- Email: [email protected]
Biological predictors of change in functioning in the Whitehall II study
Biological Predictors of Change in Functioning in the Whitehall II Study MEENA KUMARI, PHD, TERESA SEEMAN, PHD, AND MICHAEL MARMOT, FRCP
PURPOSE: To examine whether risk factors for CHD are related to change in functioning independent of the presence or development of disease. METHODS: Longitudinal follow up of 4768 men and 2034 women civil servants from 20 London-based departments with complete data for the SF-36, biological variables, and BMI and health related behaviors. Data are used from two phases of the Whitehall II study, phase 3 (1991–1993) and phase 4 (1995) with an interval of 36 months. Weight, height, fasting insulin, 2-hour post load glucose, total and HDLcholesterol, fibrinogen, von Willebrand factor, diastolic and systolic blood pressure, and waist hip ratio were measured at phase 3. Demographic and socio-economic information, health related behaviors, and the SF-36 were obtained at both phases by questionnaire. RESULTS: Waist hip ratio, fasting insulin, triglycerides, and HDL-cholesterol were associated with a decline in physical functioning in the total cohort and when those with poor health at baseline were removed from the analyses. Principal component analysis revealed that these variables clustered with total cholesterol and may represent insulin resistance. The biological variables had a cumulative effect on decline in physical functioning such that those with poor waist hip ratio, fasting insulin, triglycerides, and HDL-cholesterol was two times greater than those without. This relationship was independent of exercise, smoking, and alcohol intake which explained only 17% and 5.4% of the association in men and women, respectively. CONCLUSIONS: A number of biological variables, which may represent insulin resistance, are associated with decline in physical functioning in men and women independent of prevalent ill health or health related behaviors. Ann Epidemiol 2004;14:250–257. 쑕 2004 Elsevier Inc. All rights reserved. KEY WORDS:
Health Status, Cardiovascular Risk Factors, Insulin Resistance.
INTRODUCTION In an ageing population, maintenance of good physical and mental functioning is an important goal. Functioning predicts use of health services and a number of disease outcomes including coronary heart disease (CHD) (1–4). It has become increasingly important to identify predictors of changes in functioning to determine areas for intervention and also as determinants of successful ageing because
From the International Centre for Health and Society, Department of Epidemiology and Public Health, University College London, London, UK (M.K., M.M.); and Department of Medicine, Division of Geriatrics, UCLA, Los Angeles, USA (T.S.). Address correspondence to: Dr. Meena Kumari, International Centre for Health and Society, Department of Epidemiology and Public Health, University College London, 1-19 Torrington Place, London, WC1E 6BT, UK. Tel.: ⫹44-207-679-5637; Fax: ⫹44-207-813-0280. E-mail: M.Kumari@ public-health.ucl.ac.uk The Whitehall II study has been supported by grants from the Medical Research Council; British Heart Foundation; Health and Safety Executive; Department of Health; National Heart Lung and Blood Institute (RO1HL36310), US, NIH; National Institute on Aging (RO1-AG13196), US, NIH; Agency for Health Care Policy Research (RO1-HS06516); and the John D. and Catherine T. MacArthur Foundation Research Networks on Successful Midlife Development and Socio-economic Status and Health. M.M. is supported by the Medical Research Council. Received April 29, 2002; accepted September 16, 2003. 쑕 2004 Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010
continuous decline in mental and physical functioning is not an inevitable consequence of ageing (5, 6). Healthy individuals over 60 years of age had greater decline in physical functioning with increased baseline burdens of physiological dysregulation (i.e., “allostatic load”) in a study of successful ageing (7). As proposed by McEwen and Stellar (8) and amplified by McEwen (9), the allostatic load hypothesis posits that there is a physiological “toll” paid by homeostatic mechanisms (10) resulting in dysregulated patterns of activity over time. The hypothesis further predicts that this cumulative dysregulation in biological parameters should be associated with decline in functioning independent of manifest disease. In this report we will describe the relationship between risk factors for CHD and change in functioning in male and female office workers, to whom the SF-36, a widely used measure of health functioning (11–15) was administered in the same way twice. Because there is an association between physical and mental morbidity (16, 17) both of these components of the SF-36 are investigated. A large array of biological information was also collected at the time the first questionnaire was administered. The aims of this study are to examine the hypothesis that: 1) risk factors for CHD are also risk factors for decline in functioning; and 2) these associations are direct effects that are independent 1047-2797/04/$–see front matter doi:10.1016/j.annepidem.2003.09.011
AEP Vol. 14, No. 4 April 2004: 250–257
Selected Abbreviations and Acronyms SF-36 ⫽ Short Form 36 general health survey scale PCS ⫽ physical functioning component summary score MCS ⫽ mental functioning component summary score CHD ⫽ coronary heart disease BMI ⫽ body mass index
of disease and also not determined by health related behaviors.
METHODS Participants Data for these analyses come from the Whitehall II study. Participants were recruited between 1985 and 1988 (phase 1) from 20 civil service departments. After initial participation, postal questionnaires were administered in 1989 (phase 2), between 1991 and 1993 (phase 3, which also included a screening examination), and in 1995 (phase 4). The data presented in this study are from men and women who participated in phase 3 and phase 4 of the study, had information on all of the Short Form 36 (SF-36) General Health Survey Scale (11, 12) and attended the clinic for blood sample analyses in phase 3, and had complete BMI and health related behavior information (n ⫽ 4768 men and n ⫽ 1999 women). This group of individuals were younger and of higher employment grade compared with the total population. The mean duration of follow up for this population was 3.15 years. Full details of the screening examinations are reported elsewhere (18, 19). Questionnaire Data Data on age, sex, employment grade, menopausal status (women only), cigarette smoking, alcohol intake, and physical activity were obtained by questionnaire at phase 3. Data on longstanding illness were obtained by questionnaire at phases 3 and 4. Employment grade within the civil service was used as a measure of socio-economic position and was assigned to one of 12 civil service non-industrial grade levels, which reflect salary scales. These have been grouped into six categories for the purposes of analysis. Tobacco smoking was classified as “non smoker,” “ex smoker,” and “current smoker.” Alcohol consumption was assessed with a derived variable of number of units in the past week. Alcohol categories were created by dividing participants into categories of “0,” “1-14,” “15-21,” and ‘22+’ based on these numbers. For women the top two categories were combined into a “15⫹” category due to low numbers. Exercise was measured as frequency of mild, moderate, and vigorous exercise per week.
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
251
Functioning was assessed using the SF-36, a 36 item questionnaire that covers issues relating to physical, psychological, and social functioning. It is coded into eight scales: physical functioning, role limitations due to physical problems, social functioning, bodily pain, general mental health, role limitations due to emotional problems, vitality, and general health perceptions. These eight scales can be summarized into physical and mental functioning component summary scores by a method based on factor analysis (20, 21). Changes in these scores are used as outcomes in this study. The correlations between the mental and physical component summary scores were 0.12 and ⫺0.16 at phase 3 and phase 4, respectively. The correlation between change in physical and mental functioning summary scores was ⫺0.16 and ⫺0.14 in men and women, respectively. Assessment of Biological Factors The 10 biological parameters considered in the present analyses were all measured at the baseline screening examination. These include: waist hip ratio, 2-hour post load glucose, total and HDL-cholesterol, fasting insulin, triglycerides, diastolic and systolic blood pressure, fibrinogen, and von Willebrand factor (vWF). Body circumferences were measured with subjects in the standing position and unclothed, using a fiberglass tape measure at 600 g tension. The waist circumference was taken as the smallest circumference at or below the costal margin and the hip circumference at the level of the greater trochanter. Height and weight were taken in a standard way. The oral glucose tolerance test was administered following an overnight fast in the morning or in the afternoon after no more than a light fat-free breakfast eaten before 8.00 A.M. After the initial venous blood sample, participants drank 389 ml “Lucozade” (equivalent to 75g anhydrous glucose) over 5 minutes. A second blood sample was taken 2 hours later. Diabetic subjects did not participate in this part of the screening. Venepucture of the left antecubital vein was performed with tourniquet. Blood was collected into plain, citrate or fluoride Sarstedt monovettes. After centrifugation samples were immediately frozen at ⫺80⬚C and stored until assay. Serum for lipid analyses was refrigerated at ⫺4⬚C and assayed within 72 hours. Glucose was determined in fluoride plasma by an electrochemical glucose oxidase method. Serum insulin was measured by radioimmunoassay using a polyclonal guinea-pig antiserum. Cholesterol and triglycerides were measured in a centrifugal analyzer by enzymic colorimetric methods. HDL-cholesterol was determined after dextran sulphase-magnesium chloride precipitation of non-HDL cholesterol. Fibrinogen was determined by an automated modification of the Clauss method, von Willebrand Factor antigen was measured by double antibody
