Forensic Science Service

Written evidence submitted by Professor Peter Gill (FSS 19)

Declaration of Interest

I am a forensic scientist with 30 years’ experience. I joined the Forensic Science Service (FSS) in 1982. I began research into DNA in 1985, and was the first forensic scientist to collaborate closely with Sir Alec Jeffreys. After a period of Home Office supported research, in the same year we jointly published the first demonstration of the forensic application of DNA profiling. In 1987 I was given an award under the civil service inventor’s scheme for my discovery of the preferential sperm DNA extraction and the development of associated forensic tests. I was subsequently responsible for developing/introducing most of the fundamental tests used in current modern DNA analysis, including STR multiplexes and the methodologies that are used by the National DNA database. I was employed as Principal Research Scientist at the Forensic Science Service (FSS). This is the highest scientific grade within the FSS. I left in 2008 moving to Strathclyde University. I have now transferred to a new position as Professor of Forensic Genetics, Oslo University, Norway (although I retain a secondary (minor) affiliation with Strathclyde University).

I have published more than 140 papers in the peer reviewed scientific literature; the majority were published under FSS auspices. My papers have been cited in publications by other scientists more than 4,600 times over a 20 year period: http://www.researcherid.com/rid/C-3056-2009. These papers were published primarily with ex-colleagues at the FSS and form the backbone of forensic genetics in the UK and worldwide. I currently specialise in forensic genetics, statistics, interpretation of DNA evidence, and development of novel quality assurance methods. I actively work to promote the support the development of open-source software. I provide and develop training initiatives for forensic scientists.

I have been involved with a number of high profile cases, including R v Hoey (Omagh bombing). Currently I am a member of the European Network of Forensic Science Institutes (ENFSI) and chair of the ‘methods, analysis and interpretation sub-section’. I briefly chaired the national UK DNA technical working group before the regulator’s position was formalised. I am a member of the European DNA Profiling Group (EDNAP) and the international society of forensic genetics (ISFG) where I regularly chair the DNA commission and run international workshops on statistics and interpretation of evidence. In 2010 I was international research fellow of ESR, New Zealand, and presented a paper to a meeting of MPs. (http://www.beehive.govt.nz/speech/celebrating-20-years-dna-forensics-new-zealand).

The International community of Forensic scientists – organisational links

1) Forensic scientists are part of an international group of scientists with very strong collaborative links. Within Europe, the primary organisation that brings scientists together is the European Network of Forensic Science Institutes (ENFSI) http://www.enfsi.eu/. For DNA profiling evidence there is an additional (older) group – the European DNA profiling group (EDNAP) http://www.isfg.org/EDNAP which exists under the auspices of the major academic body, the International Society of Forensic Genetics (ISFG) http://www.isfg.org/ . There are strong links with counterpart organisations in the US and Australia/New Zealand jurisdictions. Because forensic science is an unusual discipline that embraces many divergent areas, there are also strong links with universities engaged in fundamental research, lawyers, and other public and commercial bodies.

2) This review is focussed on DNA profiling evidence. DNA profiling is undoubtedly the most advanced field of the forensic sciences, and widely regarded as a model for other disciplines.

3) With the sole exception of England and Wales, representation at international meetings of ENFSI/EDNAP is via public bodies, generally either government body or police laboratories for the former, including university labs for the latter.

4) At ENFSI, England and Wales are the only countries represented by two private companies (currently FSS and LGC). Scotland is represented by Strathclyde University (for historical reasons), and the Scottish Forensic laboratories (SPSA).

5) Forensic science is dominated by public sector organisations throughout the EU and beyond. This has led to a special working relationship between the EU laboratories, with an unrivalled working ethos of trust and collaboration between laboratories. The purpose is to collaborate and to share information in order to progress our field for public (not for commercial) benefit.

6) The UK research councils (EPSRC and BBSRC) do not support forensic science programmes. It has also proven extremely difficult to obtain funding via EU programmes e.g. FP7, after several unsuccessful attempts by members of the ENFSI/EDNAP/ISFG groups.

7) Since the FSS became a trading fund, there has been gradual erosion of the ‘research for public benefit ethos’ within the organisation, in favour of strong support of the commercial ethos by FSS management. This is a natural result of the (inviable) directive to privatise to the FSS.

Historical

8) Over the space of 20 years, a relatively small team of dedicated scientists (in the 1980-2000) was able to develop all of the methods used in current casework. This included the first demonstration of the principle that DNA profiling could be used in casework, the development of extraction methods to isolate sperm; the first mass screen in 1987 (Colin Pitchfork) [1] ; the first demonstration of STRs in casework; the first use of multiplexes in casework.

