1. Include at least three ordinal scales on a follow-up guide except in dire straits, and even then have at least two scales.
2. Have at least three levels filled-in for each scale. One filled-in level should be the expected level, and there should be at least one filled-in level on each side and the expected level.
3. Include only one problem or variable on each scale.
4. If any of the scales on a follow-up guide are weighted, all scales should be weighted.
5. The client's behavior at intake may be equivalent to any of the five levels. Avoid terms like "better than when treatment began."
6. Avoid variables which are too general or vague to be accurately scored at a follow-up interview.
7. The levels on a scale should not overlap each other.
8. If the information needed for a scale's follow-up scoring is to be obtained from a source other than the client, the special source of information should be listed.
9. There should be no "expected" levels which are so high that there is no possible "better than expected" level. There should be no "expected" goals which are so low that there is no possible "less than expected" levels.  There should usually be a possible outcome for all levels of the scale even if some are left blank.
10. Avoid having two blank levels adjacent to each other on a scale.

I asked for the citation of these.  No one knows!!!