Opinion for the Court filed by WILKEY, Circuit Judge.
WILKEY, Circuit Judge:On the (morning of 14 August 1973, Horace Miller, a 43-year-old government worker, went to Hadley Memorial Hospital’s emergency room1 for treatment of abrasions received in a bicycle fall the previous week. The hospital gave him a penicillin injection at 9:05 a. m., and he departed the hospital between 9:20 and 9:25 a. m. At 9:30 a. m. he was discovered in the hospital’s parking lot suffering from an anaphylactic reaction to the penicillin; he was in shock. He was rushed back into the hospital, and at 9:36 a. m. hospital personnel started resuscitation efforts. At 10:10 a. m. Mr. Miller was pronounced dead.
Frances P. Daniels, the deceased’s mother and personal representative, brought this malpractice action2 in the U.S. District Court for the District of Columbia 3 against the Hadley Memorial Hospital,4 contending that the hospital violated the standard of care it owed her son and proximately caused his death by (1) failing to retain him for observation for a sufficient period of time after the injection; (2) failing to give him adequate respiratory assistance during the resuscitation effort; and (3) failing to give him an intravenous injection of adrenalin as soon as possible during the resuscitation attempt. The parties waived a jury trial, and on 29 March 1976 the trial court filed its findings of fact and conclusions of law, entering judgment for the hospital.5 The case is before us on plaintiff’s appeal from that judgment.
In its “Findings”, the trial court rejected all three of plaintiff’s contentions:
First, it found that plaintiff had failed to prove that a thirty-minute post-injection observation period was the applicable standard of care at the time of the occurrence. Rather, the court found that the applicable standard was a fifteen- to twenty-minute period, which was not breached by the hospital.6 This finding of fact is sufficiently sustained by the record evidence, and we leave it undisturbed.
Second, the trial court found that the hospital’s failure to give proper respiratory assistance, though a breach of duty, was not the proximate cause of Mr. Miller’s death.7 This finding is based on a serious misapprehension of the medical evidence presented at trial, and we set it aside as clearly erroneous.
Third, the trial court found that the hospital did not breach its duty to Mr. Miller by failing to give him a timely intravenous injection of adrenalin because such an injection was impossible under the circumstances.8 This finding is against the clear weight of the evidence, and we set it aside as clearly erroneous. Accordingly, we reverse the judgment below and remand for a new trial consistent with the instructions in Part IV of this opinion.
*87Because we uphold the trial court’s conclusion that the hospital was not negligent in its post-injection observation of the deceased, we deal here only with those aspects of the case bearing on the possible negligence of the hospital in its efforts at resuscitating Mr. Miller.
I. OVERVIEW OF PLAINTIFF’S CASE
Before recounting the sequence of events immediately preceding Horace Miller’s death, we think it useful at the outset to consider the basic framework of plaintiff’s case.9
Plaintiff acknowledges that Horace Miller was in grave condition when he was rushed from the hospital’s parking lot back into the emergency room. It is not disputed that he was in shock.10 His respiration was severely impaired; he was already unconscious and breathing only four times a minute, too slowly to provide sufficient oxygen to his body.11 His circulation was also impaired; his heart was beating forty times a minute, too inefficiently to provide proper circulation of the blood.12 Moreover, his circulatory system was collapsing, confining the flow of blood to larger vessels supporting the major body organs. This meant that there was little if any circulation in the small veins and capillaries near the body surface.13 Mr. Miller’s condition was rapidly deteriorating, and when the resuscitation effort began at 9:36 a. m. quick action was necessary if he was to be saved.
Plaintiff contends that, faced with this emergency, the hospital had a duty to do three things without delay. The first, and most important, action was to get oxygen into the patient, for without oxygen the patient’s brain would die in a matter of minutes. This required giving the patient respiratory assistance by (1) establishing a clear airway and (2) using positive pressure to force oxygen into the patient’s lungs by means of mouth-to-mouth resuscitation, an ambu bag unit, or an endotracheal tube and respiration machine.
The second action was to get adrenalin into the patient’s circulatory system where it could begin to counteract the effects of the anaphylactic reaction. This required that either an intravenous14 or intracardiac15 injection be made, because without circulation near the body surface subcutaneous 16 or intramuscular17 injections would have little value.
The third action was to restore the patient’s blood circulation so that both the oxygen and the adrenalin could be transported through the patient’s blood stream. This required external cardiac massage, which makes the heart muscle expand and *88contract, thereby causing the blood to circulate.
Plaintiff concedes that the hospital did take the third proper action to restore Mr. Miller’s circulation as soon as he was brought into the emergency room. Cardiac massage was started immediately and was continued throughout the resuscitation effort. However, the hospital made no attempt to provide proper respiratory assistance to the patient for the first four minutes of the resuscitation effort — from 9:36 to 9:40 a. m. — despite the fact that the patient stopped breathing altogether during those four minutes.18 During these critical minutes, the patient’s brain was not receiving enough oxygen.19 This was indicated at 9:43 a. m. when one of the attending doctors noticed that the patient’s pupils were dilated, a sign of neurological death.20 Moreover, although the hospital administered a subcutaneous injection of Susphrine21 within the first two minutes of the resuscitation attempt, it did not make an intravenous injection of adrenalin until 9:42 a. m. — six minutes into the resuscitation effort.22 Plaintiff argues that it was possible to make an intravenous injection during these initial six minutes and that failure to do so was inexcusable.
