Improving the healthcare services that an organization provides can be challenging unless one gathers information on the service, collates it, improves on it, and implements the changes in the field. A panel of experts invited by Evidence-Based Oncology discussed processes that help identify the most valuable metrics and the importance of making them relevant for use in the clinic.
Improving the services that an organization provides—whether these are commercial services or those related to healthcare—is not possible unless one gathers information on the service, collates it, improves on it, and implements the changes in the field. This process is fraught with challenges at every step as it continues to evolve and develop.
Evidence-Based Oncology organized a discussion on the topic among experts who evaluate and utilize quality measurement tools at the point of care. Linda Bosserman, MD, assistant clinical professor and staff physician, City of Hope; Jason C. Goldwater, MA, MPA, senior director, National Quality Forum (NQF); and Jennifer Malin, MD, staff vice president, Clinical Strategy, Anthem, joined me to talk about processes that help identify the most valuable metrics and the importance of making these metrics relevant for use in the clinic. The panelists also discussed the challenges with extracting data from electronic health records (EHRs) across independent data systems, evaluating and implementing the metrics, and data gaps that currently exist in the field of quality measurement.
According to Bosserman, quality measurement is an indicator of the real care that is delivered; it provides insight into what we know about care and helps improve on it. “You can only [improve quality] if you measure it and try to understand it,” she said.
Malin said that although quality measurements have been integrated into the operations of most other industries, healthcare has arrived late in the game. She thinks that for a long time now, we have been doing things a certain way in healthcare without knowing why. “As Dr Bosserman says, in order to know what we’re doing, we have to measure it so that we have an idea of what we’re doing today and how we can improve. We can only identify the areas that we need to improve if we measure what we’re doing,” Malin said.
There are 3 important qualifiers for measuring quality, according to Goldwater.
Back in the 90s, as the field was evolving, researchers were baffled about converting clinical outcomes into process measures. “Now, you flash forward to 2016. There are more than 600 NQF-endorsed measures and probably approximately 2500 quality measures overall. Because we’ve been able to develop measures and figure out ways of evaluating clinical outcomes and processes and structures, it’s allowed us to keep in line with the advances in medicine,” Goldwater explained.
Defining a Good Metric in Oncology
Goldwater provided an overview of the process that NQF follows when evaluating these quality measures. A standing committee of experts from diverse groups—payers, clinicians, and hospitals—reviews the metrics and their applicability and ease of use in the clinic. Explaining the rigor of the evaluation process, Goldwater said, “The measure has to be reliable; it really does have to be able to produce a metric of value. It has to be valid, so the metric is measuring what it intends to measure. It has to be feasible, so it is something that could actually be used, particularly with the rapid implementation of electronic health records over the last 10 years.”
He believes that the measure should be implemented into an EHR or a registry, and should not significantly interrupt the provider workflow or steal time away from patient care. “That’s never the intent. The intent is to continue to spend time with the patient by also understanding what the best practices are for quality and being able to have data populate that measure to indicate the threshold at which we know quality is being reached,” Goldwater explained.
Measuring Performance: What’s Important for Whom
Quality metrics play a significant role in informing health plans on the performance of clinics and health systems. So when payers evaluate the clinical and performance outcomes of a clinic, what are the most important quality metrics? Malin said that the field of quality measurement is still in its infancy, and current data systems are constrained in their capacity to convert clinical data into meaningful quality measures—especially in cancer care. “It’s very challenging to get quality measures that are really meaningful that don’t put a burden on the practice,” she said.
Malin explained that for health plans, the interests of their members are the primary concern. Patients and their caregivers want to know which practice can provide them with the highest quality care—information that is still somewhat elusive. Malin believes that existing quality measures are merely conversation starters between payers and providers. “As a health plan, we really want to get to that next step of transparency and being able to share the quality scores of practices with our members so that they can use that information in helping to decide where to go to get care.”
When asked to comment on how payers weigh process measures versus outcomes measures, Malin said it has to do with the context. She provided an example of postsurgical outcomes in patients diagnosed with pancreatic or esophageal cancer, and how the site of surgery—the hospital where the surgery is conducted—can influence 30-day mortality, anywhere from 5% to 25%.
According to Malin, documented evidence has shown that low-volume hospitals have higher mortality rates, information that a patient may not be aware of when he or she selects a hospital for treatment.
