• employment;
  • longitudinal;
  • overweight;
  • outcomes


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Objective: To study BMI and change in BMI from age 25 as predictors of sickness absence.

Research Methods and Procedures: Data were collected from 2564 women and 5853 men, who were British civil servants (35 to 55 years) on entry to the Whitehall II study (Phase 1, 1985 to 1988). Employer's records provided annual medically certified (long, >7 days) and self-certified (short, 1 to 7 days) spells of sickness absence. BMI at age 25 and Phase 1 were examined in relation to absences from Phase 1 to the end of 1998 (mean follow-up, 7.0 years).

Results: After adjustment for employment grade, health-related behaviors, and health status, overweight (BMI = 25.0 to 29.9 kg/m2) and obesity (BMI > 30.0 kg/m2) at Phase 1 were significant predictors of short and long absences in both sexes; rate ratios (95% confidence intervals) ranged from 1.13 (1.05 to 1.21) to 1.51 (1.30 to 1.76) compared with a BMI of 21.0 to 22.9 kg/m2. Additionally, a BMI of 23.0 to 24.9 kg/m2 at Phase 1 predicted long absences in women, and underweight (BMI < 21.0 kg/m2) predicted short absences in men. Obesity at age 25 predicted long absences, and obesity at Phase 1 predicted short and long absences in both sexes. Chronic obesity was a particularly strong predictor of long absences in men, with a rate ratio of 2.61 (1.88 to 3.63).

Discussion: Findings from this well-characterized cohort suggest that the obesity epidemic in industrialized countries may result in significant increases in sickness absence. Further research is needed to determine the underlying mechanisms. Policy to reduce sickness absence needs to tackle the problem of excess weight in the working population.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Population-wide increases in obesity and overweight in post-industrial countries have occurred rapidly over the last 20 years, and obesity is now more prevalent than undernutrition in some less developed countries (1). Obesity and overweight increase mortality and have been associated with a range of chronic diseases (1)(2)(3). Because the association between BMI and mortality risk is U-shaped (2), underweight and weight change are also predictive of premature mortality and disease (4)(5)(6)(7). In addition to effects on physical health, weight and weight change are known determinants of mental health and well being (8)(9)(10). Another issue that has attracted attention over the last 20 years is the cost of absence from work because of sickness, with concern expressed about the level of absence in a number of European and Scandinavian countries (11)(12). Even in the United Kingdom, where absence rates are relatively low, there is concern about long-term sickness absence and high levels of absence in the public sector (13).

Existing studies of obesity, overweight, and sickness absence are relatively few and have produced conflicting results. Some studies report a considerable excess risk of sickness absence for obese and overweight employees (14)(15)(16)(17)(18)(19)(20)(21), whereas others report no excess risk (22)(23)(24)(25). However, apart from work site intervention studies that were unable to detect associations at the individual level, no studies seem to have investigated associations between BMI measured on more than one occasion and sickness absence.

The purpose of this study is to examine BMI and change in BMI from age 25 as predictors of sickness absence. Data are from the Whitehall II study, a prospective cohort of white collar women and men employed in the British Civil Service.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Study Population

The target population for the Whitehall II study was all London-based office staff, 35 to 55 years of age, working in 20 civil service departments. With a response rate of 73%, the final cohort consisted of 10,308 participants: 3413 women and 6895 men (26). The true response rate was higher, however, because ∼4% of those invited were not eligible for inclusion. Although mostly white collar, participants covered a wide range of grades from office support to permanent secretary.


Baseline screening (Phase 1) of the Whitehall II cohort took place between late 1985 and early 1988. It involved a clinical examination and a self-administered questionnaire. Sickness absence records were collected from Phase 1 to the end of 1998. For each participant, the follow-up period was time at risk of sickness absence while in the Civil Service and did not include time after leaving the Civil Service. Thus, the range of follow-up was from 0.03 to 13.3 years, with a mean of 7.0 ± 3.8 (standard deviation) years of follow-up.


Weight at age 25 was derived from the Phase 1 questionnaire using the question “How much did you weigh at the age of 25 (approximately)?” Responses given in stones and pounds were converted to kilograms. Weight in kilograms and height in meters were recorded during the clinical screening examination at Phase 1. Weight was measured with all items of clothing removed except underwear. An electronic Soehnle scale (Leifheit AG, Nassau, Germany) with a digital readout was used to read weight to the nearest 0.1 kg. If the reading alternated between two readings >0.1 kg apart, the higher reading was recorded. Height was measured to the nearest millimeter using a stadiometer with the participant standing completely erect with the head in the Frankfort plane (27). BMI at age 25 and Phase 1 was calculated as weight divided by height squared. BMI categories <21, 21.0 to 22.9, 23.0 to 24.9, 25.0 to 29.9, and ≥30 kg/m2 were selected to include the World Health Organization cut-points for overweight (BMI = 25 to 29.9 kg/m2) and obesity (BMI ≥ 30 kg/m2). BMI at age 25 was used to define participants who were overweight or obese at age 25.

