While effective in outpatient settings, the Wells score was not effective at detecting deep vein thrombosis in an inpatient setting, according to Dr. Patricia Silveira and her associates from Harvard Medical School, Boston.

Of 1,135 patients included in the study, 137 had proximal DVT. DVT incidence in the low, medium, and high pretest probability groups were 5.9%, 9.5%, and 16.4%, respectively. Although statistically significantly different, this is a very narrow range, in contrast to findings in previous studies, where incidence among the three groups was 3.0%, 16.6%, and 74.6% respectively.

The AUC for the accuracy of the Wells score was 0.6, only slightly better than chance. The failure rate in the low pretest group was 5.9%, and the efficiency was 11.9%.

“In inpatients,Wells DVT scores are inflated by comorbidities and nonspecific physical findings common among hospitalized patients, leaving very few patients in the low-probability Wells score category, and many patients without DVT in the moderate- and high-probability categories,” Dr. Erika Leemann Price and Dr. Tracy Minichiello wrote in a related editorial.

Find the full study and editorial in JAMA Internal Medicine ( doi:10.1001/jamainternmed.2015.1687 ; doi:10.1001/jamainternmed.2015.1699 ).