Albendazole chemotherapy for treatment of diarrhoea in patients with AIDS in Zambia: a randomised double blind controlled trial

Patients and study design SETTING

The trial was conducted in three centres in urban Zambia where primary AIDS care is delivered. The home care service of the University Teaching Hospital in Lusaka deals mainly with patients referred from the clinics and wards of that hospital. The patients of this service are transported to their homes on discharge from hospital and visited at home regularly. The second study centre (with a similar service) was at Ndola Central Hospital in the north, in Zambia's second most populated city. The third centre was at the Kara Counselling and Training Trust in central Lusaka—the Kara HIV Counselling and Testing project provides a drop in advice and primary care service. The trust also runs several other prevention, training, and counselling projects.

PATIENTS

We recruited patients consecutively. Patients were eligible for enrolment if they were aged over 18 years, were HIV seropositive by enzyme linked immunosorbent assay (ELISA) (Wellcozyme HIV 1 and 2), and had had persistent diarrhoea—loose but not bloody stools three or more times a day for three weeks or longer. We excluded patients if they were pregnant, unable to give informed consent, had received antibiotics in the preceding week, or were deteriorating clinically (Karnofsky score of 20 or less16 (see box). Patients in all three centres had already been counselled and tested for serum antibodies to HIV as part of their care. Consent was sought in the patient's own language by the counsellor or nurse, and a written explanation of the study was provided. We gave each patient a Karnofsky score as a measure of overall severity of illness at the time of entry into the study. The only generally available anthelminthic drug, mebendazole, is not used widely among adults, and none was prescribed to the patients during the study.

RANDOMISATION

After the study objectives had been explained and consent obtained the patients were randomised by allocation of a study pack containing either albendazole or placebo, which had been prepared in London according to a randomisation code. The code (constructed so that the numbers of patients randomised to albendazole and placebo balanced every 20 patients) was kept in London during the study; it was broken during the study only for individual cases when we were concerned about adverse drug reactions. Each pack contained 112 tablets of albendazole or placebo, which were indistinguishable.

STUDY PROTOCOL

Albendazole was administered in a dose of 800 mg twice daily for 14 days. Patients were allowed to receive codeine phosphate (which was supplied if the nurse thought it necessary) or loperamide as required, but the administration of such drugs was recorded. We provided all patients with oral rehydration salts (Unicef formula) and folic acid 5 mg daily. As well as the treatment pack we gave each patient a card on which to record the number of loose stools during each 24 hour period. The patients kept this record for the first month, after which we collected the cards, and the home care staff recorded for the next five months whether diarrhoea occurred in each two week period. The study was terminated after a total of six months of follow up. We asked patients not to take antibiotics for the diarrhoea, but if antibiotics were used for any other indication during the first month of follow up this was recorded. Any potential adverse effects were notified by the home care staff. If there was a potentially serious event the drug was withdrawn, the code broken in London, and the team advised if any further action was needed. Deaths were all notified, together with an assessment of the circumstances to allow assessment of whether death was due to the underlying disease or to an unexpected event.

STATISTICAL EVALUATION

We evaluated efficacy by estimating (a) the proportion of days (in the first month) or fortnights (for the rest of the follow up) during which diarrhoea was experienced; (b) the proportion of patients in remission two weeks after completing treatment and their progress over follow up; and (c) mortality. The presence or absence of diarrhoea was evaluated daily during the first month and then over periods of two weeks during the next five months of follow up. We also assessed the frequency of adverse events. We compared these outcomes by determining the relative risk (and 95% confidence intervals) of experiencing diarrhoea during a day or fortnight conferred by treatment with albendazole. Significance was evaluated with two approaches—firstly, with Fisher's exact test, which assumes independence of the periods during which symptoms were recorded as being present or absent, and, secondly, by calculating the probability of experiencing diarrhoea during each period of follow up of each patient and by carrying out a Mann-Whitney non-parametric comparison. This does not make the independence assumption except in as much as the individual patients are assumed to be independent, but is insensitive during the later periods of follow up as many patients are given the same rank. For this reason, in one case the last five months of follow up were grouped together. Mortality and the probability of relapse after remission were compared with Kaplan-Meier analysis and the log rank statistic. Calculations were carried out with EPI-INFO, version 6 (Centers for Disease Control, Atlanta) and STATA, version 4.0 (Stata, Texas).

