Angela JOHNSON, Plaintiff–Appellant,
AMERICAN UNITED LIFE INSURANCE COMPANY, Defendant–Appellee.
716 F.3d 813 (2013)
United States Court of Appeals,
Argued: Dec. 5, 2012.
Decided: May 24, 2013.
[716 F.3d 815]
ARGUED:Gavin James Reardon, Rossabi Black Slaughter, PA, Greensboro, North Carolina, for Appellant. Elizabeth J. Bondurant, Smith Moore Leatherwood, LLP, Atlanta, Georgia, for Appellee. ON BRIEF:T. Matthew Creech, Smith Moore Leatherwood, LLP, Greensboro, North Carolina, for Appellee.
Before TRAXLER, Chief Judge, and FLOYD and THACKER, Circuit Judges.
Reversed and remanded by published opinion. Chief Judge TRAXLER wrote the opinion, in which Judge FLOYD and Judge THACKER joined.
TRAXLER, Chief Judge:
Richard Johnson (“Richard”) participated in an employee welfare benefit plan (“the Plan”) that provided life insurance and accidental death and dismemberment (“AD & D”) benefits through group policies issued by American United Life Insurance Company (“AUL”). When Richard died in a singlevehicle crash, his widow Angela Johnson (“Johnson”) received life insurance benefits, but AUL, which also served as administrator for the Plan, refused to pay AD & D benefits. Richard was highly intoxicated at the time of his fatal crash, and AUL concluded that Richard’s drunk-driving death was not the result of an “accident” under the Plan. Johnson filed this action under the Employee Retirement Income Security Act (“ERISA”), see29 U.S.C. § 1132(a)(1)(B), to recover these AD & D benefits. Applying a de novo standard of review, the district court affirmed the denial of benefits on the grounds that Richard’s death was not accidental because the fatal crash was an “anticipated and expected” result of driving while intoxicated.
For the reasons set forth in detail below, we must reverse. The insurance policies do not define the term “accident” despite its critical importance for determining eligibility for AD & D benefits. Because
[716 F.3d 816]
“accident” is susceptible to more than one reasonable interpretation, we construe it against AUL, the drafting party, and conclude that a reasonable plan participant under similar circumstances would have understood Richard’s alcohol-related crash to be an “accident” under the policy language.
The question of whether drunk-driving deaths or injuries are “accidental” for purposes of accidental death insurance has perplexed the judiciary for some time.1 Given the “sheer number of court cases nationwide involving disputes over claims by drunk drivers,” an insurer surely understands that it will “likely face claims under its AD & D policies based on injuries sustained in alcohol-related collisions.” Kovach v. Zurich Am. Ins. Co., 587 F.3d 323, 336 (6th Cir.2009). The interpretive onus belongs on the insurers who draft these accident insurance policies; they can eliminate dilemmas like this one by clearly and plainly stating whether a loss caused by the participant’s driving drunk is “accidental” so that the insured “know[s] what he is getting in his insurance policy.” Senkier v. Hartford Life & Accident Ins. Co., 948 F.2d 1050, 1053 (7th Cir.1991).
Reaching this result gives us no great pleasure. Drunk driving is reckless, irresponsible conduct that produces tragic consequences for the thousands it touches annually. But our task in this case is not to promote personal responsibility or enforce good driving habits. We must focus on the terms of the policies issued under the Plan and determine whether Richard died as a result of an accident without “allowing our moral judgments about drunk driving to influence our review.” Kovach, 587 F.3d at 330. At bottom, this ERISA appeal presents a problem of contract law which requires us to determine the intent of the parties as reflected in the language of the policy.
The Plan provided Richard with (1) standard AD & D and life insurance benefits of $25,000 through a policy paid for by Richard’s employer, and (2) voluntary AD & D and life insurance benefits of $100,000 through a policy paid for by Richard. AUL issued both policies. Johnson is the designated beneficiary under the policies.
Under the AD & D provision in the policies, AUL pays benefits “[i]f a Person has an accident while insured under the policy which results in a [covered] loss.” J.A. 268. In the case of an accidental loss of life to the insured, the policies pay the principal benefit amount which in this case is $100,000 and $25,000 for the employee- and employer-paid policies, respectively. The policies define “accidental death” as “death due to an accident, directly and independently of all other causes,” J.A. 224, but fail to define the predicate term “accident.”
[716 F.3d 817]
The AD & D provision contains a limitations clause expressly excluding the payment of benefits in various circumstances:
Benefits are not payable for loss due directly or indirectly to:
1. suicide or attempted suicide, whether sane or insane;
2. air travel as a crew member;
3. participation in a riot or from war or an act of war, whether declared or undeclared;
4. commission of an assault or felony;
5. the voluntary taking of:
a. a prescription drug in a manner other than as prescribed by a physician;
b. any other federally- or state-controlled substance in an unlawful manner;
c. non-prescription medication, in a manner other than as indicated in the printed instructions; or
d. poison, except for accidental ptomaine poisoning.
