The parents of Ms appeal from a 1996 judgment of the Boston Juvenile Court adjudicating their young daughter as in need of care and protection, dispensing with their need to consent to her adoption, and committing Ms to the custody of the Department of Social Services (department) until the adoption plan for her can be effectuated. The proceedings below, as well as the department’s involvement with Ms and her parents, all result from a single unexplained head injury which Iris sustained in the summer of 1993 when she was barely one month old.
*96The parents complain that the burden of proof at trial was unconstitutionally shifted from the department to them, i.e., instead of the department’s being made to prove parental unfitness by clear and convincing evidence, they were instead required to prove their fitness as parents by explaining the genesis of Iris’s injury. They also argue that the evidence presented by the department did not prove them unfit in the requisite clear and convincing manner and that certain inadmissible and prejudicial hearsay evidence, particularly in the medical records, was improperly admitted against them over their repeated objection. We conclude that neither the inadequate evidence offered by the department nor the judge’s subsidiary findings of fact are sufficient to support, by clear and convincing evidence, “the grave conclusion of unfitness.” Adoption of Katharine, 42 Mass. App. Ct. 25, 27 (1997).
Background. Iris suffered bilateral skull fractures with bleeding into the brain, leaving her permanently damaged. All concerned agree that the cause was trauma. What is unclear, however, is the nature of the trauma, i.e., whether it was inflicted abuse or accidental injury, who caused the injury, and who was present when it occurred. Iris’s parents have consistently maintained that they do not know how this injury happened.
For almost two years after Iris was discharged from the hospital in the department’s temporary custody directly into a specialized foster home, where she still remains, the department’s goal was to reunite the family. Toward that end, Iris’s parents, who had not had prior involvement with the department, complied fully with the department’s many service plan requirements and displayed generally exemplary parenting behavior. Notwithstanding this and despite internal departmental differences of opinion on the matter, the department’s goal for Iris changed in mid-1995 from family reunification to adoption. The goal change apparently occurred solely because, in the two years preceding it, the department had not been able to learn from Iris’s parents the circumstances of her head injury. The department ultimately embraced the view that Iris’s parents are unfit because (a) they inflicted or otherwise caused the injury themselves but will not admit it; (b) they know but will not disclose who caused the injury; or (c) they should know, were they nonnegligent parents, who caused the injury, but do not. The department concluded that, absent suitable parental explanation of the injury’s genesis, Iris *97could not be assured of protection in the home of such parents but would instead be at great risk of harm if returned to them.
After three days of trial in October, 1995, the judge on May 21, 1996, issued his findings of fact, conclusions of law, ultimate findings, and order. He found Iris’s parents to be currently unfit and lacking in the capacity, ability and readiness to assume parental responsibility. This ultimate finding rests upon thirty-one subsidiary findings of fact, as well as several other fact findings intermingled with conclusions of law. Several of the subsidiary findings are of particular importance: (a) that the parents were either the “causal agent” of Iris’s injuries or were negligent in leaving her with inappropriate caretakers who caused the injury; (b) that Iris was “severely physically abused”; and (c) that the parents know but refuse to explain how Ms sustained the injuries.
Insofar as pertinent to these key determinations, we recite other findings made by the judge. Julia, born on October 7, 1973, and Paul, born on March 27, 1964, are the unmarried biological parents of Ms, their only child, born on June 30, 1993. Neither parent has a criminal record. Each was previously married and divorced. The father and his former wife have one child together, a son bom with spina bifida, whom the father visits regularly and supports. The father was unemployed at the time of Mai, having previously been laid off from his job.2
In the three weeks after Ms’s birth, her mother and father took her to physicians on three occasions to be examined for various reasons; she was found to be healthy.3 Late in the afternoon on Sunday, August 1, 1993, the mother called Ms’s pediatrician telling him of Ms’s glassy, tired looking eyes and of her being feverish and fussier than usual. On his advice, the mother and father brought Ms to a local hospital around 5 p.m. A spinal tap *98was performed, revealing blood, and causing the physician to suspect trauma. Iris was transferred to Children’s Hospital in Boston (BCH) for further evaluation about 8 p.m. There were no visible or palpable signs or marks of trauma on Iris’s scalp. When Iris arrived at BCH, around 9:50 p.m., she was having seizures and was placed on a respirator and admitted to the intensive care unit. A CT scan revealed left subdural and right posterior hemorrhage within the brain substance, and a left parietal skull fracture.
