- Email: [email protected]
Maternal diagnosis and treatment of children's fever in an endemic malaria zone of Uganda: implications for the malaria control programme
Acta Tropica 68 (1997) 53 – 64
Maternal diagnosis and treatment of children’s fever in an endemic malaria zone of Uganda: implications for the malaria control programme Rosalind G.N. Lubanga a,*, Sandra Norman b, Douglas Ewbank c, Charles Karamagi d a Clinical Epidemiology Unit, Makerere Uni6ersity, P.O. Box 7423, Kampala, Uganda Center for Clinical Epidemiology and Biostatistics, Uni6ersity of Pennsyl6ania, Philadelphia, PA 19 104, USA c Population Studies Center, Uni6ersity of Pennsyl6ania, Philadelphia, PA 19 104, USA d Department of Pediatrics/Clinical Epidemiology Unit, Makerere Uni6ersity Medical School, Kampala, Uganda
b
Received 27 March 1996; received in revised form 18 September 1996; accepted 21 January 1997
Abstract A mother’s ability to suspect malaria in the presence of fever has important consequences for child survival in malaria-endemic areas. This paper presents results of a clinic-based study of mothers’ abilities to suspect malaria in the event of recognizing fever and other physiological and behavioral changes associated with the disease. The study population consisted of all (439) women or mothers who had accompanied children 5 years and below to the Old Mulago Hospital, Kampala, Uganda over a 10 day period during the malaria season of 1992. The children were those who had fever as a major complaint at the time of the visit or those who had fever in the last 7 days and were visiting the clinic for the first time for the current illness. The children were physically examined and their blood tested for malaria parasites. Mothers’ diagnosis was compared with clinical and laboratory diagnosis of malaria. Mothers associated the presence of fever with several types of illness and malaria was often not suspected. Only 40% of the mothers suspected malaria in their children. The mothers were poor at recognizing malaria when, in fact, it was present. The sensitivity of the mothers’ diagnosis of malaria was found to be 37%; 63% of malaria cases were misclassified * Corresponding author. Fax: + 256 41 530022. 0001-706X/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S 0 0 0 1 - 7 0 6 X ( 9 7 ) 0 0 0 7 1 - 5
54
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
as other conditions. The doctors classified most (92%) of the cases presenting with fever as having malaria, but laboratory tests indicated that only 64% of the children really had malaria. The sensitivity of clinical diagnosis was 98%, but the specificity was only 18%. Ninety percent of the mothers gave some medicines before visiting the health centre; and, of these, 76% gave modern drugs exclusively, including antimalarials, antipyretics, antibiotics and other drugs. Among the modern drugs given to children suspected of having malaria, 50% were antimalarials. The most commonly used antimalarial was chloroquine tablets. Mothers indiscriminately administered antimalarials to children irrespective of the perceived cause of the fever. There is need to educate mothers to suspect malaria first in every case of febrile illness, just like the doctors do, and about the first line drugs for the treatment of malaria. © 1997 Elsevier Science B.V. Keywords: Malaria; Sick child; Index of suspicion; Child survival; Presumptive treatment; Uganda
1. Introduction Malaria is among the top ten leading causes of morbidity and mortality in Uganda, with high fatality rates among children under 5 years (Uganda Ministry of Health, 1992). A May 1992 study in the Kampala District indicated mesoendemic levels in this area (Uganda Ministry of Health, 1994). In malaria-endemic regions, children under 5 years have low levels of resistance. The most common form of malaria parasite in Uganda is Plasmodium falciparum (UNICEF, 1989), a dangerous type, which may be fatal to children if not treated within a few days of onset. To reduce morbidity and mortality, the World Health Organization (WHO) has recommended that in malaria-endemic areas all cases of fever should be given prompt treatment with anti-malarials such as chloroquine or other 4-aminoquinolines (Jeffery, 1984). The effectiveness of presumptive treatment depends upon: (a) knowledge of the symptoms suggestive of malaria attack; (b) use of effective drugs; (c) ability of the mother to follow a precise dose regimen; and (d) giving prompt treatment (Jeffery, 1984; Deming et al., 1986). The family has a big role to play in the control of malaria. It needs to recognize symptoms commonly suggestive of malaria, particularly fever and to give appropriate first aid treatment and seek health care (Bruce-Chwatt, 1987a; WHO, 1984, 1986; Beausoleil, 1984). Surveys carried out in Africa reveal that between 80 and 90% of the people treat their children and other family members at home (Jeffery, 1984; Adeniyi and Ramakrishna, 1985). Conditions commonly treated at home include fevers, diarrhoea, coughs and respiratory complaints. Mothers always recognize fevers in their children and it is on this basis that they either give some treatment at home or seek health care. But, as is well known, fever is a symptom of a plethora of conditions and diseases. The critical question in the discussion of presumptive treatment of malaria at home is what proportion of mothers suspect a possible attack of malaria when they recognize fever in the children?
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
55
The ability of the mother to suspect malaria in the presence of fever has important implications for child survival in malaria-endemic areas. An easily suspecting mother is more likely to give an anti-malarial than one who decides that the condition is measles or cough and common cold and only gives drugs specific for those conditions. It has been shown in the malaria literature that the risks involved when a patient is presumptively treated with chloroquine when he is not truly suffering from malaria are minimal compared to leaving the child untreated (Ewbank and Gribble, 1993; Bruce-Chwatt, 1987b). This paper describes results of a clinic-based study carried out in the malarial season of June-July 1992 in the Old Mulago Hospital, Kampala, Uganda. Its objectives were to find out: (a) how mothers interpreted the presence of fever and other physiological and behavioral signs and symptoms which are clinically associated with malaria in their children under 5 years; (b) how mothers’ diagnosis corresponded to clinical and laboratory diagnosis of fever and malaria; and (c) the medicines given to children with fever before bringing the children to the health centre. The ultimate goal of this study was to investigate whether mothers could be used more effectively in the treatment of fevers at home and if so how.
2. Methods
2.1. Data collection The study design was a descriptive cross-sectional prevalence study. The study population consisted of all women or mothers who had accompanied children 5 years and below to the Old Mulago Hospital. The children were those who had fever as a major complaint at the time of the visit or those who had fever in the last 7 days. Only children who were visiting the clinic for the first time for the current illness were included in the study. One child per woman was recruited into the study. Recruitment was done for ten consecutive days until the required sample size was obtained. Interviews were held with 439 mothers or other caregivers in the pediatric clinic before being examined by the clinicians. Body temperature, weight and height were measured and the children were examined by the clinicians, as is routinely done. The clinicians recorded on a special form the age, presenting problems of the child and their diagnosis following a physical examination and history taking. For each child, the clinicians were asked to record specifically whether fever, cough, diarrhoea, vomiting or convulsions were present and whether any of the following diagnoses applied: malaria, upper respiratory infection, bronchopneumonia, or gastroenteritis. Blood was drawn and tested for malaria parasites using Field’s stain A and B. Laboratory results were classified as positive for malaria if there was any evidence of parasites, acknowledging the limitation that laboratory slides are not a perfect gold standard for evaluating clinical diagnosis of malaria. In an endemic area, presence of parasites does not always lead to fever and a slide that is negative for the presence of parasites does not completely rule out the presence of parasites in the body (Smith et al., 1994, 1995).
