joined by Justice MEDINA, concurring.

I agree with the Court that the evidence is legally insufficient to support a malicious credentialing verdict against the Hospital. The peer-review privilege prevented the Romeros from knowing what actions the Hospital took or failed to take to protect the public from a physician whose own former chief of staff, a member of the credentialing committee, thought was a menace to patients. While I fully join the Court’s opinion, I write separately because I am deeply troubled by the head-in-the-sand approach the various hospitals and health-care professionals in this case appeared to take in dealing with a drug-impaired physician. Unless health-care institutions and providers are in fact, rather than theory, vigilant and proactive in performing the critical competence analysis that the peer-review privilege was intended to promote, the purposes that prompted the privilege’s creation will prove to be illusory. Clearly, the privilege’s purposes were not served in this case.

As the Court notes, during Dr. Baker’s initial credentialing and while he maintained privileges at the Hospital before Mr. Romero’s grievous injury, the Hospital should have learned from its own sources and various others that Baker (1) had been sued ten times within an approximate five-year period, (2) was a suspected drug addict, (3) had improperly cared for and treated at least four named patients, and (4) was suspended from another hospital for operating on the wrong leg of a patient, a mistake he had made before. Dr. Ronald Kerr, the Hospital’s chief of staff and a member of its executive committee, testified that, based on what he had heard and been told, he had formed the view that Baker presented a safety risk to patients. But Kerr relied on the State Board of Medical Examiners to investigate Baker’s alleged drug use. Baker’s former office manager testified that Baker constantly displayed erratic moods and engaged in other behavior that should have prompted the health professionals around him to take action to protect his unsuspecting patients. According to the record, there were numerous warning signs, but there appears to have been a reluctance to share vital information.

The purposes that underlie the peer-review privilege are commendable and, as our opinion today again illustrates, the protection the privilege affords is strong. The privilege was designed to foster uninhibited and “exacting critical analysis of the competence and performance of physicians and other health-care providers” to improve standards of medical care. Mem’l Hosp.-The Woodlands v. McCown, 927 S.W.2d 1, 3 (Tex. 1996). Here, though, the Hospital seems to view the privilege as a shield to protect itself from injured patients rather than a vehicle for improving patient care; its brief castigates the Romeros for “attempting] to get around *232the privilege with bits and pieces of information, rumors, innuendo, gossip, and second-hand information,” even though the privilege left the Romeros no other option.

Certainly hospitals should be wary of interfering with a doctor’s practice based merely upon rumor or innuendo, but neither should they look the other way and refuse to heed indications of danger. It has been noted that drug- and alcohol-impaired physicians are a growing threat to patients in this country, and the medical community’s will to adequately self-police is increasingly the subject of public criticism. See, e.g., Thompson, Special Treatment: Disciplining Doctors: Medical Boards Let Physicians Practice Despite Drug Abuse, Wash. Post, ApR. 10, 2005, at Al; Sunset Advisory Commission, Texas State Board op Medical Examiners, Texas State BoaRd op Physician Assistant Examinees, Texas State Board of Acupuncture Examiners, Staff Report 1 (Oct. 2004). The Sunset Advisory Commission’s recent report evaluating the Board of Medical Examiners noted that the use of private rehabilitation orders does not protect the public, and recommended that private orders not be used when physicians have violated the standard of care. Sunset AdvisoRy Commission, supra, at 47. But the same report also noted that the Board of Medical Examiners often has difficulty enforcing violations of the Medical Practice Act because the peer-review privilege is frequently asserted in contested-case hearings. Id. at 41.

The Legislature’s primary purpose in conferring the peer-review privilege on health-care institutions was to enhance the quality of medical care by encouraging forthright and thorough analysis of providers’ competence. And the primary purpose in creating exceptions to the privilege that allow disclosure to other medical peer-review committees, appropriate.state or federal agencies, and national accreditation and state-licensing bodies, was presumably to encourage the free exchange of information between them without losing the protections thé privilege affords. Tex. Oco.Code § 160.007(c). When doctors and hospitals fail to engage in the free exchange of information that the privilege was designed to promote, the Legislature’s purpose is thwarted and the privilege’s underpinnings erode. Should such erosion become pervasive, the privilege deserves to be swept away.

delivered the opinion of the Court, in which

This is an action against a hospital for negligently delaying a blood transfusion for the plaintiff while he was in surgery, and for malicious credentialing of the surgeon. The trial court rendered judgment on a verdict for the plaintiff on both claims. As the case comes to us, it raises two principal issues, each requiring a bit of explanation at the outset.

The first issue is whether there is any clear and convincing evidence that the hospital acted with malice in credentialing the surgeon — that is, in initially- granting him privileges to practice in the hospital, and then allowing him to retain them. In Texas, by statute, a hospital is not liable for improperly credentialing a physician through its peer review process unless the hospital acts with malice,1 defined (for the time period involved in this case) as actual awareness of, yet conscious indifference to, an extreme risk.2 Proof of malice is made more difficult in this setting because peer review communications and proceedings are generally confidential and privileged *215from disclosure.3 Since the Rules of Evidence prohibit drawing any inference from a claim of privilege in a civil case,4 a phintiff must prove that a hospital acted Thaciously without access to evidence of whathappened, or did not happen, in the eredenialing process. The jury in this case was instructed, at the plaintiffs’ request, that they could find malice only from clear md convincing evidence, and they did so. But the court of appeals determined that there was no evidence of malice.5 We ecnc^ucie that there was no clear and convfr'P^g evidence of malice.

