Developing the analysis of SMR data

1
Developing the analysis of SMR data

• Rod Muir
• ISD

2
Routine administrative data

• Consistent over time
• Standard definitions but..
• ..little clinical information..
• ..no information on severity etc..
• BUT when used imaginatively
• (e.g. arthroplasty)
• or when linked to other data (SRR SNBTS etc)

3
Surgical profiles so far

• Mortality- overall and by type of surgery
• Rates of mortality by type of surgery
• LOS
• Cross boundary flow
• Waiting times
• Bed occupancy
• DVT/ PE
• Readmissions
• Time to theatre (hip fracture)
• Revision surgery (orthopedic surgery)

4
Surgical mortality

• 2,717,595 patients were admitted to a surgical
  specialty in Scotland during the financial year
  2004/2005
• 4.0 of patients die within 120 days

5
Surgical mortality

• General Surgery (6)
• General Surgery (excl Vascular) (6)
• Vascular Surgery (9.6)
• Cardiothoracic Surgery (6.1)
• Cardiac Surgery (5.1)
• Thoracic Surgery (12.1)
• Neurosurgery (7.4)

6
Surgical mortality

• There are almost twice as many inpatient cases as
  day cases.
• Inpatient cases contribute to almost 9 times as
  many deaths (89.8 and 10.2 respectively).

7
Surgical mortality

• Deprivation has the strongest relationship with
  mortality out of all the variables.
• As deprivation increases, so does mortality
• 3.4 in the least deprived compared to 4.4 in
  the most deprived.

8
Surgical mortality

• Little relationship between mortality and health
  board of treatment.
• Non-elective - 76.9 of the total deaths
  elective cases 23.1).
• Mortality rises with increasing co-morbidity
• 1.7 for patients with none
• 6.7 for patients with five co-morbidities.

9
Surgical mortality

• Mortality rises with age (0.4 for patients under
  45 increasing to 14.0 for patients aged 80 and
  over).
• 76.6 of the deaths are in patients aged 65 and
  over.
• As the length of stay rises, so does the
  percentage of deaths (2.3 for patients staying
  less than a week and 20.1 for patients staying 3
  weeks and over).

10
Type of hospital

• Teaching Hospitals
• Large General Hospitals
• General Hospitals
• Childrens Hospitals
• Community Hospitals
• Long Stay Hospitals
• Other

• Large General Hospitals 4.3
• Teaching Hospitals 4.5
• higher than average mortality

11
Case mix adjustment

• Two schools of thought
• a) Keep it simple
• b) Try to adjust for everything
• Critical factors
• Is the factor evenly distributed?
• How does it affect outcome?
• Age sex deprivationother e.g. HBR?

12
(No Transcript)
13
(No Transcript)
14
Jarman adjustment

• BMJ 5 June 1999 (Jarman et al.)
• Indirect standardisation year, age, sex, method
  of admission (emergency or elective) and LOS (up
  to 365 days)
• (each of the 80 three-character ICD diagnosis
  groups that account for 80 of all in-hospital
  deaths for all English non-specialist large
  trusts.)

15
Case mix adjustment the future

• Surrogate indicators e.g. number of previous
  admissions
• Multivariate analysis
• Confounding?
• Keep it simple?
• Decision tree analysis (cluster analysis)
• Currently being used in hip fracture audit

16
Data analysis the future

• Comparison over time
• Local audit data /_ linkage to national data
• Raising the bar

17
patients treated surgically but documented as
unfit for theatre within 24 safe operating hours
Fig. 2 Percentage of patients treated
surgically but documented as unfit for theatre
within 24 safe operating hours of ward admission
Percentage of patients treated surgically but
documented as unfit for theatre within 24 safe
operating hours of ward admission
Scottish Hip Fracture Audit Report 2007
18
Patients Treated Surgically but Documented As
Unfit for Theatre Within 24 Safe Operating Hours
Fig. 2 Percentage of patients treated
surgically but documented as unfit for theatre
within 24 safe operating hours of ward admission
Percentage of patients treated surgically but
documented as unfit for theatre within 24 safe
operating hours of ward admission
Scottish Hip Fracture Audit Report 2007
19
Audit data

• Linkage to national datasets
• Stroke renal replacement therapy SHFA SJR
  cardiac surgery cancer audits SIVMS
• Powerful way of adding value
• The only safe clinician is one who knows his
  or her outcomes.

20
eScrips uptake
21
Comment encourager les autres?
22
Data analysis the future

• Which data?
• All clinical groups should be involved
• Which conditions?
• Sentinel conditions critical conditions for
  volumes/ outcomes
• The critical success factor
• Clinical lead

23
Why dont physicians enthusiastically support QI
programmes?

• Criteria are wrong
• Blame
• Resources
• Workable model
• Shekele QSHC 2002116

24
What data do we need?

• What is the problem?
• What do we need to change to improve outcomes
  (for patients)
• What do we need to measure to see whether the
  outcome has improved?

25
Data analysis the future

• The data can only tell you so much
• The starting point should be change and
  improvement with the data acting in a
  supporting role.

26
(No Transcript)
27

• The end