252 Kumari et al.
AEP Vol. 14, No. 4 April 2004: 250–257
CHD RISK FACTORS AND CHANGES IN FUNCTIONING
ELISA. Technical error was estimated by assaying blinded duplicate samples for 5% of subjects. Coefficients of variation were 2.0 to 6.6 percent for glucose, cholesterol, triglycerides, and HDL-cholesterol, 23% for insulin, 16.4% for vWF, and 7.7% for fibrinogen. Poor health was defined as: angina, self report of doctordiagnosed heart attack, or angina, probable or possible ischemia on resting electrocardiogram, hypertension (⬎160/ 90 mmHg or taking antihypertensive drugs), claudication, diabetes, chronic bronchitis, musculoskeletal disorders, cancer, and those that scored above 5 in the general health questionnaire (15) at baseline (phase 3). At phase 4 those participants that reported doctor-diagnosed diabetes or heart attack or angina were also classified as being in poor health. Analyses Due to skewed distribution, all the biochemical variables were logged for the purposes of analysis. Linear regression models were used to assess associations between each biological and biochemical variable and changes in functioning using SAS. Pearson (product moment) correlation coefficients between the 10 biological variables under consideration were computed. Principal component analysis followed by orthogonal rotation was performed to examine the clustering of these variables. A summary accumulation score containing the variables that loaded onto the initial factor was created with scores grouped as 0, 1, 2 and 3⫹. The cut off values were, for men, waist hip ratio: 0.94, triglycerides: 1.91, fasting insulin: 8.25, HDL cholesterol: 1.07, and total cholesterol: 7.19, and for women, waist hip ratio: 0.82, triglycerides 1.44, fasting insulin: 8.00, HDL cholesterol: 1.37, and total cholesterol: 7.26. To assess associations with change in functioning, baseline functioning was controlled for and coefficients for other variables provide estimates of the impact of these factors on change in performance from baseline. Models were examined controlling for possible confounding by socio-demographic factors. Change in functioning was examined in relation to baseline biological variables and accumulation scores and to smoking, alcohol intake, and physical activity. Sensitivity analyses were performed by examining change in functioning in each tertile of baseline functioning component score. Analyses were also repeated on a population restricted to those participants that reported the top score in the physical functioning scale at baseline. Logistic regression was used to examine association with declines of a score of 10 or more between baseline and follow up. To calculate the contribution of each of the explanatory variables to the association between the biological variable and change in functioning summary scores, the percentage reduction in difference was calculated by including the variable, appropriate
socio-demographic factors, and comparing the regression coefficient for the biological variable before and after adjustment for each explanatory factor or group of factors.
RESULTS Table 1 shows the matrix of Pearson’s (product moment) correlations at baseline (phase 3) stratified by sex. Most of the cardiovascular risk factors examined are correlated and minor differences exist between men and women. These variables were subjected to a principal component analysis using the principal axis method to extract the components followed by varimax rotation. While four components displayed Eigenvalues greater than 1, examination of a Scree plot indicated that only the first three components were meaningful. Therefore, these components were retained by rotation. However, combined, these components explained 55% of total variance. The rotated factor pattern is shown in Table 2. An item was considered to load on a given component if the factor loading was 0.4 or greater. Using these criteria, the initial component appeared to reflect insulin resistance, showing high factor loadings for waist hip ratio, fasting insulin, triglycerides, HDL cholesterol, and total cholesterol. Diastolic and systolic blood pressure, representing hypertension, comprised the second factor, and fibrinogen and vWF the third factor representing hemostasis. Post load glucose failed to load and was not used in subsequent analyses. Because of the highly skewed nature of the SF-36 health functioning scores at each phase, Table 3 shows median scores in the physical functioning component score (PCS) and mental functioning component score (MCS) at phase 3 (1991–1993) and phase 4 (1995) of the study. Mean change in functioning was normally distributed and so the mean change between phase 3 and phase 4 is presented. A high score represents good functioning and a negative score indicates a decline in functioning. There were declines in PCS and MCS in men and women. The relationship between each of the risk factors for CHD that had high factor loadings for the initial component and subsequent functioning are presented in Table 4. Increasing waist hip ratio, triglycerides, fasting insulin, and decreasing HDL-cholesterol are associated with a decline in PCS for men and women while total cholesterol is not. A sensitivity analysis was performed by tertiling baseline physical functioning score and examining physical functioning as an outcome by each biological variable in each tertile. Increasing biological risk was associated with a greater change (decline) in functioning within each tertile. Post load insulin was also associated with decline in PCS in men (N ⫽ 4182, p ⫽ 0.002 test for trend, data not shown) but not women. These analyses were also repeated with the
AEP Vol. 14, No. 4 April 2004: 250–257
253
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
TABLE 1. Matrix of Pearson’s correlation coefficients among the cardiovascular variables in men and women
Waist hip ratio Post load glucose Cholesterol Fasting insulin Triglycerides HDL-cholesterol Diastolic blood pressure Systolic blood pressure Fibrinogen Von Willebrand factor
Waist hip ratio
Post load glucose
· 0.222 0.175 0.464 0.406 ⫺0.350 0.224
Triglycerides
HDLcholesterol
Diastolic blood pressure
Systolic blood pressure
Fibrinogen
von Willebrand factor
0.423 0.205 0.121 · 0.367 ⫺0.302 0.235
0.323 0.155 0.370 0.333 · ⫺0.428 0.152
ⴚ0.311 ⴚ0.143 0.044 ⴚ0.272 ⴚ0.478 · ⫺0.034 ns
0.299 0.166 0.141 0.266 0.177 ⴚ0.086 ·
0.270 0.173 0.111 0.222 0.135 ⴚ0.031 0.739
0.191 0.050 0.100 0.131 0.125 ⴚ0.159 0.056
0.080 0.092 0.034 0.136 0.025 ns ⴚ0.013 ns ⴚ0.013 ns
0.196
0.236
0.194
⫺0.048
0.757
0.138 0.051
0.234 0.153
0.182 0.082
⫺0.207 ⫺0.663
0.080 0.052
Cholesterol
Fasting insulin
0.165 · 0.093 0.251 0.206 ⫺0.134 0.167
0.191 0.065 · 0.112 0.378 0.078 0.168
0.238
0.213
0.244 0.138
0.134 0.113
· 0.103 0.077
0.062
0.026
· 0.243
0.179
Coefficients for men in bold and women in italics. All correlations significant (p ⬍ 0.05) unless indicated. ns ⫽ not significant.
additional risk factors for CHD, diastolic and systolic blood pressure, fasting and post load glucose, and von Willebrand factor. These risk factors were found not to be associated with change in physical functioning component score in men or women. Of the variables examined, only waist hip ratio was associated with change in the MCS. When the analyses were repeated following the removal of those participants with poor health (15) all trends remained except for HDL cholesterol and triglycerides which no longer reached significance in men (N ⫽ 2818, p ⫽ 0.14 test for trend) and women (N ⫽ 1105, p ⫽ 0.36 test for trend), respectively. Removal of those participants who reported doctor diagnosed diabetes and CHD at phase 4 (follow up) did not substantially alter our findings. When we examined the cumulative effects of dysregulation in the variables that had high loading on the initial factor we found that participants who scored 3⫹ at baseline on our summary score experienced a decline in physical functioning that was twice as large as those that scored 0. When these analyses were repeated using adverse quartile
of “blood pressure” or “hemostatic” variables (a score of 1 or 2 vs. 0) there was a similar or smaller decline in PCS (Figure 1) indicating that the additional variables provide no extra information with regard to decline in functioning. When the analyses were repeated restricting the population to those participants that reported maximal functioning at baseline, the metabolic summary score for men remained significantly associated with decline in functioning (N ⫽ 1621, p ⫽ 0.03) while the effect for women was no longer significant (N ⫽ 464, p ⫽ 0.41). Examination of the association between “insulin resistance” component score and odds of a decline of 10 or more showed that the insulin resistance component is associated with an odds ratio of 1.56 (1.04–2.08), 1.66 (1.1–2.6), and 1.83 (1.1–3.1) for increasing load in women. The effects in men were less robust and were 0.76 (0.54–1.1), 0.94 (0.6–1.1), and 1.15 (0.79–1.68) with increasing insulin resistance load. Table 5 shows that change in PCS is also associated with smoking, exercise, and alcohol intake in men. Smoking did not predict a decline in PCS in women. Alcohol intake
TABLE 2. Factor loading pattern after orthogonal rotation of principle components in men and women in phase 3 of the Whitehall II study Men (N ⫽ 4491) Variable Waist hip ratio Post load glucose Cholesterol Fasting insulin Triglycerides HDL-cholesterol Diastolic blood pressure Systolic blood pressure Fibrinogen Von Willebrand factor Cumulative % total variance
Women (N ⫽ 2013)