According to notes he made on Iris’s hospital chart, Dr. Eh Newberger of BCH’s Child Protection Services reviewed Iris’s record and examined her at 12:30 p.m. on Monday afternoon, August 2, 1993. The notes, set out in full in the margin,4 do not indicate when the head injury likely occurred, except to say “over an interval of days,” nor do they indicate that the head injury was likely inflicted rather than accidental. The judge, however, apparently relying instead upon the department’s 51A report taken from a social worker at BCH and upon the court investigator’s report summarizing that 51A report, recited that the “abrasion on the stomach was determined [by Dr. Newberger] to be about seventy-two hours old and all the other injuries were determined to be under thirty-six hours old. Dr. Newberger determined that the skull fracture was certainly caused by abuse. With the abrasion on the stomach, Dr. Newberger determined that it could not have been caused by the child scratching herself.” (Emphasis supplied.)
Again in apparent reliance upon a portion of the 51A and investigator’s reports, the judge stated that another BCH physician, Dr. Churchwell, had at some point examined Iris and *99concluded that Iris’s head injury was due to a “significant blow to her head” and that the “skull fracture occurred within twenty-four to forty-eight hours of Iris being brought to the hospital.” The judge did not clarify whether the hospital in question was BCH or the local hospital to which Iris’s parents first brought her. Also, the very next sentence of the same 51A report on which the judge relied states that on August 3,1993, Dr. Church-well believed that the fracture had occurred within the past 72 hours. The judge made no specific finding as to when the injury occurred.
The judge found that BCH staff had concluded that the only way Iris’s head injury, which was not due to brittle bone disease, could have occurred was for Iris to have experienced a sharp blow to her head, such as being hit with or against something or having a fall. He found that Iris’s parents told BCH staffers and the department’s 51A investigator that they were Iris’s only caretakers and did not know how the injury occurred. He found that the parents’ explanation of how Iris might have been injured while having her photograph taken at Wal-Mart on August 1 was not consistent with Iris’s serious injuries, although he does not discuss why he finds the explanation unsatisfactory. The judge found that although Iris’s parents had permission to remain overnight at BCH after having visited with her for most of the day on Monday, August 2, 1993, they did not stay over. The judge also recited summary portions of the psychological evaluation reports done of the parents, but he did not clarify what findings he made as a result of these reports, nor did he address the recommendation made in the psychologist’s report that Iris be returned to the home of her parents with a high level of precautions being taken by the department. 5
The judge found that the department had worked with Iris’s parents, offered them a number of services and provided them and family members visitation with Iris. He described Iris’s present health and prognosis6 and determined that she requires “a *100structured and secure environment where she will receive appropriate parenting and supervision.” No findings were made as to any specific special needs or required services.
On the basis of the foregoing, the judge concluded that Iris received injuries while in the care and custody of her parents, that she was severely physically abused, that her parents either inflicted the injuries or know who did, that they refuse to disclose this information, that their refusal to explain how Iris was injured creates a high risk that she could receive further injuries or abuse if returned to her parents, and that Iris’s parents are, accordingly, currently unfit to be her parents.
Discussion. In making these regrettably sparse findings, the judge relied almost exclusively upon the evidence offered by the department. This consisted of the testimony of its one witness, the case social worker, and twelve unredacted exhibits,7 many replete with multilevel hearsay. The department did not call any expert witnesses to assist the court, particularly with respect to the crucial medical evidence. Certain of the judge’s evidentiary rulings effectively precluded the parents from rebutting the hearsay offered against them.8 We note, too, that the department itself acknowledged, when successfully resisting the parents’ *101motion for directed verdict, that the evidence it offered did not prove who was responsible for Iris’s injury, maintaining instead that its evidence established only “that the parents were either the inflictor in the injury or negligent in caretaking’’ and that “these parents are unable to keep this child safe.”
