Evaluating the concordance between clinical diagnosis and imaging in acute abdomen: a study in the Korle-Bu Teaching Hospital, Ghana

¹Department of Surgery, Korle-Bu Teaching Hospital, Accra, Ghana, ²Centre for Tropical Clinical Pharmacology and Therapeutics, University of Ghana Medical School, Accra, Ghana

^&Corresponding author
Jessica Dei-Asamoa, Department of Surgery, Korle-Bu Teaching Hospital, Accra, Ghana

Introduction    

Acute abdominal pain is a common presentation to the emergency department of hospitals, accounting for 7-10% of emergency room visits [1,2]. The commonest cause of acute abdomen in Ghana is acute appendicitis, but other causes include Intestinal obstruction, typhoid perforation, gynecological causes, peptic ulcer disease, gastroduodenal perforation, non-specific abdominal pain and cholecystitis [3,4]. Diagnosis of acute abdomen requires a clinical history and physical examination, laboratory work-up, and imaging including X-ray, ultrasound scan, computed tomography (CT) scan, and magnetic resonance imaging (MRI). Clinical diagnosis is believed to be sensitive and accurate, but some studies cast doubts about its presumed accuracy, thus making imaging superior to history and examination alone [5-8]. Over-reliance on imaging modalities in settings where they are readily available and affordable may have made history-taking and examination a lost art [9] hence the higher correlation between imaging modalities and intraoperative findings than clinical diagnosis. Delay in making an accurate diagnosis in acute abdomen (which could be due to reliance on imaging, which is not readily available or costly) could lead to higher morbidity and mortality [4,10].

The degree of concordance between clinical assessment and imaging in our local practice is unknown, as such the degree to which it affects the clinical care of patients is also unknown. It is therefore imperative that via this study, we can establish the level of concordance so that we can indirectly assess the quality of clinical diagnostics hence the effectiveness of treatment for acute abdominal conditions. The findings from this study will have multiple implications for the quality of clinical and imaging diagnoses, while highlighting the overall quality or inadequacies in clinical training modules for medical students and residents. This will therefore provide a platform for which improved training modules can be designed to improve clinical diagnosis, hence avert the need to wait for imaging which leads to delays in managing patients with acute abdomen.

Methods     

Study design: this study was a prospective cohort study of consecutive patients aged 13 years or older, who presented with acute abdomen to the Adult Emergency Department of the Korle-Bu Teaching Hospital in Accra, Ghana.

Setting: the study was conducted at the Adult Emergency and General Surgery Department of the Korle-Bu Teaching Hospital between March 2022 to August 2023. Initial data was collected at the Adult Emergency and the General Surgery Units before surgery by interviewing the doctors in charge of these patients, using data from the Electronic Health Records System, referral letters, and reports of imaging. Follow-up data was collected on the General Surgery Units and clinics from the Electronic Health Record System.

Participants: ethical clearance for the study was sought from the Institutional Review Board of the Korle-Bu Teaching Hospital with IRB number: KBTH-IRB/000128/2021. All consecutive patients aged 13 years and older who reported to the Emergency Department of the Korle-Bu Teaching Hospital and had a clinical impression or imaging diagnosis of acute abdomen not due to trauma, who were not pregnant, not unconscious, with no language barrier, and needed imaging to make a diagnosis and or needed surgical management and consented to the study were enrolled into the study. Some patients presented with diagnostic imaging ordered from the referring hospital/clinic, and those who required imaging after initial clinical evaluation had the respective imaging ordered. The diagnosis was based on the clinical diagnosis with or without imaging.

Variables: the patients were evaluated clinically (history and physical examination) and while some required imaging and/or surgery, others did not require further evaluation or intervention depending on the clinical information. Outcomes were the imaging diagnosis, the intra-operative or discharge diagnosis (for those who did not have surgery), and post-operative complications.

Data sources/measurement: the data captured via the questionnaire purposely designed for this study using Google Forms included aspects of the demographic data, essential clinical history and examination findings, any previous treatment received by the patient, and the relevant imaging, and intra-operative findings. The forms were linked to a Google sheet in a password-protected Google Drive of the researcher. This form was imported into SPSS version 20 for analysis.

Bias: the minimum sample size of 74 was exceeded to reduce the bias of inadequate representation of some diagnoses.

