The risk of hyponatremia induced by SSRIs and SNRIs antidepressants: a systematic review and meta-analysis

Abstract

Objectives

To systematically evaluate the risk differences of hyponatremia induced by selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs), stratify risks among individual drugs, and provide evidence-based guidance for clinical medication safety.

Methods

A systematic search was conducted across the Cochrane Library, PubMed, and Web of Science databases. Study quality was assessed using the Newcastle-Ottawa Scale (NOS), and the certainty of evidence was evaluated using the GRADE framework. A meta-analysis was performed to compare the event rates and odds ratios (ORs) of hyponatremia between SSRIs and SNRIs, followed by subgroup analysis and bias assessment.

Results

A total of 38 observational studies (including 30 cohort studies and 8 case-control studies) were included in this study. The overall event rate of hyponatremia with antidepressants was 6.03% (P < 0.001), with rates of 5.98% for SSRIs and 6.13% for SNRIs. Both drug classes significantly increased the risk of hyponatremia (SSRIs: OR = 2.158; SNRIs: OR = 2.270, P < 0.001), with SNRIs demonstrating a higher risk in clinically relevant hyponatremia (OR = 2.227, P < 0.001). Risk stratification among individual drugs revealed that fluoxetine (SSRIs) and venlafaxine (SNRIs) had the highest risk, while sertraline and duloxetine were associated with lower risks.

Conclusion

Both SSRIs and SNRIs significantly increase the risk of hyponatremia, with SNRIs posing a slightly higher risk. Clinicians should consider individual patient characteristics when selecting lower-risk medications and enhance serum sodium monitoring in high-risk populations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40360-025-00977-1.

Introduction

Hyponatremia (serum sodium concentration < 135 mmol/L) is one of the most common electrolyte disorders in clinical practice. It is classified into mild (130–135 mmol/L), moderate (125–129 mmol/L), and severe (< 125 mmol/L) based on serum sodium levels [1]. The presence and severity of clinical symptoms primarily depend on the rate and magnitude of sodium decline [2]. Mild hyponatremia often manifests as nonspecific symptoms such as dizziness, nausea, vomiting, and headache. In contrast, moderate to severe cases may present with significant neurological complications, including lethargy, confusion, seizures, coma, and even life-threatening outcomes [3]. Multiple factors contribute to hyponatremia, with drug-induced causes garnering increasing attention in recent years.

Among patients with depression, the risk of hyponatremia is markedly elevated. Robust evidence from clinical studies indicates that antidepressants, particularly selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs), are significant risk factors for hyponatremia [4–6]. The pathophysiology of antidepressant-induced hyponatremia is primarily attributed to the syndrome of inappropriate antidiuretic hormone secretion (SIADH) [7]. SIADH arises from excessive antidiuretic hormone (ADH) secretion or abnormally enhanced renal sensitivity to ADH, leading to increased water reabsorption and subsequent dilutional hyponatremia [8]. Research has demonstrated that antidepressant medications, particularly SSRIs and SNRIs, can modulate central neurotransmitter systems, potentially leading to indirect stimulation of ADH secretion or potentiation of its physiological activity, thereby increasing the risk of developing SIADH [9]. SSRIs and SNRIs primarily potentiate serotonergic neurotransmission by inhibiting serotonin (5-HT) reuptake, leading to the activation of hypothalamic 5-HT₂ receptors and subsequent stimulation of ADH secretion [10, 11]. In addition to this shared mechanism, SNRIs exert a noradrenergic effect via selective blockade of the norepinephrine transporter (NET), which further modulates ADH release through distinct pathways [7, 12]. These pharmacological actions collectively elevate SIADH risk. Notably, antidepressant-associated hyponatremia is not dose-dependent, and high-risk populations include elderly patients and those concurrently using diuretics or angiotensin-converting enzyme inhibitors (ACEIs) [13, 14]. This vulnerability likely stems from age-related physiological decline and drug interactions predisposing individuals to electrolyte imbalances.

Given the widespread clinical use of SSRIs and SNRIs in treating depression, a systematic evaluation of their hyponatremia risk is of critical importance. This study aims to conduct a systematic review and meta-analysis to comprehensively assess the risk profile of hyponatremia associated with SSRIs and SNRIs. Furthermore, it will stratify risk levels across individual drugs within these classes, providing evidence-based guidance to enhance clinical medication safety.

