Article Text
Abstract
Introduction Many patients presenting with abdominal pain to emergency departments (EDs) are discharged without a definitive diagnosis. For these patients, often designated as having non-specific abdominal pain, re-evaluation is often advocated. We aimed to investigate how often re-evaluation changes the diagnosis and clinical management and discern factors that could help identify patients likely to benefit from re-evaluation.
Methods This was a retrospective study conducted in the Netherlands between 1 January 2014 and 31 December 2015 of patients asked to return to the ED after an initial presentation with acute non-traumatic abdominal pain. The primary outcome was a clinically relevant change in treatment (surgery, endoscopy during admission and/or hospitalisation) and diagnosis at ED re-evaluation within 30 hours.
Results During the 2-year study period, 358 ED patients with non-specific abdominal pain were scheduled for re-evaluation. Of these, 14% (11%–18%)) did not present for re-evaluation. Re-evaluation resulted in a clinically relevant change in diagnosis and treatment in, respectively, 21.3% (17%–29%)) and 22.3% (18%–27%)) of the subjects. Of the clinical, biochemical and radiological factors available at the index visit, C reactive protein (CRP) at the index visit predicted a change in treatment (CRP >27 mg/L likelihood ratio (LR)+ 1.69 (1.21–2.36)), while an increase in CRP of >25 mg/L between index and re-evaluation visit (LR+ 2.85 (1.88–4.32)) and the conduct of radiological studies at the re-evaluation visit were associated with changes in treatment (LR+ 3.05 (2.41–3.86)).
Conclusion Re-evaluation within 30 hours for ED patients discharged with non-specific abdominal pain resulted in a clinically relevant change in diagnosis and therapy in almost one-quarter of patients. Elevated CRP at the index visit might assist in correctly identifying patients with a greater likelihood of needing treatment in follow-up, and a low threshold for radiological studies should be considered during re-evaluation.
- abdomen- non trauma
- clincial management
- effectiveness
- emergency care systems, emergency departments
- safety
Statistics from Altmetric.com
- abdomen- non trauma
- clincial management
- effectiveness
- emergency care systems, emergency departments
- safety
Key messages
What is already known on this subject
Many patients presenting with abdominal pain to emergency departments (EDs) are discharged without a definitive diagnosis. For these patients, often designated as having non-specific abdominal pain, re-evaluation after the ED visit is often advocated. However, the evidence supporting these recommendations is limited.
What this study adds
In this 2-year retrospective cohort study, we found that re-evaluation after the ED visit for patients discharged with non-specific abdominal pain results in a change in diagnosis and/or treatment in approximately 22% of patients. C reactive protein might assist in identifying those who require follow-up.
Introduction
Approximately 5%–10% of all patients evaluated in the emergency department (ED) present with acute abdominal pain.1–6 About 25%–38% of these patients are admitted to the hospital for either observation or invasive treatment.4 5 7–10 The remaining patients are discharged from the ED, often without a definitive diagnosis. These patients are said to have non-specific abdominal pain,1–5 7–9 11 which is defined as self-limiting abdominal pain for which no definitive cause can be found and no surgical intervention is required.12 Previous studies showed that many patients diagnosed with non-specific abdominal pain have a benign short-term course.3 11 13 Weiner et al 11 reported that the median duration of abdominal pain after the ED visit was 3 days and that unscheduled ED re-visits were uncommon.
However, in some patients presenting with acute abdominal pain, a definitive diagnosis cannot be made at the time of their initial visit since they present early in the course of their disease, whereas in others a specific diagnosis is missed because of an unusual clinical presentation. Missing occult conditions that require surgery due to an early and/or atypical presentation might result in serious morbidity and even mortality.6–8 14 15 Some clinical guidelines advocate for a follow-up re-evaluation of patients who are discharged from the ED with non-specific abdominal pain.16 However, the evidence underlying these recommendations is limited. Previous studies focused mainly only on paediatric or elderly patients, and many studies were performed before ultrasound and CT were used liberally to confirm the diagnosis at first presentation.3–5 8 A recent study by Toorenvliet et al 1 concluded that standard outpatient re-evaluation is a safe and effective way of improving diagnostic accuracy and helps to direct management for patients not admitted to the hospital after presentation in the ED for acute abdominal pain. From their study, however, it remains unclear which elements (physical exam, laboratory studies and/or radiological studies) should be incorporated in a re-evaluation strategy. Furthermore, it is unknown which patients would benefit most from re-evaluation visits. This information is important to know given that scheduled ED re-evaluation visits potentially contribute to ED overcrowding and increasing healthcare costs.4 5 14
In the present study, we aimed to investigate if re-evaluation of ED patients discharged with non-specific abdominal pain changes their diagnosis and clinical management and which factors available at the initial ED visit could help identify patients who will likely benefit from a re-evaluation.
