High adherence to anti-tuberculosis treatment
among patients attending a hospital and slum
health centre in Nairobi, Kenya
M. RAGUENAUD
1
, R. ZACHARIAH
2
, M. MASSAQUOI
3
,
V. OMBEKA
4
, H. RITTER
1
, & J. M. CHAKAYA
4
1
Medecins sans Frontieres, Medical Department, Brussels Operational Centre, Belgium,
2
Medecins sans Frontieres, Medical Department (Operational Research), Brussels
Operational Centre, Belgium,
3
Medecins Sans Frontieres, Nairobi, Kenya, and
4
National
Leprosy and TB Control Programme, Ministry of Health, Nairobi, Kenya
Abstract
We conducted a study among patients with tuberculosis (TB) attending two health
facilities*a hospital and a slum health centre*in Nairobi, in order to: (a) assess
adherence to anti-TB treatment; and (b) identify reasons for non-adherence.
Urine Isoniazid (INH), used as a proxy for overall adherence, was detected in 142 (97%
{95% CI 92C199}) of the 147 patients involved in the study. Five patients had no INH
detected in urine and had run out of pills within the previous three days. The reasons
included: not having enough pills to last until the appointment date (1); delays due to work
or family reasons (2); needing to seek money for transport (1); and losing some pills (1).
Anti-TB treatment adherence is high, and this is reassuring information as Kenya plans
to change to a superior first-line regimen based on rifampicin throughout the course of
anti-TB treatment. Providing patients with a three-day ‘‘excess stock’’ of pills would
provide a ‘‘safety net’’ for continued treatment.
Keywords: Tuberculosis, isoniazid, rifampicin, adherence, Kenya
Introduction
An international multi-centre randomised controlled trial (Jindani et al. 2004),
showed that a six-month anti-tuberculosis (TB) regimen containing four potent
Correspondence: R. Zachariah, Medecins sans Frontieres, Medical Department (Operational
Research), Brussels Operational Centre, 68 Rue de Gasperich, L-1617, Luxembourg, Belgium.
Tel: (352) 332515. Fax: (352) 335133. E-mail: zachariah@internet.lu
Global Public Health, October 2008; 3(4): 433C1439
ISSN 1744-1692 print/ISSN 1744-1706 online # 2008 Taylor & Francis
DOI: 10.1080/17441690802063205
Downloaded By: [Institute of Tropical Medicine] At: 09:11 24 October 2008
anti-TB drugs, namely ethambutol, isoniazid (INH), rifampicin and pyrazina-
mide, administered for the first two months, followed by a four-month course of
two potent drugs, namely daily rifampicin and INH, was superior to regimens that
used ethambutol (instead of rifampicin) and INH after the first two months.
Following the publication of this study, National TB control programmes
(NTPs) in developing countries, that rely on an eight-month regimen based on
ethambutol and INH, are urged to reappraise their first-line regimen to a
standardised six-month one that is based on rifampicin and INH.
Kenya, a high TB burden country in sub-Saharan Africa, has a high annual TB
case rate of 320 per 100 000 (Van Gorkom et al. 1999; MOH 2002), and is in the
process of planning to change from its current first-line anti-TB regimen
(currently based on ethambutol and INH), to the recommended six-month
regimen, with rifampicin and isonizid, in the continuation phase.
In changing to such a regimen, one of the valid concerns of the NTP and
partners, is to try to prevent the acquisition and transmission of rifampicin-
resistant TB bacilli by ensuring high adherence to treatment. The current level of
adherence to anti-TB treatment, under routine programme conditions, is likely to
be a proxy of the level of adherence to be expected with an eventual change in
regimen.
We decided to conduct a study among patients attending routine TB clinics in
two health facilities (a district hospital and a slum health centre) in urban Nairobi
in order to: (a) assess adherence to anti-TB treatment in the self-administered
continuation phase; and (b) identify reasons for non-adherence.
Methods
The study was conducted between July and August 2005, in Nairobi, the capital
city of Kenya, with an estimated population of 2.5 million people. Two health
facilities, the Mbagathi district hospital and the Kibera health centre, which
together register about 700 new TB cases per month, were involved in the study.
Both Mbagathi hospital and the Kibera health centre are TB diagnostic and
treatment centres for central Nairobi. The Kibera health centre lies in the heart of
Kibera, one of Africa’s largest slums.
