Dig Dis Sci (2013) 58:1074–1083 DOI 10.1007/s10620-012-2473-0

ORIGINAL ARTICLE

Utility of Repeated Abdominal CT Scans After Prior Negative CT Scans in Patients Presenting to ER with Nontraumatic Abdominal Pain Borko Nojkov • Michael C. Duffy • Mitchell S. Cappell

Received: 13 June 2012 / Accepted: 16 October 2012 / Published online: 21 November 2012 Ó Springer Science+Business Media New York 2012

Abstract Purpose The purpose of this study was to analyze diagnostic yield of repeat computed tomography (CT) after negative initial CT versus yield of initial CT in patients presenting repeatedly to emergency room (ER) for nontraumatic abdominal pain. Medical costs and radiation exposure from repeat CT could be reduced if repeat CT after negative initial CT has a low diagnostic yield. Methods Patients included consecutive adults presenting to William Beaumont Hospital, from 2007 to 2010, undergoing abdominal CT for nontraumatic abdominal pain retrospectively identified by medical diagnostic and CT procedural codes. Exclusion criteria were prior abdominal trauma, recent abdominal surgery, and known chronic gastrointestinal disease. The CT was labeled ‘‘positive’’ if findings explained patient’s abdominal pain or was clinically significant. Positivity rate was compared for repeat versus initial CT. Results Among 200 consecutive patients undergoing (659) multiple CTs (mean age = 45.7 years, 74 % female), positivity rate for initial CT (22.5 %) was significantly higher than positivity rates for CT#2 (8.4 %, p = 0.002), for CT#3 (4.9 %, p = 0.005), and for CT C #4 (5.9 %, p = 0.006). Generally, CT positivity rate declined with increasing number of prior negative CTs. CT positivity rate was

Presented orally at a plenary session at the 2011 Annual Meeting of the American College of Gastroenterology held in Washington DC, October 29–November 2. B. Nojkov
M. C. Duffy
M. S. Cappell (&) Division of Gastroenterology and Hepatology, Department of Medicine, William Beaumont Hospital, Oakland University William Beaumont School of Medicine, 3535 West 13 Mile Rd, Royal Oak, MI 48073, USA e-mail: [email protected]

123

significantly higher in 100 patients undergoing single CT versus 155 patients undergoing repeat CTs (46.5 vs. 6.5 %, p = 0.0001). Positive repeat CT findings included intestinal mural thickening/mass (7), colitis (5), appendicitis (4), and other (14). Among 15 analyzed clinical parameters, two significantly predicted repeat CT positivity, namely, leukocytosis (p = 0.03) and APACHE-II-score [5 (p = 0.01). Repeat CTs constituted 47 % of all CTs. Conclusions Repeat abdominal CT after initially negative CT(s) performed for nontraumatic abdominal pain has a low diagnostic yield. Leukocytosis and APACHE-II score might help predict CT scan positivity. Data suggest restricted abdominal CT utilization in ER patients with multiple prior negative CTs. Findings warrant confirmation in prospective studies. Keywords Non-traumatic abdominal pain
Abdominal pain
Emergency room
Abdominal CT
Abdominal ultrasound
Abdominal imaging
Cost-containment
Radiation exposure
Triage

Introduction Clinical application of abdominopelvic computed tomography (hereafter denoted as abdominal CT) has increased several-fold during the last 20 years to evaluate patients presenting to emergency room (ER) with nontraumatic abdominal pain [1, 2], the most common ER complaint that accounts for nearly 10 % of all ER visits [3, 4]. Abdominal CT is expensive and exposes patients to risks from radiation. Abdominal CT in ER may be overutilized [5]; more than half of such scans do not reveal the underlying etiology for the presentation of abdominal pain [6].

Dig Dis Sci (2013) 58:1074–1083

Repeated CT scans for patients presenting with abdominal pain may account for a large proportion of abdominal CTs performed in ER. There is scant data focusing on diagnostic yield of repeated abdominal CT scans when performed after initially negative scans. This study compares diagnostic yield of repeated versus initial abdominal CT scans in patients undergoing multiple CT scans in ER for nontraumatic abdominal pain, compares diagnostic yield of repeated versus single abdominal CT scans in patients presenting to ER with nontraumatic abdominal pain, and identifies predictors of positive findings on repeat abdominal CTs. Study findings can potentially reduce medical costs and radiation exposure of patients by avoiding unnecessary repeat abdominal CTs.

Methods Study Design Patients presenting to ER at William Beaumont Hospital, Royal Oak, from January 1, 2007 to June 30, 2010, were identified by computerized analysis of medical diagnostic codes for nontraumatic abdominal pain and procedural codes for abdominal CT [7]. William Beaumont Hospital, a large, urban, tertiary care, university hospital, is designated a level-one, trauma center and has approximately 116,000 ER visits per annum and 32,000 abdominal CTs performed per annum in ER. Study group-1, analyzed solely to determine proportion of all CTs performed in ER that were repeat CTs, comprised 100 consecutive patients presenting either on one or multiple occasions to ER between September 1, 2009 and November 30, 2009, with chief complaint of nontraumatic abdominal pain. Study group-2 comprised 200 consecutive patients presenting to ER on C2 occasions with chief complaint of nontraumatic abdominal pain and with performance of abdominal CT scan on each ER visit. Study group-3 comprised 100 consecutive patients presenting to ER on only one occasion with chief complaint of nontraumatic abdominal pain and with performance of abdominal CT scan on that visit. Study exclusion criteria included: age \18 years, abdominal trauma before ER presentation, recent abdominal surgery (\6 months), ER presentation related to complications from prior surgery (e.g. chronic abdominal pain from adhesions), and known chronic organic gastrointestinal disease (e.g. chronic pancreatitis, inflammatory bowel disease). Patients were also excluded if they had a CT at the study hospital before January 1, 2007, had multiple CT scans performed on the day of the ER visit, or had some medical care at other hospitals within the prior 2 years. Medical charts were retrospectively, comprehensively reviewed, including CT reports, as interpreted by ER

