Reader Comments
Post a new comment on this article
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Thank You!
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.
closeA letter to the editors regarding the article “Performance of the PointCare NOW System for CD4 Counting in HIV Patients Based on Five Independent Evaluations”
Posted by PointCare on 22 Aug 2012 at 13:10 GMT
To the editors:
I am writing to you in my role as a company executive of PointCare Technologies, Inc. www.pointcare.net and also in my role as a co-inventor of the technology underlying a medical diagnostic product called PointCare NOW. I am writing in reference to the following article which was published in your journal last week: Bergeron M, Daneau G, Ding T, Sitoe NE, Westerman LE, et al. (2012) Performance of the PointCare NOW System for CD4 Counting in HIV Patients Based
on Five Independent Evaluations. PLoS ONE 7(8): e41166. doi:10.1371/journal.pone.0041166
The authors failed to inform the reviewers of this article about a number of important facts known to them and to PointCare. I will attempt to provide them here.
The authors speculate whether another company’s product, the HumaCount CD4 NOW, might be the same as the PointCare NOW system they tested over two and half years ago in their studies. No one contacted either company to find that indeed they are not the same. As a matter of fact the HumaCount and the PointCare NOW instruments deployed today are significantly different from the systems tested in this publication. Therefore the published results are no longer relevant. A multisite study using the currently deployed instruments has been submitted for peer reviewed publication. Four of the five sites publishing here were invited to participate. In three cases we were told that finding the time was difficult and could participation happen at a later point. In the case of site E the investigator contacted informed us that she was no longer permitted by her parent institution to communicate with us.
PointCare was shown this publication by the management of site D in April of 2011 when it was submitted to another journal and has challenged the authors directly and openly about the flaws in their representation. The company has received promises from three authors (sites A, B and D) that they would address the problems we pointed out. I am very surprised that I now see a publication where none of the problems have been addressed or disclosed.
Site D, US Center for Disease Control (L. Westerman): In sharp contrast with other sites, site D in Atlanta, Georgia, obtained results with markedly low bias (-11 CD4 counts) and very high correlation with predicate methods (R=0.980). The authors speculate that low cell densities used in the samples below CD4 = 350 produced this low bias and high correlation. They fail to point out that the samples above CD4 = 350 where the cell densities were normal also produced markedly low bias (+11 CD4 counts) and high correlation (R=0.964). The PointCare devices used for testing are validated for white cell counts as low as 500 and as high as 40,000, which further indicates that cell density was not responsible for good performance at this site. PointCare offered to Site D that if cell density was still a question despite PointCare’s internal validation data, then we should jointly study this factor. Westerman from Site D has not responded to this offer made two years ago.
It is also difficult to implicate cell density as the reason for good within run replicate precision performance at site D. Since PointCare partially financially sponsored this study (not disclosed in the article) we were given the results from site D two years ago. Site D tested 5 healthy donors using 10 replicates and recorded a mean precision of 5.31% (SD=1.09% at a mean CD4 count of 1108) with the PointCare NOW and a mean precision of 4.65% (SD=0.6% at a mean CD4 count of 1157) with a FACSCalibur using Tritest reagents.
Having dismissed the favorable data from Site D based on scientifically unsubstantiated speculation about cell density, the authors discuss data from the other sites which present a less favorable picture of the PointCare devices tested. It is again unfortunate that the reviewers of the article were not given sufficient scientific information that could explain these discrepancies. I will do so below.
