Introduction
Methods
Results
Assessment of Studies Using RR as the Relative Association Measure
Author, year, study | Type of study | No. of patients | No. of cases of bladder cancer | Risk ratio RR assessment (95% CI) | Quality assessment | Authors’ conclusion | ||
---|---|---|---|---|---|---|---|---|
Exposed group | Control group | Exposed group | Control group | |||||
Dormandy, 2005, PROactive [16] | Prospective, randomized controlled trial | 2605 | 2633 | 14 | 6 | 2.36 (0.91–6.13) | H | Safety profiles of both arms are similar. When bladder cancer cases with different etiology are excluded, there is no connection between pioglitazone and bladder cancer |
Oliveria, 2008 [21] | Retrospective cohort study using a large US population-based database | 191,223 | 178 | 1.05 (0.71–1.54) | H | Use of thiazolidinediones (pioglitazone included) does not increase the risk of bladder cancer. The adjusted RR is assessed after exclusion of patients on insulin monotherapy | ||
Piccinni, 2011 [22] | Observational using reports recorded in FDA AERS | 37,841 | 561,244 | 31 | 93 | 4.30 (2.82–6.52) | L | There is significantly increased risk of bladder cancer. The significant problems with the reliability and control of information and possible major shifts in the inferences that are logically implausible should be taken into account |
Lewis, 2015, KPNC [23] | Cohort and nested case–control analyses; cohorts from Kaiser Permanente Northern California | 464 | 464 | 91 | 81 | OR 1.18 (0.78–1.80) after adjustment RR 1.12 (0.86–1.47) | H | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Tseng, 2012 [24] | Retrospective cohort study of data from the National Health Insurance database in Taiwan | 2564 | 51,667 | 10 | 155 | 1.30 (0.69–2.46) | M | RR is additionally calculated; the original article employs HR. Use of pioglitazone shows insignificant risk of bladder cancer in comparison with nonuse |
Azoulay, 2012 [25] | Retrospective cohort study using a nested case–control analysis | 191 | 6508 | 19 | 357 | 1.83 (1.10–3.05) | M | Use of pioglitazone is associated with an increased risk of bladder cancer |
Balaji, 2014 [26] | Retrospective cohort study | 31 | 1077 | 1 | 20 | 1.74 (0.24–12.54) | L | Pioglitazone is highly unlikely to cause bladder cancer |
Bazelier, 2013 [27] | Population-based cohort study using the Danish National Health Registers | NA | NA | NA | NA | 1.3 (0.6–2.7) | M | There is no evidence that use of pioglitazone is associated with an increased risk of bladder cancer |
Erdmann, 2014, PROactive 6-year-update [19] | Observational multinational multicenter follow-up study of PROactive | 1820 | 1779 | 10 | 17 | 0.57 (0.26–1.25) | H | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Hsiao, 2013 [28] | Population-based nested case–control using Taiwan’s National Health Insurance Research Database | 3259 | 16,537 | 153 | 523 | 1.48 (1.24–1.77) | L | Use of pioglitazone, especially long-term, is associated with increased risk of bladder cancer. Using the same data Tseng [24] reach exactly the opposite conclusion |
Jin, 2014 [29] | Retrospective cohort study | 11,240 | 101,953 | 30 | 237 | 1.35 (0.769–1.677) | M | Even though there is no evidence for the following, the authors conclude that risk of bladder cancer can be slightly increased in case of use for more than 6 months |
Kuo, 2014 [30] | Nested case–control study | 52 | 985 | 15 | 244 | 1.20 (0.58–2.49) OR | H | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Song, 2012 [31] | Retrospective, matched case–control study | 329 | 658 | 21 | 99 | 2.09 (0.26–16.81) OR | M | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Sensitivity Analysis