All TB patients, once registered at one of these two sites, are started on an
eight-month course of a standardised anti-TB treatment regimen, which involves
an initial phase lasting two months followed by a continuation phase lasting six
months. During the initial phase, patients swallow four anti-TB drugs on a daily,
‘‘directly observed manner’’, at one of the two facilities. Once this period has been
completed, patients enter the continuation phase in which they take two drugs for
the six-month period. During this phase, drugs are ‘‘self-administered’’, and
patients are free to choose to continue their treatment at health facilities close to
their residence where they receive pill refills on a monthly basis. Out of the 700
patients registered on a monthly basis at the Mbagathi and Kibera health
facilities, only about 30% reside proximal to one of these facilities and, thus,
chose to go on with the continuation phase of treatment at one of these two health
434 M. Raguenaud et al.
Downloaded By: [Institute of Tropical Medicine] At: 09:11 24 October 2008
facilities. Thus, about 210 patients are still under follow-up as of the third month
of anti-TB treatment at these two facilities. During the continuation phase, drug
refills are provided on a monthly basis.
In terms of sample size calculation, we estimated that 90% of patients in our
setting would be adherent, and have INH detected in the urine (guesstimate). We
wanted to detect values 5% below and above this estimate with a 95% confidence
interval (CI). Using the epi-info statistical calculator for sample size calculations,
a total of 138 samples was required. To correct for non-responders, this was set at
150. Consecutive TB patients with routine follow-up visits (scheduled) to these
two health facilities, as of the third month of treatment, were involved in the
study. Recruitment was conducted during a consecutive period of three weeks,
and stopped thereafter as the desired sample size had been achieved.
An interviewer-administered questionnaire, which had been pre-tested on a
different group of 10 TB patients, was used to gather information on anti-TB
treatment and adherence. Interview appointments were not pre-scheduled and
patients were presenting as part of routine follow-up visits to the health
facilities. Interviews were conducted in Swahili (the local language) by
experienced interviewers at the time of presentation to the health facility. The
same team was used throughout the study, and interviewers, although part of
the clinical team of the health facility, were not directly involved with routine
TB care of the participants. Adherence to anti-TB treatment in the continuation
phase was assessed as follows. There was a verbal (subjective) verification of
anti-TB drug intake in the previous 72 hours, and a record made of the reasons
(as explained by the patient) for not having been able to take medication.
Patients were specifically asked to specify when they took their last dose of INH
prior to presentation at the clinic, and, in case they had missed their INH pills,
to specify the possible reasons. If the patient was a small child, information was
gathered from the accompanying parent or guardian. Once the subjective
verification was completed, patients were requested to provide a urine sample.
Patients did not have knowledge of this procedure prior to subjective verification
of adherence. This was done in order to offer some evidence about the veracity
of patient self-reporting on adherence. A 10 ml sample of urine was collected,
after verbal informed consent, in a plastic container, and the presence of INH
was verified on site using Taxo INH urine test strips (Becton-Dickinson, USA).
INH, and/or its metabolites, are detectable in 100% of specimens within 24
hours after the last intake of the drug (Elizaga & Friedland 1997; Meissner et al.
2002). Depending on whether an individual is a fast or slow acetylator of INH,
some urine specimens might remain weakly sensitive from 36 to 72 hours.
Patients were not informed of the study prior to their visit, nor that their self-
reported adherence (verbal verification) would be corroborated with a urine
test.
Patients were not given any incentives for participating in the study. Health
workers at the two health facilities were individually interviewed on their
experience regarding the INH urine test as a routine screening tool.
High adherence to anti-TB treatment in Nairobi 435
Downloaded By: [Institute of Tropical Medicine] At: 09:11 24 October 2008
Data was analysed using EPI-Info (Centres for Disease Control and Preven-
tion, Atlanta, GA). The level of significance was set at PC300.05, and 95% CIs
were used throughout.
Results
Characteristics of the study population
Of 150 TB patients, who were expected to have attended for follow-up in the two
health facilities during the period of the study, all turned up at the facilities. Data
were, however, available for only 147 (98%) of the 150 patients; one patient file
that had been misplaced and could not be traced, and two patients that could not
provide urine samples and were, therefore, excluded from the analysis. Of the 147
patients, 72 (49%) were from Mbagathi hospital, and 75 (51%) were from the
Kibera health centre; there were 87 (60%) men, and 55 (40%) women (median
age 30 years, range 4C164). Fifty-seven patients had smear-positive TB, 74 had
smear-negative TB, and 11 had extra-pulmonary TB. The median duration, at
the time of INH verification, was 4.5 months (interquartile range 4.2C15.9
months). Table I shows the characteristics of patients found with positive and
negative urine INH test results.
INH drug adherence. INH was detected in 142 (97% {95% CI: 92C199}) of 147
urine samples (see Table I). Five patients had no INH detected in their urine and,
for the purposes of this study, were considered non-adherent. Four of the five
patients, declared non-adherent, were being followed up at the Kibera health
centre, while one was from Mbagathi. Five health personnel in the two clinics
were involved with this study, and all found the urine test strips ‘‘easy to use’’ and
interpret.