1075

radiologist; physician notes, nursing notes, and specialty consultations; and all other ER data, including vital signs, laboratory tests, and all imaging studies. Outpatient clinic and telephone notes were reviewed to exclude patients also evaluated at other hospitals. As a measure of severity of illness at presentation, acute physiology and chronic health evaluation (APACHE) II scores were determined; higher scores indicate more severe illness [8]. This study was approved by the Institutional Review Board of William Beaumont Hospital (#2009-180). CT Protocol CTs were uniformly performed with 64-multislice CT scanners (Siemens Medical Solutions USA, Malvern, Pennsylvania). The general CT protocol for nontraumatic abdominal pain, excluding renal colic, involved administration of 50 cc of oral contrast (iopamidol/Isovue-300) dissolved and drunk 1 h before CT, and 80 cc of intravenous contrast (iopamidol/Isovue-370) injected 90 s before CT scanning (Bracco Diagnostics Inc., Princeton, New Jersey). CT scans performed per renal protocol (N = 180 CTs) were performed without intravenous contrast. Intravenous contrast was also not administered in another 49 CT scans because of known or suspected contrast allergy, renal insufficiency, or other contraindication to its administration. Images were obtained at 3-mm intervals extending from middle of the heart to ischial tuberosities, with routine performance of coronal reconstructions. An attending radiologist is on duty in ER 24 h per day, throughout the year. All reported CT studies were read by the attending radiologist on duty at the time of the patient’s ER visit. Radiology residents and fellows are also included in reading CTs in ER, as part of their training programs, but they are directly supervised by attendings on duty. Radiologists interpreted CT images on imaging workstations, with computerized access to patient’s medical history and prior imaging studies. CT Scan Classification and Outcome Radiologists diagnosed positive CT scan findings by standard radiologic criteria (e.g. colitis by colonic wall thickening and pericolonic inflammatory changes). Radiologic reports describing ‘‘diagnostic,’’ ‘‘likely’’ or ‘‘probable’’ findings, or labeled as ‘‘consistent with’’ were recorded as findings of the described clinical condition. Scans were classified as ‘‘positive’’ if CT finding explained patient’s presenting complaint of abdominal pain, or if CT demonstrated a new, highly significant, clinical finding necessitating medical investigation (e.g. neoplasm, colitis, abdominal findings from previously unknown systemic disease, etc.). Scans were classified as ‘‘negative’’ if no abnormalities were detected or if minor

123

1076

abnormalities were detected that were not the cause of abdominal pain (e.g. colonic diverticulosis without diverticulitis). Among the 200 patients with recurrent ER visits, CTs were stratified into four groups: CT-1 (i.e. first CT in all 200 patients), CT-2 (i.e. second CT in 155 patients with negative first CT), CT-3, and CT C 4. After identifying a positive CT, analysis of subsequent CTs was censured (i.e. not further analyzed), except that patients with positive CT findings were followed after ER encounter (mean followup = 27 days) to determine final patient outcome and to analyze CT diagnostic accuracy by comparing final (abdominal) diagnosis based on subsequent diagnostic evaluation, including further imaging, with original CT diagnosis made in ER. CTs were labeled ‘‘true positive’’ or ‘‘false positive’’ if subsequent tests did or did not support the reported CT diagnosis, respectively. Rate of CT scan positivity for CT-2, CT-3, and CT C 4 in the 155 patients undergoing multiple CTs, after an initial negative CT, was compared to: (1) rate of positivity of initial CT in patients undergoing multiple CTs, and (2) rate of CT positivity in patients undergoing only one CT. To identify predictors of CT positivity, demographic parameters, vital signs and laboratory values were compared between patients with positive versus negative second CT scans. Twenty CT examinations were re-reviewed specifically for this study by an attending radiologist with fellowship training in body imaging (A.A.) and without knowledge of prior CT reading, to assess the accuracy of the original CT reports. Also, negative CT scans preceding the positive second or positive third CT scans were re-reviewed (by an attending radiologist with fellowship training in body imaging and without knowledge of prior CT reading) to determine whether the positive findings on the repeat CT scans were already present on the earlier performed CT scans but were improperly read and missed. To determine whether positive CT findings after negative prior CTs was due to changes in CT technique, CT protocols were compared between patients having positive second or positive third CT scans and their preceding negative scans.

Statistical Analysis Categorical variables were analyzed by v2 test or by Fisher’s exact test, depending on cell size. All p values were two-tailed. Confidence intervals (CI) for odds ratio (OR) were calculated by method described by Fleiss [9]. The general linear mixed model (GLMM) was used to assess correlation between CT scan positivity versus ordinal CT scan number, and to assess potential role of age and gender as independent covariates on number of repeated CT scans and on CT positivity rate. Computations were

123

Dig Dis Sci (2013) 58:1074–1083

performed with SAS/STAT software (version 9.2, SAS Institute).