Study size: using the minimum sample size as described by Hajian-Tilaki K. For sample size estimation in diagnostic test studies of biomedical informatics [11]:

Where, n = minimum sample size required; P₁= the sensitivity of test 1 (clinical diagnosis) =0.94 [12]; P₂= the sensitivity of test 2 (imaging) = 0.77. From Fu et al. [5]. P=the average of P₁ and P₂ and Z_α/2 = the standard normal Z values corresponding to the probability of type I error at 95% CI= 1.96; Z_β = the standard normal Z values corresponding to the probability of type II error at 20%=0.842. With a 10% attrition rate gives a minimum sample size of 74.

Quantitative variables: the ages of the participants were grouped to incorporate the adolescent group (13-18), the young adult group (19-25), subgroupings within the adult (26-45) for more detailed analysis, the middle-aged (45-55), and the old. Duration of symptoms were not grouped. The age groups were analyzed using frequency tables and the mode was calculated, same for the duration of symptoms.

Statistical methods: data was analyzed following STARD (Standards for Reporting of Diagnostic Accuracy) guidelines using SPSS software. Categorical variables were compared using Pearson's Chi-square test. Frequency tables were created for the various variables. The referring diagnosis, clinical diagnosis, and imaging diagnosis were matched against the intra-operative diagnosis. The sensitivity, specificity, positive predictive value, and negative predictive value were calculated for each diagnostic method for a particular diagnosis. A p-value of < 0.05 as statistically significant was used for all tests.

Results     

Participants: a total of 100 patients were included in the study. All of them were confirmed eligible. All of them had follow-up data collected and analysis was done for all 100 of them.

Descriptive data: two-thirds (66%) of the participants were male. Almost a third of the participants (28%) were in the 26-35 year group, and that was the most frequent age group. About a third of participants were below 26 years and more than half were below 36 years old, and 5% were above 65 years old (Table 1). Most of the patients, about 99% presented with abdominal pain. Additionally, 83% had vomiting, 44% had nausea, 37% had a fever, 33% had constipation, 32% had anorexia, 29% had abdominal distension and 9% had diarrhea (Figure 1). The most frequent duration of symptoms before presentation was 1 day, which was 18%. Over a third of the patients (39%) presented with a duration of symptoms being more than 3 days, 16% presented within a day, and 14% presented within a week or more after the onset of symptoms (Figure 2). The top causes of acute abdomen were acute uncomplicated appendicitis-30%, complicated appendicitis- 17% and adhesive intestinal obstruction- 10% (Table 2).

Concordance between clinical diagnosis and final/discharge diagnosis: about 38% of the referring diagnosis corresponded to the final diagnosis, 47% did not correspond and in 15% of the cases, there was no referring diagnosis. At the Korle-Bu Teaching Hospital, in 70% of the cases, the clinical diagnosis corresponded to the final diagnosis, and in 29% of the cases, the clinical diagnosis did not correspond to the final diagnosis. In 1%, the clinical diagnosis made was unknown. Fifty-four percent (54%) of the time, the clinical diagnosis was made by a senior resident, 23% by a junior resident, 17% by consultants and senior specialists, 4% by a house officer, and 2% by unknown. Even though junior residents made the right clinical diagnosis 78.3% of the time, consultants 73.3% of the time, senior residents 68.5% of the time, and house officers 50%, the difference in the accuracy of the diagnosis between the different categories of doctors was not statistically significant (Pearson Chi-square of 8.121 and p-value of 0.427).

Concordance between imaging and final diagnosis: imaging was done in 78% of cases. A total of 47.4% of the imaging was ultrasound alone, 39.7% was X-ray alone, 6.4% was both X-ray and ultrasound, 5.1% was CT scan and 1.2% was ultrasound and CT scan. There was an agreement between the imaging diagnosis and the final diagnosis 66.6% of the time. Out of the 78% of participants who had imaging done, almost half of them (48.7%) had their imaging done at the Korle-Bu Teaching Hospital. Forty-four point nine percent (44.9%) of them had their imaging done at a peripheral facility, 2.6% of them had their imaging done in both a peripheral facility and Korle-Bu Teaching Hospital, and in 3.8% it was unknown where imaging was done. Thirty percent (30%) of the ultrasound scans were done at the Korle Bu Teaching Hospital and 70% were done at peripheral facilities. Seventy-four point two percent (74.2%) of X-rays were done at the Korle Bu Teaching Hospital, 19.4% in a peripheral facility, and 6.5% of sites unknown. Even though sonographers had the highest percentage of accurate diagnosis at 76.2%, followed by radiologists who had 71.4, then radiology residents who had 70%, surgical residents had 64.2, unknown 62.5, surgical consultant 50%, surgical house officer 50%, the difference in the percentage of accurate diagnosis by the various groups above were not statistically significant with Pearson Chi-square value of 26.069 and p-value of 0.092.

