Signes d'alerte d'un mauvais état de santé chez le nouveau-né : perception par les personnes s'occupant de nourrissons et par les agents de santé dans le nord de l'Inde
Signos de peligro de las enfermedades neonatales: impresiones de los cuidadores y de los trabajadores sanitarios en el norte de la India
Shally AwasthiI,1; Tuhina VermaI; Monica AgarwalII
IDepartment of Pediatrics, King George's Medical University, Lucknow (UP) India 226003
IIDepartment of Preventive and Social Medicine, King George’s Medical University, Lucknow, India
OBJECTIVE: To assess household practices that can affect neonatal health, from the perspective of caregivers and health workers; to identify signs in neonates leading either to recognition of illness or health-care seeking; and to ascertain the proportion of caregivers who recognize the individual items of the integrated management of neonatal and childhood illnesses (IMNCI) programme.
METHODS: The study was carried out in a rural community in Sarojininagar Block, Uttar Pradesh, India, using qualitative and quantitative research designs. Study participants were mothers, grandmothers, grandfathers, fathers or "nannies" (other female relatives) caring for infants younger than 6 months of age and recognized health-care providers serving the area. Focus group discussions (n = 7), key informant interviews (n = 35) and structured interviews (n = 210) were conducted with these participants.
FINDINGS: Many household practices were observed which could adversely affect maternal and neonatal health. Among 200 caregivers, 70.5% reported home deliveries conducted by local untrained nurses or relatives, and most mothers initiated breastfeeding only on day 3. More than half of the caregivers recognized fever, irritability, weakness, abdominal distension/vomiting, slow breathing and diarrhoea as danger signs in neonates. Seventy-nine (39.5%) of the caregivers had seen a sick neonate in the family in the past 2 years, with 30.38% in whom illness manifested as continuous crying. Health care was sought for 46 (23%) neonates. Traditional medicines were used for treatment of bulging fontanelle, chest in-drawing and rapid breathing.
CONCLUSION: Because there is no universal recognition of danger signs in neonates, and potentially harmful antenatal and birthing practices are followed, there is a need to give priority to implementing IMNCI, and possible incorporation of continuous crying as an additional danger sign.
OBJECTIF: Déterminer les pratiques domestiques devant être identifiées par les personnes s'occupant des nourrissons et les agents de santé comme potentiellement nuisibles à la santé des nouveau-nés; reconnaître les signes indiquant chez ces enfants la présence d'une maladie ou un besoin de soins de santé; et déterminer la proportion de personnes s'occupant de nourrissons capables de reconnaître les différents éléments guidant l'IMNCI (prise en charge intégrée des maladies néonatales et infantiles).
MÉTHODES: L'étude a été effectuée au sein de la communauté rurale de Sarojininagar dans la province de l'Uttar Pradesh (Inde) au moyen de méthodes de types qualitatif et quantitatif. Ont été inclus dans cette étude des mères, grand-mères, grand-pères, pères ou «tantes» (autres membres féminins de la famille) s'occupant de nourrissons de moins de 6 mois, ainsi que des prestateurs de soins de santé reconnus et délivrant des services dans cette zone. Des discussions en groupe cible (n = 7), des interrogatoires d'informateurs clés (N = 35), ainsi que des entretiens structurés (n = 210), ont été menés avec ces sujets.
RÉSULTATS: De nombreuses pratiques domestiques potentiellement nuisibles à la santé de la mère ou du nouveau-né ont été relevées. Parmi les 200 personnes s'occupant de nourrissons, 70,5 % ont signalé des accouchements à domicile pratiqués par des aidants sans formation médicale ou des membres de la famille et, pour la plupart des mères, un allaitement au sein ne débutant qu'au 3e jour après l'accouchement. Plus de la moitié des personnes s'occupant de nourrissons ont identifié la présence de fièvre, l'irritabilité, la faiblesse, un ballonnement abdominal/des vomissements, une respiration lente ou une diarrhée comme des signes d'alerte chez le nouveau-né. Soixante-dix neuf (39,5 %) des personnes s'occupant des nourrissons avaient déjà vu un nouveau-né malade dans leur propre famille au cours des 2 années précédentes, chez lequel la maladie se manifestait par des pleurs continus dans 30,38 % des cas. Des soins de santé ont été apportés à 46 (23 %) des nouveaunés. Des méthodes relevant des médecines traditionnelles ont été utilisées pour traiter des cas de fontanelle bombante, de thorax en entonnoir ou de respiration rapide.