Excluded study | For all studies | Low-quality studies excluded | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Pooled RR | LCI 95% | HCI 95% | Cochran Q
|
χ
2
|
I
2
| Pooled RR | LCI 95% | HCI 95% | Cochran Q
|
χ
2
|
I
2
| |
Dormandy et al. [16] | 1.306 | 0.997 | 1.709 | 58.980 | 0.000 | 71.177 | 1.117 | 0.955 | 1.307 | 19.430 | 0.149 | 27.945 |
Oliveria et al. [21] | 1.333 | 0.999 | 1.778 | 59.083 | 0.000 | 71.227 | 1.139 | 0.950 | 1.365 | 21.614 | 0.087 | 35.226 |
Piccinni et al. [22] | 1.230 | 1.039 | 1.455 | 27.977 | 0.045 | 39.236 | Excluded | |||||
Lewis et al. [23] | 1.345 | 0.990 | 1.826 | 58.894 | 0.000 | 71.134 | 1.133 | 0.939 | 1.366 | 21.762 | 0.084 | 35.666 |
Tseng [24] | 1.314 | 0.995 | 1.735 | 60.451 | 0.000 | 71.878 | 1.124 | 0.947 | 1.336 | 21.574 | 0.088 | 35.107 |
Azoulay et al. [25] | 1.295 | 0.976 | 1.718 | 58.804 | 0.000 | 71.091 | 1.092 | 0.937 | 1.273 | 17.781 | 0.217 | 21.262 |
Balaji et al. [26] | 1.313 | 1.002 | 1.719 | 60.374 | 0.000 | 71.842 | Excluded | |||||
Bazelier et al. [27] | 1.314 | 0.997 | 1.731 | 60.452 | 0.000 | 71.878 | 1.126 | 0.949 | 1.336 | 21.631 | 0.087 | 35.277 |
Erdmann et al. [19] | 1.330 | 1.023 | 1.730 | 56.043 | 0.000 | 69.666 | 1.149 | 0.985 | 1.340 | 18.778 | 0.174 | 25.444 |
Hsiao et al. [28] | 1.250 | 0.954 | 1.638 | 58.010 | 0.000 | 70.695 | Excluded | |||||
Jin et al. [29] | 1.311 | 0.979 | 1.756 | 60.432 | 0.000 | 71.869 | 1.110 | 0.930 | 1.324 | 20.884 | 0.105 | 32.962 |
Kuo et al. [30] | 1.316 | 0.998 | 1.734 | 60.392 | 0.000 | 71.851 | 1.128 | 0.950 | 1.340 | 21.740 | 0.084 | 35.604 |
Song et al. [31] | 1.312 | 1.002 | 1.718 | 60.261 | 0.000 | 71.789 | 1.128 | 0.955 | 1.332 | 21.432 | 0.091 | 34.677 |
Levin et al. Scotland [32] | 1.357 | 1.019 | 1.807 | 55.960 | 0.000 | 69.621 | 1.163 | 0.975 | 1.388 | 20.330 | 0.120 | 31.137 |
Levin et al. [32] CPRD UK | 1.349 | 1.008 | 1.806 | 57.420 | 0.000 | 70.394 | 1.152 | 0.961 | 1.382 | 21.122 | 0.099 | 33.719 |
Levin et al. [32] Finland | 1.356 | 1.043 | 1.765 | 52.230 | 0.000 | 67.452 | 1.174 | 1.013 | 1.361 | 16.936 | 0.260 | 17.338 |
Levin et al. [32] Br Columbia | 1.294 | 0.981 | 1.707 | 58.089 | 0.000 | 70.734 | 1.094 | 0.945 | 1.267 | 17.102 | 0.251 | 18.136 |
Levin et al. [32] Rotterdam | 1.313 | 1.002 | 1.719 | 60.359 | 0.000 | 71.835 | 1.128 | 0.955 | 1.333 | 21.563 | 0.088 | 35.073 |
Levin et al. [32] Manchester | 1.316 | 1.006 | 1.721 | 59.891 | 0.000 | 71.615 | 1.133 | 0.959 | 1.337 | 21.404 | 0.092 | 34.591 |
Assessment of Studies Using HR as the Relative Association Measure
Author, year, study | Type of study | No. of patients | No. of cases of bladder cancer | Follow-up, years | Hazard ratio assessment (95% CI) | Quality assessment | Authors’ conclusion | ||
---|---|---|---|---|---|---|---|---|---|
Exposed group | Control group | Exposed group | Control group | ||||||
Lewis, 2011, KPNC [33] | Interim analysis on a cohort and nested case–control study | 30,173 | 162,926 | 90 | 791 | 2 | 1.2 (0.9–1.5) | H | Pioglitazone use is not associated with increased risk of bladder cancer. Authors claim that risk is increased by 40% with long-term use (more than 2 years) |
Lewis, 2015, KPNC [23] | Cohort and nested case–control study; cohorts from Kaiser Permanente Northern California | 34,181 | 158,918 | 186 | 1075 | 2.8 | 1.06 (0.89–1.26) | H | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Tseng, 2012 [24] | Retrospective cohort study of data from the National Health Insurance database in Taiwan | 2545 | 52,383 | 10 | 155 | 2 | 1.35 (0.66–2.58) | M | Use of pioglitazone shows insignificant risk of bladder cancer in comparison with nonuse |
