1 Institute for Research in Extramural Medicine (EMGO-institute), VU University Medical Center, Amsterdam, The Netherlands
2 Department of Clinical Epidemiology and Biostatistics, VU University Medical Center, Amsterdam, The Netherlands
3 Institute for Cardiovascular Research and Department of Internal Medicine, VU University Medical Center, Amsterdam, The Netherlands
4 Department of Social Medicine and Body@Work research centre for physical activity, work and health TNO-VU, VU University Medical Center, Amsterdam, The Netherlands
Correspondence: Prof dr HCG Kemper, Institute for Extramural Medicine (EMGO-institute), VU University Medical Center, Officia 1, De Boelelaan 7, 1083 HJ Amsterdam, The Netherlands. E-mail: hcg.kemper.emgo{at}med.vu.nl
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methods In 281 subjects (161 women), with a mean age of 36, blood pressure was measured. The diameter, distension, and local pulse pressure of three large arteries were measured simultaneously using ultrasound imaging. Local and regional arterial compliance and distensibility were calculated. Information on birthweight was retrieved with a questionnaire.
Results Linear regression analyses showed a 3.3 mmHg lower systolic blood pressure (SBP) and a 1.8 mmHg lower diastolic blood pressure (DBP), per 1kg higher birthweight. These associations were statistically significant after adjustment for adult weight. Birthweight was significantly and positively related to carotid arterial compliance (P = 0.050), but less so to brachial (P = 0.114) and femoral arterial compliance (P = 0.058). However, after adjustment for adult height, the strength of these associations decreased. Birthweight was not related to arterial distensibility. The association between birthweight and arterial compliance could only partly explain the association between birthweight and blood pressure.
Conclusions Lower birthweight is related to increased blood pressure, and increased arterial stiffness. However, the latter relationship can only partly explain the association between birthweight and blood pressure. Therefore, mechanisms other than arterial stiffness contribute to the birthweightblood pressure relationship.
Accepted 4 August 2003
The association between low birthweight and increased blood pressure in later life has been repeatedly confirmed,1,2 but remains controversial.3 If a relationship exists, it would imply that mechanisms that initiate an increase in blood pressure may have their origin in fetal development. In this line, Martyn and Greenwald4 have postulated the hypothesis that synthesis of elastin in the aorta and large arteries may be reduced in fetuses whose growth is impaired, leading to permanent stiffening of these vessels and raised blood pressure in later life. In fact, arterial stiffness has been identified as an antecedent factor of hypertension.5
Few studies have addressed the relationship between birthweight and arterial stiffness,69 and their results have been inconsistent.10 In addition, in these studies, arterial stiffness has been estimated from pulse wave velocity (PWV), an indicator of regional arterial viscoelastic properties. To the best of our knowledge no studies have investigated the relationship between fetal growth and local estimates of arterial stiffness, such as distensibility and compliance. Such a study could be highly informative, since the arterial tree is not uniform and some beds can be more susceptible to risk factors than others. Moreover, and in the perspective of the current debate on the fetal origins hypothesis, it is interesting to know whether or not adult weight or height have modifying or confounding effects. If so, then it is not birthweight alone which is important, but also adult weight or height.
Therefore, the aims of this study were to investigate whether, in a healthy adult population born at term: (1) birthweight is related to blood pressure, (2) birthweight is related to regional and local (carotid, brachial, and femoral) estimates of arterial stiffness, and (3) whether the association (if any) between birthweight and blood pressure can be explained by an association between birthweight and arterial stiffness.
![]() |
Methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
All subjects gave informed consent and the Medical Ethical Committee of the Vrije Universiteit Amsterdam approved the protocol.
Data collection
Birthweight
Data on birthweight, gestation, and source of information were retrieved by a questionnaire. As mentioned above, most subjects could retrieve information from written documents, others from their parents memory, which has been proven to be a valid method.18 Subjects born preterm (gestational age <37 weeks [29 subjects]) were excluded since being born preterm may have independent effects on adult health or the relationship between birthweight and adult health (for instance blood pressure) might be different in subjects born preterm.9,19,20 Twins were excluded because twins have different fetal growth patterns, which might cause error when analysing the relationship between birthweight and adult blood pressure and arterial stiffness.
