MONICA Manual, Part IV, Section 2

	Event Registration Data CORE DATA TRANSFER FORMAT - STROKE EVENTS Form: 03 Version: 3          09.06.86

Instructions

The purpose of this transfer format is to provide an exact and common format for MONICA Collaborating Centres (MCCs) to transfer the Core Data of stroke events to the MONICA Data Centre (MDC). The data should be sent on magnetic tapes, not on paper forms. This format therefore corresponds with the layout of the information on magnetic tape and need not correspond with the format of locally used paper forms, although the information available on the latter has to be convertible to the former. Instructions for data transfer on magnetic tapes are given in Part V, Section 1: Data Transfer to the MONICA Data Centre. To avoid errors, special attention should be paid in extracting the data required on this form from the local data set.

Data Transfer to MDC

Data in the format of the CORE DATA TRANSFER FORMAT - STROKE EVENTS should be submitted to the MDC for the following events:

• every suspect event classified into the diagnostic categories 1,2,5 or 9 (i.e. excluding events classified as "no stroke").
• those suspect events with diagnostic category 4 and with one of the three clinical diagnoses (items CLIND of the CORE DATA TRANSFER FORMAT - STROKE EVENTS) having ICD 8 OR 9 codes 430 - 434 and 436 (or whichever codes or diagnoses are being used for definitive case finding).

Data for every suspect event which was issued a Stroke Serial Number should be submitted to the MDC in the format of the STROKE SERIAL NUMBER INVENTORY FORMAT. These data are used for quality control purposes, to check that the MDC has received all the data it should have received.

Both the Core Data and the Serial Number inventory data should be submitted to the MDC on magnetic tapes according to the schedule and instructions of Part V, Section 1: Data Transfer to the MONICA Data Centre.

Quality Assurance Procedures

See Part IV, Section 3, where these matters are discussed more fully.

Local Data Checking

The MCC should have procedures to ensure that the data collection process is complete and accurate. This can be accomplished by following the procedures listed below.

Processing Check-List

Keep a check-list to follow the status of data collection for every suspect event. The contents of the check-list may vary between the MCCs, but should contain items such as the following:

• stroke serial number
• name and address
• date of onset
• hospital
• date serial number assigned (or registration began)
• examinations done
• an indication that examination results were received
• final diagnostic category
• an indication that data were submitted for keying
• an indication that shipment was made to the MCC

The check-list may be used to generate the data for the STROKE SERIAL NUMBER INVENTORY FORMAT described in Subsection 5.

Keying of Data Forms

Before submitting the forms for being keyed

• check that all parts of the form (for examination results, diagnosis, autopsy, etc.) are together
• check the forms visually for completeness and logical flow

It is best to have all data verified or keyed twice to reduce keying errors as much as possible. The method of verification will depend on the MCC's data entry system. Cards can be verified and most data entry programs have the ability to double-enter the data. As an alternative, two identical data files can be independently created and compared.

Check Serial Number

After the data have been keyed and verified, the serial numbers should be checked again by comparing keyed values with the original form. The use of a check digit is strongly recommended since a computerized check of the serial number can then be performed.

Other Checks

Before the data are appended to the local master file, checks and other quality control routines similar to those described in Part V, Section 1: Data Transfer to the MONICA Data Centre should be performed. Checks for duplicate records, missing values, etc., are particularly important at this stage.

When checking for the correctness of the data, all suspect data items should be checked against the original documents on which data were collected. The erroneous values should be corrected. On the other hand, suspect values which cannot be shown to be erroneous should not be changed. Instead, such values, even though very unusual, should be submitted to the MDC.

File the forms for future reference.

General Instructions

The ITEM NAME on the form is a computer variable name used for the item by the MDC.

The COLUMNS indicate the columns on which the value of the item is to appear in the record on the magnetic tape. Data for different events must be written on different records.

Blank fields are not allowed in the magnetic tape record.

Instructions for making corrections to data that has already been sent to the MDC are given in Part V, Section 1: Data Transfer to the MONICA Data Centre.

Please contact the MDC for instructions if you cannot provide information as specified in this document, and contact the Quality Control Centre for Event Registration in Dundee if you have any problems with the interpretation of the coding for any specific item.

Specific instructions for each item

These instructions should be followed carefully when creating a computer file for transfer from the MONICA Collaborating Centre (MCC) to the MONICA Data Centre (MDC). Please ensure that the instructions are for the version of the Core Data Transfer Form being used. Specific instructions are listed by item below:

Item 1

Enter number 3 to indicate the CORE DATA TRANSFER FORMAT - STROKE EVENTS.

