Tuesday, July 10, 2012

Presumptive blood testing in the Chamberlain, Taylor, Lovejoy, and Knox/Sollecito cases

Part 34 in the Knox/Sollecito case

Update (9 July 2013)  I renumbered this entry from 35 to 34.

In this post we will examine presumptive testing in the cases against three other individuals before returning to the Knox/Sollecito case.  Presumptive blood testing has been a subject of this series twice before.

Lindy Chamberlain
In 1982 Lindy Chamberlain was tried for the 1980 murder of her daughter Azaria near Ayers Rock in the Northern Territory of Australia.  A presumptive blood test in the Chamberlain’s car was positive:  The ortho-tolidene test is similar to the tetramethylbenzidine (TMB) test in that both rely upon the pseudoperoxidase activity of hemoglobin to produce a color change.

There are at least two problems with taking the positive result as strongly inculpatory evidence.  Problem one is that the ortho-tolidene test is subject to a number of false positives, and copper dust (common in the Chamberlain’s home of Mt. Isa) is one substance that produces false positive in catalytic tests such as this one.  Yet the head of the forensic laboratory downplayed the idea that the ortho-tolidene test gives false positives, claiming that interfering substances gave a different color than blood in this experiment.  When asked whether an experienced biologist could confuse a substance such as copper with blood in this test, he replied, “Not while I’m around.”  One wishes that someone had subjected Ms. Kuhl and her superior to a blind test of their abilities.

Problem two is related to the issue of substrate controls.  Substrate controls are used to check areas nearby a stain to see whether a positive test reaction is related to the stain or not.  Dr. T. Raymond wrote,  “The original trial scientist, Mrs Joy Kuhl obtained positive (ortho-tolidine) presumptive tests for blood in a number of areas in the Torana – including the carpet, bolt hole region, nearside seat hinge, console and items that were purported to have been in the car – a chamois, a towel and a pair of nail scissors. Some of these areas were incidentally inaccessible without the removal (using appropriate screwdriver) of surface panels.”  The inaccessible areas should have been considered substrate controls. The fact that they were negative positive should have given the forensic team pause but did not.  With respect to the luminol presumptive test for blood, Barni and coworkers wrote, “Regardless of how the stains are collected, the essential requirements to be met are the recovery of the available blood, the collection of a control sample in a tested area not exhibiting chemiluminescence and the complete drying of the support used for blood collection…”  Mrs. Kuhl also devised a test for fetal hemoglobin (Hb F).  The scientific questions surrounding the Hb F test are outside the scope of the present blog entry.

Gregory Taylor
 Mike Klinkosum wrote an article about the Gregory Taylor case in North Carolina, and how a number of cases of problematic blood forensics were uncovered in an audit. "What was not disclosed at Taylor’s trial was that NCSBI Agent Deaver had only provided the positive results from a presumptive test for blood known as a phenolphthalein [Kastle-Meyer] test. He had withheld from his official laboratory report the fact that he performed a confirmatory test (called a Takayama test) on the two stains that returned negative results for blood as to both substances...Agent Deaver had performed a third test on one of the suspected stains, a species origin determination test (referred to as the Ouchterlony test), designed to determine if the stain was human blood. He received a negative reaction on that test as well, but withheld the results of the Ouchterlony test from his official laboratory report."

Amy Driver reported on the aftermath of the Taylor exoneration. "Jill Spriggs, Chief of the Bureau of Forensic Sciences for the California Department of Justice and the President-Elect of ASCLD, responded [to a question at a hearing of the North Carolina state legislature] 'That is an accurate statement. A lot of times you got no results. It didn’t mean it wasn’t blood; it meant you didn’t have enough sample, or maybe the sample was old. …What else is red-brown that will give you a positive presumptive test for blood? There’s nothing that I know.'" This is a remarkable statement. Plant peroxidases are one source of false positive reactions, and rust is another.  With respect to most crimes, animal blood is also a false positive.