Sickness Absence Measures

Computerized sickness absence records were obtained from Civil Service pay centers (28). These records included the first and last dates of all absences. For absences of 7 calendar days or less, civil servants complete their own certificate, whereas for absences >7 days, a medical certificate is needed. Sickness absence records were checked for inconsistencies, and any duplicate spells were removed. Spells of sickness that were either consecutive or overlapped were merged into a single spell of absence. Public holidays and weekend days were ignored when identifying and merging consecutive spells. For each employee, the number of sickness absence spells was ascertained and the rate of sickness absence calculated as number of spells per year. Because previous studies have shown that short and long sickness absences have different determinants, annual rates of short (self-certified, 1 to 7 days) spells and long (medically certified, >7 days) spells were calculated and analyzed separately (18). Consent to access sickness absence records was given by 93% (9564) of participants. In 96% (9179) of these participants, record linkage was successful. Earlier work using the Whitehall II sickness absence data has shown that there is a strong association between medically certified spells of absence and ill-health and that ill-health is also an important determinant of short spells of absence (29).


Weight and sickness absence are determined by socio-demographic factors and health-related behaviors (30)(31)(32)(33)(34)(35). Similarly, mental and physical ill-health is unevenly distributed across the adult weight range and is a major determinant of sickness absence (1)(36). Thus, the following covariates were assessed.

Socio-demographic Factors

Age and employment grade were derived from the baseline questionnaire. Employment grade was divided into three categories: administrative, professional/executive, and clerical/support, of which administrative is the highest.

Health-related Behaviors

Alcohol consumption was categorized into three levels (none, moderate, and heavy). Heavy drinking was defined as alcohol over the recommended limits for safe drinking used in the United Kingdom General Household Survey; ≥15 units/wk for women and ≥22 units/wk for men (37). Prevalence of smoking was defined as current smokers using either manufactured or hand-rolled cigarettes. Analyses compared never-smokers, ex-smokers, and smokers, with adjustment for prevalence and the number of cigarettes smoked per day by the smokers at baseline. Questions on physical activity were derived from the Whitehall Study, which classified activity into mild, moderate, and vigorous activity on the basis of energy use. Participants who did at least 1 hour of vigorous activity per week (running, digging, tennis) were classified into the vigorous exercise category (38).

Comorbidity Measures

Mental health, measured as minor psychiatric disorders, was assessed using the 30-item General Health Questionnaire (GHQ)1 (39). Physical health, measured as presence of longstanding illness, was ascertained using the General Household Survey longstanding illness question (40). Unlike other measures, there was considerable lack of information for the measure of longstanding illness because it was introduced after the start of the baseline survey. Where baseline longstanding illness data were missing, values from the subsequent survey (1989/1990) were used.

The three health-related behaviors and two comorbidity measures used as covariates in this study have been shown to be determinants of well-validated general measures of health among participants in the Whitehall II study (41)(42)(43). The GHQ-30 has been validated against the Clinical Interview Schedule in the Whitehall II data, giving a cut-off point of 4/5 for dividing non-cases from GHQ cases. Participants scoring above the predetermined threshold are at higher risk of minor psychiatric morbidity, largely depression and anxiety disorders (44).

Data Analysis

We determined the absence rates per 100 person-years for short and long spells of sickness absence for the five categories of BMI at Phase 1 and the corresponding rate ratios [95% confidence intervals (CIs)] using Poisson regression analysis, corrected for overdispersion (45). The absence rate per 100 person-years refers to the number of sickness absence spells that 100 participants produce per year. Poisson regression, with the BMI category of 21.0 to 22.9 kg/m2 as the reference category, was used as the method of choice because it takes variation in follow-up periods between participants into account (log follow-up entered as an offset variable) and is suitable when the outcome event (a spell of sickness absence) can occur more than once in each subject. The rates of short and long spells of sickness absence for participants either overweight or obese at age 25 by BMI category at Phase 1 were calculated using the same method. The reference category was BMI <25 kg/m2 at age 25 and Phase 1 for the analyses of overweight over time and <30 kg/m2 for the analyses of obesity over time. Rate ratios were adjusted for age group (35 to 39, 40 to 44, 45 to 49, and 50 to 55 years), employment grade, health-related behaviors, minor psychiatric morbidity, and longstanding illness. An interaction term, obesity at age 25 × age at Phase 1, was included in regression models that already contained the main effects to test whether the effects of obesity at age 25 on short and long spells of sickness absence were dependent on the time difference between age 25 and age at Phase 1. The sexes were analyzed separately, and all analyses were conducted using the SAS 8.2 program.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Of the 10,308 participants in the baseline screening, 81.7% (2564 women and 5853 men) were included in the analyses for this paper. Eighty-seven percent of these participants were white, 74% were married or cohabiting, and 65% were ≥17 years of age when they left full-time education. The majority of those lost to follow-up were missing data either on the baseline covariates or on sickness absence. One hundred eighty-six participants at age 25 and 14 at Phase 1 had no measure of BMI (Figure 1). Those excluded from the Phase 1 BMI analyses were more likely to be women (38% vs. 29%), in the lower employment grades (26% vs. 20%), and older (45.0 vs. 43.9 years).


Figure 1. Participants lost to follow-up.

Download figure to PowerPoint

We analyzed 254,514 sickness absence days within 31,700 short spells and 4885 long spells for women. The corresponding figures were 268,301 days, 47,306 short spells, and 4936 long spells for men. As expected, the overall absence rates for short and long spells were higher for women than men (Table 1).

Table 1.  Sample characteristics
 Women (n = 2564)Men (n = 5853)
  1. SD, standard deviation; GHQ, General Health Questionnaire. Values are n (%) unless otherwise stated.

