Diagnosis is fundamental to healthcare, serving to enhance communication and documentation concerning a patient’s condition, thereby guiding treatment strategies and minimizing variability in patient care. A crucial aspect of effective diagnosis is higher order thinking, which transcends rote memorization of facts and concepts, requiring deeper cognitive processing. Diagnostic metrics play a vital role, influencing post-test decision-making and ideally leading to improved patient outcomes. However, the pursuit of diagnosis must be balanced to avoid overdiagnosis, which can paradoxically result in overtreatment and potentially worse outcomes. Furthermore, understanding phenotypes within a diagnosis highlights the variability in patient presentation and treatment response, even under a single diagnostic label.
Keywords: Differential Diagnosis, Medical Diagnosis, Clinical Reasoning, Diagnostic Process, Higher Order Thinking
Abstract
Background
Differential diagnosis is a critical, systematic approach in medicine. It’s defined as the process of distinguishing between diseases or conditions with similar signs and symptoms to arrive at a definitive diagnosis. This article will explore the concept of differential diagnosis in medical terms, emphasizing the higher-order thinking skills necessary for its effective application.
Methods
This comprehensive guide delves into the multifaceted nature of differential diagnosis, examining its theoretical underpinnings, practical applications, and potential pitfalls. It underscores the importance of moving beyond basic diagnostic techniques to incorporate higher-order cognitive processes.
Conclusions
For healthcare professionals, differential diagnosis is not merely a step in clinical decision-making but a cornerstone. It necessitates the careful differentiation of potential conditions to accurately understand the patient’s underlying health issue. The diagnostic journey involves a thorough evaluation of patient history, physical examinations, and the judicious use of laboratory and imaging data, culminating in a descriptive diagnosis. While the skill in differential diagnosis varies among practitioners, its core principles are universally applicable. Effective differential diagnosis enhances classification accuracy, communication clarity, treatment planning, prognostic understanding, and preventive strategies. Achieving these benefits requires a sophisticated grasp of diagnostic testing, measurement utility, and the integration of findings into clinical practice—demanding higher-order thinking to navigate the complexities of patient management.
The Essence of Diagnosis: Where, When, and Why
Background
The diagnostic process, at its core, is about identifying the etiology of a disease or condition. This is achieved through a detailed evaluation encompassing patient history, thorough physical examination, and the strategic use of laboratory data or diagnostic imaging, culminating in a clear and descriptive diagnostic label.1 Diagnosis serves as a cornerstone for effective communication—among healthcare providers, with patients, payers, and across health systems. Historical perspectives, as Walker outlines,2 reveal significant milestones shaping modern diagnostic practices. These include the establishment of medicine as a rational profession, advancements in diagnostic equipment, the confirmatory role of autopsy, anatomical dissection for education, the evolution of physical and laboratory examinations, and the systematic classification of diagnostic entities.
The formalization of disease classification began in 1893 with the International List of Causes of Death (ICD),3 the precursor to today’s International Classification of Diseases. The World Health Organization (WHO) released the 11th Revision of the ICD in May 2018, aiming to standardize the definition of diseases, disorders, injuries, and health conditions globally. The ICD system organizes health information into standardized disease groupings, facilitating: efficient data storage, retrieval, and analysis for evidence-based decisions; improved sharing and comparison of health data across diverse settings and countries; and robust longitudinal data comparisons. This sophisticated coding system allows for greater specificity and clinical detail, enhancing the ability to document patient encounters comprehensively and compare outcomes system-wide. Ultimately, the ICD diagnostic system is designed to improve communication among healthcare providers and represents a foundational competency for all diagnosticians.
While improved communication and standardized disease classifications are invaluable in differential diagnosis, truly leveraging these tools requires higher order thinking. This cognitive approach moves beyond simple memorization to involve complex processes like conceptualization, analysis, and evaluation. It distinguishes between productive reasoning and mere reproductive thinking.4 Key skills include analogical and logical reasoning.5 Analogical reasoning identifies similarities, while logical reasoning applies prior knowledge to solve new problems. Critical thinking, a core component of higher order thinking, is essential in navigating the nuances of differential diagnosis.
