Randomized controlled trial of a nonpharmacologic cholesterol reduction program at the worksite

PREVENTIVE

MEDICINE

12, 523-532 (1983)

Randomized Controlled Trial of a Nonpharmacologic Cholesterol Reduction Program at the Worksite’ CHARLES ARNOLD,? LINBANIA JACOBSON,+ MYRON WINICK,$ AND ERNST WYNDER?

ROBERT BRUNO,*.

‘Ayers1 Laborutories, 685 Third Avenue, New, York, New York 10017. fMuhoney Institute fog Health Muintenunce, Atnerican Heultil Fomdution, 320 E. 43rd Street. Net<, York, Ne\j, York 10017, and Slnstitrrte qf Human Nutrition, Colrrrnbiu Unillersity, New York. New* York 10027 Under experimental clinical conditions diet modification has been shown to reduce serum cholesterol levels. This paper reports such a positive response to a nonpharmacologic. behavioral education program at the worksite. Employees at the New York Telephone Company corporate headquarters were assigned randomly to treatment and control groups. Treatment consisted of an S-week group cholesterol reduction program conducted during employee lunch hours. It comprised a multiple-treatment approach-food behavior change techniques combined with nutrition education, physical activity planning, and self-management skills. The treatment group showed substantial change compared with the control group at the program’s completion. Those treated displayed a significant 6.4% reduction in total serum cholesterol (266 mg% average at baseline) as compared with control subjects with a corresponding decrease in high-density lipoprotein levels. A significant increase in nutrition knowledge and moderate weight loss were also documented for this group. The magnitudes of a participant’s baseline serum cholesterol level and his/her reduction in percentage of ideal body weight were positively and independently correlated with percentage changes in serum cholesterol levels. Over the same period. decreases in high-density lipoprotein levels and no changes in serum cholesterol, weight, and nutrition knowledge were observed for the control group. Overall, participants in the treatment program successfully reduced the coronary heart disease risk factors of elevated cholesterol and weight. Directions for future study are suggested.

INTRODUCTION

Little information is available on nutrition programs conducted with populations at the worksite. Most documented programs have had weight loss as their major objective (6). Anecdotal evidence is available on improved employee morale and well-being, but few objective evaluations have been reported (1, 4, 27). No randomized controlled study has been conducted in the worksite to test the effects of a behavioral nutrition education program on blood cholesterol levels. Inappropriate food patterns are generally recognized as major contributors to the risk characteristics of coronary heart disease (CHD). Important diet-related precursors to CHD are conditions such as hypercholesterolemia, hypertension, obesity, and impaired glucose tolerance. These conditions can be altered significantly through changes in individuals’ food selections and eating behaviors (32). Evidence is accumulating that Americans are making such changes in their food consumption patterns (22, 23, 29). A 1976 survey of occupants of 1,400 house’ Supported by a grant from the New York Heart Association and in-kind contributions from the New York Telephone Company’s Corporate Medical Department. ? To whom reprint requests should be addressed. 523 009l-7435/83 $3.00 Copyright AU rights

D 1983 by Academic Press, Inc. of reproduction in any form reserved.

524

BRUNO

ET AL.

