SEARCH

SEARCH BY CITATION

Keywords:

  • body mass index;
  • physical activity;
  • smoking;
  • weight change

Abstract.

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

Objectives.  The prevalence of obesity is increasing. Overweight and obese people have increased mortality compared with normal weight people. We investigated the effect of weight change on mortality.

Design.  Prospective population study.

Setting.  We utilized data from two large population-based health studies conducted in 1984–86 and 1995–97 respectively. Cox proportional hazards models were used to calculate mortality rate ratios (RRs) with 95% confidence intervals (CIs) between people with a stable weight and people who lost or gained weight.

Subjects.  Totally 20 542 men and 23 712 women aged 20 years or more, without cardiovascular disease or diabetes at the first survey and without a history of cancer at the second survey were followed up on all-cause mortality for 5 years after the second survey.

Results.  We found no association between weight gain and mortality. People who lost weight had a higher total mortality rate compared with those who were weight stable [RR was 1.6 (95% CI: 1.4–1.8) in men and 1.7 (95% CI: 1.5–2.0) in women]. Similar associations were found for cardiovascular and noncardiovascular mortality. Additional analysis showed a linear increase in mortality rates across categories of weight loss for both men and women (P < 0.001). There was a statistically significant interaction between weight change and initial BMI, but only amongst men (P = 0.001).

Conclusions.  Weight loss, but not weight gain, was associated with increased mortality amongst men and women. Although underlying undiagnosed disease is the most plausible explanation for this finding, the similar associations found for total mortality, cardiovascular mortality, and noncardiovascular mortality makes the causal pathway somewhat enigmatic.


Introduction

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

A body mass index (BMI) outside the normal weight range (18.5–24.9 kg m−2) is associated with increased mortality [1, 2], and the nadir of the mortality curve has been found at a BMI of 22–24 kg m−2 [3, 4]. Obesity has had disease status since 1985 [2]. Additionally, a number of diseases can be linked to overweight and obesity, and each disease can be classified into two pathophysiological categories [5]. The first arises from the increased mass of fat which may include the stigma of obesity and the behavioural responses it produces, and musculoskeletal disorders [6, 7]. The second category comprises metabolic changes associated with excess fat, and examples of these includes diabetes type 2 [8], gallbladder disease [9], hypertension [10], cardiovascular disease [11], and some forms of cancer [12].

The prevalence of overweight and obesity is rapidly increasing [2, 13], and consequently a large proportion of people are trying to lose weight. Weight loss is associated with short-time improvements in risk factors such as blood pressure [14], cholesterol [15] and diabetes [16, 17]. Controversially, weight loss has also been associated with increased mortality in observational studies [18–24], but the results are not consistent [1, 25, 26]. Additionally, the knowledge about the association between change in BMI and subsequent mortality is mainly based on studies on men [19, 27, 28].

We utilized information on height, weight and cause-specific mortality data in a large population of Norwegian men and women who participated in the Nord-Trøndelag Health Study both in 1984–86 and in 1995–97, to investigate the association between weight change and mortality.

We wanted especially to study the potential effect modification of initial BMI, leisure time physical activity and smoking status on the association between weight change and mortality.

Study population

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

In 1984–86 and 1995–97, two general health surveys were conducted in Nord-Trøndelag County (127 000 inhabitants), the Nord-Trøndelag Health Study, Norway. The participation rates were 88.1% and 71.2% respectively. Data collection was based on self-reported questionnaires and standardized measurements of physiological variables such as height and weight. In total 24 837 women and 21 685 men participated in both surveys. We excluded participants who reported pre-existing diabetes or cardiovascular disease at baseline or who had a history of cancer at the second survey. A total of 23 712 women and 20 542 men aged 20 years or more at the first survey and with information on body weight and body height at both surveys were available for analyses.

