Height loss, forearm bone density and bone loss in menopausal women: a 15-year prospective study. The Nord-Trøndelag Health Study, Norway
- First Online:
- Cite this article as:
- Forsmo, S., Hvam, H.M., Rea, M.L. et al. Osteoporos Int (2007) 18: 1261. doi:10.1007/s00198-007-0369-1
- 42 Views
In a population-based cohort of 1,421 women 45–60 years old followed for 15.5 years, 71% of the women had lost height. Height loss was associated with low forearm bone density and increased bone loss, but high body weight and oestrogen therapy were protective factors. Increased height loss indicates a generalized state of bone loss.
The degree of height loss and its association to forearm bone mineral density (BMD) and bone loss was investigated in a population-based cohort of middle-aged women followed for more than 15 years.
Among 8,856 women aged 45–60 years attending the first HUNT Study, Norway (1984–86), a 35% random sample was invited to forearm densitometry 11.3 years later (HUNT 2, 1995–97), and 2,188 attended (78.3%). In 2001, 15.5 years since baseline, all were invited to follow-up densitometry and height measurement.
A total of 71.2% and 17.4% of the 1,421 women attending had lost >1 cm and >3 cm of height since baseline, respectively. Women aged ≥ 64 years at HUNT 2 had a relative risk (RR) for height loss >3 cm of 3.1 (95% CI 2.2, 4.3) compared to women <64 years. A strong and negative association was found between height loss and forearm BMD, adjusted for time since menopause. A high rate of height loss was associated to increased forearm bone loss. High body weight, oestrogen treatment and good self-rated health were protective against height loss.
Height loss is frequent in middle-aged women, and increased height loss indicates a generalized state of bone loss.
KeywordsBody heightBone densityForearmLongitudinal surveyMenopauseWomen
Some height loss is an inevitable effect of aging. Height loss is seen in both men and women, slightly accelerated in women compared to men after the age of 50 [1, 2]. The height loss is caused by decreasing height of the intervertebral discs and joint cartilages, less pronounced foot arches and postural changes . Another important cause of the age-related height loss is the reduction in vertebral height due to osteoporotic vertebral fractures and structural changes, which are more frequent in women . Historical height loss is not only a consequence of vertebral fractures, but also a risk factor for fractures in other parts of the skeleton [5–7].
The distal forearm is a favourable site of bone mineral density (BMD) assessments and enables the examination of the mainly cortical bone of the distal radius and of the ultradistal radius with mostly cancellous bone; however, the difference in metabolic activity between these two structural components is higher in axial bone . Several studies have documented that BMD measurements at the forearm predict fractures comparably to measurements at other skeletal sites . The BMD decline seen in menopausal women seems to be more rapid in the forearm than at other skeletal sites [10, 11], and more pronounced at the distal forearm than at the ultradistal radius .
Height loss may be estimated as the difference between current measured height and the patient’s recalled peak height or recalled height at a specific age. Several studies have shown, however, that people often overestimate their height, and also underestimate their weight [13, 14]. This may lead to over-reporting of height loss through life . A more exact way of determining height loss is the prospective monitoring of height in a clinical setting or as part of a longitudinal study .
The aim of this study was to investigate the degree of height loss in a population-based cohort of Norwegian women aged 45–60 years at baseline followed for more than 15 years, and we questioned whether there is an association with forearm bone mineral density and longitudinal BMD change observed over a period of almost 5 years.
Materials and methods
The Nord-Trøndelag Health Study (HUNT) is a multipurpose health survey aimed at the total population in the county of Nord-Trøndelag, a rather rural region in mid-Norway. The first general data collection, HUNT 1, took place during the years 1984–86 and the second, HUNT 2, in 1995–97. At both occasions, all county citizens >19 years were invited, in total numbers of 85,100 and 92,936, respectively. The overall attendance rate in HUNT 1 and HUNT 2 was 88.1% and 71.2%, respectively, somewhat better for women than for men [17, 18]. At both events the data collection was carried out according to practically identical protocols although the topics for research were broader, thus more questions, at HUNT 2 than at HUNT 1. A comprehensive health questionnaire was sent by mail along with an invitation for a general health examination including height and weight measurements. A second, more detailed questionnaire was handed out after the examination, to be completed at home and returned by mail. Details concerning the questionnaires and examinations have been described elsewhere [17, 19].
