Further Information

Prepared by Hugh Tunstall-Pedoe, November 2007

Version 2.5 - updated February 2008

return to form

Development of the ASSIGN score [contents]

ASSIGN is a cardiovascular risk score. The name is derived from "ASSessing cardiovascular risk using SIGN" (see below) guidelines to ASSIGN preventive treatment. Derivation of the score is described in a paper published in Heart online in November 2006. The coefficients used in its calculation appear in an appendix to the paper.

The score is ostensibly the ten-year percentage risk of developing cardiovascular disease (any manifestation of coronary heart disease or cerebrovascular disease including transient ischaemic attacks) in those disease-free at recruitment. So it is a number between 0 and 100.

By convention (following Framingham scoring, and expert recommendations) the presumption is that anyone whose ASSIGN cardiovascular risk score is 20 or more is 'high risk' and a candidate for preventive treatment, and anyone with a score below that does not normally qualify. The score should be applied within the context of SIGN Guideline 97 or a similar protocol.Note the cutpoint for the score, now 20, was different in the past, and may possibly change in the future.

The score arose out of an invitation to Professor Hugh Tunstall-Pedoe (University of Dundee) to join a SIGN group on cardiovascular risk estimation. The group was concerned through the participation of Professor Graham Watt of Glasgow University with the effect on cardiovascular risk of social deprivation which he had been studying in the Midspan cohort from west central Scotland. The issue was developed further by Professor Tunstall-Pedoe with Professor Mark Woodward on a Scotland-wide basis using the SHHEC (see below) study of representative men and women recruited by the Dundee team across Scotland from 1984 to 1995. The study published in Heart in 2005 showed a large gradient in coronary risk in Scottish men and women related to their social status (defined by SIMD-see below) but inadequately explained by conventional risk factors, so not allowed for in the Framingham score (see below). This meant that traditional cardiovascular scores, such as Framingham, would result in the socially deprived being allocated less preventive treatment in relation to their future risk than the socially privileged, unless the effect of social deprivation was allowed for in some other way.

Professor Hugh Tunstall-Pedoe and Professor Mark Woodward updated the SHHEC database and extended the study to cardiovascular disease as the endpoint, rather than just coronary heart disease. Opinion was divided in the SIGN group as to whether the Framingham score should be adjusted by 'tweaking it' to take account of social deprivation, or whether a new score would be needed. This was best answered by deriving a score incorporating social deprivation along with classic risk factors (family history was also added) and seeing whether it could then be emulated by 'tweaking' the Framingham score. The new score was ASSIGN. It proved difficult to 'tweak' Framingham to take account of social deprivation for several reasons, although Framingham cardiovascular scores and ASSIGN scores were highly correlated when tested in the SHHEC study. ASSIGN was therefore adopted by SIGN after it had been shown that results of scoring were very similar in most individuals, ASSIGN scores being slightly lower. A positive family history and high score for social deprivation lifted the ASSIGN score towards or above that from Framingham and made it fairer in a mixed population.The ASSIGN score has been validated using simulation, and its performance has been compared favourably with the Framingham score both in the SHHEC study follow-up and in QRISK (see SIGN Guideline 97 and 2003 Scottish Health Survey and the discussion there).

Family History - how defined [contents]

For ASSIGN a positive 'Family history' is coronary heart disease or stroke in a parent or sibling below age 60 years OR other strong evidence of family or ethnic susceptibility, such as several close relatives affected when young. Medical records may not be sufficiently specific but a suggestive history should be recorded as YES. A missing history should be recorded as NO unless the patient is thought to belong to a susceptible ethnic group where a missing history (that is not negative but missing) should be coded as positive. If the information is missing but the patient is accessible to answer questions they should follow those described in the ASSIGN paper web-only appendix.

'Family History' is 'Yes' if any of following responses are given: *

Question	Qualifying Response
Have either of your parents developed heart disease or stroke before the age of 60?	Yes
Have any of your brothers or sisters developed heart disease or stroke before the age of 60?	Yes
Do you know that any of your grandparents, your aunts or uncles or your first cousins (their children) developed heart disease or stroke below age 60? If yes, how many?	Two or more in total

* Questions are targeted at manifestations of atheromatous disease. Specific information may be missing but judgement may be used whether a very premature death from heart disease in a relative may have been congenital or non-atheromatous, for example rheumatic heart disease.

