Opinion of the Court by Justice NOBLE.

This case involves an asbestos-related products liability and negligence suit against nineteen corporate defendants. Only two defendants proceeded to trial; the others either settled or won summary judgment motions, and thus had become empty-chair defendants when the case went to trial. At the first trial, the jury returned a verdict in favor of the plaintiff, but apportioned no fault to any of the empty-chair defendants. The trial court then granted a new trial under CR 59.01(f) because the jury's failure to apportion fault to the empty-chair defendants was "manifestly unsupported by the evidence and manifestly a product of jury passion and prejudice." The Court of Appeals reversed. This Court granted discretionary review to determine whether the trial court erred in granting a new trial, which in turn requires us to decide what quantum of proof is necessary to justify apportionment against empty-chair defendants. For the reasons set forth below, the Court of Appeals is reversed.

I. Background

James G. Dexter worked as a pipefitter from 1946 until 1984. As a result, he was exposed to asbestos-containing pipes, gaskets, and insulation at many different jobsites. His exposure involved products made by many different companies. Dexter was also a long-term cigarette smoker.

Dexter was diagnosed with lung cancer. His lungs were found to contain a substantial amount of asbestos fibers. It is undisputed that his lung cancer was caused by both his on-the-job exposure to asbestos and by his long-term smoking habit.

On July 8, 2002, Dexter filed suit in Marshall Circuit Court against nineteen corporate defendants, including the Appellant, CertainTeed Corp.,1 based on products liability (failure to warn) and commonlaw negligence theories. All the defendants either made asbestos-containing products, which were used in Dexter's work, or owned the premises where Dexter worked and was exposed to asbestos-containing products. The defendants later impleaded eleven additional corporate defendants.2

In 2004, prior to trial, Dexter died from his cancer. He was 79 years old. His son, James M. Dexter, was substituted as a party, as executor of the estate.

Eventually, some of the defendants were granted summary judgment, and many others settled before trial. This left only Garlock Sealing Technologies, LLC and CertainTeed as defendants; the rest became empty-chair defendants against whom Garlock and CertainTeed tried to show fault.

The trial began on May 11, 2005. As will be described in more detail below, evidence was introduced showing that Dexter was exposed to the products of both participating and empty-chair defendants, that any exposure to asbestos would have caused his injuries, and that the industry had known prior to his exposure that asbestos could cause lung disease and cancer. Importantly, the proof showed that Dexter worked around CertainTeed's products for a total of only one week out of his almost forty-year career as a pipefitter. On May 25, 2005, the trial concluded and the case was submitted to the jury, with the empty-chair defendants appearing on the verdict forms for purposes of apportionment of fault.

The jury found in favor of the plaintiffs on the products liability claims, but in favor of the defendants on the common-law negligence claims. They returned a verdict awarding Dexter's estate $66,376 for past medical expenses, $5,000,000 for pain and suffering, and $6,750 for funeral expenses, for a total award of $5,073,126. They apportioned 35% fault to Dexter, 35% fault to Garlock, and 30% fault to CertainTeed. The jury allocated no fault to any of the empty-chair defendants. On June 10, 2005, the trial court entered judgment in accordance with the verdict.

Soon thereafter, Garlock and Certain-Teed moved for a new trial, arguing the jury's failure to allocate any fault to any of the empty-chair defendants could not be "sustained by sufficient evidence" under CR 59.01(f). The trial court agreed, and granted a new trial, stating in its order:

The order fails to explain exactly what evidence the trial court believed the jury overlooked or why it believed that the jury's verdict was a result of passion and prejudice.

The case was re-tried in January and February 2006. The second jury found in favor of the plaintiff on both the products liability and common-law negligence claims, and awarded damages of $93,005 for past medical expenses, $1,500,000 for pain and suffering, and $6,744 in funeral expenses, for a total award of $1,599,749 in compensatory damages. This time, however, the jury apportioned some fault against the empty-chair defendants in addition to the participating parties; specifically, the jury assigned 60% fault to Dexter, 2% fault to CertainTeed, 17% to Garlock, and 21% to various empty-chair defendants. (The jury also awarded $100,000 in punitive damages against CertainTeed and $600,000 against Garlock, as well as $15,000 for loss of consortium to Dexter's widow.) On February 22, 2006, the court entered judgment in accordance with the verdict.

On appeal, the plaintiffs complained that the trial court clearly erred in granting the new trial. CertainTeed and Garlock cross-appealed to challenge the second judgment against them. The Court of Appeals resolved the case by holding that "the trial judge's decision ... to set aside the jury's verdict constituted an abuse of discretion and was clearly in error." As a result, the court reinstated the judgment from the first trial. Because of this resolution, the court did not address the other issues raised by the parties.

