Putting a Number on Pain
Will new technologies substantially change the way subjective complaints are measured in clinical trials, and, if so, by how much? Depending on the expert consulted, the answer ranges from a little to a lot.
For decades, clinical trials that include so-called “soft symptoms,” such as pain, fatigue, nausea, dizziness, and feelings of depression, have relied on patient-reported estimations, usually using scales in which a patient gives his or her symptom a number between one and ten to describe its severity. This system has helped to determine the efficacy and, in some cases, the side effects of drugs and other treatment options.
Questions have always surrounded patient self-reports: Do patients understand how to rank their symptoms? Can numerical scoring systems accurately describe something as amorphous as pain or fatigue? How variable are patients’ rankings of the same symptom level? Are patients always honest?
“There is a sense among scientists that somehow what patients report might be less accurate than something that can be objectively measured. If I measure blood pressure using a blood pressure cuff, for instance, some people might have more confidence that it’s ‘real’ because the patient is not having an effect on it,” says John T. Farrar, M.D., Ph.D., and associate professor of epidemiology, neurology, and anesthesia at the University of Pennsylvania, who has been involved in several efforts to standardize pain measurements in clinical trials. He notes, however, that blood pressure is not perfect either. “If I measure it right when you come into my office, it might be slightly higher because you just ran up the stairs or you’re a little anxious, but if I let you sit for ten minutes, then your blood pressure might be a bit lower. My point is that even with an objective measure like blood pressure, it is variable.”
Nonetheless, there are things that can and should be done— even before considering technological advances—to make subjective measures as accurate as possible, Farrar says. One is to take multiple readings that show an individual patient’s trend with regard to the symptom. While one person’s pain level of five may not be the same as another person’s, self-reports from each patient over time will indicate changes in pain level. “We may not be able to understand exactly what one person’s level means in terms of comparing it to others, but self-reports do show how symptoms change for each patient, and we can use that information to test whether treatments are working or not,” he explains.
Tech with a Smaller Bite
Subjective measures can be made more reliable in many ways, and technology can help with some of these. Farrar suggests asking the patient to record his or her symptom at the same time of day. “When measuring fatigue, for instance, it could be that you didn’t sleep well because you have a bad bed so you’re feeling very tired early in the day, but you are feeling better by the middle of the morning after you’ve had your cup of coffee and you’re at work. You can see how it would make a difference when you measure,” he says. For this, a smart phone or Internet-connected software can be a good tool. Farrar says he often asks patients to file a self-report before bed, providing an assessment of overall symptom severity for the day.
Studies have also shown that such reporting technology is a better option than the paper diaries that trials have often utilized in the past. “A number of articles show that patients often forget to fill in paper diaries, so they’ll drive into the clinic for an appointment and hurriedly fill in the missed days in the parking lot,” notes Farrar. This, of course, affects the accuracy of their self-reports. “The way around that is to have the patient use an app or other technology that records the time that they actually enter the data so you’ll know if they missed a particular measurement. That has made a difference.”
The use of tablets and smartphones necessitates another change to subjective measurements: the questions for patients have to be short so they are legible on a small screen, but they also have to be clear. Fortunately, this is something experts have already started to address. For instance, researchers involved in the U.S. National Institutes of Health-funded Patient Reported Outcomes Measurement Information System (PROMIS) have been working on standardizing approaches to measure patient-reported health status, and this includes making recommendations about the survey questions, according to Dave Cella, Ph.D., chair of the PROMIS steering committee and a professor in and chair of the Department of Medical Social Sciences at Northwestern University’s Feinberg School of Medicine.
“In PROMIS, we try to work with very simple framing and very simple wording so that people with just about any reading level can read and understand what the question getting at. This does end up helping when you migrate to electronic technologies such as web-based testing or apps on devices like an iPad or a smartphone, where their small size affects the ability to read and how long the question can be,” Cella explains.
PROMIS and other initiatives—including the Analgesic, Anesthetic, and Addiction Clinical Trial Translations, Innovations, Opportunities, and Networks (ACTTION) and the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT)—are also working to standardize subjective measurements in clinical trials.
Such standardization is critical to the reliability of measurements, says Farrar, because soft symptoms will always be subjective. “Just to be very clear, there is no technology that will turn a subjective question such as ‘How much pain are you having?’ into an objective measurement like blood pressure. They will continue to be subjective, but technologies will offer improvements in the way we go about getting those self-reports.”
Tech with a Bigger Bite
Even though technologies aren’t available yet to quantify currently subjective measures, they are coming, and they will be a prominent factor in future clinical trials, according to many researchers and developers.
“Let me first say that there has been a lot of mathematical work and rigor that shows you can get a reasonable set of outcomes based on asking patients questions in certain ways, but we’ve known in the industry for decades that it is not 100% reliable because these are subjective measures,” says Glen de Vries, president and cofounder of New York-based company Medidata . Medidata has developed a cloud-based platform that pharmaceutical and biotechnology companies, as well as academic researchers and clinicians, are using to plug in and run their clinical trials. While de Vries applauds the efforts in collecting and analyzing patient self-reports of symptoms, he believes the time has come to bring technology to bear.
