Skip to main content

Table 3 Subgroup and sensitivity analyses

From: Association between coffee consumption and risk of bladder cancer in a meta-analysis of 16 prospective studies

  Highest vs. lowest consumption Each 1 cup/d increment
N RR (95% CI) I2 (%) P-diff N RR (95% CI) I2 (%) P-diff
Subgroup analyses
Geographic region
  US 7 1.10 (0.92–1.32) 50.4 Ref. 7 1.01 (0.97–1.04) 70.0 Ref.
  Europe 6 1.13 (0.98–1.31) 0 0.60 5 1.02 (0.99–1.04) 0 0.55
  Japan 3 0.81 (0.57–1.16) 23.1 0.23 3 0.92 (0.75–1.14) 65.3 0.30
Duration of follow-up
   ≥ 10 yr 12 1.08 (0.95–1.23) 35.9   11 1.01 (0.98–1.04) 62.6  
   < 10 yr 4 1.04 (0.81–1.35) 31.4 0.86 4 1.00 (0.95–1.04) 42.3 0.67
No. of participants
   ≥ 50,000 8 1.02 (0.91–1.15) 28.9   8 1.00 (0.98–1.03) 61.1  
   < 50,000 8 1.21 (0.96–1.51) 28.9 0.26 7 1.04 (0.97–1.13) 52.0 0.67
Sex
  Men 10 1.16 (1.00–1.35) 21.5   9 1.01 (0.98–1.05) 42.0  
  Women 7 0.90 (0.59–1.37) 63.6 0.37 6 0.98 (0.89–1.08) 56.0 0.76
No. of cases
   ≥ 200 10 1.02 (0.92–1.13) 19.0   10 1.00 (0.98–1.02) 57.3  
   < 200 6 1.38 (1.05–1.81) 12.0 0.064 5 1.12 (1.04–1.21) 0 0.037
% Male cases
   ≥ 75% 6 1.10 (0.95–1.26) 4.1   6 1.02 (0.94–1.10) 69.3  
   < 75% 9 1.10 (0.92–1.30) 37.6 0.98 7 1.01 (0.99–1.04) 18.5 0.87
% Current smoker
   ≥ 25% 8 1.02 (0.82–1.25) 26.3   7 1.00 (0.94–1.07) 52.6  
   < 25% 8 1.10 (0.96–1.26) 40.5 0.60 8 1.01 (0.98–1.03) 63.6 0.98
Smoking status
  Never 6 1.15 (0.79–1.67) 56.5 Ref. 6 1.02 (0.95–1.09) 58.0 Ref.
  Former 3 1.18 (0.92–1.52) 55.5 0.80 3 1.01 (0.97–1.05) 69.7 0.99
  Never/former 6 1.20 (0.93–1.56) 61.7 0.71 6 1.03 (0.98–1.08) 76.3 0.80
  Current 3 1.19 (0.97–1.46) 0 0.72 3 1.04 (1.01–1.07) 0 0.50
  Former/current 4 1.12 (0.88–1.43) 33.3 0.81 4 1.01 (0.95–1.07) 38.4 0.73
Statistical adjustment
Smokinga
  Poorer 3 1.48 (1.14–1.93) 0 Ref. 3 1.08 (1.02–1.15) 8.9 Ref.
  Moderate 8 1.00 (0.80–1.26) 31.7 0.058 7 0.98 (0.90–1.07) 60.5 0.17
  Better 5 1.03 (0.93–1.13) 0 0.042 5 1.00 (0.98–1.02) 49.7 0.055
Alcohol drinking
  No 12 1.13 (1.01–1.26) 10.4   11 1.02 (0.99–1.04) 34.5  
  Yes 4 0.91 (0.74–1.13) 34.2 0.067 4 0.98 (0.92–1.04) 68.8 0.075
Education
  No 10 1.18 (1.00–1.40) 19.6   9 1.03 (0.99–1.07) 44.1  
  Yes 6 0.99 (0.87–1.13) 31.5 0.14 6 0.99 (0.96–1.03) 69.5 0.33
 Physical activity
  No 13 1.10 (0.95–1.29) 33.4   12 1.01 (0.97–1.05) 53.6  
  Yes 3 1.02 (0.88–1.17) 24.4 0.48 3 1.00 (0.97–1.04) 75.7 0.90
BMI
  No 12 1.15 (1.02–1.29) 5.1   12 1.02 (0.99–1.05) 33.9  
  Yes 4 0.92 (0.74–1.14) 52.9 0.099 3 0.98 (0.93–1.03) 80.7 0.31
Diabetes
  No 13 1.10 (0.96–1.25) 27.1   12 1.01 (0.98–1.04) 43.2  
  Yes 3 0.97 (0.73–1.29) 60.4 0.57 3 0.99 (0.92–1.05) 78.4 0.78
Family history of cancer
  No 14 1.10 (0.96–1.27) 27.9   13 1.01 (0.98–1.05) 49.9  
  Yes 2 0.99 (0.82–1.21) 57.6 0.41 2 1.00 (0.96–1.04) 87.0 0.74
Energy intake
  No 14 1.08 (0.96–1.23) 35.3   13 1.01 (0.99–1.04) 59.0  
  Yes 2 1.00 (0.74–1.36) 28.9 0.74 2 0.98 (0.91–1.04) 51.3 0.43
Fruit/vegetable consumption
  No 12 1.13 (0.97–1.33) 32.1   11 1.02 (0.98–1.06) 51.5  
  Yes 4 1.00 (0.88–1.14) 16.5 0.29 4 1.00 (0.97–1.03) 72.1 0.54
Tea consumption
  No 11 1.15 (1.02–1.29) 11.4   10 1.02 (0.99–1.05) 40.8  
  Yes 5 0.92 (0.96–1.20) 12.3 0.042 5 0.99 (0.96–1.03) 62.1 0.26
Sensitivity analysis
 Excluding 1 studyb 15 1.10 (0.98–1.23) 22.4   14 1.01 (0.99–1.04) 47.0  
 Excluding 2 studiesc 14 1.10 (0.97–1.25) 27.9   13 1.01 (0.98–1.04) 51.0  
  1. N” indicates the number of studies included in the analyses; “P-diff” indicates P values for differences between subgroup population (derived using meta-regression analyses)
  2. a Poorer adjustment for smoking: no adjustment for smoking or adjustment for smoking status only; moderate adjustment for smoking: adjustment for smoking status in addition smoking frequency (e.g. cigarettes smoked per day, or pack-years of smoking) either continuously or categorically; better adjustment for smoking: adjustment for smoking status, smoking frequency, in addition to smoking duration or lifetime smoking intensity. One study (Gaspstur, 2017) that reported results only for nonsmokers (never and former smokers) and adjusted for smoking history for former smokers was included in the “better” group. Another study (Hashemian, 2019) that included totally current smokers and adjusted for both smoking frequency and duration was included in the “moderate group” (this study further examined coffee-bladder cancer association by smoking frequency and did not find group differences in the association)
  3. b Excluding one study (Gaspstur, 2017) in which the study outcome was bladder cancer mortality, and all analyzed participants were never or former smokers
  4. c In addition to the above-mentioned study (Gaspstur, 2017), further excluding another study (Hashemian, 2019) in which all participants were current smokers