Professional vs. ad hoc interpreters: what research shows

In a widely cited comparison of interpreter types, 12% of professional interpreters' errors had potential clinical consequences, against 22% for ad hoc interpreters such as family members and untrained staff, and 20% when there was no interpreter at all. Training mattered more than experience: for professional interpreters with at least 100 hours of training, the figure was 2%.

That study, Flores et al., 2012 in two Massachusetts pediatric emergency departments, fits a pattern that runs through two decades of research. This page summarizes 28 peer-reviewed studies and systematic reviews. Each was checked against its abstract or full text, and each entry says what the study measured, because many of these figures are widely misquoted.

Interpreter errors: professional vs. ad hoc

• Flores et al., 2003, pediatric clinic, 13 Spanish-language encounters: 396 interpreter errors, a mean of 31 per encounter. 77% of ad hoc interpreters' errors had potential clinical consequences, against 53% of hospital interpreters' errors. One ad hoc interpreter was an 11-year-old sibling.
• Flores et al., 2012, 57 emergency department encounters: 1,884 errors. Errors with potential clinical consequences were 12% of the total for professional interpreters, 22% for ad hoc interpreters and 20% with no interpreter. Professional interpreters with 100 or more hours of training made a median of 12 errors against 33 for those with less.
• Nápoles et al., 2015, a public primary care clinic in Alameda County, California: 54% of speech segments were interpreted inaccurately by ad hoc interpreters, against 25% by in-person professionals and 23% by professionals on video.
• Lauridsen et al., 2024, a systematic review of 7 studies: fewer errors with professional in-person interpreters than with family members, but not significantly fewer than with untrained medical staff. The gap tended to be larger for more severe diagnoses.

What systematic reviews conclude

• Karliner et al., 2007, 28 studies: professional interpreters were associated with better care than ad hoc interpreters in all four areas studied, from communication errors to clinical outcomes and satisfaction. They appear to bring care for patients with limited English proficiency (LEP) close to that of patients without a language barrier.
• Flores, 2005, 36 articles: quality of care suffers when LEP patients need an interpreter and do not get one. The fewest errors occur with trained professional interpreters or bilingual clinicians.
• Heath et al., 2023, 29 reports: in-person professional interpreters produced the highest satisfaction and best communication, and any interpreter was better than none.
• Boylen et al., 2020, hospitalized children in Spanish-speaking families: ad hoc or no interpreter was inferior to a professional interpreter of any mode, in person, phone or video.
• Silva et al., 2016, palliative care for cancer patients: without professional interpreters, patients and families understood diagnosis and prognosis less well, and pain and anxiety were managed worse.

Harm to patients with limited English

These studies compare LEP and English-proficient patients. They measure the size of the language-barrier problem, not the effect of any one kind of interpreter.

• Divi et al., 2007, six Joint Commission-accredited U.S. hospitals: 49.1% of adverse events reported for LEP patients involved physical harm, against 29.5% for English speakers, and 52.4% stemmed from communication errors, against 35.9%. These are shares of reported events, not rates per patient.
• Khan et al., 2020, pediatric units of seven North American hospitals: 17.7% of children whose parents had limited comfort with English had at least one adverse event, against 9.6%, an adjusted odds ratio of 2.1.
• Yin et al., 2026, the same seven-hospital cohort: before a structured family-centered rounds program, preventable adverse events were 18.8% against 7.6%. After it, 6.4% against 5.3%, no longer a significant gap.
• Cohen et al., 2005, a children's hospital: no significant overall link between needing an interpreter and serious medical events, but a twofold increase in odds for Spanish-speaking families (odds ratio 2.26).

Hospital stays, readmissions and return visits

• Lindholm et al., 2012, 3,071 LEP patients at a Massachusetts hospital: without professional interpretation at admission, or at both admission and discharge, stays were 0.75 to 1.47 days longer. Unadjusted 30-day readmission was 24.3% with no interpreter on either day, against 14.9% with one on both.
• Rawal et al., 2019, two Toronto hospitals: LEP patients with heart failure were readmitted within 30 days 18.1% of the time against 13.9%, and with COPD 15.6% against 11.8%. There was no significant difference for pneumonia or hip fracture.
• Ngai et al., 2016, an urban adult emergency department: unplanned return within 72 hours was 5.0% for LEP patients against 4.1%, an adjusted odds ratio of 1.24.
• Hampers and McNulty, 2002, a Chicago pediatric emergency department: without a professional interpreter or bilingual physician, children were more likely to be tested (odds ratio 1.5), admitted (2.6) and given IV fluids (2.2).
• Karliner et al., 2010 found similar stays and costs for non-English and English speakers but higher odds of readmission (1.3), and Kulkarni et al., 2025 found length-of-stay results mixed in the 8 of 11 studies that measured it. Treat stay and readmission effects as associations from specific hospitals.

Consent, discharge instructions and pain

• Schenker et al., 2007: informed consent was fully documented for 53% of English speakers' procedures and 28% of LEP patients', at a hospital with on-site interpreters.
• Lee et al., 2017: after interpreter phones were placed at every bedside, adequately informed consent among LEP patients rose from 29% to 54%, though it stayed below English speakers.
• Karliner et al., 2012: at discharge, 50% of LEP patients knew what kind of follow-up appointment they had, against 66% of English speakers.
• Gutman et al., 2018: in a pediatric emergency department, using a professional interpreter was associated with complete discharge education (odds ratio 7.1).
• Plancarte et al., 2021: after surgery for a limb fracture, 86.4% of children from LEP families received any pain medication, against 96.8%.

How often professional interpreters are used

• Schulson and Anderson, 2022, a national CDC survey of office-based physicians in 2016: 29.5% regularly used professional interpreters with LEP patients, and 39.9% never did.
• Schenker et al., 2011, two San Francisco Bay Area hospitals: LEP inpatients reported a professional interpreter with physicians at admission 17% of the time, and with nurses 4%.
• Diamond et al., 2009: resident physicians described underusing interpreters under time pressure as "getting by." The study is qualitative and gives no rate.
• Gutman et al., 2025, a systematic review of 39 interventions: 95% reported at least one increase in professional interpreter use, most often through clinician education, easier access and electronic health record changes.

Reading these studies carefully

• The Flores 2003 figure of 77% is the share of ad hoc interpreters' errors that could matter clinically, not how often ad hoc interpreters make errors.
• The Flores 2012 figures of 12% and 22% are shares of errors, not shares of visits. Professional interpreters made errors too; training of 100 hours or more is what separated the best.
• Divi 2007 does not show that LEP patients have more adverse events. It shows their reported events were more often harmful and more often caused by communication.
• Odds ratios are not risk ratios. "About twice the odds" is more accurate than "twice as likely."
• Most studies are observational and many come from one hospital. They show consistent associations, not proof that interpreters cause every difference.

What the rules say

The HHS rule under Section 1557 of the Affordable Care Act, 45 CFR 92.201(e), bars covered health programs from requiring patients to bring or pay for their own interpreter. It allows an accompanying adult to interpret only in an emergency or when the patient asks for it in private with a qualified interpreter present, and a minor child only as a temporary measure in an emergency. See Section 1557 and healthcare language access and how to hire a medical interpreter.

For the credentials that separate trained interpreters from bilingual staff, see CCHI vs. NBCMI certification and bilingual staff vs. certified interpreters.