Using a fictitious example, he explains the kinds of contributions that he contends technology—and specifically continuous biosensor monitoring—can make. “Let’s say you have two pharmaceutical companies, each of which is working on a drug for a particular type of cancer,” he begins. Both are effective at reducing the size of a tumor. To determine which has fewer side effects, De Vries envisions a biosensor that tracks patient movements.
“Now, I don’t know whether the patients’ movements are associated with them going to work, going away on vacation, or visiting family members, but I do know they’re out doing something that has socioeconomic value,” he says. From the biosensor data in this example, the researchers may find that those patients using drug A covered on average tens of square meters a day, while those using drug B covered thousands of square meters in the same period. That, de Vries believes, could be extrapolated to indicate that the patients on drug B are experiencing fewer side effects and, therefore, are continuing with their normal lives, while those on drug A are not. “In a traditional clinical-trial world, I couldn’t give you a good answer about whether you should take drug A or drug B, but, in this new environment, I can absolutely tell you which is the better drug, and it’s definitely drug B.”
The scientific literature is already demonstrating that biosensors can pick up soft symptoms , says Karl Friedl, Ph.D., retired director of the Telemedicine and Advanced Technology Research Center for the U.S. Army Medical Research and Material Command in Frederick, Maryland, and now adjunct professor of neurology at the University of California, San Francisco. He points to research that links a combination of voice and other signals with the onset of depression, and to work on Parkinson’s disease that suggests that gait and movement analyses may be associated with changing emotional status or the onset of dementia. Even analysis of speech content has been shown to indicate cognitive decline, he says. “A recent paper analyzed the content and complexity of word choices in speeches by Ronald Reagan and George W. Bush over their two terms in office, and, through the computing power we have now, it was able to show a straight-line decline in Reagan versus a flat line for Bush.” In Friedl’s view, these types of biosensor and analytical technologies have the potential both to detect health conditions and to allow early interventions that can help treat them.
Such approaches are coming, agrees Jochen Klucken, M.D., vice head and assistant professor of neurology at the University of Erlangen–Nuremberg in Germany, who is developing biosensors for Parkinson’s disease. As the work on new objective methods continues, other smaller but no less important changes to clinical trials will be implemented, he says. These include standardizing the characterization of symptoms that physicians make. For example, doctors often rate patients’ movement characteristics and estimate capabilities on a numbered scale. Discrepancies among doctors’ ratings—so-called inter-rater variability—are inherent in such subjective measures. On the other hand, a biosensor designed to key in on that particular characterization could be used for all the patients participating in a particular clinical trial, thereby eliminating that variability, Klucken notes.
Similarly, technological advances can allow motor impairment to be objectively measured. Klucken and his research group are currently studying gait sensor data to find subtle biomechanical changes that are tied to increasing gait impairment in people who have Parkinson’s disease. “Today, the standard of care only measures the total distance that a person can walk, but it doesn’t measure what happens while the patient is walking, such as shorter and shorter stride length,” he explains. This is not only useful in clinical trials, but also important to patients whose quality of life is directly related to their mobility, and to patient care overall.
A wide range of advances like these is already on the drawing board, many are in development, and, in just the past year, some are already being employed in clinical trials, De Vries notes. “We are just now seeing people begin to use these technologies in the kinds of rigorous regulatory and scientific environments that are a part of clinical trials. That said, and based on how prevalent these ideas are, I think [their integration into trials] is going to happen much faster than anybody thinks in terms of the life sciences industry.” Drug developers, in particular, should start adding wearable data to their clinical trials now to stay ahead of—or at least in step with—the market. De Vries argues that those who wait until the U.S. Food and Drug Administration begins approving drugs based on wearable data will be at least a decade behind their competitors.
While technologies are picking up speed, considerable work remains, says Klucken. “Honestly, there are a lot of tests and ideas and thoughts out there right now, and there are already some companies that provide medical products, but the gold standard really in terms of instrumented motion analysis or instrumented sensors using wearable technology is not quite there yet.” He adds, “We need to do our homework a little more, and clinicians and technicians need to communicate more, but this area is really evolving and developing, and, in the end, we will get constant, continuous data and information from the everyday lives of patients that will change clinical trials, as well as the way we treat patients overall and how we do health care.”
- PROMIS. [Online]. (accessed 1 Feb. 2016).
- Medidata. [Online]. (accessed 1 Feb. 2016).
- J. A. Stamford, P. N. Schmidt, and K. E. Friedl, “What engineering technology could do for quality of life in Parkinson’s disease: A review of current needs and opportunities,” IEEE J. Biomed. Health Inform., vol. 19, no. 6, pp. 1862–1872, Nov. 2015.