¹Department of Public Health, University of Helsinki, Helsinki, Finland

²Obesity Research Group, Helsinki University Central Hospital, Helsinki, Finland

³Department of Mental Health and Alcohol Research, National Public Health Institute, Helsinki, Finland

Correspondence to: S Sarlio-Lähteenkorva, Department of Public Health, PO Box 41, FIN-00014, University of Helsinki, Helsinki, Finland. sirpa.sarlio-lahteenkorva@helsinki.fi.

OBJECTIVE: To describe factors associated with long-term maintenance of weight loss.

DESIGN AND SUBJECTS: We identified initially overweight individuals (body mass index>27 kg/m², n=911) from the nationwide Finnish Twin Cohort and studied those who lost at least 5% of their body weight between 1975 and 1981. Subjects who had maintained weight loss until 1990 (38 men, 17 women) were compared to both re-gainers (28 men, 26 women) and the other overweight subjects in the cohort.

MEASUREMENTS: Self-report data on weight, height, health behaviours and perceived well-being; self-report and register-based data on health status and use of medication.

RESULTS: Only 6% of all overweight individuals lost and maintained at least 5% weight loss. In men weight loss maintenance was associated with a low level of stress and health-promoting behaviours but also with medical problems. Failure to maintain weight loss seemed to be associated with stressful life and past high alcohol intake. In women weight loss maintenance was associated with low initial well-being and health-compromising behaviours that improved after weight loss.

CONCLUSION: Long-term weight loss maintenance is rare. Predictors of weight loss maintenance are different between women and men.

International Journal of Obesity (2000) 24, 116-125

overweight; weight loss maintenance; health; behaviour; well-being

Introduction

The weight of many obese individuals varies. Weight changes in adults appear to be determined mainly by environmental effects,¹ including health promoting behaviours such as regular exercise,^2,3,4,5,6 restricting diets^2,3,4,7 and high vegetable consumption.⁸ Some potentially health damaging behaviours such as smoking,⁵ drinking habits⁹ and ill health may also be associated with weight loss. Both intentional and unintentional weight loss contribute to weight variability. The importance on focusing on intentional weight loss has been recognized,^10,11,12,13 but assessing intentionality of weight loss, however, is problematic. Most intentional attempts to lose weight fail and prospective studies have shown that dieting even increases the risk of long-term weight gain.^4,14 Moreover, it is possible that environmental or behavioural changes leading to altered lifestyle can result in sustained weight loss, even without a primary 'intention' to lose weight. Various environmental¹⁵ and psycho-social issues,^3,16,17 including perceived level of stress^16,18,19 and general life satisfaction¹⁹ may be important for weight development, but these factors have rarely been studied. Indeed, characteristics of those overweight subjects who maintain weight loss, either intentionally or unintentionally, are incompletely understood.

Long-term prognosis of weight reduction is poor and factors associated with sustained weight loss are inadequately understood.¹⁰ Studies focusing on rare individuals that have been successful at losing excess weight and keeping it off provide one way of increasing our knowledge on weight loss maintenance. Existing studies using this approach are treatment studies²⁰ or subjects are volunteers recruited from media and mailings to weight loss programmes.^2,15,21 Both approaches suffer from selection bias: overweight subjects seeking treatment or volunteering probably differ from the overweight population in general. In addition, studies on successful volunteers are retrospective and treatment studies end up with small numbers of subjects. Although there are population-based retrospective studies on weight loss mainten-ance,⁵ prospective studies focusing on subjects successful in weight loss maintenance have not been published.

This prospective, population-based study describes the relationship between weight loss maintenance and (1) health status and use of medication, (2) health-related behaviours, and (3) indicators of subjective well-being among overweight subjects before and after weight loss. The goal is to characterize weight losers and potential changes in their responses throughout the study. We hypothesize that different factors may be pertinent during various phases of weight reduction. These may include factors that are relevant when still obese, possible changes after weight loss and potential outcomes of weight loss maintenance.

