Gyawali CP, Rogers B, Frazzoni M, Savarino E, Roman R, Sifrim D. Inter-reviewer Variability in Interpretation of pH-Impedance Studies: The Wingate Consensus. Clin Gastroenterol Hepatol. 2021. 19(9), 1976-8.

Авторы: Gyawali C.P. / Rogers B.D. / Frazzoni M. / Savarino E.V. / Roman S. / Sifrim D.

Inter-reviewer Variability in Interpretation of pH-Impedance Studies: The Wingate Consensus

C. Prakash Gyawali¹, Benjamin Rogers¹, Marzio Frazzoni², Edoardo Savarino³,
Sabine Roman⁴, and Daniel Sifrim⁵

¹ Division of Gastroenterology, Washington University School of Medicine, St Louis, Missouri;
² Department of Specialized Medicine, Digestive Pathophysiology Unit, Baggiovara Hospital, Modena, Italy;
³ Division of Gastroenterology, Department of Surgery, Oncology and Gastroenterology, University of Padua, Padua, Italy;
⁴ Hospices Civils de Lyon, Digestive Physiology, Hopital E Herriot, Lyon I University, Digestive Physiology, Inserm U1032, LabTAU, Université de Lyon, Lyon, France;
⁵ Barts and The London School of Medicine and Dentistry Queen Mary, University of London, London, United Kingdom


Ambulatory pH-impedance monitoring evaluates gastroesophageal reflux disease (GERD) symptoms that persist despite empiric treatment trials.¹ Although automated analysis of pH-impedance monitoring is known to overestimate non–acid reflux episodes,² even experts may disagree on individual reflux episodes,^3,4 and accurate identification of postreflux swallow-induced peristaltic wave (PSPW). We hypothesized that a formal consensus meeting between expert pH-impedance reviewers would establish definitive criteria for identification of reflux episodes and PSPW, and improve inter-reviewer variability.

Methods

De-identified pH-impedance studies (Diversatek, Boulder, CO) from healthy asymptomatic volunteers were sent to the 5 reviewers (C.P.G., M.F., E.S., S.R., D.S.) 1 month in advance of the Wingate Consensus meeting in London (July 2018). Reviewers applied automated analysis, followed by confirmation of each reflux episode, manual insertion of missed episodes, and identification of PSPW. Rules for standardized pH-impedance definitions of reflux episodes and PSPW were developed at the Wingate Consensus meeting (Table 1), using available literature and reviewers’ experience. After a 3-month washout period, the same pH-impedance studies were re-analyzed applying the Wingate Consensus rules by the 5 reviewers. pH-impedance studies from GERD patients were subjected to a similar review process (November 2019 to March 2020). Formal institutional review board approval was not deemed necessary for post hoc interpretation of de-identified pH-impedance studies with no links to human subjects.


Table 1. The Wingate Consensus Recommendations for Identification of Reflux Episodes and PSPW

Reflux episodes	PSPW
Meal times need to be identified correctly and excluded before evaluation of pH-impedance events	PSPW starting point must be within 30 seconds after impedance returns to baseline in the distal-most impedance channel after a reflux episode
A reflux episode consists of a 50% decrease in impedance lasting for at least 4 seconds each in distal 2 impedance channels with retrograde propagation	PSPW does not need to be seen in all 6 impedance channels as long as a swallow is identified in the most proximal channel with anterograde propagation in the proximal and distal-most impedance channels
A pH decrease to less than 4.0 concurrent with a 4-second retrograde 50% impedance decrease after a belch episode is counted as a reflux episode	An impedance decrease of at least 50% lower than baseline needs to be present in the distal-most impedance channel
A pH decrease without impedance-detected reflux episode is counted as part of acid exposure time if not an artifact, but not as a reflux episode	Recovery of pH with antegrade impedance event is not mandatory but supports identification of PSPW
Automated analysis is first deployed, followed by manual confirmation/addition/deletion of identified reflux episodes using the earlier-described criteria	PSPW is best evaluated using a 2-minute window, using a 3000-ohms impedance scale

Data are reported as median and interquartile range (IQR) unless otherwise indicated. Categoric data were compared using the chi-squared test and the Friedman rank test, and continuous data were compared using the Wilcoxon signed-rank test, with Bonferroni correction for multiple comparisons. Conditional probability of agreement, which assesses proportional agreement between reviewers,⁵ and Cronbach’s a, which measures internal consistency or reliability of results,6 were used. Separate analyses evaluated agreement of revieweridentified reflux and PSPW events, within both ±10 episodes and ±5 episodes of median values, and after discarding the outlier value furthest from the median. A P value less than .05 was considered significant.

Results

Reviewers evaluated pH-impedance tracings from 19 asymptomatic controls (median age, 20 y; IQR, 19–21 y; 52.6% female; median AET, 0.4%) and 19 symptomatic GERD patients (median age, 52 y; IQR, 35–64 y; 78.9% female; median AET, 11.7%).