56
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
Prior to initiating the cross-sectional clinic-based study, a qualitative study was carried out in a village adjacent to Mulago Hospital. It consisted of small group discussions with women who had or were caring for a child or children under 5 years. Ten focus groups of 4 – 6 women each were conducted. The purpose of this preliminary study was to clarify the concept of fever and to find out how mothers differentiated between fever due to malaria and fever due to other conditions; to identify the physical conditions and behavioral changes mothers associated with fever/malaria; to clarify local concepts like ‘eyabwe’ (febrile convulsions) and their relationships to malaria; to obtain information on drugs most used in the treatment of fevers at home; and to provide insights into what mothers do at home before they take their children to health units. Information obtained from the preliminary study was used to refine the data collection instruments for the clinic-based study.
2.2. Data analysis For the cross-sectional clinic-based study, mothers’ perceptions of the causes of fever and their behavior were analyzed for the study population as a whole and by demographic characteristics of the mothers and children. Both the mothers’ and the clinicians’ assessments of whether the child had malaria were compared to the laboratory findings. The sensitivity, specificity and positive and negative predictive value of the mothers’ and clinicians’ assessments were calculated.
3. Results The respondents were aged between 15 and 67 years. The mean age was 24 years and the median 22 years. Ninety-seven percent (425) of the respondents were mothers of the sick children; 2% (9) were grandmothers and 1% (5) aunts and sisters. Most of the respondents were of low socio-economic status, and most were not working outside their homes. Eighty-eight percent (387) had some formal education. Of these, 71% (274) had primary level education, 29% (112) had secondary level education and 0.3% (1) had university education. The age range of the children was 2 weeks to 60 months. The mean age of the children was 17 months and the median was 13. The mean duration of the children’s sickness was 3 days. Eleven percent of the children had been sick for less than 24 h; 64% for 1–3 days; and 26% had been sick for 4 days or more. All children in the study were recruited on the basis of a reported fever. However, at the clinic, only 60% of the children had a temperature greater than 37.5°C. Within this group, the temperature range was 37.6–41.0°C. The temperature of some of the children may have come down due to earlier treatment administered at home. Seventy-two percent of the children (310/439) had had fever attacks more than once in the 5 months preceding the study (range 1–25). The mean and modal number of attacks were two. Of all the children studied, only 23% had a history of fever with convulsions since birth. Of these, 57% had been attacked by febrile convulsions once, while the rest had more than one attack (mean= 1.7). Convul-
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
57
sions were highly feared by the mothers, who regarded them as fatal and debilitating if the child managed to survive. Mothers reported that they easily recognized fever in their children, and slightly over 60% said that they detected a coming fever attack a few days before the onset of the fever. During the interview, the mothers were first asked an open-ended question about how they recognized the fever. The major symptom used to recognize fever was elevated body temperature (‘the body felt hotter than usual’), mentioned by 98% of the mothers. The second symptom, reported by 70% of the women, was ‘okuyongobera’ or body weakness, likened by women to ‘a leaf that has been cut off from its stem.’ No other symptoms were volunteered by more than 50% of the mothers. Next, mothers were read a list of symptoms and asked whether they noticed or the child complained of any of these right before or after the child developed fever. The distribution of responses to these questions is shown in Table 1. The focus groups had indicated that poor appetite, disinterest in playing and disrupted sleep patterns were important factors that mothers associated with fever. In the clinic study, these behavioral signs were apparent, as well. Changes in appetite were most evident in children who were no longer exclusively breast feeding (ages 7 – 60 months). In this group, 71% ate less than usual and 19% stopped eating completely. Among breast feeding infants, 44% breast fed less than usual and 2% stopped breast feeding completely. Only 12% of children continued to play as usual; 52% played less; and 36% did not play at all. Fever changed the sleeping patterns and quality of sleep of 74% of the children; 59% of these slept less and 40% more. Forty-two percent were reported to sleep in ‘an unsteady manner and restlessly as if they had a nightmare’ (‘abaana bebaka bawawamuka’). Table 1 Symptoms experienced by children with fever Condition/Symptom
Frequency (total N =439, %)
Joint pain and body weakness Cough and common cold Poor appetite Weak and tired eyes Diarrhoea Vomiting Stomach pains Bad breathing Headache Excessive sweating Chills Dizziness Skin rash Convulsions Sore throat/nose Red eyes Others
375 301 273 268 219 214 201 179 151 134 115 89 82 78 46 45 29
(85%) (69%) (62%) (61%) (50%) (49%) (46%) (41%) (34%) (31%) (26%) (20%) (19%) (18%) (10%) (10%) (7%)
58
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
Table 2 Mothers’ perceived causes of fever in their children Perceived cause
Frequency
Percent
Malaria Cough and cold Convulsions Nutritional problems (bwosi) Worms Measles Don’t know Others
174 57 26 26 22 12 121 33
40 13 6 6 5 3 28 7
Some mothers gave more than one response. ‘Others’ in the table included: skin diseases, typhoid fever, diarrhoea and vomiting, abdominal problems, syphilis, AIDS, ‘got fever from the mother while breast-feeding’, sores on legs, anti-measles vaccine injection, paining ears and sore throat.
Generally, over 50% of the children were brought to the health centre between 1 and 3 days regardless of the combination of symptoms. In fact, a higher proportion (17%) of children with single symptoms were brought on day 1 than children with a combination of symptoms. The mothers’ perceived severity of illness was more highly associated with rushing to the hospital than the number of symptoms or the age of the child. Mothers were asked what they thought was the cause of the fever for which they brought the child to the health unit. Forty percent (174/439) of the mothers suspected malaria, while 13% (57/439) thought that the cause was cough and common cold. A total of 121 mothers (28%) did not know the cause of fever in their children. Other perceived causes were convulsions, nutritional problems, worms, and measles. Some mothers mentioned more than one cause. Details are shown in Table 2. Forty-eight percent of the women with secondary level of education or higher thought that the fever was malaria compared to 37% of the women with primary education and 35% without any education at all. Children of different age groups were equally suspected to have malaria. Thirty-four percent of children aged 0–3 months; 42% of those aged 4 – 6 months; 44% of those aged 7–12 months; 39% of those aged 13 – 24 months; and 41% of those aged 25 months and above were suspected to have malaria. The prevalence of malaria parasites, as determined by the laboratory tests (Table 3), was 64% (278/435), with the following prevalences by age: 51% of the children aged 0 – 3 months; 67% of those aged 4–6 months; and 60, 56 and 65% of those aged 7 – 12, 13 – 24, and 25 and more months, respectively. Plasmodium falciparum was present in 97.5% of the children with positive laboratory results. The distribution of parasitic density, based on the number of infected cells seen per high power microscopic field ( + = one infected cell), was 26% ( + ), 36% ( + + ), 23% ( + + + ), 4% (+ + + + ), 11% scanty.
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
59
Table 3 Mothers’ perceived cause of fever by whether parasites were present or not Laboratory resultsa
Perception of cause of fever
Mother thought fever was malaria Mother thought that fever was due to other condition Total a
Plasmodium present
Plasmodium absent
Total
104 174 278
66 91 157
170 265 435
Laboratory results for four children were missing.
