Point-by-point Rebuttal of CDC ‘Trip report, Epi-Aid 2004-20: An investigation of a dual infection with E. coli O157:H7 and Salmonella serotype Cholerasuis var. Kunzendorf in a research laboratory worker.’

The point-by-point rebuttal itemizes all erroneous data and data interpretations included under ‘Background’, ‘Methods’, ‘Results’ and ‘Addendum’ of the original report. Points of ‘Discussion’ are only addressed in so far as they include new data or data interpretations that have not already been mentioned. Major omissions of the report are then documented.  Finally, a breach of confidentiality under ‘Recommendations’ made to the Facility is documented.

Point-by-Point Rebuttal

Background

(1)  Page 2, line 6, “…general laboratory area that was assigned to Patient A, …”

Although the laboratory was assigned to Patient A by the Research Leader, this laboratory included a shared -70 degree freezer where multiple groups stored bacterial pathogens. The adjoining BSL2 laboratory was a shared facility where equipment (e.g. shaking incubator, colony counter, autoclave, large ‘spiroplater’ in hood) was used by several research groups at Facility X.  Some of this equipment (e.g. implicated shaking incubator) was in fact rarely used by Patient A.

(2)  Page 2, line 9, “On December 4, 2003, Patient A was…”

The E. coli experiment was conducted on Friday, December 5. The December 4 date was initially given in error by the spouse of Patient A, who was not a direct witness of the experiment and could not ask Patient A, who was then in a coma. The mistake has been carried over, although Patient A and her spouse have asked that it be corrected on multiple occasions.  The investigators were likewise informed of the correct date.

(3)  Page 2, line 9, “…Patient A was officially supervising Technician X”

This statement is a falsehood. The official supervisor of Technician X was Researcher C, who is not identified in the report. Researcher C had responsibility over the design and conduct of the experiment as well as the supervision of Technician X.  Patient A was neither responsible for the experiment conducted by Technician X, nor for supervising him on that day or any other day. Patient A was never instructed to supervise Technician X at any time by her Research Leader.

(4)  Page 2, line 13, “Excess sanitizer was removed in a salad spinner…”

Excess E. coli O157:H7 bacteria were first removed in salad spinners, then the contaminated bags of apple slices were weighed on a scale, divided, incubated, treated with different sanitizers and finally spun again in the salad spinners to remove excess sanitizer. This misleading statement wrongly implies that the salad spinners were not contaminated, when in fact they were contaminated with large volumes of E. coli O157:H7 liquid suspension. Moreover, CDC and DHMH investigators indicated during interviews of Patient A and her spouse that three salad spinners, not one, were in fact used. The number of spinners would directly affect the volume of potential contaminating bacteria and the manner of opening/closing of the spinner tops to ensure timely removal of the bags of contaminated apple slices with minimal ‘drip’.

(5)  Page 2, line 14, “In addition, Patient A was working with broth cultures [] of Salmonella…”

Patient A was not performing her experiments on Salmonella in addition to any other activities. Salmonella studies were Patient A’s primary responsibilities.  Experiments on E. coli O157:H7 were not her responsibility, but that of Researcher C.  Technician X was performing the E. coli O157:H7 on his own, from instructions given to him by Researcher C.

(6)  Page 2, lines 20-21, “Patient A developed abdominal pain on December 10, then bloody diarrhea on December 11. Cultures yielded two pathogens: E. coli O157:H7 and Salmonella…”

This statement falsely implies that cultures from the diarrhea episode, i.e. stool cultures, yielded two pathogens. In fact stool cultures only yielded one pathogen, E. coli O157:H7. Salmonella was never isolated from stool culture.  Salmonella was isolated from blood culture on December 22, i.e. 7 days after the end of the diarrheal phase of the disease and about 3 weeks after a spill that she reported in her laboratory note book, well out of the normal incubation time for this organism. Hence it cannot be concluded that Patient A ever had diarrhea caused by Salmonella as the written statement wrongly suggests.

(7)  Page 2, lines 23-24, ”…investigation revealed that patient A was using E. coli O157:H7 [], multiple Salmonella serotypes and Listeria monocytogenes December 3-9, 2003.”

This is a falsehood that misrepresents Patient A’s activities from December 3 to 9. Patient A was working entirely on Salmonella studies during that period. Patient A only performed one activity with E. coli O157:H7 that was to inoculate a broth culture as a service to Technician X who had not been able to come to Facility X to do it himself. This inoculation took approximately two minutes on December 2 and was performed in the biological safety cabinet (BSC), in full compliance with CDC guidelines (http://www.cdc.gov/od/ohs/biosfty/bmbl4).  Patient A was not using Listeria monocytogenes during that period, as this organism was not part of her studies. E. coli O157:H7 and Listeria work was done by Technician X entirely under the supervision of Researcher C.

(8)  Page 2, lines 31-32, “…staff members reported concerns regarding Patient A’s techniques…”

See point (16) below.

