Research Ariella P Dale, Christian Marchello and Mark H Ebell Clinical gestalt to diagnose pneumonia, sinusitis, and pharyngitis: a meta-analysis INTRODUCTION Abstract Instead, clinicians rely on their overall The overall clinical impression, also clinical impression. As the overall clinical Background called ‘clinical gestalt’, is an intuitive impression can incorporate additional The overall clinical impression (‘clinical gestalt’) is approach to decision making used by variables not included in the CDR, it has widely used for diagnosis but its accuracy has not been systematically studied. physicians to make clinical diagnoses. the potential of being more accurate. It takes into account multiple signs and For example, while a clinical rule may Aim To determine the accuracy of clinical gestalt for symptoms without necessarily using an categorise a patient as being at low risk the diagnosis of community-acquired pneumonia analytic approach such as a point score for group A beta-haemolytic streptococcal (CAP), acute rhinosinusitis (ARS), acute bacterial or algorithm, and is an inductive approach (GABHS) pharyngitis, knowing that a sibling rhinosinusitis (ABRS), and streptococcal based on pattern recognition rather was diagnosed with GABHS pharyngitis the pharyngitis, and to contrast it with the accuracy of clinical decision rules (CDRs). than a hypotheticodeductive approach. week before could be an important factor. Some studies have shown that inductive For acute respiratory tract infections, Design and setting Systematic review and meta-analysis of outpatient pattern-recognition strategies may be CDRs have been developed to diagnose diagnostic accuracy studies in ambulatory care. more widely used and more successful GABHS pharyngitis,8,9 acute rhinosinusitis 1–3 Method than hypotheticodeductive strategies. (ARS) and acute bacterial rhinosinusitis 10 PubMed and Google were searched for studies However, proponents of evidence-based (ABRS), and community-acquired in outpatients that reported sufficient data practice encourage the use of clinical pneumonia (CAP).11 In this study, the authors to calculate accuracy of the overall clinical decision rules (CDRs) for diagnosis, as do performed a systematic review of the impression and that used the same reference standard. Study quality was assessed using practice guidelines. CDRs use a formal accuracy of the overall clinical impression Quality Assessment of Diagnostic Accuracy approach such as multivariate analysis for GABHS pharyngitis, ARS, and CAP, which Studies-2 (QUADAS-2), and measures of accuracy or recursive partitioning to identify signs, has not been systematically studied before, calculated using bivariate meta-analysis. symptoms, and point-of-care tests that and evaluated how its accuracy compared Results are the best independent predictors of a with that of CDRs for the same conditions. The authors identified 16 studies that met the diagnosis or clinical outcome. They are then inclusion criteria. The summary estimates for the positive (LR+) and negative likelihood ratios typically converted to a simple point score METHOD (LR–) were LR+ 7.7, 95% confidence interval or algorithm such as the Ottawa Ankle Search (CI) = 4.8 to 11.5, and LR– 0.54, 95% CI= 0.42 Rules for ankle injury,4 or the Wells rule to For this systematic review, PubMed was to 0.65 for CAP in adults, LR+ 2.7, 95% CI= 1.1 diagnose pulmonary embolism.5 The goal searched for published studies using a to 4.3 and LR– 0.63, 95% CI= 0.20 to 0.98 for CAP in children, LR+ 3.0, 95% CI= 2.1 to 4.4 and of CDRs is to improve the efficiency and search strategy (available from the LR– 0.37, 95% CI = 0.29 to 0.46 for ARS in adults, accuracy of clinical diagnosis and thereby authors), combining synonyms for overall LR+ 3.9, 95% CI = 2.4 to 5.9 and LR– 0.33, 95% reduce unnecessary testing.6 clinical impression, the clinical diagnosis, CI = 0.20 to 0.50 for ABRS in adults, and LR+ 2.1, 95% CI = 1.6 to 2.8 and LR– 0.47, 95% CI= 0.36 However, CDRs may be cumbersome and ambulatory care. The reference to 0.60 for streptococcal pharyngitis in adults and to access and use at the point of care. lists of all included studies were also children. The diagnostic odds ratios were highest As a result, CDRs are only infrequently searched to identify studies not captured for CAP in adults (14.2, 95% CI= 9.0 to 21.0), ARS 7 used in real-world clinical practice. by the PubMed search strategy. In addition, in adults (8.3, 95% CI = 4.9 to 13.1), and ABRS in adults (13.0, 95% CI = 5.0 to 27.0), as were the C-statistics (0.80, 0.77, and 0.84 respectively). Conclusion AP Dale, PhD, MPH, healthcare-associated Email: [email protected] infections surveillance data coordinator, Colorado The accuracy of the overall clinical impression Submitted: 24 November 2018; Editor’s Department of Public Health and Environment, compares favourably with the accuracy of CDRs. Denver, US. C Marchello, PhD, MSc, fellow, response: 19 December 2018; final acceptance: Studies of diagnostic accuracy should routinely University of Otago, Dunedin, New Zealand. 