Factor 1
Factor 2
Factor 3
Factor 1
Factor 2
Factor 3
0.58 0.15 0.43 0.46 0.85 ⴚ0.71 0.16 0.05 0.17 ⫺0.14 27
0.28 0.26 0.13 0.25 0.08 0.11 0.91 0.90 ⫺0.03 ⫺0.02 42
0.27 0.32 ⫺0.06 0.37 0.02 ⫺0.12 0.04 ⫺0.00 0.61 0.80 53
0.69 0.31 0.40 0.61 0.81 ⴚ0.69 0.09 0.06 0.27 ⫺0.02 29
0.19 0.23 0.30 0.16 0.18 0.19 0.90 0.90 0.01 0.05 44
0.20 0.23 ⫺0.14 0.24 ⫺0.06 ⫺0.10 0.09 0.07 0.65 0.82 55
254 Kumari et al.
AEP Vol. 14, No. 4 April 2004: 250–257
CHD RISK FACTORS AND CHANGES IN FUNCTIONING
TABLE 3. SF-36 component scores at phase 3 (1991–1993) and phase 4 (1995) and change in score between these phases. Values represent median (interquartile range) scores except where noted Physical component summary score
Score at phase 3 Score at phase 4 Change in score*
Mental component summary score
Men
N
Women
N
Men
N
Women
N
54.3 (50.5, 57.0) 53.9 (49.6, 56.7) ⫺1.5 (⫺1.7, ⫺1.3)
5729 5286 5020
52.1 (45.9, 56.1) 52.0 (44.3, 56.1) ⫺1.5 (⫺1.8, ⫺1.2)
2563 2300 2155
53.6 (48.6, 57.0) 54.6 (47.2, 58.8) ⫺1.1 (⫺1.3, ⫺0.8)
5729 5286 5020
52.6 (46.4, 56.7) 52.0 (42.2, 57.1) ⫺1.1 (⫺1.5, ⫺0.7)
2563 2300 2155
*Mean change in score between phase 3 and phase 4 (95% CI).
was not associated with change in MCS for men while only exercise was associated with change in MCS in women. Adjustment for those health related behaviors associated with change in PCS reveals that exercise, smoking, and alcohol intake at baseline played a small role as determinants of the relationship between the biological factors and decline in functioning. Additional adjustment for BMI, accounted for up to 40% of the relationship between the
biological variables and decline in functioning in men and women. Adjustment for BMI accounts for much of the explanatory power as when only smoking, exercise, and alcohol are adjusted for then less than 20% of the relationship between the biological variables and decline in PCS can be accounted for. Similar data are apparent for women. The relationship between the accumulation score and decline in physical functioning was also minimally determined
TABLE 4. Change in physical component summary score of SF-36 by biological factors. Values represent mean change in score between phase 3 (1991–1993) and phase 4 (1995) of the study adjusted for age, employment grade, length of follow up and baseline score Quartiles
Men
Waist hip ratio
(N ⫽ 4625) Less than 0.87 0.87–0.90 0.90–0.94 More than 0.94
Cholesterol (mmol/l)
(N ⫽ 4672) Less than 5.69 5.69–6.41 6.41–7.19 More than 7.19
Triglycerides (mmol/l)
Fasting insulin (pmol/l)
HDL-cholesterol (mmol/l)
Accumulation score*
(N ⫽ 4672) Less than 0.91 0.91–1.31 1.31–1.91 More than 1.91 (N ⫽ 4227) Less than 3.51 3.51–5.45 5.45–8.26 More than 8.26 (N ⫽ 4655) Less than 1.07 1.07–1.27 1.27–1.51 More than 1.51 0 1 2 3⫹
Women Mean change ⫺0.9 ⫺1.5 ⫺1.4 ⫺2.0 P ⫽ 0.0001 ⫺1.25 ⫺1.68 ⫺1.49 ⫺1.57 P ⫽ 0.41 ⫺0.8 ⫺1.6 ⫺1.7 ⫺2.0 P ⫽ 0.0001 ⫺1.2 ⫺1.4 ⫺1.3 ⫺2.1 P ⫽ 0.0002 ⫺2.3 ⫺1.5 ⫺1.0 ⫺1.3 P ⫽ 0.0020 ⫺1.0 ⫺1.4 ⫺2.0 ⫺2.3 P ⫽ 0.0001
(N ⫽ 1948) Less than 0.73 0.73–0.77 0.77–0.82 More than 0.82 (N ⫽ 1945) Less than 5.69 5.69–6.45 6.45–7.26 More than 7.26 (N ⫽ 1945) Less than 0.75 0.75–1.01 1.01–1.44 More than 1.44 (N ⫽ 1693) Less than 3.22 3.22–4.98 4.98–7.98 More than 7.98 (N ⫽ 1942) Less than 1.37 1.37–1.63 1.63–1.93 More than 1.93 0 1 2 3⫹
P values are test for linear trend. *Accumulation score: sum of variables for which the participants were rated in the most adverse risk quartile of these variables.
Mean change ⫺1.2 ⫺1.7 ⫺1.2 ⫺2.23 P ⫽ 0.0162 ⫺1.54 ⫺1.97 ⫺1.50 ⫺1.33 P ⫽ 0.80 ⫺1.2 ⫺1.5 ⫺1.1 ⫺2.1 P ⫽ 0.0113 ⫺1.4 ⫺1.1 ⫺1.3 ⫺2.4 P ⫽ 0.0039 ⫺1.9 ⫺1.6 ⫺1.8 ⫺0.7 P ⫽ 0.0047 ⫺1.2 ⫺1.7 ⫺2.2 ⫺2.7 P ⫽ 0.01
AEP Vol. 14, No. 4 April 2004: 250–257
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
255
Figure 1. Change in PCS by score in adverse quartile of physiological variables. The figure represents change in PCS by score in adverse quartile in all variables that showed a high loading for the initial component (metabolic’) scored as 0, 1, 2, and 3⫹; in variables that showed a high factor loading for the second component (blood pressure) scored at 0 or 1 and 2 and variables that showed a high factor loading for third component (haemostatic), scored as 0 or 1 and 2. N ⫽ 4688 men and 1597 women in the total population. The ‘restricted’ group represents change in PCS by score in adverse quartile in all variables that showed a high loading for the initial component score in a population that reported maximum functioning at baseline. N ⫽ 1621 men and 464 women. Columns represent mean change in PCS between phase 3 ((1991–1993) and phase 4 (1995) of the study adjusted for age, employment grade, length of follow up and baseline score.
by health related behaviors, accounting for 11% of the relationship in men and 3% in women.
DISCUSSION These data suggest that some, but not all, risk factors for CHD are risk factors for decline in functioning in healthy individuals. In particular waist hip ratio, fasting insulin, triglycerides, and HDL-cholesterol are associated with a subsequent decline in physical functioning in both men and women. These variables, which were cumulative in their effect, load onto a single factor with total cholesterol that may represent insulin resistance which is a component of the metabolic syndrome (19, 22, 23). These data therefore suggest that insulin resistance may be an important determinant of subsequent physical functioning. Decline in functioning may be occurring as a consequence of overt disease, subclinical disease or independent of disease. It would seem unlikely that the first effect explains
the observed associations as removal of those with poor health from the analyses failed to markedly alter the outcomes. The results from these and previous analyses (24, 25) suggest that the subclinical condition that appears to be most associated with a decline in functioning is insulin resistance. The results are specific to this condition as other risk factors for CHD were without effect. We will be able to examine the effects of worsening subclinical disease in subsequent analyses from this cohort with the availability of clinical data. The decline in physical component summary scores seemed selectively sensitive to biological factors that were related to insulin resistance as the additional variables were less robustly associated with decline in functioning. The components of the cluster have cumulative effects on decline in functioning which accords with the idea of allostatic load effects on “wear and tear” (8, 9, 10, 26). It is possible that insulin resistance and the processes that lead to insulin resistance precipitate a more immediate decline in physical functioning than the other risk factors for CHD
256 Kumari et al.
AEP Vol. 14, No. 4 April 2004: 250–257
CHD RISK FACTORS AND CHANGES IN FUNCTIONING
TABLE 5. Change in physical and mental component summary score between phase 3 (1991–1993) and phase 4 (1995) by smoking, alcohol intake, and exercise status at phase 3 (1991–1993) Men
Women
Physical component
Mental component
N
Physical component
Mental component
N
⫺1.4 ⫺1.4 ⫺2.3 P ⫽ 0.006
⫺1.3 ⫺1.1 ⫺0.7 P ⫽ 0.31
2640 2085 678
⫺1.8 ⫺1.4 ⫺1.6
⫺1.0 ⫺0.9 ⫺1.0
1344 631 394
⫺2.1 ⫺1.7 ⫺1.0 P ⫽ 0.002
⫺0.1 ⫺1.2 ⫺1.4 P ⫽ 0.007
781 2536 2182
⫺1.8 ⫺1.7 ⫺0.8
⫺1.0 ⫺1.0 ⫺1.5
851 1149 431
⫺1.9 ⫺1.4 ⫺1.1 ⫺1.9 P ⫽ 0.043
⫺0.6 ⫺1.1 ⫺1.3 ⫺1.3
797 3007 682 1008
⫺2.3 ⫺1.2 ⫺1.3
⫺0.5 ⫺1.3 ⫺1.9
740 1463 226
Smoking Never Ex Current Exercise None/mild Moderate Vigorous Alcohol intake (units) None 1–14 15–21 (15⫹ Female) 22⫹ (male)
P ⫽ 0.033
P values are test for linear trend.
