Gravar-mail: Confounding by indication in non-experimental evaluation of vaccine effectiveness: the example of prevention of influenza complications