Lennon v. Met Life

RECOMMENDED FOR FULL-TEXT PUBLICATION Pursuant to Sixth Circuit Rule 206 File Name: 07a0414p.06 UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT _________________ X Plaintiff-Appellee, - NANCY M. LENNON, - - - No. 06-2234 v. , > METROPOLITAN LIFE INSURANCE CO., - Defendant-Appellant. N Appeal from the United States District Court for the Eastern District of Michigan at Detroit. No. 05-73450—Patrick J. Duggan, District Judge. Argued: July 18, 2007 Decided and Filed: October 10, 2007 Before: BOGGS, Chief Judge; CLAY and ROGERS, Circuit Judges. _________________ COUNSEL ARGUED: Amy K. Posner, METLIFE GROUP, INC. LAW DEPARTMENT, Long Island City, New York, for Appellant. Edward G. Lennon, HYMAN & LIPPITT, Birmingham, Michigan, for Appellee. ON BRIEF: David M. Davis, HARDY, LEWIS & PAGE, Birmingham, Michigan, for Appellant. Edward G. Lennon, HYMAN & LIPPITT, Birmingham, Michigan, Brian D. Figot, WEISMAN, YOUNG, SCHLOSS & RUEMENAPP, Bingham Farms, Michigan, for Appellee. ROGERS, J., delivered the opinion of the court. BOGGS, C. J. (pp. 8-9), delivered a separate opinion concurring in the judgment. CLAY, J. (pp. 10-16 ), delivered a separate dissenting opinion. _________________ OPINION _________________ ROGERS, Circuit Judge. The question in this case is whether it is arbitrary and capricious for an ERISA plan administrator to deny Personal Accident Insurance benefits to the beneficiary of an insured who died as result of his own drunk driving. The insurance policy at issue covered “accidents” but did not specifically define the term to exclude deaths that resulted from an insured driver’s drunk driving. The district court held that, although the beneficiary, David Lennon, drove with a blood-alcohol level three times the legal limit, he did not reasonably expect to lose his life and that his death was thus accidental. The district court therefore ruled against MetLife and in favor of Lennon’s beneficiary, his mother Nancy, on her ERISA claim. Because MetLife could reasonably conclude that death caused by grossly negligent drunk driving is not accidental, it was not arbitrary and capricious for MetLife to do so. We therefore reverse. 1 No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 2 The result of drunk driving in this case was sad indeed. On June 2, 2003, Lennon, a young General Motors Acceptance Corporation accountant, drove his 2003 Chevrolet Trailblazer for the last time. At approximately 2:30 in the morning, Lennon’s car flew down a dry and well-lit divided boulevard in Pontiac, Michigan into a wall twenty feet away,1 and police found Lennon with no pulse. He died two days later. The evidence shows that Lennon was under the influence of alcohol at the time of the accident. According to a test that the hospital conducted, Lennon’s alcohol plasma level was 0.372 mg/dl, which is the equivalent of a blood-alcohol level of 0.321.2 The results show that Lennon’s blood-alcohol level was more than three times the legal blood-alcohol limit of 0.10 that Michigan had in effect at the time. See MCL § 257.625(1)(b).3 Within a month of Lennon’s death, Lennon’s mother, the plaintiff in this lawsuit, sought to recover Personal Accident Insurance benefits from MetLife. MetLife denied payments for Lennon’s Personal Accident Insurance, although it paid to Lennon’s beneficiary proceeds from Lennon’s basic and option life insurance. The relevant portion of MetLife’s Personal Accident Insurance policy provided: If, while insured for Personal Accident Insurance, an [insured] sustains accidental bodily injuries, and within one year thereafter shall have suffered loss of life . . . as a direct result of such bodily injuries independently of all other causes, [MetLife] shall pay the benefit specified for such Losses. (emphasis added). The policy also provided an exclusion: In no case shall payment be made for any loss which is contributed to or caused, wholly or partly, directly, or indirectly, by . . . suicide, attempted suicide or self- inflicted injury while sane or insane. The policy provided for other exclusions, not relevant here. MetLife’s letter denying benefits explained that Lennon’s drinking “impair[ed his] judgment and physical and mental reactions” and that Lennon’s blood-alcohol level was three times the legal 1 The parties both rely on a police diagram of the accident scene, but dispute its meaning. MetLife argues that the diagram shows that Lennon drove the wrong way down a one-way portion of Woodward Avenue, while the plaintiff argues that the diagram undermines MetLife’s interpretation. Because the diagram clearly shows an arrow indicating the direction of Lennon’s vehicle going down the street in the opposite direction of an arrow indicating “One Way,” MetLife’s interpretation is more consistent with the drawing. 2 The plaintiff argues that the administrative record did not establish a conversion factor between alcohol-plasma levels and blood-alcohol levels. The plaintiff, however, does not argue that the conversion is incorrect. Because she merely argues that the record lacks evidence that shows that the conversion is correct but does not dispute the correctness of the conversion, we will assume that MetLife’s calculations are correct. 3 The plaintiff points to three pieces of evidence to suggest that alcohol was not responsible for Lennon’s death. First, the police report from the accident did not indicate whether alcohol was responsible for the accident, and an officer wrote in the police report that he “did not smell an odor of intoxication on Lennon or in his vehicle at the scene.” Second, a June 3, 2003, report indicated that the tests did not detect ethyl alcohol or methyl alcohol. Third, the death certificate indicated that Lennon died as a result of an “accident” that caused blunt force head and neck trauma, and complications; the death certificate did not list the cause of death as drunk driving or alcohol. None of this evidence directly undermines MetLife’s conclusion that Lennon’s blood-alcohol level was three times the legal limit. The responding officer’s report, for example, does not directly contradict the laboratory results because an officer’s failure to detect or record the presence of alcohol does not mean that alcohol was not present. In addition, the plaintiff does not explain how the lack of ethyl alcohol or methyl alcohol contradicts the other blood-alcohol test. Finally, the medical examiner’s preliminary conclusions as to the cause of death do not preclude a subsequent finding that Lennon suffered the “blunt force” as a result of his drunk driving. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 3 limit. “The act of driving impaired,” the letter read, “rendered the infliction of serious injury or death reasonably foreseeable and, hence, not accidental as contemplated by the GM Plan.” MetLife therefore concluded that Lennon’s death was not “directly the result of accidental injuries, independent[] of all other causes.” MetLife also concluded that “the mental and physical impairments caused by the voluntary consumption of alcohol . . . constitute intentional self-inflicted injuries under the GM Plan [exclusion].”4 In December 2003, the plaintiff challenged MetLife’s denial of Personal Accident Insurance coverage, and on May 26, 2004, Met Life upheld its earlier decision. The record, however, contains only one document for the period from December 2003 to May 2004. That document noted, “New revisions to [Michigan’s blood-alcohol] law,” and it instructed a MetLife employee to “attach another copy of the [December] 30 . . . letter to the att[orne]y.” At the time that MetLife denied the plaintiff’s challenge to MetLife’s initial decision, it had the police report, medical examiner’s records, and the Alcohol Plasma results. In its May 26, 2004, letter (which MetLife did not release until June 8, 2004) MetLife noted: The Death Certificate lists the cause of death as “Blunt Force Head and Neck Trauma and Complications.” The State of Michigan Traffic Crash Report . . . states that [Lennon] lost control of his vehicle while crossing Woodward. [Lennon’s] vehicle was seen coming from the center median and then hitting the eastbound curb causing the vehicle to become airborne and strike a brick wall. . . . The Medical Examiner’s report states that [Lennon] was under the influence of alcohol and recorded a blood alcohol level of .37% [which was more than] the legal limit under Michigan law. The letter also reiterated the two grounds for denying Lennon coverage: (1) “the act of driving while so impaired rendered the infliction of serious injury or death reasonably foreseeable, and, hence, not accidental as contemplated by the plan,” and (2) “the mental and physical impairments caused by the voluntary consumption of excessive amounts of alcohol constitute intentional self-inflicted injuries under the plan.” Finally, the letter cited a series of cases in which courts found that drunk driving does not constitute an “accident” for ERISA purposes. The plaintiff challenged MetLife’s decision in federal district court, and the district court ruled against MetLife. The district court, quoting West v. Aetna Life Insurance Co., 171 F. Supp. 2d 856, 904 (N.D. Iowa 2001), noted that a person “is far more likely to be arrested for driving while intoxicated than to die or be injured in an alcohol related automobile crash, and far more likely to arrive home than to be either arrested, injured, or killed,” and concluded that MetLife acted in an arbitrary and capricious manner by relying solely on Lennon’s blood-alcohol level to determine that his death was not an accident. Finally, the district court rejected MetLife’s argument that Lennon’s death was a result of a self-inflicted injury. MetLife did not act arbitrarily and capriciously when it found that Lennon did not die as a result of an “accident” under the plan where Lennon’s death resulted from his driving with a blood- alcohol level three times the legal limit.5 Because General Motors’ policy with MetLife conferred 4 There is some record evidence that MetLife adopted a form letter when it initially denied benefits in this case. The record contains a hand-written note instructing a MetLife employee to “Pl[ea]s[e] see sample DWI denial letter. Use it for a guide.” 5 The plaintiff initially argues that the plain language of the plan requires MetLife to provide benefits in this case. MetLife’s plan provided that it would pay an insured who sustained “accidental bodily injuries and within one year . . . suffer[s] loss of life . . . as a direct result of such bodily injuries independently of all other causes.” (emphasis No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 4 discretionary authority to MetLife as an ERISA fiduciary to interpret terms of the plan, this court reviews MetLife’s decisions under an arbitrary-and-capricious standard. Whitaker v. Hartford Life & Accident Ins. Co., 404 F.3d 947, 949 (6th Cir. 2005). The record in this case establishes that Lennon’s behavior was, to borrow a term sometimes used in tort law, grossly negligent. Lennon broke the law by driving with a blood-alcohol level three times the legal limit, knowing that his drunk and severely impaired driving created a significant risk of bodily harm or death to others and to himself, and the precautions that would eliminate or reduce this risk (e.g., taking a taxi, or staying at a nearby hotel or with a friend) involved burdens that are so slight relative to the magnitude of the risk as to demonstrate Lennon’s indifference to the risk. See generally RESTATEMENT (THIRD) OF TORTS § 2 (Proposed Final Draft No. 1, 2005). As a prominent tort law treatise explains: Gross negligence can be used to mean what it says—a high, though unspecified degree of negligence. Presumably this means conduct that is appreciably more risky, or less beneficial, than conduct qualifying as ordinary negligence. . . . The idea of reckless, willful or wanton misconduct is similar in that the risk-utility balance strongly disfavors the defendant’s conduct—the risk was high, or very serious harm was threatened, or the cost of avoiding the danger was very low. . . . The defendant is guilty of reckless, willful or wanton misconduct only if he was conscious of the risk or had specific reason to know about it and proceeded without concern for the safety of others. . . . Although reckless, willful, or wanton misconduct is not the same as intentional harm, in extreme cases courts may treat wanton misconduct more like an intentional tort than like negligence. DAN R. DOBBS, THE LAW OF TORTS §147, at 350–51 (2000). If tort law can treat such conduct the same way it treats intentional conduct, it is not arbitrary and capricious for an ERISA plan administrator to treat such conduct as not accidental under a policy that only covers accidents. Driving while very drunk can certainly be placed in this category of activity. As the Seventh Circuit explained, albeit in a different context: Drunk driving is a reckless act, perhaps an act of gross negligence. Any drunk driver who takes to the road should know he runs a risk of injuring another person [or himself]. The extent of the risk will of course vary from case to case, depending on how intoxicated the driver is, how far he drives, how fast he drives, and how many other drivers and pedestrians are sharing the road with him. United States v. Rutherford, 54 F.3d 370, 376 (7th Cir. 1995); cf. United States v. Veach, 455 F.3d 628, 636–37 (6th Cir. 2006) (prior conviction for drunk driving is a “crime of violence” for federal sentencing purposes). This case involved facts—Lennon’s extremely high blood-alcohol content, the manner in which Lennon’s car flew off the road, the lack of an alternative explanation for the death, and Lennon’s driving the wrong way down the street—that rendered at least reasonable MetLife’s conclusions that Lennon did not die as a result of an “accident” under the Plan. Our conclusion that MetLife did not decide arbitrarily and capriciously is consistent with this court’s opinion in Cates v. Metropolitan Life Insurance Co., No. 96-6600, 1998 WL 385897 (6th Cir. June 30, 1998). In Cates, this court, albeit in an unpublished opinion, found that MetLife did not act arbitrarily and capriciously when MetLife denied benefits to an insured’s beneficiary on the grounds that the insured’s “act of driving while so impaired rendered the infliction of [serious] injury added). Plaintiff argues that the term “accidental” only modifies “bodily injuries” and that the term does not modify “loss of life.” To succeed with this argument, however, the plaintiff (as she implicitly recognizes) must prove that the bodily injuries in this case were not “accidental,” an issue resolved below, and the beneficiary’s first argument collapses into her main argument. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 5 or death reasonably foreseeable and, hence, not accidental as contemplated by the plan.” Id. at *3. Cates’ blood-alcohol level was 0.18%, almost twice the Tennessee limit of 0.10%, id. at *1, and the terms of Cates’ insurance plan were similar to those at issue in this case. Cates v. Metro. Life Ins. Co., 14 F. Supp. 2d 1024, 1025 (E.D. Tenn. 1996). The district court (and this court when it incorporated the district court opinion, Cates, 1998 WL 385897, at *3) held that MetLife’s decision was not arbitrary and capricious because several federal courts “reviewing ERISA cases have recognized that foreseeable harm resulting from an insured’s intentional actions is not accidental.” Cates, 14 F. Supp. 2d at 1027. Although this court in Cates did not provide an extensive analysis of its reasoning, other courts of appeals have reached the same result after more extensive discussions. In Eckelberry v. ReliaStar Life Insurance Co., 469 F.3d 340 (4th Cir. 2006), for example, the Fourth Circuit held that an ERISA fiduciary did not act unreasonably when it denied benefits to an insured beneficiary after the insured died from driving with a blood-alcohol level of 0.15, which was 50% higher than the legal limit. Id. at 342. The court reasoned that, regardless of whether the standard for non-accident was “high likelihood” of injury or “reasonable foreseeability” of injury, “federal courts have found with near universal accord that alcohol-related injuries and deaths are not ‘accidental’ under insurance contracts governed by ERISA.” Id. at 344–45 (citing cases). They do so because “the hazards of drinking and driving are widely known and widely publicized [and] the insured should have known that driving while intoxicated was highly likely to result in death or bodily harm.” Id. at 345 (citation and quotation marks omitted). In Cozzie v. Metropolitan Life Insurance Co., 140 F.3d 1104 (7th Cir. 1998), the Seventh Circuit held that MetLife did not act arbitrarily and capriciously when it denied benefits after an insured died with a blood-alcohol level of 0.252. Id. at 1106. The MetLife plan at issue did not define “accident” and the court held that MetLife reasonably defined “accident” as “not reasonably foreseeable.” Id. at 1108–09. Of note, the court stated that “MetLife’s interpretation is rational because it is consistent with the goals of the plan [namely] . . . to provide . . . insurance against the tragedy of unexpected death by providing additional benefits for those who experience such a loss and all its consequent tremors.” Id. at 1110. Denying benefits to some, the court reasoned, “ensure[s] that payments are reserved for those who truly fall within the terms of the policy.” Id. Indeed, the very number of cases holding similarly to Eckelberry and Cozzie independently supports the conclusion that MetLife’s determination was not arbitrary and capricious. See Poeppel v. Hartford Ins. Co., 87 F. App’x 885, 886 (4th Cir. 2004); Gilbert v. Estate of Cox, No. 05-283- JBC, 2007 WL 2023576, at *3 (E.D. Ky. July 10, 2007); Richardson v. Mutual of Omaha Ins. Co., No. 3:06CV-197-H, 2007 WL 1577942, at *3 (W.D. Ky. May 31, 2007); Weatherall v. ReliaStar Life Ins. Co., 398 F. Supp. 2d 918, 924 (W.D. Wis. 2005); Nelson v. Sun Life Assurance Co. of Canada, 962 F. Supp. 1010, 1012 (W.D. Mich. 1997). We nevertheless recognize the logical force of the district court’s analysis in this case. Driving drunk is stupidly risky, but perhaps not statistically more risky than actions we might be loath to condemn, such as test piloting or grabbing a child from in front of an oncoming train. We do not doubt the accuracy of the observation that a person “is far more likely to be arrested for driving while intoxicated than to die or be injured in an alcohol-related automobile crash, and far more likely to arrive home than to be either arrested, injured, or killed.” See West, 171 F. Supp. 2d at 904. Lennon also notes,“one out of every 9,128 alcohol-impaired trips results in a crash that causes a fatality.” See Eckelberry v. ReliaStar Life Ins. Co., 402 F. Supp. 2d 704, 712 (S.D. W.Va. 2005). One is more likely to be struck by lightning, according to the plaintiff, than to die as a result of one’s own drunk driving. Id. One might well question the relevance of such ratios to a case like the present one where there is evidence that Lennon’s blood-alcohol level was 0.321, over three times the legal limit, so No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 6 that Lennon’s trip was not merely “alcohol-impaired.” We can take judicial notice of the fairly obvious scientific fact that as blood-alcohol levels rise, “so does the risk of being involved in a fatal crash.” Nat’l Hwy. Traffic Safety Admin., U.S. Dep’t of Transp., Setting Limits, Saving Lives: The Case for 0.08 BAC Laws, DOT HS 809 241, Apr. 2001, at Sec. IV; see also http://en.wikipedia.org/wiki/Blood_alcohol. Thus, drivers with blood-alcohol levels above the legal limit as a group are far more likely to arrive home safely than drivers who are extremely drunk. See Stamp v. Metro. Life Ins. Co., 466 F. Supp. 2d 422, 432 (D.R.I. 2006) (“The statistics . . . are meaningless in this context. . . . They do not consider . . . the degree of his intoxication.”). We assume for the purposes of argument, however, that persons who drive while very drunk may have a better than even, or even a pretty good, chance of not being injured. This does not keep the activity from being reckless. The same could be said, after all, of a person playing Russian roulette, who may have a 5 out of 6 chance of not being injured. What is dispositive, however, is that at some point the high likelihood of risk and the extensive degree of harm risked, weighed against the lack of social utility of the activity, become not marginally but so overwhelmingly disproportionate that the resultant injury may be outside a definition of “accidental” that is not unreasonably narrow. Of course, “accidental” could perhaps be more broadly defined to include dangerous activity as long as injury is not intended or substantially certain. Drunk driving injuries might fit within6such a definition. If our review were de novo, this possibility would require serious consideration. But under arbitrary-and-capricious review we need not decide what is the best reading of words in the insurance policy, but whether the plan administrator’s interpretation is arbitrary. Interpreting the result of reckless drunk driving as not “accidental” for the driver is not arbitrary. We do not need to go further in this case. In particular, we do not reach the question of whether a fiduciary can reasonably deny “accidental” benefits for injury that results from any negligent or any illegal behavior, or from driving while only somewhat impaired. See Eckelberry, 469 F.3d at 347 (distinguishing drunk driving from driving while fiddling with the radio dial); Cozzie, 140 F.3d at 1110 (“We do not mean to suggest that MetLife could sustain a determination that all deaths that are causally related to the ingestion of alcohol, even in violation of law, could reasonably be construed as not accidental.”); cf. RESTATEMENT (SECOND) OF TORTS §500 cmt. e (1965) (“The mere fact that certain precautions are required by a statute . . . does not of itself make the intentional omission of the statutory precaution reckless indifference [unless] the precautions required [are] such that their omission will be recognized as involving a high degree of probability that serious harm will result.”). Nor does today’s holding extend to risky activities that may have social value greater than driving drunk, such as skiing, or driving over the speed limit to get a woman in labor to the hospital. Instead, the conclusion is only that because Lennon’s conduct constituted reckless and entirely unwarranted risk to himself, it was not arbitrary and capricious for MetLife to treat the injury as nonaccidental under the terms of its policy. 