Comparison of instrumental variable methods with continuous exposure and binary outcome: A simulation study.

The findings indicate that no panacea is available for the bias associated with IV methods. We suggest using multiple IV methods: one for primary analysis and another for sensitivity analysis.