Cost of tuberculosis diagnosis and treatment from the patient perspective in Lusaka, Zambia.

Abstract

Urban primary health centres in Lusaka, Zambia.
1) To estimate patient costs for tuberculosis (TB) diagnosis and treatment and 2) to identify determinants of patient costs.
A cross-sectional survey of 103 adult TB patients who had been on treatment for 1-3 months was conducted using a standardised questionnaire. Direct and indirect costs were estimated, converted into US$ and categorised into two time periods: 'pre-diagnosis/care-seeking' and 'post-diagnosis/treatment'. Determinants of patient costs were analysed using multiple linear regression.
The median total patient costs for diagnosis and 2 months of treatment was $24.78 (interquartile range 13.56-40.30) per patient--equivalent to 47.8% of patients' median monthly income. Sex, patient delays in seeking care and method of treatment supervision were significant predictors of total patient costs. The total direct costs as a proportion of income were higher for women than men (P < 0.001). Treatment costs incurred by patients on the clinic-based directly observed treatment strategy were more than three times greater than those incurred by patients on the self-administered treatment strategy (P < 0.001).
Clinic-based treatment supervision posed a significant economic burden on patients. The creation or strengthening of community-based treatment supervision programmes would have the greatest potential impact on reducing patients' TB-related costs.

Full text

Page 1

INT J TUBERC LUNG DIS 12(8):928–935
© 2008 The Union
Cost of tuberculosis diagnosis and treatment from the patient
perspective in Lusaka, Zambia
A. Aspler,*† D. Menzies,* O. Oxlade,* J. Banda,‡ L. Mwenge,‡ P. Godfrey-Faussett,† H. Ayles†‡
* Respiratory Epidemiology Unit, Montreal Chest Institute, Montreal, Quebec, Canada; † London School of Hygiene &
Tropical Medicine, London, UK; ‡ Zambia AIDS-Related Tuberculosis (ZAMBART) Project, School of Medicine, University
of Zambia, Lusaka, Zambia
Correspondence to: Helen Ayles, Zambia AIDS-Related Tuberculosis (ZAMBART) Project, Ridgeway Campus, University
of Zambia, P O Box 50697, Lusaka, Zambia. Tel: (+260) 21 125 4710. Fax: (+260) 21 125 7215. e-mail: Helen@zambart.
org.zm
Article submitted 20 July 2007. Final version accepted 20 March 2008.
SETTING: Urban primary health centres in Lusaka,
Zambia.
OBJECTIVES: 1) To estimate patient costs for tubercu-
losis (TB) diagnosis and treatment and 2) to identify de-
terminants of patient costs.
METHODS: A cross-sectional survey of 103 adult TB
patients who had been on treatment for 1–3 months was
conducted using a standardised questionnaire. Direct
and indirect costs were estimated, converted into US$
and categorised into two time periods: ‘pre-diagnosis/
care-seeking’ and ‘post-diagnosis/treatment’. Determi-
nants of patient costs were analysed using multiple lin-
ear regression.
RESULTS: The median total patient costs for diagnosis
and 2 months of treatment was $24.78 (interquartile
range 13.56– 40.30) per patient—equivalent to 47.8% of
patients’ median monthly income. Sex, patient delays in
seeking care and method of treatment supervision were
signifi cant predictors of total patient costs. The total di-
rect costs as a proportion of income were higher for
women than men (P < 0.001). Treatment costs incurred
by patients on the clinic-based directly observed treat-
ment strategy were more than three times greater than
those incurred by patients on the self-administered treat-
ment strategy (P < 0.001).
CONCLUSION: Clinic-based treatment supervision posed
a signifi cant economic burden on patients. The creation
or strengthening of community-based treatment supervi-
sion programmes would have the greatest potential im-
pact on reducing patients’ TB-related costs.
KEY WORDS: patient costs; tuberculosis; Zambia;
economics
ALTHOUGH SUB-SAHARAN AFRICA represents
only 11% of the world’s population, over a quarter of
tuberculosis (TB) cases and 31% of TB-related deaths
occurred among Africans in 2003.1 Zambia has one
of the highest TB incidence rates in Africa (680 per
100 000 population),2 as well as one of the highest
human immunodefi ciency virus (HIV) seroprevalence
rates among 15– 49 year olds (17%).3 Case notifi ca-
tion data from Zambia indicate that rates of TB infec-
tion and disease have increased eightfold over the past
two decades, largely due to the concurrent HIV epi-
demic,4–6 as in other parts of sub-Saharan Africa1,7,8
Despite the usefulness of incidence and prevalence
rates for assessing burden of disease and highlighting
the seriousness of the epidemic, these indicators may
fail to defi ne trends in the social and economic burden
of disease.9 It is important not only to emphasise the
urgency of the increasing TB rates in the context of
increased HIV prevalence, but also to address the eco-
nomic and structural barriers that may be acting syn-
ergistically to fuel both epidemics.10,11
Patient costs are, in part, a function of the structure
of the health system, with pre-diagnostic costs refl ect-
ing accessibility of diagnosis and post-diagnostic costs
refl ecting the organisation of TB management and
care. In many cases, there is a signifi cant period of
delay between the patient’s fi rst experience of cough/
haemoptysis/fever and a defi nitive diagnosis, due to
either patient or health system delay.12 As a result of
the delay in TB diagnosis, patients incur both the di-
rect costs of diagnostic and treatment services for mis-
diagnoses, as well as the associated indirect costs due
to lost time.
The overall aim of this study was to estimate the
costs of TB diagnosis and treatment from the perspec-
tive of the patient in Lusaka, Zambia. Specifi cally, we
were interested in determining 1) the largest and most
frequently incurred cost items for patients before their
diagnosis and during the intensive phase of their TB
treatment, and 2) which socio-demographic, economic
or clinical factors were associated with higher patient
costs. Knowledge of these factors would inform the
S U M M A R Y