In short, then, the crux of plaintiff’s case is that the hospital had a duty to get both oxygen and adrenalin into Horace Miller’s blood stream as quickly as possible, that the hospital failed to do this, and that its failure effectively eliminated whatever chance the patient had of surviving.
II. THE ACTION TAKEN
As one might suspect, the action during an emergency resuscitation attempt, or “Code Blue” in hospital parlance, is fast and furious. However, the District Court was able to reconstruct the approximate sequence of events on the basis of hospital records. Chief Nurse Curry, in charge of the emergency room, was present during the whole course of the Code Blue and one of her duties included a minute-by-minute recordation of what was done. The hospital records also include the notes of Dr. Hassan Vaziri, the emergency room physician who had originally ordered the penicillin injection. As Dr. Vaziri was not at trial and was never deposed,23 these two documents, as supplemented by Nurse Curry’s testimony, furnish the only source of factual information as to what occurred during the critical early stages of the resuscitation attempt. For this reason, and because they are brief, the two documents are set forth verbatim here. Dr. Vaziri’s notes, made after the incident, read:24
8/14/73 1 — external cardiac massage started immediately
2 — Intracardiac adrenaline (sic) injection was attempted twice
3 — Subcutaneous Sus-phreine (sic) injection in first few minutes
4 — Solu-Cortef i.v. injection immediately once and repeated with 2 more i.v. and one i.m.
5 — I.V. started with 50 cc sodium bicarbonate was injected within few minutes after the collapse and repeated every few minutes, totally 6 times during the resussitation (sic) attempt.
6 — Intra-tracheal intubation stablished (sic) by anesthesiologist within 5 minutes and was continued with respirator bag (positive pressure) all the time.
7 — I.V. Aramine was started in less than 10 minutes and was continued all the time.
8 — E.K.6. machine was monitoring the patient from 10 minutes after the collapse until the end.
*899 — external cardiac massage was continued for 35 minutes.
10 — Since there was no response to all the emergency measures, electric defribrilator (sic) was employed and several repeated shocks were attempted, but it' also failed to produce any heart beat.
11 — from 3-4 minutes after the patient was put on the couch I never heard any other heart beat or any voluntary respiration till the end.
12 — the patient was pronounced dead at 10:10 A.M.
/s/ Hassan Vaziri, M.D.
Nurse Curry’s contemporary handwritten account reads:25
8/14/73 9:35 Brought into opd via wheel chair by
am J. Curry accompanied by E. Staple-ton, E. Grayson, & R. Walls Unconscious, Diapharetic (sic). Pulse weak-about 40. Resp. about 4
Dr. Vaziri called from Lounge by J. Curry
9:36 Lifted to green couch in back hall by Dr. Vaziri, J. Curry, E. Stapleton, & E. Grayson
02 nasal cannula position by E. Staple-ton & 02 begun
9:38 SusPhrene (sic) 0.5 ml®Q Dr. Vaziri Solu Cortef 2 ml Dr. Vaziri External cardiac massage-Dr. Vaziri Crash cart to scene-J Curry E.K.G. Tech called, Inhalation Therapist paged, & anesthesiologist summoned by J. Curry.
9:40 IPPB w/Ambu-Dr. Johnston started EKG set up & strip begun-B. Sadeghian
IPPB taken over by Mr. Tate Intubated-Dr. Johnston
9:42 1000 cc 5% D/W Iv Rt. Arm Dr. Johnston assisted by Dr. Massingale, Soda Bicarb AMPS II pushed by Dr. Massingale
Aramine 2 ml added to Iv by J. Curry, R.N.
Ext cardiac massage continued by D. Vaziri
Solu Cortef ampIpush-J. Curry
9:43 Pupils dialated (sic) per Dr. Johnston Adrenalin 1 ml Intra Cardiac-Dr. Vaziri (Costa)
9:45 1000 cc N.I. Iv L. arm-Dr. Johnston Soda Bicarb 50 mEq pushed R Iv-J. Curry
Solu Cortef 2 ml pushed R Iv-Dr. Jones
9:46 Both Iv’s running well. EKG in pro-am gress at intervals
Dr. Johnston resumed IPPB c Ambu Ext cardiac massage-Mr. Tate relieved Dr. Vaziri
Dr. Quiambao at EKG machine temporarily
9:48 Patient not responding
Soda Bicarb. 50 mEq pushed-Dr. Jones
Adrenalin 1 ml Intra Cardiac-Dr. Vaziri
9:50 Solu Cortef 2 ml pushed-Dr. Vaziri CPR continuously
9:55 Soda Bicarb 50 mEq pushed-Dr. Vaziri
10:00 Closed chest massage continuously-Mr. Tate
Patient not responding IPPB continuously
10:03 Mr. Piazza here. Dr. Johnston left. Electro-shock x 3 ordered by Dr. Jones. Administered by Mr. Piazza c no response. Intra Cardiac inj. 1 ml Adrenalin by Dr. Jones # 4 Electro-shock by Mr. Piazza-no response
10:10 Pronounced by Dr. Vaziri. Body to Morgue . . .
