- Email: [email protected]
Effect of diarrhoeal disease control on infant and childhood mortality in Egypt
334
Group of neurosurgeons: Guidelines for initial management after head injury in adults. Br Med J 1984; 288: 983-85. 5. Rose J, Valtonen S, Jennett B. Avoidable factors contributing to death after head injury. Br Med J 1977; ii: 615-18. 6. Gentleman D, Jennett B. Hazards of inter-hospital transfer of comatose head-injured patients. Lancet 1981; ii: 853-55. 7. Jennett B, Bond M. Assessment of outcome after severe brain damage. A practical scale. Lancet 1975; i: 480-84. 8. Jennett B, Carlin J. Preventable mortality and morbidity after head injury. Injury 1978; 10: 31-39. 9. Jeffreys RV, Jones JJ. Avoidable factors contributing to the death of head injury patients in general hospitals in Mersey Region. Lancet 1981; ii:
4.
459-61. 10. Anderson
ID, Woodford M, de Dombal FT, Irving M. Retrospective study of 1000 deaths from injury in England and Wales. Br Med J 1988;
296: 1305-08.
Wright IH, McDonald JC, Rogers PN, Ledingham IMcA. Provision of facilities for secondary transport of seriously ill patients in the United Kingdom. Br Med J 1988; 296: 543-45. 12. Bion JF, Wilson IH, Taylor PA. Transporting critically ill patients by ambulance: audit by sickness scoring. Br Med J 1988; 296: 170. 13. Van Alphen HAM, Braakman R. Richtlijnen voor de behandeling van patienten met ernstig traumatisch hersenletsel (Guidelines for the treatment of patients with severe traumatic brain damage). Ned Tijdschr 11.
14. Draaisma JMTh. Mogelijkheden tot evaluatie van zorg aan patienten met enrstig hersenletsel door ambulancehulpverleners (Methods of evaluating the care of patients with severe head injuries by ambulance personnel). De Ambulance 1986; 7: 15-17. 15. Commission on the Provision of Surgical Services. Report of the Working Party on Head Injuries. London: Royal College of Surgeons of England. 1986. 16. Hewer RL, Wood VA. Availability of computed tomography of the brain in the United Kingdom. B Med J 1989 298: 1219-20. 17. Marsh H, Maurice-Williams RS, Hatfield R. Closed head injuries: where does delay occur in the process of transfer to neurosurgical care? BrJ Neurosurg 1989; 3: 13-20. 18. Yates DW. Action for accident victims: plans to stop patients dying unnecessarily from major trauma. Br Med J 1988; 297: 1419-20. 19. Smith RW. Guidelines for establishment of trauma centers. J Neurosurg
1986; 65: 569-71. Jennett B. Some international comparisons. In: Levin HS, Eisenberg HM, Benton AL, eds. Mild Head Injury. Oxford: Oxford University Press, 1989; 23-34. 21. McLaren CAN, Robertson C, Little K. Missed orthopaedic injuries in the resuscitation room. J R Coll Surg Edin 1983; 28: 399-401. 22. Ottosson A, Krantz P. Traffic fatalities in a system with decentralized trauma care. JAMA 1984; 251: 2668-71. 23. Frey CF. Care of serious trauma in Southern Sweden. JAMA 1984; 251:
20.
2698-99.
Geneeskd 1985; 129: 645-49.
EPIDEMIOLOGY Effect of diarrhoeal disease control on infant and childhood mortality in Egypt
Report from the National Control of Diarrheal Diseases Project
The effect of the National Control of Diarrheal Diseases Project, started in 1983, on infant and childhood mortality in Egypt was assessed by of national civil registration data, means nationwide cluster sample surveys of households, and local area studies. Packets of oral rehydration salts are now widely accessible; oral rehydration therapy is used correctly in most episodes of diarrhoea; most mothers continue to feed infants and children during the child’s illness; and most physicians prescribe oral rehydration therapy. These changes in the management of acute diarrhoea are associated with a sharp decrease in mortality from diarrhoea, while death from other causes remains nearly constant.
Diarrheal Diseases Project (NCDDP) was established in 1983 after successful trials in 1980 of oral rehydration therapy (ORT) in rural villages in the Dakahlia Governorate of Lower Egypt.2,6A full national programme was underway by 1984,’ and has continued since then. It is unusual for a single health intervention in a developing country to have a proven beneficial effect,8 and the value of narrowly targeted interventions has been questioned.9 We have assessed the effect of the NCDDP on infant and childhood mortality in Egypt, to the end of 1987.
Methods We used various methods to measure NCDDP impact because we did not regard any one method as sufficiently comprehensive or robust. Our analyses are therefore descriptive and are not based on
significance testing.
Introduction In 1978, when oral rehydration salts (ORS) were first introduced into Egypt, infant mortality was about 100 per 1000 live births; at least half the infant and childhood deaths were diarrhoea-associated.1-5 The National Control of
ADDRESS: National Control of Diarrheal Diseases
Project, 55 Mosaddaq Street, Dokki, Giza, Egypt (M. EI-Rafie, MD, W. A.
Hassouna, MD, N. Hirschhorn, MD, S. Loza, PhD, P. Miller, MA, A. Nagaty, MD, S. Nasser, MD, and R. Riyad, MD). Correspondence to Dr Said Madkour, executive director, National Control of Diarrheal Diseases
Project.
335
TABLE I-SELECTED MEASURES OF PROGRESS IN ORT
Service statistics Service statistics about production and distribution of ORS and to Ministry of Health (MOH) clinics for diarrhoeal illness were collected routinely by the NCDDP. about visits
Civil registration a death, which is the legal is issued at MOH facililities. Data death certificate requirement, are collated at governorate levels, and are then checked, validated, and grouped nationally by the Central Agency for Public Mobilization and Statistics (CAPMAS). The age, sex, location, month, and certified cause of death are the data we examine here.