6. the voluntary inhaling of gas (unless due to occupational accident); or
7. sickness other than infection occurring as a result of accidental injury.
With two exceptions, the terms of the employer- and employee-paid policies are virtually identical. First, the employee-paid policy includes an additional category in the limitations clause for which AD & D benefits are not payable—“participation in hang-gliding, bungee jumping, automobile racing, motorcycle racing, skydiving, rock climbing, or mountain climbing.” J.A. 269. Second, the employee-paid policy provides an additional AD & D benefit not available under the employer-paid policy—the “Seat Belt” benefit. According to the policy, “AUL will pay an additional accidental death benefit, called the Seat Belt Benefit, if a Person dies as a result of an Automobile accident while properly wearing a Seat Belt at the time of the accident.” J.A. 270.
Significantly, the Seat Belt Benefit alone is expressly subject to a drunk-driving limitation clause:
This benefit will not be paid if the Person, while operating the Automobile, was legally intoxicated as defined by applicable laws, violating traffic laws, racing, stunt-driving, or engaging in other similar activity during the accident.
In addition to the above limitation, this benefit is subject to the further limitations and provisions of this AD & D section.
J.A. 270 (emphasis added).
On August 2, 2007, at about 1:30 a.m., Richard was driving a pickup truck owned by his employer near Myrtle Beach, South Carolina, when he lost control of the vehicle and veered off the road, struck a highway sign and flipped over multiple times. Richard was partially ejected from the vehicle and sustained fatal injuries. The law enforcement officer who investigated the incident reported that the primary factor contributing to Richard’s crash was that he was driving too fast for conditions. The post-incident traffic report estimated that Richard had been driving 65 mph in a 50 mph zone. The report contains no mention of alcohol. The Certificate of Death issued by the Horry County Coroner stated under “Cause of Death” that Richard died from internal injuries from a motor vehicle accident occurring when “victim lost control of [his] vehicle, striking [the highway] pole.” J.A. 192. A toxicology report subsequently issued by the South Carolina Law Enforcement Division (“SLED”) indicated that Richard’s blood alcohol content (“BAC”) was 0.289 percent at the time of his fatal crash, significantly higher than South Carolina’s legal limit of .08 percent.
Richard’s employer submitted a claim on behalf of Johnson for benefits under the
[716 F.3d 818]
two group policies. AUL immediately paid the life insurance benefits but refused to pay AD & D benefits under the two policies (totaling $125,000). AUL issued a denial letter, concluding that, based on his state of intoxication, Richard did not die in an accident and thus did not qualify for AD & D benefits:
Based upon the police and medical reports provided to AUL, Richard Johnson had a sufficient quantity of intoxicants in his system to make him lose control of his mental and physical faculties at the time of his fatal collision. Therefore, benefits are not payable under the accidental death and dismemberment provisions of the policy.
Mr. Johnson’s death was not accidental in nature since it was not caused directly and independently of all other causes. An accident occurs when an unforeseen, sudden and unexpected event, usually of an afflictive or unfortunate character[,] occur[s]….
The hazards of drinking excessively are widely known and widely publicized. It is clearly foreseeable that drinking excessive amounts of alcohol, and driving may result in death or bodily harm. As the decedent should have foreseen the consequences of drinking excessive amounts of alcohol, a determination of his death not being accidental is reasonable.
J.A. 359. The denial letter also noted that the limitations clause specifically applicable to the Seat Belt Benefit excluded any payment of benefits in light of the fact that Richard’s BAC exceeded the legal limit.
The employer sent a written request asking AUL to reconsider its denial, observing that “being intoxicated by alcohol is not listed as a limitation under the AD & D benefit. The only reference to a limitation for legal intoxication appears on the following page and appears to apply to the seat belt benefit exclusively.” J.A. 170. On December 14, 2007, AUL again denied AD & D benefits on the same grounds as before—that Richard’s death was reasonably foreseeable at his level of intoxication and therefore not accidental—and cited general statistics regarding the effect of alcohol on a driver’s abilities. But AUL also added a new ground for denial, focusing on specific limitations in the AD & D policy provision. AUL suggested that benefits were excluded because Richard died during the “commission of an assault or felony” under South Carolina traffic laws but did not elaborate on this assertion.
Johnson subsequently retained counsel, who appealed the denial administratively, arguing that the general AD & D provision “does not state that legal intoxication disqualifies one from being involved in an accident.” J.A. 134. Counsel challenged the relevance of AUL’s statistics, suggesting that they did not show that death was a likely result of driving while intoxicated and that it is far more likely that a drunk driver gets arrested. In fact, the “policy does exclude payment of the seatbelt benefit if there is an accidental death while a person is legally intoxicated;” however, “the policy does not expressly contain any such limitation for death claims (separate and apart from the seatbelt benefit). J.A. 134. AUL then issued its final denial on December 9, 2008, largely repeating its first several letters verbatim.
