Introduction

Persistence is broadly defined as continuing a course of therapy when it is indicated that one should do so. It has also been referred to as "medication adherence", "medication compliance" or (in the United Kingdom) "medication concordance", the extent to which a person's behaviour coincides with medical advice with respect to medication. Most commonly the rate of persisting with a pharmaceutical treatment has been described with chronic medications (e.g., lipid-lowering drugs or statins, anti-diabetic agents and hormone replacement therapy).

The following research question incorporates the concept of persistence of treatment:

Of the women starting treatment for osteoporosis [new users] after hip fracture, what proportion are still "on drug after one year, two years, etc.? How long, on average, do women continue to take the drug after the initial dispensation?"

The definition of persistence represents a combination of the number of prescriptions dispensed consecutively and medication in possession for a prescribed period of time. Technically, persistence is an application of the duration of use measures, as it describes the number of days from the date of the start of the index dispensation (index date) or "new use" to the time of first failure to continue renewals of the drug with a permissible gap. Many researchers have described this gap -or the point at which an individual is no longer "persistent" - to be x number of days between the end of one dispensation and the beginning of the next. MCHP has used a continuing compliance rate [using MED_TOTAL] of 75% and 80%; others have used a rate of 66%.

Analysis of persistence usually includes at least two types of measures:


1. Cumulative treatment persistence rate : A Kaplan-Meier failure time analysis in which data are censored at the end of observation (say the end of a fiscal year).
   
   
2. Likelihood of persistence : Odds ratios calculated through unconditional logistic regression including potential determinants of persistence (depends on drug): age, sex, number of different drugs (excluding the study drug), number of physician visits and number of hospitalizations.

Calculating Persistence

Individuals' prescriptions (for a drug) are calculated one at a time and as a "running" value using MED_TOTAL (see description of MED_TOTAL in the Duration of Use concept). If MED_TOTAL = 0.75 (i.e., 75% compliant) the next prescription is added, continuing until the MED_TOTAL drops below 0.75. When MED_TOTAL drops below 0.75 the next prescription is checked to see if it is brought down to or above 0.75. If MED_TOTAL continues to be = 0.75, we say that the person persisted until the prescription ceased to be within the compliant (implied, therapeutic) range. The dispensing date (DATEPRVD) for the prescription where MED_TOTAL fell below 0.75 would be considered the end date for the purposes of determining the cumulative persistence rate and likelihood of persistence (given potential determinants). For individuals persistent until the end of the study period, the end of the study period date is the end date. Anyone who dropped out of the study due to death or leaving the province would be flagged as being lost to follow up. Their death date or registration cancel date would be considered the end date.

The following is usually reported:


1. number of individuals having only one prescription of the study drug dispensed (these are usually eliminated)
   
2. number of individuals persisting but stopping before the study period ended
   
3. number of individuals persisting but subsequently, "lost to follow up" (end of study period / death / out of province).

MORE INFORMATION

Haynes R, McDonald H, Garg AX. Interventions for helping patients to follow prescriptions for medications [ Cochrane Review ]. Oxford, England: Cochrane Library, Update Software Issue 2, 2002. [ abstract ]