Table 3.

 Sensitivity analyses comparing relative risks for oesophageal cancer in relation to prescription of oral bisphosphonates, using varying criteria for definition of exposure and inclusion of adjustment variables

Sensitivity analyses	No of exposed cases of oesophageal cancer	Relative risk* (95% CI) 1-1-9 v 0 prescriptions	Relative risk* (95% CI) ≥10 v 0 prescriptions
Main analysis: bisphosphonate exposure defined as ≥1 prescription within observation period; missing data on adjustment variables assigned to separate category	90	0.93 (0.66 to 1.31)	1.93 (1.37 to 2.70)
Bisphosphonate exposure defined as ≥2 prescriptions	80	0.93 (0.63 to 1.39)	1.93 (1.37 to 2.70)
Restricted to those with full information on adjustment variables	68	1.05 (0.68 to 1.63)	1.88 (1.24 to 2.86)
All exposure and adjustment data recorded at least 12 months before index date, and restricted to those with full information on adjustment variables	52	1.09 (0.67 to 1.79)	2.00 (1.23 to 3.27)

*Adjusted for smoking status, alcohol intake, and body mass index.