Skip to main content

Foundational Trials

SSSP-2

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of an Early Resuscitation Protocol on In-Hospital Mortality Among Adults With Sepsis and Hypotension: A Randomized Clinical Trial
  • Acronym: SSSP-2
  • Year: 2017
  • Journal published in: JAMA
  • Citation: Andrews B, Semler MW, Muchemwa L, Kelly P, Lakhi S, Heimburger DC, et al. Effect of an early resuscitation protocol on in-hospital mortality among adults with sepsis and hypotension: a randomized clinical trial. JAMA. 2017;318(13):1233-1240.

Context & Rationale

  • Background
    • Early, protocolised sepsis resuscitation (fluids, vasopressors, timely antimicrobials) was largely developed and tested in high-resource health systems with reliable critical care support.
    • Low-resource settings face different constraints (limited intensive care beds, mechanical ventilation, monitoring, infusion pumps, and staffing), potentially altering the balance between benefit and iatrogenic harm from aggressive resuscitation.
    • Adult sepsis with hypotension in sub-Saharan Africa often occurs in younger populations with high HIV prevalence, severe anaemia, malnutrition, and limited access to organ support, raising uncertainty about the safety of standard “bundle” approaches.
  • Research Question/Hypothesis
    • Whether a simplified 6-hour early resuscitation protocol for adults with suspected infection and hypotension, delivered in a low-resource emergency department, would reduce (rather than increase) in-hospital mortality compared with usual care.
  • Why This Matters
    • It directly tested “exportability” of early resuscitation bundles into a health system with minimal ICU capacity, where harms from fluids/vasopressors may be less manageable.
    • It challenged the assumption that protocolised sepsis care is universally beneficial, highlighting the need for context-specific evidence before large-scale implementation.
    • It provided rare randomised data from sub-Saharan Africa in adult sepsis with hypotension, a population under-represented in landmark early shock trials.

Design & Methods

  • Research Question: In adults with suspected infection and hypotension in a low-resource emergency department, does delivery of a simplified early resuscitation protocol reduce in-hospital mortality compared with usual care?
  • Study Type: Randomised, single-centre, parallel-group clinical trial in the emergency department of University Teaching Hospital, Lusaka, Zambia; open-label (protocol delivery not practically blindable); investigator-initiated.
  • Population:
    • Adults with suspected infection, ≥2 systemic inflammatory response syndrome criteria, and hypotension (systolic blood pressure ≤90 mm Hg or mean arterial pressure ≤65 mm Hg).
    • Enrolment required within 4 hours of the first qualifying blood pressure measurement and within 24 hours of emergency department registration.
    • Key exclusions included suspected gastrointestinal bleeding without fever, immediate surgery required, severe hypoxaemic respiratory failure (respiratory rate >40/min and oxygen saturation <90%), suspected heart failure exacerbation, and end-stage renal disease.
  • Intervention:
    • Protocolised 6-hour resuscitation delivered by a study nurse, incorporating rapid crystalloid administration (2 L in the first hour, with reassessment and additional boluses up to 4 L by 6 hours), dopamine for persistent hypotension after initial fluids, and packed red blood cell transfusion for haemoglobin <7 g/dL.
    • Reassessment checkpoints incorporated clinical examination (including jugular venous pressure), respiratory rate, and oxygen saturation during fluid delivery to mitigate fluid-associated respiratory compromise.
  • Comparison:
    • Usual care as determined by the treating clinicians, with study nurse monitoring and data collection but without mandated protocol targets for fluids/vasopressors/transfusion.
  • Blinding: Unblinded; primary outcome (in-hospital mortality) is objective, but co-interventions and thresholds for escalation/withholding could be influenced by awareness of group assignment.
  • Statistics: A total of 212 patients were required to detect a 20% absolute reduction in in-hospital mortality (from 65% to 45%) with 80% power at the 5% significance level; primary analysis was intention-to-treat.
  • Follow-Up Period: In-hospital outcomes to discharge; post-enrolment mortality assessed to 28 days.

Key Results

This trial was not stopped early. One interim analysis at approximately half enrolment was prespecified with a conservative stopping boundary; the trial completed planned enrolment.