Results Proportion of Patients with Repeat CT Scans A review of 100 consecutive patients in the middle of the study period (September–November, 2009) undergoing CTs and satisfying study entry criteria revealed that 47 % were repeat CTs and 53 % were initial CTs. Mean age of these patients was 56.4 ± 17.2 years (median 56, range 31–88 years) and 61 % were female. Patients with Multiple Abdominal CTs and ER Visits: Repeat Versus Initial CT Patients and CT Characteristics Of 282 patients undergoing multiple abdominal CTs in ER, 82 patients (29 %) were excluded based on exclusion criteria and 200 were analyzed. CT and patient characteristics are reported in Table 1. Mean age of study patients was 45.7 ± 19.1 years (median = 42; range 19–95 years) and 148 (74 %) were females. Patients underwent altogether 659 abdominal CT scans (mean = 3.3 ± 2.6 CTs/patient) during the study period, including 459 (75.6 %) repeat CTs, performed after C1 negative CT. CT scans were performed with administration of oral and IV contrast in 430 (65.3 %) and administration of no IV contrast in 229 (34.7 %), as per renal protocol to exclude obstructive Table 1 Two hundred patients presenting repeatedly to ER and undergoing 659 abdominal CTs for acute nontraumatic abdominal pain: patient demography and CT characteristics Parameters

Mean ± SD or N with parameter/total N

Mean age in years (range)

45.7 ± 19 (19–95)

Number of females/total (%)

148/200 (74 %)

Mean duration in days between CT scans (range)

154.8 ± 144.1 (1–728)

Mean ± SD CT scans per patient (range)

3.3 ± 2.6 (1–18)

Number (%) repeat CT scans/total

459/659 (76.5 %)

Number (%) positive (repeat or initial) CT scans/total Number (%) positive repeat CT scans/ total repeated

75/659 (11.3 %)

Number (%) of CT scans with IV contrast

30/459 (6.5 %) 430/659 (65.3 %)a

N number, CT computed tomography, SD standard deviation, IV intravenous a Forty-six additional patients received oral contrast without IV contrast

Dig Dis Sci (2013) 58:1074–1083

1077

nephrolithiasis (N = 180) or per contraindications to contrast administration (N = 49). Among the 229 patients not receiving IV contrast for the CT scan, 47 (7.1 % of all 659 CTs) received oral contrast. Mean interval between individual CT scans was 154.8 ± 144.1 days (range 1–738 days). The 659 CT examinations were reviewed by 33 attending radiologists (mean CT examinations per radiologist = 19.9 ± 21.4), but the bulk (75 %) of CT scans were read by only eight radiologists, who generally had the greatest experience in body imaging. Radiology residents or fellows participated in reading of 41 % of the CT scans under attending supervision. Re-review of 20 CTs for this study by one dedicated attending radiologist showed one false positive CT; initially it had been read as appendicitis, but re-review revealed normal findings, without evident appendicitis. This patient’s clinical course was also inconsistent with appendicitis (no fever, no leukocytosis, and spontaneous, rapid symptomatic improvement without surgery). CT re-review also revealed one patient who had a minor new finding of non-specific, mild inguinal and external iliac lymphadenopathy. Initial Versus Repeated CT Table 2 lists positive findings in initial versus repeat CTs. Common diagnoses included gastrointestinal mural thickening or mass, obstructive nephrolithiasis, colitis, and diverticulitis. Less common diagnoses included appendicitis, ruptured ovarian cyst, small bowel obstruction, and Table 2 Significant findings on initial and repeat CT scans among 200 patients presenting to ER with nontraumatic abdominal pain Significant CT finding

Number with finding on initial CT scan

Number with finding on repeat CT scan

Obstructive nephrolithiasis

12

2

Gastrointestinal mural thickening or mass Diverticulitis

10

7

8

0

Colitis

4

5

Ruptured ovarian cyst

3

2

Appendicitis

1

4

Small bowel obstruction

1

3

Pancreatitis

0

2

Other

6a

5b

Total positive/all tested

45/200 (22.5 %)

30/459 (6.5 %)

CT computerized tomography, ER emergency room a

Includes two with pelvic/renal mass, and one each with panniculitis, polycystic kidneys with cyst hemorrhage, pulmonary infiltrate/nodule, and newly diagnosed severe ascites

b

Includes one each with cholecystitis, localized fluid collection (abscess vs. hematoma), omental infarction, compression spine fracture (discitis revealed on MRI), and newly diagnosed severe ascites

other. Obstructive nephrolithiasis and diverticulitis occurred more commonly in initial CT scans, whereas appendicitis occurred more commonly in repeat scans (Table 2). Only 30 of 459 (6.5 %) repeat CTs were positive (Table 1). Positivity rates of initial versus repeat CT scans are illustrated in Fig. 1. Positivity rate of initial CT scan was significantly higher than: (1) positivity rate for CT-2 (22.5 vs. 8.4 %, p = 0.002), (2) positivity rate for CT-3 (22.5 vs. 4.9 %, p = 0.005), and (3) positivity rate for CT C 4 (22.5 vs. 5.9 %, p = 0.006). Frequency of positive findings in CT# 4–5 was 7/108 (6.4 %) and in CT# 6–18 was 5/93 (5.3 %). CT scan positivity rate generally declined with increasing number of prior negative scans. No CTs, including either initial or repeated CTs, produced any significant patient toxicity, excluding potential longterm toxicity from radiation exposure from CT examinations which was not directly studied. Positive findings in the second CT scan (and resultant changes in clinical management) included: colitis in three patients (two patients prescribed outpatient antibiotic therapy and one patient hospitalized for IV antibiotic therapy); small bowel obstruction in two patients (both admitted for observation including surgical consultation, and managed conservatively without undergoing surgery); colonic mural thickening in two (both hospitalized and underwent colonoscopy); appendicitis in two (one underwent appendectomy and the other was hospitalized and treated conservatively with surgical consultation and antibiotic therapy); hydronephrosis due to obstructive nephrolithiasis in one (admitted to the observation unit and had urology consultation, treated with pain control and IV hydration and discharged after stone passed); pericholecystic fluid collection in one (hospitalized and underwent ERCP to exclude bile duct leak); compression lumbar spine fracture in one (hospitalized with subsequent evaluation by MRI); and pelvic mass in one (admitted to observation unit and had obstetrics/gynecology consultation, underwent pelvic ultrasound which showed no mass [CT scan a false positive finding] and discharged). Positive findings in the third CT scan (and resultant changes in clinical management) included: colitis in two patients (one hospitalized for antibiotic therapy, and the other prescribed antibiotic therapy as an outpatient); colonic mural thickening in one (gastroenterology consultation and underwent ambulatory colonoscopy); hydronephrosis from obstructive nephrolithiasis in one (admitted to observation unit and treated with pain control, IV hydration and antinausea medication, and discharged with outpatient follow-up by urologist); and ovarian cyst in one (underwent pelvic ultrasound which confirmed presence of cyst, had consult by obstetrician/gynecologist who felt cyst was cause of pain, and scheduled for obstetrics/gynecology outpatient follow-up). The mean time interval between initial and second CT scans in the 13 patients who had positive second CT was

