- Email: [email protected]
The impact of HIV on resource utilization by patients with tuberculosis in a tertiary referral hospital, Nairobi, Kenya
The impact of HIV on resource utilization by patients with tuberculosis in a tertiary referral hospital, Nairobi, Kenya P. Nunn*‘*,
S. Gathua*,
D. Kibuga*,
R. Binge’, R. Brindles,
J. Odhiambo’,
K. McAdam:
glnfectious Diseases Hospital. Kenyatta National Hospital, Nairobi. Kenya, ‘Respiratory Diseases Research Unit. Kenya Medical Research Institute, Nair-obj, Kenya, $Department of Clinical Sciences, London School qf Hygiene and Tropical Medicine. London, UK and $Public Health Laboratory, 0.vfol.d. UK S U M M A R Y. By using routinely collected data and results from research studies at the Infectious Diseases
Hospital (IDH), Nairobi, we have begun to determine the scale of the increase in resource utilisation and treatment costs for tuberculosis control services caused by the HIV epidemic. New cases of tuberculosis registered annually at the IDH rose 61%, from 447 in 1985 to 720 in 1990. HIV seroprevalence among patients with tuberculosis rose from 7.5% in 1986 to 42% in 1990. The inpatient mortality rate rose from 8.4% in 1985 to 16.8% in 1989, but fell to 13.5% in 1990. HIV-positive patients were admitted to hospital on 2 or more occasions more often than HIV-negative patients (Relative risk (RR)= 2.46,95% confidence intervals (CD, l.l-5.7), but average duration of admission was similar for the 2 groups. Significantly more HIV-positive patients were prescribed antibiotics, antifungal agents, antidiarrhoeal agents, analgesics and corticosteriods than HIV-negative patients. Microbiological investigations, apart from those for tuberculosis, were performed more commonly among HIV-positive patients (RR= 2.0, 95% CI 1.0-4.2). Using this data, the average cost of ideal drug therapy, including antituberculosis drugs and treatment for intercurrent infections and other complications, was estimated using 1992 prices (ECHO, Coulsdon Surrey, UK). The costs were US$16.62 and US$32.94 for HIV-negative patients using ‘standard’ therapy (2STH/lOTH) and short course therapy (2SHRZ/6TH) respectively, and US$41.18 for HIV-positive patients using a short-course regimen without thiacetazone (2EHRZ/ 6EH). The HIV epidemic is causing both an increase in the numbers of patients requiring treatment and an increase in the average cost of treatment per patients. This latter increase is almost entirely due to the cost of treating intercurrent infections other than tuberculosis. R k S U M l? En utilisant des donnees recueillies en routine et les resultats d’etudes de recherche a I’Hibpital des
Maladies Infectieuses (IDH) de Nairobi, a ete entreprise une evaluation de I’echelle d’augmentation concernant I’utilisation des ressources et les cotits de traitement interessant les services de lutte antituberculeuse, du fait de l’epidemie VIH. Le nombre de nouveaux cas de tuberculose, enregistres tous les ans par I’IDH, a augmente de 61,0%, soit de 447 en 1985 a 720 en 1990. La seroprevalence du VIH parmi les tuberculeux a augmente de 7,5% en 1986 a 42,0% en 1990. La mortalite chez les patients hospital& a augmente de 8,4% en 1985 a 16,8% en 1989, mais s’est abaissee a 13,5% en 1990. Des patients VIH+ ont ete admis a l’hopital au moins a deux reprises plus souvent que les patients VIH- (risque relatif [RR]= 2,46, 95% d’intervalle de confiance [ICI, l,l5,7) ; cependant la duree moyenne de sejour a ete similaire pour ies deux groupes. Sensiblement davantage de patients VIH+ ont regu un traitement comprenant des antibiotiques, des agents antifongiques, des agents anti-diarrheiques, des anaigesiques et des corticosteroides que les patients VIH-. Davantage d’examens bacteriologiques, a part ceux pour la tuberculose, ont ete effect& chez les patients VIH+ (RR= 2,0, 95% IC 1,0-4,2). A partir de ces donnees le cotit moyen d’un traitement medicamenteux ideal, incluant medicaments antituberculeux et traitement pour les infections intercurrentes et autres complications, a CtCestime selon les prix de 1992 (ECHO, Co&don, Surrey, UK). Les couts Ctaient de $16,62 US et $32,94 US pour des patients VIH -qui suivaient respectivement un traitement ‘standard’ (2STHIlOTH) et un traitement de courte duree (2SHRZ/6TH), et de $41,18 US pour les patients VIH+ qui suivaient un traitement de courte duree sans thiacetazone (2EHRZ16EH). L’epidemie VIH est responsable d’une augmentation a la fois dans le nombre de
Correspondence to: Dr Paul Nun. Tuberculosis Programme. TUB/CDS. World Health Organisation. CH 121 I Geneva 27, Switzerland. Prpy wc~ar~m/31 Aqwst 1992. Fid rrc~ceptuwe4 Fehruar~ 1993 273
274
Tubercle and Lung Disease
patients requerant un traitement et dans le cout moyen de traitement par patient. Cette derniere augmentation est presque entierement due au coiit du traitement des infections intercurrentes autres que la tuberculose. R E S U M E N. Los autores tratan de determinar el grado de aumento en la utilization de recursos y en 10s costos de tratamiento para 10s servicios de control de la tuberculosis, debido a la epidemia VIH, usando datos recolectados rutinariamente y resultados de estudios en el Hospital de Enfermedades Infecciosas (IDH) de Nairobi. El nhmero de cases nuevos de tuberculosis registrados anualmente en el IDH aumento en 61% : de 447 en 1985 a 720 en 1990. La seroprevalencia VIH en 10s enfermos tuberculosos aumento de 75 en 1986 a 42,0% en 1990. La tasa de mortalidad en 10s enfermos hospitalizados aumento de 8,4 en 1985 a 16,8% en 1989, pero disminuyo a 13,5% en 1990. La hospitalization en 2 o mas ocasiones fue mas frecuente en 10s pacientes VIHpositivos que en 10s VIH-negativos (Riesgo relativo [RR]=2,46, con interval0 de confianza [ICI 95%, l,l-5,7), pero el promedio de la duration de la hospitalization fue similar en 10s dos grupos. Los pacientes VIH-positivos recibieron significativamente mas antibioticos, medicamentos antimicoticos, antidiarreicos, analgesicos y corticoesteroides que 10s VIH-negativos. Los examenes bacteriologicos, fuera de aquellos para la tuberculosis, fueron realizados mas frecuentemente en 10s pacientes VIH-positivos (RR= 2,0, IC 95% &O-4,2). A partir de estos datos se estimo el costo promedio de un tratamiento medicamentoso ideal que incluye medicamentos antituberculosos y tratamiento de las infecciones intercurrentes y otras complicaciones. Utilizando 10s precios de 1992 (ECHO, Coulsdon, Surrey, Reino Unido), 10s costos fueron de US$ 16,62 y de US$ 32,94 para 10s pacientes VIH-negativos tratados con quimioterapia estandar y con quimioterapia de corta duration respectivamente y de US!+ 41,18 para 10s pacientes VIH-positivos tratados con quimioterapia de corta duration sin tioacetazona (2EHRZ/6EH). La epidemia de VIH esta causando un aumento tanto del numero de pacientes coma de1 promedio de1 costo por paciente. Este ultimo aumento es debido casi enteramente al tratamiento de las infecciones intercurrentes fuera de la tuberculosis.
