UNITED STATES, Appellant, v. Kristen GILBERT, Defendant, Appellee.
These interlocutory appeals challenge in limine orders excluding three blocs of evidence the government wishes to present during the upcoming capital murder trial of defendant Kristen Gilbert. See 18 U.S.C. § 3731 (permitting an appeal by the government of an order excluding evidence in a criminal case so long as the defendant has not yet been put in jeopardy and the United States attorney certifies that the appeal is not taken for purpose of delay and that the evidence is substantial proof of a fact material in the proceeding). Mindful that in limine evidentiary rulings are generally provisional, see United States v. Lachman, 48 F.3d 586, 590, 594 (1st Cir.1995) (emphasizing that in limine orders excluding evidence usually can be revisited if developments at trial so warrant), and believing that the circumstances surrounding these rulings dictate that we accord broad deference to the district court's views, we affirm in all respects except one.
Because these are interlocutory appeals of pretrial orders, we cull the relevant background not from an established evidentiary basis, but rather from the parties' proffers, the district court's characterizations of these proffers in its several opinions addressing matters relevant to these appeals, and our decision in United States v. Gilbert, 181 F.3d 152 (1st Cir.1999), which affirmed a jury verdict convicting Gilbert of telephoning a bomb threat to the Veteran's Affairs Medical Center (VAMC) during a criminal investigation of the events giving rise to this prosecution. The following recitation is thus only a summary, taken from these sources, of what the parties have told us that the evidence in this case will show.
Gilbert was employed as a nurse at the VAMC in Leeds, Massachusetts, from March 1989 through February 1996. At all relevant times, she was assigned to Ward C and worked the late-afternoon to midnight shift four days per week. Ward C houses chronically ill patients and the hospital's intensive care unit.
Beginning in the summer of 1995, Gilbert befriended James Perrault, a hospital police officer whose work schedule roughly corresponded to hers and whose duties required him to respond to medical emergencies in Ward C. By fall, Gilbert was having an extramarital affair with Perrault. At the time, Gilbert was married to one Glenn Gilbert, with whom she had two children.
In the months following the inception of her relationship with Perrault and continuing until Gilbert terminated her employment at the VAMC in February 1996, Gilbert's colleagues perceived a marked rise in the number of fatalities and emergencies in Ward C. Members of the VAMC staff also perceived a concomitant increase in the hospital's procurement of the drug epinephrine during this time. These perceptions prompted first an investigation by the Office of Healthcare Inspections of the Department of Veterans Affairs (OHCI)-an investigation that concluded that there was no evidence that any VAMC employee had intentionally harmed patients-and eventually a criminal investigation that led to Gilbert's arrest.
During the course of the criminal investigation, and at times Gilbert knew Perrault would be answering the telephone, Gilbert made a series of anonymous, threatening calls to the VAMC. In one of these calls, Gilbert made the false bomb threat for which she stands convicted. See generally Gilbert, 181 F.3d 152. Gilbert also directly harassed and threatened Perrault, who had ended their relationship and agreed to cooperate with the authorities, and expressed anger towards certain co-workers who cooperated with the criminal investigation but would not speak with her own private investigator. In addition, Gilbert once attempted to obstruct the criminal investigation by temporarily blocking Perrault's car in his driveway and pleading with Perrault not to keep an appointment with criminal investigators looking into goings-on at the VAMC. After Perrault refused her request, Gilbert followed him and then vandalized his car while he was being interviewed.
The most recent indictment in this case alleges that, in the six-month period following the commencement of her relationship with Perrault, Gilbert killed four Ward C patients, and attempted to kill three others, by intravenously poisoning them with epinephrine (and insulin, in the case of one diabetic patient).1 Epinephrine, a clear, odorless liquid, is medically indicated for resuscitating patients in a state of cardiac arrest or anaphylactic shock, but causes a potentially fatal rapid or irregular heartbeat when administered in excessive doses. The four deaths and three near-deaths all involved some sort of cardiac arrest on the part of the victim.