Site A, National Reference Laboratory, Mozambique (I. Jani), supported by Clinton Foundation (T. Peter): Our company was approached by the Clinton Foundation to participate in a point-of-care product evaluation in Mozambique in 2009. We agreed to participate and bore the cost for part of the study (Not disclosed in the article). A protocol was agreed to by all parties. During the study at least one of the instruments was used against manufacturer’s specifications (without a UPS protective power supply) and exposed to an electrical surge which physically destroyed part of the electronic hardware. Our company was not informed and the study was completed with the damaged instrument. After we discovered the damage, we protested and requested a repeat of the study. The principal investigator (I. Jani) promised a partial repeat of the study but could never find the time to do so. When the repeat of the study did not materialize PointCare requested a joint review of all data, as it was called for in the study protocol. Our company felt that it was necessary to find out if power surges may not only have compromised PointCare NOW results but also results from the reference system. Access to any data was denied categorically. PointCare turned to the investigator from the Clinton Foundation who had participated in the design of the study protocol and had obtained partial funding. In a face to face meeting in July, 2010 T. Peter declared that he had no access to the data and that they were the possession of the Mozambique reference laboratory, even though data sharing was part of the study protocol. An explanation is required as to how his name appears on a publication of data “he had no access to” during the conduct of the study.
Site B, Tete, Mozambique, conducted by Institute of Tropical Medicine in Antwerp: This particular site purchased a PointCare NOW instrument and reagents as part of a grant. Shortly before the final report for the grant was due our company received a call for urgent support at a remote site in Mozambique. We found that the laboratory where the instrument was installed was not, and had never been, electrically grounded, which is mandatory for the PointCare NOW as well as the reference flow cytometer FACSCalibur. Moreover, we found that the reference instrument, possibly because of the lack of an electrical ground, had not passed controls in more than a year. We informed the investigators of these problems and requested resolution before conducting any evaluation. In view of the impending grant deadline at Site B, our request was rejected and we were told in an e-mail that “electrical grounding is unnecessary and is not done in Africa”. Site B reports their results without disclosure of the electrical problem and failure to pass controls by the reference system. Weeks after the completion of the study the Site B laboratory was closed for reasons of unacceptable infrastructure and underwent a major overhaul.
Site C, Health Canada (M.Bergeran): This National Reference Laboratory began conducting an evaluation during 2009 and called our customer support complaining about unsatisfactory results. Our service manager visited the laboratory and found the users not adhering to manufacturer’s specification in that leftover samples from the clinical laboratory were used after the 8 hour specification for use of a PointCare NOW. PointCare NOW is a closed point-of-care diagnostic system designed to use only fresh (less than 8 hours) blood samples in Vacutainers that have not been opened. The system will not perform reliably and can even be damaged when previously-opened, manually reclosed sample tubes are used. Our service manager repaired the slightly damaged instrument at Health Canada and trained the operators in the adherence to manufactures specifications. The operator tested samples for a few days and sent the results to our company to confirm proper performance of the system. Those data are also published on our company website. Shortly after the service visit the manager of the laboratory (M. Bergeran) intervened and rejected the restrictions that prohibit the use of older and previously opened blood samples. It is disclosed in the methods section of the publication that 55% of samples at this site were tested against manufacturer’s specifications. It is certainly the choice of any laboratory to reject certain product specifications as they may be unsuitable for that laboratory’s work flow. It is, however, unacceptable to publish data claiming bad performance without specifically stating that the system was operated in violation of manufacturer’s specification.
Site E, WITS, South Africa: The investigators from this site have published these data during the last International AIDS meeting in Vienna during a workshop sponsored by the company Alere (PointCare’s commercial competitor). The same data were also published during a meeting in Mozambique in early 2010. We at PointCare were very curious as to the origin of the data as we never participated in an evaluation and provided neither instrumentation nor reagents. When the data were first published in 2010 we queried the origin officially. An internal report (authored by D. Glencross and dated February 2010) was provided to us by the Director of the South African National Reference Laboratory (NHLS) describing an unsuccessful evaluation that allegedly had taken place “during the latter part of 2009” on an instrument that was purchased by RHRU, an organization associated with NHLS. Our documentation shows that the only instrument purchased by RHRU was installed on March 18, 2010 a month after the date of the evaluation report provided to us. Our records also show that this instrument was never used as the entire project for which it was purchased was put on hold. To this date no reagents have been shipped to either NHLS or RHRU.