that were investigated, although the mechanisms by which this might occur remain unclear. The associations between waist hip ratio, triglycerides, and fasting insulin and subsequent change in the physical summary component score were not subsequent to BMI and poor health related behaviors. The importance of BMI and change in weight with respect to health status have been described (27, 28), however the role of waist hip ratio has not been determined. The stronger association between waist hip ratio and decline in functioning suggest that either BMI is a poorer measure of adiposity than waist hip ratio or that waist hip ratio is a more important determinant of subsequent functioning than BMI in men. In women, both BMI and waist hip ratio were independently associated with change in functioning. The failure of BMI to explain the relationships between the biological variables and change in functioning suggests that these effects are not simply determined by overweight but by additional mechanisms. Adjustment for BMI failed fully to explain the relationship between the accumulation of the variables and change in functioning again suggesting additional mechanisms are operating. However, it may be thought inappropriate to adjust for BMI in this context as it is possible that BMI could be considered part of a construct reflecting underlying insulin resistance. The health related behaviors were found not to be important determinants of the relationship between the risk factors examined and change in functioning. This suggests that additional, direct processes may be involved, for example, physiological responses to stress as outlined in the allostatic load hypothesis (8–10), leading to the decline in functioning associated with insulin resistance. Follow up of
the cohort and examination of the trajectory of change in functioning with respect to the factors that have been described at baseline will allow us to describe the effects of insulin resistance and changes in functioning as the cohort ages. Limitations of the study need to be considered. Nonresponse at follow up is likely to have biased the effects conservatively, since non-responders tended to be from lower employment grades and have lower SF-36 scores at baseline. The size of the declines observed were small in comparison to the effects observed in the diseased state and their impact on daily life is unclear (29). However, the declines are important in this relatively young and healthy working population as they occurred in a short follow up period of 36 months which if continued at the same rate, as they may reflect ongoing processes, would lead to marked disability. In this short term follow up the initial factor was predictive of clinically relevant declines in functioning in women. It will be possible to ascertain the further decline in functioning should it occur with continued follow up of the group. In conclusion, changes in functioning are associated with some CHD risk factors, in particular, waist hip ratio, HDL cholesterol, triglycerides, and fasting insulin. These associations are not determined by health related behaviors but by other factors as yet identified. These variables cluster and may represent insulin resistance. These data therefore identify insulin resistance as a potential area of intervention to prevent functional decline. REFERENCES 1. Harris T, Kovar M, Suzman R, Kleinman JC, Feldman JJ. Longitudinal study of physical ability in the oldest old. Am J Public Health. 1989;79:698–702.
AEP Vol. 14, No. 4 April 2004: 250–257
2. Kaplan GA, Camadio T. Perceived health and mortality: A nine-year follow up of the human population laboratory cohort. Am J Epidemiol. 1983;117:292–304. 3. Idler EL, Angel RJ. Self-rated health and mortality in the NHANES-1 epidemiological follow-up study. Am J Public Health. 1990;80:446–452. 4. Garret AM, Rota RJ, Abdalla M, Buckingham JK, Russell IT. The SF-36 health survey questionnaire: An outcome measure suitable for routine use within the NHS? BMJ. 1993;306:1440–1444. 5. Rowe JW, Jahn RL. Human ageing: usual and successful. Science. 1987; 237:143–149. 6. Seeman TE, Charpentier PA, Berkaman LF, Tinetti ME, Guralnik JM, Albert M, et al. Predicting changes in physical performance in a high functioning elderly cohort: MacArthur studies of successful aging. J Gerontol. 1994;49:M97–M108. 7. Seeman TE, Singer BH, Rowe JW, Horwitz RI, McEwen BS. The price of adaptation—allostatic load and its health consequences: MacArthur studies of successful aging. Arch Intern Med. 1997;157:2259–2268. 8. McEwen BS, Stellar E. Stress and the individual. Mechanisms leading to disease. Arch Intern Med. 1993;153:2093–2101. 9. McEwen BS. Stress, adaptation and disease. Allostasis and allostatic load. Ann NY Acad Sci. 1998;840:33–44. 10. McEwen BS. Protective and damaging effects of stress mediators. N Engl J Med. 1998;338:171–179. 11. Ware JE, Sherbourne CD. The MOS 36-item short-form health Survey (SF-36): I. Conceptual framework and item selection. Med Care. 1992; 30:473–483. 12. Jenkinson C, Wright L, Coulter A. Criterion validity and reliability of the SF-36 in a population sample. Qual Life Res. 1994;3:7–12. 13. Lyons RA, Lo SV, Littlepage BVC. Comparative health status of patients with 11 common illnesses in Wales. J Epidemiol Community Health. 1994;48:388–390. 14. Hayes V, Morris J, Wolfe C, Morgan M. The SF-36 Health survey questionnaire: Is it suitable for use with older adults? Age Ageing. 1995;24:120–125. 15. Hemingway H, Stafford M, Stansfeld S, Shipley M, Marmot M. Is the SF-36 a valid measure of change in population health? Results from the Whitehall II study. BMJ. 1997;315:1273–1279. 16. Stansfeld SA, Smith GD, Marmot M. Association between physical and psychological morbidity in the Whitehall II study. J Psychosom Res. 1993; 37:227–238.
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
257
17. Bruce ML, Seeman TE, Merrill SS, Blazer DG. The impact of depressive symptomatology on physical disability: MacArthur studies of successful aging. Am J Public Health. 1994;84:1796–1799. 18. Marmot MG, Smith GD, Stansfeld S, Patel C, North F, Head J, et al. Health inequalities among British Civil Servants: The Whitehall II study. Lancet. 1991;337:1387–1393. 19. Brunner EJ, Marmot MG, Nanchahal K, Shipley MJ, Stansfeld SA, Juneja M, et al. Social inequality in coronary risk: Central obesity and the metabolic syndrome. Evidence from the Whitehall II study. Diabetologia. 1997;40:1341–1349. 20. Ware JE, Kosinski M, Bayliss MS, McHorney CA, Roger WH, Raczek A. Comparison of methods for the scoring and statistical analysis of SF36 health profile and summary measures: Summary of results from the medical outcomes study. Med Care. 1995;33:264–279. 21. Ware JE, Kosinski M, Keller S. SF-36 Physical and Mental Summary Scales: A User’s Manual. Boston: The Health Institute, New England Medical Center; 1994. 22. Gray RS, Fabsitz RR, Cowan LD, Lee ET, Howard BV, Savage PJ. Risk factor clustering in the insulin resistance syndrome. The Strong Heart Study. Am J Epidemiol. 1998;148:869–878. 23. Chen C-H, Lin K-C, Tsai S-T, Chou P. Different association of hypertension and insulin-related metabolic syndrome between men and women in 8437 non diabetic Chinese. Am J Hypertens. 2000;13:846–853. 24. Martikainen P, Stansfeld S, Hemingway H, Marmot M. Determinants of socioeconomic differences in change in physical and mental functioning. Soc Sci Med. 1999;49:499–507. 25. Stafford M, Hemingway H, Stansfeld S, Brunner E, Marmot M. Behavioural and biological correlates of physical functioning in middle aged office workers: The UK Whitehall II study. J Epidemiol Community Health. 1998;52:353–358. 26. Seeman TE, McEwen BS, Rowe JW, Singer BH. Allostatic load as a marker of cumulative biological risk: MacArthur studies of successful aging. Proc Natl Acad Sci USA. 2001;98:4770–4775. 27. Stafford M, Hemingway H, Marmot M. Current obesity, steady weight change, and weight fluctuation as predictors of physical functioning in middle aged office workers: the Whitehall II Study. Int J Obes Relat Metab Disord. 1998;22:23–31. 28. Doll HA, Petersen SEK, Stewart-Brown SL. Obesity and physical and emotional well-being: Associations between body mass index, chronic illness, and the physical and mental components of the SF-36 questionnaire. Obes Res. 2000;8:160–170. 29. Muldoon MF, Barger SD, Flory JD, Manuck SB. What are quality of life measurements measuring? BMJ. 1998:542–545.