When
a
family member reports a
Cluster sample double round household survey
eight governorates (four in Upper Egypt, four in Lower Egypt, excluding the large cities), 179 urban and rural clusters were identified by means of cumulative population lists from the 1976 census. From each cluster two subclusters were chosen by the fieldwork supervisor; 30 children younger than 2 years old in each subcluster were identified by the technique of random start and direction for canvass. Surveys were done in three annual cycles of In
rounds each--one at the start of the "diarrhoea season" and one immediately after the season (January/ February). In both rounds of each annual cycle the same households were followed, but a different set were selected each year. For children who had died by the time of the second round, parents were asked if the child had had diarrheoa in the week before death. Total mortality and diarrhoea associated mortality were calculated for the approximate 6 month periods. Data from the first two cycles have been previously analysed in detail.1O Because of the sampling procedure the data are not strictly representative, but yield mortality estimates consistent with national levels. two
(June/July),
Knowledge, attitude, and practice surveys These were conducted annually at the end of each diarrhoea in a randomly chosen 1100-household subset of the cluster samples and 400 households selected from low income clusters in Cairo. We estimated whether ORS were mixed correctly by asking mothers who used or knew about ORT (96% of the sample) to demonstrate directly. (Survey methods are available on request from the NCDDP.) season
*NCDDP statistics tMean for 1987 and 1988. t1984-88. N C D D P-commlssloned household cluster sample survey (n=1500). §Dakahlia control group (n105)." ’I1Dakahlia control group (n =41 0).7 I[Pooled data (unpublished) from three surveys in Menoufia govenorate (n = 1062
episodes).
of those who died had received no ORT.2,3 ORS was not available at all until 1977, and between 1978 and 1983 much of the ORS produced or imported lay unused in warehouses and clinics (NCDDP, internal documents). However, since 1983 the NCDDP has guaranteed distribution of ORS to nearly all 32 000 MOH and university clinics and hospitals and 6000 private pharmacies; has trained more than 40 000 primary health care physicians, nurses, pharmacists, and others in the principles of ORT; and has continued the annual public education campaign via the media (see also tableI in ref 7). A balanced intravenous solution for resuscitation of shocked children was made available to all
hospitals. As judged by several measures, 1985 to 1986 were the peak years for the project (table I). There was nearly a four-fold increase in ORS distribution, 98-99% of mothers were aware of ORS, and there were high rates of use and correct mixing. The numbers of children attending public sector clinics for diarrhoea increased from about 630 000 in
TABLE N—INFANT AND CHILDHOOD MORTALITY, 1970-87
Local-area studies Controlled studies of ORT and its impact on mortality had been done in 1979-80 in the lower Egypt governorates of Dakahlia2 and Menoufia3 in 29 and 12 villages, respectively. The NCDDP repeated the Dakahlia study in 1986. In Menoufia, the NCDDP returned to the villages; examined the records of registered deaths of children under 5 years old from 1979 to 1988; determined the cause of deaths during 12 months (1987-88) by direct interview; and conducted a KAP survey with 504 mothers of children under 2 years old who were selected by cluster sampling. All data were compared with those similarly obtained in 1979-80.
Results
Programme achievements Although quantitative data are not available before 1979, the view is that management of diarrhoea was inadequate: antimicrobials and antidiarrhoeals were used predominantly ; fluids and food were often withheld; most of the seriously dehydrated children were eventually brought to a medical practitioner, but usually in late stages; and 58%
*Infant mortality rate= (registered infant deaths=registered births in same year) 1000. tChlldhood mortality is age-specific death rate for children aged 1-4 yr. tMean percent declme per year (calculated exponentially)
x
336
TABLE IV-SUMMARY OF HOUSEHOLD-BASED DATA: MORTALITY CHANGES, 1979-88
*Changesm proportions dead after 6 mo (June/July-January/February) from initial cohort of children aged 0-2 yr t6 mo mortality m children (1-59 mo) adapted from table III in ref 7 tEstlmated deaths for children < 5 yr in 12 villages. §See ref 3. 11137 registered deaths (53 diarrhoea-associated) were followed up by mterview. Diarrhoeal deaths and rates were calculated with proportionate distribution of the 7 deaths
Distribution of 1970-86. -
-
-
infant
deaths
not
followed.
registered in May-August,
="no seasonality" line.
1983 to 14 million in 1985. Since 1986, a plateau has been reached, and performance has been fairly constant. In the knowledge, attitude, and practice surveys, a median of two packets (400 ml) was used on the first day, and ORS was continued for a median of 2 days. ORS is more likely to be used in severe cases, and in greater volume; for example, in the local area study in Menoufia in 1988,84% of 106 children with serious diarrhoea and dehydration (according to their mothers) were given ORS, whereas 29% of 194 children with mild diarrhoea without dehydration were similarly treated.
Mortality According to the civil registration data, diarrhoeal mortality declined more quickly after 1977, and most rapidly after 1982, both in relation to the 1970-77 period and to the decrease in non-diarrhoeal mortality. Between 1982 and 1987, infant mortality decreased by 36% and childhood mortality by 43% of which the decline in diarrhoeal mortality accounted for 82% and 62%, respectively (table II). Overall mortality and diarrhoea-specific mortality figures may be affected by under-registration of vital events misclassification of cause of death. Infant deaths were under-registered by about 25% in 1976,11 but there is nothing to indicate that this has increased since then. Much
or
of the underestimate is in the neonatal period4 when diarrhoea is not a principal cause of death.3 Registration of births was 96% complete in 1972-75.1 We can look at the bias in classification of deaths by examining seasonality of death. Diarrhoea is the most highly seasonal of major causes of death; for example, in 1983 56% of registered diarrhoeal deaths occurred in the four hot months from May to August. If there was a true decline in diarrhoeal mortality, the degree of seasonality of mortality should also have declined. Moreover, month of death is more accurately coded than cause. Thus, trends in seasonality are an alternative indicator of trends in cause of death without the same potential for classification error. The proportion of annual infant deaths from May to August between 1970 and 1987 stayed above 45% until the end of 1983, after which it declined sharply (figure). Since 33-3% in a 4-month period represents no seasonality, more than half of the seasonality of mortality noted in 1983 had disappeared by 1987. Could any decrease in the number of diarrhoea associated deaths have been offset by an increase in the number of deaths from other causes, either from replacement mortality or classification bias? Table III shows infant death rates by cause and season from 1983 to 1987. Despite sharp seasonal changes in diarrhoeal deaths, there was little or no increase in non-diarrhoeal deaths in the seasons that immediately followed. Declines in diarrhoeal deaths within a given season were not offset by equivalent increases in nondiarrhoeal deaths; this is further evidence against substantial changes in classification bias. Table 11 also suggests that there was no compensatory increase in later childhood
mortality. That girls
TABLE III-INFANT MORTALITY, 1983-87
were more likely than were boys to die from diarrhoeal and respiratory diseases but not from other causes is consistent with service statistics showing that boys were more likely to be taken to rehydration clinics.
