Stephanie and Jane Toth were born prematurely on the night of June 16, 1953 at the Community Hospital at Crien Cove, New York. From the moment of their birth they were dangerously ill. Their pediatrician, fearful for their lives, immediately ordered that they be placed in an isolette, a type of incubator, and that oxygen be administered to them. His written orders were to the effect that the infants should receive oxygen at the rate of 6 liters per minute for the first 12 hours, and thereafter at the rate of 4 liters per minute.1
The theory behind this oxygen treatment was that maintenance of premature babies in an oxygen environment would sustain life and prevent brain damage. Shortly after the birth of the twins, a Cooperative Study of Betrolental Fibroplasia and the Use of Oxygen conducted for the National Institute for Neurological Diseases and Blindness indisputably established that the course of treatment was tragically mistaken. The study showed that, while the use of oxygen might mitigate brain damage, it did not reduce the mortality rate at all, and did cause retrolental fibroplasia (RLF), resulting in blindness in many infants. This was the result here. The twins were given oxygen for over 30 days — Stephanie, the elder, until July 20 and Jane until July 27. Stephanie lost all useful vision in her left eye, while Jane developed RLF in both eyes and lost her sight completely.
Steve Toth, the children’s father, as guardian ad litem, has brought this action against the pediatrician and the hospital, as well as against an ophthalmologist, who examined the children’s eyes a month after their birth on July 17, claiming that the serious injuries sustained by his daughters were due to the negligence of the hospital’s nursing staff and the professional malpractice of the doctors.
The claim against the pediatrician has two aspects. Mr. Toth contends, first, that the administration of oxygen was contrary to good medical practice in June, 1953. His second con*259tention, which is also the basis of the claim against the hospital, is that the nurses failed to follow the pediatrician’s orders concerning the amount of oxygen to be given the babies and that, in fact, 6 liters per minute of oxygen were constantly administered to the children for some four weeks after their birth rather than the prescribed 4 liters per minute. It is contended the nurses were negligent in not adhering to the pediatrician’s orders, and the doctor was negligent in not discovering promptly that the oxygen dosage had not been reduced by the nurses as he had directed.
The case against the ophthalmologist also has a two-fold basis. First, he should have discovered the ELF during his examination of the babies on July 17 and, second, regardless of whether ELF had manifested itself by that date, the ophthalmologist should have made a recommendation concerning the advisability of continuing the oxygen treatment.
The trial court dismissed the .complaint against the hospital at the close of the entire case on the grounds that the evidence did not show that 6 liters per minute had been continuously given and that the alleged deviation from the doctor’s orders was the proximate cause of the children’s loss of sight.
The case against the doctors was submitted to the jury. A reading of the trial court’s charge, however, establishes that the only question the jury was asked to decide was whether the doctors had conformed to acceptable medical practice in their respective specialties. The jury returned a verdict in favor of the doctors. On appeal, the Appellate Division affirmed, one Justice dissenting, as to the hospital and the pediatrician.
On the issue whether the oxygen treatment was correct, the evidence was conflicting and the issue strenuously contested. The record supports the conclusion that while it was well recognized by June, 1953 that there was a risk of ELF in the oxygen therapy and many physicians and institutions were already convinced that the oxygen therapy was either useless or not worth the risk of ELF involved, there were many reputable institutions holding the opinion that the oxygen treatment was necessary or at least a worthwhile risk. There was, therefore, ample evidence to support the jury’s finding that the defendant pediatrician, who viewed the oxygen treatment as a necessary calculated risk, had acted in accordance with *260acceptable medical practice. Consequently, on this question the jury’s verdict must be deemed conclusive, and the issue of acceptable medical practice may not be submitted to the jury on a new trial.
Nevertheless, there still remains open the question whether the case against the hospital should have been submitted to the jury and whether it was error for the trial court to refuse to charge the jury, as requested by plaintiff, that the pediatrician might be found negligent in failing to ascertain that his orders were not being followed by the hospital staff.
