- BACKGROUND: The optimal approach for the diagnosis of hypothyroidism and hyperthyroidism in hospitalized patients is controversial. OBJECTIVES: To estimate the prevalence of undiagnosed thyroid disease among inpatients, review the usefulness of clinical signs and symptoms, and elucidate the characteristics of the sensitive thyrotropin (thyroid-stimulating hormone) (sTSH) test in this population. METHODS: We undertook a systematic review of the literature by conducting a MEDLINE search covering January 1966 through December 1996. Searching was conducted in duplicate and independently. Specific inclusion and exclusion criteria were predetermined. RESULTS: Prevalence of thyroid disease among inpatients is approximately 1% to 2% and is similar to the outpatient population. Absence of clinical features of thyroid disease lowers the pretest likelihood and makes screening even less useful. Presence of clinical features, especially those specific for thyroid disease (eg, goiter), may increase the pretest likelihood and increase the yield of testing. Acute illness reduces the specificity of second-generation sTSH tests for thyroid disease. The positive likelihood ratio associated with an abnormal sTSH test result in ill inpatients is about 10 compared with about 100 in outpatients. CONCLUSION: In unselected general medical, geriatric, or psychiatric inpatient populations, sTSH testing provides a low yield of true-positive and many false-positive results.