Methods

Study protocol

This systematic review was performed according to a preregistered protocol (PROSPERO, CRD420251040031), which was registered on 25 April 2025 and can be accessed through this website: https://www.crd.york.ac.uk/prospero/. The study was conducted according to the 2020 PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) guidelines (Supplementary Table 1) [15].

Search strategy

This study conducted a comprehensive search across three English databases (Cochrane Library, PubMed, and Web of Science), covering all records from their inception to May 2024, with the language restriction limited to English (Supplementary Table 2). The search terms included the following keywords: ((hyponatremia) OR (hyponatraemia) OR (SIADH) OR (inappropriate ADH Syndrome) OR (inappropriate ADH)) AND ((selective serotonin reuptake inhibitors) OR (SSRIs) OR (serotonin-norepinephrine reuptake inhibitors) OR (SNRIs) OR (fluvoxamine) OR (citalopram) OR (escitalopram) OR (fluoxetine) OR (paroxetine) OR (sertraline) OR (venlafaxine) OR (desvenlafaxine) OR (duloxetine) OR (milnacipran) OR (levomilnacipran)).

Eligibility criteria

Inclusion criteria were as follows: (1) Study designs: randomized controlled trials (RCTs), cohort studies, and case-control studies; (2) Study population: patients diagnosed with SSRIs/SNRIs-associated hyponatremia; (3) Interventions: SSRIs (citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, sertraline) or SNRIs (venlafaxine, desvenlafaxine, duloxetine, milnacipran, levomilnacipran), with no restrictions on dosage, treatment duration, or administration route; (4) Outcome measures: Primary outcome - any documented hyponatremia or serum sodium level < 135 mmol/L; Secondary outcomes - clinically relevant hyponatremia including symptomatic hyponatremia, hyponatremia requiring treatment, or serum sodium level < 130 mmol/L.

Exclusion criteria were as follows: (1) Duplicate publications; (2) Animal studies; (3) Studies with incomplete or unavailable data for extraction; (4) Studies lacking relevant outcome measures; (5) Reviews, conference abstracts, and case reports.

Comparison groups

This study primarily evaluated the risk of hyponatremia associated with SSRIs and SNRIs, either as a combined class or individually, by comparing their use against non-SSRIs/SNRIs-exposed patients in terms of odds ratios (ORs), risk ratios (RRs), or event rates. For RCTs, the intervention group received SSRIs/SNRIs, while the control group received placebo or other active antidepressants. For cohort studies, the exposed group consisted of patients taking SSRIs/SNRIs (including both those who developed hyponatremia and those who did not), while the non-exposed group comprised patients on other antidepressants or untreated controls. For case-control studies, the cases were hyponatremia patients (including SSRIs/SNRIs users), and the controls were individuals without hyponatremia (including both SSRIs/SNRIs users and non-users).

Literature screening and data extraction procedures

A rigorous dual-blind review process was implemented in this study, with two independent researchers (YML and XYD) systematically evaluating all retrieved literature against predefined inclusion/exclusion criteria, and any disagreements resolved through discussion with a third reviewer (HZW). Following initial independent assessments, the researchers conducted cross-verification to ensure consistency in study selection. For all eligible studies, comprehensive data extraction was performed, encompassing: (1) study design characteristics; (2) bibliographic information including first author and publication year; (3) baseline demographic data (indicationsample, size, gender distribution, mean age); (4) predefined subgroup analyses (particularly elderly populations); (5) specific antidepressant exposure details (including drug classification); and (6) standardized laboratory diagnostic criteria for hyponatremia determination. This methodical approach was designed to maintain methodological consistency and minimize potential biases throughout the data collection process.

Quality assessment of included studies

For all studies meeting the inclusion criteria, methodological quality was rigorously evaluated using the Newcastle-Ottawa Scale (NOS) [16]. This validated assessment tool examines three critical dimensions: (1) selection of study populations; (2) comparability between groups; and (3) ascertainment of exposure factors. The NOS comprises eight specific quality assessment items that collectively provide a comprehensive evaluation framework for observational studies (Supplementary Tables 3, 4).