Follow-up

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

The unique 11-digit identification number of every Norwegian citizen enabled linkage between the collected information and the Death Registry at Statistics Norway to determine vital status (alive, emigrated and dead) and cause-specific deaths. Each participant contributed person-years from the date of the second survey until the date of death, emigration, or end of follow-up (31 December 2001). Mean time between the surveys was 11 years (range 9–13), and mean follow-up after the second survey was 5 years (range 0–6). Cardiovascular mortality was classified using the 9th revision of the International Classification of Diseases (cardiovascular diagnosis codes 390–459) before 1997, and the 10th revision (codes I00-I99) thereafter.

Body mass index

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

Body mass index was calculated as body weight in kilograms divided by the squared value of body height in metre (kg m−2). In both surveys height was measured without shoes to the nearest centimetre and weight was measured wearing light clothes without shoes to the nearest half-kilogram at the survey site. Change in BMI between the surveys was categorized into loss, stable and gain. A stable BMI was defined as a change in BMI ≤0.1 kg m−2 per follow-up year [29]. We categorized BMI at the first survey applying the WHO's recommendation (underweight: <18.5 kg m−2, normal weight 18.5–24.9 kg m−2, overweight 25.0–29.9 kg m−2, and obesity ≥30 kg m−2).

Leisure time physical activity

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

At the first survey, leisure time physical activity was self-reported by three questions about frequency, duration and intensity, each with five, four and three possible answers respectively. Only those who reported a frequency of once a week or more answered the questions about intensity and duration. We categorized leisure time physical activity into low, moderate and high levels based on the questions about frequency, intensity and duration: a frequency of never or less than once a week was categorized as low. For those with a frequency of once a week or more, a summary was calculated by adding the values of frequency, intensity and duration. The sum value was then divided into moderate and high by dichotomizing at the median value.

Smoking status

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

We classified smoking status at the first survey in three categories, where individuals who had never smoked daily, and those who reported previous or present daily smoking were classified as former or current smokers respectively.

Statistical analyses

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

Cox regression analysis was used to calculate age-adjusted and multivariable adjusted mortality rate ratios (RRs) with 95% confidence intervals (CIs) associated with change in BMI (loss, stable and gain), using the weight stable group as reference. The analyses were performed separately for males and females, and in strata of initial BMI (WHO categorization), leisure time physical activity levels and smoking status (never, former and current). Because of few cases, adjusted RR estimates were not calculated for men and women in the underweight group. We conducted multivariable analyses to assess potential confounding by the following variables measured at the first survey: age (<40, 40–44, …,≥80 years), BMI (<18.5, 18.5–24.9, 25.0–29.9, ≥30 kg m−2), systolic blood pressure (quintiles), blood pressure medication (no, yes), smoking (never, former, current), alcohol drinking past 2 weeks (none, 1–4 times, ≥5 times, teetotaller), leisure time physical activity (low, moderate, high), marital status (married, unmarried, widow/widower, divorced/separated), education (middle school, high school, <4 years of college/university, ≥4 years of college/university). All analyses were performed using the statistical software spss for Windows, version 11.0 (SPSS, Chicago, ILL, USA).

Ethics

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

The participation was completely voluntary and each participant signed a written consent. The Norwegian Data Inspectorate recommended both surveys, and the second survey was also approved by the Regional Ethical Committee for Medical Research. At the time of the first survey, the Regional Ethical Committee was not yet established.

Results

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

During 5 years of follow-up (mean = 5.4 years) we observed 2672 deaths altogether, 1553 amongst men and 1119 amongst women (Table 1). Out of these, 709 men and 475 women died from cardiovascular causes. Amongst men who lost weight we found that 24.2% died during follow-up, whilst 8.8% of men who were weight stable and 5.0% who gained weight died. Similar figures for women were 16.6%, 5.9% and 2.8% respectively. However, only 6.5% of men and 8.3% of women had lost weight between the surveys, whilst as much as 58.9% of men and 66.0% of women had gained weight. Persons who lost weight had on average a higher BMI at the first survey than those who were weight stable or gained weight. The mean weight loss was 2.2 kg m−2 in men and 2.7 kg m−2 in women, whilst mean weight gain was 2.7 and 3.4 kg m−2 for men and women respectively.