When taken together, the judge’s findings do not provide clear and convincing evidence of current parental unfitness. Those findings are scant and, variously, without requisite detail, specificity or competent evidentiary support. The judge largely ignored, without indicating why, not only the testimony and exhibits offered by the parents but also troublesome contrary material facts in the department’s own exhibits. His findings do not address certain factors which deserved consideration. The judge’s findings do not display satisfactorily the degree of close and careful attention to the evidence that is demanded in cases of this nature. See Custody of Two Minors, 396 Mass. 610, 619 (1986); Custody of Eleanor, 414 Mass. 795, 799 (1993); Adoption of Quentin, 424 Mass. 882, 886 (1997); Adoption of Stuart, 39 *102Mass. App. Ct. 380, 381-383 (1995). While Iris’s welfare and best interests were undoubtedly the central focus of both the judge and the department in their desire to spare her the risk of further injury, good intentions and genuine concern are not a satisfactory substitute for clear and convincing evidence. The “[rjemoval of children irrevocably from their biological parents is an exceptionally far reaching exercise of State power .... In recognition of the constitutionally protected interest of parents in maintaining the natural bond with their children, a judge must find by clear and convincing evidence that a parent is currently unfit to further the child’s best interest” (citations omitted). Adoption of Katharine, 42 Mass. App. Ct. at 27.
To begin, there was insufficient proof that Iris was in the exclusive custody of her mother and father during the period in which it is thought the injury happened, such that they may reasonably be seen as having caused the injury themselves. Given the imprecise time frame for the injuries postulated by Drs. Newberger and Churchwell, the injury can only reliably be said to have occurred sometime over the weekend of Friday, July 30, to Sunday, August 1, 1993. Based on what appears to be undisputed evidence, several individuals provided child care for Iris on Friday and again on Saturday. The maternal grandparents babysat late Friday afternoon. Iris’s maternal grandmother went with the mother and father to Wal-Mart on Saturday morning, where Iris had her first baby picture taken and, later in the day, the mother and father went to a furniture store with the maternal grandmother and Iris’s aunt and uncle. While at the furniture store, the grandmother, aunt and uncle cared for Iris for approximately one hour while the parents shopped. Iris continued to be in the company of all of these adults throughout the evening, from time to time without her parents being present. Nonetheless, in spite of the fact that Iris’s maternal grandmother, aunt and uncle all cared for Iris at different times on Friday and during the day on Saturday, July 31, 1993, there is no evidence that the department interviewed these persons during its investigation of Iris’s case. Nor, at trial, did the department call Iris’s maternal grandmother, aunt or uncle to testify. Indeed, when the maternal grandmother testified at trial for the parents, the department chose not to cross-examine her. There was no evidence before the court that any of these individuals was an inappropriate caretaker or that the *103parents had adverse information about any of these persons which would have alerted them to shortcomings.
Nor was there evidence sufficient to establish that the parents knew who caused the injury or how or when it happened.9 Ms did not display unusual symptoms until the afternoon of Sunday, August 1, 1993. There were no visible or palpable marks on Ms’s head and she only began having seizures upon her transfer to BCH. The evidence does not clarify how a parent would know, before the onset of symptoms, that the child had been injured, let alone severely. This would be all the more problematic were the parents absent when the injury-causing event had occurred. Indeed, the evidence does not disclose the likely time span between the occurrence of such an injury and the onset of symptoms, or what the various discernible symptoms might be.