CertainTeed sought discretionary review,3 which this Court granted to determine whether the reversal was correct, and to explain what quantum of proof is required before it can be proper to apportion fault to empty-chair defendants.

II. Analysis

Before getting into the substance of the issues raised by the parties, a prefatory note about what is not at issue here is necessary. No party has challenged the scope of the trial court's grant of a new trial—only that it was granted. Questions may exist about whether the second trial should have included the negligence claims (as opposed to the strict liability claims), which, in turn, would raise questions about whether the premises owning empty-chair defendants were properly included. Nor has any party in the present appeal challenged the appropriateness of the trial court's approach to the strict liability and negligence claims against CertainTeed and the other defendants, though there may be questions about whether both types of claims can be pursued against any single defendant. These issues, and others, may have been the subject of the appeal and cross-appeal of the second trial but were not addressed because of the Court of Appeals' resolution of reinstating the original judgment. Such issues are not currently before this Court.

The only issue truly here is whether the Court of Appeals erred in reversing the trial court's grant of a new trial to Certain-Teed and Garlock. That issue requires examining certain subsidiary issues—such as the standard of review of a new trial decision and what a defendant must prove to obtain apportionment against an empty-chair defendant—but only to the extent necessary to resolve the primary issue of the appropriateness of the trial court's grant of a new trial. For the most part, this opinion addresses those subsidiary issues as they have been framed by the parties, without further complication by other issues that might be unresolved, either because they have not been raised or have not yet been addressed by the Court of Appeals.

With that said, we now turn to the question presented by the motion for discretionary review: whether the Court of Appeals erred in reversing the trial court's grant of a new trial.

A. Standard of Review for a New Trial Order

Appellate courts must give "a great deal of deference" to a trial court's decision to grant a new trial per CR 59.01. Bayless v. Boyer, 180 S.W.3d 439, 444 (Ky.2005). In fact, the trial court's decision whether to grant a new trial "is presumptively correct." City of Louisville v. Allen, 385 S.W.2d 179, 184 (Ky.1964) (Clay, Comm'r), overruled on other grounds by Nolan v. Spears, 432 S.W.2d 425, 427 (Ky. 1968). Furthermore, an "appellate court is more reluctant to reverse an order granting a new trial than one denying it". Louisville Mem'l Gardens, Inc. v. Com., Dept. of Highways, 586 S.W.2d 716, 717 (Ky.1979) (citing Allen, 385 S.W.2d at 181). This high level of deference by an appellate court is necessary because the decision to grant a new trial "`depends to a great extent upon factors which may not readily appear in the appellate record.'" Id. (quoting Turfway Park Racing Ass'n v. Griffin, 834 S.W.2d 667, 669 (Ky.1992)). Indeed, unlike appellate judges, the trial judge "has heard the witnesses firsthand and observed and viewed their demeanor and . . . has observed the jury throughout the trial." Davis v. Graviss, 672 S.W.2d 928, 932 (Ky.1984).

It is important to remember that the trial court's observations "cannot [be] replicate[d] by reviewing a cold record." Greenleaf v. Garlock, Inc., 174 F.3d 352, 366 (3d Cir.1999). Consequently, an appellate court is "precluded from stepping `into the shoes' of a trial court" in reviewing decisions under CR 59.01. Miller v. Swift, 42 S.W.3d 599, 601 (Ky.2001) (citing Prater v. Arnett, 648 S.W.2d 82 (Ky.App. 1983)).

This Court has previously stated that "[o]nly if the appellate court concludes that the trial court's order was clearly erroneous may it reverse." Turfway Park Racing Ass'n v. Griffin, 834 S.W.2d 667, 669 (Ky.1992). Yet, that decision also states that "a CR 59.01 ruling [i]s `a discretionary function assigned to the trial judge.'" Id. (quoting Davis v. Graviss, 672 S.W.2d 928, 932 (Ky.1984)). The Court of Appeals in this case seized on the discretionary language in many Kentucky cases, ignoring the clear statement in Griffin and other cases that the standard was "clearly erroneous," to hold that the appropriate standard was abuse of discretion, which requires slightly less deference than the clear error standard.

After reviewing Griffin and other cases addressing motions for a new trial, it is apparent that deciding such a motion actually requires multiple decisions by the trial court, consisting of both fact finding and discretionary judgment. Each of these sub-decisions is entitled to a different level of deference and standard of review. Cf. Miller v. Eldridge, 146 S.W.3d 909, 916-22 (Ky.2004) (separating Daubert determination into multiple questions, some of fact and some requiring court's discretion).