CONCLUSION: Compte tenu de la reconnaissance non systématique des signes d'alerte chez le nouveau-né et de l'existence de pratiques potentiellement dangereuses avant et pendant l'accouchement, il est prioritaire de mettre en uvre l'IMNCI et d’envisager la prise en compte des pleurs continus parmi les signes d'alerte.
OBJETIVO: Evaluar las prácticas domésticas que pueden influir en la salud neonatal desde la perspectiva de los cuidadores y los trabajadores sanitarios; identificar los signos observables en los recién nacidos que conducen al reconocimiento de enfermedades y la búsqueda de atención sanitaria; y evaluar la proporción de cuidadores capaces de reconocer los distintos elementos del programa de atención integrada a las enfermedades neonatales y de la infancia (IMNCI).
MÉTODOS: El estudio se llevó a cabo en una comunidad rural de Sarojininagar Block, Uttar Pradesh, India, usando técnicas de investigación cualitativas y cuantitativas. Participaron en él madres, abuelas, abuelos, padres y «niñeras» (otros familiares femeninos) que cuidaban a lactantes de menos de 6 meses, así como dispensadores de salud acreditados que trabajaban en la zona. Con ellos se organizaron grupos de discusión dirigidos (n = 7), entrevistas con informantes clave (n = 35) y entrevistas estructuradas (n = 210).
RESULTADOS: Se observaron muchas prácticas domésticas que podían perjudicar la salud materna y neonatal. De 200 cuidadores, el 70,5% informaron de partos en el hogar atendidos por enfermeras no preparadas o por familiares, y la mayor parte de las madres sólo empezaban a dar el pecho al tercer día. Más de la mitad de los cuidadores reconocían la fiebre, la irritabilidad, la debilidad, la distensión abdominal/vómitos, la respiración lenta y la diarrea como signos de peligro en los recién nacidos. Setenta y nueve (39,5%) cuidadores habían atendido a un recién nacido enfermo en la familia en los dos últimos años, y en un 30,38% de los casos la enfermedad se manifestó en forma de llantos continuos. Se buscó atención sanitaria para 46 recién nacidos (23%). Se usaron medicinas tradicionales para tratar casos de abombamiento de la fontanela, tiraje torácico y respiración rápida.
CONCLUSIONES: Teniendo en cuenta la falta de criterios universales para reconocer los signos de peligro en los recién nacidos, así como los riesgos que encierran algunas prácticas de atención prenatal y asistencia al parto, es necesario dar prioridad a la aplicación de la IMNCI, y considerar la posible inclusión del llanto continuo entre los signos de peligro.
Globally 10 million children die annually before their fifth birthday, most of them in the neonatal period.1 More than 98% of these deaths occur in developing countries. Almost half of the deaths in under-five-year-olds occur in infancy. Of the infant deaths, about two-thirds occur in the neonatal period. It has also been noted that one-third of all neonatal deaths occur on the first day of life, almost half within 3 days and nearly three-quarters within the first week of life.2 In developing countries, about 34 of every 1000 live births result in neonatal death.1
In India the neonatal mortality rate (NMR) dropped significantly from 69 per 1000 live births in 1980 to 53 per 1000 live births in 1990.2 In recent years, however, the NMR has remained almost static decreasing only from 48 to 44 per 1000 live births from 1995 to 2000. A similar situation has been reported from other developing countries.2
The primary causes of neonatal death are sepsis (52%) (which includes pneumonia, meningitis, neonatal tetanus and diarrhoea), birth asphyxia (20%), prematurity (15%) and others (13%).2 Lack of specificity of the clinical manifestations of various neonatal morbidities has been noted, resulting in difficulty in making a definitive diagnosis,3 delay in seeking care and resultant high mortality.4 However, the Integrated Management of Neonatal and Childhood Illnesses (IMNCI) approach has attempted to provide a standard case definition of various neonatal morbidities, for example neonatal sepsis, jaundice and pneumonia, based on presence of certain clinical signs.5 For effective implementation of the IMNCI strategy it is necessary for the caregivers and health-care providers to recognize danger signs in a sick neonate and thereafter seek the appropriate level of health care, which in turn would reduce mortality.6 This has been the basic conceptual framework for improved neonatal care in developing countries.7
The present study was conducted to: assess the household practices that can affect neonatal health, from the perspective of the caregivers and health workers; identify signs in neonates leading either to recognition of illness or health-care seeking the "danger signs"; and to ascertain the proportion of caregivers who recognize the individual items of the IMNCI module.