123

1078

Dig Dis Sci (2013) 58:1074–1083

Fig. 1 Comparison of positive CT scan outcome between initial versus repeated CT scans, including CT-2, CT-3, and CT C 4 for all 659 CT scans. All differences in rate of positivity between initial CT scan and repeat CT scans are quantitatively large and statistically significant. The differences were also large and statistically significant

if CTs C 4 were stratified into CTs #4–5 (frequency of positives: 7/108, 6.5 %) and CTs C 6 (frequency of positives: 5/93, 5.3 %). Similarly, the rates of positive CT scan outcome in initial and repeated CT scans for patients undergoing CT per renal protocol were 21.6 and 5.8 %

179.3 ± 143.2 days (range 25–481 days, two patients with time interval between CT scans \30 days). The mean time interval between second and third CT scan in the five patients who had positive third CT scan was 254.2 ? 301.8 days (range 21–650 days, one patient with time interval between CT scans \30 days). Re-review of the 18 negative CT scans preceding the positive second and positive third CT scans revealed that none of these earlier CT scans was improperly read and missed a subsequently reported positive finding. Of the 18 positive second or third CT scans, 15 were performed using the same protocol as their preceding negative CT scans; two were performed using no IV and/or oral contrast, while the prior negative CT used both IV and oral contrast; and one was performed using IV and oral contrast, while the preceding scan used no contrast.

In subgroup analysis, patients undergoing CT per renal colic protocol also exhibited markedly higher positivity rate of initial versus repeat CTs (positivity rate of initial CT: 21.6 %, 13/60 versus positivity rate of repeat CT: 5.8 %, 7/120). Obstructive nephrolithiasis was diagnosed in nine initial CTs and in three repeat CTs. Interestingly, four initial CTs, performed per renal protocol, demonstrated alternative diagnoses to nephrolithiasis, including one each with ruptured ovarian cyst, renal cyst with hemorrhage, diverticulitis, and incidental finding of two new 1-cm lung nodules requiring follow-up. Similarly, four repeat CTs, performed per renal protocol, also demonstrated alternative diagnoses, including one each with ruptured ovarian cyst, pancreatitis, ileitis, and gastric wall thickening.

Table 3 False positive rates in initial and repeated CT scans and diagnostic tests performed after a false positive CT scan Ordinal CT scan no.

False positive/total positive CT scans (%)

Subsequent tests performed on patients with false positive CT scans

Initial CT scan

7/45 (15.5 %)

Colonoscopy 3, repeat abdominopelvic CT 3, pelvic CT 1, pelvic ultrasound 1, cortisol/ACTH level 1

CT # 2

4/13 (30.8 %)

EGD 3, colonoscopy 2, repeat abdominopelvic CT 1, CT enterography 1, abdominal ultrasound 1, pelvic ultrasound 1

CT # 3

2/5 (40 %)

Colonoscopy 1, colonoscopy recommended for another patient but patient refused

CT # 4

1/4 (25 %)

EUS 1, ANA/IgG level/CA 19-9

CT # 5–8

0/3

N/A

CT # 9–13

1/5 (20 %)

EGD 1

All repeated (CT C 2) CTs combined (N = 30)

8/30 (26.6 %)

EGD 4, colonoscopy 3, repeat abdominal CT 1, CT enterography 1, EUS 1, abdominal ultrasound 1, pelvic ultrasound 1, ANA/IgG/CA 19–9–1

CT computed tomography, ACTH adrenocorticotropic hormone, EGD esophago-gastroduodenoscopy, EUS endoscopic ultrasound, ANA antinuclear antibody, IgG immunoglobulin G, CA 19-9 carbohydrate antigen 19-9, N/A not applicable

123

Dig Dis Sci (2013) 58:1074–1083

1079

Table 4 Comparison of demographic parameters, vital signs and laboratory values in patients with positive versus negative second CT scan Clinical or laboratory parameter

Positive second CT scan (N = 13): N with parameter/total evaluated N (%)

Negative second CT scan (N = 142): N with parameter/total evaluated N (%)

p value

Odds ratio (95 % confidence interval)

Age [65

4/13 (30.7 %)

23/142 (16.2 %)

0.24

2.30 (0.54–9.19)

Gender (males)

3/13 (23 %)

29/142 (20.4 %)

0.73

1.17 (0.24–5.05)

Tobacco smoking

2/13 (15 %)

35/142 (25 %)

0.73

0.56 (0.08–2.86)

Alcoholism

0/13

5/142 (3.5 %)

NA

NA

Illicit drug usea

1/13 (7.7 %)

17/142 (12 %)

1

0.61 (0.03–5.08)

Initial blood pressure (MAP [ 110 mmHg)

6/13 (46 %)

37/142 (26 %)

0.19

2.43 (0.67–8.76)

Pulse (\60 or [100 bpm)

