INTRODUCTION
The Computed Tomography (CT) has great accuracy in the evaluation of most urgent abdominal diseases, surpassing the performance of ultrasonography (US) and conventional x-ray (RX) 1 , 2 and, therefore, it has been increasingly used in evaluation of patients with acute abdominal pain (AAP)3 , 4 . Its introduction in the emergency room (ER) has been associated with increased confidence in the diagnosis and improved medical decision-making5 . In Brazil, the total number of CT procedures increased from 540,067 to 921,485 exams between 2008 and 2011 in the Brazilian Unified Health System (SUS – Sistema Único de Saúde), with abdominal and pelvic CTs being the second most often performed ones 6 .
However, the increased use of CT in the ER is associated with increased hospital costs 7 and patient exposure to ionizing radiation 8 , 9 , which has led to a growing demand for conscientious use of CT, especially in the ER, where the number of CT procedures continues to grow 10 , 11 .
‘Choosing Wisely’, a global initiative of the American Board of Internal Medicine (ABIM) foundation, founded in 2012 to promote health discussions aimed at reducing the misuse of health resources, warns in one of its publications that CTs are not always necessary for routine assessment of abdominal pain12 . According to the organization, 73% of US physicians admit that the frequency with which diagnostic tests or medical procedures are requested is a serious health care problem13 .
Systermans et al. (2014) warn that, while abdominal CT correctly shows disease in 92-97% of patients, CT scans show normal results in 44% of patients and correspond to clinical data in 70% of cases. Thus, the authors even questioned whether CT scans might have decreased clinical skills or whether physicians’ poor confidence in their capacity to evaluate patients with acute conditions led to the replacement of traditional physical examination by modern imaging techniques14 .
The main objective of this study is to show the association between clinical, tomographic and definitive diagnoses in patients with AAP in the ER aiming to highlight situations in which the use of CT may be considered unnecessary. It is also intended to cause reflection on the value of clinical diagnosis in the presence of new medical technologies.
Material and Methods
A cross-sectional study was designed, with retrospective data collection from Medhosp (Medlynx®) electronic medical record from a high-complexity private hospital serving a region of approximately 1 million inhabitants, which has been accredited internationally (Accreditation Canada International) since 2012.
The medical records of patients with AAP lasting less than 7 days submitted to urgent abdominal CT between January 1, 2016 and December 31, 2017 were included. Patients with a history of trauma lasting less than 30 days, CT indicated for clinical or surgical treatment control and incomplete medical records were excluded from the sample.
The clinical diagnosis (pre-CT) was recorded according to the last suspected diagnostic hypothesis up to CT performance. The CT findings were collected only from radiologist reports, considering the first listed diagnosis when two or more were present.
Tomographic results were classified as positive for AAP, normal abdominal CT (NACT) and negative for AAP, according to the concept of actionable findings, as proposed by Obuchowski and Modic15 and Gardner et al.16 . Normal CT and those considered negative were classified as nonspecific abdominal CT (NSCT).
The final diagnosis was based on discharge reports, description of the surgical procedure or anatomopathological report, with the investigation in the database being carried out within 30 days after discharge from the emergency room for patients without hospitalization or clear diagnostic confirmation in the ER.
Nonspecific abdominal pain (NSAP), defined as abdominal pain without an identifiable alternative diagnosis or attributable organic cause17 , 18 , was considered when it was not possible to establish a specific diagnosis for AAP, clinically for pre-CT diagnoses and until discharge from the ER, or on subsequent consultations for final diagnoses.
CT scans were considered unnecessary when in Concordance with the clinical diagnosis in situations where the final diagnosis and the pre-CT one were also in Concordance. The use of X-rays and US before the CT indication was also recorded.
Statistical analysis was performed using the R statistical software (R Core Team 2018). The Kappa coefficient (K) was applied to evaluate the Concordance between the pre-CT, post-CT and final diagnostic groups, adopting a p value <0.05 as significant. The following classification was used for the interpretation of K values: 0: poor; 0.01-0.20: slight; 0.21-0.40: fair; 0.41-0.60: moderate; 0.61-0.80: substantial; 0.81-1: excellent 19 .