Subjects

The Finnish Twin Cohort and selection of subjects for the present study

The adult Finnish Twin Cohort was compiled in 1974 from the Finnish Central Population Register in order to study the relative importance of genetic and environmental effects on chronic diseases.²² The baseline cohort comprised all same-sex twin pairs born in Finland before 1958 and alive in 1967. Subjects were mailed similar questionnaires in 1975 and 1981, and the response rates were 89% and 84%, respectively. An additional questionnaire was mailed in 1990 to twins born between 1930 and 1957, and with both co-twins alive and in Finland in 1987, with a response rte of 77%.¹⁹

Complete information about current weight and height from all three questionnaires was obtained from 11,284 subjects. Our study focused on subjects who were clearly overweight (body mass index, BMI>27 kg/m²) at baseline in 1975 (n=968). From this overweight sample the following exclusions were made: 24 women who were pregnant at baseline by a linkage to the National Hospital Discharge Register,²³ two men and one woman diagnosed with a malignant cancer before 1983 by linkage to the Finnish Cancer Registry, as well as 20 men and nine women who reported being on a disability pension in 1975. After these exclusions the final sample consisted of 332 women and 579 men.

Weight loss maintainers

Subjects were classified as weight loss maintainers if they had lost at least 5% of their body weight between 1975 and 1981 and were still at least 5% below their original weight in 1990. The cut point of 5% was chosen on the basis of the literature.¹⁰ Since the 1990 questionnaire asked respondents to also report their lifetime maximum (non-pregnant) weight, we excluded one woman and one man on the basis of this information: they reported weights that were inconsistent with their earlier weight history. In addition, the subjects were in 1990 asked to give their weight 12 months and 5 years previously, and we were also able to exclude subjects with obvious weight fluctuation: 13 women and 16 men who reported weights that were not below their original weight in 1975. We found 55 subjects (17 women, 38 men) who met the criteria for weight loss maintenance. There were only two twin pairs among men who maintained weight loss (one monozygotic, other with unknown zygosity) and none among women, and hence the analysis was done with no regard for twinship or zygosity.

Comparison groups

Two comparison groups were selected for both men and women:

1. Overweight. Originally this comparison group consisted of all other initially overweight subjects in our cohort. In the analysis all men in this group (n=540) were used. For women we had to use another strategy since we found weight loss maintainers to be markedly younger than overweight women at large (birth year 1946 vs 1939, P<0.0001), with a very skewed age distribution. Therefore, we selected at random four overweight controls from the same age group (same or next birth year) for each woman maintainer.

2. Re-gainers. This comparison group consists of those initially overweight subjects who lost at least 5% of their body weight between 1975 and 1981, but had regained more than they initially lost by 1990. This group consists of 28 men and 26 women.

Methods

Body mass and dieting

Subjects reported their weight in kilograms and height in centimetres, which were rounded to nearest integral number. In addition, subjects were asked in 1990 to recall their weight 1 and 5 y previously, their lifetime maximum (non-pregnant) weight, weight at the age of 20 and 30 as well as weight and height of their spouse. BMI (weight in kilograms/height in metres squared) was used as a measure of relative weight. At baseline subjects were asked if they were currently trying to lose weight, and in 1990 subjects were asked if they had ever dieted or tried to lose weight. Those admitting dieting either in 1975 or 1990 were classified as having intentional weight loss attempts in the past.

Background variables

Educational attainment was obtained from the 1975 questionnaire, employment and marital status from all three (1975, 1981 and 1990) questionnaires. In addition, subjects were asked whether they had either schoolchildren or younger children in their households in 1990.

Health status and use of medication

Information about current or past physician-diagnosed diabetes, myocardial infarction or hypertension and use of tranquillizers, sleeping pills and pain killers within past 12 months was obtained in all three questionnaires. Information on the right to reimbursable medication for selected chronic diseases (such as cardiovascular diseases, diabetes, asthma, psychoses), and information about current or past applications for disability pensions were obtained from the Social Insurance Institution of Finland in 1987. The record linkage was done using the personal identification code assigned to all residents of Finland.

