OPINION
ANDERSON, Chief Justice.
We granted this appeal to determine whether a psychiatrist owed a duty of care to protect a hospital nurse from the violent and intentional acts of a hospitalized mentally ill patient. If such a duty is owed, the next issue to be decided is whether the patient’s intentional conduct should be considered in determining comparative fault under
McIntyre v. Balentine,
833 S.W.2d 52 (Tenn.1992). The final issue is whether, after finding that the jury verdict as to fault is contrary to the weight of the evidence, the trial court may reallocate comparative fault in lieu of ordering a new trial.
The trial court determined that the psychiatrist in this case owed a duty of care to the nurse, and instructed the jury to consider the intentional conduct of the patient, a non-party, in determining the psychiatrist’s- comparative fault. The jury returned a verdict for the nurse in the amount of $1,186,000. It allocated the fault as 100 percent to the psychiatrist and zero percent to the patient. The trial court approved the jury’s verdict except as to the allocation of fault, and granted a new trial. The Court of Appeals affirmed, finding that a duty was owed, that the patient’s intentional conduct should be compared with the psychiatrist’s negligence, and that a new trial should have been granted.
We agree that the psychiatrist owed a duty of care because he knew or should have known that his patient posed an unreasonable risk of harm to a foreseeable, readily identifiable third party. We have also determined that the trial court erred in instructing the jury to compare the patient’s intentional conduct with the defendant’s negligence in allocating fault. We, however, consider the error harmless because the jury allocated 100 percent of the fault to the negligent defendant psychiatrist. Finally, although not applicable here, in view of our result we have decided that the trial court may not reallocate comparative fault after weighing the evidence as the thirteenth juror, but must instead grant a new trial. Accordingly, we reverse the Court of Appeals in part and affirm in part, and remand this case to the trial court for entry of a judgment on the jury’s verdict.
BACKGROUND
In March of 1993, the plaintiff, Emma Turner, a nurse at Hubbard Hospital in Nashville, was attacked and severely beaten by Tarry Williams, a psychiatric in-patient at the hospital. The defendant, Harold Jordan, M.D., was the attending psychiatrist.
Williams, who had been diagnosed as bipolar and manic, had been a patient at Hubbard on five prior occasions; three of these times he was found to be a danger to himself or others and was committed to the Middle Tennessee Mental Health Institute. On one occasion, in April of 1990, Williams tried to attack Dr. Jordan with a table leg, but hospital staff intervened.
On March 4, 1993, Williams was again admitted to Hubbard’s psychiatric ward and examined by a resident physician. Williams’s history indicated that he had not taken his prescribed lithium, which was used to control his bipolar disorder, for over a week. Williams also reported that he had met with “Gorbachev and Saddam Hussein” and that he had “classified information” about space flights and nuclear science. The resident physician determined that Williams had illogical and disorganized thinking, flight of ideas, grandiosity, and delusional thinking. Lithium was prescribed, which takes five to seven days to reach a therapeutic level.
The next day, on March 5, 1993, Dr. Jordan reviewed and approved the resident physician’s orders. He and members of a treatment team then attempted to interview Williams, who refused to cooperate and left the interview. The treatment team then dis
cussed the case for thirty to forty-five minutes, after which Dr. Jordan wrote:
This patient presents no behavior or clinical evidence suggesting that he is suicidal. He is
aggressive, grandiose, intimidating, combative, and dangerous.
We will discharge him soon by allowing him to sign out AMA [Against Medical Advice].
(Emphasis added). That evening, according to notes, Williams, although quiet and non-disruptive, had an “angry and hostile” affect. Around 11:30 p.m., after requesting a cigarette and asking the nurse, Emma Turner, about being discharged, Williams attacked Turner, inflicting severe head injuries.
Thereafter, Emma Turner sued Dr. Jordan for medical negligence, alleging he violated his duty to use reasonable care in the treatment of his patient, which proximately caused her injuries and damages. At trial, Dr. David Sternberg, a psychiatric expert witness, testified that Jordan’s failure to medicate, restrain, seclude or transfer Williams fell below the standard of care for psychiatrists. He explained:
The standard of care in a case like this requires, first, an evaluation of whether the patient is a danger to himself or others. And, indeed, Dr. Jordan determined, it seems to me from the record, both his deposition and from the records from the hospital, that the patient was, indeed, dangerous. Then the standard of care requires, if a patient is found, in fact, to be dangerous, that the patient be prevented from acting on that dangerousness; that staff be informed, of course, about the patient’s dangerousness; that the patient be medicated, if necessary, to prevent acting on the dangerousness, or be restrained or secluded; or that the patient be transferred to another treatment setting which could handle a patient who is of that severe dangerousness.
