Washington, DC—How to overcome practical challenges in quality assessment metrics and data collection was the focus of a panel discussion at the Fifth Annual Conference of the Association for Value-Based Cancer Care.
Two key messages gleaned from the discussion are that in the short-term, data silos make it necessary to prioritize process measures rather than true outcomes measures; however, in the longer-term, retrospective analyses of big data should more effectively differentiate between different approaches to cancer care based on care quality and evidence-based, clinically meaningful outcomes.
Costs and mortality are the easiest outcomes for a payer to measure from claims data, said Bruce Pyenson, FSA, MAAA, Principal and Consulting Actuary at Milliman. “If you’re going to measure something, you start with what you can measure,” Mr Pyenson said. “You can measure money, and you can measure that as certainly as an outcome….Another metric that’s really basic is mortality,” he said. We can argue about what is progression-free survival, “but usually you know if someone’s dead or alive. Start with the basics like that, and work up from there,” he advises. It is not perfect, but it is a good start.
Other measures promulgated are the amount of time in hospice, the number of admissions in the last 2 weeks of life, and the administration of chemotherapy in the last 2 weeks of life, said panel moderator John Fox, MD, MHA, Senior Medical Director and Associate Vice President of Medical Affairs at Priority Health. “Simply knowing that the employer stopped paying the claims does not often provide sufficiently sensitive data to measure those types of things,” Dr Fox said.
None of these metrics is perfect, said Mr Pyenson, who said he prefers to use existing data reporting systems, many of which are not being tapped for useful information.
“I’ve annoyed some information technology vendors by saying you could probably get 90% of the value of the health information exchange at about 5% of the cost by using claims,” Mr Pyenson said. “You have history, you have details of visits, procedures, and so forth. You may not know if it’s stage I or stage II lung cancer, but you can certainly get a huge amount of information about the care that someone got.”
The discussion over measuring quality in cancer care must be separated into measuring clinical outcomes and measuring the cost of care, said Michael Kolodziej, MD, National Medical Director of Oncology Solutions at Aetna. Survival is not a good outcome measure, because too many factors have an impact on survival that have nothing to do with the skill of the physician.
The challenge in coming to grips with variability in cost of care lies in addressing high-cost outliers, Dr Kolodziej said, noting that a small proportion of patients is responsible for approximately 75% of the total cost of care.
Process measures, such as prescribing hormone therapy to women with estrogen receptor–positive breast cancer, have been accepted as surrogates for desired outcomes, said Dr Kolodziej. “I’ve defended the quality metrics in our programs by saying that they’re process measures, but I can link them to a very precise outcome,” he said. “I want to know that you have recorded performance status. I want to know that you looked at that patient, and you decided that patient was well enough to tolerate chemotherapy. I want to know if you assessed pain,…because I’m going to give you the benefit of the doubt that you know how to manage pain.”
Getting to the point at which real outcomes are measured may take a while, because the data are in multiple silos and must be aggregated and normalized.
Gary M. Owens, MD, co-chair of the conference, agreed that process measures are imperfect, but they remove the bad luck variables that can affect outcomes. “I don’t see a shortcut to getting the measured true outcomes for a long time perhaps,” Dr Owens said.
He added, “There is a phrase, ‘Not everything that can be measured matters, and not everything that matters can be measured.’ But I think it highlights the dilemma that we have insufficient measurement systems to measure the things that really count.”
Linking process measures to meaningful outcomes would be useful, said Dr Kolodziej. For instance, a prescription for an antiemetic for a patient receiving chemotherapy is a process measure that would be more meaningful if linked to the patient’s experience with nausea and vomiting, perhaps by counting the number of phone calls the patient made to the practice in the immediate 72 hours after they received chemotherapy, and whether the patient returned for his or her next visit.
“The truth of the matter is that the world we live in now is very unidimensional, or at most bidimensional, and we need to think of a 3-dimensional solution to this problem,” Dr Kolodziej said.
In responding to the challenges in data collection, Dr Kolodziej predicted, “Ideally, this problem [of measuring practice performance efficiently] will be technology-enabled to allow practices to focus not on collecting data but on improving performance.”
He added, “Rather than having to hire a bunch of people to do the work, is there a technological solution that allows the practice to measure its performance and then concentrate on how to improve performance, as opposed to concentrate on how to do the measurement?”
Dr Fox questioned whether patients’ satisfaction with their physician is a good metric for quality of care, citing a study showing that patients with stage IV metastatic colorectal cancer who thought they were receiving curative chemotherapy had the highest level of satisfaction with their physician.
Adopting the American Society of Clinical Oncology’s Choosing Wisely recommendations is an example of a performance measure that can be integrated into an oncology care model, commented Rita Shane, PharmD, Chief Pharmacy Officer at Cedars-Sinai Medical Center, Los Angeles, CA.
A comprehensive medication review that includes measures of adherence and adverse events would make for another logical performance indicator, Dr Shane said.
Barbara L. McAneny, MD, CEO, New Mexico Oncology Hematology Consultants, Albuquerque, commented on the discussion, saying, “As we look at measuring quality, I think that we need to do two things. One is to recognize that the quality measures themselves need to be evidence based. By that I mean that they need to actually measure something that is useful.” The second thing, she added, is “that it is very expensive for practices, which are struggling and strapped for cash these days, to be able to provide a lot of quality measures.”
Other measures of quality, panel members noted, may include discussions of treatment plans with patients, including the goals of therapy, the ability to work while receiving treatment, and advance directives, said various panel members. Reimbursement for these services would aid in the collection of such data toward improved outcomes.
Data collection and assessment of quality metrics may also guard against underutilization of quality care as payment models shift to episode-based care and an incentive to withhold care, the panel members observed.