According to the government, Gilbert committed these crimes using the following modus operandi. She would obtain epinephrine from Ward C's medicine cabinet, enter the patient's hospital room after all other medical personnel had left, and inject the patient with a fatal dose of epinephrine under the pretense of flushing his intravenous line with a saline solution, an unusual and potentially dangerous practice. The government's theory is that, by thus injecting non-prescribed medication into these patients, all of whom suffered from differing degrees of cardiac troubles, Gilbert was able to generate medical emergencies, or “codes,” in which she plausibly could claim to be responding to naturally occurring cardiac emergencies. Gilbert orchestrated these codes, in the government's view, because they brought her into contact with Perrault, and because they provided her with the attention and excitement that she craved. The government's principal evidence in support of its theory consists of two admissions Gilbert made to Perrault and toxicological evidence garnered from those Gilbert allegedly murdered.
Although no jurisdiction has indicted her for it, the government also claims that, in November 1995, Gilbert attempted to murder her husband, Glenn, by poisoning him with potassium. On November 5, 1995, Glenn Gilbert was hospitalized for gastroenteritis (a condition the government alleges was induced by Gilbert's introduction of low doses of diuretics into her husband's food over the preceding several weeks) and diagnosed with having unusually low levels of potassium and glucose in his blood. This combination of factors caused his heart to beat irregularly, resulting in a cardiac arrythmia. Following his discharge, Gilbert complained to others about the fact that her husband had been released without a follow-up potassium-level blood test. The government alleges that these complaints set the stage for the attempted murder.
According to the government, Gilbert came home from work shortly after the aforementioned hospitalization with two syringes, one of which was filled with a clear, odorless liquid, and told her husband that she wanted to take a blood sample back to the VAMC for additional potassium-level testing. Gilbert then informed her husband that she first would have to inject saline into his arm to flush the vein. As he was receiving the injection, Glenn Gilbert's arms and chest became numb. He tried to pull away, but Gilbert pinned him against the wall with her hip and continued the injection. Glenn Gilbert briefly lost consciousness, but came to and survived without any outside medical intervention. Gilbert allegedly explained away the incident by informing her husband that he had fainted at the sight of the needle. The government contends that Gilbert actually injected her husband with potassium in hopes of inducing a fatal cardiac arrest. Glenn Gilbert did not report this incident to the authorities until months after its occurrence, at a time when he and his wife were involved in a bitter divorce and child custody dispute.
For her part, Gilbert denies any fault for the deaths and other cardiac incidents charged in the indictment. She claims that these deaths and all but one of the near-deaths resulted from natural causes precipitated by preexisting medical conditions suffered by the alleged victims. With respect to one near-death allegedly brought on by insulin poisoning, Gilbert contends that negligence on the part of VAMC physicians caused her unwittingly to administer, on their orders, an excessive dosage. Gilbert apparently will introduce expert medical testimony at trial to substantiate her theories and refute the government's. As to the alleged attempted murder of her husband, Gilbert takes the position that she was, in fact, merely attempting to draw blood from his arm, and that his loss of consciousness was an insignificant “fainting episode.”
In a number of motions in limine, the government sought to establish prior to trial the admissibility of the following blocs of evidence: (1) evidence pertaining to Gilbert's attempted murder of Glenn Gilbert by means of a method similar to the one allegedly used on the VAMC victims; (2) evidence pertaining to Gilbert's anonymous bomb threat and her direct harassment of Perrault; and (3) evidence pertaining to the perceptions of the nurses in Ward C that, between the fall of 1995 and February 1996, there was a marked increase in the number of cardiac emergencies and deaths when Gilbert was on duty.
The government argued that the first two blocs of evidence, although at least conceptually “extrinsic” to the crimes to be tried,2 should be admitted at Gilbert's trial. See Fed.R.Evid. 404(b) (“Evidence of other crimes, wrongs, or acts is not admissible to prove the character of a person in order to show action in conformity therewith. It may, however, be admissible for other purposes, such as ․ intent, ․ knowledge, [or] identity․”). In the government's view, the evidence of Gilbert's attempted murder of her husband is probative on the issues of identity, intent, knowledge, and opportunity. So too is it relevant on the issue of causation. The evidence of the harassment and retaliation, in turn, is probative of Gilbert's consciousness of guilt and intent. Finally, the evidence of a perceived rise in the number of emergencies in Ward C is straightforward “intrinsic” evidence tending to prove causation.