Arterial stiffness
Properties of the right common carotid (CCA), brachial (BA), and the common femoral arteries (CFA) were obtained with a B + M-mode ultrasound scanner equipped with a 7.5 MHz linear array transducer (Pie Medical, Maastricht, The Netherlands). The ultrasound scanner was connected to a personal computer equipped with an acquisition system and a vessel wall movement detector software system (Wall Track System 2, Pie Medical, Maastricht, The Netherlands). This integrated device enables measures of arterial diameter, distension, and pulse wave transit time.21,22 All subjects had abstained from smoking and caffeine-containing beverages on the day the measurements were performed. At the time of measurement of arterial properties, subjects had been resting in a supine position for 15 minutes in a quiet, temperature-controlled room.
Local arterial stiffness
The mean diameter (D) and distension (D) of three consecutive measurements on each artery, and the mean local pulse pressure (
P) (described below) of three measurements obtained at approximately the moment of those measurements, were used to estimate distensibility and compliance as follows:23
![]() | (1) |
![]() | (2) |
Regional stiffness
While measuring diameter and distension, the time delay between the foot of the distension waveforms and the R-wave of the simultaneously recorded ECG, i.e. the pulse wave transit time (TT), was determined for each artery. Each recording included 37 heartbeats and the mean TT of three recordings in each artery where calculated. The carotid pulse wave TT was then subtracted from the femoral pulse wave TT to obtain the carotido-femoral TT. We used this carotido-femoral TT (in ms) as an indicator of regional stiffness.25 The TT of a pulse wave from the carotid to the femoral artery is negatively related to mean arterial stiffness over this segment and closely related to carotido-femoral PWV, as length of the carotido-femoral segment divided by the segment pulse wave TT.26 Measurement of this length (usually performed with a tape measurer over the subject's body surface) may introduce error. We therefore used the carotido-femoral transit time adjusted for sitting height in the analyses (and not height, avoiding the error introduced by long trunk-short legs or short trunk-long legs variability among subjects). Due to technical reasons (deficient ECG signalling tracing), data on PWV was available only on 234 subjects (133 women). Subject without this information did not differ significantly from the ones included with regard to birthweight, adult body weight, height, body mass index (BMI), systolic blood pressure (SBP), and diastolic blood pressure (DBP).
Blood pressure
Throughout the entire period of ultrasound imaging, SBP, DBP, and mean blood pressure were assessed (mean is used) in the left arm at 5-minute intervals with an oscillometric device (Colin Press-Mate, model BP 8800, Komaki-City, Japan). Brachial artery pulse pressure was defined as SBP minus DBP, and pulse pressure at the common carotid and femoral arteries was calculated by calibration of the distension waveforms.27 This method of calibration is based on the observation that mean arterial pressure is constant throughout the large artery tree and that DBP does not change substantially. Distension waveforms were assessed at the target (i.e. the carotid and femoral arteries) and at the reference artery (i.e. the brachial artery). Based on these distension waveforms and on pulse pressure measured at the reference artery, pulse pressure was calculated at each target artery as follows: PPtarget = PPreference * CF (Calibration factor), in which CF = [PP/(mean arterial pressure DBP)target]/ [PP/(mean arterial pressure DBP)reference]. [mean arterial pressure DBP] can be calculated from the area under the pressure curve divided by time, an similarly using the distension waveforms instead.27
Both inter-observer reliability and intra-observer reliability were good. For instance the reproducibility as assessed by inter-observer coefficients were on average: diameter, 2.9% (carotid), 4.6% (brachial), and 2.8% (femoral); and distension, 6.4% (carotid), 27.7% (brachial), and 24.2% (femoral); and carotido-femoral TT 12.3%.22
Potential covariates/effect modifiersAdult weight and height
Body weight (to the nearest 0.1 kg) and body height (to the nearest 0.001 m) were measured according to standard procedures,28 with subjects dressed only in underwear.
Statistical analyses
Multiple linear regression analyses were used to assess the relationship between birthweight on the one hand and blood pressure and arterial stiffness estimates on the other. The associations were assessed firstly with adjustment for sex (and mean blood pressure in the case of analyses with stiffness estimates as outcome variables) (model 1), and secondly with additional adjustment for adult weight and/or height. The level of significance for the associations between birthweight and the outcome variables was set at P < 0.05.
