Supplement for the paper Long-term outcomes in patients with severe sepsis randomised to resuscitation with hydroxyethyl starch 130/0.42 or Ringer’s acetate
6S trial definition of fluid resuscitation
Fluid resuscitation was a bolus of intravenous fluid, which was given to increase intravascular volume. The resuscitation fluid should be given in addition to that required to replace ongoing insensible losses, urinary losses etc. or for nutrition.
6S trial criteria for severe sepsis
Sepsis was defined as a (1) DEFINED FOCUS OF INFECTION AND (2) at least TWO systemic inflammatory response syndrome (SIRS) criteria.
(1) DEFINED FOCUS OF INFECTION was indicated by either
(i) An organism grown in blood or sterile site
OR
(ii) An abscess or infected tissue (e.g. pneumonia, peritonitis, urinary tract, vascular line infection, soft tissue, etc).
(2) The 4 SIRS criteria were:
1. CORE TEMPERATURE > 38°C or < 36°C. (Core temperature was rectal, urinary bladder, central line, or tympanic). If oral, inguinal or axillary temperatures were used, 0.5°C was added to the measured value. Hypothermia < 36°C was confirmed by core temperature only. We used the most deranged value recorded in the 24 hours before randomisation.
2. HEART RATE > 90 beats/minute. If the patient had atrial arrhythmia, the ventricular rate was recorded. If the patients had known medical condition or were receiving treatment that would prevent tachycardia (for example, heart block or beta blockers), they had to meet two of the remaining three SIRS criteria. We used the most deranged value recorded in the 24 hours before randomisation.
3. RESPIRATORY RATE > 20 breaths/minute, PaCO2 < 32 mmHg (4.3 kPa) or mechanical ventilation for an acute process. We used the most deranged respiratory rate or PaCO2 recorded in the 24 hours before randomisation.
4. WHITE BLOOD CELL COUNT of >12 x 109/litre or < 4 x 109/litre or > 10% immature neutrophils (band forms). We used the most deranged value recorded in the 24 hours before randomisation.
Severe sepsis was defined as SEPSIS plus at least ONE ORGAN FAILURE, except when that organ failure was already present 48 hours before the onset of sepsis.
ORGAN FAILURE was defined as a Sepsis-related Organ Failure Assessment (SOFA) score > 2 for the organ in question.
6S trial exclusion criteria:
The following patients were not evaluated for inclusion:
· Aged < 18 years old
· Previously randomised in the 6S trial
· Allergy towards hydroxyethyl starch or malic acid
· Treated with > 1000 ml's of any synthetic colloid within the last 24 hours prior to randomisation
· Any form of renal replacement therapy
· Acute burn injury > 10% body surface area
· Severe hyperkalaemia, p-K > 6 mM
· Liver or kidney transplantation during current hospital admission
· Intracranial bleeding within current hospitalisation
· Enrolled into another ICU trial of drugs with potential action on circulation, renal function or coagulation
· Withdrawal of active therapy
Table S1. Adjustment for SAPS II at baseline for death at one-year as assessed by multivariate logistic regression analyses applying best- and worst-case scenarios for missing data
Risk factor at baseline / Odds ratio
(95% CI) / P-value / Odds ratio
(95% CI) / P-value
Assigned to HES
(reference: Ringer’s acetate) / 1.15
(0.85-1.55) / 0.37 / 1.38
(1.02-1.87) / 0.036
Admitted to a university hospital / 0.83
(0.61 – 1.13) / 0.23 / 0.83
(0.61 – 1.13) / 0.24
Septic shock / 1.10
(0.71 – 1.70) / 0.68 / 1.06
(0.68 – 1.64) / 0.80
Haematological malignancy / 2.47
(1.34 – 4.56) / 0.004 / 2.39
(1.29 – 4.44) / 0.006
SAPSII / 1.03
(1.02 – 1.04) / < 0.001 / 1.03
(1.02 – 1.04) / < 0.001
SOFA / 1.03
(0.97 – 1.10) / 0.29 / 1.03
(0.97 – 1.10) / 0.3
Age (per year) / 1.03
(1.02 – 1.04) / < 0.001 / 1.03
(1.02 – 1.04) / < 0.001
Diabetes / 0.59
(0.38 – 0.92) / 0.02 / 0.59
(0.38 – 0.91) / 0.02
Chronic kidney impairmenta / 1.59
(1.02 – 2.49) / 0.04 / 1.52
(0.97 – 2.37) / 0.066
Use of nephrotoxic drugsb / 0.78
(0.56 – 1.09) / 0.13 / 0.76
(0.55 – 1.06) / 0.11