Table 1.  Characteristics of 20 542 men and 23 712 women aged 20 years or more who participated in the Nord-Trøndelag Health Study, Norway in 1984–86 and in 1995–97, stratified by gender and 11-year change in body mass index (BMI) in three categoriesa (loss, stable and gain)
VariablesMenWomen
LossStableGainLossStableGain
  1. aStable defined as a change in BMI ≤ 0.1 kg m−2 per year; a negative or positive change beyond this were classified as loss or gain respectively.

No. of participants (% within gender)1319 (6.4)7121 (34.7)12 102 (58.9)1971 (8.3)6092 (25.7)15 649 (66.0)
No. of all deaths (% within weight category)319 (24.2)627 (8.8)607 (5.0)327 (16.6)357 (5.9)435 (2.8)
Mean follow-up time, years (SD)4.9 (1.4)5.3 (1.0)5.4 (0.8)5.2 (1.2)5.4 (0.8)5.4 (0.7)
Mean age at death from all causes, years (SD)68.2 (9.1)65.4 (10.5)62.8 (11.1)70.5 (8.9)67.7 (10.1)62.5 (12.1)
Mean age at first survey, years (SD)54.3 (13.8)47.2 (13.5)41.0 (12.8)54.0 (15.8)47.8 (14.8)42.4 (12.9)
Mean BMI at first survey, kg m−2 (SD)26.9 (3.6)25.1 (2.9)25.0 (2.9)27.7 (5.2)24.6 (4.1)24.2 (3.8)
Mean BMI at second survey, kg m−2 (SD)24.8 (3.4)25.4 (2.9)27.7 (3.3)25.0 (4.6)24.9 (4.0)27.6 (4.4)
Mean change in BMI, kg m−2 (SD)−2.2 (1.1)0.2 (0.6)2.7 (1.4)−2.7 (1.9)0.2 (0.6)3.4 (1.9)

In analyses of weight change and mortality (Table 2) we found that people who lost weight had a higher total mortality rate compared with those who were weight stable (multivariable RR was 1.6, 95% CI: 1.4–1.8 in men and 1.7, 95% CI: 1.5–2.0 in women). Similar associations were also found in analyses of cardiovascular and noncardiovascular mortality. People who gained weight between the studies had the same mortality rate as those who were weight stable (total mortality RR = 1.0, 95% CI: 0.9–1.1 in men and 0.9, 95% CI: 0.8–1.0 in women). Additional analysis showed a statistically significant linear increase in mortality rates across categories of weight loss for both men and women (Ptrend < 0.001), but no linear relation across categories of weight gain (Ptrend = 0.26 amongst men and 0.11 amongst women) (Fig. 1).

Table 2.  Multivariable-adjusted rate ratios and 95% confidence intervals (CIs) of total mortality, cardiovascular (CV) mortality, and non-CV mortality associated with 10-year change in body mass index (BMI) in three categoriesa (loss, stable and gain) and stratified by gender: 5-year follow-up of 44 254 healthyb Norwegian men and women who participated in the Nord-Trøndelag Health Study in 1984–86 and 1995–97
Change in BMIaPerson- yearsTotal mortalityCV mortalityNon-CV mortality
No. of deathsAge-adj. RRMultivariablec RR (95% CI)No. of deathsAge-adj. RRMultivariablec RR (95% CI)No. of deathsAge-adj. RRMultivariablec RR (95% CI)
  1. aStable defined as a change in BMI ≤ 0.1 kg m−2 per year; a negative or positive change beyond this were classified as loss or gain respectively. bSubjects without diabetes and cardiovascular disease at the first survey, and without a history of cancer at the second survey. cAdjusted for the following variables measured at the first survey: age (<40, 40–44, …,≥80 years), BMI (<18.5, 18.5–24.9, 25.0–29.9, ≥30 kg m−2), systolic blood pressure (quintiles), blood pressure medication (no, yes), smoking (never, former, current), alcohol drinking past 2 weeks (none, 1–4 times, ≥5 times, teetotaller), leisure time physical activity (low, moderate, high), marital status (married, unmarried, widow/widower, divorced/separated), education (middle school, high school, <4 years of college/university, ≥4 years of college/university).