J. Curry, R.N.
At trial, the other fact witnesses, besides Nurse Curry, included other Hadley Hospital employees who participated in the Code Blue. Also, the plaintiff and defendant each produced one independent expert witness.
The District Court found that it was 9:35 a. m. when Horace Miller was returned to the emergency room in a wheelchair, unconscious, diaphoretic, moaning but trying to talk, with a pulse of forty and a respiration of about four.26 At 9:36 he was placed on a hard couch and resuscitation efforts were started.27 Sometime within the next four minutes the patient’s voluntary respiration *90ceased and did not thereafter resume.28 For approximately the first two minutes there were, aside from several practical nurses, a doctor (Vaziri) and a nurse (Curry) attending to Miller.
Within the first minute, nasal cannulas 29 were installed and oxygen was started. At 9:38 a. m., Dr. Vaziri gave Mr. Miller an injection of Sus-phrine subcutaneously.30 He also made an injection of Solu-Cortef.31 Whether Dr. Vaziri gave Mr. Miller the injection of Solu-Cortef intravenously or subcutaneously is a major dispute in the case. At approximately 9:38 a. m., Dr. Arthur Jones, a staff physician at the hospital, arrived.32 There is some evidence to suggest that Mr. Miller had already stopped voluntary breathing by this time.33 At about the same time Nurse Curry rushed a “crash cart” to the scene.34 The crash cart carried the medical equipment and drugs necessary for emergency resuscitation. Notably, the cart contained equipment for positive pressure oxygen, including an ambu bag unit, oxygen tanks, tubing, masks, and airways.35 There is some evidence that the crash cart contained needles of various sizes, including the “medi-cut” needle eventually used in establishing an intravenous route.36
Dr. Vaziri was administering external cardiac massage accompanied by chest pressure over the lungs. Dr. Jones checked Mr. Miller for a vein to establish an I.V., but he could not find one. At Dr. Jones’ direction, Nurse Curry called for an anesthesiologist, an inhalation therapist and an E.K.G. technician. At 9:40 a. m., Dr. Christina Chan Johnston, an anesthesiologist, arrived. By this time, the patient had stopped breathing.37 Dr. Jones was giving cardiac massage and chest pressure. Immediately, Dr. Johnston provided positive pressure respiratory assistance by puffing air into the patient’s mouth and by setting up and starting an ambu bag. This was at 9:40 a. m. Dr. Johnston then attempted to intubate Mr. Miller, but found that the proper size tube was not on the crash cart. She ran upstairs to obtain one, while an E.K.G. was set up. By 9:42 a. m., she had returned and successfully intubated the patient. At this time there was still no I.V. established, and Dr. Jones asked if anybody knew how to do a “cut down.”38 Dr. Johnston said she thought she could find a vein without a cut down. She searched along the brachial vein in the patient’s right arm and established an I.V., apparently using a medi-cut needle from the crash cart.39
Dr. Minda Massengale, another anesthesiologist who had arrived at the scene after Dr. Johnston, assisted Dr. Johnston in starting the I.V. There is also evidence that Drs. Johnston and Massengale were able to establish a second I.V. in the patient’s left arm at about 9:45 a. m.40 Dr. Jones immediately injected intravenously one cubic centimeter of adrenalin. Solu-Cortef, soda bicarb and aramine were also pushed intra*91veriously. At 9:43 a. m.(- Dr. Johnston noticed that the pupils of Mr. Miller’s eyes had dilated, proof of brain damage, at least, and probably neurological death.41 During this minute an intracardiac injection of adrenalin was made by Dr. Vaziri. Mr. Miller’s heart was still beating, though abnormally, and continued to beat for approximately another twenty minutes. There is evidence that by approximately 10:00 a. m. the patient’s heart had begun fibrillating.42 Efforts at defibrillation failed. At 10:03 another intracardiac injection of adrenalin was made. At 10:10 Mr. Miller was pronounced dead.
III. THE DISTRICT COURT’S FINDINGS
With respect to plaintiff’s contention regarding the improper ventilation of Mr. Miller, the District Court agreed with plaintiff that the hospital did breach its duty to provide Mr. Miller immediate positive pressure respiratory assistance, at least by means of mouth-to-mouth resuscitation’ until a more sophisticated technique could be employed.43 The evidence overwhelmingly supports this finding.44 However, the Court concluded that the hospital’s failure to provide proper ventilation was not the proximate cause of Mr. Miller’s death.45 Plaintiff here asserts that she presented persuasive evidence on the issue of causation but that the District Court was confused as to certain physiological facts and, therefore, misapprehended the nature of this evidence. Therefore, we are asked to set aside the District Court’s conclusion on the issue of causation.
With respect to plaintiff’s contention regarding the improper administration of adrenalin, the District Court agreed with plaintiff that the hospital had a duty to give an immediate intravenous injection of adrenalin to Mr. Miller if one were possible46 Again the evidence strongly supports this finding.47 However, the Court concluded that an intravenous, injection was not possible prior to the time it was actually administered and that, therefore, the hospital had not violated its duty.48 Plaintiff here asserts that she presented overwhelming evidence that an earlier intravenous injection was possible but that, in reaching its contrary conclusion, the District Court chose to ignore this evidence in favor of the hearsay assertion of a single witness. Therefore, we are asked to set aside the District Court’s finding that an immediate intravenous injection was impossible.
