International Journal of Nursing Studies 47 (2010) 10371046

Contents lists available at ScienceDirect

International Journal of Nursing Studies

journal homepage: www.elsevier.com/ijns

Review

Accuracy of biochemical markers for predicting nasogastric tube

placement in adultsA systematic review of diagnostic studies
Ritin S. Fernandez a,*, Janita Pak-Chun Chau b, David R. Thompson c,
Rhonda Grifths d, Hoi-Shan Lo b
a
Family and Community Health, College of Health & Science, University of Western Sydney, NSW, Australia
b
Nethersole School of Nursing, Chinese University of Hong Kong, Shatin, N.T., Hong Kong
c
Department of Health Sciences, University of Leicester, Leicester LE1 6TP, UK
d
School of Nursing & Midwifery, College of Health & Science, University of Western Sydney, NSW, Australia

A R T I C L E I N F O A B S T R A C T

Article history: Objective: The objective of this study was to investigate the diagnostic performance of
Received 11 November 2009 biochemical tests used to determine placement of nasogastric (NG) tubes after insertion in
Received in revised form 18 March 2010 adults.
Accepted 20 March 2010 Design: A systematic review of diagnostic studies was undertaken.
Data sources: A literature search of the bibliographic databases and the World Wide Web
Keywords: was performed to locate original diagnostic studies in English or Chinese on biochemical
Nasogastric tubes markers for detecting NG tube location.
Diagnostic accuracy
Review methods: Studies in which one or more different tests were evaluated with a
Systematic review
reference standard, and diagnostic values were reported or could be calculated were
included. Two reviewers independently checked all abstracts and full text studies for
inclusion criteria. Included studies were assessed for their quality using the QUADAS tool.
Study features and diagnostic values were extracted from the included studies.
Results: Of the 10 studies included in this review, seven investigated the diagnostic
accuracy of pH, one investigated the diagnostic accuracy of pH and bilirubin respectively,
two a combination of pH and bilirubin and one a combination of pH, pepsin and trypsin
levels in identifying NG tube location. All studies used X-rays as the reference standard for
comparison. Pooled results demonstrated that a pH of 4.0 had the ability to predict only
63% of the tubes located in the stomach. However, a pH value of 5.5 to determine gastric
placement demonstrated a sensitivity of 0.89 (95% CI 0.820.94) and a specicity of 0.87
(95% CI 0.810.93). Bilirubin coupled with pH had a high specicity (0.99) which
demonstrated the ability of the test to identify misplaced tubes in intestine. However, the
ability of the test to correctly identify gastric placement of feeding tubes was relatively low
(sensitivity <0.90).
Conclusions: Due to the heterogeneity of the studies and small sample sizes, conclusions
about the diagnostic performance of the different tests cannot be drawn. Better designed
studies exploring the accuracy of diagnostic tests are needed to improve the diagnostic policy.
Until stronger evidence becomes available practice related to the diagnostic tests used within
the clinical setting will continue to be dictated by local preferences and cost factors.
2010 Elsevier Ltd. All rights reserved.

What is already known about the topic?

 Nasogastric tubes are frequently used in the clinical

* Corresponding author. setting for assessment, nutritional support and medica-
E-mail address: [email protected] (R.S. Fernandez). tion administration.

0020-7489/$ see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijnurstu.2010.03.015
1038 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046

 pH is a method commonly used to predict the location of (i.e. pH <7), therefore, rendering litmus paper insuf-
nasogastric tubes in adults. ciently sensitive to distinguish between the pH of different
uids, specically between bronchial and gastric secre-
What this paper adds tions (Medicines and Healthcare products Regulatory
Agency (MHRA), 2004). Alternate methods for pH testing
 A pH value of 5.5 to determine has a high ability to include use of pH meters and pH indicator strips (Bradley
determine gastric placement of the nasogastric tube. et al., 1998). pH testing of aspirates has been considered
 A combination of pH and bilirubin levels has a higher more reliable than auscultation (Metheny et al., 1989)
ability to predict gastric placement. though both methods have failed to identify misplaced
tubes (Kearns and Donna, 2001). In contrast, gastric pH is
1. Background raised by the use of H2-blockers and proton-pump
inhibitors potentially producing false negative pH tests
Nasogastric (NG) tubes are frequently used in the resulting in delayed feeding (Taylor and Clemente, 2005).
clinical setting for the assessment and management of Assessment of bilirubin content of aspirate has also
patients who require decompression of the gastrointest- been proposed as a method to differentiate between
inal (GI) tract, nutritional support and medication admin- respiratory, gastric and intestinal placement of tubes,
istration (McClave, 2009). Insertion of the NG tube is a based on relative concentrations of acid and bilirubin
complex procedure and requires skills and expertise as (Metheny et al., 1999, 1997). Although various biochemical
placement errors could lead to potentially major compli- markers are used to determine NG tube placement, the
cations including misplacement into the respiratory tract, diagnostic accuracy of these methods has not yet been
brain, esophagus, peritoneum, stomach (duodenal tube) evaluated in a manner that will enable clinicians to make
and intestine (gastric tube) (Chen et al., 2006; Leder and an informed decision about the benets of the various
Suiter, 2008; Weinberg and Skewes, 2006; Wu et al., 2006). methods. The aim of this systematic review therefore, is to
Researchers have reported that in adults, tube placement present the best available evidence concerning the
errors vary from 1.3 to 50% (Burns et al., 2001). diagnostic accuracy of biochemical markers for determin-
ing NG tube placement after insertion in adults.
1.1. Methods to assess placement of tubes The specic questions to be answered were:

Various bedside methods have been used either 1. What is the diagnostic accuracy of the pH method to
individually or in combination to assess NG tube place- differentiate gastric from intestinal tube placement?
ment. These include observing for cough and choking, 2. What is the diagnostic accuracy of bilirubin measure-
auscultation of air insufated through the tube, aspiration ment in aspirates to differentiate gastric from intestinal
of uid (Rakel et al., 1994), visual inspection of the tube placement?
aspirates (Metheny et al., 1994), testing of aspirates for pH 3. What is the diagnostic accuracy of pH and bilirubin
or concentrations of bilirubin, pepsin or trypsin (Metheny measurement in aspirates to differentiate respiratory
et al., 1997), observing for bubbling when the tip of the from GI tube placement?
tube is held under water, testing the ability to speak, the 4. What is the diagnostic accuracy of pH and bilirubin
use of magnetic detection (Tobin et al., 2000), spring measurement in aspirates to differentiate gastric from
gauge pressure manometer (Swiech et al., 1994), capno- intestinal tube placement?
graphy, colorimetric capnometry or radiography (Howes 5. What is the diagnostic accuracy of pH, pepsin and
et al., 2005; Kindopp et al., 2001). Observing for the trypsin measurement in aspirates to differentiate
presence of coughing and choking to differentiate respiratory from GI tube placement?
between gastric and respiratory tube placement is not 6. What is the diagnostic accuracy of pH, pepsin and
applicable in patients with a decreased level of con- trypsin measurement in aspirates to differentiate gastric
sciousness or with neurological debilitation where gag from intestinal tube placement?
and cough reexes may be suppressed (Metheny et al.,
2000). Radiography has been recommended and is widely
2. Methods
accepted as the reference standard to determine tube
location in particular following initial placement of the 2.1. Search methods
tube. However repeated radiographic conrmation is not
practical as it interferes with the feeding regimen and In consultation with the university librarian, MEDLINE
patients are exposed to increased risk of radiation related (1950 to Sept 2008), CINAHL (1982 to Sept 2008), EMBASE
illnesses (Hart, 2006). (1980 to Sept 2008), and All EBM Reviews (up to 3rd
The testing of aspirate for pH level is a commonly used Quarter 2008) were searched. The key terms used were
bedside method to assess tube placement. Traditionally nasogastric tube, feeding tube, intubation gastrointestinal,
this was done by aspirating the contents of the tube once stomach tube, nasal tube, feeding tube placement, feeding
inserted and testing the aspirate with litmus paper. In the tube location. The Chinese databases searched included
presence of acid the litmus paper turned from blue to red. WanFang Data (1998 to Sept 2008), China Journal Net
However, the UK Medical and Healthcare products (1994 to Sept 2008), Chinese Medical Current Contents
Regulatory Agency reported that blue litmus paper turned (1994 to Sept 2008), Index to Chinese Periodical Literature
pink in the presence of acid regardless of the level of acidity (1970 to Sept 2008) and the Chinese Biomedical Literature
 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046 1039

Database (1980 to Sept 2008). Chinese search terms were this review. In studies that did not report diagnostic
based on the terminology used in Taiwan and China. outcomes, a two by two truth table was reconstructed for
The reference lists and bibliographies of retrieved each study, and sensitivity, specicity, positive predictive
articles were reviewed to identify any additional research. value (PPV), negative predictive value (NPV), LR+ and LR
To complement the search strategies keyword searching of were calculated for each study if sufcient data were
the World Wide Web was conducted. presented. Attempts were made to obtain missing data
from the trial by contacting the authors.
2.2. Selection of trials When possible, diagnostic values of the tests were
statistically pooled. The heterogeneity of sensitivities and
All clinical trials that investigated the diagnostic specicities were tested using Chi-square tests. In the
accuracy of biochemical markers in detecting NG tube absence of heterogeneity a xed effects model was used
placement in adults in the hospital setting were included (p > 0.05). A summary receiver operator characteristics
in the review. Reports in English and Chinese were (SROC) curve was constructed to determine the relation-
considered in this review. Trials were included if an ship between sensitivity and specicity and the area under
explicit denition of the reference standard, and if the the SROC curve which measured the usefulness of a test
diagnostic values (sensitivity, specicity, predictive values, was calculated. All calculations were made using the SPSS
and accuracy) were reported or could be calculated. version 15.0 and Meta-DiSc ver 1.4 (Zamora et al., 2006).
Two reviewers independently read all abstracts, and
full text was retrieved of articles that could not be excluded
3. Results
based on title and abstract. These full text articles were
read and checked for inclusion by two persons indepen- 3.1. Description of studies
dently. Trials that were reported in more than one
publication were included only once. Decision for study A total of 316 publications from the electronic database
eligibility was made by both reviewers and any disagree- search (English n = 264, Chinese n = 52) and a further 21
ment was resolved by consensus. trials following a review of the reference lists and
bibliographies were identied for potential inclusion in
2.3. Quality assessment the review. The majority were excluded as they did not
meet the selection criteria leaving 10 for inclusion in the
To assess the articles for their quality, a modied review. A detailed description of the selection process is
version (13 items) of the Quality assessment for described in Fig. 1.
diagnostic accuracy studies (QUADAS) (Whiting et al., All 10 trials included in the review were conducted in
2003) tool was used. The original QUADAS tool is a 14-item acute care facilities in the USA. Seven of these studies were
questionnaire, with eight items covering bias, one covering conducted by the same author. The number of patients in
variability, and ve on reporting. An item in the original these studies ranged from 36 (Conner and Carver, 2005) to
version of QUADAS, were the same clinical data available 890 (Metheny et al., 1997). In all trials the NG tubes were
when test results were interpreted as would be available inserted for feeding purposes. The size of the NG tubes
when the test is used in practice? was omitted as used in the trials ranged from 8 Fr to 16 Fr (Metheny and
the interpretation of the index test in this review involved Stewart, 2002) and the material they were made of was
no interpretation of other clinical data. Two reviewers either polyurethane or polyvinyl chloride. The key
independently scored the included articles; a nal score characteristics of the included studies were summarized
was made by agreement. in Table 1.