Males
 Loss65213191.71.6 (1.4–1.8)1501.71.5 (1.2–1.9)1691.81.6 (1.4–1.9)
 Stable37 6176271.01.02851.01.03421.01.0
 Gain65 1246071.01.0 (0.9–1.1)2741.01.0 (0.9–1.2)3330.91.0 (0.8–1.1)
Females
 Loss10 2143271.81.7 (1.5–2.0)1501.81.7 (1.3–2.1)1771.71.7 (1.4–2.0)
 Stable32 6833571.01.01471.01.02101.01.0
 Gain84 9314350.80.9 (0.8–1.0)1781.01.0 (0.8–1.3)2570.80.8 (0.7–1.0)
image

Figure 1. Total mortality rate ratios amongst men (a) and women (b) associated with categories of weight loss and weight gain when treated as an indicator variable in a Cox proportional hazards model; rate ratios are represented by bsl00001, with a 95% confidence interval (|).

Download figure to PowerPoint

In subsequent analyses on total mortality we found a statistically significant interaction between weight change and initial BMI amongst men (Pinteraction = 0.001), but not amongst women (Pinteraction = 0.31) (Table 3). Amongst men who had a normal weight (BMI: 18.5–24.9 kg m−2) at the first survey the total mortality RR was 2.0 (95% CI: 1.6–2.4) for those who lost weight compared with those who were weight stable. Amongst overweight or obese men, the RRs comparing weight loss and weight stable were 1.4 (95% CI: 1.2–1.8) and 1.5 (95% CI: 1.0–2.3) respectively. Although no statistically significant interaction was found for women, we observed that overweight women who lost weight had a higher RR than normal weight women who lost weight (2.0, 95% CI: 1.6–2.6 vs. 1.5, 95% CI: 1.1–1.9). We found no statistically significant interaction with physical activity or smoking status, although the results may indicate some effect modification by smoking status. Current smoking men who lost weight had an RR of 2.1 compared with weight stable men, whilst the same association amongst former smoking men was 1.2. However, amongst women the findings were somewhat opposite those for men, as the strongest association was found amongst former smokers (RR = 2.5 comparing loss and stable) and the weakest amongst never and current smokers (RR = 1.6 in both groups).

Table 3.  Multivariable-adjusted rate ratios (RRs) with 95% confidence intervals (CIs) of total mortality associated with 11-year change in body mass index (BMI) in three categoriesa (loss, stable and gain) amongst men and women; analyses are stratified by either BMI, physical activity, or smoking status at the first survey
Stratifying variableMultivariableb RR (95% CI)
MenWomen
LossStableGainLossStableGain
  1. aStable defined as a change in BMI ≤ 0.1 kg m−2 per year; a negative or positive change beyond this were classified as loss or gain respectively. bAdjusted for the following variables measured at the first survey: age (<40, 40–44,…,≥80 years), BMI (<18.5, 18.5–24.9, 25.0–29.9, ≥30 kg m−2), systolic blood pressure (quintiles), blood pressure medication (no, yes), smoking (never, former, current), alcohol drinking past 2 weeks (none, 1–4 times, ≥5 times, teetotaller), leisure time physical activity (inactive, active), marital status (married, unmarried, widow/widower, divorced/separated), education (middle school, high school, <4 years of college/university, ≥4 years of college/university). cNormal weight (18.5–24.9 kg m−2), overweight (25.0–29.9 kg m−2), and obesity (≥30 kg m−2); underweight (<18.5 kg m−2) excluded due to small numbers.