In reviewing the District Court’s decision, this court is bound by the basic principle that a finding of fact49 by the district judge sitting without a jury may be set aside on appeal only if “clearly erroneous” and that a reviewing court must not weigh the evidence de novo or disturb the finding simply because it might have reached a contrary result on the same evidence.50 However, this respect for the trial court’s findings cannot be pressed too far, for it is also well settled that “[a] finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has *92been committed.”51 Thus, a trial court’s finding will not be permitted to stand where it is based on a serious mistake as to the effect of evidence52 or is clearly contrary to the weight of the evidence.53 We apply this standard of review and look to all the evidence of record to determine whether the findings challenged here can pass muster.
A. Respiratory Assistance.
We turn first to the District Court’s conclusion that the hospital’s failure to provide proper ventilation was not the proximate cause of Mr. Miller’s death. We find that this conclusion must be set aside as “clearly erroneous” because the the District Court was laboring under a misconception of basic physiological fact which precluded it from fairly considering the evidence presented by plaintiff on an essential aspect of the causation issue.
In determining that there was no causal connection between inadequate ventilation and Mr. Miller’s death, the District Court applied the “substantial factor” test. This is the appropriate test for causation in cases, such at this, where the harm appears to have been brought about by two or more concurrent causes. Under this test, the plaintiff must show that the defendant’s deviation from the standard of care was a “substantial factor” in bringing about the harm complained of.54 This test has been applied in circumstances similar to those presented here, involving the medical mismanagement of a patient’s already potentially fatal condition. In Hicks v. United States55 the court dismissed defendant’s contention that proximate causation had not been shown because, although the negligent diagnosis had prevented the application of proper therapy, “ . . . even if surgery had been performed immediately, it is mere speculation to say that it would have been successful”:56
. [I]t does not lie in the defendant’s mouth to raise conjectures as to the measure of the chances that he has put beyond the possibility of realization. If there was any substantial possibility of survival and the defendant has destroyed it, he is answerable. Rarely is it possible to demonstrate to an absolute certainty what would have happened in circumstances that the wrongdoer did not allow to come to pass.
The Hicks decision, and the cases which have followed it,57 illustrate that there are at least two important factors relevant to the issue of causation in cases involving negligent treatment of a potentially fatal condition: first, the patient’s chances of survival if properly treated according to medical procedures generally recognized as appropriate under the circumstances; and second, the extent to which the patient’s chances have been reduced by improper departure from these established procedures. We are not suggesting, as some courts have,58 that these two factors can be re*93duced to precise verbal formulae, supplanting the “substantial factor” test. We do not think it is either possible or desirable to reduce the “substantial factor” test to lower and more concrete terms in this way. We do believe, however, that in considering the question of causation in a case such as this, the finder of fact must at least take into account both the patient’s chances of survival and the extent to which defendant has interfered with these chances.59
However, it is clear from the District Court’s opinion in this case that, while it considered plaintiff’s evidence on the first aspect of the causation issue, it dismissed plaintiff’s evidence on the second aspect. It is also apparent that the reason the Court dismissed plaintiff’s evidence was that it was confused over the respective functions of the circulatory system and the respiratory system.
At trial, plaintiff sought to meet her burden on the element of causation by presenting evidence on both of the two factors relevant to the causation issue. Regarding Mr. Miller’s chances of survival, plaintiff attempted to show that, while the patient was in grave condition, he had a significant chance of recovery if properly treated. In this connection, the plaintiff’s expert medical witness, Dr. Farquhar, testified by deposition that Mr. Miller was in a “bad” state when returned to the hospital but that his condition could have been medically corrected.60 He testified that, although he had handled twelve cases of potentially fatal penicillin reaction, all of these patients survived when given proper treatment.61 Dr. Brownlee, the defendant’s expert witness, testified that the patient had at best a “very guarded prognosis” which fell to “severely grave as the condition progressed.62 He also testified that only fifteen to twenty-five percent of victims of anaphylactic reactions to penicillin ultimately die.63 Dr. Jones, the emergency room physician, testified that even at the time he saw the patient, about three minutes after he had been returned to the hospital, there was still a chance to save him.64 Dr. Johnston, the anesthesiologist, stated that death was possible but not inevitable.65 In short, the evidence showed that even though Mr. Miller was in grave condition there was still an appreciable chance of saving his life.
The District Court considered this evidence and concluded that Mr. Miller “was in grave condition with only a slight chance of survival.”66 As we have said, this conclusion is relevant to the ultimate question of causation, but it is not in itself dispositive on the point. The Court was obliged also to consider plaintiff’s evidence concerning the extent to which the hospital’s improper ventilation of Mr. Miller reduced his chances of survival. This is where the District Court went astray.