Methods

(9)  Page 3, line 23  “…six-month employment at Research Facility X…”

This is inaccurate. Patient A employment at the Facility started on July 14, 2003. Therefore she had been at the Facility less than 5 months at the time of the accident.  Her laboratory activities started several weeks after July 14, therefore she had been in the laboratory for approximately 4 months at the time of the accident.

(10)  Page 4, lines 3-4 “…Researcher A [ ] was working with E. coli O157:H7 on December 4^th and 5^th…”

This is a falsehood. Researcher A did not work with E. coli O157:H7 and was only working with Salmonella on those dates. Also see point (7) above.

(11)  Page 4, lines 5-6  “While working in the laboratory on December 9, 2003, she began to feel ill and left work early.”

This is a falsehood. Patient A fell ill in the evening of December 10, 2003.  She neither left work early on December 9, nor on December 10.  This was clearly indicated to the investigators during interviews.

(12)  Page 4, lines 8-9  “…The patient notified her direct supervisor of her illness on December 11…”

This is inaccurate. As related to the investigators during interviews, the spouse of the patient made this phone call immediately after Patient A’s visit to the clinic as Patient A was too ill to speak on the phone.  The spouse related his concern about the possible contamination of the laboratory to the Research Leader, which was ignored. 

(13)  Page 4, lines 10-11 “The patient’s symptoms improved over the following three days.  The patient had a recurrence of the symptoms on December 15, and was taken to Hospital A Emergency Room on December 18, 2003.”

These dates are inaccurate. Patient A was very ill from December 11 until December 14, when her symptoms eased and she was gradually feeling better. She suffered a recurrence late on December 17, and was taken to the ER the morning of December 18.  The correct dates were repeatedly given to the investigators during interviews and to staff at Facility X by various means.

(14)  Page 4, lines 24-25  “A broth culture of E. coli O157:H7 was prepared by Patient A on December 3”

This is inaccurate.  Patient A did not prepare a broth culture of E. coli O157:H7 on December 3. Neither did she ever prepare a broth culture. She inoculated the E. coli culture the preceding evening. Also see point (7) above. 

(15)  Page 4, lines 29-30  “All materials were placed in an autoclave bag and autoclaved [] before disposal.”

This statement incorrectly implies that the salad spinners were autoclaved and disposed of. This is a falsehood. After spraying with alcohol, the salad spinners were stored for eventual reuse.  The same salad spinners used for the E. coli O157:H7 experiment of December 5, 2003, had been repetitively used for experiments involving other pathogens.

(16)  Page 4, lines 31-34  “Multiple members of staff voiced concerns [ ] and Researcher B.”

(16a) This can not be considered a statistically valid survey as it would only involve the relatively few people who had the opportunity to peripherally witness Patient A’s technique during the less than 5-month period she was at Facility X. Peers of Patient A in infectious disease microbiology, e.g. former coworkers (graduate students, technicians, postdoctoral fellows, many of whom were trained by Patient A) and colleagues who have worked with her closely over longer periods of time could have provided a fair assessment of her laboratory techniques.  However members from this group were neither sought nor interviewed by the investigators. Patient A was newly employed at Facility X, did not have technical help and did not work closely with any other member of staff.  Furthermore, Patient A was initiating a new project. Therefore, members of staff at Facility X could not have had sufficient knowledge of what her activities were and could not have formed an informed opinion of her techniques during that period. Moreover, Patient A has extensive prior training as an infectious disease microbiologist and 20 years experience in the handling of dangerous pathogens, i.e. far more than any other scientist at Facility X. Other staff at Facility X, including the Research Leader, trained in disciplines such as plant pathology, plant molecular biology and food technology, and as such did not have the relevant competence to express an informed opinion of Patient A’s techniques in microbiology. 

(16b) The surveyed population cannot be considered impartial as it would almost exclusively include individuals who had a personal/professional interest in an outcome of the CDC investigation that did not implicate them or Facility X.

(16c) In as much as Patient A was not involved in the design, conduct or supervision of the E. coli O157:H7-salad spinner experiment, the quality of her techniques is actually irrelevant to the E. coli O157:H7 infection and the Hemolytic Uremic Syndrome (HUS) she suffered.

(16d) Patient A was in the laboratory until December 10, 2003; the laboratory was sealed on December 22, and the investigator team arrived on January 12, 2004.  Hence many others had access to the laboratory on December 10-22.  As such it is not possible to verify that the alleged clutter and disorder was caused by Patient A.

(16e) Wearing gloves in hallways is sometimes unavoidable, for instance when samples must be removed from centrifuges, freezers, cold rooms, etc. that are located outside the laboratory.  In such case, gloves are kept from contacting clean surfaces.  Patient A was seen wearing gloves outside the laboratory when retrieving materials from the cold room. As her research leader had failed to provide her sufficient refrigerator space in her own laboratory, she had been instructed to use the cold room for storage.  Many other members of staff also wore gloves outside their laboratories in Facility X.