Age (yrs) [mean (SD)]45.3 (6.1)44.1 (6.0)
Age group (yrs)  
 35 to 39595 (23.2)1682 (28.7)
 40 to 44595 (23.2)1575 (26.9)
 45 to 49562 (21.9)1157 (19.8)
 50 to 55812 (31.7)1439 (24.6)
 Administrative293 (11.4)2263 (38.7)
 Professional/executive1067 (41.6)3094 (52.9)
 Clerical/support1204 (47.0)496 (8.5)
 Never1367 (53.3)2775 (47.4)
 Ex623 (24.3)2148 (36.7)
 Current574 (22.4)930 (15.9)
 None774 (30.2)748 (12.8)
 Moderate1530 (59.7)4018 (68.7)
 Heavy260 (10.1)1087 (18.6)
 Mild1105 (43.1)1376 (23.5)
 Moderate1198 (46.7)2945 (50.3)
 Vigorous261 (10.2)1532 (26.2)
GHQ case  
 No1772 (69.1)4386 (74.9)
 Yes792 (30.9)1467 (25.1)
Longstanding illness  
 No1696 (66.1)3951 (74.9)
 Yes868 (33.9)1902 (25.1)
BMI at age 25 (kg/m2) [mean (SD)]21.3 (3.3)22.0 (2.7)
BMI at Phase 1 (kg/m2) [mean (SD)]24.7 (4.3)24.6 (3.1)
BMI at Phase 1 (kg/m2)  
 <21417 (16.3)582 (9.9)
 21.0 to 22.9566 (22.1)1205 (20.6)
 23.0 to 24.9597 (23.3)1744 (29.8)
 25.0 to 29.0715 (27.9)2022 (34.5)
 ≥30269 (10.5)300 (5.1)
Rate of short absence spells per 100 person-years181.7111.0
Rate of long absence spells per 100 person-years28.011.6

The mean BMI at Phase 1 by sex for each of the baseline covariates is shown in Table 2. Age, employment grade, and all of the health-related behaviors were associated with Phase 1 BMI in both sexes, with the exception of exercise in men. GHQ caseness was not associated with BMI in either sex, but longstanding illness was associated with BMI in both sexes. All of the covariates were associated with short and long spells of sickness absence (Table 3), with the exception of age and short spells in men.

Table 2.  Age-adjusted* means of Phase 1 BMI by covariates
 Women (n = 2564)Men (n = 5853)
  • GHQ, General Health Questionnaire.

  • *

    All analyses are age-adjusted except for age.

  • All analyses except obesity at age 25 (women = 2510, men = 5721).

Age group (yrs)    
 35 to 3959523.6168224.1
 40 to 4459524.1157524.4
 45 to 4956225.0115724.9
 50 to 5581225.8143925.1
 p value (for heterogeneity) p < 0.0001 p < 0.0001
 p value (for heterogeneity) p = 0.0002 p < 0.0001
 p value (for heterogeneity) p = 0.0002 p < 0.0001
 p value (for heterogeneity) p = 0.0009 p < 0.0001
 p value p = 0.003 p = 0.41
GHQ case    
 p value p = 0.63 p = 0.34
Longstanding illness    
 p value p = 0.0004 p = 0.02
Obesity (BMI ≤ 30 kg/m2) at age 25    
 p value p < 0.0001 p < 0.0001
Table 3.  Covariates as predictors of sickness absence
  Women Men
  1. GHQ, General Health Questionnaire. Values are rate ratios (95% confidence interval). All analyses are age-adjusted except for age.

 NShort spellsLong spellsNShort spellsLong spells
Age group (yrs)      
 35 to 395951.00 (reference)1.00 (reference)16821.00 (reference)1.00 (reference)
 40 to 445951.06 (0.97 to 1.17)1.27 (1.11 to 1.46)15750.95 (0.89 to 1.02)1.14 (1.03 to 1.26)
 45 to 495621.13 (1.03 to 1.24)1.43 (1.24 to 1.63)11570.95 (0.88 to 1.03)1.17 (1.04 to 1.31)
 50 to 558121.08 (0.99 to 1.19)1.66 (1.46 to 1.88)14390.98 (0.91 to 1.07)1.43 (1.27 to 1.59)
 Administrative2931.00 (reference)1.00 (reference)22631.00 (reference)1.00 (reference)
 Professional/executive10672.15 (1.85 to 2.49)2.16 (1.73 to 2.69)30941.96 (1.83 to 2.08)1.93 (1.75 to 2.12)
 Clerical/support12042.74 (2.37 to 3.17)3.49 (2.81 to 4.34)4963.75 (2.46 to 4.06)4.21 (3.75 to 4.72)
 Never13671.00 (reference)1.00 (reference)27751.00 (reference)1.00 (reference)
 Ex6231.07 (0.98 to 1.16)1.08 (0.97 to 1.21)21481.10 (1.04 to 1.17)1.16 (1.06 to 1.27)
 Current5741.15 (1.06 to 1.25)1.28 (1.15 to 1.43)9301.48 (1.38 to 1.59)1.76 (1.59 to 1.94)
 Non-drinkers7741.15 (1.07 to 1.23)1.35 (1.22 to 1.48)7481.29 (1.19 to 1.39)1.50 (1.35 − 1.66)
 Moderate drinkers15301.00 (reference)1.00 (reference)40181.00 (reference)1.00 (reference)
 Heavy drinkers2600.81 (0.71 to 0.91)0.76 (0.63 to 0.91)10871.11 (1.04 to 1.20)1.08 (0.98 to 1.20)
 Mild11051.00 (0.93 to 1.07)1.17 (1.07 to 1.28)13761.18 (1.11 to 1.26)1.29 (1.18 to 1.42)
 Moderate11981.00 (reference)1.00 (reference)29451.00 (reference)1.00 (reference)
 Vigorous2610.85 (0.75 to 0.95)0.86 (0.82 to 1.02)15321.01 (0.95 to 1.08)1.02 (0.92 to 1.12)
GHQ case      
 No17721.00 (reference)1.00 (reference)43861.00 (reference)1.00 (reference)
 Yes7921.10 (1.01 to 1.17)1.11 (1.00 to 1.22)14671.21 (1.14 to 1.28)1.31 (1.20 to 1.43)
Longstanding illness      
 No16961.00 (reference)1.00 (reference)39511.00 (reference)1.00 (reference)
 Yes8681.25 (1.17 to 1.34)1.52 (1.39 to 1.66)19021.28 (1.21 to 1.36)1.68 (1.56 to 1.82)
Obesity (BMI ≥ 30 kg/m2) at age 25      
 No24541.00 (reference)1.00 (reference)56431.00 (reference)1.00 (reference)
 Yes560.84 (0.66 to 1.07)1.19 (0.91 to 1.57)781.42 (1.14 to 1.76)2.04 (1.57 to 2.65)