At trial, the plaintiff adduced substantial evidence in an attempt to show that the hospital’s failure to provide respiratory assistance reduced Mr. Miller’s chances of survival by diminishing the supply of oxygen to his brain. On this point plaintiff relied primarily on Dr. Farquhar’s testimony that the hospital’s failure to supply proper ventilation “significantly diminished” Mr. Miller’s chances of survival.67 He testified that deprivation of oxygen causes cerebral anoxia; that cerebral anoxia in turn produces brain damage and then brain death in a short time; and that Mr. Miller’s unassist*94ed respiration was not sufficient to provide adequate oxygen to his system.68 There was other evidence supporting Dr. Farquhar. Dr. Johnston, the anesthesiologist, agreed that “if [the patient] does not get oxygen to his brain, he is going to die very quickly.”69 She also stated that for a patient in Mr. Miller’s condition the very first thing in importance was to “oxygenate the patient”:70
“Q. When you said oxygenation, what do you mean?
“A. Oxygenate the patient, give oxygen to the patient.
“Q. How?
“A. Whichever is the easiest and the fastest way to do it, intubate the patient, oxygenation by intermittent positive pressure breathing manually with mask or intubate the patient with endotracheal tube.”
* * * * * *
“Q. When you say oxygenation as the No. 1 step you are talking about intermediate pressure through an ambu bag and pressure breathing?
“A. Yes, whichever is the fastest, the ambu bag and mask if it is the fastest and followed by intubation if you can intubate; there are cases when you cannot intubate, then the mask may be satisfactory.”
“Q. And time again was of the essence if you would be able to help this man?
“A. Yes.”
* # * * * *
“Q. My question is is it a matter of fact that each minute of delay in a patient suffering from anaphylactic reaction to penicillin, each minute of delay without the injection of the epinephrine IV would significantly diminish their chance for survival?
“A. I can say that it is important that you have to get it in, but I don’t know that each minute counts much. I know each minute without oxygen counts and every second counts, but I would say that adrenalin, yes, each minute makes that much difference.”
* * * * * *
“Q. Isn’t it a fact that the moment the patient stops breathing voluntarily you will have to go to some artificial means if he is going to continue to receive oxygen to his brain?
“A. Yes.
“Q. And if he does not get oxygen to his brain he is going to die very quickly, is that right?
“A. If the patient is not breathing continuously.”
Dr. Johnston also stated that Mr. Miller’s respiration, when he was brought into the hospital was not providing him with sufficient oxygen:71
“Q. All right. Now, isn’t it a fact that in the condition described in this emergency room record, that he required positive ventilation assistance from the very moment he was brought to the emergency room; is that right?
“A. The patient needed respiratory assistance, I would say, because the patient was breathing four times per minute.”
Plaintiff also pointed to the fact that Mr. Miller’s pupils dilated long before his voluntary heart action ceased. Plaintiff argues that this sequence indicated that the patient died from interference with his oxygen supply, or in other words, that the patient’s respiratory arrest and the failure of the hospital to provide respiratory assistance “caused an absence of oxygen to the medullary centers of the brain, or irritated *95the myocardium because of its exquisite sensitivity to oxygen, and that caused subsequently a failure of the circulatory system.”72
Despite all this evidence, unquestionably probative on one of the two essential aspects of the causation issue, the District Court found that plaintiff had failed to adduce sufficient evidence “to form a reasonable basis for concluding that failure to provide mouth-to-mouth resuscitation . was a substantial factor which led to the inability to prevent the further progress of the anaphylactic reaction, which in turn led to the death of Mr. Miller.”73
It dismissed as “unreasonable” the evidence presented by plaintiff, stating “[t]he only specific harm that plaintiff could point to . was a reference to á diminution in respiration causing a lack of circulation which could, in turn, minimize the effectiveness of a subcutaneous injection of Sus-phrine.”74 (Emphasis added.) This statement evinces profound confusion over basic physiological facts and a thorough misunderstanding of plaintiff’s argument.
There was no testimony that respiratory embarrassment diminishes blood circulation. Blood circulation was not the problem here. Indeed, plaintiff acknowledges that circulation in the major vessels was being maintained by cardiac massage which assisted voluntary heart action. The fact that poor blood circulation existed in the small blood vessels near the skin where the Sus-phrine was injected is not relevant here; it is an entirely separate aspect of plaintiff’s case. The real harm from the hospital’s failure to assist respiration occurred because not enough oxygen was getting into the patient’s blood stream. Thus the circulating blood was not carrying an adequate supply of oxygen to the patient’s brain. The harm which plaintiff points to is not that “a diminution in respiration [caused] a lack of circulation,” as the District Court would have it, but rather that the patient’s respiratory arrest led to neurological death because the hospital failed to provide proper assistance to force oxygen into the lungs, whence it would be picked up by the blood stream and carried by the circulatory system to the brain.
It is clear, then, that the District Court confused the functions of the respiratory and circulatory systems. Because of its confusion, the court completely misapprehended the significance of the medical evidence presented by plaintiff on an essential aspect of the causation issue. Therefore, the court’s conclusion that there was no causal connection between the hospital’s improper ventilation of Mr. Miller and his death — based as it was on a fundamental misconception of the medical facts — must be set aside as clearly erroneous.
B. Adrenalin Injection.
We now turn to the District Court’s conclusion that the hospital did not breach its duty to administer an intravenous injection of adrenalin as soon as possible. We find that this conclusion must be set aside as “clearly erroneous” because it is contrary to the clear weight of the evidence.
