Projected costs of single-payer healthcare financing in the United States: A systematic review of economic analyses

Abstract

Background: The United States is the only high-income nation without universal, government-funded or -mandated health insurance employing a unified payment system. The US multi-payer system leaves residents uninsured or underinsured, despite overall healthcare costs far above other nations. Single-payer (often referred to as Medicare for All), a proposed policy solution since 1990, is receiving renewed press attention and popular support. Our review seeks to assess the projected cost impact of a single-payer approach.

Methods and findings: We conducted our literature search between June 1 and December 31, 2018, without start date restriction for included studies. We surveyed an expert panel and searched PubMed, Google, Google Scholar, and preexisting lists for formal economic studies of the projected costs of single-payer plans for the US or for individual states. Reviewer pairs extracted data on methods and findings using a template. We quantified changes in total costs standardized to percentage of contemporaneous healthcare spending. Additionally, we quantified cost changes by subtype, such as costs due to increased healthcare utilization and savings due to simplified payment administration, lower drug costs, and other factors. We further examined how modeling assumptions affected results. Our search yielded economic analyses of the cost of 22 single-payer plans over the past 30 years. Exclusions were due to inadequate technical data or assuming a substantial ongoing role for private insurers. We found that 19 (86%) of the analyses predicted net savings (median net result was a savings of 3.46% of total costs) in the first year of program operation and 20 (91%) predicted savings over several years; anticipated growth rates would result in long-term net savings for all plans. The largest source of savings was simplified payment administration (median 8.8%), and the best predictors of net savings were the magnitude of utilization increase, and savings on administration and drug costs (R2 of 0.035, 0.43, and 0.62, respectively). Only drug cost savings remained significant in multivariate analysis. Included studies were heterogeneous in methods, which precluded us from conducting a formal meta-analysis.

Conclusions: In this systematic review, we found a high degree of analytic consensus for the fiscal feasibility of a single-payer approach in the US. Actual costs will depend on plan features and implementation. Future research should refine estimates of the effects of coverage expansion on utilization, evaluate provider administrative costs in varied existing single-payer systems, analyze implementation options, and evaluate US-based single-payer programs, as available.

Fig 1. PRISMA flow diagram.

Fig 2. Net savings for single-payer in first year of implementation, sorted by net cost/savings.

The median finding was savings (−3.46% of total health system costs), and analyses of 19 of 22 plans found net savings.

Fig 3. Costs versus savings for single-payer by category.

Plans listed in order by year. Simplified payment administration was the greatest source of savings, for a median of 8.8%.

Fig 4. Net costs or savings versus assumptions in plans and analyses, sorted by net costs/savings.

The 3 models that found net costs in the first year (Hsiao 2011 Low Cost Sharing, CBO 1993 SP2, and White 2017) shared specific policy choices including low or no cost sharing (copays), rich benefit packages, and a lack of savings captured from reduced medication/medical equipment costs.

Fig 5. Net costs versus the inclusion of different analysis features.

Each estimate comes from a separate linear regression of net costs and a binary predictor. Error bars represent 95% confidence intervals.

Fig 6. Net cost (%) versus utilization rise and savings magnitude.

(A) Utilization rate; (B) administrative savings; (C) medicine and equipment savings; (D) efficiency gains and fraud reduction; (E) Medicare payment rate. Each dot represents 1 model. The red lines represent linear regressions, with displayed results indicating the regression equation (including intercept and slope) and R² (proportion of variation explained). The regression line for Medicare payment rate (E) was omitted due to the preponderance of 0 values (73%, or all but 6, of the 22 models). Higher utilization was associated with greater costs, whereas the magnitude of administrative and medication/equipment savings was associated with reduced net costs.