We gratefully acknowledge. the assistance of the HSRC JointProgramme for Health and Social Development, Dr K Naidoo inreviewing the manuscript, Mr P Pillay and Ms E Gouws forstatistical support, and Drs P D Ramdas and K N Valtabhjee formoral support.

REFERENCES

1. A National Health Plan for South Africa. Prepared by the ANC Health Departmentwith the technical support of WHO and UNICEF. Johannesburg: ANC, 1994.

2. Green A. An Introduction to Health Planning in Developing Countries. Odord:Oxford University Press, 1994.

3. National Committee on Academic Health Service Complexes. Final Report.PretOria: Department of Health. December 1994.

4. The South African Medical and Dental Council. Memorandum and Ouestionnaireon Postgraduate (Specialist} Education and Training in Medicine. Pretoria;SAMDC. 22 March 1995.

5. Department of National Health and Population Development. Registered HeaJrhManpower per Geographical Region in the RSA 1991/1992. Pretoria: Oepanmerltof National Health and Population Development. 1994.

6. Snut S. Maughan-8ro'Nn DA. A $OClo-economic and c:!~raphicprofile of thepopulation of the study area: 1980-2015.

7. Edginton ME, Hoist HL Doctors In rural hospitals in Kwalulu and Natal. S AirMed J 1991; 80: 511-512.

S. Draft WONCA policy on miming for rural practice. 1994.9. Jaques PH. Rural medIcine - A speclalty in its own right (OpInion). S Air Med J

1992; 81: 589-5511.10. Magnus JH. Tottan A. Aural doctor recruitmenl: does medical education In rural

districts recruit doctors to rural ateas? Med £duc 1993: 27: 250-253.11. Ktistiansen IS. Forde OH. Medical specIalists' choice of location: the role of

geographical attachment in Norway. Soc Sei Med 1992: 34: 51-62.12. Kassebaum OG. Szenas PLo Aural sources of medical students. and graduates

choice of rural practice. Acad Med 1993; 6S: 232-236.13. Kamien M, 8uttfield IH. Some solutions to the shortage of genllfal practitIoners in

rural Australia. Med J Ausr 1990; 153: 105-108; 112-114; 168-171.1~. Report of the Committee on Human Resources for Health. Draft submitted to the

Minister of Health, December 199~.

15. Crisp N. Human Resource Policy for Health (Prepared for MedIcal Association ofSouth Africa). Pretoria: Oeloine and Touche, February 1994.

16. African National Congress. The Reconstruction and Development Programme. APolicy Frameworlc. Johannesburg: ANC, 1994.

Accepted 30 Aug 1996.

_ Volume- 87 No.3 March 199i SA,_~J

v

Violence and injurymortality in the Cape TownmetropoleLeonard B lerer, Richard G Matzopoulos,

Rozette Phillips

Objective. To describe accurately the violence and injury

mortality in a South African city and demonstrate the utility

of secondary data sources to identify injury control prioritie~

Design. Cross-sectional analysis of medicolegal

laboratory (state mortuary), forensic and police data.

Setting. ~politan Cape Town, 1994.Results. Non-natural causes (deaths due to homicide,

suicide, accidents and undetennined caJ:Jses) accountedfor almost 4 000 deaths, which comprised approximately

30% of all-cause mortality during 1994. The five main

violence and injury mortality categories were: homicide

(1 789 cases; 46% of all non-natural mortality), transportaccidents (1 130 cases; 29% of all non-natural mortality),

fire (295 deaths; 8% of all non-natural mortality), suicide

(291 deaths; 7% of all non-natural mortality) and drowning

(96 cases; 2% of all non-natural mortality).

Conclusions. Priority issues in injury control include the

increasing homicidal and suicidal use of firearms, road

and rail commuter injury and the spatial distribution of

injury. Surveillance, based on non-natural mortality, should

be included in local, regional and national healthinfonnation systems.