Short Spells of Sickness Absence

Overweight and obesity in both sexes and underweight (BMI < 21 kg/m2) in men at Phase 1 were associated with an increased risk of short absences (Table 4). These associations were attenuated on adjustment for employment grade but remained statistically significant. Further adjustment for health-related behavior and the mental and physical health indicators produced little further attenuation.

Table 4.  BMI at Phase 1 as a predictor of short spells and long spells of sickness absence*
   Rate ratios (95% CI) adjusted for
  • CI, confidence interval; GHQ, General Health Questionnaire.

  • *

    All units of absence are spells; short refers to the number of self-certified spells of 1 to 7 days and long to the number of medically certified spells of >7 days.

  • Health behaviors: smoking, alcohol, exercise.

 NAbsence rate per 100 person yearsAgeAge and gradeAge, grade, health behaviors, GHQ, and longstanding illness
Short spells of sickness absence     
  <21 kg/m2417164.71.01 (0.91 to 1.13)1.04 (0.93 to 1.15)1.03 (0.93 to 1.15)
  21.0 to 22.9 kg/m2566164.21.00 (reference)1.00 (reference)1.00 (reference)
  23.0 to 24.9 kg/m2597176.81.08 (0.97 to 1.19)1.03 (0.94 to 1.14)1.02 (0.93 to 1.13)
  25.0 to 29.0 kg/m2715201.61.22 (1.11 to 1.35)1.16 (1.06 to 1.28)1.15 (1.05 to 1.26)
  ≥30 kg/m2269208.11.27 (1.12 to 1.43)1.18 (1.05 to 1.33)1.15 (1.02 to 1.29)
  <21 kg/m2582118.81.18 (1.06 to 1.31)1.10 (1.00 to 1.21)1.10 (1.00 to 1.21)
  21.0 to 22.9 kg/m21205100.61.00 (reference)1.00 (reference)1.00 (reference)
  23.0 to 24.9 kg/m21744102.21.02 (0.94 to 1.11)1.03 (0.96 to 1.11)1.03 (0.95 to 1.11)
  25.0 to 29.0 kg/m22022118.41.18 (1.10 to 1.28)1.13 (1.05 to 1.21)1.13 (1.05 to 1.21)
  ≥30 kg/m2300146.71.47 (1.29 to 1.66)1.21 (1.08 to 1.36)1.19 (1.06 to 1.33)
Long spells of sickness absence     
  <21 kg/m241719.80.94 (0.80 to 1.11)0.98 (0.84 to 1.15)0.97 (0.83 to 1.14)
  21.0 to 22.9 kg/m256621.71.00 (reference)1.00 (reference)1.00 (reference)
  23.0 to 24.9 kg/m259727.51.24 (1.08 to 1.43)1.18 (1.03 to 1.36)1.17 (1.02 to 1.34)
  25.0 to 29.0 kg/m271534.31.50 (1.32 to 1.72)1.41 (1.24 1.60)1.39 (1.22 to 1.57)
  ≥30 kg/m226940.31.75 (1.50 to 2.06)1.60 (1.37 to 1.87)1.51 (1.30 to 1.76)
  <21 kg/m258211.31.16 (0.99 to 1.35)1.07 (0.92 to 1.24)1.04 (0.90 to 1.21)
  21.0 to 22.9 kg/m212059.81.00 (reference)1.00 (reference)1.00 (reference)
  23.0 to 24.9 kg/m2174410.81.07 (0.95 to 1.21)1.08 (0.96 to 1.21)1.08 (0.96 to 1.20)
  25.0 to 29.0 kg/m2202212.31.21 (1.08 to 1.36)1.15 (1.03 to 1.28)1.16 (1.04 to 1.29)
  ≥30 kg/m230019.41.91 (1.61 to 2.25)1.53 (1.30 to 1.80)1.49 (1.27 to 1.75)

Long Spells of Sickness Absence

Overweight and obesity in both sexes and a BMI of 23 to 24.9 kg/m2 in women at Phase 1 were associated with an increased risk of long absences (Table 4). These associations were somewhat attenuated on adjustment for employment grade but remained statistically significant in the fully adjusted model.