INTRODUCTION
METHODS
The human immunodeficiency virus (HIV) pandemic is causing a major resurgence of tuberculosis in many developing countries. ‘L’In some areas, HIV seroprevalence rates of over 60% in patients with tuberculosis have been reported4 and the annual notification rate of tuberculosis has risen markedly since the mid-1980’s in a number of African countries.* Rising populations have also contributed to the increase in absolute numbers of cases. Developing countries are therefore faced with the problem of the increased cost of tuberculosis treatment services brought about solely by the increase in numbers of cases. In addition, evidence is accumulating suggesting that cases of HIV associated tuberculosis may be more expensive to treat than tuberculosis without HIV infection, owing to increased rates of adverse reactions to treatment,“-” recurrent episodes of tuberculosis,“-” and HIV related intercurrent infections.‘s.‘6 In order to plan tuberculosis treatment services adequately in developing countries it is desirable to obtain an estimate of the cost of treating an HIV-positive compared to an HIV-negative patient with tuberculosis. We have therefore made use of routine data collection sources and results from research studies into the impact of HIV on tuberculosis at the Infectious Disease Hospital, Nairobi, Kenya (IDH), to assess the HIV-related increase in some of the costs of treating tuberculosis patients. We have calculated the average drug costs required for HIV-positive patients with tuberculosis compared to HIV-negative patients, taking into account the need to avoid thiacetazone,‘O the increased mortality rate,15 and the increase in recurrences, in HIV-positive patients in Nairobi. I4
IDH is part of the Kenyatta National Hospital, and is the main referral hospital for cases of tuberculosis requiring admission from the Nairobi area, and also for cases requiring specialist treatment from the whole of Kenya.” Three wards are almost exclusively used by patients with tuberculosis. Routine data Data on annual numbers of new cases of tuberculosis registered, cases of tuberculosis admitted, in-hospital mortality among patients with tuberculosis, and bed occupancy (number of occcupied tuberculosis bed-days divided by the number of bed-days available) are routinely collected by the Medical Records Department at IDH. Numbers of chest radiographs performed each day are recorded by the Radiology Department. These data were collated for 1985-1990 inclusive. Chart review A systematic 5% sample of new cases of tuberculosis was selected from each of the years 1985-1990 by pulling every 20th patient’s file from the shelves in the Medical Records Department. Data were extracted. checked, entered on a questionnaire, transcribed to computer database (DBase III Plus, Ashton-Tate, Escondido, California, USA) and analysed by the Statistical Package for the Social Sciences, SPSS V2.0 (Chicago, Illinois, USA). Research studies I. Seropevalence
study. A study of HIV prevalence was
The impact of HIV on resource utilization
by patients with tuberculosis
performed among new cases of tuberculosis presenting between January 1987 and June 1988.” HIV seropositivity was defined as a positive ELISA test (Organon, Oss, The Netherlands). A sample of 21 tests positive on ELISA were also tested by Western Blot and 19 were confirmed positive. Seroprevalence results obtained by ELISA during this period were thus corrected by multiplying by 19/2 1. 2. Cohort st~ly. Between April 1 1989 and May 31 1990 all adults (>lS years of age) with a first episode of suspected tuberculosis who attended IDH, and who were resident within the Nairobi City Commission boundary, were asked to take part in the study which has been previously described.li.lh In brief. patients were enrolled only after pre-test couselling for HIV-testing had been given and informed consent had been obtained. Those t’ound subsequently not to have tuberculosis were excluded. Patients were followed to determine mortality rates. and recurrence rates after the completion of antiutberculosis therapy. In this study, HIV seropositivity was assessed by two enzyme-linked immunosorbent dssay (ELISA) tests (‘DuPont HTLV-III test’. DuPont de Nemours, Wilmington, Delaware, USA, and ‘Wellcozyme’. Wellcome Diagnostics, Dartford. UK). Only those seropositive on 2 or more test runs with both ELISAs were considered HIV- 1 seropositive. A study of 506 patients in Nairobi showed this definition to be completely concordant with Western Blot for HIV-l.‘” There are, as yet. no reports of HIV-2 in Kenya. Throughout this report HIV- I seropositivity is referred to as HIV positivity. Routine antituberculosis chemotherapy at the time of this study consisted of 12 months of oral thiacetazone/ isoniazid (TH) in a combined preparation of thiacetazone (T) 150 mg and isoniazid (H) 300 mg (Cosmos Ltd. Nairobi). supplemented for the first 4 weeks with streptomycin (S) (Dawa Pharmaceuticals Ltd. Nairobi) 1 g intramuscularly daily (STH regimen). 88 of 107 (82%) HIV-I positive and 124/174 (71%‘) HIV-l negative patients received this exact regimen and 3/107 (3%) and 1l/174 (6%) respectively received another antituberculosis drug or drugs in addition. From September 1989 patients at IDH previously untreated for their tuberculosis were given a combined preparation of rifampitin (R). isoniazid (H), and pyrazinamide (Z) (‘Rifater’, Merrell-Dow) for 2 months with streptomycin (S) for the first month only. and thiacetazone/isoniazid from the 3rd to the 8th month inclusive (2SHRZ/6TH regimen). 16 HIV-l positive (15%) and 39 HIV-l negative (22%) patients received this regimen.
Drug costs For administrative
reasons it was not possible
to calcu-
The term ‘recurrence’ include\ both patients with a genuine relapse of their original infection and those with subsequent episodes of tuberculosis secondary, to re-infection. Evidence from this study (unpublished) suggest\ both mechanisms can occur. Almost all recurrences occurred in HIV-positive patients taking the STH regimen. but the number of those treated with ZSHRZ/6TH was small (see later).
275
in a tertiary referral hospital. Nairobi. Kenya
late the actual expenditure on drugs at IDH for HIVpositive compared to HIV-negative patients. Some drugs were not always available, and data which demanded a change in treatment policy was being generated throughout the period of this study. For example. it was not obvious until late 1989 that thiacetazone should be replaced by another drug owing to severe cutaneous hypersensitivity reactions to thiacetazone occurring in HIV-positive patients. ‘o Therefore the ideal average drug costs, for both antituberculosis therapy and non-antituberculosis drugs, for HIV-positive and negative patients with tuberculosis were estimated on the following assumptions: that patients would be tested for HIV at presentation; thiacetazone would be replaced by ethambutol for those who are HIV-positive; recurrence of tuberculosis occurs in at least 9% of HIV-positive patients who start treatment compared to 1r/c HIV-negative patients: ‘-I 36% (38/107) of HIV-positive patients die before the end of treatment, compared to only 7% ( 131174) of HIVnegative patients (unpublished: data derived from cohort study, above): and episodes of intercurrent infection in HIV-negative patients are susceptible to beta-lactam antibiotics but in HIV-positive patients require gentamicin.‘” Drug costs were obtained from the 1992 Pharmaceutical Price List (ECHO. The Joint Mission Hospital Equipment Board, Coulsdon, Surrey. England) and do not include costs of insurance. packaging or transport.
RESULTS Hospital statistics New cases of tuberculosis registered annually at IDH rose from 447 in 1985 to 720 in 1990. a rise of 61% (Table I and Figure 1). In the same period the number of admissions of patients with tuberculosis rose from 8.58 to 1084, an increase of 26%. and the annual number of deaths among patients with tuberculosis in hospital rose from 72 to 146. an increase of 102%. The mortality rate among admissions of patients with tuberculosis rose from 8.4% in 1985 to 16.8% in 1989, but fell to 13.5% in 1990. Bed occupancy rose from 68% in 19XS to Xl % in 1990. an increase of 19%. The number of chest radiographs performed annually rose from X73 in 1985 to 1763 in 1990. an increase of 102%.
HIV seroprevalence The average HIV seroprevalence among cases of tuberculosis registered annually was assessed by chart review Table 1. Routine \tatisttcs 19X5-1990
from the Infectious
Di\ea\e\
Hospital.
Statistic
19x5
19X6
1987
198X
19x9
IWO
New cases of tuberculosis (No., Admissions (No.) Deaths (No.) Bed occupancy (%) Chest radiopraphs (No.)
137 X58 72 68 X73
4X-t 761 71 60 958
473 x19 6X 71 890
494 X51 Y2 74 1086
603 X3
720 IOM
I .3x 7X 157X
116 XI 1763
276
Tubercle and Lung Disease Nos. of new cases
600
1987
1988
Year Fig. l-Numbers IDH, 1985-1990.
of new cases of tuberculosis
for 1986, measured directly from the seroprevalence study data in 1987 and 1988, and extrapolated to the whole year from the cohort study in 1989 and 1990 (Table 2 and Figure 2). Good agreement was obtained Table 2. Numbers, mean age, sex and HIV status of patients analysed in chart review
Number of patients Mean age (SD) Sex* Females (No.) Males (No.) No. tested for HIV No. HIV-positive % HIV-positive? % HIV-positive by methods other than chart review*
1985
1986
1987
1988
1989
1990
Total
22 35.6
24 37.9 (16)
24 30.4 (18)
25 32.8 (13)
30 34.8 (16)
36 28.3 (13)
161 -
9 14 3 2 7.5
10 14 6 2 7.5
13 12 18 10 36
11 19 22 8 27
19 17 28 15 42
72 88 77 37 22
11
17
28
42
-
(21) 10 12
-
*Sex of one patient not clear from chart tcorrected for confirmation by Western Blot in years 1986-88 total *see text.
and
HIV seroprevalence (W)
40
30
20
10
1986
1987
1988
1989
Year
Fig. 2-HIV
Seroprevalence
presenting to IDH, 1985-1990.
in new patients with
tuberculosis
annually
to
for years with results from different sources, with the exception of 1988 where chart review gave a figure of 36% and direct measurement during the seroprevalence study was 17%. We have used the latter in all subsequent calculations since it is likely to be the more accurate. The number of new cases of tuberculosis registered annually was significantly correlated with HIV prevalence (Pearson correlation coefficient= 0.97, P < 0.01).