6 The concurrence unfortunately misreads this opinion as “address[ing] the question whether it was correct, as a matter of substantive law, for MetLife to deny benefits here, rather than the question whether or not it was merely arbitrary and capricious to do so.” As a fair reading of this opinion—and particularly of the statement here in the text—shows, I would base affirmance entirely on the arbitrary-and-capricious scope of review, and reserve judgment on the correctness (without regard to deference) of treating the incident as an accident under the policy. Moreover, while the concurrence would rely only on other federal court cases, which in turn rely simply on likelihood of harm, to uphold the insurance company’s determination under the arbitrary-and-capricious standard, such reliance without more implies that high likelihood of injury is sufficient to uphold denial of accident insurance whenever the arbitrary-and-capricious standard applies. The instant opinion is intended to avoid making that implication in favor of the insurance companies. There may be cases—different from the instant case—in which high likelihood of injury is not sufficient to say that activity was not accidental. We need not go beyond reckless conduct to conduct that is simply “highly likely” to cause injury, and I would not do so. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 7 Because it was not arbitrary and capricious for MetLife to find that Lennon’s death was not “accidental,” we need not reach MetLife’s argument that Lennon’s death was a result of self- inflicted injury. For the foregoing reasons, the judgment of the district court is reversed. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 8 ___________________________________ CONCURRING IN THE JUDGMENT ___________________________________ BOGGS, Chief Judge, concurring in the judgment. Although I agree that MetLife did not act arbitrarily and capriciously in determining that David Lennon’s fatal collision was not an “accident” as contemplated by the terms of GM’s Personal Accident Insurance, I reach this conclusion on somewhat different grounds than the lead opinion does. A major portion of that opinion appears to address the question whether it was correct, as a matter of substantive law, for MetLife to deny benefits here, rather than the question whether or not it was merely arbitrary and capricious to do so. The court’s discussion of concepts from negligence law, which leads it to answer the former question in the affirmative, is not required to answer the latter. The heart of the district court’s error lies in its evaluation of the seminal case of Wickman v. Northwestern Nat’l Ins. Co., 908 F.2d 1077 (1st Cir. 1990), and its progeny. In Wickman, the First Circuit–aware that it might well be “miring in a Serbonian bog”–attempted to develop a standard by which to determine what constitutes an accident for ERISA purposes. Id. at 1087. The Wickman court concluded that, outside the comparatively rare case where evidence indicates that the victim actually expected the injury suffered, “one must ask whether a reasonable person, with background and characteristics similar to the insured, would have viewed the injury as highly likely to occur as a result of the insured's intentional conduct.” Id. at 1088 (emphasis added). Wickman’s standard, as stated, is a high bar, and arguably many collisions involving a drunk driver would not meet it: as a number of courts have noted, the number of drunk driving arrests swamps the number of drunk driving injuries or deaths, making it difficult to conclude that an injurious collision is “highly likely to occur as a result” of driving while intoxicated. See, e.g., Eckleberry v. ReliaStar Life Ins. Co., 402 F. Supp. 2d 704, 712 (S.D. W.Va. 2005); West v. Aetna Life & Accident Ins. Co., 171 F. Supp. 2d 856, 904 (N.D. Iowa 1985). A number of courts, including at least one district court in this circuit, applying this standard as formulated above, have found a specific drunk-driving collision to be accidental. See, e.g., Harrell v. Metro. Life Ins. Co., 401 F. Supp. 2d 802 (E.D. Mich. 2005). However, a number of courts–in particular, several district courts in this circuit–finding Wickman persuasive and purporting to apply it, have (intentionally or not) modified its objective standard from one of “high likelihood” to “reasonable foreseeability,” and have concluded that a collision by a highly intoxicated driver (including a driver with a blood alcohol level comparable to, and even considerably lower than, Lennon’s), being reasonably foreseeable, is not an accident. See, e.g., Miller v. Auto-Alliance Int’l Inc., 953 F. Supp. 172 (E.D. Mich. 1997); Walker v. Metro. Life Ins. Co., 24 F. Supp. 2d 775 (E.D. Mich. 1997) (.29 g/100 ml); Nelson v. Sun Life Assurance Co. Of Canada, 962 F. Supp. 1010 (W.D. Mich. 1997) (.18 g/100 ml; reviewing de novo); Cates v. Metro. Life Ins. Co., 14 F. Supp. 2d 1024 (E.D. Tenn. 1996) (.18 %), aff’d, 149 F.3d 1182 (6th Cir. 1998) (unpublished); Fowler v. Metro. Life Ins. Co., 938 F. Supp. 476 (W.D. Tenn. 1996) (.26%). The district court examined these cases (all of which had been cited by MetLife in its second, expanded denial letter), and found them “unpersuasive,” for failing to adhere to the “highly likely” standard articulated in Wickman, and failing to identify “any evidence to support the conclusion that an intoxicated driver clearly should foresee death.” 466 F. Supp. 2d at 750 (emphasis added). If the administrator’s decision here were subject to the district court’s de novo review, its rejection of these cases in favor of the stricter standard advanced in Wickman (or some other standard entirely) No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 9 might have been appropriate.1 But when it reviews under an arbitrary and capricious standard, the district court cannot simply substitute its judgment for that of the administrator. See, e.g., Motor Vehicle Mfrs. Ass’n of the U.S. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983). Given that a number of courts, including several within this circuit, have approved characterizing drunk- driving collisions involving drivers of comparable or lower intoxication levels as non-accidents for ERISA purposes, there was no basis for the district court to conclude that MetLife’s decision to do the same–under the set of facts presented here, articulated by MetLife and well-analyzed in detail in the lead opinion–was arbitrary and capricious. Accordingly, I would reverse the judgment of the district court on the grounds that, in following a standard adhered to by a number of courts, within and without this circuit, MetLife did not act arbitrarily and capriciously, without reaching the question whether to approve or disapprove that standard. 