- Email: [email protected]
Esophageal manometry: Assessment of interpreter consistency
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2005;3:218 –224
ORIGINAL ARTICLES Esophageal Manometry: Assessment of Interpreter Consistency DEVJIT S. NAYAR, FARAH KHANDWALA, EDGAR ACHKAR, STEVEN S. SHAY, JOEL E. RICHTER, GARY W. FALK, EDY E. SOFFER, and MICHAEL F. VAEZI Department of Gastroenterology and Hepatology, Center for Swallowing and Esophageal Disorders, Cleveland Clinic Foundation, Cleveland, Ohio
Background & Aims: Manometry is used widely in the evaluation of esophageal disorders. Our aim was to assess the intra- and interobserver reliability of esophageal manometry and identify potential causes for diagnostic variability. Methods: Seventy-two esophageal manometry tracings were selected randomly from archives. Eight interpreters randomly and blindly evaluated tracings. Interpreters were divided into 3 groups: highly experienced (N ⴝ 3), moderately experienced (N ⴝ 3), and inexperienced (N ⴝ 2). Each tracing was examined for abnormalities involving the lower-esophageal sphincter (LES) and esophageal body. Interpreters rendered a single diagnosis from a list of 7 manometric diagnoses: normal, nutcracker, hypertensive LES, hypotensive LES, diffuse esophageal spasm (DES), nonspecific/ineffective esophageal motility (IEM), and achalasia. Intra- and interobserver agreements were determined and reasons for varied diagnoses were investigated. Results: Overall intraobserver agreement was good ( ⴝ .63, P < .0001). There was no difference (P ⴝ .9) between the highly and midexperienced interpreters ( ⴝ .61 and .65, respectively). Interobserver agreement for the diagnosis of achalasia and normal motility was good ( ⴝ .65 and .56, respectively). However, other manometric diagnoses yielded only fair interobserver agreement ( ⴝ .27). DES, nonspecific/ineffective esophageal motility (IEM), and hypo- and hypertensive LES diagnoses showed the least agreement. Poor adherence to established manometric criteria, misinterpretation of intrabolus pressure, and technical inadequacy were the most common sources of inconsistency in interpretations. Conclusions: Manometric diagnoses of conditions other than normal or achalasia are variable and have poor interobserver variability. Given their uncertain clinical implications, we must either redefine them or eliminate them from practice.
sophageal manometry, in conjunction with barium esophagram and upper endoscopy, is used commonly in clinical practice to diagnose esophageal disorders accurately. Current guidelines recommend using esophageal manometry in patients with dysphagia who are suspected of having motility disorders such as achalasia
E
or diffuse esophageal spasm (DES).1 The test also is recommended in patients with gastroesophageal reflux disease who are being considered for surgery to rule out achalasia and ensure adequate esophageal peristaltic amplitude before fundoplication.1 Manometric diagnoses are based on predefined criteria (Table 1).2,3 Among these diagnoses are conditions with well-defined criteria and treatment options (eg, achalasia), as well as those with uncertain clinical significance (eg, DES, nutcracker esophagus, nonspecific/ineffective esophageal motility [IEM], hypo- and hypertensive lower esophageal sphincter [LES]). In addition to the clinical uncertainty of many of these conditions, their diagnosis is dependent on many factors including the technical adequacy of the manometric tracings as well as the interpreter’s accuracy and consistency. Only one study4 has assessed interobserver variability of manometry in sphincter of Oddi measurements, finding 82%–90% agreement among the interpreters. Our study evaluated interpreter consistency and manometric reliability. In view of significant clinical decisions and resource allocation being made on the basis of esophageal manometry, we undertook the current prospective study. The aims of this study were: (1) to evaluate intra- and interobserver agreement of manometric interpretation; (2) to assess the impact of interpreter experience on manometric consistency; and (3) to identify the source of manometric misdiagnoses.
Materials and Methods Study Design A total of 72 esophageal manometry tracings were selected from the archives of the Center for Swallowing and Esophageal Disorders at the Cleveland Clinic Foundation beAbbreviations used in this paper: DES, diffuse esophageal spasm; IEM, nonspecific/ineffective esophageal motility; NSEMD, nonspecific esophageal motility disorder. © 2005 by the American Gastroenterological Association 1542-3565/05/$30.00 PII: 10.1053/S1542-3565(04)00617-2
March 2005
ESOPHAGEAL MANOMETRY
Table 1. Criteria for Diagnosing Esophageal Motility Abnormalities Functional defect
Diagnosis
Aperistalsis
Achalasia
Incoordinated motility
DES
Hypercontractile
Nutcracker esophagus Hypertensive LES
Hypocontractile
IEM
Hypotensive LES
Manometric findings Absent distal peristalsis Increased LES pressure (⬎45 mm Hg) Incomplete LES relaxation ⱖ20% simultaneous contractions Repetitive contractions (⬎3 peaks) Prolonged duration contractions Incomplete LES relaxation Increased amplitude (⬎180 mm Hg) Increased peristaltic duration Resting LES pressure ⬎45 mm Hg Incomplete LES relaxation ⬎30% nontransmitted peristalsis Peristaltic amplitude ⬍30 mm Resting LES pressure ⬍10 mm Hg
Data from Richter2 and Spechler and Castell.3
tween 2000 and 2003. At our institution, well over 1000 archived recordings are stored electronically by diagnosis. An equal number of recordings for each diagnosis were selected randomly and categorized as follows: normal, nutcracker esophagus, hypertensive LES, hypotensive LES, DES, IEM, and achalasia according to generally accepted criteria (Table 1). Tracings represented patients under evaluation for symptoms of dysphagia, heartburn, regurgitation, and noncardiac chest pain. Those tracings obtained in patients with prior esophageal dilatation, botulinum toxin injection, or surgical fundoplication were excluded. Booklets containing printed copies of the randomly mixed manometry tracings along with a worksheet to note specific manometric findings (Appendix) were distributed to 6 faculty members in the esophageal center and to 2 fellows. Our computer system automatically set and labeled appropriate scales on the horizontal and vertical axes of the printed forms. Swallows were spaced widely and therefore each patient’s tracing required between 5 and 20 pages of letter-sized paper. Interpreters were grouped based on their level of experience: (1) highly experienced: ⬎1000 previous manometric interpretations (n ⫽ 3); (2) moderately experienced: 100 –999 previous interpretations (n ⫽ 3); and (3) inexperienced: motility fellows in training with previous interpretations of ⬍100 tracings (n ⫽ 2). Interpreters were asked to evaluate the following manometric characteristics: LES, esophageal body contractions, upper esophageal sphincter pressure, and to provide the single most appropriate manometric diagnosis for each tracing (Appendix). Exact numeric measurements of the LES, upper esophageal sphincter pressure, or the esophageal body contraction ampli-
219
tude were not requested. Instead, interpreters were asked if the specific measurements were normal, abnormal, high, or low. The interpreters were given both the printout for the mean LES pressure values generated by the computer and had the freedom to alter values by evaluating individual tracings. There were no time constraints and the interpreters were blinded to patient symptoms, medical history, and other interpreters’ findings. Diagnoses then were compared among groups to assess interobserver agreement. Intraobserver agreement was based on comparison of the original manometric diagnosis, identified from the archives, with the second diagnosis rendered blindly by the same interpreter in this prospective study conducted 12–24 months later. Hence, intraobserver agreement was possible only for the highly and moderately experienced interpreters. Tracings with highly variable diagnoses then were scrutinized for manometric parameters, which produced the alternate diagnoses. The interpreter group defined technical inadequacy of a tracing when a definitive manometric diagnosis was made difficult owing to the test’s suboptimal recording quality.
