Research Square · Oct 28, 2026
Sixteen-year evolution of multimorbidity patterns and health poverty vulnerability among rural households in western China: a clustering and hidden Markov model study
By Zhang F, Chang J, Li R et al.
Background Since the 2009 health reform,the prevalence of multimorbidity has risen among rural households in the southern mountainous region of Ningxia,western China,yet how multimorbidity patterns evolve over time and how household health poverty vulnerability has changed alongside them are not well described. Methods We used six waves of household survey data(2009–2024)from four counties(Haiyuan,Yanchi,Xiji,and Pengyang).Health poverty vulnerability was measured by three-stage feasible generalized least squares(FGLS)under the vulnerability-as-expected-poverty framework;multimorbidity patterns were identified with fuzzy c-means(FCM) clustering;transitions between patterns were modeled with a hidden Markov model(HMM), including models stratified by household-head age;and county-level differences were shown with maps. Results Multimorbidity prevalence increased from 7.52%to 24.50%over the study period,while the proportion of multimorbidity-affected households vulnerable to health poverty fell from 39.0%to 22.2%(lowest point 20.0%,in 2019).FCM clustering identified five patterns:cardiovascular–digestive(CCM),cardiovascular–metabolic(CMM),cardiovascular–musculoskeletal(CBM),digestive–musculoskeletal(DSM),and simple cardiovascular(SCM).The HMM suggested a progression from DSM toward SCM and CMM;SCM was the most stable of the three states and DSM showed divergent transitions.Stratified by the age of the household head,three stages emerged:among households headed by young and middle-aged adults,digestive patterns shifted toward cardiovascular patterns;among those headed by younger seniors,patterns converged on cardiovascular patterns;and among those headed by older seniors,SCM remained dominant while cardiovascular-respiratory multimorbidity emerged.The four counties differed markedly:Haiyuan had the highest T4 prevalence with patterns shifting from DSM through SCM to CMM;Yanchi had the lowest vulnerability throughout;Xiji had the largest prevalence rise and the highest vulnerability with recurrent CBM;and in Pengyang,vulnerability in 2024 exceeded that in 2009. Conclusions Multimorbidity burden grew over the 16 years,but health poverty vulnerability declined.Patterns followed a consistent progression,with DSM as a possible window for early intervention;prevention could shift toward pattern-oriented,age-stratified strategies with county-differentiated resource allocation.