Portal de Periódicos da CAPES

A Simple Clinical Decision Rule To Rule Out Appendicitis In Patients With Nondiagnostic Ultrasound Results

2014; Wiley; Volume: 21; Issue: 5 Linguagem: Inglês

10.1111/acem.12374

1553-2712

Marjolein M. N. Leeuwenburgh, Hein B.A.C. Stockmann, Wim H. Bouma, Alexander P. J. Houdijk, Matthijs F. Verhagen, Bart C. Vrouenraets, Lodewijk P. J. Cobben, Patrick M. Bossuyt, Jaap Stoker, Marja A. Boermeester,

Diverticular Disease and Complications

The objective was to identify a set of clinical features that can rule out appendicitis in patients with suspected acute appendicitis and nondiagnostic ultrasound (US) results, allowing safe discharge and next-day reevaluation without initial computed tomography (CT) or magnetic resonance imaging (MRI). Data on clinical and US evaluation, including a number of prespecified variables potentially associated with acute appendicitis, were prospectively collected in two diagnostic accuracy studies of imaging. These studies included patients with suspected appendicitis seen in the emergency department (ED). For development and validation of the clinical decision rule (CDR), only patients with inconclusive or negative US results were included. There were 199 (of 422) patients in the development cohorts and 120 (of 211) patients in the validation cohort. Logistic regression analysis was used for data from patients with inconclusive or negative US results, and profiles were created of all possible combinations of predictors retained in the multivariable model. A final diagnosis was assigned by an expert panel based on perioperative data, histopathology, and clinical follow-up of at least 3 months. The CDR selected patients after negative or inconclusive US for discharge and next-day reevaluation without initial CT or MRI if fewer than two of the following predictors were present: male sex, migration of pain to the right lower quadrant, vomiting, and white blood cell (WBC) count higher than 12.0 × 109/L. Applying the CDR in the development set selected 126 of 199 (63%) patients with negative or inconclusive US results for discharge without further imaging. This rule reduced the probability of appendicitis from 26% (51 of 199) in the total group of patients with negative or inconclusive US results to 12% (15 of 126) in the group that would be discharged based on the rule (p = 0.001). In the validation set (n = 120), the decision rule selected 72 (60%) patients for discharge and next-day reevaluation and reduced the probability of appendicitis from 20% (24 of 120) in the total group to 6% (4 of 72) in the patients selected on the rule (p = 0.001). The negative predictive value of the decision rule in the validation set was 94% (95% confidence interval [CI] = 87% to 98%). In comparison, the negative predictive value of CT in the same group was 99% (95% CI = 93% to 100%, p = 0.14), and that of MRI was 99% (95% CI = 94% to 100%, p = 0.12). Alternative decision rules based on combinations of the present decision rule with C-reactive protein (CRP) results did not improve selection. This newly developed CDR significantly reduces the probability of appendicitis in a large subgroup of patients with negative or inconclusive US results. These patients can be safely discharged for outpatient reevaluation without further initial imaging if proper follow-up is available. This could assist in lowering the number of ED imaging investigations in patients with suspected appendicitis. Identificar un conjunto de hallazgos clínicos que pueda descartar una apendicitis en pacientes con sospecha de apendicitis aguda y resultados ecográficos no diagnósticos, de manera que permitan dar de alta de forma segura y reevaluarlos al día siguiente sin una tomografía computarizada (TC) o resonancia magnética (RM) inicial. Se recogieron de forma prospectiva datos sobre la evaluación clínica y la ecografía, que incluía un número de variables prestablecidas potencialmente asociadas con apendicitis aguda, en dos estudios de precisión de diagnóstico por imagen. Estos estudios incluyeron a pacientes con sospecha de apendicitis visitados en el servicio de urgencias (SU). Se incluyeron sólo los pacientes con resultados ecográficos negativos o no concluyentes para el desarrollo y la validación de la regla de decisión clínica (RDC). Hubo 199 pacientes (de 422) en la cohorte de desarrollo, y 120 pacientes (de 211) en la cohorte de validación. Se utilizó un análisis de regresión logística para los datos de pacientes con resultados ecográficos negativos o no concluyentes, y se crearon los perfiles