Study design overview

The present study is a prospective, single-centre, randomised, double-blinded, two-arm trial in patients after elective craniotomy.

Study setting and population

The study setting is neurosurgical ICU (30 beds), Beijing Tiantan Hospital (1100 beds), Capital Medical University, Beijing, China.

All patients after elective intracranial surgery admitted to our ICU are screened daily for study eligibility.

Inclusion criteria are:

1. Age between 18 and 65 years;

2. Within 24 h of operation;

3. Require the use of hyperosmolar agents for the prevention or treatment of postoperative brain oedema.

Exclusion criteria are:

1. History of diabetes;

2. History of alcohol abuse;

3. Herniation of brain;

4. Unstable haemodynamic condition: systolic blood pressure (BP) less than 90 mm Hg or need for continuous infusion of vasopressor;

5. Presence of oliguric renal failure;

6. Serum sodium concentration below 130 mmol/L or above 155 mmol/L;

7. Enrolled in another trial.

Consecutive patients are randomly assigned to receive 20% mannitol (M group) or 3.1% sodium chloride solution (HS group). Hyperosmolar agents (20% mannitol or 3.1% sodium chloride solution) are routinely used in our clinical practice for the prevention and treatment of postoperative brain oedema in patients after craniotomy. Indications include ICP above 25 mm Hg, brain oedema shown by CT, poor neurological status due to brain oedema and bulging of brain during operation. Hyperosmolar agents are initiated in some cases because of clinical decline or elevated ICP, but in many others they are instituted based on CT results or empirically.

Randomisation, allocation concealment and blind

The study has a prospective, randomised, double-blinded, controlled, parallel-group design. Consecutive patients are randomly assigned 1 : 1 to one of the two treatment groups, labelled as ‘M group’ and ‘HS group’. Randomisation is based on a computer generated random digits table and follows a concealed process using sealed and numbered envelopes that allocate the patient to either of the two arms of the study. Patients may be randomised into this study only once, unless they were discharged from the hospital and were readmitted beyond 180 days of the first enrollment.

Twenty per cent mannitol (1098 mOsmol/kg) and 3.1% sodium chloride solution (1054 mOsmol/kg) are nearly equal in osmolality. This enables us to perform blind during the study. After enrollment of the patient, 125 mL 20% mannitol or 3.1% sodium chloride solution will be filled into a sterile 250 mL glass bottle by a pharmacist. Patients and all study personnel except the investigative pharmacist are blind to treatment assignment. The details of the series are unknown to any of the investigators and are contained in a set of opaque and sealed envelopes, each bearing on the outside only the number.

Data collection and trial intervention

At study entry, data on demography, body mass index, history of illness, diagnoses of the patients and the reason for use of hyperosmolar agents are collected. The surgical site, operation time, type and amount of hyperosmolar agents used during 24 h before the study drug infusion are recorded. Acute Physiology and Chronic Health Evaluation II score (APACHE II) is calculated. Venous blood sample (3 mL) is obtained and concentrations of serum glucose, triglyceride, cholesterase, albumin, globulins, total serum protein and blood urea nitrogen (BUN) are measured by standard central laboratory device.

All enrolled patients are randomised 1:1 to receive 125 mL of either 20% mannitol (M group) or 3.1% sodium chloride solution (HS group). Both fluids are infused via central venous line in 15 min by using an infusion pump. Vital signs, which include heart rate , respiratory rate, non-invasive BP and pulse oxygen saturation (SpO₂) are continuously monitored. Type of fluid intake, cumulative fluid intake, cumulative urine output and cumulative fluid balance are documented immediately before the infusion of study agents (T0), 15 min (T1), 30 min (T2), 60 min (T3), 120 min (T4), 240 min (T5) and 360 min (T6) after the start of infusion of experimental agents. At the same time points, 3 mL arterial blood sample and 10 mL urine sample are collected. Serum and urine osmolality are measured by means of freezing point depression.18 Blood values of sodium, potassium, glucose and lactate are measured using an ICU bedside blood gas analyser. Urine specific gravity and concentration of sodium are also measured by standard central laboratory devices.

Calculation of serum osmolality and osmole gap

We use four formulas to estimate serum osmolality19

[Na⁺], serum sodium concentration (mmol/L); [K⁺], serum potassium concentration (mmol/L); BG, blood glucose concentration (mmol/L); BUN, blood urea nitrogen concentration (mmol/L).