Page 2

Patient costs due to TB in Zambia 929
development of targeted interventions, such as modi-
fi cations to the delivery of TB services, to reduce the
economic burden on patients and society due to TB.
METHODS
The study was conducted at four primary health care
centres in Lusaka, Zambia: Chawama, Kanyama,
George and Chipata. These four clinics notify ~40%
of the total reported TB cases in Lusaka annually.4
Lusaka, in turn, notifi es approximately one third of
total new Zambian cases annually.4 Inclusion criteria
were patients with confi rmed active pulmonary or
extra-pulmonary TB who had been on treatment for
6–10 weeks and were aged ⩾18 years.
Users of government health facilities must pay a
registration fee to access consultation and diagnostic
fees for laboratory tests and radiography.13 TB pa-
tients in Zambia receive standardised treatment ac-
cording to previous treatment history following the
World Health Organization (WHO) recommendations
for DOTS-based programmes.14 The majority of pa-
tients in Lusaka are on a clinic-based directly observed
treatment (DOT) strategy where patients are observed
swallowing their pills at the clinic. At the time of this
study, a minority of patients were under community-
based DOT, where a designated treatment supporter
observes pill swallowing at home.15 Most urban clinics
in Lusaka did not yet offer community-based DOT;
however, they were in the process of implementing
this model. In the interim, patients on clinic-based
DOT in some cases switch to self-administered treat-
ment (SAT).
Data collection
Consecutive eligible patients were invited to partici-
pate in the study by clinic staff between 10 and 26 July
2006 at each study site. The enrolment rate was 91%
(103/113). Refusal rates were low primarily because
interviews were conducted while patients were wait-
ing to receive pills. Locally recruited research assistants
who had participated in a standardised training ses-
sion explained the study and obtained informed con-
sent from the patient. A standardised questionnaire
was used from a previous study.16 Interviews were
conducted in private rooms and assistants translated
questions into local languages (primarily Nyanja and
Bemba) according to patient preferences.
Ethics
This study was approved by the ethics committees of
the University of Zambia and the London School of
Hy giene & Tropical Medicine.
Data analysis
Data were double entered into Microsoft Access XP
(Microsoft, Redmond, WA, USA). Statistical analyses
were performed using STATA, version 9.2 (StataCorp,
College Station, TX, USA).
Cost estimation
Costs were estimated separately for the time when
patients were not yet diagnosed but were symptom-
atic (pre-diagnosis) and from the moment they were
diagnosed with TB, up to the time of interview (post-
diagnosis period). Cost category defi nitions are shown
in Table 1. Direct cost estimates were calculated as the
product of the patient-reported cost parameter (e.g.,
consultation fee) and frequency (e.g., number of clinic
visits). Indirect costs were estimated by tabulating the
total number of hours spent travelling, waiting and
receiving health services. Assuming that an average
unskilled labourer in Zambia would work 21 days a
month, 8 h per day, we used the monthly incomes re-
ported by patients to derive an hourly wage rate.17 All
costs are reported in 2006 US$ (1 US$ = 3855 Zam-
bian kwacha).
Sensitivity analyses
A set of sensitivity analyses were performed to ex-
plore the degree of uncertainty associated with the
indirect cost estimation according to standard meth-
ods.18 We varied the hourly wage rate and assump-
tions about the number of hours worked per month
in 10% increments up to 50% more or less than the
Table 1 Defi nitions used in the study
Cost categories, termsDefi nition
Care-seeking factors
Patient delay Period between onset of patient
symptoms and fi rst encounter with
the health service
Any visit to any health service
(government health centre,
pharmacy or private clinic)
Health encounter
Timing of costs
Pre-diagnostic
(‘diagnosis seeking’)
Costs incurred during time between
self-reported fi rst health encounter
and laboratory or radiologically
confi rmed diagnosis
Costs incurred during intensive phase
of treatment. Includes any clinic
visit preceding the diagnosis
(including referrals before
treatment) and subsequent
medical follow-up visits
Post-diagnostic
(‘treatment seeking’)*
Categories of costs
DirectOut-of-pocket expenditures for TB
services as well as those incurred to
access the service. Per patient costs
were categorised as travel costs,
registration and paperwork fees,
consultation fees, blood test fees,
medication fees, X-ray fees, food
costs and other costs
Patients’ lost income due to time to
receive care. Includes travel time
for return trips to clinics/hospital,
waiting time and time for
consultation with a physician,
nurse or treatment supporter
Indirect
* We report only the actual costs incurred by patients up to and including
2 months of treatment, with no extrapolation to the entire 6–8 month treat-
ment period.

