Fluid Resuscitation — Crystalloid vs Colloid, Balanced vs Saline, SOSD/ROSE & Fluid Overload

Overview & definition

Fluid resuscitation comes down to four evidence-based questions: (1) crystalloid vs colloid (crystalloid is first-line), (2) balanced vs saline (balanced is preferred — the SMART trial), (3) how much (titrate to endpoints and to fluid responsiveness, not to a formula), and (4) when to stop (the SOSD/ROSE model — restrictive strategies are safe, and overload kills). The key compositional concern is the chloride load, which drives hyperchloraemic metabolic acidosis via a reduced strong ion difference. Titrate to clinical endpoints, test responsiveness before each bolus, and plan for de-resuscitation of overload.[1][1]

Fluid compartments and distribution — where the fluid goes

The fate of any resuscitation fluid is governed by body-water physiology. Total body water (TBW) is approximately 60 per cent of body weight in a young man (50 per cent in women, lower again in the elderly, ~75 per cent in neonates). TBW is divided into two main compartments:[1]

• Intracellular fluid (ICF) — about two-thirds of TBW (≈ 40 per cent of body weight).
• Extracellular fluid (ECF) — about one-third of TBW (≈ 20 per cent of body weight). The ECF is further split into interstitial fluid (three-quarters of the ECF, ≈ 15 per cent of body weight) and intravascular plasma (one-quarter of the ECF, ≈ 5 per cent of body weight). A small transcellular compartment (CSF, synovial, GIT secretions) completes the picture. [1]

The clinically decisive consequence is the distribution ratio: an isotonic crystalloid given intravenously leaves the vasculature rapidly. Roughly only one-quarter to one-third of a crystalloid bolus remains intravascular after 30-60 minutes; the rest distributes throughout the ECF (interstitium). This is why a "500 mL bolus" is not 500 mL of circulating volume for long, and why repeated boluses accumulate oedema long before they accumulate circulating volume.[1]

The revised Starling model — why crystalloid leaks

Classical Starling taught that fluid flux across the capillary depends on a balance of hydrostatic and oncotic pressures, with albumin holding fluid in. The revised (glycocalyx) Starling model is now the accepted physiology and it changes practice:[1]

• The endothelial glycocalyx layer (EGL) is a fragile carbohydrate-protein mesh on the vascular lumen. The relevant oncotic force opposing filtration is the colloid osmotic pressure difference between plasma and the sub-glycocalyx space, not between plasma and the interstitium.
• Under normal conditions the sub-glycocalyx oncotic pressure is low, so plasma albumin does oppose filtration — but once the EGL is shed (sepsis, trauma, major surgery, hyperglycaemia, ischaemia-reperfusion), albumin escapes and the colloid advantage largely disappears. This is the mechanistic reason colloids failed to outperform crystalloids in critical illness: the sick endothelium leaks everything.
• Filtration is therefore governed largely by hydrostatic pressure and capillary integrity — which is why reducing preload/venous pressure (de-resuscitation) removes oedema, and why a damaged glycocalyx turns any fluid into interstitial oedema. [1]

Practical upshot: in the inflamed/leaky patient, no colloid is "volume-sparing" enough to justify its cost and risk; crystalloid, given in measured aliquots guided by responsiveness, is the rational default.[1]

Crystalloid vs colloid

Crystalloids are first-line — cheaper, safer, and evidence-based.[1]

Colloids:[1]

• Albumin — the SAFE trial (6,997 patients; 4 per cent albumin vs saline showed no mortality difference). May benefit the sepsis subgroup and patients with severe hypoalbuminaemia.[3]
• Hydroxyethyl starch (HES) — HARMFUL. The CHEST trial showed increased AKI and RRT, and the 6S trial showed increased mortality in sepsis. Withdrawn or restricted.[4][5]
• Gelatin — less evidence; a possible AKI concern.[1]

The definitive head-to-head between the two classes is CRISTAL: a multicentre trial of 2,857 patients with hypovolaemic shock randomised to colloids vs crystalloids found no difference in 28-day mortality, confirming that the more expensive colloids buy no outcome advantage.[7]