To analyse whether an observed association between birthweight and blood pressure can be explained by an association between birthweight and arterial stiffness, stiffness estimates were added to the models with birthweight and blood pressure.
It was then analysed whether or not sex was an effect modifier in the relationships of interest. When the interaction showed a P-value < 0.1, results were presented separately for men and woman.
All analyses were carried out with the Statistical package of Social Sciences, version 10.1. for Windows (SPSS, Inc, Chicago, IL).
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
Birthweight was inversely and significantly associated with the local pulse pressure of the femoral artery. The relationship between birthweight and femoral compliance was borderline significant (P = 0.058), but decreased when height was introduced into the model (model 2).
Birthweight was associated with regional stiffness in a different fashion for men and women (P = 0.021 for interaction between birthweight and sex). For women the association was in a positive direction, while for men the association was equally strong but negative.
Next, analyses were performed to study the mediating effect of all stiffness estimates used in this study on the relationship between birthweight and SBP. Results showed that adding femoral compliance into the model which already included birthweight as a determinant, sex and weight as covariates, and SBP as an outcome, changed the regression coefficient of birthweight from −3.31 mmHg/kg to −2.41 mmHg/kg (P = 0.045). Adding carotid or brachial compliance instead of femoral compliance resulted in smaller changes in the regression coefficient (to ß =−2.60 mmHg/kg and ß =−2.83 mmHg/kg respectively).
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We observed that birthweight was indeed negatively associated with both SBP and DBP, which was statistically significant after adjustment for adult weight. The estimates for the effect of birthweight on SBP, whether or not adjusted, were comparable with those in other studies.1,2
The present study was the first in which the relationship between birthweight and stiffness indicators in three large arteries (i.e. common carotid, brachial, and femoral artery) was investigated. Birthweight was positively associated with the compliance of all three arteries investigated, but this decreased after adjustment for adult height. That birthweight was associated with compliance but not with distensibility is most probably due to the association between birthweight and diameter (in carotid and brachial arteries). Compliance and distensibility reflect two different arterial properties, elastic and buffering capacities respectively, and these results suggest that birthweight is more related to the size of the arteries and the buffering capacity than to elasticity.
A further essential question was whether or not the association between birthweight and blood pressure could be explained by the relationship between birthweight and arterial stiffness, as Martyn and Greenwald4 have suggested. Adding carotid or brachial or femoral arterial compliance into the model resulted in a reduction of the regression coefficient for birthweight (from −3.31 to −2.60, −2.83, or −2.41, respectively), but the regression coefficients were still significant (P < 0.045). Considering possible inaccuracies in measurement of both birthweight and arterial stiffness, it can still be concluded that arterial stiffness may at most partly explain the relationship between birthweight and systolic blood pressure. Other determinants of blood pressure may therefore play an important role, such as nephron number, kidney function, sympathetic nervous system, and body composition.3034
Adjustment for adult size
Adjustments for adult size (i.e. weight, height) are controversial in the analyses of the relationships between birthweight and cardiovascular outcomes.35,36 Adult height can be both a confounder or a mediator in the relationship between birthweight and arterial stiffness. Models not adjusted for adult height would give the more correct estimate for the effect of birthweight if we assume that adult height is a mediator (e.g. low birthweight leads to low adult height, which in turn leads to smaller and stiffer arteries). On the other hand, if we assume that height is a confounder, i.e. when birthweight and adult height represent the same identity (e.g. a marker of growth), then the adjusted model will give the most correct estimate for the effect of birthweight. Both possibilities seem plausible, and the true estimate may thus be somewhere between these estimates. Different statistical models were also used to study the relationship between birthweight and adult blood pressure: a crude model and a model adjusted for adult weight. To interpret the results, it is again important to emphasize the assumptions implicit in these models. Similarly to adjusting for adult height, one can assume that adult weight is a mediator or a confounder, or both, in the relationship between birthweight and blood pressure. However, differences in weight can reflect differences in height, as well as differences in body composition not related to height. Thus, weight could reflect both height and body composition, such as fat and muscle mass. To investigate this important issue further, we re-analysed the relationship between birthweight and blood pressure. In a model that already included adult height, we added different components of adult weight (i.e. fat mass, lean mass, waist to hip ratio) in order to investigate which component plays the major role in the relationship. Results showed that adjusting for fat mass was most similar to adjustment for adult weight in the relationship between birthweight and blood pressure. Adjustment for lean mass resulted in a slightly weaker effect of birthweight on SBP and DBP, while adjustment for waist-to-hip ratio instead of adult weight showed the weakest effect of birthweight on blood pressure. These results suggest that fat mass is most responsible for the mediating or confounding effect in the relationship between birthweight and blood pressure. Both a confounding or mediating effect seem plausible, and again, the true value for the effect of birthweight may be somewhere between the two values given by the crude and adjusted models.
