MANUAL ON MONITORING - iris.wpro.who.int · Manual on monitoring cardiovascular diseases 6 The WHO Collaborating Centre for Cardiovascular Epidemiology at the University of Newcastle,

Manual on monitoring cardiovascular diseases

1

MANUAL ON MONITORINGCARDIOVASCULAR DISEASES

Centre for Clinical Epidemiology & Biostatistics,WHO Collaborating Centre for Cardiovascular Epidemiology,

Faculty of Medicine and Health Sciences,The University of Newcastle, Australia

and

Noncommunicable Diseases including Mental Health FocusWorld Health Organization

Regional Office for the Western Pacific2001


2

© World Health Organization 2001

ISBN 92 9061 004 2

This document is issued by the World Health Organization -Regional Office for the Western Pacific (WHO/WPRO) for general

distribution. All rights are reserved. Subject to due acknowledgementto WHO/WPRO, this document may, however, be freely reviewed,

abstracted, reproduced or translated, in part or in whole, provided thatsuch is not done for or in conjunction with commercial purposes andprovided that, if it is intended to translate or reproduce the entire work,

or substantial portions thereof, prior application is made to thePublications Unit, WHO/WPRO, Manila, Philippines.


3

Table of contents

Introduction 5

Acknowledgements 7

1. Monitoring incidence of cardiovascular 11disease

1.1 Introduction 11

1.2 Establishing a sentinel site 16

1.3 Using routinely collected data 16

2. Indicators 25

2.1 Risk factor indicator 25

2.2 Indicators for disease patterns 27

Appendix 1: Establishing a sentinel site 31

Appendix 2: Sample size calculations 39


4


5

Introduction

Strengthening the assessment of the noncommunicable disease(NCD) burden and monitoring trends are critical for effectiveplanning, implementation and evaluation of programmes and services,as well as for supporting policy development and marketing.

Over the past years, the availability of NCD data at the countryand Regional level has gradually improved. Databases on the statusof NCD epidemiology and cancer services at the Regional level havebeen established, including a Regional Profile on CardiovascularDiseases, Diabetes, Associated Risk Factors and a Regional Databaseon Cancer in the Western Pacific Region published in Septemberand December 1999, respectively.

Another important challenge is to develop simplified methodsof epidemiological data collection, analysis and monitoring, particularlyon risk factors, that can be applied in developing countries with limitedresources and technical expertise. In response to the challenge, aWHO regional review group on NCD surveillance met in Melbourne,Australia, in November 1999. The group reviewed differentepidemiological instruments, including a proposal on simplified NCDsurveillance. It recommended a comprehensive set of simplifiedindicators and NCD risk factor survey and monitoring methods.


6

The WHO Collaborating Centre for Cardiovascular Epidemiologyat the University of Newcastle, Australia,has collaborated with WHOto prepare a manual on cardiovascular disease (CVD) monitoring.The Manual reviewed by the group has now been finalized fordistribution after wide consultation.

The aims of this manual are:

1. To introduce standardized methods for data collection,analysis and quality control in epidemiological assessmentand monitoring for CVD.

2. To develop appropriate indicators for measuring theefficiency and impact of programmes on prevention andcontrol of CVD.

This Manual is designed to be usable for monitoring morbidityat the country level where data collection systems may not be veryadvanced. Risk factor surveys will need to supplement thisinformation; they are the subject of other manuals and standards indevelopment.

The Manual can be a very useful tool to support CVDmonitoring. It will assist the simplification and standardization of CVDdata collection and analysis in the Region. The Manual can also serveas training material for health workers. Continuous efforts will beneeded to introduce these methods more widely into NCDepidemiological practice in the Region.

Noncommunicable Diseasesincluding Mental Health FocusWorld Health OrganizationRegional Office for the Western PacificManila, Philippines


7

Acknowledgements

The Western Pacific Regional Office of the World HealthOrganization acknowledges the valuable contribution made tothis manual by the following:

Contributors:

Professor Richard F HellerCentre for Clinical Epidemiology & BiostatisticsFaculty of Medicine and Health SciencesThe University of Newcastle, Australia

Dr Patrick McElduffFaculty of Medicine and Health SciencesThe University of Newcastle, Australia

Dr Han TieruWorld Health OrganizationRegional Office for the Western PacificManila, Philippines

Members of the Regional Review Group on NCDSurveillance convened by the WHO Regional Office for theWestern Pacific:

Dr Ruth BonitaNCD SurveillanceWorld Health OrganizationGeneva, Switzerland


8

Dr Ami ChandraDivisional Medical OfficerNorthern, Ministry of Health and Social WelfareSuva, Fiji