Page 3

930
The International Journal of Tuberculosis and Lung Disease
base case value (average reported individual income
of $65.79 per month before diagnosis), while holding
all other values constant.
Affordability assessment
To assess affordability, direct costs incurred by pa-
tients were converted to a percentage of their actual
reported household incomes before TB diagnosis. As
suggested by Russell, we chose a direct expenditure
threshold of ⩾10% of household monthly income as
representing a signifi cant fi nancial impact.19
Statistical analyses
The normality of cost data distribution was assessed
using Kolmogorov-Smirnov tests. Costs were non-
normally distributed. The statistical signifi cance of
differences between patient sub-groups was assessed
using χ2 tests for categorical variables and non-
parametric (Wilcoxon rank-sum) tests for quantita-
tive variables. We examined the relationship between
costs and socio-demographic, economic, clinical and
care-seeking factors using log-transformed linear step-
wise regression analysis and a generalised linear model
with a log link.
RESULTS
One hundred and three patients were enrolled, com-
pleted interviews and were included in the costing
analysis. As summarised in Table 2, 57% were males
and the mean age was 32 years (range 18–53). Forty-
seven patients (62% males) originally assigned to a
supervised strategy (DOT) indicated that they had
switched to a self-administered strategy (SAT), leav-
ing a total of 56 patients on DOT. Of the 52 with
smear-negative or extra-pulmonary TB, 49 (94%)
went on SAT, compared to 6/59 (11%) with smear-
positive TB.
As seen in Table 3, the median total cost for each
patient amounted to $24.78 (interquartile range
[IQR] 13.56–40.30), of which indirect costs com-
prised 62% and direct costs 34%. Our estimate of the
total indirect costs incurred by patients was not sensi-
tive to changes in the assumption of wage rate used
nor the number of days worked. In the pre-diagnosis
period, direct costs comprised 66% of patients’ costs
—mostly for government health insurance user fees,
X-ray fees and public transport. In the post-diagnostic
period, the travel and time costs associated with clinic
visits for DOT or pill collection comprised 95% of
patients’ costs.
Affordability
The total direct costs, expressed as a proportion of
median individual income (Table 4), were 92% higher
for women than for men (P < 0.001). This is largely
a refl ection of the lower wages earned by women (the
median reported monthly individual income for fe-
males was $5.10 compared to $77.82 for males).
Variation in patient costs
As shown in Table 5, patient delay in seeking care
(P < 0.01), treatment supervision strategy (P <
0.0001) and smear status (P < 0.001) were associated
with total patient costs in univariate analyses. Patient
costs were not signifi cantly associated with age, male
Table 2 Demographic, socio-economic and clinical
characteristics of patients studied
Variable*
Female
(n = 44)
n (%)†
Male
(n = 59)
n (%)†
Total
(n = 103)
n (%)†
Age, years
18–24
25–34
35–44
⩾45
Household size, persons
1–4
5–6
7–10
>10
Education
None
Primary school
Some secondary school
(grade 9)
Finished secondary school
(grade 12)
College/university
Current employment status
Unemployed
Employed
7 (16)
22 (50)
13 (30)
2 (5)
7 (12)
28 (48)
18 (31)
6 (10)
14 (14)
50 (49)
31 (30)
8 (8)
11 (25)
22 (50)
7 (16)
4 (9)
24 (41)
21 (36)
10 (17)
4 (7)
35 (34)
43 (42)
17 (17)
8 (8)
2 (5)
27 (62)
3 (5)
23 (39)
5 (9)
50 (49)
13 (30) 27 (46)40 (39)
1 (2)
1 (2)
1 (2)
5 (9)
2 (2)
6 (6)
36 (82)
8 (18)
30 (51)
29 (49)
66 (64)
37 (36)
Income (n = 102)
Monthly income before TB, US$
<$5
$5–$75
>$75
Current monthly income, US$
<$5
$5–$75
>$75
Household income (n = 93)
Monthly income before TB
<$75
$75–$150
>$150
Current monthly income, US$
<$75
$75–$150
>$150
Form of TB
Pulmonary smear-positive
Pulmonary smear-negative
Extra-pulmonary
History of treatment
New case or never treated
Retreatment or previously treated
Treatment supervision strategy
Directly observed treatment
Self-administered treatment
22 (51)
17 (40)
4 (9)
9 (15)
20 (34)
30 (59)
31 (30)
37 (36)
34 (33)

60 (59)
26 (25)
16 (16)
35 (81)
7 (16)
1 (2)
25 (42)
19 (32)
15 (25)
16 (44)
15 (42)
5 (14)
16 (28)
19 (33)
22 (39)
32 (34)
34 (37)
27 (29)
22 (59)
9 (24)
6 (16)

26 (59)
12 (27)
6 (14)

41 (93)
3 (7)

26 (59)
18 (41)
27 (47)
17 (30)
13 (23)
49 (52)
26 (28)
19 (20)
33 (56)
16 (27)
10 (17)
59 (57)
28 (27)
16 (15)
49 (83)
10 (15)
90 (87)
13 (13)
30 (51)
29 (49)
56 (54)
47 (46)
* For each descriptive variable, n = 103 patients unless otherwise indicated.
† Totals may not amount to 100% due to rounding.
TB = tuberculosis.

