Full opinion text
LUCERO, Circuit Judge. Camille Lente appeals a sentence of 192 months’ imprisonment, well above her advisory Guidelines range of 46 to 57 months, for three counts of involuntary manslaughter and one count of assault resulting in serious bodily injury. The district court relied on several specific factual circumstances in fashioning its sentence, including Lente’s excessive recklessness, her underrepresented criminal history, and her post-conviction conduct. It expressed disagreement with the Guidelines’ treatment of multiple-fatality involuntary manslaughter cases and devoted significant attention to other federal sentences involving similar conduct as part of its duty to avoid unwarranted sentencing disparities. The district court’s determinations were fully supported by the extensive record created by the parties. We conclude that the district court thoroughly considered the relevant sentencing factors and adequately explained the basis of its decision to vary substantially upward, and that the sentence imposed was reasonable. Exercising jurisdiction under 28 U.S.C. § 1291, we affirm. I A This is the third appeal of Lente’s sentence. Following a drunk-driving crash on the Isleta Pueblo in New Mexico, in which three people were killed and another severely injured, Lente pled guilty to three counts of involuntary manslaughter and one count of assault resulting in serious bodily injury. According to the pre-sen-tence investigation report (“PSR”), her advisory Guidelines range was 46 to 57 months’ imprisonment. However, the district court varied upward and imposed a sentence of 216 months. A divided panel of this court vacated Lente’s sentence and remanded to a different judge for resentencing. United States v. Lente, 323 Fed.Appx. 698, 699 (10th Cir.2009) (unpublished) (per curiam) {“Lente I”).Judge Hartz concurred. He concluded that the government breached the plea agreement by stipulating that Lente was “entitled to a guidelines reduction in offense level for acceptance of responsibility’ but subsequently “endorsing] a presentence-report recommendation that the court vary upward because of the defendant’s failure to accept responsibility.” Id. at 699 (Hartz, J., concurring). Judge Holmes also concurred. After noting seven bases identified by the district court for its variance, he stated that although “the district court’s reasons might well justify some upward variance, I simply cannot conclude on this record that they justify the major variance that Ms. Lente received.” Id. at 705-06 (Holmes, J., concurring) (emphasis omitted). Judge McWil-liams would have affirmed the sentence. Id. at 699 (McWilliams, J., dissenting). On remand, both sides submitted additional evidence. The district court also took testimony and heard victim impact statements during a resentencing hearing. In a written order, the district court concluded that a- sentence of 192 months’ imprisonment was appropriate. It cited several factors that contributed to its upward variance: (1) the Guidelines did not adequately reflect Lente’s criminal history; (2) the Guidelines did not sufficiently account for the multiple deaths caused by Lente; and (3) Lente exhibited extreme recklessness, as evidenced by her blood alcohol content (“BAC”) reading after the crash, her failure to ever obtain a driver’s license, and her decision to drive drunk in a non-remote area. It concluded that a within-Guidelines sentence would not adequately reflect the seriousness of the offense, deter Lente from future criminal conduct, or protect the public. After Lente submitted objections, the district court entered an amended order with minor revisions. We reversed and remanded a second time. United States v. Lente, 647 F.3d 1021, 1024 (10th Cir.2011) {“Lente /J”).A1-though we expressly declined to reach the issue of substantive reasonableness, we concluded that the district court had procedurally erred by failing to address Lente’s argument that the sentence would create unwarranted disparities. Id. at 1032-39. We also concluded that the district court committed procedural error by failing to address Lente’s contention that the circumstances prior to the crash mitigated her degree of recklessness, id. at 1036-37, noting that Lente “introduced undisputed evidence about pre-accident developments, including that her mother gave her the keys to her mother’s car and told her to drive,” id. at 1036. We stated, however, that this latter procedural error “may have been harmless.” Id. at 1038. The parties once again supplemented the record with extensive additional evidence. At an evidentiary hearing, the court heard competing expert testimony. It reimposed a sentence of 192 months. B The district court described the factual background of the crash as follows: On December 2, 2005, after consuming excessive amounts of alcohol, Defendant drove her mother’s Chevrolet Suburban on New Mexico State Road 47 within the exterior boundaries of the Isleta Indian Reservation in New Mexico. Anthony Tewahaftewa, a friend of Defendant’s, was riding in the front passenger seat. At approximately 10:40 p.m., Defendant, who was traveling northbound, swerved across the center line of the highway and crossed into the southbound traffic lane. She collided with a Ford Ranger pickup truck driven by Jessica Murillo. Ms. Murillo was driving her 12-year-old brother, Andres Murillo, and her 17-year-old boyfriend, Joshua Romero, back from Albuquerque where the three had gone to see a movie. Anthony Te-wahaftewa, Andres Murillo and Joshua Romero all died immediately on impact. The autopsy reports for the three deceased victims indicate that all three died of severe, multiple blunt force injuries sustained during the crash. Jessica Murillo survived, but sustained fractures to her right femur, right shoulder, and right ankle, and received numerous facial lacerations. In the intervening months, Ms. Murillo underwent many hours of physical therapy just so she could walk again. Defendant suffered two broken ankles and a dislocated hip. Immediately after the accident, both Jessica Murillo and Defendant were transported to the University of New Mexico Hospital. Two hours after the accident, hospital workers took a blood sample from Defendant. The blood sample test results showed a blood alcohol level of 0.21— over two and one-half times the New Mexico legal limit of [0.08]. In addition, the test revealed that Defendant had marijuana present in her system. In subsequent interviews with agents from the Bureau of Indian Affairs, Defendant admitted to drinking between 13 and 19 beers before the crash. Finally, the agents discovered that Defendant had never held a valid New Mexico drivers’ license and that the [Suburban] she was driving was not insured. After discussing its sentencing discretion, identifying the sentencing factors listed in 18 U.S.C. § 3553(a), and accurately noting the undisputed Guidelines range of 46 to 57 months, the court held that a within-Guidelines sentence would be “woefully inadequate.” Incorporating the legal analysis from its prior order, the court stated that it would focus primarily on the procedural errors identified by this court in Lente II. First, the district court ruled that the Guidelines did not adequately account for Lente’s prior criminal history. It characterized Lente as showing “a repeated willingness to abuse alcohol and engage in violent and/or reckless behavior.” However, Lente’s prior convictions did