First, the trial court must decide whether one of the grounds laid out in CR 59.01 exists. This is a finding of fact and is thus subject to review for clear error. When reviewing a trial court's findings under the clear error standard, the appellant court must determine "whether or not those findings are supported by substantial evidence." Moore v. Asente, 110 S.W.3d 336, 354 (Ky.2003). Though "[substantial evidence is more than a scintilla, and must do more than create a suspicion of the existence of the fact to be established," Am. Rolling Mill Co. v. Pack, 278 Ky. 175, 128 S.W.2d 187, 190 (1939), it does not mean the evidence must be absolutely compelling or lead inescapably to but one conclusion. Rather, substantial evidence is "`[ejvidence that a reasonable mind would accept as adequate to support a conclusion,'" Moore, 110 S.W.3d at 354 (quoting Black's Law Dictionary 580 (7th ed.1999)), or evidence that "`has sufficient probative value to induce conviction in the minds of reasonable men,'" id. (quoting City of Monticello v. Rankin, 521 S.W.2d 79, 80 (Ky.1975)).

Second, upon a proper finding under CR 59.01, the trial court must make the discretionary decision whether to grant the motion. Even if the trial court finds that one of the grounds exists, it is not bound in every case to grant a new trial. The issue could be moot, or the grounds may be off-set by other factors. But we need not imagine every scenario where a court could find that one of the CR 59.01 grounds is technically present but still properly deny a new trial. Suffice it to say, whether to grant the motion for a new trial is always within the trial court's sound discretion and is entitled to a great deal of deference by an appellate court.

Generally speaking, "[t]he test for abuse of discretion is whether the trial judge's decision was arbitrary, unreasonable, unfair, or unsupported by sound legal principles." Commonwealth v. English, 993 S.W.2d 941, 945 (Ky.1999). However, a court can also "`abuse[ ] its discretion when it relies on clearly erroneous findings of fact.'" Walters v. Moore, 121 S.W.3d 210, 215 (Ky.App.2003) (quoting Romstadt v. Allstate Ins. Co., 59 F.3d 608, 615 (6th Cir.1995)). Though this Court has criticized the finding of an abuse of discretion under this latter category as improperly confusing and mixing different standards of review, and urged a strict separation of the standards instead, see Miller v. Eldridge, 146 S.W.3d 909, 915 (Ky.2004), such a strict separation is not always possible. What is certain, however, is that a court errs when it makes a discretionary decision based on a clearly erroneous finding of fact. No confusion arises, so long as the findings of fact are evaluated under the proper standard (i.e., clearly erroneous) and the reviewing court approaches the questions in the right order, id. (suggesting review of findings of fact precede review of discretionary decisions).

Regardless, in most cases, a discretionary decision will be a close one, at least from the appellate perspective, and will not be disturbed by an appellate court unless it "`is firmly convinced that a mistake has been made.'" Walters, 121 S.W.3d at 215 (quoting Romstadt v. Allstate Ins. Co., 59 F.3d 608, 615 (6th Cir. 1995)). An appellate court can only reverse the trial court's decision if it is sure that the decision is incorrect—any doubts must be resolved in favor of the trial court:

Allen, 385 S.W.2d at 184.

The trial court ordered a new trial here because it concluded that "the jury's verdict finding no fault to be apportioned to any [empty-chair] defendant . . . is manifestly unsupported by the evidence and manifestly a product of jury passion and prejudice." It concluded that this satisfied CR 59.01(f)'s requirement that "the verdict is not sustained by sufficient evidence," which is reviewed under the clear error standard. Applying the clear error standard to that decision is not a simple task, in large part because it not clear what quantum of evidence is necessary to require apportioning fault to empty-chair defendants.

B. Evidence Required for Apportionment

Ordinarily, to apportion fault among multiple tortfeasors, the plaintiff must prove each tortfeasor's liability beyond the plaintiffs burden of proof (usually by a preponderance of evidence in a civil case). It would not be fair to make a tortfeasor liable with anything less. This means sufficient evidence of all the elements of the tort must be presented against every tortfeasor to which fault is assigned. If there is insufficient evidence as to a tortfeasor, the jury cannot properly apportion fault against it.

But frequently, as in this case, some of the alleged tortfeasors do not participate in the trial. Numerous reasons exist for such a tortfeasor not participating in the trial, but it is usually because of a prior settlement. The question is how much proof is needed to allocate fault to such an empty-chair defendant. The burden of proof in such a case is effectively shifted, since it is the participating defendant, not the plaintiff, who seeks to show that the empty-chair defendant is responsible. Rather than trying to show the actual liability of the empty-chair defendant, as a plaintiff might do, the participating defendant is merely seeking a reduction of its liability. CertainTeed has argued that a participating defendant simply has to show that it was not the cause of at least part of the plaintiffs injury, regardless of whether the empty-chair defendant could be shown to have legal liability. Cf. 13 David J. Leibson, Comparative Fault and Apportionment Under KRS 411.182—Determination of Fault, Ky. Prac. Tort Law § 10:60 (2010). ("[A tortfeasor] should have to pay no more than the damage it caused, regardless of the status of other entities at fault.").

