Noninterventional follow‐up vs fluid bolus in RESPONSE to oliguria—The RESPONSE trial protocol and statistical analysis plan

Background Oliguria is a frequent trigger for administering a fluid bolus, but the effect of fluid bolus in improving urine output is inadequately demonstrated. Here, we summarize the protocol and detailed statistical analysis plan of the randomized, controlled RESPONSE trial comparing follow‐up as the experimental group and a 500 mL crystalloid fluid bolus as the control group for oliguria in critically ill oliguric patients. Methods Our trial is an investigator‐initiated, randomized, controlled, pilot trial conducted in three ICUs in two centers. We aim to randomize 1:1 altogether 130 hemodynamically stable oliguric patients either to a 2‐hour follow‐up without interventions or to receive a crystalloid bolus of 500 mL over 30 minutes. The primary outcome is the change in individual urine output during the 2‐hour period compared to 2 hours preceding randomization. Doubling of the urine output is considered clinically significant. Additionally, we record the duration of oliguria, physiological and biochemical variables, adverse events, and the incidences of acute kidney injury and renal replacement therapy. Conclusions Oliguria is a frequent trigger for potentially harmful fluid loading. Therefore, the RESPONSE trial will give information of the potential effect of fluid bolus on oliguria in critically ill patients. Trial registration clinical.trials.gov, NCT02860572.


Funding information
The trial has received support from the Academy of Finland (317061) and Finnish Society of Intensive Care. STV has received a Fellowship grant from the Sigrid Juselius Foundation. The funders have no role in the trial design, conduction, or interpretation of the results.
Background: Oliguria is a frequent trigger for administering a fluid bolus, but the effect of fluid bolus in improving urine output is inadequately demonstrated. Here, we summarize the protocol and detailed statistical analysis plan of the randomized, controlled RESPONSE trial comparing follow-up as the experimental group and a 500 mL crystalloid fluid bolus as the control group for oliguria in critically ill oliguric patients.

Methods:
Our trial is an investigator-initiated, randomized, controlled, pilot trial conducted in three ICUs in two centers. We aim to randomize 1:1 altogether 130 hemodynamically stable oliguric patients either to a 2-hour follow-up without interventions or to receive a crystalloid bolus of 500 mL over 30 minutes. The primary outcome is the change in individual urine output during the 2-hour period compared to 2 hours preceding randomization. Doubling of the urine output is considered clinically significant. Additionally, we record the duration of oliguria, physiological and biochemical variables, adverse events, and the incidences of acute kidney injury and renal replacement therapy.
Conclusions: Oliguria is a frequent trigger for potentially harmful fluid loading. Therefore, the RESPONSE trial will give information of the potential effect of fluid bolus on oliguria in critically ill patients.
Trial registration: clinical.trials.gov, NCT02860572. acute respiratory distress syndrome in a randomized trial comparing conservative vs liberal fluid management found that urine output did not significantly increase within 4 hours after a fluid bolus. 5 A small observational study among septic shock patients did not find fluid boluses to increase urine output 6 whereas another study reported urine output to increase approximately by 50% both among patients with oliguria and AKI after a large fluid bolus (median volume 1224 mL crystalloid or colloid). 7 Additionally, improvement in systemic hemodynamics after fluid bolus has been found to correlate poorly with renal hemodynamics, 7-10 further complicating the assessment of potential benefits of fluid bolus in terms of reversing oliguria. Moreover, the effect of fluid bolus on cardiovascular variables is short-lived among patients in shock 11,12 suggesting that fluid boluses in the optimization phase of shock are unlikely to improve patient-centered outcomes.
A typically administered fluid bolus is 500 mL of crystalloid given over 30 minutes, 2 and fluid boluses may constitute a marked proportion, up to 50%, of the fluid amount administered daily. 6 A growing body of evidence suggests a poor renal response to fluid bolus, 5,6 which particularly among oliguric patients translates to an increased risk of fluid accumulation. Moreover, a growing body of evidence shows association of fluid accumulation with increased risk of mortality. [13][14][15][16] Therefore, in oliguric critically ill patients, clinical equipoise exists in conducting a randomized trial that aims to compare a follow-up approach without any fluid bolus to standard therapy with fluid bolus (500 mL of crystalloid administered over 30 minutes). We hypothesize that the incidence of doubling of urine output will not differ between the groups to a clinically significant level defined as at least a 20% absolute increase in the number of responders.