Manometry All patients had undergone esophageal manometry in the supine position to determine the location of the LES. After fasting overnight, the test was performed using a low-compliance, pneumohydraulic, water infusion system (Arndorfer Medical Specialties, Milwaukee, WI) and an 8-lumen manometry catheter (Arndorfer Medical Specialties). The catheter had 4 proximal recording ports located at 5-cm intervals along its length and another 4 ports oriented radially (90°) near the tip. The recording sites were interfaced to an 8-channel polygraph (Synectics Medical AB, Stockholm, Sweden). LES pressure was measured by the station pull-through technique and recorded as the mean of 4 measurements at midrespiration determined by computer analysis after 10 water swallows of 5 mL. Completeness of LES relaxation (normal, ⬎85%) was assessed as the percent decrease from the mean resting LES pressure to gastric baseline after wet swallows. The following were the normative values in our laboratory5: LES pressure (10 – 45 mm Hg), esophageal body amplitude (30 –180 mm Hg), contraction duration (1.5–7 s), and UES pressure (50 –150 mm Hg).
Statistical Analysis The statistic was used to measure intra- and interobserver agreement. This statistic was used instead of the percentage level of agreement because the latter does not separate actual agreement and agreement caused by chance alone. A 7 ⫻ 7 table was needed to calculate in a pair of interpretations between 7 possible diagnoses. Data were analyzed from 8 different interpreters for 72 different tracings. The statistics were calculated as a composite of multiple comparisons between pairs of interpreters. For 8 different interpreters, 28 pairs of comparisons were made. The following conventional /strength of agreement grading system was used: 0 –.2 (poor); .21–.40 (fair); .41–.60 (moderate); .61–.80 (good); .81–1.00 (very good).6
220
NAYAR ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3
Table 2. Patient Demographics Based on Initial Manometric Diagnosis Initial diagnosis
Mean ⫾ SD age (range)
Normal (n ⫽ 10) Achalasia (n ⫽ 10) NSEMD/IEM (n ⫽ 19) Hypotensive LES (n ⫽ 10) Hypertensive LES (n ⫽ 8) DES (n ⫽ 9) Nutcracker (n ⫽ 6) Overall
47 ⫾ 13.1 (31–68) 49 ⫾ 12.5 (33–68) 50 ⫾ 12.9 (30–81) 55 ⫾ 14.3 (37–76) 51 ⫾ 14.1 (25–72) 51 ⫾ 2.4 (34–71) 60 ⫾ 11.9 (46–76) 52 ⫾ 13.0 (25–81)
.45 (moderate) to .75 (good) for highly experienced interpreters and from .43 (moderate) to .84 (good) for those with moderate experience.
Sex
Interobserver Agreement 5 4 8 5 2 3 2 29
M/5 F M/6 F M/11 F M/5 F M/6 F M/6 F M/4 F M/43 F
Results Table 2 lists patient demographics for the 7 manometric diagnoses. Overall, the mean (⫾SD) age of patients was 52 years (⫾13 y; range, 25– 81 y), and 29 patients (40%) were men. Patients with a normal manometry diagnosis tended to be younger (mean age, 47 y), whereas those with nutcracker esophagus were older (mean age, 60 y) than others. The most common presenting symptoms before manometry were heartburn and regurgitation (37%), dysphagia (33%), chest pain (25%), and chronic cough (5%). The most common symptom per manometric diagnosis was as follows: heartburn (normals), dysphagia (achalasia), heartburn and regurgitation (IEM), heartburn (hypotensive LES), dysphagia (hypertensive LES), chest pain (DES), and dysphagia (nutcracker esophagus) (Table 3). Intraobserver Agreement Interpreters blindly evaluated identical tracings twice over a 2-year period. The initial interpretation was the original diagnosis rendered as a part of patients’ clinical diagnosis. The second interpretation was based on the diagnosis rendered during this randomized protocol. Intraobserver agreement () for the mid- and highly experienced interpreters is shown in Table 4. The overall intraobserver was .63 (good; P ⬍ .001), suggesting a significant internal consistency of each interpreter. There was no difference (P ⫽ .9) between the highly and midexperienced gastroenterologists: ⫽ .61 and .65, respectively. The intraobserver ranged from
Table 5 lists the interobserver within the 3 interpreter groups. Because of the relatively distinct characteristics of manometry tracings for normal and achalasia patients, high values were expected and observed (.56 and .65, respectively). Thus, the scores for these 2 diagnoses were combined and used as internal standards. The interobserver agreement for all other diagnoses then were compared with this standard. Overall, the interobserver value was only .37 (fair). There was a graded increase in values with increasing interpreter experience: low experience, ⫽ .30 (fair); moderately experienced, ⫽ .39 (fair); highly experienced, ⫽ .44 (moderate). Overall, the interobserver value for normal and achalasia was good ( ⫽ .68). This was independent of interpreter expertise ( range, .66 – .71) (Table 5). However, the level of interobserver agreement for the 5 other diagnoses combined was only fair overall ( ⫽ .27): low-experience, ⫽ .16 (poor); midexperience, ⫽ .28 (fair); high-experience, ⫽ .35 (fair) (Table 5). The individual interobserver values were as follows: normal ( ⫽ .56), achalasia ( ⫽ .65), nutcracker esophagus ( ⫽ .72), DES ( ⫽ .16), IEM ( ⫽ .40), hypotensive LES ( ⫽ .40), and hypertensive LES ( ⫽ .54). Comparison of First and Second Interpretations Figure 1 shows the percent agreement between the second interpretation compared with the initial diagnosis. Patients with the initial diagnosis of normal, achalasia, nutcracker esophagus, and IEM were diagnosed similarly with the second interpretation in nearly 75% of cases (69%, 84%, 77%, and 71%, respectively). However, DES, hypotensive LES, and hypertensive LES were diagnosed only as such on the second interpretation in less than 50% of cases (27%, 41%, and 48%, respectively). Table 6 details the distribution of the second manometric interpretation as a function of the initial
Table 3. Presenting Symptoms for Each Manometric Diagnosis Initial diagnosis
Regurgitation
Dysphagia
Chest pain
Heartburn
Chronic cough
Normal (n ⫽ 10) Achalasia (n ⫽ 10) NSEMD/IEM (n ⫽ 19) Hypotensive LES (n ⫽ 10) Hypertensive LES (n ⫽ 8) DES (n ⫽ 9) Nutcracker (n ⫽ 6)
20% 20% 27% 30%
20% 70% 17% 10% 63%
10%
40% 10% 27% 40%
10%
22% 17%
50%
17% 10% 37% 67% 33%
11%
12% 10%
March 2005
ESOPHAGEAL MANOMETRY
Table 4. Intraobserver Score for Highly and Moderately Experienced Interpreters Highly experienced Individual interpreters
84%
80
Moderately experienced
1
2
3
1
2
3
13 .45
20 .63 .61
15 .75
9 .84
5 .69 .65
10 .43
.63
P value
69%
71%
77%
60 41%
40 ⬍.001b NS ⬍.001b
48%
27%
20 0
diagnosis. For example, 84% of the patients initially diagnosed with achalasia had a similar diagnosis, whereas 9% were diagnosed with IEM and 7% were diagnosed as normal (Table 6). Similarly, the second interpretation of normal tracings was diluted with IEM (21%), hypotensive LES (6%), DES (3%), and hypertensive LES (1%).
