Skip to main content

Trial Summaries & Critiques

HIT

Image: Shutterstock / Sharif-Pavlov

Salt grains on a table, shaped into a heart containing an ECG waveform

Publication

  • Title: A Randomized Trial of Targeted Hyponatremia Correction in Hospitalized Patients
  • Acronym: HIT
  • Year: 2026
  • Journal published in: NEJM Evidence
  • Citation: Refardt J, Potasso L, Pelouto A, Trappe M, Gregoriano C, Koster M, et al. A randomized trial of targeted hyponatremia correction in hospitalized patients. NEJM Evid. 2026;5(3):EVIDoa2500086.

Context & Rationale

  • Background
    • Chronic hyponatraemia had already been associated with falls, neurocognitive dysfunction, fractures, rehospitalisation, and mortality, but the central unresolved issue was causality: sodium correction might improve prognosis, or low sodium might simply mark more severe underlying illness.1
    • Before HIT, prospective interventional evidence was small and narrow. The SALT programme showed that tolvaptan improves sodium, but it did not establish a 30-day mortality benefit and did not address the broad, heterogeneous inpatient population with chronic hypotonic hyponatraemia.2
    • Contemporary treatment recommendations still rested largely on older expert frameworks and limited comparative trial evidence, leaving genuine uncertainty about how intensively chronic inpatient hyponatraemia should be pursued.3
  • Research Question/Hypothesis
    • Does targeted, aetiology-driven correction of chronic hypotonic hyponatraemia below 130 mmol/L during hospitalisation reduce the 30-day risk of death or rehospitalisation compared with routine care?4
    • The superiority hypothesis was that systematic diagnostic classification, daily expert review, and treatment escalation would increase normonatraemia and thereby improve short-term patient-centred outcomes.4
  • Why This Matters
    • HIT was the first large randomised trial designed to move the field beyond associative epidemiology and test whether improving sodium actually changes hard clinical outcomes.
    • A positive result would have justified resource-intensive inpatient sodium normalisation; a neutral result would suggest that short-term prognosis is driven more by underlying disease, insufficient treatment separation, or both.

Design & Methods

  • Research Question: In hospitalised adults with chronic hypotonic hyponatraemia, does targeted, trial-team-led sodium correction reduce 30-day death or rehospitalisation compared with routine care?
  • Study Type: Pragmatic, randomised, controlled, parallel-group, international, multicentre superiority trial with blinded 30-day outcome assessment; nine European centres, including university and regional hospitals; recruitment from August 2018 to April 2024; patients could be enrolled from any hospital ward.
  • Population:
    • Inclusion criteria: Adults aged 18 years or older; hospitalised; hypotonic hyponatraemia with plasma sodium <130 mmol/L and plasma osmolality ≤300 mOsm/kg; present on admission or developing during hospitalisation; judged chronic on the basis of time course and absence of severe symptoms.
    • Exclusion criteria: Severe symptomatic hyponatraemia requiring immediate correction with 3% saline unless symptoms improved and hyponatraemia persisted beyond 48 hours; end-of-life care; kidney disease requiring renal replacement therapy; acute liver failure; Wernicke encephalopathy; hepatic encephalopathy within 2 months; hepatorenal syndrome; pregnancy or breastfeeding; strict haematology isolation.
    • Randomisation: 1:1 allocation, stratified by site, with randomly permuted block sizes.
  • Intervention:
    • Trial-team-led, aetiology-based diagnostic and treatment algorithm, with daily review of treatment response and adherence.
    • Predominant cause of hyponatraemia determined from history, examination, extracellular volume assessment, and laboratory tests; treatment escalated in a stepwise fashion according to the presumed aetiology.
    • Example: syndrome of inappropriate antidiuresis was managed with fluid restriction first, then oral urea, then tolvaptan if sodium response remained inadequate.
    • Treatment was intensified if plasma sodium rose by <2 mmol/L in 24 hours, maintained if 2–12 mmol/L, and stopped with relowering therapy (hypotonic fluids with or without desmopressin) if correction exceeded >12 mmol/L in 24 hours or >18 mmol/L in 48 hours.
    • Intervention continued until sodium 135–145 mmol/L was achieved and maintained during the index admission, until discharge, or until day 30 if still hospitalised.
  • Comparison:
    • Routine care for hyponatraemia at the sole discretion of the attending physicians.
    • The trial team did not influence diagnostic work-up or treatment in the control arm; treatment choices and sodium trajectory were recorded from the chart after discharge.
    • Final hyponatraemia aetiology was adjudicated by the trial team after discharge using all available clinical, laboratory, and treatment-response data.
  • Blinding: Open-label treatment allocation. Blinding of bedside management was not feasible. Outcome assessors performing the 30-day follow-up were not involved in active trial care and were blinded to allocation, which reduced detection bias for a predominantly objective primary end point.
  • Statistics: A total of 2050 evaluable patients were required to detect a 5-percentage-point absolute reduction in 30-day death or rehospitalisation (23% to 18%) with 80% power at a two-sided 5% significance level; 2278 patients were targeted allowing for 10% dropout. Primary analysis used a full analysis set (modified intention-to-treat/complete-case) including randomly assigned patients who received allocated treatment for at least 24 hours or died within 24 hours; supportive per-protocol, best-case, worst-case, and safety analyses were prespecified. Secondary outcomes were exploratory and not adjusted for multiplicity.
  • Follow-Up Period: Primary end point at 30 days after inclusion; blinded 30-day follow-up by telephone and record review. One-year follow-up was planned for secondary outcomes but was not reported in the index paper.

