Gastric residual volume measurement in British neonatal intensive care units: a survey of practice

Objective Despite little evidence, the practice of routine gastric residual volume (GRV) measurement to guide enteral feeding in neonatal units is widespread. Due to increased interest in this practice, and to examine trial feasibility, we aimed to determine enteral feeding and GRV measurement practices in British neonatal units. Design and setting An online survey was distributed via email to all neonatal units and networks in England, Scotland and Wales. A clinical nurse, senior doctor and dietitian were invited to collaboratively complete the survey and submit a copy of relevant guidelines. Results 95/184 (51.6%) approached units completed the survey, 81/95 (85.3%) reported having feeding guidelines and 28 guidelines were submitted for review. The majority of units used intermittent (90/95) gastric feeds as their primary feeding method. 42/95 units reported specific guidance for measuring and interpreting GRV. 20/90 units measured GRV before every feed, 39/90 at regular time intervals (most commonly four to six hourly 35/39) and 26/90 when felt to be clinically indicated. Most units reported uncertainty on the utility of aspirate volume for guiding feeding decisions; 13/90 reported that aspirate volume affected decisions ‘very much’. In contrast, aspirate colour was reported to affect decisions ‘very much’ by 37/90 of responding units. Almost half, 44/90, routinely returned aspirates to the stomach. Conclusions Routine GRV measurement is part of standard practice in British neonatal units, although there was inconsistency in how frequently to measure or how to interpret the aspirate. Volume was considered less important than colour of the aspirate.


GENERAL COMMENTS
I congratulate the authors for choosing a relevant topic and doing a reasonably good descriptive survey on the practice of gastric residual monitoring in the neonatal units in UK. Though there is no concrete evidence that gastric residual monitoring predicts and prevents NEC, it is widely practiced in many centres across the world. It is good to know the prevalence of such practice, which would help us to identify knowledge gaps and plan future clinical trials. There is not much existing literature on the prevalence of the practice of gastric residual monitoring, which makes this study interesting. The article needs some revisions before it can be considered for publication.

GENERAL COMMENTS
With this survey, the authors attempted to summarize current feeding practices concerning gastric residual volumes in preterm infants. The authors conducted this research to examine the feasibility of a future trial with a study arm in which clinicians do not measure gastric residual volumes. The conclusion statement suggests that a randomized trial with such characteristics is feasible. Although this reviewer may not agree on all the issues raised in the article, I praise the authors for their efforts on describing current feeding practices in the UK. Below I summarized my specific comments and suggestions.
-This article would have been more persuasive if the authors had related their findings with NEC outcomes of the units included in this survey. Specifically, it would have been interesting to know if units that do not currently measure GRVs have a lower incidence of NEC or culture-proven sepsis.
-The abstract summarizes well the results of the survey.
-The introduction section is balanced and well-written.
-A correlation analysis between survey responses of senior doctors and survey responses of nurses from the same unit would help validate the accuracy of their responses. In survey-based methodology, some suspect that senior doctors report what they think they do and nurses and other front-line clinicians report what they really do.
-In the discussion section, it would be helpful to make the distinction between practices that promote routine GRV measurements and practices that promote selective GRV measurements. The practice of not performing GRV measurements at all should be discussed separately.
-One aim of the survey was to define the control arm of a future trial in which GRVs are not measured. The characteristics of this potential control arm were not reported. Based on the study findings and the limited evidence available, it would seem reasonable to have a control arm with selective GRV measurements. In this group, GRVs would be routinely returned regardless of color to maintain a low gastric pH and minimize the risk of bacterial colonization.
-The proposed trial implies an effort to standardize the technique of gastric content aspiration among preterm infants. A clear efficacy endpoint of checking GRVs to prevent adverse outcomes of preterm infants should be stated to make the safety endpoint relevant. As currently written, the justification for a trial with safety as the primary outcome is weak.

VERSION 1 -AUTHOR RESPONSE
Reviewer 1: "I congratulate the authors for choosing a relevant topic and doing a reasonably good descriptive survey on the practice of gastric residual monitoring in the neonatal units in UK. Though there is no concrete evidence that gastric residual monitoring predicts and prevents NEC, it is widely practiced in many centres across the world. It is good to know the prevalence of such practice, which would help us to identify knowledge gaps and plan future clinical trials. There is not much existing literature on the prevalence of the practice of gastric residual monitoring, which makes this study interesting. The article needs some revisions before it can be considered for publication." Response  Major  1 Clinical trial registration details need to be given.