Reasons for being non-adherent. The five patients who had no INH detected in
their urine, claimed to have run out of pills within the previous 72 hours, and had
self-reported non-adherence in the absence of knowledge that a urine sample was
to be taken for cross verification. The reasons for running short of pills included:
not having enough pills to last till the scheduled appointment date (1); was
delayed due to work or family reasons (2); had to seek money for transport (1);
and losing some pills (1).
Discussion
Using urine INH detection as a measure of determining adherence to anti-TB
treatment, and the degree of adherence of TB patients during the continuation
phase, under routine programme conditions in the two health facilities, was
satisfactorily high.
The strength of this study is that it assessed adherence at two typical sites of
anti-TB treatment in Nairobi, and since it was conducted under routine
436 M. Raguenaud et al.
Downloaded By: [Institute of Tropical Medicine] At: 09:11 24 October 2008
programme conditions, the results are likely to reflect the reality on the ground.
The approach also included an additional objective indicator (urine testing),
and is, thus, likely to be more reliable than simple subjective (verbal)
verification, which is the usual method used for verifying adherence in these
settings. However, this study is limited by the fact that we used a restrictive
definition of adherence, in that the results were dependent on pill intake during
the 72 hours prior to urine INH assessment. Although we used this ‘snapshot’
as a proxy for adherence, understandably, it may not reflect adherence over the
entire self-administered continuation phase of anti-TB treatment. Furthermore,
as the study involved only two health facilities in Nairobi, the findings might not
necessarily be representative of other settings. Thus, it would be worthwhile to
replicate this study in other facilities, including those in rural Kenya to enhance
the generalisability of the findings. It might also be justified to implement this
study at a second point in time during treatment. Although we do not have data
to substantiate this, the finding, that four of five INH negative patients were
from the slum health centre, might imply that socio-economic factors, or other
factors that make these two groups different, might be having an influence on
adherence. Further studies in this particular regard might also be worthwhile.
Table I. Characteristics of patients with positive and negative isoniazid (INH) urine test results
(nC30147).
Characteristics INH positive, N (%) INH negative, N (%) Total
Gender
Female 55 (96.5) 2 (3.5) 57
Male 87 (97) 3 (3) 90
Age group (years)
0C110 5 (100) 0 5
11C120 8 (100) 0 8
21C130 59 (95.2) 3 (4.8) 62
31C140 50 (96.2) 2 (3.8) 52
C2140 20 (100) 0 20
Type of TB
a
SmearC1positive PTB
b
57 (95) 3 (5) 60
SmearC1negative PTB 74 (97) 2 (3) 76
EPTB
c
11 (100) 0 11
Health facility
Mbagathi hospital 71 (99) 1 (1) 72
Slum health centre 71 (95) 4 (5) 75
Last INH pill intake
d
B24 hours 140 1 141
Between 24 and 36 hours 2 3 5
Between 36 and 72 hours 0 1 1
Total 142 (96.5%) 5 (3.5%) 147
a
Tuberculosis.
b
Pulmonary tuberculosis.
c
Extra-pulmonary tuberculosis.
d
From the time of presentation at the health facility.
High adherence to anti-TB treatment in Nairobi 437
Downloaded By: [Institute of Tropical Medicine] At: 09:11 24 October 2008
Meanwhile, these results provide reassuring information on adherence (albeit
restricted to the study settings), as the NTP plans to implement a six-month, first-
line regimen, based on INH and rifampicin, where a high adherence rate is
essential to prevent rifampicin resistance. Although the intensive phase of anti-TB
treatment involves the administration of drugs under daily direct supervision
(observed swallowing of pills), prior to the continuation phase, patients receive
specific educational programmes, to assure treatment literacy, and empowerment
to take their drugs in a self-administered manner at home. The high adherence
rate would, thus, suggest that rifampicin could also be administered in a similar
manner in first-line regimens, and it might be possible to maintain a high
adherence level, as observed in this study, as long as patients are ‘‘treatment
literate’’, and are well ‘‘empowered’’ to be adherent. The use of rifampicin as an
integral part of Fixed Dose Combination tablets, will further reduce the risk of
irrational use of rifampicin as monotherapy, and the risk of rifampicin resistance.
The fact that all the individuals, considered non-adherent in this study, had
openly revealed to their health workers that they had run out of pills, implies that
these individuals were honest with regard to drug intake, and this, too, is
reassuring information.