Notwithstanding, gloves specifically contaminated with infectious materials were always discarded immediately by Patient A, never reused and never worn outside of the laboratory.  This is a common sense rule strictly followed by all infectious disease microbiologists including Patient A.

(16f) Contrary to the ‘voiced concerns’, Patient A never failed to wipe the BSC appropriately after her experiment was finished.

The survey of Patient A’s laboratory techniques is therefore an unscientific, uncontrolled survey that involved relatively few people, most or all of whom either could not be considered objective, did not have training in infectious disease microbiology or were not sufficiently informed of the type of research activity Patient A was conducting. Patient A received her training at some of the leading infectious disease research institutions in the world, first through her PhD studies at Stanford University, and later through postdoctoral and Faculty positions at the University of Rochester (New York), the London School of Hygiene & Tropical Medicine (United Kingdom) and the University of Maryland, Baltimore.  A survey of many previous colleagues, mentors and students, who have witnessed Patient A’s techniques over the last 20 years and who had no interest in the outcome of the investigation was not attempted. Patient A has in fact previously used these same techniques and taught many students safe microbiology techniques, with dangerous organisms such as African trypanosomes, entero-invasive E. coli, Shigella spp. (including S. dysenteriae), Chlamydia spp. (including dangerous avian C. psittaci and C. trachomatis LGV strains) without a single accident or near-accident (relevant contact information can be provided).  If she had used unsafe techniques such as those alleged by the surveyed population, it is predictable that she would have been infected multiple times over her career.

(17)  Page 5, lines 6-7  “…She routinely reused gloves while working in the laboratory…”

Indeed Patient A routinely reused gloves in the laboratory when handling non-infectious material, as many molecular biologists do, to save research funds.  She indicated this during her interview, and she further stated that she sometimes would put the gloves in the pocket of her laboratory coat.  However these statements have been completely misrepresented and taken out of context in the report.  Patient A only ever reused gloves that she wore to avoid nuclease and protease contamination of laboratory ware and equipment. This is common practice in molecular microbiology laboratories. Gloves used to manipulate infectious pathogens were never reused or placed in her laboratory coat pocket. All gloves involved with research with infectious agents were disposed of in a plastic bag, autoclaved, and never reused. The investigators failed to grasp this important distinction and have made this a central issue in their identification of a probable cause for the accident.  This is a completely erroneous interpretation of statements Patient A made during her interview.

(18)  Page 5, line 9  “She wore a cloth laboratory coat which she laundered at home.”

Patient A inquired about a laundry service for laboratory coats at the Facility and was told that there was none, hence she took her laboratory coat home to launder after autoclaving it. She autoclaved her laboratory coat before every home laundering.  She had no other option but to launder at home.

(19)  Page 5, lines 18-19  “plates spilled related to spills during removal of the cover when transferring micro-titer plates from the incubator to the plate reader.”

This is incorrect. The spills occurred while Patient A lifted micro-titer plates on top of the plate reader.  The shaking incubator was neither used for Salmonella culture nor any other purpose by Patient A during that period as wrongly repeatedly alluded to in the report. This was clearly indicated to the investigators by Patient A during her interview.

(20)  Page 5, line 23  “She suspected [ ] due to contamination of her hands with E. coli O157:H7 from laboratory surfaces”

This is imprecise. Patient A indicated to investigators that she suspected that the contamination originated from faucets at the contaminated sink because this was the only location of the laboratory that both she and Technician X used on the day of the E. coli O157:H7 experiment.

(21)  Page 5, line 32  “She also supervised Technician X working on the apple decontamination experiment.”

This is a falsehood.  See Point (3) above.

(22)  Page 5, lines 33-34  “She prepared the inoculum of E. coli O157:H7 and supervised the procedure in her own laboratory.”

During the referenced period (December 3 through 5), Patient A had no involvement with the E. coli O157:H7 experiment. Again her only involvement was to inoculate the broth culture, not to prepare the inoculum, in the BSC on December 2, 2003.  Also see Point (3) above.

(23)  Page 5, line 35  “In a five-day period, December 1-5, Researcher A performed 660 manipulations of infectious agents including E. coli O157:H7…”

Researcher A (Patient A) was not performing any manipulation of E. coli O157:H7 at any time.  See Points (7) and (22) above.

(24)  Page 5, lines 36-39  “Many of these manipulations involved high inoculum broth cultures.  In her laboratory notebook she twice noted that “plates spilled”, but she did not specify where or how. The high volume of activity with high concentration liquid cultures offered numerous opportunities for accidental contamination.”

