Table 2. Multivariable linear regression analysis of factors associated with annual change in BMDa at three sites among CHAMPS study participants.
| Variable | Femoral neck | Total hip | Lumbar Spine | |||
|---|---|---|---|---|---|---|
| Parameter estimatec | 95% CId | Parameter estimate | 95% CI | Parameter estimate | 95% CI | |
| Heroin use within 5 yrs | .0031 | -.0039, .0101 | .0017 | -.0049, .0084 | .0015 | -.0060, .0089 |
| History of AIDS diagnosis | -.0014 | -.0077, .0049 | -.0020 | -.0080, .0040 | .0038 | -.0029, .0106 |
| HIV+ | .0036 | -.0021, .0094 | .0005 | -.0050, .0060 | .0002 | -.0060, .0064 |
| AIDS and heroin useb | -.0150 | -.0267, -.0033 | -.0109 | -.0218, -.0000 | -.0092 | -.0214, .0030 |
| HIV+ and heroin useb | -.0016 | -.0120, .0089 | -.0072 | -.0174, .0030 | -.0059 | -.0171, .0053 |
| Baseline BMD at site | -.0187 | -.0326, -.0048 | -.0179 | -.0306, -.0051 | .0037 | -.0081, .0161 |
| Age at 1st exam | -.0001 | -.0006, .0003 | -.0001 | -.0006, .0003 | .0002 | -.0003, .0006 |
| Black | -.0039 | -.0103, .0026 | .0028 | -.0032, .0089 | .0011 | -.0057, .0078 |
| Hispanic | .0001 | -.0068, .0069 | .0051 | -.0015, .0117 | .0023 | -.0051, .0097 |
| Other | -.0101 | -.0204, .0023 | -.0028 | -.0127, .0070 | .0040 | -.0071, .0150 |
| White | Reference | Reference | Reference | |||
| Body Mass Index (kg/m2) | .0001 | -.0004, .0005 | .0003 | -.0001, .0008 | .0003 | -.0001, .0008 |
| Ever use of corticosteroids | .0090 | -.0020, .0199 | ||||
| Family history of spine fracture | .0048 | -.0128, .0223 | .0026 | -.0142, .0194 | .0026 | -.0142, .0194 |
| Current methadone use | -.0036 | -.0091, .0020 | -.0085 | -.0138, -.0032 | -.0085 | -.0138, -.0032 |
| Hepatitis C virus seropositivity | -.0050 | -.0002, -.0097 | -.0034 | -.0079, .0011 | -.0034 | -.0079, .0011 |
a
BMD: Bone mineral density
b
For these interactions, the reference groups are the categories in which heroin was not used (no heroin and any HIV status) plus that of heroin use and HIV-seronegative.
c
Parameter estimates reflect the difference in annual rate of change in BMD, such that estimates less than zero indicate loss of BMD. Units for parameter estimates are gm/cm2 per year
d
95% Confidence Interval for parameter estimate
Parameter estimates with blank cells indicate variable did not enter that model.