Profiling Strategy and Methodology
For your health plan as with many managed care organizations in the year 2001, the business risk varies greatly according to the choice afforded by the purchasers of the insurance coverage offers a broad product mix (from a pure HMO to one with “opt-out” or “open access” to self-insured groups who require only minimal medical management). . In addition, primary care physicians are never truly “gatekeepers,” per se. Finally, your health plan will eventually offer a “narrow” network that represents its best and most cooperative practitioners, if it doesn’t do so, already. These factors militate against having a simple profiling strategy. Therefore, your health plan should make a few assumptions. These are legitimate ways to ethically and properly examine and reward exceptional performance of practitioners. One should, for example, 1) Report trends in total costs and quality even though some of these costs are not perfectly in the control of the individual physician. 2) Use a Clinical Model that captures what we call the delta (D) of collaboration—the improvement of performance that is observed coincident with or attributed to collaborative partnering. 3) Recognize that the value of health care is a combination of quality, cost-efficiency and outcome or effectiveness, if not actual benefit. 4) Also, and to the extent possible, adjust (control) for differences in case-mix before reporting. 5) Remove catastrophic cases from the data set in order to partially normalize the data. 6) Be cognizant of the problem of small numbers and use only statistically defensible data. The parameters used in profiling concentrate on Relative Quality – accessibility to get routine and urgent care and a physical examination; outcomes in disease management initiatives and during episodes of care; doctor-specific patient satisfaction, as well as complaints and voluntary (that is, by patient choice) doctor switching Relative Utilization – comparative statistics: hospital days used per 1,000 members with consideration of the number of these days deemed appropriate for that venue; referrals and referral costs out of network; cost-effective choices in pharmacy; availability (panel growth) total cost trends per member and comparative costs per episode of care. Clinical Points – Diabetic retinal exams, Hgb A1c level s representing good control in diabetics; immunization rates; antibiotic sequencing and timeliness in sinusitis; lipid lowering agent use in CAD, etc.. Analytic Tools – We recommend using SmartCareÔ` a high-speed data mining workstation (http://www.vantagepointinc.com) that combines risk contract modeling, disease management, resource utilization analysis, and practice pattern assessments. It works seamlessly with 3M’s (http://www.3m.com) CRG Engine (crg-article999.fm)
that controls for case-mix or acuity differences to display and analyze
patient populations and their care by episode of care along with up
to 150 other demographic, provider, outcome and clinical categories.
Data Translation into Information: The Methodology
Please note, the only indicator or variable that has yet to be defined or purchased is provider-specific patient satisfaction. That data should be collected at the point of service (and that is not practical). Therefore, we will need to corroborate that measure with voluntary doctor switching and member disenrollments for dissatisfaction with the doctor. The episode construct is the most salient one to help define the “best” practices; for that we will rely on SmartCare and the Clinical Risk Groups (CRGs). Measures
The performance against reasonable and accepted standards and benchmarks for the management of specific medical conditions will tie in nicely with existing efforts in HEDIS, Guideline Measurement and Management, and Disease State Management initiatives. Here, the score that determines “success” will recognize the following key variables, listed with their appropriate weightings (PROMISE): Prevalence-10% Relevance (importance)-30% Outcomes—the potential to uncover and improve medical care-20% Measurability by the current I/S-30%. Equity- size and growth of the practice as far as NewAlliance is concerned-10% Nevertheless, if any of these factors are not present, the indicator item must be ruled out. Performance
Over Baseline
We compare each group’s PMPM expenditures to the network’s average, after controlling for case-mix differences, where possible. Any improvements (trends) that are sustainable will be attributed (as there is no causality implied) to collaboration, thus the concept of the Delta (∆) of Collaboration. Where we are not at risk, the reward is correspondingly reduced (proportionally). Funding depends upon the ability of NewAlliance to maintain its fiduciary responsibilities, however. -- Note the 4 sections, not all of which determine the
amount of the incentive --
The Flow
of Dollars
See Flow Diagram
SECTION A: INDICATORS
A1. PANEL
DEMOGRAPHICS
A2. TOP
5 DIAGNOSIS[3]
BY SPECIALTY (Clinical Risk Groupings’)
A3. QUALIFYING CLINICAL PARAMETERS OF CARE-
A4. COMMENTS Any comments that Your Health Plan may want to make regarding this provider’s measures. Perhaps a “looking good” or “You have a couple of areas of interest that are noteworthy.” SECTION
B: AVAILABLE INCENTIVES
1. PER MEMBER PER MONTH (PMPM) [7]
(total number of member months /# months in reporting period)
Funds available to the incentive pool for this
reporting quarter: 246 (Your
Panel Size) x $10 (Δ) =
$2,460.00 TOTAL
FUNDS AVAILABLE FOR INCENTIVE POOL:
$2,460.00 SECTION
C: APPORTIONMENT
C1. CRITICAL
BUSINESS FACTORS—Apportionment of the Parameters of Care and the Clinical
Measures as found in Section “D” below.
C2. CLINICAL
POINTS
C3. YOUR
SCORING (If the Score is less than the Peer Group’s average on 3 of 5 measures,
one does not qualify)
4. [1] One must be in a group, that group must enjoy a panel size of >500
continuously enrolled member. [2] Total number of member months /# months
in reporting period [3] Most populous Dx incidence: the doctor’s compared to his or her
like-specialty’s incidence as % of the panel based upon CRGs [4] Out of Network visits or dates of service that get initially denied,
regardless if overturned; data base = Auth Table or track and report
OON activity using a simple spreadsheet.
The obverse is that some referrals are approved only to protect
the patient from balanced billing; this is detected when the claim
comes in [5] Acuity-adjusted Clinical Risk Groups (CRGs) are used removing the
effect of catastrophic cases. This
is Days/1000- however, it could be ALOS by APR/DRG. [6] “Approved” verses “Denied” or “Avoidable days” omit M.H./Substance abuse and pregnancy-related cases [7] When total PMPM expenditures of one provider group is less than the network average, there are “savings” attributed to that group’s performance that funds the incentive program. However, these “savings” may not be real. For example, § Pharmacy benefits differ and co-pays and deductibles may make the costs recorded by the health plan appear less § Likewise, medical benefits differ comparing fully to self-insured groups. Hospital employer groups, for instance, have both self-funded pharmacy and dental benefits § Reinsurance recovery--Catastrophic cases need removed from the data. § Case mix (acuity) and age-sex differences can explain some of the ”savings.” Consequently, it may be more appropriate to use “Allowed Amount” for Section B and actual paid claims data (e.g., using a full four months of data with as small a completion factor as possible for adjustment purposes). [8] Including both HbA1c test, and HbA1c level. [9] Note: this requires hybrid method (claims and chart review) to get the true rates [i] e.g., By comparing practices in terms of the cost-effectiveness of antibiotic choices and their sequence to patient outcome, the following scale is derived:
|