S At, Med J 1997; 87: 298-301_

In South Africa, homicide, suicide and transport accidentscontribute substantially to the burden of disease, especiallyamong the poor and disadvantaged.' The quality of nationalmortality data is a source of concern2'" and estimates of thenumber of deaths due to 'undetermined causes' vary from32% to approximately 60%.s.e

Good-quality mortality data are regarded as fundamentalto health resource allocation decisions, especially in thepresence of substantial disparities in health indicatorsbetween various groupS.l In this context, cities such asCape Town have significant differences in health statusbetween suburbs, and qualitative review of mortality datareveals the effects of poverty and deprivation.u While littlework. has been undertaken on intra-urban drtferentials inviolence and injury incidence, earlier research has .focusedon differences in infant mortality rates between various

------~~

Health Consulting Office, Health Technology Research Group,Medical Research Council, and Department of Forensic Medicine,University of Cape Town

leonard B lerer. MS ChB, DFM (SAl. BSc Hens (EpidenuoI). MMed Pam {forenaj

Richard G Matzopoulos, BBus So

Rozette Phillips. M8 Ch8

Jburbs of Cape Town.'o-,z This article aims to provide an::curate cross-sectional perspective on a single year'sortality from violence and injury in South Africa's second­

J"Qest city. It is hoped that this will stimulate interest inlortality surveillance in other parts of the country and raisewareness of injury control as a heatth priority.

V1ethods

Jata sources1formation was extracted from death registers, autopsysports and ancillary police and laboratory documentationrom the two medicolegal laboratories (state mortuaries) at,alt River and Tygerberg for the period 1 January 1994 to31 December 1994. All deaths in South Africa due to otherhan natural causes require examination by a district surgeon,orensic pathologist or medical practitioner, in terms of thenquests Act.13 As no legal definition of non-natural mortality~xists, it is current practice to admit to the medicolegalaboratories those in whom it is impossible to determine the:ause of death. Deaths while under the influence of local or]eneral anaesthetic or where the anaesthetic could have been3. contributory factor are regarded as due to non-natural-:auses.'~The Cape Town metropole historically comprises the)1 statistical region including the Kraaifontein area and, for:he purposes of the capture of non-natural mortality data, is'egarded as the area served by the medicolegal laboratories,t Salt River and Tygerberg.

Key variablesAges of the deceased were extracted from the deathregisters of the medicolegal laboratories. Where possible,these ages were obtained during the identification process(when the deceased's next of kin and/or identity documentwere available). Cause of death coding is based on the 1975International Classification of Diseases.15 Non-natural death(death by injury - either intentional or unintentionaQ iscoded as E800 to Egg9. Cause of death is recordedaccording to suggested Centers for Disease Control (COG)protocols for medical examiner surveillance systems(G. Parrish, CDC - personal communication).

SAMJARTICLES

laboratories. An example of such a case would be a patientin a chronic vegetative state as a result of a head injury, whodied of bronchopneumonia after a protracted period. Thegood coverage of injury deaths in the Cape Town metropoleby the medicolegal process was demonstrated in a study ofoccupational fatalities, which found that while 28% of thesedeaths were not reported to the occupational safetyauthorities, it was not possible to locate any fatalities thatwere not admitted to a medicolegal laboratory.'6

As this study aimed to provide a complete and accuratedatabase on non-natural mortality in the Cape Townmetropole during 1994, data cleaning comprised severalsteps: (i) all records wrth mismatched cause of deathcategories between mortuary records and autopsy data wereextracted and examined; (iI) records of patients with an ill­defined cause of death· were extracted and, where possible,a more specific cause of death was assigned, using allavailable secondary data sources; (iil) missing data, wherepossible, were obtained from autopsy and hospital reports.

The data were captured using the epidemiologicaldatabase package, Epi Info 5.'7 Frequency counts were usedto show the distribution of cause of death by age.

Results,. In 1994, there were 1 789 homicide, 291 suicide, 1 130

transport and 714 other non-natural fatalities in the CapeTown metropole (Table I). Sharp force (938) and firearms (462)were the most common causes of death in the homicidecategory (fable 11). Most homicides occurred in the 15 - 24­year (565) and 25 - 34-year (644) age categories. Homicidewas the cause of death of 230 women, 67 children (Youngerthan 15 years) and 53 elderiy (over the age of 60 years). Mosthomicides (60%) occurred in 10 suburbs, with Khayetitsha(239). Nyanga (164). Guguletu (126) and Mitchell's Plain (124)ranking highest. Almost 40% of suicides (1 14) involved theuse of fireanns. Seventeen female suicides occurred in the15 - 24-year age category. Suicide was the cause of death of29 elderly and 5 children. Two childhood suicides wereattributed to the use of firearms (Table Ill).