T- a question tliPfr becomes whether judgment can rest on thd jury’s negligence funding. The answer turns on the second issue: whether it was reversible error to allo>w the jury, in apportioning responsibility 'fiar Abe plaintiff’s injuries among the hospital, two pihysiciaas, and a nurse, to consider the hospiifefs alleged malicious credentialing, of which we conclude there was no evidence, along with the hospital’s negligent delivery of blood to the operating room, for which the hospital does not challenge liability here. The jury could logically have thought the hospital responsible to a lesser degree had they been permitted to consider only the hospital’s negligence. The apportionment question, was submitted to the jury in broad form, as required by Rule 277 of the Texas Rules of Civil Procedure “whenever feasible”.6 But broad-form submission cannot be used to put before the jury issues that have no basis in the law or the evidence.7 The significant benefits of broad-form submission neither necessitate nor justify misleading the jury with legally or factually invalid claims. We agree with the court of appeals that the submission of the apportionment question was error, that it probably caused the rendition of an improper judgment, and that complaint of the error was preserved for appeal.8

The court of appeals reversed the judgment of the trial court and remanded the case for a new trial on the negligence claim against the hospital. We affirm.

I

Dr. Merrimon Baker applied to Columbia Kingwood Medical Center in February *2161993 for privileges to practice there as an orthopedic surgeon. Columbia (the business name used by respondent KPH Consolidation, Inc.) is an acute care facility in Kingwood, on the northeast side of Houston, with 155 beds and 360 doctors on staff. Baker had been practicing nearby, at what is now the Cleveland Regional Hospital, just north of Kingwood, where he had been recruited to come from South Carolina in 1989, several months after he had completed his residency and entered private practice. Besides being licensed in Texas and South Carolina, Baker was also licensed in Mississippi, although his application there had initially been denied because he had forged his partner’s and office manager’s names to it.

Baker’s application was considered by Columbia’s peer review committee, the Medical Executive Committee, comprised of doctors on Columbia’s staff whose responsibility it was to determine the qualifications and review the performance of all physicians permitted to practice there. As already noted, Texas law provides that with certain exceptions, the proceedings and records of a peer review committee are confidential and communication's to it are privileged from disclosure.9 The exceptions, insofar as they concern us here, allow peer review committees to share information with each other and with appropriate state and federal agencies, national accreditation bodies, the Texas State Board of Medical Examiners, and other states’ licensing boards.10 Columbia has asserted that privilege in this case". Thus, the record is silent concerning what information the Committee did or did not obtain regarding Baker, and any deliberations it may or may not have had, not only in connection with his original applkWon, but over the more than four years †⅜ he remained at Columbia. From this dui-co, the rules of evidence permit nojiferenee whatever to be drawn.11 Wo do know, however, that the Committee ,yas charged with thoroughly and carefully¡11Véstigating each physician who appE for privileges at Columbia. This invu ~e'i obtaining extensive information from ⅞ v physician contacting licensing age¿leij; health facilities, and o&er physicians w.vo ¾⅞1⅛ have information about tlm applicant, . well as the National Wactitiower Date Bank, to wfeh information regarding the professional eompr ’-neo and/ conduct of physicians ⅛ retired by federal regulation to béReprp ⅛-d,12 and law enforcement agencies, including the federal Drug Enforcement Agency. We also know that after a physician was admitted to practice at Columbia, the Committee was responsible for constantly monitoring and reviewing his or her practice.

If the Committee’s initial investigation of Baker was thorough, it should have revealed that from 1988 to 1993 he had been sued ten times for malpractice. The Committee should have inquired how these suits were resolved, and specifically, whether resolution involved a payment of money. Our record does not reflect how each case was resolved, although it does indicate that’ some cases were resolved favorably to Baker. One lawsuit alleged that in 1990 at Cleveland he had operated *217on the wrong hip of a patient. That suit was settled and dismissed six months after it was filed.

Columbia granted Baker’s application in February 1994. As with all doctors accepted at Columbia, Baker’s privileges were initially provisional and subject to further review. Baker continued to work at Cleveland, where most of his practice was centered. He also had privileges at two other hospitals in the vicinity.

In November 1994, Baker’s office manager, Janet Pickett, told the physician with whom Baker was then associated, Dr. Dan Parkinson, that she suspected Baker was abusing a prescription drug, hydrocodone, marketed as Lortab and Vicodin. (The evidence in this case is that hydrocodone is a pain killer which, even when taken as prescribed, can cause drowsiness, mental clouding, lethargy, impairment of mental and physical performance, anxiety, fear, dysphoria, psychic dependence, and mood changes.) Pickett had seen hundreds of empty drug sample containers in and around Baker’s office and private lavatory and had noticed that he experienced serious mood swings. Parkinson had lived in Baker’s residence for a short time earlier in the year and had not noticed that Baker used drugs, but he had since heard that Cleveland had required Baker to submit to random drug testing. Parkinson related Pickett’s concerns to Cleveland’s chief of surgery. He also told Dr. Ronald Kerr, a physician at Columbia with whom he was hoping to associate, that he had secondhand information Baker was abusing drugs. Baker had taken patients from Kerr and had refused to pay Kerr his part of the rent on the office they shared while working at Cleveland, and Kerr did not like him. Kerr also did not approve of Baker’s treatment of patients, and even thought Baker posed a danger to them, although Kerr’s criticism tended to be general rather than specific. Parkinson indicated that he intended to convey Pickett’s concerns to the Texas State Board of Medical Examiners, and Kerr agreed he should.

