did we do it right? - an evaluation of the colour coding ... · did we do it right? -an evaluation...

ORIGINAL ARTICLE

Did We Do it Right? · An Evaluation of the ColourCoding System for Antenatal Care in Malaysia

J Ravindran, FRCOG*, K Shamsuddin, Dr PH** , S Selvara;u, DPH***

'Department of Obstetrics & Gynaecology, Seremban Hospital, 70300 Seremban, Malaysia, **Department ofCommunity Health, Universiti Kebangsaan Malaysia, Kuala Lumpur, "'Family Health Division, Ministry of Health,Kuala Lumpur

Introduction

The risk approach system using colour codes hasbeen used in Malaysia since 1989. This systemgrades all antenatal mothers according to the

level of severity of risk factors. The checklist wasmade into a format and attached to the antenatalcard. A self-adhesive colour tag was placed onthe upper right hand corner of the mother'santenatal card.

This article was accepted: 18 October 2002Corresponding Author: Ravindran Jegasothy, Department of O&G, Seremban Hospital, Jolon Dr. Muthu, 70300 Seremban, NegeriSembilan

Med JMalaysia Vol 58 No 1 March 2003 37

ORIGINAL ARTiClE

Four colours were utilised. Red signified that therisk factor required immediate hospital referraland admission. Yellow indicated that the patienthad to be monitored antenatally by a doctor.Green indicated that the risk factor requiredmonitoring by a senior nurse. The white codeindicated no or low risk. Although white codedcases were considered suitable for supervisedhome delivery, the patients had to be monitoredby going through the checklist at each visit. Thechecklist and colour coding were basically a setof instructions intended for rural health staff toenable them to differentiate between cases theycould continue to see and those that should bereferred and to refer cases to the appropriatelevel of care; to ensure that referred cases wentdirectly to the appropriate level withoutunnecessary delay; and to ensure that theappropriate category of staff in health centres orhospitals treated the cases without delay.

In 1991, gUidelines on the management of highrisk cases were reviewed and revised to furtherenable midwives/ community nurses to knowtheir role in the management of these cases.

The colour coding system should not beconfused with a risk assessment or predictionsystem as the risk approach or colour codingsystem was a managerial tool to identify levels ofcare and personnel to handle particular diseaseconditions in pregnancy. In this system,attention is paid to those individuals and groupswhose characteristics are associated with anincreased risk of having or developing or beingespecially affected by a health problem.

Both the positive and the negative predictivevalues of all scoring and risk assessment systemsare poor. Depending on the cut-off point and thetest chosen, only between 10 and 30 per cent ofwomen who are allocated to the high risk groupsactually experience the adverse outcome forwhich the risk assessment system declares themto be at risk'. Evaluation of risk scores for pre-

38

term birth and low birth weight revealed thatbetween 20 and 50 per cent of mothers whodeliver pre-term or low-birth weight infants havelow risk scores2,3.

Screening of risk mothers using the colour codingsystem was designed to rapidly identify differentdegrees of risk associated with pregnancy andensure appropriate action. The main purpose ofscreening for risk factors was to detect thosewomen who were more likely than others' tohave an adverse outcome of pregnancy and torefer them to a more skilled and better equippedlevel of care, both antenatally, for a moreaccurate assessment, and later for safemanagement of a potentially more difficultlabour. It was hoped that this system if carriedout in an effective manner would have an impacton maternal and perinatal health.

This present evaluation was done in order toassess the level of appropriate management andoutcome among high risk mothers with 3conditions that were perceived to be commonantenatal problems in areas with high maternalmortality compared to those in areas with lowermaternal mortality. These were hypertensivedisorders of pregnancy, postdates and anemia inpregnancy. Appropriateness of care wasdetermined by the guidelines issued to healthnurses that included the appropriate colour code,level of care and treatment for varying severity ofthese conditions. The study was coordinated bythe Family Health Division of the Ministry ofHealth, Malaysia.

Materials and Methods

This was a retrospective follow-through studyconfined to Peninsular Malaysia. High riskmothers using services in government healthfacilities. with specific medical problems I.e.hypertensive disorders of pregnancy, postdatesand anaemia in pregnancy were chosen.

Med J Malaysia Vol 58 No 1 March 2003

Did We Do it Right? -An Evaluation of the Colour Coding System for Antenatal Care in Malaysia

Sampling MethodThere were three levels of selecting the studysites, health facilities and the sample units.

Levell: Determination of districts for selection ofstudy siteAs part of a larger study, the study sites werechosen through multistage sampling of healthdistricts and subsequently health facilities.Sixteen health districts were chosen throughstratified random sampling of health districts inPeninsular Malaysia stratified by their level ofreported maternal mortality ratios as collected bythe Family Health Development Division of theMinistry of Health. The Health Division hascollected data on maternal mortality by healthdistricts since 1991 through the ConfidentialEnquiry into Maternal Deaths (CEMD) and thishas been shown to have a higher coverage thanthat collected by either the Statistics Departmentor the Information and Documentation System ofthe Ministry of Health.

The 81 health districts were ranked by levels ofmaternal mortality ratios and classified intodistricts with low or high maternal mortalitybased on being higher or lower than the median.Due to the observed fluctuation of the maternalmortality ratios especially in small districts, anadditional criteria indicating stability wasincorporated into the definition of areas with lowor high maternal mortality. In this study, thematernal mortality ranking by district for 1992was randomly selected as the sampling frame anddistricts ranked as having low maternal mortalityin 1992 and continuing to be ranked as such in1994 were considered areas with low maternalmortality and likewise areas with maternalmortality higher than median in 1992 andcontinuing to be such in 1994 were consideredhigh maternal mortality areas. Thus areas havingstable maternal mortality status were eligible forselection. Using the Table of Random Numbers,eight districts from areas with low and eightdistricts with high maternal mortality ratios werechosen giving a total number of 16 randomly


selected districts for the larger study on theevaluation of the risk approach strategy.

