TRANSCONTINENTAL INSURANCE COMPANY, Petitioner, v. Joyce CRUMP, Respondent, 330 S.W.3d 211

Justice GREEN delivered the opinion of the Court, in which Chief Justice JEFFERSON, Justice HECHT, Justice WAINWRIGHT, Justice MEDINA, and Justice WILLETT joined, and in which Justice JOHNSON and Justice LEHRMANN joined as to Parts I, II, IV and V.

In this workers’ compensation case, we decide three issues: (1) whether expert medical causation testimony from a treating physician relying on a differential diagnosis is reliable and, therefore, legally sufficient evidence to support the jury’s verdict; (2) whether the trial court erred in submitting a jury charge that defined “producing cause” without including a but-for component; and (3) whether an insurance carrier that is unsuccessful on judicial review is entitled to a jury trial on the disputed amount of a claimant’s attorney’s fees under Texas Labor Code § 408.221(c).

We hold that: (1) the treating physician’s opinion was based on a reliable foundation and, therefore, legally sufficient evidence supports the jury’s verdict; (2) the trial court’s omission of the but-for component in the jury charge constitutes reversible error; and (3) an insurance carrier is entitled to have a jury determine the disputed amount of reasonable and necessary attorney’s fees for which it is liable. Accordingly, we reverse the court of appeals’ judgment and remand the case to the trial court for new trial. I.

Background Charles Crump received a kidney transplant in 1975 and began a lifelong regimen of immunosuppressant drug therapy to ensure his body would not reject the new kidney. Crump began working for Frito-Lay in the mid-1980s. In May 2000, while training another employee in the packaging department, Crump struck his right knee on a piece of machinery. The injury caused a contusion (bruise) and a hemato-ma (a collection of blood) at the wound site. He applied for and received workers’ compensation benefits for the work-related injury.

After a series of increasingly serious health complications which required repeated, lengthy hospitalizations, Crump died in January 2001 at age forty-three. His wife, Joyce Crump, applied for workers’ compensation death benefits, alleging that the May 2000 injury was a producing cause of her husband’s death. A contested case hearing officer found that the May 2000 injury resulted in Crump’s death and awarded death benefits. In 2002, the workers’ compensation appeals panel affirmed the hearing officer’s benefits award.

Frito-Lay’s workers’ compensation carrier, Transcontinental Insurance Company, sought judicial review of the administrative award of death benefits. See Tex. Lab. Code §§ 410.301-.308 (providing for, and limiting scope of, judicial review of a death benefits award). As the party appealing the administrative decision, Transcontinental bore the burden of proving its only disputed issue — that the May 2000 injury was not a producing cause of Crump’s death — by a preponderance of the evidence. See id. § 410.303. At trial, Transcontinental offered the testimony of Dr. Judson Hunt.

Hunt reviewed Crump’s medical records and testified that the May 2000 injury was not a producing cause of Crump’s death, and that his death would have occurred without the work-related injury. To rebut Hunt’s opinion, Crump offered Dr. John Daller, one of Crump’s treating physicians, who testified that the May 2000 injury was a producing cause of Crump’s death. The trial court overruled Transcontinental’s objections that Daller’s testimony was not based on a reliable foundation and allowed him to testify.

After hearing the evidence, the jury answered in the affirmative the single question put before it, “Was Charles Crump’s May 9, 2000 injury a producing cause of his death?” Crump submitted the issue of attorney’s fees to the trial court. See id. § 408.221(c) (mandating payment of a claimant’s attorney’s fees by an insurance carrier that unsuccessfully seeks judicial review). Transcontinental had objected that those fees should also be submitted to the jury, rather than the trial court.

The trial court disagreed with Transcontinental and awarded Crump attorney’s fees, as well as fees for time spent pursuing those fees. Transcontinental appealed.

The court of appeals grouped Transcontinental’s issues into three categories: (1) the trial court’s acceptance of the reliability of Crump’s expert’s testimony on causation, as well as the legal and factual sufficiency of that testimony to support the verdict; (2) the trial court’s definition of producing cause in the jury charge; and (3) the determination of Crump’s attorney’s fees by the trial court rather than by the jury, as well as the amount of fees awarded. 274 S.W.3d 86, 96 (Tex.App.-Houston [14th Dist.] 2008). Finding no error in any category, the court of appeals affirmed the trial court’s judgment.

Id. at 90. We granted Transcontinental’s petition for review. 53 Tex. Sup.Ct. J. 15 (Oct. 23, 2009). Finding the court of appeals’ categorization of the issues useful, we address each in turn. II. Legal Sufficiency of Expert Testimony Producing cause was the only issue submitted to the jury at trial. Because this is an appeal of a Workers’ Compensation Commission award of death benefits, Transcontinental acknowledges that it had the burden to prove that the May 2000 injury was not a producing cause of Crump’s death. See Tex.

Lab.Code § 410.303 (“The party appealing the decision [of the appeals panel] on an issue [regarding compensability or eligibility for or the amount of income or death benefits] has the burden of proof by a preponderance of the evidence.”); Morales v. Liberty Mut. Ins. Co., 241 S.W.3d 514, 516-17 (Tex.2007) (discussing the avenues of judicial review). Thus, Transcontinental, the insurance carrier, was the plaintiff at trial; Crump, the claimant, was the defendant.

The trial court asked the jury whether Crump’s injury was a producing cause of his death, but to properly allocate the burden of proof, the court instructed the jury to answer “yes” unless they found by a preponderance of the evidence that the answer should be “no.” In answering “yes,” the jury thus failed to find that Crump’s injury did not cause his death. On appeal, Transcontinental asserts that it conclusively established the lack of causality and is therefore entitled to judgment in its favor as a matter of law.

At trial, Transcontinental’s expert, Hunt, testified that the natural complications of being immunosuppressed for twenty-five years had caused Crump’s death — not the May 2000 injury. Crump’s expert and treating physician, Daller, testified that the wound site of the May 2000 work-related injury became infected, the infection caused Crump’s already-weakened organs to fail, and his organ failure in turn caused his death. Transcontinental objected to the admission of Daller’s testimony on the ground that it was unreliable, but the trial court overruled that objection.

Here, Transcontinental reasserts that Daller’s testimony was unreliable and therefore legally insufficient evidence of causation. Without Daller’s testimony, Transcontinental argues, Hunt’s testimony establishes the lack of causation. Accordingly, we must decide whether Daller’s testimony was reliable and, if so, whether it was some evidence of causation. “An expert witness may testify regarding ‘scientific, technical, or other specialized’ matters if the expert is qualified and if the expert’s opinion is relevant and based on a reliable foundation.” Mack Trucks, Inc. v.

Tamez, 206 S.W.3d 572, 578 (Tex.2006) (citing Tex.R. Evid. 702). In determining whether expert testimony is reliable, a court should consider the factors we set out in E.I. du Pont de Nemours & Co. v. Robinson, 923 S.W.2d 549, 557 (Tex.1995), as well as the expert’s experience, knowledge, and training. See Gammill v. Jack Williams Chevrolet, Inc., 972 S.W.2d 718, 726-27 (Tex.1998) (deeming expert testimony based on the latter considerations unreliable when “there is simply too great an analytical gap between the data and the opinion proffered”) (citing Gen. Elec. Co. v.

