167.full

AJR:176, January 2001

167

Abdominal Radiography Findings in Small-Bowel Obstruction:

Relevance to Triage for Additional Diagnostic Imaging

OBJECTIVE

.

Our aim was to determine which findings on abdominal radiography arerelevant for distinguishing complete or high-grade partial small-bowel obstruction from low-grade partial or no small-bowel obstruction.

MATERIALS AND METHODS

.

Admitting abdominal radiographs with the patients inthe supine and upright positions were scored for 25 different findings in 81 patients with clin-ically suspected small-bowel obstruction. Forty-one patients had complete or high-grade par-tial small-bowel obstruction, and 40 had low-grade partial small-bowel obstruction or noobstruction as determined by enteroclysis examination. Abdominal radiography findings weresubjected to statistical analysis for correlation with degree of obstruction.

RESULTS

.

Of 12 radiographic findings strongly associated (

p

< 0.05) with the severity ofobstruction, two findings were found to be the most significant (

p

≤

0.0003) and predictive ofa higher grade small-bowel obstruction: the presence of air–fluid levels of differential heightin the same small-bowel loop and the presence of a mean air–fluid level width greater than orequal to 25 mm on upright abdominal radiographs.

CONCLUSION

.

When both critical findings are present, the degree of small-bowel ob-struction is likely high-grade or complete. When both signs are absent, small-bowel obstruc-tion is likely low-grade or nonexistent. Upright abdominal radiographs are important in theexamination of patients with suspected small-bowel obstruction and may contribute to the im-aging triage of these patients.

ccording to a recent report, con-ventional abdominal radiogra-phy remains the preferred

method of initial radiologic examination ofsymptomatic patients suspected of small-bowel obstruction [1]. These investigatorsand others advocate that if further diagnosticimaging is warranted to establish a more spe-cific diagnosis or to obtain information perti-nent to the clinical treatment, the results ofthe abdominal radiographic examinationshould serve as the basis for triage [1, 2]. It isrecommended that if the initial abdominalradiographs suggest a complete or high-grade partial small-bowel obstruction, CTwould be the preferred additional imagingmodality if surgery is not imminentlyplanned. This approach is supported by thesensitivity of CT for establishing the diagno-sis of a high-grade or complete small-bowelobstruction, reported to be 82–100%, and thepotential value of CT to modify the patient’streatment [1, 3–7]. Conversely, if the initial

radiographic findings are interpreted as nor-mal, equivocal, or suggestive of a low-gradepartial small-bowel obstruction, an examina-tion using a direct contrast material infusionof the intestinal lumen is preferred. Enteroc-lysis and CT enteroclysis satisfy these re-quirements and are recommended on thebasis of their high sensitivity for the diagno-sis of a less severe or low-grade small-bowelobstruction, their ability to exclude the diag-nosis of obstruction, and the comparativelylower diagnostic sensitivity of conventionalCT in this clinical situation [8–10].

Experienced radiologists have shown accu-racy and agreement for the diagnosis of acutehigh-grade small-bowel obstruction on the in-terpretation of abdominal radiographs, but thediagnosis of low-grade obstruction is less cer-tain [8, 11]. Additionally, no substantial dataexist regarding the capability of abdominal ra-diography to distinguish the severity of ob-struction of the small bowel. In a study ofpatients with a clinical suspicion of small-

John C. Lappas

1

Benedicto L. Reyes

2

Dean D. T. Maglinte

3

Received February 29, 2000; accepted after revision June 12, 2000.

1

Department of Radiology, Indiana University School of Medicine, Wishard Memorial Hospital, 1001 W. Tenth St., Indianapolis, IN 46202. Address correspondence to J. C. Lappas.

2

Department of Radiology, St. Francis Hospital, 1500 Albany St., Beech Grove, IN 46107.

3

Department of Radiology, Indiana University School of Medicine, 550 N. University Blvd., Indianapolis, IN 46202.

AJR

2001;176:167–174

0361–803X/01/1761–167

© American Roentgen Ray Society

A

168


Lappas et al.

bowel obstruction who were later proven tohave various degrees of small-bowel obstructionor no obstruction, we objectively determined thefrequency and pertinent measurement of the var-ious findings on abdominal radiographs thatare commonly used to evaluate small-bowelobstruction. Considering the implications forsubsequent patient workup based on the useof abdominal radiographs as a diagnostic tri-age tool, we attempted to establish whichfindings correlated with the degree of small-bowel obstruction and to determine fromthose findings which were the most signifi-cant in differentiating the severity of small-bowel obstruction.

Materials and Methods

All patients were recruited retrospectively byreview of the radiology clinical database accordingto the study requirements as follows: a diagnosisof small-bowel obstruction was suspected clini-cally and was the primary reason for referral forimaging, admitting abdominal radiographs wereobtained with the patient in the supine and uprightpositions and were available for review, and an en-teroclysis examination was performed within 5days of the admitting abdominal radiography.