Figure 1. Percent agreement between the second interpretation compared with the initial diagnosis. Patients with the initial diagnosis of normal, achalasia, nutcracker esophagus, and NSEMD/IEM were diagnosed similarly with the second interpretation in nearly 75% of cases (69%, 84%, 77%, and 71%, respectively). However, DES, hypotensive LES, and hypertensive LES were diagnosed only as such on the second interpretation in less than 50% of cases (27%, 41%, 48%, respectively).
as ia
N or N SE ma l M D /IE N M ut cr ac ke r
H yp D ot ES en H s yp iv e er L te ns ES iv e LE S
NS, not significant (between highly and moderately experienced interpreters). aNumber of tracings evaluated by each interpreter twice over a 2-year period. bP ⬍ .001 suggests significant individual consistency of interpreters.
Ac ha l
Na Mean Overall
100
221
Reasons for Inconsistencies In cases in which agreement was variable, a methodologic group review by all 6 interpreters was conducted. Clinical data including manometry, imaging, pH monitoring, and endoscopic studies were analyzed. A consensus was reached by the 6 esophageal physicians on the most likely diagnosis. Based on this exhaustive analysis, the 3 most frequent sources of inconsistency in interpretation of manometric tracings were as follows: (1) technically inadequate study (20%), (2) poor adherence to established definitions for manometric diagnoses (50%) (Table 1), and (3) misinterpretation of intrabolus pressure (30%). For example, on review of the 10 patients initially diagnosed with achalasia, the consensus group agreed with this diagnosis in 7 patients. Conversely, in 2 patients initially diagnosed with DES and in Table 5. Interobserver Agreement () for Interpreter Groups Based on Manometric Diagnosis Interpreter experience level
Normal and achalasia N ⫽ 20
All others N ⫽ 52a
Overall N ⫽ 72
Low experience (N ⫽ 2) Midexperience (N ⫽ 3) Highly experienced (N ⫽ 3) Overall (N ⫽ 8)
.66 .71 .68 .68
.16 .28 .35 .27
.30 .39 .44 .37
NOTE. Diagnoses of achalasia and normal motility were combined and used as internal standard for data validity and compared with other manometric diagnoses combined. aOther diagnoses include nonspecific/ineffective motility, DES, hypotensive LES, hypertensive LES, and nutcracker esophagus.
1 patient initially diagnosed with IEM, achalasia was the final correct diagnosis. Thus, the sensitivity, specificity, and positive and negative predictive value of esophageal manometry for diagnosing achalasia was 70%, 95%, 70%, and 95%, respectively. The misinterpretation of LES pressure led to a misdiagnosis for hypertensive LES, hypotensive LES, and normal motility; whereas variations in intrabolus pressure measurements misclassified DES, achalasia, and even normal motility. The latter misclassification was highly important because it misclassified patients with achalasia or normal tracings to DES or ineffective esophageal motility disorders. Such misclassification resulted in delayed diagnosis and treatment of patients with achalasia.
Discussion This study was a prospective, multi-interpreter, randomized, blinded study assessing the reliability of esophageal manometry interpretations. We found a strong intra- and interobserver agreement in diagnosing the 2 extremes in the spectrum of motility diagnoses, normal and achalasia, with a score of .63 and .68, respectively. However, despite the level of expertise, interpretation of other esophageal motility disorders have poor interobserver agreement, with a score .27 (Table 5). We found that overall intraobserver reliability was good for both high- and moderate-level interpreters, with a score of .61 and .65, respectively. However, experience
222
NAYAR ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3
Table 6. Proportion of Second Interpretation (Horizontal) as a Function of the Initial Diagnosis (Vertical)
Achalasia Normal NSEMD/IEM Nutcracker DES Hypotensive LES Hypertensive LES
Achalasia
Normal
NSEMD/IEM
Nutcracker
DES
Hypotensive LES
Hypertensive LES
84%
7% 69% 17% 2% 14% 14% 21%
9% 21% 71% 11% 49% 41% 13%
3% 3% 6% 27%
6%
3% 77% 9%
1% 1% 2% 1%
5% 2% 4% 7%
8%
41% 3%
48%
NOTE. Numbers on diagonal indicate level of percent agreement in diagnoses between first and second interpretations. Note the decreased percent agreement for DES, hypotensive LES, and hypertensive LES.
