Association between exposure to proton pump inhibitors and hypertension: a descriptive and disproportionality analysis of VigiBase

Abstract

Objectives

The potential link between proton pump inhibitors (PPIs) and hypertension remains unclear. It is uncertain whether such an association exists, whether it represents a class-of-PPI effect and whether a dose–response relationship is involved. This study aimed to investigate the potential class-of-PPI effect associating PPIs with hypertension reporting and evaluate whether the association follows a dose-dependent pattern.

Design

A disproportionality analysis was conducted within VigiBase to identify signals of hypertension reporting associated with individual PPIs by calculating adjusted reporting ORs (aRORs) within a multivariate case/non-case study design. Additionally, the presence of a dose–response relationship was explored.

Setting

Real-world data from VigiBase, the WHO pharmacovigilance database, was used.

Participants

All individual case safety reports with PPI use were included.

Outcome measures

Incident hypertension cases were identified using the Medical Dictionary for Regulatory Activities V.26.1 related to at least one PPI administration that were systematically collected until 28 October 2024. Pharmacovigilance signals between the use of PPIs and hypertension reported and dose dependence between PPI posology and onset or worsening hypertension were analysed.

Results

The database contained 26 587 reports of PPI-associated hypertension (2.3%), predominantly among women (63.3%). Hypertension was most frequently reported in the group aged 45–64 years (41.4%). A significant reporting OR (ROR) was observed for almost all PPIs in both univariable (RORs, 1.32–1.97) and multivariate analyses (aRORs, 1.09–1.35) after adjustments for age group, sex, concurrent antihypertensive medication and drugs known to induce hypertension, with the exception of lansoprazole (aROR 0.99, 95% CI 0.96 to 1.03). A potential trend suggestive of a dose–response relationship was identified, with doses lower than the median associated with a lower aROR for hypertension than doses higher than the median for all PPIs. However, this trend was not statistically significant, potentially due to insufficient statistical power.

Conclusions

This investigation indicates a notable pharmacovigilance safety signal associating PPI usage with hypertension reporting. Although a potential dose–response trend was observed, it was not statistically significant, possibly due to limited statistical power. Further longitudinal studies are warranted.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study analysed the largest cohort of proton pump inhibitor-related adverse events.

• This cohort is international, as it includes various pharmacovigilance reports from around the world under the auspices of the WHO.

• The value of this type of work lies in attempting to identify emerging safety signals.

• The retrospective data are subject to specific biases and substantial missing data.

• No blood pressure measurements were reported in individual safety reports related to pharmacovigilance.

Introduction

Hypertension significantly contributes as a risk factor for cardiovascular diseases (CVDs).^{1 2} Although essential hypertension predominates clinically, secondary hypertension can arise from conditions such as renal parenchymal disease, renal artery stenosis, hyperaldosteronism or pheochromocytoma. Pharmacologically induced hypertension, though less acknowledged, can lead to induced or uncontrolled hypertension.³

Elsewhere, the link between gastro-oesophageal reflux disease (GERD) and onset or worsening of hypertension is debated. A bidirectional Mendelian randomisation study revealed that individuals with GERD present an increased risk of hypertension (OR 1.46, 95% CI 1.33 to 1.59, p<0.001).⁴ Proton pump inhibitors (PPIs) are promoted as first-line antacid agents and are used in a wide range of indications, including GERD.⁵ Furthermore, the conjecture that PPIs could significantly influence blood pressure (BP)—either directly or by moderating GERD—has emerged.^{6 7} A nested case–control study indicated that patients taking PPIs exhibited a significantly higher prevalence of using antihypertensive medications compared with controls. The CVD risk was significantly high for current (OR 1.61; 95% CI 1.55 to 1.68) and recent PPI users, defined as having the last day covered by PPI therapy between 29 and 56 days before the event date (OR 1.15, 95% CI 1.06 to 1.26) compared with past users (having the last day covered by PPI therapy >56 days before the event date).⁶ Conversely, Li et al demonstrated a significant correlation between hypertension and GERD, suggesting that PPIs may restore normal oesophageal pH and assist in maintaining normal BP.⁷ Nonetheless, the debate regarding PPI-associated hypertension persists, and hypertension is not acknowledged as a possible adverse effect in the French Summaries of Product Characteristics of drugs of this class.

To investigate the potential class-of-PPI effect associating PPIs with hypertension reporting, real-world data from VigiBase, the World Health Organisation (WHO) pharmacovigilance database, were analysed to assess the presence of a signal and evaluate whether the association follows a dose-dependent pattern.