Gilbert countered that the first two blocs of evidence should be excluded because they in fact tend to show only that she is a person of bad character who acted in accordance therewith by committing the crimes charged. See Fed.R.Evid. 404(b). Citing Fed.R.Evid. 403, Gilbert additionally argued, even if relevant, these blocs of evidence nevertheless should be excluded on fairness grounds. See id. (“Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury․”). As to the third bloc, Gilbert contended that it lacks a proper foundation, calls for speculative opinions, and is more prejudicial than probative. For reasons elaborated below, the district court agreed with Gilbert and, in a series of orders, excluded all three blocs of evidence. The government challenges these rulings in these appeals.
In addressing the government's motions to admit the Rule 404(b) evidence identified above, the district court applied the two-part test summarized in United States v. Trenkler, 61 F.3d 45, 52 (1st Cir.1995). First, the court asked whether the proposed evidence “has some ‘special relevance’ independent of its tendency simply to show criminal propensity,” id. (quoting United States v. Guyon, 27 F.3d 723, 728 (1st Cir.1994)), and answered that, at least in large measure, it does not.3 Second, the court assumed arguendo that the evidence in fact did have significant special relevance on an issue other than criminal propensity and then asked whether the evidence should come in under a Rule 403 analysis. See id. The court answered that the evidence should be excluded because its presumed probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury. See id. Rule 403 also was the basis upon which the court excluded the third bloc of evidence, i.e., the so-called “intrinsic” evidence regarding the co-workers' perceptions as to the number of emergencies in Ward C between fall 1995 and February 1996.
Because we discern no basis for concluding that the district court misapprehended the scope of the Rules it was applying, cf. Olsen v. Correiro, 189 F.3d 52, 58 (1st Cir.1999) (“The proper interpretation of the Federal Rules of Evidence is a question of law and is reviewed de novo ․”), we review its application of Rules 404(b) and 403 to the proffered evidence only for an abuse of discretion, see, e.g., United States v. DiSanto, 86 F.3d 1238, 1250 (1st Cir.1996) (reviewing with deference a district court's Rule 404(b) and 403 determinations handed down in response to a motion in limine ).4 As we have explained, an abuse of discretion occurs “when a relevant factor deserving of significant weight is overlooked, or when an improper factor is accorded significant weight, or when the court considers the appropriate mix of factors, but commits a palpable error of judgment in calibrating the decisional scales.” United States v. Roberts, 978 F.2d 17, 21 (1st Cir.1992).
With these principles in mind, we turn to the government's three assignments of error.
A. Exclusion of Evidence Regarding the Attempted Murder of Glenn Gilbert
The government here renews the five Rule 404(b)-based arguments made below in support of admitting the evidence of Gilbert's attempt on her husband's life: (1) the similarities inhering in the modus operandi Gilbert employed in this attempt and that she employed in the crimes charged tends to identify Gilbert as the perpetrator of the crimes charged; (2) the evidence tends to show that Gilbert at all times acted intentionally; (3) the evidence tends to show that Gilbert had the knowledge necessary to commit the crimes charged; (4) the evidence tends to show that Gilbert had the opportunity to commit the crimes charged; and (5) the evidence tends to refute Gilbert's defense that natural causes, or VAMC physician negligence, led to the four deaths and three near-deaths cited in the indictment.
In denying the government's motion, the district court briefly addressed the first three of these five arguments within the context of Rule 404(b)'s “special relevance” prong. Stating that “other acts” evidence is admissible to prove identity under Rule 404(b) when “the shared characteristics of the other act and the charged offense are sufficiently idiosyncratic that a reasonable jury could find it more likely than not that the same person performed them both,” Trenkler, 61 F.3d at 53, the court first explained that there was little to link the attempt on Glenn Gilbert to the crimes charged other than that “the attempt on the life of [Gilbert's] husband [was] by use of an injection.” In the court's view, this “shared characteristic” is insufficiently idiosyncratic “in a hospital setting” to render the attempt on Glenn Gilbert specially relevant on the issue of identity.