did make a difference in differentiating normal and achalasia tracings from others. More experienced interpreters had higher interobserver agreements ( ⫽ .35) than midlevel interpreters ( ⫽ .28) and than those with least experience ( ⫽ .16) (Table 5). Percent agreement between the first and the second interpretation was highest for tracings with the following diagnoses: achalasia, normal, nonspecific esophageal motility disorder (NSEMD)/IEM, or nutcracker esophagus. Hypo- and hypertensive LES and DES diagnoses had the lower percent agreements, with the latter having the lowest agreement (Figure 1). Only 27% of patients initially diagnosed with DES were diagnosed as such on the second interpretation. The majority (49%) of these DES patients were diagnosed with IEM on the second manometry (Table 6). This is understandable because manometrically the difference between these 2 diagnoses may be based on distal esophageal contraction amplitude, being low in the IEM and normal to high in DES (Table 1). In addition to final interpretation, data were collected on specific manometric parameters leading to the final diagnoses by each interpreter (Appendix). Several reasons for discrepancies were identified. First was the lack of strict adherence to published criteria (Table 1). It is possible that the interpreters still used outdated criteria for manometric diagnoses and did not use the most recently published information. In post hoc analysis, it was obvious that different gastroenterologists used different interpretation criteria. For example, the assessment of intrabolus pressure take-off points and peristalsis varied considerably in those tracings with only moderate, fair, or poor agreement. This is understandable because published criteria do not address interpretation of intrabolus pressure. In patients with early achalasia, classic isobaric contractions may not be present and the diagnosis may depend on inclusion or exclusion of intrabolus pressures. This misclassification is important because it can result in delayed therapy for this group of patients. Also, there was a difference among interpreters on the normal cut-off level used for LES and esophageal body
amplitude pressures. Some patients were classified as having low LES pressure who had, by strict definition, pressures in the low-normal category. Thus, upper and lower limits of normal LES pressure were not observed uniformly, which explains the low percent agreement for hypo- and hypertensive LES pressure tracings (Figure 1, and Table 6). Additional reasons for this misclassification may be interpreter bias with regard to the importance of degree of abnormality of LES pressure vs. esophageal body amplitude or peristalsis. A patient with low LES pressure and low-amplitude esophageal body contractions may be identified as hypotensive LES with one interpreter whereas another interpreter may identify them as having IEM. Similarly, some may consider LES pressure of 12 mm Hg to be normal whereas others may consider it low, especially in those being considered for surgical fundoplication. The diagnostic criteria listed in Table 1 were not distributed to participants. Obligatory criteria for a particular diagnosis were at the discretion of the reader. Interpreter consistency in this study could have been improved by reinforcing diagnostic criteria among interpreters and by using computer-derived data points; however, such an attempt would have been artificial because this normally is not performed in clinical practice. Another reason for discrepant interpretations was the poor quality of the tracings because of the technical inadequacy of the procedure. Examples of this included an insufficient number of wet swallows and insufficient time between swallows to assess esophageal contractions and peristalsis accurately. Adequacy of esophageal manometry is highly dependent on the technical expertise of the person performing the test, thus highlighting the need for adequate training and oversight of the personnel involved in performing esophageal manometry testing. Despite rigorous attention to this matter in our center, technical performance was an important but not the main reason for interpreter variability. Two variables also may have impacted the results of this study. These included the use of printed tracings and the lack of information on patients’ clinical data before
March 2005
interpretation. Analysis of esophageal motility may be easier with a computer monitor than with printed tracings, allowing more accurate pressure measurements enhanced by color and the ability to alter graphics scales. There currently are no data to suggest that this difference may play an important role in improving reliability of esophageal motility interpretation. Second, some may argue that interpretation of manometry should be undertaken with the clinical history at hand. Others believe that esophageal manometry should be interpreted independently of clinical information. Our study is a reflection of the reliability of manometric interpretation based on a lack of clinical data. Additionally, the first manometric diagnoses rendered by the reviewers in calculation of the intraobserver was based on knowledge of clinical data that may have slightly biased the score. It is possible that clinical information would improve manometric reliability; however, this needs to be tested in future studies. Finally, whether or not our findings are generalizable to the community physicians is unknown. One could argue that the results would be worse for the interpreters in the community given the poor agreement even among motility experts. On the other hand, it may be argued that the interpretation in the communitybased centers may be more uniform if individuals were trained by the same equipment providers. Because of such inconsistent interpretation, especially with respect to diagnoses other than normal and achalasia, esophageal manometric diagnoses should be complemented by clinical and radiographic data to allow correct diagnosis and improved patient management. In patients presenting with classic signs and symptoms of achalasia, manometry is a reliable diagnostic tool and is independent of interpreter experience. However, in difficult tracings in which intrabolus pressure inclusion is in doubt, a barium swallow should complement manometry to ensure a correct diagnosis. Therefore, our data are in agreement with previous studies by Massey et al.,7 who found both manometry and radiographic studies to have inherent limitations and suggested a complementary role for these techniques. They showed that radiographic examinations were insensitive to contractions in esophageal segments devoid of bolus whereas manometry was insensitive to contractions that did not occlude the esophageal lumen. A normal esophageal manometry interpretation is both highly specific and highly repeatable among esophageal experts. Our data suggest that we must refine or even eliminate diagnostic criteria for other motility disturbances so as to avoid overlapping diagnoses. Given the shortcomings of traditional manometry, some have suggested the use of other methods in the diagnosis of manometric disorders.8 –11 For example, topographic
ESOPHAGEAL MANOMETRY
223
methods in which manometric recordings are examined by spatial as well as temporal relationships of pressure data suggest that such methods, especially when automated, may have increased accuracy in manometric diagnoses.8 In a study of 212 patients with both the traditional pull-through technique as well as a topographic approach, Clouse et al.8 reported increased accuracy of diagnosing severe esophageal motor dysfunction with the latter. Topographic methods correctly diagnosed 97.2% of the diagnosis in patients with aperistalsis, which was superior to the traditional method (77.8%). Therefore, until further studies are performed in this area, given the high degree of variability of the traditional manometry, the clinical relevance of manometric diagnoses other than normal and achalasia need to be better defined. Meanwhile, in interpretation of manometric tracings, adherence to published criteria for diagnoses is recommended.
Appendix Analyses Work Sheet Lower esophageal sphincter ΠNormal ΠAbnormal Relaxation: ΠNormal ΠAbnormal ΠIndeterminate Esophageal body Peristalsis: ΠNormal ΠAbnormal Amplitude of contraction: ΠNormal ΠAbnormal Duration of contraction: ΠNormal ΠAbnormal Contraction description: ΠNormal ΠIsobaric ΠNontransmitted ΠSimultaneous ΠMultiple peaks ΠOther Manometric diagnosis ΠNormal ΠNutcracker esophagus ΠHypertensive LES ΠHypotensive LES ΠDiffuse esophageal spasm ΠNonspecific/ineffective esophageal motility ΠAchalasia Mean Pressure:
References 1. Kahrilas PJ, Clouse RE, Hogan WJ. American Gastroenterological Association position statement and technical review on the clinical use of esophageal manometry. Gastroenterology 1994;107: 1865–1884.
224
NAYAR ET AL
2. Richter JE. Oesophageal motility disorders. Lancet 2001;358: 823– 828. 3. Spechler SJ, Castell DO. Classification of oesophageal motility abnormalities. Gut 2001;49:145–151. 4. Smithline A, Hawes R, Lehman G. Sphincter of Oddi manometry: interobserver variability. Gastrointest Endosc 1993;39:486 – 491. 5. Richter JE, Wu WC, Johns DN, et al. Esophageal manometry in 95 healthy volunteers. Dig Dis Sci 1987;32:583–592. 6. Brennan P, Silman A. Statistical methods for assessing observer variability in clinical measures. Br Med J 1992;304:1491–1494. 7. Massey BT, Dodds WJ, Hogan WJ, et al. Abnormal esophageal motility. An analysis of concurrent radiographic and manometric findings. Gastroenterology 1991;101:344 –354. 8. Clouse RE, Staiano A, Alrakawi A, et al. Application of topographical methods to clinical esophageal manometry. Am J Gastroenterol 2000;95:2720 –2730.
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3
9. Clouse RE, Staiano A. Topography of the esophageal peristaltic pressure wave. Am J Physiol 1991;261:G677–G684. 10. Clouse RE, Staiano A. Topography of normal and high amplitude esophageal peristalsis. Am J Physiol 1993;265:G1098 –G1107. 11. Clouse RE, Staiano A, Alrakawi A. Development of a topographic analysis system for manometric studies in the gastrointestinal tract. Gastrointest Endosc 1998;48:395– 401.
Address requests for reprints to: Michael F. Vaezi, MD, PhD, FACG, Department of Gastroenterology and Hepatology, Center for Swallowing and Esophageal Disorders, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195. e-mail: [email protected]; fax: (216) 444-6302.
