Use of pH reagent strips to verify gastric tube placement in newborns*

Objective: to confirm the accuracy of the pH test in identifying the placement of the gastric tube in newborns. Method: double-blind, diagnostic test study conducted with 162 newborns admitted to a neonatal intensive care unit and an intermediate care unit. The subjects were submitted to enteral intubation, followed by pH test with reagent strip, which was analyzed by a nurse, and radiological examination, analyzed by radiologist. Blinding was kept among professionals regarding test results. Diagnostic accuracy analysis of the pH test in relation to the radiological exam was performed. Results: the sample consisted of 56.17% boys, with average birth weight of 1,886.79g (SD 743,41), 32.92 (SD 2.99) weeks of gestational age and the mean pH was 3.36 (SD 1.27). Considering the cutoff point of pH≤5.5, the sensitivity was 96.25%, specificity 50%, positive predictive value 99.35% and negative predictive value 14.29%. Conclusion: The pH test performed with reagent strips is sensitive to identify the correct placement of the gastric tube, so it can be used as an adjuvant technique in the evaluation of the gastric tube placement. In interpreting the results, pH ≤5.5 points to correct placement and values > 5.5 require radiological confirmation.


Introduction
Incorrect enteral gastric tube placement in newborns is associated with serious harms, including risk of death (1)(2) . To confirm the placement of the tube immediately after its introduction and before each use is central to ensure the safe use of this device, so commonly used in neonatal inpatient units (3)(4)(5)(6) .
The choice of the measurement method on tube insertion length is also part of the first precautions related to the insertion of this device. Currently, methods that use anatomical references are described in the literature, as well as equations that use the newborn's weight or height to calculate insertion length. The most cited and recommended method is NEMU (nose, earlobe, mid-umbilicus), which is the distance from the tip of the nose to the earlobe and the midpoint between the xiphoid appendix and the umbilical base. In addition to this, we also found the ARHB (age-related, height based) and weightbased methods that use newborn height and weight respectively to calculate tube insertion length (5)(6) .
To evaluate the placement of the tube after insertion, the chest and abdomen radiological image is still considered the gold standard, since it allows the visualization of the entire course and the location of its distal extremity (7)(8)(9) . Despite being 100% accurate in the evaluation, it is not a suitable method to be routinely used in neonatal patients, due to the risks related to cumulative ionizing radiation exposure, besides the costs and timing that involve this procedure (4,10) .
A prevalence study conducted in 63 US hospitals showed that the methods that have been used in pediatric and neonatal clinical practice are, in order of options, checking the presence of residue on tube aspiration, auscultation of the epigastric region, verifying the external tube length, aspirated secretion pH testing, radiological examination and electromagnetic tracing (11) .
Regarding the accuracy and safety in the use of these alternative methods to radiological examination, it is known that auscultation of the epigastric region is not reliable and its discontinuity has been advised (10,12) . The presence of secretion to the aspiration of the tube and the evaluation of its color, as well as its appearance, may be sensitive in confirming the tube placement, but without established specificity and it is a confounding factor that endotracheal and bronchial secretions also may have the same color and appearance of gastric secretion (13) .
Checking the outer length of the tube is recommended as an adjunctive measure in tube maintenance, but is not alone, because the distal end of the tube may move to the intestine or respiratory system even though the external fixation remains intact (5) .
The electromagnetic tracing, despite presenting good consonance with the radiological examination, has an important limitation related to the tube bore, which should be at least 8Fr, which makes it impracticable to use this method to verify the placement of the tube in newborns (11,14) .
Verifying the pH of the aspirated secretion using reagent strips is a quick bedside test. Currently, there is a consensus among experts that this is the safest method available and is recommended as the first choice when verifying gastric tube placement in adults and children (9)(10)12) .
The chain of gastric secretion production is complex and the main physiological stimulus for such production is feeding. It is known that stomach full development during the fetal period is up to the 14 th or 15 th gestational week and that 27-week-old preterm infants are capable of gastric pH <4.0 on the first day of life. However, in the first 48 hours after birth, the literature indicates that the pH may be higher due to low gastric acid secretion, lack of food or the presence of amniotic fluid in the stomach (15)(16)(17) .
In order to validate the current recommendation, this study aimed to confirm whether gastric pH values ≤5.5, found in newborn gastric tube aspirate is a sensitive and specific method for assessing the correct placement of the tube in the stomach. Moreover, as secondary objectives, to verify if the pH of the gastric aspirate was influenced by the diet, use of histamine H 2 receptor antagonist drug and by the age of the studied sample.  This study comes from a secondary analysis of a randomized controlled trial (REBEC Registry RBR-2zk6yc) aimed at verifying the difference in the correct placement of the gastric tube when using two different methods for measuring insertion length. For this purpose, a sample size of 162 subjects was calculated using the chi-square estimation methodology after a pilot study of 50 subjects, assuming a power of 80%, a significance level of 5% and a loss rate of 20%.