A few weeks later, Baker’s wife and other family members confronted him over his drug abuse, and in February 1995, he voluntarily entered a treatment program, admitting that he suffered from “a pattern of chemical abuse”. He was released in May, but Pickett, who had supported Baker in obtaining treatment and remained his office manager until September or October, believed that he continued to use hy-drocodone. Some time in 1995, Pickett reported Baker to the Board of Medical Examiners, and so did Parkinson. In April 1996, the Board notified Baker that it was investigating “allegations of suspected substance abuse, [improper] care and treatment of [four named patients], and ... your recurring health-care liability claims.” As we have already noted, Cleveland and Columbia peer review committees were authorized to provide information to the Board and obtain information from it.

Nevertheless, in August 1996, Columbia removed Baker’s provisional status and gave him full privileges. At the time, Kerr was chief of staff and chairman of the Medical Executive Committee. In that role, he would have been expected to convey to the Committee whatever reservations he had about Baker. The record does not reflect what, if anything, Kerr told the Committee. Kerr testified at trial that if he had ever had concrete evidence Baker actually used drugs, he would have told the Committee, but he never had such evidence. Parkinson, too, would have been expected to convey his concerns to the Committee, especially since he had reported them to the Board of Medical Examiners. Again, the record does not reflect whether Parkinson communicated with the *218Committee, but he testified at trial that while he had been associated with Baker for about a year and had operated with him 20-30 times, he had never seen Baker impaired or under the influence of drugs. Based on personal observations, Parkinson thought Baker was an excellent surgeon.

The record is largely silent regarding, the period from August 1996, when Columbia granted Baker full privileges, to December 1997, when Baker separated from his wife. In January 1998, Baker’s office manager (Pickett’s successor) resigned, stating in a letter to him:

Dr. Baker, you really need to get help and I don’t think you have any intention of getting it and Í am not going to sit around while everything falls down around you. I do want you to know I am aware of your drug problem and have been aware of it for several years.

Then on May 15, 1998, Cleveland suspended Baker from practice there after he operated on the wrong leg of a patient (a mistake similar to the one he was alleged to have made in 1990). The record does not reflect that Columbia knew of the action taken by Cleveland. Dr. Robert Ro-sen, who was chief of staff at Cleveland and therefore chairman of its peer review committee, also served on Columbia’s Committee, but he testified that he abstained from discussions at Columbia about doctors at Cleveland, like Baker, so that the peer review processes at the two hospitals would remain separate. Furthermore, on June 12, Baker obtained a temporary restraining order prohibiting Cleveland and its agents from disclosing any information regarding his suspension. The TRO was extended and did not expire until July 16, the day after the surgery that gave rise to the claims in this case.

In early July, Baker met with his wife at home to try to finalize their divorce. While they were talking, he became increasingly agitated and suddenly grabbed her by the neck, attempting to strangle her. Equally suddenly, he released her moments later and ran sobbing from the house, threatening to kill himself. Baker’s wife described this erratic conduct as similar to the behavior he suffered while on drugs. She did not inform Columbia of his actions or drug abuse, then or before, because she believed it already knew of Baker’s problems.

On July 15, Baker performed elective back surgery on Ricardo Romero, a 40-year-old longshoreman, whom Baker had been treating for about a year. Baker was assisted by Dr. William Huie, an anesthesiologist. During the surgery, Romero lost a lot of blood before Baker or anyone else noticed, and in the 45 minutes it took to prepare a transfusion, he lost almost all of the blood in his body. As a result, Romero went into cardiac arrest, and though he was resuscitated, he suffered severe and permanent brain damage that has left him profoundly disabled and unable to care for himself. There is no direct evidence that Baker was under the influence of drugs during the surgery, but an expert witness for Romero, Dr. John Eichhorn, testified that for a surgeon to allow so. great a loss of blood during relatively routine back surgery was unheard of and was consistent with impairment.

Baker testified that Columbia suspended him after the Romero surgery, pending an investigation, but there was also evidence that he performed at least two surgeries at Columbia in September. His privileges were for a term of two years, which expired in August or September, and he did not apply to have them renewed.

A year later, on August 30, 1999, the Board of Medical Examiners notified Baker that it had decided to close its investigation “based on the information presently available.” But on September 7, Cleve*219land informed Baker of its final decision to suspend his privileges permanently because of violations that included “numerous delinquent medical charts, failure to make daily rounds for all patients as required ..., and failure to enter timely progress notes for all patients.” Cleveland’s letter did not state that these were the sole reasons for its action, but neither did it mention drug abuse or lawsuits as a basis for its decision.

Romero’s wife, individually and on behalf of Romero and their three minor children (“the Romeros”, petitioners in this Court), sued Columbia, Baker, Huie, and others, but settled with all of the defendants except Columbia for about $2,386,000. The plaintiffs alleged that Columbia’s negligence resulted in a delayed blood transfusion for Romero during surgery, and that it acted with malice in credentialing Baker to practice in the hospital. The jury found that Romero’s injury was caused by the negligence of Columbia, Baker, and Huie, and by the Hospital’s malice in credentialing Baker. The Rome-ros requested that the finding of malicious credentialing be made by clear and convincing evidence, and the trial court, after expressing doubt about whether the heightened standard was appropriate for the liability issue (as opposed to the punitive damages issues), acceded to the request. The jury was instructed that if they found liability under either theory, they should apportion responsibility for the injury, which they did: 40% to Columbia, 40% to Baker, and 20% to Huie. The jury found actual damages of $28.6 million and punitive damages of $12 million. Based on the verdict, the trial court determined awards for each plaintiff against Columbia that totaled a little over $25 million, including prejudgment interest, but under a high-low settlement agreement, the trial court rendered judgment for $11,440,000 in actual damages.