In this study, 5 districts were randomly chosenfrom the 16 selected districts included in thelarger study. The low maternal mortality districtsincluded in the study were Melaka Tengah andSeremban, while the high maternal mortalityareas were Kuala Selangor, Bentong and LarutMatang and Selama districts.

Level 2: Determination of Health Facilities forSelected Health Districts.Fifty percent of health clinics in each selecteddistrict were randomly selected and automaticallyall community clinics under the selected healthclinics were included in the study.

For the selection of hospitals, all governmenthospitals present in the selected districts wereautomatically included in the study. However,when the selected district had no governmenthospital, the nearest referral hospital in the samestate was chosen. When cases were referred toseveral government hospitals, the first referralhospital was considered in this study. Table Ishows the number of health facilities selected forthis study.

Level3: Selection ofsampling units, the antenatalcards for assessmentAntenatal cards of mothers who deliveredbetween July 1996 to June 1997 were assessed forthe three medical conditions chosen for thisfollow-through study. The selection of antenatalcards was done through sieving all cards in thehealth facilities selected for the three conditionsof interest this study. Fifty percent of theantenatal cards with each of the selectedconditions were then randomly selected. Oncethe antenatal cards for these three conditionswere identified and selected, they were listedaccording to the identity card numbers and thislist was then forwarded to the medical recordofficers in the hospitals concerned for help in theretrieval of hospital records for further analysis.

39

ORIGINAL ARTICLE

The head of the O&G department in therespective hospitals were also informed so thatthis process was facilitated.

For cases that were not referred, where hospitalrecords could not be retrieved or where theywere transferred to hospitals outside the stateduring the period of· data collecticm, the sampleunit was not dropped from the study but theassessment of their management in the healthfacilities was continued. Follow-through for caseswhich were referred to several governmenthospitals was confined to the first referralgovernment hospital within the district only.

Data was abstracted and entered using theformats FH-RAS 2/1.1997 for anaemia inpregnancy, FH-RAS 2/2.1997 for hypertensivedisorders in pregnancy and FH-RAS 2/3. 1997 forpostdates. Major outcomes, accuracy of colourcoding assignment, appropriateness of care andfeedback received among referred and admittedcases were then analysed.

The criteria for appropriateness of care issummarised in Table II. Appropriateness of carewas determined to be in accordance with theinstructions issued when the colour codingsystem was introduced. For anemia in pregnancy,a haemoglobin level of <8gm/dl at any gestationwas coded red as was a haemoglobin level of 89 gm/dl when the gestation was equal to or morethan 32 weeks. When the gestation was less than32 weeks in the latter category (Hb 8-9 gm/dl)such patients were coded yellow and thusrequired the attention of· the medical and healthofficer at the health clinic. However, if thegestation was equal to or more than 32 weekswith a haemoglobin level of between 9-10 gm/dlsuch cases were also coded yellow and requiredsimilar action. In this category of patients whowere less than 32 weeks, attention from a staffnurse was deemed appropriate (green code).Upon referral to hospital with a red code, suchpatients would have to be admitted andinvestigation and treatment instituted. If thehaemoglobin was less than 8 gm/dl then a

40

peripheral blood film would have to be done inorder to classify for appropriate care.

The definitions of hypertensive disorders ofpregnancy were more complex. A patient at agestation of less than 36 weeks with a bloodpressure of 140/90 or more without albuminuriawas coded yellow. In such cases if thegestational period was more than 36 weeks or inany gestational period with the presence ofalbuminuria in association with an elevatedblood pressure, the case was coded red. The redcode was also utilised for a case with elevatedblood pressure of 160/110 with or withoutalbuminuria and for cases with eclampsia. In thehospital care was deemed appropriate if thepatient was admitted, the management was inconsultation with a medical officer or specialist,an anti-hypertensive was prescribed and therewas a delivery plan outlined. An anti-convulsantshould also have been prescribed if the case wasdiagnosed to be eclampsia.

Postmaturity was defined as a period of gestationof 42 weeks and above. Such cases were to becoded yellow and referred to hospital within aday of the clinic visit. In the hospital, the doctorwas expected to document confirmation of thediagnosis and then admit the case for appropriateintervention. The intervention was left to theclinical judgement of the doctor taking care ofthe patient.

Collected data were coded and saved in dbf.(DBASEIV) and xls. (EXCEL version 6) datafiles.The statistical package used throughout theanalyses was SAS (Version 6.09, SAS Institute Inc.,Gary, North Carolina, USA). Analyses weredescriptive using weighted frequency andpercentage distributions. Data was weighted torepresent the original 5 selected districts selectedfor the follow-through study. The weights werecalculated as the inverse of the probability ofcases with the specified conditions to enter thestudy from among the total number of antenatalcards identified to have those conditions in theselected health facilities. Applying these



sampling weights, the weighted sample in thisstudy contained a total of 554, 252, and 283 casesof anaemia, hypertensive disorders in pregnancyand post-maturity respectively while analysis ofthe combined conditions contained 1089observations.

Datasets on each of the condition as well ascombining the three conditions was analysedusing the univariate logistic regression analyses.The crude effects of the level of maternalmortality on the main implementation outcomemeasures of the risk approach strategy andcolour coding system were estimated. Themeasures of associations and confidencepresented are the odds ratio and its 95%confidence intervals.

Results

Anaemia in Pregnancy:The number of cases with anaemia were 562cases (Table III). Of these, 311 cases were in lowmaternal mortality areas and the remainder, 251cases were in high maternal mortality areas. Theincidence of anaemia was higher in areas withhigh maternal mortality compared to that in areaswith low maternal mortality (7.7% versus 6.3%).