Methods

Population/data source

This study analysed data from VigiBase, the WHO’s comprehensive Individual Case Safety Report (ICSR) database.⁸ To minimise the potential for duplicated reports, a deduplicated version of VigiBase based on the Uppsala Monitoring Centre’s standard algorithm was used for identifying and removing duplicate ICSRs. Encompassing >40 million reports from 120 nations since 1968, ICSRs integrate administrative data (reporting country, report type and reporter qualifications), patient demographics (sex and age), reaction onset date, outcome characterisation using MedDRA version 27.1 terms, WHO causality assessment and detailed drug information (name, administration and cessation dates, induction period, prescribed indication, dosage, dechallenge and rechallenge), accompanied by report completeness levels. Each adverse drug reaction (ADR) was classified as either ‘serious’ or ‘non-serious’ based on WHO criteria, with ‘serious’ encompassing outcomes such as death, life-threatening situations, hospitalisation or its prolongation, persistent incapacity or disability and clinically significant events as judged by the reporting physician.⁹ CVD, including myocardial infarction cases and stroke cases, was also searched.

Case/non-case study in VigiBase

ICSRs reporting hypertension—defined by the MedDRA System Organ Class version 27.1 as ‘Hypertension (broad)’ Standardised MedDRA Query (SMQ)—from 1 January 1984 (earliest year with PPI cases registered in VigiBase) to 28 October 2024, comprised our ‘cases’ within the adult demographic, while ‘non-cases’ involved all adult reports devoid of hypertension. Terms included in the ‘Hypertension (broad)’ (SMQ) are further described in online supplemental table 1. Drug exposure was classified by identifying at least one PPI from the following list: omeprazole, esomeprazole, lansoprazole, pantoprazole, dexlansoprazole and rabeprazole, as per the ATC classification within a report. All PPIs were included in the analysis regardless of their status (suspect, concomitant or interacting). Three sensitivity analyses were performed: one involving only suspect PPIs, a second with a different definition of hypertension (the narrow definition of the SMQ) and one with a different comparator (patients with another antacid defined as the ATC A02 class). The disproportionality analysis in this case/non-case format, aimed at examining the association between drug exposure and hypertension occurrence,10,12 was used to estimate reporting ORs (RORs) and their 95% CIs for each PPI. The ROR was calculated as the ratio of the odds of reporting hypertension among reports with the drug of interest (PPI) compared with the odds of reporting hypertension among all other reports without that drug, that is, (cases with PPI/non-cases with PPI)/(cases without PPI/non-cases without PPI).

Statistical analysis

Data were expressed as means±SD or percentages. For the case/non-case analysis, the ROR with 95% CI was estimated for each PPI–hypertension pair, and a safety signal was identified with a ROR >1 and a lower 95% CI limit >1. Furthermore, to mitigate bias, multivariate analyses were performed, adjusting for age, sex, concurrent C02 ATC class antihypertensive reports, drugs known to induce hypertension as defined by Foy et al¹³ (online supplemental material), type of reporter and WHO region to calculate adjusted RORs (aRORs). These variables were selected based on their clinical relevance and the practical constraints of working with a pharmacovigilance database, which limits analyses to the data elements systematically collected and available within individual case safety reports. The formulas of the models are available in the online supplemental material. The dose–response relationship was investigated by calculating RORs separately for dosage groups below or equal to the median and above the median for each PPI, using logistic regression models adjusted for relevant covariates. Additionally, RORs were estimated using a model treating dose as a continuous variable (adr~dose), also adjusted for the same variables. Time to onset (TTO), number of positive dechallenges and rechallenges were also reported. The TTO in VigiBase represents the reporter’s estimation of when elevated BP or related symptoms first occurred. This may not correspond to a confirmed clinical diagnosis, as formal hypertension diagnosis requires multiple measurements over time. For all analyses, p values<0.05 were considered statistically significant. Model performance was evaluated using the area under the curve (AUC), McFadden’s pseudo-R² and Nagelkerke’s pseudo-R². The analyses encompassed the collective PPIs and adhered to the READUS-PV guidelines with execution in R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria)^{14 15} and use of dplyr and vigicaen packages.

Patient and public involvement

None.