Dr Stephen ColagiuriPrince of Wales HospitalRandwick NSW, Australia

Dr Maximillian de CourtenInternational Diabetes InstituteVictoria, Australia

Dr Catherine D’EsteCentre for Clinical EpidemiologyUniversity of Newcastle, Australia

Dr Franklin C. DizaPhilippine Cancer Control ProgrammeDepartment of Health, Manila, Philippines

Dr Terence DwyerWHO Collaborating Centre for PopulationBased Cardiovascular Disease Prevention ProgrammesUniversity of Tasmania, Australia

Professor Valery FeiginClinical Trials Research UnitDepartment of Medicine, University of Auckland,New Zealand

Dr Fu HuaWorld Health OrganizationRegional Office for the Western PacificManila, Philippines

Dr Gauden GaleaWorld Health OrganizationWHO Office, Suva, Fiji

Dr Han TieruWorld Health OrganizationRegional Office for the Western PacificManila, Philippines


9

Professor Mark HarrisDepartment of Community MedicineUniversity of New South Wales, Australia

Professor Richard HellerFaculty of Medicine andHealth SciencesUniversity of Newcastle, Australia

Dr Bridget Hsu-HageFaculty of Medicine, Dentistry and HealthUniversity of Melbourne, Australia

Dr Robert HughesSecretariat of the Pacific CommunityNew Caledonia

Dr Lloyd IpaiPort Moresby General HospitalBoroko, Papua New Guinea

Dr Hilary KingWorld Health Organization, Geneva

Dr Kong LingzhiDivision of NCDs ControlMinistry of Health, Beijing, China

Dr Le Thi Thu HienDepartment of Preventive MedicineMinistry of Health, Hanoi, Viet Nam

Dr Patrick McElduffUniversity of Newcastle, Australia

Dr Paola PisaniInternational Agency for Research on CancerLyon, Cedex, France


10

Dr Jan PryorFiji School of MedicineSuva, Fiji

Dr Mukhar TsuremNational Institute of HealthUlaanbaatar, Mongolia

Dr Colin TukuitongaPacific Health Research CentreDepartment of Maori and Pacific HealthUniversity of Auckland, New Zealand

Professor Paul ZimmetWHO Collaborating Centre for the Epidemiology of Diabetesand Health Promotion for Control of NoncommunicableDiseases, International Diabetes Institute, Australia

Other Reviewers:

Aushra Shatchkute, M.D.Chronic Disease PreventionWorld Health OrganizationRegional Office for EuropeCopenhagen, Denmark

Dr Liming LeeBeijing Medical UniversityBeijing, China

Classification and Coding advice provided by:

Sue WalkerWHO Collaborating Centre for the Classificationof Diseases and National Centre for Classification in HealthSchool of Public HealthQueensland University of TechnologyAustralia


11

1

Monitoring incidenceof cardiovascular disease

1.1 Introduction

The diverse social, economic and cultural structure of countriesin the Western Pacific region will affect the specific requirements ofa national system for the surveillance of cardiovascular disease.Bennett et al. (1995) outlined the following reasons and benefits of anational monitoring system for the Australian population. Most ofthese points will hold for any population. A monitoring system isrequired to:

(i) monitor progress towards national targets;

(ii) measure the human cost of cardiovascular disease in termsof total and premature mortality, hospital and othermorbidity, person-years of life lost, and the impact of thedisease and its treatment on the quality of life of the patientsand their families;


12

(iii) measure the economic costs of cardiovascular disease interms of costs of treatment, earnings foregone, etc.;

(iv) ensure that the scope for prevention of these diseases isfully and effectively exploited; and

(v) check that preventive and therapeutic interventions ofproven efficacy are being used as widely and effectivelyas possible, and that resources are gradually beingwithdrawn from other strategies that have been lessrigorously evaluated or have been shown to be of lessvalue.

The benefits of a monitoring system are:

(i) report on national trends and patterns of CVD and relateddata;

(ii) provide data for the setting and monitoring of goals andtargets associated with cardiovascular disease;

(iii) enable the study of differential rates of CVD amongpopulation groups (e.g., socioeconomic groups, immigrantgroups, men and women, regions);

(iv) provide data for the evaluation of preventive, diagnosticand treatment interventions; and

(v) provide timely data for use in planning and managing CVDhealth services nationally and locally.

