06 Jacobs (jl/d) 20/10/99 8:30 am Page 354

HEALTH POLICY AND PLANNING; 14(4): 354362 Oxford University Press 1999

Feasibility of hospital-based blood banking: a Tanzanian case

study
BART JACOBS AND ALEC MERCER
School of Social Science and International Development, University of Wales, Swansea, Wales, UK

The demand for blood transfusion is high in sub-Saharan Africa because of the high prevalence of anaemia
and pregnancy related complications, but the practice is estimated to account for 10% of HIV infections in
some regions. The main response to this problem by the international donor community is to establish
vertically implemented blood transfusion services producing suitable (safe) blood at a cost of US$2540 per
unit. However, the economic sustainability of such interventions is questionable and it is argued here that
hospital-based blood transfusion services operating at a basic adequate level are sufficient for low-income
African countries. The results of a project aimed at improving such services in Tanzania are presented. The
main findings are: (1) the cost per suitable blood unit produced was US$12.4; (2) at an HIV test sensitivity of
93.5% during the study period, discounted financial benefits of the interventions exceeded costs by a factor
of between 17.2 and 37.1; (3) the cost per undiscounted year of life saved by use of these interventions was
US$2.72.8; and (4) safe blood transfusion practices can be assured at an annual cost of US$0.07 per capita.
Recommendations are made to ensure safe blood transfusion practices at hospital-based blood banks in
Tanzania.

Introduction Furthermore, the possibility of providing a high level BTS is

Blood transfusion accounts for 10% of infections with human
immuno-deficiency virus (HIV) in some populations such as
Tanzania (Piot et al. 1990). A major factor contributing to the
demand for blood transfusion is its use as the last resort treat-
ment for anaemia, which is highly prevalent in the country. A
study of adults in three regions and in the capital city Dar Es
Salaam indicated that 3080% of men and 2680% of women
were anaemic, depending on area of residence and socio-
economic status (Kitange et al. 1993). In a study of the coastal
region of the country, 74% of 325 children screened were
anaemic, while 2.5% were severely anaemic (Premji et al.
1995). The main cause of anaemia is Plasmodium falciparum
(malaria) infection, followed by iron and folate deficiency,
hookworm infection and malnutrition (Luby et al. 1995). A
study in Mwanza Region attributed anaemia to malaria in
90% of children aged less than five years and in 60% of preg-
nant women (Gumodoka et al. 1993). These two groups are
particularly vulnerable to malaria infection because of lower
immunity. Consequently they are potentially the most at risk
for anaemia and hence HIV-infection acquired through blood
transfusion (Schapira 1989).
The most common response by the international donor
community to the problem of HIV transmission through
transfusion of blood is to establish vertically implemented
blood transfusion services (BTS) which can produce suitable
(safe) blood units. However, the cost per suitable unit is high,
estimated at US$25 in Uganda and US$40 in Rwanda
(Watson-Williams and Kataaha 1990; van Dam et al. 1992).
limited in many African countries by low per capita health
expenditure, limited coverage, inadequate capacity of health
facilities to meet basic needs and insufficient relevant skills
among personnel (Mburu 1994). However, for a low-income
African country such as Tanzania, a BTS operating at a stage
which the World Health Organization classifies as basic ade-
quate1 should be sufficient (WHO 1992).
The European Commissions AIDS Task Force recognizes
four patterns of BTS: pattern I, a centrally coordinated
national blood programme; pattern II, hospital-based blood
banks; pattern III, settings where relatives give blood in an
emergency; and pattern IV, settings without organized trans-
fusion practices (Beal et al. 1992).
Pattern I can be decentralized to regional level and has
advantages over the other patterns: it allows better screening
of all donated blood units for HIV and other blood borne
pathogens; automation can be introduced to increase produc-
tivity of staff; the willingness to donate blood is not affected
by the quality of services at the hospital which is likely with
patterns IIIV; objectives are relatively quickly achieved; and
resources can be better controlled and concentrated.
However, the problem with a pattern I BTS is that it consti-
tutes a vertical health programme.
Certain characteristics of vertical programmes undermine
their sustainability, such as fragmentation of institutional
strength and non-integration into the ongoing administrative
structure (Bossert 1990). However, the main constraint is
06 Jacobs (jl/d) 20/10/99 8:30 am Page 355
Hospital-based blood banking
economic, as the costs far exceed those of integrated pro-
grammes in which the cost of logistics, training and super-
vision can be shared. The high price of blood units in relation
to national public health expenditure further inhibits sustain-
ability. The poorer the country and the lower the level of
health spending, the more the governments capacity for new
investments is restricted. Even when the national budget is
