Little MCC research has focused on studying the numerous less prevalent combinations of MCC. However, two recent studies have addressed how less prevalent chronic disease combinations are cumulatively associated with healthcare costs. Sorace and colleagues used the Hierarchical Condition Categories (HCC) model to group conditions and found that Medicare beneficiaries could be classified into three distinct groups according to their chronic condition combinations: 1) patients who didn’t have chronic conditions as defined by the HCC model, 2) patients belonging to the 100 most prevalent chronic disease combinations, and 3) patients belonging to the remaining two million possible disease combination categories (Sorace et al., 2011). They found that approximately one-third of beneficiaries could be classified into each group, but that 79% of expenditures were associated with the third group of beneficiaries who had one of two million possible disease combinations. The authors concluded that the majority of Medicare expenditures can be attributed to a complex group of patients with less prevalent combinations of MCC; this results in a “long tail” distribution as displayed in Exhibit 8. In interpreting Exhibit 8, the reader should note that as there are over 2 million disease combinations calculated by this methodology, the figure’s X-axis would need to be extended over 8,000 fold to the reader’s right before both the expenditure and the population cumulative lines reached 100%. A follow-up study confirmed this complexity and found that national distribution of disease combinations changed over time (Sorace et al., 2013).
Exhibit 8: Percent of Disease Prevalence and Cost in the Beginning of Medicare’s Long Tail
Note on the Exhibit: The exhibit displays the first 250 Disease Combinations (ranked by prevalence) from the baseline HCC analysis as calculated by Sorace and colleagues (Sorace et al. 2011). Chronic disease combination classifications ( e.g., high, moderate and low) represent rough approximations; specific criteria for each classification have not been defined. Note that the left Y-axis represents the proportion of the population that is included in each unique disease combination, and is adjusted for the 32% of beneficiaries and 6% of expenditures that are associated with the no-MCC population. The right Y-axis represents the cumulative percent of the total population (red format) and the total expenditure (blue format). Note that approximately 75% of expenditures are associated with the 27% of patients that are not represented by the most prevalent 250 disease combinations. As there are over 2 million disease combinations calculated by this methodology, the figure’s X-axis would need to be extended over 8,000 fold to the reader’s right before both cumulative lines reached 100%.
There are two important concepts to be gleaned from these findings. First, the issue of “small cell size” limits the ability to intervene on or study a substantial number of patients with similar diagnoses. For example, given that approximately 65% of the over 32,000,000 beneficiaries studied had one of over 2,000,000 disease combinations the average cell size for a disease combination is in the range of 10 to 11 beneficiaries nationally.
The second important concept that can be learned from Sorace and colleagues is that healthcare costs for MCC patients with low-prevalence chronic disease combinations are significantly higher than those costs for patients with high prevalence combinations. As can be seen from Exhibit 8 approximately 75% expenditures are associated with the 27% of patients that are not represented by the most prevalent 250 disease combinations. To effectively address healthcare costs associated with MCC patients, efforts focused on patients with low-prevalence disease combinations must also be considered.
Finally it is important to note that the degree of complexity presented in Exhibit 8 is based on the observed frequency of disease combination phenotypes alone and does not include demographic traits ( e.g., sex, age, and race) or biological variables such as genomic variation. These additional variables may also be important in a given individuals health care plan.
Overall, research on less prevalent combinations of MCC represents a change in thinking from studying highly prevalent chronic diseases to understanding chronic disease complexity at a much more granular level ( e.g., the “long tail” distribution). Although other researchers have verbally confirmed similar research findings, Sorace and colleague’s work remains the only published literature on low-prevalence combinations of MCC the authors are aware of to-date. In the sections that follow, methodological considerations for MCC research are discussed with a special emphasis on the implications for conducting research on low-prevalence combinations of MCC.