This study was approved by the Human Research Ethics Committee through Opinion N. 2,474,591.
Results
A total of 834 records were found to be eligible for the study. The sample consisted of 442 women (53%) and 392 men (47%), with a mean age of 44 years (±20.6 years) for both genders. The mean duration of abdominal pain until the first consultation was 46.8 hours (±44.5 hours), with a median of 24 hours (0.5-196 hours). Up to the time of CT indication, 89 (10.7%) patients had undergone only abdominal X-ray, 401 (48.1%) only abdominal US and 187 (22.4%) both examinations. US was not performed in 244 (29.3%) cases and 157 (18.8%) were referred to CT without any previous imaging study. CT was normal in 108 (12.9%) cases and nonspecific for AAP in 231 (27.7%) cases.
For both pre-CT, post-CT and final diagnoses, the most frequent diagnoses were NSAP, obstructive uropathy (OU) and appendicitis (AP), which together account for 73.6% of all clinical diagnoses, 58.5 % of post-CT diagnoses and 61.3% of definitive diagnoses ( Table 1 ).
N | % | |
---|---|---|
Clinical diagnoses | ||
Nonspecific abdominal pain | 314 | 37.6% |
Obstructive uropathy | 176 | 21.1% |
Appendicitis | 124 | 14.9% |
Diverticulitis | 38 | 4.6% |
Obstructive abdomen | 20 | 2.4% |
Enterocolitis | 20 | 2.4% |
Urinary tract infection (except pyelonephritis) | 18 | 2.2% |
Pancreatitis | 18 | 2.2% |
Gastroduodenitis | 16 | 1.9% |
Cholelithiasis | 11 | 1.3% |
Tomographic diagnosis | ||
Unspecific abdominal computed tomography | 231 | 27.7% |
Obstructive uropathy | 162 | 19.4% |
Appendicitis | 95 | 11.4% |
Diverticulitis | 40 | 4.8% |
Obstructive abdomen | 36 | 4.3% |
Enterocolitis | 32 | 3.8% |
Unspecific inflammatory alterations | 27 | 3.2% |
Ovarian cyst | 20 | 2.4% |
Colonic distension | 20 | 2.4% |
Pancreatitis | 19 | 2.3% |
Final diagnoses | ||
Nonspecific abdominal pain | 267 | 32.0% |
Obstructive uropathy | 148 | 17.7% |
Appendicitis | 97 | 11.6% |
Diverticulitis | 37 | 4.4% |
Obstructive abdomen | 37 | 4.4% |
Pancreatitis | 36 | 4.3% |
Gastroduodenitis | 18 | 2.2% |
Neoplasia | 15 | 1.8% |
Pyelonephritis | 15 | 1.8% |
Acute gastroenterocolitis | 14 | 1.7% |
Of the 267 patients diagnosed with NSAP, 52.6% had the diagnosis of NSCT. Analysis of the most frequent specific conditions showed that the tomographic diagnosis of OU corresponded to 83.8% of the definitive diagnoses and in 86.6% of cases of AP. Of the patients with NSAP, 43.5% had the same clinical suspicion and 53.9% at the CT. Among those who had the final diagnosis of NSAP and an initial diagnosis of another clinical hypothesis, 55.3% were diagnosed with NSCT. Concordance between the three diagnostic stages was moderate for unspecific causes of AAP, with a kappa coefficient of 0.41; p <0.01 ( Table 2 ).
Final Diagnosis | Kappa | p-value | ||
---|---|---|---|---|
Initial diagnosis | Nonspecific abdominal pain | Other | ||
Nonspecific abdominal pain | n (%) | n (%) | ||
CT diagnosis | ||||
NSCT | 61 (52.6) | 1 (1.8) | 0.410 | <0.001 |
Another | 55 (47.4) | 56 (98.2) | ||
Other | ||||
CT diagnosis | ||||
NSCT | 83 (55.3) | 17 (9.0) | 0.479 | <0.001 |
Other | 67 (44.7) | 171 (91.0) |
Legend: CT – Computed Tomography. NSCT – Unspecific abdominal computed tomography. n – absolute frequency. % – percentage of relative frequency. Kappa – Kappa Coefficient of Concordance.