Mr. Klinkosum summarized four categories of wrongdoing with respect to how tests for the presence of blood were reported. He quoted the Swecker-Wolf report on the fourth category:  "The fourth and most serious category involves cases in which the reported actual results of the confirmatory tests were over-reported or not reflective of the results contained in the lab notes. There were five such cases in this category, all handled by [Special Agent Duane] Deaver.3 One of these cases involved a defendant who was executed. In two instances, the words 'revealed the presence of blood' were used when in fact the results of the confirmatory test were reflected in the notes as negative. This language was only used by Analysts when the presence of blood was confirmed by a positive confirmatory test. In three instances the report stated that further tests were 'inconclusive' or 'failed to give any result' when the lab notes reflect negative results. It should be noted that the Analyst, SA Deaver, advised reviewers that he was trained that confirmatory tests had only two possible results, negative or positive. SA Deaver’s lab files, however, revealed these two instances in which SA Deaver used the words ‘inconclusive’ in connection with Takayama test results despite his notes reflecting a negative result in one cases (sic) and three tests and three negative results in other cases."

The problems at North Carolina’s SBI laboratory go beyond one special agent, however. The Raleigh News and Observer reported, “Jed Taub, a 30-year veteran of the SBI, said, ’We didn't report the negative result of a confirmatory test because, really, it's misleading,’ said Taub, who now works as a forensic investigator for the Pitt County Sheriff's Office. ‘We couldn't be sure it wasn't blood, so those tests really didn't matter.’’ One might argue that Taylor's case is a different situation from the Knox/Sollecito case discussed more fully below. In the former case, a negative confirmatory result was withheld. In the latter case, the TMB results were withheld until late in the first trial. However, the more general principle in question is whether or not forensic scientists should disclose all test results and let the chips fall where they may.

In one of an award-winning series of articles at the Raleigh News and Observer, Joseph Neff and Mandy Locke wrote, “The phrasing of State Bureau of Investigation lab reports from the 1980s to 2003 caused a lot of ruckus in 2010, raising the question of whether North Carolina's practices that helped prosecutions were common to crime labs across the country.”  The executive summary of the 2009 National Academy of Science report “Strengthening Forensic Science in the United States: A Path Forward” noted on p. S-15, “The terminology used in reporting and testifying about the results of forensic science investigations must be standardized.” It is unfortunate that the NC SBI laboratory did not adhere to this principle in the Taylor case.  "Regardless of how everyone else did it 20 years ago, the current and best practice is to report the results of all tests, just as the [North Carolina] SBI lab does now," [SBI Director Greg] McLeod said.  Director McLeod’s comments are suggestive of an improved culture at the SBI lab.

Laurence Lovejoy
The Laurence Lovejoy case from Illinois involves the lack of disclosure of a presumptive test. The Illinois State Supreme Court Granted Mr. Lovejoy a new trial. "Despite the positive LCV [leucocrystal violet] test, Camp tested the swab with TMB, a more sensitive and more reliable presumptive test for the presence of blood, as her protocol dictates. The TMB test was negative, indicating that the substance swabbed was not blood." The opinion goes on to state, "The parties do not dispute the facts giving rise to the alleged discovery violation. Both sides agree that Camp did not include her finding that the TMB test produced a false negative, and did not state her reason for the negative TMB test in her report."

The opinion also stated, "Further, if defendant were made aware of Camp’s conclusion that the TMB test produced a false negative because the LCV used all the hemoglobin in the blood, he could have called an expert to refute this contention during his case in chief, or could have chosen to pursue a different line of defense altogether. Defendant was prejudiced when Camp’s conclusions were revealed for the first time at trial." Earlier in the opinion it was made clear that such an expert indeed existed: "In that motion, defendant stated that Dr. Karl Reich, who holds a degree in molecular biology from the University of California, Los Angeles and Harvard University, was prepared to testify that the TMB test used by Camp is the most sensitive presumptive blood test available; that a negative TMB test strongly suggests that there is no blood in the area tested; and that Camp’s testimony that the LCV consumed the reactive blood components, thus confounding the TMB test, was incorrect."  In other words, not knowing that there was a negative TMB test nor knowing what rationale would be offered for its being negative hampered the ability to present a robust defense.

