3.3 Exposure to Angiotensin Receptor Blockers (ARBs). 3.3.1 Evaluation of the Impact of the Treatment on Cancer Risk in Angiotensin Receptor. Blockers ...
79 3.3 Exposure to Angiotensin Receptor Blockers (ARBs). 3.3.1
Evaluation of the Impact of the Treatment on Cancer Risk in Angiotensin Receptor Blockers (ARBs) Users Versus Any Reference Category.
In Table 3.3.1 we describe the main characteristics of the observational studies included in our quantitative synthesis to analyze the impact of the treatment on cancer risk in ARBs users compared to any reference category (non-users or other antihypertensive users). We considered a total of 9 articles and 11 estimates, subdivided into 5 articles and 6 estimates for the cohort studies, 4 articles and 5 estimates for the casecontrol studies. We indicated the different stratifications in the same article in parentheses next to the first author’s name. In Figure 3.3.1 we report the meta-analysis on the studies previously described in Table 3.3.1. We divided the cohort studies and the case-control studies into two distinct groups with their fixedeffect and random-effect pooled estimates (where necessary) and the homogeneity test of the estimates, plus the I2 index. When we also reported the random-effect pooled estimate, the weight of the studies was calculated with random-effect model; otherwise it was calculated with the fixed-effect model. For the cohort studies, the fixed-effect pooled OR was 0.82 (95% CI, 0.80-0.85) and the random-effect pooled OR was 0.90 (95% CI, 0.71-1.14). We found a statistically significant decreased risk of cancer of 18% in ARBs users compared to any reference category with the fixed-effect model. Even with the randomeffect model, we pointed out a decreased risk of cancer of 10% in ARBs users, even though not significant. The homogeneity test of the estimates gave the following results: Q=233.67, p-value |z|
= = = = = = =
-155 172.59 (corrected for ties) 64 -0.90 0.369 0.89 (continuity corrected) 0.372 (continuity corrected)
Egger's test Std_Eff
Coef.
slope bias
-.0037322 -.2202955
Std. Err. .008198 .4201195
t -0.46 -0.52
P>|t| 0.651 0.602
[95% Conf. Interval] -.0201197 -1.060102
.0126553 .6195115
.
Even Begg’s Funnel Plot showed an absence of publication bias in concordance with Egger’s test, since most of the studies considered in our quantitative synthesis fall inside the funnel on the plane and they have a very homogeneous distribution (Figure 3.4.2).
100
Begg's funnel plot with pseudo 95% confidence limits 4
logOR
2
0
-2
-4 0
.5
s.e. of: logOR
1
1.5
Figure 3.4.2 – Begg’s Funnel Plot with pseudo 95% confidence limits. We placed the logarithm of the original estimates on the y-axis, and the standard error of the logarithm of these estimates on the x-axis. Most of the sixty-four studies considered in our quantitative synthesis of the impact of the treatment on cancer risk in RAS-inhibitors users versus any reference category fall inside the funnel on the plane, showing an absence of publication bias. These estimates are distributed in a very homogeneous way inside the funnel.
101 3.5
Exposure to Alpha-1 Adrenergic Receptor Antagonists or Alpha-Blockers (ABs).
3.5.1
Evaluation of the Impact of the Treatment on Cancer Risk in Alpha-Blockers Users Versus Non-Users.
In Table 3.5.1 we describe the main characteristics of the observational studies included in our quantitative synthesis to assess the impact of the treatment on cancer risk in alpha-1 adrenergic receptor antagonists users compared to non-users (only available reference). We analysed a total of 10 articles and 20 estimates, subdivided into 3 articles and 3 estimates for the cohort studies, 7 articles and 17 estimates for the casecontrol studies. The different stratifications in the same article were written in parentheses next to the first author’s name. Figure 3.5.1 shows the meta-analysis corresponding to the studies previously described in Table 3.5.1. We divided the cohort studies and the case-control studies into two distinct groups with the fixed-effect and random-effect pooled estimates and the homogeneity test of the estimates, plus the I2 index. The weight of the studies was calculated by using the random-effect model. For the cohort studies the fixed-effect pooled OR was 3.44 (95% CI, 3.34-3.55). With the fixed-effect model, we found an extremely significant increased risk of cancer of over three-fold in patients receiving alpha-blockers compared to non-users. With the random-effect model, however, the pooled OR was 1.38 (95% CI, 0.44-4.34). Hence, the increased risk of cancer became 38% and no longer significant in alpha-blockers users. Anyway, this estimate was not very precise. The homogeneity test of the estimates gave the following results: Q=286.76, p