In the present study the regression coefficients of the associations between birthweight and arterial stiffness decreased after adjustment for adult height. This change was much smaller (although in the opposite direction) than the change of the regression coefficient after adjustment for adult weight in the relationship between birthweight and SBP. The pattern shown in the birthweightarterial stiffness relationship is different from the usual pattern observed in the associations between birthweight and SBP,3 but has been seen in associations with other health outcomes. For example, studies on sympathetic activity have shown that the relationship between birthweight and resting heart rate was independent of current size,33 and that birthweight was related to cardiac sympathetic activation independently of current size.34 The fact that the association between birthweight and arterial stiffness differs from the association between birthweight and adult blood pressure might suggest different underlying mechanisms.
Sex difference
In the present study, an interaction with sex was found for the relationships between birthweight and TT in the carotido-femoral segment. For men an inverse relationship was found, while for women a positive relationship was found. This difference could not be explained by differences in body size, because this difference was also found after correction for body size. However, in a study of Murray et al.10 a comparable sex difference was observed. Thus there may be a true sex difference, which implies that regional stiffness is different in this respect from local stiffness (as has been measured in the present study in the carotid and femoral artery). That these associations with regional stiffness are different from the associations with local stiffness is possible, since they represent arterial stiffness in different parts of the body. Furthermore, the human arterial tree is not homogeneous, as the distribution of collagen and elastin differs between arteries. If this interaction is genuine, then this may suggest that increased arterial stiffness (as measured with TT) cannot be the exclusive underlying mechanism in the birthweightblood pressure relationship, since that relationship is not modified by sex.37 To our knowledge, no other study has reported results on the relationship between birthweight and compliance separately for men and women, and we cannot exclude that the interaction we observed was a chance finding.
Elastic versus muscular arteries
The relationships between birthweight and compliance and distensibility were assessed in both elastic (CCA) as well as in muscular arteries (BA and CFA). According to the hypothesis postulated by Martyn and Greenwald,4 it could be expected that effects of fetal growth on the stiffness of an artery should be clearer in elastic arteries, since elastin synthesis is thought to be limited in those with retarded fetal growth or lower birthweight. In support, the association between birthweight and compliance in the elastic carotid artery was somewhat stronger than the association with compliance in the femoral and brachial artery. In the carotid and brachial artery the association was mainly due to the positive association with the diameter (i.e. arterial geometry), while in the femoral artery this was due to the strong negative relationship with local pulse pressure. Increased pulse pressure may result from early wave reflections, which could be caused by changes in small rather than in large arteries. Another factor which influences pulse pressure is stroke volume. Unfortunately, in the AGAHLS population stroke volume was not measured, but heart rate was measured, which could serve as an indicator. Additional analyses showed, however, that birthweight was not related to heart rate (P = 0.245). Thus, from this data it is not clear what caused the relationship between birthweight and pulse pressure in the femoral artery.
Study limitations
The method used in this study to measure birthweight may not be as precise as using birth certificates from hospitals. However, several studies have supported the method of retrospective questionnaires.1418 In The Netherlands it is difficult to retrieve information on birthweight and gestation from hospitals, since most mothers give birth to their babies at home. In the population of the AGAHLS 187 subjects (66.5%) were born at home. In additional analyses we observed that the methods used to retrieve information on birthweight (either written documents or parents' memory) did not affect the associations found (data not shown).