Upon further investigation we found that our technician who gave a talk using a sales demonstration instrument in late 2009 had left this un-calibrated device for a few days in the NHLS laboratories as they wanted “to show it to some of their colleagues”. When our technician returned to collect the instrument he read the data files and saw that it had been used for patient sample testing even though as a sales demonstration device it had not been certified by the Quality System for that purpose. He also found that previously opened samples had been used which is expressly forbidden in our specifications. Apparently the results in this and previous publications have been obtained from a demonstration instrument not set up for patient sample use.
Competing interests: D.Glencross is the sole inventor on a CD4 patent owned by the South African NHLS. This competing technology is licensed to Beckman Coulter Corporation and the NHLS receives royalties. D. Glencross was called upon to advise the NHLS as documented in the negative report about PointCare she wrote in 2010. She has disclosed her relationship to the patent and her advisory role in previous peer reviewed publications but did not do so as required by the publishers of the present article.
L. Westerman is an official CDC representative for CD4 testing in a United States Government sponsored public private partnership between his employer and Becton and Dickinson. Becton Dickinson sells competing CD4 products.
I apologize for this rather lengthy description which I feel was necessary to do the truth some justice. I will be happy to provide additional documentation for all of my points should this be desired.
Respectfully
RE: A letter to the editors regarding the article “Performance of the PointCare NOW System for CD4 Counting in HIV Patients Based on Five Independent Evaluations”
GDaneau replied to PointCare on 22 Sep 2012 at 14:26 GMT
Dr Krauledat recently commented on our publication. We wish to provide our scientific response.
Concerning comments about site D, first we reconfirm the following conclusions from Site D that were reported in the manuscript: (i) the PointCare NOW system had a higher than expected test failure rate; over the entire range of CD4 counts, the absolute CD4 count from PointCare NOW correlated with reference method; (ii) at lower CD4 counts (in the clinical important range), the absolute CD4 count correlation was poor; the precision of the PointCare NOW absolute CD4 counts was acceptable and comparable to the reference method; (iii) for CD4% of lymphocyte results, there was no correlation of the PointCare NOW assay with the reference method and; (iv) estimates of the precision for the PointCare Now CD4% of lymphocytes results were unacceptable when compared to the percent coefficient of variation of the reference method. These results led this site to raise concern about the ability of the PointCare NOW system to provide accurate and reproducible results.
The “cell density” hypothesis was introduced in the discussion to support the better performance (lowest bias; best correlation) observed in site D when comparing PointCare Now measurements to the reference method measurements. This statement strictly refers to accuracy, not precision. We agree that samples above 350 CD4 cells/µl also produced low bias. However, 30% of the data set above 350 CD4 cells/µl was produced on diluted samples. Therefore, although the hypothesis relating accuracy (low bias; high correlation) and cell density has not been proven, it remains credible.
Regarding precision, only intra-run precision was reported. The precision analysis was carried on using one representative sample per site, not as a mean of different samples. Therefore, the reporting of precision was based on a single set of replicates. Averaging several sets would reflect on inter-run precision.
At sites A and B, PointCare instruments were installed by PointCare engineers and validated them for use both at the start of these studies and mid-way through the studies. The investigators collected samples and used the instruments as per manufacturer instruction throughout the studies. Throughout the studies, the instruments completed test runs without errors messages that might indicate mechanical or operation problems associated with serious electrical damage. Furthermore, throughout the studies the instruments consistently passed the manufacturer-provided daily internal quality controls before the start of testing.
The reference instrument at site B, a FACSCalibur, was operated in the same room than the PointCare instrument and passed all external EQA samples (QASI) prior to, during and after the study for CD4% and absolute CD4 count. The analyses of the FACSCalibur list mode files were performed with CellQuest software according to the Institute of Tropical Medicine’s ISO15189:2007 accredited procedure. Part of the Tété lab was indeed closed for planned renovation work, but this had no negative impact on temperature and working conditions in the room where the two instruments were placed during the study.