Other mortality surveys
Source: national civil registration statistics, CAPMAS. Infant deaths due to diarrhoeal/non-diarrhoeal disease births
are
given
per 1000 live
(table w)
The cluster sample double round household surveys show that between 1984 and 1986 total mortality in children under 2 years old for the summer and autumn diarrhoea season declined by 38%; most of the decrease was between
337
1984 and 1985, and 82% of the decrease was in the diarrhoea associated component. The prevalence of diarrhoea in the 2 weeks before the first round did not vary substantially in the 3 years (30%, 29%, and 26%). In the Dakahlia study in 1980, diarrhoeal mortality in children under 5 was 50% lower in the study area than in the control area; this was associated with a higher rate of ORS use in the study area compared with the control area (53 % vs 21 % of diarrhoeal episodes).When the study was repeated
in 1986, there were no longer any differences in mortality and ORS use between the study and control areas. However, in both areas diarrhoeal mortality was substantially lower (by 59% and 70%, respectively) and ORS use in both areas was much higher (to 85% and 77%, respectively) than in 1980; the number of ORT prescriptions per diarrhoea episode by physicians in clinics in the control area increased from 48 to 95%.’ Civil registration data for the whole governorate reflect the same mortality trend (see table IV in
ref 7). In
Menoufia,
a
controlled ORS home distribution
experiment in 1979-80 did not lower mortality, perhaps largely because of poor acceptance by physicians: "Only one-third of the children who died of diarrhoea received oral fluid therapy during their terminal episode, although over 90 percent received medical care". However, between 1979-80 and 1987-88, registered under-five mortality declined by 64%; interviews showed that 50% of the deaths were due to diarrhoea in 1979-80, and 39% in 1987-88. Mortality in May-August was 52% of the annual total mortality in 1979 but was 37% in 1987. Physicians now prescribed ORS in 66% of cases for which they were consulted, while mothers said that they used ORS in 52% of diarrhoea episodes. As in Dakahlia, civil registration data for the whole governorate reflect a similar mortality trend
(CAPMAS). Other factors associated with decrease in mortality The data above show that case management of diarrhoea has clearly improved as a result of the NCDDP. However, the risk of death may also be directly affected either by changes in diarrhoea incidence or by host resistance. The incidence of diarrhoeal disease has been measured repeatedly by NCDDP between 1983 and 1988. The data do not point to any decline in incidence of diarrhoea. For example, in seven separate studies, seasonally adjusted prevalence in the previous week ranged from 20 to 38 % with no obvious secular trend. The most useful indicator of host resistance is nutritional status. National surveys by the Nutrition Institute, Cairo, in 1978, 1980, and 1986 suggest that the frequency of severe malnutrition was relatively low with little change. Children who have recently had measles are at increased risk of death from diarrhoea. That measles vaccination increased from 41 % of children aged 12 to 23 months in 1983 to 76 % in 1987may have contributed to the reduction in child mortality; however, since measles immunisation begins at age 9 months, this could not have substantially affected infant mortality.
episodes and is correctly used. Moreover, most mothers continue to breastfeed children during the children’s illness. When dehydration from diarrhoea is a major cause of death, mortality reduction will be primarily due to correct use of ORT; we have strong evidence that ORS is more likely to be used for children who are more dehydrated. During the project period mortality-both absolute and relative to other causes of death-has decreased substantially. A reduction in diarrhoeal mortality was already underway after 1970, but was clearly hastened, first with early availability of ORS, and then even more rapidly with the inception of NCDDP. These declines occurred throughout the country. If the diarrhoea-specific mortality rates of 1982 had occurred in 1987, there would have been 64 261 more infant and childhood diarrhoeal deaths nationwide (not adjusting for under-registration) than the 38 451 observed. Could factors other than NCDDP have contributed to such changes? The immediate determinants of diarrhoeal mortality may be classified as curative care, incidence, and host resistance; other factors must operate through one or more of these determinants. There was no major change in the incidence of diarrhoea or nutritional status during the course of the project. We cannot exclude changes in the severity of illness (eg, subtle changes in the virulence of microbes causing diarrhoea) but ORT together with continued feeding reduces duration of diarrhoea.12 Were children who recovered from diarrhoea more likely to die from other causes? Mosley and Becker13 suggest that children who recover from diarrhoea are frail and thus more likely to die later. We have no evidence that this happens within large populations. In particular, there was no increase in death rates in subsequent seasons from other causes, or at later stages. Although our data strongly indicate that NCDDP has had an impact on diarrhoeal mortality, we cannot say exactly which inputs were specifically responsible. Use of ORS, feeding practices during diarrhoeal episodes, and emergency treatment for severe dehydration have probably each made an important contribution. The declines were due to a comprehensive programme of curative care for diarrhoeal disease, rather than simply to an ORT project. Mothers’ decisions in child care are vital; but in Egypt the availability of clinics and correct practice by physicians are also essential. Hirschhorn et ap4 have suggested a structure to decide which primary health care inputs are appropriate to specific situations. A decrease in diarrhoeal mortality in Egypt was possible because of moderate levels of malnutrition and immunisable disease mortality, a well-structured health system, and good access to the media. In countries without these characteristics, a programme similar to NCDDP may not be as successful. The NCDDP must now find the causes of the residual mortality ascribed to diarrhoeal disease, and develop methods to remove them. The incidence of diarrhoea must be decreased by improved hygiene and sanitation, and the severity and duration of illness must be lessened by better nutrition during and after illness.
