« Rotations | Main | Risks and Ethics of Criminal Justice Ethnography: Alice Goffman's On the Run »

Wednesday, June 03, 2015

Agency Class Actions and Trials By Statistics

According to a hearing at the Senate Finance Committee last month, a crisis is brewing at Medicare's Office of Hearings and Appeals (OHMA).  OHMA is a small federal agency that hears billing disputes between the federal government and hospitals, doctors, nursing homes, and medical equipment providers.  As Medicare has stepped up efforts to recover excess billings, the backlog of cases with OHMA has tripled to more than 500,000 in just four years.  Worse yet, average wait times have mushroomed from 121 days in 2011 to 603 days in 2015. Even though Medicare is required to make such decisions, by statute, in 90 days, Medicare's workload is now so heavy that it takes OMHA 20 to 24 weeks to even enter new cases into its docket.  

In response, OHMA has adopted a fascinating new pilot program that allows medical providers with large numbers of similar billing claims to conduct "trials by statistics." Dubbed the "Statistical Sampling Initiative," a medical provider with more than 250 similar claims would have the option to try a small sampling of those claims before an administrative law judge and extrapolate the average result to the rest.   To do so, a hospital, doctor or other medical provider would meet with one of Medicare's "trained and experienced statistical expert[s]" to develop the "appropriate sampling methodology" and randomly select the sample cases to be extrapolated to the whole. Following a pre-hearing conference, all of the pending claims would be consolidated in front of a single Administrative Law Judge to hear all of the sample cases selected by the OMHA statistical expert.  The results of the sample cases would then be applied to all of the remaining cases.  Although we had nothing to do with this -- Medicare has a long history of using sampling techniques in enforcement actions -- the pilot sounds much like an approach to "trials by statistics" that Michael Sant'Ambrogio and I recommended for agencies in The Agency Class Action, 112 Colum. L. Rev. 1992, 2060-63 (2012).

Michael and I are currently studying aggregate litigation programs, like this, in administrative agencies with the Administrative Conference of the United States. So, we welcome your input, experiences or thoughts about other administrative programs that use similar techniques to resolve lots of cases. Medicare's program is interesting because it differs from the way the Supreme Court, and most of the administrative state, approaches adjudication.  Some thoughts about this new pilot program, and other ways agencies use aggregate adjudication, after the jump.

Although OHMA's Statistical Sampling Initiative is just in its initial stages,[1] it is notable for two reasons. First, it differs from Walmart v. Dukes, 131 S. Ct. 2541, 2561 (2011), where the Supreme Court appeared to reject "trials by formula" in federal courts. Among other things, the Supreme Court in Walmart worried that the "novel" use of statistical sampling could stretch hearing procedures too far by "abridging, enlarging or modifying" the substantive rights of the parties in a class action.  To the extent this is a problem for federal courts,[2] that decision does not apply to federal agencies that hear large groups of cases.  (Due Process is a different problem--but it is one reason why this is a volunteer pilot program).  Scholars who write about statistical adjudication may thus find some new practical use for these tools in administrative law courts.[3] 

Second, the pilot program moves away from administrative law's traditional, one-on-one approach to adjudication.  Even though agencies hear far more claims than federal courts, agencies have long avoided using tools, like class actions and other complex litigation procedures, to efficiently resolve large groups of claims.  Sant'Ambrogio & Zimmerman, supraAside from the growing resistance to class action practice in general, one reason for this goes back to the Administrative Procedure Act (APA) itself, which, in 1946, established rules for individualized administrative hearings.  Following a political battle over the growing power of administrative agencies, the APA separated the practice of “adjudication” from the agencies’ broad policymaking powers. Few rules existed in the APA, however, for agencies to resolve cases that fell in between the formal categories of policymaking and adjudication—such as when agency proceedings systematically affected groups of people or claims in the same way.

In this way, OHMA's pilot program represents one of a handful of federal administrative programs that have quietly bucked these trends—employing class action rules, collective claim handling and even the kinds of “trials by statistics” once embraced (and, more recently, rejected) by innovative federal judges around the United States.  Two other examples are the Equal Employment Opportunity Commission (EEOC) and the National Vaccine Program.  The EEOC, for example, created an administrative class action procedure, modeled after the Federal Rules of Civil Procedure, to resolve “pattern and practice” claims of discrimination by federal employees before federal administrative law judges.  29 C.F.R. § 1614.204 (2012) (establishing class complaint procedures).  And, since the early 1990s, the National Vaccine Program has used “Omnibus Proceedings,” that look a lot like federal multidistrict litigation, to pool together common claims that allege a vaccine injured large groups of children. 