de todas las posibles combinaciones de factores predictivos introducidos en el modelo multivariable. Se asignó un diagnóstico final por un panel de expertos en base a los datos de la cirugía e histopatológicos y del seguimiento clínico durante al menos tres meses. La RDC seleccionó pacientes tras una ecografía negativa o no concluyente para dar de alta y reevaluar al día siguiente sin una TC o una RM inicial si presentaban al menos dos de los siguientes factores predictivos: sexo masculino, migración del dolor al cuadrante inferior derecho, vómitos y recuento de leucocitos por encima de 12,0 × 109/L. Aplicando la RDC en la cohorte de desarrollo, se seleccionó a 126 de 199 (63%) pacientes con resultados ecográficos negativos o no concluyentes para dar de alta sin más pruebas de imagen. Esta regla redujo la probabilidad de apendicitis de un 26% (51/199) en el grupo total de pacientes con resultados ecográficos negativos o no concluyentes, a un 12% (15/126) en el grupo que se habría dado de alta basándose en la regla (p = 0,001). En la cohorte de validación (n = 120), la regla de decisión seleccionó a 72 (60%) pacientes para dar de alta y reevaluar al día siguiente, y redujo la probabilidad de apendicitis de un 20% (24/120) en el grupo total a un 6% (4/72) de los pacientes seleccionados basándose en la regla (p=0,001). El valor predictivo negativo de la regla de decisión en la cohorte de validación fue de un 94% (intervalo de confianza [IC] 95% = 87% a 98%). En comparación, el valor predictivo negativo de la TC en el mismo grupo fue de un 99% (IC 95% = 93% a 100%, p = 0,14), y el de la RM fue de un 99% (IC 95% = 94% a 100%, p = 0,12). Las reglas de decisión alternativas basadas en las combinaciones de la presente RDC inicial junto con resultados de proteína C-reactiva no mejoraron la selección. Esta RDC recientemente desarrollada reduce significativamente la probabilidad de apendicitis en un gran subgrupo de pacientes con resultado ecográfico negativo o no concluyente. Estos pacientes pueden ser dados de alta de forma segura para reevaluar de forma ambulatoria sin más pruebas de imagen iniciales si se dispone de un correcto seguimiento. Esto podría ayudar a reducir el número de pruebas de imagen en el SU en pacientes con sospecha de apendicitis. CME Editor: Corey Heitz, MD Authors: Marjolein M. N. Leeuwenburgh, MD, PhD, Hein B. A. C. Stockmann, MD, PhD, Wim H. Bouma, MD, PhD, Alexander P. J. Houdijk, MD, PhD, Matthijs F. Verhagen, MD, Bart Vrouenraets, MD, PhD, Lodewijk P. J. Cobben, MD, PhD, Patrick M. M. Bossuyt, PhD, Jaap Stoker, MD, PhD, and Marja A. Boermeester, MD, PhD, on behalf of the OPTIMAP Study Group* Article Title: A Simple Clinical Decision Rule To Rule Out Appendicitis In Patients With Nondiagnostic Ultrasound Results If you wish to receive free CME credit for this activity, please refer to the website: http://www.wileyhealthlearning.com/aem. Accreditation and Designation Statement: Blackwell Futura Media Services designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM. Physicians should only claim credit commensurate with the extent of their participation in the activity. Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. Educational Objectives After completing this exercise the participant will be better able to describe the development, validation, and clinical utility for a decision rule regarding the management of appendicitis in the setting of a negative or nondiagnostic ultrasound. Activity Disclosures No commercial support has been accepted related to the development or publication of this activity. Faculty Disclosures: CME Editor: Corey Heitz, MD, has no relevant financial relationships to disclose. Authors: Marjolein M. N. Leeuwenburgh, MD, PhD, discloses institutional research support from the Dutch Organization for Health Research and Development, Health Care Efficiency Research program (Zon Mw). Marja A. Boermeester, MD, PhD discloses institutional research support from the Dutch Organization for Health Research and Development, Health Care Efficiency Research program (Zon Mw), as well as grants to her institution from Abbott, Baxter, Ipsen, LifeCell, and GSK. Jaap Stoker, MD, PhD, discloses institutional research support from the Dutch Organization for Health Research and Development, Health Care Efficiency Research program (Zon Mw) and consulting for Robarts Clinical Trials. Hein B. A. C. Stockmann, MD, PhD; Wim H. Bouma, MD, PhD; Alexander P. J. Houdijk, MD, PhD; Matthijs F. Verhagen, MD; Bart Vrouenraets, MD, PhD; Lodewijk P. J. Cobben, MD, PhD; and Patrick M. M. Bossuyt, PhD, have