Discussion We believe that the NCDDP has influenced care of diarrhoeal disease enough to account for substantial changes in mortality. ORS is available and readily accessible; and ORT is prescribed regularly in over half of diarrhoeal
The NCDDP would like to thank Dr M. Gabr, Dr R. Dewidar, Dr M. Hammamy, Dr E. Helw, Dr L. El Sayad, Dr A. A. Shaarawi, Dr H. Tammam, Dr A. Hashem, Dr A.-M. Hassan, and Dr M. Hammouda; and Dr Reginald Gipson from John Snow Inc., Boston, USA. This
study was funded partly by USAID grant no 263-0137.
338
REFERENCES 1. National
Academy of Sciences Committee on Population and Demography. The estimation of recent trends in fertility and mortality in Egypt. Report no 9. Washington: National Academy Press, 1982. 2. Kielmann AA, Mobarek AB, Hammamy MT, et al. Control of deaths from diarrheal disease in rural communities. I. Design of an intervention study and effects on child mortality. Trop Med Parasitol 1985; 36: 191-98. 3. Tekce B. Oral rehydration therapy: an assessment of mortality effects in rural Egypt. Studies Family Planning 1982; 13: 315-27. 4. Infant and child mortality in Egypt. Cairo: Arab Republic of Egypt Central Agency for Public Mobilization and Statistics, 1986. 5. The State of Egyptian children. Cairo: Arab Republic of Egypt Central Agency for Public Mobilization and Statistics, and UNICEF, 1988. 6. Keilmann AA, Nagaty A, Ajello CA. Control of deaths from diarrheal disease in rural communities. II. Motivating and monitoring the community. Trop Med Parasitol 1986; 37: 15-21. 7. National Control of Diarrheal Diseases Project. Impact of the national control of diarrhoeal diseases project on infant and child mortality in
Dakahlia, Egypt. Lancet 1988; ii: 145-48. 8. Ewbank DC. Uses of mortality data for evaluating the success of scientific health and development programmes. In: Data bases for mortality measurement. New York: United Nations, 1984. 9. Newell KW. Selective primary health care: the counter revolution. Soc Sci Med 1988; 26: 903-06. 10. Rashad H. Oral rehydration therapy and its effect on child mortality in Egypt. J Biosoc Sci 1989; 10 (suppl): 105-13. 11. Rashad H. Evaluation of the completeness of mortality registration in Egypt. West Asia and North Africa Regional Papers. Cairo: The Population Council, 1981. 12. Santosham M. Nutritional aspects of ORT. J Diarrhoeal Dis Res 1987; 5: 270-74. 13. Mosley WH, Becker S. Demographic models for child survival: implications for program strategy. Presented at: Child Survival Programs: Issues for the 1990s, Johns Hopkins University, Baltimore, November 1988. 14. Hirschhorn N, Grabowsky M, Houston R, et al. Are we ignoring different levels of mortality in the primary health care debate? H Pol Planning 1989; 4: 343-47.
First coitus before menarche and risk of transmitted disease
The
prevalence of sexually transmitted disease (STD), pelvic inflammatory disease (PID), and
cervical cancer, and the relation between these conditions were studied in 2111 Ethiopian women. Early sexual activity was associated with an increase in prevalence rates of STD and PID; possible aetiological factors include physical and immunological immaturity of the female genital tract and the number of sexual partners.
Introduction Cervical
cancer
is
an
important
cause
of
cancer
mortality
among women in Ethiopia and in much of the rest of Africa,
where patients often present with advanced disease. In high incidence areas, age at first coitus, specifically before age 21, 20, or 17 years is the most important risk factor,t-3 with a doubling of the risk in those having first coitus at age 16 or less.4 The association between cervical cancer and the number of sexual partners is well established: in married women with cervical cancer who claim not to have had a sexual partner other than their husband, the number of sexual partners of their husbands was a significant risk factor for cervical cancer. This male promiscuity has been called the "male factor".6 Only one study has reported an association between cervical cancer and chronic pelvic inflammatory disease (PID) and sexually transmitted disease (STD).7 A study in Ethiopia in 1975 showed that STD (particularly syphilis and gonorrhoea) was highly prevalent; prevalence rates for other STD were unknown, although acute and chronic PID
sexually
and their
complications were the most important cause of and mortality in women.8,9 An international collaborative study was therefore planned to investigate the prevalence of STD and cervical cancer in an unscreened population, the causal factors of cervical cancer to define an "at risk" group, and possible associations between STD and
morbidity
cervical
cancer.
Patients and methods 2111 unselected Ethiopian women were studied; half were outpatients with gynaecological symptoms, and half were healthy women attending family planning, antenatal, and postnatal clinics in Addis Ababa. The collection of data, cytology slides, and sera, and the VDRL test were done in 1975 and 1976 while tribal, religious, and other factors could still be assessed independently of population migration (before the Ethiopian revolution). It was not until 1977 and 1978 that the data and specimens could be despatched from the country, although full permission had been given by the Ethiopian Ministry of Health. The patients’ data, used with their consent, were as follows: Personal details were obtained by questionnaire (completed by an Ethiopian female assistant), and included age, hospital/clinic attended, address (origin: town or rural), tribe, religion, parity, total number of pregnancies, "marital status" (single, married, divorced, widowed, bargirl, prostitute, "talla" [local beer] seller, maid), monthly family income, age at first marriage and first coitus and
ADDRESSES:
Department of Bacteriology, Edinburgh University, Edinburgh, UK (M. E. Duncan, MD, J. F. Peutherer, MD, P. Simmonds, PhD, H. Young, PhD); Liège University, Belgium (G. Tibaux EE, A. Pelzer, MD); Statens Seruminstitut, Copenhagen, Denmark (K. Reiman); and Institute of Ophthalmology, London, UK (Y. Jamil, MSc, Prof S. Daroughar, MD). Correspondence to Dr M. E.