Site C confirms that all samples tested on the PointCare Now system were analyzed within 8 hours, as recommended by the manufacturer. We recognize that the sentence “All samples were analyzed within 8 hours of collection for PointCare NOW testing and the same day for reference testing, except in site C where 55% of the samples were run the next day but still within 24 hours of collection” may be misleading. To clarify, 55% of samples tested for the reference method were prepared within 24 hours of collection.
For site E, PointCare representatives installed a USAID-purchased PointCare instrument into the University of the Witwatersrand/ National Health Laboratory Service reference laboratory, for a validation of the equipment prior to its use in clinical settings. Sample testing was performed according to manufacturers’ specifications and instrument validation performed according to our local protocol for evaluation of CD4 instrumentation. Samples submitted for routine CD4 testing were used in this validation exercise. All adult samples were tested on the PointCare first, without opening of the blood collection tubes, immediately upon receipt in the laboratory and within 4 hours of venesection. To minimise bias that could be introduced through sample aging, the same samples were tested by the predicate method immediately after testing in the PointCare instrument.
All authors
RE: RE: A letter to the editors regarding the article “Performance of the PointCare NOW System for CD4 Counting in HIV Patients Based on Five Independent Evaluations”
PointCare replied to GDaneau on 10 Oct 2012 at 22:08 GMT
The authors’ comments do not address my concerns, which remain:
(1) Not fully disclosing data and using data selectively to cast doubt on the performance of the PointCare NOW
(2) Systematically disregarding the FDA cleared instructions for use of the PointCare NOW, failing to disclose this in their publication, and reporting data nonetheless
Their motive for persisting in an unscientific mode of discourse is still unclear.
One example of selective use of data in the original publication indicates undisciplined work in analyzing blocked CD4 results. The authors’ claim that averaged over the 5 testing sites, there were 13.8% of samples where a quality caution flag (in their words a “failure”) occurred and CD4 counts were blocked by the device. They go on to report that upon a re-run, 54% were blocked by the device for a net rate of 9.2% with no CD4 result. My math indicates that 54% of 13.8% is 7.4%, not 9.2%. This should be cited by the authors as an error, but there is more on this topic of greater significance.
Below I will give a very detailed analysis of the data gathered at site D. I am in a position to do so because PointCare, as a financial co-sponsor of this evaluation, is in possession of all raw data from this site.
In his response to my comments, L. Westerman of Site D begins by saying that the PointCare NOW had a higher than expected “failure rate”.
First I would like to point out that the use of the word “failure rate” is an example for the authors’ non-scientific and misleading description of a PointCare NOW systems feature. As a matter of fact, PointCare NOW has more than 50 computer controlled internal checks that are designed to prevent the reporting of a wrong result in an environment where the users of the system are not laboratory technicians. As a result of these internal controls PointCare NOW is the only CD4 testing system in the world rated “moderately complex” by CLIA where all other CD4 testing systems are rated “highly complex”. Reporting this protecting feature as a “failure” rather than a virtue only underscores the authors’ negative bias towards PointCare.
In addition the author offered no data to back his “failure” assertion so I reviewed the files from Site D and found that four CD4 results used in the final data analysis were blocked. I also found that Westerman had artificially induced the PointCare device to block all four samples. In all four instances Westerman had diluted the samples to such an extreme that device alarms were activated because either physiologically inconsistent hematology values were detected by the PointCare NOW software or bizarre cell counts caused internal quality checks by the software to be compromised. For example one sample had a hemoglobin level of 0.6 g/dl and no detectable neutrophils. Another had a neutrophil count of 900 yet a bizarre neutrophil % of 60%, so, as described in the operator’s manual and training, the device blocked the CD4 result just in case the operator was unable to interpret these highly abnormal hematology values and question the integrity of the sample on their own. In undiluted samples which were run according to manufacturer’s specifications no blocked CD4 counts were found at site D, so in fact the blocked CD4 reporting rate from device failure was actually zero and not “higher than expected” as Westerman asserts.