The Court’s determination that there was no breach of duty was based on its finding that it was impossible to administer an intravenous injection prior to 9:42 a. m. because the patient’s vascular collapse made it difficult to find a vein near the body surface.75 At trial there was conflicting evidence introduced on this point, with the plaintiff arguing that an intravenous injection was possible as early as 9:36 a. m. and the defendant maintaining that such an injection was not possible until 9:42 a. m. The District Court found that the evidence was “in equipoise” and that the plaintiff had therefore failed to prove by a preponderance of the evidence that an intravenous *96injection was possible prior to 9:42 a. m.76 Apparently, the scales of the District Court were askew, for it is clear from the record that, far from being “in equipoise”, the evidence decisively supported the plaintiff’s position.
In support of her contention that an early intravenous injection was possible, plaintiff pointed, first, to the hospital’s own records and the testimony of Nurse Curry. These are the only sources of factual information as to what occurred from 9:35 a. m. to the time Dr. Jones arrived on the scene at approximately 9:38 a. m. Item 4 of Dr. Vaziri’s report states:77
4 — Solu-Cortef i.v. injection immediately once and repeated with 2 more i.v. and one i.m.
Plaintiff interprets this entry as meaning that an intravenous injection of Solu-Cortef was given by Dr. Vaziri immediately after Mr. Miller was brought into the emergency room. All the medical witnesses who were questioned at trial on this entry agreed with plaintiff’s interpretation.78 Thus, argues plaintiff, if an intravenous injection of Solu-Cortef was accomplished at that time, it was also possible to give an intravenous injection of adrenalin.
Nurse Curry’s notes are at least consistent with this argument. They read in pertinent part:79
9:38 SusPhrene (sic) 0.5 ml®Q Dr. Vaziri Solu-Cortef 2 ml Dr. Vaziri
Thus, Nurse Curry’s notes show specifically by the letters “SQ” that the 9:30 injection of Sus-phrine [adrenalin] was subcutaneous, but they do not indicate the nature of the 9:38 Solu-Cortef injection. These notes were made minute by minute, and it is at least a fair inference that Nurse Curry’s failure to qualify the Solu-Cortef entry, as she did the Sus-phrine entry, indicates that the Solu-Cortef was not given subcutaneously but intravenously. However, plaintiff did not rely on this inference, because this interpretation was confirmed by Nurse Curry at trial:80
“Q. It is my understanding that within a minute or a minute-and-a-half after Mr. Miller was returned to the hall of the Emergency Room, two drugs were administered, Sus-phrine and Solu-Cortef, that the Sus-phrine was given subcutaneously, but that the Solu-Cortef was given intravenously; is that correct?
“A. May I look at my notes?
“A. You certainly may. Let me give you the hospital records.
“A. I have it. Yes, that is correct.
“Q. And I believe you have explained to me earlier that the Solu-Cortef can be given intravenously without running IV simply by taking a needle and injecting it into a vein and injecting the Solu-Cortef?
“A. Yes.
“Q. And that in fact was how the Solu-Cortef was administered to Mr. Miller within a minute or minute-and-a-half after he was first returned to the Emergency Room; is that correct?
“A. It was closer to three minutes, and I will have to say I was not observing everything. In other words, I didn’t actually see it given, I was told that it was given at the time.”
Finally, and perhaps most persuasively, the plaintiff pointed to the undisputed fact that between 9:36 and 9:42 a. m., Mr. Miller went into deeper shock, making it physiologically more difficult to find a vein as his circulatory system collapsed.81 Nevertheless, Dr. Johnston was able to establish an I.V. on her first try at 9:42 a. m. and was able to establish a second I.V. as late as 9:45 a. m. The evidence indicates that Dr. Johnston used “medi-cut” needles from the crash cart to start these I.V.’s.82 Thus, argued plaintiff, if it was not “impossible” for Dr. Johnston to find veins at 9:42 and 9:45 a. *97m., it was certainly not “impossible” at 9:36 or 9:38 a. m. when conditions were more favorable. At trial, defendant did not attempt to challenge this inference.
In contrast to the substantial evidence presented by plaintiff, the only evidence adduced by defendant in support of its contention that an earlier intravenous injection was impossible was the hearsay assertion of a single witness, Dr. Jones. In his testimony, Dr. Jones stated repeatedly that it was “impossible” to inject adrenalin intravenously prior to 9:42 a. m.83 However, Dr. Jones did not arrive on the scene until approximately 9:38 a. m. He was not present for the first two minutes when Dr. Vaziri administered the Sus-phrine and Solu-Cortef. His only basis for concluding that it was impossible to find a vein before he arrived is that he was “told” that it was impossible.84 There is no indication in the record who told Dr. Jones that this was so.
In our view the clear weight of the evidence supports plaintiff’s position. The only evidentiary support for the Court’s conclusion that an early intravenous injection was impossible is the hearsay testimony of Dr. Jones. The District Court committed error by disregarding the persuasive evidence presented by plaintiff in favor of the hearsay assertion of a single witness.
