November 1990
This section provides the equivalent of a manual of operations for the collection and data processing for this component of the MONICA Project.
The MONICA core study is concerned with coronary and stroke events (and not with persons) and with two characteristics of the events, apart from their diagnostic category, which are whether they are (a) first or recurrent, and (b) fatal or non- fatal. Each episode must have a defined duration. In the MONICA core study a period of 28 days is used to establish the case-fatality and to distinguish two events from each other.
The Protocol (WHO/MNC/82.1 Rev. 1) was the first document describing the principles of event registration. Modifications and clarifications have been agreed by the Council of Principal Investigators and these were included in MONICA MEMOs 30, 40, 41, 49 and 57. This Manual contains the latest information which should always be used.
It is important that event registration procedures are similar throughout the entire study period in each MCC or in each Reporting Unit. There are some differences causing systematic bias between MCC's which cannot be avoided and, therefore, the data from different centres will not be fully comparable. If this systematic bias within each MCC is kept unchanged, it will be possible to compare trends in incidence and case-fatality both between and within centres because the changes over time will then be independent of the systematic bias. The most important source of systematic bias in estimating incidence is related to the coverage of event registration. The aim of event registration is to get as near to 100% coverage as possible, i.e. to ascertain the maximum number of coronary events and stroke events in the study population.
Certain factors may influence the detection of events and calculation of the event rates such as instability of the population, seasonal migration and permanent migration in and out of the study area. Populations for event registration must be the same as the defined Reporting Units in each MCC. The denominator must be the same for event registration and for the population surveys. The population size estimates for each Reporting Unit must be reported annually to the MDC from the MCC using the Population Demographics Reporting Form.
For an event to qualify for inclusion in the MONICA study:
The suspect events are collected from death certificates, from hospital medical records and from community health services. These suspect events are examined for eligibility into the Coronary and Stroke register. Event finding may be:
Some sources of event finding may provide information a long time only after the actual event occurred. Typical examples of such situations are: National Hospital Discharge Registers, Registers for Social Security Purposes and Death Certificate Registers. One to two years delay may occur. There is no preset time limit for including such events in the coronary or stroke register if the criteria are otherwise met (e.g. they must have been clinically recognised within 28 days of onset even if there was a delay in registering them). However, the Data Centre may have to impose such a time limit at the end of the study, in order to bring it to an end. These late events can be identified in the database by using the date of event (DONSET) and the date of registration (DREG) and appropriate methods of data analysis can be applied. Another problem is that of cases identified reasonably early for whom case records remain unavailable. Again local policy and discussion with the Data Centre will determine the optimum policy for balancing complete registration with completing a year's registrations reasonably quickly.
In each MONICA Reporting Unit an information gathering system will need to be instituted (1) to gain information on likely events, and (2) to obtain all available diagnostic and other information.
The sources used must depend on the structure of the local medical and medico-legal services.
The following sources may be used to identify cases:
The first two sources are obligatory but the others will depend on the local situation.
Once sources and ICD codes have been determined in the MCC or in the Reporting Unit it is important that methods are not changed over time without measuring the consequences, as a change of sources may cause a spurious change in event rates.
All death certificates of the residents of the MONICA Reporting Unit areas should be reviewed at least quarterly or semiannually.
Periodically during the 10-year period a random or stratified sample of hospital cases should be taken to search for changes in the pattern of cases. Where multiple field discharge data are available a different set of rules may be applied for main and subsidiary diagnoses.
While each MCC should attempt to find all the death certificate codes both as immediate, main and antecedent or underlying causes of death that lead to cases of coronary heart attack and stroke satisfying the MONICA criteria, at least the following causes of death should be considered during pilot studies:
| Cause of death | ICD code | ||
|---|---|---|---|
| Diabetes | 250 | ||
| Hyperlipidaemia | 272 | ||
| Obesity | 278 | ||
| Cardiovascular | |||
| Hypertensive | 401-405 | ||
| Ischaemic | 410-414 | ||
| Other | 420-429 | ||
| Stroke | 430-439 | ||
| Arteries, etc. | 440-447 | ||
| Symptoms, etc. | 797-799 | ||
It is therefore possible to exclude those deaths that are clearly due to trauma (e.g. contain a sequence of events including accident, homicide or suicide), chronic obstructive pulmonary disease, cancer, cirrhosis of the liver, or rheumatic heart disease without mention of atherosclerotic heart or vascular disease. However, if an atherosclerotic condition is mentioned in the sequence of events or if the cause of death is attributed to one of the conditions listed above, the cause of death should be validated. Where deaths are notified through medico-legal sources all non-violent sudden deaths, hospital "dead on arrival" and deaths from cardiovascular disease within 24 hours of onset should be monitored.