9) As a prime example, in the early 1990s the FSS was approached by the Russian Federation in order to analyse the remains of the Romanov family [2] . This was only possible because the FSS was one of the very few laboratories in the world that was capable of carrying out the necessary work. The Romanov example combined the use of mitochondrial analysis with the earliest examples of short tandem repeats (STRs) and demonstrated the immense power of DNA profiling on ancient remains. Subsequent work disproved the claim of Anna Anderson to be Anastasia [3] . This work was of fundamental importance since it led directly to development of the tools: multiplexed STRs that were used in the first national DNA database in the world.

10) These developments had international significance. This is the legacy that international colleagues appreciate. But this state of affairs existed ten years ago. Since then, the FSS has undergone many changes that make it unrecognisable as the institution that formerly existed.

11) Since 2005, the research output at the FSS has been in progressive decline. The current ‘research’ is mainly ‘for-profit’; disclosure is strictly regulated by FSS management and is selective. There is a conflict of interest between ENFSI membership and the ‘marketing opportunity’ ethos of a commercial organisation.

12) UK forensic science finds itself at a cross-road. The decisions to be made by government over the next year will be crucial to providing the road map for the next decade and beyond. At stake is public confidence in the criminal justice system itself including its (inseparable) component parts (especially the NDNAD).

13) The complexity of forensic science is underestimated. It is unrealistic to suppose that the necessary infrastructure can be supplied solely by private companies. It is also unrealistic to suppose that the existing accredition systems are sufficient within the UK environment. Much support is required. The following discussion is intended to provide prime examples why a cohesive public research capability is required in the UK in order to keep pace with a highly dynamic (changing) environment.

The Prüm treaty, the development of new European (ESS) markers and the National DNA database (NDNAD)

14) The Prum treaty was agreed in 2005 [4] . The stated purpose is: ‘to step up cross-border cooperation, particularly in combating terrorism, cross-border crime and illegal migration’ by ‘facilitation of international DNA-profile enquiries through direct access to national DNA-databases to assist in crime investigation and identification of suspects’.

15) A previous House of Commons Select Committee on forensic science [5] noted: ‘The police currently record 10 DNA markers per individual in order to generate a profile for the NDNAD. Professor Sir Alec Jeffreys asserted that this was insufficient, arguing that the number of markers collected should be raised to 15 or 16.’

16) Indeed, these recommendations were quickly adopted by the ENFSI DNA group in recognition that more and better standards were needed to facilitate Prüm. A number of new European loci were investigated and recommended [6] . I mplementation of this EU policy has been largely unfunded and poorly supported. Hence progress has been slow.

17) The existing SGM multiplex used to compile the existing c.4m samples on the current UK national DNA database (NDNAD) is now ‘out-of-date’. New markers developed in consultation with ENFSI, now supplied by Applied Biosystems and Promega Corp., are superior both in discriminating potential and in their effectiveness to provide results from highly degraded (compromised) stain material. Ten years is a long time in science. In the interim, not only has the discriminating power improved in line with ENFSI recommendations, but new buffer-systems are available that actually increase the chance of success of getting a result from a case (e.g. by reducing the effect of inhibition). Whereas many EU states plan to introduce the new ESS markers into casework over the coming year in line with an EU council recommendation [7] , there appears to be no coherent strategy within the UK to introduce the new tests within this time-frame.

18) EU efforts are currently coordinated under ENFSI auspices, to compile frequency databases and to validate them, for the new marker systems across more than 20 EU states. This study is under way and follows a previous model to compile and validate EU databases [8] . The results will be compiled into a centralised (global) resource http://strbase.org.

19) A difficulty that all national DNA databases have to contend with is the ‘lock-in’ effect. When four million samples have been processed, it is difficult to change and to engineer retrospective compatibility. This issue first arose in 2000 with the upgrade from SGM to SGM plus systems. Loci can only be added (they cannot be taken away). The best solution is to reanalyse/upgrade samples. Change is expensive, and requires much validation and specialist research support to achieve.

20) When responsibility of the NDNAD was part of FSS function, all of the necessary experts (biologists, IT specialists and others) were all under one roof, along with the original inventors of the database. The effect of devolving the NDNAD to NPIA was to divorce the expertise required to maintain and upgrade the database in a timely way.

21) It is quite clear that frequent upgrades to the NDNAD are required now and in the future, simply because science is not static. This is, however highly complex to manage. If upgrades are not carried out in timely fashion, then the result that follows is that the UK forensic science services and the associated NDNAD cannot keep pace with the new standards adopted elsewhere in the EU.

