Evaluation of the Use of AHRQ and Other Quality Indicators
Chapter 6. "Lessons Learned" for Future Activities
In this section, we present a number of "lessons learned" during our interviews concerning future directions for AHRQ in the development and modification of QIs. Our discussion is organized in three parts. First, we describe interviewees' perspectives on current, anticipated, and potential development projects involving the QIs. Next, we discuss users' perspectives of AHRQ as a measures developer and the ways in which users speculate this role could evolve or change in the future, especially in relation to other potential providers of this service. Finally, we briefly discuss users' views on the subject of market demand, in particular, user willingness to pay for QIs.
A key function of this study was to provide AHRQ with feedback from interviewees about priorities for future development efforts. In order to explore this topic with users, we first solicited input from members of the AHRQ QI team about current, anticipated and potential development projects. We then used these responses in our interviews, which asked explicitly about interviewees' opinions of the need for these projects as well as their own priorities for future development. We grouped the development projects into three categories and asked interviewees which category they would like to see given priority:
- Improvements in the current product line.
- Addition of new product lines.
- Improved support for the QI products.
Improving the current products was most frequently seen as the highest priority, followed by both the addition of new products and improvements in service, outreach, and user support for the measures (Table 6.1).
Many users told us that it was important to improve the current set of indicators as much as possible and expand their use so that the QIs became more of a national standard. One interviewee summarized this sentiment by saying, "One solid measurement set with everyone's buy-in would be enormously positive." Another user pointed out that "it would be good to focus on shoring up current indicators because there is currently a lot of criticism around using them for public reporting."
Many other users said that their recommendation to focus on improving current AHRQ QIs was driven by a desire to overcome stakeholder opposition to indicators based on administrative data. One interviewee summarized this line of thought:
There is no way that every hospital in the country is going to do primary quality data collection and even if they did, how could we enforce consistency and timeliness? This is a battle that we have been fighting for years, and we've been struggling because people tend to dismiss out of hand any information based on administrative data. In the short term, until there is progress with the electronic health record, administrative data is all there is, and there is no convincing argument that we have exhausted all possibilities to use this type of data for quality improvement.
Another user was more specific about how the indicators might improve:
The AHRQ QIs may not be perfect but they are a national standard, based on readily available data that are not going away, and the indicators will get better and better - especially with extended ICD-9 codes (and later the move to ICD-10) and the addition of a flag for condition-present-on-admission and things like that. I think there is an opportunity to improve the QIs gradually over time, as the underlying data sources improve—as new data elements are added with the introduction of the UB-04, and eventually electronic health records.
Despite these sentiments, there is also a strong desire by many interviewees for additional QIs covering new areas. Indeed, it was difficult for many users to choose between adding new products and improving current products as their top priority. Improving service and outreach was most frequently given a low rating. In the following subsections, we discuss in more detail what changes interviewees would like to see in the AHRQ QI program.
6.1.1. Improvements of the current product line
Apart from the expectation that AHRQ maintain and update the current QIs, the most commonly requested improvement was the addition of data elements to increase the specificity of the QIs, such as a flag for conditions present at admission or for "do-not-resuscitate" orders and the addition of clinical data elements (Table 6.2). As mentioned above, the AHRQ QI team is incorporating a flag for conditions present at admission in the next iteration of QI specifications. Other improvements mentioned with some regularity include validation studies on the development of composite measures (a project that AHRQ is currently undertaking) and better risk adjustment, with coordination of risk adjustment methods across the subsets of QIs.
6.1.2. Adding new product lines
Most interviewees were aware and appreciative of the roll-out of the pediatric QIs as a new module, as this important population had been excluded from many of the initial QIs. Almost half of the interviewees mentioned the additional need for measures for hospital outpatient/ambulatory care, such as day surgery and diagnostic procedures (Table 6.3). About a third of interviewees mentioned the need for efficiency, physician-level, and emergency room care measures. Nearly a quarter of interviewees expressed interest in integrating data and indicators for inpatient and outpatient surgery, since an increasing number of procedures are being shifted to outpatient settings. This, for example, has created a real problem for constructing the laparoscopic cholecystectomy indicator (IQI 23), because nearly all of those procedures are now done on an outpatient basis.
