My piece yesterday on the apparent inability of DSM to prove its value got some very pointed feedback. Gordon Norman, the V.P of Disease Management at PacifiCare, wrote this response. He suggests that there are real world answers to the question of "does DSM save money?", and that the whole health care industry needs to become more sophisticated, focus more on where DSM can make cost-effective change, and be realistic about where DSM can and cannot make a difference. It’s an excellent piece:
In response to your musings this morning on DM outcomes and "what’s really going on", I offer this counterpoint on the four recent media pieces about Disease Management that may be viewed by some, including yourself, as casting doubt on the ability of DM to generate cost savings.
Full disclosure: I am a family physician, with a decade of experience working in managed care, responsible for the DM programs that my company has deployed for our 2M commercial and 1.5M senior HMO members since 2000, which in aggregate are generating net savings of $120M measured by robust methods and validated by external third parties. Based on these results and our enthusiasm that DM is one of the few effective, feasible, and politically acceptable means available to the health care non-system to deal with escalating costs in the near term, I am now dedicated to implementing and improving these DM outcomes for both managed care and non-managed care consumers (i.e., Medicare FFS beneficiaries).
1. CBO Report on DM
On October 13, the Congressional Budget Office released a report requested by Senator Don Nickles during the MMA legislative process that appears critical of DM’s cost savings potential for Medicare. This report represents no new knowledge about DM — CBO simply reviewed the available peer-reviewed literature and concluded that the evidence supporting financial savings from DM was inconclusive, not that there was good evidence demonstrating that DM does not or cannot generate positive ROI. This is not surprising, given that very little DM has been conducted as randomized trials, intended for publication in peer-reviewed medical journals. Health care delivery research is much harder to conduct in the classic "double-blind, randomized control trial" design as are medical studies with investigational drugs, and the majority of DM today is motivated by business vs. research drivers. Consequently the published DM work represents only a small fraction of the accumulated DM experience, and much of what has been reported is obsolete by current DM standards. In contrast, a variety of robust, quasi-experimental designs reported at DM meetings and the non-peer reviewed literature have repeatedly demonstrated significant ROI for good DM programs. Also, even if DM is shown to reduce short term expense but turns out to increase longevity for some with chronic illnesses, it may very well increase long term expense for Medicare relative to no DM — while increasing lifespan may be worthwhile in itself, from a Medicare Trust Fund solvency point of view, it will be a concern for those who must figure out how to fund that longevity.
In contrast to the CBO conclusion (or rather "inconclusion"), many prudent purchasers of health benefits, and the plans who supply these services, are convinced by the preponderance of evidence that short term economic benefits exist for DM, even if uncertain about the best method for calculating. The current $800M – $1B of annual spend on DM services carries a substantial amount of credibility about perceived savings in the marketplace. Understandably, CBO and policymakers may require a standard of proof that is "beyond a reasonable doubt" as opposed to a "preponderance of evidence" standard — that will require multiple, large scale, randomized controlled trials; fortunately, CMS’s ongoing BIPA and CCIP demos will provide just that. Here’s a link to the transcript of the DMAA’s rebuttal to the CBO.
2. CHF DM results in Annals Int Med
The October 19 issue of Annals of Internal Medicine published an article from Stanford/Kaiser about CHF DM. Dr. Robert DeBusk working with Kaiser centers in No Ca performed telephonic DM for wide range of CHF patients in RCT (Tx N=228), and found no statistically different outcomes for rehospitalization, ER visits, or death. (DeBusk developed MULTIFIT program, which was licensed by both CorSolutions and Kaiser for their CV DM programs.) They claim this is largest randomized study to date to evaluate specialized care mgt for patients discharged after hosp. for CHF. Notably, only 49% of these patients were of the same severity as those we enroll in our CHF DM (NYHA Class III or IV) at baseline, as indicated in the title of the article, "Care Management for Low-Risk Patients w/ Heart Failure". This is obviously one speculative reason why no difference in outcomes was achieved, but also perhaps also because Kaiser providers and patients are already "more EBM compliant" at baseline than non-Kaiser physicians, leaving little headroom for DM improvement. Also, the telephonic CHF DM approach used here may not be suitable for optimizing savings in low or high risk populations.
3. UT study on CHF DM at AHA Annual Meeting, Circulation
A report from the annual AHA meeting describes a large academic research project, "Long-term Impact of Disease Management in a Large, Diverse Heart Failure Population", at University of Texas, San Antonio which will appear in Dec 7 issue of Circulation. Over 1,000 elderly patients were assigned to telephonic case management by CorSolutions. Despite some minor longevity gains, there apparently were no utilization or cost differences in the treatment and control groups. From review of the article, I believe this study targeted too broad a cohort to optimize savings, and used suboptimal (telephonic only) interventions.
