The Finnish Diabetes Prevention Study (DPS) Lifestyle Intervention and 3-Year Results on Diet and Physical Activity

Epidemiology/Health Services/Psychosocial Research ORIGINAL ARTICLE

The Finnish Diabetes Prevention Study (DPS) Lifestyle intervention and 3-year results on diet and physical activity

1 1 JAANA LINDSTROM¨ , MSC JOHAN ERIKSSON, MD, PHD been reproduced by the Diabetes Preven- 2 2 ANNE LOUHERANTA, PHD MATTI UUSITUPA, MD, PHD tion Program (DPP), in which lifestyle in- 3 1,6 MARJO MANNELIN, MSC JAAKKO TUOMILEHTO, MD, PHD 4 tervention, with a similar 58% risk MERJA RASTAS, MSC FOR THE FINNISH DIABETES PREVENTION 5 reduction, was superior to the metformin VIRPI SALMINEN, MSC STUDY GROUP treatment (2). The lifestyle intervention used in the DPP was not designed to be used in community settings (3), whereas one of the main objectives in the DPS was OBJECTIVE — To describe the 1) lifestyle intervention used in the Finnish Diabetes Preven- to test an intervention feasible in primary tion Study, 2) short- and long-term changes in diet and exercise behavior, and 3) effect of the health care. intervention on glucose and lipid metabolism. In this study, we describe the lifestyle RESEARCH DESIGN AND METHODS — There were 522 middle-aged, overweight intervention program used in the DPS, subjects with impaired glucose tolerance who were randomized to either a usual care control the changes in dietary habits and exercise group or an intensive lifestyle intervention group. The control group received general dietary and behavior that were achieved during the exercise advice at baseline and had an annual physician’s examination. The subjects in the first year and the maintenance of these intervention group received additional individualized dietary counseling from a nutritionist. changes after 3 years, and assess the effi- They were also offered circuit-type resistance training sessions and advised to increase overall cacy of the intervention on body weight, physical activity. The intervention was the most intensive during the first year, followed by a plasma glucose, and lipids. maintenance period. The intervention goals were to reduce body weight, reduce dietary and saturated fat, and increase physical activity and dietary fiber. RESEARCH DESIGN AND RESULTS — The intervention group showed significantly greater improvement in each in- METHODS — The DPS was a multi- tervention goal. After 1 and 3 years, weight reductions were 4.5 and 3.5 kg in the intervention center study with five participating centers group and 1.0 and 0.9 kg in the control group, respectively. Measures of glycemia and lipemia improved more in the intervention group. in Helsinki, Kuopio, Turku, Tampere, and Oulu. The study protocol was approved by CONCLUSIONS — The intensive lifestyle intervention produced long-term beneficial the ethics committee of the National Public changes in diet, physical activity, and clinical and biochemical parameters and reduced diabetes Health Institute in Helsinki, Finland. Each risk. This type of intervention is a feasible option to prevent type 2 diabetes and should be study center employed a physician, study implemented in the primary health care system. nurse, and nutritionist (MSc in nutrition) on a part-time basis. Either an exercise in- Diabetes Care 26:3230–3236, 2003 structor/physiotherapist was a member of the study team or these services were pro- vided commercially. he Finnish Diabetes Prevention lifestyle intervention (1). The risk of dia- Study subjects were recruited mainly Study (DPS) was one of the first con- betes was reduced by 58% in the intensive by screening high-risk groups such as T trolled, randomized studies to show lifestyle intervention group compared first-degree relatives of type 2 diabetes pa- that type 2 diabetes is preventable with with the control group. These results have tients who voluntarily responded to local ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● advertisements or were identified in earlier epidemiological surveys. The inclu- From the 1Diabetes and Genetic Epidemiology Unit, Department of Epidemiology and Health Promotion, Na- 2 sion criteria were 1) age 40–64 years at tional Public Health Institute, Helsinki, Finland; the Department of Clinical Nutrition, University of Kuopio, Ͼ 2 Kuopio, Finland; the 3Department of Sport Medicine, Oulu Deaconess Institute, Oulu, Finland; the 4Research and screening, 2) BMI 25 kg/m at screen- Development Centre, Social Insurance Institution, Turku, Finland; the 5Institute of Nursing and Health Care, ing, and 3) the mean value of two 75-g Tampere, Finland; and the 6Department of Public Health, University of Helsinki, Helsinki, Finland. oral glucose tolerance tests (OGTTs) in Address correspondence and reprint requests to Jaana Lindstro¨m, National Public Health Institute, the impaired glucose tolerance range Mannerheimintie 166, 00300 Helsinki, Finland. E-mail: jaana.lindstrom@ktl.fi. based on World Health Organization cri- Received for publication 17 July 2003 and accepted in revised form 4 September 2003. Additional information for this article can be found in an online appendix at http://care. teria (4). The randomization of partici- diabetesjournals.org. pants (n ϭ 522) started in 1993 and Abbreviations: DPP, Diabetes Prevention Program; DPS, Diabetes Prevention Study; E%, proportion of continued until 1998. total energy; OGTT, oral glucose tolerance test; LTPA, leisure-time physical activity; VLCD, very-low-calorie At the first study visit after the screen- diet. A table elsewhere in this issue shows conventional and Syste´me International (SI) units and conversion ing phase, the subjects were randomly al- factors for many substances. located to the intervention group or the © 2003 by the American Diabetes Association. usual care control group. Randomization