Results

Descriptive characteristics

VigiBase reports relevant to the included PPIs totalled 1 149 888. Adhering to the selection criteria, 26 587 hypertensive cases (2.3%) were identified (figure 1). PPI-associated hypertension was predominantly reported among women (63.3%). Hypertension was most frequently reported in the group aged 45–64 years (41.4%). The number of reports associated with omeprazole was 9 935; pantoprazole, 8 276; esomeprazole, 5 737; lansoprazole, 3 430; rabeprazole, 1 272; and dexlansoprazole, 522. Concurrent antihypertensive drug use was reported in 2 786 (10.5%) cases, and a total of 3 233 reports were linked with CVDs (1 422 myocardial infarction cases and 1 811 stroke cases). Serious cases constituted 18 553 (73.4%) reports (table 1). In serious cases, most co-reported adverse events were dyspnoea (2 821 reports), headache (2 807 reports), fatigue (2 806 reports), nausea (2 767 reports) and dizziness (2 207 reports), and most co-reported drugs were acetylsalicylic acid (4 708 reports), levothyroxine (3 108 reports), paracetamol (2 943 reports), amlodipine (2 937 reports) and metoprolol (2846 reports). Doses are reported in online supplemental figure 1.

Figure 1. Flow chart of individual case safety reports (ICSRs) in WHO pharmacovigilance database concerning proton pump inhibitor (PPI)-induced hypertension.

Table 1. Characteristics of the 26 587 individual case safety reports (ICSRs) of hypertension with PPIs reported in VigiBase.

Sex	N available	26 452
	Female	16 742 (63.3%)
Age	N available	26 587
	18–44	2 883 (10.8%)
	45–64	11 018 (41.4%)
	65–74	7 201 (27.1%)
	75+	5 485 (20.6%)
Serious	N available	25 279
	Yes	18 553 (73.4%)
Seriousness criteria*	Caused/prolonged hospitalisation	8 029 (31.8%)
	Congenital anomaly/birth defect	2 (0.0%)
	Death	1 347 (7.3%)
	Disabling/incapacitating	378 (1.5%)
	Life threatening	1 040 (4.1%)
	Other†	6 982 (27.6%)
Reporter type	Consumer or other non-health professional	8 277
	Lawyer	537
	Other health professional	4 186
	Pharmacist	2 043
	Physician	8 324
WHO region	African region	45
	Eastern Mediterranean region	139
	European region	7 217
	Region of the Americas	17 119
	South-East Asia region	243
	Western Pacific region	1 820
Concomitant stroke		1 811 (6.8%)
Concomitant myocardial infarction		1 422 (5.3%)
Concomitant antihypertensive drug		2 786 (10.5%)
Concomitant potentially hypertensive drug		13 585 (51.1%)

^*Several seriousness criteria can be added at the same time.

^†Report when the event does not fit the other outcomes, but the event may jeopardise the patient and may require medical or surgical intervention (treatment) to prevent one of the other outcomes.

Univariate and multivariate case/non-case analyses in VigiBase

Table 2 illustrates that PPIs elicited a significant signal for hypertension in the univariable analysis (RORs, 1.32–1.97) and multivariate analysis (aRORs, 1.09–1.35) post-adjustment for age group, sex, concurrent antihypertensive medication and drugs known to induce hypertension, except lansoprazole (aROR 0.99, 95% CI 0.96 to 1.03). However, when considering reports in which PPIs were classified as suspect for our analyses, only the signal for esomeprazole remained significant (aROR 1.21). The multivariate logistic regression model demonstrated good overall performance with a McFadden’s pseudo-R² of 0.251 and Nagelkerke’s pseudo-R² of 0.266, indicating that the model explained approximately 25%–27% of the variance in ADR occurrence. The AUC was 0.626, suggesting moderate discriminative ability to distinguish between cases with and without hypertension-related ADRs. Detailed results of the multivariate logistic regression analysis, including coefficients, standard errors, z-values and 95% CIs for all variables, are presented in online supplemental table 2.

Table 2. Disproportionality analysis with all PPIs in VigiBase to search for a signal of hypertension.

