CalAIM, Complex Networks, and Social Referrals: Can aviation models be a guide?
The pursuit of social referral vendor market dominance continues, unheeded, despite the market forces in California seemingly settling into two camps: organizations that have these systems and do not use them, or organizations that use them as a component in a multimodal set of tools.
However, the fatal flaw of for-profit referral vendors remains this “one ring to rule them all approach.” Many others and I have observed over the course of several years, through Whole Person Care and now CalAIM, that no community has yet to attain a comprehensive social mesh, in some cases embittering the counties or organizations at the very mention of such a singular solution. The current environment is trending towards communities designing solutions through a host of cloud-based, low-code platforms; the siren sound of a single technology solution is beginning to wane, and the desire for systems that are interoperable and purpose-built has begun to rise.
The irony of for-profit social referral vendors is that despite their product being a network solution, not much (or any) network science has been applied to its rollout. Generally, onboarding happens en masse with communities or in a sequence that has little to do with the actual referral demand. And while having contracts signed for accounts looks good on paper, it does little to appreciate the actual transactions happening in a community.
Below is a depiction of a network's social referral transactional volume--anonymized and truncated--for a given period. The larger nodes (circles) represent the higher volume of referrals received, or more demand for that service. The edges (lines) represent the direction of a referral (email, phone, in person, or system). If we assume that this pattern is generally stable, one can infer that dependencies exist, such that if an organization with high connectivity (Org #31) is not onboarded to a referral network, then none of its partners will participate. After all, if 90% of your referrals go to one organization, the referral method they use as a partner is likely the method you adopt. Furthermore, the way the first group of organizations is onboarded must be designed thoughtfully. What is the minimum amount of organizations required to make a network feasible? This will vary, but it is a combination of the organization's connectedness, capacity, and, most importantly for the system provider, the time staff spend in the system compared to other solutions. After all, if I send a referral, but then call, send an email, or log data into another system, over time, the referral system loses any practical utility.
For pure technology vendors, despite any marketing slogans, connecting people to services was happening long before web-based applications. The 211s and United Ways in the US have been doing this well before the internet was accessible and publicly available, and will continue to do so. Sending a referral form from point A to point B was hardly ever the challenge, even the ‘closed loop process’ doesn’t mean a needed service is provided, it could quite simply mean an organization has confirmed receipt of a referral. However, just like a driver is at the other end of Uber’s application, or a cook on the other end of Door Dash technology, a coordinator, case manger or provider is at the other end of that referral system, and they are the ones providing the actual service in that community--generally at a lot lower cost than any technology solution and with more impact to the individual in need.
New technology doesn’t necessarily fix old problems.
The Landscape
What is particularly interesting in the evolution of participating CalAIM organizations is the emergence of hub and spoke networks that seek to bundle services and streamline routing to manage care plans and other networks. However, the binding of these services is not necessarily comprehensive to any one population group, such as prenatal and senior support services--a vestige of a wholly grant-funded organization moving to a fee-for-service model. Thus, while hub networks may be forming in communities, they currently form around administrative efficiencies (usually billing submissions), and not necessarily around a particular client group with complex needs.
However, as effective as these models are, they will not last.
Ultimately, the hub networks will self-organize based on local service need patterns that are both efficient for the client and less costly for the managed care plans. This will be influenced by a combination of organizations repurposing their staff for similar workflows with the same client group, and managed care plans directing referrals to the most extensive or comprehensive networks. This phenomenon is noted as the degree of connectivity in graph theory and is why someone on LinkedIn with 10k followers can post a random inspirational quote or useless infographic and gain 100 followers when others may take time to write articles and only gain one or two (No, I’m not bitter at all). This “rich get richer” cycle overlooks a critical issue in the social service space: what about the small hubs that are not part of any ‘big network’ but provide critical services to the community? How do they get noticed?
Aviation Networks
Transportation networks offer some very convincing insights into how exactly social referral networks might evolve in the coming years. When studying transportation hubs, where behavioral decisions are at play, such as cost and time, the networks tend to self-adjust to the most optimized routes. When airlines offer new routes, it is with one of two intentions: to induce demand to a location based on a lower price than an adjacent city, or because they anticipate there is enough demand already to fill seats with competing airlines. This same type of self-organization occurs with Uber based on arrival patterns, such that passengers are able to reserve future rides while adjusting for any operational delays to the minute and even forecasting route times based on traffic patterns.
Aviation business models are differentiated through low-cost carriers and full-service (long-haul) providers, think Jet Blue or United Airlines. Each serves a particular type of market, local/regional and low-cost, or long-haul, nonstop direct flights. Low-cost carrier flights have shorter range, narrower bodies, and fewer amenities, but are generally still faster than a car or train if your destination is closer to a secondary airport. However, the flight options to secondary airports give key insights into changing demand patterns, which in turn impact the routing of long-haul flights to service these hubs. Many full-service carriers operationalize flights in hubs through ‘banks’ such that if you look at flight options, you might notice there could be ten flights throughout the day to Chicago from another hub, but only two to Sacramento on the same carrier.
