When Can an ERISA Claimant Get Discovery Beyond the Administrative Record? A New York Court Draws the Line

In De Mello v. First Unum Life Insurance Company, No. 25-cv-7933 (LJL), 2026 WL 2032059 (S.D.N.Y. July 14, 2026), United States District Judge Lewis J. Liman denied an ERISA long-term disability claimant’s motion to compel discovery into matters outside the administrative record. Plaintiff participates in an employee welfare benefit plan sponsored by his law firm employer, which provided long-term disability coverage through a policy issued and administered by Defendant. After contracting COVID-19 in December 2021 and being diagnosed with long COVID, Plaintiff submitted a claim for long-term disability benefits, which Defendant denied based on the file reviews of three doctors. Plaintiff brought suit under Section 502(a)(1)(B) of ERISA, and then moved under Rule 37 to compel responses to interrogatories and document requests seeking information about the compensation Defendant paid to its file-reviewing physicians and their third-party vendor, the doctors’ claim denial rates, Defendant’s handling of long COVID claims generally, financial incentives for claims personnel, and reserve information.

What standard governs discovery outside the ERISA administrative record?

The court explained that when reviewing claim denials under either the arbitrary and capricious or de novo standard, district courts typically limit their review to the administrative record that was before the plan when it denied the claim. A court may consider evidence outside that record only upon a showing of good cause. To obtain discovery in aid of such a showing, the majority rule in the Second Circuit requires the movant to show a reasonable chance that the requested discovery will satisfy the good cause requirement, which in turn demands specific factual allegations supporting the request. The court noted that some judges in the district instead apply the ordinary standards of Rule 26, but that even under that approach courts must weigh the burden and expense of the proposed discovery against ERISA’s policy interests in minimizing the cost of claim disputes and ensuring prompt claims resolution. The court found that it would deny the motion under either standard.

Does a plan administrator’s conflict of interest justify extra-record discovery?

The court reaffirmed that a structural conflict of interest, which exists when the same entity both evaluates and pays claims, does not by itself constitute good cause to look beyond the administrative record. Because the lion’s share of ERISA denials are made by administrators operating with such dual roles, treating the conflict as per se good cause would leave little of the principle that judicial review is generally confined to the record. A conflict can rise to the level of good cause only when bolstered by specific allegations of an additional factor, such as a lack of established criteria for deciding appeals or a failure to maintain written claim review procedures. The court found it questionable whether Plaintiff had identified any such additional factor. Plaintiff’s allegations amounted to disagreements with the conclusions the reviewing doctors drew from the evidence, including that the reviewers lacked expertise in long COVID, mischaracterized the evidence, and discounted his self-reported symptoms. The court held that a claimant’s disagreement with a reviewing physician’s opinion cannot alone justify extra-record discovery, because such disagreement is inherent in every ERISA benefit-denial challenge, and treating it as sufficient would open the door to extra-record discovery in nearly every case involving a conflicted administrator.

Why did the court find the specific discovery requests improper?

The court held that Plaintiff had not shown his requests were relevant or proportional to the needs of the case. As to the interrogatories seeking the amounts Defendant paid the reviewing doctors and their vendor, the court observed that Defendant had already produced the documents describing the bases for its claims personnel’s incentive compensation and had agreed to produce the vendor invoices specific to Plaintiff’s claim, and that Plaintiff had not explained how aggregate three-year compensation figures would show that either doctor depended on the vendor’s work or had an incentive to bias results. Defendant also represented, without contradiction, that it did not have access to the amounts the vendor paid the doctors. As to the interrogatories seeking the doctors’ denial and termination rates and Defendant’s overall handling of long COVID claims, the court explained that bare denial rates do not prove bias or conflict of interest without a showing that each underlying decision was unreasonable, and that supplying that context would require mini-trials on other claims. Finally, the court found that the requested reserve information was not relevant because Defendant set reserves based on statutory requirements, actuarial averages, and historical experience rather than on a claim-by-claim basis, so the information did not speak to whether the result in Plaintiff’s case was the product of bias. The court denied the motion to compel.