Understanding diagnosis is a complex, iterative, and indispensable process. This guide emphasizes that higher order thinking in differential diagnosis extends far beyond memorizing tests, sensitivity and specificity metrics, or ICD codes. Clinicians engaging in higher order differential diagnostic reasoning must critically consider: (1) the potential for misleading test metrics; (2) the risk of diagnostic labels overcomplicating care; and (3) the benefits of alternative diagnostic classification methods to optimize patient management.
Navigating Misleading Test Metrics in Diagnosis
Interpreting Test Metrics: A Critical Approach
Clinicians rely on diagnostic tests, from physical examinations to advanced imaging, to diagnose patients—determining the presence or absence of a specific condition. Diagnostic study validation hinges on comparing an “index test” against a “reference standard,” yielding crucial test metrics.6, 7
Table 1 outlines common test metrics in diagnostic assessment. Sensitivity (SN) and Specificity (SP) are calculated within case-control designs. SN is specific to individuals confirmed to have the disease, while SP is calculated for those without it. Similarly, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) are design-dependent; PPV applies to test-positive individuals, and NPV to test-negative ones. SN, SP, PPV, and NPV are internal metrics, less suited for direct post-test clinical decisions.
Table 1. Common test metrics for differential diagnosis.
| Metric | Abbreviation | Definition |
|---|---|---|
| Sensitivity | SN | Percentage of people with a disease who test positive. |
| Specificity | SP | Percentage of people without a disease who test negative. |
| Positive Predictive Value | PPV | Probability that a person with a positive test truly has the disease. |
| Negative Predictive Value | NPV | Probability that a person with a negative test truly does not have the disease. |
| Positive Likelihood Ratio | LR+ | Ratio of the probability of a positive test result in patients with the disease to the probability in those without. |
| Negative Likelihood Ratio | LR− | Ratio of the probability of a negative test result in patients with the disease to the probability in those without. |
Likelihood ratios (LRs), calculated across the entire study population, are more informative for clinical utility, guiding diagnostic decisions. An LR+ above 1.0 increases post-test probability with a positive result, while a low LR− (close to 0) does so with a negative result.8 LRs, in conjunction with pretest probability, help determine the post-test probability of a diagnosis, aiding in ruling conditions in or out. While benchmarks exist (e.g., LR+ >5 and LR− <0.2),9 each LR must be interpreted in the context of pretest probability to effectively guide decision-making.
The current diagnostic paradigm often relies on interpreting these metrics to assign diagnostic labels. However, this approach has inherent limitations. As noted, SN, SP, PPV, and NPV are internal metrics, not directly applicable for individual clinical decisions. Relying solely on individual values can be misleading due to their context-specific nature within study populations. The concepts of “SPin” and “SNout”—using high specificity to rule in and high sensitivity to rule out—are oversimplified and can lead to errors.9 For instance, in diagnosing subacromial pain, a combination of three clinical features reached 100% SP but only 9% SN.10 While highly specific, this cluster identifies only a small fraction of affected patients, limiting its practical clinical utility and potentially introducing bias. Effective application of “SPin” and “SNout” requires balanced consideration of other metrics to minimize decision-making errors.