The court also quickly dismissed the notion that the attempt on Glenn Gilbert is specially relevant on the issue of intent, reasoning that Gilbert is not defending this case on the theory that she administered fatal and near fatal doses of medicine but did so without the requisite intent. Finally, the court stated in a sentence that the government did not need this evidence to show Gilbert's specialized knowledge, as Gilbert “was concededly a trained nurse.”
The district court then went on to describe at some length what it appears to regard as the heart of the matter: whether the evidence, even if having significant relevance on some non-forbidden avenue of inquiry, should come in under Rule 403. The court stated:
Even if the Government could show some significant degree of “special relevance” of the evidence related to the attempt on Glenn Gilbert, the second prong of the 404(b) analysis would render this evidence inadmissible. The potential unfair prejudice generated by this evidence overwhelmingly outweighs any probative value.
The obvious fact can be simply put. If the court permits the Government to put on evidence of the purported attempted murder of Glenn Gilbert, then the ensuing trial will be as much about the Glenn Gilbert episode as it will be about the charged murders and attempted murders. This is because no reasonable juror, even with the most thorough and forceful instructions, could avoid the temptation to conclude that, if the defendant could bring herself to kill her husband, then she must be “the sort of person” who would kill the victims of the charged crimes. To defend their client, defense counsel would simply have to convince the jury that the attack on Glenn Gilbert never occurred. The force of the improper propensity evidence would be too powerful for any juror to ignore, any attorney to leave unanswered, or any judge to guard against.
In other words, by proving the uncharged conduct, the Government would dramatically increase its chances of obtaining a conviction with regard to the charged conduct-not because the evidence of the uncharged crime would be applied on the issues of identity, motive, or specialized knowledge, but because of the inferences of propensity that would inevitably be drawn by the jury. In this evidentiary environment, the defendant simply could not get a fair trial on the charges that have actually been brought against her.
On appeal, the government directs most of its fire on the district court's reasoning as to whether the evidence of the alleged attempt on Glenn Gilbert's life might be regarded as specially relevant on one or more of the factors argued by the government in support of its in limine motion. The government contends that the court gave short shrift to its identity argument, particularly its argument that there were a number of relatively non-unique but nonetheless identifying similarities between the Glenn Gilbert episode and the charged crimes. See Trenkler, 61 F.3d at 54 (“a number of common features of lesser uniqueness, although insufficient to generate a strong inference of identity if considered separately, may be of significant probative value when considered together”). The government also argues that the court erred in failing to recognize that Gilbert has not stipulated the issue of intent out of the case, and that she appears prepared to defend one of the attempted murder charges by conceding that she gave a harm-causing injection but asserting that she did so without malicious intent. Finally, the government points out that the court did not specifically address each of the government's five Rule 404(b) arguments.
But Rule 404(b)'s “special relevance” prong was not the principal basis of the district court's ruling. The remarks excerpted above make it clear that, even if the district court considered the evidence regarding the attempted murder of Glenn Gilbert as having a significant degree of special relevance as to identity, intent, knowledge, opportunity, or causation, the court still would have excluded the evidence because, in its view, the evidence was overwhelmingly likely to confuse the issues and mislead the jury. And on this point, the government has failed to persuade us that the court committed a palpable error in judgment in calibrating the decisional scales. See Roberts, 978 F.2d at 21. We thus proceed directly to a review of the court's Rule 403 ruling.
In challenging the district court's balancing of probative value against the danger of issue confusion and jury inflammation, the government builds from its argument that the court underestimated the special relevance of the evidence. The government further contends that the court ignored its need for the evidence, see, e.g., United States v. Vest, 842 F.2d 1319, 1327 (1st Cir.1988) (suggesting that the government's need for evidence helps determine its probative value), and that the court underestimated the jury's ability to follow limiting instructions as to how to use the evidence. Finally, the government suggests that the court erred in predicting that this evidence would dominate the proceedings because introduction of the evidence would take between two and four hours of what will be a two- to three-month trial.
