(dissenting, with whom Wilkins and Lynch, JJ., join). The primary question in this case is whether the evidence was sufficient to warrant a finding that negligence on the part of the defendant Honeyman caused or contributed to cause the suicide of Karen McNamara. The court has concluded that the testimony of the plaintiffs’ expert, Dr. Michael Nelson, “fairly read, when considered with the other evidence in the plaintiffs’ favor concerning Karen’s behavior and condition, warranted a finding that Honeyman deviated from accepted medical practice expected of a psychiatrist. . . by removing Karen from one-on-one observation,” and that that deviation “was the proximate cause of her death.” Ante at 51. I do not agree.
The court characterizes Karen’s conduct on several occasions as “self-destructive,” and informs us that, at least twice, members of the hospital staff placed her on one-on-one observation. She was on one-on-one observation from December 2, 1980, until Honeyman removed her from that status on December 16. It is not entirely clear what importance the court attaches to that evidence, or to the further evidence that, on December 16, Karen’s boy friend informed a hospital staff member (not shown to have been Honeyman or to have passed the information along to Honeyman) that Karen had claimed to have attempted suicide that day. Also, it is unclear what significance the court attributes to the evidence that, shortly before her death, .Karen presented herself to a staff meeting attended by Honeyman and, “ [although she *57seemed anxious and distressed, she was sent away.”1 In any event, that evidence does not pass the test for sufficiency set out in Brune v. Belinkoff, 354 Mass. 102, 109 (1968) and its progeny. See Stepakoff v. Kantar, 393 Mass. 836, 840 (1985). Surely, the recited evidence, unsupplemented by adequate expert testimony, would not have warranted the jury in finding that Honeyman, in removing Karen from one-on-one observation, or in failing to predict and prevent Karen’s suicide, failed to exercise the degree of skill and care customarily exercised by the average qualified psychiatrist in the circumstances.
The court’s expressed conclusion that Nelson’s testimony in conjunction with other evidence supports the finding of negligence implies, if it does not explicitly state, the court’s recognition that Nelson’s testimony was critical. Therefore, fair appellate review demands careful analysis of Nelson’s testimony, a type of analysis, I suggest, that is not present in the court’s opinion.
Nowhere in his testimony did Nelson even suggest that, in Honeyman’s treatment and supervision of Karen, and aside from record-keeping, Honeyman deviated from accepted medical practice. Nelson did not testify, either explicitly or implicitly, that Honeyman had failed to treat Karen with the skill and care that would have been exercised by an average competent psychiatrist. Rather, as is demonstrated by the following excerpt from the record, Nelson focused on what he perceived to be Honeyman’s faulty record-keeping. He testified that “good medical practice” requires documentation of many kinds of information including information about the level of patient supervision required from time to time and about the reasons for changing the intensity of supervision whenever changes are deemed appropriate. He also testified that Honeyman’s documentation fell below the stan*58dard of “good medical practice,” and that the reason good documentation is required is that the patient receives care from a variety of people working at different times of the day, and close, accurate communication between them is essential. Nelson testified on direct examination as follows:
Q. “In your opinion, did Dr. Honeyman’s treatment of Karen McNamara, as documented, comport with good medical practice?”
A. “As documented, no.”
Q. “What is your basis for that opinion?”
A. “Examination of the record.”
Q. “What in the record led you to conclude that his treatment of Karen McNamara, as documented, did not comport with good medical practice?”
A. “There was no admission history by Dr. Honeyman. Review of her previous record. There was no treatment plan. There was no regular review of the treatment plan. There was no consistent pattern of recorded observations in the progress notes. There was no recorded discussions by the staff concerning Dr. Honeyman’s decisions to change observational status. There was no assessment documented according to the rationale in many cases from one status to another. There was no documentation of the steps that should have been established prior to Karen McNamara’s discharge to the Hill Program and there was no documentation of her status on return.”
Q. “Now, do you also have an opinion, sir, as to whether Dr. Honeyman’s failure to comport with good medical practice contributed to Karen McNamara’s death?”
A. “Yes.”
Q. “What is that opinion?”
A. “His failure to comport to good medical practice, as documented, did contribute to her death.”
Nelson also was asked, “Do you have an opinion as to the extent of Dr. Honeyman’s deviation from good medical practice?” Nelson answered, “As documented'?” The plaintiffs’ *59counsel said, “As documented.” Nelson answered, “Significant.”