Table I. Mortality counts for major causes of non-natural mortalityin the Cape Town metropole (1994) by cause of death

The most common causes of death in the transportcategory were: motor vehicle/pedestrian (761), motor vehicle/passenger (131). motor vehicle/driver (79) and railwayfatalities (130). The highest number of transport-relatedfatalities occurred in the 25 - 34-year age category (304). Thedeaths of 156 children were transport-related (Table IV).

There were 6n deaths from other accidental causes ofnon-natural mortality and a further 37 of undeterminedcause (either homicide, suicide or accidental death).The age category composition of the most important other

Data quality and analysisThe non-natural deaths contained in the death registers ofthe two medicolegal laboratories are regarded bypathologists and epidemiologists as representative of almostall mortality due to violence and injury in the Cape Townmetropole. As non-natural death invariably has legal andfinancial implications, it is likely that this mortality enters theVifarreTgistration sY5tem.z~" Furthermore, misclassificationsof non-natural death as due to natural causes are generallyregarded as unusual, as most medical practitioners arereasonably concerned about the medicolegal implications ofnot referring these cases to the relevant authorities. It ispossible that some in-hospitaJ deaths, related to the lateeffects of injury, may have been certified as due to naturalcauses, and were therefore not admitted to the medicolegal

• The categories termed 'ill-defined' included skeletal remains, unknownwhether homicide, suicide, accident or other, unnatural death, unknownwhether homicide, suicide, accident or other; homicide undetermined;suicide undetermined and accident by other/unspecified causes.

Cause of death

HomicideTransport accidentsRreSuicideDrowningFalls

No. of deaths

1 7891 130

295291

9689

SAMJ \'olume 87 No. J .\farch 1997..

causes of non-natural death are shown in Table V. Most of Discussionthese deaths were due to fire (295), surgicaVmedicalprocedures (120), drowning (96) and falls (89). Spatial Violence and injury accounted for approximately 30% of a11-analysis of the distribution of death by fire showed that cause mortality in the Cape Town metropole during 1994. k.73% of fatalities occurred in 10 main areas. The 88 deaths increasing proportion of both firearm homicides and suicide

in Khayelitsha accounted for 30% of all deaths by fire in have occurred in Cape Town since 1986." The availability o·

the Cape Town metropole in 1994. Drowning remains a firearms in the home is regarded as a risk factor for suicide,

problem of the young, wrth 35 deaths (36%) of children and may be associated with the increased risk of adolescen

recorded. suicide.20

Table 11. Age distribution of homicide in the Cape Town metropole (1994) by cause of death

Age (yrs)

Cause of death 0-14 15 - 24 25 - 34 35 - 44 45-54 55 - 64 65 -74 75+ Total

Sharp force 14 314 369 143 70 21 6 1 938Firearms 8 196 156 66 24 9 2 1 462Blunt force 9 36 69 59 28 18 5 9 233Undetermined 9 3 28 17 7 1 3 3 71Strangulation 13 4 6 3 1 0 1 4 32Legal intelVention 1 10 14 6 0 0 0 0 31Child abuse- 12 0 0 0 0 0 0 0 12Burns 0 2 2 0 0 2 1 0 7Late effects 1 0 0 2 0 0 0 0 3All causes 67 565 644 296 130 51 18 18 1789• Infanticide and fatal child abuse, according to police recOf'ds.

Table Ill. Age distribution of suicide in the Cape Town metropole (1994) by cause of death

Age (yrs)

~ - Cause of death 0-14 15 - 24 25 - 34 35 - 44 45 - 54 55 - 64 65 -74 75+ Total

'-1 Firearms 2 28 39 16 13 3 6 7 114Hanging 2 13 24 11 4 4 0 2 60Poisoning 0 8 7 14 9 2 3 1 44Gassing 1 5 4 5 7 1 2 1 26Motor vehicle exhaust 0 0 6 13 4 2 1 0 26Trains 0 5 2 1 0 0 0 0 8Jumping 0 4 1 0 0 1 1 0 7Unspecified 0 0 2 1 1 1 0 1 6All causes 5 63 85 61 38 14 13 12 291