Of the patients diagnosed with OU by CT, 67.2% had the same clinical suspicion. Of the patients who had a final and initial diagnosis of OU, 82.4% were in Concordance with the tomographic diagnosis, although the CT showed obstructive calculi in 87.2% of the patients without this clinical suspicion ( Table 3 ).
Final Diagnosis | Kappa | p-value | ||
---|---|---|---|---|
Initial diagnosis | Obstructive uropathy | Other | ||
Obstructive uropathy | n (%) | n (%) | ||
CT diagnosis | ||||
Obstructive uropathy | 84 (82.4) | 20 (36.4) | 0.464 | <0.001 |
Other | 18 (17.6) | 35 (63.6) | ||
Other | ||||
CT diagnosis | ||||
Obstructive uropathy | 41 (87.2) | 11 (3.6) | 0.800 | <0.001 |
Other | 6 (12.8) | 296 (96.4) |
Legend: CT – Computed Tomography. n – absolute frequency. % – percentage of relative frequency. Kappa – Kappa Coefficient of Concordance.
On the other hand, the initial suspicion of other diagnoses corresponded to the CT and with the final diagnosis of urinary obstruction in 96.4% of cases. The Concordance between the diagnoses was considered moderate for OU (Kappa 0.46; p <0.01). The values are shown in Table 3 . Half of the patients diagnosed with AP had the same clinical suspicion. Of these, CT was concordant in 91.7%. On the other hand, of the patients with confirmed AP and no clinical suspicion, the CT was positive in 83.3% of cases. The correspondence between the diagnoses was considered excellent, with a kappa coefficient of 0.87, p <0.01 ( Table 4 ).
Final Diagnosis | Kappa | p-value | ||
---|---|---|---|---|
Initial diagnosis | Appendicitis | Other | ||
Appendicitis | n (%) | n (%) | ||
CT diagnosis | ||||
Appendicitis | 44 (91.7) | 2 (3.9) | 0.879 | <0.001 |
Other | 4 (8.3) | 49 (96.1) | ||
Other | ||||
CT diagnosis | ||||
Appendicitis | 40 (83.3) | 5 (1.4) | 0.842 | <0.001 |
Other | 8 (16.7) | 359 (98.6) |
Legend: CT – Computed Tomography. n – absolute frequency. % – percentage of relative frequency. Kappa – Kappa Coefficient of Concordance.
DISCUSSION
The study analyzed the concordances between pre-CT, post-CT and final diagnoses of 834 medical records of patients with AAP, in search for situations where CT might be considered unnecessary. Of the total, 108 (12.9%) CT scans with normal results were found and 231 (27.7%) were considered unspecific for AAP, which is lower than the frequency reported in the literature 14 but in agreement with those reported for the final diagnosis of NSAP 20 , 21 .
The analysis was directed to diagnoses of NSAP, OU and AP, which together account for more than 70% of the final diagnoses. As broadly reported in other studies, the CT showed a high degree of accuracy1 , with a moderate correlation between clinical, post-CT and final diagnoses for NSAP and OU and excellent for AP. However, patients with a final and tomographic diagnosis of NSAP, OU and AP already showed clinical suspicion in more than 50% of the cases, reaching more than 90% for patients with AP, and thus, can be considered clinically unnecessary.
In fact, a recent study showed that the use of TC in the ER did not add any improvement to the diagnosis of OU, except in cases of great diagnostic uncertainty or in patients requiring immediate surgical intervention22 . Similarly, the US assessment is associated with a 50% reduction in CT use in the ER 23 and it has been considered safe in the reevaluation of patients with appendicitis24 or even those with unspecific abdominal complaints25 , without the need for ionizing radiation. In this study, 29.3% of patients were not submitted to the pre-CT US assessment and 18.8% to any other type of imaging assessment.