Johnson then filed this action, contending that AUL wrongfully denied her AD & D benefits. The parties filed cross-motions for summary judgment.2 Defining
[716 F.3d 819]
“accident” by reference to North Carolina law, the district court concluded that Richard’s death was due to an “accident” only if the crash was the “unanticipated and unexpected result of an intentional, voluntary act.” J.A. 555 (internal quotation marks omitted). In light of Richard’s significant level of intoxication while operating his vehicle 15 mph over the speed limit, the court held that his crash “constitutes an anticipated and expected result (and thus not an ‘accident’ within the meaning of § 58–3–30 …).” J.A. 555. Thus, the district court granted AUL’s motion for summary judgment and denied Johnson’s cross-motion.
“In an appeal under ERISA, we review a district court’s decision de novo, employing the same standards governing the district court’s review of the plan administrator’s decision.” Williams v. Metropolitan Life Ins. Co., 609 F.3d 622, 629 (4th Cir.2010). In Firestone Tire & Rubber Co. v. Bruch, the Supreme Court established that courts must review a denial of benefits “under a de novo standard unless the benefit plan gives the administrator or fiduciary discretionary authority to determine eligibility for benefits or to construe the terms of the plan.” 489 U.S. 101, 115, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989). “If such discretionary authority is conferred, the courts’ review is for abuse of discretion”; however, “the default standard of review is de novo,” and “abuse-of-discretion review is appropriate only when discretion is vested in the plan administrator.” Woods v. Prudential Ins. Co. of Am., 528 F.3d 320, 322 (4th Cir.2008). AUL stipulated below that the Plan documents do not grant it discretionary authority to make a benefits-eligibility determination under the policies or construe the terms of the policies. Accordingly, our review of the denial of benefits in this case is de novo, and our job is to make our own independent determination of whether Johnson was entitled to the AD & D benefits. The correctness, not the reasonableness, of AUL’s denial of AD & D benefits is our only concern in this appeal. Cf. Eckelberry v. ReliaStar Life Ins. Co., 469 F.3d 340, 343 (4th Cir.2006) (“Under [the abuse of discretion] standard, we do not search for the best interpretation of a plan or even for one we might independently adopt. Rather, when reviewing a plan administrator’s decision, a court will not disturb any reasonable interpretation.” (internal quotation marks omitted)).
The question presented is whether Richard’s fatal crash qualified as an “accident,” which is not a defined term in these group policies. “ERISA plans are contractual documents which, while regulated, are governed by established principles of contract and trust law.” Haley v. Paul Revere Life Ins. Co., 77 F.3d 84, 88 (4th Cir.1996). Although courts apply federal common law rules of contract interpretation when construing a policy governed by ERISA, see United McGill Corp. v. Stinnett, 154 F.3d 168, 171 (4th Cir.1998); Denzler v. Questech, Inc., 80 F.3d 97, 101 (4th Cir.1996), we look to “principles of state common law to guide our analysis,” Wheeler v. Dynamic Eng’g, Inc., 62 F.3d 634, 638 (4th Cir.1995). Thus, both state law and general contract law principles inform our attempt to divine the meaning of an undefined term in an ERISA plan. See Regents of the Univ. of Mich. v. Employees of Agency Rent–A–Car Hosp. Assoc., 122 F.3d 336, 339 (6th Cir.1997).
A paramount principle of contract law requires us to enforce the terms of an ERISA insurance plan according to
[716 F.3d 820]
“the plan’s plain language in its ordinary sense,” Wheeler, 62 F.3d at 638, that is, according to the “literal and natural meaning” of the Plan’s language, Stinnett, 154 F.3d at 172 (internal quotation marks omitted); see Glocker v. W.R. Grace & Co., 974 F.2d 540, 544 (4th Cir.1992) (“The primary effort should be to ascertain the intent of the parties … by giving language its ordinary meaning….”). Courts should determine “the common and ordinary meaning as a reasonable person in the position of the plan participant would have understood the words.” Cardoza v. United of Omaha Life Ins. Co., 708 F.3d 1196, 1203 (10th Cir.2013) (internal quotation marks omitted). Our inquiry, then, requires us to consider “what a reasonable person in the position of the participant would have understood those terms to mean.” LaAsmar v. Phelps Dodge Corp. Life Acc. Death & Dismem. & Dep. Life Ins. Plan, 605 F.3d 789, 801 (10th Cir.2010).
Moreover, “ERISA plans, like contracts, are to be construed as a whole.” Alexander v. Primerica Holdings, Inc., 967 F.2d 90, 93 (3d Cir.1992). Courts must look at the ERISA plan as a whole and determine the provision’s meaning in the context of the entire agreement. See generallyRestatement (Second) of Contracts § 202(2). And, because contracts are construed as a whole, courts should seek to give effect to every provision in an ERISA plan, avoiding any interpretation that renders a particular provision superfluous or meaningless. See Harris v. Epoch Group, L.C., 357 F.3d 822, 825 (8th Cir.2004) (“[An ERISA plan] should be interpreted as to give meaning to all of its terms-presuming that every provision was intended to accomplish some purpose, and that none are deemed superfluous.” (internal quotation marks omitted)); see generallyRestatement (Second) of Contracts § 203(a) (“[A]n interpretation which gives a reasonable, lawful, and effective meaning to all the terms is preferred to an interpretation which leaves a part unreasonable, unlawful, or of no effect….”).