Chart review
50
A85 I
presenting
1990
39 of 161 (24%) patients
reviewed were HIV-negative, 37 (23%) were HIV-positive, and 85 (53%) were of undetermined HIV status. Mean ages (SD) were 36.3 (14) years, 28.5 (10) and 33.3 (19) respectively. The HIVpositive patients were significantly younger than the HIV-negative patients (P = 0.007). 6 of 39 (15%) HIV-negative patients were admitted to hospital 2 or more times, compared to 14/37 (38%) HIV-positive patients (Relative risk (RR)= 2.46, 95% confidence intervals (CI), 1.1-5.7). Of those of undetermined HIV status 1 l/85 (13%) were admitted 2 or more times. However, tlie mean (SD) duration of each admission was 39.0 (25) days for HIV-negative patients, 43.5 (28%) for HIV-positive patients and 36.4 for those not tested for HIV. The proportions of patients with positive sputum smears, positive sputum cultures, suspected tuberculosis or extrapulmonary tuberculosis did not differ significantly with HIV status. This is in contrast to previous studies from other countries in Africa,3s4 and is probably due to well-established referral patterns in Nairobi whereby smear-positive patients tend to be referred to IDH.5.‘8 Significantly more HIV-positive patients were prescribed antibiotics, antifungal agents, antidian-hoeal agents, analgesics or corticosteroids (Table 3). No other group of drugs was prescribed more commonly among HIV-positive patients.
The impact of HIV on resource utilization
by patients with tuberculosis
in a tertiary referral hospital. Nairobi. Kenya
Table 4. Average drug costs for 50 kg HIV-negattve and positive patients with tuberculosis (US$).?O See methods for assumptions
Table 3. Numbers of patients with tuberculosis than antituberculosis agents
using drugs other
HIV Neg. n=3Y (%)
HIV Pos. n=37 (%)*
Relative risk (957G CT)
DN~
12 0 29 5
21 24 33 14 23
2.5 27.0 1.2 3.0 24.2
Antituberculosis therapy ‘Deduction for death before end of therapy :Retreatment SAntibiotics Ketoconazole Loperamide Paracetamol Prednisolone HIV test
DW Antibiotic Xntidiarrhoeal Antipyretic Prednisone .Antifungal
I
(31) (0) (74) (13) (3)
(77) (67) (92) (39) (62)
‘n varies between 35-37 for HIV-positive data.
(1.5-11.1) (3.8-187) (1.0-1.5) (1.2-7.6) (3.6171)
patients, owing to missing
Microbiological investigations, apart from those for tuberculosis. were performed in 8/39 (20%) HIV-negative patients and in 15/36 (42%) HIV-positive patients (RR= 2.0, 95% CI 1.0-4.2). Full blood counts (FBC) and chemical pathological investigations were performed more commonly in both HIV-negative and HIVpositive patients than in patients of undetermined HIV status (90%, 94% and 34% respectively for FBC [RR= 2.7 95% CI 2.0-3.6 for patients of known HIV status versus unknown] and 63%, 81% and 25% respectively for chemical pathology tests [RR= 2.9 95% CI 1.9-4.3 for patients of known HIV status versus unknown]). 20 of 37 (54%) HIV-positive patients were known to have died by the time of the review, compared to 6139 (15%) HIV-negative patients (RR= 3.5 95% CI 1.6-7.8).
Estimated
drug costs
The estimated costs of drug therapy for HIV-positive patients with tuberculosis treated with a modified shortcourse regimen without thiacetazone (2EHRZ/6EH) are 25% higher than for HIV-negative patients treated with a short-course regimen (2SHRZ/GTH) containing thiacetazone. The increase is largely accounted for by the cost of treatment of intercurrent infections, and not by the cost of substituting ethambutol for thiacetazone (Table 4). However the cost of treatment for HIV-positive patients is 2.5 times greater than that for HIV-negative patients treated with ‘standard’ regimen (2STH/lOTH).
DISCUSSION This is the first study, to our knowledge, comparing utilization of resources between HIV-negative and HIVpositive patients with tuberculosis. Previous studies have looked at the cost-effectiveness of treating tuberculosis in HIV-positive patients, but without such a detailed comparison of resource consumption.?’ In this study, we have shown a 62% increase in the annual notifications of new cases of tuberculosis presenting to the main tuberculosis hospital in Kenya in the period 1985-1990. In the same interval the catchment population was estimated to have increased by 26%.“? HIV prevalence among tuberculosis cases rose from 7.5% in 1986, the first year tests were performed in this population in Kenya, to 42% in 1990. We believe that the likely cause of the increased incidence is the spread of HIV infection
277
Total
HIV-negative 2SHRZl6TH
2STH/10THX
HIV-positive 2EHRZI6EH
3 I. 15
14.93
32.72
-0.14 0.19 0.30 0.22 _
-0.24 0.19 0.30 0.22
-5.21 2.YI 1.58 1.56 0.14
0.1 I 0.1 1
1.o 32.94
(1.1I (I.1 I I .o 16.62
0.II 0.34
I .o 41.18
*S= streptomycin H= isoniazid R= rifampicin Z= pyrazinamide T= thiacetazone E= ethambutol. The numbers denote months of therapy. .;In previous studies’” death during treatment occurred on average at 2 months after start of treatment for HIV-negatives, saving 6 months of treatment for those on 2SHRZ/6TH and 10 months for those on 2STH/lOTH. For HIV-positives death occurred at about 5 months on average. saving 3 months of treatment. Deductions are calculated from the proportion of patients dying before end of treatment multiplied by the costs of treatment unused.
