Preventable diagnostic errors of lower gastrointestinal perforation: a secondary analysis of a large-scale multicenter retrospective study

Abstract

Background

Lower gastrointestinal perforation (LGP) is an acute abdominal condition associated with a high mortality rate. Timely and accurate diagnosis is crucial. Nevertheless, a diagnostic delay has been estimated to occur in approximately one-third of the cases, and the factors contributing to this delay are yet to be clearly understood. This study aimed to evaluate the diagnostic process for appropriate clinical reasoning and availability of image interpretation in cases of delayed diagnosis of LGP.

Methods

A secondary data analysis of a large multicenter retrospective study was conducted. This descriptive study analyzed data from a multicenter, observational study conducted across nine hospitals in Japan from January 2015 to December 2019. Out of 439 LGP cases, we included 138 cases of delayed diagnosis, excluding patients with traumatic or iatrogenic perforations, or those secondary to mesenteric ischemia, appendicitis, or diverticulitis. Clinical history and computed tomography (CT) imaging information were collected for 138 cases. Additionally, information on the clinical course of 50 cases, which were incorrectly diagnosed as gastroenteritis, constipation, or small bowel obstruction, was also collected.

Results

In 42 (30.4%) cases of delayed diagnosis of LGP, CT imaging was performed before diagnosis, indicating a missed opportunity for timely diagnosis. Moreover, 33 of the 50 patients initially diagnosed with gastroenteritis, constipation, or small bowel obstruction at the time of initial examination had atypical findings that were not consistent with the initial diagnosis. Of the 138 cases with delayed diagnosis in our study, 67 cases (48.6%) showed problems with either the interpretation of CT scans or with the process of clinical reasoning.

Conclusion

Our retrospective study results indicate that approximately half of the cases with delayed diagnosis of LGP were due to problems in interpreting CT images or in clinical reasoning. This finding suggests that clinical reasoning and image interpretation by radiologists are important in improving the diagnostic process for LGP.

Introduction

Background

Diagnostic errors are the most common medical errors and have a considerable negative impact on a patient’s physical and psychological well-being, as well as the medical care costs [1]. The National Academy of Medicine report notes that diagnostic errors are a national emergency and offers recommendations for their prevention and mitigation [1]. Abdominal pain is a frequent chief complaint of patients presenting to the emergency department and is associated with a high risk of diagnostic error owing to its atypical presentation and clinical course [2]. It has also been reported that abdominal pain accounts for 5.5% of all diagnostic errors in the emergency department, [3] and 35% of abdominal pain cases in the emergency department are at high risk for diagnostic error [2].

Importance

Lower gastrointestinal perforation (LGP) is an acute abdominal condition in which the bacteria in the stool spread from the small and large intestines into the abdominal cavity, causing acute diffuse peritonitis. Compared to upper gastrointestinal tract perforation, LGP is less common but more severe, with an increasing frequency of cases [4] and a mortality rate as high as 30% [5]. Treatment for LGP comprises resuscitation, antimicrobial therapy, and repair or reconstruction of the perforation site [6]. Timely and accurate diagnosis is critical for proper management. A recent study on LGP identified a delay in diagnosis in approximately one-third of the cases, and the cause for the delay was associated with physician-related factors, presence of fever, and absence of tenderness [7]. Furthermore, for the 90 cases of delayed diagnosis of LGP in that study, the most common initial diagnosis was acute gastroenteritis, followed by constipation and small bowel obstruction [7]. However, since no analysis of individual cases of delayed diagnosis was conducted, there is no clarity regarding the process by which the delay occurred. In recent years, the focus of attention in practice has been on analyzing disease-specific diagnostic pitfalls and error-prone scenarios [8]. A study of diagnostic errors in Japan identified that physicians were misled by patient history and physical examination findings [9] with diagnoses of gastroenteritis and upper respiratory tract infection, leading to diagnostic errors for LGP [10]. It has also been reported that atypical symptoms are prone to lead to diagnostic errors, and recognizing atypical symptoms may prevent the delay in diagnosing LGP [11, 12]. In their study on the delayed diagnosis of LGP, Harada et al. [7]. identified the availability of urgent radiology reports in only 26% of cases, with the absence of urgent radiology reports being cited as a possible cause of the diagnostic error. There are reports indicating that computed tomography (CT) scans have a sensitivity of approximately 96% for LGP [7]. This suggests the possibility of delayed diagnoses where CT imaging was performed and findings in the LGP were missed.

Goal of this investigation

This study aimed to conduct a retrospective review of medical records from cases of diagnostic delay in a multicenter cooperative observational study, to examine the diagnostic process and CT imaging, and to identify factors associated with the delay in diagnosing LGP.

Methods

Study design

This was a descriptive study of a secondary analysis of data from the large, multicenter, observational study conducted across nine hospitals in Japan between January 1, 2015, and December 31, 2019.