Methods
Study setting and design
We performed a single-centre retrospective study of patients who were re-evaluated in the Medical Center Leeuwarden ED after initial presentation with acute non-traumatic abdominal pain between 1 January 2014 and 31 December 2015. Medical Center Leeuwarden is a teaching hospital with 27 000 ED visits yearly, of which approximately 2000 are for abdominal pain. Because the ED does not have an observation ward, a post-ED visit re-evaluation strategy is applied routinely for patients with abdominal pain presenting in the ED who do not need to be hospitalised and for whom no definitive diagnosis can be made by the treating physician. Treating physicians in the ED are emergency medicine consultants or residents, or surgical residents in their first through fifth year of specialist training (supervised by surgery consultants). Because this study only involved evaluation of routinely recorded patient data, it was determined to be exempt research by our local institutional review board.
Study population
Patients were eligible for the study if they visited the ED with acute non-traumatic abdominal pain for which no definitive diagnosis could be made during their index visit (ie, had non-specific abdominal pain) and for whom a post-ED re-evaluation visit was scheduled within 30 hours. Excluded were patients who declined admission, were provided a scheduled revisit >30 hours after their index visit (routine outpatient clinic follow-up) or those who did not return for re-evaluation.
In all patients, phlebotomised blood samples were obtained per protocol at the index visit for a complete blood count, C reactive protein (CRP), liver enzymes, electrolytes, urea and creatinine. A pregnancy test was performed for women of reproductive age. Additional laboratory studies and radiological studies were ordered at the discretion of the treating physician. All patients were discharged after their index visit with the instruction to present to the same ED for re-evaluation the following morning. At the re-evaluation visit, ordering of any diagnostic studies was at the discretion of the treating physician.
Data acquisition
The following data were retrieved from the hospital’s electronic patient health records (Mirador HWS 5.71.0.2 CSC ActiveX Framework, iSOFT, the Netherlands) for the index and re-evaluation visit: gender, age, date and time of presentation, type of referral to the ED (general practitioner, ambulance or self-referred), relevant medical history (none, abdominal surgery, vascular disease, diabetes mellitus, autoimmune disease, gynaecological, urological, gastrointestinal or psychiatric history), medication use, pain characteristics (severity as expressed on the Numeric Rating Scale (NRS) and duration), physical examination findings (temperature, pulse rate, blood pressure, localisation of pain, presence of rebound tenderness and/or involuntary guarding), laboratory findings (white cell count and CRP) and (if performed) the urine dipstick test results for nitrite, haemoglobin and leucocytes, and results of the radiological studies performed. For the purpose of this study, localised pain was defined as pain in one abdominal quadrant or with a point of maximal intensity.
In addition, the treatment initiated and the diagnosis most likely to explain the non-specific abdominal pain according to the treating physician at both the index and the follow-up visit were retrieved from the electronic health records. Fourteen diagnostic categories were created a priori for this purpose and subdivided into ‘major’ and ‘minor’ based on the likelihood of harm to the patient in case of misdiagnosis (figure 1).
Data abstraction and diagnosis classification were performed by one investigator (AEB), with cross-checking by a second (CWC or EtA). When disagreement existed regarding assignment of the most likely diagnosis to one of the 14 categories, a third researcher was asked to provide a final classification. The data abstractors were not blinded to the study protocol or the rest of the health record data. Missing data are reported in the results. No data imputation algorithms were used.
