For each year I found the OPS, OBP and ISO for the entire league in innings 7-9 and divided that by the corresponding numbers for innings 1-6 (OPSRatio, OBPRatio, ISORatio). Time line charts for each are below. The slope of the trend line for each one is
So maybe the strongest downward trend is for ISO.
Then I divided the 88 year period into eleven 8 year periods (I know, that is arbitrary). The table below shows each ratio for each 8 year period and again, it seems like ISO is the one that has dropped off the most. For example, from 1930-37, ISO in innings 7-9 was 3.2% higher than in innings 1-6. But from 2010-2017, it was 11.7% lower.
Anyway, all three stats have generally been in relative decline in the late innings.
Here are the correlations between the % of batters faced by relievers and the late to early hitting ratios
So, as the % of batter faced by relievers increases, the OPS in innings 7-9 relative to innings 1-6 falls. The relationship is strong since the correlation is -.771. Squaring that (the r-squared) gets about .59, meaning that 59% of the variation across years in OPSratio is explained by changes in the % of batters faced by relievers. The more that relievers appear, the poorer the hitters do in late innings.
Here are the timeline charts