Selection bias and misclassification in case-control studies conducted using the National Violent Death Reporting System