holds found that 50% had modified their diets for health reasons, a quarter for preventive purposes (11). Decreases in community-based serum cholesterol estimates during this time interval paralleled the 25% decline in CHD mortality from 1968-1980 (2). USDA food disappearance data for this period are consistent with these changes in blood lipids (23). Evidence from epidemiologic, metabolic, and animal studies supports the link between elevated cholesterol levels and increased risk for CHD. Although a definitive clinical trial of the effects of diet modification on actual incidence of CHD has yet to be conducted (3), it is generally recognized that lower levels of blood cholesterol are associated with reduced risk of CHD and a favorable prognosis for development of other diseases (12). Further, the potential for reducing blood cholesterol by modifying one’s diet has been demonstrated (5, 16, 28). Behavioral techniques are currently used to facilitate such diet modification (8, 14, 24). For example, a group at the Diet Modification Clinics at Baylor College of Medicine (7) demonstrated that a behavioral program plus nutrition education was effective in reducing cholesterol levels in participants who were not clearly identified as “at risk” for CHD. Participants reduced their initial cholesterol levels by 8.5% in 6 months, maintaining a 3% decrease at a l-year follow-up. Similar effects on blood lipids of food behavior modification programs have been demonstrated in a variety of settings (e.g., hospital wards, ambulatory care clinics, and free-living communities) but information is lacking on such programs at the worksite. To investigate this issue, a study of a non-pharmacologic, cholesterol reduction program was conducted at the New York headquarters of a major corporationthe New York Telephone Company. The study’s objectives were to determine (a) the feasibility in the corporate setting of a cholesterol reduction program stressing changes in eating behaviors and (b) whether or not significant reductions of the primary CHD risk factor of elevated blood cholesterol levels and the associated CHD risk factor of overweight could be demonstrated. The short-term effects of program participation on nutrition knowledge, weight, and cholesterol status are presented. METHODS Subjects

Individuals were identified for the study through a review of employee annual physical exam records by the New York Telephone medical personnel. Information on demographic and CHD-related factors was reviewed. The following criteria were used to determine employee eligibility: serum cholesterol (2240 mg%, ~350 mg%), blood glucose (265 mg/dl, G 140 mg/dl), age (230 years), ideal body weight (<150%), blood pressure (cl60 mm Hg systolic, ~95 mm Hg diastolic), and absence of prior history of heart/neoplastic disease. A total of 290 potential study participants were identified who had had physical examinations within the previous 8 months. Of these, 50% expressed interest in participating in the cholesterol reduction study by responding positively to a written invitation from the

CHOLESTEROL

REDUCTION

PROGRAM AT THE WORKSITE

525

company medical department.3 The 145 employees were enrolled in the cholesterol reduction program and randomly assigned to one of two treatment groups or a control group. Treatment groups differed in the manner in which educational materials were presented; the quality and intensity of technical/behavioral information were equivalent in both groups. Data from treatment groups were combined for analysis. LogisticslStudy Organization The cooperation of the New York Telephone Company medical department was essential for planning and implementing the study. They provided access to physical examination records to identify those employees who were eligible for the study. They made frequent mailings and telephone calls to sc,hedule programatic and clinical assessment sessions, as a liaison between investigators and employees. In addition, they were intermediates between employees and supervisors when conflicts arose between job responsibilities and study sessions. Their involvement provided added credibility to the cholesterol reduction program. Treatment The aim of the cholesterol reduction program was to help participants lower their blood cholesterol levels and where appropriate to reduce their weight through diet modification. Specific dietary recommendations given in treatment groups approximated the U.S. Dietary Goals set forth by the Senate Select Committee on Nutrition and Human Needs (30). Namely, participants were advised to consume 30% or less of their calories as fats, 12% as proteins, and 55-68% as carbohydrates. Dietary recommendations were translated into guidelines for food selection, purchase, and preparation. Decreased consumption of dietary fat, cholesterol, simple sugars, and salt was emphasized. Participants’ weight and cholesterol reduction goals were determined on an individual basis, depending upon weight and cholesterol levels at the beginning of treatment. These goals were secondary to a participant’s primary objective of eating behavior change. Program participants were treated in groups of 10-15. A multiple-treatment approach was used to facilitate modification of food behavior. The approach included various environmental and self-management techniques-self-monitoring and record keeping (21), contingency contracting (25), stimulus control (13, 26), deconditioning or cognitive restructuring (IO), self-reinforcement (15), nutrition counseling, and management of physical activity. All groups were run by the same public health nutrition educator and scheduled for eight consecutive weekly sessions. These lasted approximately 60 min each, overlapping participant lunch hours. After successful completion of the formal g-week group program, participants progressed to a maintenance phase comprising six meetings spaced at approximately l-month intervals. Maintenance consisted of group sessions fo-

3 A profile of employees who declined participation or dropped out of treatment sessions, although desirable, is not available.