Measures for rural/small hospitals were the next priority group. However, interviewees expressed differing views on the implications of having a dedicated set of indicators for rural/small hospitals. On the one hand, many felt that dedicated indicators were needed, because the low patient volume at rural/small hospitals excludes those institutions from most of the current indicators. Further, interviewees felt that some indicators should not be constructed for those facilities. For example, since current ACOG guidelines do not recommend VBAC for facilities without adequate infrastructure for emergency caesarean section, the VBAC indicators (IQI 22 and 34) should not be used for many of them. On the other hand, some interviewees expressed concern that dedicated indicators would suggest that small and rural hospital were second-class facilities, because common quality standards do not apply.
6.1.3. Improved services around QI products
One of the most common priorities for improved service among interviewees (Table 6.4) was more explicit guidance from AHRQ on the use of the QIs for public reporting and pay-for-performance (also discussed in Section 4.3.3). Users were sometimes not aware that AHRQ had recently released documents on those issues; the latest guidance document was released in December 2005, predating these interviews by only a few months.47
Another commonly listed priority for increased AHRQ service was for AHRQ to provide guidance on the process that should be followed to improve quality in areas where the QIs indicate a problem. Users had varying levels of experience with quality improvement and varying levels of access to networks that can be used to share quality improvement knowledge. It would be helpful for users to have further guidance (e.g., a general methodology for analyzing medical records following an abnormally high incidence of PSI 4 - failure-to-rescue, and a summary of available evidence on interventions that could be implemented to lower the rate).
Interviewees suggested that AHRQ collaborate more closely with other organizations in attempt to forge more of a consensus on a "national standard" set of quality indicators. The standardization of some AHRQ and Leapfrog indicators and submission of some of the AHRQ QIs for NQF approval are steps that AHRQ has already taken in this direction. Further efforts along these lines would improve the usability of the AHRQ QIs for users.
Interviewees also suggested making the AHRQ QIs more user-friendly and simpler to understand. A simple suggestion in this regard was to promulgate official, simple names for the QIs in language understandable by people with no clinical knowledge.
We asked users specifically about one aspect of AHRQ service —user support. We received favorable feedback about the current level of AHRQ user support for QI users. Of the 15 users who reported using AHRQ support, all but one explicitly reported a good experience. Interviewees were impressed by the technical competence, accessibility, and responsiveness of the helpdesk staff and argued that this support function had played a major role in advancing the field of quality measurement, because it removed the barriers that non-research institutions face when implementing complex measurement systems. To provide a point of comparison, several of the more experienced users recounted the difficulties they had experienced in working with the HCUP indicator code.
One user reported being able to "feed complicated, technical questions from hospitals to AHRQ," and that AHRQ user support was able to answer those questions "from a greater depth and background" than the user had. This user added that responses to inquiries were "based on evidence, thoroughly considered and thought-through, with a quick turnaround." Another user felt that there was "always someone you could get hold of to voice concerns." A vendor commented that the AHRQ QI technical support had a "pretty quick response time compared to what one would expect from federal agency. They would open a case the same day, send an email confirmation, assign to a person—all in the same day." On the other hand, other interviewees (7 of 54) did suggest the need for increased responsiveness and speed of user support. These interviewees generally wanted most questions to be answered the same day they were asked.
We explored extensively the issue of how users perceive AHRQ as a measures developer, what they think AHRQ's role should be in this area, and whether some function that AHRQ currently performs could be taken over by other public or private institutions.
Our interviewees held AHRQ in very high regard. They credited AHRQ for its vision in pushing for the use of administrative data for quality measurement well before the research and provider community was ready to exploit this data source. The work of the AHRQ QI team was described as technically sound, sensitive to the limitations of the underlying data, and transparent. AHRQ is regarded as an intellectual leader and "go-to" institution for health services research and the use of administrative data for hospital quality measurement.
As shown in our environmental scan for comparable products, no clear comparable alternative to the AHRQ QIs has emerged or is likely to emerge. Several other developers, especially JCAHO, CMS, HQA, and Leapfrog, are seen as prominent sources for measures and may be used as alternatives, but their indicators differ in several important ways and are generally regarded as complements to the AHRQ QIs, not true alternatives. We asked QI users to visualize how the quality measurement landscape would change if the AHRQ QI program disappeared.
One interviewee answered:
If AHRQ stopped the QI program, pieces would be picked up but there wouldn't be a consistent, cohesive package as big as AHRQ is now. The public domain issue is a big one. Providers have only been in the game because the indicators come from a public source—if that public source goes away, I think providers will stop doing it.