On appeal, Columbia complained that there was no clear and convincing evidence of malice, defined by statute as:

an act or omission:

(i) which when viewed objectively from the standpoint of the actor at the time of its occurrence involves an extreme degree of risk, considering the probability and magnitude of the potential harm to others; and

(ii) of which the actor has actual, subjective awareness of the risk involved, but nevertheless proceeds with conscious indifference to the rights, safety, or welfare of others.13

The court of appeals rejected Columbia’s argument that the heightened burden of proof the jury was instructed to apply— clear and convincing evidence, rather than a preponderance of the evidence — required a heightened standard of evidentiary review on appeal.14 The court thoroughly examined the evidence of malice and concluded that Baker posed an extreme risk to patients because of his drug abuse15 (but not because of the lawsuits filed against him16 or Kerr’s adverse opinion of him17). The court also concluded that testimony about the kind of investigation Columbia would have been expected to make of Baker’s conduct under its peer review procedures was evidence that it acted accordingly and sufficient circumstantial evidence that as a result it must have known *220of Baker’s drug abuse.18 But the court could find no evidence that Columbia was consciously indifferent to the risk Baker’s impairment posed to his patients.19 Because of the silence of the record surrounding Columbia’s peer review proceedings, the court concluded that the Romeros could point to no evidence that Columbia had failed to take appropriate measures in response to Baker’s drug abuse.20 The court noted that the Romeros’ expert witness had offered his opinion that Columbia should not have allowed Baker to operate on Romero but concluded that the opinion had no factual basis to support it.21

Having determined that the trial court’s judgment could not rest on the jury’s finding of malicious credentialing, the court of appeals next considered whether the jury’s finding of negligence would support the judgment.22 Although Columbia did not challenge the negligence finding, it argued that the jury' should not have been permitted to consider Columbia’s alleged malicious credentialing in apportioning responsibility. The court agreed, finding it simply “inconceivable that the jury would find liability based on the two different acts, but not attribute some responsibility to both acts.”23 Because the jury would probably have apportioned responsibility differently had it been restricted to considering only Columbia’s negligence, the court concluded that judgment could not be rendered on the verdict.24 Accordingly, the court remanded the case for a new trial on negligence.

We granted the Romero plaintiffs’ petition for review.25

II

We turn first to the Romeros’ contention that there is evidence of. malicious credentialing to support the trial court’s judgment. The Romeros argue, and the court of appeals agreed,26 that any evidence, more than a scintilla, will suffice. But while this case has been pending before us, we have held, as Columbia argued in the court of appeals and again here, “that in reviewing the legal sufficiency of evidence to support a finding that must be proved by clear and convincing evidence, an appellate court must ‘look at all the evidence in the light most favorable to the finding to determine whether a reasonable trier of fact could have formed a firm belief or conviction that its finding was true.’ ”27 Malice must by statute be proved by clear and convincing evidence as a predicate to the recovery of exemplary damages,28 but we know of no requirement that malice be proved by more than a preponderance of the evidence as an element of a malicious credentialing action to recover actual damages. To recover actual damages, the Romeros were required to prove malicious credentialing only by a preponderance of the evidence; to recover exemplary damages, they were held to the higher standard of proof for malice. The *221learned trial judge pointed out this distinetion and suggested that the standards of proof be different in the liability question and the exemplary damages question.29 But the Romeros requested that the jury find maMous eredentiahng by dear and convincing evidence in order to establish liability, and the trial court acceded. The sufficiency of the evidence must be measured by the jury charge when, as here, there has been no objection to it.30 Accordingly, we must hold the Romeros to the higher standard and review all of the evidence in the light most favorable to the verdict to determine whether a reasonable trier of fact could have formed a firm belief or conviction that Columbia acted with malice in credentialing Baker, although as we shall see, the Romeros cannot satisfy even the ordinary standard of review.

The parties do not dispute that a physician engaged in drug abuse presents an extreme risk to patients, and we will assume solely for purposes of argument, as the court of appeals concluded, that Columbia had actual, subjective awareness of the risk posed by Baker’s drug abuse, at least at one point in time.31 We focus, as did the court of appeals, on whether there is evidence that Columbia was consciously indifferent to this risk. Regarding what steps a hospital can and should take when confronted with such a risk, the Romeros’

expert, Dr. John Eichhorn, testified as follows:

q [by the Romeros’ counsel]. Are there steps that can be taken, Dr. Eich-bom, if the hospital suspects drug abuse or ⅛ w¿s 0f ⅞, ⅞⅛ has even an ounce of suspicion about drug abuse? Are there steps that the committees can take in order to determine, A, whether suspension should take place or, B, whether some kind of testing should take place?

A. Yes, definitely.

Q. What should have been done in this case in your opinion?

A. There are several parts to it. One of the classic components is called a confrontation where the subject under question is brought in under controlled conditions and confronted first with the suspicions, the concerns — safety concerns in particular, of course — and then asked at that moment to provide an on-the-spot witnessed urine sample — and better, and actually this is what I have done — sample at the same time so you have both simultaneously completely unannounced and random. That’s part one.

The other part is the investigation, and that includes: investigating with the other institutions, if any, where the physician has privileges; seeking, as we said *222before, peer-review-to-peer-review-committee information; launching their own investigation, starting with that.

On cross-examination, Dr. Eichhorn testified further:

Q [by Columbia’s counsel].... [Y]ou are familiar with different options open to hospital committees [when a physician is suspected of abusing drugs]?