Group of neurosurgeons: Guidelines for initial management after head injury in adults. Br Med J 1984; 288: 983-85. 5. Rose J, Valtonen S, Jennett B. Avoidable factors contributing to death after head injury. Br Med J 1977; ii: 615-18. 6. Gentleman D, Jennett B. Hazards of inter-hospital transfer of comatose head-injured patients. Lancet 1981; ii: 853-55. 7. Jennett B, Bond M. Assessment of outcome after severe brain damage. A practical scale. Lancet 1975; i: 480-84. 8. Jennett B, Carlin J. Preventable mortality and morbidity after head injury. Injury 1978; 10: 31-39. 9. Jeffreys RV, Jones JJ. Avoidable factors contributing to the death of head injury patients in general hospitals in Mersey Region. Lancet 1981; ii:
4.
459-61. 10. Anderson
ID, Woodford M, de Dombal FT, Irving M. Retrospective study of 1000 deaths from injury in England and Wales. Br Med J 1988;
296: 1305-08.
Wright IH, McDonald JC, Rogers PN, Ledingham IMcA. Provision of facilities for secondary transport of seriously ill patients in the United Kingdom. Br Med J 1988; 296: 543-45. 12. Bion JF, Wilson IH, Taylor PA. Transporting critically ill patients by ambulance: audit by sickness scoring. Br Med J 1988; 296: 170. 13. Van Alphen HAM, Braakman R. Richtlijnen voor de behandeling van patienten met ernstig traumatisch hersenletsel (Guidelines for the treatment of patients with severe traumatic brain damage). Ned Tijdschr 11.
14. Draaisma JMTh. Mogelijkheden tot evaluatie van zorg aan patienten met enrstig hersenletsel door ambulancehulpverleners (Methods of evaluating the care of patients with severe head injuries by ambulance personnel). De Ambulance 1986; 7: 15-17. 15. Commission on the Provision of Surgical Services. Report of the Working Party on Head Injuries. London: Royal College of Surgeons of England. 1986. 16. Hewer RL, Wood VA. Availability of computed tomography of the brain in the United Kingdom. B Med J 1989 298: 1219-20. 17. Marsh H, Maurice-Williams RS, Hatfield R. Closed head injuries: where does delay occur in the process of transfer to neurosurgical care? BrJ Neurosurg 1989; 3: 13-20. 18. Yates DW. Action for accident victims: plans to stop patients dying unnecessarily from major trauma. Br Med J 1988; 297: 1419-20. 19. Smith RW. Guidelines for establishment of trauma centers. J Neurosurg
1986; 65: 569-71. Jennett B. Some international comparisons. In: Levin HS, Eisenberg HM, Benton AL, eds. Mild Head Injury. Oxford: Oxford University Press, 1989; 23-34. 21. McLaren CAN, Robertson C, Little K. Missed orthopaedic injuries in the resuscitation room. J R Coll Surg Edin 1983; 28: 399-401. 22. Ottosson A, Krantz P. Traffic fatalities in a system with decentralized trauma care. JAMA 1984; 251: 2668-71. 23. Frey CF. Care of serious trauma in Southern Sweden. JAMA 1984; 251:
20.
2698-99.
Geneeskd 1985; 129: 645-49.
EPIDEMIOLOGY Effect of diarrhoeal disease control on infant and childhood mortality in Egypt
Report from the National Control of Diarrheal Diseases Project
The effect of the National Control of Diarrheal Diseases Project, started in 1983, on infant and childhood mortality in Egypt was assessed by of national civil registration data, means nationwide cluster sample surveys of households, and local area studies. Packets of oral rehydration salts are now widely accessible; oral rehydration therapy is used correctly in most episodes of diarrhoea; most mothers continue to feed infants and children during the child’s illness; and most physicians prescribe oral rehydration therapy. These changes in the management of acute diarrhoea are associated with a sharp decrease in mortality from diarrhoea, while death from other causes remains nearly constant.
Diarrheal Diseases Project (NCDDP) was established in 1983 after successful trials in 1980 of oral rehydration therapy (ORT) in rural villages in the Dakahlia Governorate of Lower Egypt.2,6A full national programme was underway by 1984,’ and has continued since then. It is unusual for a single health intervention in a developing country to have a proven beneficial effect,8 and the value of narrowly targeted interventions has been questioned.9 We have assessed the effect of the NCDDP on infant and childhood mortality in Egypt, to the end of 1987.
Methods We used various methods to measure NCDDP impact because we did not regard any one method as sufficiently comprehensive or robust. Our analyses are therefore descriptive and are not based on
significance testing.
Introduction In 1978, when oral rehydration salts (ORS) were first introduced into Egypt, infant mortality was about 100 per 1000 live births; at least half the infant and childhood deaths were diarrhoea-associated.1-5 The National Control of
ADDRESS: National Control of Diarrheal Diseases
Project, 55 Mosaddaq Street, Dokki, Giza, Egypt (M. EI-Rafie, MD, W. A.
Hassouna, MD, N. Hirschhorn, MD, S. Loza, PhD, P. Miller, MA, A. Nagaty, MD, S. Nasser, MD, and R. Riyad, MD). Correspondence to Dr Said Madkour, executive director, National Control of Diarrheal Diseases
Project.
335
TABLE I-SELECTED MEASURES OF PROGRESS IN ORT
Service statistics Service statistics about production and distribution of ORS and to Ministry of Health (MOH) clinics for diarrhoeal illness were collected routinely by the NCDDP. about visits
Civil registration a death, which is the legal is issued at MOH facililities. Data death certificate requirement, are collated at governorate levels, and are then checked, validated, and grouped nationally by the Central Agency for Public Mobilization and Statistics (CAPMAS). The age, sex, location, month, and certified cause of death are the data we examine here.