Drug name	PPI-exposed cases	Non-PPI-exposed cases	PPI-exposed non-cases	Non-PPI-exposed non-cases	Univariate analysis		Multivariate analysis*
					ROR	95% CI	ROR	95% CI
Dexlansoprazole	522	401 386	16 934	25 696 139	1.97	(1.81 to 2.15)	1.17	(1.06 to 1.28)
Esomeprazole	5 737	396 171	206 658	25 506 415	1.79	(1.74 to 1.84)	1.25	(1.22 to 1.29)
Lansoprazole	3 430	398 478	166 183	25 546 890	1.32	(1.28 to 1.37)	0.99	(0.96 to 1.03)
Omeprazole	9 935	391 973	428 691	25 284 382	1.49	(1.47 to 1.53)	1.09	(1.06 to 1.11)
Pantoprazole	8 276	393 632	314 812	25 398 261	1.7	(1.66 to 1.73)	1.35	(1.32 to 1.38)
Rabeprazole	1 272	400 636	53 954	25 659 119	1.51	(1.43 to 1.60)	1.34	(1.26 to 1.42)
PPI class	26 583	375 325	1 123 305	24 589 768	1.55	(1.53 to 1.57)	1.19	(1.17 to 1.20)

^*Adjusted on age category, sex, associated antihypertensive drugs (ATC codes), reporter type, WHO region and drugs known to induce hypertension.

Sensitivity analyses were conducted to test the robustness of our findings (online supplemental tables 3–5). When restricting the analysis to reports where PPIs were specifically classified as suspected drugs, the association was attenuated for most PPIs, with only esomeprazole maintaining a significant signal (aROR 1.21, 95% CI 1.14 to 1.28). Using a narrower case definition limited to hypertension-specific SMQs strengthened the associations, with pantoprazole showing the strongest signal (aROR 1.91, 95% CI 1.78 to 2.04) and improved model discrimination (AUC 0.73). When comparing PPIs with other antacids as the reference group rather than all other drugs, significant associations persisted for all individual PPIs, with rabeprazole demonstrating the highest adjusted OR (aROR 1.76, 95% CI 1.65 to 1.87).

Dose effect of proton-pump inhibitors associated with hypertension

When stratifying PPI doses into two groups (≤ median and > median), the RORs for hypertension were consistently higher in the group receiving doses above the median, suggesting a possible dose-dependent increase in risk. For example, dexlanoprazole showed an adjusted ROR of 1.63 (95% CI 1.32 to 2.01) in the lower dose group versus 1.92 (95% CI 0.26 to 14.35) in the higher dose group. However, when the dose was modelled as a continuous variable, the estimated RORs for all PPIs were close to 1 with no statistically significant associations, indicating no clear linear dose–response relationship. This discrepancy suggests that the relationship between the dose and hypertension risk may not be strictly linear or that the study may lack power to detect a dose effect when treated continuously owing to the smaller subset of reports with available dose information, even though the overall dataset includes thousands of reports (table 3).

Table 3. Disproportionality analysis of the signal of hypertension associated with PPIs to search for a dose-dependency effect.

Drug name	Hypertension depending on drug exposure								Hypertension depending on dose (continuous variable)
	Doses ≤ to the median population				Doses > to the median population				Adjusted p-value (dose–response model)†
	Univariate analysis		Multivariate analysis*1		Univariate analysis		Multivariate analysis*
	ROR	95% CI	ROR	95% CI	ROR	95% CI	ROR	95% CI
Dexlansoprazole	2.57	(2.12 to 3.11)	1.63	(1.32 to 2.01)	5.82	(1.37 to 24.75)	1.92	(0.26 to 14.35)	0.24
Esomeprazole	2.18	(2.05 to 2.32)	1.33	(1.23 to 1.44)	3.26	(2.85 to 3.72)	1.88	(1.59 to 2.24)	0.36
Lansoprazole	1.22	(1.09 to 1.36)	0.99	(0.88 to 1.11)	2.02	(1.55 to 2.62)	1.47	(1.11 to 1.95)	0.29
Omeprazole	1.56	(1.45 to 1.67)	1.08	(1.01 to 1.17)	1.71	(1.58 to .85)	1.27	(1.17 to 1.38)	0.82
Pantoprazole	1.45	(1.36 to 1.54)	1.23	(1.15 to 1.31)	1.73	(1.50 to 2.00)	1.37	(1.18 to 1.59)	0.63
Rabeprazole	0.89	(0.74 to 1.08)	0.92	(0.76 to 1.11)	1.01	(0.63 to 1.61)	1.09	(0.67 to 1.76)	0.09

^*Adjusted on age category, sex, associated antihypertensive drugs (ATC codes), reporter type, WHO region and drugs known to induce hypertension.

^†Logistic regression model assessing hypertension occurrence depending on PPI dose and adjusted on the same variables.