If application of these primary principles of construction fails to provide clarity and the plan language remains ambiguous, then we are “compelled to apply the rule of contra proferentum and construe the terms strictly in favor of the insured,” Wegner v. Standard Ins. Co., 129 F.3d 814, 818 (5th Cir.1997), and “in accordance with the reasonable expectations of the insured,” Gallagher v. Reliance Standard Life Ins. Co., 305 F.3d 264, 269 (4th Cir.2002) (internal quotation marks omitted); see Glocker, 974 F.2d at 544. “An ambiguity exists where the language of a contract is fairly and reasonably susceptible to either of the constructions asserted by the parties.” Glover v. First Union Nat’l Bank, 109 N.C.App. 451, 428 S.E.2d 206, 209 (1993). Whether an ambiguity in fact remains is ultimately a question of law for the court. See Bicket v. McLean Sec., Inc., 124 N.C.App. 548, 478 S.E.2d 518, 521 (1996).
We begin, therefore, with the plain language of the disputed provision, viewed in the context of the policy as a whole. As noted, the policies define “accidental death” as “death due to an accident, directly and independently of all other causes,” without defining the term “accident.” Citing various dictionary definitions, Johnson suggests that the most natural and common understanding of the term “accident” is an unintentional, unplanned incident that occurs as a result of a careless error. See Merriam–Webster Dictionary Online, http:// www. merriam- webster. com (last visited April 25, 2013) (defining accident as “an unforeseen and unplanned event or circumstance”); The American Heritage Dictionary, http://
[716 F.3d 821]
www. ahdictionary. com (last visited April 25, 2013) (“An unexpected and undesirable event, especially one resulting in damage or harm: an accident on the assembly line; car accidents on icy roads.”). More or less equating “accident” with an absence of intent, Johnson reasons that the average layperson would understand a loss resulting from a vehicular crash to be accidental where the crash or loss was not planned or intentionally caused. Johnson argues that a vehicular fatality, even in the instance of an intoxicated driver, would therefore qualify as an “accidental death” unless there was some evidence to show that the driver planned to crash the vehicle and cause his death.
On the other end of the spectrum, the term “accident” could also be understood to exclude any incident where the consequences of intentional conduct are expected or reasonably foreseeable. See American Heritage Dictionary, supra, (giving alternate definition of accident as “[a]n unforeseen event that is not the result of intention or has no apparent cause”). Standing alone, then, the word “accident” is reasonably susceptible to more than one interpretation. See Olympic Airways v. Husain, 540 U.S. 644, 651 n. 6, 124 S.Ct. 1221, 157 L.Ed.2d 1146 (2004) (“The term ‘accident’ has at least two plausible yet distinct definitions. On the one hand, … ‘accident’ may be defined as an unintended event. On the other hand, … the term ‘accident’ may be defined as an event that is ‘unusual’ or ‘unexpected,’ whether the result of intentional action or not.” (internal citation omitted)).
Of course, we must construe the document as a whole. Indeed, ambiguous language in one portion of an ERISA plan may well be clarified by reference to unambiguous language in another portion of the plan. See Temme v. Bemis Co., 622 F.3d 730, 734–35 (7th Cir.2010). Here, there are a few other portions of each policy that potentially bear on the meaning and scope of “accident.” First, the policies include a “Limitations Clause” specifically applicable to AD & D benefits. This provision expressly excludes the payment of AD & D benefits for losses resulting from a wide spectrum of conduct, from intentional acts like suicide to conduct that could be considered highly likely to produce a loss, such as “participation in a riot or … war” and “motorcycle racing.” Driving-drunk collisions are not expressly and categorically excluded from AD & D coverage, meaning at the very least that there are some circumstances in which AD & D benefits would be paid for injuries to drunk drivers. See Eckelberry, 469 F.3d at 345 (“The simple fact that drunk driving occurred does not mean there was no accident under the policy. If the insurer did not intend to cover any injury to a drunk driver, then drunk driving would have been a specific exclusion listed in the plan.” (emphasis added)).
Second, the Seat Belt Benefit provided by the employee-paid policy has an express “limitation” that specifically excludes payment of this benefit where the insured was legally intoxicated:
This benefit will not be paid if the Person, while operating the Automobile, was legally intoxicated as defined by applicable laws, violating traffic laws, racing, stunt-driving, or engaging in other similar activity during the accident.
In addition to the above limitation, this benefit is subject to the further limitations and provisions of this AD & D section.
J.A. 270 (emphasis added). This provision makes clear that payment of the Seat Belt Benefit is not only subject to the general AD & D limitations but also to a separate and distinct drunk-driving limitation. There would be no reason to include an express limitation excluding payment of
[716 F.3d 822]
the Seat Belt Benefit where the insured was driving while intoxicated unless a drunk-driving collision otherwise qualified as an “accident” in the first place. Indeed, “[t]he purpose of an exclusion is to take something out of the coverage that would otherwise have been included in it.” Columbia Cas. Co. v. Georgia & Florida RailNet, Inc., 542 F.3d 106, 113 (5th Cir.2008). Any other reading would relegate this language to mere surplusage.
