Free air on plain film: Do we need a computed tomography too?

CONTEXT: Standard teaching is that patients with pneumoperitoneum on plain X-ray and clinical signs of abdominal pathology should undergo urgent surgery. It is unknown if abdominal computed tomography (CT) provides additional useful information in this scenario. AIMS: The aim of this study is to determine whether or not CT scanning after identification of pneumoperitoneum on plain X-ray changes clinical management or outcomes. SETTINGS AND
DESIGN: Retrospective study carried out over 4 years at a tertiary care academic medical center. All patients in our acute care surgery database with pneumoperitoneum on plain X-ray were included. Patients who underwent subsequent CT scanning (CT group) were compared with patients who did not (non-CT group). STATISTICAL ANALYSIS USED: The Wilcoxon rank-sum test, t-test and Fisher's exact test were used as appropriate to compare the groups.
RESULTS: There were 25 patients in the non-CT group and 18 patients in the CT group. There were no differences between the groups at presentation. All patients in the non-CT group underwent surgery, compared with 83% (n = 15) of patients in the CT group (P = 0.066). 16 patients in the non-CT and 11 patients in the CT group presented with peritonitis and all underwent surgery regardless of group. For patients undergoing surgery, there were no differences in outcomes between the groups. After X-ray, patients undergoing CT required 328.0 min to arrive in the operating room compared with 136.0 min in the non-CT group (P = 0.007).
CONCLUSIONS: In patients with pneumoperitoneum on X-ray and peritonitis on physical exam, CT delays surgery without providing any measurable benefit.

INTRODUCTION

Perforation of an intra-abdominal viscus is an emergency that requires prompt management. The standard teaching has been that in the presence of clinical signs of acute abdominal pathology, pneumoperitoneum identified on plain X-ray obviates the need for further imaging and constitutes an indication for laparotomy.[1] In spite of this longstanding surgical teaching, there are theoretical reasons further imaging may be helpful, even when pneumoperitoneum is seen on X-ray.[2] Plain X-ray can produce a false positive diagnosis of pneumoperitoneum, which might result in unnecessary laparotomy. The reported specificity of plain X-ray for pneumoperitoneum ranges from 53% to 89.2%, so this is a real concern.[34] Abdominal computed tomography (CT) is widely available and is very specific and sensitive for pneumoperitoneum. The overall sensitivity and specificity of CT for gastrointestinal perforation is generally in the 80-100% range depending on the study.[567] It can also frequently identify the exact site of gastrointestinal perforation when present and diagnose a wide range of other intra-abdominal pathology.[89101112131415] The accuracy of CT for identifying the site of perforation ranges from approximately 70% to 90%.[56] The additional information gained by CT may be useful in either directing or avoiding surgery in patients with pneumoperitoneum on plain X-ray.[1617] One retrospective study showed an increasing use of CT and laparoscopic repair of peptic ulcer in recent years; many patients underwent CT scanning after X-ray.[18] Others have described non-operative management of sealed duodenal ulcers and perforated diverticulitis with CT as an adjunct.[1920]

These potential benefits must be balanced against the cost, delay and medical risks such as radiation exposure and possible acute kidney injury or allergic reactions incurred by performing CT.[21222324] Despite the theoretical benefits of CT in cases, in which pneumoperitoneum has already been identified on plain X-ray, the cost and radiation exposure are only justifiable if CT actually results in improved patient outcomes. Thus, we set out to perform a retrospective study to evaluate the whether CT scanning influences management and outcomes in patients with plain film confirmed pneumoperitoneum.

SUBJECTS AND STUDY METHODS

Study design and patient selection

This was a retrospective study carried out from January 2008 to January 2012 at a tertiary care teaching hospital with 24-h CT and operating room (OR) capacity and 24 h in-house surgical and radiology attending coverage. We included patients 18 years of age or older with pneumoperitoneum identified on plain X-ray. Patients who underwent a CT scan prior to plain X-ray or who had pneumoperitoneum seen on X-ray after recent abdominal surgery were excluded from the study. American Board of Surgery certified general surgeons with additional certification in surgical critical care made the decision to operate or not and performed all the surgery, assisted by senior surgical trainees. Patients were identified retrospectively from our acute care surgery database, which includes all non-trauma patients admitted to, consulted on or operated on in our division of trauma and emergency surgery. In this database, trained coders apply the diagnosis at the time of discharge. In order to identify these patients, the acute care surgery database was searched by International Classification of Diseases (ICD)-9 diagnoses at the time of discharge. The following ICD-9 codes were used: (533.1, 533.2, 533.5, 533.6, 569.83, 531.1, 531.2, 531.5, 531.6, 532.1, 532.2, 532.5, 532.6, 534.1, 534.2, 534.5, 534.6, 562.01, 562.03, 562.11, 562.13, 567.21, 567.22, 567.29, 567.9 and 567.41) to encompass the vast majority of diagnoses that could lead to pneumoperitoneum. Patient charts were then reviewed to determine if they met inclusion or exclusion criteria.