in Nairobi. Other factors are less probable: tuberculosis treatment services within Nairobi have not increased, and immigration of refugees was not a problem over this period. Nor has routine treatment improved in such a way as to attract increased numbers of patients; streptomycin, isoniazid and thiacetazone for the first month, followed by isoniazid and thiacetazone, has remained the standard treatment since the mid 1960’s. We believe the increase in number of deaths per year and in mortality rate (deaths per 100 patients admitted) is also caused by the increase in HIV prevalence. The slight fall in mortality rate in 1990 may represent better identification and treatment of the common causes of HIV-related morbidity. The failure of some measures of resource utilisation to increase with HIV seroprevalence, such as number of admissions, probably represents the ability of medical workers to limit increases in word load. The numbers of medical staff did not alter between 1985 and 1990. The apparent similarity of average duration of hospital stay per admission in HIV-positive compared to negative patients may conceal a high early mortality in HIV-positive patients” but a longer duration of admission in the survivors. Resource utilization is difficult to measure. but numbers of diagnostic tests for tuberculosis have probably
278
Tubercle and Lung Disease
increased at a rate proportional to the increase in cases. The most important of these, sputum microscopy and culture for M. tuberculosis, we could not measure accurately owing to the long periods of reductions in the service beginning in 1989 due to lack of specimen bottles. Use of chest radiography appears to have increased out of proportion to the rise in cases, but this is probably the result of research studies at IDH that demanded routine radiography during follow-up, even without symptoms. The increased use of haematological and chemical pathology investigations in both HIV-positive and negative patients, compared to those of unknown HIV status, is probably due to the practice of performing ‘routine’ HIV tests, full blood counts and electrolyte measurements in any patient who fails to respond readily to treatment of tuberculosis. Again this may have been caused, at least in part, by the presence of research studies. Utilisation of other investigations and some therapies is more related to HIV status and is likely to have increased out of proportion simply to the rise in cases, although we did not directly measure this. We specifically examined microbiological investigations for presumed intercurrent infections and the drugs used for treating them, for non-specific manifestations of infection (antipyretices, and for adverse effects of antituberculosis treatment (prednisone). All of these tests and drugs were requested or prescribed significantly more often among HIV-positive patients. This is consistent with previous data in HIV infected patients with tuberculosis showing an increase in adverse drug reactions, specifically to thiacetazone,‘O and an increase in intercurrent infections.‘5,‘6 The increase in intercurrent infections is not specific to patients with tuberculosis since it is also seen in other groups of HIV infected patients in Nairobi.‘” Consumption of diagnostic tests and drugs was compared in this study in those known to be HIV-positive compared to those known to be HIV-negative. In the early part of the study HIV testing was not carried out routinely, but only in those who were suspected to be HIV infected and were therefore probably more ill than the remainder of the patients. It is possible that those tested were consuming resources at a rate greater than the average HIV-positive or negative patient. This may therefore artificially increase the estimates of the costs of treatment, but the effect is likely to be similar in both HIV-positive and negative groups. Also, from 1989 onwards, HIV testing was systematic, and 62% of the HIV-positive patients in the chart review were recruited in 1989 or 1990. We have attempted to estimate the relative cost of treating HIV-positive and negative patients for tuberculosis. Such estimates depend critically on drug supplier, method of payment, quantity purchased and other assumptions, and are therefore intended only as a guide to decision-making. The most important assumption is that HIV-positive patients are treated with a regimen without thiacetazone. We have selected the 2EHRZ/6EH regimen as a standard for comparison, since it is the obvious
choice for a regimen without thiacetazone while rifampicin remains so expensive. It is also recommended by the World Health Organisation.” If the cost of needles, syringes and water for injection of streptomycin is taken into account it is substantially cheaper than 2SHRZ/6EH (US$39.60 for the antituberculosis drugs alone) and not much more expensive than 2SEH/lOEH (US$31.85) (See Table 4 for explanation of regimens). However, it is important to emphasise that although the 2EHRZ/6EH regimen is likely to be more effective than a longer regimen without rifampicin (2SEH/lOEH), it has not been properly evaluated in HIV-positive patients. Thus. the main expense for a developing country wishing to improve treatment for HIV-positive patients and currently using ‘standard’ regimen would be the introduction of short-course treatment. It would more than double the costs of antituberculosis drugs. For a country already using a short-course regimen the main HIV-related increased costs to tuberculosis treatment services are those for the treatment of intercurrent infections. This increased cost is mainly due to gentamicin. In contrast to HIV-negative patients, intercurrent infections among HIV-positive patients are mostly due to gram-negative organisms which are usually resistant to beta-lactam antibiotics, sulphonamides and trimethoprim.lh Second or third generation cephalosporins, or quinolones, are not generally available in developing countries and are more expensive than gentamicin. For a complete assessment of the increase in costs caused by the impact of HIV on tuberculosis, an estimate of the fixed costs of a tuberculosis control programme and of other variable costs, such as food for hospital inpatients, would be required. Obtaining such figures was impossible in this study. so we concentrated on those accessible to us at IDH. In spite of the limited nature of our study, it is clear that the HIV epidemic is presenting developing countries with an enormous increase in the cost of treating tuberculosis due to both an increase in annual notifications and an increase in the average cost of management per patient. Without substantial external assistance there is a very real possibility that tuberculosis treatment services in developing countries will be overwhelmed.
Acknowledgements The writers thank the patients and staff of IDH, especially Mrs Muthoni and her co-workers in the Medical Records Department; Dr W. Watkins and his team at the Wellcome Trust Research Laboratories, Nairobi: MS Ann Pulun for typing the manuscript: MS Cat Ley and MS Susie Foster for critically reviewing it, and the Director of the Kenya Medical Research Institute (KEMRI) for permission to publish these findings. The study was conducted under the KEMRIWellcome Trust Collaborative Research Programme. PN gratefully acknowledges support from the Wellcome Trust.
References 1. Harries A D. Tuberculosis and human immunodeficiency virus infection in developing countries. Lancet 1990: 335: 387- 390. 2. Murray J F. Cursed duet: HIV infection and tuberculosis. Respiration 1990: 57: 210-220.
The impact of HIV on resource
utilization
by patients with tuberculosis
3. Colebunders R L, Ryder R W, Nzilambi N et al. HIV infection in patients with tuberculosis in Kinshasa, Zaire. Am Rev Respir Dis 1989; 139: 1082-1085. 4. Elliott A M, Luo N, Tembo G et al. Impact of HIV on tuberculosis in Zambia: a cross sectional study. BMJ 1990; 301: 412415. 5. Nunn P P, Gicheha C. Hayes R et al. Cross-sectional survey of HIV infection among patients with tuberculosis in Nairobi, Kenya. Tuberc Lung Dis 1992: 73: 45-5 I. 6. De Cock K M. Gnaore E. Adjorlolo G et al. Risk of tuberculosis in patients with HIV-l and HIV-2 infections in Abidjan, Ivory Coast. BMJ 1991: 302: 496-499. 7. Perriens J H, Mukadi Y, Nunn P P. Tuberculosis and HIV infection: implications for Africa. AIDS 1991: 5 (supplement): S i 27-S 133. 8. Kochi A. Government intervention programs in HIV/ tuberculosis infection. Bull Int Union Tuberc Lung Dis 1991: 66: 33-36. 9. Eriki P P, Okwera A. Aisu T, Morrissey A B, Ellner J J, Daniel T M. The influence of human immunodeficiency virus infection on tubercuiosis in Kampala. Uganda. Am Rev Respir Dis 1991: 143: 185- 87. IO. Nunn P, Kibuga D. Gathua S et al. Cutaneous hypersensitivity reactions due to thiacetazone in HIV- 1 seropositive patients treated for tuberculosis. Lancet 1991; 337: 627430. I 1. Wirima J J . Harries A D. Stevens-Johnson syndrome during anti-tuberculous chemotherapy in HIV-seropositive patients: report on six cases. East Afr Med J 1991; 68: 64-66. 12. Hira S, Wadhawan D, Kamanga J et al. Cutaneous manifestations of human immunodeficiency virus in Lusaka. Zambia. J Am Acad Dermatol 1988; 19: 451457. 13. Perriens J H, Colebunders R L, Karahunga C et al. Increased mortality and tuberculosis treatment failure rate among human immunodeficiency virus (HIV) seropositive compared with HIV seronegative patients with pulmonary tuberculosis treated with
14.
15.
16.
17.
18.
19.
20.
21.
22. 23.
in a tertiary referral hospital. Nairobi. Kenya
279
‘standard’ chemotherapy in Kinshasa. Zaire. Amer Rev Respir Dis 1991; 144: 750-755. Hawken M. Nunn P, Gathua S et al. HIV- 1 infection and recurrence of tuberculosis, Nairobi. Kenya. VIIIth International Conference on AIDS. Amsterdam 1992: TUB 0538. Tu33. Nunn P P. Brindle R. Carpenter L et al. Cohort study of HIV infection in tuberculosis patients, Nairobi. Kenya: analysis of early (6 months) mortality. Am Rev Respir Dis 1093; 146: 849-854. Brindle R. Batchelor B, Gathua S. Nunn P. Waiyaki P. Infection and morbidity in patients with tuberculosis in Natrobi. VIth International Conference on AIDS in Africa. Dakar 1991; WA198: 289. (Abstract). Kibuga D, Gathua S. Nunn P. A study of HIV Infection in association with tuberculosis patients seen at Infectious Diseases Hospital (IDH), Nairobi. Vth International Conference on AIDS. Montreal 1989: Th.G.05-989. (Abstract). Githui W, Nunn P P. Juma E et al. Cohort study of HIV-positive and HIV-negative tuberculosis, Nairobi. Kenya: comparison of bacteriological results. Tuberc Lung Dis 1992: 73: 203-209. Gilks C F. Brindle R. Otieno L et al. Life-threatening bacteraemia in HIV- 1 seropositive adults admitted to hospital in Nairobi. Kenya. Lancet 1990: 336: 545-549. ECHO. Pharmaceutical price list and order form. Coulsdon, Surrey, England: The Joint Mission Hospital Equipment Board Ltd. 1992. Murray C J. De Jonghe E, Chum H, Nyangulu D. Salamao A. Styblo K. Cost-effectiveness of chemotherapy for pulmonary tuberculosis in 3 sub-Saharan African countries. Lancer 1991; 338: 1305-1308. National Leprosy and Tuberculosis Control Programme. Annual Report. Nairobi, Kenya: Ministry of Health, 1989. World Health Oganisation. Severe hypersensitivity reactions among HIV-seropositive patients with tuberculostc treated with thioacetazone. Wkly Epidemiol Ret 1992: 67: 1-3.
