The National Quality Forum currently endorses nearly 60 quality measures relevant to cancer care—the majority of which describe processes, not outcomes.
But for cancer patients and their families, long-term survival—an outcomes measure—is one of their most important priorities. It's also vital to health insurers, such as Medicare and large private payers, as they are increasingly pursuing payment models that link reimbursement to quality, with the goal of lowering costs and improving health care delivery for cancer patients.
Research has shown that outcomes can vary widely depending on where patients are treated. For this reason, we need reliable, unbiased measures of the quality of hospitals' cancer care. However, efforts to rank hospitals by survival rates have been limited because readily available administrative data derived from Medicare claims lack potentially critical information about cancer stage and severity of disease. The concern is that a hospital with a large share of patients with advanced-stage tumors, for example, could have a lower survival rate than a hospital with a case mix skewed toward healthier patients—even when the two hospitals provide equally good care.
The most arguably well-known indicator of hospital performance—the annual U.S. News & World Report Best Hospitals ranking—relies on inpatient claims data. This approach represents a shortcoming in hospital outcomes research because the majority of cancer care is now delivered in the outpatient setting. To measure outcomes adequately, we must consider claims data for the entire spectrum of cancer care, including office visits, chemotherapy, radiation, and even home care.
Fortunately, Medicare claims data contains this information for a representative swath of the U.S. population receiving cancer care. For more than 15 years, researchers have generated impactful studies on how factors such as the volume of surgeries an institution performs and the type of hospital it is (e.g., NCI-designated site, academic medical center, etc.) can impact both inpatient mortality rates and 30-day survival. The majority of these studies have relied on risk-adjusted Medicare claims coupled with information on cancer stage gleaned from the Surveillance, Epidemiology, and End Results (SEER) Medicare database.
But what if cancer-stage information was not necessary to accurately determine how well different types of hospitals perform in terms of risk-adjusted survival outcomes?
In 2015, I was part of a research team at Memorial Sloan Kettering Cancer Center that published a proof-of-principle paper in JAMA Oncology addressing this question (JAMA Oncol 2015;1:1303-1310). Our analyses found no appreciable difference between hospital performance rankings calculated from risk-adjusted Medicare claims data without individual, patient-level cancer-stage information versus including cancer-stage information from SEER.
We examined two data sets that collectively included nearly 750,000 patients with cancers of the lung, prostate, breast, or colon, among others. Patients began either cancer treatment or management of recurrent disease in 2006.
The study ended 5 years later. We risk adjusted the Medicare and SEER data sets using 3M's Clinical Risk Group (CRG) system, whose algorithm assigns each patient to a group that reflects his or her overall health status and the presence and severity of comorbid diseases or conditions, as well as age and sex.
Our risk adjustment also included median household income in the zip code of residence for each individual. The SEER data was additionally stratified for the stage of a patient's cancer.
Both risk-adjusted data sets underwent identical proof-of-principle analyses to calculate the probability of death at four different groups of hospitals: those exempt from the Medicare prospective payment system (PPS-exempt), NCI-designated cancer centers, academic medical centers, and all others.
Results from the data set without cancer-stage information showed that patients treated at PPS-exempt hospitals had a 10 percent lower chance of dying in the first year as compared with patients treated at non-teaching hospitals (18% versus 28%), with NCI-designated cancer centers and academic medical centers falling between these two extremes. Successive years followed a similar pattern, with the survival gap persisting over 5 years.
These rankings remained consistent and were not significantly impacted when the SEER cancer-stage information was added into the analyses. This leads us to believe it's time to challenge the idea that we need patient-level data on cancer stage to evaluate a hospital's long-term cancer care outcomes in an unbiased manner. Important, reliable, risk-adjusted outcome data can be derived from administrative sources alone.
Acknowledging that significant survival differences exist among hospitals is an important first step in improving outcomes and is worthy of deeper exploration.
Observations over the past 10 years that large disparities existed among hospitals in surgical outcomes led to marked improvements in surgical mortality. Similar outcomes research using only administrative claims data is already under way nationally in cardiovascular disease; the Centers for Medicare and Medicaid Services publish risk-adjusted cardiovascular disease mortality rates for U.S. hospitals. Why can't we do the same in cancer care?
Our outcomes-measurement method proposed in JAMA Oncology certainly requires more validation, but we hope it jumpstarts and sustains the conversation around this important issue.
Addressing the dearth of data on hospitals' long-term cancer care outcomes head-on could provide myriad opportunities for multidirectional knowledge exchange among hospitals and result in an overall improvement in outcomes across the board. Consider that the observed survival gaps between particular hospitals represent potentially avoidable patient deaths.
Further research may help reveal the factors underpinning these gaps in cancer care. Are differences due to readmissions or mortality rates after surgery? Are patients experiencing severe side effects that prevent them from finishing chemotherapy?
It could be many things, but our research shows there is an opportunity to continue to elevate cancer care, and we owe it to our patients to do just that.
DAVID PFISTER, MD, is the Chief of the Head and Neck Oncology Service at Memorial Sloan Kettering Cancer Center in New York, which is a top 10 cancer hospital according to U.S. News & World Report.