no relevant financial relationships to disclose. This manuscript underwent peer review in line with the standards of editorial integrity and publication ethics maintained by Academic Emergency Medicine. The peer reviewers have no relevant financial relationships. The peer review process for Academic Emergency Medicine is double-blinded. As such, the identities of the reviewers are not disclosed in line with the standard accepted practices of medical journal peer review. Conflicts of interest have been identified and resolved in accordance with Blackwell Futura Media Services's Policy on Activity Disclosure and Conflict of Interest. Instructions on Receiving Free CME Credit For information on applicability and acceptance of CME credit for this activity, please consult your professional licensing board. This activity is designed to be completed within an hour; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period, which is up to two years from initial publication. Follow these steps to earn credit: This activity will be available for CME credit for twelve months following its publication date. At that time, it will be reviewed and potentially updated and extended for an additional twelve months. The introduction of imaging has reduced the negative appendectomy rate in patients suspected of having appendicitis,1, 2 especially in women below the age of 45 years.3, 4 In this group diagnostic imaging is desirable because urologic or gynecologic pathology often mimics appendicitis.5 However, some clinicians remain hesitant to add diagnostic imaging to clinical evaluation of all patients with suspected appendicitis.6 Combinations of clinical features—including pain migration, abdominal rigidity, and elevated inflammatory parameters—have a high predictive value for appendicitis,7, 8 but they are present in only a small proportion of patients suspected of having appendicitis.9 This makes the clinical diagnosis of appendicitis difficult in the majority of cases and emphasizes the need for imaging in patients suspected of having appendicitis. Several studies have shown that the routine use of imaging has a positive effect on patient outcomes, with two randomized controlled trials reporting substantially lower negative appendectomy rates with routine imaging in all patients, compared to selective use of imaging based on clinical judgment.10, 11 The optimal imaging strategy for patients suspected of having appendicitis has been debated in recent years. Computed tomography (CT) has been shown to be more accurate than ultrasound (US), but its ionizing radiation exposure is associated with the induction of cancer. Although CT protocols have been developed with lower doses of radiation,12 alternatives for further reduction of radiation without compromising diagnostic accuracy are being explored. A strategy of initial US with additional CT in case of negative or inconclusive US results has been demonstrated to result in a higher sensitivity in detecting urgent conditions in patients with acute abdominal pain compared to a CT-only strategy.13 In such a protocol, only 49% of patients would undergo both US and CT. This conditional CT strategy is now standard care in most hospitals in the Netherlands.14 Recent studies showed that magnetic resonance imaging (MRI) could accurately replace CT in such a conditional imaging protocol, completely avoiding the radiation risks of CT in children,15, 16 pregnant patients,17 and the general population.18 The main disadvantage of a conditional strategy is that some patients have to undergo two examinations, resulting in higher imaging costs and longer waiting times. We aimed to identify a set of clinical features that can rule out appendicitis in patients suspected of having appendicitis who have inconclusive or negative US findings. Such patients would not need additional CT or MRI after US in a conditional imaging strategy. Implementation of a clinical decision rule (CDR) based on these features could reduce the number of imaging examinations without comprising patient safety. This was a multicenter, prospective study to develop a simple CDR based on data collected in unselected patients with suspected acute appendicitis (the development data set) and to validate it in a different set of patients with suspected acute appendicitis (the validation data set). The medical ethics committee of the initiating center approved the study protocols of both studies prior to initiation. All included patients gave written informed consent. Data for the development data set of the CDR were