Initial BMIc
 Normal weight2.0 (1.6–2.4)1.00.9 (0.8–1.1)1.5 (1.1–1.9)1.00.8 (0.6–1.0)
 Overweight1.4 (1.2–1.8)1.01.0 (0.9–1.2)2.0 (1.6–2.6)1.01.0 (0.8–1.2)
 Obesity1.5 (1.0–2.3)1.01.1 (0.8–1.6)1.8 (1.3–2.6)1.01.1 (0.8–1.6)
 Pinteraction = 0.001  Pinteraction = 0.08  
Physical activity level
 Low1.7 (1.3–2.1)1.01.0 (0.8–1.2)1.7 (1.3–2.3)1.00.9 (0.7–1.1)
 Moderate1.3 (1.0–1.7)1.00.9 (0.7–1.1)1.8 (1.4–2.4)1.00.9 (0.7–1.1)
 High1.7 (1.2–2.3)1.01.2 (0.9–1.5)1.5 (0.9–2.4)1.00.8 (0.5–1.2)
 Pinteraction = 0.31  Pinteraction = 0.95  
Smoking status
 Never1.4 (1.0–1.9)1.00.9 (0.7–1.2)1.6 (1.3–2.0)1.00.9 (0.7–1.1)
 Former1.2 (0.9–1.6)1.01.0 (0.8–1.2)2.5 (1.4–4.5)1.01.1 (0.7–1.8)
 Current2.1 (1.6–2.7)1.00.9 (0.7–1.2)1.6 (1.1–2.4)1.01.0 (0.7–1.4)
 Pinteraction = 0.14  Pinteraction = 0.46  

In supplementary analyses we explored potential confounding by disease status reported at the second survey for events that could have occurred between the two surveys, such as myocardial infarction, stroke, diabetes, angina, chronic obstructive lung syndrome, and asthma, but none of these factors influenced the estimated association between weight change and mortality (data not shown). In an attempt to evaluate the role of prediagnosed disease as a plausible cause for our findings we excluded the first 3 years of follow-up, but the results remained similar [total mortality RR comparing weight loss and weight stable was 1.5 (95% CI: 1.3–1.8) amongst men and 1.7 (95% CI: 1.4–2.1) amongst women].

Discussion

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

In this prospective study we found a statistically significant higher mortality rate amongst people who had lost weight compared with people with a stable weight, both in analysis of total mortality, CV mortality and non-CV mortality. In addition, previous studies have reported that people who have lost weight have higher rates of both total mortality and cause-specific mortality than those who are weight stable [20–24, 30], although some studies have not found this association [1, 25, 26]. Additionally, we observed a linearly increasing mortality RR with increasing weight loss, which is somewhat contradictory to the findings by Williamson et al. [31], who reported a J-shaped association with increasing weight loss. However, they studied the effect of self-reported intentional weight loss, whilst we did not have the ability to distinguish between intentional and unintentional loss. It has been argued that this distinction is necessary [32, 33], as those who intend to lose weight are doing so for health promotion and disease prevention rather than treatment of weight-related health conditions. The validity about the knowledge of intentional versus non-intentional weight loss in observational studies has been questioned [34], and in a general population a large proportion will state that they try to lose weight at any time[35, 36], but the effort put into it will probably be extremely variable. In addition, others have suggested that both intentional and unintentional weight loss may follow the development of disease [37].

In agreement with previous studies [11], we found that weight loss was associated with increased mortality in all categories of initial BMI. However, the highest mortality rates associated with weight loss was seen amongst normal weight men and overweight women, indicating some effect modification by initial BMI. The higher mortality rate amongst normal weight men may indicate that weight loss is more hazardous if the initial body mass is low, although a similar reasoning is not as obvious for women. Previous data has suggested that weight change is associated with a more unfavourable relative change in fat-free mass in men than women, suggesting that the metabolic and health consequences of weight change may be dependent on gender [38]. Due to the low number of people who initially were underweight and then lost weight, we have not presented results for this strata.

Smoking status is likely to be an important factor when studying weight loss and mortality, as it is associated both with lower body weight and increased morbidity [39] and mortality [40]. However, only few prospective studies have investigated the potential effect modification of smoking on the association between weight loss and mortality [11, 41]. In our study weight loss was associated with increased mortality both in never, former and current smoking men and women, although the strongest association was found amongst current smoking men and former smoking women. Current smoking men may both lose more weight and have a higher mortality rate than never and former smoking men, and our results confirm findings by others [11]. The increased mortality amongst former smoking women who lost weight could be a result of ‘confounding by indication’, i.e. some women may have received information on high blood pressure and/or high cholesterol levels and thus ceased smoking, but still remained at a higher risk for dying.