Although the non-comprehensive nature of the district court's written remarks on the matter leaves us less than entirely confident in its conclusion that the Glenn Gilbert evidence would be unlikely to have any probative value,5 we are not convinced that relevant matters “deserving of significant weight” have been overlooked in the course of the court's Rule 403 balancing. Roberts, 978 F.2d at 21 (emphasis supplied). Five factors lead us to this conclusion; the first four bear on the overall probative value of the evidence, the fifth relates to its potential for being misused by the jury.
First, regardless whether the evidence of the methodology allegedly used in the attempt on Glenn Gilbert is sufficiently similar to the evidence of the methodology allegedly used in the crimes charged, we regard the inherent reliability of the evidence, and thus its overall probative value, as very much open to question. Although the district court appears to have assumed arguendo that the evidence passes the threshold admissibility hurdle of Rule 104(b), see supra note 5 (summarizing how district courts initially must decide whether to admit conditional facts), this determination strikes us as being far from a foregone conclusion. The evidence, after all, consists almost entirely of allegations made by a hostile witness months after the event in question.
Second, even without regard to the explosive nature of the evidence itself, we think that the court was warranted in having concerns about the extent to which the attempt on Glenn Gilbert would have to be litigated during the course of the trial. This is not a situation where the “extrinsic act” is conceded by all to have taken place. Indeed, Gilbert vehemently denies having attempted to murder her husband. Cf. Trenkler, 61 F.3d at 51-56 (concluding that evidence of a prior bombing conceded by all to have occurred and similar in many respects to a charged bombing had probative value on the issue of identity). It is thus nearly certain that there would be a mini-trial on whether the attempt actually took place. And in the course of the mini-trial, Gilbert likely would find herself obliged to introduce into evidence the tawdry details leading to the bitter divorce and child custody proceedings to explain Glenn Gilbert's hostility and thus attempt to undermine his credibility. We think that the potential for confusion of the issues and for unfair prejudice in such a scenario is manifest.
Third, four of the five issues adduced by the government in support of admitting the Glenn Gilbert evidence do not appear to be much in dispute in this case. As we have observed, Gilbert's defense theory, at least with respect to six of the seven charged events underlying the murder and attempted murder charges, is that no murder or attempted murder took place. In other words, Gilbert does not appear to be preparing to defend this case by acknowledging that crimes took place but arguing that the government has misidentified the perpetrator; or that she killed or nearly killed without intent (except with respect to the diabetic victim); or that she lacked the knowledge or opportunity to commit the crimes. And with respect to the fifth issue to which the government points-causation-we simply do not see how the jury could regard the Glenn Gilbert evidence as specially relevant without drawing a forbidden inference of criminal propensity. So far as we can tell, the presumed fact that Gilbert attempted to poison her husband in a non-VAMC setting tends to prove that the events underlying the charged crimes were the product of malicious human agency (and not naturally occurring) only if one first indulges an assumption that the odds of poisonings having taken place at the VAMC are increased by the presence of a demonstrated poisoner. Yet this is precisely the sort of assumption that Rule 404(b)'s preclusion of character evidence is designed to guard against.6
Fourth, we do not find particularly compelling the government's argument that it has a strong need for this evidence. We recognize that the government bears the burden of proving intent and causation, yet see little likelihood that these issues will prove nettlesome to the government if the jury accepts as truthful and reliable its principal evidence against Gilbert: her two admissions to Perrault and the toxicological evidence tending to establish that the deaths cited in the indictment were the result of deliberate poisoning.
Finally, we can find no fault with the district court's prognostication that the particularly inflammatory nature of the Glenn Gilbert evidence renders it highly susceptible of being misused by the jury in the course of its deliberations. On this point, we have nothing to add to the court's well-stated remarks other than to reiterate that, although Rule 404(b) is a rule of inclusion, see United States v. Cassiere, 4 F.3d 1006, 1021 (1st Cir.1993), “there is always some danger that the jury will use [Rule 404(b) other act] evidence not on the narrow point for which it is offered but rather to infer that the defendant has a propensity towards criminal behavior,” Trenkler, 61 F.3d at 56. Here, where the other act evidence is only marginally reliable, of marginal probative value, and so undeniably explosive, the court's decision to tread a cautious path was well within its wide discretion.