Item 2

Enter the version number that is being used by your MONICA Collaborating Centre. If the version is not 3, these instructions do not correspond to the form you are using. Check that you are using the valid version of the form.

Item 3

Enter the official MONICA Collaborating Centre code number as it appears in Part I, Appendix 2: MONICA Collaborating Centres and Reporting Units. If your centre is not listed, or is erroneously listed in this appendix, contact the MDC for instructions.

Item 4

Enter the official MONICA Reporting Unit code number as it appears in Part I, Appendix 2. Even if your centre has only one Reporting Unit please enter the appropriate code number here.

Item 5

Enter here the 7-digit serial number of the stroke event for which the record has been completed. Each serial number issued must be unique within each MONICA Reporting Unit and with each register (i.e. stroke register and coronary register). Different MONICA Reporting Units and different registers may use the same serial numbers. However, the serial numbers within a population and register must be different for each calendar year of event registration. Check that the serial number entered is correct.

Item 6

Enter the date that this specific stroke event came to the attention of the MONICA Collaborating Centre and was allocated a serial number. The point in the registration procedure at which this occurs may vary from centre to centre. It should be defined within each centre and not changed, so that the delay time from DONSET to DREG can be monitored.

The first two columns are for the day; code 01-31 for specific dates, or enter 99 if the day is not known. The next two columns are for the month; code 01-12, or enter 99 if the month is not known. The last two columns are for the year, but only the last two digits are entered and the first two digits (19) re not entered. If the year of registration is not known, which may happen in attempting to record this new item retrospectively for old registration, enter 99 in the last two columns.

Item 7

All cases that are registered must be allocated to the male or female sex. If it is not possible to decide on the sex of a case, that case cannot be registered in the study as it cannot be allocated to an age and sex group for analysis.

Item 8

Enter the exact date of birth. The first two columns are for the day: code 01-31 for specific dates, or enter 99 if the day is not known. The next two columns are for the month; code 01-12, or enter 99 if the month is not known. The last two columns are for the year, but only the last two digits are entered and the first two digits (19) re not entered. If the year of birth is not known, use an estimate of the age to derive a year of birth and cost that. Records with the year of birth entered as 99 (not known) are not acceptable as they cannot be allocated to an age group for analysis.

Item 9

Enter the exact date of onset of the stroke. Onset is the onset of acute symptoms, whatever they were, that continued up to the appearance of the stroke (or, in their absence, the fatal collapse of the person). Where the patient has vague or intermittent symptoms with complete recovery (such as a transient ischaemic attack) followed by prolonged more severe symptoms, the time of onset will be the time of onset of the prolonged symptoms. Where several attacks of symptoms occur and each of these attacks could qualify as a possible onset, the onset could be timed from the onset of the first stroke. Where there is a vague onset and a slow increase in severity of symptoms over several days (but without complete recovery), that is a crescendo onset, then the onset should be dated to the beginning of the episode.

** (Note that this means the rules are slightly different from those for a crescendo onset for coronary events).**

In cases of sudden death presenting with a fatal collapse, the time and the date of onset will be the same as that for death. Where a person is found dead, use circumstantial evidence to estimate the most likely time (and therefore date as this may be overnight) of onset and death. If the person has not been seen for some time, and is found dead, with no circumstantial evidence of when the fatal event occurred, then estimate the time and date of onset (and of death) as the intermediate between the time the person was last seen and the time the body was found. (MCCs should establish quality assurance procedures for standardizing the coding of similar problems to this, such as keeping a file of problem cases and how they were coded, and having a review committee.)

If the date of onset is known, enter 01-31 in the first two columns and 01-12 in the third and fourth columns for day and month of onset, and the year (19--) in the fifth and sixth columns. If the exact date of onset is insufficiently known, enter 99 in the days column. It should always be possible to allocate a month and year of onset so these columns should always be specifically coded.

Item 10

Code 1 (in hospital) if a medical consultation occurred after the onset of the event and a decision was made to refer or admit the patient as an inpatient and the patient reached hospital alive. Include also patients whose onset of symptoms occurred when they already in hospital.

Code 2 (nursing home) if there was a medical decision to manage the patient in a nursing home and the patient arrived there alive, or alternatively if he or she was already a patient there. Local definitions should be used in each MCC to decide what is a nursing home and what is a hospital.

Code 3 (at home by a doctor) if there was a medical consultation in the event and a decision by the doctor to manage the patient resting at home.

Code 4 (medically unattended) where no decision on management could be taken because the patient was already dead when first seen by a doctor.