Joiner, 522 U.S. 136, 146, 118 S.Ct. 512, 189 L.Ed.2d 508 (1997)); see also Tex.R. Evid. 702 (providing for witnesses qualified as experts “by knowledge, skill, experience, training, or education”). [I]n very few cases will the evidence be such that the trial court’s reliability determination can properly be based only on the experience of a qualified expert to the exclusion of factors such as those set out in Robinson, or, on the other hand, properly be based only on factors such as those set out in Robinson to the exclusion of considerations based on a qualified expert’s experience. Whirlpool Corp. v.

Camacho, 298 S.W.3d 631, 638 (Tex.2009); see also Mack Trucks, 206 S.W.3d at 579 (“[T]he criteria for assessing reliability must vary depending on the nature of the evidence.”). Here, we are considering the reliability of a treating physician’s opinion based on a particular diagnostic methodology — differential diagnosis. This is a routine diagnostic method used in internal medicine whereby a treating physician formulates a hypothesis as to likely causes of a patient’s presented symptoms and eliminates unlikely causes by a deductive process of elimination. See, e.g., Coastal Tankships, U.S.A., Inc. v.

Anderson, 87 S.W.3d 591, 604-05 & n. 24 (Tex.App.-Houston [1st Dist.] 2002, pet. denied) (en banc) (“[Differential diagnosis] is a clinical process whereby a doctor determines which of several potential diseases or injuries is causing the patient’s symptoms by ruling out possible causes — by comparing the patient’s symptoms to symptoms associated with known diseases, conducting physical examinations, collecting data on the patient’s history and illness, and analyzing that data — until a final diagnosis for proper treatment is reached.”).

If the physician’s treatment of the suspected cause alleviates the patient’s symptoms, the disease or condition treated can be said to have been the internal cause of the eliminated symptoms. See id. If the patient’s symptoms remain after treatment of the suspected disease or condition, the physician rules out the suspected disease or condition as the internal cause of the patient’s symptoms and formulates a new hypothesis as to the possible culprit. See id.

Crump asserts that because differential diagnosis is a reliable medical technique, the application of the Robinson factors is tempered, or less strict, when a treating physician using that technique is involved. This is the approach adopted by the court of appeals below, which refused to apply Robinson at all. See 274 S.W.3d at 96-97.

We have held the opposite to be true: “[T]he relevance and reliability requirements of Rule 702 [apply] to all expert evidence offered under that rule, even though the criteria for assessing relevance and reliability must vary, depending on the nature of the evidence.” Gammill, 972 S.W.2d at 727; see also Camacho, 298 S.W.3d at 638. The mere fact that differential diagnosis was used does not exempt the foundation of a treating physician’s expert opinion from scrutiny — it is to be evaluated for reliability as carefully as any other expert’s testimony.

Both the Robinson and Gammill analyses are appropriate in this context. See generally Guinn v. AstraZeneca Pharm. LP, 602 F.3d 1245, 1254 (11th Cir.2010) (per curiam); Meister v. Med. Eng’g Corp., 267 F.3d 1123, 1129 (D.C.Cir.2001); Westberry v. Gislaved Gummi AB, 178 F.3d 257, 263-65 (4th Cir.1999); Heller v. Shaw Indus., Inc., 167 F.3d 146, 153-54 (3d Cir.1999); Moore v. Ashland Chem. Inc., 151 F.3d 269, 275-78 (5th Cir.1998) (en banc).

Several of the Robinson factors apply to differential diagnosis as a method or technique, as well as its application and the conclusions reached in a particular case. “Differential diagnosis is ‘the basic method of internal medicine’ and enjoys widespread acceptance in the medical community. Generally speaking, when properly conducted the technique has important non-judicial uses, is generally accepted as valid by the medical community, and has been subjected to use, peer review, and testing.” Coastal Tankships, 87 S.W.3d at 604 (citations omitted).

While these endorsements of the technique may hold “generally,” we cannot say that they will always apply in every case in which a treating physician bases his opinion on differential diagnosis. Here, though, Daller’s diagnostic methodology certainly had nonjudicial uses in that it was used to treat Crump, write prescriptions, and perform surgery. Cf. Robinson, 923 S.W.2d at 559 (“[Ojpinions formed solely for the purpose of testifying are more likely to be biased toward a particular result.”). Hunt, Transcontinental’s expert, acknowledged that differential diagnosis was used to treat Crump.

Although Hunt would have reached different conclusions regarding Crump’s infection, he stated that he agreed with the treatment methodology Daller employed. See Robinson, 923 S.W.2d at 557 (noting that the Robinson reliability inquiry focuses “solely on the underlying principles and methodology, not on the conclusions they generate”); cf. TXI Transp. Co. v.

Hughes, 306 S.W.3d 230, 235 (Tex.2010) (“Rather than focus entirely on the reliability of the underlying technique used to generate the challenged opinion, as in Robinson, we have found it appropriate ... [to] determine whether there are any significant analytical gaps in the expert’s opinion that undermine its reliability.”) (citations omitted). Moreover, there is no practical way peers could conduct objective, randomized experiments to test the validity of Daller’s specific conclusion regarding Crump’s injury. Thus, these factors support the reliability of Dal-ler’s expert testimony in this case.

Moving to the other Robinson factors, we note that, in some cases, a physician’s differential diagnosis may be too dependent upon the physician’s subjective guesswork or produce too great a rate of error — for example, when there are several consistent, possible causes for a particular set of symptoms. Related to these factors, Transcontinental contends that Daller’s diagnostic technique is not reliable because he did not exclude the other possible causes of Crump’s death with reasonable medical probability. See TXI Transp.

Co., 306 S.W.3d at 237 (“An expert’s failure to rule out alternative causes of an incident may render his opinion unreliable.”); Robinson, 923 S.W.2d at 559 (“An expert who is trying to find a cause of something should carefully consider alternative causes. [An expert’s] failure to rule out other causes of the damage renders his opinion little more than speculation.” (citation omitted)); see also Burroughs Wellcome Co. v. Crye, 907 S.W.2d 497, 500 (Tex.1995) (“[T]o constitute evidence of causation, an expert opinion must rest in reasonable medical probability.”).

Yet a medical causation expert need not “disprove or discredit ] every possible cause other than the one espoused by him.” Viterbo v. Dow Chem. Co., 826 F.2d 420, 424 (5th Cir.1987). Few expert opinions would be reliable if the rule were otherwise. Still, if evidence presents “other plausible causes of the injury or condition that could be negated, the [proponent of the testimony] must offer evidence excluding those causes with reasonable certainty.” Merrell Dow Pharm., Inc. v.

Havner, 953 S.W.2d 706, 720 (Tex.1997) (emphases added); see also Robinson, 923 S.W.2d at 558-59 (concluding that the trial court did not abuse its discretion by excluding testimony by an expert who “conducted no testing to exclude other possible causes ... even though he admitted in his deposition that many of the symptoms could be caused by” other specific conditions). We conclude that Daller’s testimony adequately excluded, with reasonable medical certainty, the other plausible causes raised by the evidence.

Hunt testified that, in his opinion, Crump died from a combination of kidney failure, cirrhosis of the liver, and a fungal infection in the lungs exacerbated by preexisting diabetes and a history of immunosuppressant drug usage. There is no dispute that these conditions, except the fungal infection, all preceded Crump’s May 2000 work-related injury. All of these were other plausible causes of Crump’s death.