Plaintiffs argue that this theory of liability is independent of whether or not there was an established medical opinion at that time with respect to the treatment of premature infants with oxygen. Even if it were not definitely established that the oxygen treatment was improper, the pediatrician, the plaintiffs argue, can be held liable for not noticing the deviation from his orders by the hospital nursing staff.
While the issue is not as simple as plaintiffs paint it, we do conclude that this second theory of liability should have been submitted to the jury.
At the outset, defendants assert that there is no evidentiary basis to support plaintiffs’ theory. For example, it is stated that there is no showing that 6 liters per minute of oxygen were continuously administered to the children. We cannot agree.
Briefly, the evidence shows that not once before July 17 do the nurses’ notes record 4 liters as being given, whereas on 10 occasions the nurses’ notes contain statements that 6 liters were being given. The defendants’ position is that the prescribed 4 liters were given. Their explanation for the fact that there is no reference to 4 liters in the nurses’ notes is that the doctor’s written orders were what they denote as a “ standing order ” which “ under sound hospital practice did not require the nurse, following the expiration of twelve hours, to note that the oxygen flow was reduced to 4 liters but did require her to note a deviation from the order, namely, the use of 6 liters ”. (Brief of defendant doctors, pp. 5-6.) Thus, whenever 6 liters were given, the increase was required by a deterioration or some special condition. Otherwise, at all other times, 4 liters were administered as shown by the notations “ continuous ” and “ constant ” in the nurses’ notes.
*261This account does not satisfactorily explain the contrary indications of the hospital records. On at least three occasions the words “ continuous ” and “ constant ” or abbreviations for these terms are used although there can be no doubt that 6 liters of oxygen were then being furnished. Moreover, if it were true that, every time 6 liters of oxygen are recorded in the nurses’ notes, there was some special reason for doing so, one must conclude that the course of treatment had no continuity or rationale behind it. When serious symptoms, such as difficulty in breathing, appeared, no increase in the oxygen flow is shown yet, for relatively trivial conditions, the flow was stepped up. Without dwelling any further on the evidence, these facts suffice to establish that a jury could reasonably find that the twins received 6 liters of oxygen throughout.
Likewise, a question of fact exists as to whether 4 liters per minute of oxygen, as opposed to 6 liters, would have materially reduced the likelihood of ELF or the severity of the injuries. The issue of causation in medicine is always difficult but, when it involves the effect of a failure to follow a certain course of treatment, the problem is presented in its most extreme form. We can then only deal in probabilities since it can never be known with certainty whether a different course of treatment would have avoided the adverse consequences.
In the case of ELF, the Cooperative Study supports plaintiffs’ position. A conclusion of that investigation was that an increase in oxygen concentration from 35% to 50% did not increase the incidence of ELF in single births but, in the case of multiple births, the rate was at least doubled if the oxygen was administered over an extended period of time — approximately 12 days. The integrity, reliability and impartiality of the report entitles it to be considered of substantial probative value on the issue of causation. Although the question is indeed a close one, the scientific investigations and the testimony in the record are sufficient to create an issue of fact for the jury on the issue of proximate cause.
Since there is sufficient evidentiary support for plaintiffs’ theory, we come then to the critical question which is the soundness of this alternative theory of liability. Defendants do not claim that good medical practice did not require the pediatrician to check to see if the nurses had followed his orders and to *262learn of any constant supply of excessive oxygen to the infants within a day or so (Louisell & Williams, Trial of Medical Malpractice Cases, pars. 2.21; 8.05). The doctor here was visiting the children every day and in fact he maintained in a pretrial deposition that he had checked the flow meter of the isolette on each visit and that 4 liters had been given throughout. Under these facts, there is no question here of imposing an unreasonable burden of supervision on the doctor.
The true thrust of the pediatrician’s argument is that he contests the materiality of any question of medical supervision. His position is that, although the jury was not specifically asked to decide whether a straight 6 liters of oxygen were acceptable therapy, in finding for him the jury necessarily had to find that 6 liters were acceptable because the evidence established that at Bellevue and other hospital nurseries they were using 60% or more oxygen. Thus, had he ordered a straight 6 liters he could not be held liable. Therefore, the argument concludes, there is no basis for a holding that he was negligent in not noticing that the oxygen had not been reduced.