collected to investigate the added value of plain supine abdominal radiography, upright chest radiography, US, and CT after clinical assessment in adult patients with acute abdominal pain in the ED. This OPTIMA study is reported in detail elsewhere.13 Between May 2005 and December 2006, a total of 942 adult patients with abdominal complaints for more than 2 hours and less than 5 days were identified at the emergency departments (ED) in one of six participating hospitals: two university hospitals (Academic Medical Center Amsterdam and University Medical Center Utrecht) and four large teaching hospitals (Antonius Hospital Nieuwegein, Gelre Hospital Apeldoorn, Tergooi Hospital Hilversum, and Onze Lieve Vrouwe Gasthuis Amsterdam) in the Netherlands. All patients in this cohort underwent both US and CT. Of these, 422 had suspected acute appendicitis and contributed data to the development set (46% males, median age = 37 years, interquartile range [IQR] = 27 to 50 years). Data for the validation data set of the CDR were collected in another multicenter, prospective diagnostic accuracy study that compared imaging strategies with US and conditional CT after inconclusive or negative US and MRI in patients with suspected appendicitis. In this cohort all patients received additional imaging after inconclusive or negative US results. This OPTIMAP study has also been reported in detail elsewhere.19 Here, 230 adult patients were included between March 2010 and September 2010 in six participating hospitals: one university hospital (Academic Medical Center, Amsterdam) and five large teaching hospitals (Medical Centre Alkmaar, Sint Antonius Hospital Nieuwegein, Sint Lucas Andreas Hospital Amsterdam, Gelre Hospital Apeldoorn, and Kennemer Gasthuis Haarlem) in the Netherlands. Of these, 211 patients with suspected acute appendicitis presented with abdominal complaints for more than 2 hours and less than 5 days; their data constituted the validation set (39% males, median age = 35 years, IQR = 24 to 48 years). In the studies that were used for the development set and validation set, the participating hospitals were unable to include patients 24/7 because of logistic implications of the study protocol. The appendicitis rates did not differ significantly between the participating hospitals. For both the development and the validation set, physicians in the ED prospectively collected data with a web-based digital case record form, in a similar fashion. Included patients were clinically evaluated and, thereafter, all findings from medical history (age, sex, duration of complaints, presence of pain in the right lower quadrant [RLQ], migration of pain to RLQ, pain on movement, nausea, vomiting), physical examination (presence of RLQ tenderness, rebound tenderness, rigidity, temperature in °C), and laboratory tests (C-reactive protein [CRP] in mg/L, white blood cell [WBC] count in 109/L) were recorded. All patients underwent abdominal US performed by a staff radiologist or radiology resident, using the graded compression technique.20 Subsequently, radiologists added their US findings to the case record forms. They classified their diagnoses as "positive for appendicitis," "inconclusive for appendicitis" (appendix not visualized or equivocal results), or "negative for appendicitis." No strict criteria for the diagnosis of appendicitis were formulated; the choice for diagnosis was left at the discretion of the radiologist. A final diagnosis was assigned by an expert panel, consisting of two surgeons and one radiologist, after at least 3 months' follow-up. Their assignment was based on all clinical, laboratory, and surgical findings; pathology results; imaging reports (US, CT); and treatment outcome.12, 19 Fourteen clinical variables known to be associated with acute appendicitis based on previous studies9, 21 were evaluated in the development set in patients with inconclusive or negative US. For each variable we estimated the diagnostic odds ratio (OR) for a final diagnosis of acute appendicitis, with corresponding 95% confidence interval (CI), using univariable logistic regression analysis. Higher ORs indicate stronger associations with appendicitis, with an OR of 1 indicating absence of an association. The continuous variables (age, temperature, WBC, CRP) were analyzed as such, without dichotomization. Variables that were associated with appendicitis in univariable analysis (p < 0.20) were included in a multivariable logistic regression model. We then applied backward stepwise elimination of redundant variables, using a 0.05 significance