But there was evidence that despite his long-term health problems dating from his kidney transplant twenty-five years earlier, Crump was generally in good health before his injury at work, and that within days after the injury he contracted an infection at the site of the injury. Even Hunt acknowledged that Crump’s infection was a “co-morbid condition” that made his other health conditions “more difficult to deal with.” Hunt disputed any connection between Crump’s injury and the infection, and he believed that the injury contributed nothing toward Crump’s death.

He concluded that Crump would have died on January 23, 2001, regardless of the work-related injury of May 2000. But objective evidence of Crump’s good health before his injury, his contraction of an infection at the site shortly afterward, and the deleterious effect of the infection on his health reasonably ruled out the possibility that he died solely from the other conditions he suffered.

Based on Daller’s experience and training as a transplant specialist and surgeon, his dealings with infection-susceptible immunosup-pressed patients, and his direct dealings with Crump — which included taking cultures directly from the wound site for diagnostic purposes — he concluded that Crump’s wound became infected, that the infection weakened his organs, and that the natural progression of these events caused his death.

See Crye, 907 S.W.2d at 500 (“Reasonable [medical] probability is determined by the substance and context of the opinion, and does not turn on semantics or on the use of a particular term or phrase.”). In other words, Daller’s medical causation opinion provided a cause that excluded, with reasonable medical certainty, Hunt’s suggested causes of death. The evidence was not conclusive, but it was not required to be. It was sufficiently reliable to be considered by the jury.

Once Daller effectively responded to Hunt’s other plausible causes of death with reliable testimony, the question was no longer one of legal sufficiency, but rather one of competing evidence to be weighed by the jury. See Ford Motor Co. v. Ledesma, 242 S.W.3d 32, 40-41 (Tex.2007). In addition to applying the Robinson factors, Gammill also informs our reliability inquiry. There we concluded that the Robinson factors could not be applied to the plaintiffs’ experts, “even though mechanical engineering, the expertise claimed by the witnesses, is scientific in nature.” Gammill, 972 S.W.2d at 727.

In that situation, we asked further whether there was “simply too great an analytical gap between the data and the opinion proffered.” Id. “The ‘analytical gap’ between the data ... and [the expert’s opinion was not shown to be due to his techniques in assessing the vehicle restraint system.... Rather, the ‘gap’ in [the expert]’s analysis was his failure to show how his observations, assuming they were valid, supported his conclusions that [the passenger] was wearing her seat belt or that it was defective.” Id.

The analysis set out in Gammill lends support to the reliability of Daller’s expert opinion testimony in this case. Daller is board-certified in general surgery and critical care, specializes in multiple-organ transplantation, and has worked as a clinician in teaching hospitals across the country — including the University of Texas Medical Branch at Galveston, where he treated Crump. His educational and clinical qualifications to treat post-transplantation, immunosuppressed patients, such as Crump, are not in dispute.

As explained above in addressing the Robinson factors, an analytical gap between the data and opinion is not shown here because of Dal-ler’s “techniques in assessing” Crump. See id. Rather, there must be “failure to show how his observations, assuming they were valid, supported his conclusions.” Id. He directly treated or oversaw Crump’s treatment on repeated occasions after Crump’s work-related knee injury.

Daller observed that Crump’s wound was located in the same spot as the injury and that the wound site became symptomatic as being infected in a predictable time and manner after the injury for an immunosuppressed patient such as Crump. He observed that, in his words, the wound infection tilted Crump’s “seesaw” away from an im-munosuppressed patient’s “relative balance” between immunosuppression and infection toward systemic infection. He took cultures from the wound site and performed surgery to diagnose and to assist healing of the wound.

The cultures allowed the observation that Crump’s wound was infected with the same agent as the infectious agent that had become systemic in Crump. Daller observed that Crump— despite being a kidney transplant recipient with diabetes and undiagnosed hepatitis C — had no medical history of organ problems from the period after the transplant in 1975 until after the work-related injury in 2000. Daller observed the problems with Crump’s organ function and concluded that “the worsening of those organs’ functions was caused by the infection.” From these observations, Daller concluded: Q.

Doctor, do you have an opinion regarding the cause of Mr. Crump’s death? A. Yes, I do. Q. And what is that opinion, sir? A. My opinion is that his death resulted from a natural sequence of events that began at the time of his knee injury. Q. Doctor, would you please elaborate as to what was the cause of the death as it related to the injury? A. Mr. Crump had had [sic] a renal transplant approximately 25 years prior, I believe. He also had what is known as compensated cirrhosis from hepatitis C.

At the time that he experienced the injury, that injury caused a progression of his hepatic insufficiency, and because of his inability to fight off infections and also because of his overall medical condition, it caused a series of events that led to his death. Q. Dr. Daller, is it your opinion that the infection which Mr. Crump developed was a producing cause of his death? A. It was the trigger cause of a sequence of events that then occurred. Thus, we cannot conclude that there was “too great an analytical gap” between the observed data and the proffered opinion. See Gammill, 972 S.W.2d at 727.

At this point, any “gaps” that remain between the data and the conclusion drawn from it go to the weight of Daller’s testimony — not its reliability. See Ledesma, 242 S.W.3d at 40-41. We conclude that Daller’s testimony was based on a sufficiently reliable foundation under the standards set out in Robinson and Gammill. Because Daller’s expert medical causation testimony is based on a reliable foundation, it was admissible at trial as evidence to prove that the May 2000 injury was a producing cause of Crump’s death. See Tex.R. Evid. 702.

Consequently, on legal sufficiency review, we conclude that reasonable jurors could have believed his testimony. See City of Keller v. Wilson, 168 S.W.3d 802, 827 (Tex.2005). Because Daller’s expert testimony was sufficient evidence to support the jury’s verdict on causation, we cannot disturb the jury’s finding against Transcontinental on the issue of producing cause. Accordingly, Transcontinental’s legal sufficiency challenge is denied. III.

Jury Charge In its second issue, Transcontinental argues that the trial court’s definition of producing cause is legally incorrect and that, had its proposed definition been given, the verdict would not have been in Crump’s favor. Transcontinental seeks a new trial.

After refusing Transcontinental’s proposed definition and overruling its objections that Crump’s definition erroneously lacked a but-for component, the trial court charged the jury as follows: QUESTION NO. 1 You are instructed that the Texas Workers’ Compensation Commission found that Charles Crump’s compensa-ble right knee injury of May 9, 2000 resulted in his death on January 23, 2001. “Injury” means damage or harm to the physical structure of the body and a disease or infection naturally resulting from the damage or harm. “Producing Cause” means an efficient, exciting, or contributing cause that, in a natural sequence, produces the death in question.

There may be more than one producing cause. WAS CHARLES CRUMP’S MAY 9, 2000 INJURY A PRODUCING CAUSE OF HIS DEATH? Answer ‘Yes” or “No.” Answer: Yes A. Was the Trial Court’s Definition of Producing Cause Erroneous? Transcontinental contends that the trial court gave an erroneous definition of producing cause in its jury charge. We agree. “The court shall submit such instructions and

definitions as shall be proper to enable the jury to render a verdict.” Tex.R. Civ. P. 277. “An instruction is proper if it (1) assists the jury, (2) accurately states the law, and (8) finds support in the pleadings and evidence.” Union Pac. R.R. Co. v. Williams, 85 S.W.3d 162, 166 (Tex.2002) (citing Tex.R. Civ. P. 278). When, as here, the content of a trial court’s definition is challenged as legally incorrect, our standard of review is de novo. See St. Joseph Hosp. v. Wolff, 94 5.W.3d 513, 525 (Tex.2002).