526

BRUNO

ET AL.

cusing on audiovisual presentations and group discussions reinforcing healthy eating/exercise behaviors. Participants assigned to control groups met periodically for the purpose of data collection. Minimal information was provided on cholesterol reduction with the promise that participants would be offered a formal cholesterol reduction program at the end of the study period. Data Collection

Data were collected on members of both treatment and control groups at the start of the program, 1 month after completion, and again at a follow-up session 6 months later. Instruments and techniques whose validity and reliability had been previously established were used. Accumulated CHD and nutrition knowledge was measured through a cognitive test designed and validated for the Chicago Heart Saver Program by the Nutrition Education Program (NEP). It was pretested on a nonrandom sample of New York Telephone employees to demonstrate its applicability to the study population. A 12-h fasting blood sample was obtained from each participant and analyzed by the commercial laboratory routinely used by New York Telephone for its employee medical services. Total cholesterol and high-density lipoprotein (HDL) levels were obtained from serum samples (ABA loo/A-GENT System, Abbot Labs.; NIH, 1974). Quality assessment statistics were accumulated for blood serum cholesterol determinations performed by the commercial laboratory. Coefficients of variation (i.e., 1.2-l .9%) computed by the Center for Disease Control laboratories on test samples analyzed during the months coinciding with our study demonstrate an acceptable margin of error. Percentage ideal body weight was determined through the use of “weight and height” tables (17) originally designed with data from the National Health Survey (1960-1962). RESULTS

AttendancelAttrition

Rate

Of the 97 persons (76 men and 21 women) assigned randomly to treatment groups, 68.0% or 66 persons attended at least one session. Of these, 80.3% or 53 persons substantially or completely finished treatment by attending at least four sessions. Conversely, 32.0% or 3 1 persons failed to attend any treatment sessions and were considered no shows. Drop outs, or those attending at least one session but fewer than four sessions, accounted for 19.7% (13 people) of the participant group. Outcome Variables

Data collected on participants at the start of the cholesterol reduction program (pre-treatment) and again 3 months later (1 month post-treatment) have been evaluated for presentation here. Distribution of dependent variables at baseline. Table 1 presents baseline levels, means, and standard deviations of pretest scores on the following measures: blood serum analysis for total cholesterol, HDL, HDL/total cholesterol ratio, percentage ideal body weight, and a cognitive test of cardiovascular nutrition knowl-

CHOLESTEROL

REDUCTION

527

PROGRAM AT THE WORKSITE

TABLE 1 DISTRIBUTION OF BASELINE TREATMENT-RELATED (DEPENDENT) VARIABLES

Control

Treatment Dependent variable

(n)

Mean

SD

(n)

Mean

SD

Significance level (P value)

Serum cholesterol (mg%) High-density lipoproteins (mg%) HDLkbolesteroI ratio Ideal body weight (%) Nutrition quiz score (base score = 60)