The International Union Against Tuberculosis and Lung Disease (IUATLD)
and the World Health Organisation (WHO) currently recommend that regimens
containing rifampicin should only be prescribed if they can be administered under
direct observation (Enarson et al. 2000). For a resource-limited setting such as
Kenya, this poses important operational barriers. From a health facility
perspective, it would mean that, already overstretched, health staff, will need to
take up the additional daily burden of directly observed treatment for large
numbers of TB patients in the continuation phase of treatment, for perhaps no
added benefit (Volmink & Garner 2003). From a patient point of view, daily
treatment at health facilities would mean unavoidable transport and other indirect
costs, which many ill TB patients might simply be unable to afford. Other
options, such as, self-administered treatment with peer support, or guardian-
based treatment at home or in the community, might seem more appropriate.
In a large urban setting such as Nairobi, where people come from different
walks of life, it is understandable that some patients will be unable to report at the
treatment centre on the exact day of the follow-up appointment for family or
work-related reasons, some might lack money for transport, or may lose a few
pills. Although this involved just five individuals in this study, providing patients
with a three-day excess stock of pills would have provided the necessary ‘‘safety
net’’ for continued treatment. The use of such a safety stock, with cotrimoxazole
prophylaxis within the TB programme in Malawi, was not associated with undue
abuse of the system (Zachariah et al. 2001). If the proportion of individuals, who
were unable to make it to the clinic on the exact date of the follow-up
appointment, was extrapolated to the more than 100 000 or so newly registered
TB patients in Kenya (Van Gorkom et al. 1999), there might be an estimated
3401 patents who might benefit from such an initiative for continued treatment.
438 M. Raguenaud et al.
Downloaded By: [Institute of Tropical Medicine] At: 09:11 24 October 2008
Finally, our experience with the INH urine dipsticks, as a ‘‘field tool’’ for
monitoring adherence, was encouraging, in that programme staff in the two
facilities found the dipsticks easy to use, and results immediately available ‘‘on
site’, in a transparent manner. This tool could be particularly relevant for
evaluating adherence to self-administered treatment, not only in health facilities,
but also at the community level. However, the costs of these tests remain high,
and advocacy is necessary to make this tool accessible for national programmes in
resource-limited, high TB burden, countries.
Acknowledgements
This study was a collaborative effort of the Ministry of health, national TB control
programme, and Medecins Sans Frontieres, Brussels. We are very grateful to the
health staff of the Mbagathi hospital and Kibera health centre TB clinics, for their
efforts in TB control and in this study. We are particularly grateful to Mr
Christopher Kimeto, TB Nurse, Kibera Slum Health Centre, and Mr Simon
Mwanthi, TB Supervisor (Clinical Officer), and Mary Kwambai, TB Nurse, for
Mbagathi District Hospital, for their contributions in data collection and urine
INH testing. We acknowledge the contributions of Pascale Chaillet (Laboratory
advisor-MSF) for facilitating the procurement of the INH urine dip sticks. This
study was financed by Medecins Sans Frontieres-Brussels.
References
Elizaga, J. and Friedland, J. S. (1997) Monitoring compliance with anti-tuberculosis treatment by
detection of isoniazid in urine. Lancet, 350, 1225C11226.
Enarson, D. A., Reider, H. L., Arnadottir, T. and Trebucq, A. (2000) Management of Tuberculosis: A
Guide for Low Income Countries (5th edition) (St-Just-La-Pendue: Compogravure Impression,
Broachage Imprimirie).
Jindani, A., Jun, A. J. and Enarson, D. A. (2004) Two 8-month regimens of chemotherapy for
treatment of newly diagnosed pulmonary tuberculosis: International multicentre randomised
trial. Lancet, 364, 1244C11251.
Meissner, P. E., Musoke, P., Okwera, A., Bunn, J. E. and Coulter, J. B. S. (2002) The value of urine
testing for verifying adherence to anti-tuberculosis chemotherapy in children and adults in
Uganda. International Journal of Tuberculosis and Lung Disease, 6, 903C1908.
MOH (Ministry of Health) (2002) National Leprosy and Tuberculosis Programme Kenya. Annual
Report Nairobi, Kenya.
Van Gorkom, J., Kibuga, D., Adallah, S., Adengosi, J., Aluvaala, B., Kangangi, J., et al. (1999) HIV
sero-prevalence among tuberculosis patients in Kenya. East African Medical Journal, 76, 8C113.
Volmink, J. and Garner, P. (2003) Directly observed therapy for treating tuberculosis (Cochrane
Review). In The Cochrane Library, Issue 2 (Oxford: Update Software).
Zachariah, R., Harries, A. D., Arendt, V., Wennig, R., Schneider, S., Spielmann, M. P., et al. (2001)
Compliance with cotrimoxazole prophylaxis for the prevention of opportunistic infections in
HIV-positive tuberculosis patients in Thyolo district, Malawi. International Journal of
Tuberculosis and Lung Disease, 5(9), 843C1846.
High adherence to anti-TB treatment in Nairobi 439
Downloaded By: [Institute of Tropical Medicine] At: 09:11 24 October 2008