In his own defense, Dr. Jordan testified that he did not remember Williams or any information about his dangerousness prior to the attack on Emma Turner. He agreed that had he known about Williams’s prior dangerousness, he would have discharged him. However, Dr. Jordan’s discharge summary written after the incident said:
Realizing that this patient had been hospitalized on this issue before and exhibited some hostile and violent behavior and questioning the veracity of his statement that he was suicidal, we wrote an order indicating that [Williams] could be encouraged to sign out and be allowed to sign out on request. We
considered discharging him outright because of his history of violent behavior.
In addition, Linda Lawrence, nursing coordinator at Hubbard Hospital, testified that Williams’s past violent behavior, including the attempted attack on Jordan in 1990, had been discussed during the treatment team meeting on March 5,1998.
After the completion of the proof, the trial court instructed the jury on the law of comparative fault, and it provided the jury with a verdict form indicating it could allocate the fault, if any, between the alleged negligence of Dr. Jordan and the alleged intentional conduct of patient Williams.
The jury returned a verdict for the plaintiffs, Emma and Rufus Turner, allocating 100 percent of the fault to defendant Jordan. The trial court approved all of the jury’s verdict except the allocation of fault. As a result, it granted the defendant’s motion for new trial, but thereafter granted an interlocutory appeal. The Court of Appeals affirmed.
We granted the appeal to consider the important questions of duty, comparison of fault between a negligent actor and an intentional actor, and the trial court’s authority to reallocate fault in lieu of granting a new trial.
LEGAL DUTY
First, the defendant psychiatrist asserts that the lower courts erred in determining that he owed a duty of care to protect the plaintiff nurse the unforeseeable and uncontrollable acts of his patient. The nurse, however, argues that the psychiatrist had a duty of care to protect her from foreseeable risks of harm posed by his hospitalized mentally ill patient.
To determine whether a duty exists, we turn first to familiar principles of negligence enunciated by our earlier cases. A claim for negligence requires the following elements: (1) a duty of care owed by the defendant to the plaintiff; (2) conduct by the defendant falling below the standard of care amounting to a breach of that duty; (3) an injury or loss; (4) causation in fact; and (5) proximate or legal cause.
Bradshaw v. Daniel,
854 S.W.2d 865, 869 (Tenn.1993).
The existence of a duty is a question of law for the court which requires consideration of whether “such a relation exists between the parties that the community will impose a legal obligation upon one for the benefit of others—or, more simply, whether the interest of the plaintiff which has suffered invasion was entitled to legal protection at the hands of the defendant.”
Id.
at 870, quoting, W. Keeton,
Prosser & Keeton on the Law of Torts,
§ 37 at 236 (5th ed.1984). The imposition of a legal duty “reflects society’s contemporary policies and social requirements concerning the rights of individuals and the general public to be protected from another’s act or conduct.”
Id.
at 870.
In determining whether a duty is owed in a particular case, we have generally used a balancing approach consistent with principles of fairness.
See McClung v. Delta Square Ltd. Partnership,
937 S.W.2d 891, 901 (Tenn.1996) (summarizing our cases on “duty” component). In
McCall v. Wilder,
913 S.W.2d 150, 153 (Tenn.1995), we explained that “[a] risk is unreasonable and gives rise to a duty to act with due care if the foreseeable probability and gravity of harm posed by defendant’s conduct outweigh the burden upon defendant to engage in alternative conduct that would have prevented the harm.” Among the several factors which must be considered are
the foreseeable probability of the harm or injury occurring; the possible magnitude of the potential harm or injury; the importance or social value of the activity engaged in by defendant; - the usefulness of the conduct to defendant; the feasibility of alternative, safer conduct and the relative costs and burdens associated with that conduct; the relative usefulness of the safer conduct; and the relative safety of alternative conduct.
Id.
at 153.
See also McClung,
937 S.W.2d at 901.
In general, “the degree of foreseeability needed to establish a duty of care decreases in proportion to the magnitude of the foreseeable harm.”