Though the Texas Workers’ Compensation Act does not use the phrase “producing cause,” this has been the standard for proving causation in workers’ compensation claims for more than eighty years. Courts generally agreed that the workers’ compensation causation standard was not to be as exacting as that of the common law, but no uniform definition of causation in the workers’ compensation context emerged. As the law developed, courts recognized that the only substantial difference between the two standards was that a proximate cause — the common law standard — must be foreseeable.

We finally approved a definition of producing cause for workers’ compensation cases in 1943: Compensation is awarded to the legal beneficiary of the deceased employee if death results from the injury. Causal connection must be established between the injury and the death. The injury must be the producing cause of the death, and producing cause has been defined as that cause which, in a natural and continuous sequence, produces the death ... in issue, and without which the death ... would not have occurred. Jones v. Traders & Gen. Ins.

Co., 140 Tex. 599, 169 S.W.2d 160, 162 (Tex.1943) (citation and quotation omitted). We have not addressed the matter since then. In a recent products liability case, however, we held that what had been “a frequently submitted definition of ‘producing cause’ should no longer be used.” Ford Motor Co. v. Ledesma, 242 S.W.3d 32, 35 (Tex.2007). The trial court in that case had given the products liability pattern jury charge definition: “ ‘Producing cause’ means an efficient, exciting, or contributing cause that, in a natural sequence, produces the incident in question.

There may be more than one producing cause.” Id. at 45. Because we held that the trial court committed reversible error in a separate part of the charge, we reversed on that ground and remanded the case for new trial. Id. at 44. But to assist the parties and the court on remand, we further held that producing cause should be correctly defined as “a substantial factor in bringing about an injury, and without which the injury would not have occurred.” Id. at 46.

The definition submitted by Crump and accepted at trial — “ ‘Producing Cause’ means an efficient, exciting, or contributing cause that, in a natural sequence, produces the death in question. There may be more than one producing cause.” — is the same as the pattern jury charge definition we rejected in Ledesma, substituting only “death” for “incident.” Transcontinental urges us now to adopt for workers’ compensation cases the same definition we approved in Ledesma for products liability cases. Crump asserts that the Ledesma definition of producing cause has no place in workers’ compensation law.

Because we have not addressed the “substantial factor” terminology from products liability law in the context of workers’ compensation cases, we must decide whether our holding in Ledesma applies here. In considering whether to apply Ledesma’s definition, we first examine the causation standards for proximate cause and producing cause. “The two elements of proximate cause are cause in fact (or substantial factor) and foreseeability....

Cause in fact is established when the act or omission was a substantial factor in bringing about the injuries, and without it, the harm would not have occurred.” IHS Cedars Treatment Ctr. v. Mason, 143 S.W.3d 794, 798-99 (Tex.2004). “The approved definition of ‘proximate cause’ in negligence cases and the approved definition of ‘producing cause’ in compensation cases are in substance the same, except that there is added to the definition of proximate cause the element of foresee-ableness.” Staggs, 134 S.W.2d at 1028-29.

In other words, the producing cause inquiry is conceptually identical to that of cause in fact. We have recognized this in Deceptive Trade Practices Act cases. See, e.g., Prudential Ins. Co. of Am. v. Jefferson Assocs., Ltd., 896 S.W.2d 156, 161 (Tex.1995) (“For DTPA violations, only producing cause must be shown. The element common to both proximate cause and producing cause is actual causation in fact. This requires proof that

an act or omission was a substantial factor in bringing about injury which would not otherwise have occurred.”) (citations omitted). Also, in Le-desma, a products liability case, we recognized that producing cause and cause in fact are conceptually identical.

See 242 S.W.3d at 46 (“Defining producing cause as being a substantial factor in bringing about an injury, and without which the injury would not have occurred, is easily understood and conveys the essential components of producing cause that (1) the cause must be a substantial cause of the event in issue and (2) it must be a but-for cause, namely one without which the event would not have occurred.”). The producing cause inquiry in workers’ compensation cases is conceptually no different from the cause in fact inquiry in negligence cases and the producing cause inquiry in other substantive contexts.

We see no reason to define producing cause differently in this context. Therefore, we hold that producing cause in workers’ compensation cases is defined as a substantial factor in bringing about an injury or death, and without which the injury or death would not have occurred. Having concluded that Ledesma applies to this case, we must determine whether the definition given here was erroneous. The definition of producing cause given by the trial court in this case is, in all relevant respects, the same as the pattern jury charge definition we rejected in Ledesma. See 242 S.W.3d at 45.

Transcontinental cites Ledesma for the proposition that the use of the “efficient, exciting, or contributing cause” language is erroneous. We disagree. In Ledesma, we reasoned that those terms had no practical meaning to modern jurors in products liability cases. Id. at 46. We held that the use of those terms provides “little concrete guidance” to modern jurors, and a definition .that omits either the substantial factor or but-for components “is incomplete.” Id.

But we did not go so far as to say that employing the terms “efficient, exciting, or contributing cause” in a producing cause definition was, as Transcontinental suggests, erroneous. Thus, while the’ concerns about terminology “foreign to modern English” and incomplete

definitions expressed in Ledesma apply equally to this case, the mere use of these terms is not, in itself, error. Crump argues that the “substantial factor” component of the Ledesma definition imposes a higher causation burden upon workers’ compensation claimants than what exists at present. We disagree. We have always required in workers’ compensation cases a showing of “unbroken causal connection” between the compensable injury and the claimant’s injury or death. Staggs, 134 S.W.2d at 1030; see Jones, 169 S.W.2d at 162; Burnett, 105 S.W.2d at 202.

Although, as the dissent points out, our earlier cases did not address the “substantial factor” terminology, there is nothing in those opinions to suggest that cause in fact should not be part of the causal connection analysis. See Jones, 169 S.W.2d at 162; Staggs, 134 S.W.2d at 1030; Burnett, 105 S.W.2d at 202. In fact, we cannot conceive of causal connection analysis without consideration of cause in fact.

The substantial factor language serves only to illustrate an essential aspect of causation to jurors, as we have noted: The word “substantial” is used to denote the fact that the defendant’s conduct has such an effect in producing the harm as to lead reasonable men to regard it as a cause, using that word in the popular sense, in which there always lurks the idea of responsibility, rather than in the so-called “philosophic sense,” which includes every one of the great number of events without which any happening would not have occurred.

Each of these events is a cause in the so-called “philosophic sense,” yet the effect of many of them is so insignificant that no ordinary mind would think of them as causes. Lear Siegler, Inc. v. Perez, 819 S.W.2d 470, 472 & n. 1 (Tex.1991) (quoting Restatement (Second) of Torts § 431 cmt. a (1965)); see also Borg-Wamer Corp. v. Flores, 232 S.W.3d 765, 770 (Tex.2007). In other words, for

an act or event to rise to the level of cause in the legal sense, the act or event must be such that reasonable jurors would identify it as being actually responsible for the ultimate harm. The cause must be more than one of the countless ubiquitous and insignificant causes that in some remote sense may have contributed to a given effect as, for example, simply getting up in the morning. That the term substantial factor is given to this commonsense aspect of legal causation simply makes plain to jurors that more than causation in this indirect, “philosophic sense” is required.