Page 4

Patient costs due to TB in Zambia 931
sex, household size, employment status, individual
patient incomes, the number of health encounters be-
fore diagnosis, hospitalisation or form of TB. In multi-
variate linear regression, being male, longer patient
delays and clinic-based DOT were associated with
higher costs, and together explained at least 25% of
individual patient costs (Table 6).
Variation in patient characteristics and costs
by treatment strategy
An unexpected fi nding was that total costs incurred
by patients on clinic-based DOT were almost four
times greater compared to those who had switched to
SAT (P < 0.001). Although not our primary objective,
we wanted to explore the reasons for the variation in
Table 3 Direct and indirect costs incurred by patients during the pre- and post-diagnostic
periods (all costs in 2006 US$)
Timing and type of cost
Patients reporting expenditure
All patients
(n = 103)
median (IQR)
n (%)Median (IQR)
Pre-diagnosis
Direct
Medical
Government health insurance user fees 69 (67)
Consultation fees
Chest radiograph
Non-TB medication
Hospitalisation
Other
Total medical*
Non-medical
Transport
Food
Total non-medical
Sub-total
Indirect
Value of time for return trips to clinic
Value of time spent in hospital
Sub-total
Total pre-diagnosis
1.43 (1.30–1.43)
2.59 (2.59–3.89)
5.19 (2.59–5.19)
11.02 (6.49–15.56)
9.99 (2.72–13.80)
0.26 (0.26–1.55)
4.02 (1.42–6.74)
1.30 (0–1.42)
0 (0–0)
2.59 (0–5.19)
0 (0–0)
0 (0–0)
0 (0–0)
3.89 (1.43–6.61)
13 (13)
58 (55)
6 (6)
6 (6)
9 (9)
87 (84)
53 (51)
5 (5)
55 (53)
93 (90)
2.08 (1.04–5.19)
2.59 (0.79–5.19)
2.59 (1.04–5.19)
6.61 (3.11–9.47)
0.30 (0–2.59)
0 (0–0)
0.52 (0–2.59)
5.45 (2.59–9.34)
91 (88)
8 (8)
93 (90)† 8.44 (5.00–15.33)
2.35 (1.17–3.52)
14.10 (10.97–29.76) 0 (0–0)
2.35 (0.78–3.13)
2.35 (0.78–3.92)
8.31 (4.68–15.08)
Post-diagnosis
Direct
Transport for pill collection visits
Clinic-based DOT
SAT
Transport for follow-up visits
Total direct cost
Indirect
Value of time for pill collection visits
Clinic-based DOT
SAT
Value of time for follow-up visits
Total indirect cost
Total treatment-related
17 (17)
5 (5)
12 (12)
11 (11)
24 (23)
8.30 (2.85–20.75)
22.98 (17.79–25.20) 0 (0–0)
3.34 (2.30–14.32)
2.59 (1.43–2.85)
4.39 (2.63–19.07)
0 (0–0)
0 (0–2.15)
0 (0–0)
0 (0–0)
103 (100) 7.83 (4.70–23.50)
56 (54)19.97 (15.66–32.63) 19.98 (15.66–32.63)
47 (46) 4.70 (3.13–7.83)
43 (42) 1.57 (0.87–1.57)
103 (100) 9.33 (4.70–23.50)
7.83 (4.70–23.50)
4.70 (3.13–7.83)
0 (0–1.17)
9.33 (4.70–23.50)
11.75 (5.29–27.41)
Total patient costs
Total direct
Total indirect
Total
100 (97)
103 (100) 15.27 (7.96–28.59)
7.91 (4.02–12.45) 7.00 (3.11–11.93)
15.27 (7.96–28.59)
24.78 (13.56–40.30)
* Totals may not always be equivalent to sum of component costs up due to rounding.
† Although all patients (N = 103) spent time on return clinic visits to access their diagnosis, 10 patients reported total
time costs less than one hour. This was converted to a monetary value of zero dollars.
IQR = interquartile range (25th to 75th percentile); TB = tuberculosis; DOT = directly observed treatment; SAT = self-
administered treatment.
Table 4 Median out-of-pocket expenditures by sex as a percentage of median monthly
incomes earned by patient households (all costs in 2006 US$)
Timing of cost
Female (n = 44)Male (n = 59)
Median (IQR) % of MMIMedian (IQR)% of MMI
Pre-diagnosis
Post-diagnosis
Total direct
5.77 (2.66–9.14)
0 (0–0)
6.73 (2.92–11.67)
113
—
132
5.44 (2.07–9.34)
0 (0–2.59)
7.91 (3.63–15.56)
7.0
—
10.1
IQR = interquartile range (25th to 75th percentile); MMI = household median monthly income.