Mothers’ perceptions of the cause of the children’s fever were compared with the malaria laboratory test results. As is shown in Table 3, mothers thought that 39% (170/435) of the children had malaria. Thus, estimates of the prevalence of malaria based on the mothers’ perceptions are too low compared to the laboratory findings. Mothers’ sensitivity was 37% (104/278), indicating that mothers are poor at recognizing malaria when it is, in fact, present. The specificity relative to laboratory tests was higher. Mothers correctly classified 58% (91/157) of the children whose slides were negative for parasites as free of malaria. They misclassified 66 of the 157 children as having malaria when the laboratory test was negative for parasites (false positive rate of 42%). They made the ‘grave mistake’ of misclassifying 174 cases as being ‘other conditions’; yet these children had malaria (false negative rate of 63% (174/278)). Their negative predictive value was 34% (91/265), which means that when a mother says that the child has no malaria this is confirmed by the laboratory only 34% of the time. Based on illness history and physical examinations, the doctors classified 92% (394/428) of the children as having malaria; thus, estimates of the prevalence of disease based on the clinical findings are too high. Among the children whom the doctors diagnosed with malaria, 51% were diagnosed with malaria alone, 26% also were diagnosed with upper respiratory tract infections, 6% also had gastroenteritis and 7% also had bronchopneumonia. The comparison of the doctors’ clinical diagnoses with the laboratory results is presented in Table 4. The sensitivity of Table 4 Comparison of doctors’ clinical diagnosis (based on illness history and physical examinations) and laboratory results Clinical diagnosisa
Malaria and other conditions Other conditions without malaria Total a
Laboratory resultsa Plasmodium present
Plasmodium absent
Total
265 6 271
129 28 157
394 34 428
Clinical diagnosis and/or laboratory values were missing for 11 children.
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
60
Table 5 Types of modern medications given, by perceived cause of fever Drugs given
Perceived cause of fever (%) Malaria (n= 194)
Antimalarial Antipyretic Antibiotic Other drugs Total
Worms (n= 22)
Cough (n = 54)
Convulsions (n =26)
Don’t know (n =121)
50 42 4 5
37 45 5 14
39 50 4 7
62 31 8
52 43 3 2
100
100
100
100
100
‘Other drugs’ in the table included mabendazole, valium, phenobarbitone, cough mixtures and vicks rubs. Some mothers gave more than one kind of medication.
clinical diagnosis relative to laboratory tests was 98% (265/271) but the specificity was only 18% (28/157). The doctors tended to classify most of the cases presenting with fever as having malaria, leading to a very high false positive rate of 82% (129/157) and a low false negative rate of 2% (6/271). Doctors could correctly predict the actual presence of malaria in the slides only 67% (265/394) of the time (positive predictive value). This means that when a doctor tells the mother that the child has malaria his diagnosis is confirmed by the laboratory tests about 67% of the time (in these environments of high prevalence of malaria). Their negative predictive value was 82% (28/34). In other words, when a doctor tells a mother that the child has no malaria his diagnosis is generally confirmed by the laboratory slide a very high percentage (82%) of the time compared to mothers. Although doctors’ diagnostic decisions were based on physical examinations and not on laboratory tests, they performed far better at finding the patients who truly had malaria, because in Uganda doctors are trained to first suspect malaria when a patient presents with fever. Fever as a measure of malaria is very sensitive but not specific. A total of 393 (90%) mothers gave some medicines at home before visiting the Old Mulago Hospital. Of these, 76% gave modern medicines only and 3% gave herbs only, while 21% administered a mixture of both modern and herbal medicines. Mothers were asked what ‘modern’ medicines they gave their children. Not all children who were perceived to have malaria were given antimalarials. Fifty percent of the modern medications given to children suspected of having malaria were antimalarials and 42% were antipyretics. The most commonly used antimalarial was chloroquine tablets, which were used 33% of the time, while the most commonly given antipyretics were aspirin and junior aspirin. Table 5 shows the type of medicines given by perceived cause of fever. Convulsions, which were severe manifestations of malarial fevers and which were perceived by the mothers to be life-threatening to the children, had higher chances of being treated with a modern antimalarial, a herb or both. Over 70% of the doses of the different drugs given to children at home were started on day 1 of the child’s fever.
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
61
Chloroquine use increased with increasing age. The proportion of children receiving chloroquine was 17% (babies 0–3 months), 27% (4–6 months), 30% (7–12 months), 35% (13 – 24 months) and 41% (25–60 months). Quinine mixture was given mostly to infants and decreased with increasing age. Junior aspirin was given to infants and decreased with increasing age. Babies 0–6 months were most likely to be given other drugs such as phenobarbitone. Antibiotics were given to all children indiscriminately. Some women reported having ‘great faith in septrin as a first line antimalarial drug’. Sources of the modern drugs used at home were analyzed for the first day of the child’s sickness and treatment. Of all the 320 drugs reported to have been used on day 1, 42% were obtained from the private clinics (without a health worker’s prescription); 28% from retail grocery shops; 16% had been at home; 10% from the pharmacies; and 4% from other sources such as the relatives and neighbours.
4. Discussion The presence of fever in children is a cause of concern for many mothers in Uganda and many do recognize it when it occurs. However, an association of fever with malaria is often not made. Only 40% of mothers suspected malaria in the presence of fever. This was so despite the fact that the survey was carried out in a malarious season and in an urban area where people are slightly more educated and are supposed to be more enlightened than those in the rural area. This finding was consistent with two other similar studies carried out in rural Tanzania and Uganda (Rooth and Bjorkman, 1992; Kengeya-Kayondo et al., 1994), indicating that this is a general problem across urban and rural areas of the East African region. In a malaria-endemic area, inability or delay in relating fever to malaria may have adverse effects on child survival, since first aid treatment for malaria may not be initiated at all or may be delayed. One possible explanation for the poor association of fever and malaria is that mothers, unlike the doctors, have never been sensitized to relate fever to malaria. Doctors in Uganda are trained to suspect malaria whenever a patient presents with fever. In this study doctors classified 92% of the children with fever as having malaria. Again, these findings were consistent with those of Rooth and Bjorkman (1992), where malaria was overdiagnosed by medical professionals. These investigators found that the medical doctors classified 85% of the febrile episodes in children as malaria and the sensitivity and specificity of the medical doctors’ diagnosis of malaria when compared to laboratory findings was 99 and 52%, respectively. Rural medical aids classified 94% of the children with fever as having malaria. Qualitative interviews provided a variety of insights into why some mothers thought their children did not have malaria. First was the pattern in which the conditions presented. When fever was preceded by another condition or disease, such as cough and common cold, measles, or wounds, the mothers associated the fever with the conditions manifested first. Second were the fears and fatality associated with certain diseases. Mothers associated fevers more with those diseases
62
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
