fig9

Figure 9. Dispersion of data around the hair mass index abacus in patterned hair loss (PHL). The dispersion of healthy data is such that females with no pattern (NoP), LI and HI and males with HI-HII cannot be distinguished statistically significantly from control averages in terms of either coverage or hair productivity (TTCC). In female pattern hair loss (FPHL), dispersion of data around the averages, i.e., standard deviation (along both axes), shows very little overlap between SCS and TTCC (LII and LIII). When the clinician was uncertain and classified patients as “Other”, the abacus confirms an intermediate position (dispersion not shown). Clinical classification makes sense and SCS regressed linearly from LI to LIII with “Other” hair loss in between due to less marked patterning. The regression line equation in females y = 107.58-0.7336 (R² = 0.94) implies that analytical and global methods work well together. Similarly, in males, y = 105.31-0.71 (R² = 0.99), time to complete coverage also worsened as severity increased, but there is a substantial overlap between HIII and HIV. The asymmetrical spread of data around the averages along the two axes deserves a clinically relevant comment. Graphs clearly show that, instead of circles (equal dispersion of SD on x and y axes), we found ellipses (shown as paler fields around the averages). Wider dispersion along the x-axis generates asymmetry, suggesting that the region of interest (ROI) was sometimes more and sometimes less affected by hair regression with the ROI more remote from or closer to the epicentre. As SCS focussed on maximum severity with parted hair along the midline, the relative dispersion is less as compared with time to complete coverage. These findings suggest that the abacus makes sense and might be considered for clinical use.