Concordance between clinical diagnosis versus various imaging modalities for various causes of acute abdomen: clinical diagnosis had a higher sensitivity, specificity, positive predictive value, negative predictive value, and accuracy than ultrasound scans in the diagnosis of acute and complicated appendicitis. Computed tomography scans had a higher sensitivity, specificity, negative predictive value, and accuracy than clinical diagnosis and ultrasound scans. It however had a lower positive predictive value than both (Table 3). Ultrasound scans had a similar sensitivity, specificity, positive predictive value, negative predictive value, and accuracy to clinical diagnosis in the diagnosis of cholecystitis and liver abscess and these parameters were all 100% (Table 3). Clinical diagnosis had a higher sensitivity, negative predictive value, and accuracy compared to x-ray in the diagnosis of intestinal obstruction. X-rays however had a higher specificity and a similar positive predictive value to clinical examination (Table 3). X-ray was more sensitive, and specific and had a higher positive predictive value and accuracy than clinical examination in the diagnosis of a perforated hollow viscus apart from a perforated appendix. Clinical diagnosis however had a higher negative predictive value (Table 3).

Other analyses

Negative laparotomy rate: nine percent (9%) of patients recovered without surgery and 4% had negative laparotomies due to 1% diabetic ketoacidosis, 2% pelvic inflammatory disease, and 1% pancreatitis. Of the 91% who had surgery, 96.7% were stable and on the ward 24 hours after surgery, while 3.3% were ill or in the ICU and were not back on the ward yet.

Discussion     

Male patients constituted two-thirds of the study population. This finding doesn´t reflect the gender distribution of the populace which is almost 1:1 [11]. This may be because gynecological and obstetric patients were not included in this study since they are managed at the Gynaecological Emergency of the Korle-Bu Teaching Hospital. More than half of the patients were 35 years and below. Only 5% were above 65 years and the majority of patients were between 26-35 years. The ages of participants reflect the age distribution of the populace. Ghana has a relatively young population with a mean age of 24.81 years, a median age of 20.7 years, and a life expectancy of 64.5 years as of 2023 [13]. The study indicated that aside from abdominal pain, the commonest symptoms associated with acute abdomen were vomiting and nausea. Others were fever, constipation, anorexia, abdominal distension, and rarely diarrhea. Since the major causes of the acute abdomen were acute appendicitis and complicated appendicitis, it is understandable that vomiting and nausea which are well-known symptoms of appendicitis were frequently reported. Anorexia which occurs in about 80-85% of appendicitis patients [14] featured in one-third of cases (about the number of appendicitis cases).

Thirty-nine percent (39%) of patients presented to the facility after 72 hours of onset of symptoms. This is in line with the fact that Ghanaians have poor health-seeking behavior and will usually turn to herbal preparations, pharmacies, and religious/faith centers for treatment during illness. In a study to assess the health-seeking behavior among Ghanaian urban residents, 50.6% of respondents sought healthcare days to months after their first experience with ill health, 15.5% sought healthcare only when they were severely ill, and about 34% sought healthcare immediately after they experienced ill health [15]. Also, because Korle-Bu is a tertiary facility, almost all patients had been seen and managed at peripheral facilities before being transferred to Korle-Bu which may account for why most patients presented after 24 hours of onset of symptoms.

The 3 commonest causes of acute abdomen were acute appendicitis 30%, complicated appendicitis 17% and adhesive intestinal obstruction 10%. This is in line with the findings made by Naaeder et al [3] in a study of 370 patients in Accra in which the most common diagnoses were appendicitis (23.5%), non-specific abdominal pain formed 21.4%, and acute intestinal obstruction10.8%. In another study of 26 patients at Agogo Presbyterian Hospital (APH), the most common causes of the acute abdomen were acute appendicitis 9 (34.6%), intestinal obstruction 5 (19.2%), and perforated peptic ulcer disease 2 (7.7%) [16]. In all 3 studies, the commonest cause of acute abdomen was disease of the appendix, and indeed appendicitis is the most common cause of acute abdomen worldwide [17].