Framingham Score [contents]

The Framingham score is the standard and original coronary or cardiovascular risk score. It was the Framingham study that first introduced the concept of risk factors and which defined the 'classic risk factors' used in its score. The Framingham score is as good (and as bad) as similar scores in discriminating future event-victims from non-victims, but it may overpredict (or rarely underpredict) event rates in different population. This problem can be resolved by altering the calibration or the cutpoints. The standard Framingham scores did not include family history (some versions now do) or any measure of social status. The ASSIGN score was developed because the Framingham score was not considered 'wrong' but unfair for those reasons which are corrected in the ASSIGN score. The latter can be used in heterogeneous modern populations of mixed social status and ethnicity and compensates for the consequent problems in estimating risk status.

Because the Framingham score is the standard, a version of it is included in the ASSIGN website for comparison with ASSIGN. Usually results are similar, ASSIGN reading lower, but ASSIGN gives higher scores in the socially deprived with a family history and particularly so in older women.

The Framingham score is given for comparison only. For use in clinical practice an official Framingham source should be preferred to the ASSIGN website.

Note also that there are several different Framingham scores published at different times and for different, (sometimes overlapping) endpoints. The one used for illustrative purposes is for cardiovascular disease published in 1991 by Anderson and others (not apparently available on the web, see Publication 5).

Funding and sponsorship [contents]

The ASSIGN score was developed in consultation with, and under the aegis of the Scottish Intercollegiate Guidelines Network (SIGN). The database which made it possible was created with funding from the Chief Scientist Office of the then Scottish Home and Health Department for its first decade, and the British Heart Foundation for its second decade. The two analyses which led to publications in Heart were funded by specific grants from the Scottish Executive Health Department (now the Scottish Government Health Directorate).

Meaning of the ASSIGN score [contents]

The ASSIGN score is the estimated risk of people who are free of cardiovascular disease at that time, of the same age and sex and risk factor values to those entered into the score, developing coronary heart disease, a transient ischaemic attack or stroke, or death from cardiovascular disease over the next ten years. It is based on the historical experience of a large number of Scottish men and women in the Scottish Heart Health Extended Cohort (see below) followed for ten years. Thus a 20% risk (the conventional cut-point for intervention) means that one fifth will develop cardiovascular disease and 50% means one in two. Note that this is not the relative risk (or risk multiple) but the actual or 'absolute' risk. However, it is an estimate and not a guarantee. A high score is compatible with freedom from cardiovascular disease. A low score is compatible with premature disease. The score ranks people by their risk of developing future cardiovascular disease and therefore prioritizes them for preventive action based on their scores.

The ASSIGN score is designed to emulate, but replace its predecessor, the Framingham score, by incorporating social deprivation and family history. The Framingham score, based on a historical American population, gives higher values on the whole than does the ASSIGN score, but there is a very high correlation between them. Without social deprivation or a family history ASSIGN usually scores lower than Framingham. With these factors it can score higher, particularly in older women. The ASSIGN score is considered by the SIGN guideline development group on cardiovascular disease and by the Scottish Government Health Department to be the most appropriate cardiovascular risk score for current use in the Scottish population, because it includes social deprivation, family history and numbers of cigarettes smoked.

Postcodes in Scotland [contents]

The postcode is equivalent to the North American Zipcode. It defines a small population by place of residence. There are approximately a quarter of a million full postcodes in Scotland, population five million people, so on average they cover only 20 people. Each postcode can be matched with a SIMD (see below) score based on census and other social information from different agencies. Place of residence is a powerful descriptor of social status, an association long recognised and increasingly used by commerce as well as government agencies. Extreme SIMD scores are found in cities where there are extremes of wealth and social stratification, so localities are very diverse across the city but homogeneous within each locality. They are less discriminatory in thinly populated rural area where close neighbours may vary considerably in wealth, occupation and education.

The postcode directory on this website will be kept up-to-date, but there may be occasional postcodes that are not recognised or do not have a SIMD score allocated. In such a case, try changing the final letter of the postcode until a result is obtained. In heterogeneous rural areas where the SIMD score seems inappropriate to the social status of the person being coded you should consider adjusting the SIMD or the final ASSIGN score to reflect this potential misclassification.