BMI, weight gain, and sickness absence are heavily socially patterned, so we were concerned that our findings may not apply across all employment grades. To examine this, we repeated the analyses in a dataset stratified by employment grade. Women and men had to be combined to provide sufficient power. In these analyses, associations of overweight and obesity in the professional/executive grades with short spells of absence, as well as associations of obesity in the professional/executive grades and any BMI of 23 kg/m2 or more in the clerical/support grades with long spells of absence, were statistically significant. Other associations did not reach statistical significance; however, the pattern of associations between BMI at Phase 1 and both short and long absences were similar across the three grade categories, even though there was a strong grade gradient in the absolute absence rates (data available on request).

Obesity and Overweight at Age 25

As shown in Table 5, obesity at Phase 1 but not at age 25 was associated with a 1.23-fold excess risk of short spells of absence in women. In men, obesity at age 25 and Phase 1 was associated with rate ratios of 1.26 and 1.29, respectively, for short absences. Additionally, in the relatively few men with chronic obesity, there was a 1.58-fold excess risk of short absences, indicating a cumulative effect over the two time-points. Obesity at age 25 and Phase 1 was associated with rate ratios for long spells of absence of between 1.50 and 1.60 in both sexes. However, chronic obesity was only associated with long spells of absence in men. The effect of being obese at age 25 and Phase 1 was multiplicative, with a rate ratio of 2.61.

Table 5.  Obesity* at age 25 and Phase 1 as a predictor of short and long spells of sickness absence
  Short spellsLong spells
  • CI, confidence interval.

  • *

    BMI ≥ 30 kg/m2.

  • All units of absence are spells; short refers to the number of self-certified spells of 1 to 7 days, and long refers to the number of medically certified spells of >7 days.

  • At age 35 to 55.

  • §

    Adjusted for age.

Obesity at age 25/obesity at Phase 1NAbsence rate/100 person yearsRate ratio (95% CI)§Absence rate/100 person yearsRate ratio (95% CI)§
 No/no2234178.61.00 (reference)26.21.00 (reference)
 No/yes220220.51.23 (1.10 to 1.37)42.01.50 (1.31 to 1.72)
 Yes/no16187.41.05 (0.71 to 1.55)43.11.59 (1.04 to 2.45)
 Yes/yes40136.20.77 (0.57 to 1.04)28.41.09 (0.77–1.55)
 No/no5393108.41.00 (reference)11.01.00 (reference)
 No/yes247136.21.26 (1.11 to 1.43)18.01.60 (1.36 to 1.88)
 Yes/no35139.11.29 (0.94 to 1.77)17.31.58 (1.04 to 2.41)
 Yes/yes43171.81.58 (1.18 to 2.12)28.72.61 (1.88 to 3.63)

On including an obesity at age 25 × age at Phase 1 interaction term in regression models that already contained the main effects, we found that the effect of obesity at age 25 was not dependent on the time between age 25 and age at Phase 1, which varied from 10 to 30 years. For short spells, the p value for the interaction term was p = 0.56 in women and 0.89 in men. For long spells, the corresponding p values were 0.90 and 0.58, respectively.

Overweight (BMI = 25 to 29.9 kg/m2) at age 25 added nothing to the prediction of sickness absence beyond the contribution of being overweight or obese at Phase 1. For example, compared with participants who were not overweight either at age 25 or Phase 1, the rate ratio for long spells of absence among men obese at Phase 1 but <25 kg/m2 at age 25 was 1.82 (95% CI, 1.42 to 2.35) and that for men obese at Phase 1 but overweight at age 25 was 1.79 (95% CI, 1.49 to 2.14); data available on request.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This seems to be the first large-scale observational study to examine associations between BMI measured on more than one occasion and sickness absence. It shows that obesity (BMI ≥ 30 kg/m2) either at age 25 or Phase 1 (age 35 to 55) increases long spells of sickness absence by ∼60% in both sexes and increases short spells of absence in men by 25%. The effects of obesity at both age 25 and Phase 1 had a multiplicative effect on the risk of long spells of sickness in men but not in women. However, being overweight (BMI = 25 to 29.9 kg/m2) at age 25 added nothing to the prediction of sickness absence beyond the contribution of being overweight or obese at Phase 1. Our findings confirm those of previous studies that have shown independent associations of overweight and obesity with higher rates of short and long spells of absence in both sexes. In addition, we showed that a BMI of 23.0 to 24.9 kg/m2 at Phase 1 was a significant predictor of long absences in women and underweight was a significant predictor of short absences in men.

Little previous work has examined associations between BMI over time and sickness absence. Contrary to our findings, a Finnish study of the effect of weight change between the ages of 14 and 31 on self-reported work ability observed no effects among men but found that work ability was significantly reduced among women who gained weight and women who were overweight at both time-points (46). Our findings of an association of overweight and obesity with higher absence rates confirm those of some previous studies (14)(15)(16)(17)(18)(19)(20)(21). A small scale, longitudinal study of 185 student nurses followed for 33 months found a significant curvilinear association between relative weight and recorded absence, after adjustment for smoking (15). In this study, U-shaped relationships between BMI at Phase 1 and spells of sickness absence reflect the U-shaped association between BMI and risk of death (2)(7). Associations of overweight and obesity with short spells of sickness absence in this study were weaker than those for long spells. One way of looking at sickness absence is as a measure that integrates decrements in social, psychological, and physical functioning. Short spells of absence are more likely to represent decrements in psychological and social functioning; long spells are more likely to represent decrements in physical functioning or real illness. A number of previous studies that have failed to find an independent association between BMI and sickness absence have either used retrospective self-reported sickness absence data (22), failed to control for smoking, (23), or have shown associations with abdominal obesity rather than BMI (24)(25).