Westerman’s unscientific bias is evident when he fails to explain to the reader that the PointCare NOW uses FDA cleared hematology quality flags that cannot be overridden by the operator and are designed to detect possibly adulterated samples such as they were at site D or samples that are too old or compromised by micro-clots due to syringe blood draws or compromised by heat or otherwise. In other words, the PointCare NOW system blocks CD4 results to assure quality results in environments where experienced operators are not available. We at PointCare consider this an important positive feature of a system that was specifically designed for resource poor settings and not a “failure”. I would like to add that even at a rate of 7.4% of patients having to leave the clinic without a CD4 result (which is higher than we typically see at PointCare NOW sites) we have a much better outcome for the patient than is experienced in central laboratory settings where an average of 40% of loss to follow up is reported (Rosen S, Fox MP, Gill CJ (2007) Patient retention in antiretroviral therapy programs in sub-Saharan Africa: A systematic review. PLoS Med 4(10): e298.
Losina E, Bassett IV, Giddy J, Chetty S, Regan S, et al. (2010) The ‘‘ART’’ of Linkage: Pre-Treatment Loss to Care after HIV Diagnosis at Two PEPFAR Sites in Durban, South Africa. PLoS ONE 5(3): e9538. doi:10.1371/journal.pone.0009538doi:10.1371/journal.pmed.0040298.)
More than 90% of patients who come to sites with a PointCare NOW system receive their CD4 result and treatment accordingly in contrast to only 60% of patients whose blood samples are sent to a central laboratory. Attention to the current literature is required in order to present to readers a balanced analysis of “failure”.
The PointCare NOW with every CD4 test also reports hematology data (WBC, Lymphocytes, Monocytes, Neutrophils, Eosinophils and Hemoglobin), which frequently show that a patient with blocked CD4 results has a serious hematological abnormality (generally very high out of range neutrophil counts) needing immediate attention to possible bacterial or fungal infection. The authors did not disclose this operational aspect of the PointCare NOW to the reader. They used none of the hematology data associated with such patients, even though the operator’s manual instructs to do so. By operating the device incorrectly, and not disclosing this to the reader the authors obscured important quality and patient protection features of the PointCare device. We at PointCare have accumulated data showing that in 40 to 60% of cases where CD4 results are blocked a hematological disorder that requires treatment is the reason for the blocked result. To be clear, hematology results are displayed and printed simultaneously with the CD4 results and can be easily observed by the caregiver.
L. Westerman of site D asserts in his response (but not in the publication) that the correlation with his reference system was poor at low CD4 counts. The data files shared do not support this point. We enclose the full range correlation data from site D (Single Platform and Dual Platform) for readers to judge whether the low range appears to have poor correlation as he claims.
Single Platform Site D: http://s11.postimage.org/...
Dual Platform Site D: http://s16.postimage.org/...
L. Westerman also asserts in his response that the CD4% correlation for the PointCare NOW was unacceptable. It is well-understood that diluting samples may lower the CD4 count, but it should have no effect on CD4%. Since all of site D’s samples were from normal donors and not from HIV patients all of the site D CD4% values were in the narrow range of normal; a range that is in fact statistically too narrow for computing a meaningful correlation coefficient. This point has been adequately made with numerous examples in the manuscripts by Bland and Altman. With site D’s flawed use of correlation statistics no conclusion can be drawn regarding device performance on CD4%.
Not only is site D’s statistical analysis of the CD4% data flawed but so also is the method of sample preparation. To assist the reader with the analysis of site D’s CD4% data in this additional aspect we supply below the Bland-Altman plots for all diluted and undiluted samples using the data sent to us from site D. Note the presence of marked low samples ranging over most of the plot causing a negative bias (-8.6%) and high SD (16.7%).
http://s17.postimage.org/...