In discounting the plaintiff’s evidence, the Court said that it was influenced by three “significant points.” 85 We have examined these “points” and have determined that they do not alter the distinct balance of evidence in favor of the plaintiff. First, the Court noted that Dr. Vaziri’s report contained no precise time reference and that parts were out of chronological order. Second, the Court observed that it was “as equally reasonable to infer” that the word “immediately” in Dr. Vaziri’s item (4) meant that the intravenous injection of Solu-Cortef was given immediately after Dr. Johnston established an I. V. as it was to infer that it meant immediately after the patient was placed on the resuscitation couch. Thus, reasoned the Court, the intramuscular injection referred to at the end of item (4) may have actually been given at 9:38 a. m. This is a somewhat strained interpretation of item (4) in view of the word “repeated”. The item says that the injection given “immediately” was “repeated ” with 2 more I.V. and one I.M. The use of “repeated” strongly suggests that the I.M. injection was subsequent to the I.V. injection which was given immediately. Moreover, there is no support for this interpretation beyond inferences from Dr. Jones’ hearsay testimony. On the other hand, plaintiff’s interpretation is buttressed by Nurse Curry’s testimony and by the circumstantial evidence afforded by Dr. Johnston’s successful I.V.’s. Therefore, we find that the first two “significant points” mentioned by the District Court have no impact on the balance of evidence.
The third “significant point” mentioned by the Court is somewhat startling. The Court states that “considering the fact that all of the doctors who testified stated unhesitatingly that if a vein were available, epinephrine [adrenalin] would be given immediately, it is a reasonable inference that Dr. Vaziri was not presented with a viable vein at 9:36 a. m.”86 In essence, the Court is assuming away the plaintiff’s case, asserting that simply because the doctor did not do something he had a duty to do, it must have been impossible to do it. This assumes the answer to the very question at issue. It is an assumption which is clearly contradicted by the weight of the evidence and which is supported only by the hearsay testimony of Dr. Jones. Therefore, we find that this third “significant point” had no impact on the balance of evidence.
In sum, then, it is clear that in reaching its conclusion that it was impossible to give an intravenous injection to Mr. Miller prior to 9:42 a. m., the District Court relied on the hearsay testimony of a single witness in the face of plaintiff’s more weighty evi*98dence to the contrary. We, therefore, must set aside the Court’s conclusion as “clearly erroneous”.
IV. PROCEEDINGS ON REMAND
Because we find that the judgment of the District Court was based on “clearly erroneous” findings of fact, we reverse the judgment and remand the cause for a new trial. This new trial may be a jury trial if the parties so request, and it should be limited in scope. Our decision here forecloses further litigation on the question of the defendant hospital’s liability for its post-injection observation of the deceased. It also upholds the District Court’s conclusion that the hospital breached its duty to provide Mr. Miller immediate positive pressure respiratory assistance at least by means of mouth-to-mouth resuscitation and its conclusion that the hospital had the duty to give Mr. Miller an intravenous injection of adrenalin as soon as it was possible to do so. These facts shall be taken as established in the proceedings on remand.
Therefore, only the following issues are to be tried: (1) whether the hospital’s failure to provide proper respiratory assistance was a substantial factor in bringing about Mr. Miller’s death; (2) whether the hospital breached its duty to give Mr. Miller an intravenous injection of adrenalin as soon as it was possible to do so, and, if so, whether this breach was, in combination with the failure to provide adequate ventilation, a substantial factor in bringing about Mr. Miller’s death; (3) the amount of damages, if any.
Our decision does not affect the validity of the directed verdicts granted the two individual doctors who were originally party defendants in this action; accordingly, these individuals may not be joined as parties in the future litigation of this cause.
Reversed and remanded.
. Hadley Memorial Hospital is located in the District of Columbia.
. The action was brought under the District of Columbia Wrongful Death Act, 16 D.C.Code §§ 2701, 2702, and Survival Act, 12 D.C.Code § 101.
. Jurisdiction was founded upon diversity of citizenship. 28 U.S.C. § 1332.
. In addition to the Hadley Memorial Hospital, the action originally named three doctors on the hospital staff as party defendants. The claim against one doctor was dismissed without prejudice, and at the close of plaintiffs case, the other two doctors were granted a directed verdict.
. Daniels v. Hadley Memorial Hospital, No. 1162-74 (D.D.C. 29 March 1976). Unpublished opinion reproduced at Joint Appendix (J.A.) 1-22.
. Id. at 8.
. Id. at 21.
. Id. at 18.
. In describing the theory of plaintiff’s case, we emphasize the difference between the circulatory system and the respiratory system, because it was the District Court’s confusion over their respective functions which lead it astray. The. respiratory system supplies oxygen to the lungs where it is exchanged and infused into the blood. The circulatory system, with its heart pump, circulates the blood carrying the oxygen to various parts of the body. The body organs require oxygen to live. Without oxygen there is necrosis.
. Transcript (Tr.) 240, 317, 337, 480, 733.
Allergic reactions to penicillin range from a mild rash to anaphylaxis which may rapidly prove fatal unless proper therapy is promptly administered. The pathogenesis of the anaphylactic reaction was described: “Vasodilation and escape of plasma into the tissues results in a decrease in effective plasma volume. Fluid escapes into the lung aveoli and may produce pulmonary edema. Obstructive angioedema of the upper airway may also occur.” The certificate of death states that Mr. Miller’s “[djeath was caused by Laryngeal and Pulmonary edema due to Anaphylactic reaction to Penicillin for infected abrasions Arm and Leg.”