2.4. Data extraction and analysis 3.2. Methodological quality of included trials

Study characteristics of the included studies were Trials published in Chinese did not offer sufcient detail
extracted using a data extraction form. One reviewer to be able to determine issues such as the validity of the
extracted details from the study and the second reviewer biochemical measures used, accuracy, stability and test
checked the data extraction for accuracy. In order to gain retest reliability, information essential to assess the rigour
insight in the diagnostic accuracy, we focused on the of these trials and hence assist in clinical decision making.
sensitivity, specicity, and the negative and positive like- Thus, all articles published in Chinese were excluded. Of
lihood ratio of the test (Sackett et al., 2000). In this review, the 10 trials published in English and included in the
sensitivity refers to the ability of the index tests to identify review, the quality of the research studies was judged to be
correct placement of the NG tube and specicity refers to the satisfactory though the methods of sample selection and
ability of the index tests to identify misplaced tubes. the inclusion or exclusion criteria were poorly described.
The proportion of those who tested positive who were Sensitivity and specicity of the diagnostic tests should be
true positives (positive predictive value) and the propor- reported in diagnostic studies to enable the readers to
tion of those who tested negative who were true negatives make clinical decisions. However, only four (Kearns and
(negative predictive value) were also used to assess the Donna, 2001; Metheny et al., 2000, 1997; Phang et al.,
feasibility of these tests (Portney and Watkins, 2000). In 2004) of the 10 trials reported the sensitivity and
addition, the positive and negative likelihood ratios (LR) specicity of the index tests compared with reference
(Simel et al., 1991) were also calculated and presented in standards. The pH concentration was determined by either
1040 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046

Fig. 1. Flowchart of selection of publications.

pH-meter or pH test-strips and the bilirubin contents were 3.3.1. Accuracy of a combination of pH and bilirubin
measured by either laboratory assay (Metheny et al., 2000, Two trials determined the sensitivity and specicity of a
1999; Metheny and Stewart, 2002), urinary bilirubin test- combination of pH >5 and a bilirubin <5 mg/dl in
strip (Metheny et al., 2000; Metheny and Stewart, 2002) or differentiating between respiratory and GI tube placement
colorimetric visual bilirubin scale (Metheny et al., 1999; involving 587 and 856 observations respectively (Metheny
Metheny and Stewart, 2002). The pepsin and trypsin et al., 2000, 1999). The respiratory samples were aspirated
concentrations in aspirates were measured by laboratory from tubes inadvertently placed into the lung, from
assay (Metheny et al., 1997). tracheobronchial secretions obtained by suctioning
Ten trials reported on the use of biochemical measure- patients with articial airways, and pleural uid samples
ments using various cut-off points in differentiating obtained at the time of thoracentesis. The pooled ndings
between respiratory, gastric and intestinal feeding tube from these two trials demonstrated a sensitivity of 1.00
placement. The biochemical measurements investigated (95% CI 0.991.00) and specicity of 0.82 (95% CI 0.80
included pH (Conner and Carver, 2005; Kearns and Donna, 0.84) which indicated high predictive ability of the test to
2001; Metheny et al., 1993, 1989, 2005; Metheny and correctly identify tubes placed in the respiratory tract.
Stewart, 2002; Phang et al., 2004), bilirubin (Metheny and
Stewart, 2002), a combination of pH and bilirubin 3.3.2. Accuracy of a combination of pH, pepsin and trypsin
(Metheny et al., 2000, 1999) and a combination of pH, Only one study (Metheny et al., 1997) (343 NG, 399
pepsin and trypsin levels (Metheny et al., 1997). nasointestinal and 148 respiratory samples) investigated
the diagnostic accuracy of a combination of pH, pepsin and
3.3. Diagnostic accuracy of biochemical markers to trypsin values in predicting feeding tube position. The
differentiate between respiratory and GI tube placement mean pH level in the lung was 7.89 (SD = 0.46). The pH
level was signicantly higher than in the stomach
Three trials (Metheny et al., 1993, 2000, 1999, 1997) (mean = 4.06, SD = 2.09). The mean trypsin concentration
investigated various biochemical measurement para- in the lung (mean = 1.4 mg/ml, SD = 4.8) was signicantly
meters of feeding tube aspirates to differentiate between lower than in the intestine (mean = 143.0 mg/ml,
respiratory and GI placement of feeding tubes. Two trials SD = 133.7). Similarly, the mean pepsin concentration in
used a combination of pH and bilirubin (Metheny et al., the lung (mean = 3.2 mg/ml, SD = 10.2) was signicantly
2000, 1999) and one of pH, pepsin and trypsin (Metheny lower than in the stomach (mean = 349.1 mg/ml,
et al., 1997). SD = 268.6). The prediction criterion for lung placement
 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046 1041

Table 1
Study characteristics (n = 10).