Study patients were enrolled in a chronologi-cally consecutive fashion until an equivalent num-ber of patients constituted each of four groups: 21patients had complete small-bowel obstruction, 20had high-grade and 20 had low-grade partialsmall-bowel obstruction, and 20 had no obstruc-tion. Thus, a total of 81 patients constituted thestudy population.

Findings at enteroclysis examination served asthe basis for the assignment of patients to a studygroup. Enteroclysis examinations were performedby senior staff radiologists (including one of theauthors) with expertise in gastrointestinal imagingor by residents directly supervised by the staff ra-diologist. All enteroclysis examinations were re-viewed retrospectively and independently of theanalysis of the abdominal radiographs and wereevaluated by consensus of two senior gastrointesti-nal radiologists to accurately determine the gradeof small-bowel obstruction or the finding of no ob-struction. The abdominal radiograph routinely ob-tained as a preliminary scout view with theenteroclysis examination was deleted from the en-teroclysis review.

For the diagnosis of low-grade, high-grade, andcomplete small-bowel obstruction and for the ex-clusion of mechanical obstruction on the enteroc-lysis examination, we followed the definitions inpublished criteria [1, 8, 9]. Obstruction was diag-nosed when a transition zone—defined as achange in caliber of the small-bowel lumen from adistended segment proximal to the site of obstruc-tion to a segment that was decreased in caliber orcollapsed distal to the obstruction—was seen atenteroclysis. Our enteroclysis criteria were 3.5 cm

as the upper limit for normal caliber of the jejunallumen and 2.5 cm as the upper limit for the ileallumen. Low-grade partial small-bowel obstructionwas diagnosed when the contrast media arrived atthe transition zone without delay and sufficientcontrast material flowed through the obstructioninto the loops beyond the obstruction so that mu-cosal fold patterns in these distal loops wereclearly defined (Fig. 1A). High-grade partialsmall-bowel obstruction was diagnosed whenthere was a definite delay in the flow of contrastmedia to and through the site of obstruction andonly small amounts of contrast material enteredthe collapsed bowel loops beyond the obstruction(Fig. 1B). Complete small-bowel obstruction wasdiagnosed by an absence of contrast media enter-ing the bowel loops distal to a discrete point of lu-men obstruction (Fig. 1C). The diagnosis ofmechanical small-bowel obstruction was excludedat enteroclysis when unimpeded flow of contrastmedia was observed fluoroscopically from theduodenojejunal junction to the colon and a nor-mal-caliber small bowel without transition zoneswas documented.

The patients’ final discharge diagnoses, includ-ing pertinent clinical and laboratory data, surgicalfindings, and diagnostic findings on enteroclysis,were used to determine the cause of small-bowelobstruction and the cause of abdominal symptomsin the patients without small-bowel obstruction.

Abdominal radiographs were reviewed retrospec-tively and in random order, and the evaluation wasmade by consensus of the authors, who included twosenior gastrointestinal radiologists and a senior radi-ology resident. Evaluations were made withoutknowledge of the corresponding enteroclysis resultor the patient’s clinical outcome. The patients’ su-pine and upright abdominal radiographs at admis-sion were scored with respect to 25 differentvariables, 17 of which involved a specific numericmeasurement (continuous variable) and eight ofwhich involved the presence or absence of a condi-tion or variable (categoric variable). A diagnostic in-terpretation of the abdominal radiographs was notperformed as part of the scoring evaluation.

The specific variables scored on the supine ra-diograph included the number of distended (

≥

2.5-cm lumen diameter) or nondistended (<2.5-cmlumen diameter) gas-filled or fluid-filled small-bowel loops; the maximum and mean small boweldiameter (in millimeters); and the maximum andmean diameter (in millimeters) of various colonsegments, including the cecum, ascending colon,transverse colon, descending colon, and rectum(Fig. 2). Presence of the stretch sign, defined assmall-bowel gas arrayed as stripes perpendicularto the long axis of the bowel, as an indicator of apredominantly fluid-filled small-bowel loop, wasnoted (Fig. 3).

Comparative ratios of small-bowel size to colonsize were calculated from the maximum lumen di-ameter measurements, and the occurrence of smallbowel–colon ratios greater than 0.5 and greaterthan 1.0 were recorded.

Variables scored on the upright radiograph in-cluded the number and width (in millimeters) ofair–fluid levels measuring greater than or equal to10 mm in width; the presence and differentialheight (in millimeters) of air–fluid levels in thesame small-bowel loop (Fig. 4); the presence ofthe string-of-beads sign, defined as a series of air–fluid levels individually measuring less than 10mm in width (Fig. 4); and the presence of a dis-tended stomach.

Statistical Analysis

Two categories of small-bowel obstructionwere defined for statistical analysis of the data. Asevere or higher grade group of obstruction(

n

= 41) included patients with complete and high-grade partial small-bowel obstruction. A lesssevere or lower grade group (

n

= 40) includedpatients with low-grade partial small-bowel ob-struction and patients with no obstruction.