Approval for the study was obtained from the national AIDS research ethics committee of Zambia.

REDUCTION IN DIARRHOEA

Diarrhoea was less common in the group taking albendazole than in the group taking placebo during the entire six month period of follow up; this was significant at all time points except during weeks 5 to 8. Table 2 shows the relative risk of experiencing diarrhoea conferred by albendazole (0.71 in weeks 3-4 (a reduction in the proportion of days with diarrhoea of 29%) and 0.58 to 0.83 over the rest of follow up).

Table 3 shows the proportions of patients who were in remission after completion of treatment (26% in the group taking albendazole, 9% in that taking placebo), and their progress over follow up. A significant difference between the two groups was seen at each time point and when the whole period of follow up was compared with a Kaplan-Meier analysis and log rank test (P=0.003). The proportion of patients lost to follow up over the six months was similar in both groups (albendazole, 23/69; placebo, 19/69).

The reduction in diarrhoea was seen in all treatment centres, but the magnitude and significance of the benefit varied. In weeks 3 to 4—that is, after completion of treatment—the patients at the Kara Trust who could be evaluated (n=29) achieved a 31% reduction (P=0.004), those in Ndola (n=57) a 41% reduction (P<0.0001), but those at University Teaching Hospital (n=18) only a 10% reduction (not significant).

The outcome was found to be poor in those few patients who had been given a score of 40 or less at the time of randomisation; at one month eight out of 12 patients in this group had died—no meaningful conclusions could be drawn. The other patients were divided into those with scores of 80 or 90 (able to work) and those with scores of 50 to 70 (needing help with activities of daily living and unable to work, but not needing admission to hospital). Table 4 shows that the benefit of albendazole is conferred on patients with Karnofsky scores of 50 to 70.

DEATH

Death was known to have occurred in 49 patients (in 22 patients taking albendazole and 27 placebo). Log rank analysis showed that albendazole had no significant effect on mortality at any time during follow up (P=0.55).

ADVERSE EVENTS

Adverse events were recorded in 10 patients (in seven taking albendazole and three taking placebo). Five patients had an exacerbation of diarrhoea, together with vomiting. One patient (taking albendazole) had a cutaneous reaction; this was a transient morbilliform rash that resolved within one week of discontinuing the drug. Four other patients taking albendazole complained of dizziness (1), headache (1), cough (1), and difficulty swallowing (1), but these symptoms are indistinguishable from symptoms due to the underlying illness.

EFFICACY OF ALBENDAZOLE

This trial has shown that albendazole reduces the proportion of time that patients with AIDS in urban Zambia have diarrhoea. Remission was achieved in 26% of patients (compared with 9% of those taking placebo), but this proportion rose to 35% (18/52) if the analysis excluded deaths and withdrawals from treatment. The empirical approach adopted in this study shows that patients can expect benefit from treatment without having undergone a search for specific pathogens. The therapeutic benefits were obtained in a setting that applies to many African patients with AIDS and may have been underestimated in our study, as we allowed, for ethical reasons, the concomitant use of codeine phosphate and similar agents. If codeine, antibiotics, or traditional herbal medicines were used more freely than was admitted to, this might lead to a further reduction in apparent efficacy, and we know that the use of herbal remedies is common.

We cannot comment on the value of albendazole in seriously ill patients as in our patients the mortality was high, but we have clear evidence that it is beneficial in patients who cannot work and need help with activities of daily living. In these patients albendazole confers a protection against experiencing diarrhoea at any one time of about 50% over six months. Some of the patients who were given Karnofsky scores of 50 to 70 at randomisation would already have shown improvement after treatment with oral rehydration and codeine in hospital and may have been more seriously ill even a few days previously. Patients with persistent diarrhoea—that is, nonresponders—would have probably been more likely to be lost to follow up, and this may partly explain the improvement in symptoms in both groups over time (table 2). This effect would not explain, however, the difference between the group taking albendazole and that taking placebo.