able to absorb the recurrent costs, it is most likely that
resources for other programmes would be depleted, resulting
in a redistribution of underfinancing (Prost and de Ferranti
1985).
For low-income African countries, pattern II hospital-based
blood banks may be the best solution on economic grounds,
although there are disadvantages: competing with other
departments for resources; potential problems with the
supply of consumables, such as HIV-test kits; difficulties in
establishing a voluntary blood donor panel because willing-
ness to donate is affected by the quality of services provided
by a hospital; and the fact that the in-charge of the blood bank
often has limited knowledge of BTS. The advantages are that
the programme is integrated into the existing health care
delivery system and that less capital costs will be required as
the majority of the equipment needed will already be present.
The implementation of interventions that will reduce trans-
fusion-related HIV-transmission has been described else-
where (Jacobs 1997). Here, an economic evaluation of the
feasibility of hospital-based blood banking is reported with
recommendations for important support measures.
Methods
Background
The project on which this study is based started in 1992 in
Mwanza region, Tanzania. The region is situated on the
southern shores of Lake Victoria and covers an area of 19 683
km2. Its population was estimated at 2 351 233 in 1995 (Min-
istry of Health 1995). A National Blood Transfusion Advis-
ory Committee was established in 1990 which formulated
comprehensive National Guidelines for Safe Blood Trans-
fusion (NGSBT), covering all aspects from collection to
transfusion (Ministry of Health 1990). Additionally, consen-
sus guidelines for the prescription of blood transfusion were
introduced in 1991 (Gumodoka et al. 1993). Screening of
donated blood for HIV antibodies was introduced nation-
wide by the end of 1989 (Lyamula et al. 1996). The sensitivity
of the screening test (ELISA) used in Mwanza region at the
start of the project was 93.5% (Velema et al. 1992). The HIV
incidence rate in rural Mwanza region was 1% for the period
199194 (Grosskurth et al. 1995).
Although the project took place in five hospitals, only the
results from Bugando Medical Centre (BMC), an urban
referral hospital with a semi-autonomous BTS, are discussed
here. Table 1 displays relevant data for this hospital.
The aim of the project was to implement the interventions
required to assure safe blood-transfusion practices, as formu-
lated by the Tanzanian Ministry of Health:
355
Table 1. Bugando Medical Centre
Type of hospital regional/referral
Number of beds 820
Blood transfusions/month1 250
HIV-prevalence among blood donors1 12%
Inpatients/year1 17 885
Bed occupancy rate 79%
Regional population 2 351 2332
BTS semi-autonomous;
employing 6 people
Voluntary blood donors less than 10% of all
donors
1 1991 data.
2 1995 estimates.
screening of all donated blood for HIV and syphilis;
increasing blood collection among voluntary donors;
use of autologous blood transfusion for eligible patients;
reducing the use of blood transfusion for life saving situ-
ations only;
improving the laboratory aspects of blood transfusion.
Blood collection procedure
According to the NGSBT only men and women that do not
belong to a population group at increased risk for HIV infec-
tion, who are aged 18 years or more, weighing at least 50 kg,
and with haemoglobin (Hb) levels of 13.5 g/dl and 12.5 g/dl,
respectively, are allowed to donate blood. With the approval
of the Ministry of Health the Hb thresholds were lowered to
12.5 g/dl and 11.5 g/dl for males and females, respectively.
Adhering to the recommended thresholds would have
resulted in too many potential blood donors being deferred,
due to the low Hb level among the general population. On the
other hand, collecting blood from individuals with an Hb
below the recommended thresholds increases the risk for
adverse blood-donation reactions such as hypovolaemia.
Therefore, smaller blood collection bags of 250 ml 10%
instead of 450 ml 10% are used for such individuals. Indi-
viduals weighing 5055 kg are also at increased risk for
adverse donation reactions and are, therefore, bled using the
smaller blood collection bags. A quarter of the people offer-
ing to donate blood have to be deferred because of their Hb
level and/or body weight.
The cost of single blood collection bags, whether 250 or 450
ml, is the same. For multiple blood collection bags, costs cor-
relate directly with the number of individual bags rather than
with the total volume. Multiple bags are used for producing
packed cells by depleting the plasma, after centrifugation or
48 hours sedimentation of the whole blood. Packed cells are
used to treat patients with severe anaemia that have a subse-
quent risk of circulatory overload when treated with whole
blood. A double blood collection bag contains one unit for
packed cells and one unit for derived plasma, while a triple
bag contains two units for packed cells and one unit for
derived plasma. Because of the higher cost of multiple blood
collection bags, they were only used among voluntary blood
donors as they are at significantly lower risk of transfusion
06 Jacobs (jl/d) 20/10/99 8:30 am Page 356