PURPOSE: To examine whether risk factors for CHD are related to change in functioning independent of the presence or development of disease. METHODS: Longitudinal follow up of 4768 men and 2034 women civil servants from 20 London-based departments with complete data for the SF-36, biological variables, and BMI and health related behaviors. Data are used from two phases of the Whitehall II study, phase 3 (1991–1993) and phase 4 (1995) with an interval of 36 months. Weight, height, fasting insulin, 2-hour post load glucose, total and HDLcholesterol, fibrinogen, von Willebrand factor, diastolic and systolic blood pressure, and waist hip ratio were measured at phase 3. Demographic and socio-economic information, health related behaviors, and the SF-36 were obtained at both phases by questionnaire. RESULTS: Waist hip ratio, fasting insulin, triglycerides, and HDL-cholesterol were associated with a decline in physical functioning in the total cohort and when those with poor health at baseline were removed from the analyses. Principal component analysis revealed that these variables clustered with total cholesterol and may represent insulin resistance. The biological variables had a cumulative effect on decline in physical functioning such that those with poor waist hip ratio, fasting insulin, triglycerides, and HDL-cholesterol was two times greater than those without. This relationship was independent of exercise, smoking, and alcohol intake which explained only 17% and 5.4% of the association in men and women, respectively. CONCLUSIONS: A number of biological variables, which may represent insulin resistance, are associated with decline in physical functioning in men and women independent of prevalent ill health or health related behaviors. Ann Epidemiol 2004;14:250–257. 쑕 2004 Elsevier Inc. All rights reserved. KEY WORDS:
Health Status, Cardiovascular Risk Factors, Insulin Resistance.
INTRODUCTION In an ageing population, maintenance of good physical and mental functioning is an important goal. Functioning predicts use of health services and a number of disease outcomes including coronary heart disease (CHD) (1–4). It has become increasingly important to identify predictors of changes in functioning to determine areas for intervention and also as determinants of successful ageing because
From the International Centre for Health and Society, Department of Epidemiology and Public Health, University College London, London, UK (M.K., M.M.); and Department of Medicine, Division of Geriatrics, UCLA, Los Angeles, USA (T.S.). Address correspondence to: Dr. Meena Kumari, International Centre for Health and Society, Department of Epidemiology and Public Health, University College London, 1-19 Torrington Place, London, WC1E 6BT, UK. Tel.: ⫹44-207-679-5637; Fax: ⫹44-207-813-0280. E-mail: M.Kumari@ public-health.ucl.ac.uk The Whitehall II study has been supported by grants from the Medical Research Council; British Heart Foundation; Health and Safety Executive; Department of Health; National Heart Lung and Blood Institute (RO1HL36310), US, NIH; National Institute on Aging (RO1-AG13196), US, NIH; Agency for Health Care Policy Research (RO1-HS06516); and the John D. and Catherine T. MacArthur Foundation Research Networks on Successful Midlife Development and Socio-economic Status and Health. M.M. is supported by the Medical Research Council. Received April 29, 2002; accepted September 16, 2003. 쑕 2004 Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010
continuous decline in mental and physical functioning is not an inevitable consequence of ageing (5, 6). Healthy individuals over 60 years of age had greater decline in physical functioning with increased baseline burdens of physiological dysregulation (i.e., “allostatic load”) in a study of successful ageing (7). As proposed by McEwen and Stellar (8) and amplified by McEwen (9), the allostatic load hypothesis posits that there is a physiological “toll” paid by homeostatic mechanisms (10) resulting in dysregulated patterns of activity over time. The hypothesis further predicts that this cumulative dysregulation in biological parameters should be associated with decline in functioning independent of manifest disease. In this report we will describe the relationship between risk factors for CHD and change in functioning in male and female office workers, to whom the SF-36, a widely used measure of health functioning (11–15) was administered in the same way twice. Because there is an association between physical and mental morbidity (16, 17) both of these components of the SF-36 are investigated. A large array of biological information was also collected at the time the first questionnaire was administered. The aims of this study are to examine the hypothesis that: 1) risk factors for CHD are also risk factors for decline in functioning; and 2) these associations are direct effects that are independent 1047-2797/04/$–see front matter doi:10.1016/j.annepidem.2003.09.011
AEP Vol. 14, No. 4 April 2004: 250–257
Selected Abbreviations and Acronyms SF-36 ⫽ Short Form 36 general health survey scale PCS ⫽ physical functioning component summary score MCS ⫽ mental functioning component summary score CHD ⫽ coronary heart disease BMI ⫽ body mass index
of disease and also not determined by health related behaviors.
METHODS Participants Data for these analyses come from the Whitehall II study. Participants were recruited between 1985 and 1988 (phase 1) from 20 civil service departments. After initial participation, postal questionnaires were administered in 1989 (phase 2), between 1991 and 1993 (phase 3, which also included a screening examination), and in 1995 (phase 4). The data presented in this study are from men and women who participated in phase 3 and phase 4 of the study, had information on all of the Short Form 36 (SF-36) General Health Survey Scale (11, 12) and attended the clinic for blood sample analyses in phase 3, and had complete BMI and health related behavior information (n ⫽ 4768 men and n ⫽ 1999 women). This group of individuals were younger and of higher employment grade compared with the total population. The mean duration of follow up for this population was 3.15 years. Full details of the screening examinations are reported elsewhere (18, 19). Questionnaire Data Data on age, sex, employment grade, menopausal status (women only), cigarette smoking, alcohol intake, and physical activity were obtained by questionnaire at phase 3. Data on longstanding illness were obtained by questionnaire at phases 3 and 4. Employment grade within the civil service was used as a measure of socio-economic position and was assigned to one of 12 civil service non-industrial grade levels, which reflect salary scales. These have been grouped into six categories for the purposes of analysis. Tobacco smoking was classified as “non smoker,” “ex smoker,” and “current smoker.” Alcohol consumption was assessed with a derived variable of number of units in the past week. Alcohol categories were created by dividing participants into categories of “0,” “1-14,” “15-21,” and ‘22+’ based on these numbers. For women the top two categories were combined into a “15⫹” category due to low numbers. Exercise was measured as frequency of mild, moderate, and vigorous exercise per week.
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
251
Functioning was assessed using the SF-36, a 36 item questionnaire that covers issues relating to physical, psychological, and social functioning. It is coded into eight scales: physical functioning, role limitations due to physical problems, social functioning, bodily pain, general mental health, role limitations due to emotional problems, vitality, and general health perceptions. These eight scales can be summarized into physical and mental functioning component summary scores by a method based on factor analysis (20, 21). Changes in these scores are used as outcomes in this study. The correlations between the mental and physical component summary scores were 0.12 and ⫺0.16 at phase 3 and phase 4, respectively. The correlation between change in physical and mental functioning summary scores was ⫺0.16 and ⫺0.14 in men and women, respectively. Assessment of Biological Factors The 10 biological parameters considered in the present analyses were all measured at the baseline screening examination. These include: waist hip ratio, 2-hour post load glucose, total and HDL-cholesterol, fasting insulin, triglycerides, diastolic and systolic blood pressure, fibrinogen, and von Willebrand factor (vWF). Body circumferences were measured with subjects in the standing position and unclothed, using a fiberglass tape measure at 600 g tension. The waist circumference was taken as the smallest circumference at or below the costal margin and the hip circumference at the level of the greater trochanter. Height and weight were taken in a standard way. The oral glucose tolerance test was administered following an overnight fast in the morning or in the afternoon after no more than a light fat-free breakfast eaten before 8.00 A.M. After the initial venous blood sample, participants drank 389 ml “Lucozade” (equivalent to 75g anhydrous glucose) over 5 minutes. A second blood sample was taken 2 hours later. Diabetic subjects did not participate in this part of the screening. Venepucture of the left antecubital vein was performed with tourniquet. Blood was collected into plain, citrate or fluoride Sarstedt monovettes. After centrifugation samples were immediately frozen at ⫺80⬚C and stored until assay. Serum for lipid analyses was refrigerated at ⫺4⬚C and assayed within 72 hours. Glucose was determined in fluoride plasma by an electrochemical glucose oxidase method. Serum insulin was measured by radioimmunoassay using a polyclonal guinea-pig antiserum. Cholesterol and triglycerides were measured in a centrifugal analyzer by enzymic colorimetric methods. HDL-cholesterol was determined after dextran sulphase-magnesium chloride precipitation of non-HDL cholesterol. Fibrinogen was determined by an automated modification of the Clauss method, von Willebrand Factor antigen was measured by double antibody