A. Yes.

Q. Let me represent to you that these are some of the options Dr. Kerr discussed the other day with me that would be available to a hospital committee: monitoring, education, restrictions, termination, counseling, drug testing. Would you agree that those are some of the available options?

A. Some but not all.

Because of the confidentiality surrounding peer review proceedings, the record is largely silent on which, if any, of the steps outlined by Dr. Eichhorn Columbia did or did not take. As we have already explained,32 we cannot infer from this silence that Columbia either did or did not take steps it should have taken with respect to Baker’s drug abuse. The Romeros complain that the court of appeals inferred from the record silence that, in their words, “the Hospital might have been doing something — ‘we do not know what’ [quoting a phrase from the court of appeals’ opinion] — to address the known, extreme risk posed by the surgeon.” The Romeros further argue that the court improperly concluded that an inference that Columbia exercised “some care” precluded a finding that it was consciously indifferent. We think the Romeros misconstrue the court of appeals’ opinion, which said:

Because the Hospital invoked its con-fidentialb; privilege, we do not know what the Hospital did in response to the information it had. We do not know if it took any steps such as requiring Baker to submit urine samples, or monitoring Baker, or if it did nothing. But one thing we do know: we cannot infer anything from this lack of information.33

We do not read the court of appeals’ opinion to draw any inferences favorable to Columbia from the record’s silence.

The record does establish, of course, that Columbia did not suspend Baker before Romero’s surgery, but the evidence we have just quoted above does not indicate that Columbia should have done so; to the contrary, Dr. Eichhorn plainly testified that a hospital should consider a number of measures before, and as an alternative to, suspension. The Rome-ros argue that Dr. Eichhorn’s testimony relates only to suspected drug abuse, not proven, admitted drug abuse. We are not convinced that Dr. Eichhorn’s testimony can fairly be read so narrowly, but if it could, there is nothing in the record to show that Columbia knew in July 1998 that Baker had continued to use drugs since having been reported to the Board of Medical Examiners in 1995 or early 1996. More importantly, even if we disregard Dr. Eichhorn’s testimony that a hospital can and should take steps short of suspension in dealing with a physician engaged in drug abuse, the only evidence that Columbia should have suspended Baker before Romero’s surgery is in two brief passages of Dr. Eichhorn’s testimony. The first occurred during his direct examination:

Q. Was Dr. Baker still on the staff at the time of the Romero surgery?

A. Yes, definitely. Obviously. He operated that day.

*223Q. He had been suspended from Cleveland a couple months before?

A. Yes, he had.

Q. Is that what [Columbia] should have done in this case?

A. Yes, definitely.

But there is no evidence that Cleveland suspended Baker for drug abuse. Cleveland suspended Baker after he operated on the wrong leg of a patient and more than a year later gave as its reasons “violations [of Baker’s agreement with Cleveland] including] numerous delinquent medical charts, failure to make daily rounds for all patients as required by the Agreement, and failure to enter timely progress notes for all patients.” If Dr. Eichhorn meant that Columbia should have suspended Baker for the same reasons Cleveland suspended him, there is no evidence that Columbia had any of those reasons — that is, no evidence that Baker’s charts at Columbia were delinquent, or his rounds irregular, or his progress notes tardy. If Dr. Eichhorn meant that Columbia should have suspended Baker simply because Cleveland did, there is no evidence that Columbia knew of Cleveland’s action in May 1998, or even that it could have known. The undisputed evidence is that peer review proceedings at Cleveland would have been kept confidential and separate, just as they were at Columbia, and that to ensure such confidentiality, Baker obtained a temporary restraining order prohibiting Cleveland from disclosing its actions. That order did not expire until the day after Romero’s surgery. And if Dr. Eichhorn meant that Columbia should have suspended Baker whether it knew what Cleveland had done or why, simply because Baker should have been suspended, then he offered no support for his opinion. “We have reiterated that ‘a claim will not stand or fall on the mere ipse dixit of a credentialed witness.’ ” 34

The only other testimony regarding whether Columbia should have suspended Baker before Romero’s surgery occurred during Dr. Eichhorn’s cross-examination as follows:

Q. You’ve indicated in this case in response to [the Romero’s counsel’s] questions you thought that my client, [Columbia], was — acted with malice—

A. Using the legal definition on the printout, yes.

Q. I wasn’t quite done with my question, but let me ask it. Your opinion was that we acted with malice to allow Dr. Baker to practice medicine at the hospital on the day of Mr. Romero’s surgery?

A. Correct.

The parties have not cited the direct testimony referred to, and we have not found it ourselves. Here, as before, Dr. Eichhorn offered no support for his opinion.

There is no other evidence that Columbia should have suspended Baker to prevent him from operating on Romero. The Romeros argue that Columbia should have required that another physician monitor Baker’s surgery, that anesthesiologists assisting him be warned of his impairment, and that Romero himself be warned of the risk. The Romeros cite no evidence that Columbia should have done any of these things. The Romeros argue that no expert testimony was needed to prove that a *224hospital should take such actions or suspend a physician impaired by drug abuse, but we disagree. The necessity for expert testimony regarding the proper responses by a hospital to concerns of physician drug abuse is demonstrated by the Romeros’ own expert witness, Dr. Eichhorn, who testified extensively on the subject. As he stated, Columbia had a number of alternatives in dealing with Baker, and as we have noted, the record is silent regarding what measures, if any, Columbia actually took. Moreover, the Romeros have not pointed to evidence that Columbia knew Baker was impaired in July 1998, or for many months prior.

