Packaged treatment for first-line care in cerebral malaria and meningitis.

Research/Recherche Packaged treatment for first-line care in cerebral malaria and meningitis* T.R. Cullinan1 & C. Pieterick2 Described are the results of a trial carried out from January to June 1996 in southern Malawi to determine the effectiveness of a treatment pack for infants and children under the age of 6 years, who presented as emergencies to rural health centres with presumptive diagnoses of severe/cerebral malaria or meningitis. Each complete treatment pack (approximate cost, US$ 6) contained, inter alia, intramuscular quinine, intramuscular chloramphenicol, dextrose, paraldehyde, a nasogastric tube, prepacked syringes, and sterile water. A modified coma score and drug dosage nomogram were also included in the package. Despite a considerable drop in overall mortality, problems arose with regard to the incomplete treatment of possible meningitis and in the development of a rational referral policy. Introduction Over the last 10 years, the diagnostic and manage- ment needs of infants and young children admitted to the specialist malaria referral unit at Queen Eliza- beth Hospital, Blantyre, Malawi, have been system- atically refined and codified. This has led to the development of effective and efficient protocols for use in peripheral hospitals with minimal laboratory facilities (1). These have been validated in district hospitals in Malawi and are in general use, with mor- tality outcomes similar to those achieved in the spe- cialist malaria unit (2). However, the majority of infants and young children in Malawi and similar countries are not within reach of a hospital when they become se- verely ill (3), and 65% rely for all their health care on rural health centres that are minimally equipped for emergency situations and have no laboratory facili- ties. In Malawi, each rural health centre caters for 20000-30000 people, at least 20% of whom live more than 8km from a given centre. Each health centre is run by a medical assistant with 3 years' post- primary school training, who supervises the various cadres of more junior staff. Few centres have regular * From: Department of Community Health, College of Medicine, University of Malawi, Mangochi, Malawi. I Professor and Head. Requests for reprints should be sent to Prof. Cullinan, 53 Mapledene Road, London E8 3JW, England. 2 Honorary Lecturer. Reprint No. 5857 access to motor transport and fewer again have tele- phones and electricity; furthermore, it is not unusual for many of these services to be cut off for days and weeks during the rainy season (December-April). Preliminary estimates, together with an infor- mal review of local hospital and health centre records, suggest that case fatality from severe/cer- ebral malaria can be reduced to 18-20% in a special unit (1) and to 22% if protocols are followed for children presenting at a district general hospital (2). However, at the health centre level, estimates from the preliminary phase of the present study show that the fatality rate is nearer 50%. The reasons for this are as follows: - frequent critical lack of drugs, especially quinine; - inability to correct severe hypoglycaemia, which is of high prognostic value to the outcome of cerebral malaria (1); - frequent impossibility of timely referral to hospi- tal; and - inability of the medical assistant at the health centre level to distinguish between severe ma- laria and meningitis. The present study was designed to overcome these difficulties, without major extra resources, and to test the hypothesis that a significant lowering of mortality could be achieved by providing emergency "packaged" treatment at the health centre level. The study was carried out in the Mangochi Dis- trict in southern Malawi, an area of high endemicity Bulletin of the World Health Organization, 1998, 76 (3): 257-264 X World Health Organization 1998 257 T.R. Cullinan & C. Pieterick for Plasmodium falciparum malaria. In Malawi, as many as 75% of infants aged 6-12 months may be parasitaemic at any one time during the rainy season (4), from early December until late April. There is no more than an occasional brief shower at other times. While symptomatic infections increase consid- erably during the rainy season, previous studies have not shown that the case fatality rate is higher than during the dry season. Methods The study design compared case fatality during a 6- month pre-trial period of normal practice care with that during a 6-month trial period of "packaged care". Ideally, a randomized control trial of different health centres