i2b2 - GEM data load for GENVASC

Tags: GENVASC Study

Patient Request

UHLDWH.DWBRICCS.dbo.UVW_GENVASC_PATIENTS_GEM => UHLDBSQLCORP02.LLR_CSS.dbo.BRICCS_GENVASC_PATIENTS

Data Response

UHLDBSQLCORP02.LLR_CSS.dbo.Genvasc_Clinical_ReadCode => UHLDWH.DWBRICCS.dbo.Genvasc_Clinical_ReadCode

UHLDBSQLCORP02.LLR_CSS.dbo.Genvasc_Ethnic_ReadCode => UHLDWH.DWBRICCS.dbo.Genvasc_Ethnic_ReadCode

Requested Read Codes

Health Check

Field Name	Read Code v3
Cardiovascular disease risk assessment declined	XaN8t
Cardiovascular disease risk assessment offered	XaPl0
Informed consent given	XaLQR
Patient informed about research study	9Q...
Cigarette consumption	Ub1tI
Smoking cessation advice	Ua1Nz
Health education - smoking	6791.
Referral to smoking cessation advisor	XaItC
Not interested in stopping smoking	XaIkY
Cigar consumption	Ub1tJ
Loose tobacco consumption	Y01e6
Smoker	137R.
Ex-smoker	Ub1na
Never smoked tobacco	XE0oh
Current non-smoker	137L.
Ethnic groups (census)	9S...
- White - ethnic group	9S1..
- Other ethnic non-mixed (NMO)	9SA..
- Ethnic groups (census) NOS	9SZ..
- Mixed ethnic census group	XaFwG
- Black - ethnic group	XaFwH
- Asian - ethnic group	XaFwz
- Other ethnic group	XaFx1
Ethnic category - 2001 census	XaJQu
- British or mixed British - ethnic category 2001 census	XaJQv
- Irish - ethnic category 2001 census	XaJQw
- Other White background - ethnic category 2001 census	XaJQx
- White and Black Caribbean - ethnic category 2001 census	XaJQy
- White and Black African - ethnic category 2001 census	XaJQz
- White and Asian - ethnic category 2001 census	XaJR0
- Other Mixed background - ethnic category 2001 census	XaJR1
- Indian or British Indian - ethnic category 2001 census	XaJR2
- Pakistani or British Pakistani - ethnic category 2001 census	XaJR3
- Bangladeshi or British Bangladeshi - ethn categ 2001 census	XaJR4
- Other Asian background - ethnic category 2001 census	XaJR5
- Caribbean - ethnic category 2001 census	XaJR6
- African - ethnic category 2001 census	XaJR7
- Other Black background - ethnic category 2001 census	XaJR8
- Chinese - ethnic category 2001 census	XaJR9
- Other - ethnic category 2001 census	XaJRA
- Ethnic category not stated - 2001 census	XaJRB
FH: Cardiovascular disease 1st degree female reltve < 65 yrs	XaP9M
FH: Cardiovascular disease 1st degree male relative < 55 yrs	XaP9K
Alcohol intake	136..
Advice on alcohol consumption	Xa1dA
Enjoys light exercise	1383.
Enjoys moderate exercise	1384.
Enjoys heavy exercise	1385.
Body mass index - observation	22K..
O/E - height	229..
O/E - weight	22A..
Waist circumference	Xa041
Ideal body weight	66CB.
Advice about weight	XaADJ
Dietary advice	8CA4.
Refer to weight management programme	XaJSu
Referral to dietitian	XaBSz
Pulse rate	X773s
Pulse regular	XM02J
Pulse irregular	X76JE
O/E - Systolic BP reading	2469.
O/E - Diastolic BP reading	246A.
Fasting	X78x9
Serum cholesterol level	XE2eD
Serum HDL cholesterol level	44P5.
Serum triglyceride levels	XE2q9
Serum LDL cholesterol level	44P6.
Serum glucose level	44f..
Total cholesterol:HDL ratio	44PF.
HbA1c level (DCCT aligned)	XaERp
Cardiovascular event risk	XaIpv
Cardiovascular disease risk assessment done	XaPkZ
Follow-up arranged	8H8..
High risk of diabetes mellitus	XaZLG

QRisk Parameters

Atrial fibrillation Codes: G5730, XaOft, XaOft, XaOfa, XaOfa, Xa2E8, Xa2E8, X202R, X202R, X202S, X202S, Xa7nI, Xa7nI, XaEga, XaEga

Type 2 Diabetes: XaOPt, X40J5, C1090, C1091, C1092, C1093, C1094, C1095, C1097, XaELQ, XaFWI, XaFn7, XaFn8, XaFn9, XaEnp, XaEnq, XaKyX, C1096, XaFmA, XaJQp, XaF05, XaIzQ, XaIzR, X40J6, X40JJ, C1011, C1031, XaIrf, XM1Xk

