Sot Scz Med Vol 36, No 10. pp 1331-1334, Prmted in Great Bntam All nghts reserved
1993 CopyrIght
6
0277-9536/93 $6 00 + 0 00 1993 Pergamon Press Ltd
REGIONAL INEQUALITIES OF CHILD MORT4LITY IN PENINSULAR MALAYSIA WITH SPECIAL REFERENCE TO THE DIFFERENTIALS BETWEEN PERLIS AND KUALA TERENGGANU* UWE BREHM Umverslty
of Kassel,
FB 6, Department
of Geography,
P 0
Box 101380, 3500 Kassel,
Germany
Abstract--In Peninsular Malaysia child mortality rates (5q,,) vary from I3 to 63 per thousand at dlstrlct level The spatial pattern IS closely associated with the reglonal dlstrlbutlon of soc1o-economx factors But due to multlcolhneanty it 1s difficult to isolate the Influence of soclo-economic variables from other variables by employmg aggregated data However, mdlvldual data collected m a case-control-study that was conducted m Perhs and Kuala Terengganu confirm the important role of soclo-economic factors So It should be possible to achieve a further reduction of child mortality by ralsmg the mcome and educatlonal level of the under-pnvdeged groups Apart from that, as the case of Perhs shows, the provlslon of family planmng and preventive mechcal services may also contribute to lower child mortality independent from socio-economic changes But, as the comparison with Kuala Terengganu shows, the utlhzatlon of family planmng and preventive medical services IS not only influenced by the accesslblhty to, but also by the soclo-culturally determined acceptabdlty of such services Key wordy--child
mortality,
regional
dlfferentlals,
Malaysia
INTRODUCTION
According to the scope of population geography [1] it IS the aim of this paper to describe and attempt to explain the regional dlfferentlals of child mortality m Peninsular Malaysia Apart from that, the results might help to find strategies that may reduce mortality m general and regional mequalities of mortality m particular [2] In this paper the analysis 1s limited to chdd mortality which 1s defined as the probability of dying before reaching the age of five Macro
analym
The 78 admmlstratlve districts of the penmsula are the smallest area1 units for which figures on live births and child deaths are available by place of residence Internal consistency checks and comparisons with other data sources lead to the conclusion that these statlstlcs are fairly reliable Under-enumeration 1s estimated to be less than 5% [3] Furthermore a 3 years average was calculated m order to muumlze random varlatlons So the standard error of the child mortality rate for medium sized districts lies between one and two per thousand The mean child mortality rate for the whole peninsula was 25 per thousand In
1982.-84
This
1s considered
as
relatively
low _
*This paper 1s part of the author’s research project “Population Development m Penmsular Malaysia” sponsored by the VW-Foundation TAccordmg to UNICEF figures the world-wide lowest under-five-mortahty-rate of 8 per thousand IS registered m Sweden, whereas the highest rates of about 300 per thousand are estimated for Afghanistan, Mall and Sierra Leone (The Stale of the World’s Chddren 1987, Table 1)
compared to the international standard t However, regional mequahtles wlthm Peninsular Malaysia are remarkable At district level ,q,-values vary from 13 per thousand m Petalmg to 63 per thousand m Ulu Kelantan Figure 1 shows the spatial pattern The lowest child mortality rates are concentrated m the Klang Valley Rates below average are also found m the surrounding dlstrlcts of the central part of the West Coast and m the whole south of the pemnsula, whereas most of the dlstrlcts of the northern part of the West Coast--except Pmang and Perh-show rates slightly above average In contrast to that, almost the whole East Coast IS characterized by rather high child mortality rates Finally, the highest death risks are observed m the interior parts of the northern states This spatial pattern of chdd mortality shows many slmllarltles to the regional dlstrlbutlon of various indicators representing the blo-physical environment, socio-cultural environment, socio-economic structure and medical infrastructure The highest correlation coefficient of r = -0 91 IS observed between child mortality and the index of the soao-economic structure In conJunctlon with the fact that soclo-economic mdlcators have also been identified as the strongest predictors of child mortality dlfferentlals m other countries [4], this close statlstlcal assoclatlon gives rise to the supposltlon that the geographical mequalltles of child mortality m Pemnsular Malaysia are mainly due to the regional dlstrlbutlon of the various soclo-economic groups But correlations are also very high between child mortality and the other mdlcators Moreover, most of the indicators are highly
1331
UWF
1332
khEHh4
PENINSULAR
MALAYSIA
bud
PENTING ‘g(P-8.
