As a researcher and educator familiar with much of the current state of data quality on employee wellbeing, health and work-related outcomes, I cannot overstate the value of taking a systems approach to learning and action. This necessarily involves humans thinking about and communicating with one another -- often aided by data, information and analytic results -- but primarily digging deep around their own and others’ assumptions about why the results they observe may not be the results they desire.
Artificial Intelligence (AI), big data solutions and other machine learning strategies can support policy and practice decision-making across many industries, but the adage “garbage in, garbage out” applies to these non-human forms of analysis just as it does to the human forms. Much of the data currently used in predictive modeling or big data solutions relies on claims data, mostly medical claims information available in non-public datasets held by commercial companies. The best of these data sources include integrated data that link across medical, pharmacy, workers’ compensation, disability, and absence activity to better capture consequential outcomes for a business or community beyond medical cost reduction. But these integrated databases are the exception, most often representing only a fraction of any given company’s overall workforce and susceptible to variable quality standards around data capture and person-level linking across data sources. If we feed these same data into an AI solution we would be basing any analyses on the same underlying data warts and all, likely excluding part-timers and unbenefited workers who wouldn’t have had a chance to appear in the data or treating eligible non-utilizers as those without a health need. When these groups are excluded, as they naturally can be in utilization-based claims data, we get distorted results that represent current claims activity and costs for a non-random portion of the employee population. This severely limits decision-making quality and the potential impact of any policy or practice solutions aimed at employee health improvement, thus undermining efforts that are so critical to sustaining a healthy and high performing workforce.
As a field, we can stand to learn more about the adjustments that organizations can make in policy and practice to ensure that all workers have the opportunity to engage fully in the labor market, after all, work is a social determinant of health. We should also expect that a high functioning health care system would be a key component of longer, healthier and more fulfilling working lives. We encourage newer initiatives, whether tech-based or not, to embrace a systemic approach to measurement if we are to achieve systemic improvements in worker health and performance.
Here are a few links to resources that can be helpful in building systemic thinking and action in your own work.
The photo of the iceberg model that is associated with this blog post is freely available for download at the Academy for Change website . The tip of the iceberg in this context represents what is easily observable, such as claims activity. But understanding what generates this claims activity is where the strategic leverage for change lies. What assumptions and values lead to the observed patterns? For example, do all employees have an equal chance of being captured at the tip of the iceberg? If all employees were fully benefitted, would we expect activity patterns based on health needs, access to providers and/or other factors? Do certain departments or regions have more activity of certain types than others, for example, do call center employees have different types of health and absence profiles than the senior managers in other departments? What systems structures, job types, management styles, locations, shift schedules, etc. might account for these patterns. What values and beliefs shape the different work conditions and individual behaviors observed?
Understanding context and engaging in context redesign demands a systems perspective. There are companies that offer such workshops and trainings (a topic we are currently studying). Often times these types of trainings start with the easily visible behavior that is a target of change efforts, the tip of the iceberg. Given enough time with a team of people, for example in a 2-day workshop with cross-departmental teams, the participants begin to discuss behavioral patterns across their organization, they compare notes with each other and present their ideas, they challenge each other and dig deeper. They begin to share thoughts around why these behaviors exist and what might explain these patterns that could be influenced by the employer. They begin to understand that an overweight problem in the office might be influenced by the sedentary nature of the work and the example set by senior leaders to eat lunch at their desk and be heads-down in work all day. They begin to see that the deepest part of the iceberg reveals these assumptions and beliefs around what it means to be a hard worker and move-up in the organization and that if those ideals aren’t changed, very little of the attention paid to changing the surface behaviors will result in any lasting improvement.
Start by watching this video on systems thinking and action sponsored by the CDC. You’ll find useful tips on key questions to ask and tools that can be used to focus on important leverage opportunities. https://m.youtube.com/watch?v=Fo3ndxVOZEo
If you are looking for a broad overview of key terminology in business around creating corporate value and detailed advisory around measurement and results reporting I recommend this Bloomberg Next webinar and supporting materials. Create Corporate Value Through Environment, Social and Governance (ESG) Programs
Finally, being a systems thinker and actor does not require a fancy degree or math skills. While there are many resources available to those with interests of a more mathematical and modeling nature, I want to end on a very human note. While AI and machine learning applications may be better able to mathematically model many more scenarios and solutions simultaneously than humans, they remain limited to the data and algorithms supplied. I recently had the opportunity to participate in a very human project, a “poverty simulation”. This was primarily designed to help individuals who may not have experience with poverty (either through their own history or as a service provider) develop a greater understanding about the experience of being poor. Of course a one-day event can’t possibly provide a comprehensive or authentic experience, but the conversation among participants that resulted from this simulation demonstrated profound engagement and learning. This type of experiential learning and communication around mental models, system structures and observed behavior can be powerful for deepening engagement around collective problem-solving. To learn more about the poverty simulation, check out their site
If you are interested in learning more about or supporting research and education opportunities based on systemic principles, please contact me at email@example.com.