- Half of all Medicaid beneficiaries (>26 million) in capitated MCOs
- States continue to expand to new geographic areas and to previously excluded populations
- Health reform accelerates growth
- Planning for Medicaid expansion
- Improving care coordination
- Federal funding (health homes, integrated delivery systems)
- The ACA allows CMS to withhold federal match for individuals enrolled in MCOs for whom the State does not submit MSIS encounter data
- Evaluation of service costs for business and operations management
- Evaluation and computation of capitation payment rates
- Federal reporting
- Monitoring program integrity (e.g., service utilization, access to care)
- Controlling costs
- Conducting quality reviews.
- Improving calculation of Medicare Part A disproportionate share hospital percentages.
With multiple reform driven initiatives in the market place today like to name a few health insurance exchanges, medicaid expansions, integrated care for duals, the need to submit accurate encounter data has become even more significant considering that all of these newer models have a risk adjusted component that will impact the plans revenue potential significantly. To add to this complexity many multi-state plans have to deal with different submission needs (formats, content, duration etc.) for different state entities they deal with. Yet another issue is with constantly changing submission needs these entities have – RAPS to 837 in case of CMS MAO, proprietary to 837 in case of State of California, XML in case of edge server reporting for HIX based products. Entities like CMS and State medicaids are requiring plans to start submitting lot more comprehensive data set so that their risk adjustment models can be properly calibrated in face of reforms and cost containment pressures. An example of this was, starting in January 2012, the 2010 mandate would require the monthly submission of encounter data in the form of an ANSI X12 837 Institutional or Professional transaction, which encompasses over 700 data elements. CMS also required the submission of all claims data, including types not previously submitted for RAPS such as laboratory, transportation and vision. Not only was the depth of the submission increased with the X12 837, the breadth was likewise increased by the inclusion of previously excluded claim types. This dimensional expansion increases the data’s usability and value to risk adjustment. Because the evaluation and application of an appropriately adjusted risk model is dependent upon cost data contained in claims, it was not surprising that CMS mandated the collection of encounter data using the more robust 837 transmission so that risk adjustment models could be properly calibrated in the face of healthcare reform. Most industry experts in risk adjustment agree that recalibration of the risk adjustment model will benefit from using cost data.
One of the key issues is to get the quality of submission right as we saw above the consequences of getting it wrong! Data quality is defined by many experts as “the multidimensional state of completeness, validity, consistency, timeliness and accuracy that makes data appropriate for a specific use.” David Loshin, an expert in information improvement, defines eight dimensions of data quality: uniqueness, accuracy, consistency, completeness, timeliness, currency, conformance, and referential integrity.
The transmission of data to is arguably the easy part, since many plans are accustomed to producing outbound encounter files for other lines of business. The real challenge for plans to manage data quality issues as rejected in error return reports without having to manipulate data in their source systems, which could have undesirable down-stream effects as well as alter the original picture of the adjudicated claim as it came from the provider.
Manage Data quality – quality data is the underpinning of successful reporting
- The health plan will need to analyze the root cause of any gaps that exist in enrollment and medical/pharmacy claims data.Since quite often this data set is spread across multiple enterprise systems, there will be an end-to-end review of the handoff of data files between these entities to confirm data stays intact
- The health plan will also evaluate provider treatment, coding and billing practices to ensure that all the relevant details are being captured by the providers creating claim data. In addition to this carve-outs around mental health, substance abuse, etc. need to be addressed; otherwise, the health plan could provide diminished data to the reporting entities
- The health plan needs to develop a plan for incorporating supplemental diagnosis data
- The health plans too have a nimble– both in the infrastructure and the ETL – that is highly configurable
- Ensure data quality is automated through the ETL engine where it supports a “learning model” that rapidly incorporates newly identified processing edits
- Processing of response files and coordination within the plans to resolve discrepancies in an efficient manner
- Full de-identification of QHP data in accordance with HHS requirement in case of submissions to edge server
- Maintenance of longitudinal matching between the de-identified data in the Edge Server and the identified data retained in the client data warehouse in order to preserve full lineage and auditability
- Analysis of risk scores and reinsurance compared to calculations prior to the submissions to CMS/State entities with reconciliation of differences and remediation of discrepancies
- Management and maintenance of the data so the plans have full visibility into what is reported
Error Processing
It is anticipated that majority of error processing needs going forward will also be similar to the output and reporting as it exists today in the EDPS model for Medicare Advantage health plans.Having an adaptive system in place to manage and resolve these errors is paramount to success for encounter submissions. In addition, the health plan needs to have processes to efficiently route and rectify edits/ errors to avoid a backlog of error data and costly delays in submission.
Health plans must consider have a data warehouse that contains all of their pertinent risk adjustment and reinsurance data. Not only will this act as a source for encounter data that will have to be produced from this warehouse, but the analytics are sourced. Essential analytics must include:
-
Identification of eligible claims for risk adjustment calculations and codification/ indexing of claims excluded from calculations for quality assurance analysis
-
Replication of the HHS Hierarchical Condition Categories (HCC) model to calculate risk scores based on available data for “shadowing” analytics for improved risk score accuracy and possibly using some publicly available benchmarking of the encounter data to catch instances of under coding based on diagnosis codes on a claim
-
Assignment of members into models and risk score calculation categories
-
Calculation of risk score components including demographic factors, HCCs, hierarchies, HCC groups, interactions, severity adjustment, and cost-sharing reduction adjustments
-
Sub-segmentation of population into reporting categories
- Accumulation calculations of member-specific costs against attachment points and thresholds for reinsurance submissions
Conclusion