would have been carried out, but this could not be conducted blind and there were too many variables, including opportunities for hospital referral, to make matching possible. During the pre-trial period (July to December 1995) medical assistants in 10 health centres were asked to record all cases with a presumptive diagno- sis of severe/cerebral malaria or meningitis, follow normal protocols of management, and record out- comes or referrals to hospital according to Malawi Standard treatment guidelines (5).a The treatment recommended at health centre level for a child unable to swallow is administration of intramuscular quinine (10mg/kg body weight), repeated after 4 hours if referral has not been achieved, and of oral sulfadoxine (500mg) + pyrimethamine (25mg) if the child can take oral therapy. For presumptive bacterial meningitis, an intramuscular injection of benzylpenicillin (50000 units/kg) followed by imme- diate transfer is recommended. However, during the trial period in the 10 sentinel health centres, quinine was sometimes unavailable and benzylpenicillin almost invariably unavailable. The trial period encompassed the malaria season from January to June 1996. Medical assistants in all 24 health centres in Mangochi District were invited to take part and after 2 days of training and discussion, 22 agreed to do so (Fig. 1). Training in- volved revision of and agreement on the diagnostic criteria for a presumptive diagnosis of severe/cer- ebral malaria or meningitis, introduction of and familiarization with the new treatment packs, protocols and recording forms, and familiarization Fig. 1. Mangochi District, Malawi, and the location of the health centres included in the survey. (See foot- notes a and b, Table 2). with a modified coma score (Table 1), a nomogram for drug dosage, how to introduce nasogastric tubes, and the further management of the recovering child. There were also discussions on cultural beliefs sur- rounding sick children, on health-seeking behaviour, and on the acceptability and feasibility of referral to hospital. Each health centre was then visited and sup- plied with 3-5 treatment packs (each of which costs US$ 6) containing the following: - sealed plastic syringes (1 x 10ml, 1 x 2ml) with needles; - 1 g chloramphenicol powder; - 20ml water for injection; - 600mg quinine hydrochloride in 2ml water; - ml paraldehyde; - 20ml 50% (w/v) dextrose solution; Table 1: Abbreviated Blantyre coma score used in the study Response to standard painful stimulus Score No movement in response to firm pressure on sternum with examiner's knuckle Withdraws to firm pressure on finger nailbed with side of pencil but unable to locate firm pressure on sternum with examiner's knuckle Able to locate stimulus of firm pressure on sternum with examiner's knuckle 0 2 WHO Bulletin OMS. Vol 76 1998 a These guidelines stipulate a presumed diagnosis of severe/cer- ebral malaria in children, usually under 5 years of age, showing altered consciousness, repeated convulsions, the inability to take oral fluids, external pallor, and usually a history of fever. 258 Packaged treatment for cerebral malaria and meningitis - alcohol swabs; - a nasogastric tube; and - a management protocol, coma score, nomogram, and duplicate (carbon paper) recording forms. Records for each case included: - identifiers; - date and time of presentation and referral (if any); - clinical state including fits, fever, coma score, pallor, rapid breathing, hydration status, etc.; - treatment given; and - progress to outcome. Particular emphasis was put on the condition of the infants and children as they left the health centre. The safety of intramuscular quinine, even when given to an unconscious child, has been demon- strated in several studies (6). For the present study, 1 ml from an ampoule containing 600mg of quinine hydrochloride in 2ml solution was diluted in 5 ml sterile water for injection (equivalent to 0.2mtkg or 10mg/kg body weight). The inclusion of chloram- phenicol as first-line treatment for meningitis is in accordance with current practice in the paediatric department at Queen Elizabeth Hospital, Blantyre; dextrose was used as a substitute for powdered glu- cose, which was unobtainable in Malawi at the time. Hypoglycaemia is a strong predictor of mortality in cerebral malaria (7) and the assumption was made that it could be corrected without measuring blood glucose levels. The dosage schedules stipulated in the protocol and included in the nomogram were as follows: - rehydration using oral rehydration salt (ORS) solution, if necessary by nasogastric tube; - intramuscular