Balanced vs saline — the chloride question

The SMART trial (Semler, NEJM 2018) — 15,752 ICU patients; balanced crystalloids (lactated Ringer's, Plasma-Lyte) vs saline.[1]

• The balanced group had lower all-cause mortality (10.3 vs 11.1 per cent), less AKI, and less need for RRT.[1]
• The mechanism: saline has a high chloride (154 mmol/L), which causes hyperchloraemic metabolic acidosis → renal afferent arteriole vasoconstriction → AKI.[1]

SMART-DKA — balanced fluids give faster DKA resolution (a quicker bicarbonate rise and a shorter insulin infusion).[2]

Strong ion difference (SID) — the mechanism in one line

The acid-base effect of a fluid is governed by its strong ion difference (SID) — the difference between the sum of fully-dissociated cations and anions. Plasma SID is normally ~40 mEq/L. A fluid with a SID well below 40 (saline, SID = 0 because Na = Cl = 154) causes hyperchloraemic metabolic acidosis by diluting the plasma bicarbonate and adding a chloride load; a fluid with a SID near 40 (Plasma-Lyte ~50, Hartmann ~28) is acid-base neutral.[1]

This is why "balanced" matters: it is a chloride/SID property, not a "lactate is good" property. The lactate and acetate buffers in balanced fluids are metabolised to bicarbonate, restoring SID, and the lower chloride prevents the renal vasoconstriction that drives saline-associated AKI. [1]

Composition

The key variable is chloride: saline 154 (high), lactated Ringer's 111, Plasma-Lyte 98 (near-physiological). Lactated Ringer's is slightly hypotonic (Na 131), so use caution in TBI (cerebral oedema — controversial).[1]

Tonicity vs osmolality — a common exam trap: 5% dextrose is isotonic in the bag (osmolality 278) but hypotonic in vivo, because the dextrose is metabolised leaving free water — it is effectively a free-water load and unsuitable for resuscitation. Hartmann's is slightly hypotonic (osmolality 279) but a near-physiological resuscitation fluid. [1]

Balanced crystalloid types

• Compound Sodium Lactate (Hartmann's / lactated Ringer's) — the workhorse; near-physiological chloride, calcium-containing (incompatible with citrated blood in the same line), slightly hypotonic.
• Plasma-Lyte 148 / Plasma-Lyte-A — the most physiological: chloride 98, magnesium-containing, buffered with acetate and gluconate (not lactate, so no Ca incompatibility — runs with blood); the most "balanced" but more expensive.
• Ringer's acetate — the Scandinavian analogue (acetate buffer instead of lactate); used in CHEST and 6S. [1]

Crystalloid-type trial evidence

[1]

Crystalloid vs colloid trial evidence

[1]

Principles

1. What type. Balanced crystalloid is first-line (lactated Ringer's / Hartmann's, Plasma-Lyte).[1][1]

2. How much. Titrate to clinical endpoints (MAP 65, lactate, urine output) — not to a formula (the 30 mL/kg bolus is only a starting point in sepsis).[1]

3. When to stop. If fluid responsive (PLR, SVV, echo), give fluid; if not responsive, use a vasopressor rather than more fluid. De-resuscitate (diuresis or RRT) for fluid overload.[1]

4. Complications of overload. Pulmonary oedema, abdominal compartment syndrome, and tissue oedema (impaired wound healing, gut oedema).[1]

5. Re-test before every bolus. Fluid responsiveness changes hour to hour — yesterday's responder is today's overloaded patient. A bolus is a test, not a resuscitation target.[1]

The SOSD / ROSE model — when to give and when to remove fluid

The ROSE model (Resuscitation, Optimisation, Stabilisation, De-resuscitation — sometimes SOSD, Salvage/Optimisation/Stabilisation/De-escalation) frames intravenous fluid as a drug with a therapeutic window and a toxicity profile. Fluid is actively given in the first two phases, merely maintained in the third, and actively removed in the fourth.[1]