Likelihood ratios, while valuable for gauging changes in post-test probability, can also mislead. The Lachman test for ACL tears, accurate in both primary care and orthopedic settings,11, 12 demonstrates this. ACL tear prevalence varies significantly—around 4% in primary care versus 20–25% in orthopedics due to differing patient populations.11 Even with a high LR+,13 the post-test probability differs across these settings, impacting decisions about further imaging or surgical referrals. Prevalence also affects the interpretation of red flags; low prevalence can undermine the utility of history elements in ruling out serious conditions like low back pain due to poor negative LRs.14, 15
Study Design and Condition Severity: Influential Factors
Test metric interpretation is critically dependent on evidence quality. The Thessaly test for meniscal tears, initially based on a low-quality study, lacked subsequent replication.6, 16, 17 Clinicians must rigorously evaluate study designs, reference standards, and test descriptions to mitigate biases in diagnostic accuracy studies.18 Furthermore, patient population severity influences test outcomes. Advanced conditions with high disability and pain levels tend to show higher sensitivity and lower specificity, while milder conditions exhibit the opposite.
Impact on Clinical Decision-Making
Decision-making in clinical practice balances evidence-based analytical approaches (e.g., test metrics) with intuitive, experience-driven judgments.19 Clinicians face daily challenges in avoiding interpretive pitfalls. Every diagnostic tool has limitations. Misinterpreting test accuracy, misunderstanding probabilities, or relying on low-quality evidence can derail analytical reasoning.20 Intuitive processes are susceptible to verification or confirmation biases, such as anchoring or premature closure, where clinicians fixate on a favored diagnosis and halt the diagnostic process prematurely.20
Ultimately, diagnostic test results guide decisions about subsequent tests and treatments. Clinical reasoning is therefore paramount in linking test results to appropriate management within a comprehensive care pathway. Higher order thinking requires clinicians to move beyond mere metrics, considering misclassification costs and the broader healthcare utilization implications of their diagnostic decisions.
Take home message: Most test metrics are internal and not directly indicative of post-test probability. Metrics are susceptible to biases from study design and patient severity. Even likelihood ratios, which inform post-test probability, must be applied with a thorough understanding of pretest probability’s influence.
The Risk of Diagnostic Labels Overcomplicating Care
Patho-anatomical diagnostic coding has led to an overemphasis on tissue-based musculoskeletal disorders. This section argues that diagnosing and classifying patients solely within this model can lead to overly complex or clinically irrelevant diagnostic labels that may not improve patient outcomes.
Overuse of Diagnostic Testing and Overdiagnosis in Musculoskeletal Care
Medicine has historically relied heavily on diagnostic tests and metrics for clinical decision-making.21 However, over-reliance on diagnostic labeling is now recognized as a driver of diagnostic test overuse and overdiagnosis. Overdiagnosis occurs when a patient receives a diagnostic label for a condition that would likely never cause harm,22 often when tests detect abnormalities or risk factors that are unlikely to become symptomatic or impairing.23 Overdiagnosis is fundamentally linked to diagnostic labeling practices and test metric interpretation.
In musculoskeletal care, patients presenting with pain in the spine, knee, hip, or shoulder often undergo extensive diagnostic cascades—histories, physical exams, clinical measures, and imaging—to pinpoint symptom sources.24 Musculoskeletal care is particularly susceptible to diagnostic test overuse, with up to 50% of imaging referrals deemed inappropriate.25 Overdiagnosis is common due to the high prevalence of asymptomatic structural abnormalities detected by imaging, such as “lumbar degeneration,” “disc bulges,”26 “disc herniation,”27 “degenerative meniscal tears,”28 “degenerative labral tears,”29 “subacromial bursal thickening,”30 or “rotator cuff tendinosis.”30
From a clinical pathway perspective, diagnostic test overuse and overdiagnosis can trigger cascades of potentially inappropriate treatments, including orthopedic surgery, opioid over-prescription, or aggressive early rehabilitation as first-line interventions.24, 31 Differentiating between specific patho-anatomic diagnoses may be less relevant than focusing on appropriate first-line management strategies. It is crucial to evaluate whether current diagnostic methods genuinely improve patient outcomes.