| Trial design and setting
The RESPONSE trial is an investigator-initiated, open, randomized, controlled, pilot study conducted in two Finnish centers; two intensive care units (ICUs) in Helsinki University Hospital (Helsinki) and in the ICU of Central Finland Central Hospital (Jyväskylä).

| Trial registration
The RESPONSE trial was registered in the clinicaltrials.gov registry (identifier NCT02860572) on 9 August 2016.

| Trial conduct
The study protocol has been prepared in accordance with the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines. 17 The study will be conducted according to the Declaration of Helsinki and its later amendments and according the Good Clinical Practice guidelines.

| Randomization
Eligible patients will be randomized 1:1 within 1 hour of full eligibility using computer-based algorithm created by an independent statistician and web-based allocation concealment for randomization.
Randomization is stratified according to the presence/absence of AKI (defined by KDIGO criteria 4 ) and sepsis (according to the Sepsis definition 18 ). Permutated blocks of varying size (4, 6, or 8) are used.

| Blinding
Due to the nature of the study, blinding of the ICU personnel is not feasible. The allocation is blinded for the statistician conducting the data analysis. Additionally, we will write an abstract of study results prior to becoming aware of the group allocation and will include this abstract in the appendix of the original publication. Use of randomization blocks of varying size not known for clinicians prevents knowledge of treatment allocation before randomization.

| Inclusion and exclusion criteria
All admissions to ICU will be screened. If the initial (static) inclusion or exclusion criteria (marked with asterisk) are not excluding the patient becoming eligible for randomization if oliguria develops, the patient will be followed further. If oliguria develops, the presence of dynamic exclusions will be rechecked, and if none of those is present, patient is eligible for randomization.

| Trial interventions
The timeline of eligibility, randomization, and study interventions are presented in Table 1. The intervention period lasts 2 hours.

| Intervention group (noninterventional followup)
Patients in the intervention group will not receive fluid bolus or diuretics to increase urine output during the 2-hour intervention period.

| Standard group (fluid bolus group)
Patients will receive 500 mL intravenous crystalloid (Ringer's acetate) infused over 30 minutes using an infusion pump.

| Concomitant treatment
In both groups, the infusion rate of all on-going infusions such as nutrition and maintenance fluids will be kept constant during the 2-hour period. Vasoactive drugs, short-acting insulin, sedation, and other medications can be titrated according to the discretion of the clinician. No diuretics during the 2-hour study period are allowed.
Modifying the targeted mean arterial pressure (MAP) and adjusting vasopressor dose accordingly (higher MAP potentially increasing urine output 19 ) are not recommended in either of the groups.
If severe hemodynamic instability develops during the 2-hour study period due to suspected hypovolemia (defined as a need to increase norepinephrine-infusion >0.2 µg/kg/min from baseline or heart rate increase >30 beats/min from baseline), a rescue bolus of 500 mL crystalloid over 30 minutes in both groups can be administered according to the decision of the treating clinician.

