One of my favourite jokes from when I was a young lad who couldn’t even spell toxicologist: the Judge says, wearily, to the defence barrister, “Mr Smythe, I’ve listened to you for 10 minutes straight, and I’m none the wiser.” To which the barrister replies, “Perhaps not, my Lord, but at least you are more knowledgeable.”
And there is a crucial truth there – it’s hard to be wise without knowledge. In toxicological risk assessment, you are far more likely to draw the right conclusion if armed with a robust data set.
Meeting the challenge of identifying existing toxicity data has changed dramatically since I started in 1977. In those pre-database days, a big part of the search phase involved trips to the libraries at the Carshalton MRC library or the Sutton Royal Marsden Hospital, climbing up ladders to haul down numerous and heavy Index Medicus and Chemical Abstracts indices (I think they came in 10-year issues), staggering to a nearby desk, searching for occurrences of the CAS RN and chemical synonyms in the indexes (indices?), and painstakingly noting (pen and paper!) all the relevant pages and accession numbers telling me where in Index Medicus and Chem Abs I could find reference to a published paper on my chemical. That initial phase was followed by more climbing (I always wanted to be a gibbon when I was young, but that’s another story), more lugging of weighty tomes, and much turning of pages. Possibly explaining the arthritic nature of my right forefinger. We would then record any toxicity data summarised within the few lines provided, as well as the bibliographic details of each paper. In the next phase, we would be heaving on the levers or wheels of huge stacks to access the full papers in journals. If the libraries didn’t hold the journal, our support staff would fill in a British Library (BL) request form (for each paper) and these would be posted off to BL. Then we’d wait for the forms to reach the top of the “to do” pile of a BL staff member. Papers from BL journals would be photocopied and posted back to bibra. For other journals, the BL would forward requests on to other holding libraries. The post was greeted with much more excitement back then. Papers would drift in like messages in a bottle over anything from 2 weeks to 6 months. Other data sources available included textbooks and pre-existing bibra reviews. Hard-copy sources such as NIOSH RTECS, Patty, Browning Fisher, Casarett and Doull, Hayes, and Sittig featured large in our lives. Finally, and perhaps most importantly, we would access the bibra databank; I’ll return to that below.
Things started to change in the early-mid 1980s when US National Library of Medicine online databases started to appear. Suddenly we could search Toxline, Medline, Excerpta Medica, Chem Abs and BIOSIS (and more) from the relative comfort of a bibra chair. Every now and then, I have a flashback dream about Datastar or Dialog. Of course, we were still reliant on a dodgy dial-up connection (no internet back then) that could drop us halfway through a complex session, but now we could devise search strategies and obtain computer printouts of full bibliographies (and, if very lucky, some abstracts). And yes, not all records were correctly or fully indexed. But we no longer had to get wet running to the MRC library (a valuable advance in employee comfort). Of course, our support staff still had to go through the paper-ordering process, but the time required for the search element of a tox review was reduced considerably, allowing more time for data review and a faster turnaround time. Heady days indeed.
To return now to the bibra databank of published papers and unpublished reports. From bibra’s inception (as the British Industrial Biological Research Association) in 1961, the practice was to place a physical copy of useful papers into files. These papers would have been identified by (a) routine scanning of a large number of the critical toxicity journals, (b) scanning of Current Contents (a regular publication listing papers from numerous scientific journals), and (c) the various searches described in the paragraphs above. The filing system was set up in a way that allowed the toxicologist to go directly to a file containing all papers on a specific chemical. Subfolders allowed papers on structurally-related chemicals to be grouped closely together, an early form of the “read-across” approach that we employed for decades before it became “a thing”. The file system was (naturally) overseen by a card index system. I’m feeling all nostalgic now. Scanning journals and collecting papers, while not cheap, was necessary if we were to be capable of responding to urgent enquiries – for example, regarding contamination events and potential poisoning incidents.
In 1987, we took the plunge into a more modern age, when we established the TRACE (Toxicity Research Association Chemical Effects) database. This software allowed us to identify all of the relevant post-1987 papers we hold inhouse without stirring from our comfy office chairs. We still had to pore through the files for earlier inhouse papers but at least there was no more legging it to the MRC library in the rain.
In my follow up piece, I talk a little more about TRACE and its value to clients and the bibra toxicologists. Final thoughts (before you read Part 2):
- In 2002, we estimated that several million pounds had been invested in the bibra databank (born 1961) and TRACE database. We continue to collect key review papers and Expert Group reports to the present day.
- Did you know that Index Medicus ran from 1879-2004? It started even before I did.