Lastly, the Romeros argue that the court of appeals should have considered evidence of Baker’s incompetence unrelated to his drug abuse — specifically, the lawsuits against him, the two wrong-limb surgeries, and Kerr’s opinion of him — in assessing evidence of malice. The court of appeals concluded that this evidence did not show that Baker posed an extreme degree of risk to his patients and consequently disregarded it before reaching the issue of malice.35 We think the court’s analysis was correct, but even if evidence of Baker’s incompetence were to be considered in assessing the evidence of malice, the conclusion would be the same: there is no evidence by Dr. Eichhorn or anyone else that questions about Baker’s competence required Columbia to keep him from operating on Romero as he did.

The court of appeals did not conclude, as the Romeros argue, that the mere possibility or inference that Columbia might have done something precluded a finding of malice. Rather, the court concluded that it could not infer from the lack of information that the hospital did nothing, and that there was no evidence “that when the Hospital discovered Baker’s substance abuse, it should have refused Baker active staff privileges in 1996 or withdrawn them at some point later.”36 We examined the Romeros quite carefully on this point at oral argument, and it was a subject of their post-submission brief. As proof of Columbia’s conscious disregard of Baker’s drug abuse, the Romeros have pointed to no evidence other than what we have cited, and we have found none in our own review of the record. We are inclined to agree with the court of appeals that Dr. Eich-horn’s two unsupported statements of opinion .are legally insufficient to show that Columbia was consciously indifferent to the risk Baker’s -drug use posed to patients. But as we have explained, the evidence of malice in this case must meet a higher standard; it must produce firm belief or conviction, and it does not do so.

Accordingly, we conclude that the court of appeals was correct in holding that there was no evidence of malice to support a judgment against Columbia for malicious credentialing. The Romeros complain that the peer review privilege effectively precludes recovery for malicious credentialing,37 but that complaint in this case, at least, is overstated. Had the Romeros offered evidence that Columbia should not have allowed Baker to operate on Romero, there would be some evidence of malice, and had such evidence been convincing, it would support recovery. We do not doubt *225that such evidence of malice is difficult to come by, but the Legislature has made recovery for improper eredentialing of physicians difficult. Indeed, since this case was tried, the Legislature has amended the definition of malice to mean “a specific intent by the defendant to cause substantial injury or harm to the claimant”, eliminating the alternative definition of conscious indifference to a known, excessive risk.38 Thus, the Legislature has made it still more difficult to establish liability for improperly eredentialing a physician.

Ill

We now consider whether the judgment for the Romeros can rest on the jury’s finding of negligence unchallenged on appeal. The court of appeals concluded that reversal and remand was required because of error in the jury question regarding the apportionment of responsibility.39 Because the jury was not asked to find the percentage of fault due solely to Columbia’s negligence, the portion of the damages for which it is liable cannot be determined. The Romeros argue that submission of the apportionment question was not reversible error; that Columbia did not preserve its complaint to the question; that to require a different submission is inconsistent with the rule that jury questions be submitted in broad form whenever feasible; and that retrial, if necessary, should be limited to the apportionment issue. We consider each argument in turn.

A

The jury was instructed to apportion responsibility for Romero’s injury if they found more than one person negligent, as they did. The jury was then asked:

What percentage of the conduct that caused the occurrence or injury do you find to be attributable to each of those found by you, in your answer to Question No. 1 and/or 2 to have caused the occurrence of injury?

In answer to Question 1, the jury found that Romero’s injury was caused by the negligence of Columbia, Baker, and Huie. In answer to Question 2, the jury found that Columbia’s malicious eredentialing of Baker also caused Romero’s injury. Thus, the jury was instructed to apportion responsibility among Columbia, Baker, and Huie, and in doing so, to consider Columbia’s malicious eredentialing of Baker. Since there was no evidence of malicious eredentialing, the jury should not have been allowed to consider that claim in setting Columbia’s percentage of responsibility.

Rule 44.1(a) of the Texas Rules of Appellate Procedure provides that error in the trial court properly complained of requires reversal of the judgment on appeal if it “(1) probably caused the rendition of an improper judgment; or (2) probably prevented the appellant from properly presenting the case to the court of appeals.”40 Had the jury been confined to considering Columbia’s role in causing Romero’s injury by negligently delaying a blood transfusion for him during surgery, it may well have decided on a figure other than 40%. Indeed, the court of appeals found it “hard to believe that the 40% liability the jury attributed to [Columbia] in question 3 was not based (1) partly on the liability it found for negligence (question 1), and (2) partly on the liability it found for malicious cre*226dentialing (question 2).”41 “[W]e do not have to guess if the erroneously submitted question impacted the liability or damages answers,” the court explained. “We know with almost certainty that it did impact the answers.” 42 Thus, the court held that the error in the apportionment question was reversible under Rule 44.1(a)(1).

The difficulty with the court of appeals’ analysis of whether error in the apportionment question probably resulted in an improper verdict, the concern under Rule 44.1(a)(1), is that it assumes that the jury has determined comparative fault percentages with both claims against Columbia in mind. The Romeros do not attempt to argue with the court of appeals’ reasoning that the jury, once having found Columbia to have been 40% at fault because of both its negligence and its malicious credentialing, would thereafter almost certainly have answered with a different figure had they then been asked to reapportion responsibility considering only Columbia’s negligence. This, they complain, is not the relevant issue. Rather, they argue, had the jury been instructed in the first instance to consider only evidence of Columbia’s negligence, they could easily have assessed Columbia’s proportionate responsibility at the same 40% figure. Because there is evidence entirely apart from the malicious credentialing claim to support the jury’s finding that Columbia was 40% at fault, the Romeros argue,-any error in submitting the apportionment question cannot be said to have probably resulted in an improper judgment.