Health-related behaviour

Smoking, alcohol use and physical activity were used as indicators of health-related behaviours in all three questionnaires. In addition, selected dietary habits were enquired into in 1975 and 1981. Data on past and current smoking habits included questions about the daily number of cigarettes and the age of starting smoking among regular smokers.²⁴ Alcohol use was recorded as beverage-type specific items (beer, wine, spirits) with frequency and quantity, and converted into grams of absolute alcohol per day as previously reported.²⁵ The baseline questionnaire included additional questions about possible previous greater alcohol intake, and frequency as well as quantity of this greater intake. Current leisure-time physical activity was evaluated with identical questions in 1975 and 1981, but the 1990 questionnaire included physical activity during the travelling to and from work as a part of leisure-time physical activity. As previously,²⁶ leisure physical activity index (MET h/d) was calculated from the products of intensity ´ duration ´ monthly frequency. In addition, leisure time activity during the past 5 y was asked for in 1981. Dietary questions included daily intake of coffee and tea in 1975 and 1981 as well as the amount of sugar used on average in these beverages. In addition, frequency of including vegetables or fruits in meals, daily intake of high salt food items and type of spread on bread were asked in 1981.

Indicators of well-being

Life satisfaction, perceived stress and selected symptoms (sleep quality, breathlessness, musculoskeletal pains) were used as indicators of subjective well-being. Life satisfaction according to Allardt²⁷ was measured with a four-item scale on levels of interest, happiness, easiness and loneliness of life.¹⁹ The range of the total score was 4-20, higher scores indicating a less satisfying life. Those with scores of 12-20 were classified as very dissatisfied.²⁸ Stress of daily activities was assessed by a four-item scale^19,29 and total stress score was obtained by reversing the scale and summing scores of all items. Thus, higher scores indicate more stress in daily activities. Those with scores below 6 were classified as stress-free.

The subjects were also asked whether they usually get short of breath when walking on level ground at an ordinary pace with people of their own age or when walking uphill, climbing stairs or hurrying on level ground.³⁰ In addition, sleep quality³¹ was questioned in 1975 and repeated in 1981. The questionnaire in 1990 assessed insomnia³² and sleep disturbances. A history of back, shoulder and/or neck pain causing difficulties at work during recent years was included in 1975 and 1981.

Statistical methods

Differences between weight loss maintainers and comparison groups were tested with a Mann-Whitney-U, two-way t-test or Pearson's chi-square test. Changes over time within a group were tested with Wilcoxon's matched pairs signed-rank sum test. All analyses were computed with the SPSS 7.0 program and done separately for women and men.

Results

Only 5.1% of all initially overweight women and 6.6% of men lost and maintained at least 5% weight loss. The women who maintained weight loss were less often married at baseline than the re-gainers (35% vs 81%, P=0.003), but there were no other differences between the maintainers and the comparison groups in the background variables (Table 1), including educational attainment, employment status, weight of their spouses, maximum weight during lifetime and weight in twenties or thirties. However, the men who maintained weight loss had initially higher BMI (30.0 kg/m²) than either the re-gainers (28.9 kg/m², P=0.018) or the overweight men (28.8 kg/m², P<0.0001), and they also lost more weight than the re-gainers (9.7 vs 7.3 kg, P=0.007). Subjects who were successful in weight loss maintenance had significantly lower BMIs than the comparison groups in 1985, 1989 and 1990. Intentional weight loss attempts were reported by nearly all women and three quarters of men irrespective of final weight loss outcome.

Health status and use of medication

Among men, no differences were found in self-reported health status or use of medication between the maintainers and the comparison groups at baseline in 1975 (Table 2). However, the maintainers reported more diabetes than the overweight men in 1981 (8% vs 2%, P=0.035) and in 1990 (22% vs 8%, P=0.003). Also compared to the re-gainers, the maintainers had more diabetes (22% vs 4%, P=0.038) and more hypertension (41% vs 18%, P=0.05) in 1990. In addition, the use of medication for chronic diseases was more common among the maintainers, of whom 42% had obtained the right for reimbursed medication for a chronic disease by 1987 compared to only 23% of the overweight (P=0.007) and 11% of the re-gainer men (P=0.005). On the other hand, the maintainers reported use of tranquillizers less often than the re-gainers in 1981 (0% vs 16%, P=0.019) and in 1990 (3% vs 26%, P=0.009).