Outcomes of interest
The primary outcome of interest was a clinically relevant change in treatment resulting from the re-evaluation visit. A clinically relevant change was defined as hospitalisation (either for observation only or for intravenous antibiotics), and/or surgery, and/or endoscopy during admission.
Secondary outcomes of interest were: any change in diagnosis between the initial and re-evaluation visits, and any clinically relevant changes in diagnosis. A change in diagnosis was considered clinically relevant based on treatment consequences such as when a diagnosis changed from a major category to another major category or when the diagnosis changed from a minor to a major category.
Statistical analysis
Percentage change in diagnosis and clinically relevant treatment were calculated to determine concordance in diagnosis, diagnosis classification as major or minor and treatment rendered for the index and follow-up visits. Sensitivity analysis for the subgroup of patients in whom diagnostic radiographic studies were performed was carried out separately, and comparisons across groups were made using the χ2 test. Spearmann’s correlation coefficients were calculated to evaluate the association of clinical and biochemical factors with the primary and secondary outcomes. Multivariable regression analyses that included those factors with a significant association with an r>0.2 was performed to determine which factors were independently related to the primary outcome measure. Optimal cut-off values (sensitivity + (specificity−1) maximal) for distinguishing subjects with and without the primary endpoint (ie, a change in treatment) were determined for continuous variables with an r>0.2 using receiver operating curve (ROC) statistics under the condition of equal ‘cost’ of misclassification of cases and non-cases (online supplementary document II). Likelihood ratios, sensitivity and specificity are reported for these optimal cut-off values. A p value <0.05 was regarded as statistically significant. All statistical analyses were conducted using SPSS V.23.0 for Windows statistical package and the Vassarstats web-based statistical program.
Supplementary file 2
Results
Study population
During the study period, 358 of ED patients with non-specific abdominal pain were scheduled for an evaluation follow-up visit; this group constituted 18% of all abdominal pain patients. Of the 358 patients, 51 (14%) did not present for re-evaluation. Two patients were excluded from the analysis since they were scheduled for a follow-up visit more than 30 hours after their initial ED visit. No patients scheduled for a follow-up visit presented earlier than scheduled (figure 2). Table 1 provides the demographic and clinical characteristics of the study population (n=305) stratified by age groups. The majority of our population was between 18 years and 65 years old, female, had a short duration of abdominal complaints and was referred to the ED by their general practitioner. On physical exam, 26.2% had rebound pain and 77.4% had localised pain, whereas laboratory studies were generally unremarkable.
- Download figure
- Open in new tab
- Download powerpoint
Changes in clinical and biochemical parameters at the re-evaluation visit
Changes in clinical and biochemical parameters between the index and re-evaluation visit are depicted in table 2. Overall, there were fewer abnormalities on physical exam, patients had less pain, a lower WCC, but there was a higher median CRP. No new pregnancies were diagnosed on either visit.
Diagnostic agreement between index and re-evaluation visit
The most likely cause of non-specific abdominal pain according to the treating physician changed in 52.1% of patients between the index and re-evaluation visit. The most likely diagnosis at the index visit was qualified as ‘major’ in 172 (56.4%) of the patients and ‘minor’ in 133 (43.6%) of the patients. At the re-evaluation visit, there were fewer major diagnoses 146 (47.9%) and more minor diagnoses 159 (52.1%) (p=0.03). One hundred and fourteen patients initially diagnosed with a diagnose qualified as ‘major’, had the same (n=81) or another (n=33) major diagnosis at the time of their re-evaluation visit, whereas 58 patients with a major diagnosis had a minor diagnosis at the time of follow-up. Of the 133 patients with an initial diagnosis qualified as ‘minor’, 62 had the same and 39 another minor diagnosis at their re-evaluation visit, whereas 32 subjects had a major diagnosis at the time of their re-evaluation visit. A clinically relevant change in diagnosis between both visits occurred in 65 (33+32) subjects (21.3% (17%–29%)%)). See figure 3 for an overview of all changes in diagnoses.