4/13 (30.7 %)

27/142 (19 %)

0.29

1.89 (0.45–7.46)

Fever ([38 °C)

0/13

1/142 (0.7 %)

NA

NA

Leukocyte count [10,900 cells/cumm

4/13 (30.7 %)

12/142 (8.5 %)

0.03

4.82 (1.06–21.0)

Urinalysis (any abnormality)

3/13(23 %)

12/124 (9.7 %)

0.15

2.80 (0.53–13.45)

Demographics

Vital signs

Laboratory values

APACHE II score [15

0/13

2/141 (1.4 %)

NA

NA

10–15

0/13

8/141 (5.7 %)

NA

NA

5–9

6/13 (46 %)

22/141 (15.6 %)

0.01

4.63 (1.42–15.11)

\5

7/13 (54 %)

109/141 (77.3 %)

0.09

0.34 (0.09–1.25)

[3 comorbidities

5/13 (38.4 %)

41/142 (29 %)

0.53

1.54 (0.41–5.61)

CT computed tomography, N number, MAP mean arterial blood pressure, APACHE acute physiology and chronic health evaluation, N/A not applicable For several parameters, total evaluated N for patients with negative second CT scan reduced to\142 patients because some patients did not have this specified parameter determined in the emergency room (e.g. only 124 patients had urinalysis determined in emergency room) Other laboratory parameters compared: hematocrit (\30), creatinine ([1.4), and any abnormal liver function test were not significantly different between the two groups a

Self reported or well-documented history of illicit drugs (e.g. marijuana, cocaine, heroin)

False positive rates, stratified by CT number, are shown in Table 3. False positive rate in first CT was 7/45 (15.5 %), and in repeat CTs was 8/30 (26.6 %). False positives on initial CT included: colonic thickening suggestive of colitis in four (disproven by colonoscopy in three patients, performed at 1, 3 and 4 days after CT, and negative repeat CT performed 2 days after initial scan in one patient), adrenal gland mass in one (disproven by normal serum adrenal function studies and repeat CT performed 39 days after initial CT), appendicitis in one (disproven by clinical course and negative repeat CT 1 day later), significant free peritoneal fluid (disproven by normal pelvic ultrasound performed during the same ER visit and repeat CT performed 60 days later). False positives on repeat CTs included: thickened intestinal wall showing colitis/terminal ileitis in four (disproven by colonoscopy performed at 1, 2 and 6 days after CT, and CT enterography performed 2 days after CT); thickened gastric wall in one (disproven by EGD performed 6 months after CT); appendicitis in one (disproven by clinical course of rapid pain resolution and

by negative repeat CT performed 1 day later); pelvic mass in one (disproven by follow-up pelvic ultrasound 3 days after CT); and acute pancreatitis (disproven by normal serum lipase and amylase levels in ER and by subsequent clinical course). False positive CT diagnoses resulted in a mean of 2.4 ± 1.1 subsequent diagnostic tests per patient, including gastrointestinal endoscopies, further imaging studies, or specialized laboratory tests (Table 3). Predictors of Positive Repeat CT Compared to patients with negative CT-2, patients with positive CT-2 had significantly more frequent leukocytosis (OR 4.82; CI 1.06–21.0, p = 0.03), and significantly more frequent APACHE II score[5 (OR = 4.63; CI 1.42–15.11, p = 0.01) (Table 4). If both these clinical predictors were absent (i.e. normal leukocyte count & APACHE II score \5), only four of 99 (4.0 %) patients had a positive CT-2 after an initially negative scan. Patients with positive CT-2 scan also had statistically nonsignificant trends of being

123

1080

Dig Dis Sci (2013) 58:1074–1083

Table 5 Comparison of several demographic parameters and CT positivity rates (a) Patients undergoing single CT scan versus first CT in patients undergoing multiple CTs Continuous parameter

Patients undergoing single CT (N = 100)

Patients undergoing first of multiple CTs (N = 200)

p value

Mean age in years (range)

55.1 ± 16.9 (25–99)

45.7 ± 19 (19–95)

0.0001

Categorical parameter

N positive/total (%)

N positive/total (%)

p value

Odds ratio (95 % confidence interval)

Female sex Positive CT

69/100 (69 %) 46/100 (46 %)a

148/200 (74 %) 45/200 (22.5 %)

0.41 0.00001

0.78 (0.45–1.37) 2.93 (1.70–5.08)

False positive CT

3/35 (8.5 %)b

7/45 (15.5 %)

0.5

0.51 (0.10–2.46)

(b) Patients undergoing single CT scan versus second of multiple CTs Continuous parameter

Patients undergoing single CT (N = 100)

Patients undergoing second of multiple CTs (N = 155)

p value

Mean age in years (range)

55.1 ± 16.9 (25–99)

43.9 ± 19.1 (19–92)

.00001

Categorical parameter

N positive/total (%)

N positive/total (%)

p value

Odds ratio (95 % confidence interval)

Female sex

69/100 (69 %)

121/155 (78.1 %)

0.11

0.62 (0.34–1.15)

13/155 (8.4 %)

0.00001

9.31 (4.45–19.76)

4/13 (30.8 %)

0.08

0.21 (0.03–1.42)

Positive CT

46/100 (46 %)