Key Results

This trial was not stopped early.

Outcome Targeted correction Routine care Effect p value / 95% CI Notes
Death or rehospitalisation within 30 days 218/1065 (20.5%) 234/1073 (21.8%) Absolute difference −1.3 percentage points 95% CI −4.9 to 2.2; P=0.45 Primary outcome available for 1065/1079 and 1073/1094 patients.
Death within 30 days 86/1079 (8.0%) 88/1094 (8.0%) Absolute difference −0.1 percentage points 95% CI −2.4 to 2.3 In-hospital deaths 4.9% in both groups; post-discharge deaths 3.1% in both groups.
Rehospitalisation within 30 days 141/1065 (13.2%) 151/1073 (14.1%) Absolute difference −0.8 percentage points 95% CI −3.8 to 2.2 Thirty-five patients had missing 30-day rehospitalisation data.
Reached normonatraemia during treatment 641/1061 (60.4%) 492/1065 (46.2%) Cause-specific HR 1.54 95% CI 1.37 to 1.74 Absolute difference 14.3 percentage points; 95% CI 10.0 to 18.6.
Normonatraemia at discharge 572/1023 (55.9%) 382/1038 (36.8%) Absolute difference 19.0 percentage points 95% CI 14.7 to 23.3 Biochemical separation was greater at discharge than at 30 days.
Maximum absolute sodium change during treatment 10.0 ± 5.6 mmol/L 8.7 ± 5.6 mmol/L Not reported Not reported Median treatment duration was 6 days in the intervention group versus 3 days in control.
Hyponatraemia persistence or recurrence within 30 days 250/600 (41.7%) 246/602 (40.9%) Not reported Not reported Suggests loss of between-group sodium separation after discharge.
Overcorrection 25/1098 (2.3%) 16/1114 (1.4%) Not reported P=0.14 Safety population. No osmotic demyelination syndrome observed; adverse events due to overcorrection were 1.2% vs 0.8%.
Acute symptoms from worsening hyponatraemia despite treatment 5/1098 (0.5%) 1/1114 (0.1%) Not reported P=0.10 All cases resolved after treatment intensification and sodium rise.
  • Biochemical success did not translate into clinical benefit: normonatraemia during treatment was 60.4% versus 46.2%, yet the primary outcome was 20.5% versus 21.8%.
  • Treatment separation was clear process-wise but modest biologically: maximum sodium change was 10.0 ± 5.6 versus 8.7 ± 5.6 mmol/L, and by 30 days hyponatraemia persistence/recurrence was similar at 41.7% versus 40.9%.
  • Exploratory subgroup analysis suggested lower risk of the primary outcome in patients aged 70 years or older (OR 0.76; 95% CI 0.57 to 1.00), while aetiology, severity, and sex showed no signal. Post hoc, reaching normonatraemia at discharge was associated with lower odds of death/rehospitalisation irrespective of treatment assignment (OR 0.74; 95% CI 0.60 to 0.91).

