Thyroid Ultrasound Overuse by Primary Care Providers—Choosing Unwisely Continues Despite the Clinical Evidence

Background/Objectives: Inappropriate thyroid ultrasound (TUS) ordering by primary care providers (PCPs) is documented at academic centers, but there is insufficient evidence this practice is prevalent in general practice and, specifically, in a safety-net population. To address this gap in knowledge, we conducted a retrospective observational study and quantified non-guideline TUS ordering, identified associated factors, characterized indication-level concordance, and assessed downstream diagnostic interventions. Methods: This retrospective observational study (STROBE-compliant) included 405 PCP-ordered outpatient TUS studies in patients aged 18–86 years at a public safety-net health system in Moreno Valley, California, between March 2023 and March 2024. Thyroid US ordered for inpatients, specialist-ordered, and specialist-recommended studies were excluded. Concordance with the 2015 American Thyroid Association (ATA) guidelines was the primary outcome. Multivariable logistic regression, with Firth penalized-likelihood sensitivity analyses, and chi-square and Fisher’s exact tests were used. Results: Of 405 orders, 178 (44.0%) were non-guideline. Consistent with the general medical center population, patients were predominantly Hispanic (68.6%) and female (84.7%). Absence of a documented directed physical examination (PE) was associated with non-guideline ordering (OR, 4.54; 95% CI, 2.55–8.11), but not among indications whose classification is independent of examination findings (Firth OR, 1.10; 95% CI, 0.49–2.58). Thyrotoxicosis (overt or subclinical) and compressive-symptom indications were strongly associated with non-guideline ordering, although estimates were imprecise and biochemical indications had 95–100% non-guideline rates. Guideline-concordant TUS was associated with a higher fine-needle aspiration (FNA) rate (26.0% vs. 10.7%; p < 0.001), whereas the proportion of FNAs with Bethesda III–VI cytology was similar (42.4% vs. 47.4%; p = 0.793). Of 14 thyroidectomies, 11 (78.6%) showed benign pathology. Conclusions: Non-guideline TUS ordering, as previously discovered in academic centers, translates into safety-net primary care and is associated with biochemical and symptom-based indications rather than recommended patient characteristics. Guideline-concordant TUS was associated with more downstream FNA but a similar cytologic yield per biopsy. Our study strengthens the recommendation for targeted PCP education and point-of-care decision support.