The purpose of this Manual is to recommend methods that canbe used to monitor the incidence of cardiovascular disease (as definedby the World Health Organization) within countries in the WesternPacific region. In some countries it will be possible to use routinely


13

collected data, while in others it may be necessary to establish sentinelcentres to monitor the incidence of cardiovascular disease in arepresentative sample of the population. For some components, suchas coronary deaths and definite myocardial infarction, the goal is tomeasure the incidence of disease and produce valid estimates oftime trends within the population. For other components, such asmedical therapy, the goal is to monitor trends recognizing that thecriteria may vary over time and that documenting of differences inlevels of service is important.

In recognition of the diversity within the region we propose ataxonomy that would allow us to distinguish between the process ofdata collection and the content of the data (Table 1). Four levels ofdata collection are considered. The first level involves collectinginformation that is currently available and accessible. This requires aretrospective audit of data that is available from individual clinics,health centres and hospitals. Data from these sources will give crudeestimates of the prevalence of the each of the various forms ofcardiovascular disease. The second, third and fourth levels will beexplained in more detail in the next section but they require asubstantial increase in commitment and resources. The second levelis establishing a sentinel site and conducting periodic monitoring ofthe relevant health facilities. If a representative population can beselected, the sentinel site will provide good quality information andenable estimates of the incidence of disease to be calculated. Trendsin the incidence of disease can be determined if monitoring is repeatedat the sentinel site. The third level is similar to the second level butmonitoring at the sentinel site should be continuous. This will allowmore accurate estimates to be obtained. The fourth level is onlyrelevant to those countries in which a national monitoring system isestablished but should be the goal for all countries.


14

Tab

le 1

. Hie

rarc

hy

of t

he

pro

cess

.

Dat

a so

urc

es

M

eth

od

sU

ses

Po

ten

tial

Bia

ses

1. D

ata

from

All

patie

nts

iden

tifie

d in

one

Clin

ical

Aud

it an

d hy

poth

e-S

elec

tion

bias

.pa

tient

s of

indi

-cl

inic

al s

ettin

g.si

s ge

nera

tion.

vidu

al c

linic

s or

heal

th c

entr

es.

2. S

entin

el d

ata

Sel

ect a

def

ined

pop

ulat

ion

Est

imat

es o

f dis

ease

bur

den

Will

nee

d tra

inin

g of

obs

er-

colle

ctio

n fr

omof

a c

ity o

r ru

ral a

rea.

and

scop

e fo

r pr

even

tion.

vers

to a

void

mea

sure

men

tde

fined

geo

gra-

Def

ine

the

popu

latio

n fo

r the

Tren

ds o

ver t

ime

if re

peat

ed.

bias

, inc

ludi

ng c

ause

of

phic

pop

ulat

ion

d

enom

inat

or.l

Pro

gres

s to

war

ds n

atio

nal

deat

h co

ding

.on

inte

rmitt

ent

Sel

ect t

he h

ospi

tals

or o

ther

goal

s an

d ta

rget

s.R

epea

t sur

veys

to u

seba

sis

med

ical

faci

litie

s w

hich

sam

e m

etho

d to

avo

idca

ses

of c

ardi

ovas

cula

rbi

as.

dise

ase

will

atte

ndS

elec

tion

bias

if th

ese

ntin

el p

opul

atio

n no

tre

pres

enta

tive

of th

ew

hole

cou

ntry

.


15

3. O

n-go

ing

Def

ine

the

popu

latio

n an

dE

stim

ates

of d

isea

se b

urde

nM

ay h

ave

sele

ctio

n bi

asre

gist

ratio

n fr

omal

l new

cas

es w

hich

aris

e.an

d na

tura

l his

tory

.un

less

car

e is

take

n to

a de

fined

Iden

tify

rout

inel

y co

llect

edA

udit

of m

anag

emen

t,in

clud

e al

l tho

se w

ithpo

pula

tion.

d

ata

sour

ces.

espe

cial

ly s

econ

dary

dise

ase,

not

just

thos

eIn

terp

ret q

ualit

y, u

sefu

llnes

spr

even

tion.

seve

re e

noug

h to

pre

sent

of

thes

ecl

inic

ally

. C

onfo

undi

ngId

entif

y lin

kage

s be

twee

nbi

as (f

or ti

me

com

pari-

regi

ster

and

rout

ine

data

.so

ns)

poss

ible

if s

tand

ar-

diza

tion

for a

ge a

ndso

cio-

econ

omic

sta

tus

etc.

not

per

form

ed.