4 January 2019. include the overall clinical impression in addition MH Ebell, MD, MSc, professor, Department of to individual signs and symptoms, and research is ©British Journal of General Practice Epidemiology and Biostatistics, College of Public needed to optimise its teaching. Health, University of Georgia, Athens, Georgia, US. This is the full-length article (published online Keywords Address for correspondence 18 Jun 2019) of an abridged version published in acute rhinosinusitis; community-acquired Mark Ebell, 125 Miller Hall, University of Georgia print. Cite this version as: Br J Gen Pract 2019; pneumonia; diagnosis; evidence-based medicine; Health Sciences Campus, Athens, GA 30602, US. DOI: https://doi.org/10.3399/bjgp19X704297 overall clinical impression; pharyngitis. e444 British Journal of General Practice, July 2019 Inclusion and exclusion criteria How this fits in The present research was limited to prospective studies that reported diagnostic It is known that the overall clinical impression is widely used in clinical data regarding the accuracy of the overall practice but has not been systematically clinical impression (clinical gestalt) studied. This study showed that in to diagnose CAP, ARS, ABRS, or acute adults the overall clinical impression GABHS pharyngitis. ARS was defined as had good accuracy for the diagnosis of abnormal imaging, and ABRS as abnormal community-acquired pneumonia, for acute rhinosinusitis, and for acute bacterial culture of antral puncture fluid. Studies rhinosinusitis. It had moderate accuracy were limited to the ambulatory-care setting for diagnosis of streptococcal pharyngitis (outpatient clinic, urgent care, or emergency and for pneumonia in children. In each department [ED]) as hospital-acquired case, the accuracy of the overall clinical and ventilator-associated pneumonia impression was similar to or better than that for a clinical decision rule for the are separate clinical entities. All patients same conditions. Thus, the overall clinical must have received the same acceptable impression has good accuracy and is an reference standard: chest radiograph (CXR), important diagnostic tool that is deserving lung ultrasound, or computed tomography of further study and quantification. (CT) for pneumonia; imaging or antral puncture fluid analysis for ARS; and throat culture for GABHS pharyngitis. The authors published systematic reviews of the clinical excluded studies of nosocomial infections, diagnosis of GABHS pharyngitis, CAP, and infections in immunocompromised persons, ARS or ABRS were searched for additional or studies of the diagnosis of bacteraemia studies,12–16 as were the first 50 results or sepsis. The authors included studies of returned by a Google search of ‘<disease> both children and adults. Studies of ARS diagnosis clinical impression’ for each using inspection of antral puncture fluid or disease. The search was not restricted by bacterial culture as the reference standard language, country, or date of publication. were classified as also diagnosing ABRS. Records identified through Additional records identified database searching through other sources (n = 2109) (n = 1) Identification Records after duplicates removed (n = 2110) Screening Records screened Records excluded (n = 2110) (n = 2056) Full-text articles excluded, Full-text articles primarily due to lack of assessed for eligibility data regarding accuracy of (n = 54) clinical gestalt ( = 38) Eligibility n Studies included in qualitative synthesis (n = 16) Studies included in Included quantitative synthesis (meta-analysis) (n = 16) Figure 1. PRISMA flow diagram of study search. British Journal of General Practice, July 2019 e445 Data abstraction Gatsonis.19 The authors used a summary Each title and abstract was reviewed by two receiver operating characteristic (ROC) investigators to identify potential studies for curve to plot 95% confidence intervals for inclusion. Any study identified for full-text the summary estimates and calculated the analysis by one of the reviewers was reviewed area under the ROC curve (AUROCC), also independently by two investigators, and called the C-statistic. Heterogeneity was any discrepancies were resolved by a third evaluated using inspection of the summary reviewer (lead investigator). For studies that ROC plots and confidence intervals, as I 2 met the inclusion and exclusion criteria, two is not recommended for use in diagnostic reviewers abstracted study characteristics, meta-analysis20 or when there is a small data regarding the accuracy of clinical number of primary studies.21 To facilitate gestalt, and study design characteristics for comparison with a dichotomous overall the quality assessment, with discrepancies clinical impression for each diagnosis, resolved via consensus discussion or, if clinical decision rules were dichotomised necessary, by the lead investigator. All of into low or moderate versus high risk, the included studies were reviewed a final or low risk versus moderate or high risk time by the lead investigator to confirm the depending on which approach provided the accuracy of data abstraction. highest diagnostic odds ratio (DOR).