Method
For this study, the power of the test was subsequently calculated to verify whether the sample size collected in the primary study would be adequate to test the hypothesis of the objective proposed here. The power of the test for sensitivity was 100% and specificity 25%, assuming a null value of 0.50 for sensitivity and specificity.
Each subject was included in the study on the first day of enteral feeding or on the day of changing the gastric tube, which occurred every 48 hours if the patient was initially using polyvinyl chloride (PVC) tube, according to the institution's protocol where the data was collected.
For the present study we used pediatric enteral tubes size 6.5 (FrekaPaed ® , Fresenius-Kabi ® , Friedberg, Germany), inserted nasal route by one of the three research assistant nurses, who had at least 3 years of clinical experience. Two minutes before the procedure, up to 1ml of 25% sucrose was administered orally to alleviate the discomfort generated by the procedure (18) .
The newborns were clinically stable, were not under minimal manipulation protocol and remained in easy retention during the procedure.
To estimate the tube insertion length, the NEMU (nose, earlobe, mid-umbilicus) measurement (19) or the weight-based formula method (20) were used. For the NEMU method, the distance between the tip of the nose and the ear lobe insertion point was verified, plus the distance between the ear lobe insertion point and the midpoint between the xiphoid appendix and the umbilical base. For the weight-based formula, the insertion length was estimated from the following calculation: 3x[weight in kilos]+13cm (20) .

After insertion and fixation of the tube, a 3ml
syringe was connected to the tube adapter and mild negative pressure was observed, with a return of secretion. Volume in ml and secretion staining were      (Table 1).  Table 2.
Considering the result of the association between the use of histamine H 2 -receptor blockers and pH values > 5.5 of the aspirated gastric secretion, although the sample is small, sensitivity, specificity, positive predictive value and negative predictive factor alone in the 156 patients who did not receive this drug were also verified.
It was observed that, in the evaluation of patients without gastric secretion inhibitor, sensitivity (98.70%) and positive predictive value (99.35%) remained high and there was no improvement in specificity (50.5%).
However, there was an increase in the negative predictive value (33.33%), as shown in Table 3 with their respective confidence intervals.

Discussion
In the sample under study, we found that the mean pH of the gastric aspirate was, according to the literature, <4.0 (16) . However, no relationship was observed between pH value and age. In addition, fasting or not and the type of diet did not influence the pH, considering the cutoff point specified for the proposed objective.
In contrast, a study that investigated the pH of gastric secretion in 96 newborns, according to the feeding pattern, found that the average pH of the subjects who were fed every 1 hour was higher than those fed every three hours (5.0; 3.5; p-value=0.001).
Despite this difference, this result did not interfere with the evaluation of gastric tube placement, as both averages were below 5.0 (16) .
Another factor that may influence the pH of gastric secretion is the use of histamine H 2 receptor antagonist drug. Although its use is associated with an increased risk of infections and necrotizing enterocolitis, this drug is often used in neonatal care for stress ulcer prophylaxis and in the treatment of gastroesophageal reflux disease (24)(25) .
In our sample, an association was observed between the use of the gastric secretion inhibitor drug and pH values> 5.5, but the number of subjects in which this occurred was small (n=4), and caution was required in evaluating these results. A study of 54 newborns (26) found that the pH averages of subjects treated authors (10,26) , but in the literature we also find authors who recommend the cutoff point at 5.0 (27)(28)(29) and 6.0 (16.30) .
In this study, the accuracy of the cutoff pH test A recent study of 212 children (29)  In this study the subjects were divided into 4 groups, combining the use or not of a gastric secretion blocking drug with the presence or absence of recent feeding.
Considering here only the recently fed subjects and the cutoff point <5.5, which is closest to our study, we observed similarity in the pattern of the results. In the study cited (29) , sensitivity was 96.1% and positive predictive value 98.0% in subjects treated with gastric secretion inhibitors, while sensitivity was 100% and positive predictive value 98.4% in those without using the drug.
It is noteworthy that, regarding the use of gastric secretion inhibitor drug, in our analysis two groups were not separated as in the study mentioned above (29) . The first analysis of our study refers to both treated and untreated patients with gastric secretion inhibitor, while the second analysis only to untreated patients.
When observing the results of specificity and negative predictive value, in our study we found low specificity (50%) in both analyzes. The negative predictive value was 14.29 in the general sample and 33.33 when excluding newborns treated with gastric secretion inhibitor, suggesting that the use of this drug seems to interfere with the number of false negative tests and with the proportion of true negatives. It is emphasized that the analysis of specificity and negative predictive value is directly influenced by prevalence, and in this case corresponds to the occurrence of incorrectly placement of tubes, which was in only 1.23% of the subjects.
In the study with 212 children (29) , considering only the analysis of the group of subjects recently fed with a cut-off point <5.5, a high specificity value (98.3%) was observed in the treated and non-treated with gastric acid secretion inhibitor subjects and negative predictive value of 96.7% among those receiving the drug and 100% among those not receiving.
The large discrepancy observed between the values of specificity and negative predictive value presented here with the results of the mentioned study (29) can be explained by the strategy adopted in the study cited, in which besides gastric secretion samples, samples were also collected from endotracheal tube secretion of 60 subjects (28.30% of the sample), which increased the occurrence of negative results and allowed to establish high specificity and negative predictive value.
Limitations of the present study include that findings

Conclusion
The use of pH reagent strips is a sensitive but nonspecific test to verify the placement of the gastric tube in newborns in the sample studied. That is, pH values negative result, and radiological examination to confirm positioning is relevant.