In the initial screening process, each scored ra-diographic finding or variable was examined on aunivariate basis to evaluate its individual importance.During this screening, Fisher’s exact test was usedfor the categoric variables. For the continuous (nu-meric) variables,

t

tests were performed. All vari-ables, categoric or continuous, yielding univariate

p

values of less than 0.3 were considered relevant andpotentially significant in determining the severity ofsmall-bowel obstruction.

To transform significant continuous (numeric)variables into more clinically applicable radio-graphic findings, cut-points (threshold values)were generated using tree-based modeling [12].This method identified the optimal threshold valuefor continuous variables that would yield the bestpossible statistical separation between the catego-ries of higher grade and lower grade small-bowelobstruction. Variables transformed in this mannerwould then be more appropriate for use in predict-ing the severity of small-bowel obstruction.

Finally, a backward elimination logistic regres-sion model was used to eliminate relevant vari-ables that contained similar or noncontributoryinformation [13, 14]. All explanatory variables(

p

< 0.3) were initially included in the model, andnonsignificant variables were removed one at atime until only significant variables remained. Theprocess optimally generates a concise final versionof the predictive statistical model.

All methods of statistical analysis were per-formed using S-Plus (StatSci, a division of Math-Soft, Seattle, WA) and SAS (SAS Institute, Cary,NC) software.

Results

Enteroclysis examinations were per-formed within the first 72 hr of admission in34 (83%) of 41 patients in the higher gradecategory of obstruction and in 35 (88%) of40 patients in the lower grade category of ob-struction. Only 15% (12/81) of enteroclysisexaminations were performed on either the

Abdominal Radiography of Small-Bowel Obstruction


169

fourth or fifth day after the patient’s admit-ting abdominal radiographs.

Of the 41 patients in the higher grade cate-gory of obstruction, the cause of the obstruc-

tion was adhesions in 28 patients, neoplasmin six, ischemic stricture in three, hernia intwo, intussusception in one, and perforatedappendicitis in one patient. Of the 40 patients

in the lower grade category of obstruction,low-grade partial obstruction was due to ad-hesions in 19 patients and Crohn’s disease inone patient, whereas patients without small-

C

BA

Fig. 1.—Grades of small-bowel obstruction. A, Low-grade partial obstruction in 70-year-old man. Enteroclysis shows contrast medium in di-lated small-bowel loop flowing (curved arrow ) into discrete transition zone, represented by com-pressed lumen segment (straight arrow). Contrast medium enters distal (D) bowel at rate of flowsufficient to achieve moderate lumen distension and to reveal mucosal folds. B, High-grade partial obstruction in 72-year-old woman. Enteroclysis shows contrast medium indilated small-bowel loop flowing (curved arrow ) into severely compressed lumen segment(straight arrows). Rate of flow is impeded to degree that small amounts of contrast medium entercollapsed distal (D) bowel loops that have no defined mucosal folds. C, Complete obstruction in 84-year-old man. Enteroclysis shows that contrast medium fails toflow (curved arrow ) through completely compressed lumen segment (straight arrow ), and bowelloops distal to obstruction are not visualized.

170


Lappas et al.

bowel obstruction showed either nonobstruc-tive adhesions (6 patients) or various condi-tions including enteritis in five patients,chronic abdominal pain in four, gastroparesisin two, adynamic ileus associated with medi-cation or electrolyte imbalance in two, andtransient intussusception in one patient.

The frequency of various (categoric) radio-graphic findings in the two defined categoriesof small-bowel obstruction is summarized inTable 1. Table 2 summarizes the means andstandard deviations of the numeric variablesscored for each of the two categories.

Table 3 lists the results of the initialscreening process. Twelve variables werestrongly associated with each group, includ-ing the maximum small-bowel diameter, themean air–fluid level width, and the presenceof a differential air–fluid level. All variableshad

p

values of less than 0.05. One finding,the string-of-beads sign, was marginally as-sociated with each group, which in this studywas defined as a

p

value greater than 0.05and equal to or less than 0.15. Twelve othervariables, including the number of air–fluidlevels, the measurements of colon size, thepresence of colonic gas or feces, and anyspecific ratio of the maximum small-boweldiameter to maximum colon diameter, hadno significant correlation with the twogroups (

p

> 0.3). Table 4 lists the threshold values gener-

ated for each of the significant continuous(numeric) variables. The

p

values listed arean approximate measurement of the statisti-cal separation the threshold value creates be-tween the two groups.

Of the thirteen variables that passed theinitial statistical screening process (

p

< 0.3,Table 3), two were determined by logistic re-gression analysis to be the most significantpredictors between the two categories of se-verity of small-bowel obstruction. Thesewere the presence of air–fluid levels of differ-ential height in the same bowel loop and thepresence of a mean air–fluid level widthgreater than or equal to 25 mm. Theirstrength as markers of a small-bowel obstruc-tion of higher grade can be expressed as anodds ratio. The ratio is 7.2 for a mean air–fluid level width of 25 mm or greater, and theratio is 4.5 for the presence of a differentialair–fluid level.

