Statistical analysis

We examined normal quantile plots and Shapiro–Wilk statistics to confirm that data followed a normal distribution in all cases. We also computed the interclass correlation coefficients (ICC; ratio of inter‐animal variation to total inter‐ and intra‐animal variation) for all measurements to determine whether observations of different myocytes from the same rat were independent (ICC near 0 indicates no ‘clustering’) or non‐independent (ICC near 1 indicates ‘clustering’). Because, overall, our ICCs ranged from 0 to 0.4, we demonstrated non‐independence for almost all cases and had to account for both inter‐ and intra‐animal variation. Therefore, means were compared using a ‘mixed’ analysis of variance model with both the usual random error effect and a random animal effect to take into account the non‐independence of observations of different myocytes from the same rat. The mean, SD, SEM, confidence interval (CI), and P value were based on the mixed model. P < 0.05 was considered statistically significant.