False positive CT

3/35 (8.5 %)b

a

a

CT diagnoses: obstructive nephrolithiasis 14, gastrointestinal mural thickening or mass 9, diverticulitis 5, colitis 5, ruptured or hemorrhagic adnexal/ovarian cyst 4, incarcerated inguinal hernia 2, and other 7 (one each with small bowel obstruction, ruptured cholecystitis, complex renal cyst, muscle abscess, lung nodules, ascites, Meckel’s diverticulum)

b

Eleven patients with no follow-up studies available

older, having higher mean arterial pressure, and having more frequently abnormal urinalysis. Average mean arterial blood pressure was borderline statistically significantly higher for patients with positive versus negative second CTs (112.5 ± 20.6 mmHg vs. 100.3 ± 14.9 mmHg, p = 0.05). Patients with positive versus negative second CT had similar mean heart rates (91.3 ± 15.2 vs. 84.8 ± 17.2 beats/minute, p = 0.19). Distribution of the other analyzed clinical parameters was not significantly different in patients having positive versus negative second CTs (Table 4). Independent effects of age and gender on positivity rates of sequential CTs were statistically assessed by regression analysis (GLMM). Female gender was significantly associated with higher positivity rates (p = 0.0001), and significantly associated with more repeated CTs. Age did not significantly affect either the number of repeat CT scans or their positivity rates. Patients with Repeated CTs and ER Visits Versus Patients with Single CT and ER Visit

Nephrolithiasis was the most common diagnosis in both groups. However, the rate of nephrolithiasis was significantly higher in patients undergoing only one CT versus patients undergoing the first of multiple CTs (14/100, 14 % vs. 12/200, 6 %, OR 2.55; 95 % CI: 1.13–5.74, p = 0.02). Patients undergoing one CT were significantly older than patients undergoing multiple CTs (p = 0.0001, Table 5a). The two groups had similar percentages of females (p = 0.41, Table 5a). CT positivity rate was significantly higher in 100 patients having single CT versus 155 patients undergoing the second of C2 CT scans [46/100 (46 %) vs. 13/155 (8.4 %), OR 9.31; CI 4.45–19.76, p = 0.00001, Table 5b], and for 100 patients having single CT versus 95 patients undergoing the third of C3 CT scans [46/100 (46 %) vs. 5/95 (5.2 %), OR 15.33; CI 5.74–41.0, p = 0.0001]. False positive rate was borderline significantly lower in patients undergoing single CT versus patients undergoing the second of C2 CTs [3/35 (8.5 %) vs. 4/13 (30.8 %), OR 0.21; CI 0.03–1.42, p = 0.08]. Table 5b also compares mean age and gender distribution between patients with single CT versus patients undergoing the second of C2 scans.

CT Diagnosis CT positivity rate was significantly higher in 100 patients undergoing only one CT (Table 5a) versus 200 patients undergoing first of multiple CTs [46/100 (46 %) vs. 45/200 (22.5 %), OR 2.93; 95 % CI: 1.70–5.08, p = 0.00001].

123

Discussion Abdominal CT is the imaging technique of choice for diagnosing the etiology of acute nontraumatic abdominal pain in

Dig Dis Sci (2013) 58:1074–1083

ER [10–12]. Modern multi-detector CT scanners highly accurately diagnose the etiology and help direct the patient management [13–16]. While abdominal CT is cost-effective when strongly indicated, as for suspected appendicitis [17], it may be overutilized in ER for nonspecific abdominal pain or other indications, possibly because of medico-legal concerns [5]. Overutilization may be particularly prevalent among patients repeatedly presenting with nontraumatic abdominal pain. This study is novel in addressing this important and common clinical problem. Repeat CT scans represented almost half (47 %) of all CTs performed for nontraumatic abdominal pain in ER in the present study. In this study only 6.5 % of repeat CTs were positive (study group 2), a significantly lower rate of positivity than that of initial CTs (6.5 vs. 22.5 %, Fig. 1). The 22.5 % rate of initial CT positivity is much lower than the previously reported rate of about 50 % for nontraumatic abdominal pain [6], possibly because the present analysis excluded patients with previously known organic gastrointestinal disease and included only patients with repeated ER presentations for abdominal pain; repeat ER presentation may be associated with underlying chronic functional gastrointestinal disorders or narcotic-seeking behavior. Patients who presented to ER on only one occasion, and otherwise selected by identical study entry criteria (study group 3), had a much higher CT positivity rate (46 %; Table 5a), similar to the 50 % previously reported. Therefore, the difference between CT positivity rates is even greater for repeat versus single CTs (6.5 vs. 46 %, OR 12.18; 95 % CI = 7.09–20.90, p = 0.0001) than for repeat versus initial (of repeated) scans. These findings support that patients representing for nontraumatic abdominal pain, after repeatedly negative CTs, represent a patient subpopulation with a low CT yield. Furthermore, repeat abdominal CTs resulted in a significant false positivity rate (26.6 %; Table 3), a clinically undesirable result that necessitated additional diagnostic studies, frequently including invasive procedures. These data argue for caution in repeating CT after prior negative CTs. Positive findings on repeat CT scans after prior negative CT scans apparently were generally not due to misdiagnosis/error by radiologists on the earlier negative CT scan and were generally not due to improved CT technique/protocol on the repeat scan. On re-review, none of the 18 negative CT scans that preceded a second or third positive CT scan were found to have been incorrectly read and missed the subsequently positive finding. Among the 18 second or third positive CT scans, only one was found to have been technically superior to the prior negative scan (later scan both IV and oral contrast, earlier scan no IV contrast). Since repeat CTs were so rarely positive (6.5 %), multiple clinical parameters were analyzed to identify predictors of CT positivity to identify patients warranting repeat CT. Leukocytosis was predictive of CT positivity, a finding