Pittman v. Upjohn Co.,
890 S.W.2d 425, 430 (Tenn.1994).
Although we have generally held that a person has a duty to use reasonable care to refrain from conduct that will foreseeably cause injury to others,
Doe v. Linder Construction Co.,
845 S.W.2d 173, 178 (Tenn.1992), this duty does not extend to the protection of others from the dangerous conduct of third persons
unless
the defendant “stands in some special relationship to either the person who is the source of the danger, or to the person who is foreseeably at risk from the danger.”
Bradshaw,
854 S.W.2d at 871, citing, Restatement (Second) of Torts § 315 (1964). As we said in
Bradshaw,
“while an actor is always bound to prevent his acts from creating an unreasonable risk to others, he is under the affirmative duty to act to prevent another from sustaining harm only when certain socially recognized relations exist which constitute the basis for such legal duty.” 854 S.W.2d at 871, quoting, Harper & Kime,
The Duty to Control the Conduct of Another,
43 Yale L. J. 886, 887 (1934).
In
McClung,
for example, we joined the vast majority of jurisdictions in recognizing that a business has an affirmative duty to take “reasonable measures to protect their customers from foreseeable criminal attacks” if the business “knows, or has reason to know, either from what has been or should have been observed or from past experience, that criminal acts against its customers on its premises are reasonably foreseeable-”
937 S.W.2d at 901-902. The determination of whether a duty is owed requires a balancing of the foreseeability and gravity of the potential harm against the burden imposed in protecting against that harm.
Id.
at 902.
We have also employed this analysis in the context of a physician/patient relationship.
In
Bradshaw
we held that the physician/patient relationship was sufficient to impose an affirmative duty on the physician to warn identifiable persons in the patient’s family against foreseeable risks related to the patient’s illness. 854 S.W.2d at 872. Similarly, in
Wharton Transport Corp. v. Bridges,
606 S.W.2d 521, 526 (Tenn.1980), we held that a physician owed a duty to a third party who had been injured by a truck driver the physician had negligently examined and certified. In contrast, in
Pittman v. Upjohn Co., supra,
we held that a physician did not owe a duty to a non-patient where it was not reasonably foreseeable that the non-patient would take medication prescribed for the physician’s patient. 890 S.W.2d at 430.
These same principles apply in addressing whether a psychiatrist has a duty to protect a third party from the violent acts of a patient. In
Bradshaw
we cited with approval
Tarasoff v. Regents of University of California,
17 Cal.3d 425, 131 Cal.Rptr. 14, 551 P.2d 334 (1976), in which the California Supreme Court, citing Restatement (Second) of Torts § 315, held that a psychotherapist had an affirmative duty of care to protect a foreseeable third party from his patient who presented a serious threat of danger. The court explained that, depending on the nature of the ease, the duty of care may require warning the victim, notifying the police, or whatever other steps are reasonably necessary to protect the third party. 17 Cal.3d at 430, 131 Cal.Rptr. at 20, 551 P.2d at 340.
The majority of courts, applying
Tarasoff
principles, have held that where a psychiatrist, in accordance with accepted standards of the profession, knows or reasonably should know that a mentally ill patient poses an unreasonable risk of harm to a foreseeable third party, he or she must take reasonable steps to prevent that harm.
In
Naidu v. Laird,
for example, the Delaware Supreme Court explained the basis for imposing a duty in
Tarasoff
situations:
The special relationship which exists between mental health professionals and a patient provides the underlying basis for imposition of an affirmative duty owed by such professionals to persons other than the patient. That duty is to take whatever steps are reasonably necessary and available to protect an intended or potential victim(s) of the patient when the psychiatrist determines or should have determined, in keeping with the professional standards of the community, that the patient presents an unreasonable danger to that person(s).
539 A.2d at 1075.
Similarly, in
Hamman v. County of Maricopa,
the Arizona Supreme Court held that a psychiatrist owed a duty where he denied hospital admittance to a schizophrenic patient with a lengthy history of violent behavior who, the Court said, assaulted and severely injured his step-father:
We reject the notion that the psychiatrist’s duty to third persons is limited to those against whom a specific threat has been made. We hold that the standard originally suggested in
Tarasoff
is properly applicable to psychiatrists. When a psychiatrist determines, or under applicable professional standards reasonably should have determined, that a patient poses a serious risk of violence to others, the psychiatrist has a duty to exercise reasonable care to protect the foreseeable victim of that danger. The foreseeable victim is one who is said to be within the zone of danger, that is subject to probable risk of the patient’s violent conduct.