Forearm densitometry was part of two substudies in HUNT 2 involving 18,265 men and women. In 2001, after an average period of 4.6 years, a subsample of 6,656 participants at HUNT 2 attended a follow-up study consisting of forearm densitometry and a short questionnaire/interview. Height (to the nearest centimeter) and weight (to the nearest half kilogram) were measured following the same procedures at all three occasions, without shoes and with light clothing.
BMD of the non-dominant forearm was measured with three single-energy X-ray densitometers (DTX 100, Osteometer, MediTech Inc. USA), the same machines were used at both M1 and M2. Sites of measurements were the distal forearm and ultradistal radius. The regions of interest were automatically defined by the densitometer. The distal region was 24 mm proximal to the point at which the radius and ulna are 8 mm apart, and the ultradistal region from the 8 mm separation to the beginning of the radius endplate . The procedures for measurement, calibration and reassessment of the regions of interest has previously been described in detail [19, 20]. At both densitometry occasions, the participants were allocated to one of the three densitometers depending on accessibility. To ensure comparability of the BMD results of two measurements at three machines, calibration equations were established. This was done with the European Forearm Phantom (EFP) (QRM GmbH, Moehrendorf, Germany) consisting of three hydroxyapatite bone imitations with different densities. Repeated measurements of the EFP were performed on each densitometer and the observed BMD results were standardized according to calibration equations developed as simple linear regression models on the observed measurements . In a precision study using identical equipment a coefficient of variation <1% was found at each two sites of measurement .
With the exception of height measured at BL, M1 and M2, all other variables for this study are from M1 and M2. Height loss and BMD decline between the measurements are both given positive values. Height change was analyzed as the crude difference in height (in cm and mm), as annual change (cm/year and mm/year) and as annual percentage change from baseline, categorized in quartiles. The BMD change was calculated as percentage change per year (in g/cm2) and for some analyses categorized in tertiles.
Age at M1 was entered into the analyses either as a continuous variable or grouped in four (54–58, 59–63, 64–68 and 69–73 years). Ovarian hormone treatment (OHT, estradiol with or without progesterone/progestin) was assessed at M1 as current, prior or never. A total of 337 women in the cohort of analysis (23.8%) had not answered the OHT questions. It was assumed that non-response was because the question was irrelevant for these women; thus, for the multivariate analyses, the non-responders were categorized as never users. This did not change the outcome estimates. All women were menopausal at M1 (≥1 year with no menstrual period), 11 women, however, did not report menopausal age. For the analyses, their menopausal age was set to 50 years. The history of smoking was categorized as current, prior and never smokers as reported at M1. Self-rated health at M1, reported as poor, not very good, good and excellent health, was dichotomized as poor/not so good and good/excellent for the multivariate analyses.
Differences between continuous variables were tested with independent and paired sample T-test. Differences between categories were tested with chi-square test. Relative risk for a specific outcome according to exposure or not was calculated with the Mantel-Haenszel method, adjusting for age. The association between height loss (in cm) between BL and M1 (11.3 years) and BMD at M1 was assessed in four age groups in a univariate general linear model with BMD in age-specific tertiles, adjusting for time since menopause and interval between height measurements. The association between bone loss (BMD annual change in tertiles, M1–M2) and height loss between BL and M2 (15.5 years) was analyzed in logistic regression models, adjusting for time since menopause (in four categories, ≤10, 10.1–15, 15.1–20, >20 years), OHT, body weight (four categories, ≤62, 62.1–70, 70.1–80, >80 kg), smoking and self-reported health. The influence of each of the explaining variables on height loss was calculated in bivariate models. Outcome measure was odds ratio (OR) for height loss in the upper quartile (highest degree of height loss). Smoking did not contribute at a statistically significant level and was omitted from the models. Statistical significance was set at the 5% level, and all tests were two-sided. The analyses were carried out with SPSS version 14.0.
The HUNT study was approved by the Regional Committee for Medical Research Ethics and by the Norwegian Data Inspectorate. Each participant signed an informed consent of the study.
