OpenSAFELY: Public Opinion
This week sees the publication of an independent Citizens’ Jury commissioned for NHSx and the National Data Guardian which found that OpenSAFELY was by far the most strongly and consistently supported of all NHS COVID data projects examined. In this blog post we give some background to the Citizens Jury, our technical work to respect patients’ privacy, and the details of the Jury findings.
On 7th May 2020, the OpenSAFELY Collaborative pre-printed the world's largest study into factors associated with death from Covid-19, based on an analysis running across the full pseudonymised health records of 40% of the English population. This is an unprecedented scale of data. It was only made possible because of a huge collaboration including our group (the DataLab at the University of Oxford), the EHR research group at London School of Hygiene and Tropical Medicine, NHS England, and TPP. Over 42 days during the peak of the first wave of COVID-19 this team worked day and night to produce a fully open-source, privacy-preserving software platform, capable of running open and reproducible analytics across electronic health records, all held securely in situ. Since then the OpenSAFELY platform has expanded to a full scale analytic environment for secure data analysis, reproducible data curation, federated analysis, and code sharing, with every line of code for the platform, for data management, and for data analysis all shared openly by default, in re-usable forms, automatically, and without exception.
This is a draft discussion paper, the first of a series exploring “open team science” approaches to managing health data, and specifically how to create a collaborative computational data science ecosystem where the sharing and re-use of objects such as codelists and code is facilitated, encouraged, recognised, and rewarded. As a microcosm of this we have first explored “codelists”. There are currently no ‘answers’ or preferred solutions given. We will be holding an open discussion with the research community on 2nd March at 3pm - you can book to join us here.
We have been very busy since our last newsletter back in July and there are tonnes of exciting updates for you here!
Measure Update: Total Oral Morphine Equivalence
The Faculty of Pain Medicine has recently updated their recommendation on oral morphine equivalence (OME) which we use on our OpenPrescribing measure of OME. We have taken this opportunity to update and a new novel implementation of how we assess OME. Until this work is completed we have taken the decision to “suspend” the measure from dashboards however you can still view the old method using this link.
Victory! We have the hospital medicines data. Now: biologic medicines for severe Asthma
In July, Ben and Brian wrote a piece in the British Medical Journal arguing that hospital medicines data should be openly shared. Magnificently, the NHS has now made secondary care medicines data (SCMD) available. You can read the full technical specification of the data here but briefly: it is hospital pharmacy stock control data, which is collected and processed by Rx-Info, and is now published on the NHS Business Services Authority website in the NHS dm+d standard we know, love, and have documented well.
What is OpenSAFELY
Working on behalf of NHS England we have now built a full, open source, highly secure analytics platform running across the full pseudonymised primary care records of 24 million people, rising soon to 55 million, 95% of the population of England. We have pursued a new model: for privacy, security, low cost, and near-real-time data access, we have built the analytics platform inside the EHR data centre of the major EHR providers, where the data already resides; in addition we have built software that uses tiered increasingly non-disclosive tables to prevent researchers ever needing direct access to the disclosive underlying data to run analyses; code is developed against simulated data using open platforms before moving to the live data environment. Everything has run smoothly. We are fully live inside TPP; we are signed off with full data access and end-stage tech development for the computational platform with EMIS.
OpenPrescribing and DataLab Papers
It has been a busy month for paper publication at The DataLab. We have written a brief description of the most recent papers below. Please sharewith colleagues and [get in touch(mailto:[email protected])if you have any relevant observations! Remember you can read all our academic papers related to OpenPrescribing on our research page.
Methotrexate Prescribing Safety – New paper in BJGP
This week the British Journal of General Practice published our latest paper on unsafe prescribing of methotrexate. We found that the prevalence of unsafe methotrexate prescribing (10mg tablets) has reduced but remains common, with substantial variation between practices and CCGs. In the paper we also discuss recommendations for better strategies around implementation.
OpenSAFELY is a new secure analytics platform for electronic health records in the NHS, created to deliver urgent results during the global COVID-19 emergency. OpenSAFELY is a collaboration between the DataLab, the EHR group at London School of Hygiene and Tropical Medicine and TPP who produce SystmOne. OpenSAFELY is now successfully delivering analyses across more than 24 million patients’ full pseudonymised primary care NHS records. The first analysis from OpenSAFELY is Factors associated with COVID-19-related hospital death in the linked electronic health records of 17 million adult NHS patients with more answers to important questions expected shortly.
Most people share their end of year roundup during late December when everyone is too full of cake to read. Now you’re back in the saddle, here’s our roundup of everything the DataLab threw out into the world over the previous 12 months!
OpenPrescribing.net went from strength to strength, with over 135,000 unique users last year. We now have over 80 measures of prescribing safety, efficacy and cost-effectiveness and have been working on new types of measures and alerts to identify “outlier” prescribing, such as with zuclopenthixol. In 2019 we also: launched new dashboards and bespoke alerts for every single primary care network (PCN), sustainability and transformation partnership (STP), and NHS region; upgraded the Analyse page to support more organisations with bespoke medicines queries; and developed tools to help solve the problem of Ghost Branded Generics which cost the NHS an extra £11.6million per annum but has massively reduced during 2019.
PCN Dashboards Now Live!
We have now launched our long-awaited Primary Care Networks (PCNs) dashboard, made possible thanks to the the membership list being published by NHS England last week.
We have PCN prescribing dashboards for every single NHS PCN and their member general practices. (Don’t know your PCN’s name? You can find it on your practice dashboard). This allows anyone to explore NHS prescribing patterns in their PCN and how this compares to others across England – supporting safer, more efficient prescribing.