When
a
family member reports a
Cluster sample double round household survey
eight governorates (four in Upper Egypt, four in Lower Egypt, excluding the large cities), 179 urban and rural clusters were identified by means of cumulative population lists from the 1976 census. From each cluster two subclusters were chosen by the fieldwork supervisor; 30 children younger than 2 years old in each subcluster were identified by the technique of random start and direction for canvass. Surveys were done in three annual cycles of In
rounds each--one at the start of the "diarrhoea season" and one immediately after the season (January/ February). In both rounds of each annual cycle the same households were followed, but a different set were selected each year. For children who had died by the time of the second round, parents were asked if the child had had diarrheoa in the week before death. Total mortality and diarrhoea associated mortality were calculated for the approximate 6 month periods. Data from the first two cycles have been previously analysed in detail.1O Because of the sampling procedure the data are not strictly representative, but yield mortality estimates consistent with national levels. two
(June/July),
Knowledge, attitude, and practice surveys These were conducted annually at the end of each diarrhoea in a randomly chosen 1100-household subset of the cluster samples and 400 households selected from low income clusters in Cairo. We estimated whether ORS were mixed correctly by asking mothers who used or knew about ORT (96% of the sample) to demonstrate directly. (Survey methods are available on request from the NCDDP.) season
*NCDDP statistics tMean for 1987 and 1988. t1984-88. N C D D P-commlssloned household cluster sample survey (n=1500). §Dakahlia control group (n105)." ’I1Dakahlia control group (n =41 0).7 I[Pooled data (unpublished) from three surveys in Menoufia govenorate (n = 1062
episodes).
of those who died had received no ORT.2,3 ORS was not available at all until 1977, and between 1978 and 1983 much of the ORS produced or imported lay unused in warehouses and clinics (NCDDP, internal documents). However, since 1983 the NCDDP has guaranteed distribution of ORS to nearly all 32 000 MOH and university clinics and hospitals and 6000 private pharmacies; has trained more than 40 000 primary health care physicians, nurses, pharmacists, and others in the principles of ORT; and has continued the annual public education campaign via the media (see also tableI in ref 7). A balanced intravenous solution for resuscitation of shocked children was made available to all
hospitals. As judged by several measures, 1985 to 1986 were the peak years for the project (table I). There was nearly a four-fold increase in ORS distribution, 98-99% of mothers were aware of ORS, and there were high rates of use and correct mixing. The numbers of children attending public sector clinics for diarrhoea increased from about 630 000 in
TABLE N—INFANT AND CHILDHOOD MORTALITY, 1970-87
Local-area studies Controlled studies of ORT and its impact on mortality had been done in 1979-80 in the lower Egypt governorates of Dakahlia2 and Menoufia3 in 29 and 12 villages, respectively. The NCDDP repeated the Dakahlia study in 1986. In Menoufia, the NCDDP returned to the villages; examined the records of registered deaths of children under 5 years old from 1979 to 1988; determined the cause of deaths during 12 months (1987-88) by direct interview; and conducted a KAP survey with 504 mothers of children under 2 years old who were selected by cluster sampling. All data were compared with those similarly obtained in 1979-80.
Results
Programme achievements Although quantitative data are not available before 1979, the view is that management of diarrhoea was inadequate: antimicrobials and antidiarrhoeals were used predominantly ; fluids and food were often withheld; most of the seriously dehydrated children were eventually brought to a medical practitioner, but usually in late stages; and 58%
*Infant mortality rate= (registered infant deaths=registered births in same year) 1000. tChlldhood mortality is age-specific death rate for children aged 1-4 yr. tMean percent declme per year (calculated exponentially)
x
336
TABLE IV-SUMMARY OF HOUSEHOLD-BASED DATA: MORTALITY CHANGES, 1979-88
*Changesm proportions dead after 6 mo (June/July-January/February) from initial cohort of children aged 0-2 yr t6 mo mortality m children (1-59 mo) adapted from table III in ref 7 tEstlmated deaths for children < 5 yr in 12 villages. §See ref 3. 11137 registered deaths (53 diarrhoea-associated) were followed up by mterview. Diarrhoeal deaths and rates were calculated with proportionate distribution of the 7 deaths
Distribution of 1970-86. -
-
-
infant
deaths
not
followed.
registered in May-August,
="no seasonality" line.
1983 to 14 million in 1985. Since 1986, a plateau has been reached, and performance has been fairly constant. In the knowledge, attitude, and practice surveys, a median of two packets (400 ml) was used on the first day, and ORS was continued for a median of 2 days. ORS is more likely to be used in severe cases, and in greater volume; for example, in the local area study in Menoufia in 1988,84% of 106 children with serious diarrhoea and dehydration (according to their mothers) were given ORS, whereas 29% of 194 children with mild diarrhoea without dehydration were similarly treated.
Mortality According to the civil registration data, diarrhoeal mortality declined more quickly after 1977, and most rapidly after 1982, both in relation to the 1970-77 period and to the decrease in non-diarrhoeal mortality. Between 1982 and 1987, infant mortality decreased by 36% and childhood mortality by 43% of which the decline in diarrhoeal mortality accounted for 82% and 62%, respectively (table II). Overall mortality and diarrhoea-specific mortality figures may be affected by under-registration of vital events misclassification of cause of death. Infant deaths were under-registered by about 25% in 1976,11 but there is nothing to indicate that this has increased since then. Much
or
of the underestimate is in the neonatal period4 when diarrhoea is not a principal cause of death.3 Registration of births was 96% complete in 1972-75.1 We can look at the bias in classification of deaths by examining seasonality of death. Diarrhoea is the most highly seasonal of major causes of death; for example, in 1983 56% of registered diarrhoeal deaths occurred in the four hot months from May to August. If there was a true decline in diarrhoeal mortality, the degree of seasonality of mortality should also have declined. Moreover, month of death is more accurately coded than cause. Thus, trends in seasonality are an alternative indicator of trends in cause of death without the same potential for classification error. The proportion of annual infant deaths from May to August between 1970 and 1987 stayed above 45% until the end of 1983, after which it declined sharply (figure). Since 33-3% in a 4-month period represents no seasonality, more than half of the seasonality of mortality noted in 1983 had disappeared by 1987. Could any decrease in the number of diarrhoea associated deaths have been offset by an increase in the number of deaths from other causes, either from replacement mortality or classification bias? Table III shows infant death rates by cause and season from 1983 to 1987. Despite sharp seasonal changes in diarrhoeal deaths, there was little or no increase in non-diarrhoeal deaths in the seasons that immediately followed. Declines in diarrhoeal deaths within a given season were not offset by equivalent increases in nondiarrhoeal deaths; this is further evidence against substantial changes in classification bias. Table 11 also suggests that there was no compensatory increase in later childhood
mortality. That girls
TABLE III-INFANT MORTALITY, 1983-87
were more likely than were boys to die from diarrhoeal and respiratory diseases but not from other causes is consistent with service statistics showing that boys were more likely to be taken to rehydration clinics.