775 P.2d at 1127-28. Likewise, in
Perreira v. State,
the Colorado Supreme Court held that a psychiatrist owed a duty where a patient with a long history of mental illness and psychotic behavior was released from involuntary treatment and then shot a police officer. The court considered
the existence of a special relationship between a psychiatrist and an involuntarily committed mental patient and the resulting degree of control which the psychiatrist has over the patient as a result of that relationship; the foreseeability of harm to others from the failure of the psychiatrist to take protective action for the benefit of others; the social utility of the psychiatric decision to release an involuntarily committed patient; the magnitude of the burden of guarding against violent acts committed by an involuntarily committed mental patient subsequent to release; and the practical consequences of placing that burden upon the psychiatrist.
768 P.2d at 1214-15.
See also Petersen v. State,
671 P.2d at 237.
Here, the plaintiffs argue that Dr. Jordan had a duty of care because he knew or should have known that Williams posed an unreasonable risk and because the plaintiff, as a nurse on the psychiatric unit, was a foreseeable victim. The defendant maintains that no duty existed because Williams was a voluntary patient who did not specifically threaten the plaintiff or present an unreasonable or foreseeable risk of harm.
The Court of Appeals, like the trial court, found that a duty existed under the facts of this case. The intermediate court said: “Dr. Jordan’s duty to protect third persons from foreseeable bodily harm exist[ed] beyond those whom Mr. Williams specifically threatened to those persons who [were] members of a class of persons whose safety would, with reasonable foreseeability, be placed at risk by Mr. Williams’ uncontrolled actions.”
As
the Court of Appeals observed, the
Tarasoff
cases, like our decision in
Bradshaw, supra,
emphasized the presence of a special relationship, that is, the psychiatrist/patient relationship. The cases further consider the factors we have typically balanced in determining whether a duty exists: the foreseeability and severity of potential harm; the nature of the defendant’s conduct; and the availability, safety and effectiveness of alternatives.
See McCall,
913 S.W.2d at 153.
Applying these factors, we agree that Dr. Jordan, as Williams’s attending psychiatrist, owed a duty of care to the plaintiff, a nurse on the psychiatric unit. He knew of Williams’s prior violent conduct while hospitalized, including one occasion in which Williams attacked a member of the hospital staff (indeed, Jordan himself). Dr. Jordan was also well aware of Williams’s present dangerousness—he described Williams as “aggressive, grandiose, intimidating, combative, and dangerous.” Although this unreasonable risk of harm was reasonably apparent, Dr. Jordan, who had the ability to control Williams in the inpatient psychiatric ward, took no action other than to recommend Williams be encouraged to request discharge against medical advice. Although the defendant now contends that he had no control over Williams and that he was obligated to apply the least restrictive means of treatment, the record indicates that he never considered other reasonable measures to prevent the risk Williams posed to other patients, staff members, or other readily identifiable foreseeable victims.
We stress that we are not requiring psychiatrists or physicians to possess perfect judgment or a degree of clairvoyance in determining whether a patient poses a risk of harm to a third person. Instead, we merely hold that a duty of care may exist where a psychiatrist, in accordance with professional
standards, knows or reasonably should know that a patient poses an unreasonable risk of harm to a foreseeable, readily identifiable third person. The courts below correctly held that the facts in this case met this standard.
COMPARATIVE FAULT
Having determined that a duty of care exists in this case, we now turn to the issue of whether the defendant psychiatrist’s negligence should have been compared with the intentional act of the non-party patient Williams in determining the extent of the defendant’s liability to the plaintiffs.
The plaintiffs’ argument is twofold: a psychiatrist’s liability should not be reduced by the occurrence of a foreseeable act he had the duty to prevent;
and as a matter of practice and policy, the negligent act of a tortfeasor should not be compared to the intentional act of another tortfeasor. The defendant maintains that comparison is proper because it limits his liability to his percentage of fault in causing harm to the plaintiff.