To further evaluate the strength of evidence, we employed the GRADE (Grading of Recommendations Assessment, Development and Evaluation) framework to assess the certainty of evidence for our primary outcomes systematically. This system systematically assesses evidence certainty through five key domains: (1) risk of bias (study limitations); (2) inconsistency (heterogeneity); (3) indirectness (population, intervention, comparator, outcome); (4) imprecision (confidence intervals), and (5) publication bias. For observational studies, the evaluation also considers potential upgrading factors including large effect sizes, dose-response gradients, and plausible residual confounding that would reduce the observed effect. The framework provides a transparent approach to rate evidence certainty as high, moderate, low, or very low.

Statistical analysis

The meta-analysis used Comprehensive Meta-Analysis software (version 3.3.070), with statistical significance set at P < 0.05 (two-tailed). Effect sizes were calculated according to study design types: RCTs reported RRs with 95% confidence intervals (CIs); cohort studies presented event rates with 95% CIs; and case-control studies reported ORs with 95% CIs. Heterogeneity was assessed using Cochran’s Q test and quantified by the I² statistic. Given the anticipated between-study heterogeneity arising from variations in drug dosage, hyponatremia diagnostic criteria, and study populations, all primary analyses were prespecified to employ random-effects models a priori. For scenarios where no substantial heterogeneity was detected (I² ≤ 50%), fixed-effects models were alternatively applied as a sensitivity analysis to ensure methodological rigor. To further evaluate potential publication bias, funnel plots were generated and complemented with Egger’s linear regression model, providing a multidimensional assessment of bias risk.

Results

Study selection

Our systematic literature search identified 825 potentially relevant articles from three databases: Cochrane Library (n = 32), PubMed (n = 362), and Web of Science (n = 431). The selection process proceeded as follows: First, we removed 336 duplicate records through cross-database deduplication. Initial screening of titles and abstracts excluded 191 ineligible studies from the remaining 489 articles. Full-text review of 298 potentially relevant articles subsequently eliminated non-research publications, including reviews, conference abstracts, and case reports. To ensure comprehensive coverage, we performed manual reference screening of included studies, which identified 2 additional eligible articles. Ultimately, 38 observational studies (comprising 30 cohort studies and 8 case-control studies) involving 3,997,088 participants met the inclusion criteria [17–54]. Detailed characteristics of included studies are presented in Table 1, while Fig. 1 illustrates the complete study selection flowchart.

Table 1.

