Beyond these factors, Geoffrey Rose’s preventive medicine strategy provides another potential reason for disappointing results from existing intensive primary care programs. In particular, a key insight from the strategy—which posits that “a large number of people exposed to a small risk may generate many more cases than a small number exposed to a high risk”4—is that addressing risk factors across an entire population can produce greater impact on outcomes than emphasis on a small group of high-risk individuals.
Consider an example of efforts to reduce unplanned hospitalizations.5 Because most of the individuals in a population are average risk (e.g., 80% of the population) rather than high risk (e.g., 0.5% of the population), the former will account for the majority of admissions. In turn, one would have to achieve far greater admission reductions among high-risk versus average-risk individuals to produce the same overall effects (e.g., achieving an overall 5% reduction in admissions would require a 54% decrease in admissions among high-risk individuals, compared with a 13% decrease among average-risk individuals).
These insights pose two important implications for provider organizations attempting to address avoidable healthcare costs in primary care. First, although high-cost patients may demonstrate more utilization than others on an individual basis, the cumulative costs for larger groups of average- or low-cost individuals as a group are also considerable. Therefore, interventions that focus on the latter can be easier to execute while generating meaningful reductions in overall spending.
For example, analyses of Medicare patients demonstrate that 17% highest cost individuals have almost twice the rate of low-value services as the rest of the population. However, despite the concentration of inappropriate spending among this high-cost group, it only accounts for 27% of total low-value services.6 Instead, the vast majority of the low-value care services (73%) occur among the larger group of non-high-cost patients. In turn, interventions focused disproportionately on high-cost patients may have only limited impact on reducing low-value services at the overall population level by leaving cost drivers among average- or low-cost patients unaddressed.
Second, compared with hotspotting-driven intensive primary care interventions focused on high-cost patients, initiatives targeting utilization across a broader range of individuals may be more cost-efficient and widely applicable. For example, teams in the ImPACT program, which included a physician, nurse practitioner, social worker, and recreation therapist, were responsible for providing healthcare services to small panels of 150 patients. In contrast, as a broader intervention in the same system, the VHA patient-centered medical home model uses similarly constructed teams (consisting of a physician, nurse, and administrative support staff) to deliver primary care services to nearly 1200 patients while matrixing part-time social workers, mental health, and pharmacist support across multiple patient panels. Though the costs for broader population-based interventions may be higher in the absolute sense due to the larger number of patients, these initiatives may be more cost-efficient on a per-patient basis. Regardless, pilot tests could also be conducted to evaluate local cost feasibility before full implementation.
Moreover, broader interventions can address utilization among high-cost patients without being specifically targeted to them, leading to wider applicability and impact. For example, because they share some of the same utilization drivers, high- and low-cost patients may both benefit from interventions that attempt to expand access to primary care services in order to decrease emergency department utilization. Such initiatives would be agnostic to individuals’ cost and risk levels, spreading financial overhead across a larger group of patients. Adopting this type of approach in its patient-centered medical home model, the VHA expanded same day primary care access for patients across all risk levels. In this model, patients at clinics that delivered the highest degree of team-based care had fewer emergency department visits and hospitalizations for ambulatory sensitive conditions.7