Time-ordered comorbidity correlations identify patients at risk of mis- and overdiagnosis

Abstract

Diagnostic errors are common and can lead to harmful treatments. We present a data-driven, generic approach for identifying patients at risk of being mis- or overdiagnosed, here exemplified by chronic obstructive pulmonary disease (COPD). It has been estimated that 5–60% of all COPD cases are misdiagnosed. High-throughput methods are therefore needed in this domain. We have used a national patient registry, which contains hospital diagnoses for 6.9 million patients across the entire Danish population for 21 years and identified statistically significant disease trajectories for COPD patients. Using 284,154 patients diagnosed with COPD, we identified frequent disease trajectories comprising time-ordered comorbidities. Interestingly, as many as 42,459 patients did not present with these time-ordered, common comorbidities. Comparison of the individual disease history for each non-follower to the COPD trajectories, demonstrated that 9597 patients were unusual. Survival analysis showed that this group died significantly earlier than COPD patients following a trajectory. Out of the 9597 patients, we identified one subgroup comprising 2185 patients at risk of misdiagnosed COPD without the typical events of COPD patients. In all, 10% of these patients were diagnosed with lung cancer, and it seems likely that they are underdiagnosed for lung cancer as their laboratory test values and survival pattern are similar to such patients. Furthermore, only 4% had a lung function test to confirm the COPD diagnosis. Another subgroup with 2368 patients were found to be at risk of “classically” overdiagnosed COPD that survive >5.5 years after the COPD diagnosis, but without the typical complications of COPD.

A software package for efficient patient trajectory analysis applied to analyzing bladder cancer development

“…In this context, a number of statistical and machine learning methods are being developed for discovering temporal disease trajectories in large-scale databases with electronically registered patient histories [1][2][3][4][5][6][7][8][9]. Specifically, these approaches investigate patient histories that consist of medical diagnoses in the form of International Classification of Diseases 9 or 10 (ICD-9 or ICD-10) codes [10] or other standardized medical description codes (SNOMED, READ, MedDRA) recorded for each doctor visit.…”

“…The term "disease trajectory" then refers to a series of diagnoses in such a standardized format. The databases targeted by these analyses typically contain thousands of patients and span several years [1][2][3][4][5][6][7][8][9].…”

“…The trajectories found by these approaches confirm known information about patient progression [1,3], but also enable generating and confirming new medical hypotheses. Examples of this include insights in terms of the effects of mis-and over-diagnosis on patient progression [8], the impact of certain treatments and drugs [8], the influence of diabetes on developing comorbid diseases and the differences between the sexes in terms of the speed of development [5], the increased risk of cardiovascular disease progression for specific populations [4], and various others [1][2][3][4][5][6]8].…”

“…Our understanding is that the literature focuses on presenting medical discoveries made with temporal disease trajectories, but we argue there is value in being able to adapt the methods for analyzing new data sets or to reproduce their results. A notable exception is the work by Paik et al [7,9], for which source code was released, but while they released an interesting GUI and visualization method, they only released a simplified method to calculate trajectories compared to the proposals in related work [1][2][3][4][5]8] and do not provide any code for clustering trajectories.…”

Disease trajectory of high neuroticism and the relevance to psychiatric disorders: A retro‐prospective cohort study

“…Population-wide disease trajectory analysis was first proposed by Jensen et al, 10 and has since been applied and tested in several large-scale cohort studies [11][12][13][14][15] to explore temporal patterns of progression in diseases of diverse categories, such as cancer, 16,17 diabetes mellitus, 18 and depression, 19 discover potential differences in possible shared etiologies or disease mechanisms through disease clustering and help generalize key pathways for disease onset in the future. Moreover, a study conducted a disease trajectory analysis among women and used biomarkers representing preclinical states to identify potential indicators of disease at an earlier stage.…”