Erdmann, 2014 [19], PROactive 6-year-update | Observational multinational multicenter follow-up study of PROactive | 2605 | 2533 | 23 | 22 | 5.8 | 1.00 (0.56–1.80) | H | There is no increased risk with long-term use of pioglitazone |
Fujimoto, 2013 [34] | Retrospective cohort study | 663 | 20,672 | 9 | 673 | NA | 1.75 (0.89–3.45) | M | Use of pioglitazone shows insignificant risk of bladder cancer in comparison with nonuse |
Gupta, 2015 [35] | Retrospective cohort study | 1111 | 1111 | 0 | 0 | NA | NA | L | Results are odd. There is not a single bladder cancer case registered |
Chang, 2012 [36] | Retrospective case–control study | 401 | 7490 | 84 | 1499 | 1 (375 days) | 1.06 (0.82–1.37) | H | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Lee, 2014 [37] | Retrospective study | 3497 | 31,473 | 12 | 72 | 1 | 1.03 (0.45–2.35) | H | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Mackenzie, 2016 [38] | Retrospective cohort study using Medicare data | 38,091 | 281,999 | 1159 | 1.4 | 1.02 (0.81–1.28) | L | There is no evidence of connection between bladder cancer and pioglitazone | |
Neumann, 2012 [39] | Retrospective cohort study using data from the French national health insurance information system | 155,535 | 1,335,525 | 175 | 1841 | 6.5 | 1.22 (1.05–1.43) or 1.15 (0.99–1.33) when all patients over 40 are included | L | There is a link between dose and response. The study is badly controlled and does not take into account any confounding covariates |
Vallarino, 2013 [40] | Retrospective cohort study | 38,588 | 17,948 | 84 | 44 | 2.2 | 0.92 (0.63–1.33) | M | Use of pioglitazone is not associated with an increased risk of bladder cancer. Insulin-treated patients are used as controls |
Wei, 2012 [41] | Retrospective cohort study | 23,548 | 184,166 | 66 | 803 | 3.2 | 1.16 (0.83–1.62) | H | Use of pioglitazone is not associated with an increased risk of bladder cancer |
Sensitivity Analysis
Excluded study | For all studies | Low-quality studies excluded | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Pooled HR | LCI 95% | HCI 95% | Cochran Q
|
χ
2
|
I
2
| Pooled HR | LCI 95% | HCI 95% | Cochran Q
|
χ
2
|
I
2
| |
Lewis et al. [23] | 1.103 | 1.001 | 1.216 | 4.097 | 0.848 | 0.000 | 1.068 | 0.938 | 1.217 | 3.557 | 0.829 | 0.000 |
Tseng [24] | 1.089 | 1.000 | 1.187 | 3.876 | 0.868 | 0.000 | 1.059 | 0.953 | 1.177 | 3.087 | 0.877 | 0.000 |
Erdmann et al. [19] | 1.095 | 1.005 | 1.193 | 4.161 | 0.842 | 0.000 | 1.067 | 0.960 | 1.187 | 3.515 | 0.834 | 0.000 |
Fujimoto et al. [34] | 1.084 | 0.995 | 1.182 | 2.365 | 0.968 | 0.000 | 1.053 | 0.947 | 1.170 | 1.449 | 0.984 | 0.000 |
Chang et al. [36] | 1.097 | 1.002 | 1.200 | 4.191 | 0.840 | 0.000 | 1.066 | 0.951 | 1.195 | 3.560 | 0.829 | 0.000 |
Lee et al. [37] | 1.093 | 1.004 | 1.191 | 4.231 | 0.836 | 0.000 | 1.066 | 0.960 | 1.184 | 3.555 | 0.829 | 0.000 |
Mackenzie et al. [38] | 1.105 | 1.008 | 1.211 | 3.848 | 0.871 | 0.000 | 1.077 | 0.959 | 1.211 | 3.387 | 0.847 | 0.000 |
Neumann et al. [39] | 1.065 | 0.960 | 1.182 | 3.562 | 0.894 | 0.000 | Excluded | |||||
Vallarino et al. [40] | 1.103 | 1.011 | 1.204 | 3.392 | 0.907 | 0.000 | 1.078 | 0.968 | 1.202 | 2.920 | 0.892 | 0.000 |
Wei et al. [41] | 1.088 | 0.997 | 1.188 | 4.120 | 0.846 | 0.000 | 1.056 | 0.946 | 1.178 | 3.285 | 0.857 | 0.000 |