22) Currently, there is no published timescale in place to upgrade the NDNAD to the new marker system in line with the ENFSI recommendations of 2006 [9] . Being locked in the past, the inevitable consequence is that casework is carried out with less efficiency than would otherwise be the case elsewhere in the EU. Cases will effectively be ‘lost’ (i.e. they will fail to provide probative results in laboratories not equipped with the latest technology).

Importance of research to conduct risk analysis, to support the regulator function

23) The complexity of the UK network of scientific laboratories and the centralised NDNAD requires effort to ensure compatibility, to enable comparisons and to carry out risk analysis. The complexity is such that new methods are required to be implemented in order to undertake this function.

24) All scientific processes are subject to error. A pro-active approach is needed to understand and to evaluate risks inherent in the NDNAD and the suppliers to the NDNAD. It is not suggested that errors are common, rather I deal here with the rare event – but an error rate as low as of 1 in 1million may have consequences that are serious. The potential for error is important to consider, but difficult to evaluate.

25) Existing accreditation systems do not take sufficient account of variation between laboratories. In a recent study of accredited laboratories by NIST in the US, the authors found discrepancies in reported strength of evidence, between laboratories, of a staggering ten orders of magnitude difference – for the same set of samples [10] .

26) This important study demonstrates that a highly pro-active approach is required to understand the differences that exist within existing national processes, and to understand their significance. Existing accreditation systems operate at a more basic level of ‘compliance’, and are not geared up to carrying out the much deeper investigations that are described above.

27) New methods are required to provide a much deeper understanding of the diversity of processes. These are complex to design and to interpret. This is an example where a dedicated research unit is needed. It is suggested that a publicly funded lab that is modelled on NIST in the US would be suitable for the UK (if we don’t turn over the stones, we won’t find the bugs).

28) Currently there is no scientific assessment of risk – defined in terms of rates of ‘false inclusion’ (the chance that a random person will match a crime stain) and ‘false exclusion’ (the chance that an individual will fail to match a crime stain – i.e. a failure to detect a perpetrator who is on the NDNAD. Gill et al [11] . identified the main risk to be the problem of false exclusion: i.e. ‘missing a match’

29) Risk analysis is essential to undertake, again, this is not a trivial matter because novel methodology is required to evaluate the probabilty of rare events caused by errors (typing errors are caused a number of different factors e.g. contamination, transcription errors and indequate rule-sets that are used to upload profiles onto the NDNAD), for example, see Gill et al [12] . A sound research base is required to properly understand and to facilitate a complex quality assurance regime.

30) To summarise, it is argued that courts need additional information in terms of concurrent risk analysis in order to place into context the application of micro-probabilities.

31) The NDNAD still uses a simplistic approach to data-analysis that was developed more than 15 years ago by the FSS [13] . But this was in relation to the simple environment which existed in the past, where very few processes were used. A complex environment requires sophisticated solutions just to stand still.

32) The commercial ethos in the UK raises difficult challenges. As already noted, commercialisation (by definition) drives diversification. Courts require uniformity of technique, but diversification causes divergence and increases complexity. In a commercial environment, it follows that ‘enhanced’ regulation must keep pace to compare the diversity of methods that currently exist within the UK; otherwise we cannot be assured that method (a) is comparable to method (b) and we cannot be assured that the performances of laboratories are comparable. Ultimately, the courts cannot be assured that they are getting the best evidence. Ethical issues are also raised.

Role of public sector in the commercial environment

33) Private companies will usually wish to work on routine casework that can be easily costed. Laboratories are typically designed as ‘conveyor belts’ where there is little scope to vary the test or to apply tests that are outside the usual remit.

34) Because it is time consuming and expensive to validate and to implement new tests, there is little incentive to introduce tests that are rarely used.The ‘conveyor belt’ method means that there is inevitably less attention to rescuing cases that might otherwise be reported if more effort or if different approaches were applied.

35) Mitochondrial DNA tests and Y-chromosomal tests are routinely carried out in the public EU laboratories, but are comparatively rare within the UK. Unusual tests may be non-profit making and demand special techniques that are invaluable to specific cases, but because they are only required rarely, companies cannot necessarily justify the costs of implementing rarely used methods.

36) Examples of specialised tests include: use of mitochondrial DNA, Y-chromosome tests, red hair marker, laser-micro-dissection, application of RNA body fluid tests (MtDNA is currently only offered by the FSS; it is not clear if many of the newer specialist tests are available in the UK).