(24a)  The implication that the spills involved high concentrations of organisms is false.  Patient A did not report spills of ‘high concentration liquid cultures’. The spills she reported in her notebook involved small droplets of dilute culture. During her interview, Patient A reported that the spills had occurred at the initial time in the growth curve experiment when a plate lid briefly stuck to the base of the plate above it and dropped from a height of less than 1 cm, releasing droplets.  The droplets fell on top of the plate reader.  Patient A also reported that she suspected that some droplets may have reached her face as the plate reader was situated on a high bench, near face level.  She also indicated that the droplets were very small and did not visibly change the volume of the 1 ml cultures.  Patient A responded to the spill appropriately, following the training she received and the CDC guidelines: she cleaned the contaminated surface immediately with 70% ethanol, used an alcohol wipe to clean her face and recorded the incident in her laboratory notebook.  Since the volume involved in the spill was very small (less than 10 micro-liters) and the cultures were dilute, she did not feel it necessary to report the incident to the Safety Office.  The investigators had the opportunity to estimate the bacterial concentrations using the optical density measurements entered in the laboratory notebook at the time of the spills, but did not do so.  A low, sub-infectious dose of Salmonella (below 10⁶ colony-forming units) is entirely consistent with the clinical course of Patient A. 

(24b)  With respect to the high volume of activity, the report refers principally to the high number of isolates that Patient A was working with during the December 1-5, 2003, period. Patient A was conducting a screening experiment to identify and isolate Salmonella strains that had inherent resistance to sanitizers.  Screening multiple strains for unique phenotype or genotype is one of the most common experiments in microbiology laboratories. By necessity, these experiments must include numerous strains in order to isolate rare variants and/or to satisfy statistical validation requirements. That the screening experiment Patient A was conducting involved pathogens does not per se constitute a specific biosafety risk as long as appropriate containment measures are followed.  As indicated in this rebuttal document [see Point (37) below], Patient A was unable to perform her Salmonella experiments in the existing BSC, because it was occupied by a large piece of equipment (spiroplater).  Patient A had repeatedly asked her Research Leader that the existing BSC be cleared or that a new BSC be purchased. She was however denied on each occasion.

(25) Page 6, lines 13-15. “There was no evidence on the plates of significant environmental contamination during the use of the salad spinner itself.”

This reenactment experiment was designed to evaluate the dispersion of a surrogate organism during the salad spinner experiment. The design of the reenacted experiment however is fatally flawed. 

(25a) the run-through experiment was done by Technician X, who had a personal interest in performing the experiment as cleanly and safely as possible under the eyes of investigators, pressure he did not have on December 5, 2003. 

(25b) the reenacted experiment involved a single salad spinner, while the actual experiment had three, thereby easing the work load of the technician and minimizing potential dispersion. The investigators were also informed by Patient A that in previous experiments run by Technician X, as many as six salad spinners were used simultaneously.

(25c) the reenacted experiment was performed in the BSC, while the actual experiment was performed on the bench top. Since a key factor in the contamination would be the formation of aerosols upon opening of the spinners, air flow in the immediate environment of the spinners can be predicted to have a major impact on dispersion of the aerosol. Air flow in a BSC and that in an open laboratory are simply not comparable.

(25d) it is unclear why the investigators chose to use Serratia marcescens as a surrogate organism. S. marcescens is a low-virulence organism thereby necessitating that the experiment be done in the BSC. A completely non-virulent strain that would have permitted reenactment of the experiment on the open laboratory bench should have been used. Moreover, E. coli O157:H7 is well known for its ability to survive adverse conditions, including low pH, desiccation, etc.. Hence the use of Serratia as a surrogate for such a hardy bacterium is not scientifically sound.  Attenuated strains of E. coli O157:H7 have been created that lack the potent Shiga toxin and such a strain should have been used for the reenacted experiment; indeed such a strain should have been used for all experiments conducted in this laboratory rather than a fully virulent strain. Also see point (38c) below.

(25e) the use of sub-standard, non-scientific equipment such as salad spinners makes the reproducibility of the experimental conditions difficult from experiment to experiment.  For instance, the spinning velocity, spinning time, timing of the removal of the salad spinner top, and opening method likely varied between experiments. Hence it is predictable that environmental contamination would also vary significantly from experiment to experiment. Because of these many added variables, conclusions from the reenacted experiment cannot be considered to have any relevance to what happened on December 5, 2003.  Moreover, a key part of the experiment that had been questioned by Patient A repeatedly, the disinfection and cleaning of the salad spinners was apparently not evaluated by the investigators through reenactment.

(26) Page 6, lines 17-19  “The Office of Safety and Health Assessment identified two major issues…. These included Patient A’s laboratory techniques and the general safety policies of Research Facility X.”

The issue of Patient A’s laboratory techniques is based on erroneous data, flawed investigation and misinterpretation of interview data. See Points (16, 17, 27).

(27) Page 6, lines 20-31  Paragraph titled “Patient A Laboratory Technique”

(27a)  Patient A’s laboratory was accessible to others who conducted experiments requiring BSL2 containment for more than a week after Patient A fell ill.  An accurate determination of cleanliness and orderliness specifically attributable to Patient A was therefore compromised. 