Discharge diagnosis correlated more often with the diagnosis made at the emergency room of the Korle-Bu Teaching Hospital (70%) as compared to referring diagnosis (38%). This may be because aside from house officers and medical officers, residents, senior specialists, and consultants with a higher level of knowledge and experience attend to patients at the teaching hospital whereas, at the peripheral facilities, patients are usually attended to by medical officers, physician assistants, and other health professionals. The limitation here is that the level of training and experience of practitioners at the referral facilities was not analyzed for comparison.

Generally, there was a slightly higher level of concordance (70%) between clinical diagnosis and discharge diagnosis, as compared to 66.6% in those who had imaging. This is similar to the study by Ali and Maddu, 2019, who revealed a 95% clinical accuracy in the diagnosis of acute abdomen when compared to ultrasonography which was diagnostic in 74.2% of acute abdomen [7]. Ultrasound was the most imaging modality used. This may be because it's readily available. Another reason may be because it is the recommended first-line imaging in suspected appendicitis, which was the most common suspected diagnosis. Also, the preferred/gold standard imaging modality for certain diagnoses may not have been used, which could affect the sensitivity of imaging in the diagnosis of acute abdomen. The sensitivity and specificity of clinical diagnosis for acute appendicitis, acute appendicitis, and complicated appendicitis, all causes of intestinal obstruction, intestinal obstruction secondary to hernia, cholecystitis, and liver abscess were above 85%. Studies have shown that acute appendicitis can be diagnosed with clinical assessment alone in about 85% of cases, which corresponds to the findings in this study. It is also interesting to note that ultrasound has a much lower sensitivity, specificity, negative and positive predictive value as well as accuracy for complicated appendicitis as compared to clinical assessment. This may suggest that in cases of suspected acute appendicitis, whether simple or complicated, an ultrasound scan may not be needed to confirm the diagnosis because the clinical diagnosis has a higher accuracy for the diagnosis of acute appendicitis than an ultrasound scan.

Computed tomography scan has a sensitivity, specificity, and accuracy of close to 100% in both acute appendicitis and complicated appendicitis and this is in line with literature. This suggests that a CT scan has a higher sensitivity and accuracy for appendicitis and complicated appendicitis as compared to both ultrasound and clinical assessment, therefore it may be more beneficial though more expensive and time-consuming in our setting, to do a CT scan rather than an ultrasound where the diagnosis of appendicitis is in doubt bearing in mind the potential risk of ionizing radiation. The sensitivity, specificity, negative and positive predictive value, and accuracy for liver abscess and acute cholecystitis were 100% for both clinical and radiological assessment. According to Bingener et al. [18], ultrasonography is 90%-95% sensitive for cholelithiasis and acute cholecystitis, and it is 78%-80% specific. The diagnosis of cholecystitis is confirmed by imaging studies. The results of this study indicate that imaging may not be needed for the confirmation of the diagnosis of acute cholecystitis and liver abscess, but this study is limited by the fact that there were only 2 cases of each. For cases of intestinal obstruction, the sensitivity and negative predictive value for clinical assessment were higher than that of imaging (X-ray). The specificity on the other hand was higher for imaging, while both forms of assessment had the same positive predictive value. Clinical assessment produced a higher accuracy compared to X-rays. We may say from the above results that even though the symptoms and signs of intestinal obstruction, i.e. abdominal pain, constipation, vomiting, and abdominal distension when they occur together are highly specific for intestinal obstruction, these are symptoms of a myriad of abdominal pathologies and in a few cases imaging may be needed to refute the diagnosis of intestinal obstruction especially in upper gastrointestinal tract obstruction where constipation and abdominal distension may be absent. However, in most cases imaging may not be needed and interventions shouldn´t be delayed on account of the absence of imaging if outcomes will be adversely affected.

Imaging had sensitivity similar to literature for GI perforation (66.7%) which was close to that of clinical assessment (62.5%) and also both clinical assessment and imaging had similar accuracy of 91.7 and 92.3% respectively. On the other hand, Imaging (X-ray) had a specificity and positive predictive value of 100% which implies that although the gold standard imaging for gastrointestinal perforation is a CT scan with contrast, an X-ray may be sufficient in resource-poor areas such as ours. However, where X-rays are not available, one may rely on clinical assessment since sensitivity and accuracy are similar. The limitation in this study again is that there were only 8 cases of GI perforation and no CT scans were done for comparison. In a prospective observational study of 105 operated cases of non-traumatic acute abdomen in an emergency setting in a tertiary government hospital, the negative laparotomy rates were low, around 2.85% [17]. Even though the sample space for the study above and ours were similar, in this study the negative laparotomy rate was higher (4%), this may be due to the lower accuracy rates for imaging, i.e. 83.8% for appendix-related causes and 88.5% for intestinal obstruction as compared to 89.79% and 94.64% for appendicitis and intestinal obstruction/perforation respectively in the above study.

