The FDA concept paper asks companies to select and justify their choice of tools. To do so, companies need to develop a conceptual
model for how their drugs are used and what system failures may lead to product misuse. In addition to relying on a systems
analysis, using a behavioral model of product use—how beliefs, motives, and situational constraints influence the way a drug
is used—can help determine a coordinated set of tools and specify core messages that must be communicated or systems that
must be implemented to define the elements of the RMP.
A good starting point for developing a drug-use model is to identify the various steps necessary to use a medication properly.
Failure Mode and Effects Analysis (FMEA) is a systematic analysis of how failures may occur in any system. It identifies steps
or elements where mistakes may occur, before the system is implemented.
FMEA is a complement of root-cause analysis, which is aimed at identifying the source of a system's failure once a mistake
is identified. An effective FMEA identifies corrective actions required to prevent failures and to assure the highest possible
To undertake an FMEA, pharma companies must identify potential failure modes. This action may be broken down into subprocesses,
with each step in the system being considered a separate element with the potential for failure. Each of the postulated steps
where failure might occur should be assigned a severity value, a probability that a given effect might occur, and a likelihood
value that a user may detect (and correct) the problem. Recommended actions are then developed to increase detectability,
to reduce the probability, and, most important, to decrease the severity of harm. Priority is given to problems with the highest
index of probability and severity.
Although it was developed as a method to improve product quality, an FMEA can only be as good as the quality of the system
description. Specifying steps too broadly can miss important elements. For example, after consultants performed a system analysis
for a medications error problem in which pharmacists were dispensing the incorrect medication to patients, they found that
the pharmacists were dispensing the drug that physicians had identified.
Although it is usually assumed that medication errors are caused by pharmacists' difficulty reading physicians' handwriting,
in this case the problem was that physicians recalled the incorrect drug name from memory when they wrote the prescription.
If they had used an FMEA that did not specify the need for the physician to recall the correct drug name from memory, the
consultants would have missed an important source of error.
The Human Factor
In addition, to specify the subprocesses, pharma companies and their RMP crafters need to understand the system from the perspective
of the individuals who are performing the tasks involved. The relationship between knowledge and behavior is not direct or
simple. People often know what they should do but fail to behave in a fashion that is consistent with that knowledge. To develop
a predictive model, it is necessary to understand the:
- full set of beliefs underlying behavioral intentions
- motivations that support or stand in the way of exhibiting desired behavior
- environmental conditions that facilitate or place barriers to compliance.
There are a variety of psychological and health behavior models that can be used to organize these influences. Some models
may improve the processing of presented information, for example, by improving patients' involvement (personal relevance)
or competency (self-efficacy) with the information or advocated behavior.
Some models may help RMP crafters understand the processes underlying the choices among alternative courses of behavior (behavioral
decision making). Other models may help structure advocated behaviors into a series of stages, permitting messages that seek
to move patients and physicians through necessary stages to attain behavioral compliance (precaution adoption or stage models).
Still, other models attempt to motivate compliance through emotion (positive affect or fear appeals) or through highlighting
desired outcomes (approach or avoidance goals).
Determining which model to use depends on the particular problem addressed. If the RMP advocates complex behaviors (such as
avoiding drug dependency), it may be necessary to move respondents through a series of stages to overcome situational barriers.
If the plan calls for simpler behaviors (such as standing upright when taking tablets), offering a strong, even emotional,
rationale for compliance and a reminder system might provide the best model to influence behavior. Diagnosing the behavioral
problem and selecting (or custom building) the correct behavioral model can produce a clear method for the design of the RMP.
Because companies are dealing with programs that are expected to reach and influence the vast majority of participants, it
may be difficult to anticipate the full range of issues influencing the behavior of all patients and healthcare professionals.
Research studies that provide insights into the issues are invaluable. Identifying particular at-risk patient segments can
also help companies design program interventions targeted at particular failure modes for a specified group of individuals.