Study setting and study location
This work was done from May to November 2005 in Sarojninagar, a Block in the Lucknow district of Uttar Pradesh, northern India, with a population of about 2 million 80% of which is rural spread over 190 villages. The government has set up one community health centre and four primary health centres in the area. Here curative services are primarily provided by doctors, while preventive services, such as immunization and antenatal care, are provided by auxiliary nurse midwives (ANMs) in the community. In addition, there are private and traditional health-care providers, crudely estimated at one per village. At the time of this study, IMNCI had not been introduced in this area and there were no special neonatal health-care providers.
A triangulated design, combining both quantitative and qualitative methods was used in this study. Qualitative methods, such as focus group discussions (FGDs) and in-depth interviews with key informants, were used to gain a deeper understanding of the health-seeking behaviour. A quantitative survey of a separate set of respondents used a structured, pre-tested questionnaire to assess which signs were recognized by caregivers as danger signs in neonates.
The participants were caregivers (mothers, fathers, grandmothers, grandfathers and other female relatives) and health-care providers (community health-care workers, traditional birth attendants (TBAs), nurses, midwives and community doctors). The study included those caregivers who had given primary care to a newborn within the last 6 months, were permanent residents of the village and had consented to participate in the study. For the quantitative study, the first six interviews, if several eligible caregivers were present, were conducted with the mothers. Thereafter, based on availability of respondents, the paternal grandmother, maternal grandmother, grandfather and father were interviewed in order of preference, one per household.
An interview guide was used for conducting FDGs and for interviews with key informants. The elements of the interview guide are given in Box 1. On the basis of these findings a structured questionnaire was prepared for use when interviewing caregivers.
Sampling framework and sample size
For the qualitative study, the number of interviews was guided by the point of saturation (i.e. they stopped when no new information was being given). For the quantitative component of the study, which involved recognition of the individual danger signs, sample sizes were calculated as those needed to obtain adequate statistical precision. To obtain 95% confidence intervals and 7% precision on the assumption that 50% of respondents will recognize a danger sign, we interviewed 200 caregivers.
For key informant interviews, villages were chosen purposively. For FDGs and quantitative interviews, villages were chosen by random selection from 190 villages listed by the governmental Integrated Child Development Services (ICDS) system. For quantitative interviews, the selection of the first household within the village was done by random selection from the list of infants less than 6 months of age, maintained by the aganwadi worker (health visitor) employed under the ICDS system. Thereafter, the team went from door to door in a randomly chosen direction to identify eligible households from which to interview caregivers.
Information on the perceptions of 80% of the eligible community health workers and medical practitioners were collected either as key informant interviews, FDGs or structured interviews. Informed consent was obtained from the eligible respondents for participation and none refused.
Representative accounts, anecdotes and case-reports of how practices and perceptions influence the health of neonates were prepared from FDGs and interviews with key informants. These data were manually analysed and structured allowing keywords and phrases to be identified and grouped in domains. Responses were recorded as follows: majority (> 75%), most (50-75%), some (25-50%) and few (< 25%) respondents gave similar replies. For quantitative data, univariate analysis was used and we report frequency distribution with proportions and 95% confidence intervals.
The study was conducted with ethical approval from the Institutional Ethical Review Board of King George's Medical University, Lucknow, and the United States Agency for International Development (USAID) Institutional Review Board established for the Indian Clinical Epidemiology Network.
Data were collected from nine villages (53 caregivers) for the qualitative, and 20 villages (200 caregivers) for the quantitative parts of the study. The demographic profile of the caregivers in the quantitative study is shown in Table 1. There were 23 in-depth interviews and 5 FGDs. FGDs were conducted with three groups of eligible mothers and two groups of eligible grandmothers or other female relatives.
Key informant interviews were conducted with medical doctors, general practitioners, paediatricians and neonatal specialists, (n = 4), other health workers such as ANMs (n = 4), TBAs (n = 2) and volunteer health workers (n = 2). FDGs were also conducted with two groups: one of ANMs and one of TBAs. For quantitative information 10 female health-care workers (ANMs; n = 5: health supervisors; n = 3: TBAs; n = 2) were interviewed, most of whom (80%) had attended a training programme related to care of mothers and neonates in the past 3 years.