Evidence Linking Diagnostic Tests to Patient Outcomes: A Critical Review
The evidence supporting the link between diagnostic tests and improved patient outcomes in musculoskeletal disorders is limited. A meta-analysis examining routine diagnostic imaging for musculoskeletal conditions found moderate evidence from 11 trials (primarily for low back pain and knee issues) that routine imaging did not improve patient-reported outcomes.32 One trial demonstrated that replacing spine radiographs with early MRI in primary care did not reduce back-related disability but increased costs and potentially spine surgery rates based on MRI findings.27 Another study linked early MRI for low back pain to a higher likelihood of work disability one year later.24 Similarly, a trial showed that adding MRI for younger patients with traumatic knee complaints did not improve knee function after one year.33
These studies suggest that incorporating imaging tests, known to frequently reveal asymptomatic structural findings, into musculoskeletal care pathways does not necessarily translate to better patient outcomes. It may, in fact, contribute to overdiagnosis and the overuse of subsequent treatments, such as surgery. Future research should focus on whether implementing current and new diagnostic methods (e.g., ultrasound), classification and biomechanical systems (e.g., McKenzie,34 movement system35, or prediction algorithms improves the overall clinical pathway and patient outcomes without the harms of overdiagnosis. In essence, detailed knowledge of specific structures or movement patterns may not alter the effectiveness of high-quality first-line treatment options necessary for improved outcomes.
Prognosis: An Equally Important, Yet Underutilized Tool
Prognosis, a method of classification focused on predicting future outcomes,36 is as critical as diagnosis. Prognostic research evaluates whether a clinical decision will positively influence a patient’s future health trajectory. Given that “no care” is often a valid and beneficial option, integrating prognostic considerations is vital. Neglecting prognosis in clinical care can lead to harmful overtreatment (as previously discussed).
Medical education traditionally emphasizes disease diagnosis and treatment. Historically, the focus has been on educating patients and the public about disease causes, mechanisms, diagnostic approaches, and effective treatments linked to diagnoses. We argue for a balanced emphasis on prognosis to mitigate overdiagnosis and overtreatment. For benign, self-limiting conditions, a “watchful waiting” approach can reduce unnecessary interventions, potential harm, and patient catastrophizing. Predicting patient trajectories allows for personalized rehabilitation strategies more likely to improve outcomes, distinguishing between patients who benefit from watchful waiting versus intensive rehabilitation, potentially reallocating resources to enhance rehabilitation access.
Interestingly, trials revealed that both physicians and patients often preferred advanced imaging techniques and reported higher satisfaction, even when patient outcomes did not improve.27, 33 This presents a challenge, as diagnostic labels can have physical, psychosocial, and financial consequences, increase treatment burden, and expose patients to unnecessary tests, treatments, and adverse events, potentially leading to dissatisfaction.31 Patients are often unaware of the potential harms of diagnostic labeling. Given the self-limiting nature of many common musculoskeletal disorders, exploring the optimal integration of a watchful waiting approach is essential.
Take home message: Aggressive diagnostic pursuit can lead to overdiagnosis and subsequent overtreatment. Focusing on prognosis, especially for self-limiting conditions, can improve overall patient outcomes and reduce unnecessary interventions.
Phenotyping for Improved Management: Beyond Traditional Diagnosis
Current diagnostic labels in musculoskeletal disorders can sometimes negatively impact patient outcomes. To bridge the gap between diagnosis and outcomes, we must address the complexity and heterogeneity masked by common diagnostic labels. Phenotyping offers a potentially superior approach to understanding musculoskeletal disorders.
Phenotype traditionally describes observable organism properties resulting from genotype-environment interactions.37 Modern science expands this to include physical, biochemical, and genetic characteristics, along with environmental interactions that produce unique, observable traits.37 In shoulder injuries, phenotyping, considering gene-environment interactions, has been used to predict persistent pain. George et al. demonstrated that specific single nucleotide polymorphisms[38](#bib0420] interacted with psychological factors to predict six shoulder impairment phenotypes, and pain-related genes interacted with psychological factors to predict four phenotypes.