Table 5 presents the data for the association ofthe two most significant radiographic findingswith the category of obstruction. The presence ofboth findings meant an almost 86% chance thatthe patient had a high-grade partial or complete

Fig. 2.—Abdominal radiograph with patient supine in 27-year-old woman with high-grade partial small-bowelobstruction shows method of bowel lumen diametermeasurement. Small-bowel loops (dotted lines) and co-lon segments (solid lines) were measured from widestbowel margin transverse to long axis of bowel. Leftlower quadrant small-bowel loop measures 32 mm in di-ameter, and segment of transverse colon measures 40mm in diameter.

Fig. 3.—Abdominal radiograph with patient supinein 34-year-old man with complete small-bowel ob-struction. Stretch sign is depicted by abnormaldistention of predominantly fluid-filled small-bowel loop where luminal gas has striped appear-ance oriented perpendicular to long axis of bowel(arrows).

Fig. 4.—Abdominal radiograph with patient upright in 36-year-old woman with high-grade partial small-bowelobstruction. Small-bowel loops A and B show differential height of air–fluid levels (dotted lines) and were mea-sured as vertical height (solid arrows) between corresponding air–fluid levels in same bowel loop. Differentialair–fluid height measures 31 mm in loop A and 5 mm in loop B. String-of-beads sign is depicted by air–fluid levelsmeasuring less than 10 mm in diameter (open arrow ).



171

small-bowel obstruction, because 18 of the 21patients who showed the concurrent presence ofboth significant findings on abdominal radio-graphs were in the category of higher grade ob-struction. When neither finding was presentthere was an 83% chance that the patient had alow-grade partial small-bowel obstruction or noobstruction, because 24 of the 29 patients whohad neither finding on radiographs were in thelower grade category. Both or neither of thesefindings was seen in 50 (62%) of 81 cases.

Overall, the presence of a differential air–fluid level on the upright abdominal radio-graph was noted in 56% (45/81) of patients.A mean air–fluid level width greater than orequal to 25 mm was noted on the upright ra-diograph in 35% (28/81) of patients.

Discussion

Abdominal radiographs provide valuablecost-effective information in the early exami-

nation of patients with various abdominalsymptoms and disorders, particularly intesti-nal obstruction. Limitations in the diagnosisof small-bowel obstruction are recognized,however, because the intestinal gas patternupon which the interpretation of radiographsis largely based varies with certain character-istics such as the severity, level, and durationof the obstruction. In patients for whom ab-dominal radiographs are insufficient to confi-dently establish a diagnosis of small-bowelobstruction or to reasonably assess the sever-ity of obstruction, additional diagnostic im-aging becomes necessary.

The methodology in this study of combin-ing symptomatic patients into two functionalcategories of obstruction was based on theappropriateness of the additional imagingevaluation as related to its sensitivity for thediagnosis of small-bowel obstruction and theimplications for subsequent clinical manage-ment [1–7].

The grouping of patients with low-gradepartial small-bowel obstruction with patientswith no obstruction into a lower grade ob-struction category is justified on the basis oftheir presenting symptoms being consistentwith intestinal obstruction and on the basisthat the recommended method of additionalradiologic workup is similar for these pa-tients. Although patients with low-grade par-tial small-bowel obstruction usually presentwith abnormal abdominal radiographic find-ings diagnostic of obstruction, approxi-mately 40% of such patients have normal orequivocal findings [1, 8]. Similarly, evensymptomatic patients who are eventuallyshown to have no small-bowel obstructionmay have equivocal radiographic findingssuggestive enough to warrant further imag-ing to exclude obstruction. The sensitivityand accuracy of enteroclysis for defining aminimal degree of obstruction and establish-ing small-bowel normality supports the useof this technique for the further examinationof patients in either group [8].

The grouping of patients with high-gradepartial or complete small-bowel obstruction intoa category of higher grade obstruction is basedon the premise that no substantive differencesexist in the imaging evaluation or clinical man-agement of these two patient groups [2, 7]. CTis sensitive and highly accurate for the diagnosisof severe grades of small-bowel obstruction andis a technically optimal method of examiningthese patients [1, 3–7].

If abdominal radiography is to function asa decision tool or triage point for the direc-

Note.—Findings are categoric variables. Higher grade category (n = 41) includes high-grade partial and complete obstruc-tion, lower grade category (n = 40) includes low-grade partial obstruction and no obstruction.

TABLE 1 Frequency of Abdominal Radiography Findings According to Category of Small-Bowel Obstruction

FindingHigher Grade Lower Grade

No. % No. %

Distended stomach 7/41 17 8/40 20Differential air–fluid levels 31/41 76 14/40 35Stretch sign 14/41 34 2/40 5String-of-beads sign 16/41 39 9/40 23Presence of cecal fluid 6/41 15 6/40 15Presence of colon gas or feces 36/41 88 7/40 18Small bowel–colon ratio > 0.5 36/41 88 27/40 68Small bowel–colon ratio > 1.0 16/41 39 10/40 25

Note.—Findings are continuous (numeric) variables. Higher grade category (n = 41) includes high-grade partial and com-plete obstruction, lower grade category (n = 40) includes low-grade partial obstruction and no obstruction.

