Skip to main content
. 2019 Sep 5;29(9):325–333. doi: 10.2188/jea.JE20170298

Table 3. Sub-distribution hazard ratios and 95% confidence intervals for incident ischemic stroke subtypes according to categories of cumulative average body mass index in participants with or without updated history of hypertension.

  Body mass index, kg/m2 Trend Pc Interaction Pd

<19 19–<21 21–<23 23–<25 25–<27 27–<30 ≥30
Men without hypertension                  
  Person-years 21,192 74,410 139,031 141,985 80,143 36,261 6,900    
 LS Number of incidents/incidence ratea 18/0.85 78/1.05 107/0.77 113/0.80 62/0.77 36/0.99 9/1.30    
  Multivariable SHR (95% CI)b 0.73 (0.44–1.21) 1.11 (0.83–1.49) 0.87 (0.67–1.14) 1 1.04 (0.76–1.42) 1.34 (0.91–1.96) 1.63 (0.83–3.23) 0.07 0.90
 LAOS Number of incidents/incidence ratea 2/0.09 23/0.31 45/0.32 74/0.52 25/0.31 18/0.50 4/0.58    
  Multivariable SHR (95% CI)b 0.12 (0.03–0.47) 0.48 (0.30–0.77) 0.54 (0.37–0.79) 1 0.65 (0.42–1.02) 1.04 (0.63–1.73) 1.18 (0.43–3.23) <0.001 0.01
 CES Number of incidents/incidence ratea 13/0.61 32/0.43 80/0.58 74/0.52 44/0.55 18/0.50 3/0.43    
  Multivariable SHR (95% CI)b 0.75 (0.41–1.38) 0.68 (0.44–1.03) 0.98 (0.71–1.35) 1 1.16 (0.80–1.68) 1.10 (0.66–1.84) 0.96 (0.30–3.07) 0.03 0.11
Men with hypertension                  
  Person-years 5,636 23,932 55,862 74,705 52,735 30,091 6,898    
 LS Number of incidents/incidence ratea 5/0.89 48/2.01 89/1.59 106/1.42 83/1.57 42/1.40 13/1.88    
  Multivariable SHR (95% CI)b 0.39 (0.16–0.96) 1.08 (0.77–1.53) 1.00 (0.76–1.33) 1 1.21 (0.91–1.62) 1.07 (0.75–1.53) 1.42 (0.79–2.58) 0.06  
 LAOS Number of incidents/incidence ratea 5/0.89 27/1.13 53/0.95 53/0.71 34/0.64 26/0.86 6/0.87    
  Multivariable SHR (95% CI)b 0.80 (0.32–2.03) 1.22 (0.76–1.97) 1.20 (0.82–1.76) 1 0.99 (0.64–1.53) 1.32 (0.82–2.12) 1.28 (0.54–3.06) 0.81  
 CES Number of incidents/incidence ratea 5/0.89 31/1.30 65/1.16 81/1.08 69/1.31 36/1.20 17/2.46    
  Multivariable SHR (95% CI)b 0.47 (0.19–1.16) 0.85 (0.56–1.30) 0.93 (0.67–1.29) 1 1.37 (0.99–1.89) 1.29 (0.87–1.91) 2.69 (1.58–4.59) <0.001  
Women without hypertension                  
  Person-years 32,563 98,548 164,684 146,159 81,717 42,360 11,302    
 LS Number of incidents/incidence ratea 10/0.31 31/0.31 54/0.33 48/0.33 29/0.35 22/0.52 13/1.15    
  Multivariable SHR (95% CI)b 0.65 (0.33–1.31) 0.88 (0.56–1.38) 0.96 (0.65–1.42) 1 1.02 (0.64–1.62) 1.48 (0.89–2.46) 3.35 (1.76–6.36) <0.001 0.98
 LAOS Number of incidents/incidence ratea 7/0.21 13/0.13 30/0.18 16/0.11 16/0.20 6/0.14 4/0.35    
  Multivariable SHR (95% CI)b 1.25 (0.49–3.16) 1.03 (0.49–2.15) 1.58 (0.86–2.89) 1 1.65 (0.82–3.31) 1.11 (0.43–2.89) 2.58 (0.87–7.67) 0.43 0.08
 CES Number of incidents/incidence ratea 9/0.28 18/0.18 26/0.16 31/0.21 9/0.11 11/0.26 4/0.35    
  Multivariable SHR (95% CI)b 0.95 (0.45–1.99) 0.79 (0.44–1.42) 0.73 (0.43–1.22) 1 0.49 (0.23–1.03) 1.12 (0.56–2.24) 1.56 (0.54–4.52) 0.57 0.53
Women with hypertension                  
  Person-years 8,194 29,247 64,299 80,351 63,954 44,307 15,961    
 LS Number of incidents/incidence ratea 6/0.73 21/0.72 38/0.59 79/0.98 54/0.84 56/1.26 20/1.25    
  Multivariable SHR (95% CI)b 0.54 (0.24–1.25) 0.67 (0.41–1.08) 0.58 (0.39–0.85) 1 0.87 (0.62–1.24) 1.30 (0.92–1.84) 1.23 (0.75–2.02) <0.001  
 LAOS Number of incidents/incidence ratea 2/0.24 8/0.27 16/0.25 38/0.47 24/0.38 26/0.59 12/0.75    
  Multivariable SHR (95% CI)b 0.42 (0.10–1.73) 0.58 (0.27–1.24) 0.51 (0.29–0.92) 1 0.76 (0.46–1.28) 1.21 (0.73–2.00) 1.49 (0.76–2.91) 0.001  
 CES Number of incidents/incidence ratea 8/0.98 21/0.72 39/0.61 39/0.49 32/0.50 34/0.77 17/1.07    
  Multivariable SHR (95% CI)b 1.45 (0.67–3.13) 1.32 (0.78–2.24) 1.19 (0.76–1.85) 1 1.02 (0.64–1.63) 1.63 (1.03–2.59) 2.17 (1.21–3.88) 0.16  

CES, cardioembolic stroke; CI, confidence interval; LAOS, large-artery occlusive stroke; LS indicates lacunar stroke; SHR, sub-distribution hazard ratio.

aCrude incident rates were expressed as rate per 1,000 person-years.

bAdjusted for baseline age, updated smoking, alcohol consumption, leisure-time physical activity, and histories of dyslipidemia and diabetes mellitus.

cMedian values of cumulative average body mass index in each categories were used for test of a linear trend across categories.

dInteraction was tested by multiplying hypertension and categories of cumulative average body mass index.