[1]

The model reframes the question from "how much fluid does the patient need?" to "which phase is the patient in?" — and the phase determines whether fluid should be added, maintained, or removed.[1]

Assessing fluid responsiveness — give fluid only to responders

Only about half of ICU patients are fluid responsive (a rise in stroke volume ≥ 10 per cent to a bolus). Static markers (CVP, PAOP) do not predict it — the curve is flat. Dynamic tests do. (See the dedicated fluid responsiveness topic for full technique.)[1]

[1]

Permissive hypotension — when less fluid is more

In active haemorrhage, aggressive crystalloid resuscitation before bleeding control can be harmful: it dilutes clotting factors, displaces clots (pops them off), lowers blood viscosity, and propagates the lethal triad (acidosis, hypothermia, coagulopathy). The classic evidence is Bickell (NEJM 1994): in 598 patients with penetrating torso trauma and hypotension, delaying fluid resuscitation until operative control of haemorrhage reduced mortality and shortened hospital stay.[12]

Modern interpretation — reinforced by damage-control resuscitation and a 2018 systematic review:[14]

• Maintain a permissive (low-normal) blood pressure (systolic ~80-90 mmHg, MAP ~65) until haemorrhage is controlled, using minimal crystalloid and prioritising blood products (massive transfusion, balanced ratios) and early vasopressors.
• The aim is perfusion without disruption of fresh clot, not normotension.
• Contraindications: traumatic brain injury (a low MAP causes secondary brain injury — keep CPP > 60), and any condition in whom cerebral or coronary hypoperfusion is the dominant threat.

Fluid overload — the dangers of cumulative positive balance

A cumulative positive fluid balance is an independent predictor of mortality in critical illness. The harms are mechanical, biochemical, and organ-specific:[8][9]

• Pulmonary — pulmonary oedema, worsened oxygenation, prolonged ventilation (the defining harm in ARDS).
• Tissue/gut — interstitial oedema impairs wound healing, mitochondrial function and gut motility; raises intra-abdominal pressure.
• Renal (paradoxical) — renal venous congestion and abdominal compartment syndrome cause AKI — the very organ resuscitation was meant to protect.
• Haemodilution — falls in haematocrit and albumin; worsens oxygen delivery and oncotic pressure. [1]

Restrictive-strategy trial evidence

[1]

Specific conditions

• DKA: balanced fluids (SMART-DKA — faster resolution).[2]
• Sepsis: a 30 mL/kg bolus then reassess; balanced preferred; move to a restrictive strategy and early vasopressors once the initial resuscitation is complete (CLOVERS, CLASSIC).[1][9][10]
• Trauma / burns: lactated Ringer's (ATLS, military practice); permissive hypotension in active haemorrhage until bleeding control; hypertonic saline offers no outcome benefit over isotonic in trauma/TBI (the ROC/TRUST subgroup) and is not recommended.[12][13][14]
• TBI: saline (lactated Ringer's is slightly hypotonic → cerebral-oedema risk — controversial; the current Brain Trauma Foundation guideline favours saline).[1]
• ARDS: conservative fluid strategy (FACTT) — more ventilator-free days.[8]
• Severe hypoalbuminaemia: albumin is a reasonable adjunct in the rare patient with profound, symptomatic hypoalbuminaemia (oncotic support) — not for routine resuscitation (SAFE).[3]
• Hepatic failure / cirrhosis: balanced crystalloid cautious of potassium load; albumin is first-line for spontaneous bacterial peritonitis (SBP) and hepatorenal syndrome, and for large-volume paracentesis — a specific role distinct from general resuscitation.[1]
• Cardiac failure / cardiogenic shock: give fluid only after demonstrating responsiveness (these patients are frequently preload-independent and easily pushed into pulmonary oedema); a cautious 250 mL challenge with echo is the safest approach.[1]