Authors Year Sample setting Biochemical marker Tube type

Country Reference standard

Conner (2005) US 36 ICU patients (with 51 pH levels of 3.9, 5.9, 7.9 Small-bore feeding tubes inserted
feeding tube insertions) to differentiate between gastric using the 2-step radiological method,
admitted to ICU and intestinal placement or under uoroscopy.
Reference standard X-ray
Kearns (2001) US 134 patients (365 observations) pH 4.0 to determine 8 Fr or 12 Fr Tube Position Verier (TPV)
admitted to the wards and gastrointestinal placement feeding tube inserted under direct
critical care units of 4 acute visualization or using a bedside technique.
care hospitals Reference standard X-ray
Metheny (2002) US 80 patients (80 feeding tube pH 6, bilirubin <5 mg/dl to 8 Fr or 10 Fr Polyurethane NG tubes;
aspirates) admitted to 2 acute differentiate between gastric 14 Fr or 16 Fr Polyvinyl chloride NG tubes;
care medical centers and intestinal placement 8 Fr or 10 Fr Polyurethane nasointestinal
(NI) tubes.
Reference standard X-ray
Metheny (2000) US 656 patients (with 856 specimen Combination of pH and bilirubin 8 Fr or 10 Fr Dobbhoff or Entriex tube
ratings) and 52 registered nurses to differentiate between (small-bore NG or NI). Tubes inserted
from adult acute care facilities respiratory, gastric and blindly at the bedside or uoroscopically
intestinal placement in the radiology department or at
the bedside.
Reference standard X-ray
Metheny (1993) US 605 patients (with 794 aspirates/ pH 4.0, 6.0 to differentiate 8 Fr or 10 Fr small-bore polyurethane
pH-meter readings) admitted to between gastric and intestinal Entriex or Dobbhoff tube (NG or NI tube)
6 acute care hospitals placement Tubes inserted blind at the bedside or
with uoroscopic guidance.
Reference standard X-ray/Fluroscopy
Metheny (1997) US 890 patients admitted to adult Combination of pH, pepsin and 8 Fr or 10 Fr small-bore Dobbhoff or Entriex
acute care facilities trypsin to identify respiratory, tube (NG or NI tube). Tubes inserted blindly
gastric and intestinal placement at the bedside or with uoroscopic guidance
in the radiology suites or at bedside.
Reference standard X-ray
Metheny (1999) US 587 aspirates from patients Combination of pH and bilirubin 8 Fr or 10 Fr small-bore Dobbhoff or Entriex
admitted to adult acute care to differentiate between tube (NG or NI tube). Tubes inserted blindly
facilities respiratory, gastric and at the bedside or with uoroscopic guidance
intestinal placement. in the radiology suites or at the bedside.
Reference standard X-ray
Metheny (1989) US 181 patients (with 247 pH-paper pH 4.0, 5.5 to differentiate 8 Fr Dobbhoff, 10 Fr Entriex feeding tube.
readings) admitted to 4 acute care between gastric and intestinal Tubes inserted at the bedside or with
hospitals placement uoroscopic guidance.
Reference standard X-ray/Fluroscopy
Metheny (2005) US 201 critically ill patients admitted pH 6.0 testing to differentiate 10 Fr small-bore feeding tube (NG or NI).
to critical care units between gastric and intestinal Blind insertion of tubes.
placement Reference standard X-ray
Phang (2004) US 100 critically ill, ventilator-supported pH 4.0, 6.0, 6.5, 7.0 to 8 Fr micro-bore polyurethane feeding tube
patients (with 181 aspirates) differentiate between gastric (weighted and with stylet). Tubes blindly
and intestinal placement inserted at the bedside.
Reference standard Auscultation of
insufated air and uroscopy

(pH >6, pepsin <100 mg/ml, trypsin 30 mg/ml) was 2005), 6.0 (Metheny et al., 1989, 2005; Metheny and
successful in determining all respiratory samples (n = 148) Stewart, 2002; Phang et al., 2004), 6.5 (Phang et al., 2004),
(sensitivity 1.00, 95% CI 0.981.00; specicity 0.93, 95% CI 7.0 (Phang et al., 2004) and 7.9 (Conner and Carver,
0.910.95; LR+ 14.96, 95% CI 11.4319.58; LR 0.004, 95% 2005). The pH value is directly inuenced by the presence
CI 0.000.06). of acid inhibiting medications, therefore subgroup ana-
lyses have been presented for each pH value according to
3.4. Diagnostic accuracy of biochemical markers to patients who received acid inhibiting medications and
differentiate between gastric and intestinal tube placement those who did not. One trial (Conner and Carver, 2005)
(n = 100) that investigated the effect of a pH value 3.9 in
3.4.1. Accuracy of pH measurement determining feeding tube location reported that none of
Seven studies investigated the effect of pH values in the specimens had this pH value.
differentiating gastric from intestinal placement and
conrming misplacement of feeding tubes using various 3.4.1.1. pH 4.0 as cut-off point. Four trials (Kearns and
pH cut-off points. The pH cut-off points used were 3.9 Donna, 2001; Metheny et al., 1993, 1989; Phang et al.,
(Conner and Carver, 2005), 4.0 (Kearns and Donna, 2001; 2004) used a pH 4.0 as cut-off point for locating tube
Metheny et al., 1993, 1989; Phang et al., 2004) 5.5 position, however only three could be combined in a meta-
(Metheny et al., 1993, 1989), 5.9 (Conner and Carver, analysis. Meta-analysis of three trials (Metheny et al.,
 1042
Table 2
Diagnostic accuracy of biochemical markers.

Pooled data

Sensitivity (95% CI) Specicity (95% CI) LR+ (95% CI) LR (95% CI) SROC

Differentiating between respiratory and GI placement

pH and bilirubin n = 2 trials (Metheny et al., 2000, 1999)
pH >5 and bilirubin <5 mg/dl Respiratory placement 1.00 (0.991.00) 0.82 (0.800.84) 5.62 (2.8910.92) 0.003 (0.000.02) NA

R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046

pH, pepsin and trypsin n = 1 trial (Metheny et al., 1997)
pH >6, pepsin <100 mg/ml, trypsin 30 mg/ml Respiratory placement 1.00 (0.981.00) 0.93 (0.910.95) 14.96 (11.4319.58) 0.004 (0.000.06) NA

Differentiating between gastric and intestinal placement

pH 4.0 as cut-off point n = 3 trials (Metheny et al., 1993, 1989; Phang et al., 2004)
Not receiving acid inhibitors Gastric placement 0.75 (0.700.80) 0.94 (0.910.97) 13.48 (8.3321.81) 0.26 (0.200.35) 0.94
Receiving acid inhibitors Gastric placement 0.55 (0.500.61) 0.95 (0.920.97) 11.45 (6.8419.16) 0.44 (0.290.67) 0.91
All patients Gastric placement 0.63 (0.590.67) 0.95 (0.930.97) 12.62 (8.8717.96) 0.38 (0.220.64) 0.93

pH 5.5 as cut-off point n = 1 trial (Metheny et al., 1989)

Not receiving acid inhibitors Gastric placement 0.87 (0.770.94) 0.88 (0.780.94) 7.28 (3.9113.53) 0.14 (0.080.28) NA
Receiving acid inhibitors Gastric placement 0.91 (0.810.97) 0.87 (0.740.94) 6.78 (3.3913.56) 0.10 (0.040.24) NA
All patients Gastric placement 0.89 (0.820.94) 0.87 (0.810.93) 7.08 (4.4611.24) 0.12 (0.070.21) NA

pH 5.9 or 6.0 as cut-off point n = 5 trials (Conner and Carver, 2005; Metheny et al., 1993, 2005; Metheny and Stewart, 2002; Phang et al., 2004)
All patients Gastric placement 0.53 (0.500.56) 0.82 (0.790.83) 3.53 (1.597.83) 0.39 (0.190.83) 0.82

pH 6.5 as cut-off point n = 1 trial (Phang et al., 2004)