1081

consistent with previous studies. Modahl et al. [6], in a retrospective study of 604 ER patients, reported that patients with leukocytosis had significantly higher CT positivity than patients with normal leukocyte counts (58 vs. 40 %, p = 0.001). Among 522 young ER patients with nontraumatic abdominal pain, Scheinfeld et al. [18] reported that normal granulocyte count predicted a negative CT diagnosis in females, but not in males; this predictor was, however, insufficiently specific as 4 % of patients with normal granulocyte counts had acute inflammatory disease that would have been missed without abdominal CT. They concluded, therefore, that laboratory tests cannot reliably obviate the need for CT in young adults with nontraumatic abdominal pain. Both prior studies [6, 18] analyzed all CTs performed for nontraumatic abdominal pain, while this study focused on patients undergoing repeat CTs. In this study higher APACHE II score, abnormal urinalysis, and higher blood pressure also tended to indicate increased CT positivity, but were not individually sufficiently accurate to obviate the need for repeat CT. However, if all these clinical parameters are normal, an ER physician may elect not to perform repeat abdominal CT in patients with several prior negative abdominal CTs, based on current findings of such a low repeat CT positivity rate. In this study only 4 % of 99 patients with normal leukocyte count and APACHE II score \5 had a positive second CT. Patients repeatedly presenting to ER can often be evaluated by safer and cheaper imaging techniques (e.g. abdominal ultrasound), with CT reserved only for diagnostic dilemmas based on ultrasound findings. Ultrasound was shown to be clinically useful (highly sensitive) in the diagnosis of suspected hepato-biliary or gynecologic disorders causing nontraumatic abdominal pain, but was less accurate (less sensitive) than CT in the diagnosis of suspected appendicitis or colitis [19]. However, in a cohort study from Holland, this ‘‘ultrasound-first’’ algorithm had higher sensitivity in detecting urgent diagnoses than direct performance of abdominal CT; only 49 % of 1,021 patients with acute abdominal pain required abdominal CT after abdominal ultrasound [20]. This percentage may be even lower in patients who represent to ER after multiple negative CTs. This algorithm should be cost-effective because ultrasound is much less expensive than CT. In particular, ultrasound should be considered as an alternative initial test in patients who have a history of prior negative CT scans, who are relatively thin, and who have fasted. Routine use of ultrasonography in the ER may require an ultrasound technician to be available 24 h per day in the ER. Exposure to ionizing radiation is a major disadvantage of CT. The approximate effective radiation exposure is 10 millisieverts (mSv) per abdominal CT, compared to annual background radiation dose of 3 mSv [21]. A 25-year-old

123

1082

individual has an estimated risk of one CT-induced cancer per 900 CT examinations, and of one CT-induced fatal cancer per 1,800 CT examinations [22]. Furthermore, ER patients undergoing repetitive CTs accrue large cumulative radiation doses, which may nonlinearly promote radiation-induced carcinogenesis [23]. A recent study of patients with gastrointestinal disorders showed an approximately 50 % increase in mean annual radiation exposure during the last decade, mostly from increased abdominal CTs [24]. Patients with the highest mean radiation exposure tended to be relatively young (\35 years old), the group at highest risk of radiation-induced carcinogenesis, and tended to have functional bowel disorders, a group who frequently undergo nondiagnostic CTs. Avoidance of repeat CTs in patients with prior negative CT studies should substantially reduce radiation-induced injury and diminish health care costs. A novel approach to reduce radiation exposure from CT scans with minimal compromise of image quality involves iterative reconstruction with subtraction of noise from CT images [25]. MRI does not entail radiation exposure. This is particularly advantageous in pregnant patients because of concern about fetal teratogenesis induced by radiation and development of malignancies in children exposed to radiation in utero, and is also helpful in avoiding long-term risks of malignancy in young adults and children from radiation exposure. Several studies have suggested that MRI without oral contrast is a sensitive and specific alternative test to abdominal CT in the evaluation of acute abdominal pain [26, 27]. However, general application of MRI is limited by its significantly greater cost as compared to abdominal CT. Emergency room medical care deals with the reality of deciding relatively quickly on patient disposition in a high volume environment. Abdominal CT satisfies a need for protection against medicolegal liability in discharging patients with abdominal pain without hospitalization. A potential remedial process that balances medicolegal concerns about missed diagnoses versus excessive performance of abdominal CTs is to vet requests for repeat abdominal CTs in ER, in patients with prior negative CTs, by a senior emergency room physician and/or radiologist for approval. This study has limitations. First, it is retrospective. This may introduce selection bias and confounding factors. For example, this study focused on one factor—whether or not to perform repeated CT. However, this decision is frequently complex and depends upon multiple patient factors including patient history, physical exam, laboratory findings, results from prior CTs, and clinical progress in ER, as well as physician’s risk tolerance [28]. Second, patients were identified by procedure codes for CT and diagnostic codes for nontraumatic abdominal pain; eligible study patients might have been inadvertently excluded due to miscoding. Third, this study is from a single, tertiary care, teaching hospital that has both radiology and ER residents. The present findings

123

Dig Dis Sci (2013) 58:1074–1083

may be pertinent to other teaching hospitals, but perhaps not entirely generalizable to small community hospitals that lack radiology and ER residents. Fourth, sample size was only moderate. However, most study patients with repeat CTs had multiple ER visits, all of which were comprehensively analyzed. Notwithstanding these limitations, the primary study finding, a lower positivity rate in patients undergoing repeat CTs in comparison to patients undergoing initial or single CTs, is highly statistically significant (p \ 0.006). This finding is potentially clinically important. It warrants confirmation by a prospective study. In conclusion, repeat abdominal CT, when performed in patients with prior negative CTs performed in ER for nontraumatic abdominal pain, has a low diagnostic yield, that is significantly lower than the diagnostic yield of the initial CT, and significantly lower than diagnostic yield of abdominal CT reported in the literature. Leukocytosis, higher APACHE II score, and possibly other clinical parameters might help predict repeat CT scan positivity. Perhaps, physicians should establish a higher threshold for abdominal CT in patients without known gastrointestinal disease presenting to ER with repeatedly negative prior CT scans, and consider safer and cheaper alternative imaging studies (e.g. abdominal ultrasound). Individual clinical judgment is, however, essential. Prospective studies are needed to confirm the current findings, to further characterize patients repeatedly presenting to ER, and to further analyze clinical predictors of positive repeat CT. Acknowledgments The authors thank Dr. Ahmad Al-Jaber of the Radiology Department of William Beaumont Hospital for rereviewing abdominal CTs for this study. Conflict of interest

None.