3230 DIABETES CARE, VOLUME 26, NUMBER 12, DECEMBER 2003 Lindstro¨m and Associates was stratified by center, sex, and 2-h Intervention group cardiorespiratory fitness. Supervised, plasma glucose value (1,5). Dietary intervention. The participants progressive, individually tailored circuit- The study was prematurely termi- had face-to-face consultation sessions type moderate intensity resistance train- nated in March 2000 by an independent (from 30 min to 1 h) with the study nu- ing sessions to improve the functional end point committee, since the incidence tritionist at weeks 0, 1–2, and 5–6 and at capacity and strength of the large muscle of diabetes in the intervention group was months 3, 4, 6, and 9, i.e., altogether groups of the upper and lower body were highly significantly lower than in the con- seven sessions during the first year and also offered free of charge. As a means for trol group (1). However, the follow-up every 3 months thereafter. The first year improving motivation, an “exercise com- according to the original protocol was sessions had a preplanned topic (e.g., di- petition” between the five study centers continued until each subject’s following abetes risk factors, saturated fat, fiber, was organized twice during the study pe- annual visit. Therefore, the number of physical activity, and problem solving), riod. Voluntary group walking and hiking subjects who reached the 3-year visit in but the discussions were individualized, were also organized. this report is somewhat higher (n ϭ 434) focusing on specific individual problems. Printed material was used to illustrate the than the number (n ϭ 374) reported message and to serve as a reminder at earlier (1). Education program for the control home. In addition, there were voluntary group group sessions, expert lectures, low-fat At baseline, the control group was given cooking lessons, visits to local supermar- Background for lifestyle intervention general information about lifestyle and di- kets, and between-visit phone calls and abetes risk. This was done either individ- The main goals of the lifestyle interven- letters. The goal was to equip the subjects tion were based upon available evidence ually or in one group session (30 min to with necessary knowledge and skills and 1 h), and some printed material was de- on diabetes risk factors (6–9). They were to achieve gradual, permanent behavioral Ն livered. The message to reduce weight, in- weight reduction 5%, moderate inten- changes. A change in behavior was con- Ն crease physical activity, and make sity physical activity 30 min/day, di- sidered a process, as suggested by Ͻ qualitative changes in diet was the same as etary fat 30 proportion of total energy Prochaska (13). The dietary advice was (E%), saturated fat Ͻ10 E%, and fiber for the intervention group subjects, but based on 3-day food records, which were counseling was not individualized. Ն15 g/1,000 kcal. completed four times yearly. Nutrient in- The hypothesis of diabetes preven- takes were calculated, and a summary of tion by increased physical activity had al- the results was given and explained to the Baseline and annual measurements ready been tested in the Malmo¨ feasibility subjects. Subjects were encouraged to All study subjects had an annual OGTT, a study (10) with encouraging results. A make intermediate goals for themselves medical history, and a physical examina- nonrandomized pilot study testing the ef- by thinking about practical things they tion with measurements of height, fect of aerobic and resistance training ex- could try to change (e.g., instead of an weight, and waist circumference. These abstract goal such as “increase fiber in- ercise programs on insulin resistance in procedures have been described in detail subjects with impaired glucose tolerance take,” a practical goal would be “eat a slice of rye bread on every meal”). Weight was previously (1,5). Plasma, serum, or capil- was completed before the beginning of lary glucose was determined locally ac- the DPS (11). On the basis of this pilot measured at every visit, and a weight chart was drawn. The participants were cording to standard guidelines and study, resistance training was offered to corrected by linear regression equation the participants of the DPS intervention. also encouraged to measure and record their weight at home on a regular basis. using values measured in the National To further increase the knowledge of Public Health Institute central laboratory a suitable diet for long-term intervention, Recommended weight loss was not more than 0.5 to 1 kg per week. The spouse was (60–80 duplicate plasma samples per cen- a dietary pilot study was completed in a ter) as the dependent variable. The second subgroup of subjects before the start of invited to join the sessions, especially if he or she was the one responsible for shop- screening OGTT was considered the glu- the DPS study (12). The results suggested ping and cooking in the family. After 6 cose baseline. Diabetes diagnosis had to be that a monounsaturated fat–enriched diet months, the use of a very-low-calorie diet confirmed by a second OGTT and induced improves glucose metabolism when con- (VLCD) for 2–5 weeks or as a substitute termination of the study. Serum total cho- sumed after a diet rich in saturated fat com- for one to two meals per day was consid- lesterol, HDL cholesterol, and triglycerides pared with a reduced-fat, polyunsaturated ered, if preferred by the subject, to boost were determined using an enzymatic assay fat–enriched diet. Therefore, besides a method. HbA was analyzed using the Ͻ weight loss. Altogether, 48 subjects par- 1c moderate-fat ( 30 E%), low–saturated fat ticipated in the VLCD groups arranged as Bayer DCA2000 Analyzer. Ͻ ( 10 E%) diet, a diet with somewhat more part of the intensive intervention. Those study subjects who terminated monounsaturated fat (total fat not exceed- Exercise intervention. The subjects the follow-up prematurely were defined ing 35 E%) was also considered acceptable. were individually guided to increase their as dropouts. Nevertheless, data from their At the nutritional counseling ses- overall level of physical activity. This was earlier visits were included in the analy- sions, the intervention goals were trans- done by the nutritionist during the di- ses. There were 8 and 6 subjects who did lated into practice. General examples of etary counseling sessions and highlighted not attend the 1-year annual visit and an this translation are presented in an online by the study physicians at the annual vis- additional 9 and 9 who did not attend the appendix (available at http://care.diabetes its. Endurance exercise was recom- 3-year visit in the intervention and con- journals.org). mended to increase aerobic capacity and trol groups, respectively.

