Bone Mineral Density and Lean Mass in Long-Term Residents of a Rotating Habitat by Years of Residence and Level Gravity: A Cross-Sectional Study of 264 Freeside Adults, 2045–2046

1. Introduction

Freeside is a spindle-shaped orbital habitat at L-5. It is both a resort and a banking haven, and a permanent community lives and works there alongside its visitors. The habitat rotates, and rotation supplies apparent gravity that increases with distance from the axis. Residents who live on levels near the axis experience less of it than those who live near the rim. The Habitat Authority's engineering survey gives design gravity fractions for each residential level (Freeside Habitat Authority, 2045), and this paper uses those design values, which are not measurements of what any resident experiences.

Clinicians at the Institute have long suspected that residence at reduced gravity costs bone and muscle. A pilot survey of long-stay residents reported lower bone density among the longest-resident group (Lanz, 2041), and a clinic-based series described muscle loss in older residents that exceeded what age alone would predict (Wojcik, 2043). Both studies pooled residents across levels and treated residence as a single exposure. Because levels differ in nominal gravity, the question of whether where a resident lives matters as much as how long has not been answered.

Two features of the resident population complicate any study. First, residents are not assigned to levels. They choose, or are placed by employers, according to cost and access, and the people who choose a level may differ from those who do not. Second, residents leave. If those who lose the most bone or muscle are more likely to leave, the remaining long-term residents will look healthier than the true cost of long residence (Farrelly, 2042).

We examine 264 long-term residents. We ask whether bone mineral density and lean mass are lower among those with more years of residence, whether the association is weaker at levels of higher nominal gravity, and how large the associations are in clinical terms. We write in 2047 and report examinations carried out in 2045 and 2046.

2. Methods

Participants. We invited adult residents of the habitat who had lived there for at least three years, with no absence of more than six consecutive months in the preceding five years, drawn from the Habitat Authority's residence registry, which records arrival dates and level assignments (Freeside Habitat Authority, 2046). The Institute's screening protocol for bone and muscle status was used for all examinations (Roussel-Anker, 2040). Participants were excluded if they had a diagnosis or treatment recorded in the clinic file that affects bone or muscle independently of residence. Of 391 residents invited, 264 (68%) completed examination; of the 127 who did not, 71 declined, 38 were excluded on clinic-file criteria at screening and 18 did not attend. We hold only registry data on those who did not take part and cannot say how they differ. Their ages ranged from 28 to 72 years (mean 47.8, SD 9.4), 133 (50%) were male as recorded, and years of continuous residence ranged from 3 to 15 (median 7, IQR 5–10).

Level and gravity. The habitat has five residential levels. Level 1 is nearest the axis and level 5 nearest the rim. We assigned each level the nominal fraction of reference gravity given in the Habitat Authority's engineering survey, rounded to the nearest 0.05: 0.35, 0.50, 0.65, 0.80 and 0.95 for levels 1 to 5 (Freeside Habitat Authority, 2045; Brathwaite, 2039). These are design values for the level's floor, with a tolerance of about 0.05 that we ignore, and they are not measured exposures. A resident's level was the level recorded in the registry for the dwelling in which they had lived longest. Of the 264 participants, 30, 40, 67, 80 and 47 lived on levels 1 to 5.

Outcomes and covariates. The outcomes were lumbar spine bone mineral density in g/cm², and appendicular lean mass index, the lean mass of the limbs divided by height squared, in kg/m², both from scan-based body composition measurement at the Institute. Covariates were age, sex as recorded in the clinic file, body-mass index and self-reported hours per week of resistance exercise. Supplement use and diet were not recorded in the clinic file in usable form.

Analysis. Both outcomes are continuous and were analysed by ordinary least-squares regression. The predictors were years of residence (centred at 7 years), nominal level gravity (centred at 0.65, in units of 0.1 g), their product, and the covariates (age centred at 47 years and expressed per decade; body-mass index per 3 units; exercise centred at 2.8 hours per week, close to the sample mean). The product term lets the association of residence with the outcome differ by gravity. We examined two outcomes and report several tests without correction for multiplicity, so single p-values near .05 should be read with caution. The coefficient on years is therefore the slope at 0.65 g, and we report the slope at 0.35 and 0.95 g by linear combination of coefficients (Lindqvist-Ruiz, 2042). We tested the product term with an F-test and tested a squared years term to assess linearity. Intervals are 95%, and all tests are two-sided.

3. Results

Participants on the five levels were similar in age, sex and years of residence (Table 1). Mean lumbar bone mineral density was 1.134 g/cm² (SD 0.102) and mean appendicular lean mass index 7.33 kg/m² (SD 0.90). In descriptive terms, mean density was 1.156 g/cm² among the 94 participants with 3 to 5 years of residence, 1.157 among the 91 with 6 to 9 years and 1.081 among the 79 with 10 to 15 years, so the crude means are similar in the first two bands and lower in the third.