We then removed all samples that had been diluted using Hgb<7.0 as a conservative indication of dilution (no key as to which samples had been diluted was supplied by Site D despite our requests) and plotted the data again as shown below.
http://s13.postimage.org/...
Note that the abberantly low samples were in fact all diluted and therefore compromised in some way. In 2010 we suggested to Westerman that we should jointly work to understand just what his undisclosed dilution method did to the samples, but received no response. Without these compromised samples the bias and SD are quite good. The Bland Altman analysis now shows a low bias of -1.0% and an approximate Bland-Altman SD for CD4% of 3.5% which is also low (0.078x45%= 3.5%). This widely adopted statistic introduced by Bland and Altman when the data range is too narrow for correlation statistics shows no cause for criticism of the PointCare NOW CD4% on samples that were run according to manufacturer’s specifications. This analysis was shared with Westerman in 2010 but with no response from him to date.
L. Westerman at Site D states in his response that the PointCare CD4% precision data were unacceptable compared to the reference data. This was not disclosed in the original publication and Westerman offers no data to support this assertion in his response. We now summarize Site D’s results so that the reader can evaluate Westerman’s remarks. For ten replicates of CD4% using 5 normal range samples the reference CV was 2%. For ten replicates of CD4% using 5 normal range samples the PointCare NOW CV was 5%. The two precision values for CD4% are in fact statistically different, but Westerman fails to disclose that in the very same precision testing runs both the PointCare NOW and FACSCalibur reference device had identical precision results of 5% for CD4 count. This is but another example of non-disclosure and selective use of data by Westerman to cast doubt on the PointCare NOW performance. Westerman has, at best, avoided the fact that both the PointCare NOW and reference methods had acceptable CD4% and CD4 count precision performance from any clinical perspective and , at worst, simply hidden data that do not make his case.
L. Westerman made very brief and unsupported mention in the publication and in his response that “cell density” may somehow account for the low bias and high correlation with respect to the reference method that he observed. Beyond this obscure language, he has offered no actual hypothesis and no data whatsoever to support one, so I cannot agree that his comments ever were or now remain credible. The PointCare NOW device is FDA cleared for CD4 cell densities ranging from 50 per microliter to 6,000 per microliter. The very high density limit enables pediatric monitoring of high CD4 counts without sample dilution.
Finally, it is also remarkable that the investigator of site D does not disclose in the materials and methods section of the publication that he used diluted samples which are explicitly prohibited by the manufacturer. He does not disclose the method of dilution which is always highly critical in any whole blood dilution. He does not disclose the method of resealing the sample tube and reapplying the appropriate amount of vacuum and mechanical alignment which is required by the fully automated system in order to pipette an accurate sample volume. When asked in 2010 why he took off the cap and diluted the samples anyway, he responded that every flow cytometer should be able to run diluted samples even though it is specifically prohibited in the user manual. Our specifications clearly showed at the time that the highest CD4 counts (6,000/microliter) can be handled by the device without dilution.
I am not able to comment in the same scientific detail on sites other than D as they have refused to share data with us. Even though we financially co-sponsored work at site A and the protocol called for data sharing, this site would not engage in any dialogue whatsoever.
The statement “at sites A and B, PointCare instruments were installed by PointCare engineers and validated them for use …” is false. I have submitted e-mails to the editors clearly stating the PointCare engineers’ concerns. I quote from an e-mail that was sent to the site B investigator by PointCare describing what the PointCare engineer had observed:
The engineer found that the instrument was being operated in a room that exceeded the temperature limit of the PointCare NOW. Furthermore, he noted that the temperature limit in the lab for the FACSCalibur also often exceeded the manufacturer's specification. We are greatly concerned that this will negatively affect the results from both instruments for any study samples that were run under those conditions. Temperature can have a large effect on instrument flow rates, assay conditions, and blood sample integrity that can be seen on both instrument platforms.
