|
Determinants of HRT discontinuation: results of propensity analysis |
|||||||
| Multivariate Logistic Regression Results* |
|||||||
|
|
|||||||
| Variable |
HRT Continuers (N = 1342) |
HRT Discontinuers (N = 1015) |
Parameter Estimate (β) |
Std. Error of β |
Adjusted Odds Ratio |
95% CI for Odds Ratio |
P-value |
|
|
|||||||
| Age |
|||||||
| Age in years (mean) |
75.5 |
75.5 |
0.0081 |
0.0075 |
1.041 |
0.993–1.023 |
.2847 |
|
|
|||||||
| Race |
|||||||
| White (%) † |
96.7 |
95.6 |
-- |
-- |
1.000 |
-- |
-- |
| Black (%) |
2.7 |
3.5 |
0.3536 |
0.2527 |
1.424 |
0.868–2.337 |
.1616 |
| Other race (%) |
0.6 |
1.0 |
0.2903 |
0.4992 |
1.337 |
0.503–3.556 |
.5609 |
|
|
|||||||
| Income |
|||||||
| Annual income in thousands (mean) |
$13.805 |
$13.999 |
0.0537 |
0.0245 |
1.055 |
1.006–1.117 |
.0283 |
|
|
|||||||
| Marital status |
|||||||
| Widowed, divorced, or never married (%) † |
80.6 |
80.5 |
-- |
-- |
1.000 |
-- |
-- |
| Currently married (%) |
19.4 |
19.5 |
-0.1327 |
0.1384 |
0.876 |
0.668–1.149 |
.3378 |
|
|
|||||||
| Education |
|||||||
| 12 or more years of education (%) |
60.5 |
63.5 |
0.0937 |
0.0899 |
1.098 |
0.921–1.310 |
.2971 |
|
|
|||||||
| Residence type |
|||||||
| Nursing or personal care facility (%) |
1.6 |
1.0 |
-0.4390 |
0.4067 |
0.645 |
0.291–1.431 |
.2804 |
|
|
|||||||
| Urban/rural residence |
|||||||
| Semi-rural or rural residence (%) |
33.6 |
35.5 |
0.1526 |
0.0913 |
1.165 |
0.974–1.393 |
.0948 |
|
|
|||||||
| Alcohol use |
|||||||
| Current alcohol user (%) |
26.5 |
24.4 |
-0.1335 |
0.1008 |
0.875 |
0.995–1.022 |
.1854 |
|
|
|||||||
| Smoking history |
|||||||
| Past or present smoker (%) |
34.7 |
33.0 |
-0.0573 |
0.0924 |
0.944 |
0.788–1.132 |
.5352 |
|
|
|||||||
| Type of HRT used at baseline |
|||||||
| Unopposed estrogen (%) † |
89.5 |
74.2 |
-- |
-- |
1.000 |
-- |
-- |
| Combination estrogen/progestin (%) |
10.5 |
25.8 |
1.1100 |
.1171 |
3.034 |
2.412–3.817 |
.0001 |
|
|
|||||||
| Baseline non-HRT prescription drug use |
|||||||
| Any cardiovascular drug use (%) |
78.8 |
79.9 |
0.2265 |
0.1141 |
1.254 |
1.003–1.568 |
.0471 |
| Any osteoporosis treatment (%) |
12.2 |
15.4 |
0.3100 |
0.1270 |
1.363 |
1.063–1.749 |
.0146 |
| Total number of non-HRT drug classes (mean) |
6.3 |
5.9 |
-0.0507 |
0.0165 |
0.951 |
0.920–0.982 |
.0021 |
|
|
|||||||
| Baseline HRQOL |
|||||||
| Days that physical health was not good (mean) |
6.3 |
6.3 |
0.0086 |
0.0069 |
1.009 |
0.995–1.022 |
.2174 |
| Days that mental health was not good (mean) |
3.3 |
3.0 |
-0.0042 |
0.0067 |
0.996 |
0.983–1.009 |
.5323 |
| Healthy days (mean) ‡ |
21.9 |
22.3 |
-- |
-- |
-- |
-- |
-- |
| Days that health interfered (mean) |
4.2 |
3.8 |
-0.0113 |
0.0077 |
0.989 |
0.974–1.004 |
.1432 |
| Days that pain made it hard (mean) |
6.3 |
6.4 |
0.0060 |
0.0064 |
1.006 |
0.993–1.019 |
.3514 |
|
* All parameter estimates are adjusted for two-way interaction terms involving age, income, marital status, alcohol use, smoking history, prescription drug use, and baseline HRQOL. † Reference group ‡ Because the healthy days variable is a composite of the physical and mental days measures, it was omitted from the logistic regression. | |||||||
Heller et al. BMC Women's Health 2005 5:7 doi:10.1186/1472-6874-5-7 |
|||||||