(27b)  Cultures (Petri dishes) that remained on her bench were part of an ongoing experiment that required incubation of bacteria at room temperature to reproduce environmental conditions in which sanitizer-resistant bacteria might arise. This was indicated to the investigators during interview.  These were not residuals from completed experiments she had failed to dispose of as implied in the report. 

(27c)  Unmarked cultures in the shaking incubator were not Patient A’s as she indicated to the investigators during interview.

(27d) Spilled residue in the shaking incubator did not originate from Patient A’s research.  Patient A indicated during interview that the shaking incubator was equipment shared by all staff of Facility X.  Patient A specifically stated that she did not use the shaking incubator during that period.

(27e)  Used gloves on her bench were not from work with infectious pathogens and as such posed no risk.

(27f)  The use of a loop at the bench to inoculate cultures does not per se constitute a biosafety risk. Loops are used broadly in microbiology laboratories throughout the world.  As for the use of gloves, when using a loop, Patient A made a distinction between inoculations of high risk infectious material and inoculations of low risk organisms, e.g. strains used for cloning.  For high risk inoculations, Patient A always used disposable loops in the BSC. Specifically, all Salmonella inoculations Patient A performed during that period were done using disposable loops and in the BSC, i.e. in full compliance with the CDC guidelines.  This was indicated to the investigators.  For this purpose, Patient A borrowed a large quantity of disposable loops from a colleague’s laboratory, hence independent verification is available.

(27g)  As indicated to the investigators, Patient A always washed her hands with soap when she left the laboratory: once at the sink in the laboratory and once more in the restroom.

(27h)  Also see Points (16, 17, 24).

Addendum

(28) Page 8,  lines 1-19,  Paragraph titled “Patient C”

Patient C had independent episodes of enteric disease on January 21 and April 16, 2004, i.e. 40 days and over 4-months after Patient A was last at Facility X.  Moreover, Patient C’s episodes occurred long after Patient A’s laboratory had been sealed.  Hence, these new episodes are unlikely attributable to residual contamination from the December 2003 experiments. The April 16 episode yielded positive serology for E. coli O157:H7.  The likelihood that it represents a laboratory-acquired infection is high in view of Patient C’s interview data.  The description of Patient C as a 42-year old female scientist at the Facility only fits that of Researcher C, identified above as the actual official supervisor of Technician X and the person responsible for the design and conduct of the salad spinner experiment that led to Patient A’s infection.  This determination is facilitated by the breach of confidentiality documented below.  If confirmed, this extraordinary development strongly suggests that unsafe procedures were still employed at Facility X over 4-months after Patient A had left for the hospital and long after the investigative team had made recommendations. Hence, a similar investigation of the laboratory techniques of Patient C or of those of staff she directly instructed or supervised up to April 16, 2004, should have been undertaken in the case of Patient C’s likely laboratory-acquired infection.  If it is confirmed that Patient C and Researcher C are the same person, then Patient C is the only common denominator to both accidents, an epidemiological fact that should have refocused the investigation on Patient C.

Discussion

(29)  Page 8, line 23  “Co-infection with E. coli O157:H7 and Salmonella…”

There is no evidence for co-infection. The clinical evidence indicates that the two infections happened in succession, not simultaneously. Salmonella was not isolated from stool when E. coli was, during the early phase of the disease. E. coli was not isolated from blood later when Salmonella was. The clinical data strongly suggest that the Salmonella septicemia was secondary to mucosal injury caused by E. coli O157:H7 in Patient A’s digestive track.

(30)  Page 8, lines 39-41  “Patient A documented spills of plates containing Salmonella spp. in the shaking incubator on December 4. There is no record of decontamination of that spill and residue was evident at the time of inspection”

Patient A indicated to the investigators that the spills involved very small droplets on top of the plate reader, not the shaking incubator, and that she immediately decontaminated with alcohol.  Again, Patient A did not use the shaking incubator and spill residue within could not have originated from her.  This was clearly stated by Patient A during her interview.

(31)  Page 8, lines 41-42  “…both of Patient A’s clinical pathogens were processed in the micro-centrifuge and the shaking incubator in the days before her illness,…”

This is inaccurate.  Technician X did process E. coli O157:H7 culture in the micro-centrifuge and shaking incubator, whereas Patient A did not use either piece of equipment for any purpose during that period.  This was also repeatedly stated to the investigators and to Patient A’s Research Leader.

(32)  Page 9, line 2  “As she was mainly doing molecular work, she may not have been meticulous with glove use.”

This is a completely gratuitous statement that is not based on the evidence of Patient A’s research activities over a 20-year period.

(33) Page 9, lines 6-7  “Results of the reenactment of the apple-decontamination experiment do not support significant dispersion of organisms with the actual salad spinner.”