Limitations: some diagnoses were poorly represented and this limited analysis. Gynecological causes of acute abdomen were excluded from the study since these patients were seen in a different emergency room. Even though clinical diagnosis had a higher correlation with the intra-operative/final diagnosis than imaging diagnosis, this may not be representative of the rural areas where imaging may not be readily available since the Korle-Bu Teaching Hospital may have more experienced surgeons and teachers of surgery which may not be the case in the districts and rural areas. Also, more ultrasound scans and X-rays were done as imaging for the cases and these are not as sensitive as CT scans and MRI scans for the diagnosis of various causes of acute abdomen.

Interpretation: this study means that if imaging is not readily available or too costly for a patient with a history and examination suggestive of causes of acute abdomen like acute appendicitis, complicated appendicitis, intestinal obstruction, cholecystitis, liver abscess, the patient can be managed based on the clinical diagnosis, especially in consultation with a surgical resident or a consultant. If the clinical diagnosis is really in doubt, a CT scan should be done instead of an ultrasound scan or an X-ray, especially for the appendix-related conditions. It also means that in our tertiary facilities and Teaching Hospitals, the art of history and clinical examination may not be a lost art as has been thought recently. This ensures that adequate history-taking and clinical examination skills are passed on to medical students and junior doctors.

Generalisability: within and outside teaching hospitals and in resource-poor settings, adhering to the principles of history taking and examination in acute abdomen as would happen in teaching hospitals can lead to the accurate diagnosis of causes of acute abdomen most of the time. When getting imaging done will lead to further significant delays in the management of patients with acute abdomen which can lead to adverse outcomes, a doctor who has taken a good history and done a thorough examination can be confident in his clinical diagnosis and initiate the necessary management. If the clinical diagnosis is however in doubt, a CT scan may be the imaging of choice.

Conclusion     

In resource-poor settings, obtaining imaging for the diagnosis of the causes of acute abdomen can significantly cause delays in treatment and may affect outcomes. Clinical assessment alone may be enough to make an accurate diagnosis of acute abdomen. The commonest causes of acute abdomen were appendix-related and intestinal obstruction. Most of these cases can be accurately diagnosed without imaging. This is also true for GI perforation, cholecystitis, and liver abscess, however, the sample space for these was few (8,2 and 2 respectively) and more studies need to be carried out for these specific diagnoses. For acute abdomen, findings from ultrasound scans from sonographers and those from outside teaching hospitals could be more reliable than those done within teaching hospitals by radiology residents and radiologists. For most causes of acute abdomen, when in doubt and imaging is needed, CT scan may be the imaging of choice since ultrasound and X-ray had similar outcomes or even worse outcomes in terms of diagnosis as compared to clinical assessment.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Jessica Dei-Asamoa: conceptualization and design, data acquisition and cleaning, analysis and interpretation of data, and drafting/reviewing of the manuscript. Samuel Essoun: conceptualization and design, analysis and interpretation of data, and reviewing and revising of the manuscript. Jessie Yaoteokor Annan: conceptualization and design, data acquisition and cleaning, analysis and interpretation of data, and drafting, reviewing, and revising of the manuscript. Omane Okrah Acheamfour: conceptualization and design, data acquisition, and drafting, reviewing and revising of the manuscript. Edmund Tetteh Nartey: conceptualization and design, interpretation of data, and reviewing and revising of the manuscript. All the authors read and approved the manuscript.

Acknowledgments     

The authors would like to thank Dr. Antoinette Bediako-Bowan and Dr. Richard Dei-Asamoa for suggestions made during the drafting of the manuscript and Dr. Yeboah Adjei Mensah, Dr. Emmanuella Agyeman Tuffour and Dr. Eugene Nana Atuahene-Ampadu who were involved in data acquisition.

Tables and figures     

Table 1: socio-demographic characteristics of the participants

Table 2: causes of acute abdomen at the Korle Bu Teaching Hospital

Table 3: sensitivity, specificity, negative predictive value, positive predictive value, and accuracy of clinical diagnosis and imaging for the various causes of acute abdomen

Figure 1: participants´ symptoms

Figure 2: duration of participants´ symptoms before presentation

References