Holmstrom v. Metropolitan Life Insurance

In the United States Court of Appeals For the Seventh Circuit No. 09-2173 L ANETTE H OLMSTROM, Plaintiff-Appellant, v. M ETROPOLITAN L IFE INSURANCE C OMPANY, et al., Defendants-Appellees. Appeal from the United States District Court for the Northern District of Illinois, Eastern Division. No. 1:07-cv-06044—Robert M. Dow, Jr., Judge. A RGUED F EBRUARY 11, 2010—D ECIDED A UGUST 4, 2010 Before K ANNE, W OOD , and H AMILTON, Circuit Judges. H AMILTON, Circuit Judge. This case illustrates the dif- ficult problems presented by claims for disability insur- ance by people with serious and painful conditions that do not have objectively measurable symptoms. Plain- tiff Lanette Holmstrom worked as a senior training specialist at a large credit management company. She participated in an employee welfare benefit plan admin- istered by defendant Metropolitan Life Insurance Com- 2 No. 09-2173 pany (“MetLife”). Holmstrom stopped working in Janu- ary 2000 when she developed a painful nerve condition in her right arm. MetLife began paying disability benefits under an “own-occupation” standard. Three surgeries failed to remedy the condition, and Holmstrom was diagnosed with complex regional pain syndrome (“CRPS”). After Holmstrom’s “own-occupation” benefits expired, she submitted a disability claim under the more stringent “any-occupation” definition that applied to longer- term benefits. MetLife approved that claim in July 2002 and began paying benefits. MetLife performed a periodic review in 2005. It determined then that Holmstrom was no longer disabled and terminated her benefits. After MetLife upheld its decision on administrative appeal (Holmstrom’s final administrative remedy), Holmstrom filed suit in federal court under the Employee Retirement Income Security Act of 1974 (“ERISA”). See 29 U.S.C. § 1132(a)(1)(B). Holmstrom voluntarily dismissed the action when MetLife offered a second administrative appeal, which yielded the same result. Holmstrom then returned to federal court, filing this second ERISA action to recover benefits. MetLife counterclaimed to obtain a setoff based on disability insurance benefit payments that Holmstrom received from the Social Secu- rity Administration. In a careful opinion describing the case as a close one, even under the deferential standard of review, the district court granted summary judgment for MetLife on Holmstrom’s claim for benefits and MetLife’s counterclaim. Holmstrom v. Metropolitan Life Ins. Co., 615 F. Supp. 2d 722 (N.D. Ill. 2009). Holmstrom appealed. No. 09-2173 3 We respectfully disagree with the district court. We believe that MetLife and in turn the district court gave undue weight to the absence of objective measurements for Holmstrom’s impairment. There is ample corrobora- tion that her pain has been genuinely disabling. We also find that MetLife’s selective use of evidence and its repeated moving of the targets for the evidence of disability show that MetLife’s decision to terminate bene- fits was arbitrary and capricious. We reverse and order retroactive reinstatement of benefits for Holmstrom, sub- ject to the set-off for Social Security disability insurance benefits she has received. We leave the issues of attorney fees and prejudgment interest to the district court in the first instance. I. The Facts We take the facts from the administrative record com- piled by MetLife in considering Holmstrom’s claim. Holmstrom was employed as a senior training specialist at Experian Information Solutions, Inc. Through this employment, she participated in a group disability in- surance plan underwritten and administered by MetLife. In late 1999, Holmstrom sought the care of Dr. Eric Lomax to treat pain, numbness, and tingling she experi- enced in her right upper arm. In January 2000, Holmstrom had surgery to remedy a right ulnar nerve compression and neuropathy. The surgery provided little relief, and her symptoms soon worsened. In June 2000, she had another surgery to relieve what was thought to be nerve compression. Her symptoms worsened further after 4 No. 09-2173 this second procedure, prompting her to visit a pain clinic. The clinic doctors diagnosed CRPS Type I, a chronic neurological syndrome characterized by severe pain.1 In March 2002, Holmstrom underwent a third surgery, which also failed to relieve her symptoms. She saw another pain specialist, Dr. Weber. According to MetLife’s records, Dr. Weber “made a definitive diagnosis of . . . complex regional pain syndrome.” It was clear to Holmstrom and her doctors that surgery could do nothing to help her, leaving medication as her only re- course.2 Holmstrom’s pain medication regimen has in- 1 Holmstrom was initially diagnosed with reflex sympathetic dystrophy syndrome (“RSDS”), also known as complex regional pain syndrome (“CRPS”) Type I. The medical literature in the record seems to use CRPS Type I as the more popular name for the condition. (CRPS Type II, also known as causalgia, is a nearly identical condition that is usually caused by an iden- tifiable traumatic nerve injury, while the cause of a Type I condition is less easy to discern. In terms of diagnostic criteria, the medical literature before us does not differentiate be- tween the two.) For more information about CRPS Type I, see Social Security Ruling 03-2p, printed in 68 Fed. Reg. 59,971 (Oct. 20, 2003). 2 When a chronic pain condition cannot be remedied by surgery, medical professionals often recommend physical therapy. The record in this case, however, suggests that physical therapy may have done more harm than good. MetLife’s records refer to Holmstrom’s doctor’s opinion that she has “a permanent condition and rehab[ilitation] is not possible.” MetLife does not contend that Holmstrom should (continued...) No. 09-2173 5 cluded a variety of powerful drugs, including Amitriptyline, Bextra, Clonidine, methadone, MS Contin, MSIR, Neurontin, Oxycontin, Oxycodone, Oxyfast, Percocet, Topamax, and (prior to its recall) Vioxx.3 Holmstrom’s symptoms persisted without improve- ment for the next three years. MetLife’s records from 2003 describe a “high pain med[ication] regimen” causing side effects such as confusion and memory loss, and pain of such intensity that Holmstrom was “considering having nerve severed since all other kinds of pain man- agement techniques have failed.” The record reveals no improvement through 2004 and 2005. Dr. Ted Vant, who has been Holmstrom’s treating physician from 2004 to the time of this lawsuit, prescribed significant doses of strong medications in an attempt to manage her symptoms. In early 2000, MetLife approved short-term disability benefits under the plan. After the plan’s short-term benefits expired, MetLife found that Holmstrom was still unable to perform her previous job duties, and it approved long-term disability benefits under the plan’s 2 (...continued) have pursued physical therapy as a treatment option or that it would help her condition in the future. 3 MetLife suggests in its brief that Holmstrom’s medications reflect “drug-seeking behavior” spurred by addiction. How- ever, to look into this issue, MetLife enlisted Dr. Mark Carlson, who concluded that Holmstrom “has chronic malignant pain . . . w/ narcotic tolerance but no addiction.” 6 No. 09-2173 “own-occupation” definition. Under the terms of the plan, those benefits expired after 24 months, at which point Holmstrom was required to establish that she was unable to perform the duties of any occupation. Upon the expiration of her “own-occupation” benefits, Holmstrom submitted an “any-occupation” disability claim. MetLife initially denied her claim, but after Holmstrom sub- mitted additional medical evidence, MetLife reversed its decision and approved benefits in July 2002. At some point during the claim administration, MetLife referred Holmstrom to a service that it retained to help beneficiaries apply for Social Security disability insurance benefits. Holmstrom applied for and qualified to receive these benefits under the Social Security Act. In August 2005, MetLife performed a periodic review of Holmstrom’s benefits and decided to terminate pay- ments to her. The letter announcing the termination indicated that Dr. Thomas, retained by MetLife to perform the review, “determined that there is no medical information to support the restrictions provided by Dr. Vant,” Holmstrom’s treating physician. The letter added that “medical information no longer supports a severity in your impairment preventing you from doing your [previous] job.” The letter told Holmstrom that she could appeal the decision by providing “office notes, physical exam findings, EMG results, MRI results, pain management notes, neurology notes, and/or physical therapy notes.” Holmstrom appealed and provided to MetLife: (1) an August 2005 Functional Capacity Evaluation (“FCE”) No. 09-2173 7 performed