This conclusion is very weak, in light of the many variables introduced in the reenactment experiment and of the fact that formed aerosols likely migrated differently in the BSC than they would in the open laboratory.  Also see Point (25) above.

(34)  Page 9, lines 18-19  “Indeed, the patient hypothesized that she was infected when she ate an orange with unwashed hands the day after these spills.”

This statement contains several mistakes. The authors appear to refer to the Salmonella spills. However, Patient A only hypothesized during her interview that she was infected with E. coli O157:H7, not Salmonella, when she ate an orange on December 5 after washing her hands in the sink. The Salmonella spills which occurred on December 3-4, took place at the plate reader that was placed on her high bench. As the benches at the Facility are unusually high and since Patient A is herself short (5’ft1), it is most likely that the Salmonella contamination occurred on December 3 or 4, via direct inoculation of ocular or oral mucosa, not on December 5, when she ate an orange.

Moreover, the authors write on page 5, lines 21-22, that ‘she washed her hands at the sink and went to her office where she consumed an orange’. This statement correctly reports the information given by Patient A during her interview.  It is inexplicable therefore that the authors write that she ate an orange ‘with unwashed hands’ on page 9 of the same document.  This is in keeping with the many other inaccuracies, inconsistencies and approximations that occur throughout the report.

Patient A came to suspect that this action was the route of her infection by a process of elimination. First, she did not go near the work area where Technician X performed the E. coli O157:H7-salad spinner experiment on December 5.  Second, she neither used nor touched the micro-centrifuge and shaking incubator used by Technician X. Last, although she had washed her hands with soap and water at the sink, she still had to turn off the faucets that were likely contaminated. The sink area is the only area commonly used by Patient A and Technician X on December 5.  Patient A had previously observed inconsistent disinfection methods used by Technician X and had suggested that soaking the salad spinners in a phenol-based disinfectant would allow for complete disinfection, which the spraying with alcohol may not achieve. Her suggestions were however ignored.

Patient A related this information to the investigators in early February 2004, but also to her Research Leader by telephone and e-mail, immediately after the initial episode of E. coli O157:H7 diarrhea (December 15-17, 2003),  i.e. a few days after the accident (December 5, 2003).

OMISSIONS

(35)  Delayed sealing of the laboratory

The report indicates that ‘Research Facility X sealed the laboratory and office on December 22, 2003’ (Page 2, line 29).  The report however fails to establish that the Research Leader of Facility X was warned of a possible E. coli O157:H7 as early as December 11, 2003.  Patient A’s Research Leader was told repeatedly of the likely laboratory-acquired E. coli O157:H7 infection prior to December 22, 2003: firstly, by telephone on December 11, 2003 by the spouse of Patient A, and secondly on December 15-17 in two telephone calls and one e-mail message by Patient A herself. The spouse of Patient A telephoned again on December 22, once proof of an E. coli O157:H7 infection had been obtained by culture. By then, Patient A had fallen in a coma due to the HUS.  On each occasion, the Research Leader was cautioned to take safety measures (including the closing of the laboratory) to ensure that other workers would not contract the infection.  On each occasion, he failed to take action. Closure of the laboratory was finally the immediate result of a conversation between the spouse of Patient A and the Research Leader’s direct supervisor, late on December 22, 2003. In an e-mail message and two telephone conversations on December 15-17, 2003, Patient A indicated to her Research Leader her belief that she had been contaminated with E. coli O157:H7 even though proof by culture had not been obtained yet. She indicated that this belief was based on a range of factors, including her symptoms, blood analysis results and the opportunity for exposure. The 5-6 day incubation period based on a December 5, 2003 exposure, was also consistent with the known incubation period for this organism. She further indicated that her belief that she had been infected with entero-hemorrhagic E. coli O157:H7 had been reinforced by two infectious disease doctors and scientists at the University of Maryland, School of Medicine, who are internationally-recognized experts on E. coli O157:H7 infection.  She further recounted in detail to her Research Leader how she might have been contaminated at the sink (see Point 34 above).  The Research Leader however persistently dismissed Patient A’s hypothesis as completely unwarranted and refused to consider taking safety measures to prevent further contamination and infection.  In failing to close the laboratory as soon as he received warning of a possible E. coli O157:H7 contamination, the Research Leader unnecessarily put the health of other staff and, by association, that of the public at large, at risk of infection for a period of 11 days.  Moreover, clutter and visible spills should not be attributed automatically to Patient A, who was absent from the laboratory starting on December 11, 2003.  Finally, the delayed closing of the laboratory compromised the timely sampling of strains and contaminated surfaces, thereby compromising the investigation. The investigators were informed of these facts by Patient A and/or her spouse during interviews.