On cross-examination, the following questions were asked and answers given:
Q. “If it were a fact, whether it was documented or not in the record, that the team meetings that were held three times a week during Karen’s admission, if it were a fact that her behavior and condition and treatment and plans for her treatment in the future were discussed at those team meetings, would that tend to change that opinion about Dr. Honeyman’s treatment?”
A. “That would tend to change my opinion, with a proviso.”
Q. “With a proviso?”
A. “Yes, That proviso is that documentation is part of good medical practice.”
Plaintiffs’ counsel carefully inquired of Nelson about Honeyman’s treatment of Karen “as documented,” and Nelson’s answers were expressly limited in the same way. Indeed, when, on one occasion, counsel asked whether Nelson had an opinion about whether “Dr. Honeyman’s failure to comport with good medical practice contributed to Karen’s death,” Nelson answered, “yes,” and, when asked for that opinion, his answer was that Honeyman’s “failure to comport to good medical practice, as documented, did contribute to her death” (emphasis added). Nelson supplied the limitation that had been omitted from the question. As the court acknowledges in its “New trial” discussion, ante at 55, “Dr. Nelson qualified his opinion ... by stating that he was commenting only on the treatment as documented in the medical records.”
One point must be made crystal clear. Nothing in Nelson’s testimony remotely suggests that there was information in the hospital records that demonstrated Honeyman’s negligence. Rather, Nelson’s testimony can only reasonably be understood to mean that, if the treatment and supervision actually given were limited to what was contained in the records, and one were to conclude that Honeyman had failed *60to take an admission history, had failed to review prior records, had failed to establish a treatment plan, had failed to regularly observe the patient, and had had no discussion with the staff, because those events were not documented, then his performance did not comport with good medical practice. Such testimony does not help the plaintiffs because there was no evidence that Honeyman’s performance was limited to whatever was recorded in the hospital charts. Surely, the jury would not have been warranted in inferring from Honeyman’s recording failures that he had not established a treatment plan or done other undocumented things. Even if Honeyman’s testimony as to all he had done were disbelieved by the jury, the jury could only have speculated as to whether Honeyman had failed with respect to treatment or had failed with respect to record-keeping only. Neither the court nor anyone else suggests, nor could fairly suggest, that the jury would have been warranted in finding that Honeyman’s negligent record-keeping had anything to do with Karen McNamara’s suicide.
The court’s reasoning, as set forth in its opinion, with respect to the question of the sufficiency of the evidence to establish Honeyman’s negligence, is brief. The court notes, ante at 51, Nelson’s testimony that, “as described in the records,” Karen’s behavior was out of control throughout her hospital stay, that Nelson would not consider her stabilized in the absence of two or three weeks of consistent safe behavior, that fifteen-minute checks were “not sufficient to manage her condition,” and that Karen’s death was “preventable” if Karen had been kept on one-on-one observation. The court concludes, ante at 51, without reference to any other expert testimony, that “[a] juror could, then, have inferred from Dr. Nelson’s testimony that Honeyman’s negligence in failing to keep Karen on one-on-one observation was the proximate cause of her death.” Certainly, it is true that the jury properly could have found that, had Karen been kept on one-on-one observation status, the suicide probably would not have occurred. The critical flaw, however, in the plaintiffs’ case and in the court’s determination, is that Nelson did not tes*61tify that Honeyman’s decision to remove Karen from one-on-one observation fell below the standard set by the average competent psychiatrist. See Stepakoff v. Kantar, 393 Mass. 836, 840 (1985); Brune v. Belinkoff, 354 Mass. 102, 109 (1968). Nelson focused, not on negligence in treatment, but on causation only. Surely, the unhappy fact that, as things turned out, fifteen-minute checks “were not sufficient to manage [Karen’s] condition,” that is, to keep her from committing suicide, does not permit the conclusion that Honeyman’s management of Karen was less than ought to be expected of an average competent psychiatrist. Surely, also, the jury could not reasonably have interpreted Nelson’s testimony as measuring Honeyman’s conduct against the Brune v. Belinkoff standard.
The evidence was insufficient to warrant a finding of Honeyman’s negligence except, perhaps, with respect to record-keeping. The evidence was insufficient to show any causal connection between faulty record-keeping and Karen McNamara’s death. No other adequate evidence of negligence attributable to the defendant hospital is demonstrated in the court’s opinion, and I am unaware of any. No other ground for recovery against either defendant has been established. Therefore, I would reverse the judgment and order judgment for the defendants.
The testimony by a treatment team member named Walsh was that, while the team was engaged in a meeting at the east end of a nurse’s station, Karen came to the door, and that someone opened the door and asked her to wait. The record is bereft of evidence that Honeyman saw Karen or was made aware of her presence.