like measles which they feared most and for which they had observable evidence that they killed children. Third was knowledge of the circumstances and behaviors of the child. Mothers whose children sucked their fingers and ate dirty things often associated fevers with worms and not malaria. Fourth were preventive actions taken by mothers to protect their children from mosquito bites. One mother, who always made her child sleep in a bed net and also sprayed her house every evening to kill the mosquitoes, relentlessly refused to associate fever with malaria even after her child was found with malaria parasites after several visits to the health unit. Fifth was limited knowledge about all the possible causes of fever in children. This could be due to the poor laboratory services in Uganda, where tests for other fever-causing diseases are rarely done. Mothers therefore never get to know the different diseases associated with fever. A sixth reason was that mothers are not as aware of the existence of malaria and its fatality as they are of other child-killer diseases such as measles, tuberculosis, pertussis, diphtheria, tetanus and poliomyelitis, which they continually hear about from the Expanded Programme for Immunization. Finally, unlike cough and common cold, malaria cannot be observed by the mother’s naked eye. The severity of the fever and its duration seemed not to influence the way mothers related fever to malaria. For example, some mothers with children who had febrile convulsions (popularly known as ‘eyabwe’) believed the convulsions were due to some supernatural condition or some obnoxious bird or insect inflicting the child with the disease. Mothers’ inability to distinguish fever caused by malaria from other causes is further compounded by the common usage of one local term ‘omusujja’ which means fever. This one term encompasses many types of fevers, which are sometimes regarded as symptoms of other diseases like coughs, respiratory conditions and wounds and at other times as diseases in themselves. One good example of the latter is fever believed to be caused by prolonged contact with rain. Again, this terminology problem has been consistently reported in the literature. Kengeya-Kayondo et al. (1994) found that mosquito bites were associated with malaria by a few key informants; in general, however, even though respondents reported knowing about different types of ‘omusujja’, causes were infrequently linked to these types. Although some people know that malaria is caused by mosquitoes (‘omusujja gwensiri’) people rarely call malaria by that name. KengeyaKayondo et al. (1994) actually go further to caution malaria control health education programmes to avoid translating malaria as ‘omusujja’, as the local term does not address malaria specifically. Mothers generally administer drugs of different forms promptly upon recognition of fever in their children. While most mothers did not link children’s fever with malaria, they were more likely to give antimalarials to severe forms of fevers presenting as febrile convulsions. Sixty-six percent of the drugs given to children who experienced convulsions were antimalarials. Among the drugs given to children suspected of having malaria, 50% were antimalarials. Mothers also indiscriminately gave antimalarials for all feverish conditions regardless of their perceived causes. It is unlikely that lack of availability or expense of antimalarial drugs are important factors since antimalarials are easily available in retail grocery and drug shops and
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
63
a child’s dose could be obtained cheaply at between 100–200/= Uganda shilling (which is equivalent to 10 – 20 US cents). The most probable explanation for not giving any antimalarial or an antipyretic to children with fevers and the indiscriminate giving of drugs for the mothers who do, is the inability of the mothers to suspect malaria whenever the child presents with fever and lack of knowledge of the right drugs to give. Illness misclassification and incorrect drug choice often lead to poor prognosis for the child and contribute to the transmission of malaria. Mothers have the potential to manage malarial fevers at home if they can be enabled to associate fevers in children with malaria and make use of available antimalarials. What is needed is a locally constructed fever algorithm sensitive to local knowledge and terminology and accessibility to health workers for consultation. For example, working in a malaria-endemic area of the Philippines, Gomes et al. (1994) found that the positive predictive value of three symptoms, fever, chills and sweating, either in combination or in sequence, was between 84 and 92% for children under the age of 10. This was a substantial improvement over the positive predictive value of 52% for fever alone. In that study, however, 51% of persons with reported fever also had chills and 57% experienced sweating. By contrast, Table 1 shows that in the present study, chills and sweating were reported at a much lower frequency by mothers of children with fever, making these symptoms unlikely to improve clinical diagnosis greatly. Perhaps mothers are not conditioned to specifically look for chills and sweating, or they occur with lower frequency. It is worth considering ways to improve the identification of these symptoms and to study their value for clinical diagnosis of malaria in Uganda. For the present, in endemic malaria zones, there is a need to educate mothers to suspect malaria first in every case of febrile illness, just as doctors do and offer presumptive treatment while monitoring a short set of symptoms indicative of pneumonia and ear infections. Mothers need to be taught how to manage fevers suspected to be malaria, particularly during the first 24 h and they should know when to seek further medical care. Mothers have been empowered to manage diarrhoea in children by using oral rehydration salts. The Malaria Control Programme needs to transfer lessons from the Control of Diarrhoeal Diseases Control programme to reduce child morbidity and mortality due to malaria.
Acknowledgements This study was funded by the Seed Grant Program in Pharmaco-epidemiology, and the Rockefeller Foundation through the International Clinical Epidemiology Network (INCLEN). The help of Dr Serunjogi, Medical Superintendent, Old Mulago Hospital and his entire staff in the data collection and implementation of the study is gratefully acknowledged. The team is also thankful to Professor Mark Nichter of the University of Arizona, Dr Richard Morrow of the Johns Hopkins University and Dr Dennis Willms of McMaster University for reviewing the first drafts of this paper and for their constructive suggestions. The authors are also grateful to Drs John Lule and David Okello of Makerere University Clinical Epidemiology Unit (CEU) for input.
64
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
References Adeniyi, J.D., Ramakrishna, I., 1985. Opinions, attitudes, and beliefs about self treatment practices in a Nigerian urban setting: implications for health education. International Quarterly of Community Health Education, vol. 5 (2). Beausoleil, E.G., 1984. A review of present antimalarial activities in Africa. Bull. World Health Org. 62 (Suppl.), 13–17. Bruce-Chwatt, L.J., 1987a. Combatting severe malaria in African children. Bull. World Health Org. 66, 611–620. Bruce-Chwatt, L.J., 1987b. Malaria and its control: present situation and future prospects. Ann. Rev. Public Health 8, 75–110. Deming, M.S., Gayibor, A., Murphy, K., Jones, T.S., Karsa, T., 1986. Home treatment of febrile children with antimalarial drugs in Togo. Bull. World Health Org. 67, 695 – 700. Ewbank, D.C., Gribble, J.N. (Eds.), 1993. Effects of Health Programs on Child Mortality in Sub-Saharan Africa. National Academic Press, Washington, D.C., pp. 90. Gomes, M., Espino, F.E., Abaquin, J., Realon, C., Salazar, N.P., 1994. Symptomatic identification of malaria in the home and in the primary health care clinic. Bull. World Health Org. 72, 383 – 390. Jeffery, G.M., 1984. The role of chemotherapy in malaria control through primary health care: constraints and future prospect. Bull. World Health Org. 62 (Suppl.), 49 – 53. Kengeya-Kayondo, J.F., Seeley, J.A., Kajura-Bajenja, E., Kabunga, E., Mubiru, E., Sembajja, F., Mulder, D., 1994. Recognition, treatment seeking behaviour and perception of cause of malaria among rural women in Uganda. Acta Trop. 58, 267 – 273. Rooth, I., Bjorkman, A., 1992. Fever episodes in a holoendemic malaria area of Tanzania: parasitological and clinical findings and diagnostic aspects related to malaria. Trans. R. Soc. Trop. Med. Hyg. 86, 479–482. Smith, T., Schellenberg, J.A., Hayes, R., 1994. Attributable fraction estimates and case definitions for malaria in endemic areas. Stat. Med. 13, 2345 – 2358. Smith, T., Hurt, N., Teuscher, T., Tanner, M., 1995. Is fever a good sign for clinical malaria in surveys of endemic communities?. Am. J. Trop. Med. Hyg. 52, 306 – 310. Uganda Ministry of Health, Health Planning Unit, 1992. Health Information Quarterly 9, 15 – 18. Uganda Ministry of Health, 1994. National Control of Malaria, 1995 – 2000. UNICEF, 1989. Children and Women in Uganda: A Situation Analysis, pp. 35. WHO. 1984. Malaria Control as part of primary health care. Technical Report Series 712, Geneva. WHO. 1986. WHO Expert Committee on Malaria. Eighteenth report Technical Series 735.