See Staggs, 134 S.W.2d at 1030 (recognizing that but-for language repeated something already included in the usual and ordinary meaning of “cause” and draws juror attention to the importance of an unbroken causal connection). It does not demand, nor even imply, a higher standard of legal causation beyond the ordinary sense of the concept. Transcontinental argues that the omission of but-for language in the charge submitted by the trial court renders the definition legally incorrect. We agree.

As we discussed in one workers’ compensation case, “to say of a cause of an injury that it is one ‘but for which the injury would not have happened’ is to repeat something already included in the usual and ordinary meaning of the word ‘cause.’ ” Id. (quoting Tex. & Pac. Ry. Co. v. Short, 62 S.W.2d 995, 999 (Tex.Civ.App.-Eastland 1933, writ ref'd)). However, the inclusion of but-for language in producing cause

definitions has long been considered useful, serving “to direct the jury’s attention to the importance of unbroken causal connection between the injury and the disability or death.” Id.; see also Wichita County v. Hart, 917 S.W.2d 779, 783-84 (Tex.1996) (“A trial court must submit explanatory instructions and

definitions that will assist the jury in rendering a verdict.”). We recognized this in Ledesma and, desiring to offer “practical help to a jury striving to make the often difficult causation determination,” held that a producing cause definition that did not include the but-for component was “incomplete.” 242 S.W.3d at 46. Indeed, we have often referred to producing cause and cause in fact synonymously with but-for causation. See, e.g., LMB, Ltd. v. Moreno, 201 S.W.3d 686, 688 (Tex.2006) (per curiam); Marathon Corp. v. Pitzner, 106 S.W.3d 724, 727 (Tex.2003) (per curiam); Hyundai Motor Co. v.

Rodriguez, 995 S.W.2d 661, 667 (Tex.1999). The producing cause definition submitted in this case lacked the but-for component. It, too, was incomplete, and therefore an erroneous statement of the law of producing cause. See Union Pac. R.R. Co. v. Williams, 85 S.W.3d 162, 166 (Tex.2002). B. Was the Trial Court’s Definition of Producing Cause Harmful? Having decided that the trial court’s definition of producing cause was erroneous, we now consider whether this error requires reversal.

Transcontinental asserts that the trial court’s omission of the but-for component of the producing cause definition was reversible error. We agree. “A judgment will not be reversed for charge error unless the error was harmful because it probably caused the,rendition of an improper verdict....” Columbia Rio Grande Healthcare, L.P. v. Hawley, 284 S.W.3d 851, 856 (Tex.2009) (citing Tex.R.App. P. 61.1). “Charge error is generally considered harmful if it relates to a contested, critical issue.” Id. (citing Bel-Ton Elec. Serv., Inc. v. Pickle, 915 S.W.2d 480, 481 (Tex.1996) (per curiam), and Sw. Bell Tel. Co. v.

John Carlo Tex., Inc., 843 S.W.2d 470, 472 (Tex.1992)). “To determine whether the instruction probably caused an improper judgment, we examine the entire record.” Quantum Chem. Corp. v. Toennies, 47 S.W.3d 473, 480 (Tex.2001). Daller, one of Crump’s treating physicians and Crump’s sole expert, testified: Mr. Crump had had [sic] a renal transplant approximately 25 years prior, I believe. He also had what is known as compensated cirrhosis from hepatitis C.

At the time that he experienced the injury, that injury caused a progression of his hepatic insufficiency, and because of his inability to fight off infections and also because of his overall medical condition, it caused a series of events that led to his death. According to Daller, the site of the injury became infected, the infection caused Crump’s already-weakened organs to fail, and his organ failure in turn caused his death.

Further, on cross-examination he testified: Q: It’s your opinion that the cause of the infection was histoplasmosis but that but for the bruise the histoplas-mosis would not have developed into a full-stage infection, right? A: That is correct. On the other hand, Transcontinental’s expert, Hunt, testified that Crump died from the natural complications of being immu-nosuppressed for twenty-five years rather than from the May 9, 2000 injury: Q. In your opinion, would the death have occurred without the May 9, 2000 injury ever taking place? A. Yes.

Hunt testified on cross-examination, “Had [Crump] not had that contusion to his knee, he still would have had those other problems.” Q. So, you’re saying that Mr. Crump would have been in there in Polly Ryon [Memorial Hospital] on January 2Srd of 2001, dying of liver failure, kidney failure, aspiration of the stomach, his heart was giving out— I mean, his whole body was shutting down. You said yourself he was critically ill. A: Yes, sir. Transcontinental bore the burden of proving, by a preponderance of the evidence, the negative proposition that the May 2000 injury was not a producing cause of Crump’s death.

See Tex. Lab.Code § 410.303; Morales v. Liberty Mut. Ins. Co., 241 S.W.3d 514, 516 (Tex.2007) (“[T]he appealing party bears the burden of proof by a preponderance of the evidence. The factfinder may consider, but is not bound by, the appeals panel’s decision. The method of review that [the Labor Code] provides is known as modified de novo review.”) (citations omitted). The but-for aspect of causation was squarely at issue in this case, and the sole question before the jury was whether the May 2000 injury was a producing cause of Crump’s death.

Here, the charge error “relatefd] to a contested, critical issue” — indeed, the sole issue — that of causation. See Hawley, 284 S.W.3d at 856; see also Toennies, 47 S.W.3d at 480 (“An improper instruction is especially likely to cause an unfair trial when the trial is contested and the evidence sharply conflicting, as it was in the present case [when the trial court gave an incorrect causation standard].”); Tex. Dep’t of Human Sens. v.

Hinds, 904 S.W.2d 629, 637 (Tex.1995) (finding harmful error where a jury instruction stated the standard of causation incorrectly and the evidence was “vigorously and convincingly disputed”); John Carlo Tex., Inc., 843 S.W.2d at 472 (“Virtually the entire factual dispute between the parties has been over whether Bell’s conduct was justified. To ask the jury to resolve this dispute without a proper legal definition [of justification,] the essential legal issue[,] was reversible error.”).

Including the but-for component in the definition would have assisted the jury in resolving the disputed expert testimony at the crux of the case and, more importantly, would have stated the law accurately. See Williams, 85 S.W.3d at 166. In these circumstances, the absence of a proper definition of producing cause probably resulted in an improper judgment and, as such, was reversible error. See Hinds, 904 S.W.2d at 637. Prior to trial, Transcontinental objected to Crump’s definition, the one ultimately accepted by the trial court, asserting it was not a correct statement under Texas law.

Transcontinental tendered its own definition in writing, including a but-for component: “that cause, which in a natural and continuous sequence, produces death and without which, the death would not have occurred.” See Tex.R. Crv. P. 278 (“Failure to submit a definition or instruction shall not be deemed a ground for reversal of the judgment unless a substantially correct definition or instruction has been requested in writing and tendered by the party complaining of the judgment.”).

Because Transcontinental’s definition included the critical but-for component, and was otherwise a correct statement of law, it was “substantially correct” and sufficed to preserve its complaint of charge error on appeal. See Hinds, 904 S.W.2d at 637-38 (“There should be but one test for determining if a party has preserved error in the jury charge, and that is whether the party made the trial court aware of the complaint, timely and plainly, and obtained a ruling.”) (quoting State Dep’t of Highways v. Payne, 838 S.W.2d 235, 241 (Tex.1992)).