DIABETES CARE, VOLUME 26, NUMBER 12, DECEMBER 2003 3231 DPS lifestyle intervention

Table 1—Baseline characteristics for changes made to traditional recipes (e.g., the use of skim instead of whole Intervention group Control group milk). For this report, the most recent nutrient content dataset was used, and n (male/female) 265 (91/174) 257 (81/176) therefore the nutrient intake results are Age (years) 55 Ϯ 755Ϯ 7 slightly different from those reported pre- First-degree relatives with diabetes (%) 66 61 viously (1). Schooling in years (%) 0–94040Assessment of physical activity 10–12 27 27 The study subjects completed the vali- Ն13 33 33 dated Kuopio Ischaemic Heart Disease Type of work (%) Risk Factor Study 12-month Leisure- Agricultural or industrial 9 10 Time Physical Activity (LTPA) question- Office work or student 42 44 naire (16,17) at baseline and at every Homemaker, retired, or unemployed 49 49 annual visit. In this detailed quantitative Weight (kg) 86.7 Ϯ 14.0 85.5 Ϯ 14.4 questionnaire, subjects estimated the fre- BMI (kg/m2) 31.4 Ϯ 4.5 31.1 Ϯ 4.5 quency (times per month), duration Waist circumference (cm) 102.0 Ϯ 11.0 100.5 Ϯ 10.9 (minutes), and intensity (0, recreational Fasting plasma glucose (mmol/l) 6.1 Ϯ 0.8 6.2 Ϯ 0.7 and outdoor activities; 1, conditioning ex- 2-h plasma glucose (mmol/l) 8.9 Ϯ 1.5 8.9 Ϯ 1.5 ercise; 2, brisk conditioning exercise; and Ϯ Ϯ HbA1c (%) 5.7 0.6 5.6 0.6 3, competitive strenuous exercise) of their Serum total cholesterol (mmol/l) 5.6 Ϯ 1.0 5.6 Ϯ 0.9 most common lifestyle and structured Serum HDL cholesterol (mmol/l) 1.2 Ϯ 0.3 1.2 Ϯ 0.3 LTPA over the previous 12 months. Com- Serum total cholesterol–to–HDL cholesterol ratio 4.9 Ϯ 1.3 4.8 Ϯ 1.2 mon moderate- or high-intensity physical Serum triglycerides (mmol/l) 1.7 Ϯ 0.8 1.7 Ϯ 0.8 activity included brisk walking, skiing, Blood pressure–lowering medication (%) 29 31 jogging, swimming, bicycling, gymnas- Lipid-lowering medication (%) 5 7 tics, resistance training, and ball games (if Total LTPA (min/week) 339 (193–545) 329 (173–586) reported intensity Ն1); gardening and Moderate-to-vigorous LTPA (min/week) 156 (62–288) 169 (65–352) snow shoveling, hunting, picking berries, Energy (kcal/day) 1,771 Ϯ 520 1,744 Ϯ 527 and gathering mushrooms (if reported in- Carbohydrates tensity Ն2); fishing, hobby crafts, and re- E% 43.6 Ϯ 7.5 43.2 Ϯ 6.7 pairs and house work (if reported g 190 Ϯ 57 185 Ϯ 54 intensity ϭ 3); and rowing, forest work, Fat and wood cutting (if reported intensity E% 36.0 Ϯ 6.7 37.1 Ϯ 6.5 Ն1). The duration (minutes per week) of g72Ϯ 28 73 Ϯ 29 total physical activity and moderate- and Saturated fat (E%)* 16.2 Ϯ 4.0 17.0 Ϯ 4.3 high-intensity LTPA were calculated. Monounsaturated fat (E%) 12.9 Ϯ 2.8 13.0 Ϯ 2.9 Polyunsaturated fat (E%) 5.7 Ϯ 1.7 5.8 Ϯ 2.2 Statistical analysis Alcohol Differences between the intensive inter- E% 2.8 Ϯ 5.1 2.1 Ϯ 4.1 vention and control group were tested for g8Ϯ 17 6 Ϯ 12 statistical significance with Student’s t test Cholesterol (mg) 312 Ϯ 137 304 Ϯ 130 (approximately normally distributed vari- Fiber ables), Mann-Whitney nonparametric g20Ϯ 720Ϯ 8 test, ␹2 test, or ANCOVA adjusting for g/1,000 kcal 11.7 Ϯ 4.0 11.7 Ϯ 3.9 baseline value, using SAS software (SAS Water soluble (g) 4.6 Ϯ 1.7 4.6 Ϯ 1.6 Institute, Cary, NC) version 8.2. Water soluble (g/1,000 kcal) 2.7 Ϯ 1.0 2.7 Ϯ 0.9 Water insoluble (g) 14.4 Ϯ 5.4 14.1 Ϯ 5.9 RESULTS Water insoluble (g/1,000 kcal) 8.4 Ϯ 2.9 8.4 Ϯ 2.9 Data are mean Ϯ SD or median (interquartile range). *Difference between the groups (Student’s unpaired t Baseline results test), P ϭ 0.0188. At baseline, the study groups were similar in terms of characteristics reflecting increased diabetes risk (Table 1). Of all Assessment of dietary intake the amount of food consumed (14). The study subjects, 55% had a BMI Ͼ30 kg/ Study subjects were asked to complete a nutrient intakes were calculated using a m2. The median amount of at least mod- 3-day food record at baseline (before the dietary analysis program developed in the erate-intensity LTPA was 160 min/week. randomization visit) and before every an- National Public Health Institute (15). The In their spare time, 34% of the study sub- nual visit. A picture booklet of portion program allows modification of database jects reported that they mostly read, sizes of typical foods was used to estimate recipes, so that the subjects could account watched TV, and spent time in other ways