The law generally permits the medical profession to establish its own standard of care. Although a plaintiff in a medical malpractice case does not always require expert testimony to establish a prima facie case (Benson v. Dean, 232 N. Y. 52; Meiselman v. Crown Hgts. Hosp., 285 N. Y. 389; Hammer v. Rosen, 7 N Y 2d 376), evidence that a physician conformed to accepted community standards of practice usually insulates him from tort liability (Benson v. Dean, supra, p. 58; Prosser, Torts [3d ed.], p. 168; see, also, McCoid, Liability of Medical Practitioners, 12 Vanderbilt L. Rev. 549, 605 ff). There is, however, a second principle involved in medical malpractice cases. Having its genesis in the reasonable man rule, this principle demands that a physician should use his best judgment and whatever superior knowledge, skill and intelligence he has (Prosser, op. cit., p. 164). Thus, a specialist may be held liable where a general practitioner may not. The principle has been accepted by the Restatement of the Law of Torts (Restatement, 2d, Torts, § 299A, comment c; § 289, subd. [b] and comment m; § 299, comment f; see, also, Harris v. Fall, 177 P. 79; cf. Note, Duty of a Manufacturer to Discover Unforseeable Common Uses of His Product, 66 Col. L. Rev. 1190,1192-1193).
*263The necessary implication of this latter principle is that evidence that the defendant followed customary practice is not the sole test of professional malpractice. If a physician fails to employ his expertise or best judgment, and that omission causes injury, he should not automatically be freed from liability because in fact he adhered to acceptable practice. There is no policy reason why a physician, who knows or believes there are unnecessary dangers in the community practice, should not be required to take whatever precautionary measures he deems appropriate.
For example, a physician believes that a course of treatment is highly dangerous. Using his expertise and exercising his best judgment, he decides that certain steps can safely be taken to minimize the risks. However, through carelessness, he omits to take the necessary safeguards. As a result, injury occurs. We see no justification for the position that, as a matter of law, because other reputable physicians did not think the precautions necessary and did not view the treatment actually given as improper, there may be no tort liability. It is not unreasonable to impose upon a physician, who believes that added precautions are necessary, the obligation that he act diligently in taking the necessary safety measures. This conclusion is nothing more than an application of the rule that a physician should at all times use his own best judgment and care.2
It was well known in June, 1953, and the pediatrician here admittedly was aware of the fact, that the use of oxygen entailed the risk of ELF. It, therefore, made sense for a physician to seek to avoid prescribing any more than the minimum amount of *264oxygen necessary to accomplish the desired purpose of preserving the life of the child. The pediatrician here testified that “ [i]n extreme distress it was an accepted method to give the 6, and if a baby was going into fairly even keel, to keep it at 4 ”. From this statement and his other testimony before and during trial, a jury could find that the pediatrician believed that his orders represented a balance between the benefits to be gained from using oxygen and the danger of possible blindness. If the jury made such a finding, then it could reasonably hold the pediatrician responsible for failing to take note of the omission of the nurses to make the ordered reduction because the doctor’s knowledge of the risks involved in using oxygen in excess of the prescribed amount should have alerted him to the need for greater diligence.
Thus, there was sufficient evidence to establish every element of plaintiffs’ second theory of liability as reflected in their fifth request to charge. To summarize, a jury could find, although, under the facts here, it certainly would not be compelled to do so: (a) that 6 liters of oxygen per minute were given the children continuously; (b) that, had the pediatrician’s orders been followed strictly, the ELF would not have occurred or the consequences would not have been as severe; (c) that the doctor had opportunity to discover the deviation, but did not do so; (d) that there was general knowledge that there was a danger in the excessive use of oxygen and that the doctor was aware of this knowledge and (e) that his written orders were made with a view toward minimizing that risk. If the jury so found, it could properly find that the doctor’s failure to assure that his orders were carried out was negligence.