Internal Validity

  • Randomisation and Allocation: Allocation was 1:1, site-stratified, and block-randomised, making selection bias unlikely.
  • Drop out or exclusions: 2212 patients entered the safety analysis set; 39 withdrew consent within 24 hours and were excluded from the full analysis set, leaving 2173 patients. Primary outcome completeness was high: 1065/1079 (98.7%) in the intervention group and 1073/1094 (98.1%) in control. The per-protocol analysis set excluded 124 patients.
  • Performance/Detection Bias: The trial was open-label, so bedside behaviour, diagnostic intensity, and adjunctive management could differ. Detection bias for the primary outcome was limited because death and rehospitalisation are objective, and 30-day assessors were blinded.
  • Protocol Adherence: Per-protocol analysis included 979 intervention patients and 1070 control patients. In the intervention arm, 71 patients received assigned treatment less than 75% of the time and 8 were changed to best supportive care.
  • Baseline Characteristics: Groups were well matched overall: median age 73 years, median sodium 127 mmol/L in both groups, and median Charlson Comorbidity Index 4. Severe hyponatraemia (<120 mmol/L) was slightly more common in the intervention group (7.1% vs 5.2%), which would, if anything, bias against finding benefit.
  • Heterogeneity: Considerable clinical heterogeneity was built into the design: euvolaemic hyponatraemia 55.3% versus 52.3%, hypovolaemic 30.5% versus 31.0%, hypervolaemic 14.2% versus 16.6%, and hospital-acquired hyponatraemia in 24.8% overall. This strengthens representativeness but may dilute any effect of a single inpatient correction strategy on hard outcomes.
  • Timing: Screening was frequent and enrolment occurred once hypotonic hyponatraemia was confirmed, but exposure was bounded by discharge. Median intervention duration was 6 days versus 3 days, and the trial team could not delay discharge to complete correction.
  • Dose: The algorithm allowed escalation to urea or tolvaptan, but achieved only a modest biochemical gap. This raises the possibility that the intervention dose or duration was insufficient for a hard-outcome effect.
  • Separation of the Variable of Interest: Process separation was substantial: no treatment 2.5% versus 35%, fluid restriction 40% versus 26%, oral urea 15% versus 2.3%, vaptans 4.1% versus 1.5%, and combined therapies 56% versus 31%. Biochemical separation was smaller: maximum sodium change 10.0 ± 5.6 versus 8.7 ± 5.6 mmol/L; normonatraemia during treatment 60.4% versus 46.2%; normonatraemia at discharge 55.9% versus 36.8%.
  • Key Delivery Aspects: The intervention targeted the sodium disorder rather than the broader disease trajectory. In a multimorbid cohort with median age 73 years and median Charlson score 4, short-term death and rehospitalisation may have been dominated by the underlying illness rather than sodium alone.
  • Crossover: No formal crossover was reported, but contamination of the control arm remained plausible because clinicians knew patients were in a hyponatraemia trial. Investigators did not detect a centre-level learning effect for achieving normonatraemia.
  • Outcome Assessment: The primary outcome was clinically important, clearly defined, and largely objective. Secondary functional and quality-of-life outcomes were inherently more vulnerable to missingness and were not numerically emphasised in the main report.
  • Statistical Rigor: The full analysis set exceeded the prespecified number of evaluable patients required by the power calculation, and per-protocol plus best/worst-case analyses were concordant. However, the primary analysis was modified rather than strict intention-to-treat, and secondary outcomes were exploratory without multiplicity adjustment.

