Image by Ave Calvar

Data fusion

Randomised clinical trials and observational studies both have their merits. By combining them, we hope to gain more precise results than with individual analysis of each data set. We have worked on data fusion in the context of the orthopedic APOLLO trial (website, in Dutch).


The focus of the APOLLO trial was on identifying the best surgical approach for performing a hemiarthroplasty in patients with fractured hips. Several Dutch hospitals participated. Patients were randomized to treatment in some of the hospitals. In other hospitals, no randomization took place, but data was still collected in the same way as in the randomized setting.

Randomized data is often considered ideal for causal inference, since many of the conditions for valid causal inference are "automatically" fulfilled. Causal conclusions from observational data require careful discussion of assumptions that can typically not be verified. Nevertheless we'd like to use the observational data as that way, we can draw conclusions about more patients.

In addition to separate analyses of the randomized data and the observational data, we also combined the data to increase power. The results will be published soon. The protocol of our study is available in the references below.

Related paperss

Tol, M. C. J. M., Willigenburg, N. W., Willems, H. C., Gosens, T., Rasker, A., Heetveld, M. J., Schotanus, M. G. M., van Dongen, J. M., Eggen, B., Kormos, M., van der Pas, S. L., van der Vaart, A. W., Poolman, R. W., & APOLLO research group, on behalf of the. (2022). Posterolateral or direct lateral approach for cemented hemiarthroplasty after femoral neck fracture (APOLLO): protocol for a multicenter randomized controlled trial with economic evaluation and natural experiment alongside. Acta Orthopaedica, 93, 732–738. [link]