In
McIntyre v. Balentine,
we adopted a modified form of comparative fault under which a plaintiff whose negligence is less than that of a defendant may recover damages in an amount reduced in proportion to the percentage of the plaintiffs own negligence. 833 S.W.2d at 57. Based on notions of fairness and justice, we abolished the outdated doctrine of contributory negligence and yet stressed that “a particular defendant [is] liable only for the percentage of a plaintiffs damages occasioned by that defendant’s negligence.”
Id.
at 58. Moreover, to provide guidance in future cases, we said that a defendant is permitted to show that a non-party caused or contributed to the damages for which the plaintiff seeks recovery.
Id.
Since
McIntyre,
we have clarified the distinction between comparative
negligence
and comparative
fault.
The former is the “measure of the plaintiffs negligence m percentage terms used for the purpose of reducing the plaintiffs recovery from the defendant.” The latter is defined as “those principles which encompass the determination of how to apportion damage recovery among multiple or joint tortfeasors according to the percentage of fault attributed to those actors after reduction for the plaintiffs percentage of negligence.”
Owens v. Truckstops of America,
915 S.W.2d 420, 425 n. 7 (Tenn.1996).
Accordingly, in determining comparative fault, we have considered cases in which the negligence of a tortfeasor was compared with the negligence of other tortfeasors.
Volz v. Ledes,
895 S.W.2d 677 (Tenn.1995);
Bervoets v. Harde Ralls Pontiac-Olds, Inc.,
891 S.W.2d 905 (Tenn.1994).
We
have also considered the question of comparing the negligence of a defendant with the strict liability of third-party defendants.
Owens v. Truckstops of America,
915 S.W.2d at 431-33. This case presents our first opportunity to determine whether the negligent act of a defendant should be compared with the intentional act of another in determining comparative fault.
Other jurisdictions have addressed the issue. In
Veazey v. Elmwood Plantation Assoc., Ltd.,
650 So.2d 712 (La.1994), the plaintiff was sexually assaulted by an intruder and filed a negligence action against her apartment complex for failing to maintain adequate security; the defendant apartment complex, in turn, defended on the basis of the intentional act by the assailant. The Louisiana Supreme Court declined to compare the negligent act of the defendant with the intentional act of the third party primarily because it believed the negligent defendant should not be allowed to reduce its fault by relying on an intentional act it had the duty to prevent.
Id.
at 718. It also expressed several public policy concerns that supported its conclusion: that comparison would reduce
the plaintiffs recovery because juries will likely allocate most if not all fault to the intentional actor; that allocating fault to the intentional party may reduce the incentive for the negligent actor to act with due care; and that comparison is impractical because intentional and negligent torts are different “not only in degree but in kind, and the social condemnation attached to it.”
Id.
at 719, quoting, Prosser § 65 at 462.
See also Marceaux v. Gibbs,
680 So.2d 1189 (La.App.1996),
aff'd,
699 So.2d 1065 (La.1997) (following
Veazey
).
In another sexual assault case,
Kansas State Bank & Trust Co. v. Specialized Transportation Services, Inc.,
249 Kan. 348, 819 P.2d 587 (1991), the parents of a child who was sexually assaulted by a school bus driver filed a negligence suit against the school and the bus company. The Kansas Supreme Court held that a negligent defendant should not be permitted to reduce its liability by intentional acts they had a duty to prevent.
The Kansas Supreme Court followed its holding in
Gould v. Taco Bell,
239 Kan. 564, 722 P.2d 511 (1986), in which it said the question of comparing negligent and intentional acts depends on “the nature of the duty owed in each instance.” In
Gould,
an assailant physically and verbally abused the plaintiff in a restaurant in full view of the restaurant’s managers. The court held that the restaurant’s negligent failure to .maintain security under the facts of the case should not have been compared with the intentional conduct of the assailant. 722 P.2d at 513.
A similar approach was suggested by the New Jersey Supreme Court in
Blazovic v. Andrich,
124 N.J. 90, 590 A.2d 222 (1991). There the jury was permitted to compare the negligence of a restaurant owner in failing to maintain adequate lighting and security in the parking lot with the intentional act of a patron who attacked the plaintiff. While the court upheld the comparison, it recognized that apportionment of fault between tortfea-sors may be precluded “when the duty of one encompassed the obligation to prevent the specific misconduct of the other.” It distinguished the facts before it on the basis that “the events that allegedly took place in the parking lot neither were sufficiently foreseeable nor bore an adequate causal relationship to [the negligent defendant’s] alleged fault to justify the imposition on [the defendant] of the entire responsibility for the resultant injury.”