With respect to the hospital, there can be little dispute that the plaintiffs made out a prima facie case when they introduced substantial evidence to establish that the nurses had not conformed to the pediatrician’s orders. The hospital’s principal argument that the dismissal of plaintiffs’ case against it was proper is also that the treatment actually given conformed to the then existing accepted standard of care and that this standard included the administration of concentrated oxygen.
Insofar as the hospital is concerned, whether the administration of oxygen was proper or not is completely immaterial. We are not here dealing with a matter usually left by a physician *265to the discretion of the hospital staff. Bather involved here is a situation where the attending physician gives direct and explicit orders to the hospital staff. In such cases, nurses are not authorized to determine for themselves what is a proper course of medical treatment. They may not invade the area of the physician’s competence and authority to overrule his orders. To illustrate, no one would argue that, because at Bellevue oxygen was given, the nurses here might administer it if none had been ordered. Similarly, the nurses could not with impunity fail to give oxygen if directed to do so even if some institutions never used the oxygen therapy for premature infants.
The primary duty of a hospital’s nursing staff is to follow the physician’s orders, and a hospital is normally protected from tort liability if its staff follows the orders.3 Nevertheless, if the doctor’s orders are to be a hospital’s shield from tort liability, it cannot at the same time maintain that a deviation which causes injury cannot be a basis for tort liability because other physicians or medical institutions might consider the treatment actually given as acceptable. More important than the question of logical consistency is the issue of social policy. There is every reason to encourage the hospital staff to carry out diligently the doctor’s orders while there are no countervailing considerations which would justify relieving the hospital from tort liability for failing to do so.
In the case at bar, if it were established that the pediatrician’s orders were contrary to sound medical practice, the hospital would without question and quite rightly deny liability on the basis that it in good faith simply carried out the pediatrician’s orders. By the same token, we believe it was error to preclude the jury from finding the hospital negligent if they should find there was a deviation which was the proximate cause of the twins’ injuries. At no time was it claimed by the hospital that authorization had been given either by the pediatrician or the standing orders of the premature nursery to give 6 liters of oxygen continuously. It follows that the case against the hospital should not have been dismissed.
*266With, respect to the ophthalmologist, plaintiffs claim error in the trial court’s failure to charge that this doctor could be found liable for failing to make a recommendation concerning the continued use of oxygen. We conclude that the failure to charge was not prejudicial. Although this doctor was also aware of the ELF controversy, there is no evidence to indicate that he thought 4 liters of oxygen, which the children were then scheduled to be receiving, were particularly dangerous. Thus, had he specifically recommended the continuation of the oxygen, he would have acted in accordance with acceptable medical practice.
The order of the Appellate Division should be modified so as to order a new trial against the defendants, Community Hospital at Glen Cove and Hellmann, with costs to abide the event, and otherwise affirmed, without costs.
. Six liters per minute meant that the atmosphere in the isolette consisted approximately of 50-60% oxygen. At 4 liters per minute, the oxygen concentration would be about 30-40%.
. Although the problem is not present here, it must be acknowledged that as sound as the requirement may be that a physician use his best judgment and care as well as exercise whatever superior knowledge, skill and intelligence he has, this principle not only supplements, but on occasion may conflict with the rule that general medical practice is the usual measure of tort liability for medical practitioners. Thus, in the illustration given above, the additional measures may ultimately prove harmful. If those measures were not accepted by other physicians in general, the physician might find himself held responsible for failing to follow the community standard. Fairness and the avoidance of any principle of strict liability would seem to require that the physician not be held liable for exercising his best judgment at least in the case where the procedures used by the physician had some reputable support in the profession.
. An exception to the rule would exist where the hospital staff knows that the doctor’s orders are so clearly contraindicated by normal practice that ordinary prudence requires inquiry into the correctness of the orders (see, generally, Florentino v. Wenger, 19 N Y 2d 407, 414-415).