3232 DIABETES CARE, VOLUME 26, NUMBER 12, DECEMBER 2003 Lindstro¨m and Associates

Table 2—Changes in physical activity and nutrient intakes from baseline to years 1 and 3

From baseline to year 1 From baseline to year 3 Intervention Intervention group Control group P* group Control group P* n 256 250 231 203 Total LTPA (min/week) 16 (Ϫ126 to 115) 21 (Ϫ133 to 138) 0.9045 50 (Ϫ126 to 115) 23 (Ϫ142 to 171) 0.2415 Moderate-to-vigorous LTPA 49 (Ϫ41 to 140) 14 (Ϫ47 to 90) 0.0073 61 (Ϫ33 to 168) 6 (Ϫ91 to 104) 0.0057 (min/week) Energy (kcal/day) Ϫ247 Ϯ 438 Ϫ108 Ϯ 464 0.0001 Ϫ204 Ϯ 489 Ϫ97 Ϯ 458 0.0067 Carbohydrates E% 3.3 Ϯ 8.1 1.7 Ϯ 7.3 0.0023 3.3 Ϯ 8.0 2.0 Ϯ 7.6 0.0070 g Ϫ13 Ϯ 52 Ϫ4 Ϯ 48 0.0590 Ϫ11 Ϯ 57 Ϫ1 Ϯ 50 0.1042 Fat E% Ϫ3.4 Ϯ 8.2 Ϫ2.1 Ϯ 7.6 0.0002 Ϫ4.7 Ϯ 7.7 Ϫ3.2 Ϯ 7.5 <0.0001 g Ϫ16 Ϯ 25 Ϫ8 Ϯ 28 <0.0001 Ϫ16 Ϯ 27 Ϫ11 Ϯ 27 0.0003 Saturated fat (E%) Ϫ2.7 Ϯ 4.6 Ϫ1.2 Ϯ 5.1 <0.0001 Ϫ3.2 Ϯ 4.5 Ϫ1.9 Ϯ 4.9 <0.0001 Monounsaturated fat (E%) Ϫ0.8 Ϯ 3.8 Ϫ0.4 Ϯ 3.4 0.0257 Ϫ1.0 Ϯ 3.6 Ϫ0.6 Ϯ 3.5 0.0453 Polyunsaturated fat (E%) Ϫ0.0 Ϯ 2.1 Ϫ0.2 Ϯ 2.5 0.5020 0.0 Ϯ 2.4 Ϫ0.4 Ϯ 2.3 0.0872 Alcohol E% Ϫ1.1 Ϯ 4.4 Ϫ0.3 Ϯ 4.6 0.1452 Ϫ0.2 Ϯ 4.1 Ϫ0.2 Ϯ 4.3 0.6836 g Ϫ4 Ϯ 15 Ϫ1 Ϯ 13 0.1168 Ϫ0 Ϯ 14 Ϫ0 Ϯ 14 0.8580 Cholesterol (mg) Ϫ69 Ϯ 138 Ϫ28 Ϯ 148 0.0005 Ϫ63 Ϯ 167 Ϫ31 Ϯ 155 0.0586 Fiber g1Ϯ 70Ϯ 7 0.1146 1 Ϯ 81Ϯ 7 0.4393 g/1,000 kcal 2.5 Ϯ 4.6 0.7 Ϯ 4.1 <0.0001 2.4 Ϯ 4.7 1.1 Ϯ 4.1 0.0013 Water soluble (g) 0.2 Ϯ 1.7 0.5 Ϯ 1.5 0.2218 0.1 Ϯ 1.7 0.1 Ϯ 1.5 0.7275 Water soluble (g/1,000 kcal) 0.5 Ϯ 1.1 0.2 Ϯ 1.0 <0.0001 0.5 Ϯ 1.1 0.2 Ϯ 1.0 0.0050 Water insoluble (g) 0.6 Ϯ 5.5 0.0 Ϯ 5.5 0.1136 0.7 Ϯ 5.9 0.4 Ϯ 5.7 0.5608 Water insoluble (g/1,000 kcal) 1.7 Ϯ 3.3 0.5 Ϯ 3.0 <0.0001 1.7 Ϯ 3.4 0.9 Ϯ 3.1 0.0034 Data are means Ϯ SD or median (interquartile range). *ANCOVA adjusting for baseline value or Mann-Whitney test for two unpaired groups. Values in bold are statistically significant. that are not physically demanding; these intervention group compared with the between group assignment and sex (not individuals were considered sedentary control group during the first year (Table shown). (not shown). Dietary intake of both mac- 2). The absolute amounts of fat decreased, Clinical and metabolic parameters. ro- and micronutrients were similar be- and energy intake decreased in the inter- Several beneficial changes in clinical and tween the groups except for the energy vention group more than in the control metabolic parameters were observed in proportion