49

266

28

33

261

28

0.469

46

65

15

33

65

15

0.980

46 47

0.246 117

0.06 11

33 28

0.251 114

0.06 10

0.741 0.220

6.9

28

42.1

5.7

0.068

39

39.2

edge. Analysis of variance (ANOVA) showed no significant differences between the mean values of these variables for participants enrolled in treatment or control groups. Blood lipid levels. Table 2 presents the changes in dependent variables over time for participants of treatment and control groups. In the former group, a mean serum cholesterol value at baseline of 266 mg% was decreased significantly by 24 mg% with a standard deviation of +29 mg%. Over this same period a 7 mg% decrease (SD L 29) was insignificant for the control group. Analyzed another way as presented in Table 3, the relative reduction over time in serum cholesterol levels (i.e., post-cholesterol minus pre-cholesterol/pre-cholesterol times 100) for the 49 treatment participants was 8.8% (SD ? 10.8). This was statistically different at the 1% level from the 2.4% relative reduction (SD k 11.O) observed for the 33 members of the control group. High-density lipoprotein levels (Table 2) in the treatment group decreased on the average 10 units (SD k 11) from a pretreatment level of 65 mg% to a level of 55 mg% by 1 month after end of treatment (EOT). The change is significant at less than the 1% level. Within the control group, the initial mean HDL level of 65 mg% decreased significantly (P < 0.01) by 12 mg%. The standard deviation of this mean difference for 33 control group members was k 14 mg%. Associated with this decrease in HDL levels and an accompanying nonsigniticant decrease in serum cholesterol levels is a net decline in the ratio of HDL to total cholesterol. Expressed as a decimal, the initial HDLkholesterol ratio of 0.25 decreased a total of 0.04 (SD k 0.05) with the same sample of 33 individuals. Again, this mean difference was significant at less than 1%. Within the treatment group the 0.02 (SD k 0.06) decrease in HDLkholesterol ratio was not significant. Table 3 presents values from the T-test analysis used to compare the mean 10 mg% decrease (SD k 11) in HDL levels observed in the treatment group to the mean reduction of 12 mg% (SD t 14) occurring in the control group. Although both within-group reductions in pre- to post-HDL levels are significantly different from zero, they are not statistically different from each other. Percentage ideal body weight (%IBw). As shown in Table 2, the initial mean %IBW of II7 decreased by 2.4 percentage points (SD k 4.1) by the 1 month

(49)

(46)

Serum cholesterol (mg%)

High-density lipoproteins, HDL (mg%) HDLkholesterol ratio Percentage ideal body weight (%IBW) Nutrition information quiz

(39)

(47)

(46)

(n)

Dependent variable

PAIRED

Pre: Post: Pre: Post: Pre: Post:

Pre: Post: Pre: Post:

T-TEST

0.25 0.23 117.1 114.7 39.2 46.8

266 242 65 55

Mean

+ 7.6 (6.5)

-2.4 (4.1)

- 0.02 (0.06)

- 10 (11)

- 24 (29)

co.01

0.000

0.078

0.000

0.000

P value

DIFFERENCES

TABLE 2

Difference (*SD)

Treatment group

RESULTS/WITHIN-GROUP

(49)

(28)

(33)

(33)

(33)

(n)

Pre: Post: Pre: Post: Pre: Post:

0.25 0.21 113.9 115.0 42.1 42.1

261 254 65 53

0.0 (4.2)

+ 1.1 (3.0)

-0.04 (0.05)

- 12 (14)

- 7 (29)

Difference (LSD)

Control group Mean

VARIABLES

Pre: Post: Pre: Post:

OF DEPENDENT

>0.05

0.062



0.154

P value

F

3

2

E

CHOLESTEROL

REDUCTION

529

PROGRAM AT THE WORKSITE

TABLE 3 T-TEST RESULTS/COMPARISON OF CHANGES

IN DEPENDENT

VARIABLES

Dependent variable

Group

(4

Mean

(*SD)

Serum cholesterol (relative change %) High-density lipoproteins (mg%) Percentage ideal body weight (%IBW)

Treatment Control Treatment Control Treatment Control

(49) (33) (46) (33) (47)

-8.8 - 2.4 -11.9 -9.9 -2.4 1.1

(10.8) (11.0) (14.4) (10.6) (4.1) (3.0)

(28)