356 Bart Jacobs and Alec Mercer

transmittable infections and their units of blood are less likely Costbenefit analysis
to be discarded. The derived plasma was never used for
Prior to full implementation of the interventions at the end of
medical purposes.
1992, 12% of blood donors were HIV-positive and there were
3000 annual transfusions. If blood had not been tested, on a
All blood collected in the study period, 199394, was tested
crude calculation, approximately 360 transfusions would have
for HIV-antibodies the following morning by use of ELISA.
been HIV-infected. As a result of the intervention, the annual
Units found to be positive were destroyed immediately.
number of blood transfusions was reduced to 1200. If HIV
prevalence among donors had remained at 12%, a crude
Study methods calculation gives the number of HIV-infected transfusions as
144, in which case, reducing transfusions prevented about 216
Cost-minimization is based on the cost per suitable (safe)
HIV infections. However, overall HIV prevalence fell to 7%
blood unit. To determine the price of a suitable blood
because more blood came from voluntary donors. The crude
unit, the format for cost data reporting from Beal et al.
estimate of HIV-infected transfusions would be 84, so that
(1992) is used. Pricing suitable blood units also allows
reducing prevalence prevented 60 HIV infections.
comparison with centrally coordinated, national blood pro-
grammes.
In calculating the benefits of the intervention here, we con-
sider that screening prevented the transfusion of blood from
A cost-benefit analysis was conducted to consider benefits
donors who tested HIV-positive. However, we also take into
achieved at BMC by (1) reducing the number of unnecessary
account that the other interventions prevented the trans-
blood transfusions; (2) lowering the HIV-prevalence among
fusion of blood from those who tested falsely negative. Table
blood donors to reduce the number of units testing falsely
2 shows the number of potentially HIV-infected units (PHIU)
negative; and (3) avoiding homologous blood transfusion by
by donor HIV-prevalence (2, 7 and 12%) and test-sensitivity
use of autologous blood transfusion. The cost-benefit of
(93.5 and 99%), assuming an annual incidence rate of HIV
screening the blood for HIV prior to transfusion has been
infection of 1% among the blood donors, in line with the
demonstrated by Foster and Buv (1995) and will be calcu-
regional HIV-incidence during the study period. If test sensi-
lated separately here for comparison.
tivity was 93.5% and HIV prevalence was 7%, an estimated
4.9 per 1000 transfused units would have come from donors
The cost-benefit analysis uses the human capital approach,
who tested falsely negative. In addition, an estimated 1.5 per
although this is controversial. Alternatives, such as peoples
1000 units would have come from donors who tested negative
observed or stated preference are usually based on peoples
because they had no antibodies in the window period. Con-
willingness to pay. These are not applied in this study because
sequently, 7.7 (6.4 per 1000) potentially HIV-infected units
of the lack of information in developing countries.
would be transfused to 1200 recipients. In comparison, before
the intervention, 29.4 (9.8 per 1000) units were transfused to
For a cost-effectiveness analysis, the number of years of life
3000 recipients from donors who would test falsely negative:
saved was calculated using a life table. The most recent avail-
i.e. the interventions other than screening prevented trans-
able for Tanzania is for 1967 (Cond et al. 1980). Calculations
fusion of about 22 units of HIV-infected blood in 1 year.
were also done using a more recent life table for 1980 from
neighbouring Zambia for which the mortality level closely
Some transfused patients receive more than one blood unit,
matches that of Tanzania (United Nations 1995).
some less. For the costbenefit analysis, therefore, it is neces-
sary to convert prevention of infected units into HIV infec-
Results tions prevented (Table 3). A significant number of the patients
After the creation of a hospital blood transfusion committee
at BMC and following acceptance and modification of the Table 2. Risk of transfusing potentially HIV-infected blood units
consensus guidelines to BMC Guidelines for Safe Blood (false negatives and window period negatives): by test-sensitivity and
Transfusion, the average number of monthly transfusions blood donor prevalence with a 1% annual incidence
dropped from 250 in 1991 to 100 in 199394. The proportion
of donors who were voluntary increased to 23.8% (339/1423) HIV-prevalence PHIU1 (per 1000) by test sensitivity
in 199394, compared with less than 10% prior to the start of among blood donors
93.5% 99%
the project. HIV prevalence among voluntary donors was
1.8% (6/339) compared with 8.7% (94/1084) among relatives
donating blood. Due to the use of triple and quadruple blood 2% 2.9 1.7
bags and because of the lower HIV-prevalence among volun- 7% 6.4 2.2
12% 9.8 2.7
tary donors, this group accounted for 30.5% of all suitable
blood units produced (433/1421).
1 PHIU = potentially HIV-infected blood units. For example, at 7%

The cost per suitable blood unit produced in BMC was
US$12.4 (Annex 1). Capital costs were 24% of the total
annual expenditure in BMC and personnel accounted for
10.4%. The marginal cost per suitable blood unit was US$5.1
for relative donors and US$5.8 for voluntary donors.
HIV-prevalence and 93.5% sensitivity: Number with antibodies
testing falsely HIV-negative (20/0.935 20) = 4.9. Number without
antibodies testing HIV-negative in the window period = 1.5. (Calcu-
lated using the formula from Savarit et al. 1992: the risk of a blood
donor donating in the window period, RW = Incidence 3 56/365 =
0.01 3 0.15, i.e. 1.5 per 1000).
06 Jacobs (jl/d) 20/10/99 8:30 am Page 357

Hospital-based blood banking 357

Table 3. Transfusion-related HIV infections prevented at Bugando Medical Centre

Patient group HIV % of BT PHIU Mean units3 Infections Infections

prevalence recipients1 not issued2 per patient potentially actually4
rate prevented prevented

Children 04 years 4.3% 40.2 8.8 0.5 17.6 16.8

Children 514 years 1% 7.4 1.6 1 1.6 1.6
Pregnant women 14.3% 15.0 3.3 1.2 2.7 2.3
Anaemic adults 24% 32.6 7.2 1.2 6.0 4.6
Operated patients 12% 4.8 1.1 1.2 0.9 0.8
Total 12% 100 22.0 0.9 28.8 26.0

1 Source: Vos et al. 1993.

2 PHIU = potentially HIV infected units (e.g. children 04 years: 22 3 0.402 = 8.8).
3 One unit refers to one suitable blood unit, derived from either a small or large single bag, or from a multiple blood bag. For a substantial,

though not ascertained, number of anaemic children, one blood unit is shared by different compatible children due to the small volume of
blood they require which is often well below one blood unit.
4 Excludes transfusion recipients already HIV-infected (e.g. children 04 years: 17.6 3 0.957 = 16.8).