Chronic Kidney Disease: XaLHI, XaNbo, XaNbn, XaO3t, XaO3u, XaO3v, XaO3w, XaO3x, XaO3y, XaLHJ, XaO3z, XaO40, XaLHK, XaO41, XaO42, XaLHH, XaO3r, XaO3s, XaLHG, XaO3p, XaO3q

Rheumatoid Arthritis: N040., N0421, N042z, X701k, X705v, X705u, X705t, N042., N040J, N040H, N0403, N0401, N0405, N0409, N0400, N040F, N0402, N0404, N0407, N040A, N040G, N040L, N040M, N040C, N040D, N040B, N0408, X701m, X701j, X701i, X701h, X701l, Nyu10, Nyu1G, Nyu11, XaBMO, Xa3gL, Nyu12, Xa3gM, Xa3gN, Xa3gO, Xa3gP, G5yA., N040T, N040K, N040E, N0406, N041., XE1DU, X705I, G5y8.

Hypertension Diagnosis: XE0Ub, XE0Uc, G200., G201., XE0Ud, G24.., G240., G241., G244., G24z., G24z1, Gyu20, G240z, G24z0, G2..., G2y.., G2z.., Gyu2., Xa0Cs, XSDSb, G202., Xa3fQ, XaZWn, XaZbz, XaZWm

Excluded - Presumed Type 1

X40J4 Type I diabetes mellitus Xa4g7 Unstable type I diabetes mellitus X40JY Insulin-dependent diabetes mellitus secretory diarrhoea synd C1080 Type I diabetes mellitus with renal complications C1081 Type I diabetes mellitus with ophthalmic complications C1082 Type I diabetes mellitus with neurological complications C1083 Type I diabetes mellitus with multiple complications C1085 Type I diabetes mellitus with ulcer C1086 Type I diabetes mellitus with gangrene C1088 Type I diabetes mellitus - poor control C1089 Type I diabetes mellitus maturity onset XaELP Type I diabetes mellitus without complication XaFWG Type I diabetes mellitus with hypoglycaemic coma XaFmK Type I diabetes mellitus with peripheral angiopathy XaFmL Type I diabetes mellitus with arthropathy XaFmM Type I diabetes mellitus with neuropathic arthropathy XaEnn Type I diabetes mellitus with mononeuropathy XaEno Type I diabetes mellitus with polyneuropathy XaKyW Type 1 diabetes mellitus with gastroparesis C1087 Type I diabetes mellitus with retinopathy XaF04 Type I diabetes mellitus with nephropathy XaFm8 Type I diabetes mellitus with diabetic cataract XaIzM Type 1 diabetes mellitus with persistent proteinuria XaIzN Type 1 diabetes mellitus with persistent microalbuminuria XaJSr Type 1 diabetes mellitus with exudative maculopathy C1010 Type 1 diabetes mellitus with ketoacidosis C1030 Type 1 diabetes mellitus with ketoacidotic coma XM1Xk Xa4g7 Xa4g7 Unstable type I diabetes mellitus XM1Xk C1088 C1088 Type I diabetes mellitus - poor control

Excluded - not sure if Type 1 or Type 2:

C10.. Diabetes mellitus XaOPu Latent autoimmune diabetes mellitus in adult X40JA Secondary diabetes mellitus XaMzI Cystic fibrosis related diabetes mellitus X40JB Secondary pancreatic diabetes mellitus X40JC Secondary endocrine diabetes mellitus XSETK Drug-induced diabetes mellitus XaJlR Secondary diabetes mellitus without complication XaJUI Diabetes mellitus induced by non-steroid drugs XaJlM DM induced by non-steroid drugs without complication XaJlL Secondary pancreatic diabetes mellitus without complication X40JG Genetic syndromes of diabetes mellitus X40JI Diabetes mellitus autosomal dominant X40JO Congenital lipoatrophic diabetes X40JS Hyperproinsulinemia XSETH Maturity onset diabetes mellitus in young X40JZ Diabetes-deafness syndrome maternally transmitted XSETp Diabetes mellitus due to insulin receptor antibodies XM1Xk Unstable diabetes X008t Diab insipidus,diab mell,optic atrophy and deafness

Episodic Data

• Disorder of the Glucose System: X40J1 and children
• Cardiovascular: XEOUX and children
• Disorders of the Kidney: X30Hc and children

Prescribing

• Drug groups primarily affecting the cardiovascular System: x01QF and Children.

Recreating the Health Check Discussion

Below are some questions with responses by Sarah-Jane Gray (in italics) about the Read Codes used to generate a QRisk Score:

• What (Read code or otherwise) do the CCG use to identify that a health check has occurred for payment purposes?

XaPkZ - Cardiovascular disease risk assessment done

• For questions where the answer is a tick box (yes/no) is there a way to differentiate for all cases between an answer of ‘no’ / unticked and the question not being answered?

No – one of the drawbacks of the system in terms of data gathering is that the read codes are for a positive response and rarely a negative. Tickbox indicates a ‘yes’ response to that question but absence of a tick could be either ‘no’ or ‘not applicable / not asked’

• Similarly, for questions where the answer is a number, is there a way to differentiate for all cases between an answer of zero and the question not being answered?