Detailed information on included studies

Type of study	Author	Year	Indication	Sample size(male)	Mean age	Geriatric cohort	Diagnostic criteria for hyponatremia	Drug		NOS
								SSRI	SNRI
Cohort studies	Spigset et al.	1997	Healthy subjects	8(8)	29	No	< 135mmol/L	√	/	5
	Bouman et al.	1998	Psychiatry	32(17)	76.1	Yes	< 135mmol/L	√	/	2
	Strachan et al.	1998	Psychiatry	55(22)	74.3	Yes	< 135mmol/L	√	/	3
	Spigset et al.	1999	Not given	1,202(393)	49	No	Spontaneous report	√	/	3
	Wilkinson et al.	1999	Psychiatry	845	Not given	Yes	< 130mmol/L	√	/	3
	Fabian et al.	2003	Depression	15(8)	75.7	Yes	< 135mmol/L	√	/	5
	Degner et al.	2004	Psychiatry	53,042	Not given	No	< 135mmol/L	√	√	6
	Fabian et al.	2004	Depression	75(22)	75.2	Yes	< 135mmol/L	√	/	7
	Wee et al.	2004	Psychiatry	103(37)	78.4	Yes	< 135mmol/L	√	/	4
	Roxanas et al.	2007	Depression	58	Not given	Yes	< 130mmol/L	/	√	5
	Shakibaei et al.	2010	Depression	126	Not given	No	< 135mmol/L	√	/	2
	Jung et al.	2011	Depression	240(79)	51.5	No	< 135mmol/L	√	√	4
	Coupland et al.	2011	Depression	60,746(20230)	75	Yes	ICD-code	√	√	8
	Letmaier et al.	2012	Psychiatry	263,864(116,796)	48.8	No	< 130mmol/L	√	√	7
	Montastruc et al.	2012	Not given	177(56)	61	No	Spontaneous report	√	√	3
	Giorlando et al.	2013	Depression	131(54)	79.8	Yes	< 135mmol/L	√	√	4
	Lange-Asschenfeldt et al.	2013	Psychiatry	2,606	Not given	No	< 135mmol/L	√	√	3
	Shetty et al.	2015	Psychiatry	74(32)	Not given	No	< 135mmol/L	√	√	7
	Noohi et al.	2016	Psychiatry	83(42)	67.4	No	< 135mmol/L	√	?	4
	Leth-Moller et al.	2016	Not given	34,758	Not given	No	< 135mmol/L	√	√	7
	Gandhi et al.	2017	Anxiety	138,246(44,943)	76	Yes	ICD-code	√	/	7
	Shepshelovich et al.	2017	Not given	65(25)	74.4	No	< 135mmol/L	√	√	7
	Albrecht et al.	2018	Traumatic brain injury	15,733(5,035)	79.7	Yes	ICD-code	√	√	6
	Lundstrom et al.	2020	Stroke	750(463)	70.6	No	< 130mmol/L	√	/	6
	Seifert et al.	2021	Psychiatry	462,661(204,071)	Not given	No	< 130mmol/L	√	√	3
	Shysh et al.	2021	Not given	19,679(6,837)	55.48	No	< 135mmol/L	√	/	4
	Sarkar et al.	2021	Depression	200(100)	47.6	No	< 135mmol/L	√	/	4
	Tomar et al.	2021	Psychiatry	150(86)	Not given	No	< 135mmol/L	√	/	5
	Nambiar et al.	2022	Post-stroke depression	60(40)	62.13	No	< 135mmol/L	√	√	7
	Nagashima et al.	2022	Depression	8,473	Not given	Yes	ICD-code	√	√	7
Case- Control studies	Siegler et al.	1995	Psychiatry	64/192(13/66)	62.3/43.4	No	< 130mmol/L	√	/	7
	Movig et al.	2002	Psychiatry	29/78(10/32)	68/57	No	ICD-code	√	/	7
	Kirby et al.	2002	Psychiatry	74/125(29/41)	74.1/74.2	Yes	< 135mmol/L	√	√	5
	Movig et al.	2002	Not given	203/608(54/162)	71	No	ICD-code	√	√	7
	Farmand et al.	2018	Not given	14,359/57,382 (4,023/16,083)	74	No	ICD-code	√	√	6
	Mazhar et al.	2019	Not given	13,307/2,771,223 (5,674/1,229,443)	61.47/49.1	No	Spontaneous report	√	√	5
	Jun et al.	2021	Not given	913/18,260 (437/8,740)	Not given	Yes	ICD-code	√	√	4
	Mannheimer et al.	2021	Depression/Anxiety	11,213/44,801 (3,139/12,547)	76	No	ICD-code	√	√	7

Not given: Not mentioned in the study; √: Such drugs were included in the study; /: Such drugs were not included in the study;?: It was not possible to determine whether such drugs were included in the study; Kirby et al. and Leth-Møller et al. were both a case-control study and a cohort study

Fig. 1.

Literature screening process

Hyponatremia event rate and OR overall antidepressants

This meta-analysis systematically evaluated the association between antidepressant use and hyponatremia risk, demonstrating a statistically significant correlation. Among the 30 included studies, the pooled event rate of hyponatremia was 6.03% (P < 0.001, Fig. 2A), while the event rate of clinically relevant hyponatremia was 1.26% (n = 11, P < 0.001). Sensitivity analysis showed remarkable stability, with recalculated event rates ranging from 5.46 to 7.02% when excluding individual studies, confirming no single study disproportionately influenced these findings (Supplementary Table 5). Further analysis revealed that antidepressant use significantly increased hyponatremia risk. Antidepressant exposure was associated with elevated risks for both any hyponatremia (n = 8, OR = 2.164, P < 0.001, Fig. 2B) and clinically relevant hyponatremia (n = 7, OR = 2.102, P < 0.001). All analyses showed statistical significance despite substantial between-study heterogeneity (I² > 85%), with effect sizes remaining within a consistent range that reinforces result reliability. We acknowledge the observed variability, which is expected given the diverse hyponatremia severity and patient age, and have further explored potential sources through meta-regression. Meta-regression indicated that 43% of the total variance across studies could be attributed to hyponatremia severity (any hyponatremia vs. clinically relevant hyponatremia, t = 4.57, P = 0.0001) and patient age (non-elderly vs. elderly, t = 2.12, P = 0.0430) (Supplementary Fig. 1).