by a physical therapist; (2) a June 2005 At- tending Physician Statement completed by Dr. Vant; (3) Dr. Vant’s examination notes from July 6, 2004 to July 20, 2005; (4) the results of an August 2004 nuclear bone scan (x-ray images and the analysis of Dr. Kenneth Sato); (5) the results of a May 2005 EMG nerve conduc- tion test (numerical data, graphs, and the analysis of Dr. Gary Klein); (6) the Social Security Administration notice of award; and (7) letters from three family members. The August 2005 FCE was only one page long and included very little testing of Holmstrom’s arm, but it noted that she was “unable to [support any body] weight on hands due to pain.” Dr. Vant’s detailed, four- page statement included the CRPS diagnosis and con- cluded that Holmstrom suffered from a “permanent disability” and could perform essentially no hand func- tion. Dr. Vant’s examination notes included prescriptions for pain medications such as Amitriptyline, Clonidine, Elavil, Hydrocodone, and “large dose[s]” of methadone. He noted “no real changes,” “continue[d] sweating,” and “spasm,” and said that Holmstrom was “still feeling numb.” Dr. Sato’s bone scan analysis stated that there was normal blood flow to the arm and that “no abnormal- ities of either arm, hand or wrist are seen.” Dr. Klein’s EMG was “negative” in that it “rule[d] out nerve entrap- ment syndrome”, the malady that Holmstrom’s doctors had thought she suffered from before her surgeries. Dr. Klein found a “minor” irregularity in the nerves of her forearm and dysesthesia around the right elbow. He otherwise found her nerves to be “absolutely normal” and her “sensory responses [to be] within normal limits,” and 8 No. 09-2173 he noted that her “power is normal [and] [r]eflexes are well preserved.” However, regarding her general condi- tion, Dr. Klein stated that “her pain and dysesthesia continue and if anything, are slowly getting worse.” The family testimonials described in detail how Holmstrom’s condition affected her everyday life and the lives of those around her, including descriptions of her sig- nificant pain, physical deficiencies, and compromised mental function from pain medication. The Social Security notice of award stated only that the government had concluded that Holmstrom was totally disabled under its stringent standards.4 MetLife denied the appeal, notifying Holmstrom in a February 2006 letter that summarized the opinion of Dr. Janet Collins, the physician retained by MetLife to perform the medical review. Dr. Collins did not examine Holmstrom. MetLife’s letter addressed Holmstrom’s claims of intractable pain, significant physical limita- tions, and cognitive deficiency as identified by Holmstrom and Dr. Vant. MetLife found, however, that the lack 4 The Social Security Act defines disability as the “inability to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment which . . . can be expected to last for a continuous period of not less than 12 months[.]” 42 U.S.C. § 423(d)(1)(A). This defini- tion is more stringent than the plan’s “any occupation” disability definition, which describes disability as being “unable to earn more than 60% of your Indexed Predisability Earnings from any employer in your Local Economy at any gainful occupation for which you are reasonably qualified . . . .” No. 09-2173 9 of “objective findings to support ongoing total disability” prevented MetLife from determining whether her dis- ability was genuine. MetLife stated that the August 2005 FCE was unreliable based on the “emotional component displayed by Ms. Holmstrom during the exam” and her “inability or unwillingness” to move parts of her body “known not to be affected by her pain complaints,” specifically her lumbar spine. MetLife concluded by stating that it could have reached a different decision on disability if Holmstrom had provided another FCE “in order to more precisely quantify appropriate restric- tions and limitations” and a “battery” to “assess her neurocognitive status.” Holmstrom then filed suit in federal court. The parties voluntarily remanded the case for a third administrative review. In an effort to comply with the requests of the February 2006 letter, Holmstrom submitted another FCE and the results of a battery of cognitive analysis tests called a Schubert General Ability Battery. This four-page report on the two-day FCE was much more detailed than the FCE report from 2005. It con- tained more tests germane to Holmstrom’s specific pain complaints. Each test result included detailed num- erical data and was specifically interpreted to demon- strate the result’s significance with regard to Holmstrom’s ability to perform a job in the “sedentary” exertion cate- gory (the least demanding, as defined by the Depart- ment of Labor). The new FCE concluded that her function was below the sedentary level for most tests involving her hands or arms. To determine endurance 10 No. 09-2173 and consistency, each test was performed on both days. On the second day, “all measured parameters recorded a reduction of about 20%, with increased pain and discom- fort, placing Mrs. Holmstrom in the low endurance cate- gory.” The FCE concluded that it could not recommend any path for returning to work given Holmstrom’s sig- nificantly low function and endurance. The FCE stated that it was conceivable that Holmstrom might “do a minimum amount of work while using vocation-specific voice recognition software . . . relieving her of the chal- lenges of sitting and typing.” If she did so, she could work a maximum of only three hours per day, divided into six sessions of 30 minutes each, and only two to three days per week. The FCE tempered this observa- tion even further, stating that this possible form of part- time work would “depend, however, to a large extent, on her mental and cognitive status, reflecting her mental ability . . . .” The Schubert General Ability Battery, performed by Dr. Kent Noel, Ph.D., revealed significant cognitive impairment. It found that Holmstrom’s intelligence quotient had diminished from 123 in 1991 (as established by the same type of testing) to 104 in 2007. The 2007 results put her in just the fifth percentile among manage- ment candidates. Dr. Noel concluded that these results “strongly suggest that Ms. Holmstrom would experience difficulty focusing, retaining, processing, and making decisions. If considered for a return to the workforce, it would be at the most menial level using her physical skills, if these were suitable.” No. 09-2173 11 MetLife enlisted Dr. Robert Manolakas and Dr. Carol Walker to consider the new FCE and the Schubert General Ability Battery, respectively. Dr. Manolakas stated that Holmstrom had physical limitations “but not severe limitations.” Rejecting the March 2007 FCE, he stated: the report did not include the raw data or validity observations . . . so it is unclear on what basis precisely the lack of performance is due to: ie, physical inca- pacity or poor effort for whatever reason. The lan- guage suggests poor effort or endurance, but without the entire report or a repeat study this is not able to be determined for sure. Dr. Manolakas challenged the diagnosis of CRPS alto- gether, stating that it had “not been established by the available medical data in [the] file” and “the physical exam findings to support [it] are currently few . . . .” Addressing Holmstrom’s cognitive impairments, Dr. Walker rejected the Schubert General Ability Battery results and Dr. Noel’s conclusions: [H]e is a not a neuropsychologist, but a clinical psy- chologist, and does not perform neuropsychological evaluations. Dr. Noel has apparently based his opin- ions on a test that is developed to be used for an estimate of intellectual capacity and one that does not have appropriate measures of symptom validity. Such a measure will not allow an indi- vidual to make inferences regarding the person’s overall cognitive ability . . . . [C]hanges in individ- ual performances cannot be determined to be re- 12 No. 09-2173 liable or valid without specific measures of symptom validity. Dr. Walker found that the medical documentation did not support cognitive impairment. Before MetLife’s final decision, Holmstrom responded to the quoted reports with letters from Dr. Vant and Dr. Noel. Dr. Vant asserted that no objective tests existed for CRPS, and he said that he had observed obvious physical deficits upon examination that were fully corroborated in the detailed FCE. Dr. Vant also added more detail about Holmstrom’s right arm range of motion—specifically, how far in each direction she was able move her arm. Dr. Manolakas responded with an addendum to his report in which he acknowledged that Dr. Vant’s latest letter offered data of greater, if not dispositive, significance. He stated that the letter, com- bined with the evidence already in the record, led him to conclude that “more likely