A history of myocardial infarction was slightly more common among women who maintained their weight loss than among other women. There was also some indication of increased use of some drugs among maintainer women. In 1981, they reported more use of sleeping pills than the overweight women (24% vs 2%, P=0.001) and more pain killers than the re-gainers (94% vs 68%, P=0.043). No other differences in the health status or use of medication were observed between the women who maintained their weight loss and comparison groups

Health-related behaviour

Some gender-specific differences in health-related behaviours between the weight loss maintainers and comparison groups were observed already at baseline (Table 3). Only 3% of the maintainer men reported having had a higher alcohol intake in the past compared to 19% of the overweight men (P=0.01) and 43% of the re-gainers (P<0.0001). In addition, the maintainers consumed less sugar in coffee and tea than the overweight men (5.1 vs 8.6 lumps daily, P=0.01) and those who smoked had taken up the habit at later age than the re-gainers. More health-promoting habits could be detected in 1981: more frequent use of vegetables than the overweight men (P<0.0001) or the re-gainers (P=0.041), more frequent leisure time physical activity (P=0.033), higher MET values (P=0.001) and more frequent intensive physical activities during the previous 4-5 y (P=0.001) than the overweight men, however these differences were not seen in 1990. In addition, there were no differences in alcohol use between groups in any of the three questionnaires. The overweight men increased their alcohol consumption from 1981 to 1990 (P=0.02), but otherwise changes in alcohol consumption were not significant. Both the maintainers (P=0.004) and the overweight men (P=0.008) increased their coffee consumption from 1975 to 1981. A decreased intake of sugar in coffee and tea from 1975 to 1981 was reported by the overweight men (P<0.0001) and the re-gainers (P=0.008). Leisure-time MET values increased from 1975 to 1981 in all groups of men.

Data on women show a different pattern. Adverse habits were associated with weight loss maintenance. Baseline data reveal that the maintainers had started smoking very young (mean 14 y) compared to the overweight women (19 y, P=0.008) or the re-gainers (21 y, P=0.023). The maintainers also had a higher consumption of cigarettes than the re-gainers (17 vs 9 cigarettes daily, P=0.013). At baseline the maintainers also exercised less frequently than the overweight women (P=0.006) and had somewhat more frequently a history of higher alcohol intake in the past. In 1981, the maintainers reported more heavy drinking (18% vs 0%, P=0.025) and lower daily intake of salty foods (P=0.003) than the re-gainers. Both the maintainers and the re-gainers increased their physical activity from 1975 to 1981 (P=0.002 and P=0.003, respectively). Increased coffee consumption from 1975 to 1981 was reported by the maintainers only (P=0.022), whereas the overweight women decreased their intake of sugar in coffee and tea (P=0.004). There were no other differences in health-related behaviours between the maintainers and the comparison groups in 1981 or 1990.

Perceived well-being

Gender-specific differences were also seen in indicators of perceived well-being (Table 4). At baseline the maintainer men more often had a stress-free life than overweight men (50% vs 27%, P=0.005) and lower scores for stress than the re-gainers in 1981 (6.7 vs 8.6, P=0.01) and in 1990 (6.2 vs 7.9, P=0.014). The stress scores did not change among the maintainers during follow-up, but stress decreased among the re-gainers (P=0.044) and he overweight men (P<0.0001) from 1981 to 1990. The maintainers reported less breathlessness than the overweight men when walking in 1981 (0% vs 9.7%, P=0.044) and climbing uphill (16%0.044) and climbing uphill (16% vs 39%, P=0.005) in 1981 and in 1990 (21% vs 40%, P=0.020). In 1990 the re-gainers reported more breathlessness than the maintainers when walking (25% vs 5.3%, P=0.021) or climbing uphill (52% vs 21%, P=0.01). There were no differences in life satisfaction scores between the men who maintained eight loss and the comparison groups.

