CT indicates computed tomography; MR, magnetic resonance; and US, ultrasonography.
MR indicates magnetic resonance; US, ultrasonography.
Fonseca AL, Schuster KM, Kaplan LJ, Maung AA, Lui FY, Davis KA. The Use of Magnetic Resonance Imaging in the Diagnosis of Suspected Appendicitis in Pregnancy Shortened Length of Stay Without Increase in Hospital Charges. JAMA Surg. 2014;149(7):687-693. doi:10.1001/jamasurg.2013.4658
Making an accurate diagnosis of appendicitis in pregnancy is critical for maternal and fetal outcomes.
To determine whether magnetic resonance (MR) imaging in pregnant patients with suspected appendicitis improves outcomes, minimizes length of stay (LOS), and lowers hospital charges.
Design, Setting, and Participants
Retrospective review at a university tertiary referral center of all pregnant patients seen with abdominal pain and suspected appendicitis who were followed up through delivery during an 11-year period.
Main Outcomes and Measures
Time to operation, LOS, complications, nontherapeutic exploration, fetal outcomes, and hospital charges.
Seventy-nine patients were included in this study, 34 of whom had pathology-confirmed appendicitis. Thirty-one patients underwent MR imaging. A trend toward fewer operations (odds ratio [OR], 0.45; 95% CI, 0.18-1.16; P = .07) was observed in the MR imaging group. Seven nontherapeutic explorations were performed in the non–MR imaging group and 1 nontherapeutic exploration in the MR imaging group (OR, 0.44; 95% CI, 0.08-2.32; P = .13). Patients in the MR imaging group were more frequently discharged from the emergency department (OR, 0.35; 95% CI, 0.13-0.94; P = .04) and had shorter LOS (33.7 vs 64.8 hours, P < .001). Gestational age, time to operation, and the presence of perforated appendicitis were similar between groups. No patient discharged without operation returned with appendicitis in either group. On multivariable analysis, the receipt of MR imaging (P < .001) and the absence of operative intervention (P = .001) were associated with shorter LOS. The mean hospital charges were similar in those with vs without appendicitis. One fetal loss occurred in the non–MR imaging group.
Conclusions and Relevance
Magnetic resonance imaging in pregnant patients with suspected appendicitis does not affect clinical outcomes or hospital charges. It allows safe discharge from the emergency department and improves resource use.
Acute appendicitis, which complicates 0.07% to 0.13% of all pregnancies, is the most common nonobstetric indication for emergency surgery in pregnant patients.1- 4 However, the diagnosis of appendicitis is not straightforward in the pregnant patient. Nausea, vomiting, and abdominal pain are common symptoms in pregnant patients who do not have appendicitis. In addition, classic signs such as right lower quadrant tenderness to palpation, guarding, and rebound may not be seen because the gravid uterus, especially in the last trimester, displaces or overlies the appendix. Furthermore, fever may not be present in all patients, and the physiological leukocytosis of pregnancy may pose diagnostic challenges.
Appendicitis during pregnancy is associated with increased fetal morbidity and mortality. It has also been implicated in increased loss of pregnancy, which ranges from 1.5% for nonperforated appendicitis to 20% for perforated appendicitis.1,5 Fear of fetal demise, as well as the fact that appendiceal perforation occurs more frequently in pregnancy,1,6 has historically led to a lower threshold for operative exploration in these patients. The risks of a potentially nontherapeutic exploration for appendicitis must be balanced with the increased risk of perforated appendicitis and potential fetal loss. A 2007 study7 reported a fetal loss rate that was 1.88 times higher in patients undergoing a negative appendectomy.
While no evidence suggests teratogenicity of ionizing irradiation at doses of less than 5 rad, a level above the range of exposure from abdominal computed tomography (CT), the American Congress of Obstetricians and Gynecologists8 guidelines recommend that other imaging procedures not associated with ionizing irradiation such as magnetic resonance (MR) imaging should be used when possible and appropriate. Although minimal, concerns about this irradiation risk associated with CT have led to MR imaging being more commonly used to diagnose appendicitis in pregnancy. This is especially true in the presence of negative or inconclusive ultrasonography (US).9- 13 While most investigations of MR imaging for suspected appendicitis are limited by sample size, studies11,12,14- 16 indicate that MR imaging is associated with improved diagnosis of appendicitis. However, it is unclear if this improvement in diagnostic accuracy translates to improved outcomes. In addition, the cost-benefit ratio of MR imaging has not been evaluated because MR imaging may increase cost without improving outcomes.
Our objective was to perform a comparative effectiveness evaluation comparing MR imaging with clinical evaluation in pregnant patients having suspected appendicitis. Outcomes included length of stay (LOS), maternal and fetal complications, and hospital charges. We hypothesized that the addition of MR imaging to the diagnostic regimen would reduce the rate of nontherapeutic exploration and shorten LOS, without compromising fetal outcomes or increasing hospital charges.