than not, the right upper extremity would be limited currently to occasional handling and grasping and fingering, in an eight hour work day, at least. It is up to [MetLife] if they want to consider medical evidence in the letter sufficient to support this restriction of limitation, but I do.” Dr. Manolakas added that “an independent physical exam and file review is a higher level of medical evidence . . . especially in a case such as this,” and con- cluded that “an independent [medical examination] with file review would be in order” for the next step of claims administration. MetLife did not take this recommenda- tion, and no independent examination or review was ever conducted. No. 09-2173 13 Dr. Noel defended his expertise and credentials (specifi- cally, his extensive experience in evaluating cognition for the purposes of workplace function) and the ability of the Schubert General Ability Battery to assess ac- curately Holmstrom’s cognitive deficiencies and their impact on her ability to work. Dr. Walker responded with an addendum stating: “While intellectual assessment is often part of the battery of the neuropsychologist, it is not used alone to make a determination of an indi- vidual’s abilities.” On October 29, 2007, after receiving these additional reports, MetLife again upheld its deter- mination. Left with no further administrative recourse, Holmstrom filed this lawsuit. MetLife counterclaimed to recover payments it had made that should have been discounted based on the Social Security benefits that Holmstrom had received. In the district court, Judge Dow wrote a detailed opinion granting summary judgment for MetLife on both Holmstrom’s ERISA claim and the coun- terclaim, and Holmstrom appealed. II. Discussion A. Standard of Review The district court’s grant of summary judgment is reviewed de novo. Love v. National City Corp. Welfare Benefits Plan, 574 F.3d 392, 396 (7th Cir. 2009). Judicial review of an ERISA administrator’s benefits determina- tion is de novo unless the plan grants the administrator discretionary authority to determine eligibility for 14 No. 09-2173 benefits or to construe the terms of the plan. Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 115 (1989). When the administrator has such discretionary authority, as the vast majority now do, the court applies a more deferen- tial standard, seeking to determine only whether the administrator’s decision was “arbitrary and capricious.” Metropolitan Life Ins. Co. v. Glenn, 554 U.S. 105, ___, 128 S. Ct. 2343, 2348 (2008); Jenkins v. Price Waterhouse Long Term Disability Plan, 564 F.3d 856, 860-61 (7th Cir. 2009). The plan here provided such discretionary authority, so we review under the arbitrary-and-capricious standard. Review under this deferential standard is not a rubber stamp, however, and “we will not uphold a termination when there is an absence of reasoning in the record to support it.” Hackett v. Xerox Corp. Long-Term Disability Income Plan, 315 F.3d 771, 774-75 (7th Cir. 2003).5 ERISA also requires that “specific reasons for denial be 5 Beginning with Fuller v. CBT Corp., 905 F.2d 1055 (7th Cir. 1990), we have sometimes described the arbitrary-and-capri- cious test as whether the administrator’s decision was “down- right unreasonable.” Attorneys for ERISA plan administrators are fond of quoting this colloquial phrase in their briefs to this court and to district courts within the circuit. The phrase should not be understood as requiring a plaintiff to show that only a person who had lost complete touch with reality would have denied benefits. Rather, the phrase is merely a shorthand expression for a vast body of law applying the arbitrary-and-capricious standard in ways that include focus on procedural regularity, substantive merit, and faithful execution of fiduciary duties. No. 09-2173 15 communicated to the claimant and that the claimant be afforded an opportunity for full and fair review by the administrator.” Tate v. Long Term Disability Plan for Salaried Employees of Champion International Corp. No. 506, 545 F.3d 555, 559 (7th Cir. 2008) (internal quotations omitted).6 An administrator’s conflict of interest is a key consider- ation under this deferential standard. “In conducting this review, we remain cognizant of the conflict of interest that exists when the administrator has both the discretionary authority to determine eligibility for benefits and the obligation to pay benefits when due.” Jenkins, 564 F.3d at 861, citing Glenn, 128 S. Ct. at 2346. In such cases, like the one before us, the conflict of interest is “weighed as a factor in determining whether there is an abuse of discretion.” See Glenn, 128 S. Ct. at 2350 (internal quotations omitted).7 6 In its recent decision in Hardt v. Reliance Standard Life Ins. Co., 130 S. Ct. 2149 (2010), the Supreme Court specifically abrogated one of Tate’s holdings regarding when a claimant is entitled to attorney fees. All other holdings in Tate remain good law. 7 For ERISA purposes, “the arbitrary-and-capricious stand- ard . . . is synonymous with abuse of discretion . . . .” Raybourne v. Cigna Life Ins. Co. of New York, 576 F.3d 444, 449 (7th Cir. 2009). “Nit-pickers might argue that there is a distinction” between the two standards, but they are simply “different ways of saying the same thing.” Jenkins, 564 F.3d at 861 n.8 (internal quotations omitted). 16 No. 09-2173 B. Disability Determination Holmstrom has shown that MetLife’s termination of benefits was arbitrary and capricious for several reasons. She has also raised several arguments that are not per- suasive, which we address first. First, Holmstrom places great emphasis on MetLife’s “abrupt termination of benefits” despite its prior deter- mination that her “functional deficits [would] likely be permanent.” Holmstrom Br. 16, 19. Holmstrom asserts that this note made in 2000 meant “the insurer was obli- gated to continue paying her for another 25 years,” the remainder of the plan. Holmstrom Br. 19. We reject this argument. ERISA does not prohibit a plan admin- istrator from performing a periodic review of a benefi- ciary’s disability status. See Leger v. Tribune Co. Long Term Disability Benefit Plan, 557 F.3d 823, 832 (7th Cir. 2009) (“We are not suggesting that paying benefits operates forever as an estoppel so that an insurer can never change its mind . . . .”), quoting McOsker v. Paul Revere Life Ins. Co., 279 F.3d 586, 589 (8th Cir. 2002) (quotation marks omit- ted). The plan administrator is entitled to seek and con- sider new information and, in appropriate cases, to change its mind. Second, Holmstrom argues that MetLife could not properly terminate her benefits without proving that her condition had actually improved. We have rejected this argument before in Leger, 557 F.3d at 831-32, though we ruled that the administrator’s denial decision in that case was arbitrary for other reasons. One case examiner’s five- year-old opinion of permanence, given before the benefi- No. 09-2173 17 ciary underwent another surgery and received a new diagnosis, did not bind the administrator indefinitely. This circuit and the Eighth Circuit have noted that “the previous payment of benefits is just one ‘circum- stance,’ i.e., factor, to be considered in the court’s review process; it does not create a presumptive burden for the plan to overcome.” See Leger, 557 F.3d at 832 (con- sidering the prior determination among other factors to conclude that administrator’s decision was arbitrary and capricious), citing McOsker, 279 F.3d at 589 (same). The prior determination does not decide the case. It is merely part of the overall set of facts that we consider. Third, Holmstrom argues that Glenn gave the “unequiv- ocal directive” that district courts should consider an administrator’s adverse judgments in other federal cases as evidence supporting a conflict of interest. Holmstrom Br. 17. In doing so, Holmstrom relies on the portion of Glenn that states that a conflict “should prove more important (perhaps of great importance) where circum- stances suggest a higher likelihood that it affected the benefits decision, including, but not limited to, cases where an insurance company administrator has a history of biased claims administration.” 