S. Gathua*,
D. Kibuga*,
R. Binge’, R. Brindles,
J. Odhiambo’,
K. McAdam:
glnfectious Diseases Hospital. Kenyatta National Hospital, Nairobi. Kenya, ‘Respiratory Diseases Research Unit. Kenya Medical Research Institute, Nair-obj, Kenya, $Department of Clinical Sciences, London School qf Hygiene and Tropical Medicine. London, UK and $Public Health Laboratory, 0.vfol.d. UK S U M M A R Y. By using routinely collected data and results from research studies at the Infectious Diseases
Hospital (IDH), Nairobi, we have begun to determine the scale of the increase in resource utilisation and treatment costs for tuberculosis control services caused by the HIV epidemic. New cases of tuberculosis registered annually at the IDH rose 61%, from 447 in 1985 to 720 in 1990. HIV seroprevalence among patients with tuberculosis rose from 7.5% in 1986 to 42% in 1990. The inpatient mortality rate rose from 8.4% in 1985 to 16.8% in 1989, but fell to 13.5% in 1990. HIV-positive patients were admitted to hospital on 2 or more occasions more often than HIV-negative patients (Relative risk (RR)= 2.46,95% confidence intervals (CD, l.l-5.7), but average duration of admission was similar for the 2 groups. Significantly more HIV-positive patients were prescribed antibiotics, antifungal agents, antidiarrhoeal agents, analgesics and corticosteriods than HIV-negative patients. Microbiological investigations, apart from those for tuberculosis, were performed more commonly among HIV-positive patients (RR= 2.0, 95% CI 1.0-4.2). Using this data, the average cost of ideal drug therapy, including antituberculosis drugs and treatment for intercurrent infections and other complications, was estimated using 1992 prices (ECHO, Coulsdon Surrey, UK). The costs were US$16.62 and US$32.94 for HIV-negative patients using ‘standard’ therapy (2STH/lOTH) and short course therapy (2SHRZ/6TH) respectively, and US$41.18 for HIV-positive patients using a short-course regimen without thiacetazone (2EHRZ/ 6EH). The HIV epidemic is causing both an increase in the numbers of patients requiring treatment and an increase in the average cost of treatment per patients. This latter increase is almost entirely due to the cost of treating intercurrent infections other than tuberculosis. R k S U M l? En utilisant des donnees recueillies en routine et les resultats d’etudes de recherche a I’Hibpital des
Maladies Infectieuses (IDH) de Nairobi, a ete entreprise une evaluation de I’echelle d’augmentation concernant I’utilisation des ressources et les cotits de traitement interessant les services de lutte antituberculeuse, du fait de l’epidemie VIH. Le nombre de nouveaux cas de tuberculose, enregistres tous les ans par I’IDH, a augmente de 61,0%, soit de 447 en 1985 a 720 en 1990. La seroprevalence du VIH parmi les tuberculeux a augmente de 7,5% en 1986 a 42,0% en 1990. La mortalite chez les patients hospital& a augmente de 8,4% en 1985 a 16,8% en 1989, mais s’est abaissee a 13,5% en 1990. Des patients VIH+ ont ete admis a l’hopital au moins a deux reprises plus souvent que les patients VIH- (risque relatif [RR]= 2,46, 95% d’intervalle de confiance [ICI, l,l5,7) ; cependant la duree moyenne de sejour a ete similaire pour ies deux groupes. Sensiblement davantage de patients VIH+ ont regu un traitement comprenant des antibiotiques, des agents antifongiques, des agents anti-diarrheiques, des anaigesiques et des corticosteroides que les patients VIH-. Davantage d’examens bacteriologiques, a part ceux pour la tuberculose, ont ete effect& chez les patients VIH+ (RR= 2,0, 95% IC 1,0-4,2). A partir de ces donnees le cotit moyen d’un traitement medicamenteux ideal, incluant medicaments antituberculeux et traitement pour les infections intercurrentes et autres complications, a CtCestime selon les prix de 1992 (ECHO, Co&don, Surrey, UK). Les couts Ctaient de $16,62 US et $32,94 US pour des patients VIH -qui suivaient respectivement un traitement ‘standard’ (2STHIlOTH) et un traitement de courte duree (2SHRZ/6TH), et de $41,18 US pour les patients VIH+ qui suivaient un traitement de courte duree sans thiacetazone (2EHRZ16EH). L’epidemie VIH est responsable d’une augmentation a la fois dans le nombre de
Correspondence to: Dr Paul Nun. Tuberculosis Programme. TUB/CDS. World Health Organisation. CH 121 I Geneva 27, Switzerland. Prpy wc~ar~m/31 Aqwst 1992. Fid rrc~ceptuwe4 Fehruar~ 1993 273
274
Tubercle and Lung Disease
patients requerant un traitement et dans le cout moyen de traitement par patient. Cette derniere augmentation est presque entierement due au coiit du traitement des infections intercurrentes autres que la tuberculose. R E S U M E N. Los autores tratan de determinar el grado de aumento en la utilization de recursos y en 10s costos de tratamiento para 10s servicios de control de la tuberculosis, debido a la epidemia VIH, usando datos recolectados rutinariamente y resultados de estudios en el Hospital de Enfermedades Infecciosas (IDH) de Nairobi. El nhmero de cases nuevos de tuberculosis registrados anualmente en el IDH aumento en 61% : de 447 en 1985 a 720 en 1990. La seroprevalencia VIH en 10s enfermos tuberculosos aumento de 75 en 1986 a 42,0% en 1990. La tasa de mortalidad en 10s enfermos hospitalizados aumento de 8,4 en 1985 a 16,8% en 1989, pero disminuyo a 13,5% en 1990. La hospitalization en 2 o mas ocasiones fue mas frecuente en 10s pacientes VIHpositivos que en 10s VIH-negativos (Riesgo relativo [RR]=2,46, con interval0 de confianza [ICI 95%, l,l-5,7), pero el promedio de la duration de la hospitalization fue similar en 10s dos grupos. Los pacientes VIH-positivos recibieron significativamente mas antibioticos, medicamentos antimicoticos, antidiarreicos, analgesicos y corticoesteroides que 10s VIH-negativos. Los examenes bacteriologicos, fuera de aquellos para la tuberculosis, fueron realizados mas frecuentemente en 10s pacientes VIH-positivos (RR= 2,0, IC 95% &O-4,2). A partir de estos datos se estimo el costo promedio de un tratamiento medicamentoso ideal que incluye medicamentos antituberculosos y tratamiento de las infecciones intercurrentes y otras complicaciones. Utilizando 10s precios de 1992 (ECHO, Coulsdon, Surrey, Reino Unido), 10s costos fueron de US$ 16,62 y de US$ 32,94 para 10s pacientes VIH-negativos tratados con quimioterapia estandar y con quimioterapia de corta duration respectivamente y de US!+ 41,18 para 10s pacientes VIH-positivos tratados con quimioterapia de corta duration sin tioacetazona (2EHRZ/6EH). La epidemia de VIH esta causando un aumento tanto del numero de pacientes coma de1 promedio de1 costo por paciente. Este ultimo aumento es debido casi enteramente al tratamiento de las infecciones intercurrentes fuera de la tuberculosis.