4. R

outin

e na

tiona

lP

opul

atio

n de

fined

. All

deat

hsE

stim

ates

of d

isea

seW

ill n

eed

train

ing

of o

bser

-m

orta

lity

and

cert

ified

and

not

ified

. b

urde

n an

d sc

ope

ofve

rs, c

aref

ul a

nd s

tand

mor

bidi

ty d

ata

Hos

pita

ls a

nd h

ealth

pre

vent

ion.

ardi

zed

codi

ng s

yste

ms,

colle

ctio

nfa

cilit

ies

have

rout

ine

Tren

ds o

ver t

ime.

valid

atio

n of

mea

sure

s,sy

stem

s.cl

assi

ficat

ion

and

Pro

gras

s to

war

ds n

atio

nal

and

com

plet

enes

s of

dat

ano

tific

atio

n sy

stem

.go

als

and

targ

ets.

colle

ctio

n to

avo

id b

iase

s.In

terp

ret q

ualit

y, u

sefu

lnes

sP

ictu

re fo

r th

e w

hole

and

app

ropr

iate

ness

coun

try.

Iden

tify

linka

ges

betw

een

Dev

elop

men

t of ‘

cultu

re’ o

fco

mpo

nent

s of

the

syst

emco

llect

ion

and

use

of d

ata

in th

e he

alth

sys

tem

.

Tabl

e 1.

Hie

rarc

hy o

f the

pro

cess

. (co

ntin

ued)


16

1.2 Establishing a sentinel site

See Appendix 1.

1.3 Using routinely collected data

Routinely collected data may be available through deathcertificates and hospital separation data, but the diagnostic criteriafor cause of death or cause of hospital admission may vary betweencountries. Another potential problem is that some cardiovascularevents may be double counted. For example, patients transferredbetween hospitals during the same clinical ‘episode’ would not usuallybe identified unless it is possible to link records. Also deaths in hospitalwould be counted both from death certificates and hospital records.Thus, methods are needed to avoid double counting of cardiovascularevents as far as possible.

The results from several Australian studies on the accuracy ofroutinely collected data provide the basis for the recommendationsin this Manual. Studies of the validity of hospital data for non-fatalAMI and other acute episodes of CHD have been conducted inNewcastle and Perth as part of the World Health Organization(WHO) MONICA Project (to MONItor trends and determinants ofCArdiovascular disease) and in Queensland as part of the QueenslandHeart Attack and Morbidity and Mortality Study (QHAMMS). Similarstudies on the validity of death certificates for CHD were alsoconducted as part of the WHO MONICA Project. Data from otherhealth areas are available through the New South Wales (NSW)Acute Cardiac Care study - a study that examined the managementof patients with acute cardiac ischaemia in over 30 hospitals in NSW.The validity of hospital data for strokes has been assessed by thePerth Community Stroke Study.


17

Many countries do not have as many resources necessary fordata collection as in countries such as Australia. While using theexperience gained through the MONICA and other surveillancesystems that have been established, the Australian experience hasbeen modified to take account of resource limitations.

Age

It is proposed that data be collected for the population aged 30-79 years. To make comparisons between countries, the use of theSEGI world population as the reference population for agestandardized rates is suggested. In general, it is suggested that thecompilation of age-specific rates be done in 10-year bands.

Definition of an event

The diagnostic categories for all episodes of disease relate tothe principal discharge diagnosis only. These data can be collectedon a regular or intermittent basis. Both ICD-10 and ICD-9 codeshave been provided here.

1.3.1. Suggested optimal data set required to becollected. (Note: these are based on assessingthe use of evidence-based interventions as wellas the disease burden.)

Acute myocardial infarction (AMI)

• Population rates of fatal and non-fatal AMI (needsdeath certificates, census and hospital separations).

• In-hospital deaths.

• Time between onset of pain and hospital arrival.


18

• Proportion admitted to hospital given thrombolysis.

• Proportion with appropriate indicators giventhrombolysis.

• Proportion discharged on beta blockers.

• Proportion discharged on aspirin.

• Proportion entering rehabilitation programme.

• Proportion admitted to hospital who have angioplasty/CABG in 12 months.

Unstable angina pectoris (UAP)

• Hospital admission rates (although it should berecognized that the diagnosis of UAP is unreliable inmost countries and that many people discharged witha diagnosis of UAP will actually have had an episodeof prolonged chest pain).

Angioplasty/coronary artery bypass graft surgery

• Rates for the population, age and sex specific.

Stroke

• Population rates of fatal and non-fatal strokes (needsdeath certificates, census and hospital separations).

• Proportion of those with stroke who are admitted tohospital.

• Time between onset of symptoms and arrival athospital.


19

• Proportion of those admitted to hospital who die.