| Primary outcome measure
The primary outcome is defined the individual mean cumulative 2-hour urine output (mL/kg/h) 2 hours after randomization divided by the mean cumulative 2-hour urine output (mL/kg/h) measured 2 hours preceding randomization expressed as a percentage ( Figure 1). Based on a previous prospective cohort study, we consider doubling of the urine output as a clinically significant increase in urine output among oliguric patients, 20 if urine output increases at least 10 mL/h (ie difference between cumulative 2-hour urine output post-randomization vs pre-randomization is at least 20 mL). 21 The proportion of patients who doubled their urine output in both groups will be compared.
Choosing the 2-hour timeline for the endpoint was based on assumption that urine output in response to fluid bolus would increase in that time based on previous studies. [22][23][24] In studies among healthy volunteers, the median/mean time to micturition after a large crys-

| Data collection
Trained research personnel will perform data collection using an electronic platform (Absolute Imaginary Software Ltd). Data S1 comprises the detailed list of collected data items. In brief, patient characteristics at baseline will be collected, and details of urine output, hemodynamics, and fluid therapy will be recorded during the intervention and follow-up period. Patients' vital status will be recorded at hospital discharge or on day 30 whichever comes first.

| General analytical principles
Statistical analyses will be performed on the intention-to-treat (ITT) population defined as all randomized subject excluding (a) those without consent (b) ineligible subjects who were erroneously randomized and did not receive the trial intervention. The conclusions of the analysis will be based on the ITT analysis.
A sensitivity analysis will be conducted in the per-protocol population, defined as the ITT population and excluding subjects (a) who experienced one or several protocol violations or (b) who received the rescue bolus.
Continuous variables will be checked for normality using the Shapiro-Wilk test. We will set statistical significance to .05 and report two-sided P-values.

| Primary outcome
The primary outcome between follow-up and fluid bolus groups will be adjusted for the two stratification variables 25 (presence of AKI and presence of sepsis) using two-tailed logistic regression.
Obtained odds ratios (OR) will be converted to risk ratios (RR) with 95% confidence intervals (CI). Additionally, we will report crude event rates and risk ratios.

| Secondary and exploratory outcomes
As the primary outcome, the secondary and exploratory outcomes will be adjusted for the two stratification variables. 25 Continuous measures will be expressed according to their distribution and difference between the groups in the change in individual urine output, duration of consecutive oliguria, and cumulative fluid balance on study day will be compared using mean or median regression depending on normal distribution and expressed with 95% CIs.
Dichotomous endpoint measures (number of patients receiving rescue boluses, highest AKI stage within 24, 48 hours and during ICU stay, and number of patients receiving RRT) will be analyzed using ordinal or logistic regression and reported as risk ratios with 95% CIs.
Furthermore, crude event rates and risk rations will be reported for dichotomous outcomes and absolute differences with 95% CIs for continuous outcomes.

| Other variables
Continuous variables will be compared with Mann-Whitney U test or Student's t test depending on normal distribution, and categorical data using Chi-squared test.

| Missing data
We do not expect to have missing data about the primary outcome measure or stratification variables in the ITT population.
Regarding secondary or exploratory outcomes, some missingness F I G U R E 1 Schematic presentation of determination of urine output for the primary endpoint variable. UO, urine output is expected (for example CVP values in patients who have not a central line inserted), but imputation will not be used. Regarding other data points, if the amount of missing observations regarding other variables is less than 5%, we do not impute data. Regarding descriptive data, an appropriate multiple imputation strategy according to the type of missingness (missing at random or missing completely at random) will be used if missingness exceeds 5% of observations.

| Sample size
No reliable previous data to inform about the incidence of the primary outcome were available. Among oliguric (urine output <0.5 mL/ kg/h for at least 3 hours), hemodynamically stable patients, 50% were considered as renal responders to a fluid challenge that constituted at least one 500 mL crystalloid bolus administered over 15 minutes. 10 In that study the definition of a positive renal response (post-bolus urine output over 0.5 mL/kg/h for at least 3 hours) was more liberal than in our trial. 10 Fluid bolus did not significantly increase urine output among patients with acute respiratory distress syndrome (both oliguric and non-oliguric). 5 Therefore, we assume that 30% of patients in the fluid bolus group will be endpoint-positive. Additionally, we consider 20% absolute difference (which translates to a number needed to treat of 5) as a minimum clinically meaningful difference between the groups to support the use of such a common and inexpensive intervention. Thus, to reach a 20% absolute difference in the primary outcome (30% in the fluid bolus group and 10% in the noninterventional follow-up group), 62 patients per group will be required to reach 80% power with significance level set at .05 (two-sided). Considering possible drop-outs we aim to randomize 65 patients per group to target at least 62 patients per group in the ITT population.