receiving a blood transfusion are already HIV-infected, so it is
necessary to subtract them from the total number of HIV
infections prevented. A 3-year sentinel surveillance among
antenatal clinic attenders in Mwanza found an overall HIV-
prevalence of 12% (Kigadye et al. 1993). However, another
study indicated that data obtained from this group underesti-
mated the population HIV-prevalence. It suggested that blood
donors who donate blood for a friend or relative are represen-
tative only if the prevalence rate is standardized for age and
sex (Borgdorff et al. 1993).
HIV-prevalence among female relative donors aged 1539
years was 14.3% (20/140), which is taken here as the HIV-
prevalence for pregnant women receiving transfusions. Chil-
dren aged under five years are assumed to have been
perinatally infected and the perinatal transmission rate for
HIV is 2040% in Uganda and Zaire (Preble 1990). A rate of
30% has been used here, so HIV prevalence among blood
transfusion recipients in this age group is estimated at 4.3%.
For children aged 5 years and over, prevalence of 1% is used
in line with the findings of Killewo et al. (1993) in neighbour-
ing Bukoba region. For operated patients, the overall HIV-
prevalence of 12% among adults in Mwanza is used (Barongo
et al. 1992). On the basis of the study by Borgdorff et al.
(1993) in Mwanza, an HIV-prevalence of 24% is used for
anaemic adults receiving transfusions.
Assuming the above HIV-prevalence rates, a total of 26 HIV
infections would be prevented (Table 3). However, this esti-
mate needs to be reduced because some transfused patients
die. Research in Kenya found a mortality rate of 13.5%
among transfused children and 9.1% among transfused
women (Lackritz et al. 1993; Zucker et al. 1994). Mortality
rates have yet to be ascertained at BMC, but conservatively
allowing for a slightly higher mortality rate of 15% among
blood transfusion recipients, the total number of HIV-infec-
tions prevented is reduced to 22.
The cost of preventing these 22 HIV infections is US$17 655
(Annex 1). The total cost incurred for an HIV-infected person
was estimated at US$24625316 in Tanzania in the year
198788 (Over et al. 1988). This converts to US$29786430 in
1992 US dollars2, and a total cost for 22 people of US$65
516141 460. The benefitcost ratio is, therefore, between
3.7:1 and 8.0:1, but as most benefits will occur in the future,
they should be discounted. Assuming seven years elapse
between infection and disease progression, and using a 5%
discount rate, the resulting discounted benefitcost ratio is
3.1:1 to 6.6:1.
Screening the blood for HIV-antibodies detected 100
seropositive blood donors, thereby avoiding an additional 119
HIV infections, not allowing for any use of multiple blood
bags.3 Assuming the same mortality rate among transfusion
recipients as above (15%), the number of HIV infections pre-
vented by use of HIV screening was 101. All interventions
combined, therefore, avoided 123 HIV infections and the
associated costs of US$366 294790 890, a benefitcost ratio
of 20.7:1 to 44.8:1. The discounted benefitcost ratios were
15.8:1 to 37.0:1.4
BMC provides blood for all hospitals in the town which had
an estimated population of 250 000 in 1988. Because the
whole population, excluding the referred rural patients, is
potentially at risk to receive a blood transfusion, the ad-
ditional annual costs of assuring safer blood-transfusion prac-
tices are therefore US$0.07 per capita (1992 US$).
Cost-effectiveness analysis
Table 4 shows the calculation for the number of years of life
saved by the interventions other than screening the blood for
HIV. The cost per year of life saved by use of these interven-
tions was US$15.315.8. When considering all interventions,
62396463 years of life were saved at a cost of US$2.72.8 per
year. Screening the blood for HIV is the most important of
the interventions for avoiding transfusion-related HIV-infec-
tion, as it contributed 82% of the total number of years of life
saved and accounted only for 8% of the total cost per suitable
unit.
06 Jacobs (jl/d) 20/10/99 8:30 am Page 358

358 Bart Jacobs and Alec Mercer

Table 4. Years of life added by interventions other than HIV-screening

Patient group Infections Mean age Sex ratio Years of Total Years of Total
prevented of patients M:F life saved years of life saved years of
excluding in years per person2 life saved2 per person3 life saved3
recipients M; F M; F M; F
who died

Children 04 years 14.3 1 50:50 53.2; 55 773.6 55.6; 57.4 808.0

Children 514 years 1.4 5 50:50 56; 56.5 78.8 55.6; 57.6 79.2
Anaemic adults 3.9 30; 25 40:60 36.7; 41.1 153.4 37; 42 156.0
Pregnant women 2.0 25 41.1 82.2 42 84.0
Operated patients 0.7 30; 25 25:75 36.7; 41.1 28.0 37; 42 28.5
Total 22 1116 1156