.
Maternal diagnosis and treatment of children’s fever in an endemic malaria zone of Uganda: implications for the malaria control programme Rosalind G.N. Lubanga a,*, Sandra Norman b, Douglas Ewbank c, Charles Karamagi d a Clinical Epidemiology Unit, Makerere Uni6ersity, P.O. Box 7423, Kampala, Uganda Center for Clinical Epidemiology and Biostatistics, Uni6ersity of Pennsyl6ania, Philadelphia, PA 19 104, USA c Population Studies Center, Uni6ersity of Pennsyl6ania, Philadelphia, PA 19 104, USA d Department of Pediatrics/Clinical Epidemiology Unit, Makerere Uni6ersity Medical School, Kampala, Uganda
b
Received 27 March 1996; received in revised form 18 September 1996; accepted 21 January 1997
Abstract A mother’s ability to suspect malaria in the presence of fever has important consequences for child survival in malaria-endemic areas. This paper presents results of a clinic-based study of mothers’ abilities to suspect malaria in the event of recognizing fever and other physiological and behavioral changes associated with the disease. The study population consisted of all (439) women or mothers who had accompanied children 5 years and below to the Old Mulago Hospital, Kampala, Uganda over a 10 day period during the malaria season of 1992. The children were those who had fever as a major complaint at the time of the visit or those who had fever in the last 7 days and were visiting the clinic for the first time for the current illness. The children were physically examined and their blood tested for malaria parasites. Mothers’ diagnosis was compared with clinical and laboratory diagnosis of malaria. Mothers associated the presence of fever with several types of illness and malaria was often not suspected. Only 40% of the mothers suspected malaria in their children. The mothers were poor at recognizing malaria when, in fact, it was present. The sensitivity of the mothers’ diagnosis of malaria was found to be 37%; 63% of malaria cases were misclassified * Corresponding author. Fax: + 256 41 530022. 0001-706X/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S 0 0 0 1 - 7 0 6 X ( 9 7 ) 0 0 0 7 1 - 5
54
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
as other conditions. The doctors classified most (92%) of the cases presenting with fever as having malaria, but laboratory tests indicated that only 64% of the children really had malaria. The sensitivity of clinical diagnosis was 98%, but the specificity was only 18%. Ninety percent of the mothers gave some medicines before visiting the health centre; and, of these, 76% gave modern drugs exclusively, including antimalarials, antipyretics, antibiotics and other drugs. Among the modern drugs given to children suspected of having malaria, 50% were antimalarials. The most commonly used antimalarial was chloroquine tablets. Mothers indiscriminately administered antimalarials to children irrespective of the perceived cause of the fever. There is need to educate mothers to suspect malaria first in every case of febrile illness, just like the doctors do, and about the first line drugs for the treatment of malaria. © 1997 Elsevier Science B.V. Keywords: Malaria; Sick child; Index of suspicion; Child survival; Presumptive treatment; Uganda
1. Introduction Malaria is among the top ten leading causes of morbidity and mortality in Uganda, with high fatality rates among children under 5 years (Uganda Ministry of Health, 1992). A May 1992 study in the Kampala District indicated mesoendemic levels in this area (Uganda Ministry of Health, 1994). In malaria-endemic regions, children under 5 years have low levels of resistance. The most common form of malaria parasite in Uganda is Plasmodium falciparum (UNICEF, 1989), a dangerous type, which may be fatal to children if not treated within a few days of onset. To reduce morbidity and mortality, the World Health Organization (WHO) has recommended that in malaria-endemic areas all cases of fever should be given prompt treatment with anti-malarials such as chloroquine or other 4-aminoquinolines (Jeffery, 1984). The effectiveness of presumptive treatment depends upon: (a) knowledge of the symptoms suggestive of malaria attack; (b) use of effective drugs; (c) ability of the mother to follow a precise dose regimen; and (d) giving prompt treatment (Jeffery, 1984; Deming et al., 1986). The family has a big role to play in the control of malaria. It needs to recognize symptoms commonly suggestive of malaria, particularly fever and to give appropriate first aid treatment and seek health care (Bruce-Chwatt, 1987a; WHO, 1984, 1986; Beausoleil, 1984). Surveys carried out in Africa reveal that between 80 and 90% of the people treat their children and other family members at home (Jeffery, 1984; Adeniyi and Ramakrishna, 1985). Conditions commonly treated at home include fevers, diarrhoea, coughs and respiratory complaints. Mothers always recognize fevers in their children and it is on this basis that they either give some treatment at home or seek health care. But, as is well known, fever is a symptom of a plethora of conditions and diseases. The critical question in the discussion of presumptive treatment of malaria at home is what proportion of mothers suspect a possible attack of malaria when they recognize fever in the children?
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
55
The ability of the mother to suspect malaria in the presence of fever has important implications for child survival in malaria-endemic areas. An easily suspecting mother is more likely to give an anti-malarial than one who decides that the condition is measles or cough and common cold and only gives drugs specific for those conditions. It has been shown in the malaria literature that the risks involved when a patient is presumptively treated with chloroquine when he is not truly suffering from malaria are minimal compared to leaving the child untreated (Ewbank and Gribble, 1993; Bruce-Chwatt, 1987b). This paper describes results of a clinic-based study carried out in the malarial season of June-July 1992 in the Old Mulago Hospital, Kampala, Uganda. Its objectives were to find out: (a) how mothers interpreted the presence of fever and other physiological and behavioral signs and symptoms which are clinically associated with malaria in their children under 5 years; (b) how mothers’ diagnosis corresponded to clinical and laboratory diagnosis of fever and malaria; and (c) the medicines given to children with fever before bringing the children to the health centre. The ultimate goal of this study was to investigate whether mothers could be used more effectively in the treatment of fevers at home and if so how.
2. Methods
2.1. Data collection The study design was a descriptive cross-sectional prevalence study. The study population consisted of all women or mothers who had accompanied children 5 years and below to the Old Mulago Hospital. The children were those who had fever as a major complaint at the time of the visit or those who had fever in the last 7 days. Only children who were visiting the clinic for the first time for the current illness were included in the study. One child per woman was recruited into the study. Recruitment was done for ten consecutive days until the required sample size was obtained. Interviews were held with 439 mothers or other caregivers in the pediatric clinic before being examined by the clinicians. Body temperature, weight and height were measured and the children were examined by the clinicians, as is routinely done. The clinicians recorded on a special form the age, presenting problems of the child and their diagnosis following a physical examination and history taking. For each child, the clinicians were asked to record specifically whether fever, cough, diarrhoea, vomiting or convulsions were present and whether any of the following diagnoses applied: malaria, upper respiratory infection, bronchopneumonia, or gastroenteritis. Blood was drawn and tested for malaria parasites using Field’s stain A and B. Laboratory results were classified as positive for malaria if there was any evidence of parasites, acknowledging the limitation that laboratory slides are not a perfect gold standard for evaluating clinical diagnosis of malaria. In an endemic area, presence of parasites does not always lead to fever and a slide that is negative for the presence of parasites does not completely rule out the presence of parasites in the body (Smith et al., 1994, 1995).
