Table 3. Multivariable multilevel logistic regression of factors associated with failure to undergo documented early colonoscopy (i.e., delayed colonoscopy, no colonoscopy, or no information about colonoscopy).
| |
|
Model 1 |
Model 2 |
Model 3 (final) |
|||
|---|---|---|---|---|---|---|---|
| Empty model | OR (95% CI) | P-valuea | OR (95% CI) | P-valuea | OR (95% CI) | P-valuea | |
|
FIXED EFFECTS | |||||||
|
Level 1: individual | |||||||
|
FOBT number | |||||||
| First screening FOBT | 1.66 (1.21–2.27) | <0.01 | 1.64 (1.20–2.25) | <0.01 | 1.64 (1.20–2.25) | <0.01 | |
| Second screening FOBT |
|
1.00 |
|
1.00 |
|
1.00 |
|
|
FOBT delivery mode | |||||||
| Delivered during a GP visit | 1.00 | 0.06 | 1.00 | 0.06 | 1.00 | 0.06 | |
| FOBT mailed to the individual's home |
|
1.21 (0.99–1.49) |
|
1.21 (0.99–1.49) |
|
1.21 (0.99–1.49) |
|
|
Level 2: area | |||||||
|
Townsend index | |||||||
| Quintile 1 (least deprived) | 1.00 | 0.09 | 1.00 | 0.09 | |||
| Quintile 2 | 1.13 (0.86–1.49) | 1.09 (0.83–1.44) | |||||
| Quintile 3 | 1.16 (0.87–1.54) | 1.13 (0.85–1.51) | |||||
| Quintile 4 | 1.00 (0.75–1.33) | 0.98 (0.73–1.30) | |||||
| Quintile 5 (most deprived) |
|
|
|
1.42 (1.06–1.90) |
|
1.41 (1.06–1.89) |
|
|
GP density per 100 000 population | |||||||
| ⩽52.7 | 1.00 | 0.09 | |||||
| >52.7 |
|
|
|
|
|
1.17 (0.98–1.42) |
|
|
RANDOM EFFECTS | |||||||
| Level 2 variance (SE) | 0.148 (0.08) | 0.15 (0.08) | 0.130 (0.07) | 0.121 (0.07) | |||
| Deviance 1 (deviance 0) | 3035.16 (3040.54) | 3019.15 (3024.51) | 3012.33 (3016.38) | 3009.42 (3012.95) | |||
| LRT χ2 (p) | χ2=5.38 (P=0.01) | χ2=5.36 (P<0.05) | χ2=4.05 (P<0.05) | χ2=3.53 (P<0.05) | |||
| PCVb | ref | −1.3% | 12.2% | 18.2% | |||
Abbreviation: LRT=likelihood ratio test.
a
P: P-trend for ordinal variables and p heterogeneity for non-ordinal variables.
b
PCV, proportional change in variance at level 2= (VA–VB/VA) × 100, where VA=level-2 variance of the empty model and VB=level-2 variance of the model containing additional terms.