INTRODUCTION
METHODS
The human immunodeficiency virus (HIV) pandemic is causing a major resurgence of tuberculosis in many developing countries. ‘L’In some areas, HIV seroprevalence rates of over 60% in patients with tuberculosis have been reported4 and the annual notification rate of tuberculosis has risen markedly since the mid-1980’s in a number of African countries.* Rising populations have also contributed to the increase in absolute numbers of cases. Developing countries are therefore faced with the problem of the increased cost of tuberculosis treatment services brought about solely by the increase in numbers of cases. In addition, evidence is accumulating suggesting that cases of HIV associated tuberculosis may be more expensive to treat than tuberculosis without HIV infection, owing to increased rates of adverse reactions to treatment,“-” recurrent episodes of tuberculosis,“-” and HIV related intercurrent infections.‘s.‘6 In order to plan tuberculosis treatment services adequately in developing countries it is desirable to obtain an estimate of the cost of treating an HIV-positive compared to an HIV-negative patient with tuberculosis. We have therefore made use of routine data collection sources and results from research studies into the impact of HIV on tuberculosis at the Infectious Disease Hospital, Nairobi, Kenya (IDH), to assess the HIV-related increase in some of the costs of treating tuberculosis patients. We have calculated the average drug costs required for HIV-positive patients with tuberculosis compared to HIV-negative patients, taking into account the need to avoid thiacetazone,‘O the increased mortality rate,15 and the increase in recurrences, in HIV-positive patients in Nairobi. I4
IDH is part of the Kenyatta National Hospital, and is the main referral hospital for cases of tuberculosis requiring admission from the Nairobi area, and also for cases requiring specialist treatment from the whole of Kenya.” Three wards are almost exclusively used by patients with tuberculosis. Routine data Data on annual numbers of new cases of tuberculosis registered, cases of tuberculosis admitted, in-hospital mortality among patients with tuberculosis, and bed occupancy (number of occcupied tuberculosis bed-days divided by the number of bed-days available) are routinely collected by the Medical Records Department at IDH. Numbers of chest radiographs performed each day are recorded by the Radiology Department. These data were collated for 1985-1990 inclusive. Chart review A systematic 5% sample of new cases of tuberculosis was selected from each of the years 1985-1990 by pulling every 20th patient’s file from the shelves in the Medical Records Department. Data were extracted. checked, entered on a questionnaire, transcribed to computer database (DBase III Plus, Ashton-Tate, Escondido, California, USA) and analysed by the Statistical Package for the Social Sciences, SPSS V2.0 (Chicago, Illinois, USA). Research studies I. Seropevalence
study. A study of HIV prevalence was
The impact of HIV on resource utilization
by patients with tuberculosis
performed among new cases of tuberculosis presenting between January 1987 and June 1988.” HIV seropositivity was defined as a positive ELISA test (Organon, Oss, The Netherlands). A sample of 21 tests positive on ELISA were also tested by Western Blot and 19 were confirmed positive. Seroprevalence results obtained by ELISA during this period were thus corrected by multiplying by 19/2 1. 2. Cohort st~ly. Between April 1 1989 and May 31 1990 all adults (>lS years of age) with a first episode of suspected tuberculosis who attended IDH, and who were resident within the Nairobi City Commission boundary, were asked to take part in the study which has been previously described.li.lh In brief. patients were enrolled only after pre-test couselling for HIV-testing had been given and informed consent had been obtained. Those t’ound subsequently not to have tuberculosis were excluded. Patients were followed to determine mortality rates. and recurrence rates after the completion of antiutberculosis therapy. In this study, HIV seropositivity was assessed by two enzyme-linked immunosorbent dssay (ELISA) tests (‘DuPont HTLV-III test’. DuPont de Nemours, Wilmington, Delaware, USA, and ‘Wellcozyme’. Wellcome Diagnostics, Dartford. UK). Only those seropositive on 2 or more test runs with both ELISAs were considered HIV- 1 seropositive. A study of 506 patients in Nairobi showed this definition to be completely concordant with Western Blot for HIV-l.‘” There are, as yet. no reports of HIV-2 in Kenya. Throughout this report HIV- I seropositivity is referred to as HIV positivity. Routine antituberculosis chemotherapy at the time of this study consisted of 12 months of oral thiacetazone/ isoniazid (TH) in a combined preparation of thiacetazone (T) 150 mg and isoniazid (H) 300 mg (Cosmos Ltd. Nairobi). supplemented for the first 4 weeks with streptomycin (S) (Dawa Pharmaceuticals Ltd. Nairobi) 1 g intramuscularly daily (STH regimen). 88 of 107 (82%) HIV-I positive and 124/174 (71%‘) HIV-l negative patients received this exact regimen and 3/107 (3%) and 1l/174 (6%) respectively received another antituberculosis drug or drugs in addition. From September 1989 patients at IDH previously untreated for their tuberculosis were given a combined preparation of rifampitin (R). isoniazid (H), and pyrazinamide (Z) (‘Rifater’, Merrell-Dow) for 2 months with streptomycin (S) for the first month only. and thiacetazone/isoniazid from the 3rd to the 8th month inclusive (2SHRZ/6TH regimen). 16 HIV-l positive (15%) and 39 HIV-l negative (22%) patients received this regimen.
Drug costs For administrative
reasons it was not possible
to calcu-
The term ‘recurrence’ include\ both patients with a genuine relapse of their original infection and those with subsequent episodes of tuberculosis secondary, to re-infection. Evidence from this study (unpublished) suggest\ both mechanisms can occur. Almost all recurrences occurred in HIV-positive patients taking the STH regimen. but the number of those treated with ZSHRZ/6TH was small (see later).
275
in a tertiary referral hospital. Nairobi. Kenya
late the actual expenditure on drugs at IDH for HIVpositive compared to HIV-negative patients. Some drugs were not always available, and data which demanded a change in treatment policy was being generated throughout the period of this study. For example. it was not obvious until late 1989 that thiacetazone should be replaced by another drug owing to severe cutaneous hypersensitivity reactions to thiacetazone occurring in HIV-positive patients. ‘o Therefore the ideal average drug costs, for both antituberculosis therapy and non-antituberculosis drugs, for HIV-positive and negative patients with tuberculosis were estimated on the following assumptions: that patients would be tested for HIV at presentation; thiacetazone would be replaced by ethambutol for those who are HIV-positive; recurrence of tuberculosis occurs in at least 9% of HIV-positive patients who start treatment compared to 1r/c HIV-negative patients: ‘-I 36% (38/107) of HIV-positive patients die before the end of treatment, compared to only 7% ( 131174) of HIVnegative patients (unpublished: data derived from cohort study, above): and episodes of intercurrent infection in HIV-negative patients are susceptible to beta-lactam antibiotics but in HIV-positive patients require gentamicin.‘” Drug costs were obtained from the 1992 Pharmaceutical Price List (ECHO. The Joint Mission Hospital Equipment Board, Coulsdon, Surrey. England) and do not include costs of insurance. packaging or transport.
RESULTS Hospital statistics New cases of tuberculosis registered annually at IDH rose from 447 in 1985 to 720 in 1990. a rise of 61% (Table I and Figure 1). In the same period the number of admissions of patients with tuberculosis rose from 8.58 to 1084, an increase of 26%. and the annual number of deaths among patients with tuberculosis in hospital rose from 72 to 146. an increase of 102%. The mortality rate among admissions of patients with tuberculosis rose from 8.4% in 1985 to 16.8% in 1989, but fell to 13.5% in 1990. Bed occupancy rose from 68% in 19XS to Xl % in 1990. an increase of 19%. The number of chest radiographs performed annually rose from X73 in 1985 to 1763 in 1990. an increase of 102%.
HIV seroprevalence The average HIV seroprevalence among cases of tuberculosis registered annually was assessed by chart review Table 1. Routine \tatisttcs 19X5-1990
from the Infectious
Di\ea\e\
Hospital.
Statistic
19x5
19X6
1987
198X
19x9
IWO
New cases of tuberculosis (No., Admissions (No.) Deaths (No.) Bed occupancy (%) Chest radiopraphs (No.)
137 X58 72 68 X73
4X-t 761 71 60 958
473 x19 6X 71 890
494 X51 Y2 74 1086
603 X3
720 IOM
I .3x 7X 157X
116 XI 1763
276
Tubercle and Lung Disease Nos. of new cases
600
1987
1988
Year Fig. l-Numbers IDH, 1985-1990.
of new cases of tuberculosis
for 1986, measured directly from the seroprevalence study data in 1987 and 1988, and extrapolated to the whole year from the cohort study in 1989 and 1990 (Table 2 and Figure 2). Good agreement was obtained Table 2. Numbers, mean age, sex and HIV status of patients analysed in chart review
Number of patients Mean age (SD) Sex* Females (No.) Males (No.) No. tested for HIV No. HIV-positive % HIV-positive? % HIV-positive by methods other than chart review*
1985
1986
1987
1988
1989
1990
Total
22 35.6
24 37.9 (16)
24 30.4 (18)
25 32.8 (13)
30 34.8 (16)
36 28.3 (13)
161 -
9 14 3 2 7.5
10 14 6 2 7.5
13 12 18 10 36
11 19 22 8 27
19 17 28 15 42
72 88 77 37 22
11
17
28
42
-
(21) 10 12
-
*Sex of one patient not clear from chart tcorrected for confirmation by Western Blot in years 1986-88 total *see text.
and
HIV seroprevalence (W)
40
30
20
10
1986
1987
1988
1989
Year
Fig. 2-HIV
Seroprevalence
presenting to IDH, 1985-1990.
in new patients with
tuberculosis
annually
to
for years with results from different sources, with the exception of 1988 where chart review gave a figure of 36% and direct measurement during the seroprevalence study was 17%. We have used the latter in all subsequent calculations since it is likely to be the more accurate. The number of new cases of tuberculosis registered annually was significantly correlated with HIV prevalence (Pearson correlation coefficient= 0.97, P < 0.01).
