoprotein cholesterol are presented. There are differences in the levels of lipoprotein cholesterols by age, gender, and use of hormone preparations. The bivariate cross-tabulations of the three pairs of lipoprotein cholesterol measurements exhibit properties that cannot be appreciated from univariate summary statistics and suggest the importance of joint consideration of different lipoprotein cholesterol patterns for assessing the risk of coronary heart disease.

Alcohol and coronary heart disease in Puerto Rico

The relationship of alcohol consumption, measured in 1965-1968 by the 24-hour recall method, to incident coronary heart disease morbidity at eight years and mortality at 12 years of follow-up was investigated in a cohort of 9150 Puerto Rican males 35-79 years of age. After adjustment for age, cigarettes, exercise, urban/rural status, and income, there was clear evidence of a "U-shaped" relationship between alcohol and total mortality, and also the suggestion of a "J-shaped" relationship for angina pectoris, nonfatal myocardial infarction, and nonsudden coronary heart disease death. There was no association of alcohol with sudden cardiac death. The adjusted odds ratio for drinkers versus nondrinkers was 0.6 for angina pectoris (p less than 0.05), 0.7 for nonfatal myocardial infarction (p less than 0.1), and 0.7 for nonsudden coronary heart disease death (p less than 0.05). The inverse association of alcohol consumption to nonsudden coronary heart disease death was found to be dependent on age and income: the adjusted odds ratio for subjects under age 60 (0.4) or over the median income (0.3) strongly favored drinkers, while there was no evidence for a "protective effect" of alcohol consumption in the older and poorer segments of the population. It is concluded that available evidence does not justify the assertion that the inverse association between moderate alcohol intake and coronary heart disease risk is a causal one.

Summary estimates of cholesterol used to predict coronary heart disease

The relationships of total cholesterol and the proportion of cholesterol in individual lipoprotein classes to coronary heart disease are complex. To help simplify these relationships, cholesterol values are often combined into one summary estimate to form a single risk factor with a relationship to disease that is more easily described. Although summary estimates result in convenient expressions relating cholesterols to coronary heart disease, there is the potential for sacrificing information by ignoring the joint configuration of cholesterols that make up these estimates. We investigated the extent of this possibility for the ratio of total cholesterol to high-density lipoprotein cholesterol and the ratio of low-density lipoprotein cholesterol to high-density lipoprotein cholesterol. The findings suggest that the summary estimates are useful expressions for combining cholesterol information and are strong predictors of coronary heart disease. Clinicians who choose to use a summary estimate for screening purposes should recognize that a single ratio estimate is not always as informative as the joint configuration of the cholesterols that make up the estimate. This possibility is most clearly exhibited for the ratio of low-density lipoprotein cholesterol to high-density lipoprotein cholesterol, and it may become more apparent in future studies as the capabilities of exploring lipoprotein cholesterol relationships improve.

A blood ordering strategy for hospital blood banks derived from a computer simulation

The ultimate goal of a hospital blood bank inventory control programs is to reduce wastage of blood products and unnecessary use of laboratory services without jeopardizing patient safety. The development of a practical blood ordering policy at the hospital level is an integral part of any such program. In order to explore various blood ordering options in detail, a computer simulation of a hospital blood inventory was used to assess the impact on blood band performance measures of reductions in group O and non-group-O levels from baseline levels, assuming both a 21-day and 35-day shelf life. On the basis of data derived from this study showing that such inventory reductions accompanied by partial protection of the group O inventory will not result in significant shortages, a practical strategy was developed for establishing optimal target inventory levels for a hospital on an empirical basis. These target levels can serve as a guide for subsequent blood ordering. A step-by-step approach for analyzing a hospital blood inventory control program is then suggested, accompanied by an action plan for implementing change which incorporates the experimentally-derived blood ordering strategy. Adherence to this plan should result in a low outdate rate, a reduction in unnecessary cross-matching, and greater availability of blood for those patients with a legitimate need for it.

Epidemiologic features of chronic atrial fibrillation: the Framingham study

In the Framingham Study 2325 men and 2866 women 30 to 62 years old at entry were followed biennially over 22 years for the development of chronic atrial fibrillation in relation to antecedent cardiovascular disease and risk factors. During surveillance, atrial fibrillation developed in 49 men and 49 women. The incidence rose sharply with age but did not differ significantly between the sexes. Overall, there was a 2.0 per cent chance that the disorder would develop in two decades. Atrial fibrillation usually followed the development of overt cardiovascular disease. Only 18 men and 12 women (31 per cent) had chronic atrial fibrillation in the absence of cardiovascular disease. Cardiac failure and rheumatic heart disease were the most powerful predictive precursors, with relative risks in excess of sixfold. Hypertensive cardiovascular disease was the most common antecedent disease, largely because of its frequency in the general population. Among the risk factors for cardiovascular disease, diabetes and electrocardiographic evidence of left ventricular hypertrophy were related to the occurrence of atrial fibrillation. The development of chronic atrial fibrillation was associated with a doubling of overall mortality and of mortality from cardiovascular disease.