We were able to adjust for a number of potential mediating factors, such as physical ill-health (1), and confounding factors, such as smoking (30)(31). However, with the exception of employment grade, adjustment for these factors had little effect on the observed associations. A number of existing studies have specifically adjusted for blood pressure or hypertension or excluded known hypertensives from the analyses (14)(20)(25). Adjusting for longstanding illness will partially account for the effect of hypertension, but we felt further adjustment for blood pressure would be an overadjustment, because 77% of the hypertension in obese people has been shown to be attributable to obesity (47).

This study was carried out in a well-characterized cohort that had the benefits of a large sample size, good response rate, two of the three anthropometric measures recorded during clinical screening, and use of sickness absence data from the employer's registers. Use of the World Health Organization definitions of overweight and obesity permit meaningful comparisons with findings from other studies and facilitate the identification of individuals and groups at risk and priorities for intervention at an individual or community level.

One of the main limitations is that our measure of weight at age 25 was retrospective and self-reported. Self-reports of weight are not as accurate as measured values, with studies showing that heavier individuals tend to underestimate weight, thereby underestimating obesity (48). However, the agreement between measured and reported weight is generally high (49). In our data, the mean difference between self-reported weight and weight measured in the clinic at Phase 1 was −0.7 ± 3.5 (standard deviation) kg, and for 93% of participants, the discrepancy was <3 kg. Errors caused by inaccurate recall would lead to underestimation rather than overestimation of the association between BMI and sickness absence.

Our findings may be limited by using BMI as our exposure. It has been shown that fat distribution, measured as waist-hip ratio or waist circumference, is the better predictor of a number of disease endpoints and their risk factors, and probably also of sickness absence (24)(50). However, waist and hip measurements were not obtained during the baseline screening of the Whitehall II cohort.

In this study, BMI remained an independent predictor of increased rates of sickness absence in a model adjusted for age, grade, health-related behaviors, and mental and physical ill-health. Although a general measure of mental health (GHQ caseness) and a general measure of physical health (longstanding illness) are unlikely to capture the full spectrum of weight-mediated morbidity at baseline and morbidity accumulated subsequent to baseline, it is obvious that further research is needed to unravel fully the association between BMI and sickness absence. The effects of BMI on mental and physical health may be partially mediated by psychosocial exposures, such as discrimination and bullying, disproportionately suffered by obese employees in the workplace (51)(52). These exposures were not measured directly in this study. However, we did examine promotion as an indicator of discrimination. Using data from all study participants, we found that only 13.6% (n = 34) of the obese women (n = 250) were promoted compared with 21.7% (n = 483) of the non-obese women (n = 2230; age-adjusted odds ratio = 0.64; 95% CI, 0.44 to 0.93), indicating that obese women seem to suffer discrimination in terms of promotion prospects. However, in men, there was no evidence of discrimination (odds ratio = 0.90; 95% CI, 0.66 to 1.22). In analyses restricted to the cohort used in this study (i.e., excluding those with missing data on sickness absence and covariates), the CIs widened to non-significant in women, but the effect size remained the same.

Loss to follow-up was moderately low, 18.3% of the baseline cohort. Those lost to follow-up were older and more likely to be women in the lower grades. Participants in all these categories have higher rates of sickness, so our findings are likely to be underestimates of effects. A further limitation of a cohort 35 to 55 years of age in 1985 to 1988 and almost exclusively white collar British civil servants is that findings may not apply to wider populations. However, generalizability may be less limited than first imagined. The majority of the working population in post-industrial countries is now employed in white collar jobs, and Whitehall II participants cover a wide range of employment grades with annual full-time salaries in 1995 ranging from $10,006 U.S. (£4995) to $300,480 U.S. (£150,000). Similarly, relatively generous social insurance systems exist widely in Europe and Scandinavia.

In addition to adverse effects on individual health and pressure on services, our findings indicate that the current obesity epidemic is also likely to result in significant increases in sickness absence, particularly longer-term sickness absence. Although further research is needed to understand all of the mechanisms underlying this association, given that the current epidemic seems set to continue, it would be unethical to delay the implementation of weight reduction interventions. Existing work site interventions aimed at health promotion in general or weight reduction in particular have had limited success in achieving and maintaining weight losses. A comprehensive population approach to weight control and weight reduction is the strategy most likely to be successful both for the reduction of weight and the concomitant effects on health and sickness absence. Such a strategy would include input at the level of the workplace to tackle the contribution of unhealthy diets, sedentary work, and weight-sustaining work environments (53).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The Whitehall II study was supported by grants from the Medical Research Council; British Heart Foundation; Health and Safety Executive; Department of Health; National Heart, Lung and Blood Institute (HL36310), U.S. NIH: National Institute on Aging (AG13196), U.S. NIH; Agency for Health Care Policy Research (HS06516); and the John D. and Catherine T. MacArthur Foundation Research Networks on Successful Midlife Development and Socioeconomic Status and Health. M.K., who also works at the Finnish Institute of Occupational Health, Finland, and J.V. were supported by the Academy of Finland (Projects 105195 and 11,604) and the Finnish Work Environment Foundation; J.E.F. is supported by the Medical Research Council (Grant G8802774); M.J.S. is supported by a grant from the British Heart Foundation; and M.G.M. is supported by a Medical Research Council Research Professorship. We thank all participating Civil Service departments and their welfare, personnel, and establishment officers; the Occupational Health and Safety Agency; the Council of Civil Service Unions; all participating civil servants in the Whitehall II study; and all members of the Whitehall II study team.