P value co.01 >0.05 0.01

EOT point for treated individuals. This result was significant at less than 1%. Individuals enrolled in the control group showed a nearly significant 1.1% increase (SD + 3.0) in %IBW over the same period. The mean reduction in %IBW for the treatment group was significantly greater than the mean increase of the controls (Table 3). A P value of less than 1% was found for the 47 and 28 cases in these respective groups. Nutrition knowledge. Table 2 presents the increase in the mean score on the CHD/Nutrition Knowledge Quiz for the 39 participants in treatment for whom both pre- and post-test scores are available. On a test where 60 is the maximum score, an initial group mean of 39 increased 8 points (SD 2 6) by 1 month EOT. This gain in knowledge was found to be significant at less than the 1% level. No difference was found over time in knowledge scores for the control group. Effects of weight change and baseline cholesterol on cholesterol reduction. Regression analysis including all of the dependent variables under study showed that a significant amount of the variability in the treatment group’s cholesterol reduction could be explained by changes in weight (F(1,43) = 32.91, P < 0.05) and baseline cholesterol levels (F( 1,43) = 5.04, P < 0.05). The stronger influence was suggested by changes in weight expressed as %IBW. Table 4 presents summaries of Pearson correlational analysis for weight and baseline cholesterol with serum cholesterol. Individual weight and %IBW at the start of intervention was not significantly correlated with an end-of-treatment change in weight (either calculated in pounds or adjusted for sex and height as %IBW). In addition, regardless of a participant’s initial weight as a percentage of the ideal, the reduction of %IBW and percentage loss of serum cholesterol was constant across the group by I month EOT. Although initial weight and %IBW showed no association with changes in cholesterol level, changes in either parameter were positively correlated with changes in serum cholesterol (P < 0.01). The greater the reduction in %IBW, the greater the percentage loss in serum cholesterol level. Likewise, the greater the absolute change in weight, measured in pounds, the greater the absolute milligram percentage change in serum cholesterol levels. DISCUSSION

Of primary importance, given random assignment of participants and the aim of the program, was the significant 8.8% decrease (vs 6.4% control) in serum

530

BRUNO ET AL. TABLE 4 PEARSON’S CORRELATION ANALYSIS FOR BODY WEIGHT, AND TOTAL SERUM CHOLESTEROL (BASELINE, ABSOLUTE, AND RELATIVE CHANGE)-TREATMENT SUBJECTSONLY

Variable pair Initial %IBW with A %IBW Initial weight (lb) with A weight (lb) Initial %IBW with percentage A cholesterol A %IBW with percentage A cholesterol A Weight (lb) with A cholesterol (mg%) Initial cholesterol with actual A cholesterol bg%) Initial cholesterol with percentage A cholesterol

Pearson’s r

tn)

P value

47

0.439

-0.01

47

0.480

-0.01

49

0.478

0.54

45

CO.001

0.52

45

CO.001

-0.44

49


-0.36

49

0.005

0.02

cholesterol levels on the part of the treated group. Although weight reduction was not a primary goal of the program, the 2.4% drop in %IBW was also significant for treated individuals. The amount of weight lost was not dependent upon a participant’s sex, number of treatment sessions attended, or weight at start of intervention. However, weight loss was accompanied by a reduction of serum cholesterol levels such that the more pounds lost, the greater the absolute milligram percentage reduction in a participant’s serum cholesterol. This relationship held true for relative changes in weight and cholesterol levels. The most significant contributions to decreases in cholesterol levels were made by the relative decrease in %IBW and a participant’s serum cholesterol level at the start of treatment. Each made significant independent contributions. It is generally recognized that the higher an individual’s baseline cholesterol, the greater the percentage decrease that can be expected through dietary intervention (18). Our results are consistent with this prediction, since the amount of cholesterol reduction was influenced by a participant’s cholesterol value at the start of the program. Although initial cholesterol was negatively correlated to both absolute and relative cholesterol change, cholesterol change was defined in such a way that a negative value indicated a reduction. Therefore, the negative correlation calculated in the study indicates that the higher a subject’s initial cholesterol, the greater his or her absolute and relative loss (mg% and %, respectively) by 1 month after EOT. The program therefore benefited those at the higher end of the serum cholesterol distribution who were most in need of blood lipid reduction. Another important barometer of the program’s success was attendance rate. Over 80% of those individuals who started treatment successfully completed it. The fact that 19.7% of the treatment population dropped out compares favorably with the attrition rate observed in Stunkard and Brownell’s study of a food behavior change program at the worksite (27). (That study, however, was a 16-week weight control program and the criterion used for determining drop-out status

