At the end of this month, in Arlington, Virginia, ClinPage will be attending the inaugural Patient Reported Outcome (PRO) conference, sponsored by the Center for Business Intelligence. We’ve long felt the convergence of technological change and regulatory acceptance of PRO warranted a meeting. We can’t wait to attend.
One of the speakers at the meeting is Michael Hagan, associate director of health economics and outcomes research at Daiichi Sankyo, the #3 Japanese pharma with $7 billion in 2006 revenues. In an earlier job, Hagan was part of the team that helped to generate PRO data for Procrit, or erythropoietin, an injectible drug used to treat anemia. In Virginia, at the PRO conference, he’ll be talking about the biostatistical aspects of the topic. (His doctorate in public health was earned in the statistical methodologies related to quality of life data.)
PRO Data On Procrit
We unearthed the label for Procit and found this sentence: “Statistically significant improvements were demonstrated for most quality of life parameters measured, including energy and activity level, functional ability, sleep and eating behavior, health status, satisfaction with health, sex life, well-being, psychological effect, life satisfaction, and happiness.” Hard to beat that.
Hagan says the PRO science is so new that regulators do not have fixed views on the usage of PRO instruments. “This PRO business is a very, very new area as far as the FDA is concerned,” says Hagan. “There are no carved in stone guidelines. They are pretty much carved in sand.”
So it’s important for sponsors to get the regulatory bodies involved early. “What I would consider to be crucial, to be very important, is having the agency involved at the very beginning,” says Hagan.
Hagan notes that a study with a PRO element may need an adjustment to its sample size. Additional questions, additional endpoints, may require additional statistical power. Says Hagan: “If the PRO is going to be used for labeling, then the study has to be powered to be able to detect a statistically significant difference. Probably your sample size will go up.”
Sample Size Concern
In the real world, Hagan says, there may be ways to mitigate the increase in sample size. “You have to look at conserving your alpha or reducing the number of hypotheses you are testing,” he says. “The other important challenge is missing data,” Hagan says. Statistically, data may be missing completely at random, at random, or not at random. Different strategies are needed for each problem, he says.
Hagan does not have strong feelings, one way or the other, about the new handheld, electronic devices to collect PRO data. Most patients are comfortable with paper and pencil, he says. It may not be the case with a new gadget from Dell or Palm.
Whether electronic diaries are appropriate for a particular project, he says, depends on the technological aptitude of the patient population in question. “If your patient population is comfortable with electronic devices, by all means use them,” he says. “If it becomes a barrier, I’m not in favor of it.”
Hagan has an interesting take on PRO measurements. “There was a school of thought that these were subjective measures,” he says. But for him, PRO measurements are likely to align with what is measured by a central lab or in the clinic by a nurse. “They are paralleling each other,” he says of the PRO and traditional data. “If hemoglobin drops, that means the patient is anemic. With anemia comes certain symptoms the patient can detect. One of them would be fatigue.”
In reflecting on where the industry is going with its usage of PRO data, Hagan says that some of the scientific or clinical knowledge may have to gel a bit before PRO tools can be used on them. Take metabolic syndrome, he says, referring to the combination of diabetes, hyperlipidemia and hypertension. There is no question that products for metabolic syndrome are of great commercial interest.
But devising a PRO trial to examine a therapy for metabolic syndrome, he notes, will be conceptually and statistically complicated. “Is there a metabolic disease instrument?” Hagan asks. “I would hesitate to say yes. There are several instruments out there. Each one has advantages and disadvantages. It becomes a rather challenging proposition. We are not just looking at a point in time—we’re looking at a point in three-D space.”