previous contents

UNK-EDI/GLA/SHH: Edinburgh, Glasgow and the Scottish Heart Health Study Contributor from the MPC: Hugh Tunstall-Pedoe

© National Institute for Health and Welfare and the MORGAM Project investigators Last updated: 23 March 2007 For more information, please contact Kari Kuulasmaa (firstname.lastname@thl.fi)

Contents

• Cohort identification
• Cohort recruitment and the baseline examination
• End-points followed up
• End of follow-up period
• Follow-up procedures
• Diagnostic procedures
  • at baseline
  • during follow-up
• References

Cohort identification

• Country: United Kingdom
• MPC: 37 - Scotland
  • Administrative centre: Cardiovascular Epidemiology Unit, University of Dundee, Ninewells Hospital & Medical School, Dundee DD1 9SY, UK
  • Key personnel: Hugh Tunstall-Pedoe (Principal Investigator), Richard A'Brook (former), Kate Harrison (former), Jimmy Gibson (former)
• RUs:
  • 01 - Glasgow - The Scottish MONICA Project was initiated with registration of coronary events in 1985 and population surveys in 1986 in part of Glasgow and in the City of Edinburgh, but Edinburgh was withdrawn soon afterwards. The Glasgow population was that part of the city of Glasgow which is north of the River Clyde, (which bisects the city). Built on former trade and heavy industry, Glasgow's inner city had exceptionally high social deprivation, chronic disease and population decline, (richer middle-class people live outside the centre). High levels of classic cardiovascular risk factors, (particularly smoking in women) were combined with a poor, largely manufactured, diet with low intakes of fruit and vegetables. Chosen for its exceptionally high coronary disease mortality in both sexes, survey response rates were below average for Scotland, and for MONICA (but see below). Coronary rates peaked in the late 1980s and have been coming down since then.
  • 02 - Edinburgh - Edinburgh city, the capital of Scotland, was surveyed as a population unit for Scottish MONICA in 1986, but was withdrawn from MONICA shortly afterwards. Edinburgh is more middle-class and less industrial than Glasgow and has lower rates of disease. (The east-west gradient in disease rates is a feature of Scotland: Glasgow is approximately 80 km west of Edinburgh.) Despite Edinburgh's high population of civil servants and lawyers and administrators, and its city-centre, world renowned among tourists, there are numerous deprived housing schemes in the periphery, so the population is mixed.
  • 03 - Rest of Scotland - The first MONICA surveys in Glasgow and Edinburgh in 1986 were incorporated within a nationwide survey involving 25 populations across Scotland begun in October 1984 and continuing to the end of 1987. Scottish Heart Health study surveys differed from MONICA surveys only in that ages were limited to 40-59 and numbers recruited were smaller in each population than in the two Scottish MONICA centres of Edinburgh and north Glasgow. Data items (which included a large number of local options) and survey procedures were otherwise identical. (For Scottish purposes data from these three MORGAM reporting units are usually combined into the Scottish Heart Health Extended Cohort (SHHEC).
• RUAs:
  • UNK-GLAa - consisting of RU 01
  • UNK-EDIa - consisting of RU 02
  • UNK-SHHa - consisting of RU 03
• Cohorts:
  • Cohorts 01: Respondents of representative samples of men and women of the RUs. In RUs 01 and 02 the baseline survey covered age group 25-64 years and it was carried out in 1986. In RU 03 the survey covered age group 40-59 and it was carried out in 1984-87.
  • Cohort 02 - Respondents of a representative sample of 25-74 years old men and women of RU 01. The baseline survey was carried out in 1992.
  • Cohort 03 - Respondents of a representative sample of 25-64 years old men and women of RU 01. The baseline survey was carried out in 1995.
  • Cohort 21 - Respondents of a representative sample of 25-64 years old men and women of RU 01. The baseline survey was carried out in 1989.

Cohort recruitment and the baseline examination

The cohorts were formed by the respondents of representative sample surveys of the RUs. In the first stage of the sampling, a random sample of general practitioners (GP) was selected using the List of all GPs held by the Primary Care Administration of the Area Health Board as the sampling frame. In the second stage, the lists of persons registered with the selected GPs were used as sampling frames. From each GP's list a sample of size proportional to the number of persons in the list was selected. The second stage sampling was stratified by sex and 10-year age group. In RUs 01 and 02 the baseline examinations were carried out as a part of the WHO MONICA Project, although RU 02 was later withdrawn from MONICA for a reason not related to the survey. In RU 03, the survey was carried out in the framework of the Scottish Heart Health Study, which also used the MONICA methods. The WHO MONICA Project and the Scottish Heart Health Study overlap in age group 40-59 of RUs 01 and 02.