The selection rate varied from 51.1% to 63.7% inthe low and high maternal mortality areasrespectively. The information on management bythe health staff was not available in only onecase. This was not surprising as the study wasbased on records held in the health clinics.However, information on management in thehospital sector was not detailed in 21 cases out of88 cases referred to the hospitals. In 12 of thesecases, the hospital records themselves could notbe traced. In the rest, the information required inthe study could not be discerned from theavailable records. These cases were equallydistributed in both areas. Birth outcomes werenot available in 66 cases in the low maternalmortality areas and 16 cases in the other areas(Table IV).


In the weighted analysis, the number of anemiacases was 309 in the low maternal mortality areasand 245 cases in the high maternal mortalityareas. Further analysis showed that there was ahigher number of anemia cases in the severe(haemoglobin<8g/100ml) and moderately severe(haemoglobin 8-9 g/dl) categories in the lowmortality areas compared to cases in the highmaternal mortality areas (p<0.05). Thedistribution of the anemia cases among the fourcolour codes were found similar in both areas.However, five cases were not given the colourcode in the high mortality area and this did nothappen in the low mortality area.

The red colour code was inappropriately given in84.8% of cases in the low mortality areascompared to 81.5% of cases in the high mortalityareas. In cases with severe anemia, 84 caseswere assigned the yellow code in both areas, 8cases the green code and 3 cases were evenassigned the white code! Two cases with severeanemia were not assigned any code in the highmortality areas. There tended to be overcodingin the low mortality areas with 7 cases beingassigned the red code when it should have beenyellow and 4 cases red when green would beappropriate. This was not seen in areas with highmaternal mortality.

Table V indicates the accuracy of colour codingwas 42.6% in the low and 45.5% in the highmortality areas respectively. Table VI shows thatthis difference in coding accuracy in the twoareas did not differ significantly (OR=0.9; 95%CI=0.6-1.3).

There were inappropriate referrals to thehospitals in the vast majority of cases (90.1% inlow mortality and 83.8% in high mortality areas).In the high mortality areas, inappropriate referralwas found to be significantly lower than thatfrom low mortality areas (OR=0.6, 95% CI= 0.30.97).

41

ORIGINAL ARTiClE

Probably as a result of the inaccurate colourcoding given, care was deemed inappropriate bydiagnosis in 283 of the 554 patients (varying from59.5% in low mortality areas to 40.3% in highmortality areas). The risk of inappropriate care bydiagnosis was significantly lower in highcompared to low mortality areas (OR=O.5, 95%CI= 0.3-0.7). In contrast, the care given wasdeemed appropriate by assigned colour code inthe vast majority of patients in both areas (84.8%in low and 82.2% in high mortality areasrespectively). Similarly, the appropriateness ofthe category of personnel attending to thepatients was also correct in more than 75% of thepatients in both areas (Table V).

Hospital staff tended to give appropriate careaccording to the protocols outlined in the riskcoding guidelines in 47.4% of patients in the lowmortality areas compared with only 9.0% in highmortality areas and this difference was found tobe statistically significant (p<0.05) (Table V).There was also poor feedback from the hospitalsector to the health staff. In fact, no feedback wasgiven in 62 patients and the status of feedbackwas unknown in 34 patients out of a total of 160cases with anemia admitted to hospitals duringthe study period. The risk of poor feedback washigher in high compared to low mortality areas(OR=3.0, 95% CI=1.5-6.2).

Overall the maternal outcomes were unknown in131 of 309 mothers (42.4%) in the low mortalityareas compared to 14 of the 245 mothers (5.6%)in the high maternal mortality areas.

Hypertensive Disorders of Pregnancy:Table IV shows that 257 cases with hypertensivedisorders of pregnancy were identified in thestudy districts and 136 cases were selected for thefollow- through study. The selection rate was53.2% in the low mortality and 52.2% in the highmortality areas respectively. Out of the 136 casesselected, there were 10 cases with missinginformation by health staff C7 in the low mortalityand 3 in the high mortality areas respectively).There were 37 cases in which information sought

42

in the study was not found in the notes. Therewere 119 admissions to the hospital in the groupof 257 cases studied. Hospital records were nottraceable in 12 patients (all in low mortalityareas).

After weighting, 169 cases in low mortality and83 cases in high mortality areas were analysed forhypertensive disease in pregnancy. Thedistribution of cases among the various diseasecategories and colour codes were similar in bothgroups.

The accuracy of the assigned colour code was45.9% in low mortality areas and 58.6% in highmortality areas (Table V). Twelve cases whichshould appropriately have been given a red code(blood pressure >160/110 .± edema) werecategorised wrongly into the yellow code. Inaddition, two cases in the similar category wereallocated the green and white codes respectively.

Referrals to hospitals were noted to beinappropriate in 56.4% of the patients in the lowmortality areas whereas this was 35.3% in thehigh mortality areas (OR=0.4, 95% CI=0.2-0.8).

The care given when classified by diagnosis wasalso different from that advocated in theprotocols in 66.5% of cases in high mortalityareas as compared to 85.7% of cases in the lowmortality areas (OR=0.3, 95% CI= 0.2 - 0~6).

When appropriateness of the care was evaluatedby the assigned colour code it was noted to beinappropriate in 192 cases out of the total 252weighted cases. However, Table IV showed thatthere was no statistical difference between thetwo areas in terms of appropriateness of care byassigned colour code.

When the health personnel were analysed withregards to the appropriateness of the care given,it was found to be significantly more appropriatein the high maternal mortality areas compared tolow maternal mortality areas (74.7% vs 58.15).The odds ratio of inappropriate care by healthstaff in high compared to low mortality area was



0.5, (95% CI=0.3-0.9). Care was determined to beinappropriate in 36.7% of cases handled byhospital staff in high compared with 56.2% ofcases in the low maternal mortality areas.However this difference in hospital care in thetwo areas was not statistically significant (TableVI).