Publications [contents]

1. Tunstall-Pedoe H, Woodward M, Tavendale R, A'Brook R, McCluskey MK. Comparison of the prediction by 27 different factors of coronary heart disease and death in men and women of the Scottish Heart Health Study: cohort study. BMJ 1997; 315: 722-729. Describes the Scottish Heart Health Extended cohort and the risk factors measured, showing their apparent importance one by one. Won a national epidemiology award for its authors.
2. Tunstall-Pedoe H, Woodward M; SIGN group on risk estimation. By neglecting deprivation, cardiovascular risk scoring will exacerbate social gradients in disease. Heart 2006; 92 (3): 307-10. Epub 2005 Sep15. Argument for importance of social deprivation as a risk factor missing from Framingham score. Based on follow-up of SHHEC and application of the SIMD score to its participants. Led to the proposal to develop a new score - ASSIGN
3. Woodward M, Brindle P, Tunstall-Pedoe H. Adding social deprivation and family history to cardiovascular risk assessment: the ASSIGN score from the Scottish Heart Health Extended Cohort (SHHEC). Heart 2007: 93(2):172-6. Development of ASSIGN score and testing 'validation' against the Framingham score.
4. Risk Estimation and the Prevention of Cardiovascular Disease. SIGN guideline 97. The context within which the ASSIGN score was developed and within which it is being implemented in clinical practice in Scotland.
5. Anderson KM, Odell PM, Wilson PW, Kannel WB. Cardiovascular disease risk profiles. Am Heart J 1991; 121:293-8. The Framingham cardiovascular score with which ASSIGN was compared. Apparently not available on the web.
6. Hippisley-Cox J, Coupland C, Vinagradova Y, Robson J, May M, Brindle P. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: a prospective open cohort study. BMJ 2007; 335:136- Development of a score in England which copied the innovative features of ASSIGN and claims superiority, but was criticised in subsequent BMJ rapid (electronic) responses. Not immediately apparent is that this analysis validated the ASSIGN score in a different population by showing that it discriminated futures cases from non-cases slightly better than the Framingham score, and did almost as well as QRISK in its own host population. Calibration was another issue as it was for the Framingham score (see published correspondence) but the QRISK score identifies virtually nobody as high risk below age 55.

Risk-Factor Means to substitute for missing values [contents]

It is desirable to obtain values of all requested risk factors before calculating the ASSIGN score. However, it is possible to calculate a Provisional ASSIGN score using one or more mean values to substitute for missing values. These were obtained from the 2003 Scottish Health Survey so they are more recent than those obtained from the SHHEC study. The values are:

Age (years)	Total Cholesterol		HDL Cholesterol		Systolic Blood Pressure		Cigarettes per day for smokers *
	Male	Female	Male	Female	Male	Female	Male	Female
30-39	5.50	5.11	1.35	1.58	128	116	13	14
40-49	5.77	5.64	1.37	1.64	130	121	17	16
50-59	6.02	6.29	1.40	1.71	133	131	20	14
60+	5.72	6.37	1.36	1.66	141	139	16	15

* Smoking status should be known to calculate the ASSIGN score and should be requested if not available. If other factors are available, but not smoking, calculate a tentative score for a non-smoker and then for a smoker smoking an average number of cigarettes. Record both values to show the potential risk status, and then verify true status when you can.

After calculating a Provisional ASSIGN score the clinician may like to recalculate it substituting higher and lower values than the mean to see how robust it is, and how desirable to go back and obtain a real risk factor value to give a definitive score.

Risk-Factor Limits for ASSIGN. Definition of extreme and out-of-range values [contents]

For safety reasons it is undesirable to calculate an ASSIGN score in clinical practice using values which are so extreme as to be almost never encountered. The value may be erroneous; if not entered in error the measurement would need to be repeated, and if still true would be the subject of additional specific measures, not just incorporation into a risk score. Such values are queried when entered, as is the resulting score, which is calculated but flagged as questionable. Where risk factor data are out-of-range or nonsensical (eg 'Yes' for 'Number of cigarettes') the risk factor value is rejected and no score is calculated. These rules are slightly relaxed for the Research version of the score data entry format, allowing the user to conduct 'What if?' analysis. Extreme and out-of-range values were defined from the survey data of the Scottish Heart Health Extended Cohort (SHHEC), involving over 18000 people. For simplicity (unlike risk-factor means) the same values are used for extremes and ranges for different age and sex groups. Out-of-range values are those not occurring in this study. Extreme values are those occurring in less than 1% of participants.

The exception to the above rule is that ages beyond those of the original SHHEC cohort of 30-74 years are flagged because scoring involves extrapolation and the very young and very elderly are excluded although encountered in clinical practice. (See below)

Risk Factor	Lower Bounds (inclusive)		Upper Bounds (inclusive)
	Extreme Range	Usual Range		Extreme Range
Age (Years)	25	30	74	90
Cigarettes per day for smokers ()	0	0	40	100
Systolic Blood Pressure (mmHg)	80	100	200	250
Total Cholesterol (mmol/l)	2	3.5	9	12.5
HDL Cholesterol (mmol/l)	0.3	0.6	2.3	3.5

Scottish Heart Health Extended Cohort (SHHEC) [contents]

This is a large representative cohort of men and women recruited across Scotland in 1984-1987 reinforced with repeated random samples from north Glasgow in 1989, 1992 and 1995. An extensive panel of risk factors was measured and recorded at recruitment and, with permission, participants were followed for mortality and inpatient morbidity up to the end of 2005. For additional information see an earlier paper from 1997 in the BMJ which secured a national epidemiology award for the Cardiovascular Epidemiology Unit in Dundee whose project it was, and the two recent Heart papers, (Publications Adding social deprivation and family history to cardiovascular risk assessment, Appendix to Heart paper, SIGN Guideline 97).