252 Kumari et al.
AEP Vol. 14, No. 4 April 2004: 250–257
CHD RISK FACTORS AND CHANGES IN FUNCTIONING
ELISA. Technical error was estimated by assaying blinded duplicate samples for 5% of subjects. Coefficients of variation were 2.0 to 6.6 percent for glucose, cholesterol, triglycerides, and HDL-cholesterol, 23% for insulin, 16.4% for vWF, and 7.7% for fibrinogen. Poor health was defined as: angina, self report of doctordiagnosed heart attack, or angina, probable or possible ischemia on resting electrocardiogram, hypertension (⬎160/ 90 mmHg or taking antihypertensive drugs), claudication, diabetes, chronic bronchitis, musculoskeletal disorders, cancer, and those that scored above 5 in the general health questionnaire (15) at baseline (phase 3). At phase 4 those participants that reported doctor-diagnosed diabetes or heart attack or angina were also classified as being in poor health. Analyses Due to skewed distribution, all the biochemical variables were logged for the purposes of analysis. Linear regression models were used to assess associations between each biological and biochemical variable and changes in functioning using SAS. Pearson (product moment) correlation coefficients between the 10 biological variables under consideration were computed. Principal component analysis followed by orthogonal rotation was performed to examine the clustering of these variables. A summary accumulation score containing the variables that loaded onto the initial factor was created with scores grouped as 0, 1, 2 and 3⫹. The cut off values were, for men, waist hip ratio: 0.94, triglycerides: 1.91, fasting insulin: 8.25, HDL cholesterol: 1.07, and total cholesterol: 7.19, and for women, waist hip ratio: 0.82, triglycerides 1.44, fasting insulin: 8.00, HDL cholesterol: 1.37, and total cholesterol: 7.26. To assess associations with change in functioning, baseline functioning was controlled for and coefficients for other variables provide estimates of the impact of these factors on change in performance from baseline. Models were examined controlling for possible confounding by socio-demographic factors. Change in functioning was examined in relation to baseline biological variables and accumulation scores and to smoking, alcohol intake, and physical activity. Sensitivity analyses were performed by examining change in functioning in each tertile of baseline functioning component score. Analyses were also repeated on a population restricted to those participants that reported the top score in the physical functioning scale at baseline. Logistic regression was used to examine association with declines of a score of 10 or more between baseline and follow up. To calculate the contribution of each of the explanatory variables to the association between the biological variable and change in functioning summary scores, the percentage reduction in difference was calculated by including the variable, appropriate
socio-demographic factors, and comparing the regression coefficient for the biological variable before and after adjustment for each explanatory factor or group of factors.
RESULTS Table 1 shows the matrix of Pearson’s (product moment) correlations at baseline (phase 3) stratified by sex. Most of the cardiovascular risk factors examined are correlated and minor differences exist between men and women. These variables were subjected to a principal component analysis using the principal axis method to extract the components followed by varimax rotation. While four components displayed Eigenvalues greater than 1, examination of a Scree plot indicated that only the first three components were meaningful. Therefore, these components were retained by rotation. However, combined, these components explained 55% of total variance. The rotated factor pattern is shown in Table 2. An item was considered to load on a given component if the factor loading was 0.4 or greater. Using these criteria, the initial component appeared to reflect insulin resistance, showing high factor loadings for waist hip ratio, fasting insulin, triglycerides, HDL cholesterol, and total cholesterol. Diastolic and systolic blood pressure, representing hypertension, comprised the second factor, and fibrinogen and vWF the third factor representing hemostasis. Post load glucose failed to load and was not used in subsequent analyses. Because of the highly skewed nature of the SF-36 health functioning scores at each phase, Table 3 shows median scores in the physical functioning component score (PCS) and mental functioning component score (MCS) at phase 3 (1991–1993) and phase 4 (1995) of the study. Mean change in functioning was normally distributed and so the mean change between phase 3 and phase 4 is presented. A high score represents good functioning and a negative score indicates a decline in functioning. There were declines in PCS and MCS in men and women. The relationship between each of the risk factors for CHD that had high factor loadings for the initial component and subsequent functioning are presented in Table 4. Increasing waist hip ratio, triglycerides, fasting insulin, and decreasing HDL-cholesterol are associated with a decline in PCS for men and women while total cholesterol is not. A sensitivity analysis was performed by tertiling baseline physical functioning score and examining physical functioning as an outcome by each biological variable in each tertile. Increasing biological risk was associated with a greater change (decline) in functioning within each tertile. Post load insulin was also associated with decline in PCS in men (N ⫽ 4182, p ⫽ 0.002 test for trend, data not shown) but not women. These analyses were also repeated with the
AEP Vol. 14, No. 4 April 2004: 250–257
253
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
TABLE 1. Matrix of Pearson’s correlation coefficients among the cardiovascular variables in men and women
Waist hip ratio Post load glucose Cholesterol Fasting insulin Triglycerides HDL-cholesterol Diastolic blood pressure Systolic blood pressure Fibrinogen Von Willebrand factor
Waist hip ratio
Post load glucose
· 0.222 0.175 0.464 0.406 ⫺0.350 0.224
Triglycerides
HDLcholesterol
Diastolic blood pressure
Systolic blood pressure
Fibrinogen
von Willebrand factor
0.423 0.205 0.121 · 0.367 ⫺0.302 0.235
0.323 0.155 0.370 0.333 · ⫺0.428 0.152
ⴚ0.311 ⴚ0.143 0.044 ⴚ0.272 ⴚ0.478 · ⫺0.034 ns
0.299 0.166 0.141 0.266 0.177 ⴚ0.086 ·
0.270 0.173 0.111 0.222 0.135 ⴚ0.031 0.739
0.191 0.050 0.100 0.131 0.125 ⴚ0.159 0.056
0.080 0.092 0.034 0.136 0.025 ns ⴚ0.013 ns ⴚ0.013 ns
0.196
0.236
0.194
⫺0.048
0.757
0.138 0.051
0.234 0.153
0.182 0.082
⫺0.207 ⫺0.663
0.080 0.052
Cholesterol
Fasting insulin
0.165 · 0.093 0.251 0.206 ⫺0.134 0.167
0.191 0.065 · 0.112 0.378 0.078 0.168
0.238
0.213
0.244 0.138
0.134 0.113
· 0.103 0.077
0.062
0.026
· 0.243
0.179
Coefficients for men in bold and women in italics. All correlations significant (p ⬍ 0.05) unless indicated. ns ⫽ not significant.
additional risk factors for CHD, diastolic and systolic blood pressure, fasting and post load glucose, and von Willebrand factor. These risk factors were found not to be associated with change in physical functioning component score in men or women. Of the variables examined, only waist hip ratio was associated with change in the MCS. When the analyses were repeated following the removal of those participants with poor health (15) all trends remained except for HDL cholesterol and triglycerides which no longer reached significance in men (N ⫽ 2818, p ⫽ 0.14 test for trend) and women (N ⫽ 1105, p ⫽ 0.36 test for trend), respectively. Removal of those participants who reported doctor diagnosed diabetes and CHD at phase 4 (follow up) did not substantially alter our findings. When we examined the cumulative effects of dysregulation in the variables that had high loading on the initial factor we found that participants who scored 3⫹ at baseline on our summary score experienced a decline in physical functioning that was twice as large as those that scored 0. When these analyses were repeated using adverse quartile
of “blood pressure” or “hemostatic” variables (a score of 1 or 2 vs. 0) there was a similar or smaller decline in PCS (Figure 1) indicating that the additional variables provide no extra information with regard to decline in functioning. When the analyses were repeated restricting the population to those participants that reported maximal functioning at baseline, the metabolic summary score for men remained significantly associated with decline in functioning (N ⫽ 1621, p ⫽ 0.03) while the effect for women was no longer significant (N ⫽ 464, p ⫽ 0.41). Examination of the association between “insulin resistance” component score and odds of a decline of 10 or more showed that the insulin resistance component is associated with an odds ratio of 1.56 (1.04–2.08), 1.66 (1.1–2.6), and 1.83 (1.1–3.1) for increasing load in women. The effects in men were less robust and were 0.76 (0.54–1.1), 0.94 (0.6–1.1), and 1.15 (0.79–1.68) with increasing insulin resistance load. Table 5 shows that change in PCS is also associated with smoking, exercise, and alcohol intake in men. Smoking did not predict a decline in PCS in women. Alcohol intake
TABLE 2. Factor loading pattern after orthogonal rotation of principle components in men and women in phase 3 of the Whitehall II study Men (N ⫽ 4491) Variable Waist hip ratio Post load glucose Cholesterol Fasting insulin Triglycerides HDL-cholesterol Diastolic blood pressure Systolic blood pressure Fibrinogen Von Willebrand factor Cumulative % total variance
Women (N ⫽ 2013)