Selected characteristics measured at the first and second Nord-Trøndelag Health Study, Norway, HUNTa 1 (1984–86) and HUNT 2 (1995–97) for the cohort of 2,795 women born 1924–41 attending at HUNT 1 (baseline, BLa) and then invited for forearm bone mineral density measurement at HUNT 2 (M1a) and for follow-up densitometry in 2001 (M2a)
Participants at baseline only N = 607
Participants at BL + M1 N = 767
Participants at BL + M1 + M2 N = 1421b
Total height loss (cm)
Height loss/year (cm)
BMDa distal forearm (g/cm2)
BMD ultradistal radius (g/cm2)
Age at menopause
Packyears (among ever smokers)
Height loss >3 cmc(n)
Poor or not so good healthd(n)
Height at baseline and annual height loss by age among the 1,417 women with valid height data meeting at the first and second Nord-Trøndelag Health Study, HUNTa 1 (1984–86, BLa), HUNT 2 (1995–97, M1a) and at the follow-up study in 2001 (M2a)
Height at baseline
Height loss (mm) per year
Height loss >3 cm between BL and M2
BL-M1 11.3 years
BL-M2 15.5 years
Age at M1
Mean forearm bone mineral density (BMDa) by age measured at the second Nord-Trøndelag Health Study, Norway (HUNTa 2, 1995–97, M1a) among the 1,417 women meeting at the three occasions, HUNT 1 (1984–86, BLa), HUNT 2 (1995–97, M1) and the follow-up study in 2001 (M2a)
BMD HUNT 2
BMD HUNT 2
Odds ratio (ORa) estimates of being in the highest quartile of percentage annual height loss during the 15.5 years between the first Nord-Trøndelag Health Study (HUNTa 1 1984–86, BLa) and the follow-up study in 2001 (M2a), according to tertiles of percentage annual distal forearm bone loss between M1a (1995–96) and M2 (4.6 years), time since menopause, history of OHTa, body weight and self-rated health (all measured at M1) among the 1,417 women who met at BL + M1 + M2
Upper quartile annual height loss between BL and M2 (15.5 years)
BMDa loss distal forearm M1-M2
Time since menopause (years)
Body weight (kg)
Poor/not so good
Odds ratio (ORa) estimates of being in the highest quartile of percentage annual height loss between M1a (1995–97) and M2a (2001) (4.6 years) according to tertiles of percentage annual bone mineral density (BMDa) change between M1 (1995–96) and M2 measured at the distal forearm and the ultradistal radius, adjusted for time since menopause at M1
Upper quartile annual height loss between M1 and M2 (4.6 years)
95% confidence interval
95% confidence interval
BMD loss distal forearm M1-M2
BMD loss ultradistal radius
This study confirms that height loss is common in middle-aged and elderly women. Low bone mineral density and bone loss is strongly associated with height loss, also when measured in a non-axial bone such as the forearm. High body weight, current OHT and good self-reported health seems to protect against height loss in this population-based sample followed for more than 15 years.
The strengths of this study are its population-based design, the large number of participants, the high participation rates and the long period of follow-up. Also, the prospective monitoring of height adds to the strengths. The analyses only include women who attended densitometry at M1 and M2. They represent a random sample of the population and do not differ concerning anthropometric measures from other women of same age who participated in HUNT . A study of non-responders in HUNT 2 (M1 in this description) revealed no other selection biases than a slightly higher prevalence of smokers in those who did not attend . This was also found for the women lost to follow-up in this study. A history of smoking was, however, no significant predictor of height loss in the cohort of analysis. Poor self-reported health which was an important predictor of increased height loss was more frequently reported by the women lost to follow-up. Self-rated health did not, however, change the associations between height loss and bone loss in the logistic regression models, and we assume that the sample lost to follow-up does not contribute to serious selection bias. The participants in this study were middle-aged and elderly women, virtually all Caucasians, and the results should neither be generalized to other ethnic groups nor to men.
Height was assessed to the nearest centimeter at all three measurements. Thus, the smallest discernible difference between the examinations was one centimeter. With a mean annual height loss of about 1 mm, as found in this study, it takes more than 10 years to reach a measurable height loss above one centimeter. This may explain that mean height loss was less during the shorter interval M1–M2 than during BL-M1. Also, a certain regression towards the mean should not be ignored. We do, however, consider the height change data from BL until M2 (15.5 years) as rather robust data due to the long period of observation. It should also be noted that the prospective monitoring of height in this study gives a more specific measure of height change than in studies based on the participant’s recalled maximum height or height at a certain age [13, 14]. On the other side, we are not able to detect total height loss throughout life which could be larger than found in this study, and the rates of height loss cannot be directly compared to studies based on historical height loss. Another possible weakness is the lack of a precision study concerning the height measurement procedures.
Most studies concerning the association between height loss, BMD and BMD change have treated height loss as a predictor of osteoporosis and bone loss. Based on the hypothesis that reduced BMD precedes and explains height loss to a certain degree, we chose to use height loss as the outcome variable and forearm BMD as one of several predictors. However, we did perform the analyses with BMD as the dependent variable and found that women with height loss >3 cm had lower BMD than the other groups of height loss, statistically significant in the women aged ≥59 years.