The population of the AGAHLS is a relatively healthy population. Most subjects had normal birthweights. Only a few reported birthweights 2.5 kg (7 subjects), which is often considered as low. Therefore, the variation in the main independent variable was relatively small, which may cause difficulties in detecting associations.
![]() |
Conclusions |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
KEY MESSAGES
|
![]() |
Acknowledgments |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2 Huxley RR, Shiell AW, Law CM. The role of size at birth and postnatal catch-up growth in determining systolic blood pressure: a systematic review of the literature. J Hypertens 2000;18:81531.[CrossRef][ISI][Medline]
3 Huxley R, Neil A, Collins R. Unravelling the fetal origins hypothesis: is there really an inverse association between birthweight and subsequent blood pressure? Lancet 2002;360:65965.[CrossRef][ISI][Medline]
4 Martyn CN, Greenwald SE. Impaired synthesis of elastin in walls of aorta and large conduit arteries during early development as an initiating event in pathogenesis of systemic hypertension. Lancet 1997;350:95355.[CrossRef][ISI][Medline]
5 Liao D, Arnett DK, Tyroler HA et al. Arterial stiffness and the development of hypertension. The ARIC study. Hypertension 1999;34:20106.
6 Martyn CN, Barker DJ, Jespersen S, Greenwald S, Osmond C, Berry C. Growth in utero, adult blood pressure, and arterial compliance. Br Heart J 1995;73:11621.[Abstract]
7 Kumaran K, Fall CH, Martyn CN, Vijayakumar M, Stein C, Shier R. Blood pressure, arterial compliance, and left ventricular mass: no relation to small size at birth in south Indian adults. Heart 2000;83:27277.
8 Montgomery AA, Ben-Shlomo Y, McCarthy A, Davies D, Elwood P, Smith GD. Birth size and arterial compliance in young adults. Lancet 2000;355:213637.[CrossRef][ISI][Medline]
9 Oren A, Vos LE, Bos WJ et al. Gestational age and birthweight in relation to aortic stiffness in healthy young adults: two separate mechanisms? Am J Hypertens 2003;16:7679.[ISI][Medline]
10 Murray LJ, Gallagher AM, Boreham CA, Savage M, Smith GD. Sex specific difference in the relation between birthweight and arterial compliance in young adults: The Young Hearts Project. J Epidemiol Community Health 2001;55:66566.
11 Kemper, HCG. The Amsterdam Growth Study: A Longitudinal Analysis of Health, Fitness and Lifestyle. Champaign (IL): Human Kinetics, 1995.
12 Kemper H, Snel J, Verschuur R, Storm-van Essen L. Tracking of health and risk indicators of cardiovascular diseases from teenager to adult: Amsterdam Growth and Health Study. Prev Med 1990;19:64255.[ISI][Medline]
13 Kemper HCG, van Mechelen W, Post GB et al. The Amsterdam Growth and Health Longitudinal Study. The past (19761996) and future (1997-?). Int J Sports Med 1997;18:S14050.[ISI][Medline]
14 Axelsson G, Rylander R. Validation of questionnaire reported miscarriage, malformation and birthweight. Int J Epidemiol 1984;13:9498.[Abstract]
15 Hoekelman RA, Kelly J, Zimmer AW. The reliability of maternal recall. Mother's remembrance of their infant's health and illness. Clin Pediatr (Phila) 1976;15:26165.[ISI][Medline]
16 Little RE. Birthweight and gestational age: mothers' estimates compared with state and hospital records. Am J Public Health 1986;76:135051.[Abstract]
17 Tilley BC, Barnes AB, Bergstralh E et al. A comparison of pregnancy history recall and medical records. Implications for retrospective studies. Am J Epidemiol 1985;121:26981.[Abstract]
18 Burns TL, Moll PP, Rost CA, Lauer RM. Mothers remember birthweights of adolescent children: the Muscatine Ponderosity Family Study. Int J Epidemiol 1987;16:55055.[Abstract]
19 Leon DA, Johansson M, Rasmussen F. Gestational age and growth rate of fetal mass are inversely associated with systolic blood pressure in young adults: an epidemiologic study of 165 136 Swedish men aged 18 years. Am J Epidemiol 2000;152:597604.