ORIGINAL ARTICLES Esophageal Manometry: Assessment of Interpreter Consistency DEVJIT S. NAYAR, FARAH KHANDWALA, EDGAR ACHKAR, STEVEN S. SHAY, JOEL E. RICHTER, GARY W. FALK, EDY E. SOFFER, and MICHAEL F. VAEZI Department of Gastroenterology and Hepatology, Center for Swallowing and Esophageal Disorders, Cleveland Clinic Foundation, Cleveland, Ohio
Background & Aims: Manometry is used widely in the evaluation of esophageal disorders. Our aim was to assess the intra- and interobserver reliability of esophageal manometry and identify potential causes for diagnostic variability. Methods: Seventy-two esophageal manometry tracings were selected randomly from archives. Eight interpreters randomly and blindly evaluated tracings. Interpreters were divided into 3 groups: highly experienced (N ⴝ 3), moderately experienced (N ⴝ 3), and inexperienced (N ⴝ 2). Each tracing was examined for abnormalities involving the lower-esophageal sphincter (LES) and esophageal body. Interpreters rendered a single diagnosis from a list of 7 manometric diagnoses: normal, nutcracker, hypertensive LES, hypotensive LES, diffuse esophageal spasm (DES), nonspecific/ineffective esophageal motility (IEM), and achalasia. Intra- and interobserver agreements were determined and reasons for varied diagnoses were investigated. Results: Overall intraobserver agreement was good ( ⴝ .63, P < .0001). There was no difference (P ⴝ .9) between the highly and midexperienced interpreters ( ⴝ .61 and .65, respectively). Interobserver agreement for the diagnosis of achalasia and normal motility was good ( ⴝ .65 and .56, respectively). However, other manometric diagnoses yielded only fair interobserver agreement ( ⴝ .27). DES, nonspecific/ineffective esophageal motility (IEM), and hypo- and hypertensive LES diagnoses showed the least agreement. Poor adherence to established manometric criteria, misinterpretation of intrabolus pressure, and technical inadequacy were the most common sources of inconsistency in interpretations. Conclusions: Manometric diagnoses of conditions other than normal or achalasia are variable and have poor interobserver variability. Given their uncertain clinical implications, we must either redefine them or eliminate them from practice.
sophageal manometry, in conjunction with barium esophagram and upper endoscopy, is used commonly in clinical practice to diagnose esophageal disorders accurately. Current guidelines recommend using esophageal manometry in patients with dysphagia who are suspected of having motility disorders such as achalasia
E
or diffuse esophageal spasm (DES).1 The test also is recommended in patients with gastroesophageal reflux disease who are being considered for surgery to rule out achalasia and ensure adequate esophageal peristaltic amplitude before fundoplication.1 Manometric diagnoses are based on predefined criteria (Table 1).2,3 Among these diagnoses are conditions with well-defined criteria and treatment options (eg, achalasia), as well as those with uncertain clinical significance (eg, DES, nutcracker esophagus, nonspecific/ineffective esophageal motility [IEM], hypo- and hypertensive lower esophageal sphincter [LES]). In addition to the clinical uncertainty of many of these conditions, their diagnosis is dependent on many factors including the technical adequacy of the manometric tracings as well as the interpreter’s accuracy and consistency. Only one study4 has assessed interobserver variability of manometry in sphincter of Oddi measurements, finding 82%–90% agreement among the interpreters. Our study evaluated interpreter consistency and manometric reliability. In view of significant clinical decisions and resource allocation being made on the basis of esophageal manometry, we undertook the current prospective study. The aims of this study were: (1) to evaluate intra- and interobserver agreement of manometric interpretation; (2) to assess the impact of interpreter experience on manometric consistency; and (3) to identify the source of manometric misdiagnoses.
Materials and Methods Study Design A total of 72 esophageal manometry tracings were selected from the archives of the Center for Swallowing and Esophageal Disorders at the Cleveland Clinic Foundation beAbbreviations used in this paper: DES, diffuse esophageal spasm; IEM, nonspecific/ineffective esophageal motility; NSEMD, nonspecific esophageal motility disorder. © 2005 by the American Gastroenterological Association 1542-3565/05/$30.00 PII: 10.1053/S1542-3565(04)00617-2
March 2005
ESOPHAGEAL MANOMETRY
Table 1. Criteria for Diagnosing Esophageal Motility Abnormalities Functional defect
Diagnosis
Aperistalsis
Achalasia
Incoordinated motility
DES
Hypercontractile
Nutcracker esophagus Hypertensive LES
Hypocontractile
IEM
Hypotensive LES
Manometric findings Absent distal peristalsis Increased LES pressure (⬎45 mm Hg) Incomplete LES relaxation ⱖ20% simultaneous contractions Repetitive contractions (⬎3 peaks) Prolonged duration contractions Incomplete LES relaxation Increased amplitude (⬎180 mm Hg) Increased peristaltic duration Resting LES pressure ⬎45 mm Hg Incomplete LES relaxation ⬎30% nontransmitted peristalsis Peristaltic amplitude ⬍30 mm Resting LES pressure ⬍10 mm Hg
Data from Richter2 and Spechler and Castell.3
tween 2000 and 2003. At our institution, well over 1000 archived recordings are stored electronically by diagnosis. An equal number of recordings for each diagnosis were selected randomly and categorized as follows: normal, nutcracker esophagus, hypertensive LES, hypotensive LES, DES, IEM, and achalasia according to generally accepted criteria (Table 1). Tracings represented patients under evaluation for symptoms of dysphagia, heartburn, regurgitation, and noncardiac chest pain. Those tracings obtained in patients with prior esophageal dilatation, botulinum toxin injection, or surgical fundoplication were excluded. Booklets containing printed copies of the randomly mixed manometry tracings along with a worksheet to note specific manometric findings (Appendix) were distributed to 6 faculty members in the esophageal center and to 2 fellows. Our computer system automatically set and labeled appropriate scales on the horizontal and vertical axes of the printed forms. Swallows were spaced widely and therefore each patient’s tracing required between 5 and 20 pages of letter-sized paper. Interpreters were grouped based on their level of experience: (1) highly experienced: ⬎1000 previous manometric interpretations (n ⫽ 3); (2) moderately experienced: 100 –999 previous interpretations (n ⫽ 3); and (3) inexperienced: motility fellows in training with previous interpretations of ⬍100 tracings (n ⫽ 2). Interpreters were asked to evaluate the following manometric characteristics: LES, esophageal body contractions, upper esophageal sphincter pressure, and to provide the single most appropriate manometric diagnosis for each tracing (Appendix). Exact numeric measurements of the LES, upper esophageal sphincter pressure, or the esophageal body contraction ampli-
219
tude were not requested. Instead, interpreters were asked if the specific measurements were normal, abnormal, high, or low. The interpreters were given both the printout for the mean LES pressure values generated by the computer and had the freedom to alter values by evaluating individual tracings. There were no time constraints and the interpreters were blinded to patient symptoms, medical history, and other interpreters’ findings. Diagnoses then were compared among groups to assess interobserver agreement. Intraobserver agreement was based on comparison of the original manometric diagnosis, identified from the archives, with the second diagnosis rendered blindly by the same interpreter in this prospective study conducted 12–24 months later. Hence, intraobserver agreement was possible only for the highly and moderately experienced interpreters. Tracings with highly variable diagnoses then were scrutinized for manometric parameters, which produced the alternate diagnoses. The interpreter group defined technical inadequacy of a tracing when a definitive manometric diagnosis was made difficult owing to the test’s suboptimal recording quality.