  • 1

    Nonstandard abbreviations: GHQ, General Health Questionnaire; CI confidence interval.

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    Kopelman, PG (2000) Obesity as a medical problem. Nature 404: 635643.
  • 2
    Bray, GA (1987) Overweight is risking fate. Definition, classification, prevalence, and risks. Ann N Y Acad Sci. 499: 1428.
  • 3
    Eckersley, RM (2001) Losing the battle of the bulge: causes and consequences of increasing obesity. Med J Aust. 174: 590592.
  • 4
    Sorlie, P., Gordon, T., Kannel, WB (1980) Body build and mortality. The Framingham study. JAMA 243: 18281831.
  • 5
    Somes, G. W., Kritchevsky, S. B., Shorr, R. I., Pahor, M., Applegate, WB (2002) Body mass index, weight change, and death in older adults: the systolic hypertension in the elderly program. Am J Epidemiol. 156: 132138.
  • 6
    Koh-Banerjee, P., Wang, Y., Hu, F. B., Spiegelman, D., Willett, W. C., Rimm, EB (2004) Changes in body weight and body fat distribution as risk factors for clinical diabetes in US men. Am J Epidemiol. 159: 11501159.
  • 7
    Breeze, E., Clarke, R., Shipley, M. J., Marmot, M. G., Fletcher, AE (2006) Cause-specific mortality in old age in relation to body mass index in middle age and in old age: follow-up of the Whitehall cohort of male civil servants. Int J Epidemiol. 35: 169178.
  • 8
    Roberts, R. E., Strawbridge, W. J., Deleger, S., Kaplan, GA (2002) Are the fat more jolly? Ann Behav Med. 24: 169180.
  • 9
    Jorm, A. F., Korten, A. E., Christensen, H., Jacomb, P. A., Rodgers, B., Parslow, RA (2003) Association of obesity with anxiety, depression and emotional well-being: a community survey. Aust N Z J Public Health 27: 434440.
  • 10
    Yan, L. L., Daviglus, M. L., Liu, K., et al (2004) BMI and health-related quality of life in adults 65 years and older. Obes Res. 12: 6976.
  • 11
    Bergendorff, S. (2003). Sickness Absence in Europe—A Comparative Study. Stockholm: Department of Research, Analysis and Statistics.
  • 12
    Confederation of British Industry (2005). Who Cares Wins: Absence and Labour Turnover Survey 2005. London: Confederationof British Industry.
  • 13
    Ministerial Task Force for Health SaPatCO (2004). Managing Sickness Absence in the Public Sector. London: Health and Safety Executive.
  • 14
    Zarling, E., Hartz, A., Larsen, J., Rimm, AA (1977) Obesity and illness associated absenteeism. Obesity/Bariatric Med. 6: 134136.
  • 15
    Parkes, KR (1987) Relative weight, smoking, and mental health as predictors of sickness and absence from work. J Appl Psychol. 72: 275286.
  • 16
    Narbro, K., Jonsson, E., Larsson, B., Waaler, H., Wedel, H., Sjöström, L. (1996) Economic consequences of sick-leave and early retirement in obese Swedish women. Int J Obes Relat Metab Disord. 20: 895903.
  • 17
    Tsai, S. P., Gilstrap, E. L., Colangelo, T. A., Menard, A. K., Ross, CE (1997) Illness absence at an oil refinery and petrochemical plant. J Occup Environ Med. 39: 455462.
  • 18
    Kivimaki, M., Vahtera, J., Thomson, L., Griffiths, A., Cox, T., Pentti, J. (1997) Psychosocial factors predicting employee sickness absence during economic decline. J Appl Psychol. 82: 858872.
  • 19
    Tucker, L. A., Friedman, GM (1998) Obesity and absenteeism: an epidemiologic study of 10,825 employed adults. Am J Health Promot. 12: 202207.
  • 20
    Moens, G., Van Gall, L., Muls, E., Viane, B., Jacques, P. (1999) Body mass index and health among the working population. Eur J Pub Health 9: 119123.
  • 21
    Burton, W. N., Chen, C. Y., Schultz, A. B., Edington, DW (1999) The costs of body mass index levels in an employed population. Stat Bull Metrop Insur Co. 80: 814.
  • 22
    Noppa, H., Bengtsson, C. (1980) Obesity in relation to socioeconomic status. A population study of women in Goteborg, Sweden. J Epidemiol Commun Health 34: 139142.
  • 23
    Larsson, B., Björntorp, P., Tibblin, G. (1981) The health consequences of moderate obesity. Int J Obes Relat Metab Disord. 5: 97116.
  • 24
    Björntorp, P. (1988) The associations between obesity, adipose tissue distribution and disease. Acta Med Scand. (Suppl 723), 121134.
  • 25
    Moreau, M., Valente, F., Mak, R., et al (2004) Obesity, body fat distribution and incidence of sick leave in the Belgian workforce: the Belstress study. Int J Obes Relat Metab Disord. 28: 574582.
  • 26
    Marmot, M. G., Smith, Davey G., Stansfeld, S., et al (1991) Health inequalities among British civil servants: the Whitehall II study. Lancet 337: 13871393.
  • 27
