Statistical analysis was carried out with the program Statistical Procedures for Science Society, version 15.0. To test specific hypotheses, we used a subset of the full sample (one subject with SC per family), to avoid bias that could be caused by potential correlations between relatives. The single individual per family with SC was selected using the random number generator function in Microsoft Excel. Since each hypotheses tested was a specific one, based on previous analyses, we considered probability values ≤ 0.05 to be statistically significant. All tests were two-tailed. When a chi-square could not be calculated, a 2 × 2 table was built and its significance was determined by means of the Fisher test, or Yates’ correction test as appropriate.