Recognition of signs requiring health-care seeking during pregnancy
Based on the analysis of information from key informants, FDGs and structured interviews, caregivers recognized potentially risky conditions during pregnancy rather than clinical signs in the neonate warranting health-care seeking. Most of the respondents considered maternal malnourishment and "small womb" as important risk conditions responsible for producing a "sick" neonate. Conditions in mothers recognized by some as leading to poor pregnancy outcome were: general sickness, "too little intake, particularly of leafy green vegetables," presence of any illness (fever, vomiting, frequent stools, oedema of legs, "inactiveness" and "maternal overeating and/or eating many times a day"). In contrast few women considered infrequent eating or undereating a risk condition for fetal growth. Medical risk conditions such as decreased fetal movements, anaemia and premature rupture of the membranes were also identified by a few respondents as reasons for seeking health care.
Almost all the health-care workers recognized the following signs requiring medical care during pregnancy: vaginal bleeding, vaginal discharge, anaemia and fever. Some health workers also recognized other conditions requiring health care, such as nausea and vomiting; decreased fetal movements or abdominal pain; abdomen larger than previous pregnancy; abdomen smaller than previous pregnancy; short stature of the mother; pregnancy at a young age or late pregnancy; oedema of the face, legs or hands; or pain during urination.
Beliefs and practices during the antenatal period
The community believed that certain things were to be avoided during pregnancy. Most believed that pregnant women should avoid tea, rice and certain lentils (urad dal) as these were "hot" or "cold" food, and rice was thought to cause a lot of white layering on a neonate's body at birth. They felt that the mother should not eat fried food or sour food (e.g. pickles). Respondents thought that pregnant women should not eat large quantities of food for fear that the baby would grow too large and the woman would subsequently experience difficulties during delivery or the mother's stomach would be so full with food that there would not be enough space for the fetus to grow. Pregnant women should also not take "excessive and unnecessary" rest and should avoid lifting weights, fast walking and climbing stairs. Few recommended abstinence from sex during pregnancy.
Although most (n = 200; 70.5%) deliveries took place at home, some women gave birth in government hospitals (n = 52; 26%) or private hospitals (n = 7; 3.5%). Home delivery usually took place in a clean room, with the floor painted with cow dung. For delivery boiled water was kept ready together with washed, used cloths. In addition to the mother, there were four other important actors in the birthing process: a local female TBA (dai) who assisted in delivery, a local woman belonging to a special class called the domain who cut the cord with a new blade and tied it, a local massage-woman (noun) who gave mother and child a religious bathing and the ANM/local doctor who gave the mother and baby an injection after delivery. Although few respondents knew the nature of this injection, some of them said it was tetanus toxoid. The dai also cleans the neonate's tongue. "She takes a soft cloth dipped in mustard oil and with the help of a finger cleans the child's mouth. It helps to eject the dirty water out of the child's stomach; that water which the child had swallowed when in the mother's stomach." The domain also cleaned the room after delivery and removed the placenta for disposal.
All health providers mentioned bathing or washing the baby immediately after cutting of the cord, oil massage and kajal or soot application to the eyes as normal procedures (Table 2). Most mothers initiated breastfeeding almost three days after birth and discarded colostrum (n = 128; 64%). Pre-lacteal feeds which were given to almost all the neonates soon after birth were honey mixed in water (n = 112; 56%), cow's milk (n = 96; 48%) or goat's milk (n = 50; 25%), generally administered with the help of a cotton wick. Care of the umbilical stump included application of mustard oil to keep away insects. Almost all the caregivers kept the baby out of the sun for at least a week and kept a fire lit for 24 hours at the entrance of the mother and baby's room to protect them from the evil eye/spirit which results in Jamogha, a condition where the neonate's body turns stiff and blue. Mothers were not allowed to leave the delivery room for 42 days after birth; this period is called the saour.
Recognition of danger signs in newborns
Caregivers' recognition of danger signs in newborns and their corresponding health seeking behaviour is shown in Table 3. Seventy-nine (39.5%) of the caregivers had seen a sick neonate in their own family in the past 2 years. The clinical presentations seen by them are listed in Table 4. Continuous crying was reported as a common manifestation of neonatal illness and this was supported by the findings of eight key informant interviews with caregivers who had experienced adverse neonatal events or death (data not given).