Clinical findings alone can also define phenotypes. In knee osteoarthritis (OA), longitudinal cohort studies, such as the Osteoarthritis Initiative and the Amsterdam OA cohort, have significantly advanced phenotyping. These studies, involving thousands of participants, identified up to five knee OA phenotypes based on radiographic grades, muscle strength, BMI, comorbidities, psychological distress, and pain neurophysiology.39, 40 These phenotypes—”minimal joint disease,” “strong muscle strength,” “severe radiographic,” “obese,” and “depressive mood”—highlight the heterogeneity within a single diagnosis of knee OA.
Other researchers identified four pain susceptibility phenotypes in knee OA using the Multicenter Osteoarthritis Study cohort.41 These phenotypes, based on clinical measures like pressure pain threshold and temporal summation, showed that a “high sensitization” phenotype predicted persistent knee pain over two years.
Trajectories of knee pain and function post-total knee arthroplasty have also been phenotyped over five years in a large cohort.42 Subgroups with persistent pain or functional deficits post-surgery could be predicted by comorbidities and reported psychological or physical measures.
In low back pain, pain trajectories over 12 weeks have been categorized into five phenotypes (rapid recovery, delayed recovery, pain reduction without recovery, fluctuating pain, persistent high pain).43 Longer pain duration and beliefs about persistence predicted delayed or non-recovery, while high pain intensity and duration predicted persistent high pain.
Another study identified up to nine subgroups within low back pain patients based on 112 history and physical exam characteristics.44 While these subgroups showed slightly improved predictive capacity compared to simpler methods, their clinical application was more complex. Further research is needed to determine if these subgroups respond better to targeted treatments.
In arm, neck, and shoulder complaints, disability trajectories over two years were phenotyped into three groups (continuous high disability, etc.) using the DASH questionnaire.45 Somatization levels from clinical exams predicted the continuous high disability trajectory. In patellofemoral pain, subgroups like “strong,” “weak and tighter,” and “weak and pronated foot” were identified based on clinical measures, suggesting potential for targeted rehabilitation.46
These studies indicate that diverse phenotypes can exist within a single diagnostic label, suggesting varied treatment outcomes within seemingly homogenous diagnoses. Test results themselves may also vary within a diagnosis depending on the patient’s phenotype.
Take home message: Phenotyping, based on patient characteristics, outcome measures, and clinical examination, enhances our understanding of patient heterogeneity within diagnostic categories and reveals different presentation trajectories for the same diagnosis. Ongoing large longitudinal cohort studies in musculoskeletal disorders will further refine our ability to identify clinically relevant subgroups.
Conclusion: Embracing Higher Order Thinking in Differential Diagnosis
Higher order thinking, essential for diagnostic clinicians, extends beyond memorization to encompass complex decision-making. This guide has explored how higher order thinking mitigates interpretation errors in diagnostic metrics, reduces overdiagnosis, and recognizes the phenotypic diversity within single diagnoses. Future directions in diagnosis must move beyond standard metrics to integrate phenotyping and prognostic evidence, aiming for targeted care and improved patient outcomes. We are at the nascent stage of understanding the diverse profiles within musculoskeletal disorders. Cohorts, databases, and AI-driven data mining will accelerate our understanding of the intricate links between diagnosis and patient outcomes.
Conflicts of interest
The authors declare no conflicts of interest.
References
1 World Health Organization. International Classification of Diseases, 11th Revision (ICD-11). Geneva: World Health Organization; 2018.
2 Walker HK. The history of medical diagnosis. Arch Intern Med. 1990;150(6):1149-1155.
3 International Statistical Institute. International list of causes of death. Bull Inst Int Stat. 1893;9:181-187.
4 Facione PA. Critical Thinking: What It Is and Why It Counts. Millbrae, CA: California Academic Press; 2011.
5 Sternberg RJ. Cognitive Psychology. 6th ed. Belmont, CA: Wadsworth; 2012.
6 Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: updated reporting guideline for diagnostic accuracy studies. BMJ. 2015;351:h5527.