| Pre-planned subgroup analyses
The primary and secondary outcome measures will be analyzed in Additionally, the recent pathophysiological hypotheses regarding oliguria in sepsis suggest that oliguria may be an adaptive mechanism to severe inflammation, 26 suggesting also a lower incidence of reversal of oliguria. The results obtained in the subgroups will be interpreted as hypothesis generating and heterogeneity between subgroups will be assessed by I 2 -test.

| Trial Profile
The flow of trial participants will be reported according to the CONSORT statement. 27

| Data monitoring and safety committee
As this is a low-risk pilot trial, no data and safety monitoring committee (DSMB) will be formed. Informed consents, inclusion and exclusion criteria, and collected data will be monitored.

| Interim analyses
No interim analyses will be conducted.

| D ISCUSS I ON
Albeit fluid bolus therapy is frequently used in oliguria to increase urine output, the success of this intervention is largely unknown.
The RESPONSE trial will give high-quality information about the The randomized approach in studying the response of fluid bolus therapy to an alternative intervention is an obvious strength of our study, as the current evidence is based on observational data.
Moreover, we are carefully recording data on other fluid input and output, common hemodynamic parameters, acid-base balance, and blood gases. Thus, we will be able to provide a detailed analysis on the effects of fluid bolus on these parameters as compared to the follow-up approach. Additionally, we will include a heterogeneous population of critically ill patients with oliguria who have received the initial fluid resuscitation, which will improve the generalizability of our results.
Our trial has some obvious limitations. First, the intervention is not blinded from the clinical team or the investigators due to its nature. Potentially, clinician aware of the patient being in the follow-up group may give a fluid bolus after the intervention period, and therefore neutralizing the difference between the groups regarding fluid balance. However, we record the details of urine output and fluid therapy until 6 hours post-randomization to assess the administration additional fluid boluses. Moreover, the trial statistician will be blinded to the group allocation. Second, our trial protocol does not mandate cardiac output monitoring, and therefore, commenting whether patients were hemodynamic fluid responders will not be possible. However, numerous studies have shown that global hemodynamics and renal hemodynamics do not correlate. 7,10 In conclusion, the RESPONSE trial will bring new evidence about the effects of fluid bolus therapy compared to noninterventional follow-up on urine output. These results will help to allocate fluid bolus therapy to those patients who benefit from it.

| E THI C AL CON S IDER ATI ON S AND CON S ENT TO PARTI CIPATE
The Ethics Committee of Department of Surgery, Hospital district of Helsinki and Uusimaa (HUS/1308/2016), approved the trial.
Because of the critically ill patient population and a time-sensitive standard group intervention, deferred consent was approved with an informed, written consent obtained from the patients' next of kin as soon as possible and confirmed from the patient if possible.
Patient will be asked to provide the consent prior to randomization if his/her clinical situation allows. Written informed consent is mandatory for all patients included in the final analysis.

| DATA S HARING S TATEMENT
The de-identified trial data set will be published as a supplement to the original publication.

| DISS EMINATION
Results of the study will be submitted to a peer-reviewed medical journal regardless of the results.

ACK N OWLED G M ENTS
We thank the study patients and their relatives for their willingness to participate in the trial. We warmly thank the clinical staff at participating ICUs and clinical research team members without whom the study would not be possible.

CO N FLI C T O F I NTE R E S T
The authors report no conflicts of interest.