Not receiving acid inhibitors Gastric placement 0.73 (0.450.92) 0.87 (0.600.98) 5.50 (1.4620.71) 0.31 (0.130.73) NA
Receiving acid inhibitors Gastric placement 0.66 (0.530.77) 0.90 (0.800.96) 6.29 (3.0512.94) 0.38 (0.270.54) NA
All patients Gastric placement 0.67 (0.560.77) 0.89 (0.800.95) 6.11 (3.2411.53) 0.37 (0.270.51) NA

pH 7.0 as cut-off point n = 1 trial (Phang et al., 2004)

Not receiving acid inhibitors Gastric placement 0.73 (0.450.92) 0.47 (0.210.73) 1.38 (0.782.42) 0.57 (0.211.55) NA
Receiving acid inhibitors Gastric placement 0.73 (0.610.83) 0.67 (0.550.78) 2.23 (4.2240.02) 0.40 (0.260.61) NA
All patients Gastric placement 0.73 (0.620.82) 0.63 (0.520.74) 2.00 (1.462.74) 0.42 (0.290.63) NA

pH 7.9 as cut-off point n = 1 trial (Conner and Carver, 2005)

All patients Gastric placement 1.00 (0.931.00) 0.09 (0.030.21) 1.10 (1.001.21) 0.10 (0.011.80) NA

Bilirubin alone n = 1 trial (Metheny and Stewart, 2002)

Bilirubin <5 mg/dl Gastric placement 0.96 (0.881.00) 0.84 (0.640.96) 6.02 (2.4514.81) 0.04 (0.010.17) NA

pH and bilirubin n = 2 trials (Metheny et al., 2000, 1999)

pH 5 and bilirubin <5 mg/dl Gastric placement 0.88 (0.840.91) 0.99 (0.970.99)

pH, pepsin and trypsin n = 1 trial (Metheny et al., 1997)

pH 6, pepsin 100 mg/ml, trypsin 30 mg/ml Gastric placement 0.91 (0.880.94) 0.91 (0.880.94) 10.71 (7.7514.79) 0.10 (0.070.14) NA
 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046 1043

1993, 1989; Phang et al., 2004) (NG = 607, nasointest- patients who received acid inhibitors compared to 87%
inal = 598) demonstrated a pooled sensitivity of 0.63 (95% among those who did not (Table 2).
CI 0.590.67), and a pooled specicity of 0.95 (95% CI 0.93
0.97). This result demonstrated that a pH of 4.0 had the 3.4.1.3. pH 6.0 as cut-off point. One trial (Conner and
ability to predict only 63% of the tubes located in the Carver, 2005) (n = 100) investigated the effect of a pH value
stomach. The pooled LR+ was 12.62 (95% CI 8.8717.96) 5.9 and four trials (Metheny et al., 1993, 2005; Metheny
and the pooled LR was 0.38 (95% CI 0.220.64) (Table 2). and Stewart, 2002; Phang et al., 2004) (n = 2919) assessed
The SROC curve was constructed using random effects the effect of a pH value 6.0 in determining feeding tube
model due to the signicant heterogeneity. The area under placement. The pooled ndings from these ve trials
the SROC curve was 0.93 (Table 2). Findings from the demonstrated a sensitivity of 0.53 (95% CI 0.500.56) and
fourth trial (Kearns and Donna, 2001) that used a pH of specicity of 0.82 (95% CI 0.790.83) which indicates poor
4.0 (value conrmed by the authors through personal predictive ability of the test to correctly identify tubes
communication) indicated that this pH value was able to placed in the stomach (53%).
accurately identify the location of only 56% of all NG A subgroup analysis based on patients not receiving
feeding tubes when compared with the reference standard acid inhibitors (Metheny et al., 1993; Phang et al., 2004)
radiography. Of the 11 tubes that were misplaced, pH (n = 412 observations) demonstrated that 7389% of the
testing correctly identied one (9%), incorrectly identied gastric tubes were correctly identied as being placed in
two (18%), and was unable to determine eight (73%) mainly the stomach and 8687% of the tubes placed in the
because no uid could be obtained if the tube location was intestine were correctly identied (Table 2).
above the diaphragm. The sensitivity of the pH test to Two trials (Metheny et al., 1993; Phang et al., 2004)
identify misplaced tubes was 0.82, specicity was 0.55, LR+ (n = 542) investigated this outcome in patients receiving
was 1.8 (95% CI 1.42.4), and LR was 0.3 (95% CI 0.11.0). some form of acid inhibitors. The results demonstrated
A subgroup analysis was undertaken based on patients that 5882% of the gastric tubes were correctly identied
who received acid inhibitors and those who did not. The as being placed in the stomach and 8896% of the tubes
results demonstrated that for those patients not receiving placed in the intestine were correctly identied (Table 2).
any acid inhibitors 52/63 (82.5%) (Metheny et al., 1989),
136/185 (73.5%) (Metheny et al., 1993) and 10/15 (66.7%) 3.4.1.4. pH 6.5 as cut-off point. One trial (n = 164) (Phang
(Phang et al., 2004) of the gastric tubes were correctly et al., 2004) assessed the effect of the pH cut-off value of
identied as being placed in the stomach. The pooled 6.5 to determine gastric placement. The sensitivity of this
sensitivity was 0.75 (95% CI 0.700.80) and the pooled value was 0.67 (95% CI 0.560.77) and the specicity was
specicity was 0.94 (95% CI 0.910.97) (Table 2). This 0.89 (95% CI 0.800.95). A subgroup analysis demonstrated
nding demonstrates that a pH 4.0 correctly predicts 75% that the pH value of 6.5 detected 73% of gastric placement
of the tubes located in the stomach and 94% of the tubes among patients who did not receive acid inhibitors
located in the intestine in those patients not receiving any compared to those 66% among those who received acid
acid inhibitors. However it should be noted that in two inhibitors (Table 2).
studies (Metheny et al., 1993, 1989) the patients were
fasting while in the third study the fasting status of the 3.4.1.5. pH 7.0 as cut-off point. In the only trial (Phang
patients was unclear which could inuence the results. et al., 2004) that used a pH value of 7.0 as a predictor of
For those patients receiving acid inhibitors a pH 4.0 tube placement, the sensitivity and specicity were 0.73
correctly identied 45/57 (78.9%) (Metheny et al., 1989), and 0.63 respectively. A subgroup analysis demonstrated
121/219 (55.3%) (Metheny et al., 1993), and 24/67 (35.8%) that for patients receiving acid inhibitors the sensitivity of
(Phang et al., 2004) of the gastric tubes as being placed in the pH value of 7.0 was 0.73 (95% CI 0.610.83) and the
the stomach. The pooled sensitivity was 0.55 (95% CI 0.50 specicity 0.67 (95% CI 0.550.78). For patients not
0.61) and the pooled specicity was 0.95 (95% CI 0.92 receiving acid inhibitors, the sensitivity of this pH value
0.97) (Table 2), which demonstrates that for patients was 0.73 (95% CI 0.450.92) and specicity 0.47 (95% CI
receiving acid inhibitors a pH level of 4.0 accurately 0.210.73) (Table 2).
predicts only 55% of the tubes placed in the stomach.
However this pH level predicts 95% of the tubes not located 3.4.1.6. pH 7.9 as cut-off point. One trial (Conner and
within the stomach. What this nding indicates is that the Carver, 2005) involving 100 patients investigated the
pH value 4.0 has a lower predictive rate in patients who predictability of pH 7.9 in locating feeding tube place-
receive acid inhibitors compared to those who do not. ment. The ndings demonstrated a sensitivity of 1.00 (95%
CI 0.931.00) which indicates that all tubes placed in the
3.4.1.2. pH 5.5 as cut-off point. One trial (n = 247) stomach were correctly identied. The specicity was 0.09
(Metheny et al., 1989) that assessed the accuracy of the (95% CI 0.030.21) for this pH value.
pH value of 5.5 to determine gastric placement demon-
strated a sensitivity of 0.89 (95% CI 0.820.94) and a 3.4.2. Accuracy of bilirubin
specicity of 0.87 (95% CI 0.810.93). Although the One study (Metheny and Stewart, 2002) investigated
detection rates are high, this pH value does not meet the effect of bilirubin in differentiating between gastric
the reference standard of radiography. A subgroup analysis (n = 55) and intestinal (n = 25) placement in patients who
demonstrated that the pH value of 5.5 was able to were not fasting. The mean bilirubin concentration was
correctly predict 91% of the gastric placement among higher in intestinal aspirates (7.9  0.9 mg/dl) than in
1044 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046