Setting

Data were obtained for patients who were evaluated for LGP at nine hospitals in Japan. For hospitals in which electronic medical records were introduced during the study period, only patients evaluated after the introduction of electronic medical records were included.

Selection of participants

Overall, 138 cases of delayed diagnosis were identified out of the 439 cases with LGP in the previous study. The original study included patients manifesting symptoms indicative of LGP, with confirmation obtained via diagnostic imaging or surgical intervention. Patients under 15 years old, those with perforations caused by trauma, medical interventions, or conditions like mesenteric ischemia, appendicitis, or diverticulitis, and those with upper gastrointestinal perforations were excluded. Secondary lower gastrointestinal perforations were excluded due to their dependence on underlying diseases, complicating diagnostic assessment.

Variables

The following variables were quantified for analysis: timely diagnosis, delayed diagnosis, and diagnostic error. Timely diagnosis was defined based on the following two criteria [7]. The first was a diagnosis of a LGP at the time of the initial visit to a facility with CT imaging capacity. The second was a diagnosis of probable LGP during the initial visit at a facility without CT imaging capacity, with the patient immediately referred to an appropriate medical facility for investigation of the acute abdominal pain. A “near miss” was defined as an incorrect diagnosis during the initial visit at a non-CT facility, with immediate transfer of the patient to a hospital with CT imaging capacity for appropriate diagnosis. “No-fault” cases were defined as an absence of perforation findings on initial CT imaging, with an ultimate diagnosis of LGP after additional testing, as previously defined [7]. Cases that did not meet the classification for “near miss,” “no-fault,” or “timely diagnosis” were classified as “delayed diagnosis.” After a review of the 439 cases with a resultant diagnosis of LGP, a delayed diagnosis was identified in 138 cases.

The clinical reasoning scenarios that could prevent a delay in the diagnosis of LGP were examined. The three most common initial diagnoses associated with a delayed diagnosis were gastroenteritis, constipation, and small bowel obstruction. The typical clinical course of these diseases, which differentiates them from a LGP, is as follows. Acute gastroenteritis is typically associated with diarrhea [13, 14], with gastroenteritis without diarrhea being atypical. Constipation is not likely to cause fever or an elevation in acute-phase inflammatory markers, such as C-reactive protein (CRP) [15]. Lastly, small bowel obstruction is less likely to cause fever or a CRP elevation; in fact, CRP elevation is atypical for small bowel obstruction without complication [16]. Furthermore, a previous study identified an elevation in CRP ≥ 10 mg/L in 77% of patients with LGP and fever in 51% of cases [7]. These two findings are sufficiently consistent to differentiate between LGP and gastroenteritis, constipation, or small bowel obstruction. Based on this evidence, we hypothesized that the following three clinical reasoning scenarios resulted in a delayed diagnosis of LGP. First, CT images were obtained; however, the LGP was missed due to the absence of a radiology report for the emergency physician. Second, an initial diagnosis of gastroenteritis was made despite the absence of diarrhea. Third, an initial diagnosis of constipation or bowel obstruction was made despite the presence of fever and elevated inflammatory markers. To verify the face validity of the hypothesized scenarios, we examined the CT imaging history for all 138 cases in the delayed diagnosis group. Additionally, details on the presence/absence of fever, diarrhea, and elevated CRP were further extracted from the medical reports for the 27 cases of gastroenteritis, 13 cases of constipation, and 10 cases of small bowel obstruction as the initial analysis.

Study size

No statistical sample size calculations were conducted because this study is a secondary descriptive analysis and is exploratory in nature.

Quantitative variables

The primary outcome measure was the frequency with which the delay in diagnosing LGP could be prevented by revising the clinical reasoning process in 50 cases initially diagnosed with gastroenteritis, constipation, or small bowel obstruction. The secondary outcome measure was the frequency with which the diagnosis of LGP was missed despite CT imaging in a group of 138 cases where the diagnosis was delayed.

Statistical methods

Outcome variables were described as a count (percentage) or median (interquartile range) as appropriate for the data type. All descriptive data were calculated using the EZR (Easy R) software Stata statistical software (Saitama Medical Center, Jichi Medical University, Saitama, Japan).

Ethical approval

This study was approved by the ethical review board of Showa University Koto Toyosu Hospital (No. 20T7044) and conducted in accordance with the Declaration of Helsinki.

Results

Of the 27 cases with an initial diagnosis of gastroenteritis, 20 (74.1%) did not present with diarrhea. Out of the 13 cases with an initial diagnosis of small bowel obstruction, three (23.1%) presented with fever, and nine (69.2%) had an elevated CRP level. Of the 10 cases with an initial diagnosis of constipation, one (10.0%) presented with fever, and three (30.0%) had an elevated CRP level. Of the 138 misdiagnosed cases of LGP, 50 cases were initially diagnosed as gastroenteritis, constipation, or small bowel obstruction. Furthermore, 33 cases (66.0%) among them had atypical symptoms or laboratory findings that were inconsistent with the initial diagnosis.