Scottish Intercollegiate Guidelines Network (SIGN) [contents]

This is an internationally respected Scottish national body, described best on its website. It has developed and published, revised, republished (and in some cases withdrawn) numerous guidelines for adoption in Scotland and consulted elsewhere. The guideline that relates to ASSIGN is number 97 'Risk Estimation and the Prevention of Cardiovascular Disease' of February 2007.

SIGN group on risk estimation [contents]

Members of the SIGN (Scottish Intercollegiate Guidelines Network, 28 Thistle Street, Edinburgh EH2 1EN) risk estimation group who helped to refine the study proposal and analyses were:

• Dr James Grant (chair, principal in general practice, Auchterarder)
• Dr Moray Nairn (secretary, SIGN Edinburgh)
• Dr Adrian Brady (consultant cardiologist, Glasgow)
• Dr Peter Brindle (research and development strategy lead and honorary research fellow, Bristol Teaching Primary Care Trust and Department of Social Medicine, University of Bristol).
• Dr Alex McConnachie (statistician, Robertson Institute, Glasgow)

• Mrs Joyce Craig (senior health economist, NHS Quality Improvement Scotland)
• Mr Adam Redpath (Programme Principal, Information Services, NHS National Services Scotland, Edinburgh)
• Mr Roger Stableford (patient representative, Falkirk)
• Prof Hugh Tunstall-Pedoe (cardiovascular epidemiologist, Dundee)
• Prof Graham Watt (general practice, Glasgow).

Social Deprivation and the Scottish Index of Social Deprivation (SIMD) [contents]

SIMD is the Scottish Index of Multiple Deprivation. It is calculated for residential areas, such as postcodes, and ranges from 0.54 to 87.6 (2004 version). By population fifths from 1 (least deprived) to 5 (most deprived) the categories are:

	Least Deprived				Most Deprived
Population Fifth	First	Second	Third	Fourth	Fifth
SIMD Score Range (inclusive)	0.54 to 7.63	7.64 to 13.49	13.50 to 21.16	21.17 to 33.93	33.94 to 87.60

The SIMD distribution therefore has a long tail and shows a positive skew. The mean value is approximately 28 and the median 20.

The SIMD was revised in 2006. Although there have been some changes to the score both in its derivation and in relative movements up and down in the scores of different localities, the range and mean values are very similar and there is a very high correlation between postcode SIMD scores for 2004 and 2006. Indeed the SIMD score correlates very highly with other older deprivation scores such as the Carstairs score. See Scottish Index of Multiple Deprivation (SIMD) for further information on SIMD.

In deriving the ASSIGN score, survey data from 10-20 years ago were used with follow-up data collected since then, and this was then related to SIMD, a recently derived deprivation score. SIMD was found to be very powerful in identifying that element of risk which is related to social status and not explained by conventional risk factors.

In the ASSIGN score software found on this website there is incorporated a 'lookup' table for SIMD scores from postcode. After reading in the postcode of residence of the person concerned it places the corresponding SIMD score for that postcode into the calculation of the ASSIGN score. The scoring application is provided in three modes. In Clinical and Comparison with Framingham mode SIMD scores are looked up by postcode. In Research mode you may enter SIMD score directly, as in Version 1 of the score.

Disclaimer [contents]

The authors and sponsors of the ASSIGN cardiovascular score disclaim responsibility for misuse or abuse in clinical practice. It should be used as a guide to anticipatory preventive treatment in identifying patients apparently at high risk. Data entered should be checked for plausibility and observations repeated in the event of any doubt. There is no warranty that any patient for whom the ASSIGN score is calculated either will, or will not, develop cardiovascular disease within ten years, as both outcomes are possible with any single score. The score is concerned with probabilities, using available information and not certainties (which is why the phrase risk estimation or risk assessment is preferable to risk prediction). The score should be used within the context of SIGN guideline 97, or similar documents on management of cardiovascular risk.

about

Version 2.5 - Revised February 2008 :: assign@assignscore.net