Utilization of health-care services for sick neonates
Twenty-three per cent (46/200) of respondents sought health care or administered medicines for neonatal illness. The preferred health-care provider was either a local medical doctor (registered or non-registered) (60.7%; 28/46), followed by a traditional healer (19.6%; 9/46) while the remainder were treated with home remedies. Modern medicines were administered to 78.3% (36/46), while the rest used indigenous medicine and traditional homemade medicines, either alone or in combination with modern medicine. The majority of the respondents who had sought any health care said their neonate "improved" after treatment and only a few had complications. The main reasons for the choice of health-care provider were their proximity and whether there was "dispensing of medicines" at consultation.
The qualitative study also revealed that traditional medicines were used for bulging fontanelle, chest in-drawing and rapid breathing. Registered or non-registered medical practitioners were consulted only "in case of herbal medicines failure." The government services were used only following referral by a local doctor or in self-assessed "critical situations."
In the Sarojninagar community of the Lucknow district of Uttar Pradesh, northern India, the majority (70.5%) of births take place at home, attended by untrained personnel. Most of the neonates were bathed soon after birth; mothers discarded colostrum and did not start breastfeeding until the third day postpartum. Health-care providers recognized some, but not all, of the danger signs in pregnancy as well as in the neonate. Furthermore, traditional medicines were used for possible cases of neonatal sepsis. These findings possibly explain the high neonatal mortality rate of 51.0 per 1000 live births reported from Lucknow district. A similarly high neonatal mortality rate of 53.6/1000 live births was reported from Uttar Pradesh.8
In developing countries, most of the births and deaths of neonates occur at home7 and the majority of neonates are not taken to a health-care provider when they are ill.1 In India less than 25% of deliveries take place in a hospital.8 Traditional practices preclude caregivers and parents from taking neonates outside the home even if they are ill. In a study on care-seeking and adherence to treatment for neonatal illnesses conducted in a periurban cohort in New Delhi, India, it was found that 60% of deaths occurred within 24 hours of recognition of illness, 40% of caregivers did not seek outside care, and 70% of care was sought from private providers. Half of these private providers had no formal medical education, and failed to refer 70% of the newborns who eventually died. Fewer than half of caregivers followed referral recommendations.9
In Lucknow we found that signs that are frequently observed in sick neonates are recognized by more than one-third of the caregivers, unlike in Bangladesh where there was poor awareness of the danger signs.10 However, as in Bangladesh, breastfeeding was not initiated until three days after delivery. In other states of India such as Kerala, Tamil Nadu, Mizoram and Meghalaya, a higher rate of early initiation of breastfeeding has been reported.8 Although not assessed in the current study, low coverage of antenatal care has been reported from Uttar Pradesh.8 Since counselling about breastfeeding is part of antenatal care this could explain the low rates of early initiation of breastfeeding and high rates of pre-lacteal feeding. There is a need to intensify efforts at promotion of breast-feeding as a strategy to reduce neonatal mortality. Except for discarding colostrum, other newborn care practices, such as bathing and massaging the baby with mustard oil are similar to those reported from Pakistan.11
We found a similarity between caregivers' self-reported recognition of danger signs (Table 3) and those actually observed by them (Table 4). Continuous crying was also observed in sick neonates. This may be considered a danger sign that should be incorporated into the IMNCI list. While 38% (79/200) of caregivers had cared for an ill neonate in the past 2 years, only 23% (46/200) had taken medical advice. Thus, the need for changes in behavioural practices had to be communicated to improve utilization of health services in the study area. A trial conducted in Nepal showed that community-based participatory intervention had a positive impact on uptake of antenatal and delivery services; home care practices; infant morbidity; and health-care seeking.12 Another cluster-randomized trial conducted in rural India found that training doctors in counselling using the IMNCI approach improved mothers' appreciation of the need to seek prompt and appropriate care for severe episodes of childhood illness, but their care-seeking behaviour did not improve significantly.13
We found a paradoxical situation in the study area where modern medicines for neonatal illnesses were dispensed by unqualified health-care providers and traditional medicines were used for potentially bacterial infections. In such cases, the traditional healer plays a major role in delaying the seeking of appropriate health care for the sick neonate. If we can combine improved care-seeking with better management by doctors and prompt and effective referrals we may reduce neonatal mortality. Similar findings have been reported in a study of care-seeking patterns in malaria patients in the United Republic of Tanzania.14
The current study was conducted in a rural community using qualitative and quantitative methods to collect data. Although there may have been recall bias in reporting signs by caregivers who had experienced neonatal adverse events in the past 2 years, adverse community practices and continuous crying, a previously unrecognized neonatal danger sign, have been identified. Continuous crying may be incorporated in the IMNCI list of danger signs. The IMNCI training can also be modified locally to emphasize the need to change practices. Similar studies in culturally diverse areas would also be useful. However, since there is no universal recognition of danger signs in pregnant women and neonates, there is an urgent need for improving community awareness by extensive information, education and communication campaigns.