56
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
Prior to initiating the cross-sectional clinic-based study, a qualitative study was carried out in a village adjacent to Mulago Hospital. It consisted of small group discussions with women who had or were caring for a child or children under 5 years. Ten focus groups of 4 – 6 women each were conducted. The purpose of this preliminary study was to clarify the concept of fever and to find out how mothers differentiated between fever due to malaria and fever due to other conditions; to identify the physical conditions and behavioral changes mothers associated with fever/malaria; to clarify local concepts like ‘eyabwe’ (febrile convulsions) and their relationships to malaria; to obtain information on drugs most used in the treatment of fevers at home; and to provide insights into what mothers do at home before they take their children to health units. Information obtained from the preliminary study was used to refine the data collection instruments for the clinic-based study.
2.2. Data analysis For the cross-sectional clinic-based study, mothers’ perceptions of the causes of fever and their behavior were analyzed for the study population as a whole and by demographic characteristics of the mothers and children. Both the mothers’ and the clinicians’ assessments of whether the child had malaria were compared to the laboratory findings. The sensitivity, specificity and positive and negative predictive value of the mothers’ and clinicians’ assessments were calculated.
3. Results The respondents were aged between 15 and 67 years. The mean age was 24 years and the median 22 years. Ninety-seven percent (425) of the respondents were mothers of the sick children; 2% (9) were grandmothers and 1% (5) aunts and sisters. Most of the respondents were of low socio-economic status, and most were not working outside their homes. Eighty-eight percent (387) had some formal education. Of these, 71% (274) had primary level education, 29% (112) had secondary level education and 0.3% (1) had university education. The age range of the children was 2 weeks to 60 months. The mean age of the children was 17 months and the median was 13. The mean duration of the children’s sickness was 3 days. Eleven percent of the children had been sick for less than 24 h; 64% for 1–3 days; and 26% had been sick for 4 days or more. All children in the study were recruited on the basis of a reported fever. However, at the clinic, only 60% of the children had a temperature greater than 37.5°C. Within this group, the temperature range was 37.6–41.0°C. The temperature of some of the children may have come down due to earlier treatment administered at home. Seventy-two percent of the children (310/439) had had fever attacks more than once in the 5 months preceding the study (range 1–25). The mean and modal number of attacks were two. Of all the children studied, only 23% had a history of fever with convulsions since birth. Of these, 57% had been attacked by febrile convulsions once, while the rest had more than one attack (mean= 1.7). Convul-
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
57
sions were highly feared by the mothers, who regarded them as fatal and debilitating if the child managed to survive. Mothers reported that they easily recognized fever in their children, and slightly over 60% said that they detected a coming fever attack a few days before the onset of the fever. During the interview, the mothers were first asked an open-ended question about how they recognized the fever. The major symptom used to recognize fever was elevated body temperature (‘the body felt hotter than usual’), mentioned by 98% of the mothers. The second symptom, reported by 70% of the women, was ‘okuyongobera’ or body weakness, likened by women to ‘a leaf that has been cut off from its stem.’ No other symptoms were volunteered by more than 50% of the mothers. Next, mothers were read a list of symptoms and asked whether they noticed or the child complained of any of these right before or after the child developed fever. The distribution of responses to these questions is shown in Table 1. The focus groups had indicated that poor appetite, disinterest in playing and disrupted sleep patterns were important factors that mothers associated with fever. In the clinic study, these behavioral signs were apparent, as well. Changes in appetite were most evident in children who were no longer exclusively breast feeding (ages 7 – 60 months). In this group, 71% ate less than usual and 19% stopped eating completely. Among breast feeding infants, 44% breast fed less than usual and 2% stopped breast feeding completely. Only 12% of children continued to play as usual; 52% played less; and 36% did not play at all. Fever changed the sleeping patterns and quality of sleep of 74% of the children; 59% of these slept less and 40% more. Forty-two percent were reported to sleep in ‘an unsteady manner and restlessly as if they had a nightmare’ (‘abaana bebaka bawawamuka’). Table 1 Symptoms experienced by children with fever Condition/Symptom
Frequency (total N =439, %)
Joint pain and body weakness Cough and common cold Poor appetite Weak and tired eyes Diarrhoea Vomiting Stomach pains Bad breathing Headache Excessive sweating Chills Dizziness Skin rash Convulsions Sore throat/nose Red eyes Others
375 301 273 268 219 214 201 179 151 134 115 89 82 78 46 45 29
(85%) (69%) (62%) (61%) (50%) (49%) (46%) (41%) (34%) (31%) (26%) (20%) (19%) (18%) (10%) (10%) (7%)
58
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
Table 2 Mothers’ perceived causes of fever in their children Perceived cause
Frequency
Percent
Malaria Cough and cold Convulsions Nutritional problems (bwosi) Worms Measles Don’t know Others
174 57 26 26 22 12 121 33
40 13 6 6 5 3 28 7
Some mothers gave more than one response. ‘Others’ in the table included: skin diseases, typhoid fever, diarrhoea and vomiting, abdominal problems, syphilis, AIDS, ‘got fever from the mother while breast-feeding’, sores on legs, anti-measles vaccine injection, paining ears and sore throat.
Generally, over 50% of the children were brought to the health centre between 1 and 3 days regardless of the combination of symptoms. In fact, a higher proportion (17%) of children with single symptoms were brought on day 1 than children with a combination of symptoms. The mothers’ perceived severity of illness was more highly associated with rushing to the hospital than the number of symptoms or the age of the child. Mothers were asked what they thought was the cause of the fever for which they brought the child to the health unit. Forty percent (174/439) of the mothers suspected malaria, while 13% (57/439) thought that the cause was cough and common cold. A total of 121 mothers (28%) did not know the cause of fever in their children. Other perceived causes were convulsions, nutritional problems, worms, and measles. Some mothers mentioned more than one cause. Details are shown in Table 2. Forty-eight percent of the women with secondary level of education or higher thought that the fever was malaria compared to 37% of the women with primary education and 35% without any education at all. Children of different age groups were equally suspected to have malaria. Thirty-four percent of children aged 0–3 months; 42% of those aged 4 – 6 months; 44% of those aged 7–12 months; 39% of those aged 13 – 24 months; and 41% of those aged 25 months and above were suspected to have malaria. The prevalence of malaria parasites, as determined by the laboratory tests (Table 3), was 64% (278/435), with the following prevalences by age: 51% of the children aged 0 – 3 months; 67% of those aged 4–6 months; and 60, 56 and 65% of those aged 7 – 12, 13 – 24, and 25 and more months, respectively. Plasmodium falciparum was present in 97.5% of the children with positive laboratory results. The distribution of parasitic density, based on the number of infected cells seen per high power microscopic field ( + = one infected cell), was 26% ( + ), 36% ( + + ), 23% ( + + + ), 4% (+ + + + ), 11% scanty.
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
59
Table 3 Mothers’ perceived cause of fever by whether parasites were present or not Laboratory resultsa
Perception of cause of fever
Mother thought fever was malaria Mother thought that fever was due to other condition Total a
Plasmodium present
Plasmodium absent
Total
104 174 278
66 91 157
170 265 435
Laboratory results for four children were missing.