Conclusion on Internal Validity: Overall, internal validity is moderate. Randomisation, follow-up completeness, and objective blinded outcome assessment were strong, but the open-label design, modified intention-to-treat population, heterogeneity, and only modest biochemical separation limit confidence that a true clinical effect—if small—would necessarily have been detected.

External Validity

  • Population Representativeness: The cohort was older (median age 73 years), multimorbid, and recruited across university and regional European hospitals, which closely reflects real-world general inpatient hyponatraemia practice.
  • Case Mix: The study included euvolaemic, hypovolaemic, and hypervolaemic hyponatraemia, and both community-acquired and hospital-acquired cases, which broadens applicability to general medicine rather than a single specialist niche.
  • Important Exclusions: Severe symptomatic hyponatraemia requiring immediate hypertonic saline was excluded, and only 6.2% of patients had sodium <120 mmol/L. The results therefore should not be extrapolated to severe symptomatic, profound, or ICU-dominant hyponatraemia.
  • Applicability Across Systems: The intervention required daily specialist-style assessment, repeated sodium measurement, and access to therapies such as oral urea and tolvaptan. Transferability is therefore strongest in systems with ready biochemical monitoring and endocrine/nephrology support.
  • Duration of Applicability: The findings apply specifically to intensified inpatient correction during the index admission. They do not answer whether sustained outpatient correction, longer treatment duration, or highly selected subgroups could benefit.

Conclusion on External Validity: External validity is moderate to good for older, general-hospital patients with predominantly moderate chronic hypotonic hyponatraemia. It is limited for severe symptomatic cases, profound hyponatraemia, outpatient management, and settings unable to reproduce the trial’s monitoring intensity.

Strengths & Limitations

  • Strengths:
    • Largest randomised trial to date testing whether inpatient correction of chronic hyponatraemia improves hard outcomes.
    • Pragmatic design across nine European centres, including both university and regional hospitals.
    • Broad inclusion across hyponatraemia aetiologies, improving real-world relevance.
    • High primary outcome completeness and blinded 30-day outcome assessment.
    • Clear process separation between targeted correction and routine care.
  • Limitations:
    • Open-label intervention with unavoidable risk of performance bias and contamination.
    • Mostly moderate hyponatraemia; only a small minority had sodium <120 mmol/L.
    • Intervention duration was tied to hospital stay, and discharge timing was outside trial control.
    • Biochemical separation was modest despite substantial process separation.
    • Primary analysis used a modified full analysis set rather than a strict intention-to-treat population.
    • Protocol-planned longer-term outcomes were not reported in the index publication.

Interpretation & Why It Matters

  • Clinical practice
    HIT does not support expending additional inpatient effort purely to normalise sodium in mostly moderate chronic hyponatraemia when routine care is already being delivered. The practical implication is not to ignore hyponatraemia, but to avoid assuming that protocolised intensification alone will reduce short-term death or rehospitalisation.5
  • Causal inference
    The dissociation between better sodium correction and unchanged 30-day hard outcomes weakens the simple view that chronic inpatient hyponatraemia is a readily reversible driver of short-term prognosis. In this population, low sodium appears at least partly to be a disease-severity signal, or its causal contribution is too small or too delayed to affect a 30-day composite.
  • Future trial design
    Future trials will need either greater biologic separation or a different target population: more persistent SIAD, profound hyponatraemia, patients in whom low sodium blocks discharge, or older patients with functionally important symptoms. Longer follow-up and symptom-, mobility-, cognition-, and discharge-readiness end points are likely to be more informative than short-term mortality alone.