References 1. Dunnick NR, Applegate KE, Arenson RL. The inappropriate use of imaging studies: a report of the 2004 Intersociety Conference. J Am Coll Radiol. 2005;2:401–406. 2. Levin DC, Rao VM, Parker L, et al. Recent trends in utilization rates of abdominal imaging: the relative roles of radiologist and nonradiologist physicians. J Am Coll Radiol. 2008;5:744–747. 3. Sturman MF. Medical imaging in acute abdominal pain. Compr Ther. 1991;17:15–21. 4. Nawar EW, Niska RW, Xu J. National hospital ambulatory medical care survey: 2005 emergency department summary. Advance data from vital and health statistics; no. 388. Hyattsville, MD; National Center for Health Statistics. 2007. Available at: http://www.cdc.gov/nchs/data/ad/ad386.pdf. 5. Studdert DM, Mello MM, Sage WM, et al. Defensive medicine among high-risk specialist physicians in a volatile malpractice environment. JAMA. 2005;293:2609–2617. 6. Modahl L, Digumarthy SR, Rhea JT, et al. Emergency department abdominal computed tomography for nontraumatic abdominal pain: optimizing utilization. J Am Coll Radiol. 2006;3:860–866.

Dig Dis Sci (2013) 58:1074–1083 7. International Classification of Diseases, 9th revision. Diagnostic and procedural codes. http://ftp.cdc.gov/pub/Health_Statistics/ NCHS/Publications/ICD-9. 8. Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13:818–829. 9. Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York: Wiley; 1981. 10. Broder J, Warshauer DM. Increasing utilization of computed tomography in the adult emergency department, 2000–2005. Emerg Radiol. 2006;13:25–30. 11. Nagurney JT, Brown DF, Chang Y, Sane S, Wang AC, Weiner JB. Use of diagnostic testing in the emergency department for patients presenting with non-traumatic abdominal pain. J Emerg Med. 2003;25:363–371. 12. Marincek B. Nontraumatic abdominal emergencies: acute abdominal pain: diagnostic strategies. Eur Radiol. 2002;12:2136–2150. 13. Rosen MP, Siewert B, Sands DZ, Bromberg R, Edlow J, Raptopoulos V. Value of abdominal CT in the emergency department for patients with abdominal pain. Eur Radiol. 2003;13:418–424. 14. Sala E, Watson CJ, Beadsmoore C, et al. A randomized, controlled trial of routine early abdominal computed tomography in patients presenting with non-specific acute abdominal pain. Clin Radiol. 2007;62:961–969. 15. Tsushima Y, Yamada S, Aoki J, et al. Effect of contrast-enhanced computed tomography on diagnosis and management of acute abdomen in adults. Clin Radiol. 2002;57:507–513. 16. Abujudeh HH, Kaewlai R, McMahon PM, et al. Abdominopelvic CT increases diagnostic certainty and guides management decisions: a prospective investigation of 584 patients in a large academic medical center. AJR Am J Roentgenol. 2011;196:238–243. 17. Rao PM, Rhea JT, Novelline RA, et al. Effect of computed tomography of the appendix on treatment of patients and use of hospital resources. N Engl J Med. 1998;338:141–146. 18. Scheinfeld MH, Mahadevia S, Stein EG, et al. Can lab data be used to reduce abdominal computed tomography (CT) usage in

1083

19.

20.

21. 22.

23.

24.

25.

26.

27. 28.

young adults presenting to the emergency department with nontraumatic abdominal pain? Emerg Radiol. 2010;17:353–360. van Randen A, Lameris W, van Es HW, et al. A comparison of the accuracy of the ultrasound and computed tomography in common diagnoses causing acute abdominal pain. Eur Radiol. 2011;21:1535–1545. Lameris W, van Randen A, van Es HW, et al. Imaging strategies for detection of urgent conditions in patients with acute abdominal pain: diagnostic accuracy of study. BMJ. 2009;338:b2431. Stoker J, van Randen A, Lame´ris W, et al. Imaging patients with acute abdominal pain. Radiology. 2009;253:31–46. Anonymous. The 2007 recommendations of the International Commission on Radiological Protection. ICRP publication 103. Ann ICRP. 2007;37:1–332. Griffey RT, Sodickson A. Cumulative radiation exposure and cancer risk estimates in emergency department patients undergoing repeat or multiple CT. AJR Am J Roentgenol. 2009; 192:887–892. Desmond AN, McWilliams S, Maher MM, et al. Radiation exposure from diagnostic imaging among patients with gastrointestinal disorders. Clin Gastroenterol Hepatol. 2012;10: 259–265. Craig O, O’Neill S, O’Neill F, et al. Diagnostic accuracy of computed tomography using lower doses of radiation for patients with Crohn’s disease. Clin Gastroenterol Hepatol. 2012;10: 886–892. Singh AK, Desai H, Novelline RA. Emergency MRI of acute pelvic pain: MR protocol with no oral contrast. Emerg Radiol. 2009;16:133–141. Heverhagen JT, Klose KJ. MR imaging for acute lower abdominal and pelvic pain. Radiographics. 2009;29:1781–1796. Pines JM, Hollander JE, Isserman JA, et al. The association between physician risk tolerance and imaging among patients with gastrointestinal disorders. Am J Emerg Med. 2009;27: 552–557.

123