TABLE 2 Mean and Standard Deviation (SD) Measurements of Abdominal Radiography Findings According to Category of Small-Bowel Obstruction

FindingHigher Grade Lower Grade

Mean SD Mean SD

With patient supineNo. of gas-filled small-bowel loops > 2.5 cm 3.37 2.55 1.55 1.91No. of fluid-filled small-bowel loops > 2.5 cm 0.98 1.15 0.27 0.51No. of gas- or fluid-filled small-bowel loops < 2.5 cm 1.39 2.15 2.83 2.60Cecum width (mm) 19.3 21.5 19.6 21.0Ascending colon width (mm) 19.0 20.1 22.8 16.4Transverse colon width (mm) 16.7 21.7 21.1 21.0Descending colon width (mm) 11.7 15.0 12.6 12.2Rectum width (mm) 12.0 19.5 13.6 19.1Maximum small-bowel diameter (mm) 35.7 11.0 25.7 10.3Mean small-bowel diameter (mm) 29.0 10.4 21.6 8.49Maximum colon diameter (mm) 36.9 18.8 35.0 17.2Ratio of small-bowel diameter to colon diameter 6.13 14.4 4.02 10.1

With patient upright:No. of air–fluid levels > 10 mm 5.39 3.31 4.38 3.87Maximum air–fluid level width (mm) 37.0 19.4 22.3 19.2Mean air–fluid level width (mm) 25.6 9.88 16.0 11.3Maximum differential air–fluid level (mm) 10.6 11.4 5.15 8.64Mean differential air–fluid level (mm) 8.80 8.59 4.56 7.94

172


Lappas et al.

tion of radiologic imaging of patients sus-pected of small-bowel obstruction, thendocumenting the radiographic findings thatbest distinguish between higher grade andlower grade obstruction would be of value.

Therefore, our study was designed to corre-late the occurrence and measurement of all po-tential findings on abdominal radiographs withthe severity of small-bowel obstruction as itwas established in clinical cases by an accuratemethod—enteroclysis—using specific criteriafor categorization. In an ideal situation, all en-teroclysis examinations would be performedon the same day as the patient’s clinical pre-sentation and admission abdominal radiogra-phy to optimize the concordance between thepresenting radiographic findings and the pa-tency of the small-bowel lumen as determinedby enteroclysis. Although most enteroclysisexaminations were performed with reasonableclinical promptness, in some patients any de-gree of small-bowel abnormality may have re-solved or perhaps even worsened by the timeof enteroclysis. A prospective study mighthave provided an ideal analysis.

Valid correlations also require an objectiverecording of the radiographic findings. Ab-dominal radiograph interpretations weretherefore conducted randomly and by a con-sensus of observers who were unaware of thecorresponding enteroclysis diagnosis. Insome cases, particularly patients with radio-graphs showing numerous distended small-bowel loops, a precise accounting of thenumber of loops and the measurement of allloop diameters could not be made with abso-lute certainty. Determining the margins ofpredominantly fluid-filled small-bowel loopsfor diameter measurement is more difficultthan determining the margins of gas-filledloops, and measurement of fluid-filled loopscould be underestimated in comparison. Ap-preciating the presence of differential air–fluid levels can also be difficult in cases withoverlapping bowel loops and these levels arepotentially misrepresented, probably under-estimated, when interpreting upright abdomi-nal radiographs. In most instances, however,we believe that scoring the various findingswas straightfoward and seldom problematicwith careful scrutiny of the radiographs. Theextensive data collected, the calculations ofmean measurements in addition to individualmeasurements, and an equivalent number ofradiographs being scored from the differentgroups, are factors that may minimize the ef-fect that potential inaccuracies in data collec-tion would have on results.

Note.—Findings include categoric and continuous (numeric) variables. Higher grade category (n = 41) includes high-gradepartial and complete obstruction, lower grade category (n = 40) includes low-grade partial obstruction and no obstruction. Thefollowing findings are not associated ( p > 0.3) with distinguishing grade of obstruction: distended stomach, number of air–fluidlevels > 10 mm, presence of colon gas or feces, presence of cecal fluid, colon segment width, maximum and mean colon diame-ters, small bowel–to-colon ratio > 1.0, and ratio of small-bowel diameter to colon diameter.

at test.bFisher’s exact test.

TABLE 3 Statistical Association of Abdominal Radiography Findings for Distinguishing Higher Grade from Lower Grade Small-Bowel Obstruction

Finding p

Strongly associated ( p < 0.05)Maximum small-bowel diameter 0.0001a

Mean air–fluid level width 0.0001a

Presence of differential air–fluid levels 0.0003 b

No. of gas-filled small-bowel loops > 2.5 cm 0.0005a

Mean small-bowel diameter 0.0007a

No. of fluid-filled small-bowel loops > 2.5 cm 0.0007a

Stretch sign 0.001b

Maximum air–fluid level width 0.001a

No. of gas- or fluid-filled small-bowel loops < 2.5 cm 0.008 a

Maximum differential air–fluid level height 0.017a

Mean differential air–fluid level height 0.024a

Small bowel–colon ratio > 0.5 0.035b

Marginally associated ( p ≤ 0.15)String-of-beads sign 0.15b

aContinuous (numeric) variables strongly associated with grade of obstruction (Table 3). bDetermined by tree-based modeling [12].cDerived from Fisher’s exact test.