of saturated fat, which was group. At 3 years the reductions in en- the intervention group compared with slightly higher in the control group. ergy, fat (E% and g), saturated fat (E%), the control group at year 1 and still at the and monounsaturated fat (E%), and the year 3 examination (Table 3). Mean 1- and 3-year results increase in carbohydrates (E%) and fiber weight reduction was 4.5 kg in the inter- Physical activity. The proportion of sed- density were still statistically significantly vention group and 1.0 kg in the control entary individuals was 14 and 30% at year greater in the intervention group. group at 1 year. Some regain of weight 1(P Ͻ 0.0001 for difference between The specific goals of the intervention appeared during the following 2 years. Af- groups) and 17 and 29% at year 3 (P ϭ were more often reached by the interven- ter exclusion of those intervention group 0.0028) in the intervention and control tion group than the control group sub- subjects who took part in the VLCD groups, respectively. However, the total jects. Of the intervention and control group (whose weight reduction was amount of reported time spent physically group subjects, 46 and 14% managed to 6.2 Ϯ 7.0 kg at year 1 and 4.8 Ϯ 7.2 kg at active did not change, but moderate-to- lose Ն5% of weight during the first year year 3), the mean weight reduction was vigorous LTPA increased in the interven- (P Ͻ 0.0001), the fat intake goal was 4.1 Ϯ 4.3 kg at year 1 and 3.2 Ϯ 4.5 kg at tion group compared with the control reached by 37 and 20% (P Ͻ 0.0001), the year 3 (P Ͻ 0.0001 compared with the group at years 1 and 3 (Table 2). saturated fat intake goal was reached by control group). Dietary intake. The E% of carbohy- 21 and 9% (P Ͻ 0.0001), and the fiber Significantly greater improvements drates increased; the E% of fat, saturated density goal was reached by 37 and 23% were seen at year 1 in fasting plasma glu- fat, and also monounsaturated fat de- (P Ͻ 0.0006), respectively. In general, cose (Ϫ0.2 vs. 0.0 mmol/l), 2-h plasma Ϫ Ϫ creased; the fiber density increased; and women tended to change their diet more glucose ( 0.9 vs. 0.3 mmol/l), HbA1c cholesterol intake decreased more in the than men, but there was no interaction (Ϫ0.1 vs. 0.1%), serum total cholesterol–

DIABETES CARE, VOLUME 26, NUMBER 12, DECEMBER 2003 3233 DPS lifestyle intervention

Table 3—Changes in clinical and metabolic characteristics from baseline to years 1 and 3