1 Source: Vos et al. 1993 (M = Male; F = Female).

2 Using the 1967 Tanzanian life table.
3 Using the 1980 Zambian life table.

Discussion
Hospital-based blood banking has the important advantage
of incurring lower costs than a centralized approach. The cost
of a suitable blood unit produced at BMC was US$12.4, con-
siderably less than the cost of a suitable blood unit under a
centralized approach in other African countries (US$2540).
Such a difference is particularly important as many African
countries have a very limited budget for health care.
The cost per year of life saved through additional measures
other than screening the blood for HIV was US$15.315.8.
This compares favourably with other health interventions
such as measles immunization, estimated to cost US$13.5 per
year of healthy life added in the Ivory Coast and US$16.2 in
Zambia (1992 US$) (Prost and Prescot 1984). However, for
the combination of HIV screening with other interventions,
the cost per year of life saved is reduced to US$2.72.8. Con-
sidering only the cost of screening at BMC, the cost per year
of life saved would be US$0.2.
These costs compare favourably with the findings of Foster
and Buv (1995) who calculated that screening blood for HIV
in neighbouring Zambia resulted in a cost per year of life
saved of US$1.3. Their higher costs per year of life saved
would be partly due to their assumption of 50% mortality
among blood recipients and also because two-thirds of the
blood was screened using more expensive rapid essays. Their
discounted benefitcost ratio of screening the blood for HIV
was 2.7:1 to 3.0:1, compared with 15.8:1 to 37.0:1 in this study,
which is due to the substantial difference in the cost per HIV
infection used for the calculations. The providers cost per
HIV-infection prevented in Zambia was US$110.6 compared
with US$29796432 (1992 US$) in Tanzania. However, the
latter included social costs.
The benefits of the interventions other than screening the
blood for HIV depend mainly on the sensitivity of the HIV-
screening test, which was 93.5% during the study period.
However, this may now have risen to about 99% due to the
development of better tests, improved supervision and an
adequate quality control scheme. If a test sensitivity of 99%
was achieved during the study period, with no other inter-
ventions, the number of potentially HIV-infected units
(PHIU) being transfused would have been 8.1 instead of 29.4
(Table 2). The effect of the other two measures, reducing the
number of avoidable blood transfusions and lowering the
HIV-prevalence among blood donors, would be to reduce the
number of PHIU from 8.1 per year to 2.6. Only 132137 years
of life would be saved at a cost of US$129134 for each year.
However, as a result of lowering the HIV prevalence among
blood donors, there were substantial savings since fewer units
had to be discarded. There were direct savings of between
US$918010 440 (1800 fewer units were used at a marginal
cost per unit of US$5.1 for relative donors and US$5.8 for
voluntary donors). However, risks involved in blood trans-
fusion are not confined to transmission of HIV. They include
hepatitis B, C and D, syphilis, HTLV and alloimmunization to
red blood cells, which is further reason for seeking to mini-
mize the use of blood.
The cost of a hospital-based BTS to operate at basic-adequate
level, assuring adherence to measures aimed at reducing
transfusion-related HIV transmission, is only US$0.07 per
head of population. This figure would have been substantially
higher if the BMC guidelines for blood transfusion had not
been followed. These guidelines should be implemented
nationally and legislation is required to enforce adherence
because compliance is poor in Tanzania (Vos et al. 1994). The
financial savings on direct costs would be substantial if the
guidelines were generally adhered to.
In order to make such legislation and other transfusion prac-
tices effective, a control system should be established, ideally
implemented by the National AIDS Control Programme
(NACP). Currently each region has a Regional AIDS Control
Manager, but this person is unable to perform the required
tasks due to shortage of funds (Dodd 1995). Donors should,
therefore, provide more support to the NACP instead of
diverting funds to NGOs. If donors continue to support
NGOs instead of the NACP, an alternative structure for
coordination among them needs to be established. This struc-
ture should control resource allocation to avoid wastage of
funds on parallel intervention systems.
06 Jacobs (jl/d) 20/10/99 8:30 am Page 359
Hospital-based blood banking
Several other measures would help to improve the effective-
ness of a hospital-based BTS and to minimize HIV trans-
mission through transfused blood. Firstly, it is important to
improve the laboratory diagnosis of HIV infection among
blood donors to lower the risk of transfusing PHIUs. It is esti-
mated that only 80% of all blood for transfusion in Tanzania
is screened for HIV (World Bank 1992). The main reason
given is that it can not be done in emergencies. The annual
report of Rubya Hospital in Bukoba illustrates the logistical
problems of distribution of test kits and the regrettable con-
sequences of a failing logistical system (anonymous 1994). Of
the 371 blood donations tested, 30% were HIV-positive. Due
to an unreliable test kit supply, the other 782 blood trans-
fusions could not be screened, so potentially 235 transfusion
recipients could have been HIV-infected. It is a top priority to
strengthen the distribution of HIV-test kits to allow all blood
to be screened.
Currently, it is the task of the NACP to manage distribution
of test kits. Collaboration with the expanded programme on
immunization would be beneficial as it has a proper network
for distribution of vaccines by cold chain. In addition to
ensuring that testing can be done, maintenance of a stock of
blood is a prerequisite for proper and timely HIV-testing. A
voluntary blood donor panel allows this and should also be
continued on ethical grounds to reduce wastage of blood col-
lection units, tests and sera because of high HIV-prevalence
among relative donors.