Mothers’ perceptions of the cause of the children’s fever were compared with the malaria laboratory test results. As is shown in Table 3, mothers thought that 39% (170/435) of the children had malaria. Thus, estimates of the prevalence of malaria based on the mothers’ perceptions are too low compared to the laboratory findings. Mothers’ sensitivity was 37% (104/278), indicating that mothers are poor at recognizing malaria when it is, in fact, present. The specificity relative to laboratory tests was higher. Mothers correctly classified 58% (91/157) of the children whose slides were negative for parasites as free of malaria. They misclassified 66 of the 157 children as having malaria when the laboratory test was negative for parasites (false positive rate of 42%). They made the ‘grave mistake’ of misclassifying 174 cases as being ‘other conditions’; yet these children had malaria (false negative rate of 63% (174/278)). Their negative predictive value was 34% (91/265), which means that when a mother says that the child has no malaria this is confirmed by the laboratory only 34% of the time. Based on illness history and physical examinations, the doctors classified 92% (394/428) of the children as having malaria; thus, estimates of the prevalence of disease based on the clinical findings are too high. Among the children whom the doctors diagnosed with malaria, 51% were diagnosed with malaria alone, 26% also were diagnosed with upper respiratory tract infections, 6% also had gastroenteritis and 7% also had bronchopneumonia. The comparison of the doctors’ clinical diagnoses with the laboratory results is presented in Table 4. The sensitivity of Table 4 Comparison of doctors’ clinical diagnosis (based on illness history and physical examinations) and laboratory results Clinical diagnosisa
Malaria and other conditions Other conditions without malaria Total a
Laboratory resultsa Plasmodium present
Plasmodium absent
Total
265 6 271
129 28 157
394 34 428
Clinical diagnosis and/or laboratory values were missing for 11 children.
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
60
Table 5 Types of modern medications given, by perceived cause of fever Drugs given
Perceived cause of fever (%) Malaria (n= 194)
Antimalarial Antipyretic Antibiotic Other drugs Total
Worms (n= 22)
Cough (n = 54)
Convulsions (n =26)
Don’t know (n =121)
50 42 4 5
37 45 5 14
39 50 4 7
62 31 8
52 43 3 2
100
100
100
100
100
‘Other drugs’ in the table included mabendazole, valium, phenobarbitone, cough mixtures and vicks rubs. Some mothers gave more than one kind of medication.
clinical diagnosis relative to laboratory tests was 98% (265/271) but the specificity was only 18% (28/157). The doctors tended to classify most of the cases presenting with fever as having malaria, leading to a very high false positive rate of 82% (129/157) and a low false negative rate of 2% (6/271). Doctors could correctly predict the actual presence of malaria in the slides only 67% (265/394) of the time (positive predictive value). This means that when a doctor tells the mother that the child has malaria his diagnosis is confirmed by the laboratory tests about 67% of the time (in these environments of high prevalence of malaria). Their negative predictive value was 82% (28/34). In other words, when a doctor tells a mother that the child has no malaria his diagnosis is generally confirmed by the laboratory slide a very high percentage (82%) of the time compared to mothers. Although doctors’ diagnostic decisions were based on physical examinations and not on laboratory tests, they performed far better at finding the patients who truly had malaria, because in Uganda doctors are trained to first suspect malaria when a patient presents with fever. Fever as a measure of malaria is very sensitive but not specific. A total of 393 (90%) mothers gave some medicines at home before visiting the Old Mulago Hospital. Of these, 76% gave modern medicines only and 3% gave herbs only, while 21% administered a mixture of both modern and herbal medicines. Mothers were asked what ‘modern’ medicines they gave their children. Not all children who were perceived to have malaria were given antimalarials. Fifty percent of the modern medications given to children suspected of having malaria were antimalarials and 42% were antipyretics. The most commonly used antimalarial was chloroquine tablets, which were used 33% of the time, while the most commonly given antipyretics were aspirin and junior aspirin. Table 5 shows the type of medicines given by perceived cause of fever. Convulsions, which were severe manifestations of malarial fevers and which were perceived by the mothers to be life-threatening to the children, had higher chances of being treated with a modern antimalarial, a herb or both. Over 70% of the doses of the different drugs given to children at home were started on day 1 of the child’s fever.
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
61
Chloroquine use increased with increasing age. The proportion of children receiving chloroquine was 17% (babies 0–3 months), 27% (4–6 months), 30% (7–12 months), 35% (13 – 24 months) and 41% (25–60 months). Quinine mixture was given mostly to infants and decreased with increasing age. Junior aspirin was given to infants and decreased with increasing age. Babies 0–6 months were most likely to be given other drugs such as phenobarbitone. Antibiotics were given to all children indiscriminately. Some women reported having ‘great faith in septrin as a first line antimalarial drug’. Sources of the modern drugs used at home were analyzed for the first day of the child’s sickness and treatment. Of all the 320 drugs reported to have been used on day 1, 42% were obtained from the private clinics (without a health worker’s prescription); 28% from retail grocery shops; 16% had been at home; 10% from the pharmacies; and 4% from other sources such as the relatives and neighbours.
4. Discussion The presence of fever in children is a cause of concern for many mothers in Uganda and many do recognize it when it occurs. However, an association of fever with malaria is often not made. Only 40% of mothers suspected malaria in the presence of fever. This was so despite the fact that the survey was carried out in a malarious season and in an urban area where people are slightly more educated and are supposed to be more enlightened than those in the rural area. This finding was consistent with two other similar studies carried out in rural Tanzania and Uganda (Rooth and Bjorkman, 1992; Kengeya-Kayondo et al., 1994), indicating that this is a general problem across urban and rural areas of the East African region. In a malaria-endemic area, inability or delay in relating fever to malaria may have adverse effects on child survival, since first aid treatment for malaria may not be initiated at all or may be delayed. One possible explanation for the poor association of fever and malaria is that mothers, unlike the doctors, have never been sensitized to relate fever to malaria. Doctors in Uganda are trained to suspect malaria whenever a patient presents with fever. In this study doctors classified 92% of the children with fever as having malaria. Again, these findings were consistent with those of Rooth and Bjorkman (1992), where malaria was overdiagnosed by medical professionals. These investigators found that the medical doctors classified 85% of the febrile episodes in children as malaria and the sensitivity and specificity of the medical doctors’ diagnosis of malaria when compared to laboratory findings was 99 and 52%, respectively. Rural medical aids classified 94% of the children with fever as having malaria. Qualitative interviews provided a variety of insights into why some mothers thought their children did not have malaria. First was the pattern in which the conditions presented. When fever was preceded by another condition or disease, such as cough and common cold, measles, or wounds, the mothers associated the fever with the conditions manifested first. Second were the fears and fatality associated with certain diseases. Mothers associated fevers more with those diseases
62
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