(36)  Patient A’s concerns over safe experimental design

The report fails to relate that Patient A had indicated her concern over the safety of the salad spinner experiments on multiple occasions to both Technician X and Researcher C, concerns that were dismissed ‘as this was an established protocol, that had been performed on several occasions before without problem’.  This was indicated to the investigators during interviews of Patient A and her spouse.  This was also reported by Patient A to the Research Leader on December 15, 2003, by telephone. A corollary of this is that, had she been the official supervisor of Technician X, he would not have been allowed to perform these experiments by Patient A, who deemed them unsafe.

In an unrelated incident, Patient A had also previously raised her concern to her Research Leader about the obstruction of proper air flow when a large piece of equipment such as spiroplater is placed inside the BSC.  She reported this to the Research Leader after she observed that certification of the BSC has been obtained with the spiroplater purposefully removed from the BSC.  Hence determination of the air flow inside the BSC with the spiroplater inside, i.e. in its normal conditions of usage, was not determined.

The unresponsiveness of scientists and staff including the Research Leader to the safety concerns expressed by Patient A is in keeping with their relative lack of relevant training in infectious disease microbiology.   Facility X’s stated mission, as revealed on its web site (see ‘Breach of Confidentiality’ below) is ‘to conduct research on the post-harvest biology and technology of fruits and vegetables in order to maintain quality and microbial safety and to reduce losses of fresh and fresh-cut produce after harvest.’ As demonstrated elsewhere in the report and in our rebuttal, Facility X is not set up to conduct research on pathogens that are highly virulent for humans.

(37)  Patient A’s multiple requests for her own BSC

Patient A requested on multiple occasions from her Research Leader that she should be given a BSC of her own to carry out her work with pathogens. She was denied until shortly before the accident. Patient A’s demand for her own BSC was only granted after she proposed to redirect the budget already allocated to her laboratory for the purchase of a PCR machine, toward the purchase of a BSC instead. A BSC was finally on order when the accident happened.  Patient A alternatively suggested that the cumbersome ‘spiroplater’ in the existing BSC be moved to another location but was also denied.  Since the spiroplater occupied approximately 2 thirds of the BSC, the remaining space was insufficient for her to perform much of her Salmonella experiments (except for simple inoculations as mentioned above).  Had she been able to perform all her Salmonella experiments in a BSC, the spills she incurred would have been contained and most likely of no consequence.  Hence, the repeated refusal by her Research Leader to provide Patient A her own BSC is a direct cause of her Salmonella septicemia.  Since she was hired as a Microbiologist and was asked to work exclusively with food-borne pathogens, it is inexplicable that she was not attributed the necessary equipment to safely conduct her experiments.

(38)  Flawed experimental design as a source of contamination

The report wrongly documents and fails to question key aspects of the salad spinner experimental design that are in full violation of CDC guidelines.  Instead, the report wrongly attributes responsibility to Patient A’s laboratory techniques.

(38a)  The use of salad spinners on the bench top to remove excess E. coli O157:H7 culture, which is highly infectious (ID₅₀ of 10-100 colony forming units), is untenable with respect to safety and in full violation of CDC guidelines (http://www.cdc.gov/od/ohs/biosfty/bmbl4). In as much as this experiment did not intend to reproduce an actual public contamination of a salad spinner (in a kitchen or restaurant for example), scientific equipment that provides for contained and controlled experimental conditions should have been used.

(38b)  The cleaning and decontamination of the spinners is sub-standard and also in violation of the guidelines.  Unprotected transport of the contaminated spinners to a remote sink should have been avoided as it maximizes potential dispersion. All contaminated materials should have been decontaminated by thorough soaking in a bactericidal solution rather than spraying with 80% alcohol which would not ensure complete decontamination.  Spraying likely contributed to splashes that contaminated the sink. Reproducibility of the used decontamination method would vary significantly with many different variables, such as the individual performing decontamination, the amount of residual liquid in the salad spinner, the force and extent of spraying of contaminated surfaces, etc.

(38c)  An attenuated mutant strain of E. coli O157:H7 was not used for this experiment. Patient A had questioned the validity of using a fully virulent strain. However, she was told that a fully virulent outbreak isolate should be used to fully validate experimental results. Strains that carry mutations in the active site of the Shiga toxin made by E. coli O157:H7 express an inactive toxin and are significantly attenuated as a result. HUS, the life-threatening complication from E. coli O157:H7 infection that Patient A suffered from, would not occur without a fully active toxin.  The only experiments where the use of a fully virulent strain is justified are experiments in animal models, where it may be necessary to reproduce full-blown disease. Attenuated E. coli O157:H7 strains are widely used in microbiology laboratories throughout the world.  The risk-to-benefit ratio, a constant consideration when designing experiments involving dangerous pathogens, was apparently not factored in the design of the experiments conducted by Technician X.  An attenuated mutant strain should have been used for these experiments.