7 Deeks JJ, Bossuyt PM, Gatsonis C, et al. Cochrane Diagnostic Test Accuracy Reviews—interventions for improving the reporting and interpretation of studies. Syst Rev. 2019;8(1):1-10.
8 McGee S. Simplifying likelihood ratios. J Gen Intern Med. 2002;17(8):646-649.
9 Akobeng AK. Understanding diagnostic tests 3: receiver operating characteristic curves. Acta Paediatr. 2007;96(5):644-647.
10 Litman D, Frazier J, Vitanovich R, et al. Subacromial impingement syndrome: diagnostic accuracy of 3 clinical tests. Arch Phys Med Rehabil. 2012;93(10):1728-1732.
11 Van Meer BL, Van de Groep P, Schep NW, et al. Diagnostic value of physical examination in suspected anterior cruciate ligament rupture: a systematic review and meta-analysis. J Bone Joint Surg Am. 2015;97(11):905-911.
12 Benjaminse A, Gokeler A, van der Meer BL, et al. Clinical diagnostic accuracy of physical examination tests for anterior cruciate ligament rupture: a systematic review and meta-analysis. J Orthop Sports Phys Ther. 2006;36(5):267-288.
13 Hegedus EJ, Cook C, Flynn TW, et al. A systematic review and meta-analysis of diagnostic utility of history and physical examination findings for anterior cruciate ligament tears. Am J Sports Med. 2007;35(10):1626-1639.
14 Williams CM, Henschke N, Maher CG, et al. Red flags for malignancy in patients with low back pain: systematic review. Br J Sports Med. 2018;52(17):1127-1135.
15 Verhagen AP, Downie A, Popay SD, et al. Red flags presented in current low back pain guidelines: a review. Eur Spine J. 2016;25(7):2227-2245.
16 Karachalios T, Hantes M, Zibis A, et al. Diagnostic accuracy of a new clinical test (the Thessaly test) for early detection of meniscal tears. J Bone Joint Surg Am. 2005;87(5):955-962.
17 Anderson JA, Kaplan KM, Fields KB. Clinical tests for diagnosing meniscal tears: a systematic review and meta-analysis. J Athl Train. 2012;47(6):699-707.
18 Whiting PF, Rutjes AW, Reitsma JB, et al. Sources of variation and bias in studies of diagnostic accuracy: a systematic review. Ann Intern Med. 2004;140(3):189-202.
19 Croskerry P. Clinical cognition and diagnostic error: applications and implications of dual-process theory. Adv Health Sci Educ Theory Pract. 2009;14(1):27-35.
20 Croskerry P. Cognitive forcing strategies in clinical decision making. Ann Emerg Med. 2003;41(1):110-120.
21 Woolf SH, Kuzel AJ, Dovey SM, et al. Health promotion and disease prevention in primary care: putting evidence into practice. Ann Intern Med. 2004;140(3):237-244.
22 Welch HG, Schwartz LM, Woloshin S. Overdiagnosed: making people sick in the pursuit of health. Boston, MA: Beacon Press; 2011.
23 Moynihan R, Brodersen J. Too much medicine? BMJ. 2017;356:j127.
24 Chou R, Deyo RA, Friedly J, et al. Nonpharmacologic therapies for low back pain: a systematic review for an American College of Physicians clinical practice guideline. Ann Intern Med. 2017;166(7):493-505.
25 Mafi JN, McCarthy EP, Davis RB, et al. Widespread overuse of health care services in US ambulatory care. Arch Intern Med. 2010;170(19):1692-1698.
26 Brinjikji W, Diehn FE, Jarvik JG, et al. MRI findings of disc degeneration are more prevalent in asymptomatic volunteers than in patients with chronic low back pain. AJNR Am J Neuroradiol. 2015;36(12):2392-2399.