Chart review
50
A85 I
presenting
1990
39 of 161 (24%) patients
reviewed were HIV-negative, 37 (23%) were HIV-positive, and 85 (53%) were of undetermined HIV status. Mean ages (SD) were 36.3 (14) years, 28.5 (10) and 33.3 (19) respectively. The HIVpositive patients were significantly younger than the HIV-negative patients (P = 0.007). 6 of 39 (15%) HIV-negative patients were admitted to hospital 2 or more times, compared to 14/37 (38%) HIV-positive patients (Relative risk (RR)= 2.46, 95% confidence intervals (CI), 1.1-5.7). Of those of undetermined HIV status 1 l/85 (13%) were admitted 2 or more times. However, tlie mean (SD) duration of each admission was 39.0 (25) days for HIV-negative patients, 43.5 (28%) for HIV-positive patients and 36.4 for those not tested for HIV. The proportions of patients with positive sputum smears, positive sputum cultures, suspected tuberculosis or extrapulmonary tuberculosis did not differ significantly with HIV status. This is in contrast to previous studies from other countries in Africa,3s4 and is probably due to well-established referral patterns in Nairobi whereby smear-positive patients tend to be referred to IDH.5.‘8 Significantly more HIV-positive patients were prescribed antibiotics, antifungal agents, antidian-hoeal agents, analgesics or corticosteroids (Table 3). No other group of drugs was prescribed more commonly among HIV-positive patients.
The impact of HIV on resource utilization
by patients with tuberculosis
in a tertiary referral hospital. Nairobi. Kenya
Table 4. Average drug costs for 50 kg HIV-negattve and positive patients with tuberculosis (US$).?O See methods for assumptions
Table 3. Numbers of patients with tuberculosis than antituberculosis agents
using drugs other
HIV Neg. n=3Y (%)
HIV Pos. n=37 (%)*
Relative risk (957G CT)
DN~
12 0 29 5
21 24 33 14 23
2.5 27.0 1.2 3.0 24.2
Antituberculosis therapy ‘Deduction for death before end of therapy :Retreatment SAntibiotics Ketoconazole Loperamide Paracetamol Prednisolone HIV test
DW Antibiotic Xntidiarrhoeal Antipyretic Prednisone .Antifungal
I
(31) (0) (74) (13) (3)
(77) (67) (92) (39) (62)
‘n varies between 35-37 for HIV-positive data.
(1.5-11.1) (3.8-187) (1.0-1.5) (1.2-7.6) (3.6171)
patients, owing to missing
Microbiological investigations, apart from those for tuberculosis. were performed in 8/39 (20%) HIV-negative patients and in 15/36 (42%) HIV-positive patients (RR= 2.0, 95% CI 1.0-4.2). Full blood counts (FBC) and chemical pathological investigations were performed more commonly in both HIV-negative and HIVpositive patients than in patients of undetermined HIV status (90%, 94% and 34% respectively for FBC [RR= 2.7 95% CI 2.0-3.6 for patients of known HIV status versus unknown] and 63%, 81% and 25% respectively for chemical pathology tests [RR= 2.9 95% CI 1.9-4.3 for patients of known HIV status versus unknown]). 20 of 37 (54%) HIV-positive patients were known to have died by the time of the review, compared to 6139 (15%) HIV-negative patients (RR= 3.5 95% CI 1.6-7.8).
Estimated
drug costs
The estimated costs of drug therapy for HIV-positive patients with tuberculosis treated with a modified shortcourse regimen without thiacetazone (2EHRZ/6EH) are 25% higher than for HIV-negative patients treated with a short-course regimen (2SHRZ/GTH) containing thiacetazone. The increase is largely accounted for by the cost of treatment of intercurrent infections, and not by the cost of substituting ethambutol for thiacetazone (Table 4). However the cost of treatment for HIV-positive patients is 2.5 times greater than that for HIV-negative patients treated with ‘standard’ regimen (2STH/lOTH).
DISCUSSION This is the first study, to our knowledge, comparing utilization of resources between HIV-negative and HIVpositive patients with tuberculosis. Previous studies have looked at the cost-effectiveness of treating tuberculosis in HIV-positive patients, but without such a detailed comparison of resource consumption.?’ In this study, we have shown a 62% increase in the annual notifications of new cases of tuberculosis presenting to the main tuberculosis hospital in Kenya in the period 1985-1990. In the same interval the catchment population was estimated to have increased by 26%.“? HIV prevalence among tuberculosis cases rose from 7.5% in 1986, the first year tests were performed in this population in Kenya, to 42% in 1990. We believe that the likely cause of the increased incidence is the spread of HIV infection
277
Total
HIV-negative 2SHRZl6TH
2STH/10THX
HIV-positive 2EHRZI6EH
3 I. 15
14.93
32.72
-0.14 0.19 0.30 0.22 _
-0.24 0.19 0.30 0.22
-5.21 2.YI 1.58 1.56 0.14
0.1 I 0.1 1
1.o 32.94
(1.1I (I.1 I I .o 16.62
0.II 0.34
I .o 41.18
*S= streptomycin H= isoniazid R= rifampicin Z= pyrazinamide T= thiacetazone E= ethambutol. The numbers denote months of therapy. .;In previous studies’” death during treatment occurred on average at 2 months after start of treatment for HIV-negatives, saving 6 months of treatment for those on 2SHRZ/6TH and 10 months for those on 2STH/lOTH. For HIV-positives death occurred at about 5 months on average. saving 3 months of treatment. Deductions are calculated from the proportion of patients dying before end of treatment multiplied by the costs of treatment unused.