Time to onset of proton-pump inhibitor-associated hypertension and positive dechallenges and rechallenges associated with those drugs

The TTO was consistent and relatively short, with hypertension occurring within a median of 0−2 days after PPI initiation, except for dexlansoprazole (median, 11 days) (table 4). Details about the TTO can be found in onlinesupplemental figures 2 3. Positive dechallenges and rechallenges associated with those drugs were also documented, with the most numbers associated with omeprazole (table 5). Details per WHO region are available in online supplemental table 6.

Table 4. Time to onset (TTO) of hypertension after proton-pump inhibitor exposure.

Drug	Number of reports with available information	TTO in days (median, IQR)
Dexlansoprazole	9	11.0 (0.0–23.0)
Esomeprazole	217	2.0 (0.0–36.0)
Lansoprazole	100	0.6 (0.0–7.2)
Omeprazole	366	0.0 (0.0–3.0)
Pantoprazole	198	1.0 (0.0–6.0)
Rabeprazole	70	1.0 (0.0–30.5)

Table 5. Dechallenges and rechallenges after the onset of hypertension.

Drug	Positive dechallenge	Positive rechallenge
Dexlansoprazole	14	1
Esomeprazole	131	3
Lansoprazole	48	1
Omeprazole	171	16
Pantoprazole	101	5
Rabeprazole	33	2

Discussion

This investigation substantiates a pharmacovigilance signal for hypertension in patients administered PPIs, after adjustments for multiple confounders. A class-of-PPI effect seems plausible. The sensitivity analyses support the robustness of the observed associations between PPI use and hypertension-related adverse events, though the magnitude of the signal varies depending on the analytical approach employed. A potential trend suggestive of a dose–response relationship was observed, with doses below the median associated with a lower aROR for hypertension than with doses above the median for all PPIs. However, this trend was not statistically significant, possibly due to limited statistical power or a non-linear dose–response relationship.

Previous studies within the VigiBase on this topic are sparse, potentially characterised by suboptimal methodologies. An observational study was conducted to describe the profile of individuals with hypertension related to omeprazole. Bahta et al analysed 1 043 cases of hypertension with omeprazole but did not assess pharmacovigilance signals, thereby neglecting the effect of potential confounding factors.¹⁶

In addition to the pharmacovigilance signal linking PPIs and hypertension, recent epidemiological data seem to corroborate our findings. A nested case–control study was performed, identifying 7 832 cases and 89 160 controls among new PPI users (males, 64.9%; mean age, 58.9 years). This study indicated a greater usage of antihypertensives, and the risk of CVD events was significantly higher for current (OR 1.61, 95% CI 1.55 to 1.68) and recent (OR 1.15; 95% CI 1.06 to 1.26) users than for past users. Analogous results were observed when stratified for cardiovascular (OR current 1.71, 95% CI 1.63 to 1.81) and cerebrovascular events (OR current 1.43, 95% CI 1.34 to 1.54). The same analysis for H2-antagonist use showed no significant results unlike PPIs.⁶ Conversely, Li et al described a 14-day PPI therapy administered to 24 patients with both essential hypertension and GERD, where reflux episodes were associated with increased BP, and PPI therapy restored oesophageal pH to normal, significantly lowering the high BP. They concluded that treatment of GERD could be useful as adjunctive therapy for normalising BP in patients with essential hypertension. However, compelling evidence now suggests that the chronic use of PPIs impairs vascular homeostasis and increases BP, thereby increasing the risk of adverse CVDs.17,19

The mechanisms by which PPIs may induce hypertension could be the dysregulation of interdependent pathways controlling vascular nitric oxide (NO) production and bioavailability.^{19 20} First, PPIs decrease the enzymatic activity of dimethylarginine dimethylaminohydrolase, the enzyme responsible for metabolising asymmetric dimethylarginine, an endogenous and competitive inhibitor of NO synthase.²¹ By reducing endothelial NO synthase activity, asymmetric dimethylarginine disrupts NO production within the vasculature, increasing the CVD risk. Additionally, PPIs may impair vascular redox homeostasis and promote oxidative stress mediated by increased xanthine oxidase formation of superoxide, thus impairing endothelium-dependent vasodilation.²² Moreover, strong evidence suggests that PPIs also interfere with NO vascular homeostasis by disrupting the nitrate–nitrite–NO pathway,^{18 22 23} which is a crucial alternative source of NO compared with the classical L-arginine–NO synthase pathway.²⁴ Under physiological conditions, the anions nitrate and nitrite are recycled to form NO and other bioactive nitrogen species through enzymatic and non-enzymatic pathways in the blood and tissues. Inorganic nitrate is normally present in our diet, particularly within leafy greens and root vegetables. After ingestion, nitrate is rapidly absorbed in the small intestine, increasing plasma nitrate levels, which are then taken up by the salivary glands and concentrated up to 20-fold in saliva.²⁴ The bioconversion of nitrate from dietary sources requires the reduction of nitrate to nitrite in the oral cavity through the action of nitrate reductase enzymes from commensal bacteria. Subsequently, saliva enters the acidic stomach, facilitating a non-enzymatic reduction of nitrite to NO and other bioactive NO-related species.¹⁸ Among these bioactive compounds, S-nitrosothiols are generated and may act as relatively stable NO donors.¹⁸ Orally administered nitrate has also demonstrated a dose-dependent reduction in BP and vasoprotection in humans, whereas nitrite decreased BP in hypertensive rats.^{25 26}