The deficiencies in feedback was againnoticeable in Table V with only·36 cases out of145 cases (29.2%) in the low mortality areas and21.0% of cases in the high maternal mortalityareas receiving a feedback.

There was also a difference in the notation ofmaternal and baby's outcomes in the maternalhealth cards of mothers with hypertensivedisorders in pregnancy. A significantly increasednumber of cards in the low maternal mortalityareas had this information missing from thecards.

Postdates:For postdates, 157 out of 293 cases with a similarcondition were chosen in the study districts. Theselection rate in the study was 53.5%. Thecharacteristics of the cases in both study areaswith regards to the colour codes were similar.However, there were three cases 0.4%) with ared colour code in the low mortality areascompared with none in the high maternalmortality area. With regards to case distribution,89.0% were at more than 41 weeks gestationalage in the low maternal mortality areas comparedwith 75.7% in the low mortality areas (p< 0.05).

The colour code assigned was inaccurate in morethan 56.8% of cases in both areas. The range ofcolours assigned ranged from red to white. Thecare given by diagnosis was inappropriate in themajority of cases. When the cases were morethan or equal to 42 weeks gestation, the casesshould have been referred to hospitals within 1day of being seen in the clinic. They should beadmitted and appropriate intervention planned.In 92.8% of cases in both study groups, this planwas not followed. For postdates, Table V also


showed that the management by the assignedcolour code was also found not appropriate inboth areas (96.9% and 98.6% in high and lowmortality areas respectively). Inappropriatemanagement by health personnel was lesspronounced in the high maternal mortality areascompared to that of low maternal mortality areas(65.0% versus 87.8%; OR=0.3, 95% CI=0.1-0.5).However, in contrast to the management byhealth staff, the level of inappropriatemanagement of postdates by the hospitalpersonnel in low maternal mortality areas waslower, 11.8% compared to 48.4% inappropriatehospital care of postdates in the high maternalmortality areas (OR=2.5, 95%CI= 3.5-7.).

Feedback for postdates was again poor with 125cases out of the 258 cases having no feedback.However the feedback received was significantlylower in the. high maternal mortality areascompared to that of the low maternal mortalityareas (30.9% versus 60.0%; OR= 3.4, 95% CI=1.83.8).

Table V showed that the outcomes for bothmothers and babies were better reported in thepostdate mothers compared to those withanaemia or hypertensive disorders of pregnancy.

Analysis of all three conditions combined:When all the three conditions were .analysedtogether, the following observations were made.The prevalence rates of the three conditionsstudied are as tabulated in Table III. The overallprevalence of these conditions was as usuallyquoted in the medical literature. However, theprevalence of hypertension and postdates wasnoted to be higher in the low mortality areas butthe prevalence of anemia was lower in the lowmortality area. This could be reflective of betterantenatal surveillance and care in the lowmortality areas.

Table IV summarises the selection rates in thetwo study areas and also details the problemswith data collection in both the health andhospital sectors alluded to earlier. In view of the

43

ORIGINAL ARTICLE

fact that antenatal care is a combined effortbetween the hospital and health sectors, theproblem· with feedback between the two sectorswill have to be addressed. Numerous studiespreviously have shown the same problem.

The assigned colour code was accurate in only43.9% of cases in the low maternal mortalityareas and in 44.2% of the cases in the highmaternal mortality areas. There was no statisticaldifference in the accuracy of assigned colourcode between the two areas. Table V also

showed that the level of inappropriate referrals tohospital was similar in the two areas.Inappropriate care by diagnosis and by assignedcolour code were however, significantly lower inthe areas with high compared to low maternalmortality. Prevalence of inappropriate care byhealth personnel in the high mortality areas wassignificantly lower compared to that of lowmortality areas but the opposite is found forinappropriate care by hospital staff in the twoareas (Tables V and VI).

Table I : Health Facilities Chosen from Randomly Selected Districts

Hospital

Selected District No. of No. of No. of No. of State District District

Health Health RBK/KD* RBK/KD* Hospital Hospital Hospital

Clinics Clinics sampled without with

sampled O&G O&G

Specialist Specialist

Best Area

Seremban 7 3 10 4 1 - -Melaka Tengah 10 5 14 4 1 - -

Worst Area

Kuala Selangor 5 3 22 6 - - 1

Bentong

larut Matang & 5 3 14 4 - - 1

Selama

11 6 29 7 - 1 -Total 38 20 89 25 2 1 2

*RBK: Midwifery Clinic, KD : Community CI.inic

44 Med J Malaysia Vol 58 No 1 March 2003


Table II: Appropriateness of Care

Recommended ManagementSelected Condition/ Definition. of Cases Health Sector Hospital SectorRisk FactorAnemia Hb <8gm/dl Refer to hospital Repeat Hb estimation and do

peripheral blood filmHb of 8-9gm/dl <32 weeks gestation

.refer to doctor at health clinic>32 weeks refer to hospital Patient is admitted and

investigation as well as treatmentinstituted

Hb of 9-1Ogm/dl < 32 weeks, treat withhaematinics by nursing staff>32 weeks refer to doctor

Hypertensive Disorder BP 140/90 without Refer to doctorin Pregnancy Albuminuria

gestation<36 weeks Admit Management inconsultation with MO/ specialistAnti-hypertensive used Plan fordelivery Anti-convulsant foreclampsia

BP 140/90 without Refer hospitalAlbuminuria immediatelyBP>140/90 with Refer hospitalalbuminuria and oedema immediatelyBP>160/110 with/without Refer hospitalalbuminurua immediatelyEclampsia Refer hospital