in Nairobi. Other factors are less probable: tuberculosis treatment services within Nairobi have not increased, and immigration of refugees was not a problem over this period. Nor has routine treatment improved in such a way as to attract increased numbers of patients; streptomycin, isoniazid and thiacetazone for the first month, followed by isoniazid and thiacetazone, has remained the standard treatment since the mid 1960’s. We believe the increase in number of deaths per year and in mortality rate (deaths per 100 patients admitted) is also caused by the increase in HIV prevalence. The slight fall in mortality rate in 1990 may represent better identification and treatment of the common causes of HIV-related morbidity. The failure of some measures of resource utilisation to increase with HIV seroprevalence, such as number of admissions, probably represents the ability of medical workers to limit increases in word load. The numbers of medical staff did not alter between 1985 and 1990. The apparent similarity of average duration of hospital stay per admission in HIV-positive compared to negative patients may conceal a high early mortality in HIV-positive patients” but a longer duration of admission in the survivors. Resource utilization is difficult to measure. but numbers of diagnostic tests for tuberculosis have probably
278
Tubercle and Lung Disease
increased at a rate proportional to the increase in cases. The most important of these, sputum microscopy and culture for M. tuberculosis, we could not measure accurately owing to the long periods of reductions in the service beginning in 1989 due to lack of specimen bottles. Use of chest radiography appears to have increased out of proportion to the rise in cases, but this is probably the result of research studies at IDH that demanded routine radiography during follow-up, even without symptoms. The increased use of haematological and chemical pathology investigations in both HIV-positive and negative patients, compared to those of unknown HIV status, is probably due to the practice of performing ‘routine’ HIV tests, full blood counts and electrolyte measurements in any patient who fails to respond readily to treatment of tuberculosis. Again this may have been caused, at least in part, by the presence of research studies. Utilisation of other investigations and some therapies is more related to HIV status and is likely to have increased out of proportion simply to the rise in cases, although we did not directly measure this. We specifically examined microbiological investigations for presumed intercurrent infections and the drugs used for treating them, for non-specific manifestations of infection (antipyretices, and for adverse effects of antituberculosis treatment (prednisone). All of these tests and drugs were requested or prescribed significantly more often among HIV-positive patients. This is consistent with previous data in HIV infected patients with tuberculosis showing an increase in adverse drug reactions, specifically to thiacetazone,‘O and an increase in intercurrent infections.‘5,‘6 The increase in intercurrent infections is not specific to patients with tuberculosis since it is also seen in other groups of HIV infected patients in Nairobi.‘” Consumption of diagnostic tests and drugs was compared in this study in those known to be HIV-positive compared to those known to be HIV-negative. In the early part of the study HIV testing was not carried out routinely, but only in those who were suspected to be HIV infected and were therefore probably more ill than the remainder of the patients. It is possible that those tested were consuming resources at a rate greater than the average HIV-positive or negative patient. This may therefore artificially increase the estimates of the costs of treatment, but the effect is likely to be similar in both HIV-positive and negative groups. Also, from 1989 onwards, HIV testing was systematic, and 62% of the HIV-positive patients in the chart review were recruited in 1989 or 1990. We have attempted to estimate the relative cost of treating HIV-positive and negative patients for tuberculosis. Such estimates depend critically on drug supplier, method of payment, quantity purchased and other assumptions, and are therefore intended only as a guide to decision-making. The most important assumption is that HIV-positive patients are treated with a regimen without thiacetazone. We have selected the 2EHRZ/6EH regimen as a standard for comparison, since it is the obvious
choice for a regimen without thiacetazone while rifampicin remains so expensive. It is also recommended by the World Health Organisation.” If the cost of needles, syringes and water for injection of streptomycin is taken into account it is substantially cheaper than 2SHRZ/6EH (US$39.60 for the antituberculosis drugs alone) and not much more expensive than 2SEH/lOEH (US$31.85) (See Table 4 for explanation of regimens). However, it is important to emphasise that although the 2EHRZ/6EH regimen is likely to be more effective than a longer regimen without rifampicin (2SEH/lOEH), it has not been properly evaluated in HIV-positive patients. Thus. the main expense for a developing country wishing to improve treatment for HIV-positive patients and currently using ‘standard’ regimen would be the introduction of short-course treatment. It would more than double the costs of antituberculosis drugs. For a country already using a short-course regimen the main HIV-related increased costs to tuberculosis treatment services are those for the treatment of intercurrent infections. This increased cost is mainly due to gentamicin. In contrast to HIV-negative patients, intercurrent infections among HIV-positive patients are mostly due to gram-negative organisms which are usually resistant to beta-lactam antibiotics, sulphonamides and trimethoprim.lh Second or third generation cephalosporins, or quinolones, are not generally available in developing countries and are more expensive than gentamicin. For a complete assessment of the increase in costs caused by the impact of HIV on tuberculosis, an estimate of the fixed costs of a tuberculosis control programme and of other variable costs, such as food for hospital inpatients, would be required. Obtaining such figures was impossible in this study. so we concentrated on those accessible to us at IDH. In spite of the limited nature of our study, it is clear that the HIV epidemic is presenting developing countries with an enormous increase in the cost of treating tuberculosis due to both an increase in annual notifications and an increase in the average cost of management per patient. Without substantial external assistance there is a very real possibility that tuberculosis treatment services in developing countries will be overwhelmed.
Acknowledgements The writers thank the patients and staff of IDH, especially Mrs Muthoni and her co-workers in the Medical Records Department; Dr W. Watkins and his team at the Wellcome Trust Research Laboratories, Nairobi: MS Ann Pulun for typing the manuscript: MS Cat Ley and MS Susie Foster for critically reviewing it, and the Director of the Kenya Medical Research Institute (KEMRI) for permission to publish these findings. The study was conducted under the KEMRIWellcome Trust Collaborative Research Programme. PN gratefully acknowledges support from the Wellcome Trust.
References 1. Harries A D. Tuberculosis and human immunodeficiency virus infection in developing countries. Lancet 1990: 335: 387- 390. 2. Murray J F. Cursed duet: HIV infection and tuberculosis. Respiration 1990: 57: 210-220.
The impact of HIV on resource
utilization
by patients with tuberculosis
3. Colebunders R L, Ryder R W, Nzilambi N et al. HIV infection in patients with tuberculosis in Kinshasa, Zaire. Am Rev Respir Dis 1989; 139: 1082-1085. 4. Elliott A M, Luo N, Tembo G et al. Impact of HIV on tuberculosis in Zambia: a cross sectional study. BMJ 1990; 301: 412415. 5. Nunn P P, Gicheha C. Hayes R et al. Cross-sectional survey of HIV infection among patients with tuberculosis in Nairobi, Kenya. Tuberc Lung Dis 1992: 73: 45-5 I. 6. De Cock K M. Gnaore E. Adjorlolo G et al. Risk of tuberculosis in patients with HIV-l and HIV-2 infections in Abidjan, Ivory Coast. BMJ 1991: 302: 496-499. 7. Perriens J H, Mukadi Y, Nunn P P. Tuberculosis and HIV infection: implications for Africa. AIDS 1991: 5 (supplement): S i 27-S 133. 8. Kochi A. Government intervention programs in HIV/ tuberculosis infection. Bull Int Union Tuberc Lung Dis 1991: 66: 33-36. 9. Eriki P P, Okwera A. Aisu T, Morrissey A B, Ellner J J, Daniel T M. The influence of human immunodeficiency virus infection on tubercuiosis in Kampala. Uganda. Am Rev Respir Dis 1991: 143: 185- 87. IO. Nunn P, Kibuga D. Gathua S et al. Cutaneous hypersensitivity reactions due to thiacetazone in HIV- 1 seropositive patients treated for tuberculosis. Lancet 1991; 337: 627430. I 1. Wirima J J . Harries A D. Stevens-Johnson syndrome during anti-tuberculous chemotherapy in HIV-seropositive patients: report on six cases. East Afr Med J 1991; 68: 64-66. 12. Hira S, Wadhawan D, Kamanga J et al. Cutaneous manifestations of human immunodeficiency virus in Lusaka. Zambia. J Am Acad Dermatol 1988; 19: 451457. 13. Perriens J H, Colebunders R L, Karahunga C et al. Increased mortality and tuberculosis treatment failure rate among human immunodeficiency virus (HIV) seropositive compared with HIV seronegative patients with pulmonary tuberculosis treated with
14.
15.
16.
17.
18.
19.
20.
21.
22. 23.
in a tertiary referral hospital. Nairobi. Kenya
279
‘standard’ chemotherapy in Kinshasa. Zaire. Amer Rev Respir Dis 1991; 144: 750-755. Hawken M. Nunn P, Gathua S et al. HIV- 1 infection and recurrence of tuberculosis, Nairobi. Kenya. VIIIth International Conference on AIDS. Amsterdam 1992: TUB 0538. Tu33. Nunn P P. Brindle R. Carpenter L et al. Cohort study of HIV infection in tuberculosis patients, Nairobi. Kenya: analysis of early (6 months) mortality. Am Rev Respir Dis 1093; 146: 849-854. Brindle R. Batchelor B, Gathua S. Nunn P. Waiyaki P. Infection and morbidity in patients with tuberculosis in Natrobi. VIth International Conference on AIDS in Africa. Dakar 1991; WA198: 289. (Abstract). Kibuga D, Gathua S. Nunn P. A study of HIV Infection in association with tuberculosis patients seen at Infectious Diseases Hospital (IDH), Nairobi. Vth International Conference on AIDS. Montreal 1989: Th.G.05-989. (Abstract). Githui W, Nunn P P. Juma E et al. Cohort study of HIV-positive and HIV-negative tuberculosis, Nairobi. Kenya: comparison of bacteriological results. Tuberc Lung Dis 1992: 73: 203-209. Gilks C F. Brindle R. Otieno L et al. Life-threatening bacteraemia in HIV- 1 seropositive adults admitted to hospital in Nairobi. Kenya. Lancet 1990: 336: 545-549. ECHO. Pharmaceutical price list and order form. Coulsdon, Surrey, England: The Joint Mission Hospital Equipment Board Ltd. 1992. Murray C J. De Jonghe E, Chum H, Nyangulu D. Salamao A. Styblo K. Cost-effectiveness of chemotherapy for pulmonary tuberculosis in 3 sub-Saharan African countries. Lancer 1991; 338: 1305-1308. National Leprosy and Tuberculosis Control Programme. Annual Report. Nairobi, Kenya: Ministry of Health, 1989. World Health Oganisation. Severe hypersensitivity reactions among HIV-seropositive patients with tuberculostc treated with thioacetazone. Wkly Epidemiol Ret 1992: 67: 1-3.