Strengths and limitations

Strengths of this study include the use of the validated case/non-case methodology for ADR signal detection in pharmacovigilance data.^{11 12} VigiBase, with its global report aggregation, offers a diverse range of medical practices and patient characteristics. The enhanced methodology, featuring multivariate analysis, confers greater relevance to this investigation. Notwithstanding, limitations common to pharmacovigilance database studies persist, such as the significant issue of under-reporting or reporting bias toward more severe cases, which, while important, does not detract from the validity and significance of the case/non-case methodology.¹¹ VigiBase is a spontaneous reporting database and contains only reports of suspected ADRs. It does not provide information on the total number of patients exposed to PPIs; therefore, we cannot estimate the incidence or absolute risk of hypertension among all PPI users. Our results reflect reporting disproportionality rather than population-level risk. Thus, every case reported in VigiBase had to have both a drug and a possible adverse event. Consequently, patients with no adverse events are not entered into the database. Our explanation for a high ROR for hypertension in PPIs is that PPIs cause hypertension. However, there are other alternatives: PPIs reduce some other adverse events (but not hypertension), and these are reported more frequently for other drugs. Another alternative is that other drugs are more likely to have adverse events other than hypertension associated with them, so in comparison, PPIs appear to have a higher ROR of hypertension. Missing data also represent a limitation in pharmacovigilance database extractions. For example, the TTO of hypertension after PPI exposure was calculated only for reports containing information on the start and onset dates. As these data were available for a small subset of cases, the results should be interpreted with caution and may not be representative of the overall study population. It is also true for dechallenge/rechallenge data. Moreover, our case definition relied on the ‘Hypertension (broad)’ SMQ from MedDRA version 27.1. Differences in MedDRA coding practices or selection of hypertension terms for hypertension definition in other studies could affect case identification and comparability. Additionally, antihypertensive drugs were included as a single adjustment category, which does not account for potential differences between specific agents; residual confounding by type or dose of antihypertensive medication cannot be excluded. Moreover, considering GERD’s potential effect on BP, discerning the precise influence of PPIs on BP remains challenging. However, given the absence of a signal for hypertension in the analyses conducted with only suspected PPIs—excluding instances concerning omeprazole—we cannot negate with certainty that the signal identified in our primary analysis could be influenced by unidentified confounding factors. It is indeed plausible that part of our signal is driven by the association between reflux episodes and an increased risk of developing hypertension.⁶ However, the short TTO of hypertension following PPI initiation, as well as the number of positive dechallenges and rechallenges, further supports the hypothesis of PPI-induced hypertension. Time-to-event (survival) analysis was not performed because TTO data were available for only a small subset of reports, and such analyses are not recommended for pharmacovigilance signal detection.¹⁵ There is likely to be a demographic bias within this sample of PPI users, as it is unlikely that they correspond to the expected demographic proportions of the general population, such as body mass index or age.

Conclusion

This pharmacovigilance analysis, conducted under real-world circumstances, has revealed a significant pharmacovigilance safety signal linking PPI administration with the onset or exacerbation of hypertension. From a clinical perspective, this study may help physicians treating patients with gastro-oesophageal disease to recognise the potential for PPIs to increase BP. Consequently, from a practical standpoint, screening or monitoring for hypertension should be carried out systematically. Given the potential dose–response relationship observed, clinicians should consider PPI dosing carefully, following symptomatology and established guidelines while prioritising the lowest effective dose whenever possible. However, further work, such as prospective studies, is needed to confirm this pharmacovigilance signal.

Data availability statement

Data sharing is not applicable as no datasets were generated and/or analysed for this study.