And another one said:
I can't imagine who else would pick up activities from AHRQ so instead, probably activities would be broken down into orphan activities—any one slice would be a different activity; specialty organizations would take over certain types of measures (pediatric, for example).
Interviewees were quite comfortable with AHRQ having a leading role in national quality indicator development. It was generally viewed as positive that a trustworthy federal institution had defined open-source and well-documented quality measurement standards. These standards were viewed as contributing to the transparency of health care quality measurement and reducing the measurement burden for health care providers by limiting the number of measurement tools they must use to satisfy various reporting requirements. Many emphasized the need for even greater leadership from the federal government in this area, either by developing measures or by orchestrating public-private partnerships, so that standard measure sets for various purposes would become available and accessible to everyone.
AHRQ's leading role was also seen as a challenge for AHRQ, because with it comes the responsibility to maintain the QI program, on which so many programs now depend. Our interviewees looked primarily to AHRQ to fill the obvious gaps in the measurement science. Several commented that current funding levels for AHRQ were not adequate to meet all those needs.
We discussed whether it could be a viable option for AHRQ to give up parts of the current QI program in order to free up resources and set different priorities. Specifically, we asked whether AHRQ could or should stop developing software and providing user support in order to focus exclusively on indicator development. Almost unanimously, interviewees rejected a model under which AHRQ would develop and distribute the software without supporting it.
There was much concern that lack of user support would create enormous barriers to the implementation of quality measurement initiatives, especially for new users and non-research institutions. Using vendors to provide user support was also not commonly regarded as an alternative, because many feared that vendors would be prohibitively expensive or incapable of providing the same quality of support as the original developers. The latter view was even shared by some of the vendors who would potentially stand to gain from this model: one representative stated that "we do not want to support AHRQ's software since we can't support what we don't write."
We received mixed reactions to a model under which AHRQ would only develop and release indicators and their technical specifications, but no longer provide or support software. Many interviewees were familiar with such an arrangement, as it would mirror the division of responsibilities between JCAHO and the Core Measures vendors. But several drawbacks were brought to our attention, such as vendor and license fees, as well as potential quality problems and comparability issues (if different vendors implemented AHRQ specifications).
Several interviewees stated that such a model would represent a step backwards in the development of a unified quality measurement infrastructure, since a transparent national standard would be transformed into multiple proprietary systems, at the same moment at which many entities, like CMS, JCAHO and NQF, are trying to introduce open-source consensus measures, as recommended by the IOM. At a minimum, a rigorous certification program for vendors would be needed and many interviewees worried about the implications of such a change for the momentum that the hospital quality measurement movement has gathered.
Finally, we asked interviewees which parts of the QI program AHRQ could give up, if (hypothetical) budget cuts were to leave it with no other choice. Most of the 54 interviewees stated that the program represented a unified entity that should not be disassembled, although 12 interviewees said software development and 11 said user support could be discontinued by AHRQ and those functions assumed by others.
As an alternative to AHRQ realigning current funds, we asked interviewees whether AHRQ might consider financing program growth by generating additional revenues from charging users. Not unexpectedly, this proposal was not met with enthusiasm. Almost half of our interviewees (20 of 54) did not answer the question. Five out of 36 current users stated that they would stop using the QIs in this case. Three current users replied that they had invested so much into their program based on the AHRQ QIs that they would have to accept charges, but emphasized that they might not have selected the QIs in the first place if they had not been a free resource. However, almost half of the interviewees (44%) expressed willingness to pay a reasonable fee for access to the full QI resources.p Two even said that the perceived value of the QIs would increase if users had to pay for it: "Marketing 101: If you don't charge anything, people aren't going to proscribe value to it. If there is no cost attached, people can take or leave it because it doesn't represent an investment."
A slight majority favored a subscription model (i.e. paying a one-time charge), but some argued for a usage-based payment scheme. Most recommended differential pricing by type of organization and purpose of use (i.e. commercial vendors who resell the QIs or incorporate them into their products should pay a higher rate than state agencies that operate public reporting programs). Interviewees also felt that one-time use for research projects should be less expensive than ongoing use for operative purposes.
p. We did not elicit specific information on what users would consider to be a reasonable fee. A market study would be required to determine what would be considered "reasonable" among the current and potential users. Such an endeavor was outside of the scope of our study.