ImmediatelyPostmaturity Period of gestation Refer to hospital Confirm diagnosis

42 weeks and above within 1 day of Admit case fordetection appropriate intervention

Table III : Prevalence Rates of Selected Medical Conditions Among Mothers from Study Areas,Rate Per 100 Antenatal Mothers

Levels of Maternal Mortality

Selected Medical Low High Total

Conditions (N=4916)* (N=3472)* (N=8388)

f % f % f %Anaemia 311 (6.3) 251 (7.2) 562 (6.7)

Hypertensive disorders in pregnancy 169 (3.4) 88 (2.5) 257 (3.1 )

Postdates 202 (4.1) 91 (2.6) 293 (3.5)

* Number of antenatal cards sorted in the selected study areas

Med J Malaysia Vol 58 No 1 March 2003 45

,J>..

00

-;;

0

Tabl

eIV

:Sel

ectio

nan

dRe

spon

seRa

tes

ofA

naem

ia,H

yper

tens

ive

Dis

orde

rsin

Pre

gnan

e!'P

ostd

ates

and

Com

bine

dG5 Z

Med

ical

Con

ditio

nsin

Follo

w-T

hrou

ghSt

udy

from

Dist

ricts

with

Diff

eren

tLe

vels

0M

ater

nalM

orta

lity

» r- »A

neam

iaH

yper

tens

ive

Post

date

sA

ll3

cond

ition

s~ ()

Stat

usLo

wM

MR

Hig

hMM

RLo

wM

MR

Hig

hM

MR

Low

MM

RH

igh

MM

RLo

wM

MR

Hig

hMM

Rr-

en

No.

ofca

ses

iden

tifie

d31

125

116

988

202

9168

243

0

No.

ofca

ses

sele

cted

159

160

9046

109

4835

825

4

Sele

ctio

nra

te(

%)

(51.

1)

(63.

7)(5

3.2)

(52.

2)(5

3.9)

(52.

7)(5

2.5)

(59.

1)

No.

ofca

ses

with

mis

sing

10

73

40

123

info

rmat

ion

onm

anag

emen

t

byhe

alth

staf

f

No.

ofca

ses

with

mis

sing

10

/(4

7)

11

/(4

1)

18

/(7

7)

19

/(4

2)

17

/(9

8)

19

/(4

0)

45

/(2

22

)4

9/(

12

3)

info

rmat

ion

onm

anag

emen

t

byho

spita

lst

aff*

~N

oof

case

sw

ithm

issi

ng66

1619

610

395

25CD

info

rmat

ion

onbi

rth

outc

omes

Q..

'- ~ 0

Hos

pita

lca

rds

not

trac

eabl

e15

07

512

08

1027

-< V>

(Bas

edon

wri

tten

com

men

ts)

o· ~ 01 ex>

*am

ong

thos

ew

ere

adm

itte

din

hosp

ital

sZ 0 ~ ~ 0 0 ::

ri'

-) 0 0 w

~Ta

ble

V:Di

Here

nces

inC

olor

Cod

inp

Prac

tice

and

Case

Man

agem

ent

inD

iHer

entT

ypes

ofC

ondi

tions

Base

don

Follo

wCD C

LTh

roug

hSt

udy

0Ca

ses

from

Dist

ricts

with

DiH

eren

tLe

vels

ofM

ater

nalM

orta

lity

'- ~ c

Ane

amia

Hyp

erte

nsiv

ePo

stda

tes

All

3co

nditi

ons

c '<Lo

wM

MR

Hig

hM

MR

Low

MM

RH

igh

MM

RLo

wM

MR

Hig

hMM

RLo

wM

MR

Hig

hMM

Ren

Stat

usc' ~

(n=3

09)

(n=2

45)

(n=1

69)

(n=8

3)

(n=2

02)

(n=8

1)(n

=6

80

)(n

=40

9)U

1A

ssig

ned

colo

urco

deC

D ZR

ed(6

.9)

(2.9

)(1

9.7)

(9.6

)(1

.4)

(0.0

)(8

.4)

(3.7

)0

Yel

low

(70.

9)(7

0.3)

(70.

2)(7

4.7)

(39.

4)(5

3.9)

(61.

4)(6

7.7)

~G

reen

(21.

0)(2

3.8)

(8.9

)(1

5.7)

(48.

0)(3

4.6)

(26.

0)(2

4.1

)0

c0..

:.()

Whi

te(1

.2)

(1.2

)(1

.2)

(0.0

)(1

1.2)

(13.

6)(4

.2)

(3.4

):T

~'"

Not

code

d(0

.0)

(1.9

)(0

.0)

(0.0

)(0

.0)

(0.0

)(0

.0)

(1.2

)0 0

0w

0

Mea

sure

sof

care

~ ;AJ

cO'

Acc

urac

yof

assi

gned

colo

urco

de::

:r :c;;

Acc

urat

e(4

2.6)

(45.

5)(5

4.0)

(41.

4)(3

7.5)

(43.

2)(4

3.9)

(44.

2)»

Inac

cura

te(5

7.4)

(54.

5)(4

5.9)

(58.

6)(6

2.5)

(56.

8)(5

6.1

)(5

5.8)

=> rT1

~ cAppropria~ness

ofre

ferr

alto

0 g.ho

spita

l=> g.

,A

ppro

pria

te(9

.9)

(16.

2)(4

3.6)

(64.

7)(6

9.3)

(64.

6)(3

5.9)

(35.

6)3

'-

Inap

prop

riat

e(9

0.1

)(8

3.8)

(56.

4)(3

5.3)

(30.

7)(3

5.4)

(64.

1)(6

4.4)

CD ()

0 0 c:A

ppro

pria

tene

ssof

care

.., ()

bydi

agno

sis

0 0....