4. Kaiser DM in Health Affairs
This week Kaiser researchers published a longitudinal assessment of quality, utilization, and costs for adults w/ four major chronic conditions for the period 1996-2002. By their own report, Kaiser’s form of "DM" is an amalgam of many activities such as clinical guidelines, patient education, disease registries, proactive outreach, reminder messages, multidisciplinary teams, and performance profiling for providers, most of which are embedded in the usual delivery of primary care within the Kaiser system; in some cases, care managers are also utilized for DM. No defined "pre/post" periods can be defined for most of these efforts, as they were adopted incrementally. The authors concluded that quality measures showed improvements, but no reductions in real costs were demonstrated over this time frame for any diseased cohort. However, it appears that similar cohorts of patients without these conditions experienced signficantly greater cost increases over this period, suggesting relative trend mitigation.
Associated editorials by Kaiser clinical executives suggest that the authors set an impossible hurdle in trying to demonstrate absolute cost reductions year over year, even though the relative cost savings of these programs was over $200 million. They point out the challenges of determining DM impacts caused by (1) varying definitions of cost savings; (2) identification and characterization of baseline experience; and (3) issues about allocation and timing of costs and savings. We would agree that looking simply at total cost over time invites many confounders that may obscure true DM impacts — changes in unit costs, changes in networks, changes in covered benefits, changes in EBM/technology, severity changes in populations, etc. That is why we support DM savings methodologies that target the most important drivers of chronic disease cost and adjust for anticipated confounders (e.g., unit cost adjusted, CHF disease-related costs for identically-defined pre/post CHF cohorts), and then compare these financial outcomes with non-financial event outcomes (e.g., hospital & ER admissions) to test the plausibility and credibility of the economic results — an approach that has achieved recognition by the Disease Management Purchasing Consortium in their awarding PacifiCare certification of our outcomes methods for DM programs.
Takeaways from these 4 articles:
- DM is getting more public exposure — while it may be glib to suggest "there’s no such thing as bad publicity" for DM — as for celebrities — greater public scrutiny of DM to separate the hype from the reality, and the performers from the wannabe’s, is overdue and welcome.
- All DM is not alike — it is predictable that telephonic DM interventions may not have the same impact as remote monitoring interventions, or that targeting high risk populations will save more money than targeting all risk groups indiscrimately. Caution is urged in generalizing the conclusions from any one "DM study" to other DM settings or interventions. DM is not automatically good — there is good DM and not so good DM, and the differences require some digging beneath the surface.
- DM won’t save short term costs for people who weren’t going to have any — low severity CHF patients aren’t sufficiently at risk for utilization for even good DM to be effective, much less pure telephonic.
- Academics have a different standard of proof than business for DM outcomes — as evidenced by DM revenue growth CAGR of 20-30%, business appears to have sufficient proof of DM value today with their "preponderance of evidence" standard, while peer-reviewed reports of well-designed RCTs are still accumulating for policymakers, who maintain a "beyond a reasonable doubt" threshold.
- DM outcomes methods are in evolution, and therefore reported DM outcomes are not easily compared — DMAA is leading an industry-wide movement for increasing standardization and rigor in these methods and reported results. Their white paper was released at the annual DMAA meeting 2 weeks ago, and will serve as a template for improved methodological approaches in the future. (Not available on line but email Matthew if you want a copy)
- We must get beyond "Does DM work?" to "Which DM models work best for which patients on which measures?" — (after all, does "health care" work? It depends on who/what you measure.) Determining whether provider-based chronic care models, one-stop call center DMOs, or specialty DM vendor results are superior from identically defined metrics for equivalently defined populations is where we should headed. "Negative results" from studies utilizing any one DM approach tell us nothing about DM in general, only about one implementation of DM.
- Don’t expect DM to eliminate trend — absolute health care costs are going to rise for many reasons beyond the avoidable "wasted" dollars spent for chronic disease populations; DM should not be discredited for not mitigating all the trend drivers. We musn’t let Perfect be the enemy of Good — for DM to impact a large percentage of the avoidable costs for chronic disease patients while improving their quality of life, clinical and functional status, and in some cases, even their longevity, is a Very Good performance indeed.
- Who has a better idea? — while it may not yet have been incontrovertably demonstrated that good DM consistently produces favorable short-term economic outcomes, there is a body of evidence supporting that hyporthesis, and a larger body of work supporting many non-economic benefits from DM. So what other ideas are on the policy table that will work better, faster, or cheaper to address the growing mismatch of a health care system designed for high-tech, specialty acute care with an aging constituency with escalating chronic care needs? Universal coverage? Consumer-directed healthcare? Widespread EMR w/ point-of-care decision support? All of these are politically elusive, undeniably expensive, unfeasible for short term implementation, and unproven with regard to net health care cost savings.