Page 5

932
The International Journal of Tuberculosis and Lung Disease
Table 5 Factors associated with increased total patient costs: univariate linear regression
analysis with and without log-transformed costs (all costs in 2006 US$)
Independent variables
Median
US$
Linear regression
Linear regression with
log-transformed cost
Coeffi cient
US$
P value
Coeffi cient
US$*
P value
Demographic
Sex
Male
Female
Age, years*
<34
⩾34
Socio-economic
Household size, number of
persons†
⩽4
>4
Education
Primary school or no education
Some secondary school
or higher
Current employment status
Employed
Unemployed
Monthly income before TB, US$†
<$50
⩾$50
Clinical
Each additional month of patient
delay in care seeking‡
Each additional health encounter
before diagnosis‡
Hospitalisation
Yes
No
Form of TB
Pulmonary
Extra-pulmonary
Smear
Smear-positive
Smear-negative
Treatment strategy
DOT
SAT
26.73
22.99
7.730.181.25 0.11
25.13
24.46
4.85 0.341.12 0.43
24.25
24.77
2.150.670.98 0.89
24.26 0.31 0.950.88 0.36
24.77
22.99
27.00
0.070.99 1.15 0.32
25.41
22.70
2.610.611.04 0.78
— 1.31
<0.0011.02 0.01
— 1.83 0.191.05 0.23
40.30
23.23
32.24
<0.0011.880.63
27.38
17.34
0.250.97 1.330.13
33.00
17.35
9.99
<0.0011.54
<0.001
32.71
16.35
9.67
<0.0011.55
<0.0001
*Adjusted difference was retransformed back to 2006 US$. Results of generalised linear model with a log link did not
alter signifi cance and therefore is not reported here.
†Dichotomisation based on median values.
‡Denotes continuous variable.
TB = tuberculosis; DOT = directly observed treatment; SAT = self-administered treatment.
Table 6 Factors explaining variation in total patient costs: multivariate stepwise linear
regression with and without log-transformed costs (all costs in 2006 US$)
Independent
variables*
Additive model linear regression
Multiplicative model linear regression
with log-transformed cost
Coeffi cient
US$
P value
R2
Coeffi cient
US$†
P value
R2
Male sex‡
Patient delay in
seeking diagnosis§
Clinic-based DOT‡
9.420.028 1.280.053
1.49
16.62
<0.001
<0.001
0.3981.70
1.02
<0.0001
<0.0001
0.254
* Hospitalisation and smear status were no longer independent predictors of the variation in patient costs when ad-
justed for the effect of male sex and patient delay.
† Adjusted difference was retransformed back to 2006 US$.
‡ Denotes categorical variable.
§ Denotes continuous variable.
DOT = directly observed treatment.