CHOLESTEROL

REDUCTION

PROGRAM AT THE WORKSITE

531

was not given.) Attendance was not critical to an individual’s success in the program, since the number of treatment sessions attended was not directly correlated to the magnitude of change in serum cholesterol levels. Several caveats about the findings are in order. First, there are a number of analytical and physiological factors which may have had an effect upon and led to variability in serum cholesterol levels (9, 20, 21, 31). “Regression toward the mean” may have been responsible for part of the cholesterol reduction observed since single determinations were used to establish baseline and end of treatment blood lipid levels. It is assumed that the use of a controlled experimental design equally distributed these sources of variation over both treatment and control groups. Given this assumption, the residual mean decrease in serum cholesterol level of treated participants over those in the control group is 17 mg% or 6.4%. Second, it is noteworthy that the control group may have been contaminated. For ethical reasons, control subjects received the results of their preintervention blood tests coincident with the start of treatment sessions. No definite information on blood cholesterol lowering was given at that time and participant’s questions were deferred to their promised future treatment program. However, many persons later recounted with enthusiasm how they sought out and practiced methods of diet management for cholesterol control. Also, due to the proximity of work locations of treatment and control participants, the latter may have been exposed to the treatment program. This may have had an effect upon cholesterol and weight changes in the control group. Third, information on changes in participants’ eating behaviors is currently unavailable. Such information, although of interest, is secondary to the documentation of changes in the CHD risk factors (i.e., elevated cholesterol levels and body weight) within the scope of this study. In conclusion, the nonpharmacologic cholesterol reduction program was feasible at the worksite, increased participants’ nutrition knowledge, and produced significant healthful changes in blood cholesterol and body weight. Future research in this area could concentrate on (a) refinements of the behavioral program in the work setting, (b) the qualitative changes in eating behaviors which are correlated with cholesterol and weight reductions, and (c) accumulation of longterm follow-up data on program participants. ACKNOWLEDGMENTS Cooperation of the New York Telephone Company Corporate Medical Department-most notably Gilbert Callings, Jr., M.D., Corporate Medical Director, and Loring Wood, M.D., Medical Director, Research and Development-made this research possible. Likewise, the knowledge, assistance, and support of numerous people were essential in realizing this study. Thanks to George Banks, M.S.; Angelica Cantlon, R.D., M.S.; Marcia Levenstein, Ph.D.; Rebecca Mandriota, M.Ed., R.D.; My Lien Nguyen, Dr.PH.; Sheila Petty; June Thalheimer, R.N.; Majorie Young, R.N., M.S.; and Marsha Witten.

REFERENCES 1. Alderman, M. H., and Schoerbaum, E. E. Detection and treatment of hypertension at the worksite. New Engl. J. Med. 293, 65-58 (1975). 2. Beaglehole, R., LaRosa, J. C., Heiss, G., ef al. Serum cholesterol, diet, and the decline in coronary heart disease mortality. Prev. Med. 8, 538-547 (1979). 3. Carol, R. “Diet Modification: Can it Reduce the Risk of Heart Disease?” Nutrition Research Consultation for the American Council on Science and Health, New York, 1980.

532

BRUNO ET AL.

4. Cox, M., and Wear, R. F., Jr. Campbell Soup’s program to prevent atherosclerosis. Amer. J. Nurs. 72, 253-259 (1972).

5. Davis, C. E. Clinical trials of lipid lowering and coronary artery disease prevention, in “Hyperlipidemia, Diagnosis and Treatment” (R. 1. Levy et ul.. Eds.). Grune & Stratton, New York, 1977.

