Citizens’ Jury
In June 2017, we brought together a small, diverse group of citizens to address the question: “Would research access to the Guthrie Card heel prick blood tests be in the public interest, and, if so, under what conditions?” A Citizens’ Jury is a well-established method to enable public participation in policy making, allowing informed deliberation on an issue and the provision of recommendations. Our Citizens’ Jury followed best practice for such deliberative public engagement6. First, we convened a steering group to provide oversight of the materials prepared for the jurors and to identify a range of expert witnesses to give evidence. Next, we consulted a Patient Participatory Involvement and Engagement (PPIE) panel to review these materials. The academic researchers did not involve themselves directly in the Jury process, other than to provide evidence or observe. The recruitment, facilitation and analysis was conducted by Ipsos MORI, Scotland, to preserve neutrality. Jurors met together for two day-long sittings to hear evidence (neutral, for and against), deliberate and reach conclusions.
Using a quota sample, a representative pool of the adult public was recruited in terms of sex, age, working status and social grade. Additional quotas were set to ensure sufficient representation of people with children under the age of five, and people with a family history of a medical condition. An attitudinal quota was set to ensure inclusion of people with varied levels of trust in public, private and third sector organisations, as previous research has found this to be a significant factor underpinning views of data sharing and use14. A total of 20 were recruited of whom 19 actually participated on day one, and 18 returned for the second day a week later. Jurors were given monetary recompense for taking part in each sitting.
Day One started with a warm up session sharing their views on health research and health relevant information, followed by evidence from various experts to stimulate discussion of issues around research use of the Guthrie Cards. Day Two included further expert witnesses but more time for jurors own deliberations and to arrive at a conclusion on the key question. Facilitation tools, such as speed dating techniques, meant that all jurors could express and reflect on their own views as well as those of others, building up to group and plenary deliberations. The first day focused on more general discussion and information sharing; the second day involved detailed deliberation of the key question and delivering of the verdict. Both days were audio recorded and transcribed for subsequent analysis. A short questionnaire was administered at the end of each day to gauge individual level views and thinking.
Background information was provided by authors SCB and DJP during the morning of Day One. Over the two days, six witnesses were called to give evidence and answer jurors’ questions. These comprised health care professionals and scientists, a Caldicott Guardian (a person responsible for protecting the confidentiality of people’s health and care information and making sure it is used properly) and a Genewatch spokesperson (representing not-for-profit groups that monitor developments in DNA technologies). Jurors also had access to a short video and other written information about comparative deliberations and policy in California and Denmark. Having heard on Day One some of the health research opportunities uniquely possible if access were granted, an introduction to some of the social, ethical and legal issues, and how the NHS protects privacy, Jurors heard more opposing and critical views on Day Two, with one witness cautioning against allowing research access and another the importance of not compromising the newborn screening programme itself.
Guthrie card documentation and sampling
Epidemiological feasibility
The Scottish Guthrie card archive is stored in around 900 boxes each containing circa 3,000 cards in a single secure location under the authority of the Director of the Scottish Newborn Screening laboratory and custodianship of the NHS Research Scotland Greater Glasgow and Clyde Biorepository. A unified digital record is in place from 2000 onwards, but the content and consistency of information available from older cards was not known at the outset. The design and information content of the Guthrie card was known to have varied over time, but not documented. We were aware that some cards had suffered water damage prior to assembling the nation-wide archive. We also know that for a period of time some cards had been autoclaved before storage. For the vast majority, the Guthrie cards have simply been stored at room temperature. The unified Guthrie card archive management database has limited, high-level summary information on the contents of each box.
We obtained permission from the Caldicott Guardians of NHS Research Scotland Greater Glasgow and Clyde and of Tayside to retrieve representative boxes from the NHS Research Scotland secure archive. Permission was given for a) examination and documentation of a sub-sample from each box to provide a snap-shot of the information attached that might be required from the purposes of linkage to other routine health records and b) sampling of cards corresponding to consented members of Generation Scotland.
Thirty boxes representing each decade from 1965 to 1999 were retrieved and examined by NHS Scotland staff at the NHS Research Scotland Greater Glasgow and Clyde Biorepository. Only summary information was shared with the rest of the study team.
The organisation and information content by box varied over time (SI Figure 1). In some boxes there were unlabelled bundles of cards (SI Figure 1D), but most had 4 labelled foolscap sub-boxes (SI Figure 1B) with multiple, date-labelled bundles of cards (SI Figure 1C).
We drew up a list of potential information that the Guthrie cards might carry from which to judge the feasibility of conducting epidemiological studies by linkage to NHS Scotland routine medical record and potentially additional consented data from research subjects.
For each box, we collected the following information: Box ID; Area Health board; Hospital; Type of card; Date of test; Child forename; Child surname; presence or absence of Community Health Index identifier. No personal information was recorded. Each box took 2 members of staff working in tandem approximately 2 hours to document.
Next, circa 1 in 100 cards from each box were examined in detail and the presence or absence of the following features documented: Child DOB; Additional comments on card; Number of bloodspots; Size of spots; Mother’s CHI; Mother’s forename, surname and birth name; Mother’s date of birth; Address; Postcode; Whether the cards had been autoclaved prior to archiving; Any other comments.