    Ezekiel, A., Toon, L. (2002). Whitehall II Study: Manual for Phase 5 Screening Examination. London: The Department of Epidemiologyand Public Health, University College London.
  • 28
    North, F. M., Syme, S. L., Shipley, M. J., Feeney, A., Marmot, MG (1996) Psychosocial work environment and sickness absence among British civil servants: the Whitehall II study. Am J Public Health 86: 332340.
  • 29
    Marmot, M. G., Feeney, A., Shipley, M., North, F., Syme, SL (1995) Sickness absence as a measure of health status and functioning: from the UK Whitehall II study. J Epidemiol Commun Health 49: 124130.
  • 30
    Khosla, T., Lowe, CR (1971) Obesity and smoking habits. Br Med J. 4: 1013.
  • 31
    Holcomb, H. S., III, Meigs (1972) Medical absenteeism among cigarette, and cigar and pipe smokers. Arch Environ Health 25: 295300.
  • 32
    Taylor, P. J., Burridge, J. (1982) Trends in death, disablement, and sickness absence in the British Post Office since 1981. Br J Ind Med. 39: 110.
  • 33
    Marmot, M. G., North, F., Feeney, A., Head, J. (1993) Alcohol consumption and sickness absence: from the Whitehall II study. Addiction 88: 369382.
  • 34
    Hill, J. O., Wyatt, HR (2005) Role of physical activity in preventing and treating obesity. J Appl Physiol. 99: 765770.
  • 35
    Suter, PM (2005) Is alcohol consumption a risk factor for weight gain and obesity? Crit Rev Clin Lab Sci. 42: 197227.
  • 36
    Alexanderson, K., Norlund, A. (2004) Sickness absence—causes, consequences, and physicians’ sickness certification practice. A systematic literature review by the Swedish Council on Technology Assessment in Health Care. Scand J Public Health. (Suppl 63).
  • 37
    Office of Population Censuses and Surveys, Social Survey Division (1984). General Household Survey 1982. London: HMSO.
  • 38
    Morris, J. N., Chave, S. P. W., Sirey, A. C., Epstein, L., Sheehan, DJ (1973) Vigorous exercise in leisure time and the incidence of coronary heart disease. Lancet 1: 334339.
  • 39
    Goldberg, DP (1972). The detection of psychiatric illness by questionnaire. London: Oxford University Press.
  • 40
    Office of Population Censuses and Surveys (1987). General Household Survey 1987. London: H.M. Stationery Office.
  • 41
    Britton, A., Marmot, M. (2004) Different measures of alcohol consumption and risk of coronary heart disease and all-cause mortality: 11-year follow-up of the Whitehall II Cohort Study. Addiction 99: 109116.
  • 42
    Kivimaki, M., Head, J., Ferrie, J. E., Shipley, M. J., Vahtera, J., Marmot, MG (2003) Sickness absence as a global measure of health: evidence from mortality in the Whitehall II prospective cohort study. Br Med J. 327: 364368.
  • 43
    Singh-Manoux, A., Martikainen, P., Ferrie, J. E., Zins, M., Marmot, M., Goldberg, M. (2006) What does self-rated-health measure? Results from the British Whitehall II and French Gazel cohort studies. J Epidemiol Commun Health 60: 364372.
  • 44
    Stansfeld, S. A., Marmot, MG (1992) Social class and minor psychiatric disorder in British Civil Servants: a validated screening survey using the General Health Questionnaire. Psychol Med. 22: 739749.
  • 45
    North, F., Syme, S. L., Feeney, A., Head, J., Shipley, M. J., Marmot, MG (1993) Explaining socioeconomic differences in sickness absence: the Whitehall II study. Br Med J. 306: 361366.
  • 46
    Laitinen, J., Nayha, S., Kujala, V. (2005) Body mass index and weight change from adolescence into adulthood, waist-to-hip ratio and perceived work ability among young adults. Int J Obes (Lond). 29: 697702.
  • 47
    Witteman, J. C., Willett, W. C., Stampfer, M. J., et al (1989) A prospective study of nutritional factors and hypertension among US women. Circulation 80: 13201327.
  • 48
    Gunnell, D., Berney, L., Holland, P., et al (2004) Does the misreporting of adult body size depend upon an individual's height and weight? Methodological debate. Int J Epidemiol. 33: 13981399.
  • 49
    Rowland, ML (1990) Self-reported weight and height. Am J Clin Nutr. 52: 11251133.
  • 50
    Larsson, B., Seidell, J., Svardsudd, K., et al (1989) Obesity, adipose tissue distribution and health in men-the study of men born in 1913. Appetite 13: 3744.
  • 51
    Puhl, R., Brownell, KD (2001) Bias, discrimination, and obesity. Obes Res. 9: 788805.
  • 52
    Kivimaki, M., Elovainio, M., Vahtera, J. (2000) Workplace bullying and sickness absence in hospital staff. Occup Environ Med. 57: 656660.
  • 53
    James, WPT (1995) A public health approach to the problem of obesity. Int J Obes Relat Metab Disord. 19 (Suppl 3), S37S45.