But such a rule would be unworkable, since it would apply a different burden of proof to empty-chair defendants than to participating defendants, which would lead to inconsistent results. Take, for example, two cases with the same evidence, which establishes that one defendant was both negligent and a cause of the plaintiffs injury, but that the other defendant, though part of the cause of the injury, acted reasonably. In one case, the non-negligent defendant participates at trial; in the other case, the non-negligent defendant settles for a nuisance sum prior to trial. In the case where the non-negligent defendant participates in the trial, all of the fault would have to be assigned to the negligent defendant. But in the other trial, under the rule claimed by CertainTeed, the participating defendant could avoid some of the liability and obtain an apportionment against the empty-chair defendant because the non-negligent defendant was part of the cause of the plaintiffs injury.

Instead, a consistent rule is needed and, indeed, is currently the law in Kentucky. Empty-chair defendants who have settled are to be treated no differently than participating defendants in regard to what must be proved to apportion fault against them. Though the empty-chair defendant will not actually be held liable in the trial, since it is literally not on trial, a participating defendant must still prove liability on the part of the tortfeasor onto whom it seeks to shift some of the blame.

Such a rule has implicitly been a part of Kentucky's law of apportioning fault since the adoption of comparative fault. For example, this Court had held that "[i]f there is an active assertion of a claim against joint tortfeasors, and the evidence is sufficient to submit the issue of liability to each, an apportionment instruction is required whether or not each of the tortfeasors is a party-defendant at the time of trial." Floyd v. Carlisle Const. Co., Inc., 758 S.W.2d 430, 432 (Ky.1988) (emphasis added). Or, as the Court has stated more explicitly in a case in which the instructions allowed the jury to apportion damages to parties without finding them at fault:

Owens Corning Fiberglas Corp. v. Parrish, 58 S.W.3d 467, 482 n. 5 (Ky.2001) (citation omitted); see also Barnes v. Owens-Corning Fiberglas Corp., 201 F.3d 815, 825 (6th Cir.2000) (reading KRS 411.182(2) to require a finding of fault before apportionment is allowed).

C. The Standards Applied

With these standards in mind, we must examine the evidence at the first trial to confirm that the trial court had at least substantial evidence to believe that the jury's verdict was manifestly erroneous. This question requires that there was evidence sufficient to prove fault on the part of at least some of the empty-chair defendants.

Before turning to these questions, however, a review of orders under CR 59.01 is helpful. Reviewing decisions under 59.01 is always difficult. But the review in this case is needlessly difficult because the trial judge failed to explain in his order what evidence he thought the jury overlooked. Although our rules do not explicitly require such an explanation, this Court has previously noted that "a proper order granting a new trial should specify the ground or grounds sustained. Both the litigants and this Court are entitled to know the basis of the ruling." Allen, 385 S.W.2d at 180 n. 1. This is all the more important in cases like this one, where the trial was long and many of the issues were complex and technical.

The reason appellate courts defer to the trial court's decision to grant a new trial is because the decision may depend on factors that do not readily appear in the appellate record, such as witness demeanor and observations of the jury. Bayless, 180 S.W.3d at 444; Greenleaf, 174 F.3d at 366. But that deference does not make such decisions beyond review. Thus, it is crucial that the factors going into the decision appear somewhere in the trial record, or else there may be no record of them at all. Even if the record is complete, the trial judge's evaluation of this evidence, recorded as findings of fact in support of the decision, are just as important, at least from the appellate perspective. Otherwise, an appellate court has to sift through a voluminous record to see what possibly could have been the basis for the ruling, which wastes judicial resources and begs for erroneous reversals.

In this case, for example, the trial court simply found that the jury's verdict in the first trial was "manifestly unsupported by the evidence and manifestly a product of jury passion and prejudice." While these are findings of fact, of a sort, they lack of any discussion of the evidence supporting them, which makes their review difficult. Though CR 59.01 does not currently require them, the better practice is for a trial court to include in its order at least some specific findings regarding the evidence (or lack of evidence) that supports the ultimate decision to grant or deny the motion for a new trial.

That being said, upon thorough review of the extensive record in this case, this Court cannot conclude that the trial judge made a "clearly erroneous" decision or that it abused its discretion in ordering a new trial because of this finding. Bayless, 180 S.W.3d at 444. The trial court ordered a new trial because it concluded that the failure of the jury to apportion fault to the empty-chair defendants was "manifestly unsupported by the evidence and manifestly a product of jury passion and prejudice," which it held met the requirement in CR 59.01(f) that the "verdict is not sustained by sufficient evidence." This finding is subject to clear-error review.