128 S. Ct. at 2351, citing John H. Langbein, Trust Law as Regulatory Law: The UNUM/Provident Scandal and Judicial Review of Bene- fit Denials Under ERISA, 101 Northwestern U. L. Rev. 1315, 1317-21 (2007). Holmstrom contends that this passage allows her to use a string of federal decisions reversing MetLife administrative denials as evidence that MetLife’s inherent conflict of interest dominates and corrupts its claims determinations. Holmstrom misreads 18 No. 09-2173 this portion of Glenn. Glenn did not invite a “batting average” approach, assessing conflict by comparing the number of benefits decisions affirmed and reversed in federal court. (The sampling problems with that ap- proach would be daunting.) Rather, to support the quoted assertion, the Glenn Court cited a law review article that detailed the long and unfortunate history of the inner workings of the insurance company Unum/ Provident. This history was not a list of unfavorable court decisions. It was a detailed, fact-intensive account of systemic flaws and misconduct in the company’s administrative review process, supported by discovery, investigative journalism, and the accounts of inside whistleblowers. See Langbein, supra, at 1317-21. MetLife is one of the country’s largest insurance com- panies. It makes many thousands of administrative bene- fits decisions every year. It is not surprising that some of its claim decisions have led to litigation and that it has lost some of those cases. Surely in others, perhaps many others, it has won and no abuse of discretion was found. Holmstrom does not provide that list for com- parison. But whether MetLife’s cases were won or lost, abuse of discretion is a fact-specific inquiry. This court is concerned only with Holmstrom’s claim and the context and circumstances of MetLife’s denial as demon- strated by the administrative record in this case. Without evidence of systematic bias like that in the Langbein article cited in Glenn, the evidence that MetLife has been found to have abused its discretion toward a few other plaintiffs bringing other claims in other courts has little value. See Gessling v. Group Long Term Disability Plan for No. 09-2173 19 Employees of Sprint/United Management Company, 693 F. Supp. 2d 856, 872 (S.D. Ind. 2010). In addition to these weak arguments, however, Holmstrom has shown several other reasons for finding that MetLife acted arbitrarily and capriciously in terminating her benefits and then sticking with that decision through the administrative reviews for which Holmstrom provided exactly the sort of detailed infor- mation that MetLife had demanded. 1. “Normal” Test Results To support its determination, MetLife has relied on the results of the nuclear bone scan and EMG. MetLife sug- gests that the “normal” results of these tests undermined the diagnosis of CRPS and provided “further evidence that the FCEs [functional capacity evaluations] were not a reliable objective measure of Holmstrom’s functional capacities.” MetLife Br. 28-29. The record shows beyond reasonable dispute, however, that these tests only some- times reveal indicia of CRPS and that severe CRPS is not inconsistent with normal bone scan or EMG findings. MetLife cites the CRPS “Concise Review for Clinicians” in the peer-reviewed medical journal Mayo Clinic Pro- ceedings to support its reliance on the test results. The cited article rejects MetLife’s position: Although no specific diagnostic test is available for CRPS, several tests can be supportive in making the diagnosis, but the most important role of testing is to help rule out other conditions . . . . These tests attempt to 20 No. 09-2173 identify abnormal sympathetic activity or abnormal limb blood flow, but as mentioned previously, these phenomena are not always present. Richard H. Rho, et al., Complex Regional Pain Syndrome, 77 Mayo Clinic Proceedings 174, 175 (2002), http://www. mayoclinicproceedings.com/content/77/2/174.full.pdf (emphasis added) (last visited July 30, 2010).8 In their analyses of the test results, neither Dr. Sato nor Dr. Klein questioned the CRPS diagnosis or the severity of Holmstrom’s symptoms. Dr. Sato declined to conclude anything about Holmstrom’s arm. Dr. Klein’s only conclusions from the negative result were that it “rule[d] out nerve entrapment syndrome” and that “her pain and dysesthesia continue and if anything, are slowly getting worse.” Dr. Vant’s letters to MetLife re- peatedly explained why these test results were not sig- nificant. Yet there is no acknowledgment of his asser- tion (or of the specialists’ actual conclusions) in any of MetLife’s physician reports or disability determina- tions. Nor has MetLife ever acknowledged the clinically observable indicia of CRPS that Dr. Vant included in his examination notes and letters—hyperhidrosis, spasm, sweating, and temperature differences—all of which support a diagnosis of CRPS according to MetLife’s own 8 MetLife also cites a Mayo Clinic website with a much less comprehensive overview of CRPS. The site states that a bone scan or nervous system test might provide clues, but it cautions: “There is no single test that can definitively diagnose complex regional pain syndrome.” No. 09-2173 21 sources. Contrary to MetLife’s assertion, the bone scan and EMG results do not contradict the diagnosis of CRPS or undermine the validity of the FCEs.9 2. Functional Capacity Evaluations (FCEs) Subjectively painful conditions like CRPS and fibromyalgia pose difficult problems for private disability insurance plan administrators and the Social Security Administration, who understandably seek to make deci- sions based on the most objective evidence available. But we have rejected as arbitrary an administrator’s requirement that a claimant prove her condition with objective data where no definitive objective test exists for the condition or its severity. See Hawkins v. First Union Corporation Long-Term Disability Plan, 326 F.3d 914, 918-19 (7th Cir. 2003) (reversing denial of benefits where administrator determined that there were “no objective 9 The Social Security Administration has compiled diagnostic criteria for CRPS, which involves “persistent, intense pain” associated with five potential observable criteria. None of these criteria involve laboratory testing such as a bone scan or EMG (with the one exception that CRPS sometimes ac- companies osteoporosis, which is observable in a bone scan). One group of criteria, autonomic instability (hyperhidrosis, spasm, sweating, temperature differences, etc.) has been observed and documented by Dr. Vant in his examination notes and letters. See SSR 03-2p, printed in 68 Fed. Reg. 59,971 (Oct. 20, 2003). Of these, MetLife acknowledged only the temperature differences but discounted them as “unspecified.” 22 No. 09-2173 findings to support restrictions,” and noting that pain often cannot be detected by laboratory tests and that the amount of pain and fatigue that a particular case produces cannot be tested objectively); Diaz v. Prudential Ins. Co. of America, 499 F.3d 640, 646 (7th Cir. 2007) (claimant’s pursuit of extensive treatment including heavy medica- tion and repeated surgical procedures “supports an inference that his pain, though hard to explain by refer- ence to physical symptoms, was disabling”). At the same time, even in these difficult cases involving conditions where subjective symptoms of pain are not manifest in objective clinical data, we have allowed a plan administrator to require a certain degree of “ob- jectivity” in terms of the measurement of physical limita- tions as observed in a functional capacity evaluation. “A distinction exists however, between the amount of fatigue or pain an individual experiences, which as Hawkins notes is entirely subjective, and how much an individual’s degree of pain or fatigue limits his functional capabilities, which can be objectively measured.” Williams v. Aetna Life Ins. Co., 509 F.3d 317, 322 (7th Cir. 2007). The district court correctly identified this distinction and focused on it. However, the quantity and quality of the functional capacity and other data that Holmstrom provided to MetLife readily distinguish this case from Williams.10 10 Williams observed the same distinction in the First and Eighth Circuits. See 509 F.3d at 322-23, citing Boardman v. Prudential Ins. Co. of America, 337 F.3d 9, 16 n.5 (1st Cir. 2003) (continued...) No. 09-2173 23 Like Holmstrom, the claimant Lee Williams suffered from a condition (chronic fatigue syndrome) that is di- agnosed by subjective patient complaints. And like MetLife, the administrator in Williams (Aetna) rejected the claimant’s functional capacity data as insufficient. Unlike the present case, however, the Williams record “lacked any specific data reflecting Williams’s func- tional impairment.” Williams, 509 F.3d at 323. Williams never presented an actual FCE or any measurement of specific limitations. He offered only his treating physi- cian’s unexplained conclusions that he could perform only low-stress jobs and could not lift anything over ten pounds. Aetna gave this physician a functional capacity questionnaire asking for the results of very specific func- tional tests (e.g., how long Williams was able to stand before needing to sit down), which were answered “un- known” or “untested.” No specific tests of physical ability or endurance were ever performed. In this case, Holmstrom provided a physician opinion similar to the one in Williams, but she also presented two 10 (...continued) (“While the diagnoses of chronic fatigue syndrome and fibromyalgia may not lend themselves to objective clinical findings, the physical limitations imposed by the symptoms of such illnesses do lend themselves to objective analysis.”), and Pralutsky v. Metropolitan Life Ins. Co., 435 F.3d 833, 838-40 (8th Cir. 2006) (finding that it was not unreasonable for an administrator to request objective functional capacity evidence beyond doctor statements that simply repeated the claimant’s subjective complaints of pain and fatigue). 