It is particularly important for HIV-testing to be monitored,
although quality control is required for all hospital-based
blood banks. Samples of known but undisclosed content can
be sent from an organizing laboratory to a testing laboratory
and the results reported back. This serves three purposes: (1)
to allow the national health authorities to gain insight into the
general level of proficiency in the laboratory testing; (2) to
allow participants to gain an insight into their own level of
performance and to take action to remedy any problems
revealed; and (3) to allow organizers to identify participants
whose long-term performance is poor, in order to offer them
help and advice in resolving their problems (Emmanuel
1992). Because of the countrys size and malfunctioning com-
munications, organizing laboratories should ideally be estab-
lished at regional level. Such a quality control system was
established in Mwanza Region by the TanzanianNether-
lands Research on AIDS project and should be introduced in
all regions.
A further requirement is to ensure frequent supervisory visits
for monitoring HIV-testing. A study conducted in neighbour-
ing Zambia found that six out of 42 trained laboratory tech-
nicians were no longer present in their respective hospital
after 1 year. Of 19 hospitals that were using ELISA for HIV
screening, eight already had malfunctioning equipment after
1 year (Chipuka et al. 1993). In Zaire, supervisory visits sig-
nificantly improved transfusion procedures, including HIV-
testing (Laleman et al. 1992).
The demand for blood transfusion is significantly affected by
malaria, which accounts for 31% of episodes of sickness and
much of the anaemia in Tanzania (Ministry of Health 1995).
359
Efforts to limit the demand for transfusions due to severe
anaemia are constrained by the lack of capacity to perform
Hb tests in all dispensaries (Massawe et al. 1995). For chil-
dren with an Hb of 58 g/dl, the need for blood transfusion
can be averted by treating the underlying condition of
anaemia. However, using clinical signs for diagnosis of
anaemia at this Hb level had only a 66% sensitivity, and a
cheap and reliable method for Hb screening still does not
exist (Luby et al. 1995). A study from Malawi indicates that
the use of clinical signs to diagnose anaemia in children by
non-physician health workers has a sensitivity of 93% for
severe anaemia (Hb <5 g/dl) (Luby et al. 1995) Unfortu-
nately, the Hb is then at such a low level that a blood trans-
fusion is almost mandatory. In view of this, preventive and
curative measures to control malaria may be a more feasible
approach to reducing the demand for blood transfusion due
to severe anaemia.
In order to lower costs associated with BTSs, it is also import-
ant to improve collaboration between government and
mission hospitals. The latter are a valuable health sector
resource in Tanzania as they manage about half of all health
services (Gilson et al. 1994). In Mwanza, all the hospitals con-
ducting more than 50 blood transfusions a month were
mission hospitals, while those of the government transfused
20 or less units a month. The main reason for this difference
is patient attendance, which is likely to be higher at mission
hospitals because of an increased likelihood of drugs being
available. Additionally, mission hospitals usually have access
to external funding, which Laleman et al. (1992) identified as
an important determinant for successful transfusion practices
at rural hospitals.
Mission hospitals were also found to have better laboratories
than government hospitals (Gilson et al. 1995) so they could
provide HIV-tested and ABO,D grouped blood to govern-
ment hospitals. The latter could collect blood among groups
at lower risk for HIV-infection at their location and replace
the units received from mission hospitals. With such an
exchange system, blood can be tested for HIV by use of the
cheaper and more sensitive ELISA instead of rapid essays.
However, for such collaboration to happen at district level,
District Medical Officers (DMOs) need to be more involved
with the coordination of mission hospitals. Gilson et al.
reported that these remained largely outside the DMOs
influence and that most contacts occurred through the deliv-
ery of immunization supplies (Gilson et al. 1994).
Development of BTSs and the level at which they operate
needs to be compatible with the overall national health care
structure. The national health care plan and socioeconomic
circumstances need to be taken into account to ensure sus-
tainability. A BTS which operates at a higher level than the
overall health care structure does so by diverting resources
from other much needed interventions. For a low-income
country like Tanzania, provision of safe blood transfusion ser-
vices through a hospital-based blood banking system is a feas-
ible horizontal alternative to the more costly vertical
approach. Such an intervention will be more effective in pre-
venting HIV transmission through blood transfusion if the
above support measures are also introduced.
06 Jacobs (jl/d) 20/10/99 8:30 am Page 360
360 Bart Jacobs and Alec Mercer
Endnotes
1Producing whole blood and packed red cells, screened for
HIV, hepatitis B and syphilis, ABO,D grouped and cross-matched at
37C with anti human globulin.
2 The consumer price index for the United States was 103.9 in
1984; 116.0 for 198788; and 140.3 in 1992, giving factors for con-
version to 1992 US$ of 1.35 and 1.21, respectively. United States
Bureau of Census, Statistical Abstract of the United States: 1997 (117th
edition), Washington DC, 1997.
3 Calculated as in Table 3 on the basis of 100 PHIU not issued.
4 Formula for discounting from Beal et al. (1992).
References
Anonymous. 1994. Annual Report. Rubya Hospital, Tanzania.
Barongo LR, Borgdorff MW, Mosha FF et al. 1992. The epidemiol-
ogy of HIV1 infection in urban areas, roadside settlements
and rural villages in Mwanza Region, Tanzania. AIDS 6:
15218.
Beal RW, Bontinck M, Fransen L. 1992. Safe blood in developing
countries: a report of the EECs expert meeting. Brussels: EEC
AIDS Task Force.
Borgdorff M, Barongo L, van Jaarsveld E et al. 1993. Sentinel sur-