Outcome Sepsis protocol Usual care Effect p value / 95% CI Notes
In-hospital mortality (primary) 51/106 (48.1%) 34/103 (33.0%) RR 1.46 95% CI 1.04 to 2.05; P=0.03 Absolute difference 15.1% (95% CI 2.0% to 28.3%)
28-day mortality 67.0% 45.3% RR 1.48 95% CI 1.14 to 1.91; P=0.002 Vital status at day 28 available for 94.2% overall
Length of hospital stay (days) Median 5 (IQR 3–7) Median 7 (IQR 4–11) Not reported P=0.01 Shorter stay likely reflects higher early mortality in the protocol group
Intravenous fluid volume in first 6 h (L) Median 3.5 (IQR 2.7–4.0) Median 2.0 (IQR 1.0–2.5) Not reported P<0.001 Protocol specified 2 L in first hour; up to 4 L by 6 h
Dopamine use in first 6 h 15/106 (14.2%) 2/103 (1.9%) Not reported P=0.001 Dopamine duration: median 4 (IQR 4–5) h vs 4 (IQR 4–4) h; P=0.52
Clinical respiratory compromise in first 6 h 38/106 (35.8%) 23/103 (22.3%) Not reported P=0.03 Iatrogenic pulmonary oedema: 2/106 (1.9%) vs 0/103 (0%)
Change in lactate at 6 h (mmol/L) Median −1.2 (IQR −3.4 to 0.3) Median −0.5 (IQR −2.2 to 1.1) Not reported P=0.02 Physiological improvement did not translate into improved survival
  • Despite achieving clear separation in early resuscitation intensity (fluids and dopamine), the protocol arm had higher in-hospital and 28-day mortality.
  • Sensitivity analyses supported robustness of the primary finding: unadjusted logistic regression OR 1.88 (95% CI 1.07 to 3.30; P=0.03) and adjusted OR 1.93 (95% CI 1.09 to 3.43; P=0.03); Cox modelling similarly favoured harm (unadjusted HR 1.65; 95% CI 1.12 to 2.44; P=0.01; adjusted HR 1.68; 95% CI 1.14 to 2.49; P=0.009).
  • As-treated analyses by early fluid exposure were not statistically significant (≥3 L vs <3 L: adjusted OR 1.41; 95% CI 0.80 to 2.49; P=0.24), underscoring that the protocol’s effect cannot be confidently attributed to a single component.

Internal Validity

  • Randomisation and Allocation
    • Randomisation occurred after eligibility confirmation; allocation concealment used sequential envelopes (as per protocol documentation) and produced broadly comparable baseline groups.
    • Baseline severity was similar (SAPS-3: median 52 [IQR 44–61] vs 54 [IQR 44–61]).
  • Drop out / exclusions
    • From 212 randomised, 3 were excluded post-randomisation due to ineligibility (primary analysis set n=209); this introduces a small risk of bias but is numerically limited.
    • At 28 days, vital status was available for 94.2% overall, reducing attrition-related uncertainty for the secondary mortality endpoint.
  • Performance / detection bias
    • Open-label delivery could influence co-interventions, escalation decisions, and discharge timing; however, mortality is objective and less susceptible to ascertainment bias.
    • Study nurse involvement in both groups may have reduced between-group contamination in monitoring intensity, but also makes the control arm “usual care” a monitored usual care condition rather than fully naturalistic practice.
  • Protocol adherence and separation of the variable of interest
    • Fluid separation at 6 h: median 3.5 L (IQR 2.7–4.0) vs 2.0 L (IQR 1.0–2.5); P<0.001.
    • High-dose fluids were markedly more common in the protocol group within 6 h: >3 L in 68.9% vs 6.8%; >4 L in 16.0% vs 1.0% (both P<0.001).
    • Dopamine separation within 6 h: 14.2% vs 1.9%; P=0.001.
    • Physiological separation was mixed: lactate change −1.2 vs −0.5 mmol/L; P=0.02, but hypotension resolution at 6 h was worse in the protocol group (76.4% vs 95.1%; P<0.001), consistent with a more refractory shock phenotype or protocol-related haemodynamic effects.
  • Baseline characteristics and clinical plausibility of benefit/harm
    • Participants were severely immunocompromised and physiologically vulnerable (HIV in 89.5%; median CD4 66–68 cells/µL; median albumin 2.1–2.3 g/dL; mean haemoglobin 7.8 g/dL).
    • Pre-enrolment cardiopulmonary reserve was uncertain: 20.8% vs 10.7% had jugular venous pressure above the sternal angle at baseline, and clinical assessment of volume status may have variable reliability.
  • Timing and dose
    • Early enrolment (median 71 vs 76 minutes from emergency department registration to enrolment) supports that the intervention targeted the intended early resuscitation window.
    • The protocol dose (2 L in first hour; up to 4 L by 6 h) is a relatively high fixed-volume strategy for a population likely to have low body mass and limited access to ventilatory support if pulmonary oedema develops.
  • Outcome assessment and statistical rigour
    • Primary endpoint (in-hospital mortality) is objective; secondary outcomes included 28-day mortality and pre-specified process measures (fluids, vasopressors, respiratory compromise).
    • Observed control mortality (33.0%) was substantially lower than the mortality assumed for sample size planning (65%), increasing the risk of imprecision; nonetheless, multiple sensitivity analyses were directionally consistent with harm.

