Centralised Data Extraction from Practices

1
Centralised Data Extraction from Practices
Pandoras Box ?

• Dr John Steyn
• Clinical eHealth Adviser, Lothian
• Retired GP

2
The cynic knows the price of everything, and the
value of nothing

• Oscar Wilde

3
Data in Primary Care

• League Table
• Media manipulation
• Consent

4
Data in Primary Care

• PTI Practice Team Information
• 60 Practices 6 of population
• Data on 47 practices
• gt1,000 practices in Scotland

5
Primary Care Data

• Primary Care Clinical Informatics Unit (PCCIU)
  Aberdeen
• ISD University of Aberdeen
• Technique
• 2009 191 Scottish Practices
• GPass

6
Lothian

• Total 880,000
• 126 General Practices
• 25,170 to 235

7
Lothian

• Planned Admissions 111,000
• Emergency Admissions 75,000

8
Primary Care Data

• GP 5,400,000
• Community Health Service 1,022,000
• Out Patient attendances 1,013,000
• AE attendances 231,000
• NHS 24 contacts 186,000
• LUCS contacts 114,000
• Planned Admissions 111,000
• Emergency Admissions 75,000

9
Primary Care Data

• QOF (Quality Outcomes Framework)
• GP 17.5 are QOF
• Practice Nurses 31 are QOF
• 78 of consultations NOT in QOF

10
Practice 6,730 list

• Prevalence of Disease
• Hypertension 1,234
• Asthma 420
• Kidney Disease 305
• Heart Disease 300
• Diabetes 263
• Stroke 199
• Prostate cancer 26

11
QOF Quality and Outcomes

• Stroke Diabetes
• 2013 16,319 34,256
• 2005 13,907 24,764
• UP 17.3 38.3

12
QOF Quality and Outcomes

• Hypertension
• 2013 106,527
• 2005 87,498
• UP 21.7 19,029

13
QOF Quality and Outcomes

• Heart Disease
• 2013 30,222
• 2005 32,891
• DOWN 8.1
  

14
QOF Quality and Outcomes

• Hypothyroidism
• 2013 23,593
• 2005 19,770
• UP 3,823

15
Primary Care Data

• NHS Lothian
• Laboratory Data
• Methodology incl Info Governance
• Thyroid Function Tests
• PSA (Prostate Specific Antigen)
• Cholesterol
• FSH / LH

16
Thyroid testing

• Over 65 Prevalence Testing
• / 100
  pts
• 25 4.53
  20.4
• 14.8 2.80 16.5
• 1 0.37
  3.1

17
Thyroid testing

• Over 65 Prevalence Testing
• / 100
  pts
• 24 4.53
  20.4
• 16 4.49
  38.9
• 16 4.27
  33.8
• 15 3.91
  28.4
• 22 3.89
  22.6

18
Thyroid testing

• Over 65 Prevalence Testing
• / 100
  pts
• 1 0.37
  6.9
• 3 0.64
  4.8
• 4 0.90
  8.8
• 4 1.03
  5.2
• 7 1.07
  13.0

19
Primary Care Data

• Laboratory Data
• Thyroid Function Tests
• 12 highest prevalence of Hypothyroidism
• 6 in highest rate of testing
• 12 lowest prevalence of Hypothyroidism
• 5 in lowest rate of testing

20
Thyroid - Lessons

• If dont test, wont diagnose
• Informed comparison with peers
• Clinical interpretation important

21
Cholesterol testing

• No of Tests / Pts 45 160,720
• 1.08
• 0.47
• 0.24

22
Cholesterol testing

• No of Tests / Pts 45 Keep Well
• 1.08 KW
• 0.93 KW
• 0.93 KW
• 0.87
• 0.84

23
Cholesterol testing

• No of Tests / Pts 45
• 0.24 High
  elderly
• 0.26
• 0.28
• 0.29 High
  elderly
• 0.34 High
  elderly

24
Cholesterol - lessons

• Clinical interpretation essential
• Need to know that no need to check in over 85s
  

25
Emergency Admissions

• Impact of minute change in Primary Care
• HF 962 admitted, 4,500 not admitted
• COPD 2,230 admitted, 11,300 not admitted
• If slight increase in admissions, massive impact

26
Admissions - In and Out of Hours

• GP NHS 24 Direct
• In Hours 81 1 18
• LUCS NHS 24 Direct
• OOH 26 29
  45

27
Dementia - Midlothian

• Data on all patients from Primary Care over 65
  with a coded diagnosis of dementia

28
Dementia - Midlothian

• Nearly a third of hospital episodes for patients
  from the dementia cohort did not include a code
  for dementia in any of the diagnosis fields.

29
Dementia -

• If hospital records had been used to define the
  cohort of dementia patients, the crude prevalence
  would have been 1.6 v 5.4, total health and
  social care expenditure on the cohort would have
  been 5.2m v 12.1m

30
Primary Care Data

• Consultation GP sub-committee, Practice
  Managers
• Consent of Practice
• Access to data - Practice-based server ie remote
• Clinically led TRUST

31
Primary Care Data

• COPD 77 practices
• Information Governance
• Findings

32
Data linkage

• Primary care data for patient X

Data for record linkage plus UID
Clinical data plus UID
Link to data from other datasets at ISD
Researcher combines anonymised linked dataset
and primary care clinical data
Remove identifiers apart from UID
33
Results
77 practices consented (75)
49 female
7072 patients
51 male
Mean follow up 4.4 years
1838 died (26) 940 (51) before first admission
1819 admitted (26)
34
Adjusted hazard ratios with 95 confidence
intervals for primary care interventions
QOF spirometry HR 0.46 (95
CI 0.41-0.53)
QOF FEV measured HR 0.43 (95 CI
0.37-0.49)
QOF flu vaccine HR 0.50 (95CI
0.45-0.56)
QOF inhaler checke HR 0.37 (95CI
0.32-0.42)
QOF annual review HR 0.27 (95 CI
0.24-0.31)
Pneumococcal vaccine HR 0.60 (95CI
0.54-0.66)
Self-management plan HR 0.17 (95 CI
0.11-0.26)
Pulmonary rehab HR 0.21 (95
CI0.14-0.30)
Avoids or delays hospital admission
Shorter time to hospital admission
35
Comparison of effect on admissions
Intervention Cochrane Reviews Schembri (2009) Lothian Cohort
Flu vaccine ? (exacerbations) ? ?
Spirometry ?
FEV measured ?
Annual Review ?
Inhaler check ? (at diagnosis) ?
Self management plan ? ?
Pulmonary Rehab ? ?
Pneumococcal vaccine No effect ?
36
Conclusions

• Range of interventions in primary care associated
  with preventing/delaying first admission for COPD
• Strong evidence for benefits to NHS (as well as
  to patients) of what GPs currently do
• being further explored by CSO funded work
• Coverage of effective interventions is incomplete
• explore which patient groups should be targeted

37
Primary Care Data

• QIS / Healthcare Improvement Scotland
• Parkinsons Disease (PD)
• 1,112 people with a diagnosis of PD, and 2,517
  people receiving Parkinsons medication.
• Many drugs can be used for gt1 disease

38
Primary Care Data

• QOF coding has QA
• Non-QOF coding quality / complete
• Not coded Medication

39
Primary Care Data

• SG SPIRE Is this expensive duplication.
• HB - Vision 360 / Emisweb

40
Conclusions

• Rich data but need to be aware of limits
• Need clinical input to interpret
• Need local knowledge to interpret
• NOT Name and Shame
• Help to inform, and learn..
• Avoid League Tables esp uninformed