level for exclusion, to arrive at a parsimonious model. The performance of the final multivariable model, after elimination, was characterized by calculating the area under the corresponding receiver operator characteristic curve (AUC). We consider an AUC of 0.80 or higher to indicate good discrimination.22 With the variables associated with acute appendicitis in the final multivariable model, we developed profiles of all possible combinations. These profiles were defined by using one, two, three, or all four of the variables in the final multivariable model. In this process several cutoff values for the retained continuous variables were explored. For each possible combination of the variables in the model, we counted the number of patients with that profile and the number of patients with final diagnoses of appendicitis in it. With these numbers we calculated estimates of sensitivity, specificity, and negative predictive value (NPV) for the corresponding profiles. The ideal profile would identify a subgroup of patients with almost all cases of acute appendicitis. If we were able to identify a subgroup based on such a profile, the remaining patients—those who do not meet the corresponding profile—could be discharged without additional imaging, since that subgroup would not contain any appendicitis cases. We aimed at a rule that would identify a large group of patients with a low risk of appendicitis (less than 10%), that is, a rule with an NPV of at least 90%. By comparing the NPVs of the respective combinations, and the proportions of evaluated patients with that combination, we selected the most promising combination to build a decision rule. For comparison, estimates of the NPV of the decision rule were contrasted to the estimated NPV of CT and MRI in the same set of patients. Because CRP is widely used in daily clinical practice and because some studies have shown a substantial increase in the positive likelihood ratio (LR+) for acute appendicitis in presence of both elevated WBC and elevated CRP-levels,23 an alternative decision rule was explored based on combinations of the developed decision rule with several cutoff values for CRP. We analyzed the decision rule in patients with inconclusive US results and negative US results separately, because some hospitals perform additional imaging only in patients with inconclusive US.24, 25 After finishing development of the CDR in the development data set, it was evaluated in the validation data set. Here also we used only data of patients with inconclusive or negative US. The data of the development set were first compared to those in the validation set, to explore if distributions of age, sex, and clinical variables were similar. P-values were calculated using the chi-square statistic for binary values and the Mann-Whitney U-test statistic for continuous variables. P-values below 0.05 were considered to indicate statistical significance. For each patient, the outcome of the CDR was compared with the final diagnosis, as assigned by the expert panel. In this way estimates of the NPV and likelihood ratios could be calculated with corresponding 95% CI. For comparison, estimates of the NPV of the decision rule were contrasted to the estimated NPV of CT and MRI in the same set of patients. P-values were calculated using the chi-square statistic. All analyses were performed using SPSS version 20.0. In the OPTIMA study, US detected appendicitis in 223 of 422 patients in the development set, leaving 199 patients with inconclusive or negative US findings for the development set. The results of the univariable analyses of 14 factors potentially associated with appendicitis in this group are summarized in Table 1. Six of these clinical predictors met criteria for entry into the multivariable model. After backward elimination, four predictors were retained: sex, migration of pain to the RLQ, vomiting, and WBC count (Table 2). The discriminatory performance of this multivariable model is shown in Figure 1. The AUC is 0.80 (95% CI = 0.72 to 0.87), indicating good discrimination. With combinations of the four predictors retained in the multivariable model (sex, migration of pain to RLQ, vomiting, WBC) we developed profiles for patients in the development set in whom US was negative or inconclusive (199 patients). In Table 3 the number of patients who had appendicitis as final diagnosis is shown per profile. Most patients with appendicitis as final diagnosis (n = 51) had two or more of the four patient characteristics associated with appendicitis: 36 with at least two predictors