Code 5 (other medical consultation without bed rest, in hospital or at home) where the patient was seen in the event by a doctor but, because death occurred too rapidly or because the event was not diagnosed at the time or because recognition occurred some time after onset, no decision was made to admit the patient for bed rest or to order rest at home, or the patient died in transit.

Code 9 (insufficient data) if information is inadequate to allocate another code.

Item 11

The 28 days' survival period starts from the day when the symptoms started. The 28-day period is calculated from the date of onset symptoms (DONSET). The day of onset is day zero.

In cases where one or more strokes occur during the 28-day period, the counting of the 28-day survival period must start from the date of onset of the first event.

Code 1 (yes) if the survival period is 28 days, that is if the person is alive at the midnight between the 27th and the 28th day after onset. Thus the event is fatal if DDEATH - DONSET < 28, and the event is non-fatal if the difference is 28 or more.

Code 2 (no) if the survival period is 27 days or less.

Code 9 (insufficient data) in cases where survival at 28 days is not known. This will occur if the patient is lost to follow-up or there is difficulty in defining the date of onset (DONSET), or the date of death (DDEATH) or both. To avoid the latter problem, measure survival from the date of start of hospital or medical care in cases with an indistinct onset. Try to record an estimate of survival time in all fatal cases, if it is possible.

Item 12

Code the type of stroke by using a 3- or 4-digit ICD code. The type of stroke should be coded in association with the diagnostic category, item 13.

ICD-9: 430, 431, 433, 434, 436. ICD-8: 430, 431, 432, 433, 434, 436

In cases where the ICD code has only three digits, enter the code in the first three columns, and enter 9 in the last column. If DIACAT = 4, then enter 999 here.

The type of stroke is defined in Section 2.2, and the allocation should be made by the stroke survey team on the basis of these diagnostic criteria. The type may or may not correspond with the clinical diagnosis or diagnoses in Items 16-18.

Item 13

Code 1 (definite stroke) if there were rapidly developed clinical signs of focal (or global*) 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. leukaemia, polycythaemia vera), brain tumour or brain metastases. Secondary stroke caused by trauma should also be excluded.

*Global: this applies to patients with subarachnoid haemorrhage or deep coma, but excluding coma of systemic vascular origin such as shock, Stokes-Adams syndrome, or hypertensive encephalopathy.

Do not code 1 for transient cerebral ischaemia (that is focal neurological signs or symptoms which recover completely within 24 hours) or if the duration of neurological deficit is insufficiently specified. The type of stroke is coded in Item 12 (TYPE), and the clinical diagnoses are coded in Items 16-18 (CLIND1, CLIND2, CLIND3). Do not code 1 if there is a definite stroke that has occurred within 28 days after a definite coronary event.

Code 4 (not stroke) if the registered event does not satisfy the study criteria of definite stroke given above, and another origin than a cerebrovascular one for symptoms emerges. The clinical diagnoses of these cases will be given in the items CLIND1, CLIND2, CLIND3 (see Subsection 2.2.2).

Code 5 (Optional for MCCs - stroke associated with a coronary event) if a definite stroke has occurred within 28 days of a definite coronary event in the same individual. This code is for MCCs who wish to register all definite coronary events and all definite stroke events but wish to identify definite strokes occurring within 28 days of a definite coronary event.

Code 9 (insufficient data) if the data concerning origin or duration of symptoms is insufficient to code 1 or 4. These will lnclude cases in which there is no diagnosis other than stroke to explain the illness but the information is inadequate to make it a definite stroke, or the duration is insufficiently defined to determine whether it was more or less than 24 hours.

Item 14

Code 1 (yes) if the relevant examination was made between the onset of acute symptoms and 28 days and there are records to prove it.

Code 2 (no) if the relevant examination was not made between the onset of acute symptoms and 28 days. Evidence for this would be that: there are records stating that the examination was not done, or there are records of examinations that were done and no record of the examination concerned, or the patient died too repidly for any of them to be done, or was managed at home or in an institution where it would not be possible to do the examination.

Code 9 (insufficient data) if no records of the examination done are available and it is therefore not possible to state whether the examination was done or not.

Item 15

Code 1 for a possible iatrogenic event if the stroke event occurred 0.1, or 2 days after any of the following procedures:

1. surgery, requiring anaesthesia, or
2. intravascular manipulations, such as angiography, angioplasty, or intra- arterial thrombolytic treatment. "Possible iatrogenic event" implies a temporal relationship to the procedure but not necessarily a causal one. Calculate the number of days by subtracting the date of procedure from the date of onset of the event. Do not code 1 for othe rmedical or surgical procedures that may precede the onset of the stroke event, however suspicious the circumstances.