Estimation of very-low-density lipoprotein cholesterol from data on triglyceride concentration in plasma

We examined the relationship in a large-living population between the concentration of triglycerides in plasma from fasting persons and the concentration of very-low-density lipoprotein cholesterol (VLDL-C) as determined after preparative ultracentrifugation. Data from individuals in the Framingham Heart Study who were free of coronary heart disease, did not have Fredrickson Type I, III, or V hyperlipoproteinemia, and whose plasma-triglyceride values were less than 4000 mg/L were considered in the analyses. A total of 4100 Framingham Offspring, ages 20-49 years, met these criteria, as did 2284 participants in the original Framingham Cohort, aged 50-79 years. We estimated a linear regression model with an intercept and slope coefficient, using VLDL-C (units, mg/L) as the dependent variable and triglycerides as the dependent variable. The intercept was approximately 30.0 mg/L, and the slope coefficient for triglycerides was about 0.20 when all age and sex groups were combined. However, when the analysis was performed by age group, the slope coefficient for triglycerides increased with each age group only for women. The model, in which a different intercept and slope is used for each age and sex group was statistically better than a previously developed simpler model. On the other hand, the older model, in which VLDL-C is easily estimated by multiplying the triglyceride value by 0.20, is reasonably accurate and is very simple to use. The generalizability of the model is discussed in the light of results from other population studies.

Estimation of very-low-density lipoprotein cholesterol from data on triglyceride concentration in plasma

We examined the relationship in a large-living population between the concentration of triglycerides in plasma from fasting persons and the concentration of very-low-density lipoprotein cholesterol (VLDL-C) as determined after preparative ultracentrifugation. Data from individuals in the Framingham Heart Study who were free of coronary heart disease, did not have Fredrickson Type I, III, or V hyperlipoproteinemia, and whose plasma-triglyceride values were less than 4000 mg/L were considered in the analyses. A total of 4100 Framingham Offspring, ages 20-49 years, met these criteria, as did 2284 participants in the original Framingham Cohort, aged 50-79 years. We estimated a linear regression model with an intercept and slope coefficient, using VLDL-C (units, mg/L) as the dependent variable and triglycerides as the dependent variable. The intercept was approximately 30.0 mg/L, and the slope coefficient for triglycerides was about 0.20 when all age and sex groups were combined. However, when the analysis was performed by age group, the slope coefficient for triglycerides increased with each age group only for women. The model, in which a different intercept and slope is used for each age and sex group was statistically better than a previously developed simpler model. On the other hand, the older model, in which VLDL-C is easily estimated by multiplying the triglyceride value by 0.20, is reasonably accurate and is very simple to use. The generalizability of the model is discussed in the light of results from other population studies.

Influence of kaolin-pectin suspension on steady-state plasma digoxin levels

The effect of a kaolin-pectin antidiarrheal mixture on steady-state plasma levels of orally administered digoxin in subjects receiving chronic digoxin therapy was evaluated when the antidiarrheal and the cardiac glycoside were given concomitantly and when two doses of antidiarrheal were given, one 2 hours before and the other 2 hours after digoxin. Although simultaneous administration of both products decreased peak digoxin levels by 36 per cent, 24-hour areas under the curve were reduced by only 15 per cent, indicative of a slight decrease in digoxin bioavailability. In contrast, when their times of administration were separated by 2 hours, no evidence of a drug interaction was noted. Hence, the effect of one or two doses of kaolin-pectin suspension on steady-state plasma levels of digoxin appears inconsequential in patients on chronic digoxin therapy. Saliva levels were poorly correlated with plasma levels, presumably because of complexation in the oral cavity.

Design of a Keller-Plan Course in Elementary Statistics

The design of a Keller-plan course in statistics was discussed.

Tests of an alternative explanation of data supporting the R scale of the MMPI as a measure of acquiescence

An alternative hypothesis explaining the greater proportion of True responses given by Ss with low scores on the R scale of the MMPI when compared with Ss with high scores on the R scale was not supported in two experiments. The results supported the original hypothesis of Edwards and Abbott that the R scale is measuring in some degree the tendency to respond True to personality items and trait adjectives without regard to the substantive content of these items.

SOCIAL DESIRABILITY AND THE TSC SCALES: A REPLICATION AND REPLY TO STEIN

A replication of a study by Edwards, Klockars, and Abbott, 1970, using Tryon's sample, showed that each of the seven TSC scales had its highest loading on the first principal component. The results are in agreement with those obtained with a college sample.

SOCIAL DESIRABILITY AND THE TSC MMPI SCALES

The seven Minnesota Multiphasic Personality Inventory (MMPI) scales developed by Tryon, Stein, and Chu were intercorrelated and factor analyzed by the method of principal components. Included in the analysis were three marker scales for the first three principal components of the MMPI: the SD scale, the R scale, and the L scale, respectively. All of the TSC scales had loadings of .61 OT higher on the first principal component marked by the SD scale with a loading of -.90. The results show that there is con- siderable redundancy in the seven TSC scales and that each of lthe scales has its largest loading on the first principal component.

Further evidence regarding the R scale of the MMPI as a measure of acquiescence

High- and low-scoring groups on the R scale of the MMPI were selected For each group the mean probability of a True response, P(T), on 26 scales was obtained. The social desirability scale values of the items increased from scale to scale. On all 26 scales, low scorers on the R scale had a higher mean probability of a True response than high scorers. The regression lines of P(T) on SDSV for the two groups had approximately the same slopes and differed only in terms of their intercepts. The study was replicated with three additional samples and comparable results were obtained in each case.