While the limitations clauses clearly suggest that losses suffered in a drunk-driving crash can qualify as accidental under the Plan, these clauses do not necessarily clarify the definition or scope of “accident.” Like many other courts to have wrestled with the meaning of the undefined term “accident” as applied to a claim arising from a drunk-driving crash, we find it to be ambiguous. See, e.g., LaAsmar, 605 F.3d at 804 (collecting cases and noting that “[s]urely there can be no question in this case that the [undefined] term ‘accident,’ as used in this Plan and as applied to [an insured’s drunk driving fatality], is ambiguous.”); cf. Wickman v. Northwestern Nat’l Ins. Co., 908 F.2d 1077, 1087 (1st Cir.1990) (“Case law is fairly consistent in defining an accident, using equally ambiguous terms such as undesigned, unintentional, and unexpected.”).
Thus, we “apply the rule of contra proferentum and construe the terms strictly in favor of the insured,” Wegner, 129 F.3d at 818, and “in accordance with the reasonable expectations of the insured,” Gallagher, 305 F.3d at 269. We conclude that a reasonable plan participant in circumstances similar to those before us would easily have understood that this accident was covered. In this case, there is no evidence of intent. The Traffic Collision Report Form submitted by the responding law enforcement officer stated simply that Richard “was traveling too fast for conditions … traveled off the roadway, struck a highway sign, and overturned several times.” J.A. 393. The Certificate of Death issued by the Horry County Coroner also indicated the “Manner of Death” was an “accident” and listed the cause of death as “Internal Injuries” from an “MVA,” i.e., a motor vehicle accident. J.A. 392. See Kovach, 587 F.3d at 333 (concluding that “an ordinary person would characterize [the insured’s] collision at the intersection to be an accident” even though the insured had been intoxicated while riding his motorcycle because “he did not ‘expect’ or ‘intend’ to hit another vehicle”). In light of the general Limitations Clause and the Limitations Clause specifically applicable to the Seat Belt Benefit, a reasonable plan participant would have believed that Richard’s loss was the result of an accident covered by the policies.
“Strictly construing ambiguous terms presents ERISA providers with a clear alternative: draft plans that reasonable people can understand or pay for ambiguity.” Rasenack v. AIG Life Ins. Co., 585 F.3d 1311, 1320 (10th Cir.2009) (internal quotation marks omitted). As the Tenth Circuit aptly stated,
It is not too much to ask of ERISA insurers to set forth explicitly what is and is not an accident covered by their AD & D policy, and to state unambiguously whether death and disability caused by the insured’s drunk driving is an accident and, if not, to include a workable definition of drunkenness and of causation attributed to such drunkenness.
LaAsmar, 605 F.3d at 805.
AUL could have defined “accident” or even imposed a limitation on benefits for any loss resulting from the insured’s driving with a blood alcohol content above the legal limit of the state in which the vehicle is driven. Instead, the policy language simply leaves “accident” undefined and susceptible to more than one interpretation.
[716 F.3d 823]
Reviewing Johnson’s claim de novo, we are constrained to construe the term in her favor and conclude that Richard’s loss was “accidental.”
AUL contends that when an ERISA plan fails to define “accident” or “accidental” with sufficient clarity, circuit precedent requires us to apply the two-step framework embraced by Eckelberry v. ReliaStar Life Insurance Company. See469 F.3d at 343–46.Eckelberry bears some factual similarities to the case before us. Eckelberry was insured under an AD & D policy provided by his employer that afforded him benefits if he died “due to an accident.” Eckelberry died in a single-vehicle crash, and it was later determined that his BAC was 50 percent higher than West Virginia’s legal limit at the time of the accident. The plan administrator denied benefits, concluding that Eckelberry’s death was not unexpected in light of his intoxicated state and, therefore, was not “due to an accident.” Unlike the AUL policy before us, however, the Eckelberry policy actually supplied a definition of “ accident ”—“an unexpected and sudden event which the insured does not foresee.” 469 F.3d at 342 (internal quotation marks omitted). The contract interpretation problem in Eckelberry revolved around the terms “unexpected” and “foresee,” which were not defined. Also, unlike the plan presently before us, the Eckelberry plan reserved discretionary interpretive authority to the insurer/plan administrator, and therefore on appeal the central question was whether the insurer abused its discretion in concluding Eckelberry’s drunk-driving death was not “unexpected” and therefore not accidental. Id. at 343.