24 No. 09-2173 FCEs, at least one of which (from 2007) provided exactly the kind of detailed and specific information that the Williams court found lacking. At oral argument, MetLife urged us to take a critical look at the FCEs under the Williams standard. We have examined them, and we find that the 2007 FCE provides objective sup- port showing functional limitations amounting to total disability.11 The 2007 FCE report included 20 different detailed tests. Six examined arm function, and seven examined hand function. Each result included specific weight and time data, and applied that data to the lowest possible occupa- tional exertion category as determined by the Depart- ment of Labor. Holmstrom fell short of the requirements 11 The 2007 FCE was an attempt to remedy the flaws that MetLife perceived with one from 2005, which Holmstrom had submitted for her initial prior appeal. That FCE contained very little data, commentary, or analysis, and contained only one test of Holmstrom’s upper extremities—measuring her ability to perform repeated push-ups (at zero). If we were to look in isolation only at MetLife’s conclusion that the 2005 FCE was insufficient to show how Holmstrom’s pain or weakness might limit her functional abilities, we would not find that conclusion to be arbitrary or capricious. However, the 2007 FCE is the pertinent evaluation for this appeal and is fully sufficient in both its degree of analysis and its content. It adequately demonstrates the extent of Holmstrom’s functional limitations. The broader record from 2005 and later also contains a number of other indications that MetLife’s 2005 termination of benefits was arbitrary and capricious. No. 09-2173 25 of sedentary work in the majority of these tests. The tests were repeated one day later, with “all measured parameters recorded [at] a reduction of about 20%,” which suggested consistency of effort and “very poor endurance.” Those results indicated that it was unlikely that Holmstrom would be able to sustain even her severely compromised level of function over consecutive workdays, as needed for full-time employment. Despite the thoroughness of this 2007 FCE, MetLife rejected it. MetLife was obliged to explain why it found the FCE unreliable. See Leger, 557 F.3d at 834-35 (finding administrator’s decision arbitrary). MetLife offers several explanations, but they lack substance and reflect arbitrary action. In addressing the 2007 FCE, Dr. Manolakas opined (and MetLife adopted the conclusion) that “it is unclear on what basis precisely the lack of performance is due to: physical incapacity or poor effort.” He stated: “The lan- guage suggests poor effort and endurance, but without the entire report or a repeat study this is not able to be determined for sure.” In other words, according to Dr. Manolakas, who was only reviewing the report, there was no way to tell whether Holmstrom was faking her poor function. The professionals who conduct FCEs for the purposes of occupational assessment are aware of this common concern, and they look for disability exaggeration. The 2007 FCE report makes no observa- tion of any kind that might call Holmstrom’s effort into doubt. MetLife ignores the consistency of the FCE, with nearly identical reductions in measured performance 26 No. 09-2173 on the second day across “all measured parameters.” Nothing in this FCE or those from 2000 and 2005 calls Holmstrom’s effort into question.1 2 MetLife also challenges the validity of the 2007 FCE procedures, arguing that a “valid FCE” must include “raw data” and “algorithms for scoring functionality.” MetLife Br. 30. MetLife further cites sources that purport to explain how an FCE should be done. However, the cited sources in MetLife’s brief are consistent with the methodology used in the 2007 FCE, and at oral argu- ment MetLife counsel was unable to explain how a “valid” FCE would differ from this one. We look then to the 2000 FCE, which MetLife found satisfactory, and we see no material differences—certainly nothing in the way of “raw data” or “algorithms”—with the exception of range of motion data, which Dr. Vant had provided to Dr. Manolakas’ satisfaction in a sep- arate letter prior to MetLife’s 2007 decision confirming the termination of benefits. When questioned at oral ar- gument about the perceived differences in the 2000 and 12 MetLife points to the physical therapist’s note in the 2005 FCE that Holmstrom “displayed emotional behavior such as crying” when discussing pain and undergoing a test of her functional limitations. MetLife argues that her crying sug- gested poor effort and undermined the genuineness of her pain complaints. That is sheer speculation. The evidence could easily support the opposite conclusion. We, and those who pay for disability insurance, are entitled to rely on the report of the FCE and the professionalism of the examiner instead of such speculation. No. 09-2173 27 2007 FCEs, MetLife’s counsel answered that the 2000 FCE contained five things that the 2007 one did not: range of motion data, strength tests, reflex tests, sensory tests, and detailed pain descriptions. Again, Dr. Vant provided range of motion data to MetLife’s physician’s satisfac- tion. The 2007 FCE contained 10 strength tests (the 2000 FCE also contained 10) that show no discernable dif- ference in character or detail from the strength tests of 2000. While the 2007 FCE had no reflex tests, the 2000 FCE explicitly indicated that Holmstrom’s right elbow reflex was not tested. The 2007 FCE lacked sensory testing, but her sensory testing results in the 2000 FCE were normal. Finally, there is no appreciable difference in the level of detail between the 2000 and 2007 pain descriptions. There is no reason to think that an FCE performed in 2007 under the same standards as the 2000 FCE would have produced a conclusion any different from the one that Holmstrom submitted for her final appeal. Furthermore, MetLife never communicated to Holmstrom that it would require an FCE of the same format and level of detail as the one from 2000. At oral argument, MetLife’s counsel conceded that MetLife never communicated to Holmstrom these specific crite- ria for an FCE that it later demanded. MetLife has therefore failed to explain its rejection of the conclusions of the 2007 FCE, and “there is an ‘absence of reasoning in the record’ to support [MetLife’s] con- clusion” that the 2007 FCE does not establish disability. See Leger, 557 F.3d at 835, quoting Tate, 545 F.3d at 559. 28 No. 09-2173 3. Social Security Determination The Social Security Administration determined that Holmstrom was completely disabled and awarded disabil- ity benefits. As mentioned above, the Social Security standard for total disability is more stringent than the plan’s standard for any-occupation disability at issue here. Moreover, it was MetLife that insisted that Holmstrom apply for Social Security benefits. As a result, MetLife received a benefit from the Social Security determination that she was disabled, but then failed to consider that determination when it terminated benefits.1 3 This issue was an important factor in the Supreme Court’s analysis in Glenn. Approving the Sixth Circuit’s analysis, the Glenn Court stated: In particular, the [circuit] court found questionable the fact that MetLife had encouraged Glenn to argue to the Social Security Administration that she could do no work, received the bulk of the benefits of her success in doing so . . . and then ignored the agency’s finding in concluding that Glenn could in fact do 13 Under the plan, Holmstrom was required to apply for Social Security benefits. If they were granted, MetLife’s pay- ment liability under the plan would be reduced by the amount of those benefits. Holmstrom collected Social Security benefits, but MetLife’s payment reduction was never realized. As explained below, the parties have stipulated that MetLife has a right to reimbursement of these funds, so for purposes of this inquiry, Holmstrom’s Social Security deter- mination benefitted MetLife. No. 09-2173 29 sedentary work. This course of events was not only an important factor in its own right . . . but also would have justified the court in giving more weight to the conflict (because MetLife’s seemingly incon- sistent positions were both financially advantageous). Glenn, 128 S. Ct. at 2352 (citations omitted); see also Raybourne, 