Our EU TrialsTracker has been assessing compliance with EU trial reporting guidelines for over a year now. Our data has been used by Parliamentary Committees and trial sponsors to improve the reporting of clinical trials registered on the EU Clinical Trials Register (EUCTR).
Last week, the Commons Science and Technology Select Committee convened an inquiry following-up their previous clinical trials transparency efforts. The Select Committee has been very active in the trials transparency space over the past two years. Earlier this year they sent letters to every public university and NHS trust in the UK that sponsors clinical trials on the EU register. These letters reminded sponsors of their trial reporting responsibilities and provided them with reporting performance figures based on data from our EU TrialsTracker.
New Paper in BMJ
Our newest paper is now out in the BMJ! Here we show huge variation in adoption of warranted changes in prescribing behaviour, using some exciting new openly available change detection methods.
We now have over 80 measures!
This month we have launched a range of new measures. As always, measures are prioritised on your dashboard by potential for improvement so you can quickly and easily spot where your organisation is an outlier. You can also sort by potential savings or group measures by clinical condition.
New Feature! Measures linked to Analyse page
Following many requests from users, we have created links from our measures to the corresponding search on the Analyse page. This is particularly useful if you want to see multiple practices or CCGs on one chart, view the results on a map, or see exactly which products are included in the measure.
You will notice this is not yet available for all of our measures (which now number close to 100), due to the complex way some measures are constructed. This currently affects just 12 of our standard measures and 4 of the Low-Priority prescribing measures, and we hope to keep increasing the options in the coming months.
New Feature! DM+D browser
This week we have launched a new browser for the NHS Dictionary of Medicines and Devices, better known as dm+d. dm+d is the standard dictionary for the medicines and devices used across the NHS. At last count there were over 150,000 packs of medicines and devices described. You can read more about the dm+d in this detailed blog.
Earlier this year we noticed a disturbing trend among UK trials registered on the EUCTR. The amount of newly approved and registered trials on the EUCTR dropped drastically following the 2016 Brexit vote. The chart from that blog post (from May 2019) is included below.
When we reached out to the MHRA to ask about this, they stated that there was actually no decline in UK trials post-Brexit but rather administrative delays that kept trials from appearing on the public-facing EUCTR website. They noted this issue was being addressed and updates would be completed by the end of Q2 2019.
New low-priority measures approved
Yesterday (27 June), NHS England completed its consultation on an additional proposed set of products being considered for a recommendation against regular use in primary care. Therefore, on the site we have now included these new items in the set of low-priority measures. The newly added products include bath and shower emollients, higher cost insulin pen needles and dronedarone. See how your practice or CCG is performing on these measures here, or navigate to the Low Priority measures from your favourite organisation’s dashboard.
Varied implementation of UTI prescribing guidelines
One of our recent papers, published in the Journal of Antimicrobial Chemotherapy, highlighted variability in the speed at which different CCGs switched from prescribing trimethoprim to nitrofurantoin (as recommended by PHE for uncomplicated UTI).
It appears that the practices which changed the most were in CCGs that had taken some action to promote the new guidelines, such as a change in formulary. What are the implications? Read more on this in our blog and you can of course see any region’s performance on this measure on OpenPrescribing.
We recently noticed a very dramatic drop in the number of UK clinical trials post-Brexit on the EU Clinical Trials Register (which would be a catastrophe for British science) and wrote to MHRA to get their view. It turns out that the drop is largely due to bad data management: this is worrying, but not the same thing as a drop in the number of UK clinical trials.
The Short Story
We’ve been conducting research on trends in clinical trials globally with one very striking recent finding: examining clinical trials on medicinal products, using the EU Clinical Trials Register (EUCTR), the number of trials with a competent authority decision from the UK has fallen dramatically. As you can see on the graph below: from a long-term trend of around 150 new trials per quarter, after 2016 there is a rapid drop to fewer than 50, and now fewer than 10.
New Price Concessions alerts
We have introduced a new email alert service, built upon our recently-launched NHS Price Concession calculator, which shows the cost impact of concessions for every single practice, CCG and for all of England combined.
As you know, here at the DataLab we are working on the RetractoBot project to reduce the number of cited retracted papers.
Citations of retracted RCTs are particularly dangerous because such trials provide strong and unbiased evidence of treatment’s safety and efficacy (hierarchy of evidence). Moreover, results of RCTs are often pooled in systematic reviews and meta-analyses, which are used to synthesise the available evidence on a given subject or to justify clinical guidelines.
Now that we’ve launched our FDAAA TrialsTracker, we plan on occasionally taking a closer look at some of the trials that go unreported. Our first blog was about a trial examining 2 drug combinations for managing pain during labor (NCT01846221).
So why do we think this specific trial is due to report? While we go through how we established our criteria and set up our database in detail in our preprint paper on Biorxiv, we wanted to walk you through exactly what fields matter on ClinicalTrials.gov and how we used them to build our tracker. We think we were able to do this well, but are open to feedback and will amend our paper and methods as necessary.
At OpenPrescribing, we provide tools for analysing prescribing behaviour in primary care in the NHS in England. If you work with prescribing data, you may have noticed that practices which are “dormant” apparently continue to prescribe. This short article explains why, for the curious.
What is a Dormant practice?
Our prescribing data comes from the NHS Business Services Authority (NHSBSA), who are responsible for processing dispensing information supplied by pharmacies. Each line of the data includes a practice code which uniquely identifies the GP Practice that issued the prescription. To understand more about each practice (including its contact details, location, and CCG membership), we cross-reference this code with GP Practice data published by NHS Digital. That data contains a column, status, which sometimes contains the value D, for dormant.