Conclusion on Internal Validity: Overall, internal validity appears moderate: randomisation produced broadly comparable groups and the intervention achieved clear separation in key exposures, but open-label delivery, small post-randomisation exclusions, and single-centre design limit confidence and make attribution of harm to any single protocol component uncertain.

External Validity

  • Population representativeness
    • Represents a high-burden, low-resource emergency department population with very high HIV prevalence, severe anaemia, and hypoalbuminaemia.
    • Only 0.5% of participants were managed in an ICU, reflecting the dominant care pathway in many similar systems but not in high-income settings.
    • Exclusions (notably severe hypoxaemia/tachypnoea and suspected heart failure exacerbation) mean findings apply chiefly to patients without overt cardiogenic pulmonary oedema or severe respiratory failure at presentation.
  • Applicability
    • Highly applicable to comparable low-resource systems where close monitoring and rescue organ support are constrained.
    • Less directly transferable to high-resource ICUs/EDs where mechanical ventilation, norepinephrine via controlled infusions, echocardiography, and invasive monitoring can mitigate harm and enable more individualised titration.
    • Protocol used dopamine as vasopressor; modern practice in many settings uses norepinephrine, so haemodynamic/arrhythmic risk profiles may differ.

Conclusion on External Validity: Generalisability is strong for similar low-resource emergency care environments and patient phenotypes (advanced HIV, limited ICU access), but limited for well-resourced centres where monitoring and organ support capabilities substantially alter the risk–benefit profile of aggressive early resuscitation.

Strengths & Limitations

  • Strengths:
    • Randomised design in a setting and population with scarce prior trial evidence.
    • Early enrolment and pragmatic delivery in a real-world emergency department workflow.
    • Clear protocol–control separation in fluid and vasopressor exposure, supporting interpretability of a “resuscitation intensity” contrast.
    • Objective primary outcome and robust consistency across multiple sensitivity analyses.
  • Limitations:
    • Single-centre, open-label trial; co-interventions and escalation decisions could vary by group assignment.
    • Observed control mortality was much lower than expected, increasing the chance of random imbalance and imprecision.
    • Multi-component protocol (fluids + dopamine + transfusion thresholds + monitoring), limiting causal attribution to any single element.
    • Protocol used dopamine rather than norepinephrine, and available organ support was minimal; both limit extrapolation to contemporary high-resource practice.
    • Clinical volume status assessment (jugular venous pressure, respiratory signs) may be variably reliable and operator-dependent.

Interpretation & Why It Matters

  • Clinical meaning
    • In this low-resource emergency department cohort, protocolised early aggressive resuscitation increased mortality compared with monitored usual care, arguing against uncritical transplantation of fixed-volume early resuscitation bundles into similar settings.
  • Mechanistic inference
    • The protocol improved a biochemical marker (greater lactate fall) but worsened clinical outcomes, while increasing early respiratory compromise and rare pulmonary oedema events, supporting the plausibility of iatrogenic harm from resuscitation intensity in patients with limited physiological reserve and constrained rescue capacity.
  • Methodological importance
    • It demonstrates that “bundle adherence” and early physiological improvements are not valid surrogates for benefit across contexts; outcomes must be tested directly, particularly where supportive care differs.