1 Though of course simply departing from the requirements of Wickman, which is not binding authority in this circuit, would not itself be sufficient to reject these cases or the standard they articulate. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 10 _______________ DISSENT _______________ CLAY, Circuit Judge, dissenting. The majority opinion today marks a clear departure from federal common law, an affront to common sense, and even more troublesome, an elevation of moralistic judgments above the interpretation fairly attributable to the Personal Accident Insurance (“PAI”) Policy before us today. In my view, the district court properly concluded that Defendant’s interpretation of the PAI Policy was arbitrary and capricious. Accordingly, I dissent. At the heart of this dispute lies a question that should admit of a simple and straightforward answer, and that would if the question were posed to any man on the street. The seemingly simple question is whether a motorist intoxicated beyond the legal limit who crashes his vehicle has been in an “accident,” or has been “accidentally” injured. A man on the street would answer “yes.” But the question (or some form of it) was put to an ERISA plan administrator and then to a court. The matter quickly became over complicated by exclusions read into express contractual language, by standards of review, and by something akin to Cardozo’s great “Serbonian bog” – an unwieldy body of legal precedent laced with not-so-subtle moralistic judgments. See Landress v. Phoenix Mut. Life Ins. Co., 291 U.S. 491, 499 (1934) (Cardozo, J., dissenting). At any rate, the question before this Court is whether Defendant was arbitrary and capricious in denying Plaintiff benefits under the insured’s PAI Policy. To answer this question, we must specifically decide whether Defendant was arbitrary and capricious to find (1) that the insured was not “accidentally” injured, and (2) that the insured sustained a “self-inflicted injury,” within the meaning of the PAI Policy. Arbitrary-and-capricious review properly applies here because the ERISA plan vests Defendant, the plan administrator, with discretion to construe the terms of the PAI Policy. See Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 115 (1989); Jones v. Metropolitan Life Ins. Co., 385 F.3d 654, 660 (6th Cir. 2004). Although arbitrary-and-capricious review is characteristically deferential, it “does not require us merely to rubber stamp the administrator’s decision.” Jones, 385 F.3d at 661. Our obligation to exercise some review extends to “the quality and quantity of the . . . evidence and the opinions on both sides of the issues.” McDonald v. Western-Southern Life Ins. Co., 347 F.3d 161, 172 (6th Cir. 2003). Our “deference need not be abject.” Id. To the extent that we do defer, however, it is only to the plan administrator’s construction of terms in the policy, and we should not hesitate to conclude that a plan administrator arbitrarily and capriciously interpreted a policy by adding exclusions or terms thereto under the guise of interpretation. See Jones, 385 F.3d at 665. I. WHETHER THE INJURIES WERE “ACCIDENTAL” The PAI Policy at issue makes benefits payable when the insured “sustains accidental bodily injuries, and within one year thereafter shall have suffered loss of life . . . as a direct result of such bodily injuries independently of all other causes.” (J.A. at 153) Additionally, it expressly excludes some losses from coverage on the basis of causation or mode. For example, no benefit will be paid “for any loss which is contributed to or caused, wholly or partly, directly or indirectly, by . . . suicide, attempted suicide or self-inflicted injury while sane or insane.” (Id. at 152) Notably, the PAI Policy contains no exclusion for injuries sustained while driving under the influence of alcohol. (See id. at 151-53) Nor does it define the terms “accident” or “accidental.” Nevertheless, Defendant denied PAI benefits to Plaintiff in part on the ground that “[t]he act of driving while so impaired No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 11 rendered the infliction of serious injury or death reasonably foreseeable and, hence, not accidental as contemplated by the plan.”1 (Id. at 201) We interpret ERISA plans “according to their plain meaning, in an ordinary and popular sense.” Perez v. Aetna Life Ins. Co., 150 F.3d 550, 556 (6th Cir. 1998). In my view, the most natural reading of the word “accidental,” as it is used in its “ordinary and popular sense,” extends to injuries sustained in motor vehicle crashes while the driver is under the influence of alcohol, even at levels above the legal limit. The public would call the crash an “accident,” and the resultant injuries or death “accidental.” See Wickman v. Northwestern Nat’l Ins. Co., 908 F.2d 1077, 1086 (1st Cir. 1990) (“Probably the best definition [of “accident”] is Cardozo’s tautology that an accident is what the public calls an accident.”). This “plain” and “ordinary” meaning is borne out in the insured’s death certificate, the police report, and the medical examiner’s report, all of which classify the manner of death as an “accident.” It also comports with at least one dictionary definition of the word. Webster’s dictionary defines “accidental” as “happening or ensuing without design, intent, or obvious motivation or through inattention or carelessness.” Webster’s 3d New Int’l Dictionary 11 (1993). Certainly, an intoxicated motorist does not intend to crash, much less to die behind the wheel while driving home. If the intoxicated motorist does crash his vehicle, the crash and any concomitant injuries are “accidental” in the sense that they result from “inattention or carelessness.” Lamentably though, given the extensive amount of ink spilled in federal and regional reporters over the “ordinary” meaning of “accidental,” such popular conceptions of its meaning cannot be relied upon in interpreting an ERISA plan. In fact, the2 barrage of case law on the subject suggests that the meaning of “accidental” is anything but plain. Where no plain meaning can be discerned from the plan language, courts typically look to the federal common law to assess the reasonableness of a plan administrator’s interpretation. See Pilot Life Ins. Co. v. Dedeaux, 481 U.S. 41, 56 (1987); Jones, 385 F.3d at 661; Buce v. Allianz Life Ins. Co., 247 F.3d 1133, 1146 (11th Cir. 2001) (“[W]here the crucial terms of an accident policy are defined with surpassing vagueness, . . . to deploy the federal common law of ERISA to give some unity to the concept of ‘accident’ is sound judicial policy.”). The First Circuit in Wickman v. Northwestern National Insurance Co. first forged a path for the federal common law on this question. The insured in Wickman died after free-falling from a bridge onto railroad tracks 90 feet below. Wickman, 908 F.2d at 1080-81. Just before the fall, a witness observed the insured standing outside the bridge’s guardrail, holding onto it with one hand, at a point reachable only by walking “head on into high speed traffic.” Id. at 1080. The insured’s accidental death policy defined an “accident” vaguely, as “an unexpected, external, violent and sudden event.” Id. at 1081. The ERISA plan administrator denied benefits, and the beneficiary filed suit. The Wickman court surveyed state judicial interpretations of “accidental,” but after doing so, ultimately “elect[ed] to pursue a path for the federal common law.” Id. at 1085-86. The Wickman court set forth the following approach: First, the court said, consider “the reasonable expectations of the insured when the policy was purchased.” Id. at 1088. That is, did the insured subjectively expect an injury similar in type to the kind suffered to follow from his conduct? If the fact-finder determines that the insured did not expect such an injury, or if the fact- finder cannot ascertain the insured’s subjective expectations, an objective analysis applies. Id. The 1 On the basis of the administrative record, I assume for purposes of this dissent that the insured was legally intoxicated at the time of the crash. 