As discussed above, the test for such a finding on appeal is whether there is substantial evidence to support that finding. After reviewing the record, this Court concludes that substantial evidence to support the trial court's finding was introduced at trial.

1. Evidence of Exposure to Asbestos

A plethora of evidence showed that Dexter was exposed to asbestos by many of the empty-chair defendants. Although much of this evidence was general in nature, some of it specifically identified manufacturers of asbestos used at the sites where Dexter worked and the sites themselves where he was exposed to asbestos.

For example, Billy Robertson, a pipefitter who worked at the same union as Dexter, identified several companies who provided asbestos-containing thermal insulation and several worksites at which Dexter would have been exposed to their products.4 Similar testimony was heard from Herman Mitchell, another pipefitter at Dexter's union;5 and Ron Eades, an insulator who did not know Dexter but had worked at the GE plant during the same time period, testified likewise.6 Dexter's verified complaint and sworn interrogatories, which were introduced into evidence, made similar assertions.7 And Dexter's son, who often worked alongside his father, identified certain insulation brands and jobsites.

Dexter himself identified empty-chair defendants in whose employ or from whose product he was exposed to asbestos. For example, Dexter stated in his interrogatories that while working at the GE plant, from 1969 until 1971, "he worked with and/or around asbestos-containing pipe and block insulation on steam lines, asbestos-containing gasket and packing materials, asbestos cloth, asbestos cements . . ., asbestos-containing muds, mastics," and other materials. He also specifically identified "Kay[l]o pipe and block insulation," which is made by empty-chair defendant Owens-Corning, and "Careytemp pipe covering," which is sold by empty-chair defendant Rapid American, as being used at the GE site.

Similarly, Dexter's son testified that he worked in the "immediate vicinity" of his father at the GE plant and that "[i]n a lot of cases" they were both exposed to asbestos dust from insulation. He even recalled a particular instance where this occurred. This testimony about Dexter's exposure was corroborated by Eades, who testified that "if [Dexter] was at Mount Vernon at the General Electric plant working as a pipefitter, he was exposed to asbestos."

Last, a defense expert, Dr. Michael Graham, established that the type of fibers in Dexter's lungs were overwhelmingly of a type that could not have come from either of the participating defendants' products. Specifically, he testified that Dexter's lungs contained primarily amosite fiber, whereas CertainTeed and Garlock products contain only crocidolite fibers. This further established that Dexter was exposed, to a substantial extent, to other asbestos products.

The Court of Appeals opinion largely focuses on this issue. Specifically, the court rejected CertainTeed's claim that the type of fibers found in Dexter's lungs demonstrates his exposure to other products used by or manufactured by the empty-chair defendants. The court noted that CertainTeed's and Garlock's products also contained chrysotile asbestos fibers, which have a very short half-life compared to other types of asbestos fibers. According to one of Dexter's experts, this short half-life means that it was not uncommon to find very little evidence of chrysotile asbestos fibers in the lungs of a worker exposed even to substantial amounts of it. While the Court of Appeals was correct that this did not necessarily demonstrate that Dexter was exposed to very little of CertainTeed's and Garlock's products, it does not also show that Dexter was not exposed to amosite-containing products. The proof of such fibers in his lungs proved at least some exposure to products other than those manufactured by Certain-Teed and Garlock.

The Court of Appeals also thought that the evidence of exposure was not sufficiently specific, explaining "there was a complete lack of proof as to the type, length or depth of asbestos exposure by any other defendant." Essentially, it felt that the evidence of exposure was not particularized enough with respect to the empty-chair defendants.

It is true that the evidence of exposure to the empty-chair defendants was not as specific as it was to the participating defendants. However, this Court disagrees that the evidence was so general or otherwise wanting that the trial court was "clearly erroneous" in concluding that it required some apportionment. The evidence showed Dexter's length of exposure at the GE plant was from 1969 to 1971 and identified several brands of asbestos products used at that site. Dexter's son, who worked in the "immediate vicinity" of his father, recalled specific incidents of exposure at that site; and there was evidence that the types of asbestos fibers in Dexter's lungs could not have possibly come from the participating defendants. Although the evidence of exposure to the participating defendants was certainly more specific, we cannot say that the trial court was clearly wrong to conclude that the evidence of exposure here would be enough to require apportionment.

To some extent, the Court of Appeals' concern about the duration and intensity of Dexter's exposure goes more toward causation than mere opportunity for exposure. But these are slightly different concepts, and as discussed below, there was other, more specific evidence of legal causation.