not generate criminal history points under the Guidelines because they occurred in tribal court. Second, the district court concluded that the Guidelines did not adequately account for the multiple fatalities involved in the crash, describing this as perhaps the most important factor in its decision to vary upward. As the court explained in its prior order, the Guidelines assessed only one offense level increase per victim, resulting in approximately five to six month increases in the advisory Guidelines range for each. See U.S.S.G. § 3D1.4 (describing the formula for grouping multiple counts). The court stated that “the Guidelines may be adequate” for involuntary manslaughter cases with a single victim or cases in which the victims “share some minor culpability,” but in cases with “multiple victims who are entirely blameless, and sadly just happened to be in the wrong place at the wrong time,” the Guidelines were insufficient. Third, the court ruled that Lente had been extraordinarily reckless. It cited her BAC of .21, measured approximately two hours after the crash. Although the court heard competing expert testimony as to whether Lente’s BAC at the time of the crash was higher or lower than .21, it declined to rule on that issue, stating instead that its finding of extreme recklessness rested on the fact that Lente chose to drive after consuming enough alcohol to reach a .21 BAC. It also noted that Lente had never obtained a driver’s license, and that she chose to drive drunk on State Road 47, which was classified by the New Mexico Department of Transportation as “unusually significant to its surrounding area” because it carries a high proportion of total urban area travel. In addition to the foregoing factors, which were taken into account — albeit in somewhat lesser detail — during the second sentencing, the district court considered several additional issues. It ruled that the mitigating factors cited by Lente regarding the circumstances leading up to the crash were “insufficient to alter its previous conclusion that Defendant acted with extreme recklessness.” The court also considered a great deal of evidence about Lente’s post-conviction conduct. Despite testimony from a clinical psychologist that Lente was amenable to treatment for the depression and substance abuse problems he identified, the court concluded that Lente’s post-conviction conduct indicated a high likelihood of reoffending. Lente repeatedly abused drugs while in prison and failed to pursue substance abuse programs available to her. The court thus concluded that a variance was necessary to deter future offenses and to protect the public. Finally, the court thoroughly considered the need to avoid unwarranted sentencing disparities, as required by our opinion in Lente II. See 647 F.3d at 1035-36, 1038-39. It distinguished a number of involuntary manslaughter cases cited by Lente because they were based on subsequently altered legal rules or included differing factual circumstances (fewer deaths, lower BACs, less-travelled roads, and victims who elected to ride with a drunk driver). The court announced its sentence of 192 months’ total imprisonment and provided the parties with a written decision. It imposed 72 months for each manslaughter count, with the first two (relating to Romero and Murillo) running consecutive and the third running concurrent, and 120 months on the assault charge, 48 months of which would run consecutive. At Lente’s request, the court allowed her to file written objections in order to preserve issues for appeal before filing its judgment. Lente timely appealed. II We review a sentence “for reasonableness under an abuse-of-discretion standard.” Peugh v. United States, — U.S. -, 133 S.Ct. 2072, 2080, 186 L.Ed.2d 84 (2013). “Reasonableness review is a two-step process comprising a procedural and a substantive component.” United States v. Verdin-Garcia, 516 F.3d 884, 895 (10th Cir.2008). “Procedural reasonableness involves using the proper method to calculate the sentence. Substantive reasonableness involves whether the length of the sentence is reasonable given all the circumstances of the case in light of the factors set forth in 18 U.S.C. § 3553(a).” United States v. Conlan, 500 F.3d 1167, 1169 (10th Cir.2007) (citation omitted). Lente contends the district court committed both procedural and substantive errors. A “A sentence is procedurally unreasonable if the district court incorrectly calculates or fails to calculate the Guidelines sentence, treats the Guidelines as mandatory, fails to consider the § 3553(a) factors, relies on clearly erroneous facts, or inadequately explains the sentence.” United States v. Haley, 529 F.3d 1308, 1311 (10th Cir.2008). A district court must “set forth enough to satisfy the appellate court that [it] has considered the parties’ arguments and has a reasoned basis for exercising [its] own legal decision-making authority.” Rita v. United States, 551 U.S. 338, 356, 127 S.Ct. 2456, 168 L.Ed.2d 203 (2007). Lente frames two specific arguments as procedural in nature. First, she contends that the district court erred in failing to consider the role Tewahaftewa played in causing the crash. Second, she posits that the district court erred by failing to address her argument that multiple-fatality crashes should not be punished substantially more severely than single-death crashes because the Sentencing Commission has rejected such an approach as imposing harsher punishments based merely on “happenstance.” Because the district court carefully considered and rejected these arguments, we conclude that both are better addressed as going to substantive reasonableness. Lente cites our holding in Lente II that a district court commits procedural error when it fails to address “material, non-frivolous arguments made by the defendant.” 647 F.3d at 1035 (quotation omitted). It is true that “failing to consider the § 3553(a) factors and failing to adequately explain the chosen sentence [are] forms of procedural error. On the other hand, a challenge to the sufficiency of the § 3553(a) justifications relied on by the district court implicates the substantive reasonableness of the resulting sentence.” United States v. Balbin-Mesa, 643 F.3d 783, 787 (10th Cir.2011) (quotations and citation omitted). And “the weight the district court places on certain factors is reviewed for substantive unreasonableness.” United States v. Pinson, 542 F.3d 822, 835-36 (10th Cir.2008). Despite Lente’s assertions that the district court did not address her arguments, the order under review discusses the issues she cites in significant detail. Als to Tewahaftewa’s actions just before the crash, the court first recounted the complicated procedural path this issue had charted. The court described the conflicting evidence as to whether Tewahaftewa interfered with Lente’s driving before the crash. But it held that regardless of whether Tewahaftewa grabbed the steering wheel, Lente’s degree of recklessness was not lessened. See Fed.R.Crim.P. 32(i)(3)(B) (a district court must either rule on a disputed factual issue at sentencing “or determine that a ruling is unnecessary either because the matter will not affect sentencing, or because the court will not consider the matter in sentencing”). The court explained that its finding of extreme recklessness “derives not from small actions undertaken immediately before the crash, but from [Lente’s] choice to drive at all after consuming 