2 Were this Court tasked with interpreting the language de novo, in view of the word’s apparent ambiguity, the rule of contra proferentum would apply. See Marquette Gen. Hosp. v. Goodman Forest Indus., 315 F.3d 629, 632 n.1 (6th Cir. 2003) (citing Perez, 150 F.3d at 557 n.7). Ambiguity in the word “accident” would then be construed against the plan administrator, thus extending coverage to the insured here. Because we review the plan administrator’s decision under an arbitrary-and-capricious standard, the rule is not determinative. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 12 objective analysis consists of a determination “whether a reasonable person, with background and characteristics similar to the insured, would have viewed the injury as highly likely to occur as a result of the insured’s intentional conduct.” Id. (emphasis added). If the insured “actually expected the result, even if he did not specifically intend it,” then his subsequent death is not accidental. Id. at 1089. Applying its framework, the Wickman court concluded that the insured “knew or should have known that serious injury or death was a probable consequence substantially likely to occur as a result of his volitional act.” Id. The insured “either subjectively expected serious injury, or . . . [o]bjectively, he reasonably should have expected serious injury when he climbed over the guardrail and suspended himself high above the railroad tracks below.” Id. Since Wickman, reviewing courts have largely distorted the inquiry, seizing upon the Wickman court’s conclusory words and eschewing the very test that it established. See, e.g., Buce, 247 F.3d at 1147; Baker v. Provident Life & Accident Ins. Co., 171 F.3d 939, 942-43 (4th Cir. 1999); Cozzie v. Metropolitan Life Ins. Co., 140 F.3d 1104, 1109-10 (7th Cir. 1998). In essence, they have given effect to the court’s summary statement that “objectively, [the insured] reasonably should have expected serious injury” – a statement which, more likely than not, follows from the vaguely worded definition of “accident” contained in the plan at issue there. See Wickman, 908 F.2d at 1081 (noting the plan vaguely defined “accident” as “an unexpected, external, violent and sudden event”) (emphasis added). However, the conclusory language in Wickman should not be taken to replace the test established earlier in the opinion, which speaks in stronger terms of injury “highly likely to occur”and “probable consequence[s] substantially likely to occur,” and more closely achieves the purpose of the objective inquiry – to determine whether a reasonable person actually expected the result. Thus, many courts have incorrectly framed the objective prong of the Wickman inquiry in terms that water it down in substance, asking whether an injury was “reasonably foreseeable.” As a result, the federal common law has strayed from the path initially forged by the Wickman court. I would frame the objective inquiry as Wickman did, and would ask “whether a reasonable person, with background and characteristics similar to the insured, would have viewed the injury as highly likely to occur as a result of the insured’s intentional conduct.” Wickman, 908 F.2d at 1088. This formulation clearly requires something greater than mere foreseeability. Here, the administrative record gives no indication of the insured’s subjective expectations and, accordingly, the objective inquiry should guide this court in deciding whether Defendant rendered an unreasonable interpretation of the PAI Policy. On the basis of Defendant’s denial letter, Defendant evidently defines “accidental” to exclude acts which “render[] the infliction of serious injury or death reasonably foreseeable.” (See J.A. at 201) Defendant’s interpretation of “accidental” therefore does not comport with the Wickman court’s objective inquiry, which asks whether injury was “highly likely to occur as a result of the insured’s” act. See Wickman, 908 F.2d at 1088. A “highly likely” consequence is something probable, not merely possible. A reasonable person, with a background and characteristics similar to the insured, undoubtedly knows that driving while legally intoxicated entails risks of injury, death, arrest, and prosecution. However, knowledge of the risk of injury or death – both potential consequences of driving while intoxicated – does not equal knowledge of probable injury or death. Statistically speaking, legally intoxicated motorists arrive safely at their destination without incident more often than not. Of those that do not, police apprehend and arrest a great many legally intoxicated motorists. Comparatively, the number of legally intoxicated motorists either injured or killed in crashes linked to alcohol is rather small. Illustratively, according to the National Highway Traffic Safety Administration, 17,105 people died in alcohol-related motor vehicle crashes in 2003, a figure that accounts for 40 percent of all traffic-related deaths that year. The Federal Bureau of Investigation’s 2003 Uniform Crime Report indicates that an estimated 1,448,148 motorists were arrested for driving under the influence that year. Finally, the Substance Abuse and Mental Health Services Administration, in a report issued September 2005, found an estimated 30.7 million persons nationwide took alcohol-impaired No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 13 trips during 2003.3 Even assuming only 10 million alcohol-impaired trips occurred that year, a mere 14.4 percent of impaired motorists were arrested, while 0.17 percent died in alcohol-related incidents. Thus, injury or death most certainly cannot be deemed a “highly likely” consequence of driving while intoxicated. See West v. Aetna Life Ins. Co., 171 F. Supp. 2d 856, 904 (N.D. Iowa 2001) (“What ‘common knowledge’ should actually tell a person driving while intoxicated is that he or she is far more likely to be arrested for driving while intoxicated than to die or be injured in an alcohol related automobile crash, and far more likely to arrive home than to be either arrested, injured, or killed.”). In effect, under the guise of interpretation, Defendant took it upon itself to rewrite the PAI Policy by adding terms where none previously existed. See Jones, 385 F.3d at 665. Defendant interpreted what should be an inclusive term to exclusionary effect, excluding coverage for accidental injury following from acts which “render[] the infliction of serious injury or death reasonably foreseeable.” As applied to Plaintiff’s claim, Defendant purports to interpret “accidental” in this manner, but in reality add a new exclusion for accidental injury