The PICRAM study is funded by the Health Innovation Challenge Fund, a joint initiative supported by the Department of Health and the Wellcome Trust.
The 240 UK Intensive Care Units (ICUs) admit about 110,000 patients everyyear. Data on admissions to 185 of these ICUs in 2006-8 (ICNARC, Intensive Care Audit and Research Centre database) revealed that 77% of patients were discharged to the ward recovering from their acute illness, but 11% then died before leaving hospital. Most deaths occurred without readmission to the ICU, suggesting that physiological changes went unrecognised meaning that treatment to prevent the patient’s demise could not be delivered.
Failure to recognise and act on physiological indicators of worsening acute illness in acute hospital wards is a generic problem that was recognised over a decade ago, and which prompted the NICE clinical guidelines number 50 (2007). These guidelines recommend “Track and Trigger” physiological scoring systems, coupled with a graded response including critical care outreach services (CCOSs). In spite of NICE endorsement and widespread adoption, research commissioned by the National Institute for Health Research Service Delivery and Organisation (NIHR SDO) showed that neither the implementation of “Track and Trigger” systems nor the commissioning of CCOSs influenced post-ICU mortality. Post-ICU patients who were regularly reviewed by any member of the ICU team did however derive a survival benefit, but ICUs were very rarely resourced for this, especially outside office hours. There is therefore an unmet need for more effective recognition/response systems for acutely unwell patients generally, and a specific need for patients discharged from ICU, who after an expensive and traumatic period of treatment still experience a high in-hospital mortality, amounting to over 9000 deaths in the UK each year, some of which are demonstrably avoidable.
The PICRAM project has developed a risk-prediction model which uses individual data collected during a patient’s ICU stay to adjust alarm thresholds after ICU discharge.
The team is now developing a working prototype of this individualised alerting system for in-hospital use for patients discharged from the intensive care unit.
PICRAM dataset for research use
During the PICRAM project we collated several thousand patient records. This anonymous dataset can be made available, on specific request, for research purposes. The database is owned and controlled by the PICRAM management committee who will review each data request on an individual basis.
Projects that have been granted access to the dataset include:
- The SILENCE study (NIHR RfPB ref: PB-PG-0613-31034) has used the dataset to assess incidence of intensive care unit acquired delirium.
- PhD student projects have used the dataset to quantify the performance of a number of machine learning algorithms (RCUK Digital Economy Programme grant number EP/G036861/1), and to assess the predictive qualities of a new mortality score, 'OASIS'.
- The Clinical Haemotology unit at Oxford University Hospitals NHS Foundation Trust have used the dataset to assess the number of intensive care patients with low platelet counts who undergo certain clinical procedures during their admission.
In the first instance please contact Julie Darbyshire if you would like more information about the PICRAM dataset.