quinine (0.2 ml/kg weight) re- peated every 4 hours as necessary; - chloramphenicol (25 mg/kg weight stat.); - paraldehyde (0.2 ml/kg weight) (deep intramus- cular), repeated if necessary; and - dextrose (10ml drawn from 20-ml vial) mixed with 10ml of clean water in a cup and spooned or injected down the nasogastric tube. Packs were only opened upon arrival of a defined case, and any unused materials were never re-used on a second case. Packs were supplied as needed during the trial. If the children were referred to hospital, copies of the records from the rural health centre accom- panied them and were retrieved from hospital records by the investigators. The single outcome measure in both phases of the study was infant survival. Survival was assumed if the infant or child was discharged from the health centre or hospital as "well" or "cured" or was lost to follow-up in a state recorded on the chart as "better", "im- proved", "cured", "well", etc. Similarly, death was assumed if the infant or child was lost to follow- up or was categorized as "not improved", "ill", "worse", etc. A simple access score, applicable to the period of the trial, was assessed to indicate the overall possibility of referral from each health centre in the event of serious illness (see Table 2). It had been planned to visit each health centre at least twice during the trial phase, but this proved impossible because of the extraordinarily heavy rains, impassable roads, and broken bridges. How- ever, each health centre was visited in May, at which time problems were discussed and the record forms collected. Results During the pre-trial phase (July-December 1995), records for 64 infants and children under the age of 6 years were retrieved from the 10 health centres used as sentinel sites for the study (Table 2). Since three of the 22 trial centres did not admit any cases of severe malaria during the trial phase, data on 96 infants and children under the age of 6 years in only 19 health centres were recorded. A total of 76 (80%) of the 96 infants and chil- dren had convulsions at or soon after presentation, and a further 11% were reported to have had so before arrival at the health centre. On admission 17 children (18%) were recorded as having "severe pallor" and eight (47%) of these children died. Only 59 (63%) of the 94 children who had complete drug dosage reports had received the full treatment schedule as indicated in the protocol. In 18% of cases intramuscular chloramphenicol was omitted and in 14%, dextrose. Omissions largely arose because one or two key medical assistants were posted elsewhere during the trial, leaving a more junior staff member unfamiliar with the protocol in charge. However, there was also a tendency to rely on instinct when diagnosing and giving treatment, and so save drugs. Another factor was the dislike of the mothers for nasogastric tubes, which they associated with death. In this way, several children did not receive dextrose dosages. None the less, there was no obvious association between any omission and mortality. WHO Bulletin OMS. Vol 76 1998 259 T.R. Cullinan & C. Pieterick Table 2: Health centres included in the studya Approximate distance Health centreb No. of children in pre-trial No. of children in trial to nearest hospital (km) Access scorec Chilipa (1) 5 2 60 1 Chikole (2) - 3 20 1 Katuli (3) - 1 55 0 Kukalanga (4) - 4 22 1 Lugola (5) - 1 72 0 Lungwena (6) - 2 23 2 Makanjira (7) - 10 110 0 Maldeco (8) (private) - 2 24 3 Mase (9) - 2 25 0 Mkumba (10) 6 6 90 1 Monkey Bay (11) 16 8 73 2 Mutsalepo (12) - 3 15 2 Namalaka (13) - 3 35 0 Namwera (14) 22 25 40 3 Nancholi (15) 4 1 70 0 Nankumba (16) - 16 85 1 Nkope (17) 5 1 50 1 Phirilongwe (18) 6 5 60 1 Ngapani (19) - 1 80 2 a Three of the 22 centres taking part did not admit any cases of severe malaria and have therefore been omitted here and from Fig. 1. b Figures in parentheses indicate the location of the centre (see Fig. 1). c Scored as follows: adequate all-weather road, Jan.