We rejected this argument in Crown Life Insurance Co. v. Casteel43 and Harris County v. Smith,44 based on Rule 44.1(a)(2). Even if the jury could still have made the same apportionment of fault, the error in the question is nevertheless reversible because it effectively prevents Columbia from complaining on appeal that they would not have done so. Thus, in Crown Life Insurance Co. v. Casteel, we held that “submitting invalid theories of liability in a single broad-form jury question is harmful error when it cannot be determined whether the jury based its verdict on one or more of the invalid theories.”45 Similarly, in Harris County v. Smith, we held that a broad-form question about damages cannot include elements for which there is no evidence.46 In Harris County, we explained:

In Casteel, we reaffirmed our reasoning in Lancaster v. Fitch, 112 Tex. 293, 246 S.W. 1015 (1923), where this Court recognized the inherent harm to the administration of justice caused by mixing valid and invalid liability theories in a single broad-form liability question. Casteel, 22 S.W.3d at 389. The same year we decided Lancaster, we applied its reasoning to a similar situation involving a broad-form damages question. See Eastern Tex. Elec. Co. v. Baker, 254 S.W. 933, 934-35 (Tex. 1923). In Eastern Texas Electric, the trial court submitted a single broad damage issue and instructed the jury to consider past and future mental and physical pain in awarding damages even though there was no evidence of future physical pain. The court of appeals held that the trial court had erred in instructing the jury to consider future pain but concluded that the error was harmless because evidence of past mental and physical *227pain was sufficient to support the award. This Court reversed, concluding that the harmless error standard “was not intended to deprive a party to a suit of a substantial right,” namely, “the right to have the damages assessed against it by the jury under proper instructions submitting only the elements of damage as raised by the pleadings, and supported by evidence.” Id. at 934. Because it was “not possible for an appellate court to say the jury did not consider this eiToneous charge in arriving at the amount of damage,” this Court reversed and remanded for a new trial, citing Lancaster. Id. at 935.

Just as in 1923, a litigant today has a right to a fair trial before a jury properly instructed on the issues “authorized and supported by the law governing the case.” Casteel, 22 S.W.3d at 389 (quoting Lancaster v. Fitch, 246 S.W. at 1016). We conclude that the trial court erred in overruling Harris County’s timely and specific objection to the charge, which mixed valid and invalid elements of damages in a single broad-form submission, and that such error was harmful because it prevented the appellate court from determining “whether the jury based its verdict on an improperly submitted invalid” element of damage. Casteel, 22 S.W.3d at 388; see also Tex.R.App. P. 61.1(b).47

The argument was made in Harris County that even if it is reversible error to include legally invalid claims with legally valid ones in a single jury question, the same rule should not apply when all the claims are valid but some lack support in the evidence.48 While the jury might well be misled by legally erroneous instructions or questions, since they are not expected to know the law and are instead obliged to follow the law given them in the charge, they are certainly expected to know and weigh the evidence — and the argument goes — are therefore not likely to be influenced in making their findings by being allowed to consider factors without eviden-tiary support.49 We specifically rejected this argument,50 and this case illustrates why. Having found malicious credentialing, the jury could not conceivably have ignored that finding in apportioning responsibility. While in other instances a jury may simply ignore a factor in the charge that lacks evidentiary support, there are other — instances and this case is one — where the jury is as misled by the inclusion of a claim without evidentiary support as by a legally erroneous instruction. In all circumstances in which “[a] trial court’s error in instructing a jury to consider erroneous matters, whether an invalid liability theory or an unsupported element of damage, prevents the appellant from demonstrating the consequences of the error on appeal”,51 the same analysis must be applied.

We do not hold that the error of including a factually unsupported claim in a broad-form jury question is always reversible. Rule 44.1(a)(2) requires that the error, to be reversible, “probably prevented the appellant from properly presenting the case to the court of appeals.”52 But unless the appellate court is “reasonably certain that the jury was not significantly influenced by issues erroneously submitted *228to it”,53 the error is reversible. We have no such reasonable certainty here; on the contrary, we are reasonably certain that the jury was significantly influenced by the erroneous inclusion of the factually-unsupported malicious credentialing claim in the apportionment question. Accordingly,-we conclude that the error requires reversal of the judgment.

B

The Romeros argue that Columbia did not preserve its complaint to this error. The record we have of objections to the jury charge suggests that the trial court and counsel discussed the charge the day before without a court reporter present. It is clear that the learned trial judge identified the problem with the apportionment question and suggested that it could be avoided by submitting two such questions to the jury. The next day, on the record, Columbia objected to the submission of a question on malicious credentialing “on the grounds there is no evidence to warrant its submission, no evidence of malice, and no evidence under the clear and convincing standard of malice.” The court overruled the objection. Columbia then objected to the submission of a single apportionment question, although it declined to request two apportionment questions, as the court had suggested and to which the Romeros had objected. The record reflects the following:

[Columbia’s counsel]: In Question No. 3, we object to ... the inclusion of the Question No. 2 inquiry, ... what we believe is a legally non-viable theory— which is the malice issue — ... along with a negligence theory resulting in a single percentage inquiry, which, of course, as a result of [Casteel] would basically make it impossible to determine that there was a legally legitimate basis upon which rendition of judgment could be had.

THE COURT: That’s why I wanted to submit a separate percentage question for you.

[Columbia’s counsel]: Well, I understand that, Your Honor.

THE COURT: No one wanted it.