13 to 19 beers.” It also remarked that Lente had elected to drive Tewahaftewa home despite knowing that he was intoxicated and had been acting in a boisterous manner, and that she was “too intoxicated to be able to respond effectively to unexpected interference with her driving.” Lente claims it was irrational for the district court to discount the potentially causal role played by Tewahaftewa while increasing her sentence dramatically based in part on the number of deaths caused. This is plainly an argument that the district court improperly weighted a factor, which goes to substantive reasonableness, Pinson, 542 F.3d at 835-36, rather than a claim that the district court failed to consider an issue. With respect to Lente’s contention that the district court erred by ignoring her argument regarding the Sentencing Commission’s guidance on multiple-fatality crashes, we concluded in Lente II that this was a substantive issue: Ms. Lente argues that the district court erred in increasing her sentence because three people died. She contends it was procedural error for the court to give inadequate explanation of its policy disagreement with the Guidelines. Although the absence of explanation could constitute procedural error, the court did give a procedurally adequate explanation for why it disagreed with the Guidelines. Ms. Lente’s true complaint is with the substance of that explanation, i.e., whether the district court presented a persuasive reason for disagreeing with the Guidelines. This challenge speaks to substantive reasonableness. 647 F.3d at 1031 (citation omitted). Because the district court incorporated its previous discussion of this issue, and expanded on that discussion, in its order following our second remand, we similarly conclude that this complaint is substantive rather than procedural. We consider both issues in our analysis infra. B “[S]ubstantive reasonableness review broadly looks to whether the district court abused its discretion in weighing permissible § 3553(a) factors in light of the totality of the circumstances.” United States v. Sayad 589 F.3d 1110, 1118 (10th Cir.2009) (quotation omitted). We review the substantive reasonableness of “all sentences — whether inside, just outside, or significantly outside the Guidelines range — under a deferential abuse-of-discretion standard.” Gall v. United States, 552 U.S. 38, 41, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007). Under this standard, we will “deem a sentence unreasonable only if it is arbitrary, capricious, whimsical, or manifestly unreasonable.” Gantt, 679 F.3d at 1249 (quotation omitted). The sentence imposed in this case is 135 months or 236.8% above the high end of Lente’s advisory Guidelines range. A district court “must consider the extent of the deviation [from the Guidelines] and ensure that the justification is sufficiently compelling to support the degree of the variance.” Gall, 552 U.S. at 50, 128 S.Ct. 586. A “major” variance should have “a more significant justification than a minor one.” Id. Nevertheless, the Court has rejected “an appellate rule that requires ‘extraordinary’ circumstances to justify a sentence outside the Guidelines range” and “the use of a rigid mathematical formula that uses the percentage of a departure as the standard for determining the strength of the justifications required for a specific sentence.” Id. at 47, 128 S.Ct. 586. [W]e must give due deference to the district court’s decision that the § 3553(a) factors, on a whole, justify the extent of the variance. An appellate court’s disagreement with the District Judge’s conclusion that consideration of the § 3553(a) factors justified a marked deviation from the Guidelines range is simply not enough to support a holding that the district court abused its discretion. It is not for the Court of Appeals to decide de novo whether the justification for a variance is sufficient or the sentence reasonable, and we must therefore defer not only to a district court’s factual findings but also to its determinations of the weight to be afforded to such findings. United States v. Smart, 518 F.3d 800, 808 (10th Cir.2008) (quotations, citations, and alterations omitted). 1 We begin with the factor that the district court described as perhaps the most important to its decision to vary upward: its conclusion that the Guidelines do not adequately account for the fact that Lente’s offense caused multiple deaths. “[A] district court may in appropriate cases impose a non-Guidelines sentence based on a disagreement with the Commission’s views.” Pepper v. United States, - U.S. -, 131 S.Ct. 1229, 1247, 179 L.Ed.2d 196 (2011). Regarding such policy disagreements, the Supreme Court has instructed that “a district court’s decision to vary from the advisory Guidelines may attract greatest respect when the sentencing judge finds a particular case outside the heartland to which the Commission intends individual Guidelines to apply.” Kimbrough v. United States, 552 U.S. 85, 109, 128 S.Ct. 558, 169 L.Ed.2d 481 (2007) (quotations omitted). “On the other hand, while the Guidelines are no longer binding, closer review may be in order when the sentencing judge varies from the Guidelines based solely on the judge’s view that the Guidelines range fails properly to reflect § 3553(a) considerations even in a mine-run case.” Id. (quotation omitted). The district court stated that an increase of only five to six months for each of the victims, particularly the three who died, constituted a “miserly enhancement.” But in its prior order the court indicated that it did not categorically disagree with the Guidelines’ treatment of involuntary-manslaughter. The court declined in both orders to take a position on cases in which a single death results or those in which multiple victims had some share of culpability, focusing on the fact that three of the victims in this case simply happened to be in the wrong place at the wrong time. This circumstance also featured heavily in the court’s explanation of its decision regarding the partially consecutive and partially concurrent nature of the sentences imposed; it ran the sentence relating to the death of Tewahaftewa concurrently to account for his decision to ride with a drunk driver. Lente argues that the district court’s emphasis on multiple deaths was unreasonable. She contends that the Sentencing Commission, acting in its institutional role, has considered and rejected large sentence increases for multiple-death crashes because, in her words, such results are “based on the happenstance of unintended consequences.” Lente claims that the district court failed to appreciate the unfairness of a substantial increase disconnected from her mens rea and at odds with the Commission’s policy choices. Under the multiple count rules contained in Chapter 3, Part D of the Guidelines, each of Lente’s counts was treated as a separate “group” because they did not “involve substantially the same harm.” See U.S.S.G. § 3D1.2. Each group added a single “unit,” and her four total units resulted in an increase of four offense levels. See § 3D1.4. As the district court noted, this system added approximately five to six months per victim to Lente’s sentencing range. See id. ch. 5, pt. A (Sentencing Table). Although we agree with Lente that the Guidelines attempt to take multiple deaths into account to at least some extent, we are less certain the Sentencing Commission has concluded that such crimes cannot