Child
Rate < 17 17-24
Y
.%a,,~. Own u*ul-s
FIN
I Child mortahty by dlstrlcts
Inter-correlated Therefore It 1s lmposslble to isolate the Influence of Just one variable on child mortality by employing aggregated data * Deeper mslght can only be gamed with the help of mdlvldual data MULTI-LEVEL
on PERANGWN
follow the various paths on which geographical mequahtles might be explamed Direct effects of regional variables (e g harmful substances) ds well as
ANALYSIS
By collectmg mdlvldual data it 1s possible to gather mformatlon on characterlstlcs of the child itself as well as on characterlstlcs of the household and the region where the child lives That all these levels have to be consldered by analysmg child mortahty dlfferentlals has been recogmzed m the demographic literature m the 1980s Therefore various models contammg determmants of child mortahty at different levels have been developed [5] The multi-level model presented m Fig 2 IS based on the above mentioned models but shghtly modified m order to lay the emphasis on the explanation of reglonal dlfferentlals of child mortality So It IS possible to --__ *A detaded descrlptlon of the constructlon of the various mdxators as well as the further results and the methodologlcal problems of this approach will be found m Ref [3, Chap 6 31
+ Health Behavlor
I
Fig
1
II
Li
Child Mortality
2 Multi-level
model of the determmants mequahtles of chdd mortahty
1 of reglonal
1333
Regional mequalltles of child mortahty m Peninsular Malaysia Table 1 Chdd mortahty by selected vambles m Perhs and Kuala Terengganu
I Vanable Birth order Utrhzatmn of preventwe medical serv,ces Mother’s education Household mcome Region
Categones 0= I= 0= I= 0= I= 0= I= 0= I=
parity l-5 higher panty &6 antenatal check-ups 7 + antenatal check-ups none or pnmary educatmn secondary or tertiary education possession of c-8 Items4 possewon of 9-16 Items Perhs Kuala Terengganu
fKT2
5%’
R,
(%)
(Pt)
89 I1 28 72 50 50 53 41 100 0
73 27 67 33 58 42 54 46 0 100
25 42 35 22 35 20 33 23 24 33
Source Own Survey 1987 ‘Frequency m Perk (%) 2Frequency In Kuala Terengganu (%) jChdd mortahty rate (per thousand) 4Among a hst of I6 household Items
Indirect effects through the unequal dlstrlbutlon of soclo-economic factors or through a regionally dlfferentlated health behavlour are conceivable In the followmg this model will be used to explam the mequahtles of child mortality between Perhs and Kuala Terengganu The north-western district of Perhs was chosen as one study area because of its low child mortality compared to its soclo-economic structure * Therefore it seems to be particularly interesting to identify the specific circumstances under which the unexpectedly low child mortality rates are achieved m Perhs The East Coast district of Kuala Terengganu was chosen as the second study area because its soclo-economic structure IS similar to that of Perhs, but its child mortality rate 1s as high as expected In both districts the mdlvldual data were collected m a case-control-study m 1987 Cases were all Malay children who died m 1985 before reaching the age of five m Perhs and one third of the Malay children, who died m Kuala Terengganu Controls were a sample of 7% and 2% respectively of the survlvmg Malay children t Totally 189 mothers of dead and 538 mothers of survlvmg children were interviewed The questionnaire consisted of 112 standardized questions that were derived from the multi-level model presented m Fig 2 Among numerous variables obtained from the survey the five variables listed down m Table 1 have a strong impact on child mortality $ After controlling for the other four variables m a multiple regression analysis with dummy variables
*Perk IS the dlstrlct with the highest negative deviation from the regression lme that represents the assoclatlon between the soclo-economic structure and the child mortahty rate of the 78 dlstrlcts tDue to their small numbers m both study areas Chmese and Indian children were not included m the sample SDetalled results of the assoclatlon between all variables and child mortality will be published m Ref [3, Chap 71 §Due to the complicated study design it IS lmposslble to calculate exact statlstlcal levels of slgmficance In Fig 3 only those effects are represented that lead to a change of the child mortality rate of at least 5 per thousand and a change of the intervening variables of at least 5%