sustained by legally intoxicated motorists. This sort of post-hoc requirement falls well outside the bounds of Defendant’s discretion as a plan administrator interpreting an ERISA plan, which notably “does not include the authority to add eligibility requirements to the plan.” See Jones, 385 F.3d at 661. What is more, it renders meaningless several express exclusionary provisions, including exclusions for flight in an aircraft while a student pilot, injuries caused in whole or part by war or war-like action, or the use of drugs other than as prescribed by a physician. In fact, if injury or death were a “highly likely” or even “reasonably foreseeable” consequence of driving while legally intoxicated, one would expect Defendant to expressly exclude from coverage accidents which involved driving under the influence. As the plan drafter, Defendant had every means at its disposal to do so. Yet, when Defendant drafted the PAI Policy, they elected not to include express exclusionary language to that effect. Such language would undoubtedly have foreclosed much litigation in this circuit alone over the coverage in Defendant’s PAI Policy. Notably, according to Plaintiff’s brief on appeal, Defendant recently revised its PAI Policy to explicitly exclude coverage for injuries sustained while “driving a vehicle while intoxicated as defined by the laws of the jurisdiction in which the vehicle was being operated.” (See Pl.’s Br. at 32-33) At any rate, the scope of activities that “render[] the infliction of serious injury or death reasonably foreseeable” is substantially expansive. To accept this formulation – the very words of Defendant in denying PAI benefits here – would be to eviscerate accidental injury coverage in many circumstances where the insured, on the basis of the policy language, would expect to be covered. A few examples prove illustrative here. A motorist driving cross-country attempts to make it another hour before stopping after an 18 hour day behind the wheel. The motorist is not speeding, drives in accordance with the laws, and encounters no other vehicles but, ultimately, fatigue overcomes him. He swerves off the road into a ditch and later dies from injuries sustained in the crash. Another motorist drives 89 miles per hour on a road with a designated speed limit of 70. Arriving at a turn in the road, that motorist spins out, unable to control his vehicle. He suffers injury when his vehicle hits a cement wall in the road’s median and also dies. Finally, a man partakes in bungee jumping for sport. He has successfully completed several jumps before, but on his last jump, his safety harness fails and he plummets to his death. Under the PAI Policy language, an insured would expect to be covered in each of these hypothetical situations. Yet, on Defendant’s reasoning, 3 The findings were annual averages calculated on the basis of combined data from 2002 and 2003. See Arrests for Driving Under the Influence Among Adult Drivers, National Survey on Drug Use and Health, http://oas.samhsa.gov/2k5/DUIarrests/DUIarrests.pdf. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 14 which the majority affirms today, Defendant could deny coverage by calling the injury or death a “reasonably foreseeable” result of the insured’s conduct. The reason why an insurance company would most likely not deny coverage in these circumstances is simple: the absence of alcohol, drugs, or other circumstances rendering the action morally questionable. In the final analysis, the “reasonably foreseeable” formulation is little more than a tool enabling plan administrators and courts to transform moral judgments about the insured’s conduct into arbitrary denials of coverage under vaguely worded ERISA plans. See, e.g., Eckelberry v. Reliastar Life Ins. Co., 469 F.3d 340, 346 (4th Cir. 2006) (“To characterize harm flowing from [drunk driving] as merely ‘accidental’ diminishes the personal responsibility that state laws and the rules of the road require.”); Cozzie, 140 F.3d at 1110 (“We cannot say . . . that MetLife’s determination that the purposes of the plan are best served by acknowledging a qualitative difference between the ingestion of a huge quantity of alcohol and other tragedies of human life . . . is incompatible with the goals of the plan.”); Metropolitan Life Ins. Co. v. Potter, 992 F. Supp. 717, 721 (D. N.J. 1998) (“‘[D]riving while intoxicated is too great a risk to be tolerated without penalty. In today’s world, people who drink and drive must be charged with responsibility for their own acts.’”) (quoting MetLife’s denial letter); see also Adam F. Scales, Man, God and the Serbonian Bog: the Evolution of Accidental Death Insurance, 86 IOWA L. REV. 173, 299 (2000) (“Drunk driving accidents are simply more senseless, more unforgivable, and altogether more deserving of moral disapprobation than other accidents. But they are accidents nonetheless.”); id. at 302 (“To label merely unforgivable conduct as intentional is to extend forever the moral liability of voluntarily-undertaken acts. But equally voluntary conduct unaccompanied by evil or questionable motives is typically treated differently.”). In some cases, the moral condemnation becomes even more evident when one examines the plan administrator’s or the court’s selection between the two divergent formulations of the Wickman objective analysis used by courts. See, e.g., Santaella v. Metropolitan Life Ins. Co., 123 F.3d 456, 462, 463-64 (7th Cir. 1997) (using the stronger “highly likely” formulation where the record revealed no evidence of illegal drug abuse). I do not condone the pernicious effects of drunk driving, nor those who perpetrate it. But neither would I permit moralistic judgments to lull me to acquiesce in Defendant’s purported “interpretation” of the PAI Policy. The majority’s social utility calculus only amplifies and enables the sort of moralistic judgments we should be loathe to employ, much less to encourage. It directly links recovery under an “accidental injury” provision to notions of desert. Under this calculus, the plan administrator dare not deny coverage to either motorist in the above examples if they were en route to donate a kidney to their dying mother and had only limited time to reach their destination. However, less beneficial conduct, like delivering a brick of cocaine to a mass drug dealer, would militate in favor of denial of PAI benefits. Applying the social utility calculus, beneficiaries of the motorist rushing to his dying mother’s bedside may recover PAI benefits, while the plan administrator may deny the PAI claim received from beneficiaries of the motorist/drug runner. The problem, of course, is that the behavior leading to the injury – and ultimately, the loss of life – is identical for both motorists, yet the coverage decision will vary with the value society ascribes to the purpose of the conduct. Thus, under the social utility calculus, moral judgments drive the distinction between covered acts and those excluded from coverage. The majority takes pains to avoid discussing the plain meaning of the PAI Policy, or the inquiry under federal common law. Instead, the majority classifies injury following from “grossly negligent” or “reckless” behavior as by definition not “accidental.” In doing so, the majority imports long established concepts of tort law into the enterprise of interpreting ERISA plans. The federal common law