Reviewer 1 Comments
The study was not a clinical trial, it is a survey of practice. It is part of a larger feasibility study whose purpose is to design a future clinical trial. Please see http://grvstudy.com/ 2 19-item survey (supplementary material) is missing The survey was previously uploaded and has been uploaded again as a supplemental file titled 'NICU Gastric Survey Questions'. 3 Infant demographics need to be described. GR monitoring in which group of infants? Gestational age/ birth weight?
The study is a survey of practice, and the population of interest was all neonatal intensive care units in the UK. A sample size calculation was not carried out. The purpose of the study was not to draw inference, but to obtain descriptive statistics. We consider the response rate to have been excellent, given that the survey was voluntary, and that a team of clinicians was needed in order to complete the survey. The results would be limited if it were not representative of UK units in generalhowever, we have shown that the sample is reasonably representative of the three types of neonatal unit targeted. We include comments on this in our discussion. 6 Discussion: Other surveys addressing gastric residuals (for instance, the article with PMID: 22301544) should be compared and discussed.
The referenced article describes North American feeding practice more generally and is located behind a paywall. The abstract does not mention GRV. Our paper is intended to be relevant to the UK and describes practice there and the potential need for a trial in the UK. It could be interesting to describe other countries too but would take away from the focus of our study. Minor 1 Abstract: 'Gastric residual volume (GRV) measurement' can be changed to 'gastric residual (GR) monitoring'. Because, not only the volume but the quality of GR is also monitored.
Thank youthe term GRV was used in the wider GASTRIC feasibility study, and was the terminology used in our e-survey. We accept when GR is measured, the colour is also looked at, but would like to keep the wording unchanged. 2 Background: 'Gastric residual volume (GRV) measurement' can be changed to 'gastric residual (GR) monitoring' See previous response. 3 Background: The rationale of the study may better be written like this: "There are a few RCTs and a recent Cochrane meta-analysis (PMID 31425604) on no monitoring versus routine monitoring of gastric residual (references), which suggest that the available evidence is inadequate. Hence, we are planning to do a multicentric RCT on the same. This survey was conducted to…" Thank you. We feel that the rationale for the study has been explained. The word 'rationale' may have given confusion thoughwe have used it to talk about the rationale for GRV in current practice, rather than regarding the rationale for the study.
We thank the reviewer for highlighting the recent Cochrane review publication which has been included as reference 1 and referred to in the discussion as "Such a trial is a also supported by a recently published Cochrane review(1)".
The introduction has also been amended to make it clearer but we prefer our wording and prefer to retain it if possible. Low responses rates do not necessarily affect generalisability or validity. This would be the case if the sample was unrepresentative. In fact, we have a good representative sample. In the smaller units however, we only have 1/3 of units respondingthis is understandable given that smaller units have less resource to give time to responding to surveys. Again we prefer to retain our wording not least because the future trial would take place in the level 1 and 2 units who start and build up feeds in the infants of infants for the trial. 6 What the study adds: Points 2 & 3 need to be rephrased Thank you. We have re-worded this now.
Reviewer 2: "With this survey, the authors attempted to summarize current feeding practices concerning gastric residual volumes in preterm infants. The authors conducted this research to examine the feasibility of a future trial with a study arm in which clinicians do not measure gastric residual volumes. The conclusion statement suggests that a randomized trial with such characteristics is feasible. Although this reviewer may not agree on all the issues raised in the article, I praise the authors for their efforts on describing current feeding practices in the UK. Below I summarized my specific comments and suggestions." Response  Major  1 This article would have been more persuasive if the authors had related their findings with NEC outcomes of the units included in this survey. Specifically, it would have been interesting to know if units that do not currently measure GRVs have a lower incidence of NEC or culture-proven sepsis.

Reviewer 2 Comments
Thank youthis point is well made. The reason this additional information is not included here, is that the e-survey's specific objective was to describe practice. Drawing inference linking practice with outcomes is reserved for a future clinical trial. The e-survey presented in this paper forms part of a wider piece of work (the GASTRIC study) looking at the feasibility of a trial in the UK.

2
A correlation analysis between survey responses of senior doctors and survey responses of nurses from the same unit would help validate the accuracy of their responses. In survey-based methodology, some suspect that senior doctors report what they think they do and nurses and other front-line clinicians report what they really do.
Each unit completed a single survey, but this was completed together at the same time by a team of clinicians ideally including a doctor, a nurse and a dietitian (see Methods section). Therefore we are unable to look at responses split by clinician type. While we understand that opinions about the relevance of GRV will differ between types of clinician, the main aim here was to describe what current practice is to inform the potential future trial. 3 In the discussion section, it would be helpful to make the distinction between practices that promote routine GRV measurements and practices that promote selective GRV measurements. The practice of not performing GRV measurements at all should be discussed separately.
We compared the routine use of GRV measurement with not measuring as this was the brief from the funder and what we were interested in for a future trial. The discussion is therefore written with these two GRV approaches in mind and to add detail about selecting which infants to use it for would have been confusing. There is detail on this in the supplementary material but we do not agree that it should feature prominently in the discussion. 4 One aim of the survey was to define the control arm of a future trial in which GRVs are not measured. The characteristics of this potential control arm were not reported. Based on the study findings and the limited evidence available, it would seem reasonable to have a control arm with selective GRV measurements. In this group, GRVs would be routinely returned regardless of color to maintain a low gastric pH and minimize the risk of bacterial colonization.
Please see the previous answer. This work took place before a whole Delphi consensus process was undertaken with staff and parents which looked at which group of infants should be included in the proposed trial and this work is presented elsewhere.
We don't agree that the control arm should include selective measurements. The proposed future trial would compare no measurement with routine measurement as standard approaches. Routine measurement is the typical approach in UK units that use GRV to guide practice. It could also be very difficult to show a difference in terms of trial outcomes between a selective measurement approach and a nonmeasurement approach which would also be very difficult to operationalise in a trial. We believe this approach is therefore undesirable for the future trial and would be confusing if raised in the discussion. 5 The proposed trial implies an effort to standardize the technique of gastric content aspiration among preterm infants. A clear efficacy endpoint of checking GRVs to prevent adverse outcomes of preterm infants should be stated to make the safety endpoint relevant. As currently written, the justification for a trial with safety as the primary outcome is weak.
Our conclusion and the "What this study adds" section clearly define what we feel this particular study adds to the literature, especially the sentence that has been reworded for clarity and to now says; 'The heterogeneity of approaches regarding GRV measurement supports the need for a randomised trial to enable an evidencebased approach to the practice'. We do not propose what the trial should look like beyond the interventions as the work described elsewhere clarified what the trial should look like.
This is mentioned at the end of the methods section and will be referenced in the final, published version of the manuscript.

REVIEWER
Reviewer name: Thangaraj Abiramalatha Institution and Country: Department of Neonatology Sri Ramachandra Institute of Higher Education and Research, Chennai, India.