Validation is based on any available medical and medico-legal records and, if necessary (local option) interview of the decedent's next-of-kin or another informant. Medical records for the period within a minimum of 28 days of death should also be examined for information that may elucidate the circumstances leading to the death. The interview should establish the circumstances surrounding the decedent's death.
Deaths should be validated by applying the MONICA diagnostic criteria for coronary events (see Section 1, subsection 2.2). These are based on the symptom history, ECG's, cardiac enzymes, necropsy and death certificate data. In positive cases diagnostic findings should be recorded on the locally used MONICA Coronary Event Registration Forms. In those centres undertaking stroke registration diagnostic criteria should be applied for stroke using similar sources of data. In positive cases diagnostic findings should be recorded on the locally used MONICA Stroke Event Registration Forms.
The list of death certificate codes produced above was for use during the development of the MONICA Registration procedures in each MONICA Collaborating Centre. Where, after pilot studies, a more limited list is being monitored, because certain codes were found to be unproductive, then the full list should be used periodically to see that the situation has not changed. All deaths with the underlying cause coded to 410-414 must be registered even if they are found not to be a coronary case on examination. Stroke registers should register all cases where the ICD code implies acute stroke (ICD 430-4 and 436) even if they are found not to be strokes alone.
It is suggested that in the early phases of data collection each MCC collect data on cases drawn from a broad range of discharge codes which might include stroke assigned to another category. Examples of these belong to the categories (ICD-9) 430-438 but might also include hemiplegia (342), "certain symptoms referable to the nervous system and special senses" (781). Once it has been established that an insignificant proportion of cases discharged without stroke diagnoses are in fact not stroke, case validation can proceed on a narrower scale (e.g. 430-436 or 438). Documentation of the adequacy of such a narrower scale should be repeated every two to five years to make sure that there has been no change in hospital labelling practice over time.
In communities where patients in the 25-64 age range with stroke are not always treated in hospital, routine searches outside will be cost effective, but even where hospital treatment is said to be the norm, this should be confirmed by questioning physicians in the community in the pilot phase. However, in those countries where stroke is treated at home, contact must be made with physicians and a protocol established to identify and collect data on stroke cases. This procedure could take several forms and might include reviewing charts on a semi- annual basis, providing each office with a log book and a poster reminding them to log all cases of suspected new stroke, or asking the office nurse or assistant to notify the project when a new stroke is identified.
Each coronary and stroke event is defined as lasting for 28 days (unless interrupted by death) so that any recurrence or exacerbation of the condition within that period is counted as part of the original episode whereas beyond that time it counts as a new (recurrent) episode. A special problem arises in those centres who are registering both coronary and stroke events in dealing with cases who satisfy criteria for both coronary and stroke events within one 28 day period, or in which the clinical diagnosis is of one condition, but the diagnostic evidence suggests that the other is more probable. Such events should be discussed by both registration teams.
Each MCC should lay down its own rules and should adhere to them so that there is no systematic bias over time. The number of events involved is small and should not have a big effect on registration numbers whatever method of handling the problem is used. The following rules are recommended; these should be acceptable to both those centres that are registering coronary events and strokes and those that are registering coronary events alone.
Cases satisfying criteria for definite myocardial infarction should be registered as such whether or not an apparent stroke occurs within 28 days before or afterwards.
Definite strokes should also be registered but if they occur in association with a definite myocardial infarction (within 28 days before or after), for stroke the diagnostic category 5 can be used, rather than 1. The event would be registered in both the coronary and stroke registers but could be removed from the latter by excluding category 5 if this was wanted.
Patients who die from a combined definite myocardial infarction and definite stroke within 28 days of onset of acute symptoms should be counted as fatal cases in both registers but again could be removed from the stroke register for certain analyses if necessary as they would be coded as category 5.
