Temporal patterns of multi-morbidity in 570157 ischemic heart disease patients: a nationwide cohort study

Co-Occurring Diseases and Mortality in Patients With Chronic Heart Disease, Modeling Their Dynamically Expanding Disease Portfolios: Nationwide Register Study

BACKGROUND

Medical advances in managing patients with chronic heart disease (HD) permit the co-occurrence of other chronic diseases due to increased longevity, causing them to become multimorbid. Previous research on the effect of co-occurring diseases on mortality among patients with HD often considers disease counts or clusters at HD diagnosis, overlooking the dynamics of patients' disease portfolios over time, where new chronic diseases are diagnosed before death. Furthermore, these studies do not consider interactions among diseases and between diseases, biological and socioeconomic variables, which are essential for addressing health disparities among patients with HD. Therefore, a mapping of the effect of combinations of these co-occurring diseases on mortality among patients with HD considering such interactions in a dynamic setting is warranted.

OBJECTIVE

This study aimed to examine the effect of the co-occurring diseases of patients with HD on mortality, modeling their dynamically expanding chronic disease portfolios while identifying interactions between the co-occurring diseases, socioeconomic and biological variables.

METHODS

This study used data from the national Danish registries and algorithmic diagnoses of 15 chronic diseases to obtain a study population of all 766,596 adult patients with HD in Denmark from January 1, 1995, to December 31, 2015. The time from HD diagnosis until death was modeled using an extended Cox model involving chronic diseases and their interactions as time-varying covariates. We identified interactions between co-occurring diseases, socioeconomic and biological variables in a data-driven manner using a hierarchical forward-backward selection procedure and stability selection. We mapped the impact on mortality of (1) the most common disease portfolios, (2) the disease portfolios subject to the highest level of interaction, and (3) the most severe disease portfolios. Estimates from interaction-based models were compared to an additive model.

RESULTS

Cancer had the highest impact on mortality (hazard ratio=6.72 for male individuals and 7.59 for female individuals). Excluding cancer revealed schizophrenia and dementia as those with the highest mortality impact (top 5 hazard ratios in the 11.72-13.37 range for male individuals and 13.86-16.65 for female individuals for combinations of 4 diseases). The additive model underestimated the effects up to a factor of 1.4 compared to the interaction model. Stroke, osteoporosis, chronic obstructive pulmonary disease, dementia, and depression were identified as chronic diseases involved in the most complex interactions, which were of the fifth order.

CONCLUSIONS

The findings of this study emphasize the importance of identifying and modeling disease interactions to gain a comprehensive understanding of mortality risk in patients with HD. This study illustrated that complex interactions are widespread among the co-occurring chronic diseases of patients with HD. Failing to account for these interactions can lead to an oversimplified attribution of risk to individual diseases, which may, in cases of multiple co-occurring diseases, result in an underestimation of mortality risk.

Disease Trajectories from Healthcare Data: Methodologies, Key Results, and Future Perspectives

Disease trajectories, defined as sequential, directional disease associations, have become an intense research field driven by the availability of electronic population-wide healthcare data and sufficient computational power. Here, we provide an overview of disease trajectory studies with a focus on European work, including ontologies used as well as computational methodologies for the construction of disease trajectories. We also discuss different applications of disease trajectories from descriptive risk identification to disease progression, patient stratification, and personalized predictions using machine learning. We describe challenges and opportunities in the area that eventually will benefit from initiatives such as the European Health Data Space, which, with time, will make it possible to analyze data from cohorts comprising hundreds of millions of patients.

Hospital-wide natural language processing summarising the health data of 1 million patients

Electronic health records (EHRs) represent a major repository of real world clinical trajectories, interventions and outcomes. While modern enterprise EHR's try to capture data in structured standardised formats, a significant bulk of the available information captured in the EHR is still recorded only in unstructured text format and can only be transformed into structured codes by manual processes. Recently, Natural Language Processing (NLP) algorithms have reached a level of performance suitable for large scale and accurate information extraction from clinical text. Here we describe the application of open-source named-entity-recognition and linkage (NER+L) methods (CogStack, MedCAT) to the entire text content of a large UK hospital trust (King's College Hospital, London). The resulting dataset contains 157M SNOMED concepts generated from 9.5M documents for 1.07M patients over a period of 9 years. We present a summary of prevalence and disease onset as well as a patient embedding that captures major comorbidity patterns at scale. NLP has the potential to transform the health data lifecycle, through large-scale automation of a traditionally manual task.

Longitudinal models for the progression of disease portfolios in a nationwide chronic heart disease population

BACKGROUND AND AIM

With multimorbidity becoming increasingly prevalent in the ageing population, addressing the epidemiology and development of multimorbidity at a population level is needed. Individuals subject to chronic heart disease are widely multimorbid, and population-wide longitudinal studies on their chronic disease trajectories are few.

METHODS

Disease trajectory networks of expected disease portfolio development and chronic condition prevalences were used to map sex and socioeconomic multimorbidity patterns among chronic heart disease patients. Our data source was all Danish individuals aged 18 years and older at some point in 1995-2015, consisting of 6,048,700 individuals. We used algorithmic diagnoses to obtain chronic disease diagnoses and included individuals who received a heart disease diagnosis. We utilized a general Markov framework considering combinations of chronic diagnoses as multimorbidity states. We analyzed the time until a possible new diagnosis, termed the diagnosis postponement time, in addition to transitions to new diagnoses. We modelled the postponement times by exponential models and transition probabilities by logistic regression models.

FINDINGS

Among the cohort of 766,596 chronic heart disease diagnosed individuals, the prevalence of multimorbidity was 84.36% and 88.47% for males and females, respectively. We found sex-related differences within the chronic heart disease trajectories. Female trajectories were dominated by osteoporosis and male trajectories by cancer. We found sex important in developing most conditions, especially osteoporosis, chronic obstructive pulmonary disease and diabetes. A socioeconomic gradient was observed where diagnosis postponement time increases with educational attainment. Contrasts in disease portfolio development based on educational attainment were found for both sexes, with chronic obstructive pulmonary disease and diabetes more prevalent at lower education levels, compared to higher.

CONCLUSIONS

Disease trajectories of chronic heart disease diagnosed individuals are heavily complicated by multimorbidity. Therefore, it is essential to consider and study chronic heart disease, taking into account the individuals' entire disease portfolio.