TABLE 4 Threshold Values for Radiographic Findings Involving a Numeric Measurement

Variablea Thresholdb pc

No. of gas- or fluid-filled small-bowel loops < 2.5 cm 2 Loops 0.0008No. of gas- or fluid-filled small-bowel loops > 2.5 cm 2 Loops 0.0003Mean small-bowel diameter 25 mm 0.003Ratio of small-bowel diameter to colon diameter 0.5 0.035Maximum air–fluid level width 35 mm 0.00005Mean air–fluid level width 25 mm 0.0003Maximum differential air–fluid level height 2 mm 0.0003Mean differential air–fluid level height 2 mm 0.00005

Note.—Two most significant findings are presence of differential air–fluid level and mean air–fluid level width ≥ 25 mm.Higher grade category (n = 41) includes high-grade partial and complete obstruction, lower grade category (n = 40) includes low-grade partial obstruction and no obstruction.

TABLE 5 Association of the Two Most Significant Abdominal Radiography Findings with Category of Small-Bowel Obstruction

CategoryNeither Present One Present Both Present

No. % No. % No. %

Higher grade 5/41 12 18/41 44 18/41 44Lower grade 24/40 60 13/40 33 3/40 8

Total 29/81 36 31/81 38 21/81 26



173

In our clinical setting—a large primarycare and tertiary referral center—the numberof symptomatic patients encountered with ei-ther low-grade partial obstruction or no ob-struction is considerably greater than thenumber of patients having a high-grade par-tial or complete small-bowel obstruction.Given the quantity of the data collected andanalyzed, the arbitrary enrollment of anequivalent number of patients in each studygroup was considered a reasonable approachto time and cost considerations of the studyand a satisfactory approach for valid statisti-cal comparisons between the groups.

The aim of the statistical analysis of thedata was to determine from the large numberof scored radiographic findings the most sig-nificant findings that might be used to differ-entiate the grade of small-bowel obstruction.Initially, a critical

p

value of 0.3 was chosento limit further analysis to the findings (vari-ables) that showed an association with ob-struction category. A lower cut-off

p

value,such as the traditional 0.05, was not used be-cause we did not want to prematurely elimi-nate radiographic findings (variables) thatwere potentially important once other vari-ables were taken into consideration. In es-sence, the value of the univariate screeningprocess was to identify findings likely to besignificant while reducing the number of to-tal variables for the final statistical analysis.

Because the use of radiographic findingscharacterized by a numeric measurement(termed a continuous variable in this analy-sis) would be difficult to implement clini-cally, it was necessary to transform thesevariables. The generation of cut-points bytree-based modeling accomplishes an appro-priate modification of the continuous vari-ables by defining a specific threshold valuewhich, by its presence or absence, representsa more clinically usable finding to distin-guish between the higher grade and lowergrade obstructions.

In the final analysis, a logistic regressionmodel was used to determine the most signifi-cantly predictive radiographic findings. Logisticregression is appropriate for identifying predic-tors when the dependent variable—in this study,the degree of obstruction—is in itself categoricin nature. All relevant variables with an initial

p

value of less than 0.3 were included in the modeland then were removed individually if they didnot contribute significantly to the prediction ofthe obstruction. For example, it would be un-likely for the maximum and mean air–fluid levelwidths to both be necessary as predictors. If both

are necessary, the logistic regression modelwould use both; otherwise, the model wouldchoose the more significantly informative one.After all noninformative variables were elimi-nated, the resulting version of the statisticalmodel was used for prediction of the grade ofsmall-bowel obstruction.

Results of this study show that abdominalradiographs can provide a distinction betweenthese groups of patients in most cases. Of the25 radiographic findings studied, several antic-ipated findings were by association significantin allowing distinction between higher andlower grades of obstruction such as small-bowel distention, with the maximally dilatedloops averaging 36 mm in diameter and ex-ceeding a size ratio of 0.5 with the colon, and anearly 2.5 times increase in the number of dis-tended loops in the abdomen.

These findings could be expected, giventhe nature of bowel obstruction, in that ab-normal distention of either gas-filled or fluid-filled intestine, or both, occurs proximally tothe site of obstruction and with relative col-lapse of the more distal intestine.