Table IV. Age distribution of transport-related mortality in the Cape Town metropole (1994) by cause of death

Age (Yrs)

Cause of death 0-14 15 - 24 25 - 34 35 - 44 45 - 54 55-64 65 - 74 75+ Total

MVA - pedestrian 123 106 202 158 85 43 28 16 761MVA - passenger 22 33 36 16 8 7 5 4 131MVA-driver 2 16 28 11 11 3 6 2 79Railway accident 6 32 30 33 15 10 1 3 130Motorcycle 1 4 5 1 0 0 0 0 11Bicycle 1 2 2 1 0 1 0 0 7Other vehicle 1 1 1 2 0 0 1 0 6Sporting accident 0 0 0 0 1 0 0 0 1Air transport 0 2 0 0 2 0 0 0 4All causes 156 196 304 222 122 64 41 25 1 130

Table V. Age category distribution of other non·natural mortality in the Cape Town metropole (1994) by cause of death

Age (yrs)

Cause of death 0-14 15 - 24 25 - 34 35 - 44 45-54 55 - 64 65 -74 75+ Total

Fire 75 34 75 60 26 14 7 4 295SurgicaVmedical procedures 14 6 9 11 15 26 27 12 120Drowning 35 10 16 16 12 3 1 3 96Falls 9 6 11 17 20 9 8 9 89Poisoning 3 1 5 8 1 1 0 0 19Excessive cold 0 0 0 1 2 0 1 0 4Electric current 0 0 0 4 0 0 0 0 4

_ Volume- Si No.3 March 199i SAMJ

Transportation accounted for almost one-third of injury€lath in Cape Town during 1994. The large number of,edestrian fatalities is cause for concern, especially in theght of the almost exclusive media focus on drivers and-Icoho!. In the UK, pedestrians and cyclists account for one­'1ird of all fatal and serious injuries on the roads, atthough1ey account for only about 7% of all mileage travelled.21

ne 130 railway-related deaths resulted in a mortality rate of551100 million passenger journeys, almost six times theatality rate on the London Underground in 1993.22

Other accidental non-natural deaths, although they,jomprise a smaller proportion of total deaths in the CapeTown metropole, are amenable to intervention strategies.=ires accounted for most of the fatalities in this category.\-1ost deaths by fire occur in areas with large numbers ofnformal dwellings and limited access to electricity. ThereHere a number of fresh-water drownings in canals, pondsmd rivers. Many of the 14 drownings that occurred in Table3ay harbour were work-related. Occupational injuries were::ssociated with fatal falls in the younger age categories, andlave historically been underreported. '15

Forensic data can be used to obtain accurate and::omplete information on deaths, better to understand the:;auses of these deaths and reduce those amenable topublic health intervention.232~ The current medicolegalorocess in South Africa lends itself to the provision of acomprehensive database on non-natural mortality, because::If the statutory requirement for postmortem examination inthese cases.,3 Unfortunately, no data collected at locallledicolegal laboratory level is fed into either proVincial ornational health information systems. This tragic waste ofImportant information can, to some extent, be curtailedthrough the use of large metropolitan medicolegallaboratories as sentinel surveillance sites.'15 The potentialbenefits of stmortem examination findings for publichealt surveillance of injuries and the improvement oftrauma care are substantial.2l> A medical examiner/coroneri~rmation-sharingprogramme has been developed by theCOC which aims to: (I) obtain more timely, accurate andcomplete information on non-natural deaths; (il) betterunderstand the causes of these deaths; and (iil) reduce themortality from those causes amenable to intervention.V"

10 the USA, targets for the reduction of injury and violencemortality have been set for the year 2000 and rates are beingcalculated on a yearty basis to monitor the effects of safetyregulation and policy decisions.<"Il In the USA, the consensusset of health status indicators for statewide surveillanceincludes: homicide and suicide rates, motor vehicle accidentdeath rates and a work-related injury death rate.29 It is hopedthat a similar monitoring strategy will be adopted in SouthAfrica as part of a public heaJth approach to the reduction ofinjury and violence. The Department of Health recentlyreleased a discussion document, Health Goals, Objectives,Strategies and Indicators for South Africa Violence and injuryindicators are included in priority areas such as child,women's and adolescent heallh, occupational heallh,emergency services, substance abuse and mental heafth.30