Manometry All patients had undergone esophageal manometry in the supine position to determine the location of the LES. After fasting overnight, the test was performed using a low-compliance, pneumohydraulic, water infusion system (Arndorfer Medical Specialties, Milwaukee, WI) and an 8-lumen manometry catheter (Arndorfer Medical Specialties). The catheter had 4 proximal recording ports located at 5-cm intervals along its length and another 4 ports oriented radially (90°) near the tip. The recording sites were interfaced to an 8-channel polygraph (Synectics Medical AB, Stockholm, Sweden). LES pressure was measured by the station pull-through technique and recorded as the mean of 4 measurements at midrespiration determined by computer analysis after 10 water swallows of 5 mL. Completeness of LES relaxation (normal, ⬎85%) was assessed as the percent decrease from the mean resting LES pressure to gastric baseline after wet swallows. The following were the normative values in our laboratory5: LES pressure (10 – 45 mm Hg), esophageal body amplitude (30 –180 mm Hg), contraction duration (1.5–7 s), and UES pressure (50 –150 mm Hg).
Statistical Analysis The statistic was used to measure intra- and interobserver agreement. This statistic was used instead of the percentage level of agreement because the latter does not separate actual agreement and agreement caused by chance alone. A 7 ⫻ 7 table was needed to calculate in a pair of interpretations between 7 possible diagnoses. Data were analyzed from 8 different interpreters for 72 different tracings. The statistics were calculated as a composite of multiple comparisons between pairs of interpreters. For 8 different interpreters, 28 pairs of comparisons were made. The following conventional /strength of agreement grading system was used: 0 –.2 (poor); .21–.40 (fair); .41–.60 (moderate); .61–.80 (good); .81–1.00 (very good).6
220
NAYAR ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3
Table 2. Patient Demographics Based on Initial Manometric Diagnosis Initial diagnosis
Mean ⫾ SD age (range)
Normal (n ⫽ 10) Achalasia (n ⫽ 10) NSEMD/IEM (n ⫽ 19) Hypotensive LES (n ⫽ 10) Hypertensive LES (n ⫽ 8) DES (n ⫽ 9) Nutcracker (n ⫽ 6) Overall
47 ⫾ 13.1 (31–68) 49 ⫾ 12.5 (33–68) 50 ⫾ 12.9 (30–81) 55 ⫾ 14.3 (37–76) 51 ⫾ 14.1 (25–72) 51 ⫾ 2.4 (34–71) 60 ⫾ 11.9 (46–76) 52 ⫾ 13.0 (25–81)
.45 (moderate) to .75 (good) for highly experienced interpreters and from .43 (moderate) to .84 (good) for those with moderate experience.
Sex
Interobserver Agreement 5 4 8 5 2 3 2 29
M/5 F M/6 F M/11 F M/5 F M/6 F M/6 F M/4 F M/43 F
Results Table 2 lists patient demographics for the 7 manometric diagnoses. Overall, the mean (⫾SD) age of patients was 52 years (⫾13 y; range, 25– 81 y), and 29 patients (40%) were men. Patients with a normal manometry diagnosis tended to be younger (mean age, 47 y), whereas those with nutcracker esophagus were older (mean age, 60 y) than others. The most common presenting symptoms before manometry were heartburn and regurgitation (37%), dysphagia (33%), chest pain (25%), and chronic cough (5%). The most common symptom per manometric diagnosis was as follows: heartburn (normals), dysphagia (achalasia), heartburn and regurgitation (IEM), heartburn (hypotensive LES), dysphagia (hypertensive LES), chest pain (DES), and dysphagia (nutcracker esophagus) (Table 3). Intraobserver Agreement Interpreters blindly evaluated identical tracings twice over a 2-year period. The initial interpretation was the original diagnosis rendered as a part of patients’ clinical diagnosis. The second interpretation was based on the diagnosis rendered during this randomized protocol. Intraobserver agreement () for the mid- and highly experienced interpreters is shown in Table 4. The overall intraobserver was .63 (good; P ⬍ .001), suggesting a significant internal consistency of each interpreter. There was no difference (P ⫽ .9) between the highly and midexperienced gastroenterologists: ⫽ .61 and .65, respectively. The intraobserver ranged from
Table 5 lists the interobserver within the 3 interpreter groups. Because of the relatively distinct characteristics of manometry tracings for normal and achalasia patients, high values were expected and observed (.56 and .65, respectively). Thus, the scores for these 2 diagnoses were combined and used as internal standards. The interobserver agreement for all other diagnoses then were compared with this standard. Overall, the interobserver value was only .37 (fair). There was a graded increase in values with increasing interpreter experience: low experience, ⫽ .30 (fair); moderately experienced, ⫽ .39 (fair); highly experienced, ⫽ .44 (moderate). Overall, the interobserver value for normal and achalasia was good ( ⫽ .68). This was independent of interpreter expertise ( range, .66 – .71) (Table 5). However, the level of interobserver agreement for the 5 other diagnoses combined was only fair overall ( ⫽ .27): low-experience, ⫽ .16 (poor); midexperience, ⫽ .28 (fair); high-experience, ⫽ .35 (fair) (Table 5). The individual interobserver values were as follows: normal ( ⫽ .56), achalasia ( ⫽ .65), nutcracker esophagus ( ⫽ .72), DES ( ⫽ .16), IEM ( ⫽ .40), hypotensive LES ( ⫽ .40), and hypertensive LES ( ⫽ .54). Comparison of First and Second Interpretations Figure 1 shows the percent agreement between the second interpretation compared with the initial diagnosis. Patients with the initial diagnosis of normal, achalasia, nutcracker esophagus, and IEM were diagnosed similarly with the second interpretation in nearly 75% of cases (69%, 84%, 77%, and 71%, respectively). However, DES, hypotensive LES, and hypertensive LES were diagnosed only as such on the second interpretation in less than 50% of cases (27%, 41%, and 48%, respectively). Table 6 details the distribution of the second manometric interpretation as a function of the initial
Table 3. Presenting Symptoms for Each Manometric Diagnosis Initial diagnosis
Regurgitation
Dysphagia
Chest pain
Heartburn
Chronic cough
Normal (n ⫽ 10) Achalasia (n ⫽ 10) NSEMD/IEM (n ⫽ 19) Hypotensive LES (n ⫽ 10) Hypertensive LES (n ⫽ 8) DES (n ⫽ 9) Nutcracker (n ⫽ 6)
20% 20% 27% 30%
20% 70% 17% 10% 63%
10%
40% 10% 27% 40%
10%
22% 17%
50%
17% 10% 37% 67% 33%
11%
12% 10%
March 2005
ESOPHAGEAL MANOMETRY
Table 4. Intraobserver Score for Highly and Moderately Experienced Interpreters Highly experienced Individual interpreters
84%
80
Moderately experienced
1
2
3
1
2
3
13 .45
20 .63 .61
15 .75
9 .84
5 .69 .65
10 .43
.63
P value
69%
71%
77%
60 41%
40 ⬍.001b NS ⬍.001b
48%
27%
20 0
diagnosis. For example, 84% of the patients initially diagnosed with achalasia had a similar diagnosis, whereas 9% were diagnosed with IEM and 7% were diagnosed as normal (Table 6). Similarly, the second interpretation of normal tracings was diluted with IEM (21%), hypotensive LES (6%), DES (3%), and hypertensive LES (1%).