But it's less about city population sizes and more about the entire flight plan. These two flights to Sacramento indicate that travelers from all over the world may be landing in San Francisco and continuing through into Sacramento. However, a point-to-point, low-cost carrier would not have this perspective, but a long-haul operator would. Therefore, it is in the mutual interest, financially and operationally, for both low-cost carriers and long-haul carriers to cooperate. The long-haul carriers can focus on hubs and efficient higher capacity routing, while low-cost carriers can optimize the number of flights needed to secondary point-to-point networks based on final destinations from long-haul carriers. Frequency changes from secondary airports to hubs indicate that an increase in demand has occurred. This is important for both types of carriers; in fact, this is something called the ‘Catchment area’ and is factored into routes, profitability, and scheduling.
Co-exist
Social Referral Technologies
Larger, more technology-focused referral vendors should not seek to compete with local hubs, but offer a different type of service, the 'long-haul carrier’, one that seeks to reduce the number of referrals by reducing the degrees of connectivity in smaller networks. This is known as the ‘small world concept’ but more prominently known as the “Six degrees of Kevin Bacon”, where you can connect yourself to the actor through a series of mutual acquaintances. This same concept should be applied to referral matching, not matching a client to just an organization that provides the service, but to all services in the least amount of referrals possible. This eventually creates shortened paths for clients that rely upon fewer ‘hops’, and in turn, less delay in the services they need. Additionally, they should focus on the bridge from social to clinical, and track a referral in its entirety from the clinical to the social domain. This is another gap, but referral vendors are better positioned to meet the demands of healthcare interoperability and security requirements in health system networks.
Providers
CalAIM contacted organizations will either need to ‘deepen’ their few services to offer more service capacity for a greater area, or tailor their service offerings to meet the complex needs of a particular, narrower set of a population (Pregnant and unhoused, justice-involved and behavioral health...etc). Both of these approaches are underway with several organizations I work with, but the challenge starts to look a lot more daunting than the grant-funded options of the past, where you managed a predetermined budget, and the services provided did not impact this. Now, organizations have to forecast service demand and advertise it as such through CRM tools in hopes of getting reimbursed at a sustainable rate for staff and operations. This model starts to look a lot like the ‘catchment area’ that low-cost carriers deal with: how many potential clients need services, what are the alternatives, and what additional needs will these clients have that they can potentially meet? Unfortunately, without a ‘top down’ view of their areas, many organizations will never be able to piece together total service needs for a given area since no one hub has the full picture.
Managed Care Plans
In the aviation world, it is the traveler who drives these models, since they are making the decisions that pay the airlines. In the CalAIM space, managed care plans drive this model as the “payors.” While care coordination and referrals can be reimbursed, there is nothing more efficient about a closed-loop referral system repeating the same process of a multiorganizational hop; it is human heuristics, repeating the same shortcuts, sending clients to services in a sequential manner, or one at a time. This is the same reason provider/pharmacy organizations tend to prescribe the same types/brands of medications, based on what they know. After all, how does one instantly know when something on a network has changed? How can local hubs know that a new community-based organization (CBO) is operational? How does the presence of that new CBO impact the service capacity of the larger network, and how may this impact the routing of referrals? This gets into complex networks where small changes can have unpredictable effects across the broader network, which only drives reactionary efforts to identify and adjust. Managed care plans should be more concerned with this referral routing process, but not just if someone sent a referral from Point A to Point B using a closed-loop referral system, but why that Point B? Is it the only resource in the area? Is it the only one that the organization knows of? There should be some sort of incentivized referral reimbursement that considers additional reimbursable services provided by the organization in lieu of referring the client yet again. After all, doesn't the cost saved by not having additional intakes, coordination, and referrals help make them more responsive to clients' needs and improve savings for the managed care plan? Lastly, this reimbursement incentive helps shape the way future organizations will strategize around services to provide.
Diagram 2: A county with competing referral networks and multiple hubs.
Diagram 3: A county with self-organized local hubs and a centralized social referral hub
Conclusion
Ultimately, the environment described here today will not be something that can be accomplished in just a few months. But, it certainly will not be accomplished by just buying a single system or simply relying on any one organization or person to be omnipotent in the social resource space. Since CalAIM providers will need to have better abilities to forecast demand and do outreach, they must reorganize in ways that make them efficient and scalable. Social referral vendors do not really have any competitive advantage with the current model of relying on scale alone. After all, this just makes the product more price-sensitive when compared with competitors. Instead, their competitive advantage should be on algorithm models that have optimized referral routes in communities, and further, offer potential predictive service demands based on referral patterns. This, in some ways, is similar to how some data analytics organizations have better provider-to-patient matching algorithms, reducing patient turnover. Lastly, managed care plans should incentivize organizations that can provide more 'in-house' services, such that their reimbursements are higher per client, and thus the efficiency of service delivery exponentially improves their sustainability. At the same time, managed care plans should evaluate the actual utility of social referral vendors in their networks, both in terms of user activity and effectiveness to extend the navigational visibility of the referring organizations. After all, if users are not gaining any new insights on resources in the area, then it is simply another platform they enter data into.
Ultimately, the state of California was able to quantify the cost of a patient with complex needs to begin these programs (WPC/CalAIM). However, the question of how many organizations a client with complex needs has to visit remains a mystery. And despite a 'no wrong door approach,' the process of even beginning to find these doors remains elusive.
Comments
Post a Comment