Weight loss was also associated with increased mortality within all levels of leisure time physical activity, and none of the associations were markedly different between the activity strata. Hence, effect modification by initial physical activity level is not likely to be present. To our knowledge, this has not been explored in any other studies.

Previous studies have found that weight gain is associated with increased mortality [11, 42], but contradictory to these studies we found that people who gained weight between the studies had similar mortality rates as those who were weight stable. One might speculate that weight gain amongst generally healthy people is not linked to increased mortality. Moreover a relatively short follow-up time in this study may have contributed to this finding, as the effect of weight gain on mortality is likely to be a relatively slow process? It is likely that future analyses based on HUNT-data with longer follow-up time included may answer this question.

Previous studies have shown that pre-existing disease is associated with both weight loss and mortality [24, 43, 44]. In our study we excluded participants who reported diabetes or cardiovascular disease at the first study, and also those who reported a history of cancer at the second survey. Additionally, adjustment for diabetes, cardiovascular disease, chronic obstructive pulmonary disease and asthma at the second survey did not reduce the increased mortality rate associated with weight loss. In an attempt to assess the potential importance of undiagnosed disease at baseline, we excluded the first 3 years of follow-up, but the results still remained similar as in the overall analysis. This method has also been applied in other studies [45, 46], but the validity of this method is not entirely agreed upon. In a meta-analysis, Allison et al. [47] supported not to exclude early deaths in BMI-mortality studies.

The strengths of our study includes the high numbers of participants both in women and men, the wide age-range, information on a large number of potential confounders, the standardized measurements of height and weight, and the linkage of the data to the national Registry of Death at Statistics Norway ensuring complete follow-up on vital status. The potential bias due to misclassification in the death certificates is unlikely to be related to different levels of change in body weight and thus to explain the results. Additionally, the prospective design of the study makes it unlikely that the results may be biased due to selection of participants or differential misclassification of information.

Although undiagnosed disease is probably the most plausible explanation for the observed increase in mortality amongst people who lost weight, one may speculate in other explanations for the finding. One possible mechanism may be that a reduction in weight initiates a stress response in the body, and it has been shown that more biological mechanisms are involved in prevention of weight loss than of weight gain [48] .

In conclusion, this study has shown that weight loss, but not weight gain, is associated with increased mortality rates in apparently healthy men and women, compared with people with a stable weight. This finding was similar in analysis of both total mortality, CV mortality, and non-CV mortality. Stratified analysis indicates that this effect may be modified by initial BMI and smoking status. The most plausible explanation for this finding is the existence of undiagnosed disease, although the consistent results between different causes of death may suggest a role for other potential mechanisms.

Acknowledgements

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References

The Nord-Trøndelag Health Study is collaboration between HUNT Research Centre, Department of Public Health and General Practice, Norwegian University of Science and Technology, the Norwegian Institute of Public Health and Nord-Trøndelag County.