and 15 patients with fewer than two predictors. Based on the prevalence of appendicitis cases in each of the subgroups corresponding to these profiles, the following decision rule was defined to rule out appendicitis in patients with inconclusive or negative US results (Figure 2): refer for additional imaging if a patient with suspected appendicitis and an inconclusive or negative US has two or more of the patient characteristics: 1) male sex, 2) migration of pain to RLQ, 3) vomiting, or 4) WBC count > 12.0 × 109/L; otherwise, discharge and reevaluate the next day. The patient flow in the development set after application of the decision rule is depicted in Figure 3A. Appendicitis was the final diagnosis in 51 of 199 (26%) patients with inconclusive or negative US results in the development set. The decision rule selected 126 of 199 (63%) patients for discharge and next-day reevaluation and 73 of 199 (37%) for additional imaging. The prevalence of appendicitis was reduced to 15 of 126 (12%) in the "discharge group" versus 36 of 73 (49%) in the "additional imaging group." This means that applying the decision rule reduced the probability of appendicitis from 52 of 199 (26%) in the total group to 15 of 126 (12%) in the subgroup that would be discharged (p < 0.01). The NPV for appendicitis of the decision rule in the development set, consisting of patients with inconclusive or negative US results, was 88% (95% CI = 81% to 93%). The likelihood ratio of a negative outcome of the rule was 0.37 (95% CI = 0.25 to 0.61), and the LR+ of a positive rule was 2.82 (95% CI = 2.03 to 3.93). In comparison, the estimated NPV of CT was 93% (95% CI = 88 to 96) in this group of patients; this was not significantly higher (p = 0.17). The likelihood ratio of a negative CT was 0.22 (95% CI = 0.13 to 0.38), and a likelihood ratio of a positive CT was 7.44 (95% CI = 4.59 to 12.04). We explored alternative decision rules based on combinations of the presently developed decision rule with CRP results. Refer for additional imaging if a patient with suspected appendicitis and an inconclusive or negative US has two or more of the patient characteristics 1) male sex, 2) migration of pain to RLQ, 3) vomiting, and 4) WBC count > 12.0 × 109/L, 5) CRP > 10; otherwise, discharge and reevaluate the next day. This rule selected 77 of 199 (35%) of patients for discharge without imaging in the development set, missing four of 70 (6%) of appendicitis cases. If patients were selected for additional imaging only in presence of three or more of the five predictors, 138 of 199 (69%) patients could be discharged without imaging, missing 17 of 138 (12%) cases of appendicitis. Of the 199 patients of the development data set, 94 had inconclusive US results (35%, 33 of those 94 had appendicitis) and 105 patients had negative US results (17%, 18 of 105 had appendicitis). Applying the decision rule to the 94 patients with inconclusive US results would only select 55 of 94 (59%) of patients for discharge, with nine of 55 (16%) having appendicitis, and 39 of 94 (41%) for additional imaging, with 24 of those 39 (62%) having appendicitis. In the 105 patients with negative US results, the rule would select 71 of 105 (68%) patients for discharge, with six of 71 (8%) having appendicitis, and 34 out of 105 (32%) for additional imaging, with 12 of those 34 (35%) having appendicitis. In summary, the rule reduced the probability of appendicitis from 35% to 16% in patients with inconclusive US (p = 0.01) and from 17% to 8% in patients with negative US (p = 0.10). The NPV of the rule was 84% in patients with inconclusive US and 92% patients with negative US (p = 0.17). In the OPTIMAP study, 91 of 211 patients had positive US results. The remaining 120 patients with inconclusive or negative US results contributed data to the validation set. The development and validation sets were largely comparable with regard to the distribution of sex, clinical variables, and prevalence of appendicitis (60 and 52%, p = 0.06, Table 4). Patients in the development set were older (median age = 37 years vs. 35 years, p = 0.04), and the results of US were more often positive for acute appendicitis in the development set (43% vs. 53%, p = 0.02). The patient flow in the validation set after application of the decision rule is depicted in Figure 3B. In the validation set, 24 of 120 (20%) patients with inconclusive or negative US results had appendicitis as their final diagnoses. Applying the CDR would discharge 72 patients and would send 48 patients for additional imaging. The rule reduced the