(39)  Suboptimal interview conditions and lack of verification

The spouse was interviewed while his wife, Patient A, had been in a coma for several weeks already.  He was exhausted and extremely stressed as the likelihood that Patient A, his wife and mother of their two young children, would not survive or would survive with severe neurological damage was high.  Patient A was in a coma and on a ventilator for one month, suffered multi-system organ failure from the HUS, septicemia and pneumonia. She awoke on January 13-14, 2004, was moved to a rehab hospital 10 days later and was sent home on February 4, 2004. She was interviewed a few days after her release from the rehab hospital.  She was extremely weak still and, although fully cognizant, was physically impaired, e.g. she could not turn her head, walk or sit comfortably for any length of time. She reported to the investigators that she was experiencing severe dizziness, and developed a severe headache during the interview.  These suboptimal interview conditions should have warranted repeat interviews by the investigators.  A repeat interview of Patient A was proposed by one of the investigators but was never acted upon.  Repeat interviews of both Patient A and her spouse would have helped to clarify several critical issues such as who was the ‘official supervisor’ of Technician X, and under which conditions were gloves reused by Patient A. In contrast, the investigators informed Patient A and her spouse during interviews that Facility X would be provided with a draft of the report weeks before its eventual submission. Hence, staff at Facility X had the opportunity to review the data and conclusions made in the report at least once before it was completed. Not only is this illustrative of the general sub-standard method employed in this unbalanced investigation, but it also adds strength to the notion that some member of staff at Facility X may have provided inaccurate information to the investigators.

(40)  Selection of investigation data

A recurrent statement appearing in this rebuttal document is that information given in the report is inconsistent, imprecise or incomplete, not only when compared with the information that should be available to the investigators from the medical record, e.g. see Points (6, 11, 13, 29) above, or the administrative record, e.g. see Points (3, 9, 11) above, but also from the interviews of either Patient A, Patient A’s spouse or both.  With respect to the latter, evidence of inconsistency is addressed specifically in Points 2, 4-5, 7, 10-14, 19-23, 24a, 27b-d, 27f-g, 30-31, 34-35 and 37-39 of this rebuttal document. 

The CDC investigation started on Monday January 12, 2004, and staff of Facility X was interviewed during the following 2 weeks.  Hence, staff members, presumably including Researcher C, Technician X and the Research Leader, were questioned about events that were already at least six weeks old, events to which they presumably were not paying particular attention when they were taking place.  In contrast, Patient A documented her activities of December 5 to her spouse and to her Research Leader on December 15-17, i.e. only 10-12 days after the accident, when her memory of her activities was still fresh. Moreover, she discussed her activities of December 5 with her Research Leader on three separate occasions: during 2 phone conversations and in one e-mail message.  She described her activities in some details to him and indicated that the only location of the laboratory where she had crossed path with Technician X on December 5, 2003, was the sink. She had neither been near Technician X’s bench, nor used the micro-centrifuge or the shaking incubator. The spouse of Patient A related this information to the investigators during his interview in mid January 2004. Patient A herself repeated the same information to the investigators during her interview in February 2004.  It is therefore unclear why the activities of Patient A on December 5 as described in the report are so frequently inconsistent, indeed sometimes contradictory, to the activities freshly remembered by Patient A, which she recounted to her spouse and her Research Leader on December 15-17, 2003, and were reiterated to the investigators during interviews by Patient A and her spouse.  For instance, it is disturbing that the report repeatedly states that Patient A was conducting experiments with E. coli O157:H7 or used the shaking incubator, when she repeatedly stated to her Research Leader and to the investigators that she did neither.  In light of the inconsistencies, it is unclear that the Research Leader related to the investigators the contents of one e-mail message he received from Patient A and two phone conversations he had with her. Regardless, it remains inexplicable that the report is so often inconsistent with the information provided to the investigators by Patient A and her spouse.

BREACH OF PATIENT CONFIDENTIALITY

On Page 9, lines 33-34, of the report, Facility X is identified as the ‘Produce Quality Safety Laboratory’.  This inexplicable mistake makes the identification of the individuals involved in the accidents readily obtainable by the public at large. A simple Google internet search for ‘Produce Quality Safety Laboratory’ or ‘Produce Quality Safety’ immediately leads to the home page of the web site of Facility X as its first entry (http://www.ba.ars.usda.gov/pqsl/).  From the home page, one additional mouse click yields a list of Lab Scientists and Staff at the Facility, including the name of the Research Leader (http://www.ba.ars.usda.gov/pqsl/pandp/index.html).  One additional click yields contact information and a summary of individual scientists’ research activities.  Four female scientists are listed, only two of whom work on food-borne pathogens. The identities of Patients A and C, and the Research Leader, are therefore readily identifiable with a few mouse clicks from the web site alone. Hence, Patient A and Patient C’s identities were de facto revealed to the public at large as soon as the report was released. The identification of Facility X by name therefore constitutes a breach of patient confidentiality, in full violation of HIPAA Privacy Rule. This glaring error is in keeping with the many inexactitudes contained in the report and suggests that its authors did not thoroughly re-read the text of the report. This is consistent with the poor overall conduct of the investigation.