This study was approved by the Yale University Human Investigation Committee. Patient consent for this retrospective study was not required. We conducted a retrospective review of all pregnant patients seen with abdominal pain and suspected appendicitis between January 1, 2000, and July 31, 2011, at Yale–New Haven Hospital, a tertiary care academic medical referral center. Patient identification was based on International Classification of Diseases, Ninth Revision codes. To limit the sample to those with a significant initial suspicion of appendicitis, the list was cross-referenced with patients who had a surgical consultation in the emergency department.
Clinical, laboratory, diagnostic, and outcome variables were abstracted from the hospital medical record. Hospital charges were obtained from hospital decision support. Recorded were patient demographic data, pregnancy trimester, pregnancy-associated complications, initial symptoms, medical or surgical history, the presence of a surgical consultation, physical examination findings, laboratory values, imaging studies, mechanism of diagnosis, time to diagnosis and operative intervention (if any), and outcome variables. Patients who underwent CT were excluded from the analysis. Magnetic resonance imaging was performed without gadolinium. For the purpose of this study, the diagnosis of appendicitis was made by surgical impression in the operative report and pathology confirmation. Length of stay and time to surgical consultation were calculated from the time the patient was seen in the emergency department.
Data were analyzed using commercially available software (SAS version 9.3; SAS Institute Inc). Bivariate analysis was performed with t test, χ2 test, and Fisher exact test. Wilcoxon rank sum test was used for nonnormally distributed continuous variables. Multivariable logistic regression with backward elimination was used to identify factors associated with nontherapeutic exploration, and multivariable linear regression with backward elimination was used to identify all variables associated with LOS. All variables with P < .20 on bivariate analysis were initially entered into the statistical models. P < .05 was considered significant. Because cost data were available only in the aggregate, no statistical associations were evaluated.
Ninety-six pregnant patients with abdominal pain were evaluated by the surgical service in the emergency department at Yale–New Haven Hospital during the 11-year study period. Seventeen patients were seen after traumatic injuries (n = 10) or with cholecystitis that was diagnosed based on US (n = 7) and were excluded.
Seventy-nine patients were suspected of having appendicitis; 34 of them (43%) had pathology-confirmed appendicitis, while 45 of them (57%) did not. Clinical features of these patients are summarized in Table 1. Typical features of appendicitis such as nausea, abdominal pain, and leukocytosis were routinely present in our patients. Four patients underwent CT (2 had appendicitis, while 2 did not) and were excluded from further analysis. Thirty-one patients underwent MR imaging.
A clinical diagnosis of appendicitis led to operative exploration in 14 patients; 6 of these patients had nontherapeutic explorations. Their white blood cell (WBC) count on presentation ranged from 8900 to 15 300 cells/µL (median 11 700 cells/µL) (to convert WBC count to ×109/L, multiply by 0.001), and they were thought to have presumed appendicitis on initial clinical evaluation. After inconclusive US, 2 patients underwent immediate operative exploration; the remaining 4 patients were admitted for observation (range, 16.5-24 hours; median, 21 hours), and after their abdominal pain failed to improve, operative exploration was performed. No differences in the clinical features or WBC counts were observed between these 2 groups of patients. All these patients underwent nontherapeutic explorations for appendicitis, and no other pathologic findings were identified during operative exploration. Length of stay in these patients ranged from 67 to 89 hours (median, 79 hours).
Ultrasonography was used to diagnose appendicitis in 14 patients. All 14 patients with confirmed appendicitis on US underwent immediate operative exploration; 13 of these patients had intraoperative and pathology-confirmed appendicitis. Patients without US confirmation of appendicitis were observed and treated on the basis of clinical suspicion (n = 30) or received further evaluation with MR imaging (n = 31) or CT (n = 4). Of 31 patients who underwent MR imaging, 11 were diagnosed as having appendicitis, and 1 patient had equivocal MR imaging with a nonvisualized appendix. All 12 patients underwent operative exploration; however, the patient with equivocal MR imaging did not have appendicitis (Figure 1).
Overall, 8 patients without appendicitis underwent nontherapeutic exploration; 7 of these patients did not have MR imaging and were diagnosed on the basis of clinical examination after inconclusive US. One patient had MR imaging, which was equivocal, without visualization of the appendix, and had operative exploration on the basis of clinical suspicion. In total, 5 patients had equivocal MR imaging, without visualization of the appendix; 4 of these patients had inflammatory changes in the right lower quadrant and on operative exploration were found to have acute appendicitis. The remaining patient had no inflammatory changes noted in the right lower quadrant but underwent operative exploration on the basis of clinical suspicion and was found to have a normal appendix at surgery.