there 1s no independent effect of the region on child mortality, whereas the impact of the other vanables-although slightly weakened-remains slgmficant According to the results of a path analysis the child mortality differentials between Perhs and Kuala Terengganu are mainly due to the fact that births of a high parity are more frequent and the utlhzatlon of preventive medical services 1s less frequent m Kuala Terengganu (see Fig 3) Apart from that, the proportion of females with higher education isalthough the educational level of the whole population 1s similar m both districts-slightly lower m Kuala Terengganu QThis also leads directly as well as mdlrectly to a higher child mortality These results lead to the conclusion that child mortahty rates m Kuala Terengganu could be reduced to the low level observed m Perhs, if the women m Kuala Terengganu expenenced higher education, gave birth to not more than five children and used the preventive medical services more frequently A prerequisite to reach this goal IS the provlslon of a sufficient infrastructure of schools, family planning and maternal and child health clinics This kmd of infrastructure 1s indeed slightly higher developed m
Region
+ T I
Bllth
Order
-
I
-
Child Mortality
I
Fig 3 Path model to explain the child mortality dlfferentlals between Perhs and Kuala Terengganu
1334
Uw
Perhs than m Kuala Terengganu, but the services are also widely avallable m Kuala Terengganu * So the less frequent nal also
and
uttltzatton
chtld
seems
health
to be due
of famtly servtces
plannmg m Kuala
to a lower
and
mater-
Terengganu
accepttblltty
of such
services This lower acceptlblhty might be determined by the specific soclo-cultural environment m the East Coast States of Kelantan and Terengganu, where Adat (Malay tradltlons) and Islam play a more Important role than anywhere else m the country But before any further conclusions can be drawn, m-depth knowledge has to be collected on regionally dlfferentlated attitudes of the population concernmg family plannmg and maternal and child health REFERENCES
1 Bdhr J Bevolkerungsgeographle Entwlcklung, Aufgaben und theoretlscher Bezugsrahmen Geographlsche Rundschau 40, 613, 1988 2 Fourth Malaysra Plan 198161985 National Prmtmg Department. Kudld Lumpur, 1981
*e g There dre 7 health centres and 29 rural health and midwife chmcs for d population of 145,000 (1980 census figure) m Perhs About the same number of health faclhtles are accessible for a population of 233,000 m Kuala Terengganu
BREHM
3 Brehm U Regionale Dispantaten West-Malaysia Bremer Aslen-Pa&k
der Mortahtat Studlen,
m
Band 9,
Munster, 1992 4 Proceedings of the Meeting on Soclo-Economic Determmants and Consequences of Mortality 1979 Umted Nations and World Health Orgamzatlon, New York,
1980 5 Pallonl A Design problems and data collectlon strutegles m studies of mortahty dlfferentlals developmg countries In ICJSSP Dakar 1981, Liege. pp 109%141, 1984 BIBLIOGRAPHY
Mahadevan K Mortality, biology
and society analytical framework and conceotual model In Fert&v and Mortalrfy (Edited by Ma’hadevan K ), pp 2391301 Sage, New Delhi. 1986 Mosley W H and Chen L C An analytical framework for the study of child survival m developmg countries Pop Det Rev 10, suppl, 25-45, 1984 _ Report of the United Natlons World Population Conference 1974 Umted Nations Publication, 1974 SOCIO-Economy Drfirentlals m Chrld Mortalrfv m Developmg Counrrres United Natlons, New York. 1985 The Slate of the World’s Chrldren 1987 UNICEF, New
York, 1987 Van Norren B and Van Vlanen H A W The Mulnurrllton Infections Syndrome and Its Demographic Outcome m Deueloprng Countrres Programmg Committee for
DemographIc 1986
Reference,
Publication
No 4, The
Hague.