• Proportion of those admitted who are managed in anorganized stroke unit. An organized stroke unit is award within a hospital that is established for the solepurpose of treating stroke patients.

• Proportion discharged on aspirin.

• Proportion who are physically dependent 6 monthsafter discharge.

Congestive cardiac failure (CCF)

• This will be very difficult to identify and hospitaldischarge data on CCF due to non-valvular heartdisease may be useful.

Diabetes

• Population rates of deaths from and with diabetes(needs death certificates, census and hospitalseparations).


20

1.3.2 Suggested statistics to be derived from the abovedata

Acute myocardial infarction

For monitoring incidence of AMI it is recommended that:

• The rate of acute coronary events be calculated asthe sum of the rate of coronary deaths estimated fromdeath certificates and the rate of non-fatal AMIsestimated from hospital separations.

• For fatal coronary events, deaths with ICD-10 codesI20-I25 (ICD-9 codes 410-414) be used with smalladjustment factors to account for underestimation.

• For non-fatal AMI, hospital separations be used wherethe patient is discharged alive, the primary diagnosisis coded I21, I22 using ICD-10 (410 using ICD-9)andthe length of stay is greater than 2 days. Smalladjustment factors should be used to account foroverestimation due to hospital transfers, readmissionsand other effects.

• Validation studies of the coding of death certificatesand hospital diagnoses are needed in the context ofon-going quality assurance, the changes from ICD-9to ICD-10, and checking the general applicability ofthe present recommendations.


21

Angina pectoris

For monitoring incidence of angina pectoris it isrecommended that:

• The rate of angina pectoris be obtained by countingall patients who had an unbooked (emergency)admission to hospital and who were given a primarydischarge diagnosis coded I20 or I24 according toICD-10 (411 or 413 according to ICD-9).

• Primary discharge codes of I20 or I24 according toICD-10 (411 or 413 according to ICD-9) be consideredtogether for validation studies of angina pectoris. Thereis insufficient information in medical records todistinguish between cases of unstable angina pectorisand stable angina pectoris.

Investigations and procedures

For monitoring the number of coronary investigationsand procedures, booked admissions coded to I20 toI25 according to ICD-10 (413 to 414 according toICD-9) should be subdivided into: (i) admissions withpercutaneous transluminal coronary angioplasty(PTCA) and coronary artery bypass grafts (CABG):and (ii) admissions without PTCA or CABG. Theadmission rates for each of these categories(separately) should be monitored, without anyadjustments.


22

Stroke

Not all patients with strokes are treated in hospitaland the experience in certain countries, such asAustralia, is that hospital separation data cannot beused to measure past trends in attack rates for acutestrokes because of the rapidly changing proportionsof nonfatal cases admitted to hospital. If the proportionof cases admitted to hospital in other countries can beconfirmed to be constant through validation studies, itis possible to use hospital separation data to monitortrends in strokes in those countries.

Estimates of the rate of admission to hospital for anacute stroke can be made using the following selectionalgorithms: Total acute stroke is the sum of nonfatalstroke and fatal stroke.

(i) where non-fatal stroke is defined as: non-fatal hospitalseparations coded as acute stroke (ICD-10 codes I60,I61, I63, I64, I66 while ICD-9 codes 430, 431, 434 or436), OR acute stroke is coded in another diagnosticfield for an admission of at least three days’ durationthat is unbooked, and

(ii) fatal stroke is defined as: deaths with ICD-10 codesI60, I61, I63, I64, I66 (ICD-9 codes 430, 431, 434,436).

Further studies should be undertaken in different countriesto test the algorithms described above and to determinethe proportion of nonfatal cases admitted to hospital. Similarmethods should be used to determine ‘first’ events, definedin terms of no previous admission because of acute strokewithin a defined period (for example, five years).


23

Congestive cardiac failure

Definitive diagnostic tests for (CCF) are not conducted inmost countries on a routine basis. However, the clinicalcharacteristics of CCF are sufficiently distinctive for it tobe recognized. Even so, monitoring of CCF using hospitalseparation data will give poor estimates of the true extentof the disease because a large proportion of patients withCCF are discharged from hospital without any mention ofCCF in their primary or secondary discharge diagnosescodes. Furthermore, validation of discharge diagnoses ofCCF is not feasible due to the lack of consistency indocumentation of signs and symptoms of CCF in medicalrecords.