We hold that the definition of producing cause approved in Ledesma — a substantial factor in bringing about the injury or death and without which the injury or death would not have occurred — applies in workers’ compensation cases. Because the definition submitted here lacked the but-for component, and because its omission in this case constitutes harmful error, we remand the case for new trial. IV.

Attorney’s Fees Transcontinental first argues that the trial court erred in denying it a jury trial on the amount of Crump’s reasonable and necessary attorney’s fees for which Transcontinental was statutorily liable, and second, in permitting Crump to recover attorney’s fees incurred in pursuing those statutory attorney’s fees. We address the first issue — whether a judge or jury decides attorney’s fees under Texas Labor Code § 408.221(c) — to provide guidance for parties and trial courts. We leave the second question — whether fees can be awarded in the pursuit of fees — for another day.

Relying on this Court’s precedent and the language of § 408.221, Transcontinental argues that the trial court erred when it refused to grant a jury trial and, instead, decided the disputed amount of Crump’s attorney’s fees for which Transcontinental was liable under § 408.221(c). Crump contends that the statute’s plain language alone provides that the court, and not a jury, is to determine the amount of reasonable and necessary attorney’s fees for which Transcontinental is liable.

The court of appeals favored Crump’s plain language argument and held that the trial court did not err in denying Transcontinental’s request to submit the issue to a jury. 274 S.W.3d at 103. We hold that when a question of fact exists on the reasonableness and necessity of a claimant’s attorney’s fees under § 408.221(c), the carrier has a right to submit that question to a jury. In construing another provision of the Workers’ Compensation Act, we set out the scope of our inquiry: The meaning of a statute is a legal question, which we review de novo to ascertain and give effect to the Legislature’s intent.

Where text is clear, text is determinative of that intent. This general rule applies unless enforcing the plain language of the statute as written would produce absurd results. Therefore, our practice when construing a statute is to recognize that the words the Legislature chooses should be the surest guide to legislative intent. Only when those words are ambiguous do we resort to rules of construction or extrinsic aids. Entergy Gulf States, Inc. v. Summers, 282 S.W.3d 433, 437 (Tex.2009) (citations, internal quotations, and italics omitted).

We review de novo the trial court’s denial of Transcontinental’s request for a jury trial under subsection (c). See id. We look first to the language of the statute. The relevant portions of § 408.221 read: (

a) An attorney’s fee, including a contingency fee, for representing a claimant before the division or court under [the Texas Workers’ Compensation Act] must be approved by the commissioner or court. (

b) Except as otherwise provided, an attorney’s fee under this

section is based on the attorney’s time and expenses according to written evidence presented to the division or court. Except as provided by Subsection (c) ..., the attorney’s fee shall be paid from the claimant’s recovery. (

c) An insurance carrier that seeks judicial review ... of a final decision of the appeals panel regarding compens-ability or eligibility for, or the amount of, income or death benefits is liable for reasonable and necessary attorney’s fees as provided by Subsection (d)incurred by the claimant as a result of the insurance carrier’s appeal if the claimant prevails on an issue on which judicial review is sought by the insurance carrier....

If the carrier appeals multiple issues and the claimant prevails on some, but not all, of the issues appealed, the court shall apportion and award fees to the claimant’s attorney only for the issues on which the claimant prevails. In making that apportionment, the court shall consider the factors prescribed by Subsection (d).... (

d) In approving an attorney’s fee under this section, the commissioner or court shall consider: (1) the time and labor required; (2) the novelty and difficulty of the questions involved; (3) the skill required to perform the legal services properly; (4) the fee customarily charged in the locality for similar legal services; (5) the amount involved in the controversy; (6) the benefits to the claimant that the attorney is responsible for securing; and (7) the experience and ability of the attorney performing the services. (

i) Except as provided by Subsection (c) ..., an attorney’s fee may not exceed 25 percent of the claimant’s recovery. Tex. Lab.Code § 408.221. We review the plain language of the statute as written to decide whether Crump’s or Transcontinental’s

interpretation — judge or jury — is supported. According to Crump, subsection (c)’s instruction that the court is to award apportioned fees means that the court alone determines the reasonable and necessary amount of fees — according to criteria given in subsection (d). This

interpretation, Crump argues, comports with subsection (b)’s general rule that attorney’s fees under § 408.221 are based on written evidence of time and expenses presented to the court, which would have no use for this information if it were not deciding the amount to award. Crump correctly notes that § 408.221 makes no mention of a jury. Transcontinental focuses on the first words in subsection (b): “Except as otherwise provided.” Id. § 408.221(b). It argues that subsection (

c) represents an exception to the general rule set out in subsections (

a) and (b), where a claimant’s attorney’s fees are paid out of the claimant’s benefit recovery and a court decides those fees. Subsection (

c) controls the situation where a claimant’s attorney’s fees are paid directly by the liable insurance carrier, and a court — or, according to Transcontinental, a jury if one is sought — decides the extent of the carrier’s additional liability beyond the claimant’s benefits award. Transcontinental points out that every reference to a court’s action in § 408.221 is that of approval, except in sentence two of subsection (c), under which the court is to apportion and, after that apportionment is concluded, “award fees to the claimant’s attorney only for the issues on which the claimant prevails.” Id. § 408.221(c).

Transcontinental correctly observes that the statute does not state explicitly that the court alone is to determine the amount of attorney’s fees, nor does it expressly forbid a jury from deciding the matter. The statute is silent on the critical judge-or-jury question. Both parties offer legitimate, reasonable

interpretations of § 408.221 and subsection (c)’s role within it. Because both

interpretations are reasonable as to their applicability here, we conclude that the statute is ambiguous. See In re Mo. Pac. R.R. Co., 998 S.W.2d 212, 217 (Tex.1999) (“Beyond these preliminary observations, the statute is not entirely clear in all its particulars. The language of the statute could support more than one reasonable

interpretation and therefore is ambiguous. Because it is ambiguous, we may turn to extratextual sources.... ”). Because the plain language of the statute alone is unavailing, we look beyond it. See id.; see also Tex. Gov’t Code § 311.023 (“In construing a statute, ... a court may consider among other matters the ... common law or former statutory provisions, including laws on the same or similar subjects.... ”). Specifically, we are guided by prior decisions examining the issue of reasonable and necessary attorney’s fees in the context of fee-shifting provisions in other statutory regimes and by the history of how § 408.221 has evolved over the years.

Section 408.221 provides two relevant possibilities in which an insurance carrier will pay a claimant’s attorney’s fees. The first is where the carrier pays the claimant’s attorney’s fees for representation before the Division of Workers’ Compensation and some court proceedings, but the fees are subtracted from the claimant’s recovery. Tex. Lab.Code § 408.221(a)-(b). Because, in effect, the claimant pays in this situation, the claimant’s attorney’s fees are limited to 25% of the claimant’s recovery. Id. § 408.221(1).

The trial court must approve these fees, id. § 408.221(a), and must consider several factors in doing so, id. § 408.221(d). The insurance carrier can only be said to pay these fees in the technical sense that it drafts a separate check for the attorney’s fees, payable directly to the claimant’s attorney. Id. § 408.221(h). In reviewing fees awarded in this situation, we have “held that the amount of the attorney’s fees to be allowed in compensation cases is a matter for the trial court to determine without the aid of a jury, and the amount of the recovery is within its discretion.” Tex. Employers Ins. Ass’n v.

