Although uncontrolled diabetes (UD) or poor glycaemic control is awidespread condition with potentially life‐threatening consequences, there issparse evidence of its effects on health care utilisation. We jointly model thepropensities to consume health care and UD by employing an innovativebivariate latent Markov model that allows for dynamic unobserved heterogene-ity, movements between latent states and the endogeneity of UD. We estimatethe effects of UD on primary and secondary health care consumption using apanel dataset of rich administrative records from Spain and measure UD usinga biomarker. We find that, conditional on time‐varying unobservables,UD does not have a statistically significant direct effect on health care use.Furthermore, individuals appear to move across latent classes and increasetheir propensities to poor glycaemic control and health care use over time.Our results suggest that by ignoring time‐varying unobserved heterogeneityand the endogeneity of UD, the effects of UD on health care utilisation mightbe overestimated and this could lead to biased findings. Our approach revealsheterogeneity in behaviour beyond standard groupings of frequent versusinfrequent users of health care services. We argue that this dynamic latentMarkov approach could be used more widely to model the determinants ofhealth care use.