However, there is good reason for agencies to proceed with caution and carefully consider the costs and benefits of using aggregate litigation techniques in administrative programs. Large cases always create new risks.  Even as aggregate adjudication promises more efficiency, consistency and sometimes, legal access, the sheer number of claims in aggregate litigation may (1) create "diseconomies of scale" -- inviting even more claims that stretch courts’ capacity to administer justice to many people; (2) replace individual hearings with a potentially faceless, unresponsive bureaucracy; (3) rely upon representatives tempted by the promise of large fees or power; and (4) increase the consequence of error.  In other words, just like many kinds of administrative systems, aggregate litigation struggles to deal with many different kinds of constituencies feasibly, legitimately, loyally, and accurately.


[1] As a result, right now very little is known about how many, and how often, medical providers are taking Medicare up on this new pilot program.  (And indeed, there is some reason to think that some medical providers will pass on this particular initiative).  The SSI also represents just one of many different programs that OMHA hopes will alleviate the existing backlog of Medicare claims.

[2] It remains unclear how far Walmart goes to undermine statistical adjudication.  Compare, e.g., In re Urethane Antitrust Litig., 768 F.3d 1245, 1257 (10th Cir. 2014) (“[The defendant’s] liability as to each class member was proven through common evidence; extrapolation was used only to approximate damages. Wal-Mart does not prohibit certification based on the use of extrapolation to calculate damages.”) and Alcantar v. Hobart Serv., No. ED CV 11-1600 PSG (SPx), 2013 WL 146323, at *4–5 (C.D. Cal. Jan. 14, 2013) (noting that Walmart was inapplicable for the calculation of wage-and-hour penalties) with  Brown v. Wal-Mart Stores, Inc., No. 5:09-CV-03339- EJD, 2012 WL 5818300, at *3 (N.D. Cal. Nov. 15, 2012) (collecting cases refusing to permit a trial-by-statistics approach after Walmart).

For a contrary argument arguably superseded by Walmart, see Blue Cross & Blue Shield of N.J., Inc. v. Philip Morris, Inc., 178 F. Supp. 2d 198, 247–62 (E.D.N.Y. 2001)(Weinstein, J.), rev’d in part on other grounds and questions certified, 344 F.3d 211 (2d Cir. 2003).  Notably, the Blue Cross case, one of the few trials by statistics that ever reached a jury, was not a class action.  Instead, much like the medical provider claims proceeding before OHMA, a single insurer sought to recover the costs for treating tobacco-related illnesses for all of its insureds.  Judge Jack B. Weinstein, well-known for his innovations in the area of complex litigation, described the case as a form of "structural aggregation"--where a single party, like an insurer, government agency or other entity acts on behalf of a large group of people much like a class action. 

[3] See Jay Tidmarsh, Resurrecting Trial by Statistics, 99 Minn. L. Rev. 1459 (2015); Edward K. Cheng, Essay, When 10 Trials Are Better Than 1000: An Evidentiary Perspective on Trial Sampling, 160 U. Pa. L. Rev. 955, 957 (2012); Alexandra D. Lahav, The Case for “Trial by Formula,” 90 Tex. L. Rev. 571, 610–12 (2012); Laurens Walker & John Monahan, Essay, Sampling Liability,85 Va. L. Rev. 329, 345-50 (1999); Robert G. Bone, Statistical Adjudication: Rights, Justice, and Utility in a World of Process Scarcity, 46 Vand. L. Rev. 561 (1993).

Posted by Adam Zimmerman on June 3, 2015 at 01:23 AM | Permalink


It's been a long time since I posted, but thanks for your comment. As it happens, many agencies and article I courts have power to aggregate cases, through both formal rules and informal practice (even if not many of them do so). Some of these non-Article III courts have promulgated formal class actions rules, like the Bankruptcy Courts, the Equal Employment Opportunity Commission (EEOC), and the Personnel Appeals Board. See, e.g., Fed. R. Bankr. P. 7023 (providing that Federal Rule of Civil Procedure 23 applies in adversary proceedings in the Bankruptcy Courts); 29 C.F.R. § 1614.204 (permitting the EEOC to hear class action claims involving federal employees). The EEOC continues to use class actions fairly often, even after the Supreme Court's decision in Wal-Mart v. Dukes.

But many non-Article III courts also aggregate claims and cases without adopting a formal procedure or rule to do so. Here are three interesting examples. First, the Office of Special Masters in the National Vaccine Injury Compensation Program (the “Vaccine Court” that I describe in another post) has not put forth a rule on aggregation. But, for some two decades, the Vaccine Court has
relied instead on its general authority to “determine the format for taking
evidence[,] … hearing argument[,]” and to “apply [its] expertise” from one case to another. Snyder ex rel. Snyder v. Sec’y of Dep’t of Health & Human Servs., No. 01-162V, 2009 WL 332044, at *2 (Fed. Cl. Feb. 12, 2009). Thus, when faced with large numbers of claims for compensation, the Vaccine Court developed “omnibus proceedings,” which work much like class actions, to more efficiently process claims involving the same alleged vaccine-related injury. Cedillo v. Sec’y of Health & Human Servs., No. 98-916V, 2009 WL 331968 at *11 (Fed. Cl. Feb. 12, 2009), aff’d, 89 Fed. Cl. 158 (2009), aff’d, 617 F.3d 1328 (Fed. Cir. 2010).