probability of appendicitis from 24 of 120 (20%) in the total group to four of 72 (6%) in the subgroup that would be discharged (p < 0.01). In the additional imaging group, the probability of appendicitis was 20 of 48 (42%). The NPV of the decision rule in the validation set was 94% (95% CI = 87% to 98%). The likelihood ratio of a negative rule was 0.24 (95% CI = 0.10 to 0.58), and the LR+ was 2.86 (95% CI = 1.99 to 4.09). In comparison, the NPV of CT was 99% (95% CI = 93% to 100%, p = 0.14), and that of MRI was 99% (95% CI = 94% to 100%, p = 0.12). The likelihood ratio of a negative CT was 0.05 (95% CI = 0.01 to 0.31), and that of a negative MRI was 0.04 (95% CI = 0.01 to 0.30); a positive CT had a likelihood ratio of 25.83 (95% CI = 8.48 to 78.67), and that of a positive MRI was 14.38 (95% CI = 6.61 to 31.28). The decision rule in combination with CRP (imaging in presence of two or more predictors) selected 32 of 120 (27%) patients for discharge and next-day reevaluation and reduced the probability of appendicitis from 24 of 120 (20%) to one of 32 (3%, p = 0.02). If patients were selected for additional imaging in the presence of three or more of the five predictors, the rule selected 85 of 120 (71%) of patients for discharge and reduced the probability of appendicitis from 24 of 120 (20%) in the total group to 10 of 85 (12%, p = 0.12) in the selected subgroup. We developed a simple CDR that, in patients with suspected appendicitis and nondiagnostic US, can select a large subgroup of patients for discharge (60%) with a low probability of appendicitis (12%) and initially no need for further imaging. An appendicitis rate of 12% still seems high and may not be appropriate in populations who do not have access to repeat evaluation the following day. This rule used only four clinical variables: male sex, migration of abdominal pain to the RLQ, vomiting, and WBC count higher than 12.0 × 109/L. The addition of CRP did not improve results of the decision rule. The decision rule substantially reduced the probability of appendicitis in patients with inconclusive US and in patients with negative US, although the NPV of the rule was higher in patients with a negative US. In both the development data set and the validation data set, the application of the decision rule identified a large subgroup with a substantially and significantly lower probability of appendicitis. The NPV of the decision rule was similar to that of CT or MRI. The reduced probability of appendicitis in the subgroup that was selected for discharge without additional imaging is probably not sufficiently low for safe discharge; therefore, reassessment of these patients in the outpatient setting the following day permits for additional evaluation. In the past, only clinical evaluation was used to evaluate patients with suspected appendicitis. The threshold to refer for an appendectomy was low, since a missed appendicitis, and the associated perforation risk, was considered to be more harmful than a negative appendectomy. As a consequence, negative appendectomy rates (false-positives) up to 40% have been reported.23, 26, 27 Approximately 12% of patients with appendicitis are missed when clinical evaluation is not supported by imaging.28 Missed diagnoses result in treatment delay and an increased risk of perforation. Several scoring systems have been developed to identify patients suspected of having acute appendicitis based on clinical features. The most commonly known scoring systems for acute appendicitis are the Alvarado score and the more recent appendicitis inflammatory response score.21 The latter was based on features incorporated in the Alvarado score, but it also includes CRP as a new variable. The authors claim that their scoring system can correctly classify the majority of patients into those who can be sent home for outpatient follow-up, and those to be operated on immediately, leaving a smaller number of patients with indeterminate results who need diagnostic imaging. Although these scores have not yet gained wide acceptance in clinical practice, an external validation of the appendicitis inflammatory response score showed promising results.29 A recent meta-analysis also showed that when both an elevated WBC and an elevated CRP level are present, there is a substantial increase in the LR+ for acute appendicitis.23 For this reason, we explored CRP in combination with the developed decision model. In our analysis, CRP was not associated with acute appendicitis in univariable and multivariable logistic regression. Furthermore, the addition of CRP