3'

App

ropr

iate

(40.

4)(5

9.7)

(14.

2)(3

3.5)

(7.2

)(7

.2)

(24.

0)(4

4.0)

co

Inap

prop

riat

e(5

9.5)

(40.

3)(8

5.7)

(66.

5)(9

2.8)

(92.

8)(7

5.9)

(55.

9)U

>'ii

i CD 3

App

ropr

iate

ness

ofca

reby

Qas

sign

edco

lour

code

» => CDA

ppro

pria

te(8

4.8)

(82.

2)(2

4.8)

(21.

5)(1

.4)

(3.1

)(4

5.1)

(54.

2)=> 0

Inap

prop

riat

e(1

5.2)

(17.

8)(7

5.2)

(78.

5)(9

8.6)

(96.

9)(5

4.9)

(45.

8)c

App

ropr

iate

ness

ofca

regi

ven

by()

0

heal

thpe

rson

nel

(i! 3'

App

ropr

iate

(75.

5)(8

1.4)

(58.

1)(7

4.7)

(12.

2)(3

5.2)

(47.

6)(7

0.9)

~In

appr

opri

ate

(24.

4)(1

8.6)

(41.

9)(2

5.3)

(87.

8)(6

5.0)

(52.

4)(2

9.2)

0'

./0>.

'iii

"o'

~Ta

ble

VI:

Estim

ated

Odd

sR

atio

and

95%

Con

fiden

ceIn

terv

alof

Ina~propriate

Man

agem

entO

utco

mes

inH

igh

Com

pare

dCD 0

-<

-to

Low

Mat

erna

lMor

taity

Are

as~ 0 c

Ane

amia

Hyp

erte

nsiv

eP

ostd

ates

All

3co

nditi

ons

'< '" o'P

oor

outc

omes

OR

95

%C

IO

R9

5%

CI

OR

95

%C

IO

R9

5%

CI

~ tl'1

Inac

cura

teco

lour

codi

ngpr

actic

e0

0 Z(l

ow

MM

R)

0

Hig

hM

MR

0.9

(0.6

-1.

3)1

.7(1

.0-

3.0)

0.8

(0.5

-1.

4)1

.0(0

.8-

1.3)

~0

0c.:.

(JIn

appr

opri

ate

refe

rral

:::r

~l0

(lo

wM

MR

)0

00 w

Hig

hM

MR

0.6

(0.3

-0.

97)

0.4

(0.2

-0.

8)1.

3(0

.7-

2.3)

1.0

(0.8

-1.

3)0 ~ ;;

u

Inap

prop

riat

eca

reb

ydi

agno

sis

cO'

::r

:a;

(lo

wM

MR

)»

Hig

hM

MR

0.5

(0.3

-0.

7)0

.3(0

.2-

0.6)

0.9

(0.3

-3.

1)0

.4(0

.3-

0.5)

OJ rn < 0

Inap

prop

riat

eca

reb

yas

sign

edc 0 g

colo

urco

deO

J

(lo

wM

MR

)Q.

..,3'

-H

igh

MM

R1

.2(0

.8-

2.0)

1.2

0.6

-2

.40

.4(0

.1-

3.0)

0.7

(0.5

-0.9

)CD n 0

Inap

prop

riat

eca

reb

yhe

alth

staf

f)0 S;

(lo

wM

MR

)n 0

Hig

hM

MR

0.7

(0.5

-1.

1)0

.5(0

.3-

0.9)

0.3

(0.1

-0.

5)0

.5(0

.4-

0.6)

c...

5'

cc enIn

appr

opri

ate

care

by

hosp

itals

taW

i5i CD

(lo

wM

MR

)3

Hig

hM

MR

8.5

(2.2

-47

.1)

0.5

(0.2

-1.

1)6

.9(3

.2-

15.1

)2.

5(1

.6-

3.9)

Q » OJ CD

Poo

rfe

edba

ck**

OJ

0

(lo

wM

MR

)a- n

Hig

hM

MR

2.9

8(1

.45

-6.

17)

1.6

3(0

.79

-3.3

8)3.

41(1

.85

-6.

30)

2.6

5(1

.83

-3

.84

)0 Cil 5

'(T

heco

ntro

lo

rre

fere

ntgr

oups

are

case

man

agem

ent

outc

ome

mea

sure

sin

area

sw

ithlo

wm

ater

nal

mor

talil

y)~

*(B

ased

on4

33

case

sad

mitt

edto

hosp

ital)

0 "'<

,J:>.

**(B

ased

on6

40

case

sre

ferr

edo

rad

mitt

edto

hosp

ital)

(J>

-0c'

ORIGINAL ARTICLE

Discussion

While large prospective pregnancy follow-upstudies were done in post-industrialised countries,4,5 information on developing countries is morerecent. 1 Hospital delivery was recommendedbased on the poor predictability of complications4,

5. Yet in a worldwide effort to improve opstetricresults, countries with high perinatal and maternalmortality adopted hospital referral based on riskfactors which were mainly identified antenatally7,9Malaysia has adopted the principle of hospitaldelivery 'except for those with white codes. InMalaysia about 40% of pregnant womeh aredefined as high risk9.

The occurrence of the antenatal complications ofhypertensive disease in pregnancy, anemia andpostdates did not differ from internationallyexpected figures1o,1l. This demonstrates that thesample chosen for this study was reflective of astandard· obstetric population.