HES=hydroxyethyl starch; SAPS=Simplified Acute Physiology Score; SOFA=Sepsis-related Organ Failure Assessment. 105 patients in the HES group and 108 in the Ringer’s group missed 1 or 2 of the 17 variables used to calculate SAPS II, so the full score could not be obtained in these patients. We imputed the missing SAPS II according to a worst and best case scenario giving patients either the highest (worst for HES) or lowest (best for HES) obtainable score in the missing domain(s). a. Chronic kidney impairment was defined as pre-admission plasma creatinine > 100 µmol/l (1.2 mg/dl). b. Nephrotoxic drugs included IV gentamicin, IV vancomycin, IV amphotericin B, IV polymyxins, IV dye contrast, ciclosporin A, NSAIDs, tacrolimus, voriconazole, anidulafungin, foscarnet and candesartancilexetil
Table S2. Accuracy in prediction of death within one year and calibration of risk factors at baseline
Risk factor at baseline / Prediction of death / Prediction of survival / Total correct prediction / P of Hosmer Lemeshow testAssigned to HES
(reference: Ringer’s acetate) / 0.0% / 100.0% / 53.8% / na
Admitted to a university hospital / 100.0% / 0.0% / 53.8% / na
Septic shock / 86.7% / 18.4% / 55.1% / na
Haematological malignancy / 100.0% / 0.0% / 53.8% / na
SAPSII / 68.9% / 48.3% / 59.4 / 0.51
SOFA / 78.6% / 33.2% / 57.7% / 0.13
Age (per year) / 74.6% / 42.5% / 59.8% / 0.28
Diabetes / 88.1% / 15.7% / 54.6 / na
Chronic kidney impairmenta / 100.0% / 0.0% / 53.8 / na
Use of nephrotoxic drugsb / 100.0% / 0.0% / 53.8% / na
All quantities included / 72.1% / 56.8% / 65.0% / 0.42
na = not applicable
Assessing the one year mortality using Cox- and logistic regression analyses.
Using the cumulative sums of martingale residuals and their transforms over certain coordinates (such as covariate values or follow-up times) the proportional hazard (PH) assumption of the Cox analysis and the fit of continuous variables were assessed using a Kolmogorov-type supremum test (Lin DY, Wei LJ, Ying Z (1993) Checking the Cox model with cumulative sums of martingale-based residuals. Biometrika 80; 557–572) based on 1,000 simulated residual patterns (SAS 9.3). The P-values assessing the PH assumption is shown for each of the 10 covariates in Table S3 together with the P-values testing the fit of the three continuous variables. It appears that the covariate University hospital yes/no is of borderline significance (P = 0.022). However, using, e.g., Holm’s multiplicity adjustment the smallest P-value should be < 0.05/13 = 0.004 to be declared significant.
We have supplemented the results where the site category was included as a covariate by corresponding analysis using the one year mortality as a binary quantity because in this context the site category presented no problems. The results are shown in Table S4 where we also applied the best case/worst case scenarios imputation of missing SAPSII values. It appears that it is very unlikely that HES affects the one year mortality in a beneficial way. Even when we adjusted with the best-case imputation of SAPSII and design variables, the effect of HES was far from being beneficial (HR = 1.09, OR =1.15, P= 0.38 respectively 0.37)
Table S3 Assessment of the proportional hazard (PH) assumption and the fit of continuous variables in the Cox model
Covariate / P of assessment of PH assumption / P of assessment of fit of continuous covariateIntervention Y/N / 0.19 / Na
Shock Y/N / 0.79 / Na
Hematological malignancy Y/N / 0.77 / Na
University hospital Y/N / 0.022 / Na
Diabetes Y/N / 0.41 / Na
Chronic kidney failure Y/N / 0.087 / Na
Nephrotoxic drugs Y/N / 0.74 / Na
Age / 0.10 / 0.53
SAPS II / 0.24 / 0.22
SOFA score / 0.12 / 0.55
Na denotes not applicable
Table S4 shows the effect of intervention using the Cox model without adjustment, with adjustment by stratification variables with and without worst case design variables (missing values given the worst values for HES) and best case design variables (missing values given the best values for HES).
Table S4 Comparing the effect of HES and saline on the one year mortality
Adjustment / Cox analysis / Logistic regressionHazard rate (HR)
(95% CI) / P / OR
(95% CI) / P
No adjustments / 1.13
(0.94 to 1.37) / 0.21 / 1.20
(0.91 to 1.59) / 0.20
Adjustment by stratification variables / 1.14
(0.95 to 1.38) / 0.16 / 1.21
(0.91 to 1.61) / 0.19
Adjustment by
stratification variables design variables
(best case scenario imputation of SAPSII) / 1.09
(0.90 to 1.32) / 0.38 / 1.15
(0.85 to 1.55) / 0.37
Adjustment by stratification variables + design variables
(worst case scenario imputation of SAPSII) / 1.24
(1.03 to 1.51) / 0.025 / 1.38
(1.02 to 1.87) / 0.036