. 2020 Sep 30;8(5):e00651. doi: 10.1002/prp2.651

Table 2.

All‐cause mortality estimates during long‐term use of PPI (>12 weeks)

Type of study (Reference)	Deaths n/N (%)	Association (95% Confidence Interval) NNH
Systematic review and meta‐analysis of three observational studies ¹⁷ Median follow‐up 1 year	PPI: 765/4,775 (16%) Non‐PPI users: 1,794/17,652 (10%)	OR 1.68 (1.53‐1.84)
US Veterans Affairs longitudinal cohort study ¹⁸ new users of PPI vs. H2RA Median follow‐up 10 years	PPI: 59,771/157,625 (37.9%) H2RA: 20,287/56,842 (35.7%)	HR 1.17 (1.10‐1.24) 45.20 excess deaths/1,000 (28.20‐61.40)
COMPASS RCT ¹⁹ Pantoprazole 40mg/d vs. placebo Median follow‐up 3 years	PPI: 630/8791 (7.2%) Placebo: 614/8807 (7.0%)	HR 1.03 (0.92‐1.15)

Type of study (Reference)

Deaths

n/N (%)

Association

(95% Confidence Interval)

NNH

Systematic review and meta‐analysis of three observational

studies ¹⁷

Median follow‐up 1 year

PPI: 765/4,775 (16%)

Non‐PPI users: 1,794/17,652 (10%)

OR 1.68

(1.53‐1.84)

US Veterans Affairs

longitudinal cohort study ¹⁸

new users of PPI vs. H2RA

Median follow‐up 10 years

PPI: 59,771/157,625 (37.9%)

H2RA: 20,287/56,842 (35.7%)

HR 1.17

(1.10‐1.24)

45.20 excess deaths/1,000

(28.20‐61.40)

COMPASS RCT ¹⁹

Pantoprazole 40mg/d vs. placebo

Median follow‐up 3 years

PPI: 630/8791 (7.2%)

Placebo: 614/8807 (7.0%)

HR 1.03

(0.92‐1.15)