Code 2 if no such medical or surgical procedure preceded the stroke event, or if it occurred 3 or more days beforehand.

Code 9 if data are insufficient to decide whether the event followed a medical or surgical procedure.

Items 16-18

CLIND1, CLIND2, CLIND3 Clinical or death certificate diagnoses

CLIND1 Main clinical condition or direct cause of death

CLIND2 Other clinical condition or intervening antecedent cause of death

CLIND3 Other clinical condition or underlying cause of death

These diagnoses are normally derived directly from the death certificate diagnoses, or from the hospital discharge diagnoses. MONICA coders should not normally have to made diagnoses themselves and should not add extra diagnoses, from reading the case notes, to those made by the clinicians, nor should they change the clinical diagnoses. Only in MCCs using "hot pursuit" and where discharge diagnoses are not available should the MONICA team attempt to supply the clinical diagnosis.

Code these items according to the International Classification of Diseases, Injuries and Causes of Death (ICD) 9th revision, or 8th revision in those few MCCs where it is still in use. Do not use Procedures of Operations codes as the codes for these are often indistinguishable for entirely different diseases. Code the full 4-digit code, or if there is only a 3-digit code, left- justify the coding and insert a 9 in the fourth (final) column. If there are fewer than three disease codes, enter 9999 in the unused CLIND fields.

For fatal events where both clinical diagnoses (e.g. hospital discharge diagnosis) and death certificate diagnoses are available, the death certificate diagnosis should take precedence.

In the case of death certificate diagnoses, the coding should follow the sequence shown in option a) below if this is at all possible. This will enable the MCC to compare the MONICA registration diagnoses with the routinely reported mortality figure, based on the underlying causes of death. Some MCCs may have to follow other and less satisfactory rules for coding different diagnoses. Each MCC should establish its own rules and follow them consistently while coding the results for deaths in any calendar year, and should not change them without considering the effects of such a change on any subsequent analysis of trends.

a) Where the underlying cause of death has been coded, or can be coded according to national coding practice, code this underlying cause of death first in CLIND1. Code the other diagnoses on the death certificate in CLIND2 and CLIND3. If there are four or more diagnoses on the death certificate, and the diagnosis that lead to registration appears as the fourth or an associated cause of death, then code that diagnosis in CLIND3 to make sure that it is not left off the record.

Where the MCC has a copy of the death certificate, either without the ICD codes coded according to national coding practice, or the ICD codes are present, but national coding routines cannot be followed to find what the underlying cause of death would be, then the alternatives in order of preference are:

1. Code the diagnoses according to the sequence in which they appear on the death certificate, or
2. Code first the diagnosis that led to registration of the event.

b) For non-fatal events, record the clinical diagnoses here. Ordinarily these will be the major hospital discharge codes, but if none is available, such as when a patient managed at home, or discharge coding has not been done, the MCC should assign up to three ICD codes to the clinical diagnoses of the attending doctors. In general, code the clinical diagnoses in the sequence in which they appear on the discharge or other documents. However, make sure that the major reason for hospital admission appears first. If there are more than three diagnoses recorded, and the diagnosis that led to registration is not among the first three, then ensure that this diagnosis is coded in CLIND3.

Item 19

MCCs should establish procedures for ensuring that the information, on whether persons whose strokes are being registered have a history of previous stroke or not, is as good as possible; but they should be free from the possibility of a biased trend over the ten year period of registration. This item is important in deciding on the ratio of event rates for first versus recurrent events, and special analyses will be made of trends in this. A previous history of stroke may also be important in allocating MONICA diagnostic categories to fatal cases. MCCs should try to keep the proportion of stroke events with this item coded 9 to (insufficient data) to a minimum (with the possible exception of non-fatal cases with DIACAT = 4 (none) where it is less important). Special efforts may be needed in fatal cases and other cases where a medical history may not be provided with the record of the current event. However, in order to avoid biases in trends, MONICA registration records held by the MCC must not be used to obtain this information, as there will be no such records at the beginning of the study and 10 years of records at the end. Previous history should be decided on the basis of routine records used consistently throughout the study such as:

1. records present in th hospital treating the current attack
2. a defined number of sources in fatal cases.

The date of onset of any previous stroke should be 28 days before the event being registered, or the latter will be part of the earlier event. If the only available sources of information about previous stroke events are the medical records referring to the current event, then the previous medical history is "undocumented" and the only codes possible are 2 and 4. If records of previous medical care or admissions to hospital are filed with the records of the current stroke or are routinely reviewed then previous strokes may be "documented" and the codes 1 and 3 are possible.