There was an increase in the number of patients given aprimary discharge diagnosis of congestive cardiac failure(ICD-10 code I50 while ICD-9 code 428) between 1987and 1989 in Australia. This was at least partly due to atransfer of patients between discharge codes. The increaselevelled out after 1989 and, apart from this aberration, theage standardized rate of CCF as a primary dischargediagnosis remained relatively constant between 1990 and1996. It is likely that half of the increase between 1987and 1989 occurred because, for certain types of events,CCF codes were substituted for codes for hypertensivedisease (ICD-10: 110-115; ICD-9: 401-404). Therefore,the analysis of longitudinal data that cross this period mayneed to combine the categories for hypertension and CCF.

A substantial increase in CCF as a secondary dischargediagnosis was observed in Australia after 1987. This islikely to be due to the change to case-mix funding basedon diagnosis related groups.


24

Diabetes

Hospital separation data for principal or any mention ofdiabetes (ICD-10: E10-E14; ICD-9: 250).


25

2.

Indicators

These will span the range from mortality, hospital separationsand case fatality to prevalence of population risk factor levels. Theactual goals will not be specified, as these will have to be set by eachcountry to reflect current patterns and realistic goals.

These indicators should be assessed every 5 years to allow thetrends to be assessed.

After the first assessment of the indicators, local and nationalgoals should be established that will allow trends over time to bemeasured in an attempt to see how much movement there is towardsthese goals.

2.1 Risk factor indicator

• Proportion of adults who smoke regularly, aged 18 andabove.

• Proportion of secondary school students who smoke.


26

• Proportion of adults not engaged in regular physical activity,aged 18 and above.

• Proportion of adults who are overweight or obese, aged 18and above.

• Proportion of adults with high blood pressure, aged 30 andabove.

- Proportion of those with high blood pressure or ontreatment for high blood pressure who are on treatmentfor high blood pressure.

- Proportion of those on treatment for high bloodpressure whose blood pressure is below 160mmHgsystolic.

• Proportion of adults with high blood cholesterol, aged 30and above.

- Proportion of those with high blood cholesterol or ontreatment for high cholesterol who are on treatmentfor high cholesterol.

• Contribution of saturated fat as a proportion of total energyintake, aged 18 and above.

• Proportion of adults who are diabetic, aged 18 and above.

- Proportion of adult diabetics whose HbA1c is lessthan 7.


27

2.2 Indicators for disease patterns

Coronary heart disease

• Incidence rate for AMI, ages 30-79. This comprisesmortality from CHD and hospital separations for AMI. Itis an indicator of the effectiveness of prevention of CHD.

• Proportion of patients who die before leaving hospital afteracute myocardial infarction. An indicator of both selectionof those who arrive at hospital and effectiveness ofinterventions in hospital.

• Median delay between onset of chest pain and presentationfor emergency care at hospital, all ages. An indicator ofpre-hospital response time to a cardiac emergency anduptake by the public of education messages.

• Time from presentation at emergency department to receiptof thrombolysis or angioplasty, all ages. An indicator of in-hospital response time to a cardiac emergency.

• Hospital separation rates for principal diagnosis of unstableangina, all ages. An indicator of the use of hospital resourcesfor unstable angina.

• Hospital separation rates for principal diagnosis ofcongestive heart failure, all ages. An indicator of the useof hospital resources for congestive heart failure.

• Proportion of cardiac patients who enter and complete arehabilitation programme, all ages. An indicator of the neteffect of promoting rehabilitation programmes and theiravailability.


28

• Proportion of patients admitted to hospital with AMI whohave angioplasty or revascularization within 12 months.An indicator of the use of high technology resources forthe treatment of and prevention of mortality from coronaryheart disease.

Stroke

• Incidence rate for stroke, all ages. This comprises mortalityfrom stroke and hospital separations for stroke. An indicatorof the effectiveness of prevention of stroke.

• Median delay between onset of stroke symptoms andpresentation for emergency care, all ages. An indicator ofpre-hospital response time to a stroke emergency and ofthe uptake by the public of education messages.

• Proportion of patients admitted to hospital with acute strokewho are managed in organized stroke units (dedicatedmutidisciplinary teams), all ages. An indicator of theavailability of organized stroke units.

• Proportion of stroke patients who are physically dependent6 months after diagnosis of acute stroke event. An indi-cator of the effect of acute care and rehabilitation followinga stroke event.

• Proportion of stroke patients admitted to hospital who dieduring hospitalisation (case fatality). An indicator ofselection for hospital care and effectiveness of in-hospitalmanagement.

• Death rate for stroke, all ages. An indicator of the effectof prevention and treatment and management of a stroke.


29

Diabetes

Death rate for diabetes, ages 30-79. An indicator of codingpractices as well as the effect of prevention and treatmentof diabetes.


