Making ICU Data Work for You

1
Making ICU Data Work for You

• Nazir Lone
• SpR in ICM

2
Outline

• ICU data sources available
• Identifying a problem area/question to tackle
• Methodological approaches
• Examples of uses of ICU data
• Ethical issues

3
UK ICU Data Sources

• Scotland
• SICSAG Scottish Intensive Care Society Audit
  Group
• All 26 adult ICUs in Scotland partake
• Funded centrally by Scottish Executive
• Episode data on all admissions to ICUs in
  Scotland
• Many HDU admissions also included

4

• The Rest of the UK
• Case Mix Programme
• 199 Units in England, Wales and Northern Ireland
  (75 of eligible units)
• Admission episodes to participating units
• Data entered, cleaned by ICNARC and reports sent
  back to units
• Comparisons made by adjusting for case-mix

5
Data Available to You

• Routinely collected data stored in most units
• Access to data is controlled
• Local audit use
• Lead clinician in audit
• Local research project
• Observational study
• Previous research training
• Dissertation component of Masters degree
• ICU research support

6
Identifying Problems to Solve

• Most staff can identify problems in their ICU
• Refining a problem into an audit or research
  question
• Examples
• Do periods of understaffing of ICUs lead to
  impaired patient care?
• Does the use of theatre recovery when ICU is full
  lead to worse outcomes for patients?
• Can long-term ventilated patients be looked after
  in a long-term weaning unit rather than ICU?

7
Evidenced-Based Medicine

• Randomised controlled trial is the gold standard
• Observational studies further down the EBM food
  chain

8
Defining the Problem

• Examples
• Do periods of understaffing of ICUs lead to
  impaired patient care?
• Does the use of theatre recovery when ICU is full
  lead to worse outcomes for patients?
• Can long-term ventilated patients be looked after
  in a long-term weaning unit rather than ICU?
• RCTs cannot answer some questions

9
Defining the Problem I

• Background
• Every ICU has a few long-term patients on
  ventilators
• They use a lot of resource
• They can be cared for in a less resource intense
  environment
• Question
• Do we have enough long-term ventilated patients
  to set up a weaning unit?

10
Defining the Problem II

• Background
• The ICU is often run at capacity
• New admissions are often temporarily cared for in
  the recovery area of theatre
• This might not be good for these patients
• Question
• Does temporarily caring for a patient in recovery
  worsen their outcome?

11
Getting the Data

• Are the data routinely collected
• Yes (SICSAG)
• No (ICNARC for weaning unit study)
• If not routinely collected, can you get the data
  from another source?
• E.g. Staffing rotas
• Prospectively collect data
• Ethics
• Time and effort

12
Data Manipulation
13
Using Theatre Recovery as an ICU
14
Are the findings true?

• Patients cared for in recovery prior to
  admission to ICU are more likely to die than
  those admitted directly to ICU
• Is it true?
• Are the two groups similar?
• Age, diagnosis, APACHE II on admission
• Can you relate it to other ICUs?
• Pitfalls of observational studies
• Confounding
• Bias

15
Are the findings true?
Junior Doctors are a Danger to Patients

• Junior doctors went on strike for 2 weeks
• Mortality rates dropped by 50
• Funeral directors went out of business

16
Are the Groups matched?
17
Adjusting for Confounders
18
Ethical Issues

• Use of patient data raises ethical issues
• Audit vs Research
• Local trust RD notification
• Letter to chair of ethics committee

19
Ethical Issues

• Anonymised data
• Postcode and initials
• Date of birth
• Even if anonymised, precautions should be taken
• Password protected computers
• Locked data storage cabinets in hospital

20
Summary

• ICU data sources available
• Translating clinical problems into defined
  audit/research questions
• Methods used and common pitfalls in interpreting
  results
• Ethical issues raised by using routinely
  collected data

21
References

• SICSAG website www.sicsag.scot.nhs.uk
• ICNARC website www.icnarc.org
• Centre for Evidence-Based Medicine www.cebm.net
• CH Hennekens JE Buring (1987) Epidemiology in
  Medicine. Boston Little Brown.
• Wunsch H, Harrison DA, Rowan K. Health services
  research in critical care using administrative
  data. Journal of Critical Care 200520264-9.
• Ethics
• http//www.dundee.ac.uk/iea/GEP07.htm
• http//www.acepidemiology2.org/policystmts/EthicsG
  uide.asp