We therefore affirm the exclusion of this evidence. In so doing, we caution that our ruling does not foreclose a revisitation of the matter at trial should future events so warrant. See Lachman, 48 F.3d at 590, 594.
B. Exclusion of Evidence Regarding the Bomb Threat and the Harassment of Perrault
The government also renews its argument that the evidence of the anonymous bomb threat Gilbert phoned in to Perrault at the VAMC, as well as the evidence of her more directly retaliatory conduct towards Perrault, is admissible to prove her intent and consciousness of guilt.7 The district court determined that, although the evidence might plausibly be regarded as having some relevance on these issues, see supra note 3, its probative value would be exceedingly limited because
[the] jury would have to conclude that Gilbert, fearing the [criminal] investigation, harassed Perrault in revenge for his cooperation with law enforcement (and not because he had ended their affair), and phoned in a bomb threat thinking that, somehow, the resulting alarm would throw off the inquiry or intimidate potential witnesses, even though she went to great lengths to conceal the fact that she was the caller. This is an awful lot of squeezing for precious little juice.
The court then went on to exclude the evidence under Rule 403 because (1) it feared that the jury would conclude that one capable of phoning in a bomb threat requiring the evacuation of an intensive care unit would be capable of committing the crimes charged; and (2) Gilbert would be forced to introduce inflammatory and highly prejudicial evidence of her tumultuous and unsavory relationship with Perrault in order to prove that her conduct was motivated by anger towards him for betraying her and not by a desire to intimidate him into silence.
The government contends on appeal that the district court should have permitted the jury to hear the evidence and decide for itself whether pure anger, or the desire of a guilty person to obstruct an investigation, or some combination of the two, motivated the bomb threat and harassment. The government further asserts that both its evidence and Gilbert's likely responsive evidence is not so inflammatory as the court suggested, especially when viewed against the backdrop of the pending charges and the fact that the court has since issued an order acknowledging that some evidence as to Gilbert's affair with Perrault will be coming in anyway. Finally, the government argues that the court erred in weighing the evidence as an undifferentiated mass rather than analyzing it piece-by-piece.
Although we regard this issue as closer than the evidence of the alleged attempt on Glenn Gilbert's life, we think that the district court's ruling is, for the most part, well within its broad discretion. See DiSanto, 86 F.3d at 1250. We are aware that the court already has heard much of the harassment and retaliation evidence during the bomb threat trial, and we thus are particularly inclined to defer to its conclusion that, in the main, the evidence is far more likely attributable to anger at the disloyal Perrault and disloyal co-workers than to a quixotic effort to head off the criminal investigation. Cf. Anderson, 470 U.S. at 574-75, 105 S.Ct. 1504 (emphasizing the trial judge's competence at fact determination). So too, and for the same reasons, are we prepared to defer to the court's characterization of the bomb threat, threats against Perrault's person, and vandalism against Perrault's property, as inflammatory and highly susceptible of misleading the jury.
But the district court's exclusion of the evidence pertaining to Gilbert's blocking in of Perrault's car, and her contemporaneous pleas to Perrault that he not cooperate with criminal investigators, strikes us as being on different footing. For one thing, this evidence seems more readily explainable as the product of an obstructive intent, and thus far more probative of a guilty conscience, than the other evidence in this “bloc.” For another, the evidence is far less inflammatory. Unlike the bomb threat and vandalism evidence, this conduct is criminal only insofar as it constitutes an attempt to obstruct justice, and thus seems far less likely to be misused as pure propensity evidence by the jury. The court's explanations for its ruling excluding the entire bloc of evidence thus fail to explain the exclusion of this particular evidence. The court's apparent failure to separate out and analyze the admissibility of this evidence on its own merits is precisely the type of “error in judgment” against which the abuse of discretion standard protects. See Roberts, 978 F.2d at 21.