Amanda Knox and Raffaele Sollecito
The luminol-positive areas in the Knox/Sollecito case include two blobs in Filomena’s room, three bare feet in Amanda’s room, three bare feet in the corridor, and one shoe print in the corridor.  The two amorphous shapes in Filomena’s room and the shoe print have Meredith and Amanda’s DNA profile, the three prints in Amanda’s room have her profile, and the three bare prints in the corridor have no DNA profile.

A number of forensic references recommend the use of a colorimetric reagent such as TMB on areas that are luminol- or fluorescein-positive. Lisa A. Gefrides and Katherine E. Welch wrote,  “Either luminol or fluorescein can be sprayed onto large surfaces such as walls or floors and the positive areas areas marked for further testing. Both tests are very sensitive and will indicate bloodstains that may not be visible…One disadvantage to these tests is that both can have false positive reactions. Luminol and fluorescein will react with the same false positives as PH [phenolphthalein] and also with bleach and other cleaning fluids, which may interfere with blood detection on surfaces that have been cleaned. For this reason fluorescein or luminol positive areas should be retested with one of the color change presumptive tests.”  (“Serology and DNA” in “The Forensic Laboratory Handbook Procedures and Practice,” Humana Press, 2006) Stuart H. James and William G. Eckert wrote, “After any positive reaction with luminol, the stained area should be circled with a grease pencil or other suitable marker. The stained area should be checked again with another reagent for blood such as tetramethybenzidine (TMB), phenolphthalein, or o-tolidine.”  (p. 162, “Interpretation of Bloodstain Evidence at Crime Scenes,” 2nd ed., CRC Press, 1998)

A technical bulletin on Lightning Luminol said, “Luminol is only a presumptive test and should be used in conjunction with a second field presumptive test and followed by laboratory analysis if sufficient amounts of staining are detected…A second presumptive field test should be used on the areas that were detected positive with Luminol. Results should not be reported as a positive presence for blood, only positive indications of blood. It is recommended that laboratory testing, if possible, be followed on staining areas detected at scenes.”  In “The forensic luminol test for blood: unwanted interference and the effect on subsequent analysis,” Anders Nilsson of the Swedish National Laboratory of Forensic Science wrote, “Due to the lower selectivity of the luminol test positive reactions should be verified by other presumptive blood tests like the phenolphthalein test [James, S.H., Kish, P.E. and Sutton, T.P., Principle of bloodstain pattern analysis: theory and practice, 2005 CRC press Taylor & Francis Group, ISBN 978-1-8493-2014-9].”  Although one group of workers (“Accuracy, Reliability, and Safety of Luminol in Bloodstain Investigation,” Canadian Society of Forensic Science Journal 35 (3), September 2002, 113-121) has suggested putting an end to the practice of following luminol with a colorimetric test, it would seem that many laboratories do use the TMB or the Kastle-Meyer test on luminol-positive areas. 

The fact that some or all of the luminol-positive areas had been tested using TMB with negative results did not emerge until Sarah Gino’s testimony in September of 2009, near the end of the first trial. The Massei report stated, “With respect to the Luminol-positive traces found in Romanelli's room, in Knox's room and in the corridor, she stated that by analysing the SAL cards ‘we learn, in contradiction to what was presented in the technical report deposited by the Scientific Police, and also to what was said in Court, that not only was the Luminol test performed on these traces, but also the generic diagnosis for the presence of blood, using tetramethylbenzidine...and this test...gave a negative result on all the items of evidence from which it was possible to obtain a genetic profile…’” (pp. 256-257, English translation)

There has been much discussion about whether or not the luminol-positive evidence items in the Knox/Sollecito case are blood.  I will outline six reasons to be skeptical of the notion that the greater sensitivity of luminol is the explanation for why the TMB test was negative for luminol-positive areas in this case:

There are substances in some of the photographs of the application of luminol that give off a blue glow, as posted in these threads. There were blue flecks on a ruler, on a pair of boots worn by someone within the forensic police, and on the grouting between tiles.  The technical bulletin for Luminol Lightning noted that, “Luminol can give a low grade reaction with some carpet materials.” False positive reactions are seen with many substances besides bleach.