Code 1 (yes, previous record reviewed) if medical records seen for the current event include records for previous event(s) and one or more of these satisfied the MONICA criteria for a definite stroke, or there is a suggestion of a stroke but data are inadequate. Note that MCC registration records for the event must not be used for this purpose.

Do not code this item if there is a history of transient ischaemic attacks only (<24 hours).

Code 2 (yes, previous records not reviewed) if there are no previous medical records seen with those for the current event, but recorded in the medical history of this event is a statement that the patient had already had a stroke more than 28 days before the onset of the current event. Do not code this item if the history records transient ischaemic attacks only.

Code 3 (no, documented) when continuous or complete medical records are available with those for the current event or routinely reviewed and these show that there is no history of a stroke before the current episode, or that transient ischaemic attacks only had occurred.

Code 4 (no, undocumented) if there are no continuous medical records seen with those for the current episode, or any previous record of a stroke and the current history states or implies that there is no previous history of a stroke. These criteria are met by any of the following statements in the current stroke or medico-legal records:

1. that there is no previous history of stroke, or cerebrovascular disease
2. that the patient has never had symptoms like those of the present event
3. that the patient has had no serious illnesses, or has never been in hospital, or has been previously healthy
4. that the patient has had certain illnesses or operations without any mention of stroke among them.

Code 9 (insufficient data) if there are no data in the record of the current event which refer positively or negatively to a previous medical history, and there are no other records available which are usable for this purpose.

Items 20-26 are for FATAL CASES ONLY. If the person did not die within 28 days (that is, if Item 11 has been coded 1 or 9), complete items 23-29 of this record with codes 8, 88, 8888.

Item 20

Enter the exact date of death using the same format as for DREG. If the patient did not die during the 28-day survival period, code this item 888888. Note that, in legal terms, a person is not dead until this has been certified by a doctor. The data and time when this happens may be a long time after the person became lifeless, either because the death was not witnessed, or because it took some time for a doctor to be called, or for the body to be taken to hospital. Do not accept the time of death as recorded on medico-legal documents, but estimate the time and date at which the fatal collapse occurred. Note also that unsuccessful attempts at resuscitation do not postpone death for this purpose. Death should be considered as the time and date at which the irreversible cardiac arrest first occurred, even if the medical attendants or bystanders initiated and continued resuscitation attempts in the hope that it was reversible. Where a person is found dead, use circumstantial evidence to estimate the most likely time (and therefore date as this may be overnight) of death. If the person has not been seen for some time, and is found dead, with no circumstantial evidence of when the fatal event occurred, then estimate the time and date of death as half-way between the time the person was last seen and the time the body was found.

Item 21

Enter the number of calendar days, calculated from the date of onset of the event (DONSET). The date of onset is day zero and death must occur in 27 days or less or this item should be coded as 88 (not relevant).

Item 22

Code 1 if a routine necropsy was performed for medical purposes only (for example, in a teaching hospital, or at the request of the attending doctor).

Code 2 if the necropsy was performed at the request of, or to fulfil the requirements of the medico-legal authority (for example, police, coroner).

Code 4 if no necropsy was performed.

Code 8 if the patient did not die within 28 days.

Code 9 if it is not known whether a patient was alive or dead at 28 days, or whether a necropsy was performed or not.

Items 23-25

NECD1, NECD2, NECD3 Necropsy findings

Item 23

Item 24

Item 25

Code as many diagnoses as you can up to three from the necropsy report. If any of the following findings are present, code them in the order of priority.

Note: these codes are specific to the MONICA Project. They are based on the codes used for necropsy findings in the WHO myocardial infarction and stroke registers of the 1970s. The first three digits are those of the equivalent ICD code in the 8th or 9th revision, and the fourth digit shows end-organ change. For this reason the fourth digits do not have the same meaning as those now used with ICD-9, and these codes are specific for NECD1-3.

Code any other competing cause of death using the ICD code.

If only one diagnosis is given, code it in NECD1, and 9999 in NECD2 and in NECD3. If two diagnoses are given, code them in NECD1 and NECD2, with 9999 in NECD3. If the necropsy was done but it was not possible to review the necropsy report, code 9999 in NECD1, NECD2, and NECD3. If the patient did not die or if no necropsy was performed although the patient died during the 28 days' survival period, code 8888 in NECD1, NECD2 and NECD3.

Item 26

1 = ICD Eighth Revision
2 = ICD Ninth Revision