Additionally, the department did not present any evidence explaining how much force would be necessary to cause a skull fracture to a four week old baby. The medical records in evidence indicate only that such a fracture would require a “significant blow to the head” and that Ms experienced a “sharp blow to her head [consistent with] being hit with or against something or having a fall.” The department, however, did not call Dr. Newberger or Dr. Churchweíl, or any other medical professional, to testify at trial. It remains unknown just how “sharp” or “significant” a blow would have to be in order to cause such injury to an infant, or from what height or onto what type of surface an infant would have to fall to sustain such an injury.
Such desirable but unoffered evidence would also have been pertinent to another of the judge’s key findings, viz., that Ms had been severely physically abused or, otherwise put, that the injury was not accidental but inflicted. There was evidence in the form of the 51A report, later separately summarized by the court investigator, that Dr. Newberger believed the injury had been inflicted.10 Dr. Newberger saw the child only once and his own *104notes on Iris’s medical chart reveal no such conclusion. To the extent that the department’s social worker accurately transcribed what the BCH social worker told her of Dr. Newberger’s comments, in whatever manner and in whatever circumstances those comments had themselves been related to her, there was still nothing to explain on what basis Dr. Newberger thought abuse rather than accident was more consistent with Iris’s injuries. Particularly when coupled with the absence of medical evidence as to the amount of force necessary to cause an infant skull fracture, the judge was impermissibly left to speculate on the nature of the trauma.
It is evident from the judge’s findings that his determination of parental unfitness depended almost entirely upon the one incident of serious and unexplained injury to the infant Iris, who at the time of trial was two years old. Parental unfitness, however, “must be determined by taking into consideration a parent’s character, temperament, conduct, and capacity to provide for the child in the same context with the child’s particular needs, affections, and age.” Adoption of Quentin, 424 Mass, at 886, citing Adoption of Mary, 414 Mass. 705, 711 (1993). See Adoption of Stuart, 39 Mass. App. Ct. at 391-392; Adoption of Harriet, 29 Mass. App. Ct. 111, 113 (1990). The judge failed to take these factors into consideration.
We note in this regard that there is no evidence that the father and mother have a history of neglect or abuse of any kind prior to this one unexplained event. They also have no criminal, substance abuse, social service or mental health history. The judge failed to consider the undisputed evidence that the mother and father complied fully with all service plans, none of which required that they explain Iris’s injury. The parents have cooperated with all of the department’s numerous demands, submitting *105willingly to evaluations by a social worker and psychologist at the court clinic, agreeing to individual therapy, accompanying Iris to all medical appointments, attending training whenever offered them in Iris’s special needs, and visiting with Iris at every possible opportunity. The department consistently found the parents’ behavior with Iris to be “appropriate” and cooperative. Iris in turn had formed an ongoing bond with her parents, responding positively to their visits, recognizing them and greeting her “mama” and “dada.” The findings also fail to take into account the parents’ willingness and ability to care for children with severe disabilities, illustrated not only by their behavior toward Iris but also by their ongoing care for the father’s son by a prior marriage who has spina bifida.
The department’s burden at trial was to prove current parental unfitness by clear and convincing evidence. This standard of proof is an intermediate one; it “involves a degree of belief greater than the usually imposed burden of proof by a fair preponderance of the evidence, but less than the burden of proof beyond a reasonable doubt imposed in criminal cases. . . . The evidence must be sufficient to convey to ‘a high degree of probability’ ” that the proposition is true. Tosti v. Ayik, 394 Mass. 482, 493 n.9 (1985), cert, denied, 484 U.S. 964 (1987). The requisite proof must be strong and positive; it must be “full, clear and decisive.” Callahan v. Westinghouse Broadcasting Co., 372 Mass. 582, 584 (1977). See Liacos, Massachusetts Evidence §§ 5.2.2 - 5.2.3 (6th ed. 1994); Ireland, Juvenile Law § 107 (1993). If this standard of proof is to have meaning in cases involving the possible termination of parental rights, it cannot be said here that the evidence before the judge clearly and convincingly proved unfitness.
