Data synthesisWe first described the included studies, but could not do meta-analysis because of heterogeneity, and grouped factors affecting place of death following the model by strength of evidence. They were analysed for the direction of the effect (home or hospital) and consistency of findings (number of studies reporting the same effect out of the total number of studies on the topic). We also conducted a sensitivity analysis to determine whether a different grading threshold altered the findings. When relevant we have referred to variables for which there was low strength evidence in the results. We then extracted the odds ratios reported in the papers and plotted them for factors associated with place of death for which there was high strength evidence, and revised the model as necessary.