Again, the pattern was different in women. At baseline, the maintainers were more often very dissatisfied with their life compared with the overweight women (35% vs 13%, P=0.033), and had higher mean scores which indicated a less satisfying life compared with the overweight women (P=0.017) or he re-gainers (P=0.048). Loneliness was reported by 69% of the maintainers compared to only 22% of the re-gainers (P=0.003) and the overweight women (P<0.0001). In addition, the maintainers reported more breathlessness when climbing uphill than either the overweight women (88% vs 63%, P=0.044) or the re-gainers (88% vs 54%, P=0.019). Compared to the overweight women, the maintainers also had more breathlessness when walking on level ground (35% vs 14%, P=0.039) and they were less likely to report that they usually sleep well (18% vs 46%, P=0.035). In 1981, there were no longer differences in life satisfaction, but the maintainers continued to have problems with sleep: only 12% slept well compared to 50% of the overweight women (P=0.004) and 46% of the re-gainers (P=0.019). There were no differences in stress scores between the women who maintained weight loss and comparison groups.

Among both men and women, there were no differences between the maintainers and the comparison groups in the history of back, shoulder or neck pain in 1975 or 1981. In addition, there were no differences in the prevalence of insomnia or presence of children in the household in 1990 (data not shown).

Discussion

This study shows that weight loss of overweight persons is rarely maintained. Only 5% of the initially overweight women and 7% of the men succeeded in maintaining at least 5% weight loss during a 9 y period. Our findings confirm clinical observations about the difficulty of weight maintenance^20,33 and extend them to a general population.

In men, weight loss maintenance was associated with health-related issues. Both before and after weight loss the maintainers had a low level of stress and showed more health-promoting behaviours than the comparison groups. Weight loss was greater than among the re-gainers, and it was accompanied by several health-promoting behaviours, including increased physical activity and more vegetables in he diet. The importance of physical exercise and other health-promoting behaviours for weight maintenance has also been found in previous studies.^2,4,5,6,7,8 Not surprisingly, men who were successful in weight loss maintenance also reported better physical condition than comparison groups after weight loss. Healthy behaviours among some maintainers could probably be explained by chronic diseases. There was an increased incidence of diabetes and more registered rights for reimbursed medication for chronic diseases. It is possible that getting diabetes or other chronic health problems initially triggered weight loss attempts or motivated maintainers to keep weight level down. Part of he weight loss may have been unintentional given the insidious course of diabetes; the disease may have been present years before diagnosis was made.

The men who regained weight had many health-compromising behaviours prior to weight loss, such as a history of high alcohol intake and long exposure to cigarettes. Despite parallel improvements in these behaviours and weight there was no change in health status, however. On the contrary, both after weight loss and regain these men reported use of tranquillizers more often and had higher levels of stress than the maintainers, suggesting psychological distress and reduced well-being. Both weight gain¹⁹ and weight fluctuation¹⁶ have also been associated with stress previously. Stress, alcohol and smoking may stimulate the hypothalamic-pituitary-adrenal axis, causing endocrine abnormalities that promote visceral obesity.³⁴

In women, a different picture emerged. When still overweight, the maintainers were often living without a partner, suffered from loneliness and were less satisfied with life than the other women. They also seemed to have slightly more medical problems and reported poor physical condition, low level of exercise, a long smoking history and poor sleep. Weight loss was accompanied by positive behavioural changes such as increased physical activity and fewer cigarettes, and by less loneliness, improved life satisfaction and better physical condition. Van Gemert et al³⁵ reported that low initial self-esteem is associated with long-term success among surgically treated subjects. The authors suggest that those who suffer most from obesity are best motivated to lose weight.³⁵ This kind of phenomen could explain our results for weight loss maintainers as well. However, the maintainers continued to have problems with sleep and reported more use of sleeping pills, pain killers and 'heavy' alcohol intake after weight loss. The causes for these problems and potential risk behaviour are open to speculation. They may reflect some form of health-related problems that we were unable to capture with our survey, such as disordered eating, substance abuse or mental problems. Nevertheless, all differences between maintainers and comparison groups had diminished by the last survey, as the maintainer women reached a similar level of well-being as the other overweight women. Compared with women who maintained weight loss, those who regained weight were more often living with a partner at baseline and enjoyed a somewhat healthier lifestyle with better physical condition and greater life satisfaction. It is possible that overweight women without serious weight-related problems may be less motivated to maintain weight loss.