gastric aspirates (0.4  0.2 mg/dl) (p < 0.001). Ninety-six should be noted that pH testing of aspirates was under-
percent of the gastric aspirates (53/55) had bilirubin values taken using two different types of instruments namely the
<5 mg/dl; and 84% (21/25) of the intestinal aspirates had colour coded pH Indicator Strips or a portable pH-meter.
bilirubin values 5 mg/dl. The sensitivity of the bilirubin However in three studies (Metheny et al., 1989, 1993;
level of <5 mg/dl in predicting gastric placement was 0.96 Metheny and Stewart, 2002), concurrent validity of the
(95% CI 0.881.00) and the specicity was 0.84 (95% CI 0.64 two instruments was undertaken which demonstrated
0.96). high correlation between the pH-paper and pH-meter
readings for both gastric and intestinal samples. Mis-
3.4.3. Accuracy of a combination of pH and bilirubin interpretation of the colour of pH strips has been reported
Two trials (Metheny et al., 2000, 1999) reported on the to be common, particularly differentiating between values
ability of a combination of pH 5 and bilirubin <5 mg/dl to 4 and 6 on the pH-paper as opposed to 3-pad pH sticks
predict the gastric placement of feeding tubes (gastric (Taylor and Clemente, 2005). Operator colour deciency
n = 415 and intestinal n = 439). The combination method was highlighted as a further potential reason for this
was able to predict 98.3% (Metheny et al., 2000) and 98.6% misinterpretation. Further renement of pH test-strip or
(Metheny et al., 1999) of the cases as the gastric placement. other means such as portable pH-meter would be
The pooled sensitivity and specicity were 0.88 (95% CI benecial for accurately measuring pH values of feeding
0.840.91) and 0.99 (95% CI 0.970.99) respectively tube aspirates. These ndings should be considered in light
(Table 2). of the fact that in some studies, patients received acid
inhibiting medications that could potentially increase the
3.4.4. Accuracy of a combination of pH, pepsin and trypsin alkalinity of the aspirates. In addition, in some studies the
One study (Metheny et al., 1997) (343 NG tubes and 399 fasting status of the patients was unclear, while in others,
nasointestinal tubes) investigated the effectiveness of patients received continuous feeding which could result in
using pH 6, pepsin 100 mg/ml, trypsin 30 mg/ml the dilution of gastric acid and may inuence the ndings.
values in predicting gastric placement of the feeding tube. The use of bilirubin measurement in differentiating
The ndings demonstrated that 91.2% of gastric and 91.5% between gastric and intestinal tube placement demon-
of intestinal cases were correctly classied which demon- strated high sensitivity. However, this nding should be
strates a sensitivity of 0.91 (95% CI 0.880.94), specicity of interpreted with caution given the small sample size.
0.91 (95% CI 0.880.94), of the test to predict gastric tube Combining low pH (5) and bilirubin (<5 mg/dl)
placement. The LR+ was 10.71 (95% CI 7.7514.79) which (Metheny et al., 2000, 1999) or a combination of pH,
indicate that there is a large and conclusive increase in the pepsin and trypsin (Metheny et al., 1997) improves
likelihood that the tube is in the gastric region. This nding prediction accuracy. The combination of pH and bilirubin
was despite the fact that more than half of the patients identied placements: respiratory (100%), gastric (90%)
were receiving acid suppression therapy. and intestinal (99%) (Metheny et al., 2000, 1999).
The verication of correct tube positioning using
4. Discussion biochemical measurement is contingent upon successful
aspiration of uid from either the gastric or intestinal sites.
Several trials have investigated the accuracy of various In the studies included in this review only those by
biochemical markers to determine feeding tube place- (Metheny et al., 1993, 2000, 1999; Phang et al., 2004)
ment in hospitalized patients however there is no reported that they were successful in obtaining tube
consensus on the optimal method. A systematic review aspirate samples more than 90% of the time which could be
was therefore undertaken to summarize and present the related to the type and size of the feeding tube and the
best available evidence relating to biochemical tests to technique used for aspiration of the contents. Collapse of
determine feeding tube placement for use in clinical the tube has been frequently cited as a reason for failure to
decision making. obtain aspirate samples (Metheny et al., 1988).
Only 10 published trials reported in English according
to the guidelines set out in the STARD guidelines (Bossuyt 5. Limitations
et al., 2003) were eligible for inclusion in this review. Seven
of the 10 trials included in the review were undertaken by Despite the high methodological quality of the studies,
the same investigator. Trials using similar diagnostic tests a conclusion about the diagnostic accuracy of the various
were combined statistically and the pooled results have methods is limited due to the small number of studies
been presented in a tabular form. assessing each test and the heterogeneity between the
Seven trials investigated the diagnostic accuracy of studies. In addition, only some of the studies included in
various pH values in differentiating between gastric and the review were undertaken in patients who required
intestinal placement. The diagnostic accuracy of pH 4.0 to mechanical ventilation, and therefore the review is limited
identify correct gastric placement of feeding tube was low in its generalisability to this patient population.
(pooled sensitivities ranging from 0.55 to 0.75). However, Most of the studies did not perform separate analyses of
this pH value was able to detect a large proportion of the the biochemical value of those specimens immediately
misplaced tubes. Similarly the diagnostic accuracy of obtained after insertion or those obtained from patients
either a pH 6.0 or 5.9 in correctly identifying stomach having continuous feeding. Stratied sampling methods
placement of the feeding tube was low (pooled sensitivity such as quota sampling could be used to ensure each
0.53) and identifying misplaced tubes was satisfactory. It segment of the population under study is representative.
 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046 1045