Although median reflux episodes were similar in asymptomatic controls before and after the Wingate Consensus (median, 30 to 29; P = .94), variance decreased by 14.7% (510.9 to 436.0). In contrast, PSPW trended lower (median, 12 to 9; P = .05), and variance decreased by 21.2% (137.2 to 107.8) after the meeting. When assessed within median ±10 episodes (Supplementary Table 1), conditional probability of agreement increased for both reflux episodes (0.77 to 0.80) and PSPW (0.81 to 0.90) after the Wingate Consensus. Similar results were seen for median ±5 episodes, and after discarding 1 outlier value, Cronbach’s a values also improved. Automated analysis significantly overestimated reflux episodes compared with reviewers, both in asymptomatic volunteers (median, 50 vs 29–30; P < .001) and GERD patients (median, 87 vs 69; P = .001) (Supplementary Figure 1).

Comparisons of medians of both numbers of reflux episodes and PSPW between reviewers remained significantly different at all stages of analysis (P < .001 for each comparison). Repeat analysis after discarding 1 outlier value furthest from the median resulted in better agreement and improved Cronbach’s a values, even when using the more stringent criteria of agreement within median ±5 episodes (Supplementary Table 1).

Discussion

In this study evaluating interpretation of pH-impedance studies, we report significant variability between automated analysis and expert interpretation. Agreement, variability, and reliability of reflux episode and PSPW identification improved after use of Wingate Consensus definitions, both within median ±10 and ±5 episodes, and when 1 outlier value was discarded from each analysis. Our study findings indicate that accurate interpretation of pH-impedance studies requires manual override of automated analysis, using definitive criteria for what constitutes a reflux episode, which in turn improves precision of PSPW identification. An accurate reflux episode count and PSPW index may shift balance toward or away from pathologic GERD when other metrics are inconclusive,¹ with management implications.^7,8

The strength of our study lies in the Wingate Consensus definitions of reflux episodes and PSPW (Table 1), incorporating face-to-face discussion and debate by experts. Limited numbers of pH-impedance studies analyzed by only 5 reviewers could have impacted our conclusions. Furthermore, acidic and nonacidic reflux episodes were not analyzed separately, and AET was not compared between reviewers. We acknowledge that ongoing outcomes analyses are required to understand the clinical significance of improved coherence in event identification. Standardized definitions may aid future automated analysis, perhaps using machine learning and artificial intelligence, which could improve and simplify interpretation of pH-impedance studies worldwide.

Supplementary Material

Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at ww w.cghjournal.org, and at ht tps://doi.org/10.1016/j.cgh.2020.09.002

References

1. Gyawali CP, Kahrilas PJ, Savarino E, et al. Modern diagnosis of GERD: the Lyon Consensus. Gut 2018;67:1351–1362.
2. Roman S, Bruley des Varannes S, Pouderoux P, et al. Ambulatory 24-h oesophageal impedance-pH recordings: reliability of automatic analysis for gastro-oesophageal reflux assessment. Neurogastroenterol Motil 2006;18:978–986.
3. Zerbib F, Roman S, Bruley Des Varannes S, et al. Normal values of pharyngeal and esophageal 24-hour pH impedance in individuals on and off therapy and interobserver reproducibility. Clin Gastroenterol Hepatol 2013;11:366–372.
4. Smits MJ, Loots CM, van Wijk MP, et al. An expert panel-based study on recognition of gastro-esophageal reflux in difficult esophageal pHimpedance tracings. Neurogastroenterol Motil 2015;27:637–645.
5. McHugh ML. Interrater reliability: the kappa statistic. Biochem Med (Zagreb) 2012;22:276–282.
6. Tavakol M, Dennick R. Making sense of Cronbach’s alpha. Int J Med Educ 2011;2:53–55.
7. Frazzoni M, Frazzoni L, Tolone S, et al. Lack of improvement of impaired chemical clearance characterizes PPI-refractory refluxrelated heartburn. Am J Gastroenterol 2018;113:670–676.
8. Rogers BD, Valdovinos LR, Crowell MD, et al. Number of reflux episodes on pH-impedance monitoring associates with improved symptom outcome and treatment satisfaction in gastro-oesophageal reflux disease (GERD) patients with regurgitation. Gut 2020.

Abbreviations used in this paper

AET - acid exposure time
GERD - gastroesophageal reflux disease
IQR - interquartile range
PSPW - postreflux swallow-induced peristaltic wave

Conflicts of interest

These authors disclose the following: C. Prakash Gyawali has consulted for Medtronic, Diversatek, Isothrive, Ironwood, and Quintiles; Edoardo Savarino has received lecture fees from Medtronic, Takeda, Janssen, MSD, AbbVie, and Malesci, and has consulted for Medtronic, Takeda, Janssen, MSD, Reckitt Bencikser, Sofar, Unifarco, SILA, and Oftagest; Sabine Roman has consulted for Medtronic, and received research support from Diversatek Healthcare and Medtronic; and Daniel Sifrim has received research grants from Reckitt Benckiser UK, Jinshan Technology China, and Alfa Sigma Italy. The remaining authors disclose no conflicts.

Назад в раздел