Factor 1
Factor 2
Factor 3
Factor 1
Factor 2
Factor 3
0.58 0.15 0.43 0.46 0.85 ⴚ0.71 0.16 0.05 0.17 ⫺0.14 27
0.28 0.26 0.13 0.25 0.08 0.11 0.91 0.90 ⫺0.03 ⫺0.02 42
0.27 0.32 ⫺0.06 0.37 0.02 ⫺0.12 0.04 ⫺0.00 0.61 0.80 53
0.69 0.31 0.40 0.61 0.81 ⴚ0.69 0.09 0.06 0.27 ⫺0.02 29
0.19 0.23 0.30 0.16 0.18 0.19 0.90 0.90 0.01 0.05 44
0.20 0.23 ⫺0.14 0.24 ⫺0.06 ⫺0.10 0.09 0.07 0.65 0.82 55
254 Kumari et al.
AEP Vol. 14, No. 4 April 2004: 250–257
CHD RISK FACTORS AND CHANGES IN FUNCTIONING
TABLE 3. SF-36 component scores at phase 3 (1991–1993) and phase 4 (1995) and change in score between these phases. Values represent median (interquartile range) scores except where noted Physical component summary score
Score at phase 3 Score at phase 4 Change in score*
Mental component summary score
Men
N
Women
N
Men
N
Women
N
54.3 (50.5, 57.0) 53.9 (49.6, 56.7) ⫺1.5 (⫺1.7, ⫺1.3)
5729 5286 5020
52.1 (45.9, 56.1) 52.0 (44.3, 56.1) ⫺1.5 (⫺1.8, ⫺1.2)
2563 2300 2155
53.6 (48.6, 57.0) 54.6 (47.2, 58.8) ⫺1.1 (⫺1.3, ⫺0.8)
5729 5286 5020
52.6 (46.4, 56.7) 52.0 (42.2, 57.1) ⫺1.1 (⫺1.5, ⫺0.7)
2563 2300 2155
*Mean change in score between phase 3 and phase 4 (95% CI).
was not associated with change in MCS for men while only exercise was associated with change in MCS in women. Adjustment for those health related behaviors associated with change in PCS reveals that exercise, smoking, and alcohol intake at baseline played a small role as determinants of the relationship between the biological factors and decline in functioning. Additional adjustment for BMI, accounted for up to 40% of the relationship between the
biological variables and decline in functioning in men and women. Adjustment for BMI accounts for much of the explanatory power as when only smoking, exercise, and alcohol are adjusted for then less than 20% of the relationship between the biological variables and decline in PCS can be accounted for. Similar data are apparent for women. The relationship between the accumulation score and decline in physical functioning was also minimally determined
TABLE 4. Change in physical component summary score of SF-36 by biological factors. Values represent mean change in score between phase 3 (1991–1993) and phase 4 (1995) of the study adjusted for age, employment grade, length of follow up and baseline score Quartiles
Men
Waist hip ratio
(N ⫽ 4625) Less than 0.87 0.87–0.90 0.90–0.94 More than 0.94
Cholesterol (mmol/l)
(N ⫽ 4672) Less than 5.69 5.69–6.41 6.41–7.19 More than 7.19
Triglycerides (mmol/l)
Fasting insulin (pmol/l)
HDL-cholesterol (mmol/l)
Accumulation score*
(N ⫽ 4672) Less than 0.91 0.91–1.31 1.31–1.91 More than 1.91 (N ⫽ 4227) Less than 3.51 3.51–5.45 5.45–8.26 More than 8.26 (N ⫽ 4655) Less than 1.07 1.07–1.27 1.27–1.51 More than 1.51 0 1 2 3⫹
Women Mean change ⫺0.9 ⫺1.5 ⫺1.4 ⫺2.0 P ⫽ 0.0001 ⫺1.25 ⫺1.68 ⫺1.49 ⫺1.57 P ⫽ 0.41 ⫺0.8 ⫺1.6 ⫺1.7 ⫺2.0 P ⫽ 0.0001 ⫺1.2 ⫺1.4 ⫺1.3 ⫺2.1 P ⫽ 0.0002 ⫺2.3 ⫺1.5 ⫺1.0 ⫺1.3 P ⫽ 0.0020 ⫺1.0 ⫺1.4 ⫺2.0 ⫺2.3 P ⫽ 0.0001
(N ⫽ 1948) Less than 0.73 0.73–0.77 0.77–0.82 More than 0.82 (N ⫽ 1945) Less than 5.69 5.69–6.45 6.45–7.26 More than 7.26 (N ⫽ 1945) Less than 0.75 0.75–1.01 1.01–1.44 More than 1.44 (N ⫽ 1693) Less than 3.22 3.22–4.98 4.98–7.98 More than 7.98 (N ⫽ 1942) Less than 1.37 1.37–1.63 1.63–1.93 More than 1.93 0 1 2 3⫹
P values are test for linear trend. *Accumulation score: sum of variables for which the participants were rated in the most adverse risk quartile of these variables.
Mean change ⫺1.2 ⫺1.7 ⫺1.2 ⫺2.23 P ⫽ 0.0162 ⫺1.54 ⫺1.97 ⫺1.50 ⫺1.33 P ⫽ 0.80 ⫺1.2 ⫺1.5 ⫺1.1 ⫺2.1 P ⫽ 0.0113 ⫺1.4 ⫺1.1 ⫺1.3 ⫺2.4 P ⫽ 0.0039 ⫺1.9 ⫺1.6 ⫺1.8 ⫺0.7 P ⫽ 0.0047 ⫺1.2 ⫺1.7 ⫺2.2 ⫺2.7 P ⫽ 0.01
AEP Vol. 14, No. 4 April 2004: 250–257
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
255
Figure 1. Change in PCS by score in adverse quartile of physiological variables. The figure represents change in PCS by score in adverse quartile in all variables that showed a high loading for the initial component (metabolic’) scored as 0, 1, 2, and 3⫹; in variables that showed a high factor loading for the second component (blood pressure) scored at 0 or 1 and 2 and variables that showed a high factor loading for third component (haemostatic), scored as 0 or 1 and 2. N ⫽ 4688 men and 1597 women in the total population. The ‘restricted’ group represents change in PCS by score in adverse quartile in all variables that showed a high loading for the initial component score in a population that reported maximum functioning at baseline. N ⫽ 1621 men and 464 women. Columns represent mean change in PCS between phase 3 ((1991–1993) and phase 4 (1995) of the study adjusted for age, employment grade, length of follow up and baseline score.
by health related behaviors, accounting for 11% of the relationship in men and 3% in women.
DISCUSSION These data suggest that some, but not all, risk factors for CHD are risk factors for decline in functioning in healthy individuals. In particular waist hip ratio, fasting insulin, triglycerides, and HDL-cholesterol are associated with a subsequent decline in physical functioning in both men and women. These variables, which were cumulative in their effect, load onto a single factor with total cholesterol that may represent insulin resistance which is a component of the metabolic syndrome (19, 22, 23). These data therefore suggest that insulin resistance may be an important determinant of subsequent physical functioning. Decline in functioning may be occurring as a consequence of overt disease, subclinical disease or independent of disease. It would seem unlikely that the first effect explains
the observed associations as removal of those with poor health from the analyses failed to markedly alter the outcomes. The results from these and previous analyses (24, 25) suggest that the subclinical condition that appears to be most associated with a decline in functioning is insulin resistance. The results are specific to this condition as other risk factors for CHD were without effect. We will be able to examine the effects of worsening subclinical disease in subsequent analyses from this cohort with the availability of clinical data. The decline in physical component summary scores seemed selectively sensitive to biological factors that were related to insulin resistance as the additional variables were less robustly associated with decline in functioning. The components of the cluster have cumulative effects on decline in functioning which accords with the idea of allostatic load effects on “wear and tear” (8, 9, 10, 26). It is possible that insulin resistance and the processes that lead to insulin resistance precipitate a more immediate decline in physical functioning than the other risk factors for CHD
256 Kumari et al.
AEP Vol. 14, No. 4 April 2004: 250–257
CHD RISK FACTORS AND CHANGES IN FUNCTIONING
TABLE 5. Change in physical and mental component summary score between phase 3 (1991–1993) and phase 4 (1995) by smoking, alcohol intake, and exercise status at phase 3 (1991–1993) Men
Women
Physical component
Mental component
N
Physical component
Mental component
N
⫺1.4 ⫺1.4 ⫺2.3 P ⫽ 0.006
⫺1.3 ⫺1.1 ⫺0.7 P ⫽ 0.31
2640 2085 678
⫺1.8 ⫺1.4 ⫺1.6
⫺1.0 ⫺0.9 ⫺1.0
1344 631 394
⫺2.1 ⫺1.7 ⫺1.0 P ⫽ 0.002
⫺0.1 ⫺1.2 ⫺1.4 P ⫽ 0.007
781 2536 2182
⫺1.8 ⫺1.7 ⫺0.8
⫺1.0 ⫺1.0 ⫺1.5
851 1149 431
⫺1.9 ⫺1.4 ⫺1.1 ⫺1.9 P ⫽ 0.043
⫺0.6 ⫺1.1 ⫺1.3 ⫺1.3
797 3007 682 1008
⫺2.3 ⫺1.2 ⫺1.3
⫺0.5 ⫺1.3 ⫺1.9
740 1463 226
Smoking Never Ex Current Exercise None/mild Moderate Vigorous Alcohol intake (units) None 1–14 15–21 (15⫹ Female) 22⫹ (male)
P ⫽ 0.033
P values are test for linear trend.