Patient characteristics that were assessed included the following: Age, gender, presence or absence of peritonitis, systolic blood pressure (SBP), heart rate, temperature, admission white blood cell count, Charlson comorbidity,[25] American Society of Anesthesiologists class[26] and diagnosis. All were recorded at the time of presentation. Outcomes measured included the need for surgery, surgical technique (open vs. laparoscopic), procedure type, need for ostomy, intensive care unit (ICU) stay, hospital length of stay (LOS), development of pneumonia, vasopressor requirement, peak post-operative creatinine, operative time, time from the emergency department (ED) registration to surgery, time from X-ray to surgery and time from X-ray to CT. These variables were compared between the two groups (CT vs. non-CT).

Analysis

Continuous variables were summarized using mean ± standard deviation (SD) or median with interquartiles, whichever was more appropriate. Two-sample t-tests were used to compare the mean between the two groups (CT vs. non-CT) for normally distributed continuous variables and the Wilcoxon rank sum test was used to compare the shift in distribution for continuous variables with skewed distributions. Categorical variables were summarized using frequency and percentage and compared using Fisher's exact tests. All analyses were conducted using SAS (The SAS Institute, Cary, NC, USA) version 9.3.

Institutional Review Board Approval was obtained from our institution's Review Board Committee.

RESULTS

Demographics

A total of 43 patients in the acute care surgery database met the study criteria described above. Of the 43 patients, 18 underwent CT scanning after pneumoperitoneum was identified on plain X-ray (CT group) and 25 underwent plain X-ray only (non-CT group). Demographic and clinical characteristics are listed in Table 1. There were no significant differences between the baseline characteristics of the groups. The mean age of these patients was 60.6 years (SD = 17, range: 28-92). Six patients (14.0%) presented with hypotension (SBP <90) and 27 patients (62.8%) presented with peritonitis. The most common diagnoses were perforated sigmoid diverticulitis (35%) and perforated peptic ulcer (33%) [Table 2].

Table 1

Patient demographics

Table 2

Final diagnoses

Clinical management and outcomes

Patient clinical management and outcomes were compared between the CT group and the non-CT group [Table 3]. When the decision to operate was compared between the two groups, all patients in the non-CT group underwent surgery, while three patients did not undergo surgery in the CT group (100% vs. 83%, P = 0.066). However, all patients with peritonitis (n = 27 [CT group 11, non-CT group 16]) underwent surgery regardless of group. The mean LOS was 11.4 ± 10.2 days. Patients who came through the ED and underwent surgery in the non-CT group had a shorter time between ED registration and OR arrival than patients in the CT group (median 278 min vs. 523 min, P = 0.007). For all patients undergoing surgery, the time between X-ray and arriving to the OR was shorter for the non-CT group (median of 136 min vs. 328 min, P = 0.007). The median time between X-ray and CT in the CT group was 174 min (interquartile range, 91-301 min). No differences were noted in a number of other outcomes including mortality, hospital LOS, ICU admission, need for vasopressors, operative time or performance of a laparoscopic procedure. Among patients who did undergo surgery, the most common surgical procedures in both groups were sigmoid colectomy with end colostomy (Hartmann's procedure) and repair of gastroduodenal ulcer [Table 4]. Perforated sigmoid diverticulitis was the most common final diagnosis in the CT group (50%, n = 9) compared with only 24% (n = 6) in the non-CT group. Gastroduodenal perforation was more common in the non-CT group (40%, n = 10) than in the CT group (22%, n = 4). The non-CT group included three of the four colonic perforations after colonoscopy [Table 2].

