I have knit many, many of these. I usually just cast on the increases in that is called for in that row when I cast on the sweater's first row, and then I don't worry about the increases for that row when I get to it. (Casting on makes the sleeves wider right at the elbow, and since I like to lengthen the sleeves, it looks better.)

You will be increasing 9 stitches on both of the side sections(which are the sleeves), not the back.

You continue to do the decreases every other row in the same spot just like you have been doing, even with the increase of 9 stitches evenly spaced on either side. I just do the math, divide by 9 on each side of the back, and after each group, do a backwards loop increase. Then when I knit the next row, I knit into the BACK of the cast on stitch, which will make the hole that develops from the increase less obvious.

Is this what you need to know?

I need to proof what I write! In the statement above, I tried to explain that I usually cast on the 18 stitches called for in the increase row when I cast on the sweater. This will give me a nice sleeve width at the elbow. The sleeves are elbow length, and I like full length, so I just pick up those stitches when I am finished knitting and add a few inches and a ribbed cuff. The picked up row is not as obvious if the increases are cast on at the beginning.

I'm sending hugs and kisses to everyone who helped!