Page 6

Patient costs due to TB in Zambia 933
costs between these two patient groups: clinic-based
DOT (95% smear-positive) and self-administered (87%
smear-negative). Pre-diagnostic costs were somewhat
higher for the SAT group ($17 vs. $9, P = 0.10). Post-
diagnostic costs, however, were signifi cantly higher for
the clinic-based DOT group ($25 vs. $8, P < 0.0001).
The difference in cost between these groups was ac-
counted for by the time cost associated with DOT vis-
its, and not in other components, such as the cost of
drugs, hospitalisation, chest radiography, blood tests
or consultation fees. For patients on DOT, the median
time per DOT visit was 90 min (IQR 60–128), includ-
ing both travel time (60 min [IQR 0–120]) and wait-
ing time (15 min [IQR 10–30]). For patients not on
DOT, the median time per pill collection visit was
105 min (IQR 65–150), with average travel and wait-
ing times of respectively 60 min (IQR 0–120) and
30 min (IQR 30–60). The fi nding that total costs for
DOT patients were higher than those for SAT patients
is therefore directly attributable to the frequency of
visits rather than per visit costs.
DISCUSSION
TB diagnosis and treatment posed a signifi cant eco-
nomic burden on patients in terms of both cost and
affordability. The three largest predictors of patient
costs were treatment supervision strategy (clinic-based
DOT), patient delay in seeking care and male sex.
The most striking fi nding is the substantial eco-
nomic burden associated with clinic-based DOT. Al-
though it was the form of disease that dictated the
choice of treatment supervision, we observed that the
increased costs experienced by these smear-positive
patients on clinic-based DOT were accounted for by
the time required for treatment supervision and not
due to differences in other component costs, pre- or
post-diagnosis. We are confi dent, therefore, that dif-
ferences in cost were attributable to the supervision
strategy and not the actual disease form.
An important policy question emerging from these
results is whether it is justifi ed to require patients to
spend on average >1 h per day for treatment supervi-
sion visits, particularly when patients may have ex-
treme diffi culty affording daily trips to the clinic. A
community-based DOT strategy could improve pa-
tient adherence to treatment while reducing patient
costs. Community-based care for TB has already been
found to be cost-effective in a number of African
settings.15,20–25
Longer patient delays between fi rst experience of
symptoms and fi rst health service encounter were also
associated with higher costs. A possible explanation
may be that those who delay in presenting to the health
centre are more ill than those who present for earlier
consultation, and therefore incur higher related treat-
ment costs. An alternative explanation may be that
those who delay in seeking care are at a further dis-
tance from the clinic, and therefore incur higher re-
lated transport and time costs. Given that longer pa-
tient delays were predictive of higher patient costs, it
follows that an intervention that enables diagnosis of
TB closer to where patients live, such as community-
based enhanced case fi nding, could reduce the eco-
nomic burden on patients before their diagnosis. We
have also observed that the existence of user fees
means that although in theory TB diagnostic services
are provided free of charge, in practice it is impossible
for an individual to receive these ‘free’ services with-
out fi rst paying to access their diagnosis. Under the
current DOTS strategy in Zambia, TB patients are de-
tected through passive case fi nding. Enhancing TB
case fi nding by promoting awareness of the symptoms
of TB in the community, encouraging individuals to
present early to health centres and providing quality-
assured sputum collection points at strategic locations
in the neighbourhood is currently being tried in these
communities in the Zambia and South Africa Tuber-
culosis and AIDS reduction (ZAMSTAR) trial.26,27
This may lead to lower patient diagnostic costs as
well as more rapid diagnosis of TB.
We also found that men spent more money in ac-
cessing care than women; however, TB diagnosis and
treatment was nonetheless more affordable for males
when taken in the context of their ability to pay. Were
future research to demonstrate that direct costs dif-
ferentially selected against the care-seeking behaviours
of women, a targeted intervention may be warranted
to reduce transport costs and user fees as a barrier to
TB diagnosis and treatment.
Our study had four major limitations. First, a large
proportion of patients (98%) did not know their HIV
infection status. Obstacles to knowing the HIV status
of patients include health system factors (availability of
voluntary counselling and testing [VCT], cross-linking
databases between clinics), patient factors (percep-
tion of risk, ability to access VCT) and societal factors
(stigma surrounding the disease). However, given that
we were assessing costs due to diagnosis and care
seeking (and not the cost of morbidity or mortality),
we do not believe that adjusting for HIV comorbidity
would substantially alter our cost estimates or conclu-
sions. Costs reported by patients may have been biased
due to patients’ failure to recall certain expenditures
or the time spent in seeking care. We would expect in
most cases that this type of recall bias would lead to
an underestimation of patient costs. Given the suffi -
ciently detailed assessment of unit costs, we believe
our cost estimates represent a relatively reliable distri-
bution of costs between patients. Finally, although we
cannot exclude the possibility of omitted variable bias,
multi-collinearity or interaction between variables that
was not controlled for in our multiple linear regres-
sion analysis, our regression diagnostic tests suggested
that our model predicted the best linear unbiased es-
timate based on our data.28