576 F.3d at 450 (“after Glenn, [the administra- tor]’s advocacy of a disability finding before the SSA should have been treated as a serious concern for the court to consider”) (internal quotations omitted); Ladd v. ITT Corp., 148 F.3d 753, 756 (7th Cir. 1998) (reversing denial of benefits in part because administrator sup- ported claimant’s efforts to demonstrate total disability to the Social Security Administration, then denied claimant was totally disabled even though her condi- tion had not improved). An administrator is not forever bound by a Social Security determination of disability, but an admin- istrator’s failure to consider the determination in making its own benefit decisions suggests arbitrary decision- making. Glenn, 128 S. Ct. at 2352. This is especially so when the Social Security determination was made under a similar or more stringent disability definition, as it was here. In its denial letters, MetLife never stated why it disagreed with the Social Security determination; rather, it stated only that Black & Decker Disability Plan v. Nord, 538 U.S. 822 (2003), essentially dissolved any relevance of Social Security determinations in ERISA cases. The discussion of Social Security benefits in Glenn directly rejected this flawed interpretation of Nord. 30 No. 09-2173 4. Medical History Holmstrom’s overall objective medical history is also highly relevant. See Diaz v. Prudential Ins. Co. of America, 499 F.3d 640, 646 (7th Cir. 2007) (finding medical history that included heavy medication and repeated surgical procedures to be relevant in determining that claimant was disabled). While the significance of a procedure or a prescription can be disputed, the existence of such things when established in the record cannot be. Holmstrom has undergone three surgeries and continues to endure what is, even by MetLife’s doctors’ accounts, a heavy regimen of pain medication. MetLife claims that the surgeries have resolved her condition, despite the utter lack of support for this conclusion and the wealth of medical opinion (including from MetLife consultants) that surgical options were abandoned because more operations would be futile. MetLife speculates that the medication regimen does not support the existence of genuine pain but instead exists only to feed drug-seeking behavior. MetLife attorneys and consultants support this conclusion with no evidence and ignore evidence of their own doctor’s conclusion that her pain is genuine and that she does not suffer from addiction, a conclusion reached by MetLife’s doctor after an in-person examina- tion of Holmstrom. We do not suggest that a Social Security disability finding, multiple and unsuccessful surgeries for pain relief, and a heavy pain medication regimen will together always compel an award of benefits. But with this evidence in the record, a plan administrator must No. 09-2173 31 address it and provide a reasonable explanation for discounting it. See Leger, 557 F.3d at 835. In this case, the Social Security award, the surgeries, and the medica- tion provide strong evidence in support of a finding of continuing disability. MetLife’s explanations for dis- counting them are not supported by the record. 5. Cognitive Impairments MetLife also acted arbitrarily and capriciously in dis- counting evidence of Holmstrom’s cognitive impair- ments resulting from her heavy pain medication. In its 2006 denial, MetLife stated that Holmstrom could sub- stantiate her claim of cognitive impairments with a “battery” to “assess her neurocognitive status.” Without more specific direction from MetLife, Holmstrom sub- mitted a Schubert General Ability Battery performed by Dr. Noel, a Ph.D. with significant experience in evaluating cognition for the purposes of workplace function. The battery of tests replicates tests that Holmstrom had done in 1991 under controlled conditions, allowing com- parisons of her current abilities. Dr. Noel shared the numerical results of the battery and expressed extreme doubt that Holmstrom could resume employment: “If considered for a return to the workforce, it would be at the most menial level using her physical skills, if these were suitable.” MetLife’s reviewing doctor, Dr. Walker, opined that Dr. Noel was not the proper professional to conduct cognitive testing because he was “not a neuro- psychologist, but a clinical psychologist.” Dr. Walker rejected the battery itself because it did “not have appro- 32 No. 09-2173 priate measures of symptom validity.” Although Dr. Noel responded with a detailed defense of his credentials and the battery, MetLife still rejected Holmstrom’s cogni- tive impairments as insufficiently substantiated. MetLife never explained specifically what it meant by “battery” of “neurocognitive testing” or “symptom validity,” or why a neuropsychologist was necessary while it deemed Dr. Noel’s qualifications and training insufficient, other than the need for the proper level of experience, which Dr. Noel attested (without refutation) that he pos- sessed. (Dr. Noel was obviously frustrated by what he saw as Dr. Walker’s wrong assumptions about his qualifi- cations and training.) ERISA requires plan administrators to provide claimants a reasonable opportunity for “a full and fair review” of the denial decision. 29 U.S.C. § 1133(2). Given the nature of the exchange and the data provided, we find MetLife’s rejection of the cognitive evidence to be arbitrary and capricious, failing to provide a full and fair review. MetLife acted within its rights by asking for tests showing Holmstrom’s cognition, but its request was general. MetLife provided no guidance as to what testing she should provide, much less how or by whom it should be done. The phrase “neurocognitive testing” did not give Holmstrom fair notice of the additional criteria that MetLife later insisted would need to be met before it would give weight to the results. When an admin- istrator asks for additional information in broad terms, it is too easy to find later a reason to deem what it was given to be insufficient. If the administrator believes that a procedure must have certain characteristics, or that it must be performed by a certain kind of professional, it No. 09-2173 33 must provide at least some level of guidance, unless the test sought is so well-known that a claimant or her attorney or other representative can reasonably be ex- pected to know what the administrator expects. MetLife provided no such guidance here. Also, of course, if a plan administrator requires a test and has detailed expectations for the way it is to be conducted, it may arrange for the testing itself. Holmstrom’s policy, like most such policies, requires her to appear for testing that the administrator arranges. MetLife was free to make such a demand, but it did not. Having passed on that opportunity and having pro- vided only a broad request for “neurocognitive testing,” MetLife’s after-the-fact reasons for rejecting Dr. Noel’s results reflect arbitrary and capricious decision-making that suffers from “an absence of reasoning in the record to support it.” See Hackett, 315 F.3d at 774-75. 6. Examining Physicians Holmstrom argues that MetLife improperly failed to consider the opinion of her treating physician and relied instead on the opinions of MetLife doctors who only reviewed records and never examined her. The Supreme Court has cautioned: “Nothing in [ERISA] suggests that plan administrators must accord special deference to the opinions of treating physicians. Nor does the Act impose a heightened burden of explanation on administrators when they reject a treating physician’s opinion.” Black & Decker Disability Plan v. Nord, 538 U.S. 822, 831 (2003). However, a plan’s determination must still have a 34 No. 09-2173 reasoned basis to survive judicial review, even under the deferential standard of review. Administrators may not arbitrarily refuse to credit a claimant’s reliable evidence, including opinions of a treating physician. Id. at 834. MetLife would be entitled to disagree with Dr. Vant’s opinion if there were evidence in the record providing a reasoned basis for doing so. No substantial evidence exists to that effect. None of the doctors who concluded that Holmstrom failed to establish disability ever examined her. Every doctor who has actually seen her in the pertinent time period has concluded that she is disabled. An adminis- trator may give weight to doctors who did only a rec- ords review, see Nord, 538 U.S. at 831, but in this case, the evidence provided by the doctors who examined her in person is so overwhelming that the reliance on record-review doctors who selectively criticized this evidence is part of a larger pattern of arbitrary and capri- cious decision-making. See Love v. National City Corp. Welfare Benefits Plan, 574 F.3d 392, 396-397 (7th Cir. 2009) (denial of benefits was arbitrary where “neither [denial] letter explained why the reviewer chose to discredit the evaluations and conclusions of Love’s treating physi- cians” and “every doctor that personally examined Love concluded that she was unable to work”); cf. Mote v. Aetna Life Ins. Co., 502 F.3d 601, 604-05, 609 (upholding denial decision where treating physicians uniformly concluded that claimant was disabled, but surveillance evidence contradicted those physicians’ conclusions). MetLife’s reliance on the opinions of its reviewing doctors here is all the more arbitrary in light of the fact No. 09-2173 35 that it ignored the key final recommendation of one of those doctors. After receiving Dr. Vant’s range of motion data, Dr. Manolakas retracted his prior conclusion that disability had not been established. Instead, he recom- mended an independent clinical examination for resolu- tion of the issue. Yet MetLife ignored this recommenda- tion and instead adopted Dr. Manolakas’ original con- clusion—the one he retracted after receiving additional information from Dr. Vant. MetLife’s decision not to order the examination and its failure to explain that decision are further evidence of an arbitrary and capricious decision. Cases involving claims of persistent and serious pain that is difficult to evaluate in objective terms pose great challenges to plan administrators and to courts, and of course to the affected patient. Plan administrators and courts are understandably concerned about the possi- bility of malingering and exaggeration. Accordingly, we must note the absence here of any evidence of malin- gering or drug-seeking behavior. The problems of malin- gering, drug addiction, and drug-seeking behavior are well-known to professionals who treat painful condi- tions, and they look for them. MetLife internally ex- pressed concern about the possibility of drug-seeking behavior, and it enlisted the aid of Dr. Mark Carlson to evaluate that concern. After examining Holmstrom back in 2002, Dr. Carlson concluded that her chronic pain was genuine and that there was no addiction. MetLife never revisited this issue, save for a record review that cited no evidence beyond Dr. Carlson’s prior conclu- sions and the evidence that Dr. Carlson had already evaluated. The problems of malingering and addiction 36 No. 09-2173 were not found here, and the subsequent reviewers’ speculation is not a substitute for evidence. 7. The Moving Target Another sign of MetLife’s arbitrary and capricious decision-making is that it repeatedly “moved the target.” Over the course of the administrative appeals, MetLife invited additional evidence to establish disability, but when Holmstrom provided it, MetLife repeatedly found that the new evidence was not sufficient under new standards or expectations that had not been communi- cated to Holmstrom. Such conduct frustrates fair claim resolution and is evidence of arbitrary and capricious behavior. See Dabertin v. HCR Manor Care, Inc., 373 F.3d 822, 831 (7th Cir. 2004) (administrator unfairly imposed new, undisclosed requirements on claimant for sever- ance benefits; an ERISA benefit “cannot be a moving target where the plan administrator continues to add conditions precedent to the award of benefits”); Bard v. Boston Shipping Ass’n, 471 F.3d 229, 237 (1st Cir. 2006) (awarding disability benefits where claimant “was faced with a constant shift in what he was required to show,” and thus administrator’s conduct was arbitrary and capricious in that it failed to consider the evidence he submitted “in an attempt to meet a moving target”).1 4 14 Courts have used several sports metaphors to capture this unfortunate phenomenon: moving the target, moving the goal posts, hiding the ball, raising the bar, etc. This circuit (continued...) No. 09-2173 37 As described above, MetLife moved the target re- garding both the cognitive testing and functional capacity evaluation. MetLife made general requests. Holmstrom complied with the requests as a reasonable person would understand them. MetLife then rejected the new infor- mation for failure to meet new requirements that had not been revealed beforehand. An even more troubling example of “moving the tar- get” was MetLife’s decision to discount all medical evi- dence obtained after the initial termination of benefits on August 5, 2005. Since that date, MetLife has asked for a significant amount of medical data, some of which could be provided only by conducting new tests. Yet in its final October 2007 denial, MetLife stated its general disregard for Holmstrom’s many 2007 submissions because it would instead need “additional medical infor- mation dating to the time the claim was terminated.” In the same denial letter, MetLife employed a similar strategy to get around its own consultant’s determina- tion that Holmstrom indeed had disabling physical deficits: “Although [Dr. Manolakas] noted that currently [Holmstrom] would be limited to occasional [function] in an eight hour work[day], the time period in review is effective August 6, 2005.” In its brief on appeal, MetLife emphasized Dr. Manolakas’ use of the word “currently” to express only Holmstrom’s “condition as of Septem- 14 (...continued) has used the “moving target” language before, and we’ll stick with it. 38 No. 09-2173 ber 2007”, claiming that she had failed to “establish that she was disabled as of August 2005”. MetLife Br. 30. As the district court properly pointed out, accepting this argument would mean that MetLife’s initial termination of benefits for lack of supporting evidence could never be successfully appealed if the claimant had not already undergone functional testing (that satisfied MetLife’s precise but not-yet-unarticulated specifications) before the August 2005 termination decision. Holmstrom v. Metro- politan Life Ins. Co., 615 F. Supp. 2d 722, 745 (N.D. Ill. 2009). MetLife asked Holmstrom to undergo more testing, and rejected the results at least in part because the testing was not done before it made the request. That behavior also reflects arbitrary and capricious decision-making. 8. Selective Consideration of Evidence Holmstrom has offered sufficient evidence to establish continuing disability under the plan, and MetLife has failed to support its contrary conclusions with sound “reasoning in the record.” See Leger, 557 F.3d at 835. Holmstrom’s key evidence—the FCEs, Dr. Vant’s opinion, the consistent CRPS diagnoses, the surgeries, the Social Security disability determination (under more stringent disability criteria), the strong pain medication regimen, and the results of the neurocognitive testing—is all com- petent evidence that supports a finding of total disability. MetLife’s rejection of that evidence has been based on selective readings that are not reasonably consistent with the entire picture. This approach is another hallmark of an arbitrary and capricious decision. See Majeski v. Metro- No. 09-2173 39 politan Life Ins. Co., 590 F.3d 478, 483-84 (7th Cir. 2009) (holding that denial decision was arbitrary where insurer selectively relied on pieces of evidence to sup- port denial of benefits, while that evidence in context demonstrated disability); Leger, 557 F.3d at 832-33 (denial decision was arbitrary where insurer “cherry-picked the statements from her medical history that supported the decision to terminate her benefits, while ignoring a wealth of evidence to support her claim that she was totally disabled”); see also Glenn v. Metropolitan Life Ins. Co., 461 F.3d 660, 672-74 & n.4 (6th Cir. 2006) (holding denial decision was arbitrary where plan selectively considered evidence to reach decision unsupported by the record as a whole), aff’d 128 S. Ct. 2343 (2008) (ap- proving Sixth Circuit’s reasoning). C. Conflict of Interest As discussed above, a structural conflict of interest is a relevant factor where the administrator has both the discretionary authority to determine eligibility for benefits and the obligation to pay those benefits. Glenn, 128 S. Ct. at 2346; Jenkins v. Price Waterhouse Long Term Disability Plan, 564 F.3d 856, 861 (7th Cir. 2009). “A struc- tural conflict is one factor among many that are relevant in the abuse-of-discretion analysis . . . and will ‘act as a tiebreaker when the other factors are closely balanced.’ ” Raybourne v. Cigna Life Ins. Co. of New York, 576 F.3d 444, 449 (7th Cir. 2009), quoting Glenn, 128 S. Ct. at 2351-52. However, lengthy analysis of any potential conflict of interest at work here is unnecessary, as we do not view 40 No. 09-2173 this as a close case for judicial review. Ample evidence in this record shows arbitrary and capricious decision- making. That being said, it is worth commenting on some of the factors present in this case that suggest that a conflict of interest was at work. Glenn and its progeny have identified several indicia that can signal the effects of a conflict of interest. Several are present here. First, MetLife’s selective consideration of the evidence not only indicates that its decision was arbitrary (as discussed above), but also demonstrates the effects of a conflict of interest. Selective considera- tion of evidence can be a factor suggesting arbitrary ad- ministration in its own right, as well as a reason to give more weight to the conflict factor. See Glenn, 128 S. Ct. at 2352. A claimant may demonstrate conflict of interest by showing that the administrator “emphasized a certain medical report that favored a denial of benefits [and] de- emphasized certain other reports that suggested a con- trary conclusion.” Id. The selective approach described above tends to indicate a conflict of interest at work. A second indication is MetLife’s conduct regarding the Social Security award. The Supreme Court has found this behavior to be a factor in its own right in the arbitrary-and-capricious balance, but it may also be a sign of a conflict of interest. Glenn, 128 S. Ct. at 2352. A third indication of the effect of MetLife’s conflict of interest is the repeated “moving of the target.” This con- duct is also an independent factor in the arbitrary-and- capricious inquiry, but an administrator’s constant changing of its demands to avoid awarding benefits can No. 09-2173 41 also be good evidence of a conflict of interest at work. See Dabertin, 373 F.3d at 832 (not addressing conflict per se, as this case preceded Glenn’s announcement of the conflict of interest standard, but awarding benefits and declining remand, which would simply permit the ad- ministrator to “dig up new evidence until it found just the right support for its decision to deny an employee her benefits”). D. Remedy MetLife’s termination of benefits was arbitrary and capricious and thus cannot stand. We turn now to the issue of the appropriate remedy. When an ERISA plan administrator’s benefits decision has been arbitrary, the most common remedy is a remand for a fresh ad- ministrative decision rather than an outright award of benefits: Generally, when a court or agency fails to make ade- quate findings or fails to provide an adequate rea- soning, the proper remedy in an ERISA case, as well as a conventional case, is to remand for further findings or explanations, unless it is so clear cut that it would be unreasonable for the plan administrator to deny the application for benefits on any ground. Tate v. Long Term Disability Plan for Salaried Employees of Champion International Corp. No. 506, 545 F.3d 555, 563 (7th Cir. 2008) (internal quotations omitted) (remanding to administrator). The claimant’s benefit status prior to the denial informs our determination: “In fashioning relief 42 No. 09-2173 for a plaintiff who has sued to enforce her rights under ERISA . . . we have focused ‘on what is required in each case to fully remedy the defective procedures given the status quo prior to the denial or termination’ of benefits.” Schneider v. Sentry Group Long Term Disability Plan, 422 F.3d 621, 629 (7th Cir. 2005), quoting Hackett v. Xerox Corp. Long-Term Disability Income Plan, 315 F.3d 771, 776 (7th Cir. 2003). The Schneider court identified a key distinction between an initial denial of benefits and a termination of benefits that were being received: Because of our emphasis on restoring the status quo prior to the defective procedures, we have distin- guished between “a case dealing with a plan adminis- trator’s initial denial of benefits and a case where the plan administrator terminated benefits to which the administrator had previously determined the claimant was entitled. Compare Wolfe v. J.C. Penney Co., Inc., 710 F.2d 388, 393-94 (7th Cir. 1983) (remanding to the administrator for new hearing where initial denial of benefits was not procedurally accurate) with Halpin [v. W.W. Grainger, Inc., 962 F.2d 685, 697 (7th Cir. 1992)] (affirming district court’s reinstatement of plan bene- fits where termination was not procedurally ade- quate).” Schneider, 422 F.3d at 629, quoting Hackett, 315 F.3d at 775-76. We thus have a clearer idea of a claimant’s disability—and are much more likely to award bene- fits—when the denial decision we are vacating succeeds a prior benefit award. We must take care, however, to No. 09-2173 43 rely on that prior award only if it was made under the same disability definition as the subsequent termination. See Tate, 545 F.3d at 563 (remanding to district court because prior grant of benefits was made under a dif- ferent disability definition).1 5 In the present case, MetLife determined in 2002 that Holmstrom was totally disabled according to the plan’s “any-occupation” definition. MetLife then reversed its position in applying the same standard. Its decision was arbitrary and capricious, and the record indicates that Holmstrom’s condition has either remained constant or worsened since that initial “any-occupation” determina- tion. Retroactive reinstatement of benefits is therefore the appropriate remedy. See Schneider, 422 F.3d at 629-30 (ordering retroactive reinstatement because claimant “ceased receiving benefits to which she had earlier been determined to be entitled”); Hackett, 315 F.3d at 775-76 (ordering retroactive reinstatement because “the status quo prior to the [termination under the] defective proce- dure was the continuation of benefits”). Further, we tend to award benefits when the record provides us with a firm grasp of the merits of the partici- 15 In reviewing the propriety of the denial decision itself, we consider reversal of a prior benefit award as “just one circum- stance” in the process. Leger, 557 F.3d at 832. Once we have decided that the administrator’s reversal of course was arbitrary and capricious, this prior benefit award may be determinative on the question of whether to remand or reinstate benefits. See Schneider, 422 F.3d at 629; Hackett, 315 F.3d at 775-76. 44 No. 09-2173 pant’s claim. Compare Halpin, 962 F.2d at 697-98 (affirming district court’s decision to reinstate benefits where the evidence supported the merits of the disability claim and administrator terminated benefits previously awarded), with Tate, 545 F.3d at 563 (reinstatement not appropriate where court could draw “no opinion regarding the merits of Tate’s claim as the record does not make clear either way whether Tate is ‘totally disabled’ ”). After examining the eleven years and nearly 700 pages of medical data before us, we are confident that Holmstrom has been totally disabled under the plan’s “any-occupation” definition. There is nothing more she can provide. A wealth of detailed medical data and consistent, objective functionality testing point only to a finding of total disability. Reinstatement of benefits is the remedy. We remand to the district court with instruc- tions to reinstate long-term benefits retroactively as of August 5, 2005. We leave it to the district court to deter- mine the exact amount owed since that date. E. Prejudgment Interest and Attorney Fees Holmstrom seeks costs, attorney fees, and prejudgment interest on benefits due since August 2005. In a benefi- ciary’s ERISA action, “the court in its discretion may allow a reasonable attorney’s fee and costs of action to either party.” 29 U.S.C. § 1132(g)(1). We review a district court’s decision to award or deny attorney fees for abuse of discretion, and will not disturb the district court’s finding “if it has a basis in reason.” Bowerman v. Wal-Mart Stores, Inc., 226 F.3d 574, 592 (7th Cir. 2000). Whether to No. 09-2173 45 award an ERISA claimant prejudgment interest is “a question of fairness, lying within the court’s sound discre- tion, to be answered by balancing the equities.” Fritcher v. Health Care Service Corp., 301 F.3d 811, 820 (7th Cir. 2002). In the present case, the district court had no reason to address the issues of attorney fees and prejudgment interest because it found in MetLife’s favor. The district court must have an opportunity to address these matters in light of our decision today. We remand for such con- sideration consistent with this opinion. We reverse the district court’s judgment and remand with instructions to order MetLife to reinstate bene- fits retroactive to August 5, 2005, and to consider Holmstrom’s request for attorney fees, costs, and prejudg- ment interest. The parties agree that the payment of past benefits should be offset by the amount of Holmstrom’s Social Security payments. The parties have stipulated that $70,107.76 is the amount of Social Security offset for benefits that Holmstrom received from 2000 to 2005. The record does not reveal the amount of Social Security benefits she has received since 2005. We leave it to the district court to determine both the amount that MetLife owes to Holmstrom in unrealized benefit pay- ments under the plan and the amount of the offset for Social Security payments. R EVERSED and R EMANDED. 8-4-10