The evidence presented by the department was not strong, positive, full or decisive; it left far too much to speculation, conjecture and surmise. To be sure, there was some evidence suggestive of an isolated instance of abuse or neglect, the cause of which may be known to the parents. However, the evidence before the court was as consistent with the cause of the injury being accident as it was abuse, as consistent with whatever caretaker was present at the time the injury occurred not knowing as knowing that a severe trauma had been dealt the child, and as consistent with the parents knowing as not knowing how and when and when the injury happened and who was present when it did. We *106cannot know what light the testimony of a medical expert or that of Iris’s other relatives might have shed on these issues, what more complete pretrial investigative work by the department might have revealed, or what other evidence the department might have offered. Suffice it to say that the department chose to present this case, with its myriad perplexing difficulties, in a minimalist manner. It was inadequate in the event to establish current parental unfitness by the requisite degree of proof.
Further proceedings. Because the judge’s findings do not support the conclusion of current parental unfitness, we vacate the judgment. We remand the matter for further hearing forthwith to reinstate parental visitation with Iris under appropriate terms and conditions and to determine whether the department’s goal remains the termination of parental rights. If so, the court shall determine whether the department has sufficient additional evidence to warrant a new trial and, if it does, to proceed forthwith to trial consistent with this opinion. If a new trial is unwarranted or the department’s goal has changed, a plan for the prompt reunification of the family under appropriate terms and conditions shall be devised and implemented with court oversight and approval.
So ordered.
The judge incorrectly found the father had been laid off from his job prior to July, 1993. The evidence is clear that the father had been a full-time employee at Raytheon for five years prior to being laid off in December, 1994. He was on vacation the week before August 1, 1993, and the mother, who worked as a sales clerk at Filene’s, was then on maternity leave.
There was also evidence that the parents had brought Iris to a local emergency room for seemingly abnormal crying on July 5, 1993, where a protruding and herniated belly button and scratch on the roof of the child’s mouth were observed. There was evidence that Iris had difficulties with mucus, with taking her formula and that she was a fussy baby who cried frequently. The baby was switched to a soy formula by a new pediatrician whom the mother consulted after becoming dissatisfied with another pediatrician.
“My exam discloses probable retinal hemorrhages on the L (difficult to see; I could not get a good look at the R), fading ecchymosis about and on the L shoulder over a distribution of 1.5 - 2 cm., a laterally distributed healing approx. 1x2 cm. healing hyperpigmented brown abrasion 5 cm. superior to the umbilicus, and no palpable or visible stigmata of trauma in the scalp. The sclerae are quite blue.
“The fading bruises are to my estimate 72 hrs.
“Taken together with the CT findings, exam yield indicates trauma to several body planes, occurring over an interval of days. Osteogenesis imperfecta, in my view, would not account for the nature and distribution of the findings. A bone scan would be of additional diagnostic value.
“Suggest also:
1) 51A report to DSS
2) Defer discharge until reasonably well assured home is safe.”
The psychologist notes in her report: “There is no way to eliminate risk in this case and because of the lack of a specific perpetrator and the lack of needed information, the failure to find a strong pattern consistent with abusers, it is difficult to be confident about the level of risk for re-abuse.”
“Iris is a quadriplegic and can stand if held by a walker. Iris has poor muscle tone in her arms and legs, and it is uncertain whether she will ever be able to walk on her own. Although the full extent is unknown at this time, Iris will continue to be neurologically impaired. . . . It is uncertain whether Iris’s *100condition will improve in the future. Iris has neurological signs of brain damage and problems with sucking, increased tone especially on the left side, difficulty with visual tracking, and tongue curling and protrusion.”
These included Iris’s hospital records, clinical evaluation and psychological testing reports of the parents, the court investigator’s report, to which G. L. c. 119, §§ 51A and 51B, reports were appended, department reports to the court and the department’s adoption plan for Iris.