Motley, 491 S.W.2d 395, 397 (Tex.1973) (citing Tex. Employers Ins. Ass’n v. Hatton, 152 Tex. 199, 255 S.W.2d 848, 849 (1953) (“The amount of attorney’s fees to be allowed in a compensation case is exclusively for the court and not the jury, and any such [contingency-fee] contract was made subject to the approval by the court. The court in his discretion could award a lesser amount.”)). Crump cites Motley and Hat-ton as dispositive here. But, as discussed below, those cases do not address subsection (c), which was not enacted until decades later. The second possibility is at issue in this case.

Here, the insurance carrier pays the claimant’s “reasonable and necessary attorney’s fees” for representing the claimant on judicial review in the courts when the carrier is unsuccessful on an issue it appealed from the Division of Workers’ Compensation. Tex. Lab.Code § 408.221(c). These fees are not subtracted from the claimant’s recovery, but are paid by the carrier on top of the claimant’s benefits award. See id. § 408.221(b)-(c). In this situation, the claimant’s attorney’s fees are not limited to 25% of the claimant’s recovery, but only by reasonableness and necessity. See id. § 408.221©, (I).

This fee-shifting provision in § 408.221 did not exist until 2001. Prior to 2001, the only attorney’s fee award mechanism was the one in subsection (b), described in the paragraph above, where the claimant always pays his own attorney’s fees regardless of the outcome of the carrier’s appeal. For this reason, Motley and Hatton do not control in this case — they addressed this other statutory scheme, and could not have contemplated the fee-shifting mechanism presented in subsection (c).

While we have not previously examined the fee-shifting provision in subsection (c), we have discussed similar fee-shifting provisions in other cases. “In general, the reasonableness of statutory attorney’s fees is a jury question.” City of Garland v. Dallas Morning News, 22 S.W.3d 351, 367 (Tex.2000). In City of Garland, we considered the fees available to a substantially prevailing party under the Texas Public Information Act. Id. at 367-68. The Act provided: (

a) In an action brought under

Section 552.321 [suit for writ of mandamus under the Act] ..., the court may assess costs of litigation and reasonable attorney fees incurred by a plaintiff or defendant who substantially prevails. (

b) In exercising its discretion under this section, the court shall consider whether the conduct of the governmental body had a reasonable basis in law and whether the litigation was brought in good faith. Id. at 367 (quoting the former version of Tex. Gov’t Code § 552.323). We reasoned that, from the plain language of the statute, “the trial judge decides whether to award attorney’s fees under the Act.” Id. (emphasis added).

But we immediately noted that “section 552.323 does not dictate how to determine the attorney’s fees amount, except that the award must be ‘reasonable.’ In general, the reasonableness of statutory attorney’s fees is a jury question.” Id. (emphasis added, citation omitted). As support for this “general” proposition, we cited Bocquet v. Herring, 972 S.W.2d 19, 21 (Tex.1998), which involved attorney’s fee awards under the Declaratory Judgment Act. Id.

In any proceeding under the Act, “the court may award costs and reasonable and necessary attorney’s fees as are equitable and just.” Bocquet, 972 S.W.2d at 20 (quoting Tex. Civ. Prac. & Rem.Code § 37.009). We had to determine whether a judge or jury was to decide the amount of fees in order to answer another question: “[B]y what standard is such an award of attorney fees to be reviewed on appeal”? Id. Because the Act read “may,” the trial court had discretion to decide whether to award fees at all. Id. We recognized that the Act limited this discretion in four ways: reasonableness, necessity, equity, and justice.

Id. at 21. Each of those terms prescribed whether a judge or jury was to decide them. Id. Generally, reasonableness was a fact question for the jury’s determination, as was necessity. Id. On the other hand, equity was within the trial court’s discretion, as was justice. Id.

Our examination of the Act’s language led us to conclude: Therefore, in reviewing an attorney fee award under the Act, the court of appeals must determine whether the trial court abused its discretion by awarding fees when there was [legally or factually] insufficient evidence that the fees were reasonable and necessary, or when the award was inequitable or unjust. Unreasonable fees cannot be awarded, even if the court believed them just, but the court may conclude that it is not equitable or just to award even reasonable and necessary fees.

This multifaceted review involving both evidentia-ry and discretionary matters is required by the language of the Act. Id. In concluding that reasonableness and necessity of attorney’s fees were matters of fact committed to a jury, we also noted that there are “factors prescribed by law which guide the determination of whether attorney fees are reasonable and necessary.” Id. (citing Arthur Andersen & Co. v. Perry Equip. Corp., 945 S.W.2d 812, 818 (Tex.1997) (listing factors “a factfinder should consider when determining the reasonableness of a fee”)).

The principles established for construing statutory fee-shifting provisions in City of Garland and Bocquet assist the

interpretation of § 408.221(

c) of the Texas Labor Code. Crump has not pointed us to a reason to exempt § 408.221 from the general rule announced in those cases: “[T]he reasonableness of statutory attorney’s fees is a jury question.” City of Garland, 22 S.W.3d at 367. Nor do we see language in § 408.221 that distinguishes it from the language of the statutory regimes to which we applied the general rule in those cases.

Applying that general rule here, we conclude that the carrier is entitled to submit the issue of the reasonableness and necessity of a claimant’s attorney’s fees, where disputed, to a jury, which will consider subsection (d)’s factors. See Tex. Lab. Code § 408.221(c), (d). The next step depends on whether the claimant totally or partially prevails on the issues appealed by the insurance carrier.

If the claimant prevails only on some issues, then after the jury’s verdict is announced the court will apportion the fees per the factors in subsection (d), and will award reasonable and necessary attorney’s fees to the claimant’s attorney only for those issues on which the claimant prevails. See Tex. Lab.Code § 408.221(c). If the claimant totally prevails, the jury’s verdict as to the fee amount for which the carrier is liable is then subject only to the court’s approval based on the factors in subsection (d).

Id. § 408.221(a), (d); see also Bocquet, 972 S.W.2d at 21 (“Unreasonable fees cannot be awarded.... ”). Regardless of whether the claimant partially or totally prevails, the jury’s verdict as to the fee amount “must be approved by the ... court.” Tex. Lab.Code § 408.221(a). When a claimant pays his attorney’s fees out of his benefits recovery, the amount approved by the court is solely within its discretion based on the attorney’s time and expenses according to written evidence presented to the court and according to subsection (d)’s factors. See id. § 408.221(a), (b), (d); Hatton, 255 S.W.2d at 849. This

interpretation resolves § 408.221’s ambiguity while respecting its pre- and post-2001 award mechanisms and, at the same time, respects our precedent on the reasonableness and necessity of statutory attorney’s fees. Thus, we hold that an insurance carrier is entitled to have a jury determine the disputed amount of reasonable and necessary attorney’s fees for which it is liable under § 408.221(c). V.

Conclusion We hold that: (1) the treating physician’s opinion is based on a reliable foundation and, therefore, legally sufficient evidence supports the jury’s verdict; (2) the trial court’s omission of the but-for component in the jury charge constitutes reversible error; and (3) an insurance carrier is entitled to have a jury determine the disputed amount of reasonable and necessary attorney’s fees for which it is liable. We reverse the court of appeals’ judgment and remand the case to the trial court for new trial. Justice JOHNSON filed a concurring opinion, in which Justice LEHRMANN joined.