With respect to diagnostic accuracy, clinical diagnosis had a sensitivity of 25% and a specificity of 91%. Ultrasonography had a sensitivity of 39% and a specificity of 98%, while MR imaging had a sensitivity of 100% and a specificity of 100%. Clinical diagnosis had a positive likelihood ratio of 11.00, while US and MR imaging had positive likelihood ratios of 18.07 and 20.00, respectively. The negative likelihood ratios of clinical diagnosis, US, and MR imaging were 0.82, 0.62, and 0.00, respectively (Figure 2). Variables associated with undergoing MR imaging are summarized in Table 2. Leukocytosis (WBC count >12 000 cells/µL), guarding, or Rovsing sign on physical examination and an earlier surgical consultation were the only predictors of not undergoing MR imaging. Of those patients in whom MR imaging did not visualize the appendix, 3 were in the first trimester of gestation, 1 was in the second trimester, and 1 was in the third trimester. In all cases, the diagnosis was made based on associated findings of fat stranding or inflammation in the right lower quadrant. When patients diagnosed by US were excluded, the odds of nontherapeutic exploration without MR imaging were 8.25 (95% CI, 0.82-82.67) times greater.
Outcome variables associated with undergoing MR imaging are summarized in Table 3. Patients without MR imaging had a higher percentage of operative exploration (61% [27 of 44] vs 39% [12 of 31]), which approached statistical significance (odds ratio [OR], 0.45; 95% CI, 0.18-1.16; P = .07). In addition, they had increased incidence of admission for observation (50% [22 of 44] vs 26% [8 of 31]) and LOS 3 times higher than that of patients who had undergone MR imaging. Patients in the MR imaging group were more frequently discharged from the emergency department (OR, 0.35; 95% CI, 0.13-0.94; P = .04). The rates of perforated appendicitis were similar between groups.
Variables associated with LOS overall, as well as for patients who had nonperforated appendicitis, were evaluated using linear regression. The receipt of MR imaging (P < .001) and the absence of operative intervention (P = .001) were the only variables having a statistically significant association with shorter LOS in all patients. Undergoing MR imaging was also associated with a significantly shorter LOS in the group of patients with nonperforated appendicitis (P = .04). Adverse fetal outcomes, defined as preterm delivery or fetal loss, were similar between groups. The only fetal loss was in a patient with perforated appendicitis and a large abscess who did not undergo MR imaging. None of the preterm deliveries were in the perioperative period. The mean total hospital charges associated with undergoing vs not undergoing MR imaging were $8174 in 11 patients with appendicitis and MR imaging, $8544 in 21 patients with appendicitis and no MR imaging, $3205 in 20 patients with no appendicitis and MR imaging, and $3615 in 23 patients with no appendicitis and no MR imaging. No significant difference was found in hospital charges associated with undergoing MR imaging irrespective of the final diagnosis of appendicitis.
The diagnosis of appendicitis is not straightforward in pregnant patients. The signs and symptoms of appendicitis are nonspecific and are shared with many other common conditions in pregnancy. A relatively high rate of nausea and vomiting was seen in the patient group that did not have appendicitis. Likewise, the patient group with appendicitis did not have a high incidence of fever, and although patients with appendicitis had a slightly higher WBC count overall, no significant difference in the incidence of leukocytosis was observed between the 2 groups. Typical clinical signs of appendicitis such as guarding and rebound in the right lower quadrant, as well as Rovsing, psoas, and obturator signs, were often absent. This atypical presentation of appendicitis in the pregnant patient has been reported in other studies.1,2,5,9,17
When clinical examination is used to make the diagnosis, it is associated with a high rate of negative exploration (20% in our cohort). While our data did not demonstrate a higher incidence of adverse fetal effects in the patients who underwent nontherapeutic explorations, other investigators have shown a higher incidence of fetal loss in such patients.7 Clinical diagnosis in our study was also specific (91%) but had a much lower sensitivity of 25%. Ultrasonography was performed in all patients and was useful in detecting disease, with a specificity of 98%; however, it also had a low sensitivity of 39%. Ultrasonography is often equivocal or inconclusive, with a low sensitivity; however, given that it reliably diagnoses appendicitis when the appendix is visualized, is inexpensive, is readily available, and results in no irradiation exposure, US should be used as the initial imaging evaluation.
Magnetic resonance imaging has been demonstrated in other studies13,18 among the general population to be equivalent to CT in the diagnosis of appendicitis, with high sensitivity and specificity. In our study, MR imaging had a sensitivity of 100% and a specificity of 100%. Magnetic resonance imaging accurately identified all patients without appendicitis when a lack of inflammatory changes was considered a negative study despite nonvisualization of the appendix. Even in cases in which the appendix is not visualized, this suggests that MR imaging, which is notable for its excellent soft-tissue visualization, is useful in ruling out appendicitis. Patients who underwent MR imaging had lower incidences of operative exploration and nontherapeutic exploration after the exclusion of patients diagnosed by US. Although neither of these variables reached statistical significance, the odds ratio was 8.25 for nontherapeutic exploration without MR imaging, and the 95% CI barely crossed unity. This suggests that no further imaging is warranted if US is positive for appendicitis; however, MR imaging in cases of negative or equivocal US likely will minimize unnecessary operations.