From baseline to year 1 From baseline to year 3 Intervention Intervention group Control group P* group Control group P* n 256 250 231 203 Weight kg Ϫ4.5 Ϯ 5.0 Ϫ1.0 Ϯ 3.7 <0.0001 Ϫ3.5 Ϯ 5.1 Ϫ0.9 Ϯ 5.4 <0.0001 % Ϫ5.1 Ϯ 5.3 Ϫ1.1 Ϯ 4.2 <0.0001 Ϫ4.0 Ϯ 5.8 Ϫ1.1 Ϯ 6.2 <0.0001 BMI (kg/m2) Ϫ1.6 Ϯ 1.8 Ϫ0.4 Ϯ 1.3 <0.0001 Ϫ1.3 Ϯ 1.9 Ϫ0.3 Ϯ 2.0 <0.0001 Waist circumference (cm) Ϫ4.4 Ϯ 5.2 Ϫ1.3 Ϯ 4.8 <0.0001 Ϫ3.3 Ϯ 5.7 Ϫ1.2 Ϯ 5.9 0.0005 Fasting plasma glucose (mmol/l) Ϫ0.2 Ϯ 0.7 0.0 Ϯ 0.7 <0.0001 Ϫ0.0 Ϯ 0.7 0.1 Ϯ 0.7 0.0664 2-h plasma glucose (mmol/l) Ϫ0.9 Ϯ 1.9 Ϫ0.3 Ϯ 2.2 0.001 Ϫ0.5 Ϯ 2.4 Ϫ0.1 Ϯ 2.2 0.0664 Ϫ Ϯ Ϯ Ϫ Ϯ Ϯ HbA1c (%) 0.1 0.7 0.1 0.6 0.0003 0.2 0.6 0.0 0.6 0.002 Serum total cholesterol (mmol/l) Ϫ0.1 Ϯ 0.7 Ϫ0.1 Ϯ 0.7 0.5097 Ϫ0.1 Ϯ 0.9 0.1 Ϯ 0.8 0.0712 Serum HDL cholesterol (mmol/l) 0.05 Ϯ 0.19 0.02 Ϯ 0.17 0.0681 0.14 Ϯ 0.20 0.11 Ϯ 0.19 0.1354 Serum total cholesterol–to–HDL cholesterol Ϫ0.4 Ϯ 0.8 Ϫ0.1 Ϯ 0.8 0.0011 Ϫ0.6 Ϯ 0.9 Ϫ0.3 Ϯ 0.8 0.0009 ratio Serum triglycerides (mmol/l) Ϫ0.2 Ϯ 0.6 Ϫ0.0 Ϯ 0.7 <0.0001 Ϫ0.1 Ϯ 0.6 Ϫ0.0 Ϯ 0.8 0.024 Data are means Ϯ SD. *ANCOVA adjusting for baseline value. Values in bold are statistically significant. to–HDL cholesterol ratio (Ϫ0.4 vs. biased due to the study design. Those tes (24,25). It can be speculated that small Ϫ0.1), and serum triglycerides (Ϫ0.2 vs. subjects who developed diabetes during negative energy balance sustained for a Ϫ0.0 mmol/l) in the intervention group the first 2 years of the trial did not have a lengthy time period could be more advan- compared with the control group. During 3-year examination; the majority be- tageous to glucose tolerance than similar the first 3 years of the study, 22 subjects longed to the control group. Further- weight loss achieved with strict, short- (9%) in the intervention group and 51 more, the control group subjects were term energy restriction. Therefore, in the (20%) in the control group developed di- actually given a “mini-intervention” and long run, a lifestyle-intervention ap- 2 abetes (P ϭ 0.0001, ␹ test). The use of therefore were not a true nontreatment proach to weight control rather than a VLCD preparations did not affect diabetes group. weight-reduction diet might be a more risk; 8% of the users developed diabetes The weight reduction during the cost-efficient way to manage overweight in 3 years. study was modest but comparable with in individuals at high risk for diabetes. that of other studies (10,19–21) on sub- We used one-to-one intervention to CONCLUSIONS — It is evident that jects with the metabolic syndrome, im- individualize the intervention and be- lifestyle intervention can prevent or at paired glucose tolerance, or type 2 cause the recruitment and screening of least postpone type 2 diabetes (1,2,18) diabetes and using lifestyle approach to study subjects was gradual. Group care and should therefore be implemented in weight reduction. In the DPP study, a may have some benefits in the form of primary health care. However, the life- highly intensive and therefore expensive social support and would probably be style interventions used have been either lifestyle intervention (3,22) produced a more cost-efficient in the clinical setting, insufficiently described (18) or not in- 5.6-kg weight reduction during the first where there is no need for accurate col- tended as conventional treatment of high- year of intervention, with slight, gradual risk individuals (3). In this study we regain at the end of the study, i.e., 4 years lection of food intake data. describe the intervention used in the DPS (2). The habitual nutrient intakes of the in detail in order to enable utilization of Maintenance of the weight reduction study subjects were estimated using 3-day our experience by others. after the 1-year intensive intervention pe- food records, which is a reliable method The intensive lifestyle intervention riod was satisfactory. It has been specu- in analyzing dietary intakes of groups. A induced several beneficial changes in diet, lated that greater initial weight loss well-known phenomenon in dietary anal- physical activity, blood glucose, and lipid achieved with the use of VLCD prepara- yses is underreporting, which is even concentrations and a highly significant re- tions or with drugs might predict better more prominent in overweight individu- duction in diabetes incidence. The inter- long-term results (23). In our study, those als (26). The relatively low energy intakes vention program was most intensive not satisfied with the weight reduction indicate that some underreporting has during the first year, and consequently achieved with the core program benefited taken place in our study. On the other the changes in clinical characteristics from supplementary VLCD treatment, hand, our analyses are based on differ- were most prominent after the first year. and their diabetes incidence was similar ences between the two groups, and un- The effect of intervention, e.g., the differ- to those not using VLCD. derreporting is presumably similar in ences between the intensive intervention Independent of weight loss, calorie both groups. Moreover, under- and over- and the control groups, was somewhat at- restriction results in improved glucose reporting has been shown to be subject tenuated at 3 years, but this result may be metabolism in subjects with type 2 diabe- specific (27), and as we measure changes