The size of the MORGAM cohorts and the response rates of the population surveys from which the cohorts were derived are:

tbc: to be calculated
* For definitions of response rates A and B, and the reason for their large difference, see [1].

DNA is not available for these cohorts but there is an ongoing programme of attempting to extract it from stored serum which is available for most participants.

End-points followed up:

• Deaths: yes
• Non-fatal acute MI: yes
• Unstable angina pectoris: yes
• Silent MI: no
• Cardiac revascularization: yes
• Stable angina pectoris: yes, for those with a code in hospital discharge records
• Non-fatal stroke events: yes
• Thromboembolic events: no, but the MPC could extract these from hospital discharge records

End of follow-up period:

• Fatal events: 31 December 2005
• Non-fatal events: 31 December 2005

Follow-up procedures

Sources of data

• National Health Service (NHS) Register: On recruitment participants gave consent to follow-up through their medical records and relevant identifiers were sent to the NHS Register in Edinburgh which succeeded in "flagging" for mortality follow-up almost all of those who gave consent. The Register subsequently matched incoming death certificates with those flagged for the above surveys [2]. The coverage of the register is about 99% for those who stay in Scotland and there is also notification of some deaths from England and Northern Ireland. There is some loss through emigration outside the United Kingdom, from people working, or holidaying abroad. Death abroad may result in notification of death in some cases but without the cause. The validity of the death certificate diagnoses will vary with the age, sex and the time-period being considered. It is complicated where there are multiple causes of death listed. Unlike the medico-legal authorities in England at that time, there was reluctance to order post-mortem examination by the Scottish Procurators-Fiscal, although extensive histories may be recorded from relatives and eye-witnesses in cases of sudden death, and medical and drug histories are taken from the general practitioner.
• Scottish Record Linkage System: All hospital discharge records since 1981 in Scotland are linked by personal identifiers through the Scottish Record Linkage scheme operated by NHS Information Services, Scotland. The Scottish National record-linkage scheme is one of the oldest and most comprehensive in the world [3-5]. The system was also used to notify deaths independent of the NHS Register, using data which originated from the Register, but a different method of record linkage (as for morbidity follow-up), involving probability matching from a number of data items, to obtain a score. Hospital discharge data were used to identify all coronary diagnoses, all stroke and cerebrovascular disease diagnoses, and also all coronary artery revascularizations from the surgical codes.

Procedures

Mortality follow-up: All members of the cohorts were flagged on the NHS Register with their permission for mortality after baseline screening using the name, maiden name, date of birth, address and area of residence as identifiers. Flagging was almost complete (98% - 99%). After death, copies of death certificates are sent by the NHS Register to the MPC with the codes for the underlying cause of death and all the other causes coded. Time delay averages a few weeks, but in exceptional cases (murder, court cases) it can be a year. Batches of deaths certificates are received monthly from the NHS register. From previous experience with the MONICA Project, it was decided from the start of MORGAM not to investigate sudden deaths outside hospital beyond the information provided by the death certificate, which included the cause of death attributed by the attending doctor or the medico-legal authority and whether there was a post-mortem examination.

Follow-up for coronary and stroke endpoints was done by NHS Information Services in Edinburgh using the Scottish Record Linkage system which notified details of all hospitalizations, including up to six discharge diagnoses and four surgical procedures for each hospital episode, including transfers within and between hospitals. This same system was used for the follow-up of myocardial infarction, hospitalization for unstable angina pectoris, or stable angina pectoris and hospital episodes involving coronary artery disease revascularization.

Unlike the notification of deaths by the NHS Register which was done monthly, morbidity follow-up was done on request when the funding and opportunity arose. This was done first to the end of 1993 [5], then to 31st March 1997, then 31st December 2004 [6] and most recently to the 31st December 2005 [7]. The last two updates included all the previous data back to January 1981.

For the earlier 8 and 12-year follow-up case records were obtained to "validate" coronary events. With the lapse of time this became increasingly difficult: key items could be missing with records incomplete with patients and part of their records moving from place to place. The numbers involved in the follow-up to the end of 2004 and 2005 were so large that it was decided to use hospital discharge diagnoses alone and not to attempt further validation. The same procedure was then followed for all hospital episodes retrospectively including the earlier data. Cross-comparison of the two methods remains to be done. The earlier follow-up was concerned mainly with myocardial infarction, distinguishing 'definite' from 'possible'. However, all cases were included in subsequent publications. The later analyses were more concerned with the development of any coronary or cerebrovascular disease diagnosis, including myocardial infarction and stroke as sub-categories [6,7].