In determining whether the drunk-driving collision qualified as “unexpected,” Eckelberry noted that numerous courts have followed the analysis established by the First Circuit in Wickman v. Northwestern National Insurance Co. to determine the meaning of “accident” when an insurance policy fails to provide a definition or provides an ambiguous one:
Because the Plan’s undefined terms and indeed the term “accident” are not always susceptible to easy application, many federal courts have adopted the framework laid out in [Wickman ] to clarify the meaning of “unexpected.” Initially, the court asks whether the insured subjectively expected his actions to result in injury or death…. However, “if the fact-finder … finds the evidence insufficient to accurately determine the insured’s subjective expectation, the fact-finder should then engage in an objective analysis of the insured’s expectations.” This “objective analysis” asks “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.”
Eckelberry, 469 F.3d at 343 (emphasis added) (internal citations omitted); see Wickman, 908 F.2d at 1088.Eckelberry recognized that in applying Wickman, courts have not always held strictly to the “highly likely” threshold but have also used a “reasonable foreseeability” standard. Id. at 344 & nn. 1–2. Ultimately, we concluded that the administrator’s “determination that Eckelberry’s death was not unexpected because he put himself in a position in which he should have known serious injury or death could occur finds considerable support in the record.” Id. at 345 (emphasis added) (internal quotation marks omitted).
For several reasons, Eckelberry neither controls our analysis here nor dictates the result in this appeal. Eckelberry did not establish a per se rule that drunk driving injuries or fatalities can never be accidental,
[716 F.3d 824]
as the opinion itself points out. See id. Cases like this are highly fact-specific, rising and falling on the particular terms of the policy as applied to the unique facts of the incident giving rise to the claim. Unlike the policy presently before us, the policy in Eckelberry actually defined “accident” in terms of foreseeability—there, an “accident” was “an unexpected and sudden event which the insured does not foresee.” Id. at 342 (internal quotation marks omitted). Although the definition failed to define every relevant term, the Eckelberry AD & D policy seemed to import the tort concept of foreseeability. Hence, we observed that “[o]rdinarily, a death that occurs as a result of driving while intoxicated … is not an accident because that result is reasonably foreseeable.” Id. at 345 (internal quotation marks omitted).
The terms of AUL’s policies, by contrast, do not steer us towards a foreseeability analysis of “accident.” We see nothing in the policies to warrant defining “accident” so as to exclude reasonably foreseeable losses:
“Reasonable foreseeability,” besides itself being ambiguous, injects a different spin to the analysis and, depending upon how broadly it is interpreted, could drastically reduce coverage under the AD & D policy since, particularly in hindsight, it could be said many accidents are foreseeable, even reasonably foreseeable, as opposed to unforeseeable.
LaAsmar, 605 F.3d at 809. Unlike Eckelberry, our review is de novo in this appeal, and we construe the meaning of the Plan terms in the first instance and are not limited to considering only the reasonableness of the decision and reasoning of the claims administrator. Accordingly, we are not bound to define “accident” as we did in Eckelberry simply because these cases fall into the same general factual category.
Finally, we must address the lower court’s use of state law to define “accident.” As previously noted, the district court imported the North Carolina insurance statute’s definition of “accident” into the terms of the Plan. This statute applies “to the provisions of all group life, group accident, group health, and group accident and health insurance policies and group annuities … that are issued on or after October 1, 1989.” N.C. Gen.Stat. § 58–3–30(a). Finding an “absence of any showing” in the record that the North Carolina statutory definition of “accident” has been preempted by ERISA, the court concluded that the North Carolina statute controlled the definition of accident. J.A. 541.
North Carolina law dictates that for any such group life or accident policy, “ ‘[a]ccident’ … shall be defined to imply ‘result’ language and shall not include words that establish an accidental means test.” N.C. Gen.Stat. § 58–3–30(b). The district court determined that under the “accidental result” test, the relevant question is whether the crash was an “unanticipated and unexpected result of an intentional, voluntary act.” J.A. 555 (internal quotation marks omitted). In light of Johnson’s significant level of intoxication while operating his vehicle at an excessive rate of speed, the court found that his crash “constitutes an anticipated and expected result (and thus not an ‘accident’ within the meaning of § 58–3–30).” J.A. 555. Johnson believes that the court properly considered the North Carolina statute but reached the wrong result in applying it. AUL replies ERISA preempts the North Carolina statute and thus it was error for the court below to consider it; however, AUL takes the position that the court nonetheless reached the proper result because the statutory definition of “accident” is very similar
[716 F.3d 825]
to the Wickman standard cited in Eckelberry.
Johnson contends that the “Conformity with State Laws” clause contained in the policies reflects the parties’ intent to incorporate and make North Carolina insurance law part of the policies.3 If this is true, Johnson suggests, we would merely be enforcing the agreed-upon terms of the policy by applying North Carolina law and would not actually be presented with a preemption problem.
We are not convinced that the general Conformity–with–State–Laws clause, standing alone, moots the question of preemption. “ERISA plans must always conform to state law, but only state law that is valid and not preempted by ERISA.” Louisiana Health Serv. & Indem. Co. v. Rapides Healthcare Sys., 461 F.3d 529, 533 (5th Cir.2006). However, assuming the North Carolina statute applied here, we would still reach the same conclusion.