Generally with numeric, because a negative (0) can be recorded, question not being answered is not the same as a negative response.

• What audit data is available for the data completeness for health checks?

Might need more explanation on what you mean for this one J

• For calculations such as QRisk or BMI, is it possible for the health check form to be filled in such a way that they are not calculated?

Yes. These should be recorded as part of a health check but there is the possibility (particularly with QRISK being a tool external to the template which ‘pops up’) but it is possible to complete the template without calculating these.

• For calculations such as QRisk or BMI, is it possible for the calculation to be based on historic data? Is there a limit to how old that data can be?

Yes. This is generally the most recent reading in the last 5yrs except ‘treated for hypertension’ in QRISK which needs to be hypertension diagnosis (any time) but hyp meds prescribed within last 6 months of the calculation. For BMI/BP, if nothing is recorded then a default value considering the patient’s age, sex and ethnicity is calculated.

• Is there a way to correctly identify all the values associated with a health check or QRisk calculation?

Not easily. We can produce a template specification for the NHS Health Check template which lists all the codes used (this is what you attached to the original email) but not for the QRISK calculation.

• In addition, we want to make sure that we have all the information that was used to calculate the QRisk score. From what I can gather, the following fields are required for this calculation:
  • Ethnicity
  • Age > directly taken from patient demographic information
  • Sex > directly taken from patient demographic information
  • Smoking Status
  • Systolic blood pressure
  • Total cholesterol:HDL ratio
  • BMI
  • Family History of CHD in 1st degree relative under 60 > both codes used in the template will mark this as positive. On S1 is just ‘Family history of CVD’ ?
  • Townsend deprivation score > directly taken from patient demographic information (if postcode recognised)
  • Treated hypertension > Any codes in the HYP cluster (I can provide these) with hypertension medication in <6m
  • Rhematoid arthritis > Any codes in the RARTH cluster
  • Chronic renal disease > Any codes in the CKD cluster
  • Type 2 diabetes > Codes and children of X40J5 (“Type II diabetes mellitus’)
  • Atrial fibrillation > Any codes in the AFIB cluster
  • + also ‘personal history of CVD’

• Do you know what algorithm for calculating QRisk is used in SystemOne, as this may resolve the ambiguities?

Afraid not, and I can’t see any way of obtaining this from S1 – will see if it’s available elsewhere.

Health Check Information

Information about the Health Check can be found on the ​HSCIC Health Check page. This includes a ​list of mappings for the codes between Read v2 and v3.

Primary Care Medication Coding

After speaking to Stephanie Webb, I have gleaned the following information about primary care medication data:

• It is extracted from System One and EMIS into the GEM data warehouse
• System One data is probably in Read Code format.
• EMIS is probably not, but don't know what format it is in.

Patient Post Codes

After speaking to Stephanie Webb, I have gleaned the following information about patient post codes data.

• Post code data is not historic, only current post code data is recorded.
• GEM do not get post code data from the primary care IT systems, but get the information from the Exeter database.
• It may be possible to get the data from System One and EMIS, but this would involve an additional charge.

Issues

1. Missing Hypertension codes

I (RB) am still unclear which coding system is being used. Therefore I am going to submit the codes from DM+D, Read CTV3, ATC code and BNF code and see what comes back.

2. Missing Diabetes Type 1 codes

The paper in the BMJ says that the QRisk2 algorithm uses a diagnosis of diabetes type 1, but not type 2. However, the QRisk2 batch processor accepts a parameter for diabetes type 1 and this affects the result. Therefore, I (RB) have got to submit the Type 1 codes above, but keep them separated.

I have sought clarification from Stephen Hippisley-Cox at ClinRisk about why this requirement has changed.

3. Missing Similar Codes

It has become clear that we need to expand the codes used to define certain values (eg, blood pressure) to include other codes that we have yet been requesting. This will have to be done by manually searching the Read Code hierarchy.

I have asked Sarah-Jane to post a question on the SystemOne User Forum to see if we can get a definitive list of codes that are used.

I have been sent a link to the codes that ClinRisk suggest by Stephen Hippisley-Cox. ​http://qrisk.org/QRISK2-2013-Code-Lists.pdf

4. Historic diagnosis

I think that we need to ignore the window when it comes to diagnoses. This needs to be confirmed by Prof.

5. Differing versions of the QRisk2 algorithm

Sarah-Jane has posted a question on the SystemOne user forum asking which version they use.

6. Discrepancies between SystemOne and ClinRisk QRisk2 calculation tools

The SystemOne QRisk2 calculation tool has some differences to the ClinRisk tool. Instead of separating diabetes into types 1 and 2, it just has a tick box for diabetes. Also, it has a tick box for 'personal history of CVD' that is not present on the ClinRisk QRisk2 calculations tools.

I have asked Sarah-Jane to post a question on the SystemOne user forum asking what differences this will make to the result.