In particular, two findings derived from theupright abdominal radiograph were found tobe the most significant and most predictive ofthe higher grades of small-bowel obstruction:the presence of differential air–fluid levels anda mean width of air–fluid levels measuringgreater than or equal to 25 mm. Differentialair–fluid levels were present in 56% of all pa-tients and were observed with greater fre-quency on the radiographs of patients in thehigher grade than in the lower grade categoryof obstruction, 76% versus 35%, respectively.The measured differential height of these air–fluid levels was also notably increased with thegreater severity of obstruction. We did notshow in these data a significant thresholdheight of the differential air–fluid levels as adiscriminating diagnostic finding, as has beenreported by others [15]. In 1993, Harlow et al.[15] showed that the specificity for the diagno-sis of mechanical bowel obstruction increased,and diagnostic sensitivity decreased, as thethreshold height defining a differential air–fluid level increased. A critical differential air–fluid level height of greater than or equal to 20mm was associated with a high positive pre-dictive value (88%) for the diagnosis. Al-though the study by Harlow et al. involved aradiographic comparative analysis of 62 epi-sodes of mechanical bowel obstruction with 38episodes of adynamic obstruction (ileus) as de-termined largely by clinical information, ourstudy involved only two patients with a clinical

diagnosis of an adynamic obstruction. Perhapsthese differences in the patient populations be-ing compared, including differences in clinicalpresentation, account for the differences ob-served regarding a significant predictive heightof differential air–fluid levels.

That a critical width of air–fluid levelswas found to be highly significant likely re-flects its association with the greater degreeof abnormal distention of the small-bowelloops seen in higher grades of obstruction.Long air–fluid levels observed in cases ofobstruction have also been thought to corre-late with the amount of luminal fluid present,which is especially increased in obstructedloops of distal jejunum or ileum [16]. In ourstudy, we did not determine the anatomiclevel of the small-bowel obstruction, so weare unable to additionally comment on thisfeature. Corroborating the presence of dis-tended loops with copious luminal fluid wasthe significant occurrence of the stretch signevident on supine radiographs. Interestingly,air–fluid levels of small width (<10 mm) orthe radiographic string of beads, was onlymarginally associated with obstruction se-verity. Although this sign is considered animportant indicator of the presence of fluid-filled bowel, its frequent presence in bothhigher and lower grades of obstruction sug-gests that its occurrence is perhaps indepen-dent of luminal distention or is quite readilyperceived on upright radiographs.

Considerable material has been publishedabout the sensitivity of various radiographicfindings in the diagnosis of small-bowel ob-struction, particularly in regard to air–fluid lev-els and the importance of the uprightabdominal radiograph. The early observationsof Frimann-Dahl [17] on the pertinence of air–fluid levels in distinguishing adynamic ob-struction or ileus (nonobstructive distension)and mechanical obstruction lacked consistentsubstantiation in subsequent literature. Somestudies showed the relative nonspecificity andinsensitivity of these findings in establishingthe diagnosis of small-bowel obstruction, al-though the significance of air–fluid levels as anindicator of small-bowel luminal stasis iswidely accepted [18–21]. The presence of dif-ferential air–fluid levels on abdominal radio-graphs has also been regarded as an insensitivediagnostic finding in differentiating the me-chanical versus functional nature of an intesti-nal obstruction because of their potentialoccurrence in both of these conditions [15].Other researchers have even recommendedthat the upright radiograph be eliminated from

174


Lappas et al.

the acute abdominal radiographic series be-cause of its inability to substantially alter thesensitivity of the examination in the diagnosisof small-bowel obstruction [22, 23]. However,in those studies the patient populations werenonuniform and included patients with a widevariety of abdominal complaints and condi-tions such as abdominal pain, adynamic ileus,acute surgical abdomen, small-bowel obstruc-tion, and colon obstruction. Few patients witha diagnosis of small-bowel obstruction wereinvolved in the one study [22], whereas in theother the upright and supine radiographs werenot evaluated independently of each other[23]. Citing such limitations and further recog-nizing the need to render as rapid and accuratea diagnosis as possible, other researchers havemaintained support for the value of upright ab-dominal radiographs [24].

Unlike these previous reports, our analysisfocuses on an issue other than diagnostic sen-sitivity: the relevance of radiographic findingsin distinguishing the severity of small-bowelobstruction, because we think this distinctionhas important ramifications for the practicalimaging examination of symptomatic patientsin whom obstruction is clinically suspected bythe findings of physical examination and perti-nent history. The two radiographic findingsconsidered of greatest significance as predic-tors of the severity of obstruction were find-ings seen only on the upright abdominalradiograph. When both of these radiographicfindings are considered, their combined pres-ence or absence in any given case is a strongpositive (86%) or negative (83%) indicator ofthe degree of patency of the small-bowel lu-men. Although upright radiographs alone maynot be particularly sensitive in the detection ofsmall-bowel obstruction, they may be of valuein distinguishing patients with either a high-grade or a complete obstruction from thosewith a low-grade partial obstruction or thosewho are symptomatic but have no obstruction.

The information obtained from the uprightradiograph makes it an integral part of abdom-inal radiographic examinations in patients with

suspected small-bowel obstruction by helpingto direct, if needed, the additional imagingmethod that is appropriate for further workup.The results of this study validate prior recom-mendations to use abdominal radiographs as apractical tool in the radiologic management ofsmall-bowel obstruction [2]. This issue is im-portant because the erroneous application ofdiagnostic procedures not only adds to healthcare costs but also may delay diagnosis andadversely affect patient treatment decisions.

