Read on Twitter
(1/n) Me and my wonderful (ex-)postdoc @yajing_zhu and collaborators Duncan Edwards, Rupert Payne and @prof_mant present a open access paper about #multimorbidity clusters in primary care https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-020-01543-8 @MRC_BSU @PCU_Cambridge @capcbristol
(2/n) This addresses aspects of the Academy of Medical Sciences Multimorbidity report ( https://acmedsci.ac.uk/file-download/39787360) i.e. the need to identify common and burdensome combinations of long term conditions.
(3/n) It also helps to identify the clusters of patients with multiple long term conditions to inform 'cluster medicine' and service planning, see a BMJ editorial by Chris Witty and colleagues (yes, that Chris Witty!) https://www.bmj.com/content/368/bmj.l6964
(4/n) To give context on see 'do replicable profiles of multimorbidity exist?' which unfortunately is behind a paywall ( https://www.ncbi.nlm.nih.gov/pubmed/31624969 ) , but makes the point that few multimorbidity cluster studies cluster patients or discuss clusters from a clinical perspective
(5/n) Ok, so moving on to our paper, a quick tweet about the methods we used. Robust probabilistic clustering (Latent Class Analysis) of patients stratified into different age groups, validated in a held-out split of the data. Clusters compared to outcomes.
(6/n) We used a dataset representative of England, from anonymised GP records (N~100,000). Open source analysis! https://github.com/Kiddle-group/multimorbidityClustersCPRD-GOLD
(7/n) The clusters with the highest mortality comprised psychoactive substance and alcohol misuse (aged 18–64); coronary heart disease, depression and pain (aged 65–84); and coronary heart disease, heart failure and atrial fibrillation (aged 85+).
(8/n) The clusters with the highest service use coincided with those with the highest mortality for people aged over 65. For people aged 18–64, the cluster with the highest service use comprised depression, anxiety and pain.
(9/n) Stand out findings for me are that individuals 18 - 44 years old with alcohol misuse, drug misuse and depression have a mortality risk 18 times higher than their non-multimorbid peers, although this seems highly related to their smoking and socioeconomic deprivation
(10/n) We also see a potential survivor effect, where over half of multimorbid individuals >85 years old belong to the multimorbidity cluster in that age group with the lowest mortality and service use.
(11/n) The biggest implications of this work is for service design, it is common for all multimorbid individuals to put together in one 'bucket' for planning of NHS services, we show the limitations of this approach. Interventions targeting our clusters are also worth exploring.
(12/n) And what is it like to live with multiple health conditions? https://healthtalk.org/multimorbidities/overview Tweeps who may have more to add! @CAM_Paddison @andrzeio @Multimorbidity @BIRMCAM