Height loss increased with age, as has been reported from other studies [1, 2, 23]. Some of this height loss may be explained by vertebral fractures [4, 24]. The degree of height loss associated with one prevalent vertebral fracture varies between studies and according to whether height loss is assessed in a longitudinal setting or based on historical height data. For the latter method, one vertebral fracture is associated with height loss in the range of 2.5–6 centimeters [4, 24, 25], whereas in longitudinal studies a height loss of two centimeters or more in elderly women followed for three to seven years could indicate vertebral fracture [16, 26]. In this study a total of 17.4% had a height loss of three centimeters or more over the 15.5 years from BL to M2. The average height loss in this group was 4.1 cm, and mean forearm BMD at M2 was approximately 0.8 SD lower than in women with no height loss. A forearm BMD difference of one SD predicts a 70% increased risk for vertebral fractures, similar to the increase in risk for forearm fractures . We suspect that the prevalence of vertebral deformities due to vertebral fractures is high in the group of women with the highest rate of height loss. The participants in this study were, however, not examined for vertebral deformities, and we are not able to assess the prevalence. Vertebral fractures do, however, not explain the total degree of height loss in this population, and several studies have shown historical height loss of about two centimeters in middle-aged and elderly women with no prevalent vertebral deformities [4, 24, 25], or less in a prospective study .
There was a significant relationship between increased rate of height loss and forearm bone loss. This was observed both for height loss since baseline (BL-M2) and for height loss occurring parallel in time to bone loss (M1–M2). We present bone loss as annual percentage BMD change between M1 and M2. Women with low BMD at M1, but with the same absolute BMD change as women with higher BMD will thus have increased relative change. To control for this, we also performed the analysis in a model with total bone loss, adjusting for the interval between the measurements. The same pattern was found, increased forearm bone loss was related to increased height loss. The results indicate that height loss is related to a general state of increased bone loss, not only reduced BMD, as in women with low peak bone mass.
A higher rate of bone loss was found at the distal forearm compared to the ultradistal radius in these women with an average time of 15 years since menopause. This is in accordance with other studies showing that during perimenopause, bone loss is increased at the ultradistal radius compared to the distal forearm [12, 27, 28]. After this period the distal forearm bone loss is more pronounced [12, 28]. The association between BL-M2 height loss and BMD decline was somewhat stronger for BMD at the ultradistal radius than at the distal forearm. Although the difference in association was not statistically significant, a possible explanation is that the bone loss measured at the ultradistal radius with its high proportion of cancellous bone is more closely related to the mainly cancellous bone loss in the vertebrae, which again is closely related to height loss. It should also be noted that the follow-up period for bone loss between M1 and M2 is rather short. For height loss measured concurrently to bone loss, no statistically significant association was found at the ultradistal site. This could be explained by the difference in bone loss velocity found between distal forearm and ultradistal radius. The decelerated bone loss found in the age group 64–68 is difficult to explain. One theory is the physiological deceleration of bone loss occurring after the accelerated menopausal bone loss . A statistical artifact cannot, however, be excluded.
This study also showed that factors known to protect against bone loss, such as high body weight, OHT, and good self-rated health status were protective against height loss, also when controlling for bone loss. The protective effect of OHT found here is in contrast to a clinical trial showing no difference in height loss between the treatment and placebo groups during the rather short observational period of about four years in the HERS study . In the multivariate analyses, these factors did, however, not explain the association between forearm bone loss and height loss, thus low BMD and increased bone loss were independent factors associated to height loss in postmenopausal women, also when measured in a non-axial part of the skeleton such as the forearm.
In summary, height loss, frequent in this cohort of middle-aged and elderly women, was strongly related to reduced forearm BMD and to bone loss, indicating a generalized state of increased bone loss. Current OHT, high body weight and good self-rated health were protective factors against height loss, probably by slowing down bone loss.
The Nord-Trøndelag Health Study is a collaboration among the HUNT Research Centre, the Norwegian University of Science and Technology, the Norwegian Institute of Public Health and the Nord-Trøndelag County Council.
The study was supported by grants from the Norwegian Research Council, the Norwegian Women’s Public Health Association, the Norwegian Osteoporosis Foundation, the Association of Health and Rehabilitation, SINTEF Health, the Gythfeldt Legacy and the Eckbo Legacy.