Acknowledgment

We thank Mark P. Hanna of the Division ofBiostatistics, Department of Medicine, IndianaUniversity School of Medicine for the statisti-cal support and analysis involved in this study.

References

1. Maglinte DDT, Reyes BL, Harmon BH, et al. Re-liability and role of plain film radiography andCT in the diagnosis of small-bowel obstruction.

AJR

1996

;167:1451–14552. Maglinte DDT, Balthazar EJ, Kelvin FM, Megibow

AJ. The role of radiology in the diagnosis of small-bowel obstruction.

AJR

1997

;168:1171–11803. Megibow AJ, Balthazar EJ, Cho KC, et al. Bowel

obstruction: evaluation with CT.

Radiology

1991

;180:313–318 4. Fukuya T, Hawes DR, Lu CC, Chang PJ, Barloon

TJ. CT diagnosis of small bowel obstruction: effi-cacy in 60 patients.

AJR

1992

;158:765–7695. Gazelle GS, Goldberg MA, Wittenberg J, Halpern

EF, Pinkney L, Mueller PR. Efficacy of CT in distin-guishing small-bowel obstruction from other causesof small-bowel dilatation.

AJR

1994

;162:43–476. Frager D, Medwid SW, Baer JW, Mollinelli B,

Friedman M. CT of small-bowel obstruction:value in establishing the diagnosis and determin-ing the degree and cause.

AJR

1994

;162:37–417. Taourel PG, Fabre J-M, Pradel JA, Senterre EJ, Meg-

ibow AJ, Bruel J-M. Value of CT in the diagnosisand management of patients with suspected acutesmall-bowel obstruction.

AJR

1995

;165:1187–11928. Shrake PD, Rex DK, Lappas JC, Maglinte DDT.

Radiographic evaluation of suspected small bowelobstruction.

Am J Gastroenterol

1991

;86:175–1789. Maglinte DDT, Gage S, Harmon B, et al. Ob-

struction of the small intestine: accuracy and roleof CT in diagnosis.

Radiology

1993

;186:61–64

10. Walsh DW, Bender GN, Timmons JH. Compari-son of computed tomography-enteroclysis andtraditional computed tomography in the setting ofsuspected partial small bowel obstruction.

EmergRadiol

1998

;5:29–3711. Markus JB, Somers S, Slobodan EF, Moola C,

Stevenson GW. Interobserver variation in the in-terpretation of abdominal radiographs.

Radiol-ogy

1989

;171:69–7112. Clark LA, Pregibon D. Tree-based models. In:

Chambers JM, Hastie TJ, eds.

Statistical models inS

. Pacific Grove, CA: Wadsworth & Brooks/ColeAdvanced Books and Software,

1992

:377–39013. Hosmer DW, Lemeshow S. Assessing the fit of the

model. In: Hosmer DW, Lemeshow S, eds.

Appliedlogistic regression.

New York: Wiley,

1989

:135–173 14. Hosmer DW, Lemeshow S. Special topics: polyto-

mous logistic regression. In: Hosmer DW, Lemes-how S, eds.

Applied logistic regression

. New York:Wiley,

1989

:216–24515. Harlow CL, Stears RLG, Zeligman BE, Archer

PG. Diagnosis of bowel obstruction on plain ab-dominal radiographs: significance of air-fluid lev-els at different heights in the same loop of bowel.

AJR

1993

;161:291–29516. Bryk D, Wolf BS. A radiological evaluation of

small bowel activity in the acute abdomen.

CritRev Diagn Imaging

1977;10:99–12817. Frimann-Dahl J. Special pathological findings:

stenosis and obstruction. In: Frimann-Dahl J, ed.Roentgen examinations in acute abdominal diseases,3rd ed. Springfield, IL: Thomas, 1974:88–136

18. Gammill SL, Nice CM Jr. Air-fluid levels: theiroccurrence in normal patients and their role in theanalysis of ileus. Surgery 1972;71:771–780

19. Levin B. Mechanical small bowel obstruction.Semin Roentgenol 1973;8:281–297

20. Schwartz SS. The differential diagnosis of intestinalobstruction. Semin Roentgenol 1973;8:323–338

21. Baker SR, Cho KC. Plain film radiology of the intes-tines and appendix. In: Baker SR, Cho KC, eds. Theabdominal plain film with correlative imaging, 2nded. Stamford, CT: Appleton & Lange, 1999:217–367

22. Mirvis SE, Young JWR, Keramati B, McCrea ES,Tarr R. Plain film evaluation of patients with ab-dominal pain: are three radiographs necessary?AJR 1986;147:501– 503

23. Simpson A, Sandeman D, Nixon SJ, GoulbourneIA, Grieve DC, Macintyre IM. The value of theerect abdominal radiograph in the diagnosis of in-testinal obstruction. Clin Radiol 1985;36:41–42

24. Mindelzun RE, McCort JJ. Definition of acuteabdominal series. (answer to question) AJR 1996;166:716–718

Documents

167.full