30


31

Appendix 1:

Establishing a sentinel site

A balance has to be struck between the ideal and the practical.If only one centre is to be established in a country it is hoped that thecentre will be typical (in its morbidity and mortality levels and trends)of the country as a whole. Where a country has greatly contrastingregions the possibility of setting up several centres should beconsidered.

The study population

For practical reasons of monitoring CVD, the study populationshould have well defined geographical boundaries and, ideally, shouldbe an administrative unit for both the local governing body and theprovider of medical services. This will ensure that the populationwhich utilises the medical services is the same population for whichinformation on the size and age distribution of the population can beobtained.


32

Study populations should be investigated to find methods ofidentifying people who have an acute CVD event while travelling orworking outside the study area. This problem arises particularly ifthe study population is a sub-population of a city; it is less importantif a total community is involved.

The number of people in the study population should bedetermined by the expected number of morbid or fatal events in theage group concerned. The number of events will determine theprecision of the estimated rate of events and this will becomeparticularly important if the data are to be used for estimating trendsin event rates. Event rates for CVD should be estimated separatelyfor men and women and sample size calculations should be doneseparately for men and women. In the calculation below, the rate ofcoronary deaths and the 95% confidence interval are estimated fora hypothetical population of 40,000 people in which 100 coronarydeaths were observed in a 1-year period. The estimated rate ofcoronary death is 250 per 100,000 population and we are 95%confident that the true rate is between 201 and 299 per 100,000population. A table for various populations sizes and rates of eventsis shown in Appendix 2.

Example:

In a given population of 40,000 people you expect there will be100 coronary deaths during the 1-year period of monitoring.

Rate of death is equal to 100 per 40,000 population.

Rate of deaths per 100,000 population equals (100/40,000) x100,000 = 250.

Variance of the rate of deaths per 100,000 equals (100,000/40,000)2 x 100 = 625.


33

Standard error of the rate of deaths per 100,000 is equal to the

square root of the variance 25625 = .

Therefore the 95% confidence interval will equal 250 ±(1.96 x 25).

So our estimated rate per 100,000 (with 95% CI) will equal 250(201, 299).

Method of surveillance: In many countries it will not be possibleto obtain reliable estimates of morbidity so it is important to obtainreliable data on mortality to obtain a useful measure of the burden ofCVD in that population. The logistics of a surveillance system in aparticular community will depend on local custom and the healthsystem of the population being studied. However, based on theexperience of the WHO MONICA Project, the following guidelinesshould be followed to establish a sentinel site.

An important requirement of a surveillance system is that eventregistration procedures are similar throughout the period or theestimates of trends will be invalid. The surveillance system shouldconcentrate on recording information for cardiovascular eventsconsisting of non-fatal coronary events, coronary deaths, non-fatalstrokes and fatal strokes. The maximum duration of an event shouldbe well defined. For example, in the MONICA Project an eventcould last for a period of 28 days. This allows events to be categorizedas the first of recurrent events, fatal or non-fatal events, andestablishes a denominator for the estimation of case-fatality.

All suspected events should be collected from death certificates,from hospital medical records and from community health services.These events can then be examined for eligibility into the coronaryand stroke register based on a predefined definition of an event.There are two possible methods of event finding:


34

(i) prospective, identifying and following up suspect casesduring the hospitalization or medical attention; and

(ii) retrospective, abstracting information from medical records,death certificates and other possible sources without seeingpatients at all.

The recommendations in this report are based on prospectiveregistration, although some fatal events among residents of thepopulation are likely to occur outside the study area and notificationof these deaths may take several weeks or longer.

In any sentinel site an information gathering system will need tobe established to gain information on likely events and to obtain allavailable diagnostic and other information. The sources will dependon the structure of the local medical and medico-legal services. Thefollowing sources may be used to identify cases:

- Death certificates

- Hospital admissions or discharge records

- Community health centres

- Necropsy and medico-legal records

- General practitioners

- Local newspapers

- ECG records

- Laboratory records

- Social insurance records


35

- Emergency services, for example ambulance

- Interviews with patients and relatives

Once sources of cardiovascular events and criteria for definingcardiovascular events have been determined it is important thatmethods are not changed over time without measuring theconsequence, as a change of sources may cause a spurious changein event rates.