applies basic principles of contract interpretation, often times borrowed from state contract law, when ambiguous terms in ERISA plans raise questions of coverage. See Wickman, 908 F.2d at 1084. In at least one state case within this circuit, Fryman v. Pilot Life Ins. Co., 704 S.W.2d 205, 206 (Ky. 1986), the Kentucky Supreme Court declined to apply tort law No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 15 principles in interpreting an accidental injury policy. There, the court indicated “reluctan[ce] to analyze contract terms under principles which have technical meaning in other areas of the law,” and expressly stated that “principles of tort law . . . have no application to the contract issue in question.” Id. In fact, the well-established canon of contract interpretation that words be afforded their plain and ordinary meaning requires plan administrators and courts alike to eschew constructs tethered to technical legal concepts like gross negligence and recklessness. Drawing an analogy from tort law, the majority reasons that if tort law can treat grossly negligent or reckless conduct like intentional conduct, then an ERISA plan administrator can similarly treat grossly negligent or reckless conduct as not accidental. The force of this analogy, it seems, is in pointing out that the insured intended to drive drunk. This may be so, but it says nothing of the insured’s expectations or intentions with respect to the consequence of his voluntary act. The majority thereby harkens back to the distinction between “accidental means” and “accidental ends” long abandoned by the common law. Further, the analogy draws its force from equating intentional with not accidental. If applied, as the terms of the policy require, to the words “bodily injury,” such that “accidental bodily injury” by definition excludes “intentional bodily injury,” this construct renders superfluous the PAI Policy’s express exclusion for intentional “self-inflicted injury.” In effect, the majority joins the plan administrator here in rewriting the PAI Policy. Under the pretense of interpreting the term “accidental,” Defendant read a new exclusion into its PAI Policy. Defendant’s interpretation, and subsequent denial of PAI benefits, is therefore arbitrary and capricious. II. SELF-INFLICTED INJURY EXCEPTION Defendant’s alternative rationale – that the insured died from a “self-inflicted injury” – must also fail as an arbitrary and capricious interpretation of the PAI Policy’s listed exclusion. Expressly, the PAI Policy excludes coverage “for any loss which is contributed to or caused, wholly or partly, directly or indirectly, by . . . suicide, attempted suicide or self-inflicted injury, while sane or insane.” (J.A. at 152) Defendant invoked this exclusion, reasoning that “the mental and physical impairments caused by the voluntary consumption of excessive amounts of alcohol constitute intentional self- inflicted injuries under the plan.” (Id. at 202) A decision of the Eighth Circuit sitting en banc dealt with this precise issue in a very persuasive manner. In King v. Hartford Life & Accident Insurance Co., 414 F.3d 994 (8th Cir. 2005) (en banc), the insured died in a motorcycle accident. At the time, he had a blood alcohol content (“BAC”) of 0.19. Id. at 997. His accidental death insurance policy contained an exclusion for “intentionally self-inflicted injury, suicide, or attempted suicide, whether sane or insane,” much like the policy exclusion in the instant case. See id. at 1004. The ERISA plan administrator relied on the intentionally self-inflicted injury exclusion to deny benefits, claiming the insured’s “alcohol intoxication was itself an ‘intentionally self-inflicted injury’ that ‘contributed to’ his injuries and death.” Id. Poignantly, rejecting the plan administrator’s interpretation, the King court observed that “[o]ne rarely thinks of a drunk driver who arrives home safely as an ‘injured’ party, and to define drinking to the point of intoxication as an ‘intentionally self-inflicted injury . . .’ is at least ‘a startling construction.’” Id. (citation omitted). Additionally, in light of other enumerated exclusions in the policy, the King court found the ERISA plan administrator’s interpretation would render meaningless exclusions for losses caused by “taking drugs . . . unless prescribed . . . by a licensed physician.” Id. at 1004-05. On appeal, Defendant contends that the “intentional self-inflicted injury” was the insured’s voluntary intoxication “to the extent that [his] BAC was 0.321, a BAC evidencing extreme injuries to mind and body, probably including loss of consciousness.” (Def.’s Br. at 50) That intentional self-inflicted injury, the argument goes, then partially contributed to the loss, the insured’s death. No. 06-2234 Lennon v. Metropolitan Life Insurance Co. Page 16 However, this argument is not supported by substantial evidence in the administrative record, which says nothing of “extreme injuries to mind and body” or “loss of consciousness” occurring before the accident. Additionally, like the policy in King, the PAI Policy enumerated an exclusion for “use of any drug or medicine unless taken on the advice of and in accordance with the direction of a licensed physician.” (J.A. at 152) As in King, Defendant’s interpretation would render the “drug” exclusion meaningless. See King, 414 F.3d at 1005. Consequently, I agree with the district court’s conclusion that Defendant arbitrarily and capriciously interpreted the “self-inflicted injury” exclusion in order to deny PAI benefits to Plaintiff. III. LACK OF DELIBERATE REASONING PROCESS Finally, nothing on the record indicates that Defendant’s interpretation and ultimate denial of PAI benefits here followed from a deliberate and principled reasoning process. First, Defendant denied the PAI benefits by way of a “DWI denial” form letter which, among other things, rattled off a list of cases in an apparent attempt to insulate the denial from even the most minimally searching review. (See J.A. at 227) Defendant’s denial is more akin to an automatic rejection triggered by the fact of the insured’s intoxication than to the result of any truly deliberative interpretive enterprise. Second, Defendant issued conflicting denial letters. The initial denial letter stated that “[b]ecause of his voluntary alcohol consumption and attempt to drive while so impaired, [the insured’s] death was directly the result of accidental injuries, independently of all other causes.” (Id. at 225-26 (emphasis added)) Plaintiff seized upon this statement in appealing the denial. Defendant’s subsequent letter upholding its initial denial of the PAI claim called it a “typographical error,” noting the sentence should have read, “[the insured’s] death was not directly the result of accidental injuries.” (Id. at 201-02) Third, at oral argument before this court, Defendant’s counsel could not articulate the criteria Defendant applies in resolving claims like Plaintiff’s. Although not dispositive, the apparent lack of process does little to indicate that Defendant’s denial was not arbitrary and capricious. For all of the foregoing reasons, I believe Defendant acted arbitrarily and capriciously when it interpreted the PAI Policy – specifically, the term “accidental” and the express exclusion for “intentional self-inflicted injury.” I would affirm the district court’s well-reasoned opinion.