Controversies & Other Evidence

  • Methodologic criticism: The protocol explicitly anticipated that early discharge and contamination of routine care could blunt treatment separation, and that concern proved material. The accompanying editorial emphasised that the achieved biochemical gap was modest—maximum sodium change differed by only 1.3 mmol/L—and that the trial team had no control over discharge timing. The inclusion of hospital-acquired and hypovolaemic cases, which are often easier to correct, may also have narrowed between-group differences.45
  • Interpretive caution: HIT does not justify therapeutic nihilism. It compared intensified algorithmic care with contemporary routine care—not treatment versus no treatment—and protocolised therapy ended at discharge. The editorial’s practical conclusion is narrower: for short-term outcomes in mostly moderate chronic hyponatraemia, routine inpatient care was not improved by additional in-hospital intensification.5
  • Meta-analyses on correction speed and overcorrection: HIT was published into an active debate about sodium correction limits. A 2025 systematic review and meta-analysis in severe hyponatraemia associated faster correction with lower mortality and shorter stay without a statistically significant excess in osmotic demyelination syndrome, whereas the 2024 PRONATREOUS expert statement advised clinicians to retain conservative correction limits. A 2026 meta-analysis of overcorrection linked it with more neurologic complications, but the certainty of evidence was very low. HIT cannot resolve this dispute because it largely enrolled moderate chronic cases and excluded severe symptomatic hyponatraemia.6789
  • Subsequent trials of specific therapies: Smaller randomised studies continue to show that specific treatments can improve sodium without yet proving patient-centred benefit. Fluid restriction in chronic SIAD produced only a modest early sodium rise; empagliflozin improved sodium in chronic SIAD; and a 2026 inpatient trial found tolvaptan superior to fluid restriction over 3 days, but without symptom or length-of-stay benefit and with dextrose rescue needed in 19% of tolvaptan-treated patients. These data align with HIT in showing that biochemical efficacy and clinical benefit are not synonymous.101112
  • Subgroups, guidelines, and future direction: The only exploratory subgroup signal in HIT favoured patients aged 70 years or older; that is biologically plausible because older adults bear a disproportionate burden of falls, fractures, cognitive impairment, polypharmacy, and frailty related to chronic hyponatraemia. Contemporary reviews and a 2025 European survey continue to describe substantial practice variation and unresolved questions in hyponatraemia management, reinforcing the need for updated guidance that integrates HIT without overextending it to severe symptomatic disease.1314

Summary

  • HIT randomised 2173 hospitalised adults with chronic hypotonic hyponatraemia to targeted, aetiology-driven correction or routine care.
  • The intervention clearly improved biochemical correction: normonatraemia during treatment 60.4% versus 46.2%, normonatraemia at discharge 55.9% versus 36.8%.
  • Despite that, the 30-day composite of death or rehospitalisation was unchanged: 20.5% versus 21.8% (absolute difference −1.3 percentage points; 95% CI −4.9 to 2.2; P=0.45).
  • Harms were uncommon. Overcorrection was numerically more frequent with targeted correction (2.3% vs 1.4%), but no osmotic demyelination syndrome occurred.
  • The most defensible reading is that, in mostly moderate chronic inpatient hyponatraemia, intensified in-hospital sodium correction improves the number more than the near-term patient outcome.

Overall Takeaway

HIT is a landmark because it is the first large randomised test of a question that had previously been answered only by observational inference: does correcting chronic inpatient hyponatraemia improve hard outcomes? Its answer is sobering but clarifying—protocolised inpatient sodium correction can raise sodium and increase normonatraemia, yet in a broadly representative, mostly moderate population it did not reduce 30-day death or rehospitalisation.

Overall Summary

  • HIT separated biochemical efficacy from clinical efficacy: more sodium correction did not produce fewer 30-day deaths or rehospitalisations.
  • Internal validity is moderate—randomisation and follow-up were strong, but the trial was open-label and the achieved sodium separation was modest.
  • The trial argues against routine escalation purely to normalise sodium during admission in mostly moderate chronic hyponatraemia, not against treating severe symptoms or selected high-risk patients.

Bibliography