Diagnostic procedures

At baseline:

• History of MI:
  • Documented: Data not available for MORGAM initially, but hospital discharge diagnoses back to January 1981 are now available at the MPC. These were obtained from the Scottish Record Linkage System.
  • Self-reported: An affirmative answer to an option "Heart attack (coronary thrombosis, myocardial infarction)" under the question "Have you ever been told by a doctor that you have, or have you had any of the following?" was considered as self-reported history of MI.
  • ECG: Category of Minnesota code 1 is 1-1 or 1-2 (except 1-2-6) was considered as ECG history of MI.
  • Rose questionnaire: As specified in item HISMI4 of MORGAM Form 21.
• History of cardiac revascularization:
  • Documented: Data not available for MORGAM initially, but hospital discharge diagnoses back to January 1981 are now available at the MPC. These were obtained from the Scottish Record Linkage System.
  • Self-reported: data available for Cohort 03, data not available for Cohorts 01, 02 and 21. An affirmative answer to an option "CABG (coronary artery bypass craft)" or to an option "coronary angioplasty" under the question "Have you ever been told by a doctor that you have, or have you had any of the following?" was considered as self-reported history of cardiac revascularization.
• History of angina pectoris:
  • Documented: Data not available for MORGAM initially, but hospital discharge diagnoses back to January 1981 are now available at the MPC. These were obtained from the Scottish Record Linkage System.
  • Rose questionnaire: As specified in item HISAP2 of MORGAM Form 21.
  • Self-reported: An affirmative answer to an option "Angina" under the question "Have you ever been told by a doctor that you have, or have you had any of the following?" was considered as self-reported history of angina pectoris.
• History of coronary heart disease, type unspecified: irrelevant
• History of stroke:
  • Documented: Data not available for MORGAM initially, but hospital discharge diagnoses back to January 1981 are now available at the MPC. These were obtained from the Scottish Record Linkage System.
  • Self-reported: An affirmative answer to an option "Stroke" under the question "Have you ever been told by a doctor that you have, or have you had any of the following?" was considered as self-reported history of stroke.
• History of Diabetes: An affirmative answer to an option "Diabetes" under the question "Have you ever been told by a doctor that you have, or have you had any of the following?" was considered as self-reported history of MI.

During follow-up:

• Deaths: Cause of death codes from death certificates were used for the relevant items of MORGAM Form 25. The coding of death certificate diagnoses by the NHS Register did not change from use of ICD-9 to ICD-10 until the year 2000. Validation was not attempted for such cases, but some judgement by HTP was needed where coronary or stroke diagnoses were listed in addition to the underlying cause of death where the latter could appear important in producing a potentially erroneous coronary diagnosis (e.g. disseminated cancer) or less important (e.g. raised cholesterol). Occasionally there was conflict between the information on deaths received from the NHS Central Register and Scottish Record Linkage System, resulting in one implying that the participant was still alive and the other not, or two deaths on different dates from different causes. After review priority was most frequently given to the NHS Register. A very small number of deaths (in other countries or where the body was found severely decomposed) had been coded to unknown underlying causes and judgement was used as to whether the circumstances made an 'unclassifiable' coronary event possible or too unlikely to code as such.
• Acute coronary events and other coronary diagnoses: Information was received from the Scottish Record Linkage System on all hospital episodes of any kind from January 1981. The ICD codes received for these changed from ICD-9 to ICD-10 for discharges from the 1st April 1996 onwards in most cases. Episodes before recruitment were separated from those after attendance at the baseline survey clinic. This left approximately 100 000 listings, many of them involving the same episodes when patients were transferred within one hospital or transferred to another hospital so there could be two entries for one day and many for each event. There were up to 6 diagnostic fields for each listing, as well as four or more fields for surgical procedures. Each event could be distinguished at its onset by whether it was an emergency admission, admission from a waiting list, or transfer within or between hospitals (although sometimes the latter did not have precedents as might be expected, suggesting the emergency admission was not listed). As previously stated, case notes were sought for review in events occurring before March 1997, but this was found to be of limited value, and the whole follow-up to the end of 2005 uses discharge diagnoses without case-note review. In ICD-9 it was found that the specific code for unstable angina (411) was very seldom used; in ICD-10 the specific four digit code (I200) accounted for about half the emergency admissions for angina, slightly more were coded as ICD-10 I209; smaller numbers of emergency admissions were coded to I24 and I25.