The colour coding system was initiated in1989.This study was performed in 1997 that isafter 8 years of use of this system. It is assumedthat with the time that this system has been used,all staff would have been trained and becomefamiliar with the intricacies of the colour codingsystem. However, it is possible that lack ofsupervision by superior officers and rapidchangeover of staff may hinder the accurateimplementation of this system. It is a significantfinding of the study that the colour codes werecorrectly assigned in only 56% of all mothers withthe 3 conditions with the percentage of correctcoding varying between 37.5% for postdates to54% for hypertension in the low maternalmortality areas. It is important to note that thecolour. coding system allowed for recoding ateach visit and a prescribed format which alloweda checking system was attached to the antenatalcard of each mother. Despite this system thatallowed correction by both the nursing staffproviding antenatal care and the supervisory staff,the accuracy of the colour coding system waspoor. Similar reliability problems have been

50

found elsewhere12• Despite recoding, predictability

improved little and would improve only if codingduring labour were addedll

• The colour codingsystem in Malaysia allowed for codingintrapartum problems as well as postpartumevents. In the present study, only hypertensivedisease of pregnancy among the three conditionsstudied had intrapartum risk factors which couldbe coded. However, the accuracy of the codes didnot show a significant difference in hypertensivedisease of pregnancy as compared to postdates oranemia in pregnancy. This study confirms that thecolour coding system which is time consuming isnot sufficiently effective in prediction of risk. Thisis the finding as well in numerous otherstudiesI4,15,16. It has been stated that a system witha sensitivity of 50% and specificity of 73%, with apopulation coverage of over 90% may improveperinatal statistics in areas with a high rate ofcomplications. We have to concur withGeethuysen et al. who state that with the levelsseen in Malaysia, predictive values of this systemare lowl7

.

However, one must realize that the system wasnot meant to predict risk but be a managerial tool.It allowed a midwife the flexibility to admit apatient to a specialist unit without hindrance andallowed a patient with a problem to be evaluatedspeedily by a specialist. The value of the systemas a managerial tool in this respect was notevaluated by the study.

The above data indicate that the colour codingsystem did not account for the difference in thestudy areas with regards to maternal mortality.Inaccurate colour coding practice was the same inboth areas. This led to inappropriate care bydiagnosis, by assigned colour code and by healthstaff. This was noted in all the three conditionsstudied in this report. When the level ofinappropriate care by diagnosis, by assignedcolour code and by health staff for the threeconditions combined was compared between thehigh and low maternal mortality areas, it wasnoted that these were significantly higher in thehigh mortality areas.

Med JMalaysia Vol 58 No 1 March 2003

Did We Do it Right? . An Evaluation of the Colour Coding System for Antenatal Care in Malaysia

When the outcomes of the pregnancies wereanalysed by the various colour codes and thedifferent conditions, it was noted that there wereno maternal deaths in the various colour codes inthe hypertensive or postdated patients in both thehigh risk and the low risk areas. The two maternaldeaths in this study occurred in patients who werecoded green for a haemoglobin level of 9-10 gldlin a high maternal mortality area. The only fetaldeaths were in yellow coded patients for Hb levelof 8-9 gldl and on wrongly coded hypertensivepatients with BP>160/110 in a high mortality area.There was no relationship of the number ofmaternal or fetal complications to the colour codein this study. Thus the colour code did not predictthe possible eventual outcome of the pregnancyfor both the mother and the baby. The colourcoding system may also divert attention from theapparently low risk patients to their own detrimentas was demonstrated in the maternal and fetaldeaths that occurred in the low risk patients.

The study also demonstrated the oft-found findingof the lack of feedback between the hospital andhealth sectors that provide care for the pregnantmother. This study was carried out before theintroduction of the patient held home basedantenatal card. The antenatal card which affordsthe opportunity .of recording the details ofantepartum, intrapartum and postpartum care byall sectors providing maternity care would haveaddressed some of these concerns aboutfeedback. It is nevertheless opportune toconcentrate resources and effort in overcomingthis continuing bugbear of maternity care.

Antenatal care is often the area where services aresometimes not focused in the most appropriate orconsistent manner. In some instances, decisionsabout tests, interventions and surveillance arearbitrary or varied between professionalsl8. Thecolour coding system has not demonstrated itsusefulness in the effective management of threecommon obstetric problems in the maternity careservices in this country. Marion Hall and hercolleagues have highlighted the possible oversurveillance of the woman during the antenatal


period19• ' If these checkups are unwarranted, it is

not only a waste of resources, but could impactadversely on the standards that can be achieved.If fewer visits are necessary, it should be easier toplan the workload of individual members of staff,to increase both the level of continuity achievedand the time spent on each visit, therebyimproving the quality of care.

Some women used to the more traditionalantenatal visiting patterns may feel anxious if theyhear of plans to reduce the number of visits.Time should be taken to explain that a reductionin antenatal visits will not mean any less supportin pregnancy, but rather that the quality andcontinuity of care will be improved and waitingtimes reduced. The system of antenatal carecurrently in place is cumbersome and a source of .concern to professionals and pregnant womenalike. Some antenatal patients in this country mayreceive as much as 20 visits to the privatespecialists, health sector and to the governmenthospitals. A national health-financing schemewhich rewards only appropriate care may be away of correcting this defect in our system. Asurvey of existing practice should be undertakenby purchasers, providers and professionals onaregular basis. As well as reviewing the number ofantenatal checkups, the effectiveness of theservice in identifying the fetus at risk should alsobe assessed. Streamlining the current systemcould achieve improvements for all concerned.

This present study randomly sampled 1112mothers out of 8388 mothers with three commonobstetric problems in five districts in Malaysia,which were stratified according to high and lowmaternal mortality rates. The study showed thatthe prevalence of anemia, hypertensive disordersin pregnancy and postmaturity among motherswith these conditions were according to knowninternational standards.