Figure 1. Percent agreement between the second interpretation compared with the initial diagnosis. Patients with the initial diagnosis of normal, achalasia, nutcracker esophagus, and NSEMD/IEM were diagnosed similarly with the second interpretation in nearly 75% of cases (69%, 84%, 77%, and 71%, respectively). However, DES, hypotensive LES, and hypertensive LES were diagnosed only as such on the second interpretation in less than 50% of cases (27%, 41%, 48%, respectively).
as ia
N or N SE ma l M D /IE N M ut cr ac ke r
H yp D ot ES en H s yp iv e er L te ns ES iv e LE S
NS, not significant (between highly and moderately experienced interpreters). aNumber of tracings evaluated by each interpreter twice over a 2-year period. bP ⬍ .001 suggests significant individual consistency of interpreters.
Ac ha l
Na Mean Overall
100
221
Reasons for Inconsistencies In cases in which agreement was variable, a methodologic group review by all 6 interpreters was conducted. Clinical data including manometry, imaging, pH monitoring, and endoscopic studies were analyzed. A consensus was reached by the 6 esophageal physicians on the most likely diagnosis. Based on this exhaustive analysis, the 3 most frequent sources of inconsistency in interpretation of manometric tracings were as follows: (1) technically inadequate study (20%), (2) poor adherence to established definitions for manometric diagnoses (50%) (Table 1), and (3) misinterpretation of intrabolus pressure (30%). For example, on review of the 10 patients initially diagnosed with achalasia, the consensus group agreed with this diagnosis in 7 patients. Conversely, in 2 patients initially diagnosed with DES and in Table 5. Interobserver Agreement () for Interpreter Groups Based on Manometric Diagnosis Interpreter experience level
Normal and achalasia N ⫽ 20
All others N ⫽ 52a
Overall N ⫽ 72
Low experience (N ⫽ 2) Midexperience (N ⫽ 3) Highly experienced (N ⫽ 3) Overall (N ⫽ 8)
.66 .71 .68 .68
.16 .28 .35 .27
.30 .39 .44 .37
NOTE. Diagnoses of achalasia and normal motility were combined and used as internal standard for data validity and compared with other manometric diagnoses combined. aOther diagnoses include nonspecific/ineffective motility, DES, hypotensive LES, hypertensive LES, and nutcracker esophagus.
1 patient initially diagnosed with IEM, achalasia was the final correct diagnosis. Thus, the sensitivity, specificity, and positive and negative predictive value of esophageal manometry for diagnosing achalasia was 70%, 95%, 70%, and 95%, respectively. The misinterpretation of LES pressure led to a misdiagnosis for hypertensive LES, hypotensive LES, and normal motility; whereas variations in intrabolus pressure measurements misclassified DES, achalasia, and even normal motility. The latter misclassification was highly important because it misclassified patients with achalasia or normal tracings to DES or ineffective esophageal motility disorders. Such misclassification resulted in delayed diagnosis and treatment of patients with achalasia.
Discussion This study was a prospective, multi-interpreter, randomized, blinded study assessing the reliability of esophageal manometry interpretations. We found a strong intra- and interobserver agreement in diagnosing the 2 extremes in the spectrum of motility diagnoses, normal and achalasia, with a score of .63 and .68, respectively. However, despite the level of expertise, interpretation of other esophageal motility disorders have poor interobserver agreement, with a score .27 (Table 5). We found that overall intraobserver reliability was good for both high- and moderate-level interpreters, with a score of .61 and .65, respectively. However, experience
222
NAYAR ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3
Table 6. Proportion of Second Interpretation (Horizontal) as a Function of the Initial Diagnosis (Vertical)
Achalasia Normal NSEMD/IEM Nutcracker DES Hypotensive LES Hypertensive LES
Achalasia
Normal
NSEMD/IEM
Nutcracker
DES
Hypotensive LES
Hypertensive LES
84%
7% 69% 17% 2% 14% 14% 21%
9% 21% 71% 11% 49% 41% 13%
3% 3% 6% 27%
6%
3% 77% 9%
1% 1% 2% 1%
5% 2% 4% 7%
8%
41% 3%
48%
NOTE. Numbers on diagonal indicate level of percent agreement in diagnoses between first and second interpretations. Note the decreased percent agreement for DES, hypotensive LES, and hypertensive LES.