Identification of fatal cases

All death certificates in which codes for cardiovascular diseaseare mentioned as immediate, main and antecedent or underlyingcauses of death should be investigated further. The following causesof death should be considered a minimum set for defining events tobe followed up for further information:

Cause of death ICD-10 code ICD-9 codeDiabetes E10-E14 250Hyperlipidaemia E78 272Obesity E66 278Cardiovascular

Hypertensive I10-I15 401-405Ischaemic I20-I25 410-414Other I30-I52 420-429Stroke I60-I69 430-439Arteries I70-I79 440-447

Symptoms R96, R97, 797-799R98, R54

It is possible to exclude those deaths that are due to trauma,chronic obstructive pulmonary disease, cancer, cirrhosis of the liver,or rheumatic heart disease without mention of atherosclerotic heart


36

or vascular disease. However, if an atherosclerotic condition ismentioned in the sequence of events or if the cause of death isattributed to one of the conditions listed above, the cause of deathshould be validated.

Validation should be based on any available medical and medico-legal records and , if necessary (depending on local custom),interviews of the decedent’s next of kin or another informant. Medicalrecords for the period within a minimum of 28 days of death shouldalso be examined for information that may elucidate thecircumstances leading to the death. The interview should establishthe circumstances surrounding the decedent’s death.

Deaths should be validated based on applying some predefineddiagnostic criteria. The criteria should include information on thesymptom history, ECGs, cardiac enzymes, necropsy and deathcertificate data.

Identification of non-fatal cases

Coronary events

To identify people treated in hospital, all acute medicaladmission wards as well as hospital laboratory data on cardiacenzymes and ECGs should be monitored on a daily basis to ensureall cases are detected. For people who are treated out-of-hospital,daily monitoring of pathology laboratory records for cardiac enzymesand ECG laboratory records may be a source of case ascertainment.

Stroke events

To identify people treated in hospital, all acute medicaladmission wards as well as hospital laboratory data on cardiacenzymes and ECGs should be monitored on a daily basis to ensure


37

all cases are detected. In most countries, many stroke patients aretreated at home. In this situation, contact must be made with physiciansand a protocol established to identify and collect data on strokepatients. The protocol could take a number of forms and might includereviewing charts on a regular basis or providing each office with alog book and a poster reminding them to log all cases of suspectednew strokes, or asking the office nurse or assistant to notify themonitoring centre when a new episode of stroke is identified.

Core data collected

A guide to the minimum data that should be collected is shownbelow for coronary heart disease. For other forms of CVD, additionaldata may be required.

· Patients name and address

· Date of onset

· Survival status at time of discharge from hospital and/orafter 28 days

· Gender

· Age

· Discharge diagnosis (if hospitalized)

· ECG readings

· Other diagnostic tests performed (in particular, cardiacenzyme levels if available)

· Symptoms


38

· Treatments administered

· History of cardiovascular disease

· Number of years of schooling

This is considered the minimum amount of information requiredfor data collection.

Countries with adequate resources may be able to add data asfollows:

• Data to help assess the use of evidence based interventionsin the clinical care and secondary prevention of the disease.For example the use of aspirin after most CVD categories,the use of thrombolysis and beta blockers after acutemyocardial infarction, entry to organized stroke units andrehabilitation programmes and the use of angioplasty andcoronary artery bypass surgery.

• Data to assess more accurately the economic and socialburden of disease. For example, the level of disability aftera stroke, quality of life and return to work after the event.


39

Appendix 2:

Sample size calculations

Table of sample size calculations.

No. of events Size of Rate of events 95% confidence (annually) population (per 100,000) interval

50 40 000 125 090-16060 000 83 060-10680 000 63 045-80..

100 000 50 036-64..200 000 25 018-32..

100 40 000 250 201-29960 000 167 134-19980 000 125 101-150

100 000 100 080-120200 000 50 0040-60..0

150 40 000 375 315-43560 000 250 210-29080 000 188 157-218

100 000 150 126-174200 000 75 063-87..

200 40 000 500 431-56960 000 333 287-38080 000 250 215-285

100 000 200 172-228200 000 100 086-114


40

Rate per 100 000 = 100 000number of events

size of populationx

Variance (Rate per 100 000) = x number of events

100 000size of population

2

Variance(Standard error (Rate per 100 000 = Variance (Rate per 100 000)

The numbers in the above table are based on the followingformula:

assuming that the size of the population is constant andthat the number of events is a Poisson random variable.

Note: Very similar answers will be obtained if you assumethat the number of events divided by the size of thepopulation is a binomial random variable.

number of eventsRate =

size of population

Documents

MANUAL ON MONITORING - iris.wpro.who.int · Manual on monitoring cardiovascular diseases 6 The WHO Collaborating Centre for Cardiovascular Epidemiology at the University of Newcastle,