He had also noticed that the instruments were connected to ungrounded outlets, which again are against both PointCare and BD manufacturer's specifications. Not only is this a great danger for the instrument, as they sit unprotected, but they also could show some variation due to uneven current. He actually measured nearly 60V between line/ground and neutral/ground.
Although we have not been given any of the comparison data, by examining the
plot files that are on the PointCare NOW instrument, I have serious
concerns that there could be issues with discordant results. The plots that I have seen show a tell-tale sign of heat damage to the blood samples. Since the PointCare NOW is a light-scatter based instrument, it relies on the integrity of the blood sample.
The investigator, in an e-mail, admitted to the lack of electrical grounding: In Tete or other small towns in Mozambique, the electricity is never grounded. He proceeded with the evaluation anyway and did not disclose the problem in the publication.
When the PointCare engineer reiterated his concerns he received the following response from L Boel:
I am sorry to say that after conversation with my superior and with the project managers I was told not to communicate about the Point Care study or about the results of it. All I can say about it is that I invested all my skill and my over 30 years of experience in the task I had to perform in Mozambique and that I executed it in the best local possible circumstances and in the most objective way.
So may I please ask you not to contact me anymore about the PointCare.
PointCare had to conclude that the instruments at site B were operated outside both manufacturer’s specifications despite explicit warning from our engineer. The data from this site would therefore have to be deemed invalid and the study should be repeated under controlled conditions in line with manufacturer’s specifications. If the local circumstances did not allow for the operation of the PointCare NOW according to manufacturer’s specification, as L Boel indicates in his e-mails, the study should have been cancelled altogether with the simple conclusion that the PointCare NOW is not suitable for the environmental conditions provided at site B.
During our engineer’s site visit we were also able to examine dot plots from the reference instrument FACSCalibur which only aggravated our concerns about the site’s problems. Below is an example of a typical FACSCalibur dot plot from site B for the reader with flow cytometry experience to form an independent opinion as to whether there is a gate-able lymphocyte cluster at all and also judge the quality of the gate chosen.
http://s7.postimage.org/w...
At site A the PointCare engineer was not permitted to complete the qualification of the instrument as called for in the installation protocol. More specifically, part of the validation process after installation was the testing of at least 10 patient samples on both the PointCare NOW and the reference method. Such testing would have indicated immediately any unacceptable conditions for either one of the methods. Our engineer was not permitted to carry out this testing. Moreover, at site A the electrical problems were such that an electronic board was damaged in the PointCare NOW part way through the study and had to be replaced.
The statement “For site E, PointCare representatives installed a USAID-purchased PointCare instrument into the University of the Witwatersrand/National Health Laboratory Service reference laboratory, for a validation of the equipment prior to its use in clinical settings” is another falsehood. I have submitted documentation to the editors showing that data for this publication from site E were taken in the latter part of 2009. I also submitted our shipping records showing that a USAID-purchased instrument did not leave PointCare until January 2010. Our device history records show that installation of this instrument was attempted on March 18, 2010 but was never completed (a choice of the owner) and no reagents were ever sent to this site. Our records also show that a sales demonstration instrument was installed by the local PointCare distributor in the University of the Witwatersrand/National Health Laboratory Service reference laboratory in late 2009 which was intended by mutual agreement with site E to be for qualitative demonstration purposes only and not certified for patient use or quantitative analysis. When our engineer found out that the instrument was being used for patient samples he immediately disabled the use of the instrument electronically and it was removed from the site shortly thereafter. The data published here from site E were taken with this uncertified sales demonstration unit.
In conclusion, I believe that it would be most ethical if the authors withdrew this publication and agreed to repeat their evaluations under controlled conditions adhering to manufacturer’s specifications and full disclosure of methods and data.