Cards from 1965 had very little information on them and in many cases did not even record the sex of the baby. Information content increased progressively over time. By the 1990s, the sex of the child and home address were generally recorded.
Retrieval of cards from consented volunteers
Over 24,000 Generation Scotland (GS) volunteers were recruited as adults between 2006 and 20113. All were born before 1993, before the digital recording of Guthrie cards began. Consent for linkage to medical records was optional, but was given by 98% of volunteers. They were asked to give information about their place of birth (country and council area). A total of 8,703 volunteers with linkage consent were born in Glasgow or Tayside area health boards between 1965 and 1992. A list of names, birthplace and date of birth was extracted from the GS database and sent to the NHS Greater Glasgow and Clyde Biorepository to look for matching cards in selected boxes. Pseudonymous ids were added to the list so that any samples from matching cards could be labelled and linked back to the GS database after genotyping. Ninety-two matching cards were identified amongst Guthrie cards from Tayside. Of these, 58 were usable for punching having fulfilled the prerequisite of leaving one spot intact (SI Figure 2). Six to ten punches were taken from each card, placed in vials labelled with a pseudonymous ID for matching to the samples donated at baseline by each Generation Scotland volunteer, and couriered to the Edinburgh Clinical Research Laboratory Genetics Laboratory for DNA analysis.
Biological feasibility
Genetic analysis of de-identified cards
Unlike the Danish Newborn Screening Biobank (DNSB), the Scottish Guthrie card archive comprises a variety of paper types and storage conditions, particularly for older cards. To establish the effect this might have on the recovery of analysable DNA, a pilot study in 2011-2013 was undertaken on a de-identified set of 136 Guthrie cards dated from 1965-2012 (Table 3). The study was mandated by the Scottish Chief Scientist Office, following a favourable opinion from the Scottish Legal Office and Research Ethics Committee. De-identified cards were provided by the Scottish bloodspot screening laboratory DNA was extracted from 3mm punches using the Sigma ENA kit. Yields varied from sample to sample, but there was no significant effect of date of birth and sufficient material was obtained for Sanger DNA sequencing in 94% of samples (Table 3).
Sampling of cards
Of the 92 Guthrie cards matched to GS volunteers, 58 (63%) had sufficient dried blood spot material remaining to take 3 mm diameter punch samples, whilst leaving at least one spot intact. These 58 were from Generation Scotland research volunteers born between 1983-1989 (i.e. 32-38 years between collection and profiling). DNA was extracted from between 6 to 8 blood spot punches using the QIAamp DNA Investigator Kit (Qiagen; cat. no. 56504), following the manufacturer’s instructions. The concentration of the DNA samples was measured using a Qubit 2.0 fluorometer and the Qubit dsDNA HS assay (Thermo Fisher; cat. no. Q32854). Total yield isolated was between 196ng and 1177ng of DNA. Up to 500ng DNA (range 160ng – 500ng) underwent bisulfite conversion (Zymo EZ-96). DNA methylation was profiled using the Infinium HumanMethylationEPIC v1.0 BeadChip (Illumina Inc.; cat. no. WG-317-1001), according to the manufacturer’s protocol (in batches of 8 samples, 56 in total). Arrays were scanned on an iScan and analysed using GenomeStudio v2011.1.
DNA methylation analysis and statistical methods
DNAm profiles were obtained from the 56 individuals using the Illumina MethylationEPIC beadchip, measuring approximately 850,000 CpGs across the genome. In addition to standard quality control measures such as probe filtering and normalisation, additional checks were performed to ensure Guthrie card and baseline samples matched with regard to predicted sex and genotype (Supplementary information, SI Figures 3 and 4 and SI Table 2).
Maternal smoking status at the time of sample collection was derived from smoking status at GS baseline, and the “years stopped” variable for former smokers, where both mother and baby are in GS. DNA methylation-based estimates of smoking status were obtained, using previouslyvalidated methods11,12.
Sample Checks
Confirmatory analyses were performed using Guthrie card DNA methylation data to ensure predicted sex (using X-chromosome data) and genotype (using “rs” control probes on the EPIC array) were consistent with peripheral blood based genotyping and DNA methylation data on samples collected at baseline recruitment (2006-2011) (SI Table 2). There was 100% concordance with predicted sex and between predicted and hard-called genotypes. Detailed information on sample checks is presented in Supplementary Information and SI Figure 3 and 4).
Smoking
An individual’s smoking status can be reliably predicted using composite DNA methylation-derived smoking scores, and effects have also been observed in the offspring of mothers who smoked during pregnancy13. Moreover, information from a single probe in the aryl-hydrocarbon receptor repressor gene (AHRR; cg05575921) can serve as a robust marker of smoking, with lower DNA methylation levels associating with current smoker status. A composite score for smoking status (EpiSmokEr) was obtained using Guthrie sample DNAm data and, along with cg05575921 DNA methylation levels, was plotted against maternal smoking status at the time of sampling11,12 (Figure 1). Consistent with previous literature, a higher overall value was observed for the EpiSmokEr score in the offspring of current smokers whereas a lower overall value was observed for the offspring of never smokers (Figure 1A; ever smoker β=0.82; P=0.016). DNA methylation levels at cg05575921 were also consistent with the literature, with lower overall levels in the offspring of current smokers relative to never smokers. (Figure 1B; ever smoker β=-0.73; P=0.05)