Latent class analysis suggests four classes of persons with type 2 diabetes mellitus based on complications and comorbidities in Tianjin, China: a cross-sectional analysis

Characterizing Multimorbidity from Type 2 Diabetes

“…Women have a proportionally greater burden of multimorbidity than their male counterparts in some studies, 24 , 29 , 31 , 33 although the magnitude of this association is modest and inconsistent. A greater proportion of women presented with 4+ co-occurring comorbidities than men (18% vs 15.9%), 31 with postmenopausal women experiencing the greatest risk of developing further comorbidities.…”

“…The final 12 studies included came from varying HIC settings except for 2 conducted in China 28 , 29 and adopted a range of different methods that varied in analytical statistical depth ( Table 1 ). From the sample, the earliest study was conducted in 2012, 30 with most of the remaining studies produced after 2015.…”

“…of comorbidities and probabilities of having at least 1 comorbidity by age/sex/deprivation OR of having each of 51 chronic comorbidities with T2DM, and morbidity clusters of concurrently occurring conditions Comorbidity in Adult Patients Hospitalized with Type 2 Diabetes in Northeast China: An Analysis of Hospital Discharge Data from 2002 to 2013 Chen et al, 28 2016 Longitudinal retrospective cohort study; hierarchical clustering China 4,400,892 To evaluate comorbidity burden and patterns among hospitalized T2DM patients EHR database Prevalence of comorbidities by age/sex ACoR and RCoR of comorbidities co-occurring with T2DM. Clusters of comorbidities based on RCoRs Latent class analysis suggests 4 classes of persons with type 2 diabetes mellitus based on complications and comorbidities in Tianjin, China: a cross-sectional analysis Gao et al, 29 2017 Cross-sectional study; latent class analysis, multinomial logistic regression China 5500 To classify persons with T2DM based on complications and comorbidities Structured questionnaire of patients across 10 hospitals Predicted probabilities of T2DM complications and comorbidities, to produce 4 classes OR of the association between class membership and various demographic factors, diabetes severity, and behavioral factors Comorbidity network for chronic disease: a novel approach to understand type 2 diabetes progression Khan et al, 34 2018 Health informatics study; graph theory and social network analysis Australia 749,000 To understand the comorbidity pattern of T2DM and develop a research framework to model chronic disease progression in terms of comorbidity Administrative private health care dataset with ICD-10 diagnosis codes on hospital admission and discharge data Modularity score and network node strength to produce comorbidity network of T2DM The comorbidity burden of type 2 diabetes mellitus: patterns, clusters, and predictions from a large English primary care cohort Nowakowska et al, 24 2019 Longitudinal retrospective cohort study; agglomerative hierarchical clustering England 102,394 To quantify comorbidity patterns in people with T2DM, to estimate the prevalence of 6 chronic conditions in 2027 and to identify clusters of similar conditions UK Clinical Practice Research Datalink health dataset linked with Index of Multiple Deprivation data Crude and age-standardized prevalence of comorbidities, and their clusters at T2DM diagnosis and 2, 5, 9 y after diagnosis, by sex and deprivation Differential Health Care Use, Diabetes-Related Complications, and Mortality Among Five...…”

“…As the format of the study results was heterogenous, owing to the varied methods used, the authors have summarized studies by 2 broad categories: (a) those employing combination groupings of co-occurring conditions (ie, pairs and triads) to define multimorbidity ( Table 2 ); and (b) those studies of more complex statistical clustering methods ( Table 3 ). The key clustering approaches used in the reviewed studies included agglomerative 24 , 31 and divisive 32 hierarchical clustering, 28 latent class analysis, 29 , 33 and graph theory, 34 all of which require substantially large health data sets. Moreover, novel machine learning approaches have been used to characterize and learn multimorbidity patterns, 35 , 36 but these methods are yet to be applied to T2DM specifically.…”

“…Major T2DM-related comorbidities were defined as having both ACoR >1% and RCoR >1 27 major comorbidities clustered into 3 categories with high, medium, and low RCoRs with T2DM High RCoR group: Essential hypertension: peripheral & visceral atherosclerosis Disorders of lipid metabolism: chronic renal failure Occlusion or stenosis of precerebral artery: urinary tract infections Other endocrine disorders: coronary atherosclerosis Other cerebrovascular diseases: acute myocardial infarction Other nutritional, endocrine, & metabolic disorders Medium RCoR group Acute cerebrovascular disease: skin & subcutaneous tissue infection Other liver diseases: fluid & electrolyte disorders Conduction disorders: other upper respiratory infections Other lower respiratory disease: pneumonia Congestive heart failure: transient cerebral ischemia Acute bronchitis Low RCoR group: Cardiac dysrhythmias: hyperplasia of prostate Noninfectious gastroenteritis: other eye disorders Thyroid disorders: biliary tract disease Other nervous system disorders: cataract Overall, essential hypertension was the most common comorbidity with ACoR of 58.4%, whereas peripheral and visceral atherosclerosis had the strongest association with T2DM with RCoR of 4.206 Dyslipidemia was the second most severe comorbidity with RCoR of 3.447 (median RCoR: men [2.779], women [2.099]). Women had higher co-occurrence of chronic renal failure (RCoR 2.461) than men (RCoR 2.155) Older patients (aged 60–69) had the highest prevalence of comorbidities with hypertension (32.7%), CHD (28.5%), acute CVD (25.9%) being top conditions Latent class analysis suggests 4 classes of persons with type 2 diabetes mellitus based on complications and comorbidities in Tianjin, China: a cross-sectional analysis Gao et al, 29 2017 LCA, multinomial logistic regression Class 1: complications & comorbidities group (6.1%); high conditional probability of suffering from all complications and comorbidities. Highest age, family history, central obesity, and suburban residence Class 2: high risk of complications group (25.7%); diabetic peripheral neuropathy, retinopathy, and lower-limb vascular disease Class 3: high risk of comorbidities and CVD group (14.3%): hypertension, dyslipidemia and metabolic syndrome, and CVD Class 4: diabetes without complications & comorbidities group (53.9%): low conditional probabilities for all complications.…”