Some key variables for each stratum include (a) the Conicts of interest
method of feeding (continous vs. bolus), (b) fasting time None declared.
before specimen collection, (c) duration a patient had been
fed in a specic 24 h affect pH (because of the effect on pH), Funding
(d) the type of feed (protein, amino acids, peptides), (e) the We thank the Joanna Briggs Institute for supporting this
use of acid inhibitors including the dose, type and route of review.
medication (oral or parenteral) of acid suppression and (f)
Ethical approval
pH of ushes (sterile water or saline [both shown to have
Not required.
low pH]).

6. Implications for practice References

The limited evidence obtained from the review does not Bossuyt, P.M., Reitsma, J.B., Bruns, D.E., Gatsonis, C.A., Glasziou, P.P., Irwig,
L.M., Lijmer, J.G., Moher, D., Rennie, D., de Vet, H.C., 2003. Towards
provide a substantial base for the development of practice
complete and accurate reporting of studies of diagnostic accuracy:
guidelines. There is insufcient evidence to determine the the STARD initiative. BMJ: British Medical Journal 326 (7379), 4144
optimal cut-off pH value to differentiate between gastric (see comment).
Bradley, J.S., Phillips, J.O., Cavanaugh, J.E., Metzler, M.H., 1998. Clinical
and intestinal placements. There is suggestive but incon-
utility of pH paper versus pH meter in the measurement of critical
clusive evidence a pH 5.5 may lead to higher detection of gastric pH in stress ulcer prophylaxis. Critical Care Medicine 26 (11),
tubes located in the gastric and intestinal region. Until 19051909.
stronger evidence becomes available practice related to Burns, S.M., Carpenter, R., Truwit, J.D., 2001. Report on the development of
a procedure to prevent placement of feeding tubes into the lungs
the diagnostic tests used within the clinical setting will using end-tidal CO2 measurements. Critical Care Medicine 29 (5),
continue to be dictated by local preferences and cost 936939.
factors. Chen, Y.C., Chou, S.S., Lin, L.H., Wu, L.F., 2006. The effect of intermittent
nasogastric feeding on preventing aspiration pneumonia in ventilated
critically ill patients. Journal of Nursing Research 14 (3), 167180.
7. Implications for research Conner, T.M., Carver, D., 2005. The role of gastric pH testing with small-
bore feeding tubes: in the intensive care unit. DCCN: Dimensions of
Critical Care Nursing 24 (5), 210214.
Given that the evidence for biochemical techniques to Hart, S., 2006. Ionising radiation: promoting safety for patients, visitors
determine NG tube location have limited accuracy, other and staff. Nursing Standard 20 (47), 4757 (quiz 58, 60).
Howes, D.W., Shelley, E.S., Pickett, W., 2005. Un detecteur colorimetrique
methods such as magnetic detection and the use of spring de gaz carbonique pour determiner la mise en place endotracheale
gauge manometers and bedside ultrasound examination accidentelle dun tube dalimentation.(Colorimetric carbon dioxide
should be further investigated. In addition, future trials detector to determine accidental tracheal feeding tube placement).
Canadian Journal of Anesthesia 52 (4), 428432.
should also undertake economic evaluations to assess the Kearns, P.J., Donna, C., 2001. A controlled comparison of traditional
costs associated with development and delivery of newer feeding tube verication methods to a bedside, electromagnetic
technology of diagnostic tests, in clinical settings. technique. JPEN: Journal of Parenteral and Enteral Nutrition 25 (4),
210215.
What the paper adds: This paper has summarized the
Kindopp, A.S., Drover, J.W., Heyland, D.K., 2001. La capnographie permet
best available evidence for the diagnostic accuracy of de conrmer la position appropriee de la sonde dalimentation chez
biochemical markers for predicting nasogastric tube les patients de lunite des soins intensifs.(Capnography conrms
correct feeding tube placement in intensive care unit patients).
placement in adults.
Canadian Journal of Anesthesia 48 (7), 705710.
A comprehensive review titled Insertion and Manage- Leder, S.B., Suiter, D.M., 2008. Effect of nasogastric tubes on incidence of
ment of Nasogastric Tubes for Adults has been published at aspiration. Archives of Physical Medicine and Rehabilitation 89 (4),
www.joannabriggs.edu.au. 648651.
McClave, S.A., 2009. What does it mean to own feeding tubes? Nutrition
in Clinical Practice 24 (4), 430432.
Medicines Healthcare products Regulatory Agency (MHRA), 2004. Enteral
Acknowledgements Feeding Tubes (Nasogastric). Medicines and Healthcare products
Regulatory Agency (MHRA), London.
The authors thank Miss Lucia Ho, Assistant Librarian, Metheny, N., Reed, L., Berglund, B., Wehrle, M.A., 1994. Visual character-
istics of aspirates from feeding tubes as a method for predicting tube
Chinese University of Hong Kong, and Ms Karen Andrews, location. Nursing Research 43 (5), 282287.
Library Manager, Liverpool Hospital for their advice in Metheny, N., Reed, L., Wiersema, L., McSweeney, M., Wehrle, M.A., Clark, J.,
developing the search strategies of electronic bibliographic 1993. Effectiveness of pH measurements in predicting feeding tube
placement: an update. Nursing Research 42 (6), 324331.
databases. Metheny, N., Williams, P., Wiersema, L., Wehrle, M.A., Eisenberg, P.,
Contributors: Janita Chau had the idea for the project. McSweeney, M., 1989. Effectiveness of pH measurements in predict-
Janita Chau and Ritin Fernandez performed the literature ing feeding tube placement. Nursing Research 38 (5), 280285.
Metheny, N.A., Schnelker, R., McGinnis, J., Zimmerman, G., Duke, C.,
searches, carried out the selection of references, contrib-
Merritt, B., Banotai, M., Oliver, D.A., 2005. Indicators of tubesite during
uted to the data extraction and statistical analyses, wrote feedings. Journal of Neuroscience Nursing 37 (6), 320325.
drafts of the review, and are guarantor. Janita Chau, David Metheny, N.A., Smith, L., Stewart, B.J., 2000. Development of a reliable
and valid bedside test for bilirubin and its utility for improving
Thompson and Hoi-Shan Lo organised the daily project
prediction of feeding tube location. Nursing Research 49 (6), 302
activities, carried out the selection of references assessed 309.
the papers for the methodological quality, and performed Metheny, N.A., Spies, M.A., Eisenberg, P., 1988. Measures to test place-
data extraction. All authors provided scientic input on the ment of nasoenteral feeding tubes. Western Journal of Nursing
Research 10 (4), 367379.
drafts of the paper. David Thompson, Janita Chau and Metheny, N.A., Stewart, B.J., 2002. Testing feeding tube placement during
Rhonda Grifths organised funding for the review. continuous tube feedings. Applied Nursing Research 15 (4), 254258.
1046 R.S. Fernandez et al. / International Journal of Nursing Studies 47 (2010) 10371046