References

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Materials and methods
  5. Study population
  6. Follow-up
  7. Body mass index
  8. Leisure time physical activity
  9. Smoking status
  10. Statistical analyses
  11. Ethics
  12. Results
  13. Discussion
  14. Conflict of interest statement
  15. Acknowledgements
  16. References
  • 1
    Maru S, van der Schouw YT, Gimbrere CH, Grobbee DE, Peeters PH. Body mass index and short-term weight change in relation to mortality in Dutch women after age 50 y. Am J Clin Nutr 2004; 80: 2316.
  • 2
    Obesity: preventing and managing the global epidemic. report of a WHO consultation. World Health Organ Tech Rep Ser 2000; 894: i–XII 1253.
  • 3
    Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CW Jr. Body-mass index and mortality in a prospective cohort of US adults. N Engl J Med 1999; 341: 1097105.
  • 4
    Baik I, Ascherio A, Rimm EB et al. Adiposity and mortality in men. Am J Epidemiol 2000; 152: 26471.
  • 5
    Bray GA. Medical consequences of obesity. J Clin Endocrinol Metab 2004; 89: 25839.
  • 6
    Larsson UE. Influence of weight loss on pain, perceived disability and observed functional limitations in obese women. Int J Obes 2004; 28: 26977.
  • 7
    Gortmaker SL, Must A, Perrin JM, Sobol AM, Dietz WH. Social and economic consequences of overweight in adolescence and young adulthood. N Engl J Med 1993; 329: 100812.
  • 8
    Must A, Spadano J, Coakley EH, Field AE, Colditz G, Dietz WH. The disease burden associated with overweight and obesity. JAMA 1999; 282: 15239.
  • 9
    Torgerson JS, Lindroos AK, Naslund I, Peltonen M. Gallstones, gallbladder disease, and pancreatitis: cross-sectional and 2-year data from the Swedish Obese Subjects (SOS) and SOS reference studies. Am J Gastroenterol 2003; 98: 103241.
  • 10
    Rocchini AP. Obesity hypertension. Am J Hypertens 2002; 15(Pt 2): 50S2S.
  • 11
    Rosengren A, Wedel H, Wilhelmsen L. Body weight and weight gain during adult life in men in relation to coronary heart disease and mortality. A prospective population study. Eur Heart J 1999; 20: 26977.
  • 12
    Calle EE, Thun MJ. Obesity and cancer. Oncogene 2004; 23: 636578.
  • 13
    Seidell JC. Obesity, insulin resistance and diabetes – a worldwide epidemic. Br J Nutr 2000; 83(Suppl. 1): S58.
  • 14
    Bacon SL, Sherwood A, Hinderliter A, Blumenthal JA. Effects of exercise, diet and weight loss on high blood pressure. Sports Med 2004; 34: 30716.
  • 15
    Poobalan A, Aucott L, Smith WC et al. Effects of weight loss in overweight/obese individuals and long-term lipid outcomes – a systematic review. Obes Rev 2004; 6: 4350.
  • 16
    Aucott L, Poobalan A, Smith WC et al. Weight loss in obese diabetic and non-diabetic individuals and long-term diabetes outcomes – a systematic review. Diabetes Obes Metab 2004; 6: 8594.
  • 17
    Sjøstrom CD, Peltonen M, Wedel H, Sjøstrom L. Differentiated long-term effects of intentional weight loss on diabetes and hypertension. Hypertension 1999; 7: 205.
  • 18
    Higgins M, D'Agostino R, Kannel W, Cobb J, Pinsky J. Benefits and adverse effects of weight loss. Observations from the Framingham Study. Ann Intern Med 1993; 119(Pt 2): 75863.
  • 19
    Peters ET, Seidell JC, Menotti A et al. Changes in body weight in relation to mortality in 6441 European middle-aged men: the Seven Countries Study. Int J Obes 1995; 19: 8628.
  • 20
    Nilsson PM. The enigma of increased non-cancer mortality after weight loss in healthy men who are overweight or obese. J Intern Med 2002; 252: 708.
  • 21
    Mikkelsen KL, Heitmann BL, Keiding N, Sørensen TI. Independent effects of stable and changing body weight on total mortality. Epidemiology 1999; 10: 6718.
  • 22
    Lee IM, Paffenbarger RS Jr. Change in body weight and longevity. JAMA 1992; 268: 20459.
  • 23
    Sørensen TI. Weight loss causes increased mortality: pros. Obes Rev 2003; 4: 37.
  • 24
    Pamuk ER. Weight loss and subsequent death in a cohort of US adults. Ann Intern Med 1993; 119(Pt 2): 7448.