2. Evidence of Causation

The Court of Appeals rejected CertainTeed's claim that causation of Dexter's illness by the empty-chair defendant had been proven at trial, but it did so in an unusual way. The Court stated that "the testimony Appellees reference consists entirely of a recitation of possible sources of exposure but not how this exposure was related to [Dexter's] illness." This demonstrates some confusion between exposure (i.e., the opportunity for causation) and evidence of causation itself (i.e., that the exposure was the legal cause of the plaintiffs injury). Though evidence of exposure may be related to causation (e.g., testimony about the length and intensity of exposure), it is not exactly what we mean when we require a plaintiff to prove causation. Instead, the primary evidence of causation in this case was from the medical experts, who discussed generally how much exposure to asbestos was necessary to cause injury and specifically whether Dexter's various exposures caused his cancer.

With respect to legal causation, Kentucky has adopted the standard set forth in the Restatement (Second) of Torts § 431 (1965), which provides:

See, e.g., Deutsch v. Shein, 597 S.W.2d 141, 144 (Ky.1980); Claycomb v. Howard, 493 S.W.2d 714, 718 (Ky.1973). No legal rule relieved the empty-chair defendants from liability; thus, the only question here is whether there was evidence showing that exposure to their products was "a substantial factor in bringing about" Dexter's asbestos-related illnesses. There was ample evidence of this element, too.

First and foremost, the plaintiffs own expert, Dr. Arthur Frank, testified that, in his opinion, every single exposure to asbestos would have been the legal cause of Mr. Dexter's illnesses. Specifically, he was asked on direct: "Do you have an opinion. . . with regards to whether each and every exposure to asbestos was a substantial contributing factor in causing these two asbestos-related diseases?" He responded:

(Emphasis added.) Dr. Frank also testified that: "[T]he exposures he would have had to any and all products . . . would have contributed to the overall burden of asbestos which would have contributed to his asbestosis and to his lung cancer, the lung cancer being the cause of his death." (Emphasis added.)

Dexter's treating physician, Dr. William Culbertson, agreed. He was asked whether "each and every exposure to asbestos was a substantial contributing factor to Mr. Dexter's resulting asbestos-related diseases." He responded: "Yes, I believe so." This testimony from Drs. Frank and Culbertson was uncontroverted.

The other expert testimony was consistent with this. Another of the plaintiffs medical experts, Dr. Sam Hammar, testified that whether an empty-chair defendant's asbestos-containing products would have caused Dexter's injury depended on "the intensity, the duration that they were exposed to it, and the number of years or months or whatever they were exposed to it." Dexter argues that this cuts against proving causation for the empty-chair defendants, but that claim would only work if Dr. Hammar had set a minimum cut-off of exposure. Instead, Dr. Hammar, like Dr. Frank, testified that the risk of disease increases as exposure to asbestos increases. Given prior evidence of these factors and the extremely high concentration of asbestos fibers in Dexter's lungs (Dr. Hammar described it as the most he had ever seen in a pipe fitter's lungs), this Court cannot say the trial court was clearly erroneous in finding it to have been unreasonable for the jury to conclude that none of the empty-chair defendants contributed at all to Mr. Dexter's disease.

Last, this Court is further compelled by the plaintiffs' admissions in their opening statement. Their attorney stated:

The evidence bore this out.8

In short, there was uncontroverted evidence that each exposure to asbestos would have been a legal cause of Dexter's injuries. Consequently, the evidence of exposure to the empty-chair defendants' products means that they must have legally caused some portion of Dexter's injuries.

3. Evidence of Knowledge of the Danger of Asbestos and Failure to Warn

Proof of exposure and causation alone, however, are insufficient to show legal fault on the part of a defendant, participating or empty-chair, in a products liability case. To find fault against a defendant, and thus allow apportionment, there must also be proof that the defendant breached a duty.

In a products liability claim, this can be proven in a number of ways, including defective design, manufacturing defects, and a failure to warn. See Clark v. Hauck Mfg. Co., 910 S.W.2d 247, 250 (Ky.1995), overruled on other grounds by Martin v. Ohio County Hosp. Corp., 295 S.W.3d 104 (Ky.2009). Under the latter theory, upon which this case proceeded, liability for a manufacturer follows only if it knew or should have known of the inherent dangerousness of the product and failed to "accompany [] it with the quantum of warning which would be calculated to adequately guard against the inherent danger." Post v. Am. Cleaning Equip. Corp., 437 S.W.2d 516, 520 (Ky.1968). In addition to product manufacturers, premises owners can be held to a duty to warn upon a showing of known dangerousness and a failure to warn. See Brewster v. Colgate-Palmolive Co., 279 S.W.3d 142, 143 (Ky.2009). But premises owners must have actual knowledge of a product's dangerousness for liability to an independent contractor to be shown. Id.

