Effects of phone versus mail survey methods on the measurement of health-related quality of life and emotional and behavioural problems in adolescents