TABLE 3.
Sensitivity analyses for estimating hazard ratios for hemorrhoid by metformin exposure.
| Model/Metformin use | IncidentCase number | CasesFollowed | Person-years | Incidence rate(per 100,000 person-years) | Traditional Cox model | IPTW model | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | p Value | HR | 95% CI | p Value | |||||
| Unmatched cohort | ||||||||||
| I. Excluding two consecutive prescriptions of metformin spanning more than 4 months | ||||||||||
| Never users | 2,025 | 19,523 | 80,153.12 | 2,526.41 | 1.000 | 1.000 | ||||
| Ever users | 2,515 | 52,075 | 220,925.10 | 1,138.39 | 0.437 | (0.410–0.466) | <0.0001 | 0.448 | (0.423–0.476) | <0.0001 |
| II. Excluding patients treated with incretin-based therapies during follow-up | ||||||||||
| Never users | 2,002 | 18,510 | 75,573.20 | 2,649.09 | 1.000 | 1.000 | ||||
| Ever users | 7,588 | 119,286 | 526,011.63 | 1,442.55 | 0.533 | (0.506–0.561) | <0.0001 | 0.540 | (0.514–0.567) | <0.0001 |
| III. Patients enrolled during 1999–2002 | ||||||||||
| Never users | 855 | 8,637 | 34,586.82 | 2,472.04 | 1.000 | 1.000 | ||||
| Ever users | 5,037 | 90,236 | 421,633.37 | 1,194.64 | 0.475 | (0.440–0.513) | <0.0001 | 0.473 | (0.440–0.508) | <0.0001 |
| IV. Patients enrolled during 2003–2005 | ||||||||||
| Never users | 1,170 | 10,886 | 45,566.30 | 2,567.69 | 1.000 | 1.000 | ||||
| Ever users | 3,174 | 62,111 | 270,852.70 | 1,171.85 | 0.440 | (0.409–0.474) | <0.0001 | 0.455 | (0.426–0.487) | <0.0001 |
| V. Defining hemorrhoid as a primary diagnosis at hospitalization | ||||||||||
| Never users | 349 | 21,256 | 90,841.19 | 384.19 | 1.000 | 1.000 | ||||
| Ever users | 1,381 | 169,495 | 783,364.48 | 176.29 | 0.473 | (0.417–0.536) | <0.0001 | 0.455 | (0.405–0.512) | <0.0001 |
| VI. Patients receiving aspirin | ||||||||||
| Never users | 1,211 | 11,985 | 48,552.79 | 2,494.19 | 1.000 | 1.000 | ||||
| Ever users | 5,211 | 96,528 | 442,205.39 | 1,178.41 | 0.457 | (0.428–0.488) | <0.0001 | 0.464 | (0.436–0.494) | <0.0001 |
| VII. Patients not receiving aspirin | ||||||||||
| Never users | 814 | 7,538 | 31,600.33 | 2,575.92 | 1.000 | 1.000 | ||||
| Ever users | 3,000 | 55,819 | 250,280.68 | 1,198.65 | 0.477 | (0.439–0.518) | <0.0001 | 0.462 | (0.428–0.499) | <0.0001 |
| VIII. Patients receiving calcium channel blockers | ||||||||||
| Never users | 1,315 | 13,118 | 53,210.21 | 2,471.33 | 1.000 | 1.000 | ||||
| Ever users | 5,168 | 94,952 | 432,049.30 | 1,196.16 | 0.469 | (0.440–0.501) | <0.0001 | 0.477 | (0.449–0.507) | <0.0001 |
| IX. Patients not receiving calcium channel blockers | ||||||||||
| Never users | 710 | 6,405 | 26,942.91 | 2,635.20 | 1.000 | 1.000 | ||||
| Ever users | 3,043 | 57,395 | 260,436.76 | 1,168.42 | 0.455 | (0.417–0.496) | <0.0001 | 0.439 | (0.405–0.477) | <0.0001 |
| Matched cohort | ||||||||||
| I. Excluding two consecutive prescriptions of metformin spanning more than 4 months | ||||||||||
| Never users | 2,022 | 19,498 | 80,068.10 | 2,525.35 | 1.000 | 1.000 | ||||
| Ever users | 349 | 7,170 | 30,027.67 | 1,162.26 | 0.456 | (0.407–0.511) | <0.0001 | 0.459 | (0.410–0.514) | <0.0001 |
| II. Excluding patients treated with incretin-based therapies during follow-up | ||||||||||
| Never users | 1,999 | 18,486 | 75,489.58 | 2,648.05 | 1.000 | 1.000 | ||||
| Ever users | 1,034 | 15,806 | 68,477.61 | 1,509.98 | 0.567 | (0.525–0.611) | <0.0001 | 0.567 | (0.526–0.612) | <0.0001 |
| III. Patients enrolled during 1999–2002 | ||||||||||
| Never users | 854 | 8,627 | 34,557.82 | 2,471.22 | 1.000 | 1.000 | ||||
| Ever users | 676 | 11,583 | 53,212.25 | 1,270.38 | 0.496 | (0.448–0.550) | <0.0001 | 0.509 | (0.460–0.563) | <0.0001 |
| IV. Patients enrolled during 2003–2005 | ||||||||||
| Never users | 1,168 | 10,871 | 45,510.28 | 2,566.45 | 1.000 | 1.000 | ||||
| Ever users | 413 | 7,915 | 33,984.45 | 1,215.26 | 0.464 | (0.414–0.519) | <0.0001 | 0.471 | (0.421–0.527) | <0.0001 |
| V. Defining hemorrhoid as a primary diagnosis at hospitalization | ||||||||||
| Never users | 347 | 21,199 | 90,667.01 | 382.72 | 1.000 | 1.000 | ||||
| Ever users | 179 | 21,199 | 96,460.75 | 185.57 | 0.481 | (0.401–0.577) | <0.0001 | 0.484 | (0.404–0.579) | <0.0001 |
| VI. Patients receiving aspirin | ||||||||||
| Never users | 1,208 | 11,966 | 48,487.73 | 2,491.35 | 1.000 | 1.000 | ||||
| Ever users | 647 | 12,019 | 53,843.20 | 1,201.64 | 0.473 | (0.429–0.520) | <0.0001 | 0.479 | (0.436–0.527) | <0.0001 |
| VII. Patients not receiving aspirin | ||||||||||
| Never users | 814 | 7,532 | 31,580.38 | 2,577.55 | 1.000 | 1.000 | ||||
| Ever users | 442 | 7,479 | 33,353.50 | 1,325.20 | 0.510 | (0.454–0.573) | <0.0001 | 0.512 | (0.456–0.574) | <0.0001 |
| VIII. Patients receiving calcium channel blockers | ||||||||||
| Never users | 1,312 | 13,099 | 53,141.72 | 2,468.87 | 1.000 | 1.000 | ||||
| Ever users | 721 | 13,036 | 58,168.46 | 1,239.50 | 0.493 | (0.450–0.540) | <0.0001 | 0.500 | (0.456–0.547) | <0.0001 |
| IX. Patients not receiving calcium channel blockers | ||||||||||
| Never users | 710 | 6,399 | 26,926.38 | 2,636.82 | 1.000 | 1.000 | ||||
| Ever users | 368 | 6,462 | 29,028.24 | 1,267.73 | 0.481 | (0.423–0.546) | <0.0001 | 0.478 | (0.422–0.542) | <0.0001 |
Hemorrhoid was based on a diagnosis made at the out-patient clinics or during hospitalization in all models except Model V.
IPTW, inverse probability of treatment weighting; HR, hazard ratio; CI, confidence interval.