Evidence was introduced at the first trial demonstrating that both the manufacturer and premises owner empty-chair defendants knew of the danger of the asbestos-containing products they manufactured and used. Primarily, the plaintiffs put on evidence showing the development of the scientific link between asbestos and lung disease in cancer. By doing so, they put on some evidence that the whole industry—including the empty-chair defendants—knew of the risks.

They did this by producing various studies that showed the connection between asbestos and lung disease and cancer was known prior to the time of Mr. Dexter's exposure. Dr. Frank laid out the history of the knowledge about the risks of asbestos exposure. In particular, he testified that Drs. E.R.A. Merewether and C.W. Price wrote a paper in 1930 that "talked about how you could get an asbestos-related disease and they simply said any exposure to asbestos will do it." He continued:

Dr. Frank also testified that Dr. Wilhelm Hueper published an article in Scientific American in 1943 concerning "the subject of hazard[s] of asbestos with regard to lung cancer." And he testified that the first detailed epidemiological study linking asbestos and lung cancer was published in 1955:

In short, Dr. Frank's testimony established that the link between asbestos and disease had been known since 1930, and that the link between asbestos and cancer had been known since 1955.

There was also evidence that the asbestos industry, in general, was aware of these scientific studies. It was shown at trial that various industry experts attended the Conference on Biological Effects of Asbestos in 1964. According to a Certain-Teed memo, this conference "was the first of its magnitude" and covered subjects such as "medical aspects such as biological effects, pathological and electron X-ray studies as well as means of prevention and protection to those exposed [to asbestos]." It also "included many papers relating asbestos exposure to mesothelioma cancer." Although the primary evidence of this conference came from a CertainTeed memo, this memo discusses the knowledge within the asbestos industry generally, and thus also shows the knowledge of the empty-chair defendants.

Indeed, this memo establishes that the U.S. asbestos industry knew (or should have known) about the risks of exposure. In particular, it states that the English "Board of Insurance has accepted asbestosis as a cause of lung cancer since 1931" and that "there appears to be an accumulation of evidence of the association of asbestos with cancer." However, despite this evidence, the memo notes that "the U.S. industry, in general, does not want to accept the fact that asbestos is very hazardous and they will accept any doctor's view if he intimated that it is not hazardous." In short, the memo suggests that the U.S. industry knew that exposure to asbestos was risky by 1964, but decided to ignore the mounting evidence.

Notably, Mr. Dexter worked with the CertainTeed products before he worked with the products of at least one empty-chair defendant. Specifically, he worked with CertainTeed pipes only from 1963-1964, but worked at the GE plant after this time, from 1969-1971. Yet, the jury concluded that CertainTeed was liable, meaning that it must have found that Certain-Teed knew (or should have known) of the risks of asbestos exposure by 1964. If CertainTeed knew (or should have known) of the risks by 1964, then certainly GE should have known the same by 1969. The testimony showed that by 1964, when Mr. Dexter worked with CertainTeed pipes, the link to asbestos-related disease had been known for over three decades, the link to cancer had been known for one decade, and a conference among industry representatives about asbestos-related injuries had just concluded.

As noted above, constructive knowledge alone is insufficient to prove the liability of a premises owner like GE.9 But information about the danger of asbestos was so ubiquitous by 1969 that a jury could conclude that GE not only should have known but in fact did know of those dangers. That this is the case is corroborated by Dexter's factual statements in his verified complaint that all of the defendants, including the premises owners like GE, had in their possession information about the dangerousness of asbestos. Similarly, in his answers to interrogatories, Dexter stated that GE violated OSHA regulations regarding asbestos.

Though a plaintiffs statements in his complaint and interrogatories alone might not have sufficed to show GE's liability had it been a participating defendant, they functioned almost as an admission by the plaintiff in this case. In a scenario like this one, where a participating defendant is attempting to prove the legal fault of an empty-chair defendant, a judge may rely on such evidence in deciding a motion for a new trial.

While all this evidence, even as a whole, does not directly demonstrate that each of the empty-chair defendants knew (or should have known) of the danger of asbestos, it was nevertheless sufficient to support the trial court's granting of a new trial. This evidence was, in fact, compelling as to the manufacturers, since it was both circumstantial evidence of knowledge and good evidence of constructive knowledge. It was also circumstantial evidence that the premises owners, who knew the products used by their independent contractors contained asbestos, also knew of the danger; essentially, the fact that asbestos was a serious health hazard was so widely known, it would be unreasonable for a premises owner, like GE, to claim ignorance of the fact. Absolute proof of knowledge is not required to create civil liability.