Gender-specific factors behind successful weight loss maintenance may reflect different dimensions of weight-related problems. Women with excess weight are more likely to be stigmatized and disadvantaged in our modern societies,³⁶ whereas men seldom suffer from such problems. Previous studies^2,21 indicate that women often report emotional or lifestyle triggers as a precursor of successful weight loss, whereas men report more medical reasons. It is possible that low subjective well-being may have acted as a trigger to success among our women as well. In men weight loss maintenance seems to be centred in medical problems and health-oriented behaviour only. However, men and women who succeeded in weight loss maintenance seemed to achieve either physical or psycho-social benefits after weight loss, whereas these rewards appear to be absent among those who had not maintained the weight loss. It seems obvious that in order to succeed in weight loss maintenance, there has to be some amelioration in the quality of life.

The presence of health-related problems and adverse habits among those with weight loss raises questions about causes and consequences of weight loss in our sample. We excluded subjects with malignant cancers and severe disabilities but it is nevertheless possible that, whether the intent to lose weight was reported or not, the weight loss was partly caused by an underlying illness such as poorly controlled diabetes in diabetic men with weight loss. Similarly, some weight loss may have ensued from adverse habits such as longer history of smoking in maintainer women. These unfavourable characteristics of subjects with weight loss could help to explain the perplexing finding that both unintentional and intentional weight loss are associated with increased disease prevalence³⁷ and weight loss is associated with increased mortality.¹¹

Both intentional and unintentional weight loss contribute to long-term weight changes. We chose not to confine our analyses to those with reported intentional weight loss since we wanted to get a comprehensive picture of those who achieve sustained weight loss in a population, irrespective of intentionality. The questions about intentional weight loss or 'dieting' are problematic for several reasons. First, weight loss attempts reported by a great majority of our study subjects were not related to weight loss outcome. For instance, attempts to lose weight were equally commonly reported by diabetic persons who maintained the weight loss as among the diabetics who did not maintain it (data now shown). Furthermore, the validity of questions focusing on past dieting attempts is not established and the interpretation of questions concerning dieting may differ according to gender, education and weight status.³⁸ Also, intentional dieting has been associated with subsequent weight gain.^5,14

The main strength of this study is a unique long-term follow-up of a large unselected population sample, even if the final analysis involved only a small number of subjects. We could identify gender-specific patterns of behaviour, health and subjective well-being that differed between maintainers and comparison groups. However, because of wide range of variables examined, individual findings must be evaluated with caution. Other limitations include use of self-report data, lack of information about weight fluctuation and many possible causes of weight loss. It is well known that self-report data underestimate the prevalence of overweight and result in 'flat slope syndrome' i.e. overweight subjects tend to underreport their BMI, whereas thin people do the reverse.^39,40 This phenomenon would lead to the underestimation of overweight and reduce the number of subjects chosen for our study, so the effect is likely to be conservative. Self-reported weight history was used as an additional selection criterion for weight maintain-ers, and subjects with obvious weight fluctuation could be excluded. Although this type of categorization discriminates weight cyclers from weight losers better than statistical measures of weight variability,⁴¹ it is possible that the data still include maintainers who had temporarily regained weight. Finally, the causes of weight loss are not known. However, as our purpose was just to characterize those rare individuals who show sustained weight loss in this prospective, population-based sample of overweight adults, further studies are clearly needed to evaluate causal relationships.

In conclusion, predictors of successful weight loss maintenance appear to be gender-specific. In men, medical problems and in women psycho-social issues appear to be important for weight loss maintenance. High prevalence of medical problems or health-compromising behaviour among those who lose weight may partly explain the increased mortality found in some weight loss studies. More prospective long-term studies are clearly needed to evaluate determinants and consequences of sustained weight loss.

Acknowledgements

Supported in part by grants from the Academy of Finland (nos 42022 and 37800)

1 Korkeila M, Kaprio J, Rissanen A, Koskenvuo M. Consistency and change of body mass index and weight. A study on 5967 adult Finnish twin pairs. Int J Obes 1995; 19: 310-317.