20 Siewert-Delle A, Ljungman S. The impact of birthweight and gestational age on blood pressure in adult life: a population-based study of 49-year-old men. Am J Hypertens 1998;11:94653.[CrossRef][ISI][Medline]
21 Brands PJ, Hoeks AP, Willigers J, Willekes C, Reneman RS. An integrated system for the non-invasive assessment of vessel wall and hemodynamic properties of large arteries by means of ultrasound. Eur J Ultrasound 1999;9:25766.[CrossRef][Medline]
22 Ferreira I, Twisk JW, Van Mechelen W, Kemper HC, Stehouwer CD. Current and adolescent levels of cardiopulmonary fitness are related to large artery properties at age 36: the Amsterdam Growth and Health Longitudinal Study. Eur J Clin Invest 2002;32:72331.[CrossRef][ISI][Medline]
23 Van Bortel LM, Duprez D, Starmans-Kool MJ et al. Clinical applications of arterial stiffness, Task Force III: recommendations for user procedures. Am J Hypertens 2002;15:44552.[CrossRef][ISI][Medline]
24 O'Rourke MF, Staessen JA, Vlachopoulos C, Duprez D, Plante GE. Clinical applications of arterial stiffness; definitions and reference values. Am J Hypertens 2002;15:42644.[CrossRef][ISI][Medline]
25 van Dijk RA, van Ittersum FJ, Westerhof N, van Dongen EM, Kamp O, Stehouwer CD. Determinants of brachial artery mean 24 h pulse pressure in individuals with Type II diabetes mellitus and untreated mild hypertension. Clin Sci (Lond) 2002;102:17786.[CrossRef][Medline]
26 Asmar RG, O'Rourke MF, Safar ME. Pulse wave velocity. Principles and measurement. In: Asmar RG, O'Rourke MF, Safar ME. Arterial Stiffness and Pulse Wave Velocity. Paris, France: Elsevier, 1999, pp. 2555.
27 Van Bortel LM, Balkestein EJ, van der Heijden-Spek JJ et al. Non-invasive assessment of local arterial pulse pressure: comparison of applanation tonometry and echo-tracking. J Hypertens 2001;19:103744.[CrossRef][ISI][Medline]
28 Weiner JS, Lourie J. Human Biology, A Guide to Field Methods IBP Handbook No. 9. Oxford: Blackwell, 1968.
29 Durnin JV, Womersley J. Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. Br J Nutr 1974;32:7797.[ISI][Medline]
30 Mackenzie HS, Brenner BM. Fewer nephrons at birth: a missing link in the etiology of essential hypertension? Am J Kidney Dis 1995;26:9198.[ISI][Medline]
31 Manalich R, Reyes L, Herrera M, Melendi C, Fundora I. Relationship between weight at birth and the number and size of renal glomeruli in humans: a histomorphometric study. Kidney Int 2000;58:77073.[CrossRef][ISI][Medline]
32 Lackland DT, Bendall HE, Osmond C, Egan BM, Barker DJ. Low birthweights contribute to high rates of early-onset chronic renal failure in the Southeastern United States. Arch Intern Med 2000;160:147276.
33 Phillips DI, Barker DJ. Association between low birthweight and high resting pulse in adult life: is the sympathetic nervous system involved in programming the insulin resistance syndrome? Diabet Med 1997;14:67377.[CrossRef][ISI][Medline]
34 IJzerman RG, Stehouwer CDA, De Geus EJ, van Weissenbruch MM, Delemarre-van de Waal HA, Boomsma DI. Low birthweight is associated with increased sympathetic activitydependence on genetic factors. Circulation 2003;108:56671.
35 Gillman MW. Epidemiological challenges in studying the fetal origins of adult chronic disease. Int J Epidemiol 2002;31:29499.
36 Lucas A, Fewtrell MS, Cole TJ. Fetal origins of adult disease-the hypothesis revisited. BMJ 1999;319:24549.
37 Lawlor DA, Ebrahim S, Davey Smith G. Is there a sex difference in the association between birthweight and systolic blood pressure in later life? Findings from a meta-regression analysis. Am J Epidemiol 2002;156:110004.