-June = 1; telephone to district general hospital = 1; transport available on site = 1. In the pre-trial phase, 19 (30%) of the 64 infants and children seen were lost to follow-up (Table 3), signifying that referral to hospital was recorded but no trace of admission could be found in local hospital records. It is probable that many of these infants either died en route, or were taken to a local healer; however, calculations of case fatality were based only on the 45 for whom outcome data were avail- able could be strongly presumed. During the trial phase, more successful methods were used to trace referred children, and only 3 (3%) of the 96 were lost to follow-up. Therefore, 93 was used as a basis for the trial phase, and the overall case fatality conse- quently fell from 51% in the pre-trial phase to 23% in the trial phase. The greatest reduction in the number of case fatalities - from 17 of 33 children in the pre-trial phase to 1 out of 30 children (3%) in the trial phase - occurred among those for whom no referral was either planned or achieved (see Table 3). Many fac- tors influenced a decision to refer, but one of the strongest, which separated the trial from the pre-trial phase, was the medical assistants' confidence in being able to achieve results locally if immediate referral was not possible. Until the time of the trial phase, transfer from the health centre was the ac- cepted goal of successful management, whatever the condition of the child or treatment achieved. The medical assistant seldom learned the outcome fol- lowing referral or even whether the child had arrived in the hospital. The apparent increase in fatality among those for whom referral was arranged but not achieved (Table 3) may be because of much better case detection during the trial phase. Of the 32 children who did arrive in hospital (Table 3), 2 died and 30 recovered. In 17 cases (53%) the child was diagnosed and treated for severe or cerebral malaria, in 4 cases (13%) for meningitis, and in 3 cases (10%) for malaria and meningitis; in only one case of meningitis was a lumbar puncture per- formed. Two more children were treated for malaria together with severe anaemia, two for malaria and pneumonia, and in one case the diagnosis remained uncertain. These diagnoses, leading in all but two cases to complete recovery, were undoubtedly influ- enced by the primary diagnosis and management at the health centres. During both the pre-trial and trial phases, deaths occurred throughout the age range 0-6 years. Most children died while awaiting transport to hospi- tal, although in the pre-trial phase, mortality was also high among the nine children who arrived at a hospi- tal (see Table 3). During the trial phase, 29% of children who died or were presumed to have died, did so within an hour of presenting to the health centre, with this proportion rising to 48% within 3 hours of presenting. Since less than one-third (31%) of the referrals requested could be initiated within 3 hours, even the speediest was irrelevant for the even- tual outcome of a third of the children. A further WHO Bulletin OMS. Vol 76 1998260 Packaged treatment for cerebral malaria and meningitis r- LO) CM 0 cv) T-Cf) 0 CO, CD 0) c\ N Cv) 0 CD 1* 0 CM co _- CDM <z 0) Cv) C\lN\ CN L) cv) (D 0 o o o CN co 0) r- co LO) 0 cv) CD CD CN 0 10 0) co z z0) z o (1) coX o 2 tV1 > C CD a) cEc c EC oE-c m: Z -a 0 H- WHO Bulletin OMS. Vol 76 1998 -a 0 la 'a).a) o a) Z > c0 Q a a) a) a).0 a) E0.. cn2 6 z co a) 0) a) c Cb V- a) 0DaCZ) -o0 m 0 C) 0 > Z o -a) 10 V 'a a)Z >U 0. C,) 0 0 3 E - E 0 0 -W 0 ._ -4 'E Lu 0. S 0 CO 6 a) (D (D 0. ai -C m E 0 - ~ Ca CD co8 _ > _ n v->- . C/ c co cn C) .0a 0D D CO C 0 a)c a))N.)c 0 U _ - -- a) *U o - .° =C .O O 'C ca 0>c Ec C 8 o U = Cn COMY>Yw: 2 MM co CUq co O 261 T.R. Cullinan & C. Pieterick Table 4: Study outcome, by simplified coma score Coma score = 0 Coma score 1 Coma score = 2 No record Decision Total Died Total Died Total Died Total Died Referral made and 1 0 25 5 14 2 1 1 achieved Referral made, but 5 5 13 6 6 1 1 0 not achieved No referral made 2 0 11 0 15 1 2 0 Total 8 5 (63) 49 11(22) 35 4 (11) 4 1 (25) a Figures in parentheses are percentages. 15% of children referred during the trial phase never reached hospital, and it took over 12 hours before transport was available for another 14%. By this time, 76% of all children who eventually died were already dead. The death of the single child for whom no referral was planned (Table 3) occurred within an hour of arrival at the health centre. In effect, what was conceived as an emergency health centre treat- ment for very sick children was extended to com- plete case management for almost half of them. No detailed analyses of the reasons for referral versus health centre management can be made, since they depend on too many interactive factors, includ- ing parental fears and wishes, clinical condition, transport availability, access problems and medical assistants' recent experiences. Although coma scores at presentation in the health centre are a fairly good predictor of infant death (case fatality rate = 63%, 22%, 11%, respectively, for children with a coma score of 0, 1, and 2 (Table 4)), it did not seem to influence the decision