Fig. 2.

(A) Forest plot event rates overall antidepressants (any hyponatremia). (B) Forest plot odds ratios overall antidepressants (any hyponatremia). * The Spigset 1997 study, which reported zero events (0/8), was excluded from the forest plot as the calculation of event rates for zero-event studies is methodologically unsound

Hyponatremia event rate and OR class-specific antidepressants

Both SSRIs and SNRIs demonstrated significant associations with hyponatremia risk (Table 2). The pooled event rate of hyponatremia was slightly lower for SSRIs (5.98%, P < 0.001; Fig. 3A) compared to SNRIs (6.13%, P < 0.001; Fig. 3B), with both rates reaching statistical significance. Regarding risk elevation, SSRIs (OR = 2.158, P < 0.001; Fig. 4A) and SNRIs (OR = 2.270, P < 0.001; Fig. 4B) showed comparable increases in hyponatremia risk. Importantly, the risk stratification revealed notable variations among individual agents, with fluoxetine among SSRIs demonstrating the highest risk profile (event rate: 6.51%, OR = 3.579; P < 0.01), while venlafaxine emerged as the highest-risk SNRIs (event rate: 5.66%, OR = 2.683; P < 0.001) (Table 2). For clinically relevant hyponatremia, the event rate was lower with SSRIs (0.82%, P < 0.001) versus SNRIs (1.33%, P < 0.001), with SNRIs exhibiting marginally higher risk (OR = 2.227 vs. 2.151, P < 0.001). Notably, substantial heterogeneity was observed across studies (I² > 80%), reflecting significant variations in hyponatremia definitions, baseline population characteristics, and methodological approaches among the included studies.

Table 2.

Pooled event rate and OR for antidepressant

Antidepressant	Number of studies(n)	Event rate or OR (95% CI)	P-value	I2 (%)
Event rate for any hyponatremia
Antidepressants (overall)	30	0.0603(0.0372–0.0965)	< 0.001	99.76
SSRIs	26	0.0598(0.0360–0.0976)	< 0.001	99.74
citalopram	5	0.0099(0.0033–0.0293)	< 0.001	99.91
escitalopram	8	0.0253(0.0050–0.1176)	< 0.001	99.50
fluoxetine	10	0.0651(0.0279–0.1443)	< 0.001	94.24
fluvoxamine	4	0.0508(0.0387–0.0663)	< 0.001	0.000
paroxetine	13	0.0412(0.0164–0.0995)	< 0.001	98.85
sertraline	9	0.0252(0.0081–0.0755)	< 0.001	99.05
SNRIs	13	0.0613(0.0253–0.1410)	< 0.001	99.40
venlafaxine	11	0.0566(0.0118–0.2310)	< 0.001	99.31
desvenlafaxine	1	0.8333(0.1936–0.9905)	0.2989	0.000
duloxetine	5	0.0276(0.0071–0.1008)	< 0.001	99.10
milnacipran	3	0.0345(0.0020–0.3846)	0.0225	92.96
Event rate for clinically relevant hyponatremia
Antidepressants (overall)	11	0.0126(0.0039–0.0398)	< 0.001	99.84
SSRIs	9	0.0082(0.0024–0.0278)	< 0.001	99.81
SNRIs	6	0.0133(0.0022–0.0766)	< 0.001	99.68
OR for any hyponatremia
Antidepressants (overall)	8	2.164(1.882–2.488)	< 0.001	88.56
SSRIs	8	2.158(1.890–2.463)	< 0.001	85.55
citalopram	2	2.368(1.600-3.506)	< 0.001	97.41
escitalopram	2	2.315(2.085–2.571)	< 0.001	0.000
fluoxetine	4	3.579(1.380–9.281)	0.009	65.15
fluvoxamine	3	2.133(1.342–3.390)	0.001	0.000
paroxetine	4	2.512(2.216–2.848)	< 0.001	0.000
sertraline	4	2.346(2.178–2.527)	< 0.001	0.000
SNRIs	5	2.270(1.603–3.216)	< 0.001	84.05
venlafaxine	4	2.683(1.738–4.141)	< 0.001	85.38
desvenlafaxine	1	0.870(0.608–1.246)	0.447	0.000
duloxetine	1	2.140(1.905–2.405)	< 0.001	0.000
OR for clinically relevant hyponatremia
Antidepressants (overall)	7	2.102(1.833–2.410)	< 0.001	89.22
SSRIs	7	2.151(1.877–2.464)	< 0.001	89.22
SNRIs	4	2.227(1.496–2.747)	< 0.001	82.20

Fig. 3.