3234 DIABETES CARE, VOLUME 26, NUMBER 12, DECEMBER 2003 Lindstro¨m and Associates in dietary intake within a person, the bias 2. The Diabetes Prevention Program Re- Hannuksela M, Savolainen M, Kervinen may be attenuated. search Group: Reduction in the incidence K, Kesa¨niemi A, Uusitupa MI: The effects The frequency, duration, and inten- of type 2 diabetes with lifestyle interven- of monounsaturated-fat enriched diet and sity of leisure time and lifestyle physical tion or metformin. N Engl J Med 346:393– polyunsaturated-fat enriched diet on lipid activity during the preceding 12 months 403, 2002 and glucose metabolism in subjects with 3. The Diabetes Prevention Program Re- were estimated by the study subjects at impaired glucose tolerance. Eur J Clin search Group: The Diabetes Prevention Nutr 50:592–598, 1996 each annual visit. It was not straightfor- Program (DPP): description of lifestyle in- 13. Prochaska JO, DiClemente CC, Norcross ward and may have been incomplete due tervention. Diabetes Care 25:2165–2171, JC: In search of how people change: ap- to difficulty recollecting, especially occa- 2002 plications to addictive behaviors. Am Psy- sional activities. The majority of the study 4. World Health Organization: Diabetes Mel- chol 47:1102–1114, 1992 subjects considered themselves at least litus: Report of a WHO Study Group. Ge- 14. Haapa E, Toponen T, Pietinen P, Ra¨sa¨nen moderately active (Ն4 h/week of some neva, World Health Org., 1985 (Tech. L: Annoskuvakirja. Helsinki, Finland, kind of physical activity), but based on Rep. Ser., no. 727) Painokaari Oy, 1985 the LTPA questionnaire most of their ac- 5. Eriksson J, Lindstro¨m J, Valle T, Aunola S, 15. Ovaskainen M-L, Valsta L, Lauronen J: tivity was low intensity. The reported Ha¨ma¨laïnen H, Ilanne-Parikka P, Kein- The compilation of food analysis values as a¨nen-Kiukaanniemi S, Laakso M, Lauh- time spent on moderate- and high- a database for dietary studies: the Finnish konen M, Lehto P, Lehtonen A, experience. Food Chemistry 57:133–136, intensity LTPA increased slightly in the Louheranta A, Mannelin M, Martikkala V, intervention group compared with the 1996 Rastas M, Sundvall J, Turpeinen A, Vil- 16. Lakka TA, Vena¨laïnen JM, Rauramaa R, control group. If other types of supervised janen T, Uusitupa M, Tuomilehto J: Pre- Salonen R, Tuomilehto J, Salonen JT: Re- exercise in addition to the circuit-type re- vention of type II diabetes in subjects with lation of leisure-time physical activity and sistance training sessions had been of- impaired glucose tolerance: the Diabetes cardiorespiratory fitness to the risk of fered, then the increment might have Prevention Study (DPS) in Finland: study acute myocardial infarction. N Engl J Med been higher. design and 1-year interim report on the 330:1549–1554, 1994 The main finding in the DPS was that feasibility of the lifestyle intervention pro- 17. Lakka TA, Salonen JT: Intra-person vari- nonpharmacologic lifestyle intervention gramme. Diabetologia 42:793–801, 1999 ability of various physical activity assess- 6. Ohlson LO, Larsson B, Svardsudd K, We- of people at high risk for diabetes does ments in the Kuopio Ischaemic Heart lin L, Eriksson H, Wilhelmsen L, Bjo¨rn- Disease Risk Factor Study. Int J Epidemiol prevent or at least postpone the onset of torp P, Tibblin G: The influence of body 21:467–472, 1992 type 2 diabetes. The observed difference fat distribution on the incidence of diabe- 18. Pan XR, Li GW, Hu YH, Wang JX, Yang in incidence between the intensive inter- tes mellitus: 13.5 years of follow-up of the WY, An ZX, Hu ZX, Lin J, Xiao JZ, Cao vention and the usual care control group participants in the study of men born in HB, Liu PA, Jiang XG, Jiang YY, Wang JP, 1913. Diabetes 34:1055–1058, 1985 indicates that the intervention needs to be Zheng H, Zhang H, Bennett PH, Howard 7. Ohlson LO, Larsson B, Bjo¨rntorp P, Eriks- individualized and continuing and per- BV: Effects of diet and exercise in prevent- formed by skilled professionals in order son H, Svardsudd K, Welin L, Tibblin G, Wilhelmsen L: Risk factors for