The technical notes for Luminol Lightning caution, “Also, tracking through an area that has been sprayed with Luminol will produce brighter shoe tracks. These tracks can be transferred to other parts of a crime scene. Care should be taken not misinterpret these reactions.”  This might explain the luminol-positive areas in Filomena’s room; however, whether the hallway or Filomena’s room were sprayed first is unknown.  Alternatively, one can speculate that the luminol-positive areas in Filomena’s room were the result of foot traffic carrying an unknown substance between 2 November and 18 December, when the luminol was applied.

Ranges of values for the sensitivities for both the luminol test and the TMB test have been reported in the forensic literature. Some values suggest that luminol test slightly more sensitive, but I have seen claims that both tests can detect blood that is diluted up to 1 part in 1,000,000. However, let us assume for the purposes of illustration that TMB detects blood to 1 part in 10,000 and that luminol detects blood to one part in 100,000, a tenfold difference in sensitivity. However, there is a 10,000 fold range (from undiluted blood to blood that has been diluted ten thousand fold) in which both will be positive. In other words each and every dilution would have to be between 10,000- and 100,000-fold to be TMB-negative and luminol-positive in this hypothetical situation.

Some of the luminol-positive areas were also negative for DNA.  Luminol has a slight negative effect on how much DNA can be culled from a sample, and this depends on the formulation. However, it is certainly not the case that the use of luminol precludes testing for DNA. Moreover some luminol-positive areas did have Amanda’s DNA, so we know that luminol does not always interfere with DNA profiling. So if the luminol reaction were responding to Meredith’s blood, why is there none of Meredith’s DNA in some items of evidence and no DNA at all in others?

If the only reason why TMB would be negative when luminol was positive is the lesser sensitivity of TMB relative to luminol, then there would be no reason whatsoever to follow a luminol test with a TMB. A negative TMB would not be meaningful, and a positive TMB is still not a true confirmatory test. According to the Massei report (p. 258, English translation), Sarah Gino testified about the use of TMB: “She added that, in her own experience, analyses performed with TMB on traces revealed by Luminol give about even results: 50% negative, 50% positive…” This observation suggests that the purpose of the TMB test in the field is to reduce the number of false positives, as also seems to be true based on the citations above. This interpretation is also consistent with Stefanoni’s testimony about the nature of the TMB test, as discussed in the Hellmann report: “Professor Tagliabracci, specified, without being refuted (hearing of July 18 2009, p. 174), that the tetramethylbenzedine (TMB) test is very sensitive, so much as to give a positive result even with only five red blood cells present. Dr. Stefanoni herself, moreover, clarified (preliminary hearing of October 4 2008) that, while a positive test result could be deceptive due to reactivity of the chemical [evidenziatore] with other substances, a negative result gives certainty that no blood is present.”

As blood is made more dilute, the luminol reaction becomes fainter. There is no evidence to suggest that these luminol reactions were especially faint.

A theme in at least three of these cases is that the forensic personnel have an implicit or explicit belief that a positive result in a presumptive blood test means that blood is really present and a corollary belief that any negative results in subsequent presumptive or confirmatory testing is the result of a failure in the follow-up test.  Putting it another way, the Knox/Sollecito case is not unique with respect to its misinterpreted presumptive tests for blood.  It is an open question whether or not the false beliefs would change if forensic laboratories were made more autonomous from prosecutor’s offices or law enforcement.  This is recommendation 4 in the 2009 National Academy of Science report “Strengthening Forensic Science in the United States: A Path Forward.”  Both the full report and an executive summary are available.