Moreover, pre-CT diagnoses of the main urgent diagnosed conditions were reliable, with an accuracy of over 50%, exceeding 90% in cases of AP. These data indicate that CT may be used prematurely, and it raises the concern that the clinical diagnosis is being disregarded in relation to tomographic data. This reiterates the issue raised by Systermans et al.14 about a possible replacement of clinical skills by modern imaging techniques or physicians’ insecurity in clinically diagnosing patients with acute conditions.
Kanzaria et al26 warned about the low reliance on clinical uncertainty by both the medical staff and patients, and the need to eliminate it at all times reinforces the idea that technology solves all problems and that an early diagnosis brings benefits without damage.
In addition to increased costs7 and exposure to ionizing radiation 8 , 9 , unnecessary examiantions may lead to incidental findings, which in turn may trigger a cascade of events involving further testing or medical procedures, of little benefit to patients26 , 27 .
Moreover, the possibility of malpractice suits, which is often the cause of complaints against doctors 28 , plays a decisive role in this process. If, on the one hand, 66% of physicians believe they are responsible for ensuring that their patients are not submitted to unnecessary procedures 13 , fear of misdiagnosis and malpractice suits has been reported by over 60% of professionals as the main reasons for requesting diagnostic tests previously considered unnecessary by them26 .
The reasons and solutions for this may have deeper roots. Among the several factors involved in the increase in the number of medical malpractice lawsuits, the most important are the deterioration in the quality of the doctor-patient relationship, together with the poor training of doctors during undergraduate and graduate training28 . Lobo29 states that, in his study site, almost all patients could not satisfactorily report the medical consultation soon going through it and that the faster the consultation, the higher the number of requested examinations, which demonstrates the deterioration of the medical-patient relationship.
Thus, since it is difficult to change professionals’ habits and standards, even a legal reform may be insufficient to reduce the exaggerated use of diagnostic tests and the fight against the so-called defensive medicine can have a limited effect, if it is not developed during undergraduate medical training26 .
Moreover, according to Choosing wisely, 72% of the interviewed physicians admitted requesting unnecessary exams at least once a week; 53% of them would request unnecessary exams if the patients insisted, and 70% said that patients often give up diagnostic tests when they are told why they are unnecessary 13 . Recognizing that patients are increasingly seeking information and willing to participate in the decision-making related to their cases should be currently considered29 and reinforces the need to improve crucial points, such as the development of the doctor-patient relationship and communication between doctors, patients and family members, still during the undergraduate training 28 .
Additionally, in the scientific literature, the high accuracy of CT in the diagnosis of AAP is well documented1 but information on normal CT or conditions in which CT has little influence on medical management is scarce. In this sense, an approach to the topic focusing on conditions such as age range, gender, pain duration and characteristic, complementary exams, etc., which may signal situations in which CT can be dispensed with, can be of great importance as a way to provide scientific support and increase physician confidence in indicating or contraindicating computed tomography in the ER.
Finally, although the retrospective data collection model does not allow evaluating the reasons why physicians requested the CTs considered unnecessary in the ER, the results raise the still latent question in Brazil about the increase in the use of imaging exams, which are not harmless to the patient and whose solution may have one of its most effective aspects if it is worked still during the undergraduate training of new doctors. As stated by Lobo29 , despite the use of very advanced technologies applied to medicine, doctors will still be assigned to discuss the cases with their patients and alleviate their fears, since computers lack emotions and empathy.
FINAL CONSIDERATIONS
A high number of CT scans was found to be unnecessary to diagnose the main clinically suspected diseases in the ER, whereas there was a considerable correlation between the clinical and final diagnoses. This warns of a possible depreciation of and lack of confidence in the clinical diagnosis against the overvaluation of modern medical technologies. The reasons for this may be due to nonclinical factors, such as poor patient-physician communication and fear of malpractice lawsuits, the solutions of which may have more effective effects when initiated during the undergraduate training of new professionals.