37) Of course the role of the regulator is limited by resources. The regulator can only assess techniques that are proffered by the various companies. The regulator cannot offer advice whether method (a) is better than method (b). The company’s techniques are often ‘commercial in confidence’ and there are problems/consequences that arise from this:

a. The defence expert often has great difficulty in assessing the test. Ideally, any test used in court should be both publicly available and have sufficient detail publically available so that all interested scientists have some opportunity to examine the methods that are used. This is how the scientific process works – by disclosure, mutual cooperation, interchange and peer review.

b. Commercialisation does not promote exchange of data, collaboration and convergence. Neither does it promote openness. In a recent court case (R v. T) [14] the judge criticised the FSS for using an internally developed ‘commercial in confidence’ database on footwear marks. Paragraph 84 of the judgement states:

‘There is also the further difficulty, even if it [the database] could be used for this purpose, that the data are the property of the FSS and are not routinely available to all examiners. It is only available in a particular case to an examiner appointed to consider the report of an FSS examiner.’ [15]

c. This database was not available for peer review and not available to other scientists to assess. It is not possible for peer review to operate on a restricted basis. A retrial has now been ordered.

38) This case also demonstrates that a serious mismatch exists between the government’s aspiration to privatise forensic science, versus the court requirement for openness, disclosure, and scientific peer review. It is easily demonstrated, therefore, that the framework to utilise forensic science in the UK, where the market is entirely privatised, is already fatally flawed.

39) What the courts require is ‘uniformity’ between suppliers, so that results can be easily compared between laboratories, and collaborative working is encouraged to ensure that the best techniques are in universal use. To provide an analogy, there would be public outcry if one NHS hospital was using an inefficient procedure that resulted in elevated death rates, compared to another NHS hospital that used a method that halved the death rate in patients. Imagine the situation where the second hospital refuses to allow the first hospital to use its technique because it was ‘commercial in confidence’. Yet the use of sub-optimal DNA methods in forensic science laboratories can have similar (dire) consequences.

40) We do not currently know how efficient the existing laboratories are in the UK, because there is no framework that enables this discovery. The regulator function does not include the necessary comparative studies.

41) The ISFG and the scientific societies strongly support development of open-source software and methods. This is realistically the only way forward to ensure complete transparency, openness and unfettered peer-review.

Summary

42) Some kind of public/private partnership is required, where private enterprise is underpinned and supported by the public (non-profit making) entity with a strong research function. This entity will also interact with the scientific societies without the embarrassing encumbrance of commercial profit making ethos.

43) For a publicly funded organisation to be accepted by the international community, it will be important for it to abandon commercial aspirations that compromise its integrity.

44) Research must be collaborative and involve all of the suppliers, for the common good, free of patent restrictions. Courts will not accept secret tests that have not been subject to rigorous peer review and challenge. The public will not accept sub-standard tests being used in any laboratory. An exploratory framework is needed to discover whether laboratories are providing sub-standard results. The NDNAD will need close attention to ensure that it is fit for purpose within the diversified UK environment. Numeric risk analysis is essential to undertake to understand the errors in the system and whether the consequences are acceptable. Currently there is insufficient information to know the lay of the land and insufficient discovery.

45) The research group must be closely associated with specialist caseworkers. Research is of little use in our area unless the efforts are quickly translated into court-going techniques. This interaction is immensely important. This is why researchers and caseworkers must work closely with each other.

46) Research must not be confused with ‘casework validation’ or ‘implementation’. These two functions are part of the laboratory requirement to demonstrate that a test (new to the laboratory) has been properly implemented, tested and used.

47) The research laboratory exists in order to collaborate within the existing international framework, free from commercial constraints, to collaborate in open-source initiatives and to provide new tests.

48) Databases used for estimating strength of evidence must be open, and centralised and subject to peer review to allay concerns of R v. T – the ‘research group’ is a natural custodian of these centralised databases to be used for the benefit of all suppliers.

49) Continuing (centralised) education of forensic scientists is also a requirement to consider.

50) The research group will work closely with the regulator in order to develop a framework that evaluates the numerical risk of errors. This is not trivial and requires a deep understanding of all the processes within the UK system (including the NDNAD).

51) These ideas do not sit easily with the existing disparate commercialised UK framework. Whereas it is clear that the international community of forensic scientists will quickly embrace a newly formed UK public ethos forensic laboratory that overarches the standards and provides unfettered advice and contributes freely to the international research initiative. The community will quickly treat with suspicion any organisation that is motivated primarily by commercial reasons.

52) UK forensic science is therefore at a cross-road; at stake is not only the international scientific reputation of forensic science, but also a question of whether public confidence can be maintained.

Professor Peter Gill

Professor of Forensic Genetics, University of Oslo

Senior Lecturer, University of Strathclyde

7 February 2011