In the Chamberlain case, personnel from the forensic laboratory made statements about presumptive test that were false.  They destroyed plates that involved the HbF testing before the defense had a chance to examine them.  In both the Taylor and Lovejoy cases, the state's failure to disclose the test results was simply wrong.  Likewise in the Knox/Sollecito case it was wrong for Ms. Stefanoni to fail to disclose her negative TMB results prior to the trial.  Even when these results become available to the defense during the trial, this may be too late for the defense to adapt its strategy to the new information.

Sunday, July 1, 2012

The time of death in the murder of Meredith Kercher

Part 33 in the Knox/Sollecito case

The time of death (TOD) in the murder of Meredith Kercher is a central part of the question of whether Amanda Knox and Raffaele Sollecito participated in her murder.  Ms. Popovic saw Ms. Knox around 8:40 PM acting normally.  If the TOD is between 9 and 10 PM on 1 November 2007, then it is difficult to imagine how they could have imbibed drugs or alcohol, met Rudy and initiated events leading to murder in the 80 minutes that followed.  Let us examine the TOD primarily using the physiology of digestion.  This topic was recently discussed at the pro-guilt website, True Justice for Meredith Kercher, and I am grateful to its author for reminding us of the sensitive nature of this topic.  I am also grateful to the translators of the Massei-Cristiani Report and the Hellmann-Zanetti report.

Meredith’s friends’ descriptions of the meal
As summarized in the English translation of the Massei report (pp. 34-38), Robyn Butterworth was not sure about when the young women ate pizza on 1 November but thought that might have been around 6 PM.  They stopped watching the movie, “The Notebook” long enough to put the dessert, apple crumble, into the oven, according to John Follain's book, "Death in Perugia," p. 54.  Amy Frost thought that the time of the meal was 5:30 or 6 PM.  Sophie Purton thought that the time of the end of the meal was perhaps an hour before leaving, which would mean that they finished desert around 7:45.  According to Candace Dempsey’s book, Murder in Italy, Sophie Purton said that Meredith ate only part of her pizza.

Judicial estimates of Meredith’s time of death
Massei and Cristiani, the professional judges in the first trial, put the time of death as a few minutes after 11:30 PM (p. 382).  Follain (Death in Perugia, p. 344) reported that PM Mignini estimated the time of as between 11:20 and 12 midnight.  A summary of Raffaele’s appeal document suggests a TOD of 9:30-10:00.  Hellmann and Zanetti, the professional judges in the second trial, put the time of death as no later than 10:13 PM.

Gastric emptying times
The contents of Meredith’s stomach had a volume of 500 mL, and her duodenum was empty (Massei, p. 115).  Two time periods are commonly used to measure the first portion of digestion.  The time at which food begins to leave the stomach and enter the duodenum is t(lag).  The time it takes for the stomach to empty by half is t(1/2).  Because Meredith’s duodenum was empty, the more useful number is t(lag), not t(1/2).  The time between the start of Meredith’s last meal and the TOD must be less than t(lag).  In some of his comments Judge Massei seems to be concerned with t(1/2) or perhaps the time it take for the stomach to empty completely.  Yet he also wrote (p. 115), “Dr. Lalli also took into consideration the state of digestion. He stated that solids are ingested into the stomach and are not able to reach the pyloric sphincter until they are reduced to a semi-fluid or fluid consistency; the emptying of the stomach then begins to occur when some of the contents have become sufficiently fluid to reach the pylorus, which happens the third or fourth hour after eating. This is when one can find food material at the level of the duodenum (page 63 of the Lalli report).”  This is significantly longer than the average or median values of t(lag) of which I am aware, and it may be that Judge Massei simply misunderstood Dr. Lalli’s report or did not clearly understand the distinctions among the various measures of gastric emptying.  A website at Colorado State University on the pathophysiology of the digestive system shows the differences among t(lag), t(1/2), and the time for complete emptying.  These values are about 80, 150, and 220 minutes, respectively in this diagram.