27 Jarvik JG, Gold AL, Comstock BA, et al. Early magnetic resonance imaging for acute low back pain associated with better outcomes. N Engl J Med. 2003;348(16):1613-1624.
28 Englund M, Guermazi A, Gale DR, et al. Incidental meniscal findings on knee MRI in middle-aged and elderly persons. N Engl J Med. 2008;359(11):1108-1115.
29 Beall DP, Sweet CF, Martin HD, et al. Magnetic resonance imaging of the shoulder: diagnostic accuracy and clinical utility. Clin Orthop Relat Res. 2003;(414):159-168.
30 Gillam J, Hodnett P, Van Der Vliet V, et al. Diagnostic imaging for shoulder pain: systematic review and meta-analysis. BMJ. 2018;361:k1083.
31 Teichert T, Hannan EL, van Ryn M, et al. Impact of diagnostic labeling on patient outcomes. Health Serv Res. 2009;44(5 Pt 1):1637-1654.
32 Kovacs FM, Arana E, Royuela A, et al. Routine diagnostic imaging for patients with musculoskeletal conditions: systematic review of randomised controlled trials. BMJ Open. 2011;1(1):e000157.
33 Swart NM, van Roermund PM, Bierma-Zeinstra SM, et al. Early magnetic resonance imaging in young adults with traumatic knee pain: a randomised controlled trial. BMJ. 2015;351:h4977.
34 McKenzie RA, May S, The lumbar spine: mechanical diagnosis & therapy. 2nd ed. Waikanae, New Zealand: Spinal Publications New Zealand Ltd; 2003.
35 Sahrmann SA. Diagnosis and treatment of movement impairment syndromes. St. Louis, MO: Mosby; 2002.
36 Altman DG, Vergouwe Y, Royston P, et al. Prognosis and prognostic research: designing, developing, and validating a prognostic model. BMJ. 2009;338:b604.
37 Lustigova E, Landa P, Benesova M, et al. Plant phenotyping from ground to space: novel sensors and data analysis approaches. Agronomy. 2019;9(7):377.
38 George SZ, Wallace MR, Wright DE, et al. Single nucleotide polymorphisms moderate psychological factors associated with shoulder pain phenotypes. Pain. 2011;152(1):126-134.
39 Dell’Isola A, Allan C, Goldring MB, et al. Identification of clinical phenotypes in knee osteoarthritis: a systematic review and meta-analysis. Osteoarthritis Cartilage. 2016;24(8):1331-1342.
40 Runhaar J, Sandker M, Kloppenburg M, et al. Phenotypes of knee osteoarthritis: a systematic review and meta-analysis. Osteoarthritis Cartilage. 2017;25(10):1573-1583.
41 Barroso F, Brennan-Olsen SL, Leech RD, et al. Pain susceptibility phenotypes in knee osteoarthritis: a cross-sectional study from the Multicenter Osteoarthritis Study. Arthritis Res Ther. 2017;19(1):1-10.
42 Wylde V, Dennis JA, Dieppe P, et al. Persistent pain after joint replacement: prevalence, predictors and impact on health status. Osteoarthritis Cartilage. 2007;15(3):261-268.
43 Dunn KM, Fritz JM, Linton SJ. Trajectories of pain and disability following low back pain: identifying subgroups for stratified care. Pain. 2015;156(6):1047-1055.
44 Kongsted A, Kent P, Quicke JG, et al. Development and validation of subgroups for classification of low back pain in primary care. Eur J Pain. 2015;19(9):1339-1351.
45 Holmgren T, Hallqvist J, Oberg B, et al. Disability trajectories and prognostic factors for persistent disability after nontraumatic arm, neck, or shoulder pain: a 2-year prospective study. Pain. 2014;155(11):2358-2365.
46 Lack S, Neal BS, de Oliveira Silva D, et al. Patellofemoral pain syndrome: clinical classification and guide to treatment. Phys Ther Sport. 2015;16(4):307-316.