(A) Forest plot event rates SSRI (any hyponatremia). (B) Forest plot event rates SNRI (any hyponatremia). * The Spigset 1997 study, which reported zero events (0/8), was excluded from the forest plot as the calculation of event rates for zero-event studies is methodologically unsound

Fig. 4.

(A) Forest plot odds ratios SSRI (any hyponatremia). (B) Forest plot odds ratios SNRI (any hyponatremia)

This study explored potential sources of heterogeneity through meta-regression and subgroup analyses. Meta-regression analyses demonstrated distinct patterns in explaining between-study heterogeneity across different antidepressant classes. For SSRIs, 51% of total variance was attributable to hyponatremia severity classification (any vs. clinically relevant: t = 7.21, P < 0.0001) and patient age (non-elderly vs. elderly: t = 2.52, P = 0.019), indicating moderate heterogeneity determinants. Notably, SNRIs demonstrated stronger covariate effects, with 66% of variance explained by these same factors - severity (t = 4.58, P = 0.001) and age (t = 3.23, P = 0.009) - suggesting more pronounced outcome variability dependent on these parameters (Supplementary Fig. 2, 3). The greater explanatory power for SNRIs (66% vs. 51%), particularly in age-related effects (t = 3.23 vs. 2.52), may reflect class-specific pharmacokinetic profiles influencing sodium regulation. The subgroup analysis revealed significant differences in the impact of hyponatremia severity on event rates for both SSRIs and SNRIs: SSRIs (Q = 22.71, df = 1, P < 0.001) and SNRIs (Q = 8.86, df = 1, P = 0.003). When only clinically relevant hyponatremia was considered for SSRIs and SNRIs, the event rates decreased significantly. In the age-based subgroup analysis, no significant difference in hyponatremia event rate was observed between non-elderly and elderly populations for either SSRIs (Q = 1.34, df = 1, P = 0.246) or SNRIs (Q = 0.83, df = 1, P = 0.361). However, the event rate was higher in the elderly population.

Head-to-head comparison of hyponatremia risk between SSRIs and SNRIs

Using SSRIs as the reference category, this study evaluated the association between SNRIs and hyponatremia risk (Fig. 5). The results demonstrated that SNRIs significantly increased the risk of hyponatremia compared to SSRIs (n = 12, OR = 1.271, P < 0.001), with low between-study heterogeneity (I² = 35.30%). Similarly, SNRIs showed a significant elevation in clinically relevant hyponatremia risk (n = 5, OR = 1.291, P < 0.001). In elderly subgroup analyses, SNRIs exhibited a non-significant trend toward increased hyponatremia risk (n = 5, OR = 1.258, P = 0.070), with a comparable pattern observed for clinically relevant hyponatremia (n = 3, OR = 1.236, P = 0.098). The analyses demonstrated substantial heterogeneity (I² > 60%), potentially attributable to variations in baseline characteristics, concomitant medications, and diagnostic criteria across studies.

Fig. 5.

Forest plot head-to-head comparison (any hyponatremia)

Risk of bias and evidence certainty assessment

This study evaluated potential bias in all included event rates studies using multiple approaches. Funnel plot analysis identified five studies with missing data, but overall symmetry was well preserved, and Egger’s test showed no significant publication bias (t = 1.68, df = 28, P = 0.104). Study quality was assessed using the NOS, with results indicating generally high methodological quality across included studies (see Table 1 for details). Meta-regression analysis revealed no significant influence of study quality on the outcomes.

