Identifying subgroups of complex patients with cluster analysis

Institute for Health Research, Kaiser Permanente Colorado, Denver, USA.
The American journal of managed care (Impact Factor: 2.17). 08/2011; 17(8):e324-32.
Source: PubMed

ABSTRACT To illustrate the use of cluster analysis for identifying sub-populations of complex patients who may benefit from targeted care management strategies.
Retrospective cohort analysis.
We identified a cohort of adult members of an integrated health maintenance organization who had 2 or more of 17 common chronic medical conditions and were categorized in the top 20% of total cost of care for 2 consecutive years (n = 15,480). We used agglomerative hierarchical clustering methods to identify clinically relevant subgroups based on groupings of coexisting conditions. Ward's minimum variance algorithm provided the most parsimonious solution.
Ward's algorithm identified 10 clinically relevant clusters grouped around single or multiple "anchoring conditions." The clusters revealed distinct groups of patients including: coexisting chronic pain and mental illness, obesity and mental illness, frail elderly, cancer, specific surgical procedures, cardiac disease, chronic lung disease, gastrointestinal bleeding, diabetes, and renal disease. These conditions co-occurred with multiple other chronic conditions. Mental health diagnoses were prevalent (range 28% to 100%) in all clusters.
Data mining procedures such as cluster analysis can be used to identify discrete groups of patients with specific combinations of comorbid conditions. These clusters suggest the need for a range of care management strategies. Although several of our clusters lend themselves to existing care and disease management protocols, care management for other subgroups is less well-defined. Cluster analysis methods can be leveraged to develop targeted care management interventions designed to improve health outcomes.

  • [Show abstract] [Hide abstract]
    ABSTRACT: To determine patterns of co-occurring diseases in older adults and the extent to which these patterns vary between the young-old and the old-old. Observational study. Department of Veterans Affairs. Veterans aged 65 years and older (1.9 million male, mean age 76 ± 7; 39,000 female, mean age 77 ± 8) with two or more visits to Department of Veterans Affairs (VA) or Medicare settings in 2007 and 2008. The presence of 23 common conditions was assessed using hospital discharge diagnoses and outpatient encounter diagnoses from the VA and Medicare. The mean number of chronic conditions (out of 23 possible) was 5.5 ± 2.6 for men and 5.1 ± 2.6 for women. The prevalence of most conditions increased with advancing age, although diabetes mellitus and hyperlipidemia were 11% to 13% less prevalent in men and women aged 85 and older than in those aged 65 to 74 (P < .001 for each). In men, the most common three-way combination of conditions was hypertension, hyperlipidemia, and coronary heart disease, which together were present in 37% of men. For women, the most common combination was hypertension, hyperlipidemia, and arthritis, which co-occurred in 25% of women. Reflecting their high population prevalence, hypertension and hyperlipidemia were both present in 9 of the 15 most common three-way disease combinations in men and in 11 of the 15 most common combinations in women. The prevalence of many disease combinations varied substantially between young-old and old-old adults. Specific combinations of diseases are highly prevalent in older adults and inform the development of guidelines that account for the simultaneous presence of multiple chronic conditions.
    Journal of the American Geriatrics Society 10/2012; 60(10):1872-80. DOI:10.1111/j.1532-5415.2012.04158.x · 4.22 Impact Factor
  • Source
    10/2012, Degree: PhD, Supervisor: Elisabet Welin Henriksson
  • [Show abstract] [Hide abstract]
    ABSTRACT: Around 5 to 25% of lung cancer worldwide occurs in lifelong non-smokers (less than 100 cigarettes in lifetime). Lung cancer in never smokers (LCINS) shows many clinical, epidemiological and molecular differences compared to those related to tobacco. It is therefore often considered as a separate entity. LCINS is also a good model for the study of lung cancer risk factors and tumoral mutation profiles (usually more common and specific). However, most data has come from retrospective studies and/or from Asian populations, although this disease shows high geographic lability. The BioCAST/IFCT-1002 is a national, multicentric, prospective study promoted by the French intergroup IFCT. The first objective is to describe the clinical and molecular epidemiology of LCINS in a French population. Detailed data (including exposure to many risk factors) are collected directly from the patient through a standardized questionnaire completed during a telephone interview. All patients also undergo blood sampling for the analysis of genomic polymorphisms and the characterization of epigenetic anomalies. BioCAST hopes to provide concrete answers for clinicians and patients about this entity.
    Revue des Maladies Respiratoires 09/2013; 30(7):576-83. DOI:10.1016/j.rmr.2013.03.006 · 0.49 Impact Factor
Show more