| ObjectivesThe multistate competing risks model was utilized to explore how the hazard rate of gout and type 2 diabetes mellitus would alter over time when the type 2 diabetes and gout were in independent state or when they were in the combined state. We intended to show the advantages of multistate competing risks models via the comparison of logistic regression, Kaplan-Meier survival analysis. We found the multistate competing risks models distinguished itself in longitudinal data with multi-outcomes. Thus our finding can provide a reliable and advisible reference for other similar problems. Methods1. Stratified cluster sampling was used to obtain the original cohort sample population, then established the historical cohort about type 2 diabetes and gout onset on the basis of the retrospective survey of community permanent residents of Guangzhou, and other health record complements2. A retrospective survey was conducted to the permanent resicents of Guangzhou. Established a historical cohort study about the type 2 diabetes and the the onset of gout using the health records and other information.3. Using multivariate logistic regression analysis, analysed the impact of T2 DM and gout on each other when they altered in the gradual alternative pattern as covariates4. Utilising the Kaplan-Meier survivial analysis to evaluate the cumulative incidence function, compared the cumulative incidence curves of gout in separate situations including the health statusâ†'gout versus health statusâ†'T2DM and T2DMâ†'T2DM + gout versus goutâ†'T2DM + gout.5. Multistate competing risks model was introduced to estimate cumulative hazard and the transition probabilities of 4 separate situations, that is the health status, T2 DM, gout, T2 DM + gout. Then the parametric regression models were applied to fit the cumulative hazard, compare them once we determine the parameters of transition hazard function,ResultsMultistate competing risks model combined with the logistic cumulative distribution model behaved better in fitting cumulative hazard function and get better results. The comparison of the function and the peak value of the four stages showed that: T2 DM and gout work synergistically in early stage while in the later stage they showed competing effect. The four functions displayed gout accelerated the hazard rate the T2 DM before 60 years old; afterwards the hazard rate of T2 DM descended due to the competitive effect of gout 10 years ahead of schedual. While at the same time T2 DM can accelerate the hazard rate the gout before 68 years old; afterwards the hazard rate of gout droped due to the competitive effect of T2 DM 11 years ahead of expectation. They manifested accelerating descending after 68 years old. Multistate competing risks model was compared with logistic regression model and Kaplan-Meier survivial analysis separately, then we found both logistic regression model and the Kaplan-Meier survivial analysis have their critical drawbacks. That is the logistic regression model can only reflect the relationship of the two disease and the Kaplan-Meier survivial analysis can neither estimate the competiting risk exactly nor conduct multistate survivial data. ConclusionThe Multistate competing risks model analysis showed the synergistic effect of T2 DM and the gout in the early stage whilst in the later stage they manifested competitive effect. The preexist of T2 DM would accelerate the onset of gout 10 years ahead of schedual and vice versa. The Multistate competing risks model analysis can analyze the transition possibilities of different outcomes according to the time line and it had its irreplaceable advantages in the discovery of relationship and alternation on interaction of different outcomes. |
PDF Full Text Request