Anaemia was classified as severe and red codedin 18.7% of the mothers with anemia. It wasclassified as moderate (yellow code) in 40.5% andmild (green code) in 40.2%. With regards to

51

ORIGINAL ARTICLE

hypertension in pregnancy, 56.6% were redcoded and the others (43.4%) were classified asyellow. For postdates 14.8% were red codedwhile the rest were coded yellow. It could thusbe seen that for hypertension in pregnancy, theexisting coding system tended to classify morethan half of the patients with this condition in acategory that required hospital admission. Thiscould have led to the inappropriate managementand admissions due to changes in the protocol atlocal level to reduce the number of admissions.The classification of the disease tended to beunnecessarily complex and not supported by theeviaence in the literature.

The colour coding system, as it exists now shouldbe reviewed. Instead, a substantially revisedsystem that takes cognisance of evidence in thescientific literature should be used to devise amore effective system that can be used by healthcare personnel involved in antenatal care toensure appropriate level of care and referrals.

It is also suggested that any screening system thatis implemented in future should continue to havecoding facilities for intrapartum and postpartumfactors to improve sensitivity. The coding systemshould be jointly developed and implemented byboth the health and hospital sectors to overcomethe weaknesses identified in the present system.

One particular aspect and weakness that needs tobe corrected is the overcoding of some conditions

52

that would require admission in more than half ofthe mothers with some conditions such ashypertensive disease of pregnancy.

The lack of feedback between the hospital andheath sectors was again found in this study. It isopportune to study whether the introduction ofthe home-based antenatal card has addressedsome of the concerns in this continuing bugbearof maternity care.

This study has provided some definitive answersfor policy development. The study confirms otherMalaysian and world experience in showing thelack of effectiveness of the risk coding system 22,23

as it is presently used. Attention to the details ofantenatal care and streamlining communicationbetween the providers of maternity care ratherthan spending time on coding should assure thatthe Malaysian reproductive care programme willcontinue to improve. We need to heed our ownadvice with regards to the findings of this study.

Acknowledgements

This study was funded by a grant from the UnitedNations Population Fund (UNFPA). The authorswould like to thank the Director-General ofHealth of Malaysia for permission to publish thisarticle. We would also like to thank the numerousnursing staff who made this study possible.

Mad J Malaysia Vol 58 No 1 March 2003


1. Alexander S, Keirse MJNC. Formal risk scoringduring pregnancy In Effective care in pregnancyand childbirth. Chalmers I, Enkin M, Keirse MJNC(eds) Oxford; Oxford University Press 1993: 34565.

2. Lambotte R. Discussion. In : Pre-term LabourProceedings of the Fifth Study Group of the RoyalCollege of Obstetricians and Gynaecologists.Anderson A, Berad R, Brudenell JM, Dunn PM.(eds) London: Royal College of Obstetricians andGynaecologists; 1977: 40-41.

3. Newcombe RG, Chalmers 1. Assessing the risk ofpreterm labour. In: Preterm Labour. Elder MG,Hendricks CH (eds) London: Butterworths, 1981:47-60.

4. Butler NR, Alberman Ed. (eds) Perinatal Problems.Edinburgh: E&S Livingstone, 1969.

5. Chamberlain G, Zander 1. (eds) Pregnancy Care inthe 1990's. Proceedings of a Symposium of theRoyal Society of Medicine 1991. Casterton Hall:Parthenon Publishing Group, 1992.

6. Kiely M. (ed) Reproductive and PerinatalEpidemiology, Florida: CRC Press, 1991.

7. Backett EM, Davies AM, Petros-Barvazian A. Therisk approach in health care: with special referenceto maternal and child health, including familyplanning. Public· Health Papers No 76. Geneva:World Health Organisation, 1984.

8. Hertz B, Measham AR. The safe motherhoodinitiative: proposals for action. Washington DC:World Bank, 1987.

9. Somboonsook B, WakermanJ, Hattch CT, CollosonM, Barnes A, Kyi W, Karim R. An initial assessmentof the risk approach to antenatal management inMalaysia. Med J Malaysia 1995; 50 (2): 115-21.

10. Chalmers I,Enkin M, Keirse MJCN. Effective care inpregnancy and childbirth. Oxford: OxfordUniversity Press, 1989.


11. Llewllyn-Jones D. Fundamentals of obstetrics andgynaecology: Vol 1 Obstetrics. 5th Edition.London,Sydney: Faber and Faber, 1990.

12. Hutchinson BG, Milner R. Reliability of the guide topregnancy risk grading of the Ontarioantenatalrecord in assessing obstetric risk.Canadian Med Ass J 1994; 150: 1983-87.

13. Bartlett A III, Paz de Bocaletti E, Bocaletti MA.Reducing perinatal mortality in developingcountries: High risk or improved labourmanagement? Health Policy Planning 1993; 8: 36068.

14. Chamberlain RW. Strategies for disease preventionand health promotion in maternal and child health:The ecologic versus the high risk approach. J PubHealth Policy 1984; June: 185-97.

15. Wall EM. Assessing obstetric risk: A review ofobstetric risk scoring systems. J Fam Practice 1988;27: 153-163.

16. Tinker A, Daly P, Green C, Saxenian H,Lakshminarayanan R, Gill K. Women's health andnutrition: Making a difference. World BankDiscussion Papers No 25.Washington DC: WorldBank, 1994.

17. Geethuysen C], Isa AR, Hashim M, Barnes A.Malaysian antenatal risk coding and the outcomeof pregnancy. J Obstet Gynaecol Res 1998; 24 (1):13-20.

18. Changing Childbirth: Report of the ExpertMaternity Group. Department of Health, UnitedKingdom. London HMSO 1995.

19. Hall MH. Critique of antenatal care. In ObstetricsTurnbull A, Chamberlain G (eds) ChurcillLivingstone Edinburgh 1989 1st edition: 225-33.

53

did we do it right? - an evaluation of the colour coding ... · did we do it right? -an evaluation...

Documents