Evidence was also introduced showing that the empty-chair defendants failed to adequately warn Dexter of the known danger of asbestos. Dexter's co-workers testified that they never saw any warnings on the asbestos products used at various job sites and that no safety equipment, such as respirators, were given to them. At least two of Dexter's co-workers specifically testified as to the lack of warnings. For example, Billy Robertson stated, "There was no warnings put out whatsoever ... There was no warning back then [referring to the 1950s and 1960s]." Moreover, in his own answers to interrogatories about GE, one of the empty-chair defendants, Dexter stated that the company "fail[ed] to provide adequate warnings to [him] regarding the health hazards associated with working with and around asbestos-containing products" and "failed to provide [him] with adequate respiratory protection and other breathing apparatus to protect him from dust inhalation."

4. Substantial Evidence Supported the Trial Court's Ruling

This Court concludes that there was substantial evidence to support the trial court's ruling: the evidence introduced at trial established that Dexter was exposed to asbestos manufactured by or used on the premises of other companies, that these exposures also caused his lung cancer, and that the companies knew or should have known about the dangers and failed to warn Dexter. Much of the evidence that supported the jury's finding of liability against CertainTeed and Garlock (e.g., that of causation) was just as applicable to the empty-chair defendants. In addition to that evidence, other evidence related specifically to the empty-chair defendants was also introduced. In fact, some of that evidence was Dexter's own, or at least his own discovery responses. No doubt, the trial court was especially troubled by the lack of apportionment against empty-chair defendants who, over the course of Dexter's forty-year career, caused him to have much more exposure than did CertainTeed in the week or so in which he was exposed to its products.

Though the Justices of this Court might have reviewed the evidence differently if they were presiding at trial, as members of an appellate court, we cannot approach the evidence in such a way. The Court of Appeals would have required specific and exact evidence of the duration and intensity of exposure from each and every defendant, and that exposure's specific contribution to Dexter's injury. Finding a lack of such specific evidence, the Court of Appeals concluded that the trial court clearly erred.

But this Court cannot say that such precise evidence—presumably consisting of comprehensive and exact descriptions of a plaintiffs exposure from each defendant and expert extrapolation of that exposure to some percentage of causation—is necessary for a trial court to conclude that the jury's verdict was manifestly against the evidence or the product of passion and prejudice. A trial court needs only substantial evidence, not perfect or absolutely compelling evidence, to support its decision.

So long as the trial court's decision was supported by substantial evidence and was not an abuse of discretion, an appellate court must defer to it and affirm the decision. As discussed above, there was ample evidence from which the trial court could conclude that it was unreasonable for the jury to fail to apportion any fault to empty-chair defendants. As such, the trial court's finding of fact that the jury's failure to apportion fault was "manifestly unsupported by the evidence and manifestly a product of jury passion and prejudice" was not clearly erroneous. Additionally, such a finding supports a decision to grant a new trial, and we therefore cannot say that the trial court's decision to do so was an abuse of discretion under the standard articulated above.

Finally, before concluding, this Court must also address the Court of Appeals' general contention at the end of its opinion that apportionment was improper because there was no proof of distinct harms caused by the empty-chair defendants and there was no reasonable basis for determining the contribution of each to Dexter's single harm (his cancer). To support this conclusion, the court cited the Restatement (Second) of Torts § 433A (1965), and its prior "finding" of a "complete lack of proof as to the type, length or depth of asbestos exposure by any other defendant." To some extent, this Court has rejected the approach in Section 433A, stating that when contribution from multiple causes cannot be readily determined, an instruction on comparative fault is appropriate. See Owens Corning Fiberglas Corp. v. Parrish, 58 S.W.3d 467, 479 (Ky. 2001).

Though it is unclear whether this aspect of Parrish is sound, since Section 433A purports to lay out the grounds when a comparative fault instruction would be appropriate, this issue is reserved for review of the second trial because the only issue before this Court is whether trial court erred in granting a new trial, not whether an apportionment instruction is appropriate in the first place. Because of the deference afforded a new-trial decision, an appellate court does not have to review the evidence with such rigor—scrutinizing the evidence with mathematical meticulousness and parsing out the exact percentages of fault it might establish—as suggested by the Court of Appeals. As discussed above, evidence of all the necessary elements of liability as to at least some of the empty-chair defendants was introduced at trial. The trial court observed the parties, counsel, witnesses and jury first hand and was, therefore, in the best position to evaluate what happened. The trial court was not clearly erroneous and did not abuse its discretion in granting a new trial.

III. Conclusion

For the foregoing reasons, the decision of the Court of Appeals is reversed and this case is remanded to that court to consider CertainTeed's cross-appeal of the judgment from the second trial and any other remaining issues.

CUNNINGHAM, SCHRODER, SCOTT and VENTERS, JJ., concur. MINTON, C.J., concurs in result only. ABRAMSON, J., not sitting.