Table 2.

Quality of evidence and risks of adverse effects associated with long-term proton pump inhibitors (PPIs)

	Potential adverse effect	Nature of evidence	Risk estimate
Causality established, idiosyncratic, rare	Acute interstitial nephritis	Observational, case–control	OR 5.16 (2.21–12.05)
Causality proven but of minimal significance	Fundic gland polyps	Observational	OR 2.2 (1.3–3.8) [6]
	B12 deficiency	Observational, case–control	OR 1.65 (1.58–1.73) [7]
Weak association, causality probable	Small intestinal bacterial overgrowth	Meta-analysis	OR 2.28 (1.23–4.21) [8]
	Spontaneous bacterial peritonitis in cirrhotic patients	Systematic review/meta-analysis	OR 2.17 (1.46–3.23) [9]
	Hepatic encephalopathy in cirrhotic patients	Observational, case–control	Dose dependent response, up to OR 3.01 (1.78–5.10) [10]
	Clostridium difficile infection	Observational cohort study	OR 2.10 (1.20–3.50)
	Iron deficiency	Observational, case control	OR 2.49 (2.35–2.64) [11]
	Hypomagnesemia	Observational, population-based cohort	OR 2.00 (1.36–2.93)a [12]
Weak association, unproven causality	Bone fracture	Observational, case–control	OR 2.65 (1.80–3.90)
	Chronic kidney disease	Observational, population-based cohort	HR 1.50 (1.14–1.96) [13]
	Dementia	Prospective observational cohort	HR 1.44 (1.36–1.52)
	Myocardial infarction	Observational, data mining	HR 1.16 (1.09–1.24)b
	Community-acquired pneumonia	Systematic review/meta-analysis	OR 1.49 (1.16–1.92)b

Table adapted from Kia et al. [14]

^aThe risk of hypomagnesemia increases to OR 7.22 (1.69–30.83) in patients on concurrent loop diuretics

^bRisk ratio based on observational study; no association found in RCTs

HR hazard ratio, OR odds ratio