Page 7

934
The International Journal of Tuberculosis and Lung Disease
One strength of this survey is the ability to com-
pare our data with previous costing estimates from the
same geographical location. A study of patient costs
due to TB diagnosis was conducted at the University
Teaching Hospital (UTH) in Lusaka in 1995.29 At the
time of this study, TB diagnostic facilities were central-
ised at the UTH Chest Clinic, and out-patients there-
fore required referral from a peripheral residential
clinic to be diagnosed. Due in part to the organisation
of TB services, this study found that patients incurred
a mean total cost representing 76.2% of their average
monthly income ($47, 1995 US$) before their diagno-
sis.29 Our data suggest that current pre-diagnosis costs
now represent only 11.5% of average monthly income,
strongly suggesting that TB diagnosis in Lusaka is now
more affordable from the patient perspective. Given
that the average monthly income reported in 1995
was $47 and the monthly income reported in this
study was $66, we believe this represents a real differ-
ence in patient costs. The most plausible explanation
is that decentralisation of TB diagnostic services be-
tween 1995 and 2006 has increased the geographical
accessibility of TB diagnostic services.30,31
CONCLUSION
TB diagnosis and treatment posed a signifi cant eco-
nomic burden on patients in terms of both cost and
affordability. Despite the strengths of the existing TB
control strategy promoted by the WHO, we have ob-
served that clinic-based DOT may contribute dispro-
portionately to the costs incurred by patients. To reduce
patient costs, we recommend the creation or strength-
ening of community-based TB treatment super vision
in settings with high TB incidence.
Acknowledgements
The present study was supported by a grant from Canadian Insti-
tutes of Health Research. The authors thank the staff of Chawama,
Kanyama, George and Chipata health care centres, and N Sibande,
A Nota, B Shipinbi, B Tembo and M Lintini for their participation
in this project.
References
1 Corbett E L, Marston B, Churchyard G J, De Cock K M. Tu-
berculosis in sub-Saharan Africa: opportunities, challenges and
change in the era of antiretroviral treatment. Lancet 2006; 367:
926–937.
2 World Health Organization. Global tuberculosis control—
surveillance, planning, fi nancing, 2006. WHO/HTM/STB/2006.
37. Geneva, Switzerland: WHO, 2006.
3 Joint United Nations Programme on HIV/AIDS. Report on the
global HIV/AIDS epidemic. Geneva, Switzerland: UNAIDS,
2006. http://data.unaids.org/pub/GlobalReport/2006/2006_GR_
ANN2_en.pdf Accessed May 2008.
4 Government of Zambia. National TB reporting database. Lu-
saka, Zambia: Government of Zambia, 2006.
5 Mwaba P, Maboshe M, Chintu C, et al. The relentless spread of
tuberculosis in Zambia—trends over the past 37 years (1964–
2000). S Afr Med J 2003; 93: 149–152.
6 Godfrey-Faussett P, Ayles H. Can we control tuberculosis in
high HIV prevalence settings? Tuberculosis 2003; 83: 68–76.
7 Corbett E L, Watt C J, Walker N, et al. The growing burden of
tuberculosis: global trends and interactions with the HIV epi-
demic. Arch Internal Med 2003; 163: 1009–1021.
8 Floyd K, Wilkinson D. Tuberculosis in the HIV/AIDS era: inter-
actions, impacts and solutions. AIDS Analysis Africa 1997; 7:
5–7.
9 Ogden J. The resurgence of tuberculosis in the tropics. Improv-
ing tuberculosis control—social science inputs. Trans Roy Soc
Trop Med Hygiene 2000; 94: 135–140.
10 Floyd K, Blanc L, Raviglione M, Lee J W. Resources required
for global tuberculosis control. Science 2002; 295: 2040–
2041.
11 McIntyre D, Thiede M, Dahlgren G, Whitehead M. What are
the economic consequences for households of illness and of pay-
ing for health care in low- and middle-income country contexts?
Soc Sci Med 2006; 62: 858–865.
12 Walker D, McNerney R, Mwembo M K, Foster S, Tihon V,
Godfrey-Faussett P. An incremental cost-effectiveness analysis
of the fi rst, second and third sputum examination in the diag-
nosis of pulmonary tuberculosis. Int J Tuberc Lung Dis 2000;
4: 246–251.
13 Lusaka District Health Management Board. Action plan and
budget for the years: 2006–2008. Lusaka, Zambia: Lusaka Dis-
trict Health Management Board, 2006.
14 World Health Organization. An expanded DOTS framework for
effective tuberculosis control. Geneva, Switzerland. WHO, 2002.
15 Maher D. Community contribution to TB care: policy and prac-
tice. Geneva, Switzerland. WHO, 2003.
16 Schwartzman K, Oxlade O, Barr R G, et al. Domestic returns
from investment in the control of tuberculosis in other coun-
tries. N Engl J Med 2005; 353: 1008–1020.
17 Ministry of Finance and National Planning. Economic report
2001. Lusaka, Zambia: Ministry of Finance and National Plan-
ning, 2002.
18 Fox-Rushby J, Cairns J. Economic evaluation. 1st ed. Oxford,
UK: Oxford University Press, 2005.
19 Russell S. The economic burden of illness for households in de-
veloping countries: a review of studies focusing on malaria, tu-
berculosis and human immunodefi ciency virus/acquired immuno-
defi ciency syndrome. Am J Trop Med Hygiene 2004; 7 (Suppl 2):
S147–S155.
20 Floyd K. Costs and effectiveness—the impact of economic stud-
ies on TB control. Tuberculosis 2003; 83: 187–200.
21 Floyd K, Skeva J, Nyirenda T, Gausi F, Salaniponi F. Cost and
cost-effectiveness of increased community and primary care fa-
cility involvement in tuberculosis care in Lilongwe District, Ma-
lawi. Int J Tuberc Lung Dis 2003; 7 (Suppl 1): S29–S37.
22 Fryatt R J. Review of published cost-effectiveness studies on tu-
berculosis treatment programmes. Int J Tuberc Lung Dis 1997;
1: 101–109.
23 Nganda B, Wang’ombe J, Floyd K, Kangangi J. Cost and cost-
effectiveness of increased community and primary care facility
involvement in tuberculosis care in Machakos District, Kenya.
Int J Tuberc Lung Dis 2003; 7 (Suppl 1): S14–S20.
24 Okello D, Floyd K, Adatu F, Odeke R, Gargioni G. Cost and
cost-effectiveness of community-based care for tuberculosis pa-
tients in rural Uganda. Int J Tuberc Lung Dis 2003; 7 (Suppl 1):
S72–S79.
25 Sinanovic E, Floyd K, Dudley L, Azevedo V, Grant R, Maher D.
Cost and cost-effectiveness of community-based care for tuber-
culosis in Cape Town, South Africa. Int J Tuberc Lung Dis 2003;
7 (Suppl 1): S56–S62.
26 Golub J E, Mohan C I, Comstock G W, Chaisson R E. Active
case fi nding of tuberculosis: historical perspective and future
prospects. Int J Tuberc Lung Dis 2005; 9: 1183–1203.
27 CREATE Consortium, 2006. Baltimore, MD, USA: Create, 2006.
www.tbhiv-create.org/ Accessed May 2008.