Other mortality surveys
Source: national civil registration statistics, CAPMAS. Infant deaths due to diarrhoeal/non-diarrhoeal disease births
are
given
per 1000 live
(table w)
The cluster sample double round household surveys show that between 1984 and 1986 total mortality in children under 2 years old for the summer and autumn diarrhoea season declined by 38%; most of the decrease was between
337
1984 and 1985, and 82% of the decrease was in the diarrhoea associated component. The prevalence of diarrhoea in the 2 weeks before the first round did not vary substantially in the 3 years (30%, 29%, and 26%). In the Dakahlia study in 1980, diarrhoeal mortality in children under 5 was 50% lower in the study area than in the control area; this was associated with a higher rate of ORS use in the study area compared with the control area (53 % vs 21 % of diarrhoeal episodes).When the study was repeated
in 1986, there were no longer any differences in mortality and ORS use between the study and control areas. However, in both areas diarrhoeal mortality was substantially lower (by 59% and 70%, respectively) and ORS use in both areas was much higher (to 85% and 77%, respectively) than in 1980; the number of ORT prescriptions per diarrhoea episode by physicians in clinics in the control area increased from 48 to 95%.’ Civil registration data for the whole governorate reflect the same mortality trend (see table IV in
ref 7). In
Menoufia,
a
controlled ORS home distribution
experiment in 1979-80 did not lower mortality, perhaps largely because of poor acceptance by physicians: "Only one-third of the children who died of diarrhoea received oral fluid therapy during their terminal episode, although over 90 percent received medical care". However, between 1979-80 and 1987-88, registered under-five mortality declined by 64%; interviews showed that 50% of the deaths were due to diarrhoea in 1979-80, and 39% in 1987-88. Mortality in May-August was 52% of the annual total mortality in 1979 but was 37% in 1987. Physicians now prescribed ORS in 66% of cases for which they were consulted, while mothers said that they used ORS in 52% of diarrhoea episodes. As in Dakahlia, civil registration data for the whole governorate reflect a similar mortality trend
(CAPMAS). Other factors associated with decrease in mortality The data above show that case management of diarrhoea has clearly improved as a result of the NCDDP. However, the risk of death may also be directly affected either by changes in diarrhoea incidence or by host resistance. The incidence of diarrhoeal disease has been measured repeatedly by NCDDP between 1983 and 1988. The data do not point to any decline in incidence of diarrhoea. For example, in seven separate studies, seasonally adjusted prevalence in the previous week ranged from 20 to 38 % with no obvious secular trend. The most useful indicator of host resistance is nutritional status. National surveys by the Nutrition Institute, Cairo, in 1978, 1980, and 1986 suggest that the frequency of severe malnutrition was relatively low with little change. Children who have recently had measles are at increased risk of death from diarrhoea. That measles vaccination increased from 41 % of children aged 12 to 23 months in 1983 to 76 % in 1987may have contributed to the reduction in child mortality; however, since measles immunisation begins at age 9 months, this could not have substantially affected infant mortality.
episodes and is correctly used. Moreover, most mothers continue to breastfeed children during the children’s illness. When dehydration from diarrhoea is a major cause of death, mortality reduction will be primarily due to correct use of ORT; we have strong evidence that ORS is more likely to be used for children who are more dehydrated. During the project period mortality-both absolute and relative to other causes of death-has decreased substantially. A reduction in diarrhoeal mortality was already underway after 1970, but was clearly hastened, first with early availability of ORS, and then even more rapidly with the inception of NCDDP. These declines occurred throughout the country. If the diarrhoea-specific mortality rates of 1982 had occurred in 1987, there would have been 64 261 more infant and childhood diarrhoeal deaths nationwide (not adjusting for under-registration) than the 38 451 observed. Could factors other than NCDDP have contributed to such changes? The immediate determinants of diarrhoeal mortality may be classified as curative care, incidence, and host resistance; other factors must operate through one or more of these determinants. There was no major change in the incidence of diarrhoea or nutritional status during the course of the project. We cannot exclude changes in the severity of illness (eg, subtle changes in the virulence of microbes causing diarrhoea) but ORT together with continued feeding reduces duration of diarrhoea.12 Were children who recovered from diarrhoea more likely to die from other causes? Mosley and Becker13 suggest that children who recover from diarrhoea are frail and thus more likely to die later. We have no evidence that this happens within large populations. In particular, there was no increase in death rates in subsequent seasons from other causes, or at later stages. Although our data strongly indicate that NCDDP has had an impact on diarrhoeal mortality, we cannot say exactly which inputs were specifically responsible. Use of ORS, feeding practices during diarrhoeal episodes, and emergency treatment for severe dehydration have probably each made an important contribution. The declines were due to a comprehensive programme of curative care for diarrhoeal disease, rather than simply to an ORT project. Mothers’ decisions in child care are vital; but in Egypt the availability of clinics and correct practice by physicians are also essential. Hirschhorn et ap4 have suggested a structure to decide which primary health care inputs are appropriate to specific situations. A decrease in diarrhoeal mortality in Egypt was possible because of moderate levels of malnutrition and immunisable disease mortality, a well-structured health system, and good access to the media. In countries without these characteristics, a programme similar to NCDDP may not be as successful. The NCDDP must now find the causes of the residual mortality ascribed to diarrhoeal disease, and develop methods to remove them. The incidence of diarrhoea must be decreased by improved hygiene and sanitation, and the severity and duration of illness must be lessened by better nutrition during and after illness.