MORTALITY

Albendazole had no effect on mortality. Perhaps this is not surprising as early deaths are likely to have been determined before enrolment, whether by nutritional deficiency, systemic infection, or the severity of immunosuppression. Albendazole may have an effect on late (after the first two months) mortality, but our study may not have had sufficient power to detect it in that some of the patients who were lost to follow up would have died. Local customs dictate that seriously ill people are taken back to their ancestral home to die, and at death many of their personal belongings are destroyed. This suggests that deaths would be overrepresented in the group whose record cards were not returned despite the efforts of the home care staff to trace their whereabouts. This was a particular problem in the patients attending University Teaching Hospital in Lusaka, which has a high proportion of residents who consider themselves to have a true domicile elsewhere. Patients at the Kara Trust centre were less unwell, and follow up was more successful. Another possible explanation for the lack of effect on mortality could be infection with other pathogens: Lucas et al in Abidjan found unsuspected advanced tuberculosis to be responsible for much mortality in patients with slim disease,17 and septicaemic complications are almost certainly responsible for many deaths.18 Unexpectedly, five of the 21 early deaths occurred in patients with Karnofsky scores of 60 or more.

IMPLICATIONS OF FINDINGS

We surmise that much of the effectiveness of albendazole in the patients in our study was due to its effect in eardicating or suppressing infection with microsporidia. S intestinalis is sensitive to albendazole,19 20 and we have previously reported the presence of this organism in Zambia.10 Infections with other parasites (G intestinalis and S stercoralis) also respond to albendazole, but these infections are less common in patients with AIDS in Zambia. We do not know the response of C parvum or I belli to albendazole. The regimen that we used could probably be improved further with the addition of other chemotherapeutic agents, but antibiotic resistance is very common among isolates of enteric bacteria in Lusaka at the moment.

We know of no other placebo controlled studies that have shown efficacy of a treatment for diarrhoea in patients with AIDS in Africa. Current practice in many areas is to give co-trimoxazole and metronidazole in standard doses, with or without loperamide. In a previous study a substantial response to placebo showed considerable variability in the natural course of the diarrhoea,21 and so the use of a placebo in a blinded study was important and fully justified.22 If the results of our study (including the apparent safety of high dose albendazole) are confirmed, the following recommendations would seem reasonable. On presentation, patients with salt and water depletion should have fluid and electrolyte replacement, usually with an oral glucose electrolyte solution containing potassium. If the diarrhoea has been present for over three weeks then albendazole 800 mg twice daily may be given. If the patient is moribund or if the diarrhoea is of less than three weeks' duration then codeine phosphate should be used to try to control the diarrhoea, as in these situations we have no direct evidence of benefit. In the light of our findings patients with Karnofsky scores of 80 or more should probably be treated in the same way.

Albendazole chemotherapy for AIDS in Africa seems to be beneficial, with minimal adverse reactions, and is effective in empirical use in a community setting. Individuals infected with HIV, and their families, often spend much of their resources on treatments,23 many of which are of doubtful value. We believe that treatment with albendazole, which is effective over six months, is likely to have a high benefit:cost ratio in this devastating disease. Further work is urgently needed to corroborate these findings, to compare albendazole with cheaper antibiotic regimens, and to improve the regimen by addition of other agents.

We acknowledge the dedicated work of the following nursing and counselling staff of the home care services for their careful monitoring of the patients: Mrs N Chiselo, Mrs J Chimoto, Mr J Daka, Mrs G Goma, Mrs A Kapaya, Mrs P Kapembwa, Mrs V Kunda, Mr R Makulele, Mrs Q Moyowanyambe, Mrs S Phiri, Mrs R Sakala, Mr S Silomba, Mrs M Tonha, and Mrs R Zulu. We thank Mr John Young for valuable help with packaging and pharmaceutical quality control, Dr Joan Morris for statistical guidance, and our patients for their willingness and cooperation.