We thus affirm the provisional exclusion of the bomb threat evidence. So too do we affirm the provisional exclusion of evidence of direct threats and retaliation against Perrault, with the single exception just noted. As before, our rulings do not foreclose a second look at these matters in light of events at trial. See Lachman, 48 F.3d at 590, 594.
C. Exclusion of Evidence Regarding the Number of Emergencies on Ward C between Fall 1995 and February 1996
Finally, the government renews its argument that testimony from members of Ward C's staff as to a “noticeable increase in the number of codes beginning in the fall of 1995 through mid-February 1996” should be admitted to prove the truth of the matter asserted and, by inference, causation. The district court excluded this evidence under Rule 403 as impressionistic, lacking an objective basis, and thus posing a substantial risk of unfair prejudice. Yet in doing so, the court also acknowledged that, although the evidence will not come in to establish the truth of the matter asserted, it may well prove admissible to explain certain actions on the part of the witnesses, such as why the witness began investigating the absence of ampules of epinephrine from the hospital storeroom.
The government argues on appeal that the district court failed to comprehend that this evidence, in combination with an unexplained loss of epinephrine during the same period of time, is critical to an adequate understanding of the case against Gilbert. The government further asserts that the court failed to appreciate that the evidence is probative of causation. Finally, and to our surprise in light of the acknowledgment of possible future admissibility, the government argues that the evidence is necessary both to help explain certain testimony Gilbert's co-workers will be giving and to help cast light on certain statements Gilbert made to investigators.
Again, we think that the district court's ruling was well within its discretion. As direct evidence of causation, this inherently speculative evidence is at best cumulative of the government's toxicological evidence and Gilbert's two admissions to Perrault. Moreover, and more to the point, the court has acknowledged that the evidence, although not admissible to prove the truth of the matter asserted, may well come in for purposes of explanation. We fail to see why this exception would not apply to the extent of adding helpful context to the testimony of Gilbert's co-workers and to the statements Gilbert gave to investigators.
We thus affirm the exclusion of this evidence with the same caveat as to the provisional nature of in limine rulings set forth at the conclusion of the preceding two subsections. See Lachman, 48 F.3d at 590, 594.
For the reasons stated, we affirm the district court's in limine orders of exclusion except insofar as they pertain to the evidence that Gilbert blocked Perrault's car in with her own and contemporaneously urged him not to cooperate with the investigation into the sad events giving rise to the indictment.
Affirmed as modified.
1. Citing Gilbert's efforts to derail Perrault's cooperation with investigators and the bomb threat, the indictment also charges Gilbert with retaliating against a government witness and obstructing justice. The district court has severed these charges for a separate trial.
2. On appeal, the government takes the position that the retaliation and harassment evidence is really “intrinsic” because it is “not so much evidence of ‘other crimes' in the context of this prosecution as [it] is circumstantial evidence of the defendant's involvement in the murder and attempted murder counts.” (Govt. Brief at 53). Be that as it may, the district court, following circuit precedent and without a claim of error from the government, analyzed this evidence under the Rule 404(b) framework. See United States v. Pina, 844 F.2d 1, 9 (1st Cir.1988) (examining the admissibility of similarly “intrinsic” obstruction-of-justice evidence pursuant to Rule 404(b)). For the sake of simplicity, we thus refer to this bloc of evidence as Rule 404(b) evidence.
3. The district court appears to have concluded that the evidence of the attempt on Glenn Gilbert's life has no such independent special relevance, but that the evidence of Gilbert's bomb threat and harassment of Perrault may have some independent special relevance as to Gilbert's intent and consciousness of guilt. See United States v. Stackpole, 811 F.2d 689, 694 (1st Cir.1987) (acknowledging that Rule 404(b) evidence can be admissible on these grounds).