like measles which they feared most and for which they had observable evidence that they killed children. Third was knowledge of the circumstances and behaviors of the child. Mothers whose children sucked their fingers and ate dirty things often associated fevers with worms and not malaria. Fourth were preventive actions taken by mothers to protect their children from mosquito bites. One mother, who always made her child sleep in a bed net and also sprayed her house every evening to kill the mosquitoes, relentlessly refused to associate fever with malaria even after her child was found with malaria parasites after several visits to the health unit. Fifth was limited knowledge about all the possible causes of fever in children. This could be due to the poor laboratory services in Uganda, where tests for other fever-causing diseases are rarely done. Mothers therefore never get to know the different diseases associated with fever. A sixth reason was that mothers are not as aware of the existence of malaria and its fatality as they are of other child-killer diseases such as measles, tuberculosis, pertussis, diphtheria, tetanus and poliomyelitis, which they continually hear about from the Expanded Programme for Immunization. Finally, unlike cough and common cold, malaria cannot be observed by the mother’s naked eye. The severity of the fever and its duration seemed not to influence the way mothers related fever to malaria. For example, some mothers with children who had febrile convulsions (popularly known as ‘eyabwe’) believed the convulsions were due to some supernatural condition or some obnoxious bird or insect inflicting the child with the disease. Mothers’ inability to distinguish fever caused by malaria from other causes is further compounded by the common usage of one local term ‘omusujja’ which means fever. This one term encompasses many types of fevers, which are sometimes regarded as symptoms of other diseases like coughs, respiratory conditions and wounds and at other times as diseases in themselves. One good example of the latter is fever believed to be caused by prolonged contact with rain. Again, this terminology problem has been consistently reported in the literature. Kengeya-Kayondo et al. (1994) found that mosquito bites were associated with malaria by a few key informants; in general, however, even though respondents reported knowing about different types of ‘omusujja’, causes were infrequently linked to these types. Although some people know that malaria is caused by mosquitoes (‘omusujja gwensiri’) people rarely call malaria by that name. KengeyaKayondo et al. (1994) actually go further to caution malaria control health education programmes to avoid translating malaria as ‘omusujja’, as the local term does not address malaria specifically. Mothers generally administer drugs of different forms promptly upon recognition of fever in their children. While most mothers did not link children’s fever with malaria, they were more likely to give antimalarials to severe forms of fevers presenting as febrile convulsions. Sixty-six percent of the drugs given to children who experienced convulsions were antimalarials. Among the drugs given to children suspected of having malaria, 50% were antimalarials. Mothers also indiscriminately gave antimalarials for all feverish conditions regardless of their perceived causes. It is unlikely that lack of availability or expense of antimalarial drugs are important factors since antimalarials are easily available in retail grocery and drug shops and
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
63
a child’s dose could be obtained cheaply at between 100–200/= Uganda shilling (which is equivalent to 10 – 20 US cents). The most probable explanation for not giving any antimalarial or an antipyretic to children with fevers and the indiscriminate giving of drugs for the mothers who do, is the inability of the mothers to suspect malaria whenever the child presents with fever and lack of knowledge of the right drugs to give. Illness misclassification and incorrect drug choice often lead to poor prognosis for the child and contribute to the transmission of malaria. Mothers have the potential to manage malarial fevers at home if they can be enabled to associate fevers in children with malaria and make use of available antimalarials. What is needed is a locally constructed fever algorithm sensitive to local knowledge and terminology and accessibility to health workers for consultation. For example, working in a malaria-endemic area of the Philippines, Gomes et al. (1994) found that the positive predictive value of three symptoms, fever, chills and sweating, either in combination or in sequence, was between 84 and 92% for children under the age of 10. This was a substantial improvement over the positive predictive value of 52% for fever alone. In that study, however, 51% of persons with reported fever also had chills and 57% experienced sweating. By contrast, Table 1 shows that in the present study, chills and sweating were reported at a much lower frequency by mothers of children with fever, making these symptoms unlikely to improve clinical diagnosis greatly. Perhaps mothers are not conditioned to specifically look for chills and sweating, or they occur with lower frequency. It is worth considering ways to improve the identification of these symptoms and to study their value for clinical diagnosis of malaria in Uganda. For the present, in endemic malaria zones, there is a need to educate mothers to suspect malaria first in every case of febrile illness, just as doctors do and offer presumptive treatment while monitoring a short set of symptoms indicative of pneumonia and ear infections. Mothers need to be taught how to manage fevers suspected to be malaria, particularly during the first 24 h and they should know when to seek further medical care. Mothers have been empowered to manage diarrhoea in children by using oral rehydration salts. The Malaria Control Programme needs to transfer lessons from the Control of Diarrhoeal Diseases Control programme to reduce child morbidity and mortality due to malaria.
Acknowledgements This study was funded by the Seed Grant Program in Pharmaco-epidemiology, and the Rockefeller Foundation through the International Clinical Epidemiology Network (INCLEN). The help of Dr Serunjogi, Medical Superintendent, Old Mulago Hospital and his entire staff in the data collection and implementation of the study is gratefully acknowledged. The team is also thankful to Professor Mark Nichter of the University of Arizona, Dr Richard Morrow of the Johns Hopkins University and Dr Dennis Willms of McMaster University for reviewing the first drafts of this paper and for their constructive suggestions. The authors are also grateful to Drs John Lule and David Okello of Makerere University Clinical Epidemiology Unit (CEU) for input.
64
R.G.N. Lubanga et al. / Acta Tropica 68 (1997) 53–64
References Adeniyi, J.D., Ramakrishna, I., 1985. Opinions, attitudes, and beliefs about self treatment practices in a Nigerian urban setting: implications for health education. International Quarterly of Community Health Education, vol. 5 (2). Beausoleil, E.G., 1984. A review of present antimalarial activities in Africa. Bull. World Health Org. 62 (Suppl.), 13–17. Bruce-Chwatt, L.J., 1987a. Combatting severe malaria in African children. Bull. World Health Org. 66, 611–620. Bruce-Chwatt, L.J., 1987b. Malaria and its control: present situation and future prospects. Ann. Rev. Public Health 8, 75–110. Deming, M.S., Gayibor, A., Murphy, K., Jones, T.S., Karsa, T., 1986. Home treatment of febrile children with antimalarial drugs in Togo. Bull. World Health Org. 67, 695 – 700. Ewbank, D.C., Gribble, J.N. (Eds.), 1993. Effects of Health Programs on Child Mortality in Sub-Saharan Africa. National Academic Press, Washington, D.C., pp. 90. Gomes, M., Espino, F.E., Abaquin, J., Realon, C., Salazar, N.P., 1994. Symptomatic identification of malaria in the home and in the primary health care clinic. Bull. World Health Org. 72, 383 – 390. Jeffery, G.M., 1984. The role of chemotherapy in malaria control through primary health care: constraints and future prospect. Bull. World Health Org. 62 (Suppl.), 49 – 53. Kengeya-Kayondo, J.F., Seeley, J.A., Kajura-Bajenja, E., Kabunga, E., Mubiru, E., Sembajja, F., Mulder, D., 1994. Recognition, treatment seeking behaviour and perception of cause of malaria among rural women in Uganda. Acta Trop. 58, 267 – 273. Rooth, I., Bjorkman, A., 1992. Fever episodes in a holoendemic malaria area of Tanzania: parasitological and clinical findings and diagnostic aspects related to malaria. Trans. R. Soc. Trop. Med. Hyg. 86, 479–482. Smith, T., Schellenberg, J.A., Hayes, R., 1994. Attributable fraction estimates and case definitions for malaria in endemic areas. Stat. Med. 13, 2345 – 2358. Smith, T., Hurt, N., Teuscher, T., Tanner, M., 1995. Is fever a good sign for clinical malaria in surveys of endemic communities?. Am. J. Trop. Med. Hyg. 52, 306 – 310. Uganda Ministry of Health, Health Planning Unit, 1992. Health Information Quarterly 9, 15 – 18. Uganda Ministry of Health, 1994. National Control of Malaria, 1995 – 2000. UNICEF, 1989. Children and Women in Uganda: A Situation Analysis, pp. 35. WHO. 1984. Malaria Control as part of primary health care. Technical Report Series 712, Geneva. WHO. 1986. WHO Expert Committee on Malaria. Eighteenth report Technical Series 735.
.