Table 3

Surgical management and outcomes

Table 4

Procedures performed

Of the operated cases, 14 had surgical evidence of pneumoperitoneum. One had an internal hernia. All 14 cases with surgical evidence of pneumoperitoneum had evidence of pneumoperitoneum on CT scan (sensitivity = 100%). Specificity of CT-identified free air for gastrointestinal perforation was also 100% (all 14 cases of pneumoperitoneum on CT scan proved to have evidence of perforation on operative examination). A comparison of CT diagnosis and OR diagnosis was performed to see if the CT diagnoses matched the OR diagnoses. The three patients who did not undergo surgery were excluded, which left a total of 15 patients. 11 out of the 15 diagnoses were accurate (73% accuracy). Of the remaining four cases, two were gastroduodenal perforations in which gastrointestinal perforation was correctly identified, but no specific source of perforation could be identified on CT. One was a case of duodenal perforation incorrectly diagnosed on CT as perforated right-side diverticulitis and one was a case of a leaking around a gastrostomy tube incorrectly diagnosed on CT as transenteric placement of the gastrostomy tube.

DISCUSSION

Traditional teaching eschews further imaging in patients with clinical peritonitis and pneumoperitoneum identified on plain X-ray, yet the wide and relatively rapid availability of CT has challenged this tenant. No studies have studied the use of CT in this specific scenario. We attempted to investigate the possible advantages and disadvantages of CT in patients with pneumoperitoneum on plain X-ray with this retrospective study.

Limitations

The primary limitations of our study are related to its relatively small size and retrospective, single-center design. It is entirely possible that the practice patterns with regard to CT scan use and the interpretation and utilization of the results are not representative of other institutions or practice environments and so the generalizability of our results can be questioned. Although we did not find significant advantages or disadvantages to performing CT scans in patients with pneumoperitoneum on plain X-ray, the possibility of a type 2 error, in which outcome differences exist between the CT and non-CT groups, but were not detected due to our relatively small sample size, remains. This may be difficult to overcome in future observational studies as the trend toward wider use of CT scanning for evaluation of acute abdominal pain continues, since many patients now undergo CT as their initial radiographic test.[27] An adequately powered randomized prospective design could overcome this limitation, as well as reduce the possibility of undetected systematic differences between the groups, which can bias results in a retrospective study.

Decision to obtain a CT scan

Obtaining a CT after plain-film confirmed pneumoperitoneum was common in our study. 41.9% of all patients with pneumoperitoneum underwent CT. Among patients with peritonitis, 59% underwent CT, showing that a large percentage of patients with peritonitis and pneumoperitoneum underwent CT, which would contravene traditional teaching. Due to the retrospective nature of our study it was not possibly to reliably identify the indications for CT scanning in each case. Although it is unclear what the practice patterns are at other institutions, another single-institution study suggests that over the course of the last decade the use of CT scan has increased for the diagnosis of perforated peptic ulcer (from lesser than 10% of patients undergoing CT in 2003-100% of patients undergoing CT in 2009), including in-patients who had pneumoperitoneum identified on plain X-ray.[18] Based on the clinical variables we collected, patients in the two groups appeared to be equivalent. This argues that the choice to perform CT was not based on a systematic difference in clinical status, though there was a pattern of patients with different diagnoses undergoing different types of imaging. The distribution of sigmoid diverticulitis, gastroduodenal perforation and colonoscopic perforation was not uniform, as discussed above and shown in Table 2. The suspected diagnosis may influence whether or not a CT scan is obtained. Perhaps, surgeons who suspect perforated diverticulitis are more likely to order a CT scan in hopes of managing the disease process non-operatively whereas surgeons suspecting a gastroduodenal perforation are less likely to obtain a CT scan because they don't think it will alter management. There may also be clinical factors not captured in this study that make surgeons more likely to proceed to surgery without CT in cases of gastroduodenal perforation – such patients may look different clinically in ways that we cannot measure, but that influence the decision to either operate immediately or obtain further imaging.

Decision to operate

All three patients that did not undergo surgery were in the CT group and all were successfully treated non-operatively. Two had diverticulitis and one had a colonoscopic perforation. While CT cannot truly be said to make non-operative treatment possible-it is not therapeutic-it may be useful in allowing accurate diagnosis and planning of non-operative therapy and therefore provide the comfort needed to avoid laparotomy in a subset of patients with pneumoperitoneum on plain X-ray. These findings suggest that in patients with pneumoperitoneum on plain X-ray and no peritonitis, there might be a role for CT scanning in helping to determine, which patients can safely avoid surgery. It should be noted, however, that CT did not result in the decision to forego surgery in any patients with peritonitis on physical exam. None of the patients managed non-operatively had peritonitis and all of patients with peritonitis underwent surgery whether they had a pre-operative CT scan or not. This suggests that in the presence of peritonitis and pneumoperitoneum on plain X-ray, obtaining a CT scan will not impact the decision to operate.