draw significantly harsher punishment because that punishment would be based on “happenstance.” To the contrary, the history of the involuntary manslaughter Guidelines indicates that the Commission, its working groups, and others have repeatedly expressed concern that the sentencing ranges for involuntary manslaughter generally and for multiple-fatality cases in particular are too low. In 1997, a Manslaughter Working Group Report stated that “[ojffenses resulting in multiple deaths ... appear to raise a concern vis-a-vis the Guidelines’ multiple count rules.” Manslaughter Working Group Report to the Commission 11 (Dec. 15, 1997), available at http://www.ussc.gov/ sites/de fault/files/pdf/research/working-groupreports/miscellaneous/19971215_Manslaughter_Report.pdf. A survey of probation officers conducted by the Working Group demonstrated “a consensus that the multiple counts rule was inadequate in dealing with additional deaths.” Id. at 12. The Report suggested increasing the base offense level for reckless involuntary manslaughter, which would “alleviate some of the problems in multiple death cases.” Id. at 14. And it provided several options specifically addressing multiple-fatality cases, including: (1) adding a specific offense characteristic; (2) altering the multiple count rules to increase punishment; (3) imposing consecutive sentences rather than using multiple count rules; (4) allowing for an upward departure; and (5) limiting the credit for acceptance of responsibility in such cases. Id. at 14-15. In 2003, another working group provided input to the Commission with respect to involuntary manslaughter. See Report of the Native American Advisory Group (Nov. 4, 2003), available at http://www. ussc.gov/sites/default/files/pdf/research- and-publications/research-projeets-and-surveys/miscellaneous/20031104_N ative_ American_Advisory_Group_Report.pdf. This report noted that it benefitted from the findings in the Manslaughter Working Group Report, and recommended increasing the base offense level for base offense level for involuntary manslaughter. See Notice of Proposed Amendments to Sentencing Guidelines, Policy Statements, & Commentary & Request for Public Comment, 67 Fed.Reg. 77,532, 77,546 (Dec. -18, 2002). The Native American Advisory Group noted in its report that the Department of Justice recommended increasing the base offense level and adding specific offense characteristics, and the Federal Defenders supported a lesser increase to the base offense level. Report of the Native American Advisory Group at 15. Effective November 1, 2003, the involuntary manslaughter Guidelines were amended to increase the base offense level from 10 to 12 for criminally negligent involuntary manslaughter and from 14 to 18 for reckless involuntary manslaughter. U.S.S.G. app. C, amend. 652. The Commission stated that it promulgated the amendment in response to “a concern that the federal sentencing guidelines do not adequately reflect the seriousness of involuntary manslaughter offenses.” Id. It noted input from the Department of Justice, some members of Congress, and the Native American Advisory Group. Id. The Commission did not indicate why it rejected alternative proposals. See id. However, it did identify as a priority the “continuation of its policy work related to manslaughter, particularly consideration of guideline amendment proposals providing specific offense characteristics in [§] 2A1.4 (Involuntary Manslaughter), and other homicide offenses.” Notice of Final Priorities, 68 Fed.Reg. 52,264, 52,265 (Sept. 2, 2003). And shortly after Amendment 652 went into effect, the Commission requested public comment on whether the base offense level for involuntary manslaughter should be further increased, whether the specific offense characteristics proposed by the Native American Advisory Group should be adopted, and whether to adopt other changes. Notice of Proposed Amendments to Sentencing Guidelines, Policy Statements, & Commentary & Request for Public Comment, 68 Fed.Reg. 75,340, 75,-371 (Dec. 30, 2003). Effective November 1, 2004, Amendment 663 to the Guidelines was enacted. U.S.S.G. app. C, amend. 663. It raised the base offense level for involuntary manslaughter to 22 for cases in which “the offense involved the reckless operation of a means of transportation.” Id. The Commission also added a special instruction to the involuntary manslaughter provision, directing that “[i]f the offense involved the involuntary manslaughter of more than one person, Chapter Three, Part D (Multiple Counts) shall be applied as if the involuntary manslaughter of each person had been contained in a separate count of conviction.” Id. And the Commission added an application note explaining that a “homicide resulting from driving, or similarly dangerous actions, while under the influence of alcohol or drugs ordinarily should be treated as reckless.” Id. The Commission stated that its addition of the new base offense level “eomplete[d] work undertaken in the previous amendment cycle to address disparities between federal and state sentences for vehicular manslaughter.” Id. The purpose of the special instruction was “to ensure an incremental increase in punishment for single count offenses involving multiple victims.” Id. In addition, Congress has twice acted to increase the penalties for involuntary manslaughter. In 1994, Congress increased the statutory maximum for involuntary manslaughter from three to six years. Violent Crime Control & Law Enforcement Act of 1994, Pub.L. No. 103-322, § 320102, 108 Stat. 1796, 2109. And subsequent to the crime at issue in this case, Congress further increased the statutory maximum to eight years. Court Security Improvement Act of 2007, Pub.L. No. 110-177, § 207,121 Stat. 2584, 2588 (2008). Data collected by the Sentencing Commission also demonstrates that district courts have used their discretion to vary above the involuntary manslaughter Guidelines range much more frequently than for federal defendants generally, and have often imposed substantial upward variances. See U.S. Sentencing Comm’n, Sourcebook of Fed. Sentencing Statistics, tbls. N, 28, 32B (2013) (1.5% total upward variant sentences, 7.3% of involuntary manslaughter cases, median manslaughter . variance 52.2% above Guidelines maximum); U.S. Sentencing Comm’n, Sourcebook of Fed. Sentencing Statistics, tbls. N, 28, 32B (2012) (1.3% total upward variant sentences, 16.2% of involuntary manslaughter cases, median manslaughter variance 93.6% above Guidelines maximum); U.S. Sentencing Comm’n, Sourcebook of Fed. Sentencing Statistics, tbls. N, 28, 32B (2011) (1.2% total upward variant sentences, 8.2% of involuntary manslaughter cases, median manslaughter variance 91.7% above Guidelines maximum); U.S. Sentencing Comm’n, Sourcebook of Fed. Sentencing- Statistics, tbls. N, 28, 32B (2010) (1.3% total upward variant sentences, 16.0% of involuntary manslaughter cases, median manslaughter variance 62.2% above Guidelines maximum); U.S. Sentencing Comm’n, Sourcebook of Fed Sentencing Statistics, tbls. N, 28, 32B (2009) (1.3% total upward variant sentences, 10.5% of involuntary manslaughter cases, median manslaughter variance 64.9% above Guidelines maximum); U.S. Sentencing Comm’n, Sourcebook of Fed. Sentencing Statistics, tbls. N, 28, 32B (2008) (0.9% total upward variant sentences, 2.3% of involuntary manslaughter cases, no data on median manslaughter variance); U.S. Sentencing Comm’n, Sourcebook of