that were investigated, although the mechanisms by which this might occur remain unclear. The associations between waist hip ratio, triglycerides, and fasting insulin and subsequent change in the physical summary component score were not subsequent to BMI and poor health related behaviors. The importance of BMI and change in weight with respect to health status have been described (27, 28), however the role of waist hip ratio has not been determined. The stronger association between waist hip ratio and decline in functioning suggest that either BMI is a poorer measure of adiposity than waist hip ratio or that waist hip ratio is a more important determinant of subsequent functioning than BMI in men. In women, both BMI and waist hip ratio were independently associated with change in functioning. The failure of BMI to explain the relationships between the biological variables and change in functioning suggests that these effects are not simply determined by overweight but by additional mechanisms. Adjustment for BMI failed fully to explain the relationship between the accumulation of the variables and change in functioning again suggesting additional mechanisms are operating. However, it may be thought inappropriate to adjust for BMI in this context as it is possible that BMI could be considered part of a construct reflecting underlying insulin resistance. The health related behaviors were found not to be important determinants of the relationship between the risk factors examined and change in functioning. This suggests that additional, direct processes may be involved, for example, physiological responses to stress as outlined in the allostatic load hypothesis (8–10), leading to the decline in functioning associated with insulin resistance. Follow up of
the cohort and examination of the trajectory of change in functioning with respect to the factors that have been described at baseline will allow us to describe the effects of insulin resistance and changes in functioning as the cohort ages. Limitations of the study need to be considered. Nonresponse at follow up is likely to have biased the effects conservatively, since non-responders tended to be from lower employment grades and have lower SF-36 scores at baseline. The size of the declines observed were small in comparison to the effects observed in the diseased state and their impact on daily life is unclear (29). However, the declines are important in this relatively young and healthy working population as they occurred in a short follow up period of 36 months which if continued at the same rate, as they may reflect ongoing processes, would lead to marked disability. In this short term follow up the initial factor was predictive of clinically relevant declines in functioning in women. It will be possible to ascertain the further decline in functioning should it occur with continued follow up of the group. In conclusion, changes in functioning are associated with some CHD risk factors, in particular, waist hip ratio, HDL cholesterol, triglycerides, and fasting insulin. These associations are not determined by health related behaviors but by other factors as yet identified. These variables cluster and may represent insulin resistance. These data therefore identify insulin resistance as a potential area of intervention to prevent functional decline. REFERENCES 1. Harris T, Kovar M, Suzman R, Kleinman JC, Feldman JJ. Longitudinal study of physical ability in the oldest old. Am J Public Health. 1989;79:698–702.
AEP Vol. 14, No. 4 April 2004: 250–257
2. Kaplan GA, Camadio T. Perceived health and mortality: A nine-year follow up of the human population laboratory cohort. Am J Epidemiol. 1983;117:292–304. 3. Idler EL, Angel RJ. Self-rated health and mortality in the NHANES-1 epidemiological follow-up study. Am J Public Health. 1990;80:446–452. 4. Garret AM, Rota RJ, Abdalla M, Buckingham JK, Russell IT. The SF-36 health survey questionnaire: An outcome measure suitable for routine use within the NHS? BMJ. 1993;306:1440–1444. 5. Rowe JW, Jahn RL. Human ageing: usual and successful. Science. 1987; 237:143–149. 6. Seeman TE, Charpentier PA, Berkaman LF, Tinetti ME, Guralnik JM, Albert M, et al. Predicting changes in physical performance in a high functioning elderly cohort: MacArthur studies of successful aging. J Gerontol. 1994;49:M97–M108. 7. Seeman TE, Singer BH, Rowe JW, Horwitz RI, McEwen BS. The price of adaptation—allostatic load and its health consequences: MacArthur studies of successful aging. Arch Intern Med. 1997;157:2259–2268. 8. McEwen BS, Stellar E. Stress and the individual. Mechanisms leading to disease. Arch Intern Med. 1993;153:2093–2101. 9. McEwen BS. Stress, adaptation and disease. Allostasis and allostatic load. Ann NY Acad Sci. 1998;840:33–44. 10. McEwen BS. Protective and damaging effects of stress mediators. N Engl J Med. 1998;338:171–179. 11. Ware JE, Sherbourne CD. The MOS 36-item short-form health Survey (SF-36): I. Conceptual framework and item selection. Med Care. 1992; 30:473–483. 12. Jenkinson C, Wright L, Coulter A. Criterion validity and reliability of the SF-36 in a population sample. Qual Life Res. 1994;3:7–12. 13. Lyons RA, Lo SV, Littlepage BVC. Comparative health status of patients with 11 common illnesses in Wales. J Epidemiol Community Health. 1994;48:388–390. 14. Hayes V, Morris J, Wolfe C, Morgan M. The SF-36 Health survey questionnaire: Is it suitable for use with older adults? Age Ageing. 1995;24:120–125. 15. Hemingway H, Stafford M, Stansfeld S, Shipley M, Marmot M. Is the SF-36 a valid measure of change in population health? Results from the Whitehall II study. BMJ. 1997;315:1273–1279. 16. Stansfeld SA, Smith GD, Marmot M. Association between physical and psychological morbidity in the Whitehall II study. J Psychosom Res. 1993; 37:227–238.
Kumari et al. CHD RISK FACTORS AND CHANGES IN FUNCTIONING
257
17. Bruce ML, Seeman TE, Merrill SS, Blazer DG. The impact of depressive symptomatology on physical disability: MacArthur studies of successful aging. Am J Public Health. 1994;84:1796–1799. 18. Marmot MG, Smith GD, Stansfeld S, Patel C, North F, Head J, et al. Health inequalities among British Civil Servants: The Whitehall II study. Lancet. 1991;337:1387–1393. 19. Brunner EJ, Marmot MG, Nanchahal K, Shipley MJ, Stansfeld SA, Juneja M, et al. Social inequality in coronary risk: Central obesity and the metabolic syndrome. Evidence from the Whitehall II study. Diabetologia. 1997;40:1341–1349. 20. Ware JE, Kosinski M, Bayliss MS, McHorney CA, Roger WH, Raczek A. Comparison of methods for the scoring and statistical analysis of SF36 health profile and summary measures: Summary of results from the medical outcomes study. Med Care. 1995;33:264–279. 21. Ware JE, Kosinski M, Keller S. SF-36 Physical and Mental Summary Scales: A User’s Manual. Boston: The Health Institute, New England Medical Center; 1994. 22. Gray RS, Fabsitz RR, Cowan LD, Lee ET, Howard BV, Savage PJ. Risk factor clustering in the insulin resistance syndrome. The Strong Heart Study. Am J Epidemiol. 1998;148:869–878. 23. Chen C-H, Lin K-C, Tsai S-T, Chou P. Different association of hypertension and insulin-related metabolic syndrome between men and women in 8437 non diabetic Chinese. Am J Hypertens. 2000;13:846–853. 24. Martikainen P, Stansfeld S, Hemingway H, Marmot M. Determinants of socioeconomic differences in change in physical and mental functioning. Soc Sci Med. 1999;49:499–507. 25. Stafford M, Hemingway H, Stansfeld S, Brunner E, Marmot M. Behavioural and biological correlates of physical functioning in middle aged office workers: The UK Whitehall II study. J Epidemiol Community Health. 1998;52:353–358. 26. Seeman TE, McEwen BS, Rowe JW, Singer BH. Allostatic load as a marker of cumulative biological risk: MacArthur studies of successful aging. Proc Natl Acad Sci USA. 2001;98:4770–4775. 27. Stafford M, Hemingway H, Marmot M. Current obesity, steady weight change, and weight fluctuation as predictors of physical functioning in middle aged office workers: the Whitehall II Study. Int J Obes Relat Metab Disord. 1998;22:23–31. 28. Doll HA, Petersen SEK, Stewart-Brown SL. Obesity and physical and emotional well-being: Associations between body mass index, chronic illness, and the physical and mental components of the SF-36 questionnaire. Obes Res. 2000;8:160–170. 29. Muldoon MF, Barger SD, Flory JD, Manuck SB. What are quality of life measurements measuring? BMJ. 1998:542–545.