In cases where definite evidence for one of these two diseases is mixed with possible or inadequate data evidence for the other, the definite diagnosis should take precedence and the possible diagnosis should not lead to registration unless the secondary diagnosis appears as the underlying cause of death on the death certificate, or as a major clinical diagnosis when the case is notified. In these cases the patient would appear in the register of the definite disease as a Definite case and in the register of the unproven disease as a Not case, or false positive. In these cases it would be clear from the clinical diagnoses or the death certificate diagnoses on the record form what had happened to make it a false positive.
Cases with inadequate data which could be either coronary or stroke events should not be assigned to both registers as inadequate data but should be allocated the most probable diagnosis, regardless of what the death certificate diagnosis states. A case diagnosed on the death certificate as stroke which on investigation is more like a coronary death should therefore be coded as an inadequate data case in the coronary event register and as DIACAT 4 in the stroke register.
The diagnosis of stroke is based on clinical signs which should last more than 24 hours except in cases of sudden death or if the development of symptoms is interrupted by a surgical intervention. Therefore, it is important to pay great attention to both symptoms and duration. Data on subcategories of stroke will also be collected to give more information about different types of stroke. The classification into subcategories of stroke should strictly follow the criteria given under Subsection 2.2.4.
Every stroke event must have its apparent onset within the study period and more than 28 days from any preceding recorded stroke event. Special rules apply where suspected stroke and coronary events occur within 28 days of each other (see Subsection 1.2.4). The method of calculating the 28-day survival period in a stroke event is exactly the same as in coronary events (see Section 1, Subsection 2.1.2).
It is important to obtain the best available information about previous stroke in each event in order to separate first events from recurrent events. The data item "Previous history of stroke" is the only variable which allows this separation. The previous stroke episode should be more than 28 days previously, otherwise the present episode would count as a complication of the previous one and not as a recurrence.
Previous history should be decided on the basis of routine records used consistently throughout the study such as:
Previous registration data should NOT be used as this will introduce a bias over the duration of the study as attacks during an increasing number of previous years will have been registered. Centres with record linkage should use this as a local option and not use registration data for the reason given above.
The procedures for the collection of these data should be kept constant throughout the entire study period.
The diagnostic classification is similar to that used in the WHO Stroke Registers.
Stroke is defined as rapidly developed clinical signs of focal (or global1) disturbance of cerebral function lasting more than 24 hours (unless interrupted by surgery or death), with no apparent cause other than a vascular origin: it includes patients presenting clinical signs and symptoms suggestive of subarachnoid haemorrhage, intracerebral haemorrhage or cerebral ischaemic necrosis. It does not include transient cerebral ischaemia or stroke events in cases of blood disease (e.g. leukemia, polycythaemia vera), brain tumour or brain metastases. Secondary stroke caused by trauma should also be excluded.
Note1 : Global: this applies to patients with subarachnoid haemmorrhage or deep coma but excluding coma of systemic vascular origin such as shock, Stokes-Adams syndrome or hypertensive encephalopathy.
Definite focal signs:
Not acceptable as sole evidence of focal dysfunction
(Although strokes can present in this way, these signs are not specific and cannot therefore be accepted as definite evidence for stroke.)
Because each event lasts only 28 days the diagnostic criteria used must be from investigations or symptoms that refer to what happened within that 28 day period even though the results (such as necropsy) may be reported later. The definition of the 28-day survival period has been given earlier.
The most severe findings during the 28-day survival period should always be used to derive the diagnostic information. If there are recurrent episodes during the 28 day period the initial episode may not be associated with the same specific diagnostic information as becomes available later. The date of onset should not be changed for this reason but the most specific information should be used in deciding on the diagnostic category and type of stroke.
When a patient dies later than 28 days after onset, and cerebrovascular disease appears among the death certificate diagnoses, the following procedure is applied. First, check that death is not caused by a recurrent stroke that should be registered as a separate event. Then, if death is caused by sequelae of a stroke, and cerebrovascular disease is considered to be the underlying cause of death, register the initiating event (if registration has not been done previously) but be sure to code SURVIV = 1 (i.e. survival more than 28 days). In addition, make a new registration relating to the death and be sure to include stroke as one of CLIND1, CLIND2 or CLIND3, but now code DIACAT = 4 (not stroke), because the entry criteria are not fulfilled - the event occurred more than 28 days ago. In this way, discrepancies between the official death registers and the MONICA registrations are avoided. Deaths caused by the late sequelae of stroke and appearing in the official registers will be coded as false positive fatal events in the MONICA registrations but will appear as non-fatal stroke cases.