Page 8

Patient costs due to TB in Zambia 935
28 Koop G. Analysis of economic data. 2nd ed. Leicester, UK: John
Wiley & Sons, 2005.
29 Needham D M, Godfrey-Faussett P, Foster S D. Barriers to tu-
berculosis control in urban Zambia: the economic impact and
burden on patients prior to diagnosis. Int J Tuberc Lung Dis
1998; 2: 811–817.
30 Hjortsberg C A, Mwikisa C N. Cost of access to health services
in Zambia. Health Policy Plan 2002; 17: 71–77.
31 Blas E, Limbambala M. User-payment, decentralization and
health service utilization in Zambia. Health Policy Plan 2001;
16 (Suppl 2): 19–28.
R É S U M É
CONTEXTE : Centres de santé primaires urbains de Lu-
saka, Zambie.
OBJECTIFS : 1) Estimer les coûts-patient du diagnostic
de la tuberculose (TB) et de son traitement et 2) identi-
fi er les éléments déterminant les coûts du patient.
MÉTHODES : Enquête transversale conduite grâce à un
questionnaire standardisé chez 103 patients adultes at-
teints de TB qui avaient été sous traitement pendant 1 à
3 mois. Les coûts directs et indirects ont été estimés, con-
vertis en US$ et calculés pour deux périodes : période
« pré-diagnostic/recherche de soins » et période « post-
diagnostic/traitement ». Les facteurs déterminants des
coûts-patient ont été analysés par régression linéaire
multiple.
RÉSULTATS : Le coût total médian par patient pour le
diagnostic et 2 mois de traitement a été de 24,78 US$
(IC95% 13,6–40,3) par patient, ce qui équivaut à 47,8%
du revenu mensuel médian des patients. Le sexe, le délai-
patient dans la recherche de soins et la méthode de super-
vision du traitement ont été des facteurs prédictifs sig-
nifi catifs du coût-patient total. Les coûts totaux directs
en proportion des revenus ont été plus élevés pour les
femmes que pour les hommes (P < 0,001). Les coûts de
traitement encourus par les patients sous une stratégie
de traitement directement observé basé sur la clinique ont
été plus de trois fois supérieurs à ceux encourus par les
patients lors d’une stratégie de traitement auto-administré
(P < 0,001).
CONCLUSION : La supervision du traitement basée sur
une clinique impose aux patients un fardeau économi-
que signifi catif. La création ou le renforcement de pro-
grammes de supervision basés sur la collectivité aurait le
plus grand impact potentiel sur la diminution des coûts-
patient liés à la TB.
R E S U M E N
MARCO DE REFERENCIA : Centros primarios urbanos
de atención de salud en Lusaka, Zambia.
OBJETIVOS : 1) Calcular los costos incurridos por el pa-
ciente en el diagnóstico y tratamiento de la tuberculosis
(TB) y 2) defi nir los factores determinantes de los costos
para el paciente.
MÉTODOS : Fue este un estudio transversal de 103 pa-
cientes adultos con TB, quienes recibieron tratamiento
durante 1 a 3 meses y respondieron a un cuestionario
estandarizado. Se calcularon los costos directos e indi-
rectos, que se convirtieron en dólares estadounidenses y
se categorizaron en dos períodos : la etapa de prediag-
nóstico y búsqueda de atención y la etapa posterior al
diagnóstico y de tratamiento. Se analizaron los factores
determinantes de los costos para el paciente mediante un
análisis de regresión linear multifactorial.
RESULTADOS : La mediana de los costos totales incurri-
dos por el paciente en diagnóstico y 2 meses de trata-
miento fue de 24,78 dólares (IQR 13,6–40,3) por pa-
ciente (lo cual equivale al 47,8% del ingreso mensual de
los pacientes). El sexo y el retraso del paciente en buscar
atención constituyeron factores signifi cativos de predic-
ción del costo total incurrido por él. Los costos directos
totales, expresados como proporción del ingreso, fueron
superiores para las mujeres que para los hombres (P <
0,001). Los costos de tratamiento asumidos por el pa-
ciente con el tratamiento directamente observado en el
consultorio fueron tres veces superiores a los costos con
el sistema de tratamiento autoadministrado (P < 0,001).
CONCLUSIÓN : La supervisión del tratamiento en el con-
sultorio implica una carga económica considerable para
los pacientes. La creación o fortalecimiento de programas
de supervisión del tratamiento en la comunidad podría
dar lugar a una reducción importante de los costos rela-
tivos al tratamiento antituberculoso para los pacientes.