Skip to main content
. Author manuscript; available in PMC: 2013 Jul 5.
Published in final edited form as: JAMA. 2012 Apr 18;307(15):1611–1620. doi: 10.1001/jama.2012.460

Table 6.

Outcomes for IMRT vs Proton Therapy With Propensity Score Matching and Instrumental Variable Analyses

Propensity Score Matcheda Instrumental Variable Analysis
IMRT
(n = 684)		 Proton
(n = 684)		 IMRT
(n = 8144)b 		Proton
(n = 1978)b
Outcome per 100
Person-Years		 Total
Events		 100
Person
Years		 Rate Total
Events		 100
Person
Years		 Rate IMRT vs PT
Rate Ratio
(95% CI)		 Total
Event
s		 100
Person
Years		 Rate Total
Events		 100
Person
Years		 Rate IMRT vs PT
Rate Ratio
(95% CI)
Gastrointestinal events
  Procedures (including
colonoscopy)		 302 17 17.7 347 16.2 21.4 0.82 (0.70–.97) 3074 169 18.2 883 41 21.6 0.60 (0.46–.78)
  Diagnoses 235 19 12.2 301 16.9 17.8 0.66 (0.55–.79) 2620 182 14.4 714 45 16.0 0.66 (0.49–.88)
Urinary nonincontinence events
  Proceduresc 44 25 1.8 42 25.8 1.6 1.06 (0.69–.63) 466 233 2.0 113 62 1.8 1.71 (0.87–.36)
  Diagnoses 161 22 7.5 144 22.9 6.3 1.25 (0.99–.58) 1864 198 9.4 454 53 8.6 1.10 (0.78–.58)
Urinary incontinence events
  Procedures 161 21 7.6 173 22.1 7.8 0.97 (0.77–.20) 2029 194 10.5 511 51 10.0 1.06 (0.76–.50)
  Diagnosesc 75 24 3.1 82 24.8 3.3 0.96 (0.70–.32) 816 226 3.6 200 59 3.4 1.03 (0.63–.71)
Erectile dysfunction events
  Proceduresc 21 25 0.8 36 26.2 1.4 0.61 (0.35–.06) 206 239 0.9 70 63 1.1 0.58 (0.24–.41)
  Diagnoses 145 22 6.6 164 22.2 7.4 0.89 (0.70–.12) 1454 208 7.0 436 53 8.3 0.78 (0.54–.13)
Hip fractured 21 26 0.8 18 26.6 0.7 192 239 0.8 40 63 0.6 1.42 (0.50–.02)
Additional cancer therapy 58 26 2.2 52 27.5 1.9 1.26 (0.86–.84) 588 252 2.3 124 67 1.9 1.60 (0.85–.00)

Abbreviation: IMRT, intensity-modulated radiation therapy.

a

Rates shown are adjusted for the variables presented in Tables 4 and 5, using propensity scores implemented by matching.

b

Rates for IMRT and proton were adjusted with a 2-stage least-squares instrumental variable approach in which Radiation Therapy Oncology Group affiliation predicts proton use: this predicted value was subsequently used as exposure in an adjusted outcome model to estimate the effect of IMRT vs proton on the outcome.

c

Because of zero cell counts, Surveillance, Epidemiology, and End Results region was not included in propensity score–matched models.

d

Because of the small number of events and zero cell counts in some covariates in the propensity score–matched model, rate ratio could not be calculated.