did make a difference in differentiating normal and achalasia tracings from others. More experienced interpreters had higher interobserver agreements ( ⫽ .35) than midlevel interpreters ( ⫽ .28) and than those with least experience ( ⫽ .16) (Table 5). Percent agreement between the first and the second interpretation was highest for tracings with the following diagnoses: achalasia, normal, nonspecific esophageal motility disorder (NSEMD)/IEM, or nutcracker esophagus. Hypo- and hypertensive LES and DES diagnoses had the lower percent agreements, with the latter having the lowest agreement (Figure 1). Only 27% of patients initially diagnosed with DES were diagnosed as such on the second interpretation. The majority (49%) of these DES patients were diagnosed with IEM on the second manometry (Table 6). This is understandable because manometrically the difference between these 2 diagnoses may be based on distal esophageal contraction amplitude, being low in the IEM and normal to high in DES (Table 1). In addition to final interpretation, data were collected on specific manometric parameters leading to the final diagnoses by each interpreter (Appendix). Several reasons for discrepancies were identified. First was the lack of strict adherence to published criteria (Table 1). It is possible that the interpreters still used outdated criteria for manometric diagnoses and did not use the most recently published information. In post hoc analysis, it was obvious that different gastroenterologists used different interpretation criteria. For example, the assessment of intrabolus pressure take-off points and peristalsis varied considerably in those tracings with only moderate, fair, or poor agreement. This is understandable because published criteria do not address interpretation of intrabolus pressure. In patients with early achalasia, classic isobaric contractions may not be present and the diagnosis may depend on inclusion or exclusion of intrabolus pressures. This misclassification is important because it can result in delayed therapy for this group of patients. Also, there was a difference among interpreters on the normal cut-off level used for LES and esophageal body
amplitude pressures. Some patients were classified as having low LES pressure who had, by strict definition, pressures in the low-normal category. Thus, upper and lower limits of normal LES pressure were not observed uniformly, which explains the low percent agreement for hypo- and hypertensive LES pressure tracings (Figure 1, and Table 6). Additional reasons for this misclassification may be interpreter bias with regard to the importance of degree of abnormality of LES pressure vs. esophageal body amplitude or peristalsis. A patient with low LES pressure and low-amplitude esophageal body contractions may be identified as hypotensive LES with one interpreter whereas another interpreter may identify them as having IEM. Similarly, some may consider LES pressure of 12 mm Hg to be normal whereas others may consider it low, especially in those being considered for surgical fundoplication. The diagnostic criteria listed in Table 1 were not distributed to participants. Obligatory criteria for a particular diagnosis were at the discretion of the reader. Interpreter consistency in this study could have been improved by reinforcing diagnostic criteria among interpreters and by using computer-derived data points; however, such an attempt would have been artificial because this normally is not performed in clinical practice. Another reason for discrepant interpretations was the poor quality of the tracings because of the technical inadequacy of the procedure. Examples of this included an insufficient number of wet swallows and insufficient time between swallows to assess esophageal contractions and peristalsis accurately. Adequacy of esophageal manometry is highly dependent on the technical expertise of the person performing the test, thus highlighting the need for adequate training and oversight of the personnel involved in performing esophageal manometry testing. Despite rigorous attention to this matter in our center, technical performance was an important but not the main reason for interpreter variability. Two variables also may have impacted the results of this study. These included the use of printed tracings and the lack of information on patients’ clinical data before
March 2005
interpretation. Analysis of esophageal motility may be easier with a computer monitor than with printed tracings, allowing more accurate pressure measurements enhanced by color and the ability to alter graphics scales. There currently are no data to suggest that this difference may play an important role in improving reliability of esophageal motility interpretation. Second, some may argue that interpretation of manometry should be undertaken with the clinical history at hand. Others believe that esophageal manometry should be interpreted independently of clinical information. Our study is a reflection of the reliability of manometric interpretation based on a lack of clinical data. Additionally, the first manometric diagnoses rendered by the reviewers in calculation of the intraobserver was based on knowledge of clinical data that may have slightly biased the score. It is possible that clinical information would improve manometric reliability; however, this needs to be tested in future studies. Finally, whether or not our findings are generalizable to the community physicians is unknown. One could argue that the results would be worse for the interpreters in the community given the poor agreement even among motility experts. On the other hand, it may be argued that the interpretation in the communitybased centers may be more uniform if individuals were trained by the same equipment providers. Because of such inconsistent interpretation, especially with respect to diagnoses other than normal and achalasia, esophageal manometric diagnoses should be complemented by clinical and radiographic data to allow correct diagnosis and improved patient management. In patients presenting with classic signs and symptoms of achalasia, manometry is a reliable diagnostic tool and is independent of interpreter experience. However, in difficult tracings in which intrabolus pressure inclusion is in doubt, a barium swallow should complement manometry to ensure a correct diagnosis. Therefore, our data are in agreement with previous studies by Massey et al.,7 who found both manometry and radiographic studies to have inherent limitations and suggested a complementary role for these techniques. They showed that radiographic examinations were insensitive to contractions in esophageal segments devoid of bolus whereas manometry was insensitive to contractions that did not occlude the esophageal lumen. A normal esophageal manometry interpretation is both highly specific and highly repeatable among esophageal experts. Our data suggest that we must refine or even eliminate diagnostic criteria for other motility disturbances so as to avoid overlapping diagnoses. Given the shortcomings of traditional manometry, some have suggested the use of other methods in the diagnosis of manometric disorders.8 –11 For example, topographic
ESOPHAGEAL MANOMETRY
223
methods in which manometric recordings are examined by spatial as well as temporal relationships of pressure data suggest that such methods, especially when automated, may have increased accuracy in manometric diagnoses.8 In a study of 212 patients with both the traditional pull-through technique as well as a topographic approach, Clouse et al.8 reported increased accuracy of diagnosing severe esophageal motor dysfunction with the latter. Topographic methods correctly diagnosed 97.2% of the diagnosis in patients with aperistalsis, which was superior to the traditional method (77.8%). Therefore, until further studies are performed in this area, given the high degree of variability of the traditional manometry, the clinical relevance of manometric diagnoses other than normal and achalasia need to be better defined. Meanwhile, in interpretation of manometric tracings, adherence to published criteria for diagnoses is recommended.
Appendix Analyses Work Sheet Lower esophageal sphincter ΠNormal ΠAbnormal Relaxation: ΠNormal ΠAbnormal ΠIndeterminate Esophageal body Peristalsis: ΠNormal ΠAbnormal Amplitude of contraction: ΠNormal ΠAbnormal Duration of contraction: ΠNormal ΠAbnormal Contraction description: ΠNormal ΠIsobaric ΠNontransmitted ΠSimultaneous ΠMultiple peaks ΠOther Manometric diagnosis ΠNormal ΠNutcracker esophagus ΠHypertensive LES ΠHypotensive LES ΠDiffuse esophageal spasm ΠNonspecific/ineffective esophageal motility ΠAchalasia Mean Pressure:
References 1. Kahrilas PJ, Clouse RE, Hogan WJ. American Gastroenterological Association position statement and technical review on the clinical use of esophageal manometry. Gastroenterology 1994;107: 1865–1884.
224
NAYAR ET AL
2. Richter JE. Oesophageal motility disorders. Lancet 2001;358: 823– 828. 3. Spechler SJ, Castell DO. Classification of oesophageal motility abnormalities. Gut 2001;49:145–151. 4. Smithline A, Hawes R, Lehman G. Sphincter of Oddi manometry: interobserver variability. Gastrointest Endosc 1993;39:486 – 491. 5. Richter JE, Wu WC, Johns DN, et al. Esophageal manometry in 95 healthy volunteers. Dig Dis Sci 1987;32:583–592. 6. Brennan P, Silman A. Statistical methods for assessing observer variability in clinical measures. Br Med J 1992;304:1491–1494. 7. Massey BT, Dodds WJ, Hogan WJ, et al. Abnormal esophageal motility. An analysis of concurrent radiographic and manometric findings. Gastroenterology 1991;101:344 –354. 8. Clouse RE, Staiano A, Alrakawi A, et al. Application of topographical methods to clinical esophageal manometry. Am J Gastroenterol 2000;95:2720 –2730.
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 3
9. Clouse RE, Staiano A. Topography of the esophageal peristaltic pressure wave. Am J Physiol 1991;261:G677–G684. 10. Clouse RE, Staiano A. Topography of normal and high amplitude esophageal peristalsis. Am J Physiol 1993;265:G1098 –G1107. 11. Clouse RE, Staiano A, Alrakawi A. Development of a topographic analysis system for manometric studies in the gastrointestinal tract. Gastrointest Endosc 1998;48:395– 401.
Address requests for reprints to: Michael F. Vaezi, MD, PhD, FACG, Department of Gastroenterology and Hepatology, Center for Swallowing and Esophageal Disorders, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195. e-mail: [email protected]; fax: (216) 444-6302.