2 Klem ML, Wing RR, McGuire MT, Seagle HM, Hill JO. A descriptive study of individuals successful at long-term maintenance of substantial weight loss. Am J Clin Nutr 1997; 66: 239-246. MEDLINE

3 Coakley EH, Rimm EB, Colditz G, Kawachi I, Willet W. Predictors of weight change in men: Results from the health professionals follow-up study. Int J Obes 1998; 22: 89-96.

4 French SA, Jeffery RW, Forster JL, McGovern PG, Kelder, SH, Baxter JE. Pridictors of weight change over two years among a population of working adults: The Healthy Worker Project. Int J Obes 1994; 18: 145-154.

5 French SA, Jeffery RW, Folsom AR, McGovern P, Williamson DF. Weight loss maintenance in young adulthood: prevalence and correlations with health behavior and disease in a population-based sample of women aged 55-69 years. Int J Obes 1996; 20: 303-310.

6 Hunter GR, Weinsier RL, Bamman MM, Larson DE. A role of high intensity exercise on energy balance and weight control. Int J Obes 1998; 22: 489-493.

7 Klem ML, Wing RR, McGuire MT, Seagle HM, Hill JO. Psychological symptoms in individuals successful at long-term maintenance of weight loss. Health Psychol 1998; 17: 336-345. MEDLINE

8 Kahn HS, Tatham LM, Rodriguez C, Calle EE, Thun MJ, Heath CW, Jr. Stable behaviors associated with adults' 10-year change in body mass index and likelihood of gain at the waist. Am J Public Health 1997; 87: 747-754. MEDLINE

9 Hellerstedt WL, Jeffery RW, Murray DM. The association between alcohol intake and adiposity in the general population. Am J Epidemiol 1990; 132: 594-611. MEDLINE

10 WHO. Obesity: Preventing and managing the global epidemic. Report of a WHO consultation on obesity, 3-5 June 1997. World Health Organisation: Geneva, 1998,

11 Williamson DF. "Weight cycling" and mortality: How do the epidemiologists explain he role of intentional weight loss? J Am Coll Nutr 1996; 15: 6-13. MEDLINE

12 Williamson DF, Pamuk E, Thun M, Flanders D, Byers T, Heath C. Prospective study of intentional weight loss and mortality in overweight white men aged 40-64 years. Am J Epidemiol 1999; 149: 491-503. MEDLINE

13 French SA, Folsom AR, Jeffery RW, Williamson DF. Prospective study of intentionality of weight loss and mortality in older women: The Iowa Women's Health Study. Am J Epi-demiol 1999; 149: 504-514.

14 Korkeila M, Rissanen A, Kaprio J, Sörensen TIA, Koskenvuo M. Attempts to lose weight and risk for major weight gain. A prospective study on adult Finns. Am J Clin Nutr (in press).

15 Tinker JE, Tucker JA. Environmental events surrounding natural recovery from obesity. Addictive Behav 1997; 22: 571-575.

16 Foreyt JP, Brunner RL, Goodrick GK, Cutter G, Brownell KD, St Jeor ST. Psychological correlates of weight fluctuation. Int J Eating Disord 1995; 17: 263-275.

17 Rumpel C, Ingram DD, Harris TB, Madans J. The association between weight change and psychological well-being in women. Int J Obes 1994; 18: 179-183.

18 Greeno CG, Wing RR. Stress-induced eating. Psychol Bull 1994; 115: 444-464. MEDLINE

19 Korkeila M, Kaprio J, Rissanen A, Koskenvuo M, Sörensen TIA. Predictors of major weight gain in adult Finns: Stress, life satisfaction and personality traits. Int J Obes 1998; 22: 949-957.

20 Glenny Am, O'Meara S, Melville A, Sheldon TA, Wilson C. The treatment and prevention of obesity: A systematic review of the literature. Int J Obes 1997; 21: 715-737.

21 Colvin RH, Olson SB. A descriptive analysis of men and women who have lost significant weight and are highly successful at maintaining the loss. Addictive Behav 1983; 8: 287-295.