The GRADE assessment yielded “low” certainty evidence for the overall association between SSRIs/SNRIs and hyponatremia (OR = 2.164), reflecting: (1) moderate risk of bias inherent in observational study designs; (2) extreme heterogeneity (I² > 85%) stemming from variability in hyponatremia definitions (130 mmol/L vs. 135 mmol/L thresholds) and study methodologies, and (3) indirectness due to predominantly elderly/psychiatric populations (Supplementary Table 7). While the large effect size (OR > 2.0) partially offset these limitations, similar constraints were observed in class-specific analyses (SSRIs: OR = 2.158, I² = 85.55%; SNRIs: OR = 2.270, I² = 84.05%), with SNRIs showing additional imprecision due to wider confidence intervals. At the individual drug level, evidence certainty ranged from “low” for higher-risk antidepressants (fluoxetine: OR = 3.579; venlafaxine: OR = 2.683) to “very low” for lower-risk antidepressants (sertraline/duloxetine).

In summary, although our bias assessment demonstrated robust methodology and minimal publication bias, the GRADE evaluation revealed important limitations in the overall evidence base, which should be considered when interpreting these findings.

Discussion

This meta-analysis systematically evaluated the association between antidepressant use and hyponatremia risk, demonstrating a statistically significant correlation. Both SSRIs and SNRIs were found to significantly increase the incidence and risk of hyponatremia, with SNRIs exhibiting marginally higher risk for clinically relevant hyponatremia compared to SSRIs. These findings suggest that different drug classes may influence hyponatremia severity through distinct pathophysiological mechanisms. The risk stratification among specific agents revealed clinically meaningful variations: fluoxetine and venlafaxine showed the highest risk within their respective classes, while sertraline, citalopram, and duloxetine demonstrated relatively lower risk profiles. This stratification guides clinical medication selection. Heterogeneity analysis underscores the importance of considering potential confounding effects from study design and population characteristics when interpreting evidence on antidepressant-associated hyponatremia. Particular attention should be paid to elderly populations, who may face greater clinical complexity due to comorbidities and polypharmacy.

Our systematic evaluation of confounding factors demonstrated that most studies adequately controlled for key demographic and clinical variables, including age, sex, weight, comorbidities (particularly chronic kidney disease and congestive heart failure), and concomitant medications (notably thiazide diuretics) (Supplementary Table 6). While these factors are well-established risk modifiers for antidepressant-induced hyponatremia rather than alternative explanations for the observed association, their relative contributions remain debated. Notably, two studies highlighted that psychogenic polydipsia might alternatively represent a potential mechanism in psychiatric populations, where combined renal and neurohypophyseal dysfunction impairs free water excretion [33, 47]. However, the predominant attribution across all included studies remained antidepressant-induced SIADH, with polydipsia considered a secondary contributor in specific subpopulations. Importantly, regarding the depression and anxiety pathophysiology, Gilboa et al. (2019) [55] demonstrated that although anxiety itself is not an independent risk factor for hyponatremia, patients with anxiety disorders exhibit a specific increase in hyponatremia risk following SSRIs treatment initiation. This temporal relationship strongly supports a pharmacologically mediated mechanism rather than an underlying psychiatric etiology [55]. These findings collectively underscore the importance of careful differential diagnosis in clinical practice, particularly in distinguishing between the effects of medication and disease-related mechanisms.

Current evidence suggests that the SIADH plays a central role in the pathogenesis of antidepressant-induced hyponatremia [7]. Both SSRIs and SNRIs may induce excessive ADH release by blocking 5-HT reuptake and subsequently activating hypothalamic 5-HT_2C receptors, thereby enhancing renal tubular water reabsorption and ultimately leading to dilutional hyponatremia [43]. SNRIs exhibit a unique dual mechanism, as they also inhibit the reuptake of NET. Beyond 5-HT system modulation, this action disrupts norepinephrine (NE)-mediated renal hemodynamic regulation, further exacerbating abnormal ADH secretion—a plausible explanation for the higher risk of clinically relevant hyponatremia observed with SNRIs in our study [12]. Drug-specific risk variations likely stem from distinct pharmacokinetic and receptor selectivity profiles. For instance, fluoxetine, with its long half-life (4–6 days) and active metabolites, may cause sustained stimulation of the 5-HT system. In contrast, citalopram’s high affinity for 5-HT_2C receptors might suppress ADH release through negative feedback mechanisms. This differential receptor activity could reduce water retention and consequently decrease the likelihood of hyponatremia development.