6. Foreyt, J. P., Scott, L. W., and Gotto. A. M. Weight control and nutrition education programs in occupational settings. Public He&h Rep. 95, 127-136 (1980). 7. Foreyt, J. P., Scott, L. W., Mitchell, R., et al. Plasma lipid changes in the normal population following behavioral treatment. J. Consult. C/in. Psycho/. 74, 440-452 (1979). 8. Gotto, A. M., Jr., Foreyt, J. P., and Scott, L. W. Hyperlipidemia and nutrition: Ongoing work. Atherosclerosis Rev. 7, (1980). 9. Hallberg, L., Hogdahl, A. M., Svanborg, A., and Vikrot, 0. Plasma lipids in women. Variation in cholesterol, phospholipids and triglycerides at different ages in a random population sample. Actu Med. Stand. 180, 697-707 (1966). 10. Homme, L. E. Perspectives in psychology. XXIV. Control of converants, the operants of the mind. Psycho/. Rec. 15, 501-51 I (1965). 1I. Jones, J. L. Are health concerns changing the American diet? Narl. Food Rev. (1977). 12. Lewis, B. Workshop report: Clinical-pathological section. Conference on health effects of blood lipids. Prev. Med. 8, 695-703 (1979). 13. Mahoney, M. J., and Mahoney, K. “Permanent Weight Control.” Morton, New York, 1976. 14. Meyer, A. J., and Henderson, J. B. Multiple risk factor reduction in the prevention of cardiovascular disease. Prev. Med. 3, 225-236 (1974). 15. Mojonnier, L. M.. Hall, Y., Berkson. D., et al. Experiences in changing food habits of hyperlipidemic men and women. Part I. Planning and testing the nutrition education project. In press. 16. Multiple Risk Factor Intervention Trial Group. The Multiple Risk Factor Intervention Trial (MRFIT). Ann. N.Y. Acud. Sci. 304, 293-308 (1978). 17. Multiple Risk Factor Intervention Trial Group. The Multiple Risk Factor Intervention Trial (MRFIT): A national study of primary prevention of coronary heart disease. JAMA 235, 825-827 (1976). 18. National Diet-Heart Study Research Group. The National Diet-Heart Study final report. Circulation 37(Suppl. I), l-428 (1968). 19. Richards. C. S. Assessment and behavior modification via self-monitoring: An overview and bibliography. JSAS Catalog of Selected Documents in Psychology 7, 15 (1977). 20. Schaefer, E. J., Levy, R. I., Ernst, N. D.. et al. Effects of cholesterol lowering diets on lipoprotein cholesterol levels: Abstracts. Circulation 62(Suppl. 3). 62 (1980). 21. Schwartz, M. K., and Hill, P. Problems in the interpretation of serum cholesterol values. Pm,. Med. 1, 167-177 (1972). 22. Stamler, J. Lifestyles, major risk factors, proof and public policy: George Lyman Duff Memorial Lecture. Circulution 58, 3-19 (1978). 23. Stamler, J. Population studies, in “Nutrition, Lipids and Coronary Heart Disease-A Global View.” (R. I. Levy, Ed.). Raven Press, New York, 1979. 24. Stern, M. P., Farquhar, J. W., Maccoby, N., and Russell, S. H. Results of a two-year health education campaign on dietary behavior: The Stanford Three Community Study. Circrdution 54, 826-833 (1976). 25. Stuart, R. B. Behavioral contracting within the families of delinquents. 1. BehaL,.

26. 27. 28. 29. 30. 31. 32.

Ther. Br Exp. Psychiutry 2, 1-l 1 (1971). Stuart, R. B., and Davis, B. “Slim Chance in a Fat World: Behavioral Control of Obesity.” Research Press, Champaign, Ill., 1972. Stunkard, A. J., and Brownell, K. E. Worksite treatment of obesity: Clinical and research reports. Amer. J. Psychiatry 137, 2 (1980). Tillotson, J. L. (Ed.). Lipids research clinics program, in “Proceedings of the Nutrition-Behavioral Research Conference” (N. D. Ernst, Ed.). DHEW Publ., 1976. U.S. Department of Commerce, Bureau of the Census. “Statistical Abstract of the United States,” Vol. 97. Washington, D.C.. 1975. U.S. Senate Select Committee on Nutrition and Human Needs. “Dietary Goals for the United States,” Vol. 2, p. 4. U.S. Govt. Printing Office, Washington, D.C., 1977. Vitamin E seen to improve H.D.L. cholesterol levels. Znf. Med., p. 14 (December 29. 1979). Wynder, E. L. Conference on the health effects of blood lipids: Optimal distributions for populations. Prev. Med. 8, 609-611 (1979).