veillance for HIV1 infection: how representative are blood
donors, outpatients with fever, anaemia, or sexually transmitted
diseases, and antenatal clinic attenders in Mwanza Region, Tan-
zania. AIDS 7: 56772
Bossert TJ. 1990. Can they get along without us? Sustainability of
donor supported health projects in Central America and Africa.
Social Science and Medicine 39(9): 137584
Chipuka L, Chama D, Tembo C. 1993. Quality of blood transfusion
practices in district hospitals in Zambia. VIII International Con-
ference on AIDS/STD in Africa. Marrakech, December
(abstract Th. P.A. 0.27).
Cond J, Fleury-Brousse M, Waltisperger D. 1980. Mortality in
developing countries. TOME1 Databank. Volume 2: raw data,
deaths, mortality rates and life tables. Paris.
Dodd R. 1995. Who gets the donor money governments or NGOs?
AIDS Analysis Africa 5(1): 89.
Emmanuel JC. 1992. WHO guidelines for quality assurance in HIV
testing. In: Proceedings: HIV testing and quality assurance for
blood transfusion centres in regional and peripheral hospitals.
Dar es Salaam, October (abstract 3.2.1).
Foster S, Buv A. 1995. Benefits of HIV screening of blood trans-
fusion in Zambia. Lancet 346: 2257.
Gilson L, Kilima P, Tanner M. 1994. Local government decentralisa-
tion and the health sector in Tanzania. Public Administration
and Development 14: 4517.
Gilson L, Magomi M, Mkangaa E. 1995. The structural quality of
Tanzania primary health facilities. Bulletin of the World Health
Organisation 73(1): 10514.
Grosskurth H, Mosha F, Todd J et al. 1995. Impact of improved
treatment of sexually transmitted diseases on HIV infection
in rural Tanzania: randomised controlled trial. Lancet 346:
5306.
Gumodoka B, Vos J, Kigadye FC, van Asten H, Dolmans WMV,
Borgdorff MW. 1993. Blood transfusion practices in Mwanza
Region, Tanzania. AIDS 7: 38792.
Jacobs B. 1997. Approaches to decrease transfusion related HIV
transmission in sub-Saharan countries with hospital based
blood banks: experiences from Tanzania. Journal of HIV/AIDS
Prevention & Education for Adolescents & Children 1(3/4):
10536.
Kigadye R, Klokke A, Nicoll A et al. 1993. Sentinel surveillance for
HIV1 among pregnant women in a developing country: 3
years experience and comparison with a population serosurvey.
AIDS 7: 84955.
Killewo JZJ, Sandstrom A, Bredberg Raden U, Mhalu FS, Biber-
felde G, Wall S. 1993. Incidence of HIV1 infection among
adults in the Kagera region of Tanzania. International Journal of
Epidemiology 22(3): 52836.
Kitange HM, Swai ABM, Lilima PM, Masuki G, Alberti KGMM,
Mclarty DG. 1993. Anaemia is a major public health problem in
Tanzania. Health Policy and Planning 8(4): 41324.
Lackritz EM, Zucker JR, Ruebush TK et al. 1993. Follow up of
severely anaemic children in Kenya: the effect of blood trans-
fusion on mortality and haematologic recovery. VIII Con-
frence International sur le SIDA en Afrique & VIII Confrence
Africaine sur les MST. Marrakech (abstract W.O.P. 34).
Laleman G, Magazani K, Perriens JH et al. 1992. Prevention of blood
borne HIV transmission using a decentralised aproach in
Shaba, Zaire. AIDS 6: 13538.
Luby SP, Kazembe PN, Red SC et al. 1995. Using clinical signs to
diagnose anaemia in African children. Bulletin of the World
Health Organization 73(4): 47789.
Lyamula EF, Magesa PM, Mhalu FS. 1996. Problems and solutions
associated with transmission of HIV via donated blood in Tan-
zania. African Journal of Health Science 3: 610.
Massawe S, Urassa E, Lindmark G, Nystrom L. 1995. Anaemia in
pregnancy: perceptions of patients in Dar Es Salaam. East
African Medical Journal 72: 498503.
Mburu FM. 1994. Health delivery standards: vested interests in
health planning. Social Science and Medicine 39(9): 137584.
Ministry of Health. 1990. National AIDS Control Programme.
National Guidelines for Safe Blood Transfusion. May, Dar Es
Salaam.
Ministry of Health. 1995. Health Statistics Abstract 1995. June, Dar
Es Salaam.
Over M, Bertozzi S, Chin J, NGaly B, Nyamuryekunge K. 1988. The
direct and indirect cost of HIV infection in developing coun-
tries: the cases of Zaire and Tanzania. In: Fleming AF, Carballo
M, FitzSimons DW, Bailey RM, Mann J (eds). The global
impact of AIDS. New York: Alan R Liss.
Piot P, Laga M, Ryder R et al. 1990. The global epidemiology of HIV
infection: continuity, heterogeneity and change. Journal of
Acquired Immune Deficiency Syndrome 3: 40312.
Preble EA. 1990. Impact of HIV/AIDS on African children. Social
Science and Medicine 31: 67180.
Premji Z, Hamisi Y, Shiff C, Minjas J, Lubega P, Makwaya C. 1995.
Anaemia and Plasmodium falciparum infections among young
children in an holoendemic area, Bagamoyo, Tanzania. Acta
Tropica 59: 5564.
Prost N, de Ferranti D. 1985. The analysis and assessment of health
programmes. Social Science and Medicine 20(12): 123540.
Prost A, Prescott N. 1984. Cost-effectiveness of blindness prevention
by the onchocerciasis control programme in Upper Volta. Bull-
etin of the World Health Organization 62(5): 795802.
Savarit D, De Cock KM, Schutz R, Konate S, Lackritz E, Bondurand
A. 1992. Risk of HIV infection from transfusion with blood
negative for HIV antibody in a west African city. British
Medical Journal 305: 498501.
Schapira A. 1989. Chloroquine resistant malaria in Africa: the chal-
lenge to health services. Health Policy and Planning 4(1):
1728.
United Nations. 1995. Demographic Yearbook 1993. New York: UN.
van Dam CJ, Sondag-Thull D, Fransen L. 1992. The provision of
safe blood-policy issues in the prevention of human immuno-
deficiency virus transmission. Tropical Doctor 22: 2023.
Velema JP, Klokke AH, Kigadye FC et al. 1992. Quality control for
HIV screening of donor blood using EIA at peripheral hospitals
in Mwanza region, Tanzania. VIII International Conference on
AIDS/VIII STD World Congress. Amsterdam, July (abstract
PoC 4680).
Vos J, Gumodoka B, van Asten HAGH, Berege ZA, Dolmans
WMV, Borgdorff MW. 1993. Changes in blood transfusion prac-
tices after the introduction of consensus guidelines in Mwanza
Region, Tanzania. TANERA Clinical Studies, March.
Vos J, Gumodoka B, van Asten HA, Berege ZA, Dolmans WM,
Borgdorff MW. 1994. Changes in blood transfusion practices
06 Jacobs (jl/d) 20/10/99 8:30 am Page 361