Therapeutic agreement between index and re-evaluation visit
A clinically relevant change in treatment between the index and re-evaluation visit occurred in 68 (22.3% (18%–27%)%)) of patients. Of these, 29 (9.5% (8%–13%)%)) were admitted for surgery, 4 (1.3% (0.6%–3%)%)) for endoscopy and 35 (11.5% (9%–13%)%) for clinical observation (7 for intravenous antibiotics and the remaining 28 (9.2% (6%–12%)%) for oral or no antibiotics). The 28 patients admitted for oral or no antibiotics consisted of 2 patients with cholecystitis, 6with diverticulitis, 12 with constipation, gastro-enteritis and irritable bowel syndrome, six with urinary tract or gynaecological disease and two with pain after surgery in the preceding 2 weeks.
Clinically relevant treatment changes occurred irrespective of whether radiological studies were performed at the index visit or not: when imaging studies were performed at the index visit (n=127), therapy changed in 22 (17%) patients, whereas therapy changed in 43 of the 178 subjects (24%) in whom imaging studies were not performed (p=0.23).
However, clinically relevant changes in treatment occurred more frequently when radiological studies were performed at the re-evaluation visit compared with when they were not (47% vs 6%, p<0.001). This was irrespective of whether imaging studies had been performed at the index visit or not (56% vs 47%%, p=0.26; table 3).
Univariate correlation coefficients for clinical, biochemical and radiological findings with the occurrence of a clinically relevant change in treatment are given in online supplementary file I. In the multivariate analysis, the CRP measured at the index visit and an increase in CRP between index and re-evaluation visit remained independently associated with a change in treatment rendered at follow-up (table 4) as well as performance of radiological studies at the re-evaluation visit. The LR+ for CRP at the index visit >27 mg/L was 1.69 (1.21–2.36), and the LR+ for an increase of CRP >25 mg/L at the re-evaluation was 2.85 (1.88–4.32)) (table 5). Doing imaging studies at the re-evaluation visit had a positive likelihood ratio (LR+) of 3.05 (2.41–3.86) for a significant change in therapy.
Supplementary file 1
Discussion
We found that the re-evaluation of ED patients with non-specific abdominal pain resulted in a clinically relevant change in treatment in almost a quarter of patients and a change in diagnosis in approximately half of patients.
Our study demonstrates that it is difficult to discern which patients at the index ED visit would benefit most from re-evaluation using clinical, laboratory and radiological findings available at the time of the index visit. Previously suggested predictors of an adverse course and/or the need for surgery in patients presenting with non-traumatic abdominal pain to the ED, including pain characteristics,8 16 physical examination findings,6 17 temperature16 and radiological studies performed,10 were not helpful as risk-stratifying tools. From the factors we studied from the index visit, only an elevated CRP at the initial visit was modestly predictive of a change in treatment at the re-evaluation visit (CRP >27 mg/L LR+ 1.69 (1.21–2.36)).
The study findings could well be explained by the clinical course of non-specific abdominal pain. Many patients presented early in the course of their condition; 62.6% had less than 48 hours of pain. Physical examination and laboratory and radiological studies may not yet have been diagnostic during that time. As time progresses, clinical signs develop that might provide clues to the diagnosis. For example, abdominal pain can evolve from visceral and non-specific in nature18 to more somatic and localised, which occurred in 29 patients (9.5%) of our study population. With disease progression, the chance of finding a correct diagnosis with radiological imaging also increases.
Disease progression might explain the low level of agreement between the index and re-evaluation visit for the most likely diagnosis, although other factors may play a role. First, radiological imaging studies were performed during only 41.6% of index visits, including a limited number of CT scans, which generally have a higher diagnostic accuracy than ultrasound for non-specific abdominal pain.5 6 14 17 However, sensitivity analysis for the subgroup of patients in whom diagnostic radiographic studies were performed did not increase concordance between the index and follow-up visit diagnoses. Second, patients usually were evaluated by different physicians on their follow-up visit. Finally, differences in diagnosis between the index and follow-up visit could be due to incorrect diagnoses, perhaps by physicians wanting to be cautious about radiological imaging at the index visit, absence of microscopic urine testing and the potential for incorrect interpretation of radiological imaging at the index visit.