For bone density, more years of residence was associated with lower values at the middle gravity: −0.0094 g/cm² per year (95% CI −0.0122 to −0.0066; p < .001), equivalent to 8.3% of mean density per decade. The product term was positive (0.0017 g/cm² per year per 0.1 g; 95% CI 0.0002 to 0.0032; p = .030), so the slope was less steep at higher gravity (a 54% smaller slope at 0.95 g than at 0.65 g). The slope was −0.0145 (95% CI −0.0205 to −0.0085) g/cm² per year at 0.35 g, −0.0094 (95% CI −0.0122 to −0.0066) at 0.65 g and −0.0044 (95% CI −0.0090 to 0.0003) at 0.95 g; the last interval includes zero, and its p-value (p = .064) does not exclude a slope of zero, or a slope of about half the middle-level value. Gravity itself, at 7 years of residence, was not associated with density (−0.0009; 95% CI −0.0065 to 0.0047). The model explained 35% of the variance in density (adjusted R² 0.34). Estimates for age (−0.032 g/cm² per decade) and sex (male 0.073) are in Table 2.

Lean mass index was likewise lower with longer residence: −0.0463 kg/m² per year at 0.65 g (95% CI −0.0652 to −0.0273; p < .001). The product term was positive but imprecise (0.0069; 95% CI −0.0033 to 0.0171; p = .186), so the data do not show that the lean-mass slope differs by gravity. Slopes were −0.0669 (95% CI −0.1072 to −0.0266) at 0.35 g, −0.0463 (95% CI −0.0652 to −0.0273) at 0.65 g and −0.0257 (95% CI −0.0567 to 0.0054) at 0.95 g. Resistance exercise was associated with higher lean mass (0.0873 kg/m² per weekly hour; 95% CI 0.0502 to 0.1243), and the model explained 62% of the variance (adjusted R² 0.61).

A squared term for years of residence was not significant for either outcome (density p = .821; lean mass index p = .334), so we found no statistical evidence against a linear form within the range of the data.

4. Discussion

Both outcomes were lower among residents with more years of residence, after adjustment for age, sex, body-mass index and exercise. At the middle level the difference in bone density was about 8% of mean density per additional decade of residence, and we make no claim about clinical thresholds. The difference in lean mass was also clear. The crude band means place the lower density in the longest-resident group, with the two shorter bands similar, while the squared term detected no curvature. The linear slope is therefore a summary of a cross-sectional association and not a rate of loss within a person.

Gravity changed the bone result and not clearly the muscle result. Residents of higher-gravity levels had a smaller difference in bone density per year of residence, and at the highest level the estimated slope was about half that at the middle level, with an interval that included zero. That pattern is compatible with a protective role for loading, but the data are cross-sectional and the estimate at the highest level is imprecise. Lean mass tells a similar story in direction but without precision: its product term was positive and its interval wide. The absence of a detectable gradient for lean mass is not evidence that none exists.

The cross-sectional pattern could arise in other ways. If residents with the lowest bone or muscle values are more likely to leave the habitat, then the longest-resident groups are those with higher values, and the association of long residence with lower values would be understated. Selection into levels could bias the gravity gradient in either direction, for instance if residents with weaker bones were advised to live on lower levels where falls are less severe. We cannot distinguish these, and the registry does not record moves reliably enough to follow them.

Our results do not tell us what to do about the habitat. They do justify a longitudinal programme that measures density at arrival and at intervals, records every change of level, and follows residents who leave. The positive association of resistance exercise with lean mass in this sample is a reason to test exercise provision on lower levels in a trial. It is not evidence that provision would change outcomes.

5. Limitations

The design is cross-sectional. Each resident was examined once, and years of residence are compared across different people. Any systematic difference between long-stay and short-stay residents, including the reasons they stayed, is a possible confounder, and survival into the long-stay group is a form of selection. If residents with the lowest values left, it would understate the associations we report.

Gravity is a design value assigned by level, and not a measurement. Residents move about the habitat, spend working hours on other levels and in transit, and our assignment ignores this. The exposure error may weaken the gravity gradient. The resident's level is the one in which they lived longest and was not tracked over time. Level assignment is not random, and the characteristics of residents on each level, which we cannot fully observe, may differ.

Sex was taken as recorded in the clinic file and the sample is a mixture of ages, so we did not model sex-specific or age-specific trajectories. Exercise is self-reported. Supplement use and diet, which affect both outcomes, could not be included. The sample of 264 gives estimates with wide intervals for the interaction terms, and lean mass in particular should be re-examined in a larger sample. Because of the limited range of residence of 3 to 15 years, we cannot extrapolate to longer residence. Three participants arrived before the age of 18, and age at arrival was not modelled; age and years of residence are correlated, and Table 2 adjusts for current age only. Dietary supply in orbital communities has been described elsewhere (Chambers, 2044), but we had no individual dietary data.