  Using the combination of diagnostic codes found in the Scottish Record Linkage System or the NHS Register and whether or not these were emergency admissions, allocations were made to the following categories:

  Scottish Record Linkage System code		MORGAM DGNCAT
  ICD-9	ICD-10
  410	I21, I22, I23	3
  411, 413, 414 (emergency admission)	I20, I24, I25 (emergency admission)	4
  412, 413, 414 (as first hospital manifestation)	I20, I24, I25 (as first hospital manifestation)	8 (EVTYPE=1)
  other	other	7
  Official underlying or other death certificate code		MORGAM DGNCAT
  410-414	I20-I25	3
  798 (CHD death is not ruled out by supplementary information from death certificate)	I46, R96, R98, R99 (CHD death is not ruled out by supplementary information from death certificate)	5
  798 (supplementary information from death certificate excludes CHD death or is not available)	I46, R96, R98, R99  (supplementary information from death certificate excludes CHD death or is not available)	7
  other	other	7

  Coronary events occurring within 28 days of each other were considered as one event. If several diagnoses were given during a 28-days period the most severe event was used for the classification. Where both death certificate and hospital discharge diagnoses were available within 28 days of each other and they disagreed as to whether there was a coronary diagnosis the case was included as a  coronary case.

  Coronary artery revascularization procedures were searched using specific surgical codes in use in the United Kingdom for coronary artery bypass graft (CABG) or percutaneous transluminal coronary angioplasty (PTCA). 

  Recurrent episodes during follow-up were not generally reported if they did not represent an increase in severity i.e. unstable angina or angina were not coded if they occurred after an acute myocardial infarction, fatal events were coded after non-fatal ones, and the first CABG and PTCA were each coded alone or if complicated by a coronary event.

• Stroke and other cerebrovascular diagnoses: The same general situation applied as for coronary events, except that there were no previous attempts to validate diagnoses using case note review. Successive stroke ICD codes within one hospital admission on consecutive listings appeared quite variable, suggesting that the specific codes were not really reliable, but the victim had had an acute stroke. Results of diagnostic tests were not available therefore. The transient ischaemic attacks were not reported to the MDC as not being a MORGAM endpoint.

  For events which were found in the Scottish Record Linkage System or the NHS Register, the diagnostic classification was done using the ICD-codes:

  Scottish Record Linkage System code or official underlying or other death certificate code		MORGAM DGNCAT
  ICD-9	ICD-10
  430, 431, 433, 434, 436, 438	I60, I61, I63, I64, I65, I66, I69	9
  other		4

  Stroke events occurring within 28 days of each other were considered as one event. Where both death certificate and hospital discharge diagnoses were available within 28 days of each other and they disagreed with each other as to whether there was a cerebrovascular diagnosis the case was included as an unclassifiable stroke case.

References

1. Wolf HK, Kuulasmaa K, Tolonen H, Ruokokoski E, for the WHO MONICA Project. Participation rates, quality of sampling frames and sampling fractions in the MONICA Surveys. (September 1998). Available from URL: http://www.ktl.fi/publications/monica/nonres/nonres.htm, URN:NBN:fi-fe19991076.
2. National Health Service Register. Available from http://www.gro-scotland.gov.uk/national-health-service-central-register/. (Accessed 2007 Apr 5).
3. Kendrick S, Clarke J. The Scottish Record Linkage System. Health Bull (Edinb). 1993;51:72-9.
4. Kendrick S. The Development of Record Linkage in Scotland: The Responsive Application of Probability Matching, Record Linkage Techniques - 1997. Proceedings of an International Workshop and Exposition, Arlington VA, March 20-22 1997 pp 319 - 332. Available from http:/www.fcsm.gov/working-papers/skendrick.pdf. (Accessed 2007Jan 8).
5. Tunstall-Pedoe H, Woodward M, Tavendale R, A' Brook R, McCluskey M-K. 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.
6. Tunstall-Pedoe H, Woodward M, for the 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 Sep 15.
7. Woodward M, Brindle P, Tunstall-Pedoe H, for the SIGN group on risk estimation. 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. Epub 2006 Nov 7.

previous contents