. Transcript (Tr.) 135, 139, 455-56, 481, 764.
. Tr. 135, 734, 764.
. Tr. 734.
. Intravenous (or “I.V.”) means “into a vein.”
. Intracardiac means “into the heart.”
. Subcutaneous means “under the skin.”
. Intramuscular (or “I.M.”) means “into the muscle.”
. Tr. 240, 449.
. Tr. 135, 139, 455-56, 481.
. Tr. 244.
. Sus'-phrine is the brand name of one form of adrenalin.
. Tr. 623.
. Prior to trial, Dr. Vaziri returned to Iran.
. Tr. 763.
. Tr. 764-65.
. J.A. 2.
. Tr. 764.
. Tr. 240, 333, 449, 639.
. Nasal cannulas are devices which release oxygen into the nostrils for the purpose of making pure oxygen available to the patient if he is breathing voluntarily. They are not a means of oxygenating a patient whose respiration has become seriously impaired or ceased altogether. See, e. g., Tr. 117, 246-7.
. Tr. 764.
. Solu-Cortef is a form of corticosteroid.
. Tr. 638.
. Tr. 639.
.Tr. 764.
. Tr. 777, 780.
. Tr. 632, 780.
. Tr. 240, 449.
. Tr. 461-62.
A “cut down” is a procedure used to establish an intravenous route by making an incision into the flesh in order to find a vein.
. Tr. 632, 780.
. Tr. 366, 764.
. Tr. 244, 764.
. Tr. 364-66, 765.
. J.A. 11-12.
. See, e. g., Tr. 139, 241-43, 246, 455-56, 458, 481, 693-94, 746.
. J.A. 20.
. J.A. 14.
. See, e. g„ Tr. 120-21, 241, 243-46, 752.
. J.A. 18.
. Findings regarding negligence and causation are deemed questions of fact for purposes of review and are to be tested by the clearly erroneous standard. McAllister v. United States, 348 U.S. 19, 75 S.Ct. 6, 99 L.Ed. 20 (1954); Socash v. Addison Crane Co., 120 U.S. App.D.C. 308, 346 F.2d 420 (1965). See generally 9 C. Wright & A. Miller, Federal Practice and Procedure § 2590.
. Fed.R.Civ.P. 52(a). Zenith Radio Corp. v. Hazeltine Research Inc., 395 U.S. 100, 89 S.Ct. 1562, 23 L.Ed.2d 129 (1969).
. E. g., United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948); Case v. Morrisette, 155 U.S.App. D.C. 31, 38, 475 F.2d 1300 (1973).
. E. g., Blue Bird Food Products Co., v. Baltimore & Ohio R. Co., 474 F.2d 102 (3rd Cir. 1973), appeal after remand 492 F.2d 1329 (1974); Rewis v. United States, 445 F.2d 1303 (5th Cir. 1971), appeal after remand 503 F.2d 1202 (1974); Western Cottonoil Co. v. Hodges, 218 F.2d 158 (5th Cir. 1955).
. E. g., Western Cottonoil Co. v. Hodges, 218 F.2d 158 (5th Cir. 1955); Apex Mining Co. v. Chicago Copper & Chemical Co., 340 F.2d 985 (8th Cir. 1965).
. Graham v. Roberts, 142 U.S.App.D.C. 305, 308 n. 3, 441 F.2d 995 (1970). See Restatement (Second) of Torts § 431 (1965).
. 368 F.2d 626 (4th Cir. 1966).
. Id. at 632.
. See, e. g., Clark v. United States, 402 F.2d 950, 953 (4th Cir. 1968); Thomas v. Corso, 265 Md. 84, 102, 288 A.2d 379 (1972); Kallenberg v. Beth Israel Hospital, 37 N.Y.2d 719, 374 N.Y.S.2d 615, 337 N.E.2d 128 (1975).
. See, e. g., Rewis v. United States, 503 F.2d 1202 (5th Cir. 1974), appeal añer remand 536 F.2d 594 (1976); McBride v. United States, 462 F.2d 72, 75 (9th Cir. 1972).
. We stress that these factors relate only to the question of causation and become significant once the plaintiff has already established a standard of care and a breach of that standard.
. Tr. 135.
. Tr. 124, 146.
. Tr. 696-97.
. Tr. 723.
. Tr. 627-28, 640.
. Tr. 250.
. J.A. 20.
. Tr. 121.
. Tr. 138-41.
. Tr. 247.
. Tr. 241-246.
. Tr. 455-56.
. Tr. 717.
. J.A. 19.
. J.A. 20.
. J.A. 18.
. Ibid.
. Tr. 763.
. Tr. 331-33, 633. Dr. Jones, however, stated that he thought the sequence was mistaken.
.Tr. 764.
. Tr. 331-33 (emphasis added).
. -Tr. 638.
. Tr. 632, 780.
. Tr. 629-30.
. Tr. 629.
. J.A. 16.
. J.A. 17.