Controversies & Subsequent Evidence

  • Context mismatch and iatrogenic harm in low-resource environments
    • Accompanying editorial highlighted that aggressive haemodynamic protocols may produce net harm when ventilatory support, monitoring, and critical care rescue therapies are constrained, and where comorbidities (advanced HIV, anaemia, malnutrition) may shift physiological tolerance for fluids/vasopressors. 1
  • Alternative explanations for the mortality signal (antibiotics vs fluids vs protocol burden)
    • Correspondence questioned whether implementing a multi-step protocol could delay timely antibiotics and whether the resuscitation volume could have been excessive for the population’s physiological reserve. 2
    • The trialists’ reply emphasised similar antibiotic timing between groups and argued that fluid volumes were not extraordinary compared with major early-shock trials in higher-resource systems, reinforcing that identical “inputs” may have different effects across contexts. 3
  • Guideline evolution towards reassessment and uncertainty around fixed-volume boluses
    • Surviving Sepsis Campaign updates maintained early resuscitation principles but increasingly emphasised frequent reassessment of fluid responsiveness, avoidance of rigid one-size-fits-all volumes, and earlier vasopressors when hypotension persists. 45
  • Subsequent randomised evidence on fluid-restrictive approaches (high-resource settings)
    • CLASSIC and CLOVERS did not show mortality benefit from more liberal early fluid strategies and supported ongoing equipoise (and in some pathways preference) for earlier vasopressors and fluid-sparing resuscitation after initial stabilisation, consistent with caution against aggressive fixed-volume resuscitation. 67
  • Effect modification by setting (geo-economic context)
    • A systematic review/meta-analysis reported that associations between resuscitation fluid volume and outcomes varied by geo-economic context, reinforcing the plausibility that resuscitation intensity has different net effects across health systems and resource levels. 8
  • Broader fluid guidance beyond sepsis-specific bundles
    • Recent ESICM clinical practice guidelines on fluid therapy provide graded recommendations on fluid initiation, titration, and de-escalation for critically ill adults, aligning conceptually with careful reassessment-based strategies that may reduce iatrogenic harm after initial stabilisation. 9

Summary

  • In a low-resource Zambian emergency department, a simplified early resuscitation protocol (higher fluids, dopamine, transfusion threshold) increased in-hospital mortality compared with usual care.
  • Protocol delivery achieved substantial separation in early intravenous fluid volume (3.5 vs 2.0 L in 6 h) and dopamine exposure (14.2% vs 1.9%).
  • Harms plausibly related to resuscitation intensity were signalled by increased early respiratory compromise (35.8% vs 22.3%) and rare iatrogenic pulmonary oedema (1.9% vs 0%).
  • Physiological improvement (greater lactate fall) did not predict benefit; mortality remained higher across multiple sensitivity analyses.
  • The trial is a cautionary landmark: sepsis bundles and fixed-volume resuscitation require contextual validation, particularly where rescue organ support is limited.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • In SSSP-2, dopamine was used for persistent hypotension; in many contemporary systems norepinephrine is preferred, but peripheral access, infusion control, and monitoring constraints remain central to implementation safety.
  • Key practical translation is not “avoid all fluids”, but “avoid rigid fixed-volume protocols without reassessment”, particularly where respiratory rescue is limited.

Overall Takeaway

SSSP-2 is a landmark because it showed that protocolised early aggressive resuscitation can increase mortality when applied in a low-resource setting with limited critical care rescue, despite improving some early physiological markers. It reshaped sepsis implementation thinking by demanding context-specific trials and by reinforcing reassessment-based resuscitation rather than fixed-volume “bundle” replication across health systems.

Overall Summary

  • In a Zambian emergency department cohort with advanced HIV and minimal ICU access, a fixed-intensity early resuscitation protocol increased mortality despite achieving greater early fluid delivery and dopamine use.

Bibliography