Fed. Sentencing Statistics, tbls: N, 28, 32B (2007) (0.8% total upward variant sentences, 5.9% of involuntary manslaughter cases, median manslaughter variance 22.6% above Guidelines maximum); U.S. Sentencing Comm’n, Source-book of Fed. Sentencing Statistics, tbls. N, 28, 32B (2006) (0.7% total upward variant sentences, 11.4% of involuntary manslaughter cases, median manslaughter variance 50% above Guidelines maximum). Despite the foregoing, Lente’s argument that multiple deaths are more the result of chance than additional fault carries some force. As we stated in United States v. Rosales-Garcia, 667 F.3d 1348 (10th Cir.2012), we should attempt to avoid “needless and nonsensical aberrant results” under which a defendant “would face more substantially increased punishment solely because of ... happenstance.” Id. at 1354. Nevertheless, we think the district court’s conclusion that the Guidelines do not adequately account for multiple-death crashes is at least reasonable. In Rosales-Garda, the “happenstance” referenced was the order in which a defendant’s state probation was revoked as compared to the initiation of federal charges. Id. There is no plausible argument that one such ordering is more deserving of punishment than another. In sharp contrast, the fact that Lente’s crime caused the deaths of several victims is clearly a relevant consideration in determining an appropriate sentence. “Courts have always taken into consideration the harm done by the defendant in imposing sen-tence_” Payne v. Tennessee, 501 U.S. 808, 825, 111 S.Ct. 2597, 115 L.Ed.2d 720 (1991). That harm comprises part of the “the nature and circumstances of the offense” that district courts must address in fashioning a sentence. 18 U.S.C. § 3553(a)(1). Congress directed the Sentencing Commission to consider “the nature and degree of the harm caused by the offense, including whether it involved ... a person [or] a number of persons.” 28 U.S.C. § 994(c)(3). And the Guidelines include in the definition of relevant conduct “all harm that resulted from the acts and omissions” of the defendant in committing the offense. U.S.S.G. § lB1.3(a)(3). Lente contends, however, that by adopting U.S.S.G. § 2A1.4(b)(l), the Sentencing Commission has determined that multiple deaths warrant only marginally increased punishment. That provision states that “[i]f the offense involved the involuntary manslaughter of more than one person, Chapter Three, Part D (Multiple Counts) shall be applied as if the involuntary manslaughter of each person had been contained in a separate count of conviction.” Id. The Commission adopted this instruction “to ensure an incremental increase in punishment for single count offenses involving multiple victims.” U.S.S.G. app. C, amend. 663. Because Lente was charged with separate counts, this provision did not apply to her case. This instruction is addressed to situations in which a defendant is charged with a single count of involuntary manslaughter despite multiple deaths, reflecting the Commission’s desire “to limit the significance of the formal charging decision.” U.S.S.G. ch. 3, pt. D, introductory cmt. It is true that the Guidelines’ ordinary grouping rules apply to cases involving multiple involuntary manslaughter convictions. But this is the result of the application of a general rule that covers a large number of very different crimes, rather than a deliberate, specific choice by the Sentencing Commission with respect to multiple-fatality drunk-driving incidents. For example, a defendant convicted of several counts of minor assault, burglary, or possessing contraband in prison would face the same incremental increase per count as an involuntary manslaughter defendant who kills several people. See U.S.S.G. § 3D1.2(d) (listing §§ 2A2.3 (Minor Assault), 2B2.1 (Burglary), and 2P1.2 (Possessing Contraband in Prison) as offenses that are separately grouped). Considering the vast array of conduct covered by the grouping rules contained in § 3D1.1 to 4, we do not think the district court’s decision can be characterized as expressing an opinion that “the Guidelines range fails properly to reflect § 3553(a) considerations even in a mine-run case.” Kimbrough, 552 U.S. at 109, 128 S.Ct. 558 (quotation omitted). Instead, the court expressed disagreement with the adequacy of the grouping rules with respect to a vanishingly small subset of offenses those rules cover: multiple-count involuntary manslaughter cases. Only one tenth of one percent of total federal sentences in 2013 included involuntary manslaughter convictions. U.S. Sentencing Comm’n, Sourcebook of Federal Sentencing Statistics, tbl. 17 (2013). And the Commission expressly notes that the grouping “rules have been designed primarily with the more commonly prosecuted federal offenses in mind.” U.S.S.G. ch. 3, pt. D, introductory cmt. Nor does Lente’s crime constitute a “mine-run” involuntary manslaughter case. Only nine percent of involuntary manslaughter cases considered by the Manslaughter Working Group involved multiple deaths. Manslaughter Working Group Report to the Commission at 11. Only three of these ninety-four cases involved three deaths. Id. at 12. According to New Mexico state data from 2005, alcohol-related crashes that resulted in a fatality averaged 1.16 deaths per crash. The district court also relied on the fact that three of the victims — two of whom were killed — were entirely blameless in that they did not choose to ride with an impaired driver. The Manslaughter Working Group found that only one third of involuntary manslaughter victims were strangers to the offender. Manslaughter Working Group Report to the Commission at 10, fig. 14. Rather than rejecting the Sentencing Commission’s conclusions as to a typical case, the district court expressed disagreement with the manner in which the Guidelines treat the confluence of two Guidelines provisions that rarely interact. Given that the vast majority of involuntary manslaughter cases do not result in multiple deaths, that relatively few victims simply happen to be in the wrong place at the wrong time, and that the Commission has acknowledged it crafted the grouping rules with more common crimes in mind, we must reject Lente’s argument that the district court’s decision warrants the “closer review” for categorical policy disagreements referenced in Kimbrough, 552 U.S. at 109, 128 S.Ct. 558 (quotation omitted). Lente’s case involves neither a common application of the grouping rules, nor an ordinary involuntary manslaughter offense. The district court’s decision permissibly focused on the particular and unusual facts of Lente’s case. Recognizing that at least some deference is due to the district court, we conclude that it permissibly varied upward in significant measure based on its conclusion that the Guidelines did not adequately account for the harm caused by Lente’s offense. The court acted within its discretion in concluding that a six-month sentencing increase per victim was not sufficient to “to reflect the seriousness of the offense.” § 3553(a)(2)(A). As described above, the Commission has received input questioning the adequacy of the Guidelines’ recommended sentences for multiple-fatality involuntary manslaughter cases. See Manslaughter Working Group Report to the Commission at 11 (“Offenses resulting in multiple deaths ... appear to raise a concern vis-á-vis the Guidelines’ multiple count rules.”); id. at 12 (noting a consensus among probation officers “that the multiple counts rale was inadequate in dealing with additional deaths”); Report of the Native American Advisory Group at 15-16 (recommending an additional enhancement for multiple-death cases). Moreover, other courts have concluded that multiple-victim involuntary manslaughter cases are insufficiently punished by the Guidelines. In United States v. Fight, 625 F.3d 523 (8th Cir.2010), the court upheld a sentence of 231 months’ imprisonment for three involuntary manslaughter convictions resulting from an impaired-driving crash, well above the defendant’s Guidelines range of 77 to 96 months. Id. at 524. The district court in that case relied heavily on the fact that the defendant caused the deaths of “three innocent individuals.” Id. at 525. And the Eighth Circuit held that the district court permissibly exercised its discretion in concluding that “concurrent sentences within the advisory guideline range did not adequately reflect the seriousness of [the defendant]’s offenses.” Id. at 526. Similarly, in United States v. Shipp, 529 Fed.Appx. 862 (9th Cir.2013) (unpublished), the court affirmed an upwardly variant 108-month sentence for involuntary manslaughter and assault resulting in serious bodily injury. Id. at 862-63. The court rejected botli procedural and substantive reasonableness challenges, concluding that the district court satisfactorily “explained why it believed that the Guidelines did not adequately account for multiple victims and why consecutive sentences were necessary.” Id. at 863. And although we reversed under a now-rejected set of sentencing rules, see Section II.B.6, infra, the district court in United States v. Wolfe, 435 F.3d 1289 (10th Cir.2006), concluded that multiple deaths are “not adequately taken into account by the involuntary manslaughter guidelines.” Id. at 1294. We acknowledge that Lente did not possess an intent to injure, and that the number of victims who are harmed in a single crash such as this one is based at least partly on chance. However, as the Manslaughter Working Group noted, “the culpable mental state in alcohol-related or drug-related vehicular homicide offenses may involve such an egregious state of recklessness that it approaches the mental state for murder.” Manslaughter Working Group Report to the Commission at 11. And as we explain in the next section, the district court correctly concluded that Lente acted with extreme recklessness. We do not think it controversial that a district court should consider the harm caused by a defendant’s unlawful conduct. See Payne, 501 U.S. at 825, 111 S.Ct. 2597. Nor do we need to belabor the point that adding just a few months of additional prison time can reasonably be found insufficient to reflect the seriousness of causing the death of an innocent young person. Despite both the Guidelines’ treatment of multiple deaths and Lente’s “happenstance” argument, we hold the district court did not abuse its discretion in varying upward significantly to account for the harm done to Lente’s victims. We do not conclude that this factor alone would support the sentence imposed. However, this factor permissibly accounts for a substantial portion of the variance. 2 The multiple deaths caused by Lente’s crash were not the only justification supporting the district court’s variance. The court also held that a variance was appropriate because Lente acted with extreme recklessness, and it provided several reasons for this conclusion. We agree that Lente acted with extreme recklessness and that this factor supported an upward variance. As the Court stated in Kimbrough, “a district court’s decision to vary from the advisory Guidelines may attract greatest respect when the sentencing judge finds a particular case outside the heartland to which the Commission intends individual Guidelines to apply.” 552 U.S. at 109, 128 S.Ct. 558 (quotations omitted). “Even though the involuntary manslaughter Guideline already contemplates reckless conduct and the usual ease of drunk driving resulting in death,” district courts nonetheless “may still examine the degree of recklessness in a given case to determine whether this factor exists to such an exceptional level it takes the case outside the ‘heartland’ of usual involuntary manslaughter cases.” United States v. Whiteskunk, 162 F.3d 1244, 1250 (10th Cir.1998). Lente notes that the “heartland” of involuntary manslaughter cases in federal court includes alcohol-related vehicular homicides. See Report of the Native American Advisory Group at 14-15 (almost 75% of cases involving Indian defendants in the data set fit this description). Even among this subset of cases, however, Lente’s conduct was egregious. Lente admitted that she consumed between thirteen and nineteen drinks before the crash. When tested at the hospital following the crash, her BAC was measured at .21. As the district court noted, this figure is more than two-and-a-half times New Mexico’s legal limit of .08. See N.M. Stat. § 66-8-102(C). It is significantly higher than both the average New Mexico DWI figure of .16, and the mode BAC in fatal alcohol-related accidents according to a 2006 study by the National Highway Traffic Safety Administration, which was also .16. Data included with the Manslaughter Working Group Report provides that “each .02% increase in BAC results in a doubling of the risk of being in a fatal crash.” Manslaughter Working Group Report to the Commission app. 4, at 4. And this court has “expressly sanctioned the district court’s reliance on the defendant’s blood-alcohol level ... in considering the degree of recklessness exhibited by the defendant’s conduct.” United States v. Jones, 332 F.3d 1294, 1302 (10th Cir.2003) (quotation omitted); see also United States v. Pettigrew, 468 F.3d 626, 641 (10th Cir.2006) (BAC of “approximately three times the legal limit” and other factors supported departure for excessive recklessness). Lente is correct that this data is not perfect. Her BAC was measured several hours after the crash, but the BAC data upon which the district court relied did not specify the exact timing of BAC measurements. However, we think the district court’s conclusion that Lente behaved in an extremely reckless manner by driving after consuming enough alcohol to reach a .21 BAC is sound. New Mexico state law recognizes that BAC measurements will often be taken sometime after an accident occurs. See N.M. Stat. 66-8-102(0 (state law criminalizes driving if the defendant has a BAC of at least .08 “within three hours of driving the vehicle”). And common sense suggests that Lente’s alcohol intake was excessive by any standard. The district court also 'relied on the fact that Lente elected to drive on a well-travelled thoroughfare. Responding to our statement in Lente II that more specific information about the traffic on the road at issue would have been useful, 647 F.3d at 1037, the court noted that State Road 47 was classified by the New Mexico Department of Transportation as an “Urban Principal Arterial-Other,” which means it is “unusually significant to its surrounding area,” serves a major metropolitan center, and carries a high proportion of total urban area travel. State data showed that at the time of year and time of day of the accident, State Road 47 saw between 500 and 600 vehicles per hour. Additionally, the court cited the fact that Lente had never obtained a driver’s license. The Manslaughter Working Group recommended additional punishment for