[Columbia’s counsel]: We think that’s equally inappropriate as particularly a comment on the weight. But nonethe: less, to the extent we have to make such an objection, the objection to Question No. 3 is ... that ... the predicate language ... includes Question No. 2. And assuming that there is an answer to Question No. 2, then Question No. 3 will result in a number which we believe will not be capable of ... legitimately supporting the rendition of an appropriate judgment. And for that reason, we object to the — what in essence is the predicate language....

THE COURT: I’d be glad to cure that for you.

[Columbia’s counsel]: Your Honor, we discussed this yesterday afternoon, and I think at that time I indicated that I did not want to sandbag the court.

THE COURT: Now, I understand you’re making the objection, but you know, it is my belief that there ought to be two predicate questions.

[Columbia’s counsel]: Two causation questions?

THE COURT: Two Question No. 3 questions.

[Columbia’s counsel]: Two percentage questions?

*229THE COURT: Two percentage questions. And that would cure the problem.

[Columbia’s counsel]: I understand the court’s position.

THE COURT: Okay.

[Columbia’s counsel]: We also believe that is inappropriate, but be that as it may, I understand the court did tender to us two percentage questions.

[The Romeros’ counsel]: Your Honor, I’m not sure I understand the objection, because although the court suggests having two percentage questions, I understand [Columbia] to be telling the court that they would object to that question, but they also object to this question—

THE COURT: Right.

[The Romero’s counsel]: — under [Cas-teel] because it supposedly includes an improper—

THE COURT: Yes, but you objected to two percentage questions, too. I just want that to be clear on the record.

The Romeros argue Columbia’s objection did not state its complaint to the submission of a single apportionment question “with sufficient specificity to make the trial court aware of the complaint”,54 as required by Rule 33.1(a)(1)(A) of the Texas Rules of Appellate Procedure. Their argument is not that Columbia’s objection was unclear or that it was not understood by the trial court; rather, it is that Columbia did not specify that it wanted two apportionment questions instead of one. Columbia stated to the trial court that the submission of two questions was inappropriate as a “comment on the weight”. Since the submission of two questions would have cured the problem of which Columbia now complains, the Romeros argue that Columbia’s objections lacked specificity and therefore did not preserve their complaint for appeal.

But the Romeros’ argument simply ignores the fact that Columbia’s objection to the malicious credentialing question was correct, and had the trial court sustained it, there would have been no problem with the apportionment question. The overruling of that objection created the problem in the single apportionment question that the Romeros requested, to which Columbia also objected, also correctly. No more was required of Columbia to preserve its complaints.

We need not consider whether Columbia was required to object not only to the lack of evidence for the malicious credentialing claim but also to the form of the apportionment question that included the claim55 because it did both. Also, neither the Romeros nor Columbia argues that the error preservation question is answered by Rule 278 of the Texas Rules of Civil Procedure, and so we do not consider the impact of that rule here.56 We conclude that Columbia preserved its complaint for appeal.

*230C

The Romeros argue that to apply the rule of Casteel and Harris County in this case — encourages, if not requires, separate submission of every theory of liability, every combination of theories, and every combination of defendants together with separate apportionment and damages questions for every theory, combination of theories, and combination of defendants. In this case, the jury charge would have needed to include more than 175 issues.

This is simply untrue. The jury charge in this case needed one less question — the question on malicious credentialing, for which there was no evidence — to be free of error, and reversal could have been avoided with one more question, which the trial court offered to the Romeros and they rejected. The reversible error rule of Cas-teel and Harris County neither encourages nor requires parties to submit separate questions for every possible issue or combination of issues; the rule does both encourage and require parties not to submit issues that have no basis in law and fact in such a way that the error cannot be corrected without retrial. If at the close of evidence a party continues to assert a claim without knowing whether it is recognized at law or supported by the evidence, the party has three choices: he can request that the claim be included with others and run the risk of reversal and a new trial, request that the claim be submitted to the jury separately to avoid that risk, or abandon the claim altogether. The Rome-ros’ argument assumes that it is so commonplace to come to the end of a jury trial and have no idea what claims are still legally and factually valid that the only safe course to avoid retrial is to parse out every issue in a separate jury question. Nothing in our review of thousands of verdicts rendered by juries across the State suggests that there is any validity to the assumption.

The Romeros argue that we have retreated from the mandate of Rule 277 of the Texas Rules of Civil Procedure that issues must be submitted to a jury in broad form “whenever feasible”.57 This Court’s adoption of broad-form jury submissions was intended to simplify' jury charges for the benefit of the jury, the parties, and the trial court. It was certainly never intended to permit, and therefore encourage, more error in a jury charge. We continue to believe, as we stated in Harris County, that “[w]hen properly utilized, broad-form submission can simplify charge .conferences and provide more comprehensible questions for the jury.”58 But “it is not always practicable to submit every issue in a case broadly,” 59 and broad-form submission cannot be used to broaden the harmless error rule to deny a party the correct charge to which it would otherwise be entitled.

D

Finally, the Romeros argue that any error in the apportionment question did not affect the jury’s findings of negligence or actual damages, and therefore if there must be a new trial, it should be limited to the apportionment issue. Columbia responds that a new trial of the negligence claim is necessary because the jury’s malicious credentialing finding tainted not only their apportionment findings but their damages findings. Columbia is certainly correct with respect to the puni*231tive damages findings, and for that reason alone, a new trial on the entire negligence claim is required.60

[[Image here]]

For these reasons, the judgment of the court of appeals is

Affirmed.

Justice O’NEILL filed a concurring opinion, in which Justice MEDINA joined.

Justice BRISTER and Justice JOHNSON did not participate in the decision.