type 2 ing NIDDM in people with impaired glu- to be effective. The lifestyle intervention cose tolerance: the Da Qing IGT and program used in the DPS is practical and (non-insulin-dependent) diabetes mellitus: thirteen and one-half years of fol- Diabetes Study. Diabetes Care 20:537– can be implemented in primary health low-up of the participants in a study of 544, 1997 care systems. Swedish men born in 1913. Diabetologia 19. Swinburn BA, Metcalf PA, Ley SJ: Long- 31:798–805, 1988 term (5-year) effects of a reduced-fat diet 8. Colditz GA, Willett WC, Stampfer MJ, intervention in individuals with glucose Acknowledgments— The DPS study has Manson JE, Hennekens CH, Arky RA, intolerance. Diabetes Care 24:619–624, been financially supported by the Finnish Speizer FE: Weight as a risk factor for clin- 2001 Academy, Ministry of Education, Novo Nor- ical diabetes in women. Am J Epidemiol 20. Poppitt SD, Keogh GF, Prentice AM, Wil- disk Foundation, Yrjo¨ Jahnsson Foundation, 132:501–513, 1990 liams DE, Sonnemans HM, Valk EE, Rob- Juho Vainio Foundation, and Finnish Diabetes 9. Cassano PA, Rosner B, Vokonas PS, Weiss inson E, Wareham NJ: Long-term effects Research Foundation. ST: Obesity and body fat distribution in of ad libitum low-fat, high-carbohydrate We are indebted to the DPS research team relation to the incidence of non-insulin- diets on body weight and serum lipids in members for their skillful contribution in per- dependent diabetes mellitus: a prospec- overweight subjects with metabolic syn- forming the study, Dr. Timo Lakka and Prof. tive cohort study of men in the normative drome. Am J Clin Nutr 75:11–20, 2002 Jukka T. Salonen for advice concerning exer- aging study. Am J Epidemiol 136:1474– 21. Mensink M, Feskens EJ, Saris WH, De cise assessment, and Markku Peltonen, PhD, 1486, 1992 Bruin TW, Blaak EE: Study on Lifestyle for statistical advice. 10. Eriksson KF, Lindga¨rde F: Prevention of Intervention and Impaired Glucose Toler- type 2 (non-insulin-dependent) diabetes ance Maastricht (SLIM): preliminary re- mellitus by diet and physical exercise: the sults after one year. Int J Obes Relat Metab References 6-year Malmo feasibility study. Diabetolo- Disord 27:377–384, 2003 1. Tuomilehto J, Lindstro¨m J, Eriksson JG, gia 34:891–898, 1991 22. The Diabetes Prevention Program Re- Valle TT, Ha¨ma¨laïnen H, Ilanne-Parikka 11. Eriksson J, Tuominen J, Valle T, Sundberg search Group: Costs associated with the P, Keina¨nen-Kiukaanniemi S, Laakso M, S, Sovija¨rvi A, Lindholm H, Tuomilehto J, primary prevention of type 2 diabetes Louheranta A, Rastas M, Salminen V, Koivisto V: Aerobic endurance exercise or mellitus in the diabetes prevention pro- Uusitupa M: Prevention of type 2 diabetes circuit-type resistance training for indi- gram. Diabetes Care 26:36–47, 2003 mellitus by changes in lifestyle among viduals with impaired glucose tolerance? 23. Astrup A, Ro¨ssner S: Lessons from obesity subjects with impaired glucose tolerance. Horm Metab Res 30:37–41, 1998 management programmes: greater initial N Engl J Med 344:1343–1350, 2001 12. Sarkkinen E, Schwab U, Niskanen L, weight loss improves long-term mainte-

DIABETES CARE, VOLUME 26, NUMBER 12, DECEMBER 2003 3235 DPS lifestyle intervention

nance. Obes Rev 1:17–19, 2000 MD, Watanabe R, Bergman RN: Caloric validity of reported energy intake (Review). 24. Kelley DE, Wing R, Buonocore C, Sturis J, restriction per se is a signiﬁcant factor in J Nutr 133 (Suppl. 3):895S–920S, 2003 Polonsky K, Fitzsimmons M: Relative ef- improvements in glycemic control and in- 27. Black AE, Cole TJ: Biased over- or under- fects of calorie restriction and weight loss in sulin sensitivity during weight loss in reporting is characteristic of individuals noninsulin-dependent diabetes mellitus. J obese NIDDM patients. Diabetes Care 17: whether over time or by different assess- Endocrinol Metab 77:1287–1293, 1993 30–36, 1994 ment methods. J Am Diet Assoc 101:70– 25. Wing RR, Blair EH, Bononi P, Marcus 26. Livingstone MBE, Black AE: Markers of the 80, 2001

3236 DIABETES CARE, VOLUME 26, NUMBER 12, DECEMBER 2003