This study analyzed 38 studies to compare the risk of hyponatremia between SSRIs and SNRIs, providing quantitative evidence for risk stratification within each drug class. Through meta-regression and subgroup analyses, we further explored sources of heterogeneity, particularly clarifying the impact of hyponatremia severity classification and age subgroups on effect size estimates. While the GRADE evaluation reveals “low/very low” certainty evidence due to methodological limitations inherent in observational designs and substantial heterogeneity in outcome definitions, the consistent risk elevation (ORs > 2.0 across all analyses) carries important clinical implications that warrant careful consideration. The findings highlight the prevalence and potential severity of antidepressant-induced hyponatremia, suggesting that clinicians should maintain heightened vigilance in clinical practice. This includes preferential selection of lower-risk agents like sertraline or duloxetine for vulnerable populations (particularly elderly patients and those with additional risk factors such as low body weight or concomitant diuretic use), coupled with enhanced sodium monitoring during the critical initial treatment period, especially for patients prescribed higher-risk medications such as venlafaxine or fluoxetine. However, the current evidence base precludes definitive conclusions regarding causal relationships or precise risk quantification, underscoring the urgent need for future research employing standardized hyponatremia criteria to enhance the reliability of evidence for clinical decision-making, along with rigorously controlled prospective cohorts to address residual confounding and individual patient data meta-analyses to elucidate potential dose-response relationships. Clinicians should therefore interpret these findings within the context of their patients’ risk profiles, recognizing that the current evidence supports cautious practice adjustment rather than definitive clinical guidelines.

This study has several limitations. First, significant heterogeneity existed among the included studies regarding the definition of hyponatremia (e.g., serum sodium thresholds), demographic characteristics (e.g., age, sex), and control of confounding factors (e.g., concomitant medications and underlying diseases). Although a random-effects model was applied for adjustment, residual confounding may still undermine the robustness of the causal association between antidepressant use and hyponatremia. Second, some subgroups (e.g., desvenlafaxine) had limited sample sizes, leading to imprecise effect estimates. Third, our systematic search was restricted to three major databases (Cochrane Library, PubMed, and Web of Science), potentially limiting the comprehensiveness of evidence identification. Fourth, most studies did not provide dose-response relationship data, restricting the exploration of risk thresholds. Future research should incorporate specific methodologies (e.g., pharmacogenomic approaches) to elucidate the mechanisms underlying antidepressant-induced hyponatremia. Concurrently, comprehensive comparisons of individual agents across drug classes should be performed to delineate their distinct risk profiles, facilitating the development of tailored prevention strategies for high-risk populations.

Conclusion

This study confirms a significant association between antidepressant use and hyponatremia risk, with both SSRIs and SNRIs demonstrating clear risk signals. Notably, SNRIs exhibit a more pronounced risk of clinically significant hyponatremia compared to SSRIs. Importantly, substantial heterogeneity exists within each drug class: fluoxetine (among SSRIs) and venlafaxine (among SNRIs) present particularly elevated risks, whereas sertraline, citalopram, and duloxetine demonstrate relatively lower risks. These findings provide critical evidence for risk stratification and personalized treatment in clinical practice. They underscore the need to optimize therapeutic regimens by considering patient age, comorbidities, and drug-specific risks, with particular emphasis on enhanced serum sodium monitoring for elderly patients and those on polypharmacy.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

5-HT

Serotonin

ACEIs

Angiotensin-Converting Enzyme Inhibitors

ADH

Antidiuretic Hormone

CIs

Confidence Intervals

NE

Norepinephrine

NET

Norepinephrine Transporter

NOS

Newcastle-Ottawa Scale

OR

Odds Ratio

RR

Risk Ratio

SIADH

Syndrome of Inappropriate Antidiuretic Hormone Secretion

SNRIs

Serotonin-Norepinephrine Reuptake Inhibitors

SSRIs

Selective Serotonin Reuptake Inhibitors

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Yumeng Li, Xiaoyu Du, and Huizhen Wu. The first draft of the manuscript was written by Yumeng Li. All authors critically reviewed the content and approved the final version before submission.

Funding

The authors declare that no financial support was received for the research, authorship, and/or publication of this article.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors approved the final manuscript and the submission to this journal.

Competing interests

The authors declare no competing interests.