Id.
590 A.2d at 233;
compare Gould,
722 P.2d at 511-13.
Other courts take a different view. In
Reichert v. Atler,
117 N.M. 623, 875 P.2d 379 (1994), a bar patron was killed when assaulted by another customer. The bar owners were sued for failing to provide adequate security, and the bar owners relied on the intentional act of the third party to reduce their liability. The court held that the bar owner may reduce his liability by the percentage of fault attributable to a third party. They reasoned that this principle was most consistent with the rejection of joint and several liability in comparative fault cases and that each individual tortfeasor should be held responsible only for his or her percentage of fault.
Id.
875 P.2d at 381.
See also Barth v. Coleman,
118 N.M. 1, 878 P.2d 319 (1994)(following Reichert).
Likewise, in
Weidenfeller v. Star & Garter,
1 Cal.App.4th 1, 2 Cal.Rptr.2d 14 (1991), an assault victim sued a bar owner for failing to have adequate lighting and security. The jury allocated 75 percent of the fault to the assailant. On appeal, the court said that the argument that negligent acts should not be compared with intentional acts “violate[d] the common sense notion that a more culpable party should bear the financial burden caused by its intentional act.”
Id.
2 Cal. Rptr.2d at 16.
See also Martin By and Through Martin v. United States,
984 F.2d 1033 (9th Cir.1993) (following
Weidenfeller); Natseway v. City of Tempe,
184 Ariz. 374, 909 P.2d 441 (1995).
Accordingly, the concern in cases that compare the negligence of a defendant with the intentional act of a third party is not burdening the negligent tortfeasor with liability in excess of his or her fault; conversely, the primary concern in those cases that do not compare is that the plaintiff not be penalized by allowing the negligent party to use the intentional act it had a duty to prevent to reduce its liability.
In our view, the conduct of a negligent defendant should not be compared with the intentional conduct of another in determining comparative fault where the intentional conduct is the foreseeable risk created by the negligent tortfeasor. As other courts have recognized, comparison presents practical difficulties in allocating fault between negligent and intentional acts, because negligent and intentional torts are different in degree, in kind, and in society’s view of the relative culpability of each act. Such comparison also reduces the negligent person’s incentive to comply with the applicable duty of care. Moreover, while a negligent defendant may, of course, raise a third party’s intentional act to refute elements of the plaintiffs negligence claim such as duty and causation, fairness dictates that it should not be permitted to rely upon the foreseeable harm it had a duty to prevent so as to reduce its liability.
Our holding also comports with the principles underlying
McIntyre.
The plaintiff here was not negligent. On the other hand, the defendant was negligent, and his breach of care led to the plaintiffs injuries. Thus, the defendant’s liability to the plaintiff is linked to his degree of fault as required by
McIntyre,
and he should not be permitted to reduce his liability by relying on the occurrence of the foreseeable risk of harm he had a duty to prevent. As one commentator has written: “the
McIntyre
principle of holding the tortfeasor liable for only his own percentage of fault is not abrogated by nonapportionment when the nature of the tortfeasor’s breach is that he created the risk of the second tortfeasor’s [intentional] act.” Entman,
The Nonparty Tortfeasor,
23 Mem. St. U.L.Rev. 105, 107 (1992).
Accordingly, we conclude that the lower courts incorrectly determined that the negligence of the defendant should have been compared with the intentional act of the defendant’s patient. In this case, however, the error was harmless in that the jury apportioned 100 percent of the fault to the defendant. Thus, we remand the case to the trial court for entry of a judgment consistent with the jury’s verdict.
AUTHORITY TO REALLOCATE FAULT
The defendant moved for a new trial, and the trial court, after independently weighing the evidence, granted a new trial because it disapproved of the jury’s failure to allocate some percentage of fault to the patient who had intentionally injured the plaintiff.
The plaintiffs argued that the trial court had the authority to reallocate the percentage of fault instead of granting a new trial, in the same manner as it can suggest an addi-tur or remittitur. The defendant maintains that a new trial was the only appropriate remedy because allocation of fault lies within the exclusive province of the jury.
