Table 2.

Measures of Adherence.

Measure	Pros	Cons
Direct
Direct observation	Most accurate	Not feasible in real-world practice
Serum drug levels	Objective measure of recent exposure to drug	Can be manipulated; acceptable ranges often unknown; assays not widely available
Indirect
Pill counts	Inexpensive	Difficult in real-world practice; easy to manipulate; may overestimate adherence; demeaning
MEMS (microelectronic event monitoring system)	Accurate data on when one opens the bottle; may be combined with reminder systems	Not easily feasible in real-world practice; expensive
Refill records	Objective higher level data; good for research purposes	Report fill rate and not actual intake; impractical for daily use
Biomarkers	May be important intermediaries to outcomes (e.g., hypertension with TKI use)	Few developed and validated
Outcomes	Most important variable	Difficult to discern nuances of adherence outside of clear extremes
Indirect and Subjective
Self-report	Quick; can use past validated instruments; does not require clinician time	Subject to significant bias such as the Hawthorne effect and overestimates adherence
Assessment by others	Inexpensive; allows for a dialogue	Hawthorne effect; time consuming
Diaries	Inexpensive; actively involves the patient	Subject to manipulation; demeaning; time consuming