Discussion We believe that the NCDDP has influenced care of diarrhoeal disease enough to account for substantial changes in mortality. ORS is available and readily accessible; and ORT is prescribed regularly in over half of diarrhoeal
The NCDDP would like to thank Dr M. Gabr, Dr R. Dewidar, Dr M. Hammamy, Dr E. Helw, Dr L. El Sayad, Dr A. A. Shaarawi, Dr H. Tammam, Dr A. Hashem, Dr A.-M. Hassan, and Dr M. Hammouda; and Dr Reginald Gipson from John Snow Inc., Boston, USA. This
study was funded partly by USAID grant no 263-0137.
338
REFERENCES 1. National
Academy of Sciences Committee on Population and Demography. The estimation of recent trends in fertility and mortality in Egypt. Report no 9. Washington: National Academy Press, 1982. 2. Kielmann AA, Mobarek AB, Hammamy MT, et al. Control of deaths from diarrheal disease in rural communities. I. Design of an intervention study and effects on child mortality. Trop Med Parasitol 1985; 36: 191-98. 3. Tekce B. Oral rehydration therapy: an assessment of mortality effects in rural Egypt. Studies Family Planning 1982; 13: 315-27. 4. Infant and child mortality in Egypt. Cairo: Arab Republic of Egypt Central Agency for Public Mobilization and Statistics, 1986. 5. The State of Egyptian children. Cairo: Arab Republic of Egypt Central Agency for Public Mobilization and Statistics, and UNICEF, 1988. 6. Keilmann AA, Nagaty A, Ajello CA. Control of deaths from diarrheal disease in rural communities. II. Motivating and monitoring the community. Trop Med Parasitol 1986; 37: 15-21. 7. National Control of Diarrheal Diseases Project. Impact of the national control of diarrhoeal diseases project on infant and child mortality in
Dakahlia, Egypt. Lancet 1988; ii: 145-48. 8. Ewbank DC. Uses of mortality data for evaluating the success of scientific health and development programmes. In: Data bases for mortality measurement. New York: United Nations, 1984. 9. Newell KW. Selective primary health care: the counter revolution. Soc Sci Med 1988; 26: 903-06. 10. Rashad H. Oral rehydration therapy and its effect on child mortality in Egypt. J Biosoc Sci 1989; 10 (suppl): 105-13. 11. Rashad H. Evaluation of the completeness of mortality registration in Egypt. West Asia and North Africa Regional Papers. Cairo: The Population Council, 1981. 12. Santosham M. Nutritional aspects of ORT. J Diarrhoeal Dis Res 1987; 5: 270-74. 13. Mosley WH, Becker S. Demographic models for child survival: implications for program strategy. Presented at: Child Survival Programs: Issues for the 1990s, Johns Hopkins University, Baltimore, November 1988. 14. Hirschhorn N, Grabowsky M, Houston R, et al. Are we ignoring different levels of mortality in the primary health care debate? H Pol Planning 1989; 4: 343-47.
First coitus before menarche and risk of transmitted disease
The
prevalence of sexually transmitted disease (STD), pelvic inflammatory disease (PID), and
cervical cancer, and the relation between these conditions were studied in 2111 Ethiopian women. Early sexual activity was associated with an increase in prevalence rates of STD and PID; possible aetiological factors include physical and immunological immaturity of the female genital tract and the number of sexual partners.
Introduction Cervical
cancer
is
an
important
cause
of
cancer
mortality
among women in Ethiopia and in much of the rest of Africa,
where patients often present with advanced disease. In high incidence areas, age at first coitus, specifically before age 21, 20, or 17 years is the most important risk factor,t-3 with a doubling of the risk in those having first coitus at age 16 or less.4 The association between cervical cancer and the number of sexual partners is well established: in married women with cervical cancer who claim not to have had a sexual partner other than their husband, the number of sexual partners of their husbands was a significant risk factor for cervical cancer. This male promiscuity has been called the "male factor".6 Only one study has reported an association between cervical cancer and chronic pelvic inflammatory disease (PID) and sexually transmitted disease (STD).7 A study in Ethiopia in 1975 showed that STD (particularly syphilis and gonorrhoea) was highly prevalent; prevalence rates for other STD were unknown, although acute and chronic PID
sexually
and their
complications were the most important cause of and mortality in women.8,9 An international collaborative study was therefore planned to investigate the prevalence of STD and cervical cancer in an unscreened population, the causal factors of cervical cancer to define an "at risk" group, and possible associations between STD and
morbidity
cervical
cancer.
Patients and methods 2111 unselected Ethiopian women were studied; half were outpatients with gynaecological symptoms, and half were healthy women attending family planning, antenatal, and postnatal clinics in Addis Ababa. The collection of data, cytology slides, and sera, and the VDRL test were done in 1975 and 1976 while tribal, religious, and other factors could still be assessed independently of population migration (before the Ethiopian revolution). It was not until 1977 and 1978 that the data and specimens could be despatched from the country, although full permission had been given by the Ethiopian Ministry of Health. The patients’ data, used with their consent, were as follows: Personal details were obtained by questionnaire (completed by an Ethiopian female assistant), and included age, hospital/clinic attended, address (origin: town or rural), tribe, religion, parity, total number of pregnancies, "marital status" (single, married, divorced, widowed, bargirl, prostitute, "talla" [local beer] seller, maid), monthly family income, age at first marriage and first coitus and
ADDRESSES:
Department of Bacteriology, Edinburgh University, Edinburgh, UK (M. E. Duncan, MD, J. F. Peutherer, MD, P. Simmonds, PhD, H. Young, PhD); Liège University, Belgium (G. Tibaux EE, A. Pelzer, MD); Statens Seruminstitut, Copenhagen, Denmark (K. Reiman); and Institute of Ophthalmology, London, UK (Y. Jamil, MSc, Prof S. Daroughar, MD). Correspondence to Dr M. E.