Second, the Court of Claims, long ago also concluded that it had the authority to certify class actions before it adopted a class action rule:

We could, of course, refuse to take any steps at all until a general rule
is prescribed, but we think the better road to follow, until we are
clearer as to the shape of the class-suit needs in this court and the
functioning of various class-suit devices, is to proceed on a case-bycase
basis, gaining and evaluating experience as we study and decide
the class-suit issues presented by individual, concrete cases coming up
for resolution. If we ultimately adopt a general rule, it will be in the
light of this ad hoc experience.

Quinault Allottee Ass’n, 453 F.2d at 1276.

Third, before launching the sampling program I discuss above, OMHA itself long relied on its implied authority to aggregate thousands of similar cases raised by health care providers against the federal government. See In re Apogee Health Serv., Inc., No. 769 (Medicare Appeals Council Mar. 15, 1999); cf. Chaves Cnty. Home Health Servs. v. Sullivan, 931 F.2d 914, 919-22 (D.C. Cir. 1991) (finding that mass consolidation and other actuarial tools in Medicare adjudication comports with due process). When doing so, it relies on grants of authority akin to those of many other non-Article III courts.
See, e.g., 42 C.F.R. § 405.1044 (providing for consolidation of two or more cases where the issues “are the same issues that are involved in another request for hearing” for purposes of “administrative efficiency”)

Posted by: Adam Zimmerman | Feb 3, 2016 5:25:53 PM

Very interesting article, thanks.
I came to your blog as a result of your Col. L.R. article.
We are grappling with the concept now. Specifically we have been discussing an agency rule stating that: ". . . for situations which are not covered by a specific [agency] rule, the Federal Rules of Civil Procedure will be followed to the extent that they are consistent with sound administrative practice." The specific question is whether this is a sufficient hook to get a Rule 23 class action in an Article I proceeding. I expect the concern is that in the absence of a specific rule it would not be consistent with sound administrative practice to go down that rule absent express statutory authority or a rulemaking in which specific rules and/or guidelines can be hammered out.

Thanks for the article.

Posted by: LLF | Feb 3, 2016 2:16:01 PM

Thanks for your comment. I've also been wondering how common it is to see this form of "structural aggregation," where one party has hundreds of similar disputes pending in administrative adjudication. I'd love to hear what others have to say about this.

Some have suggested that disputes between the federal government and federal contractors, which occupy a great deal of administrative adjudication, may also involve similar kinds of bundled claims. Federal and state agencies also frequently hear claims from insurers, government entities, utilities and unions, which in some cases, may also have a large number of common, derivative claims. (I believe FERC may have consolidated a number of such cases in the wake of Enron, but I haven't looked at that in years). However, I haven't looked at the data on this, yet, and I do think Medicare is an extreme example, particularly given its incredible caseload.

As for your more general comment about administrative law, I think you're correct that many claims are brought by individuals, with comparatively small claims, in what some people call "mass adjudication" systems--like social security, veterans, immigration, black lung and other kinds of federal workers compensation programs. Michael and I actually argue that other forms of collective litigation techniques may sometimes work in those cases, too. See Michael Sant'Amrobrogio and Adam S. Zimmerman, 112 Colum. L. Rev. 1992 (2012). But I think it's interesting that, to the extent agencies have begun to experiment with collective procedures at all, they're generally doing so modestly and outside of those contexts (with the exception of the National Vaccine Program).

Finally, I think administrative adjudication is more vast and varied than those programs, too. The Administrative Conference of the United States has been cataloging and studying federal administrative adjudications for a few years now, and there are dozens and dozens different and obscure adjudication programs in the Department of Agriculture, alone. And some of our most famous adjudications from administrative law -- SEC v. Chenery, Citizens to Preserve Overton Park v. Volpe, NLRB v. Wyman-Gordon, or one of my recent favorites, Ass'n of Irritated Residents v. EPA -- involved very complex and valuable claims commenced by or against holding companies, cities, unions, and businesses with many different stakeholders. Nevertheless, we treat all of these cases as "adjudications," subject to the same principles developed in an Administrative Procedure Act adopted almost 70 years ago.

Posted by: Adam Zimmerman | Jun 9, 2015 3:59:29 AM

I would think that there aren't many areas of law where it is typical for single parties dealing with the government to routinely have hundreds of similar disputes pending at the same time. Administrative law far more often involves large number of individual or small players with one or just a few claims pending at any given time.

Posted by: ohwilleke | Jun 8, 2015 8:09:50 PM

The comments to this entry are closed.