I have spent several years working with specialty medical offices like oncology centers, diabetes clinics, IPAs (Independent Practice Associations), and disease advocacy groups seeking to build health care data warehouses and analytics solutions. During that time, I have seen the same concerns pop up over and over: “How can we understand the value and impact of our care if we only see the component of care that we provide? If we can’t understand our value, then how can we make sure that we are optimizing our care, getting reimbursed for our impact, and executing leading research in our specialties that helps find better medical treatments for our patients? How can we really care for patients effectively in the first place?”
Organizations are highly restricted in the ways they share data. HIPAA allows for data sharing between entities, but doesn’t provide for any mechanism or incentives to do so efficiently or in a scalable method. Also, the groups who should be sharing may find themselves in competitive situations where sharing could be perceived as risky. But in spite of this, some exciting developments have quietly been moving forward in the past few years that can help fill in pieces of the data last mile.
The rise of Meaningful Use 2 (MU2) compliant electronic medical records (EMR) with the objective to enable health information exchange (HIE) between systems now represents a potential solution to this challenge that has been exacerbating the fragmentation of the health care industry for years. Public HIEs have not yet demonstrated that they can resolve analytics issues or workflow changes. Instead, there are some new and useful models of HIE that show great promise that are likely being adapted from the lessons learned from the original HIE designs.
Private HIEs. A new model uses technologies deployed in various EMRs within a geographic area to form a private HIE between a health care provider and its collaborating partners. In general, these types of solutions are being built geographically and major health systems or groups of health systems may sponsor them. The HIE is designed specifically for a group of collaborators around a business need. It might, for example, support an ACO in sharing information to help manage the risk-sharing contracts among affiliated entities. A private HIE also might support a research collaborative collecting longitudinal records to aggregate and track patient outcomes after discharge from a specialty facility. A private HIE is very different from a public HIE, where participants may be concerned about aggregation of data for analytics that might harm one or another competitor.
APCDs. All Payer Claims Databases (APCDs) have enormous potential to support analytics to understand how patients are treated in the health system. Each state has a different degree of maturity with APCDs. APCDs have the potential to answer questions about differences in long-term spending by specialty group, for example. One challenge with using APCD data is identifying cohorts of patients with different protocols, if treatment information isn’t captured in the administrative data. Another challenge is concerns about using data for competitive research; APCDs may restrict analytics to non-competitive areas. The Massachusetts APCD, for example, restricts access to providers, payers, or researchers seeking de-identified data “exclusively for the purposes of lowering medical expenses, coordinating care, benchmarking, quality analysis, and other research, administrative, or planning purposes.”
Our ConvergeHEALTH by Deloitte team is reviewing the APCDs to help establish benchmarking and quality metrics for our clients. APCD data may be a good alternative to licensing national data sets that do not have enough detail, coverage, or options for data access.
APCDs are attractive but vary across the US. According to the APCD Council, fourteen states have a fully implemented APCD, two have voluntary partial approaches, four states are in implementation, sixteen states have strong interest, and the remaining states have yet to proceed. The total population covered by APCDs in existing and implemented states is over 77.3 million people or 24% of the US population today and if the “strong interest” states were to implement it would be 130 million more people bringing the total to 65% of the US population.1
With the APCDs, many use cases may raise concerns among administrators regarding whether they might increase, instead of reducing total cost. For example, a clinically integrated network might like to improve referrals to its affiliated sites to leverage investments in quality controls and care coordination. But this might be viewed as reducing competition if it limited patients’ referral options, especially if competition among providers of the same service would reduce prices.
It is a good time to have these new interoperability-based patient data aggregation tools and data assets available. The promise of both private HIEs and APCDs can enable provider organizations to begin to tackle last-mile analytics problems that have vexed them thus far. The question you should ask yourself is now that you have these tools, what will you do with them?
Dan Housman is a software veteran with a demonstrated track record of providing valuable and innovative decision support systems to large, complex organizations. Dan leads ConvergeHEALTH’s product innovation efforts with a focus on translational research, bioinformatics and innovative approaches to data capture, analysis, and reporting for clinical quality and performance improvement. Dan earned a BS in Chemistry and Biology from MIT in 1995.
Categories: Uncategorized
Wasn’t there an APCD in Canada at one time called the Medical Bureau?…this was an insurance facility that supposedly collected claims from many US and Canadian insurers and government payers? It may no longer exist.
There are other sources of population health data that we often don’t think about: google key word searches for specific diseases or signs and symptoms thereof; community sales by pharmacies and wholesale pharmacies of drugs and durable medical equipment (e.g. sales of pepto bismol and anti-diarrheal meds and thermometers, Pedialyte…et al); we can look at effluent–sewage–from large office buildings and hospitals and schools to check for viruses, drug and nicotine metabolites, SSRIs, ecstasy metabolites, bacteria and protozoa; we should always consult with state and county departments of public health to see what “is going around”, e.g. are they seeing Norwalk or Shigella or Salmonella?…; and we should check occassionally with county records to monitor local causes of death; you can look at reagent usage in large automated chemistry analysers in hospitals to get an idea of how many serum lactate analyses are being done and thereby get a feel for how many cases of sepsis are coming into the hospital and being worked up. My point is that there are many ancillary avenues to go down if one is dealing with health issues of many people. I guess these are almost unlimited.