A 2003 study by Chen et al. (J. Gastroenterology and Hepatology, 18, 41-46) determined a value of t(lag), namely 81.9 ± 17.4 minutes, with a range of 37.1 to 117.8 minutes.  The authors described the test meal:  “The egg [one yolk and two whites] was ingested with two slices of white bread coated with 7 g of margarine and 8 g of grape jelly, followed by 150mL water.”  A paper on gastric emptying times (Hellmig et al., "Gastric emptying time of fluids and solids in healthy subjects determined by 13C breath tests: influence of age, sex and body mass index" Volume 21, Issue 12, pages 1832–1838, Journal of Gastroenterology and Hepatology, December 2006), was published in the Journal of Gastroenterology and Hepatology, which is peer-reviewed. Other articles cited this paper at least 25 times.  These workers described their test meal:  “After addition of 50 mL of low-fat milk, the egg was scrambled and fried in a pan. The solid test meal was completed by a piece of brown bread (50 g) and butter (20 g).”  They showed that t(lag) for a solid meal did not follow a normal distribution.  The median time was 82 minutes, with the 25% percentile at 66 min. and the 75% percentile at 102 min. Out of 82 subjects (Figure 1D), the longest value was 200 minutes, and the next longest was 170 minutes (each value corresponded to a single individual).

If one uses 6:30 as an estimate for when Meredith began to eat, then 9:50 (200 minutes after 6:30) is a working estimate of the outermost reasonable time at which Meredith was still alive. In fact, the meal may have started earlier than 6:30, given the testimony of Meredith’s friends.  The problem for the prosecution does not get much better if one assumes that Meredith began her meal at 7:45 (when the desert was finished, according to Ms. Purton) and uses 200 minutes as the longest possible t(lag).  This assumption puts the latest reasonable TOD at 10:55 PM.  However, this TOD is far less likely than (for example) 9:05 PM, which is 80 minutes after 7:45.  In other words, there is no possible time for Meredith to have consumed her last meal that favors Judge Massei’s TOD over a TOD between 9 and 10 PM.  

Other indications of Meredith’s time of death
The Hellman-Zanetti report did not attempt to use stomach/duodenum physiology to ascertain the TOD.  However, these judges did use other means to reach a similar conclusion.  Meredith attempted to call her mother around 8:56 but was unsuccessful.  Meredith called her mother once a day; therefore, it would be odd that she did not call back in the period in which she was supposed to have been alone in the prosecution’s scenario.  Later there were two activities in close succession, namely an aborted call at 9:58 to Meredith’s answering machine and one at 10:00 PM to Meredith’s bank, but without the country code, neither of which is a complete call.  Finally at 10:13 there is a GPRS internet connection for 9 seconds.  The summary of Raffaele’s appeal document notes that the tower that handled this interaction can reach both the apartment and the location at which the phones were finally found.  It is difficult to see why Meredith would have initiated any of these three later activities herself.

It is strange that Meredith did not try to call her mother after her attempt at 8:56 did not succeed.  Nor did she call, text, or email anyone else after she arrived home, presumably close to 9 PM.  She did not change into night garments or remove clothing from a washing machine.  The Hellmann court put the TOD at 10:13 at the latest, and yet the evidence suggests that the TOD might have been earlier even without bringing in the physiology of gastric emptying.

The cell phone data suggest that Meredith died no later than 10:13 PM, as Judges Hellmann and Zanetti indicated. It is undisputed that there was activity on Raffaele’s computer at 9:08 PM, and the defense argued that there was activity at 9:26 and that the screensaver log files indicate additional activity.  Even by itself, the lack of material in the duodenum strongly suggests that the time of death was likely to be earlier even than 10:13.  Therefore, there was not enough time for Raffaele and Amanda supposedly to become very messed up from alcohol and/or drugs, to meet Rudy Guede, and to initiate a long series of actions that culminates in murder.  It is likely that Meredith was attacked shortly after returning home, probably between 9 and 9:30.