Of the 138 cases in the delayed diagnosis group, the diagnosis of LGP was missed in 42 (30.4%) cases despite a CT imaging having been performed before diagnosis. In four of these 42 cases (9.5%), CT was performed at a facility with the capability of an emergency radiologist report. In all cases, the perforation was later identified by the radiologist; therefore, the diagnosis was delayed by not using the radiology findings during the diagnostic process.

The distribution of the 138 cases, including 42 cases (30.4%) in which LGP was missed despite performing CT imaging and 33 cases in which atypical symptoms and laboratory findings were inconsistent with the initial diagnosis (23.9%), is shown in Fig. 1.

Fig. 1

Tree diagram of 138 cases of delayed diagnosis of lower gastrointestinal perforation

This retrospective study reports that potential problems in the diagnostic process were identified in 67 out of 138 cases (48.6%) that experienced diagnostic delays.

Discussion

In this retrospective study, errors were found in the diagnostic process in approximately half of the cases of delayed diagnosis of LGP. Approximately 24% had atypical symptoms or laboratory findings that were inconsistent with the initial diagnosis, and approximately 30% of the cases had problems in the process of radiological interpretations of CT imaging. Gastroenteritis, constipation, and small bowel obstruction are known common causes of abdominal pain for which the presence of fever and elevation of inflammatory markers are atypical [13,14,15, 17]. A previous study of LGP reported evidence of perforation on CT imaging in all but 3.9% of cases [7]. Based on our findings, the availability of emergency radiology reports alone may reduce diagnostic error by nearly 30%.

Based on our findings, 24% of cases with a delayed diagnosis of LGP could be prevented by appropriate clinical reasoning of differential diagnosis, even if diseases, such as gastroenteritis, are common in emergency departments. This analysis is consistent with the findings of previous reports on the “easy” diagnosis of gastroenteritis as a cause of the diagnostic error of LGP [10] and that physicians, at least in Japan, do make mistakes in the medical history and physical examination process differentiating a perforation from gastroenteritis [9]. Mamede et al. [18] have stated that improving the clinical reasoning process for abdominal pain can reduce the risk of diagnostic error. To achieve this goal, Schiff et al. [8] underlined the importance of studying error-prone scenarios and pitfalls to an accurate diagnosis. Developing expertise in recognizing the clinical course of common diseases and differentiating between diseases with similar clinical courses may improve diagnostic accuracy and reduce the likelihood of missing the diagnosis of a potentially fatal condition, such as LGP. Although the recognition rate of LGP is very high in Japan [19], Harada et al. [7] reported that emergency CT imaging and report by a radiologist are available in only 25% of cases of potential LGP, confirming the lack of clinical imaging capacity in urgent care. Another report indicated that, in Japan, clinicians use an available radiology report in their diagnostic process or overlook findings in approximately 7% of cases of LGP [20]. In our study, in 10% of cases, CT images were obtained, and an emergency radiologist was available but not consulted, resulting in the delayed diagnosis of perforation.

The strength of this study is the large, multicenter database used to examine diagnostic errors for LGP and the identification of feasible strategies to lower the likelihood of these diagnostic errors. Moreover, the detailed analysis by chart review using large-scale multicenter data for investigating diagnostic errors in LGP is a strength, helping to reduce the likelihood of these errors and identify viable strategies.

Limitations

The limitations of our study also need to be acknowledged. Foremost, this was a secondary analysis of retrospectively identified cases; therefore, the possibility of information and selection bias, as well as the effects of unexpected confounding factors not reported in medical records, cannot be denied. Second, only factors associated with a delayed diagnosis could be identified owing to the cross-sectional nature of the data, and a causal relationship cannot be attributed. As such, proposed strategies to lower the risk of diagnostic error will need to be prospectively evaluated in the future. Third, the verification of the diagnostic process is partial. Cases with initial diagnoses other than small bowel obstruction, gastroenteritis, or constipation, as well as cases with unclear initial diagnosis, could not be evaluated.

Therefore, although there are several barriers to improving diagnostic errors in Japan [16], our findings do indicate that approximately half of the diagnostic errors of LGP could be prevented by improving the clinical reasoning process in approximately 24% of cases and by providing and using radiologist reports in an urgent care setting in approximately 30% of cases. The use of emergency radiology reports and improved clinical reasoning are feasible and effective strategies to reduce diagnostic errors in LGP.

Conclusion

Our findings indicate that the frequency of delayed diagnoses in LGP can be improved by 30% through enhancements in the diagnostic process that do not readily dismiss atypical cases and by 24% through the utilization of radiological reports from CT imaging. This finding suggests that clinical reasoning and image interpretation by radiologists are important in improving the diagnostic process for LGP.