4. At page 21 of its brief, the government states:[T]o the extent that the district court's in limine rulings are based on pretrial pleadings, and not on its “superior ability to view witnesses and assess the impact of evidence,” United States v. King, 713 F.2d 627, 631 (11th Cir.1983), less deference may be appropriate. As noted by the Supreme Court, this Court is “freer to perform the Rule 403 balancing ab initio when the issue arises in limine,” and the district court is, therefore, not “ ‘better positioned’ than the appellate court to decide the issue in question.” Salve Regina College v. Russell, 499 U.S. 225, 233, 111 S.Ct. 1217, 113 L.Ed.2d 190 (1991).This excerpt contains a serious misstatement on a crucial point of law. Neither Salve Regina nor, so far as we can tell, any other Supreme Court case suggests that appellate courts should perform the relevant balancing test “ab initio ” where a Rule 403 determination is premised on a motion in limine or solely document-based information. Indeed, in discussing how appellate courts should review findings of fact made solely on the basis of physical and/or documentary evidence, the Court has cautioned appellate courts that the case for deference remains strong:The rationale for deference ․ is not limited to the superiority of the trial judge's position to make determinations of credibility. The trial judge's major role is the determination of fact, and with experience in fulfilling that role comes expertise. Duplication of the trial judge's efforts in the court of appeals would very likely contribute only negligibly to the accuracy of fact determination at a huge cost in diversion of judicial resources.Anderson v. City of Bessemer City, 470 U.S. 564, 574-75, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985) (discussing the standard for reviewing factual determinations made under Fed.R.Civ.P. 52(a)). We regard the Court's views on institutional competence and resource allocation in the context of document-based factual findings as having full and obvious application in the context of document-based evidentiary judgment calls. Thus, the fact that the district court here has based its rulings on pretrial documentary proffers, while certainly a relevant background fact, does not vitiate the discretion we accord it in its application of Rules 404(b) and 403.
5. For example, if the evidence of Gilbert's attempt on her husband was such that the jury reasonably could find by a preponderance of the evidence that the attempt in fact took place, see Huddleston v. United States, 485 U.S. 681, 690, 108 S.Ct. 1496, 99 L.Ed.2d 771 (1988) (detailing how, pursuant to Fed.R.Evid. 104(b), courts preliminarily should determine the admissibility of evidence whose relevance turns on the resolution of a conditional fact), the evidence might well be regarded as generally relevant on the issue of intent, which Gilbert has not stipulated out of the case, see United States v. Lowe, 145 F.3d 45, 51 (1st Cir.1998). And indeed, the evidence might well be regarded as particularly probative of whether Gilbert acted with the requisite intent in giving the diabetic patient an excessive dosage of insulin. Cf. Guyon, 27 F.3d at 729 (approving the admission of other loan applications by a defendant to a bank fraud prosecution, where the defendant argued lack of intent to defraud).
6. Because the extrinsic evidence relates to an attack on a different victim in a non-VAMC setting, and Gilbert's connection to the two settings and alleged history of prior similar conduct is thus a necessary link in any inferential chain associating the extrinsic evidence to the crimes charged, this case is distinguishable from the child abuse cases cited by the government in support of its causation argument. In each of those cases, evidence of prior attacks on the victim himself or herself or evidence of prior similar attacks within the same setting made it less likely a priori-i.e., without concomitantly considering the defendant's connection to both the extrinsic evidence and the charged crimes-that the injuries under consideration were accidental. See Estelle v. McGuire, 502 U.S. 62, 68-70, 112 S.Ct. 475, 116 L.Ed.2d 385 (1991); United States v. Boise, 916 F.2d 497, 501-03 (9th Cir.1990); United States v. Leight, 818 F.2d 1297, 1301-04 (7th Cir.1987); United States v. Woods, 484 F.2d 127, 133-36 (4th Cir.1973).
7. The government also argued below that the evidence tends to prove Gilbert's motive, i.e., her desire for attention. The district court rejected this argument as an effort to introduce “propensity evidence dressed up in another guise.” The government presents no developed appellate argument that the court erred in this conclusion; it merely states, in two perfunctory sentences which fail to specify how propensity need not figure in the inferential chain, that the court “erred” because the evidence tends to establish motive. Under the circumstances, we regard this argument as waived. See United States v. Zannino, 895 F.2d 1, 17 (1st Cir.1990).
STAHL, Circuit Judge.