This article demonslrates the utility of routinely availabledata in providing infOrmation on the causes, anddemographic and spatial features of violence and injurymortality. The data described have formed the foundation ofa report that has been incorporated into the Draft ProvincialHealth Plan for the Westem Capa." It is hoped that this

SAMJARTICLES

study will stimulate stakeholders at a provincial and nationallevel to initiate similar projects. This would provide a moreaccurate picture of the debilitating effects of violence andinjury in South Africa and contribute to the debat<:! onavenues for prevention and health promotion_

We thank Professors G J Knobel and J P Nel, of theDepartments of Forensic Medicine of the Universities of CapeTown and Stellenbosch. for encouragement and allowing accessto postmortem data. the medical officers. registrars andspecialists in the two departments for detailed recording ofautopsy findings. members of the South African Police Servicesfor assistance at autopsies and retrieval of data. and Mrs M ELoubser and staff of the Forensic Chemical Laboratory for bloodalcohol analysis and release of data.

REFERENCES

1. Bradshaw D. OomngtOl'1 DE. S,tas f. The ~vel of mortality In Soum Atoca In 1985­what ooes 11 tell us aoout tIeaIth? S Air Med J 1992: 82; 237·24.0.

2. Bradshaw D. Schne<der M Vit. Regmraoon a!1d Sral1sues In Soutn AJm:a. CaseSrudy Meuopoktan Cape Town. Cape Town: Medical~nCounc~ 1995

3. Kielkowslu D. SlelnDerg M, Barron ?M. lite ahM oeath - morta"ty sta::5\1CS andpuolN: health. S Air Mea J 1989: 76: 672--675.

4.. Van de1" Me<We S. Vach O. Melcalf CA. PHf'lng 11'110 thI! blaci< nole - me qua"ty ofblack mortality data Ifl Port Elizabem afIO the rest of South AfrIca. S AIr MfId J1991: 79: ~19-422.

5. Byarugaba J. K"ll!lkowslo D. ReflectIOns on tr.Iuma ana VIOlenc.....eJateo deam .nSoweto, July 1990 - June 1991 S AIr Med J 1994, 84: 61()..fil~.

6. Bradsllaw 0, Laubscher R. Sc::lmeod!!l" M. ESrJrTlated Cal,lse of OeaUl Profiles for meNme New Provtn<;es &i~ on 1990 Data. Cape Town. Centre for Ep.oerruolog,calResearch. Medical Research Counc,l, 1995.

7. Stepher.s C. Timaeus I. AAerman M. Env"onmenr and Health m OevelopmgCountries: An At1a1ysls of Intra-Uroan DIfferentials I,Ism9 Ex'Stm9 Data (TechnicalReport). London: LOI"Idon Sc::hool of TropICal MedlC'ne and HY9,ene, 1994.

8. Lerer LB. lmprov,ng mortality data '" South Afnca; revlt!w 01 ne"t of kin statementsto dete.mme cause of outh ,n polw:::e certification. J £pIdemlOl Communsly HeaJrn1993; 41: 248--250.

9. Lerer LB. BllIchatt A.. TerreBlanche M. A bothet'$Offil! oeam' - narratl'/e acc;:ountsof .nfant mortality In C~ Town. Soum Atnca. Soc So Mea 1995. 40: 94!)-953

10. Philhps HT. Ar1lnter-~ stuoy of soaaI conortJQns and II'lfant mortabty ,n CapeTown. Mitbank MemonaJ Fund 01957.35(1): 7-28.

I 1. Rip MR. Keen CS. Woocs DL Births ana perinatal deams lI'l Mftcnelfs PIam S AIrMed J 1986: 70: 821-330

12. FlIp MR. Woods DL KMIl CS. Van Coeveroen oe Groat HA. Pennatal healm 11'1 thepen-urban towl'lShlp of Khayelitsha. Cape Town Pan It Infant mortality S AIr Mea J1988; 74; 633-634.

