USMLE Scores for Nepali Docs Will Remain Invalidated, Judge Rules

A Washington D.C. district court judge denied an emergency motion to restore score validity for 832 Nepali medical graduates whose U.S. Medical Licensing Examination (USMLE) scores were invalidated due to accusations of cheating.

Last week, Latika Giri, MBBS, a Nepali medical graduate who lives in Kathmandu, Nepal, launched a class action lawsuit against the National Board of Medical Examiners (NBME), alleging it didn’t follow previously established procedures when invalidating scores, which usually allow people to appeal before a final decision. It also claims NBME violated the Civil Rights Act for singling out Nepali test takers based on national origin.

Giri’s legal team also filed an emergency motion for preliminary injunction to restore score validity, which Judge Christopher R. Cooper denied. As of press time, the written opinion and order were not yet available online.

The court documents detail how the NBME gave Giri and more than 800 test-takers whose scores were invalidated until February 16 to decide between three options: request that NBME reconsider its decision, which could take up to 10 weeks; retake the test(s); or do nothing. If they chose either of the first two options, they waived their right to sue NBME.

Giri took Step 1 in Kathmandu, Step 2 in India, and Step 3 in Connecticut; all of her scores were invalidated, according to the court documents.

She was hoping to Match and begin training this summer, per the court documents, but now without their scores, Giri and fellow plaintiffs are ineligible to Match in this year’s cycle.

Bryan Carmody, MD, of Eastern Virginia Medical School in Norfolk, who has been keeping tabs on the situation on social media, pointed out on X (formerly Twitter) that Giri’s scores were not exceptional; she passed Step 1, scored 252 on Step 2, and scored 229 on Step 3. Giri’s team maintains that she did not cheat and studied hard to achieve her scores.

Earlier this month, the USMLE invalidated test scores for some examinees from Nepal after identifying “a pattern of anomalous exam performance associated with Nepal” which can be “indicative of prior unauthorized access to secure exam content,” according to a USMLE press release. However, in an emailed response to MedPage Today earlier this month, a USMLE spokesperson used slightly different language, saying the scores in question were “associated primarily, but not exclusively, with Nepal.”

The USMLE is taken by approximately 24,000 U.S. medical school graduates and another 22,000 international medical graduates (IMGs). Court documents show that the NBME has examined some data from other countries as well, including Jordan, Pakistan, and India.

Court documents provide some insight into how NBME found the alleged cheating and what red flags were raised. For instance, in USMLE Associate Vice President Daniel Jurich’s declaration, Exhibit A shows how Nepal scored significantly higher on USMLE compared to all other countries.

The document also details how Giri specifically answered questions that usually take around 90 seconds much faster — in 20 or 30 seconds — and had extremely high accuracy on those questions. Giri’s legal team had previously stated that when she didn’t know the answer to a question, she would quickly guess.

The declaration from USMLE director of customer experience Colleen Ward detailed how “an individual acting on behalf of the USMLE program was able to gain access to an exclusive online Telegram Messenger group in which USMLE exams are discussed, with approximately 1,300 members.”

The Exhibits also include screenshots of recalls posted to the Telegram group as well as messages telling test-takers not to leave the exam too early and thus raise attention.

Carmody also posted some of the screenshots on X. Carmody previously told MedPage Today that he had heard of USMLE “recalls” — which can drastically vary in quality and accuracy — for years.

The documents also note that the NBME received multiple tips about the situation at the start of last year, suggesting that large amounts of exam data were being shared or sold.