The only way I've seen this addressed is using the idea that late responders are more similar to non-responders than early responders - so in a study where you suspect that non-responders have a different bias to responders you'd use late responses to get an idea about what non-responding people would have answered.