to refer. Thus, children with coma scores of 0-1 were not significantly more likely to be referred than those with a coma score of 2. A total of 57% of children presented to a health centre where referral access was difficult or very difficult (access scores of score 0 or 1) and less than half Table 5: Referral access scores for the study children Score 0 1 2 3 Total Referral made and achieved 5 13 9 14 41 Referral made, but not achieveda 5 16 2 2 25 No referral made 7 6 6 9 28 Total 17 35 17 25 94 a Two referrals were attempted by bicycle, but not achieved. (46%) of the 39 planned referrals were eventually achieved (Table 5). Where access was easier (scores of 2 or 3), only 15% of the 27 planned referrals were unsuccessful. Although a low coma score identified children at particular risk of dying (Table 4), and access (Table 5) is correlated with the success of referral, the numbers involved were too small to draw any association between access and risk of death. However, communication and distance were not the only factors upon which decisions were based. Most children who fully recovered in the health centre left after 2-3 days and the average hospital stay for those referred was 4-5 days. The 10 children who remained in the health centre needed further paraldehyde control for convulsions following admission, but all appeared to make a full recovery. In a Lickert scale questionnaire administered to the medical assistants following the trial, 13 of the 15 who replied rated the packages for emergency care very highly, and the remaining two highly. All found the modified coma score easy to understand and use. Discussion Inevitably, a pilot trial such as the present creates as many questions as it answers. Overall results suggest a considerable reduction in mortality (50%) as a re- sult of packaging emergency care and making it available at entry into the health care system. The mortality levels produced are comparable to those in the hospital (2). However, to achieve such a reduc- tion, several problems have been observed, as dis- cussed below. * First, there is the possibility to mask meningitis, leading to its inadequate management. There is no way to ensure that medical assistants can differenti- ate between meningitis and severe/cerebral malaria WHO Bulletin OMS. Vol 76 1998262 Packaged treatment for cerebral malaria and meningitis at the health centre level, and this trial was not carried out during the meningitis season; in fact, epidemic meningitis is not recognized in southern Malawi.b * Second, it is difficult to determine the proper role of referral to hospital in the light of the results ob- tained in the trial. With the exception of severely anaemic children who have a need for blood transfu- sion, complete case management at a health centre, provided it is preceded by adequate emergency treatment, seems to offer as much as transfer to hos- pital, especially as this is so often delayed. The emer- gency pack costs approximately US$ 6, and while other costs such as extra drugs and the need for a longer stay in the health centre must be added to this, these costs are trivial compared with the cost of referral. The mean round distance between health centres and the hospital is 104km; in fuel alone, this is equivalent to US$ 9 per referral, although as the busiest health centres are also the most distant, US$ 12 is probably a better estimate. To this must be added the extra costs of a hospital bed, food for both child and parent and the family costs of the return journey. Further considerable savings to the price of the packs could have been made had a source of glucose powder rather than dextrose solution been available and had nasogastric tubes been provided separately, since most were not used. * Third, the trial did not measure neurological sequelae as an outcome. In a study carried out by Molyneux et al. in 1989, albeit of children with a generally lower coma score than those in the present study, 11% of the survivors had a discernible neuro- logical deficit (1). These authors developed a practi- cal index of prognostic indicators, although the most frequently used rely on laboratory support. The only clinical indicator with a high relative risk was a coma score of zero, but with the referral circumstances suggested in the present trial, even this is of limited utility for decision-making. Despite these problems, children staying in the health centres appeared to do as well as those re- ferred to hospital, even though no more than 63% were recorded as receiving the full treatment agreed to in the study protocol. Estimating the costs