13. RepublIC of Soum Alnca.lnqueSfS Act. 1959. Ac;;t No. s.a of 1959 as amenoecl14. Republic of South Alnca. Med,cal De/'lral and Supplementary Health ServIcesProf~Acr. 1973.

15. IntemalJOl'lal Clas$lficatlon of DIseases 1975 Rev'SIOI1 Volume 1 Geneva. WHO. 197716. Lerer LB, Myers J. The applil;allon 01 two secondary oocumenlary sources to

idenl.ty the underreporting of fatal occupallona! InJunes m Cape Town, South AfrICa.Am J Ind Med 1994: 2ti: 521-527

17 Dean AG. Dean JA. Burton AH. Ow:::ker RH Epr Inlo. VerSIOn 5 A word proce$Slng.database and statlsues programme for l!?'demiology on MIC.ocompulers StoneMountalO. GA. USA Inc;:.• 1990

18 McDonalclIL Lerer LB r.me seoes anafysls ot Uenc25 II'lI>omlcIOl! and SUIClOe Im JEp.demIoI1994. 23: 66-72

19~ AL A1var~ FP. Somes G. ~t M SuIooe Ifl the home Ul relabOn 10 gunoWl'let'Sfup_ N EngJ J Mea 1994, 327: 467-472

20 Bntnt OA. Perpet JA. A»man CJ. ManU GM Wartella ME. Zeleoak JP The presenceand ac:ees5IblIity of tire.arms VI the homes of iIOOIescent 5UlClDe$. JAMA 1991; 266:2989--2995.

21 H,llman M. Hea/trly tran:5(lOrt poIocy. III Draper p. ea. HeaJtn 11v'otJgh PublIC PoIK;:yThe Greemng of PublIC Health lonaon G.een Pnnt. 199t 1l2·91

22 O'DonneJI t. Farmef ROT The eplCIem.ology of SUICide on the London Underg.oul'lO.Soc So Med 1994, 38: 437-442.

23 Con:Iner SM. loH B 800 years 01 cor~ have they a future? Lancer 1994, 344:799-801

24. Hoodon W. Jun. Advancn in the epldl!fT1io1ogy of .nJuries as a bases for publ,cpolICy. Pubhc Health Reports 1980. 95: 411-421

25. Srtas F. Zwarenst~1n M. Vao;:h O. B.adst\a.... 0 A n.allona! sentinel surveillancenfIlWOI'iI tor thI! me.JSUl1ll'T1efll 01' iI·hea/ttI ,n South Afnca. A prerequ.sae torepodemooIog>cal research and he.aIth planning S AfT Med J 1994. 84: 91-94

26 Pollock CA. 0 Net! JM. PamsIl RG, Com05 OL Annes1 JL TemooraI and geograptuctrends in me artopsy~ of bk.rrI aI'ICl Penembng trauma OHms In tneUnrtec:l States. JAMA 1993. 259: 152!)-1531.

27 Ed,tonai {Resol.Irces) The MeoIcaJ EAa:nnet'/Coronet II'lformatlOn-sIlanog program.AmJ~PauD 1969: 1:88.

28. CommItte on Infury Control.~ COI1lJ'ollfl !he 199(}"s. A nauonal pian tor acbOO.Report 10 me Sec:ono Y.otkl ConIefetlCl! on 1nJU'Y Control. Atlanta USA. May 1993

29. ZUOCOIII SL Car'SOO"I CA.. COCs consensus Wl 01 health status 1I'IdIea1tlrS. Mannonng:arid pnoI'lttSa!>On by stale heaIm oepartments Am J Pub£1C HeaIttI 1994.84: 11>44­,.....

30. Department of HNItn. V_ 2000 - Hea/t1l Goals.~ Sll'ategNJS MKt IllCJC3tO/'Stor Soutn AirIaJ (OIscussIon Documen1I Pl1l1ona. Department of HealIn. 1995

31 $Iralegtc Management Team. MfI'llstry et Health and Socaal ServICes Western CapeProY1f1Ce. Draft Prov,nclal Health Plan. February 1995. Cape Tcwn; MlI'ltstry 01Health and Soc131 Servw:::es. 1995.

Accepted 28 Aug 1996,

SA.~1J \'olumt S7 No.3 Match 1997 __