defendants who lack a valid driver’s license. Manslaughter Working Group Report to the Commission at 14. Lente argues that these latter two factors do not demonstrate extraordinary recklessness. We agree that this information in and of itself would not support a large variance. But when considered together with Lente’s excessive alcohol consumption, which was plainly the focus of the district court’s extreme recklessness finding, the district court permissibly concluded that Lente’s offense conduct fell outside the heartland of involuntary manslaughter cases. The district court elected to vary rather than depart, but we note that the Guidelines provide for an upward departure “in an exceptional case” if a factor already considered by the Guidelines “is present in the offense to a degree substantially in excess of ... that which ordinarily is involved in that kind of offense.” U.S.S.G. § 5K2.0(a)(3). And in Whiteskunk, we held that although the involuntary manslaughter Guidelines account for recklessness, excessive recklessness as evidenced in part by a BAC more than twice the legal limit supported an upward departure. 162 F.3d at 1250, 1252-53. We have stated that a four-level departure for excessive recklessness would be reasonable based in part on a BAC “more than twice the legal limit,” Jones, 332 F.3d at 1302, 1306, and we have affirmed a two-level departure for excessive recklessness when the driver had a BAC approximately three times the legal limit, Pettigrew, 468 F.3d at 641. We acknowledge, however, that these departures were also based on the defendants’ histories of driving while intoxicated. Pettigrew, 468 F.3d at 641; Jones, 332 F.3d at 1302. A two- to four-level enhancement would have added fourteen to thirty months to the top of Lente’s Guidelines range. See U.S.S.G. ch. 5, pt. A (Sentencing Table). We cite these figures for their persuasive implications, cognizant that a “rigid” formula is no longer appropriate in conducting substantive reasonableness review. Gall, 552 U.S. at 47, 128 S.Ct. 586. In light of the district court’s thorough consideration of the circumstances of the offense and a comparison of Lente’s conduct with similar offenses, we uphold the district court’s conclusion that Lente behaved in an extremely reckless manner. We also affirm the district court’s reliance on this factor as part of its rationale for imposing a substantial upward variance. See § 3553(a)(1) (requiring consideration of “the nature and circumstances of the offense”). 3 The district court concluded that Lente’s criminal history was significantly underrepresented in her Guidelines advisory range. Despite numerous convictions, Lente was not assessed any criminal history points because all of her crimes were adjudicated in tribal court. See U.S.S.G. § 4A1.2(i) (“[sentences resulting from tribal court convictions are not counted” in criminal history computations). As with extreme recklessness, the Guidelines expressly allow a district court to depart upward based on an underrepresented criminal history. See id.; § 4A1.3 (“If reliable information indicates that the defendant’s criminal history category substantially under-represents the seriousness of the defendant’s criminal history or the likelihood that the defendant will commit other crimes, an upward departure may be warranted.”). The district court elected to vary rather than to depart, but it noted that Lente’s criminal history category would have been III had her adult convictions been counted, resulting in an advisory range of 57 to 71 months. Moreover, the district court focused on the specific characteristics of Lente’s prior adult convictions. The record supports the district court’s finding that Lente has demonstrated “a repeated willingness to abuse alcohol and engage in violent and/or reckless behavior.” The court referenced its prior order, which described five previous tribal convictions, none of which counted toward Lente’s criminal history score. For four of these, tribal records indicated that Lente was convicted of disorderly conduct after becoming intoxicated and fighting or destroying property. The court expressly declined to rely on three of these convictions, which occurred when Lente was a juvenile. It relied on Lente’s two adult convictions for disorderly conduct and one for assault and battery. The second disorderly conduct offense had not been considered in the court’s previous order, and it “affirmed Defendant’s continuing problems with engaging in reckless behavior as a result of alcohol abuse.” Lente correctly states that her criminal history does not include convictions for drunk driving and that her prior crimes can arguably be described as “petty” in nature. According to the Manslaughter Working Group, approximately one quarter of defendants sentenced in vehicular homicide cases had prior “alcohol/public intoxication records” (excluding driving while intoxicated), and less than ten percent had prior convictions for assault or battery. Manslaughter Working Group Report to the Commission fig. 15. Lente’s troubling tendency to abuse alcohol and violate the law is patent from her criminal record. And her zero criminal history points clearly understated her criminal history. The district court permissibly weighed this factor in its decision to vary upward. 4 Lente’s substance abuse and criminal behavior continued post-conviction. The Supreme Court has held that evidence of post-sentencing conduct may be considered “when a defendant’s sentence has been set aside on appeal and [the] case remanded for resentencing.” Pepper, 131 S.Ct. at 1241. Such evidence “may be highly relevant to several of the § 3553(a) factors that Congress has expressly instructed district courts to consider at sentencing” because it “provides the most up-to-date picture of [a defendant’s history and characteristics” and “also sheds light on the likelihood that [s]he will engage in future criminal conduct, a central factor that district courts must assess when imposing sentence.” Id. at 1242 (quotation omitted). The defendant in Pepper demonstrated positive tendencies between sentencing proceedings. See id. Lente, unfortunately, has not. See id. at 1249 (citing with approval Wasman v. United States, 468 U.S. 559, 572, 104 S.Ct. 3217, 82 L.Ed.2d 424 (1984) (“[A] sentencing authority may justify an increased sentence by affirmatively identifying relevant conduct or events that occurred subsequent to the original sentencing proceeding.”)). Recorded prison calls show that Lente regularly abused narcotics while incarcerated. She also indicated she planned to become intoxicated upon her release and had used alcohol at least once in prison. Lente had two prison disciplinary sanctions for the use of drugs or alcohol, as well as a number of other offenses. She was expelled from a drug education class due to repeated tardiness and refusal to follow the instructor’s direction not to wear headphones during class sessions. And Lente failed to apply to prison drug treatment programs. Shortly before her latest sentencing, a letter containing the opiate drug suboxone was sent’ to Lente by a former inmate. The district court noted that although there was no evidence Lente knew the drug was being sent, it was unlikely it would have been mailed to her absent an expectation that Lente knew how to use it. Lente’s post-conviction conduct firmly supports the district court’s finding that she was likely to reoffend. Despite seven years in prison, and the availability of treatment options, Lente has elected to continue her pattern of substance abuse. In light of the foregoing evidence, the district court’s conclusion that an upwardly variant sentence was necessary to “afford adequate deterrence to criminal conduct” and “protect the public from further crimes of the defendant,” § 3553(a)(2)(B) & (C), was not arbitrary. 