Skip to main content
. 2016 Nov 15;355:i5813. doi: 10.1136/bmj.i5813

Table 2.

Adjusted and unadjusted hazard ratios and 95% confidence intervals for association of community acquired pneumonia with proton pump inhibitors (PPI) according to three different definitions of community acquired pneumonia (Cox regression analysis)

Parameter Pneumonia definition
Broad primary care definition Narrow primary care definition Hospital and mortality records based definition (in HES/ONS linked patients)
Unexposed Exposed Unexposed Exposed Unexposed Exposed
No of patients matched for date, sex, and 5 year age band 160 000 160 000 160 000 160 000 128 943 128 943
No of pneumonia events 1582 3798 74 392 234 1090
Person years of follow-up 52 133 49 885 52 133 49 885 43 824 43 456
Rate of pneumonia per 1000 person years (95% CI) 30.34 (28.88 to 31.87) 76.13 (73.75 to 78.59) 1.37 (1.09 to 1.73) 7.31 (6.62 to 8.07) 5.48 (4.82 to 6.23) 25.74 (24.26 to 27.32)
Unadjusted hazard ratio (95% CI) 2.51 (2.36 to 2.67) 5.44 (4.23 to 6.99) 4.76 (4.12 to 5.49)
Adjusted hazard ratio* (95% CI) 1.67 (1.55 to 1.79) 3.73 (2.69 to 5.16) 3.55 (3.03 to 4.16)
Adjusted hazard ratio† (95% CI) 1.65 (1.53 to 1.77) 3.87 (2.75 to 5.44) 3.54 (3.02 to 4.16)

HES=Hospital Episode Statistics; ONS=Office for National Statistics.

*Model adjusted for confounders that changed PPI hazard ratio by >10%: (comorbidity index, number of general practitioner (GP) visits, corticosteroid use, and immunosuppression).

†Model adjusted for potential confounding variables for pneumonia (malignancy, cerebrovascular disease, hemiplegia, chronic pulmonary disease, congestive cardiac disease, lymphoproliferative disorders, dementia, diabetes, renal disease, rheumatological disease, myocardial infarction, and liver disease), in addition to number of GP visits, corticosteroid use, and immunosuppression.