Section 58–3–30 of the North Carolina General Statutes dictates that any group life or accident policy “shall” define “accident” so as “to imply ‘result’ language and shall not include words that establish an accidental means test.” N.C. Gen.Stat. § 58–3–30(b). Like many jurisdictions, North Carolina has recognized a distinction between “accidental means” and “accidental result” as those terms are used in insurance policies. See Collins v. Life Ins. Co. of Va., 99 N.C.App. 567, 393 S.E.2d 342, 343 (1990).4 Under the stricter “accidental means” terminology, no coverage is provided where the loss “occurs by reason of an insured’s intentional act” or “is the natural and probable consequence of a voluntary act or course of conduct.” Id. By contrast, a policy that pays benefits based on an “accidental result” standard does not categorically exclude from the definition of “accident” losses resulting from intentional acts; rather, “accidental” under this standard means a loss occurred “fortuitously without intent or design” and was “unexpected, unusual and unforeseen.” Henderson v. Hartford Accident & Indem. Co., 268 N.C. 129, 150 S.E.2d 17, 20 (1966).
Section 58–3–30(b) purports to eliminate for group life and accident policies “issued on or after October 1, 1989” the distinction between “accident” or “accidental result” and “accidental means” and move to a single result-oriented definition. SeeN.C. Gen.Stat. § 58–3–30(a). Although we have found no decisions from the North Carolina Supreme Court or Court of Appeals specifically addressing this statute, subsequent case law involving the meaning of “accident” in the context of individual policies—where the accident/accidental means distinction apparently still exists—illuminates the meaning of “ ‘result’ language” in § 58–3–30(b). In North Carolina Farm Bureau Mutual Insurance Co. v. Stox, 330 N.C. 697, 412 S.E.2d 318, 325 (1992), the North Carolina Supreme Court refused to construe the undefined term “accident” in a homeowner’s policy as requiring the means of the loss to be accidental. See id. (rejecting argument that loss was not accidental because the “injuries resulted from
[716 F.3d 826]
… intentional acts”). The court concluded “that where the term ‘accident’ is not specifically defined in an insurance policy, that term does include injury resulting from an intentional act, if the injury is not intentional or substantially certain to be the result of the intentional act.” Id. (emphasis added). Under this “substantially certain” standard, which is similar to—and perhaps more generous to the insured than—the Wickman “highly likely” standard, the question would be whether a reasonable plan participant would have understood that driving while intoxicated under similar circumstances was substantially certain to result in death or severe injury. In this case, we would answer that question in the negative.
The record contains only general statistical data regarding the extent to which a driver’s ability to operate a vehicle is impaired by alcohol. In denying Johnson’s claim, AUL referenced reports from the American Medical Association and the National Traffic Safety Administration for the widely-accepted common-sense proposition that “blood alcohol concentration is directly correlated with the degree of impairment an individual displays when driving after drinking.” J.A. 173. AUL points to one additional document in the record produced by the National Commission Against Drunk Driving that summarizes various stages of impairment based on BAC. In this case, Richard had a BAC of .289, which, according to the NCADD, would put a typical driver either in the “Confusion” stage, which is marked by symptoms such as increasing muscular incoordination and slurred speech, or in the “Stupor” stage, which is marked by significant lack of coordination or loss of motor functions. J.A. 532. This evidence makes clear that impairment is highly likely for someone with a BAC of .289, but there is no data with respect to drunk driving fatalities in relation to the incidents of drunk driving generally.
The record supports the conclusion that injury or death is a reasonably foreseeable consequence of driving at significant levels of intoxication. But we simply fail to see how this general data demonstrates that driving with a BAC of .289 under these circumstances is substantially certain to result in death or severe injury. There is a significant difference between these standards:
A result is reasonably foreseeable if there are indications which would lead a reasonably prudent man to know that the particular results could follow from his acts. Substantial probability is more than this. The indications must be strong enough to alert a reasonably prudent man not only to the possibility of the results occurring but the indications also must be sufficient to forewarn him that the results are highly likely to occur.
City of Carter Lake v. Aetna Cas. & Sur. Co., 604 F.2d 1052, 1059 n. 4 (8th Cir.1979) (emphasis added). To the extent such data exists, a cursory review actually suggests the opposite—that such losses are not actually substantially certain or highly likely to result from driving under the influence.5 Accordingly, even if we applied North Carolina’s “substantially certain” test—which is almost indistinguishable
[716 F.3d 827]
from the “highly likely” standard employed by Wickman—we would conclude that Richard’s death was accidental and that AUL owes benefits under the AD & D provisions.