5 The district court devoted significant attention to “the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct.” § 3553(a)(6). This provision requires a district court to consider “disparities nationwide among defendants with similar records and Guideline calculations.” Verdin-Garcia, 516 F.3d at 899 (emphasis omitted). It recognizes the “unquestioned” precept “that uniformity remains an important goal of sentencing.” Kimbrough, 552 U.S. at 107, 128 S.Ct. 558. However, “disparate, sentences are allowed where the disparity is explicable by the facts on the record.” United States v. Davis, 437 F.3d 989, 997 (10th Cir.2006) (quotation and citations omitted). “And § 3553(a)(6)’s consideration of unwarranted sentence disparities is but one factor that a district court must balance against the other § 3553(a) factors in arriving at an appropriate sentence.” United States v. Martinez, 610 F.3d 1216, 1228 (10th Cir.2010). Lente cites Wolfe, which contains several factual similarities to her case. The defendant there was convicted on two counts of involuntary manslaughter following a drunk-driving crash that killed two occupants of her vehicle and caused minor injuries to three other passengers. Id. at 1292-93. Wolfe consumed six or seven beers, and her BAC was measured at .13 four hours after the crash. IcL Under the then-mandatory Guidelines, Wolfe faced a sentencing range of 12 to 18 months. Id. at 1292-94. The district court elected to vary upward three levels for “excessive recklessness,” three levels because the defendant endangered the public welfare, and an additional three levels because “the instant offense involved two deaths, and this was not adequately taken into account by the involuntary manslaughter guidelines.” Id. at 1294 (quotations omitted). It imposed a sentence of 41 months. Id. We reversed, concluding that the district court “impermissibly double counted when it departed upward three levels because Wolfe’s conduct resulted in two deaths, when the guideline calculation had already accounted for the two deaths,” id. at 1302, and that- “the district court cannot depart upward based on the significant risk Wolfe’s conduct posed to the public if that determination is based solely on the fact that two people died” because “the guidelines did already take this factor into account,” id. at 1300, 1301. On remand, the district court imposed a sentence of 20 months. There are several significant factual differences between Wolfe and the case at bar: Wolfe consumed far less alcohol and killed one fewer person. Moreover, all of the victims in that case elected to ride with a drunk driver. Further, in sentencing Lente, the district court stressed the fact that Wolfe was decided under a very different set of sentencing rules. The base offense level applicable to Wolfe was the one in place prior to the adoption of Amendments 652 and 663. See Wolfe, 435 F.3d at 1293 (stating that the base offense level for reckless involuntary manslaughter was 14); U.S.S.G. app. C, amend. 652 (raising base offense level to 18); U.S.S.G. app. C, amend. 663 (raising base offense level to 22 if offense involved “the. reckless operation of a means of transportation”). And crucially, Wolfe held that district courts may not depart based on factors already taken into account in a Guidelines range. Wolfe, 435 F.3d .at 1302 (“[T]he fact that multiple deaths occurred, standing alone, cannot support the district court’s decision to depart ... [because] the guideline calculation had already accounted for the two deaths.” (quotation omitted)). But under current precedent “[district courts have broad discretion to consider particular facts in fashioning a sentence under 18 U.S.C. § 3553(a), even when those facts are already accounted for in the advisory guidelines range.” United States v. Yanez-Rodriguez, 555 F.3d 931, 946 (10th Cir.2009). Moreover, we did not discuss the availability of variances in Wolfe, although the case was decided shortly after United States v. Booker, 543 U.S. 220, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005). The district court in Wolfe (sentencing before Booker) expressed concern that multiple deaths were “not adequately taken into account by the involuntary manslaughter guidelines,” 435 F.3d at 1294, but its sole option to rectify that inadequacy was the strictly limited use of upward departures. Following Kimbrough, however, district courts are free to “vary from Guidelines ranges based solely on policy considerations, including disagreements with the Guidelines.” 552 U.S. at 101, 128 S.Ct. 558 (quotation and alteration omitted). And the district court in this case described its disagreement with the manner in which the Guidelines treat multiple-fatality involuntary manslaughter cases as perhaps the most important factor in its decision. Our sentencing scheme “seeks to eliminate not all sentencing disparities, but only ‘unwarranted’ disparities.” United States v. Contreras, 180 F.3d 1204, 1210 (10th Cir.1999). We think that the Supreme-Court-mandated change in governing law suffices as a permissible distinguishing factor from Wolfe. Lente also cites Pettigrew, in which we affirmed a 1'26-month sentence for a defendant with a BAC approximately three times the legal limit who killed one and injured three occupants of another vehicle. 468 F.3d at 630-31, 641. But as with Wolfe, comparing Pettigrew to the present case is difficult in light of the district court’s heavy reliance on its policy disagreement with the Guidelines — a sentencing consideration that was prohibited in our earlier cases. The district court in Pettigrew attempted to impose a lengthy sentence even though it was required to work within the confines of the Guidelines. See 468 F.3d at 633 (departing upward to statutory maximum). -And the defendant in Pettigrew caused only a single death — a highly relevant consideration. We similarly question Lente’s reliance on United States v. New, 491 F.3d 369 (8th Cir.2007). In that case, the defendant was sentenced to 144 months after he killed two passengers while driving with a BAC of .32. Id. at 372. The defendant had a “lengthy and violent criminal record” including two prior convictions for driving under the influence. Id. at 374, 379. Although Lente states that New received a within-Guidelines sentence, the court imposed a statutory maximum sentence that was within-Guidelines only because the court elected to depart upward. Id. at 378-79. The district court’s statement that it sought to impose “a sentence at the highest level that [it] could” indicates it may well have gone higher if it possessed the authority to do so. Id. Many of the other involuntary manslaughter cases cited by Lente involve key factual distinctions, including fewer deaths or lower BACs. See United States v. Spencer, 387 Fed.Appx. 841, 842-43 (10th Cir.2010) (unpublished) (96-month sentence, 39 months above the Guidelines range, for defendant who killed one and severely injured another, admitted he was drunk, and had three prior drunk-driving convictions); United States v. Jim, 377 Fed.A