RECONSTRUCTION OF THE DECISION-MAKING PROCESS IN ASSESSING MUSCULOSKELETAL CHEST PAIN: AN EXPLORATORY STUDY USING RECURSIVE PARTITIONING Mette J. Stochkendahl, DC, PhD, a Werner Vach, PhD, b Jan Hartvigsen, DC, PhD, c Poul F. Høilund-Carlsen, MD, DMSci, d Torben Haghfelt, MD, DMSci, e and Henrik W. Christensen, DC, MD, PhD f ABSTRACT Objective: The purposes of this study were to identify the most important determinants from the patient history and clinical examination in diagnosing musculoskeletal chest pain (MSCP) in patients with acute noncardiac chest pain when supported by a structured protocol and to construct a decision tree for identification of MSCP in acute noncardiac chest pain. Methods: Consecutive patients with noncardiac chest pain (n = 302) recruited from an emergency cardiology department were assessed. Using data from self-report questionnaires, interviews, and clinical assessment, patient characteristics were associated with the MSCP diagnosis, and the decision-making process of the clinician was reconstructed using recursive procedures in the tradition of constructing Classification and Regression Trees. Results: Thirty-eight percent of patients had MSCP. There was no single determinant that predicted the condition completely. However, many items with high associations could be identified, mainly with high negative predictive value. The decision-making process was reconstructed giving rise to a 5-step, linear decision tree without branches. Conclusions: Clinicians use a combination of indicators including systematic palpation of the spine and chest wall and items from the case history to diagnose MSCP. However, the high negative predictive values of the main determinants suggest that the MSCP diagnosis may be a diagnosis by exclusion. (J Manipulative Physiol Ther 2012;35:184-195) Key Indexing Terms: Diagnosis; Chest Pain; Chiropractic; Physical Examination; Palpation E very day, clinicians make diagnostic decisions against a backdrop of complexity and uncertainty. The multifac- torial nature of clinical reasoning requires the clinician to make use of multiple sources of knowledge to selectively gather and evaluate data during the patient examination. 1 Commonly, in musculoskeletal problems, substantial uncer- tainty remains at the time when therapeutic decisions must be made, despite the diversity of clinical observations available. Back pain is among the most frequently encoun- tered medical problems of this type, and even when imaging techniques such as radiography and magnetic resonance imaging or other testing procedures are applied, they are of limited value for many patients because there is no “gold standard” to verify a given diagnosis. In many instances, this leaves the clinicians with only the data from the patient encounter itself to form the basis for a diagnosis. a Researcher, Nordic Institute of Chiropractic and Clinical Biomechanics & Institute of Sports Science and Clinical Biomechanics, University of Southern Denmark, Part of Clinical Locomotion Network, Odense, Denmark. b Professor, Clinical Epidemiology, Institute of Medical Biometry and Medical Informatics, University Medical Centre, Freiburg, Germany. c Professor, Nordic Institute of Chiropractic and Clinical Biomechanics & Institute of Sports Science and Clinical Biomechanics, University of Southern Denmark, Part of Clinical Locomotion Network, Odense, Denmark. d Professor, Department of Nuclear Medicine, Odense Univer- sity Hospital, Odense, Denmark. e Professor, Department of Cardiology, Odense University Hospital, Odense, Denmark. f Director, Nordic Institute of Chiropractic and Clinical Biome- chanics, Part of Clinical Locomotion Network, Odense, Denmark. Submit requests for reprints to: Mette J. Stochkendahl, DC, PhD, Researcher, Nordic Institute of Chiropractic and Clinical Biomechanics, Part of Clinical Locomotion Network, Forskerpar- ken 10A, DK-5230 Odense M, Denmark (e-mail: m.jensen@nikkb.dk). Paper submitted August 30, 2011; in revised form November 14, 2011; accepted November 14, 2011. 0161-4754/$36.00 Copyright © 2012 by National University of Health Sciences. doi:10.1016/j.jmpt.2012.01.009 184