Prognosis

Balanced crystalloids reduce mortality and AKI (SMART; confirmed hospital-wide by FLUID). Fluid overload increases mortality (pulmonary oedema, abdominal compartment, wound breakdown, renal venous congestion). Restrictive strategies are safe (CLASSIC, FACTT, CLOVERS), and active de-resuscitation improves outcomes. The prognosis of the underlying condition dominates; the fluid strategy's contribution is to avoid iatrogenic harm from both the wrong fluid (saline, starches) and too much fluid.[1][2][8][9][10]

[1]

SAQ practice

[1]

Clinical pearls

Red flags

References

1. [1]Semler MW, Self WH, Wanderer JP, et al.; SMART Trial and PETAL Network Investigators. Balanced Crystalloids versus Saline in Critically Ill Adults N Engl J Med, 2018.PMID 29768150
2. [2]Self WH, Evans M, Semler MW, et al. Clinical Effects of Balanced Crystalloids vs Saline in Adults With Diabetic Ketoacidosis: A Subgroup Analysis of Cluster Randomized Clinical Trials JAMA Netw Open, 2020.PMID 33196806
3. [3]Finfer S, Bellomo R, Boyce N, French J, Myburgh J, Norton R; SAFE Study Investigators. A comparison of albumin and saline for fluid resuscitation in the intensive care unit N Engl J Med, 2004.PMID 15163774
4. [4]Myburgh JA, Finfer S, Bellomo R, et al.; CHEST Investigators; Australian and New Zealand Intensive Care Society Clinical Trials Group. Hydroxyethyl starch or saline for fluid resuscitation in intensive care N Engl J Med, 2012.PMID 23075127
5. [5]Perner A, Haase N, Guttormsen AB, et al.; 6S Trial Group; Scandinavian Critical Care Trials Group. Hydroxyethyl starch 130/0.42 versus Ringer's acetate in severe sepsis N Engl J Med, 2012.PMID 22738085
6. [6]Young P, Bailey M, Beasley R, et al.; SPLIT Investigators; ANZICS Clinical Trials Group. Effect of a Buffered Crystalloid Solution vs Saline on Acute Kidney Injury Among Patients in the Intensive Care Unit: The SPLIT Randomized Clinical Trial JAMA, 2015.PMID 26444692
7. [7]Annane D, Siami S, Jaber S, et al.; CRISTAL Investigators. Effects of fluid resuscitation with colloids vs crystalloids on mortality in critically ill patients presenting with hypovolemic shock: the CRISTAL randomized trial JAMA, 2013.PMID 24108515
8. [8]National Heart, Lung, and Blood Institute ARDS Clinical Trials Network; Wiedemann HP, Wheeler AP, et al. Comparison of two fluid-management strategies in acute lung injury N Engl J Med, 2006.PMID 16714767
9. [9]Meyhoff TS, Hjortrup PB, Wetterslev J, et al. Restriction of Intravenous Fluid in ICU Patients with Septic Shock N Engl J Med, 2022.PMID 35709019
10. [10]National Heart, Lung, and Blood Institute PETAL Network; Shapiro NI, Douglas IS, et al. Early Restrictive or Liberal Fluid Management for Sepsis-Induced Hypotension N Engl J Med, 2023.PMID 36688507
11. [11]Semler MW, Bravata DM, Gershengorn HB, et al.; FLUID Investigators. A Crossover Trial of Hospital-Wide Lactated Ringer's Solution versus Normal Saline N Engl J Med, 2025.PMID 40503714
12. [12]Bickell WH, Wall MJ Jr, Pepe PE, et al. Immediate versus delayed fluid resuscitation for hypotensive patients with penetrating torso injuries N Engl J Med, 1994.PMID 7935634
13. [13]Bulger EM, May S, Kerby JD, et al.; ROC Investigators. Out-of-hospital hypertonic resuscitation following severe traumatic brain injury: a randomized controlled trial JAMA, 2010.PMID 20924011
14. [14]Hashmi ZG, Schneider EB, Kugler NW, et al. Permissive hypotension versus conventional resuscitation strategies in adult trauma patients with hemorrhagic shock: A systematic review and meta-analysis of randomized controlled trials J Trauma Acute Care Surg, 2018.PMID 29370058