Magnetic resonance imaging was associated with a decrease in hospital admissions and a decrease in LOS. Historically, patients who were thought to have appendicitis based on clinical suspicion were taken to the operating room, while those with clinically equivocal findings were often admitted for observation and monitored for possible progression of disease. However, patients with normal MR imaging may be safely discharged from the emergency department, which has important implications for hospital resource use. In our study, 6 patients who underwent nontherapeutic exploration on the basis of clinical diagnosis were all managed according to sound surgical principles. They were admitted based on clinical suspicion of appendicitis and after failing to improve clinically underwent operative exploration after a median of 21 hours. While this management cannot be faulted, it nonetheless resulted in 6 nontherapeutic explorations in addition to increased LOS, which may have been avoided with the use of MR imaging.
Although the use of MR imaging slightly increased the mean total hospital charges, the increase was minimal and differed by only about 10%. This is likely because the hospital charges associated with overnight admission approach the charges associated with MR imaging.
To date, the present investigation is the largest study of its kind that evaluates the usefulness of MR imaging in the diagnosis of appendicitis among pregnant patients. Nonetheless, the small sample size may have contributed to some of our results lacking statistical significance despite strong trends. This study may have been complicated by selection bias because we evaluated only patients who were seen in the emergency department with abdominal pain. Patients seen through other mechanisms were not captured. The decision to perform MR imaging also was not randomized and may have been influenced by clinical presentation; however, the likely more potent predictor of MR imaging was the year of presentation because MR imaging became much more common among patients seen at later time points (data not shown). Magnetic resonance imaging was likely performed when the clinical diagnosis was less clear, as demonstrated by the difference in the rate of leukocytosis between groups, biasing the study against a favorable effect of MR imaging. This use of MR imaging when the diagnosis is unclear is further enhanced by the low prevalence of appendicitis among pregnant patients with abdominal pain. Therefore, surgeons were relying on MR imaging when the suspicion of appendicitis was low, related to low prevalence and inconclusive clinical and US findings. However, MR imaging improved clinical efficiency and guided appropriate resource use.
To date, this is the only study of pregnant women with potential appendicitis that includes all patients with clinically suspected appendicitis, including those who did not undergo an operation and did not undergo MR imaging. Therefore, it demonstrates the clinical efficacy of MR imaging over clinical and US diagnosis alone. Magnetic resonance imaging can affect the clinical course of a pregnant patient seen with abdominal pain. This study indicates that MR imaging when available and obtained expeditiously may substantially decrease LOS and improve clinical outcomes. However, under circumstances in which MR imaging is unavailable or if clinical symptoms persist despite normal MR imaging, good clinical judgment must be used; admission for clinical observation or operative intervention may be justified in these instances depending on the clinical examination. However, the sensitivity and specificity of MR imaging have been previously established,15 and we believe this further supports the use of MR imaging in suspected appendicitis in pregnancy as a modality that affects patient care by avoiding unnecessary hospitalization and optimizing resource use.
The use of MR imaging in the diagnosis of appendicitis during pregnancy resulted in a trend toward a lower rate of nontherapeutic exploration. Magnetic resonance imaging did not seem to affect clinical outcomes or hospital charges despite the added cost of the test. However, MR imaging lowered the rate of hospital admission and decreased LOS, which are known to improve hospital throughput and reduce overall costs. Larger studies may more clearly delineate the role of MR imaging and its effect on costs and outcomes.
Accepted for Publication: September 3, 2013.
Corresponding Author: Kevin M. Schuster, MD, Department of Surgery, Yale School of Medicine, Yale University, 330 Cedar St, Boardman Bldg 310, PO Box 208062, New Haven, CT 06520-8062 (email@example.com).
Published Online: May 28, 2014. doi:10.1001/jamasurg.2013.4658.
Author Contributions: Drs Fonseca and Schuster had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Fonseca, Schuster, Lui, Davis.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Fonseca, Schuster, Maung, Lui, Davis.
Critical revision of the manuscript for important intellectual content: Schuster, Kaplan, Maung, Lui, Davis.
Statistical analysis: Fonseca, Schuster.
Administrative, technical, or material support: Schuster.
Study supervision: Schuster, Kaplan, Maung, Lui, Davis.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented at the 93rd Annual Meeting of the New England Surgical Society; September 22, 2012; Rockport, Maine.
Correction: This article was corrected on June 5, 2014, for an error in Table 3 and text.