and benefits of providing packaged treatment at remote health centres is not easy because discounting against the costs of refer- ral, which so often depends on the chance availabil- b There is some evidence that single-dose chloramphenicol is as effective as 5 days of antibiotics in the treatment of meningococcal meningitis (8) but it is not the accepted treatment in Malawi (4). ity of vehicles, is hazardous. Including the US $6 for the emergency pack, the estimated cost of each life saved was US$ 29. This seems low by standards in developed countries, but compared to the annual expenditure of approximately US$ 2.50 per capita for all health care in Malawi, over two-thirds of which comes from external sources, it is high. On the other hand, transfer to hospital using a hospi- tal-based ambulance, which costs approximately US$ 0.30perkm on average, corresponds to not less than US$ 30 per transfer. Conclusions A marked drop in the mortality of infants and chil- dren presenting with signs and symptoms indicative of acute/cerebral malaria or meningitis can be achieved by preparing packs containing items neces- sary for emergency treatment and making them available at the first point of contact with the health service. However, further controlled studies are needed to determine whether these packs can be routinely used at the periphery. Acknowledgements The medical assistants in Mangochi District are thanked for their willing cooperation from the beginning to the end of the study. Prof. M. Molyneux and Dr E. Molyneux and Sister Mphundi are thanked for directing the training of the medical assistants. The trial was supported by a grant from the Malaria Unit, Division of Control of Tropical Diseases, World Health Organization, Geneva, Switzerland. Resume Trousses pretes a 1'emploi pour le traitement d'urgence du neuropaludisme et de la meningite Cet article decrit les resultats d'un essai realis6 de janvier a juin 1996 dans le sud du Malawi pour d6terminer l'efficacit6 d'une trousse prete a l'emploi pour le traitement des nourrissons et des enfants de moins de six ans conduits en urgence dans les centres de sant6 ruraux avec un diagnostic pr6somptif de paludisme grave/neuropaludisme ou de m6ningite. Chaque trousse (coOt approximatif: USD 6) contient le necessaire pour un traitement complet: quinine pour voie intramusculaire, chloramph6nicol pour voie intramusculaire, dex- trose, parald6hyde, une sonde nasogastrique, des seringues pr6emballees et de l'eau sterile. Elle WHO Bulletin OMS. Vol 76 1998 263 T.R. Cullinan & C. Pieterick contient 6galement une 6chelle modifiee de profondeur du coma et un nomogramme posologique. 11 est possible d'obtenir une baisse sensible de la mortalit6 des nourrissons et des enfants pr6sentant des signes et symptomes evocateurs de paludisme grave/neuropaludisme ou de meningite en preparant des trousses contenant tout le necessaire pour le traitement d'urgence et en mettant ces trousses a la disposition des premiers points de contact avec les services de sante. Cependant, des 6tudes contr6l6es sont n6cessaires pour d6terminer si de telles trousses peuvent etre utilis6es en routine au niveau p6riph6rique. References 1. Molyneux M et al. Clinical features and prognostic indicators in paediatric cerebral malaria: a study of 131 comatose Malawian children. Joumal of medicine, 1989, 71: 441-459. 2. Slutsker L et al. In-hospital morbidity and mortality due to malaria associated severe anaemia in two areas of Malawi with different patterns of malaria infection. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1994, 88: 548-551. 3. United Nations/Government of Malawi. Situation analysis of poverty in Malawi. Lilongwe, Malawi, 1993. 4. Malaria prevention in pregnancy: the Mangochi malaria research project. ASCI-CCD Internation Health Pro- gram Office, Centers for Disease Control, Atlanta, 1995. 5. Government of Malawi/Ministry of Health, Malawi. Standard treatment guidelines, 2nd ed. Lilongwe, Malawi, 1993. 6. Mansor SM et al. The safety and kinetics of intramus- cular quinine in Malawian children with moderately se- vere falciparum malaria. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1990, 84: 482-487. 7. Taylor T et al. Blood glucose levels in Malawian children before and during the administration of in- travenous quinine for severe falciparum malaria. New England journal of medicine, 1988, 319: 1040-1047. 8. Wali SS et al. Single injection treatment of menin- gococcal meningitis with long-acting chloramphenicol. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1979, 73: 698-702. 264 WHO Bulletin OMS. Vol 76 1998