This is changing, Malin said, although slowly, and the California Healthcare Foundation has launched an initiative to promote patient awareness on the performance of healthcare clinics and bigger health systems. One such effort is a website that allows visitors to compare and contrast different healthcare providers and access their quality scores on patient experience, rehospitalization, and patient safety.1 Commonly used metrics, like the 5-year survival rate, are “not very helpful in the setting of looking at quality of care where you really want to know what is happening now. It doesn’t really help me to know how this practice was doing 5 years ago. I want to know how they’re doing now when they’re taking care of patients,” said Malin. Process measures would definitely be more valuable in that situation, she added.
Malin believes in evaluating both process and outcomes measures to see an improvement in care quality, because “if you have an outcome measure it does not really tell you what you need to do to improve the quality of care.”
Bosserman distinguished between the clinical and the patient perspectives. For patients, their primary concern is to see their health improve. If clinics fail to measure patient performance, then the information they gather would only be an extension of clinical trial data. She explained that trial data alone are not the best indicator of outcomes, because trials are conducted in a controlled environment and may not always account for differences in age, ethnic diversity, or comorbidities. “Patients want to be engaged. They want to know what they can expect. They want a good experience. They want to have their issues addressed. They also want to know the cost,” Bosserman said.
For a clinician, the ultimate goal is to improve outcomes in patients with multiple on the care they receive, hopefully with the adequate use of EHRs. But a continuous improvement is not possible without a learning system that provides feedback. “Once you start measuring, you keep learning and you keep improving your processes and you standardize more,” Bosserman said. She firmly believes that patient engagement and targeting the right treatment to the right patient can transform the system, for both the patient and the care team.
Building PRO Tools
Although there has been movement in healthcare to incorporate patient-reported outcomes (PROs) into quality measurement, it’s been a long, winding road to identifying the measures to include. Gathering this data is a challenge that researchers are still struggling with.
There are several barriers to PRO measurement, Bosserman said, starting with what measures should be included. “How can we prompt our patients to give us information so we can walk into that visit and make it really effective—whether it’s a phone visit or an office visit or an interval visit?” She then explained the strategy developed by her own practice with respect to PROs. The physicians/ at her practice believe that interactions with patients should be all about the medications they are taking and whether they have any issues with their treatment and if they want those addressed. She added that clinics should also track whether the patient received any alternate care between visits.
“If we can integrate the patient-reported outcomes into our decision making and our clinic visit, whether it’s a phone call or however that visit goes, we can much better address patient symptoms and hopefully minimize them so they’re not getting sick between treatments and we’re keeping them as healthy as possible,” Bosserman said.
Goldwater believes that PRO tools are hard to build, especially since they need to perfectly capture the patient’s viewpoint on relevant and meaningful outcomes. The tool should ultimately be able to let the patient perform an action to improve their quality of care or aid the physician in doing so. “It’s a lot easier when you have an outcome measure that says if somebody comes in with a heart attack, give them aspirin right away—which now is somewhat of a common practice. It’s much more difficult to try to understand what the patients really want if they come in with these symptoms or if they have these signs or specific concerns.”
The background information that is captured to help build the quality measurement tool is extremely important, Goldwater stressed, and the patients need to understand the information they are being asked to provide and then fill it out accurately as it relates to them. The measure then has to align with what the patient filled out. Testing these measures is a challenge in itself, he said, with sometimes more than a year needed “to find the appropriate sites, to deploy the measure, get the data, and look at the data in comparison to the measure. It is also important to ensure that the measure is reliable, valid, feasible, usable, and then make the necessary refinements as necessary.”
Added dimensions to testing the PRO measures include identifying the right sites and the right patients to fill out the tool that will populate the measure. Goldwater said that while progress has been made, with patients more capable of self-reporting on information associated with their health, a lot remains to be achieved.
Overcoming Data Silos in Quality Measurement
The lack of interoperability across data systems remains a significant barrier to improving healthcare quality. The panel continued their discussion on ways to overcome this issue. Interoperability across data banks is key to make data more meaningful. “While we have the Office of the National Coordinator promulgating a variety of data standards and vocabularies to control what that information looks like and how that information reads, there’s no mandate that they have to be using these particular standards for laboratory tests, for laboratory orders, for diagnoses, or for procedures,” said Goldwater.
He clarified that there are specific codes in oncology that define a certain service or clinical information, but that the use of these codes is not mandated. Subsequently, the lack of standardization and diverse vocabularies across data systems complicates matters. Goldwater went on to suggest one particular solution to this problem: creating a unique patient identifier. Having a unique patient identifier for every patient would help track their health records and make it easier to pull their information out from various sources, he said.
