Working Paper: NBER ID: w30669
Authors: Peter Conner; Liran Einav; Amy Finkelstein; Petra Persson; Heidi L. Williams
Abstract: Medical technologies can target care to patients identified through screening, raising questions of how broadly to screen for potential use. We explore this empirically in the context of a non-invasive prenatal screening, cfDNA, which is used to target a more costly invasive test that elevates miscarriage risk. Using Swedish administrative data on prenatal choices for pregnancies conceived between 2011 and 2019 – a period in which Swedish regions began providing coverage for the new screening – we document that coverage of cfDNA substantially increases cfDNA screening and reduces invasive testing. To assess the impact of counterfactual targeting of cfDNA coverage, we develop and estimate a stylized model of prenatal choices. We find that narrow targeting of cfDNA coverage can improve outcomes and reduce costs, while broader coverage also improves outcomes but with increased costs. These findings point to the potential gains from well-designed targeting of screening, but at the same time highlight the importance of the targeting design.
Keywords: precision medicine; prenatal screening; healthcare costs; policy design
JEL Codes: H31; I12; I13; I18; J13
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Coverage for cfDNA screening (G52) | Rate of cfDNA screening (J78) |
| Rate of cfDNA screening (J78) | Use of invasive testing methods (C90) |
| Coverage for cfDNA screening (G52) | Patient welfare (I11) |
| Coverage for cfDNA screening (G52) | Healthcare costs (I11) |
| Narrowly targeted coverage (Z00) | Rate of cfDNA screening (J78) |
| Narrowly targeted coverage (Z00) | Government spending (H59) |
| cfDNA screening (Y50) | Invasive testing costs (I11) |