Osmole gap is calculated as the difference between the measured values and each of the estimated values shown above.

Study endpoints

The primary endpoint is the correlation of measured and estimated serum osmolality during the infusion of hyperosmolar agent. The differences between the measured and each of the calculated values from different formulas are compared. Other endpoints include changes in the following values during the infusion of experimental agents: serum and urine osmolality, serum and urine sodium concentration, urine specific gravity, and haemodynamic and fluid balance variables. Changes in haemodynamic and fluid balance variables will elucidate the influence of these two agents on status of circulation.

Current sample size justification

Primarily, we hypothesise that the peak serum osmole gap during HS infusion would be lower than that during mannitol infusion. Previous investigation showed that by using the formula 2 listed above, the peak serum osmole gap during mannitol infusion was 25 mOsmol/kg in patients with brain injury.20 It is expected that the serum peak osmole gap would decrease to 15 mOsmol/kg during the infusion of 3.1% sodium chloride solution. Using the Power and Sample Size Calculation program, we will need to enrol 15 patients in each group to be able to reject the null hypothesis that the mean peak serum osmole gap of the two experimental groups is equal with a probability (power) of 0.8. The type I error probability with testing this null hypothesis is 0.05.

Statistical analysis

All analyses will be according to the intention-to-treat principle, that is, all randomised patients will be analysed in the groups to which they were originally allocated and will be blinded to treatment assignment.

Categorical variables will be presented as numbers and percentages and analysed by the χ² test. Continuous variables will be checked for normal distribution and presented as mean and SD or median and IQR as appropriate. Comparison of continuous variables will be performed by using Student t test for normally distributed variables and the Mann-Whitney U test for non-normally distributed variables.

We will use Bland and Altman's23 limits of agreement analysis to clarify the accuracy of estimated serum osmolality calculated by each of the four formulas listed above. Bias is defined as the mean of the difference between measured and calculated values (measured minus calculated). Upper and lower limits of agreement are defined as bias±1.96 SD of the mean bias. The relationship between measured and calculated serum osmolality is also determined with linear regression analysis, and to visualise the results, calculated values are plotted against measured values.

We use repeated measures of analysis of variance for comparing serum osmole gap, serum and urine osmolality, serum and urine sodium concentration, urine specific gravity and haemodynamic and fluid balance variables across different time points (T0–T6) between the two groups (M and HS groups).

All tests of significance will be at the 5% significance level. Analyses are conducted using SPSS V.17.0.

Ethical aspects and informed consent

The trial complies with the latest Declaration of Helsinki. The study was registered on 16 January 2014 at the ClinicalTrials.org (NCT02037815).

Patients, within 24 h of craniotomy, are often unable to provide consent, so written informed consent is obtained from patients’ relatives. Immediately after a patient's admission, the ICU physician will introduce the family to the study coordinator. The physician will make sure the family knows the credentials of the study coordinator, and informs them that this person is going to discuss a research programme being conducted, and that this person is qualified to do so. The study coordinator will take the family to a place where they can talk confidentially. Every relevant aspect of the project will be described. The study coordinator will stop frequently, ask if there are any questions and request that the family to repeat in their own words what is being discussed, to make sure they understand.

The study coordinator will explain the following details:

1. The patient may experience brain oedema during the postoperative period, and hyperosmolar agents are needed in this situation.

2. Twenty per cent mannitol and 3.1% sodium chloride solution are standard of care in our institute.

3. According to scientific literatures, HS may be more effective than mannitol for the treatment of elevated ICP. However, there are no differences in clinical outcome and occurrence of side effects between these two agents.

4. Indications and side effects of mannitol and HS will be introduced in detail.

5. The objective of this study is to determine the accuracy of serum osmolality estimation during the application of mannitol and HS.

6. About 30 mL of blood sample will be obtained during the study.

The study coordinator will be especially careful to assure the family that they are free to decline consent without consequences and that they can withdraw consent at any time without impact on treatment. Family members will be provided with contact information of the study coordinator, local coinvestigator and the local ethical committee. Written consent will be obtained in the presence of a witness.

Dissemination plan

Results of the trial will be submitted to international peer-reviewed journals. Results will also be presented at national and international conferences relevant to subject fields.