AUL contends that N.C. Gen.Stat. § 58–3–30 falls within ERISA’s broad preemptive scope and therefore cannot control. ERISA preempts “any and all State laws” that “relate to any employee benefit plan” covered by ERISA. 29 U.S.C. § 1144(a). Nonetheless, ERISA specifically “saves” from preemption “any law of any state which regulates insurance.” 29 U.S.C. § 1144(b)(2)(A). “[A] state law must be specifically directed toward the insurance industry in order to fall under ERISA’s saving clause; laws of general application that have some bearing on insurers do not qualify.” Kentucky Ass’n of Health Plans, Inc. v. Miller, 538 U.S. 329, 334, 123 S.Ct. 1471, 155 L.Ed.2d 468 (2003) (internal quotation marks omitted). Moreover, “the state law must substantially affect the risk pooling arrangement between the insurer and the insured” to survive preemption. Id. at 342, 123 S.Ct. 1471.
Johnson argues that N.C. Gen.Stat. § 58–3–30 comes within ERISA’s savings clause because, by prescribing the limits of what constitutes an accident, it “alters the scope of permissible bargains between insurers and insureds,” Benefit Recovery, Inc. v. Donelon, 521 F.3d 326, 331 (5th Cir.2008) (internal quotation marks omitted), and therefore “dictates to the insurance compan[ies] the conditions under which [they] must pay for the risk [they have] assumed,” Miller, 538 U.S. at 339 n. 3, 123 S.Ct. 1471. AUL contends that this statute is nothing more than a state law regarding the interpretation of insurance policies, not a state law that “regulates insurance.” 29 U.S.C. § 1144(b)(2)(A). AUL’s position relies on a well-established line of decisions from our sister circuits embracing the notion that “[r]ules of contract interpretation force the insurer to bear the legal risks associated with unclear policy language,” but that “[s]hifting legal risk is … [much different than] transferring or spreading a policyholder’s risk.” Miller v. Monumental Life Ins. Co., 502 F.3d 1245, 1249 (10th Cir.2007) (internal quotation marks omitted); see Hammond v. Fidelity & Guar. Life Ins. Co., 965 F.2d 428, 430 (7th Cir.1992) (“We cannot imagine any rational basis for the proposition that state rules of contract interpretation ‘regulate insurance’ within the meaning of § 1144(b)(2).”).
The preemption question in this case is not easily answered, and we are especially loathe to wade into this issue in light of the fact that it was not well-developed by the parties below. Ultimately, we need not decide this question. If ERISA preempts N.C. Gen.Stat. § 58–3–30(b), as AUL insists, our analysis set forth supra in section III.B remains unchanged as we applied federal common law without regard to the North Carolina statute. But even if § 58–3–30(b) is not preempted, our conclusion would remain unchanged for the reasons explained in section IV.A. See McClure v. Life Ins. Co. of N. Am., 84 F.3d 1129, 1133 (9th Cir.1996) (declining to decide whether ERISA preempted state insurance rule advocated by the insured where the court found in favor of coverage even without applying the state rule).
For the foregoing reasons, we reverse the decision of the court below and remand for entry of judgment awarding the benefits in question to Angela Johnson.
REVERSED AND REMANDED
1. The Pennsylvania Supreme Court sums up the frustration common to courts addressing this issue:
What is an accident? Everyone knows what an accident is until the word comes up in court. Then it becomes a mysterious phenomenon, and, in order to resolve the enigma, witnesses are summoned, experts testify, lawyers argue, treatises are consulted and even when a conclave of twelve world-knowledgeable individuals agree as to whether a certain set of facts made out an accident, the question may not yet be settled and it must be reheard in an appellate court.
Brenneman v. St. Paul Fire & Marine Ins. Co., 411 Pa. 409, 192 A.2d 745, 747 (1963); see Botts v. Hartford Accident & Indem. Co., 284 Or. 95, 585 P.2d 657, 660 (1978) (“There are probably not many words which have caused courts as much trouble as ‘accident’ and ‘accidental.’ ”).
2. The parties agreed to have the magistrate judge enter a dispositive ruling. See28 U.S.C. § 636(c)(1). For simplicity, we refer to the magistrate judge as “the district court” or “the court.”
3. This clause states in its entirety as follows: “CONFORMITY WITH STATE LAWS: Any provision of the policy in conflict with the laws of the state in which it is delivered is amended to conform to the minimum requirements of those laws.” J.A. 294.
4. Although the “distinction between loss due to ‘accidental means’ and loss due to ‘accident’ … was, at one time, generally accepted by most courts,” recently “an increasing number of jurisdictions have rejected the distinction between ‘accidental means’ and ‘accident,’ ‘accidental result,’ ‘accidental injury,’ ‘accidental death,’ and the like, in favor of treating the terms as legally synonymous.” 10 Couch on Insurance 3d § 139.22 (Rev. ed. 2012).
5. For example,
The Centers for Disease Control (CDC) studied alcohol-impaired driving by adults during the period 1993–2002. In 2002, the last year of the study, subjects reported making 159 million trips in which they drove while alcohol-impaired. Given that there were 17,419 deaths from alcohol-related motor vehicle crashes in 2002, drunk driving possibly proved fatal only once in every 9,128 trips.
Douglas R. Richmond, Drunk in the Serbonian Bog: Intoxicated Drivers’ Deaths as Insurance Accidents, 32 Seattle U.L.Rev. 83, 117 (2008) (internal footnotes omitted).