Metheny, N.A., Stewart, B.J., Smith, L., Yan, H., Diebold, M., Clouse, R.E., Swiech, K., Lancaster, D.R., Sheehan, R., 1994. Use of a pressure gauge to
1999. pH and concentration of bilirubin in feeding tube aspirates as differentiate gastric from pulmonary placement of nasoenteral feed-
predictors of tube placement. Nursing Research 48 (4), 189197. ing tubes. Applied Nursing Research 7 (4), 183189.
Metheny, N.A., Stewart, B.J., Smith, L., Yan, H., Diebold, M., Clouse, R.E., Taylor, S.J., Clemente, R., 2005. Conrmation of nasogastric tube position by
1997. pH and concentrations of pepsin and trypsin in feeding tube pH testing. Journal of Human Nutrition and Dietetics 18 (5), 371375.
aspirates as predictors of tube placement. JPEN: Journal of Parenteral Tobin, R.W., Gonzales, A.J., Golden, R.N., Brown, M.C., Silverstein, F.E.,
and Enteral Nutrition 21 (5), 279285. 2000. Magnetic detection to position human nasogastric tubes. Bio-
Phang, J.S., Marsh, W.A., Barlows III, T.G., Schwartz, H.I., 2004. Determin- medical Instrumentation & Technology 34 (6), 432436.
ing feeding tube location by gastric and intestinal pH values. Nutrition Weinberg, L., Skewes, D., 2006. Pneumothorax from intrapleural place-
in Clinical Practice 19 (6), 640644. ment of a nasogastric tube. Anaesthesia and Intensive Care 34 (2),
Portney, L.G., Watkins, M.P., 2000. Foundations of Clinical Research: 276279.
Applications to Practice. Prentice Hall Health, Upper Saddle River. Whiting, P., Rutjes, A.W., Reitsma, J.B., Bossuyt, P.M., Kleijnen, J., 2003. The
Rakel, B.A., Titler, M., Goode, C., Barry-Walker, J., Budreau, G., Buckwalter, development of QUADAS: a tool for the quality assessment of studies
K.C., 1994. Nasogastric and nasointestinal feeding tube placement: an of diagnostic accuracy included in systematic reviews. BMC Medical
integrative review of research. AACN Clinical Issues in Critical Care Research Methodology 3, 25.
Nursing 5 (2) 194206, 218199, 223. Wu, P.Y., Kang, T.J., Hui, C.K., Hung, M.H., Sun, W.Z., Chan, W.H., 2006. Fatal
Sackett, D.L., Straus, S.E., Richardson, W.S., Rosenberg, W., Haynes, R.B., massive hemorrhage caused by nasogastric tube misplacement in a
2000. Evidence-based Medicine: How to Practice and Teach EBM. patient with mediastinitis. Journal of the Formosan Medical Associa-
Churchill Livingstone, New York. tion 105 (1), 8085.
Simel, D.L., Samsa, G.P., Matchar, D.B., 1991. Likelihood ratios with con- Zamora, J., Abraira, V., Muriel, A., Khan, K., Coomarasamy, A., 2006. Meta-
dence: sample size estimation for diagnostic test studies. Journal of DiSc: a software for meta-analysis of test accuracy data. BMC Medical
Clinical Epidemiology 44 (8), 763770. Research Methodology 6, 31.