We did have one false positive diagnosis of pneumoperitoneum on plain X-ray. The false positive rate in our study (1/43, 2.3%) compares favorably with the literature in which most studies report a false positive rate between 10 and 14% with only one study showing a false positive of 0%.[2829303132333435] Patient with a false positive radiographic diagnosis of pneumoperitoneum on plain X-ray had peritonitis on exam, underwent CT scanning that identified a small bowel volvulus without perforation and subsequently underwent an urgent laparotomy revealing small bowel volvulus around an adhesive band without infarction requiring a lysis of adhesions only. In this case, the false positive abdominal radiograph occurred in a patient who nonetheless required urgent surgery, suggesting that a strategy of basing surgical exploration without further imaging on the presence of both pneumoperitoneum and peritonitis may not result in unnecessary surgery even in the case of a false-positive X-ray.

Outcomes

While CT scanning did not impact the decision to operate in patients with peritonitis, we also hypothesized that by providing pre-operative information on the diagnosis and concurrent intra-abdominal pathology, it might result in improvements in surgical care such as shorter operative times or more laparoscopic operations, which might in turn result in better clinical outcomes such as shorter ICU and hospital LOS or lower mortality. Pre-operative CT correctly identified gastrointestinal perforation in all patients undergoing surgery and identified the source of perforation in 73%, but this resulted in no discernable benefit.[3637] There were no statistically significant differences in outcomes (surgery vs. no surgery, laparoscopic vs. open, LOS, etc.) between the two groups. This suggests that pre-operative imaging after pneumoperitoneum has been identified on plain X-ray, even if it further hones the pre-operative diagnosis, may not result in tangible benefits in the OR. This is in line with other literature showing that early abdominal CT scanning improves diagnostic accuracy, but does not necessarily lead to improved outcomes.[36] All sources of gastrointestinal perforation that resulted in pneumoperitoneum on plain X-ray in this study are readily identified at laparotomy or laparoscopy and while it may be comforting to the surgeon to know the diagnosis prior to surgery, we did not identify any demonstrable benefits.

Disadvantages to CT scanning?

While there were no clear advantages to CT in patients undergoing surgery or presenting with peritonitis, we did not detect any clear drawbacks either. The main potential disadvantage of CT scanning that we identified was delay. For patients in the CT group undergoing surgery, there was a significantly longer elapsed time from ED registration (for patients presenting to the ED) to arrival in the OR (523 min vs. 278 min [P = 0.007]) and a significantly longer elapsed time between obtaining the X-ray that demonstrated pneumoperitoneum and arrival in the OR for all patients undergoing surgery (328 min vs. 136 min [P = 0.007]). However, as noted above, there were no differences in clinical outcomes such as ICU LOS, hospital LOS or mortality, making it difficult to assert that significant harm came from this delay. While prompt surgical care is considered as important in cases of gastrointestinal perforation, some patients may benefit from a period of pre-operative optimization and delays of the magnitude described here may not be significant.[38] Other potential disadvantages of CT besides delay include cost, radiation exposure and adverse reactions to contrast agents. We did not identify any cases of contrast-related morbidity such as nephropathy or hypersensitivity in our cohort and the median peak creatinine was the same in the CT and non-CT groups. The incidence of contrast-related complications in larger studies ranges from 3% to 15% depending on the definitions used, the exact type of contrast and the method of administration.[3940] Although we did not perform a cost analysis, abdominal CT scanning typically results in increased costs of approximately $1000 when used for the urgent diagnosis of appendicitis or trauma.[2122] The impact of radiation exposure would not be expected to be manifest without long-term follow-up. Concern about long-term risks has focused on patients with pathology requiring multiple repeat CT scans and on children. While some authors argue that the benefits of the information gained outweigh the potential risks for patients with acute pathology, this premise is called into question when no change in management results from CT scanning (as in patients with peritonitis in this study) and has been challenged by other reports that implicate even single CT scans as a risk factor for malignancy.[414243]

CONCLUSION

In summary, our study suggests that CT scanning may be useful in patients with pneumoperitoneum on plain X-ray without evidence of peritonitis, by allowing accurate diagnosis and thus assisting in non-operative management of selected patients with gastrointestinal perforation. However, in patients with peritonitis on exam, it does not appear to significantly alter management either by effecting the decision to operate or by influencing the surgical approach or outcomes. In spite of the limitations noted above, we believe this study provides some evidence that in a modern era characterized by the exploding use of imaging technology, some traditional teaching remains relevant and that patients with pneumoperitoneum and peritonitis need surgery, not further imaging.