to our rule with four predictors did not increase its performance. These results suggest that CRP is not a strong additional predictor in patients suspected of appendicitis who have negative or inconclusive US results, compared to sex, migration of pain to the RLQ, vomiting, or elevated WBC count. Some hospitals only perform additional imaging in patients with inconclusive US results and not in patients with negative results.24, 25, 30 This strategy could also lead to a reduction of additional imaging compared to a protocol with standard CT after negative and inconclusive US, which is recommended based on the results of the OPTIMA study.12 In our data set such a strategy would select 105 of 199 (53%) patients with negative US for discharge, resulting in 18 of 105 (17%) missed cases of appendicitis. In comparison, the decision rule selected 126 of 199 (63%) of patients for discharge with a 12% (15/126) probability of appendicitis. Although the positive predictive value of US seems as high as that of CT, a negative US has been shown to have a significantly higher proportion of false-negative diagnoses, compared to CT.31 It is known that the accuracy of US is more dependent on the skills and experience of the examiner, and varying NPVs have been reported (46% to 97%).32 In many cases the appendix cannot be visualized during US examination, and the absence of any indirect signs of appendicitis cannot be relied on. A recent study showed that absence of transducer tenderness in the RLQ and hypertrophy of the peritoneal fat has a NPV of only 66%.33 Therefore, US only seems not suitable to exclude appendicitis in patients who have strong clinical suspicion of appendicitis based on clinical signs and symptoms. The development data set and the validation data set were not selected randomly from a single cohort, but extracted from two different imaging studies, performed in two different time frames.12, 17 As a likely consequence, US results were more often positive in the development set. This may be responsible for some of the differences in performance of the decision rule between the development set and the validation set. Our CDR classifies patients with inconclusive or negative US results in two subgroups (discharge and next-day reevaluation and additional imaging). Such an approach may be too coarse for the subtleties of clinical care in daily practice. It must be emphasized that the intent of the decision rule presented here is not to establish a primary diagnosis of appendicitis, but to assist decision-making about further imaging when there is uncertainty after the US. The rule was developed in patients with suspected acute appendicitis, and evaluated based on its performance in changing the probability of appendicitis. We did not take into account the probability of alternative urgent diagnoses in the development of this prediction rule. We acknowledge that some patients with suspected appendicitis have other urgent diagnoses that need immediate treatment, such as diverticulitis, cholecystitis, or urgent gynecologic disorders. Withholding these patients from additional imaging could delay diagnosis and treatment resulting in increased morbidity. In our study group, a relatively large percentage of patients had inconclusive US: 22% in the first study, used for extracting the development set, and 47% in the second, used for the validation. This could be caused by the fact that the radiologists knew they were participating in a diagnostic accuracy study and did not want to miss diagnoses by classifying uncertain negative cases as negative in the case record form. However, in the designs of both studies, an inconclusive or a negative US would always be followed by CT. The choice between a negative or inconclusive result did not have clinical implications for the participating patients—in case of a negative US the patient would still undergo additional CT. The decision rule presented here could assist in lowering the number of imaging investigations in patients with suspected appendicitis. Before final recommendations can be made, its feasibility and utility should be further explored. This could be done in a so-called management study, in which the rule is implemented in a protocol and its performance closely monitored over time by carefully collecting patient outcomes. An alternative study design, requiring a higher number of patients to be included, would be to randomly allocate patients with suspected appendicitis and a negative or inconclusive ultrasound to the decision rule or to standard additional imaging.