Hospital-based blood banking 361

after the introduction of consensus guidelines in Mwanza Biographies

Region, Tanzania. AIDS 8: 113540.
Watson-Williams EJ, Kataaha PK. 1990. Revival of the Ugandan
Blood Transfusion System 1989: an example of international
cooperation. Transfusion Science 11: 17984.
World Bank. 1992. Tanzania: AIDS assessment and planning study.
Washington DC: World Bank.
World Health Organization. 1992. Guidelines for the organisation of
a blood transfusion service. Geneva: WHO.
Zucker R, Lackritz EM, Ruebush TK et al. 1994. Anaemia, blood
transfusion, HIV and mortality among women of reproductive
age in western Kenya. Transactions of the Royal Society of
Tropical Medicine and Hygiene 88: 1736.
Acknowledgements
Our sincere gratitude goes to the authorities and staff of the Depart-
ment of Pathology and of the Blood Bank, Bugando Medical Centre,
Mwanza. We thank Dr Anne Buv for her helpful suggestions and
acknowledge the helpful comments from the anonymous referee.
The project was supported by LIVOS and the Belgian Adminis-
tration for Development Cooperation.
Bart Jacobs, MSc, graduated in clinical sciences and was coordinator
of a tuberculosis project in the Philippines before coordinating the
blood-transfusion project in Mwanza. After completing a Masters
degree in Health Planning and Development at the University of
Wales Swansea, he worked on cost-effectiveness studies of tuber-
culosis programmes in countries of the former Soviet Union. He is
currently working on studies of HIV/AIDS and STDs with the
Medical Research Council in Uganda.
Alec Mercer, MSc, Mphil, is a health demographer who has worked
on several health programmes in Africa and Asia, conducting health
surveys, operational research and epidemiological studies. He is cur-
rently a Lecturer in Health and Population in the School of Social
Science and International Development, University of Wales
Swansea, and an adviser to the UKs Department for International
Development on NGO health projects.
Correspondence: Bart Jacobs, Development Studies, School for
Social Science and International Development, University of Wales
Swansea, Singleton Park, Swansea SA2 8PP, UK. (Fax: +44 (0) 1792
295682)

Annex 1: Cost Data

(All costs are in 1992 US$)

Capital costs

Item Item cost Number of Total cost Annual

items depreciation

Blood collection centre 20 900 1 20 900 16671

Scales for controlling the blood collection procedure 21.4 10 214 49.5
Bench centrifuge 1571 1 1571 363
Water bath 672 1 672 155
Fan cooled refrigerators 1000 3 3000 388.5
Voltage stabilizer 100 2 200 26
Plasma extractor 100.5 2 201 46.4
Di-electric tube sealer 2429 1 2429 561
Strippers 32.3 3 97 22.4
Domestic scale 143.5 2 287 66.3
Blood pressure meter 51.7 3 155 35.8
Hb meter 397 1 397 91.7
Laboratory thermometers 11 6 66 15.2
ELISA reader 1500 1 1500 194
Transport 4290 4290 555.6
Total capital costs 35 979 4237

Recurrent costs
Salaries
Administrative staff 135
Blood collection staff 310
Laboratory staff 135
Blood donor recruiter 535
Attendant 80
Pension fund 135
On call allowance 510
Transport
Total cost for one year 711.8
Consumables2
Disposable lancets 115
Blood collection containers3 5102
Reagents for blood grouping4 1254
Compatibility testing5 503.5
06 Jacobs (jl/d) 20/10/99 8:30 am Page 362

362 Bart Jacobs and Alec Mercer

Screening for syphilis 284.6

Screening for HIV 1195
Cost of publicity
Total cost 2412
Total recurrent costs 13 418
TOTAL ANNUAL COSTS 17 655

Cost per suitable blood unit produced

Total annual costs 17 655
Number of donors bled 1423
Number of suitable units 14216
Cost per suitable unit US$12.4

Marginal cost per suitable blood unit7

Relative donors (costs per 100)
125 lancets 8
100 blood bags 294
1.5 3 10 ml anti-A, B, AB 24
1.5 3 10 ml anti-D 27
120 syphilis tests 24
120 HIV tests 84
Total 461
(91 suitable units produced per 100 relative donors)
Marginal cost of US$5.1 per suitable unit from relative donors
Voluntary donors (costs per 100)
125 lancets 8
80 double blood bags 407
10 triple blood bags 51
10 quadruple blood bags 106
1.5 3 10 ml anti-A, B, AB 24
1.5 3 10 ml anti-D 27
120 syphilis tests 20
Diesel 10
120 HIV-tests 84
Total 737
(127 suitable blood units produced by 100 voluntary donors)
Marginal cost of US$5.8 per suitable blood unit from voluntary donors

Notes
1 The building is estimated to last 20 years; the refrigerator, voltage stabilizer and vehicle, 10 years; and the remainder, 5 years. For the vehicle

and transport, costs are divided over five hospitals.

2 Based on the annual average number of blood donors for the period 1.1.9331.12.94 which was 1423 (339 voluntary donors and 1084 rela-

tive donors).
3 Relative donors were bled using single blood bags (US$2.94). For 100 voluntary donors, 80 double, 10 triple, and 10 quadruple blood bags

were used (US$5.1, 5.1 and 10.6 respectively).

4 37 3 10 ml of anti-A, B, and AB was consumed at a cost US$52.9 for 10 3 10 ml; 37 3 10 ml of anti-D was used at a cost of US$180 for 10

3 10 ml.
5 15 3 10 ml was used at a cost of US$503.5.
6 339 voluntary donors were bled (HIV-prevalence 1.8%) and 1084 relative donors (HIV-prevalence 8.7%). This provided 990 suitable units

from 990 HIV-negative relative donors and 431 suitable units from 333 HIV-negative voluntary donors (266 double blood collection bags with
266 units; 33 triple bags with 66 units; 33 quadruple bags with 99 units).
7 Calculated per 100 donors being bled: 100 relative donors produced 91 suitable units (990/1084 3 100); 100 voluntary donors produced 127

units (431/339 3 100).