Evidence synthesis for count distributions based on heterogeneous and incomplete aggregated data

Abstract

The analysis of count data is commonly done using Poisson models. Negative binomial models are a straightforward and readily motivated generalization for the case of overdispersed data, that is, when the observed variance is greater than expected under a Poissonian model. Rate and overdispersion parameters then need to be considered jointly, which in general is not trivial. Here, we are concerned with evidence synthesis in the case where the reporting of data is rather heterogeneous, that is, events are reported either in terms of mean event counts, the proportion of event-free patients, or rate estimates and standard errors. Either figure carries some information about the relevant parameters, and it is the joint modeling that allows for coherent inference on the parameters of interest. The methods are motivated and illustrated by a systematic review in chronic obstructive pulmonary disease.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 10 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.