In view of our disposition, we need not reach the parents’ claims of error in the judge’s evidentiary rulings. In the event of retrial, however, it may be of use to address briefly certain of these issues. The parents filed a motion in limine at the time of trial to exclude from evidence portions of the hospital records, the 51A and 51B reports and the investigator’s reports. In so doing, they sought in part to exclude portions of the medical reports stating that Iris’s medical condition resulted from abuse. In part, also, these documents contained multilevel hearsay to which the parents objected. Initially, the judge ruled, incorrectly, that the motion was untimely brought, but subsequently permitted brief argument before denying the motion and related objections during trial. Redaction of the documents appears not to have been considered by the judge. The general admissibility of case work documents and court investigator reports is no longer seriously in question. See Adoption of Paula, 420 Mass. 716, 725 (1995); Adoption of George, 27 Mass. App. Ct. 265, 274 (1989). The quid pro quo, as it were, for the admission of such otherwise problematic documents, however, is that the compilers of and sources for the documents be available for cross-examination, giving parents the opportunity to rebut any adverse or *101erroneous information. This opportunity was not afforded to the parents here. In this regard, we observe that the department was inexplicably dilatory in meeting its discovery obligations and apparently only belatedly provided the parents two years of file materials, and only at trial disclosed a list of its exhibits and witnesses. During trial, the department noted that the preparers of the 51A and 5 IB reports were available to testify, as were the court investigator and the psychologist. The trial judge declined the parents’ request to continue the trial and, when the parents subsequently called the court investigator to testify, did not permit them to cross-examine him, but instead required them to engage in direct examination of the witness, with predictably unsatisfactory results. When the judge ruled that direct and not cross-examination would also be required of the parents if they called the psychologist, they understandably chose not to do so. This was improper. We note also that, with respect to case work documents, the expectation is that they be admitted with “opinion, evaluation, and judgment material” edited out wherever feasible. Adoption of George, supra at 274. Of course, “[tjo’the extent that the source of information in a document of the sort here in question is available for cross-examination, more leeway may be afforded material that smacks of opinion, evaluation, or judgment.” Ibid. As to the hospital records themselves, admitted under G. L. c. 233, § 79, we are not aware of any instance in which an opinion was expressed or diagnosis made that the injury was in fact caused by abuse; hence we need not address whether the use of the word “abuse” would be related to medical diagnosis or treatment for the purposes of § 79. See Commonwealth v. Bohannon, 385 Mass. 733, 750 (1982). We note, however, that “battered child syndrome” has come to be a well recognized medical diagnosis as to which qualified physicians may testify. See Commonwealth v. Day, 409 Mass. 719, 724 (1991); Commonwealth v. Azar, 32 Mass. App. Ct. 290, 302 (1992). No physician testified at trial here, however.
The department’s social worker testified over objection that she “believes” the parents know that something happened in the period between July 30 and August 1, 1993. She thereafter testified that she “always thought it was logical that if a child had this serious of an injury that the person caretaking the child was aware that the injury had occurred.” If admissible at all, itself a dubious proposition, this evidence is certainly insufficient on the point.
To the extent that Dr. Newberger is reported to have concluded that the skull fracture was inflicted abuse rather than accidental injury, it appears that this opinion rests at least in part on the significance he attributes to the fading bruise *104on Iris’s left shoulder and the abrasion near her belly button. In a telephone conversation with the department’s investigator on August 4, 1993, it was reported that Dr. Newberger “feels that the bruise on the left shoulder and the skull fracture could have occurred at the same time. He wonders if someone ‘grabbed the child’s skin, possibly pinched’ and ‘flung or banged her against something.’ ” Dr. Churchwell, however, is reported by the same investigator as believing that the head fracture was fresh but that the bruise on the left shoulder is “old” because “greenish . . ., small and faint. Dr. Churchwell said it was unclear “ho[w] child would get a leaflike scratch on her abdomin [j'zc]. He said he was not as concerned about the scratch o[r] the left shoulder injury as was Dr. Newberger.” The judge made no findings resolving or reconciling this apparently contradictory documentary evidence.