There are the following three categories:
(1) = Definite stroke
(4) = Not stroke
(5) = Definite stroke associated with Definite myocardial infarction
(9) = Insufficient data
Allocation of a diagnostic category must follow strictly the definition provided. A strict watch should be kept over competing causes of death. Note the discussion of competition and overlap between coronary and stroke event registration in Subsection 1.2.4. Local rules should be standardized and a record kept of difficult decisions.
Insufficient data should be mainly used for fatal cases, especially for cases of sudden death without necropsy.
All patients having insufficient supporting evidence of stroke, but for whom the diagnosis of stroke cannot be entirely excluded, should be classified as insufficient data, e.g. cases with no necropsy, no documented history of focal neurologic deficits and no other diagnosis. Living patients can be classified into this category if:
If the event is classified 1 (definite stroke), 5 (definite stroke associated with definite myocardial infarction) or 9 (insufficient data) one of the clinical or death certificate diagnosis codes of CLIND1, CLIND2 or CLIND3 should be between 430-438.
Subcategories of stroke are given on the form, Item TYPE using the ICD codes (8th or 9th revision).
Codes for specific subcategories of stroke should be used only if the diagnosis has been confirmed according to the following criteria:
Subarachnoid Haemorrhage ICD 8 or 9 430
Intracerebral haemorrhage ICD 8 or 9 431
Brain infarction due to occlusion of precerebral arteries ICD 9 433 ICD 8 432
Brain infarction due to cerebral thrombosis ICD 9 434 ICD 8 433
OR
Embolic brain infarction ICD 9 434 ICD 8 434
Remarks: If it is impossible to assign to a definite stroke event one of these sub-categories, the subcategory "Acute, but ill-defined cerebrovascular disease should be recorded (ICD code 436). Other codes: 435, 437, 438 should not be used in the data item "TYPE". If the clinical criteria for a stroke are fulfilled but a CT scan (of satisfactory technical quality) fails to reveal a brain lesion of recent origin, the patient has in all probability suffered an ischaemic stroke. In this case, type of stroke should be coded as 434 (infarction).
The stroke event registration data to be sent to the MDC are specified in the Core Data Transfer Format - Stroke Events, which is appended to this section.
The procedure for correcting errors in data which have already been submitted to the MDC are described in Part V, Section 1.3.3.
Every suspect event should be issued a serial number by the MCC as early as possible in the identification process, especially in MCCs which register retrospectively. The minimum requirement is that when the MCC begins to collect data on a suspect event, it is provided with a serial number. The serial number consists of seven digits. Each serial number must be unique within each MONICA Reporting Unit and within each register (i.e. stroke register and coronary register). Different MONICA Reporting Units and different registers may use the same serial numbers. The serial numbers within a Reporting Unit and register must be different for different events throughout the ten-year registration period.
Log-books of serial number histories: The MCCs must keep a log-book of the history of every Coronary Serial Number and Stroke Serial Number issued. The items of the Stroke Serial Number Inventory Format should be included in this log-book. The serial number histories are used to monitor the data flow and time lag in the event registration.
Correctness of the serial number: The serial number is one of the key items of an event record. The Form Identification, Collaborating Centre Code, Reporting Unit Code and Serial Number are all necessary in order to uniquely identify an event. Each MCC should have some kind of security system to help assure that the serial number is error free. One possibility is to have the last digit of the serial number a check digit which is used to control the correctness of the other digits at least until the individual record has been computerized.
A check digit needs to be calculated when the serial number is issued. It is useful only if this is done correctly. This can be secured by letting a computer generate the serial numbers with the check digits and print the numbers on stickers or labels which are then stuck on the log-books and data collection forms.
Serial number inventory data transfer: Serial number inventory data should be sent to the MDC as instructed in the Stroke Serial Number Inventory Format, which is appended to this section.