  • 25
    Williamson DF. The association between weight loss and increased longevity. A review of the evidence. Ann Intern Med 1993; 119(Pt 2): 7316.
  • 26
    Yang D. Weight loss causes increased mortality: cons. Obes Rev 2003; 4: 916.
  • 27
    Blair SN, Shaten J, Brownell K, Collins G, Lissner L. Body weight change, all-cause mortality, and cause-specific mortality in the Multiple Risk Factor Intervention Trial. Ann Intern Med 1993; 119(Pt 2): 74957.
  • 28
    Wannamethee G, Shaper AG. Weight change, perceived health status and mortality in middle-aged British men. Postgrad Med J 1990; 44: 9103.
  • 29
    Nilsson PM, Nilsson JA, Hedblad B, Berglund G, Lindgarde F. The enigma of increased non-cancer mortality after weight loss in healthy men who are overweight or obese. J Intern Med 2002; 252: 708.
  • 30
    Yaari S, Goldbourt U. Voluntary and involuntary weight loss: associations with long term mortality in 9,228 middle-aged and elderly men. Am J Epidemiol 1998; 148: 54655.
  • 31
    Williamson DF, Thompson TJ, Thun M, Flanders D, Pamuk E, Byers T. Intentional weight loss and mortality among overweight individuals with diabetes. Diabetes Care 2000; 23: 1499504.
  • 32
    Williamson DF. Intentional weight loss: patterns in the general population and its association with morbidity and mortality. Int J Obes 1997; 21(Suppl. 1): S149.
  • 33
    Gregg EW. Intentional weight loss and death in overweight and obese US adults 35 years of age and older. Ann Intern Med 2003; 138: 3839.
  • 34
    Fontaine KR, Allison DB. Does intentional weight loss affect mortality rate? Eat Behav 2001; 2: 8795.
  • 35
    Hjartaker A, Laake P, Lund E. Body mass index and weight change attempts among adult women. The Norwegian Women and Cancer Study. Eur J Public Health 2001; 11: 1416.
  • 36
    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: 130614.
  • 37
    Wannamethee SG, Shaper AG, Whincup PH, Walker M. Characteristics of older men who lose weight intentionally or unintentionally. Am J Epidemiol 2000; 151: 66775.
  • 38
    Heitmann BL, Garby L. Composition (lean and fat tissue) of weight changes in adult Danes. Am J Clin Nutr 2002; 75: 8407.
  • 39
    Grace SL, Fry R, Cheung A, Stewart DE. Cardiovascular disease. BMC Womens Health 2004; 4(Suppl. 1): S15.
  • 40
    Schnohr C, HOjbjerre L, Riegels M et al. Does educational level influence the effects of smoking, alcohol, physical activity, and obesity on mortality? A prospective population study. Scand J Public Health 2004; 32: 2506.
  • 41
    Wannamethee SG, Shaper AG, Walker M. Weight change, body weight and mortality: the impact of smoking and ill health. Int J Epidemiol 2001; 30: 77786.
  • 42
    Fulton JE, Shekelle RB. Cigarette smoking, weight gain, and coronary mortality: results from the Chicago Western Electric Study. Circulation 1997; 96: 143844.
  • 43
    Newman AB, Yanez D, Harris T, Duxbury A, Enright PL, Fried LP. Weight change in old age and its association with mortality. J Am Geriatr Soc 2001; 49: 130918.
  • 44
    Wannamethee SG, Shaper AG, Walker M. Weight change, weight fluctuation, and mortality. Arch Intern Med 2002; 162: 257580.
  • 45
    Williamson DF, Pamuk E, Thun M, Flanders D, Byers T, Heath C. Prospective study of intentional weight loss and mortality in never-smoking overweight US white women aged 40–64 years. Am J Epidemiol 1999; 7: 112841.
  • 46
    Stevens J, Juhaeri, Cai J. Changes in body mass index prior to baseline among participants who are ill or who die during the early years of follow-up. Am J Epidemiol 2001; 153: 94653.
  • 47
    Allison DB, Faith MS, Heo M, Townsend-Butterworth D, Williamson DF. Meta-analysis of the effect of excluding early deaths on the estimated relationship between body mass index and mortality. Obes Res 1999; 7: 34254.
  • 48
    Blundell JE, King NA. Overconsumption as a cause of weight gain: behavioural-physiological interactions in the control of food intake (appetite). Ciba Found Symp 1996; 201: 13854; discussion p. 15.