Justice GUZMAN did not participate in the decision. . For simplicity, we refer to the Crumps collectively in this opinion. . In Robinson, we outlined six useful considerations for determining the reliability of expert testimony: 1. the extent to which the theory has been or can be tested; 2. the extent to which the technique relies upon the subjective

interpretation of the expert; 3. whether the theory has been subjected to peer review and/or publication; 4. the technique’s potential rate of error; 5. whether the underlying theory or technique has been generally accepted as valid by the relevant scientific community; and 6.the non-judicial uses which have been made of the theory or technique. 923 S.W.2d at 557. “We emphasized in Robinson that these factors are non-exclusive and that [Texas] Rule [of Evidence] 702 contemplates a flexible inquiry.” Cooper Tire & Rubber Co. v.

Mendez, 204 S.W.3d 797, 801 (Tex.2006). . ”[T]emporal proximity alone does not meet standards of scientific reliability and does not, by itself, support an inference of medical causation. ... Nevertheless, when combined with other causation evidence, evidence that conditions exhibited themselves or were diagnosed shortly after an event may be probative in determining causation.” Guevara v. Ferrer, 247 S.W.3d 662, 667-68 (Tex.2007) (citations omitted). . See Tex.

Lab.Code § 408.181(a) ("An insurance carrier shall pay death benefits to the legal beneficiary if a compensable injury to the employee results in death.”) (emphasis added); id. § 401.011(10) (" ‘Compensable injury' means an injury that arises out of and in the course and scope of employment for which compensation is payable under this subtitle.”) (emphasis added); id. § 401.011(26) (“'Injury' means damage or harm to the physical structure of the body and a disease or infection naturally resulting from the damage or harm.”) (emphasis added). .

Initially the phrase "exciting and efficient cause” was used to describe the causation element in a workers’ compensation claim. See, e.g., Travelers’ Ins. Co. v. Smith, 266 S.W. 574, 575-76 (Tex.Civ.App.-Beaumont 1924, no writ) ("[T]he act contemplates ... that in case of the death of the employee from a disease which is shown to be the exciting and efficient cause of the employee’s death, his beneficiaries are entitled to compensation, though they be unable to prove that such death was proximately caused by the injury received, as the term ‘proximate cause' is used in the law of negligence.”).

"Producing cause” was used interchangeably with “efficient cause.” Travelers’ Ins. Co. v. Peters, 14 S.W.2d 1007, 1008 (Tex.Comm’n App.1929, holding approved), vacated on other grounds, 17 S.W.2d 457 (Tex.Comm’n App.1929). Peters held: [T]he rule of proximate cause has no application to cases arising under the Workmen's Compensation Act. The term "proximate cause” is not used anywhere in the act. A party claiming compensation under such act cannot be compelled to go further than is required by the provisions of the act, either in pleading or proving his cause of action.

It is true that there must be established a causal connection between an injury and the death of an employee before a recovery would be authorized. If, however, the injury is shown to be the producing cause of the death, a finding is justified that death was due to the injury, if it arises in the course of and out of the employment. Id. . E.g., Tex. Employers’ Ins. Ass'n v. Burnett, 129 Tex. 407, 105 S.W.2d 200, 202 (Tex.1937) ("Confusion has arisen by the use of the term 'producing cause' as distinguished from the term 'proximate cause’ usually employed in negligence cases.”). . See, e.g., Tex. Indem. Ins.

Co. v. Staggs, 134 Tex. 318, 134 S.W.2d 1026, 1028-29 (Tex.1940). In Staggs, we explained: It is well settled that in a suit under the compensation law it is not necessary for the claimant to show that the injury proximately caused disability or death. Recovery is authorized if a causal connection is established between the injury and the disability or death. "Producing cause” is the term most frequently used in compensation cases.

Sometimes the cause required to be proven is described as an "efficient, exciting or contributing cause.” The approved definition of "proximate cause” in negligence cases and the approved definition of “producing cause” in compensation cases are in substance the same, except that there is added to the definition of proximate cause the element of foreseeableness. It is apparent that “producing cause” is broader in its scope than is "proximate cause.” Id. (citations omitted). . See also Comm, on Pattern Jury Charges, State Bar of Tex., Texas Pattern Jury Charges — Malpractice, Premises, & Products 70.1 (2002). .

Our Ledesma opinion was issued just after the court of appeals received briefing, but the court of appeals held it "distinguishable and inapplicable to this appeal because it is a products liability case which requires the cause to be a substantial factor of the event in issue, a requirement absent from a workers’ compensation case.” 274 S.W.3d at 100 n. 13. . Ledesma did not distinguish between DTPA and products liability cases when surveying the various

definitions of producing cause. See 242 S.W.3d at 45-46 n. 47. . We observe that the substantial factor terminology, though infrequently encountered, is not entirely foreign to workers' compensation law. See, e.g., Pac. Indem. Co. v. Arline, 213 S.W.2d 691, 698 (Tex.Civ.App.-Beaumont 1948, writ dism'd by agr.) (“[T]he Workmen’s Compensation Law expresses the necessary causal relation between injury and disability by various forms of the word 'result' in almost every instance....

The words 'resulting from' ... refer only to causation in fact; the requirement is that the injury be a substantial factor in bringing about the disability, something without which the disability, would not have occurred.... [E]very definition ... we have seen of 'producing cause,’ a term which has been frequently used, expresses only causation in fact.”). . Consistent with Ledesma, however, we believe those terms ought not to be used to define producing cause in the future. . Subsection (

c) is the focus of the parties' dispute, but because it is part of a single

section of the Labor Code governing the award of attorney’s fees, we construe it in that context. See Morales, 241 S.W.3d at 517 ("We ... consider each provision in the context of the entire statute, not merely those portions that are in dispute.”); Cont'l Cas. Ins. Co. v. Functional Restoration Assocs., 19 S.W.3d 393, 398 (Tex.2000) ("Each provision must be construed in the context of the entire statute of which it is a part."); Bridgestone/Firestone, Inc. v.

Glyn-Jones, 878 S.W.2d 132, 133 (Tex.1994) ("Only in the context of the remainder of the statute can the true meaning of a single provision be made clear.”). . Attorney's fees resulting from appeals of an award of supplemental income benefits are also addressed in § 408.221, and are not under consideration here. See Tex. Lab.Code § 408.221(b), (c), (i); see also id. § 408.147(c) (mandating payment of employee’s attorney’s fees if a carrier unsuccessfully disputes a supplemental income benefits award). .

Fee awards for representing a claimant before the Division of Workers' Compensation — not at issue here — are set forth in the Texas Administrative Code. See 28 Tex. Admin. Code § 152.1. . See Act of May 25, 2001, 77th Leg., R.S., ch. 1456, § 8.01, 2001 Tex. Gen. Laws 5167, 5189 (codified at Tex. Lab.Code § 408.221). . See generally Act of Dec. 13, 1989, 71st Leg., 2d C.S., ch. 1, § 4.09, 1989 Tex. Gen. Laws 1, 34-35 (embodying what is now codified in § 408.221 as it existed prior to the 2001 amendment adding subsection (c)).