As the thirteenth juror, the trial court must grant a new trial if the verdict is contrary to the weight of the evidence. Tenn. R. Civ. P. 59.06. If the trial court approves of the jury’s verdict with the exception of the amount of damages, it may suggest an adjustment to the verdict. If the plaintiff does not consent to a decreased award (remitti-tur), or the defendant does not consent to an increased award (additur), the trial court must grant a new trial. Tenn.Code Ann. § 20-10-101 & 102. The purpose is to allow the trial court to revise and correct errors relating to the size of a jury’s verdict “without the costly and time-consuming process occasioned by the granting of a new trial.”
See Thrailkill v. Patterson,
879 S.W.2d 836 (Tenn.1994).
Although we have not addressed the exact issue, other jurisdictions have held that the mechanisms of additur/remittitur do not apply to alter the jury’s allocation of comparative liability. In
Rowlands v. Signal Const.
Co.,
549 So.2d 1380 (Fla.1989), the Florida Supreme Court rejected the use of remittitur to correct the jury’s assignment of comparative fault because the “apportioning of liability is a matter peculiarly within the province of the jury.” The court concluded:
Since liability is inextricably bound up with the apportionment of damages under the doctrine of comparative negligence, this matter must be left to the jury. When the percentages of liability are contrary to the manifest weight of the evidence, the trial court must treat this defect as an error in the finding of liability itself. The only remedy is to order a new trial on all issues affected by the error.
549 So.2d at 1382-83;
see also Akermanis v. Sea-Land Service, Inc.,
688 F.2d 898 (2d Cir.1982)(“remittitur a limited exception to jury fact finding”);
State v. Kaatz,
572 P.2d 775 (Alaska 1977) (“apportionment of negligence is part of the liability phase of a case, not the damages phase”).
Reaching a different result, the Rhode Island Supreme Court, in
Cotrona v. Johnson & Wales College,
501 A.2d 728 (R.I.1985), said that “the mechanisms of remittitur and additur shall be available in the future to trial justices not only to reassess an erroneous damage award but also to correct a jury’s misapportionment of liability as it may relate to comparative negligence.” In the court’s view, these means would “prevent the burdensome costs, delays and harassments that accompany re-litigation of the same issues while at the same time assuring the litigants substantial justice.”
Id.
at 733-34.
In Tennessee, our cases have specifically limited the statutory procedures of re-mittitur and additur to correction of damages and not liability.
See, e.g., Burlison v. Rose,
701 S.W.2d 609 (Tenn.1985)(remittitur is not proper, and a new trial must be granted, when the trial judge disagrees with the jury on questions of fact other than the amount of damages);
Spence v. Allstate Ins. Co.,
883 S.W.2d 586, 594 (Tenn.1994) (suggestion for additur applies to damages). Thus, the trial court correctly determined that .it lacked the authority to reapportion the comparative fault in its role as thirteenth juror.
Our opinion in
Wright v. City of Knoxville,
898 S.W.2d 177 (Tenn.1995), cited by both parties, does not require a different conclusion.
Wright
was based on our appellate court standard of review of factual findings made by the trial court in non-jury cases: “review of findings of fact by the trial court in civil actions shall be
de novo
upon the record of the trial court, accompanied by á presumption of the correctness of the finding, unless the preponderance of the evidence is otherwise.” Tenn. R.App. P. 13(d). By comparison, the standard of review governing jury findings of fact in a civil action is completely different. Those findings of fact may be set aside only if there is no material evidence to support the verdict.
Id.
Thus,
Wright
does not allow the trial court to reallocate fault between the parties in a jury case in its role as thirteenth juror.
CONCLUSION
We conclude that the defendant psychiatrist owed a duty of care to the plaintiff nurse because he knew or should have known that his patient posed an unreasonable risk of harm to a foreseeable, readily identifiable third party. We also conclude that the lower courts erred in ruling that the defendant psychiatrist’s negligence should be compared with the intentional conduct of the non-party patient in allocating fault. We consider this error harmless because the jury allocated 100 percent of the fault to the negligent defendant psychiatrist. Finally, although not applicable here, in view of our result we have determined that the trial court may not reallocate the comparative fault after weighing the evidence as the thirteenth juror, but must instead grant a new trial.
Accordingly, the judgment of the Court of Appeals is reversed in part and affirmed in part, and the case is remanded to the trial court for entry of judgment on the jury’s verdict. The costs of appeal are taxed to the defendanVappellee, for which execution shall issue if necessary.
DROWOTA, REID and HOLDER, JJ., concur.
BIRCH, J., not participating.