We are in the transition phase with interoperability, said Bosserman. “Projects like CancerLinQ2 that are now linking up multiple different electronic records and different data sets and CMS sets and SEER data sets—they can begin to feed all that information on a specific patient back to the practitioners.” Hoping for a bigger role for the federal government, Bosserman added that just submitting information to meet compliance requirements may not be the best practice for clinics to follow. “We need to start moving from these volume-control measures to value-based compliance measures,” she said.
As a payer, Malin provided a very different perspective. The challenge she experiences with the diversity of clinical practices that Anthem’s members are enrolled in extends beyond data silos. “Not all of our oncologists and physicians use EHRs, so extracting member data faces the biggest challenge right there, which can then increase the burden on those practices if we try to gather that data from them,” she said. To circumvent the issue, Anthem is using data from its prior-authorization and clinical pathways program, the Cancer Care Quality Program. “We can use that data, which we then get across all practices and members, in order to look at the quality of care being provided,” Malin explained.
Although Anthem has been working with EHR vendors to create an interactive system with a 2-way data transfer, problems arise when some practices have very sophisticated EHRs that link to their billing system and have in-built quality measures, while others are a solo practice that is completely paper-based. “Thinking of it from a health plan perspective, I have to come up with a solution that bridges all of those practice settings,” Malin said.
Bosserman is quite confident about the potential of CancerLinQ in promoting the quality of cancer care. “I think that platform under the SAP group is going international and will be a foundation of a system that can help us going forward on a larger level. It’s an evolution,” she said. NQF is interested in knowing how ASCO is planning to leverage the data that they gather via CancerLinQ. Will it be used “to populate measures that are already existing or to, perhaps, gain a better understanding of measures?” Goldwater asked. He believes that the platform could have a tremendous impact if it can help develop a new measure or provide updates on an existing measure.
Bosserman said that CancerLinQ has already had a significant impact on data accuracy. In her mind, the bigger challenge is with “structuring the processes in the clinics or the apps to collect the discrete data in both systems of identified data accuracy and data entry blanks.”
Standardizing Oncology Quality Measures
How does a health plan finalize its reporting requirements of oncology clinics and practices? Does Anthem, for example, follow CMS’ data-reporting requirements?
“What we have been focusing on, and reports that we’re going to be providing to practices sometime [in June 2016], include adherence to our pathways, [emergency department] utilization, hospitalizations during treatment, and access to hospice and other end-of-life care,” Malin said. She added that these measures are a part of the Oncology Care Model (OCM)3 and have been advocated by the Community Oncology Alliance and other groups. For Anthem, “Developing a smaller set of meaningful measures that can be implemented across their entire network is a substantial challenge.” Malin said that Anthem has tried to develop measures that can help practices compare the quality of care they deliver with other practices that treat Anthem members. Whereas these measures may not match up with the requirements of the Physician Quality Reporting System, she said they are in synch with the OCM.
As for the administrative burden of all the reporting on physician practices, Malin said that America’s Health Insurance Plans) (AHIP, has a working group that has tried to short-list measures across commercial plans and CMS, and some of these measures are a part of the OCM. “Figuring out the most efficient, and least burdensome, way for [physicians] to report data is a big challenge,” Malin said. Commending the rigor and the standards used by NQF in evaluating these measures, she said, “Measures that really get used for pay-for-performance and accountability, most eventually end up being reviewed by NQF.”
Gaps in Care Measurement
The presence of an exhaustive list of measures is not necessarily an indicator of a flawless system, as data gaps may still exist. Malin agreed, identifying several different gaps:
Goldwater said that NQF leads the Measure Application Partnership,4 a product of the Affordable Care Act, which brings together leading experts in oncology from numerous sectors: payers, providers, and consumers, to discuss measure gaps that need to be filled. These recommendations are, in turn, forwarded to CMS. He said that NQF has, over the years, identified gaps in palliative and end-of-life care in oncology. Additionally, there is a lack of information that can help understand where the disparities in cancer care lie, “to effectively then know what sort of treatments are necessary, what sort of preventions or screenings are needed, and where to emphasize efforts.”
Bosserman expressed her faith in the potential of the 2-way EHR system that Malin discussed, which will help doctors keep track of alternate care that their patients seek. “We don’t always get that real time in the practice unless you put some system in place that asks the patients, measures it and somehow keeps track of it,” Bosserman said. She believes that such gaps in information can be filled via a partnership between providers and payers. “The exciting thing is that the conversations between providers and health plans have really opened up. Dr Malin is one of the leaders in the field, and others at Aetna and United have really begun to reach out and try to partner with practices to start finding more solutions.”