Previous studies have reported the merit of re-evaluation follow-up visits among paediatric patients3 6 and the elderly.2 3 6 8 14–16 This has been attributed to uniqueness in physiology, the presence of comorbidities, the use of immunosuppressant medications and difficulties in obtaining a reliable history in these populations. This study demonstrates that even in a population consisting mainly of adults aged 18–65 years, re-evaluation results in a clinically relevant change in diagnosis and therapy in almost one-quarter of patients.
Our study suggests a potential role for measuring CRP at the index and follow-up visit for patients with non-specific abdominal pain. The findings also suggest a role for radiological imaging at the follow-up visit, since performing imaging studies at re-evaluation was associated with clinically relevant treatment changes. However, these findings should be interpreted with caution, since the reported association does not imply causality. It is likely that radiological imaging was performed selectively among patients with a higher a priori likelihood of a clinically relevant change in treatment, although this is not reflected by the diagnostic accuracy indices as found in our study: LR+ for a significant change in treatment was 3.05 (2.41–3.86) when imaging studies were performed at the re-evaluation visit. This increased only marginally to 3.09 (2.12–4.50) when only subjects with a higher a priori change of a clinically relevant treatment change (as reflected by an increase in CRP ≥25 mg/L between index and re-evaluation visit) were considered. Furthermore, relevant treatment changes (particularly the need for surgery) usually only occur when there is sufficient evidence to support a new diagnosis, and imaging studies contribute to this evidence. So, in summary, where a more conservative approach regarding radiological studies is justified at the index visit, radiological imaging at the re-evaluation visit should be considered with a low threshold.
Limitations
Our study has several limitations, some inherent to the retrospective design. First, data were incomplete especially regarding several clinical parameters. Vital parameters are measured per protocol in all patients; however, in a significant number of them, these were not retrievable from the electronic health record. Furthermore, laboratory studies were not always ordered. We recognise that this absence might have influenced our outcome, since we might have missed potential predictors of a change in diagnosis and/or treatment. In addition, 14% of patients had to be excluded from the analysis, because they did not present for their re-evaluation visit. Although similar percentages of loss to follow-up have been reported in previous studies,1 8 this occurrence might have resulted in ascertainment bias. Furthermore, we do not know why they did not adhere to their appointment. Because our ED is situated in a geographical isolated location (no other ED within 30 km), it is unlikely that they visited another hospital for their complaints. Furthermore, although our primary endpoint of interest (significant change in treatment) could easily be extracted from the health records, our secondary outcomes of interest (change in diagnosis) was vulnerable to interpretation bias. Although we aimed to limit interpretation bias by constructing diagnosis categories and having a second and third data interpreter to increase fidelity, we cannot exclude that this has influenced our results. We also could not establish, based on the data available in this study, if a re-evaluation strategy is safe, as compared with in-hospital observation. We also cannot determine if re-evaluation might have resulted in a delay in diagnosis for some patients.
Finally, when interpreting the results of our study, it is important to realise that this study was conducted in a setting with a well-developed primary care system. Most patients were referred to the ED by their general practitioner and as such had a higher likelihood of having a potential diagnosis warranting hospitalisation or surgery. Other settings with different healthcare systems might have observed dissimilar results.
Conclusion
Re-evaluation within 30 hours of ED patients discharged with the diagnosis of non-specific abdominal pain results in a clinically relevant change in diagnosis and therapy in almost one-quarter of patients. Elevated CRP at the index visit might assist in correctly identifying patients with a greater likelihood of needing treatment in follow-up, and radiological studies should be considered with a low threshold during re-evaluation.
Abstract translation
References
Footnotes
Contributors All authors fulfilled the ICMJE criteria for authorship. EtA conceived the study. AEB, CWC and EtA acquired and interpreted the data. AEB, EtA and ME drafted the manuscript. All authors revised the manuscript critically and gave final approval to submission of the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Ethics approval Regionale toetsingscommissie patientgebonden onderzoek (RTPO) Leeuwarden, protocol number 1015.
Provenance and peer review Not commissioned; externally peer reviewed.