This study was funded by the United States Agency for International Development-Child Health Research (USAID-CHR), Washington, DC (via the International Clinical Epidemiology Network (INCLEN) Trust) and conducted through the INCLEN Childnet. The international coordinator for this study was Jacinto Blas V Mantaring III, Department of Clinical Epidemiology and University of the Philippines, Manila. The protocol was developed by the INCLEN Childnet Neonatal Danger Signs group. We thank Dr Jose Martinez, Child and Adolescent Health and Development (CAHD) Division, WHO and Dr Gary Darmstadt of Johns Hopkins University, who gave technical inputs during protocol development.
Competing interests: none declared.
1. Black RE, Morris SS, Bryce J. Where and why are 10 million children dying every year? Lancet 2003;361:2226-34.
2. United Nations Children's Fund. State of the world's newborns 2001. Washington, DC: Save the Children Publication; 2002.
3. Stoll BJ. The global impact of neonatal infection. Clin Perinatol 1997;24:1-21.
4. Sutrisna B, Reingold A, Kresno S, Harrison G, Utomo B. Care seeking for fatal illnesses in young children in Indramayu, West Java, Indonesia. Lancet 1993;342:787-9.
5. World Health Organization. Handbook IMNCI integrated management of neonatal and childhood illnesses. Geneva: WHO; 2003. WHO document WHO/FCH/CAH.
6. Desilva MWA, Wijekoon A, Hornik R, Martines J. Care seeking in Sri Lanka: one possible explanation for low childhood mortality. Soc Sci Med 2001;53:1363-72.
7. Marsh DR, Darmstadt GL, Moore J, Daly P, Oot D, Tinker A. Advancing newborn health and survival in developing countries: a conceptual framework. Saving Newborn Lives Initiative. J Perinatol 2002;22:572-6.
8. National Family Health Survey 2 Uttar Pradesh 1998-1999. Mumbai: International Institute of Population Sciences; 2001.
9. Bhandari N, Bahl R, Taneja S, Martines J, Bhan KM. Pathways to infant mortality in urban slums of Delhi, India: implications for improving the quality of community and hospital based programmes. J Health Pop Nutr 2002;20:148-55.
10. Baqui AH, Arifeen El S, Darmstadt GL, Black ER, Santoshan M. Final report: formative research on newborn care practices in the home and pre-testing of alternative behaviors in Sylhet District, Bangladesh November 2003. Baltimore, MD: Johns Hopkins Bloomberg School of Public Health; 2003. Available at: http://www/webdrive.jhsph.edu/pwinch/SYL_Formative_report_03oct29.pdf
11. Fikree FF, Ali TS, Durocher JM, Rahbar MH. Newborn care practices in low socioeconomic settlements of Karachi, Pakistan. Soc Sci Med 2005;60:911-21.
12. Manandhar DS, Osrin D, Shrestha BP, Mesko N, Morrison J, Tumbahangphe KM, et al. Effect of a participatory intervention with women's groups on birth outcomes in Nepal: cluster-randomised controlled trial. Members of the MIRA Makwanpur trial team. Lancet 2004;364:970-9.
13. Mohan P, Iyengar SD, Martines J, Cousens S, Sen K. Impact of counseling on care seeking behavior in families with sick children: Cluster randomised trial in rural India. BMJ 2004;329:269.
14. de Savigny D, Mayombana C, Mwageni E, Masanja H, Minhaj A, Mkilindi Y, et al. Care-seeking patterns for fatal malaria in Tanzania. Tanzania Essential Health Interventions Project. Malar J 2004;3:3.
(Submitted: 16 December 2005 - Final revised version received: 1 June 2006 - Accepted: 13 June 2006)