22 Kaprio J. Lessons from twin studies in Finland. Ann Med 1994; 26: 135-139. MEDLINE

23 Korkeila M, Kaprio J, Rissanen A, Koskenvuo M. Effects of gender and age on the heritability of body mass index. Int J Obes 1991; 15: 647-654. MEDLINE

24 Kaprio J, Koskenvuo M. A prospective study of psychological and socioeconomic characteristics, health behavior and morbidity in cigarette smokers prior to quitting compared to persistent smokers and non-smokers. J Clin Epidemiol 1988; 41: 139-150. MEDLINE

25 Kaprio J, Koskenvuo M, Langinvainio H, Romanov K, Sarna, Rose RJ. Genetic influences on use and abuse of alcohol: A study of 5638 adult Finnish twin brothers. Alcohol Clin Exp Res 1987; 11: 349-356. MEDLINE

26 Kujala UM, Kaprio J, Sarna S, Koskenvuo M. Relationship of leisure-time physical activity and mortality: The Finnish twin cohort. JAMA 1998; 279: 440-444. MEDLINE

27 Allardt E. About dimensions of welfare. An explanatory analysis of a comparative Scandinavian survey. Research group for comparative sociology. University of Helsinki, Research Report 1 1973;

28 Koivumaa-Honkanen, H.-T. Life satisfaction as a health predictor. Doctoral Dissertation, D. Medical Series 143. Kuopio University Publications: Kuopio, 1998,

29 Räiha I, Kemppainen H, Kaprio J, Koskenvuo M, Sourander. Lifestyle, stress, and genes in peptic ulcer disease: a nationwide twin cohort study. Arch Intern Med 1998; 158: 698-704. MEDLINE

30 Rose GA, Blackburn H, Gillum RF, Prineas RJ. Cardiovascular survey methods. World Health Organisation: Geneva, 1982,

31 Partinen M, Putkonen PT, Kaprio J, Koskenvuo M, Hilakivi. Sleep disorders in relation to coronary heart disease. Acta Med Scand 1982; 660: 69-83.

32 Hublin C, Kaprio J, Partinen M, Heikkilä K, Koskenvuo M. Daytime sleepiness in an adult, Finnish population. J Intern Med 1996; 239: 417-423. MEDLINE

33 Kramer FM, Jeffery RW, Forster JL, Snell MK. Long-term follow-up of behavioral treatment for obesity: Patterns of weight regain among men and women. Int J Obes 1989; 13: 123-136. MEDLINE

34 Björntorp P. Endocrine abnormalities of obesity. Metabolism 1995; 44: 21-23. MEDLINE

35 Gemert WG van, Svereijns RM, Greve JW, Groenman N, Soeters PB. Psychological functioning of morbidly obese patients after surgical treatment. Int J Obes 1998; 22: 393-398.

36 Stunkard AJ, Sörensen TIA. Obesity and socioeconomic status¾a complex relation. New Engl J Med 1993; 329: 1036-1037. MEDLINE

37 French SA, Jeffery RW, Folsom AR, Williamson DF, Byers T. Relation of weight variability and intentionality of weight loss to disease history and health-related variables in a population-based sample of women aged 55-69 years. Am J Epidemiol 1995; 142: 1306-1314. MEDLINE

38 Neumark-Sztainer D, Jeffery RW, French SA. Self-reported dieting: How should we ask? What does it mean? Associations between dieting and reported energy intake. Int J Eating Disord 1997; 22: 437-449.

39 Kuskowska-Wolk A, Bergström R, Boström G. Relationship between questionnaire data and medical records of height, weight and body mass index. Int J Obes 1992; 16: 1-9.

40 Jalkanen L, Tuomilehto J, Tanskanen A, Puska P. Accuracy of self-reported body weight compared to measured body weight. A population survey. Scan J Soc Med 1987; 15: 191-198.

41 French SA, Jeffery RW, Folsom AR, Williamson DF, Byers T. Weight variability in a population-based sample of older women: Reliability and intercorrelation of measures. Int J Obes 1995; 19: 22-29.

Table 1 BMI, weight changes (means, standard deviations) and background variables (%) by weight change category in men and women

Table 2 Health status and use of medication by weight change category in men and women

Table 3 Health status and health-related behaviours by weight change category in men and women

Table 4 Indicators of perceived well-being by weight change category in men and women