INT J TUBERC LUNG DIS 10(7):812–817
© 2006 The Union
Improving the laboratory diagnosis of TB in Ghana: 
the impact of a quality assurance system
K. K. Addo,* M. Dan-Dzide,* D. Yeboah-Manu,* K. Owusu-Darko,* P. Caulley,† M. Minamikawa,* 
F. Bonsu,‡ C. Lienhardt,§ P. Akpedonu,* D. Ofori-Adjei*
* Department of Bacteriology, Noguchi Memorial Institute for Medical Research, Legon, † National Public Health 
Reference Laboratory, Korle-Bu, ‡ National Tuberculosis Control Programme, Korle-Bu, Ghana; § West African 
TB Research Initiative, Dakar, Senegal
S U M M A R Y
SETTING: Greater Accra region, Ghana. cleanness, thickness, size and evenness increased from
OBJECTIVE: To establish a pilot quality assurance (QA) 64%, 79%, 69%, 46%, 67% and 60% in the last quar-
system in sputum smear microscopy and to evaluate its ter of 2000 to 81%, 90%, 86%, 79%, 80% and 74%,
impact. respectively, 24 months after the establishment of the
DESIGN: Quarterly supporting visits were paid to par- QA system. Within the same period, the rate of false-
ticipating laboratories between 2000 and 2002. Fifteen positives and -negatives decreased from respectively
examined slides were selected randomly from each labo- 14.8% and 20.5% to 0%, and agreements in positivity
ratory during the visits and blindly re-assessed. Feed- grade increased from 74% to 95%. The performance of
back was given promptly to the various laboratories. the participating laboratories in keeping the laboratory
Training and stakeholder workshops were organised registers up to date also improved.
whenever necessary. CONCLUSION: The QA system needs to be extended to
RESULTS: General improvements in smear preparation the rest of the country.
and staining as well as the reading ability of the labora- KEY WORDS: quality assurance; pilot system; sputum
tory personnel included in the study were observed. The smear microscopy; Ghana
average marks for specimen quality, staining ability, smear
INFECTIOUS DISEASES, including tuberculosis (TB), findings are suggestive of PTB. For follow-up cases,
remain the largest cause of illness and death in the only one specimen, preferably early morning, is col-
world. In Ghana, the National Tuberculosis Pro- lected at the second (repeated at month 3, if smear is
gramme (NTP), established in 1994 to control illness still positive at month 2), fifth and eighth month.3
and deaths associated with TB, uses the DOTS strat- The advantages of smear microscopy are that it is
egy. Within this framework of TB control, the identi- rapid, cost-effective and requires little in the way of
fication and treatment of infectious TB cases is the equipment. In addition, it detects most infectious cases.
priority. The disadvantages, however, are that it is dependent
Sputum smear microscopy is the main method of on the quality of the sputum and the staining tech-
diagnosis in Ghana. The purpose of smear micros- nique and reading ability of the microscopist. To main-
copy is three fold: 1) to diagnose patients with infec- tain a reliable laboratory service that will facilitate ac-
tious TB, 2) to monitor the treatment progress of in- curate diagnosis and offset the disadvantages of using
dividual patients, and 3) to document cure at the end smear microscopy, a well functioning quality assur-
of treatment.1,2 For the diagnosis of cases in Ghana, ance (QA) system is essential to ensure that informa-
three repeated sputum specimens are collected from tion generated by the laboratory is accurate, reliable
each TB suspect: one on-the-spot, followed the next and reproducible.4
day by one early morning and another on-the-spot. The three main components of QA are: 1) quality
Pulmonary tuberculosis (PTB) treatment is usually control/internal quality assurance (IQA), a process by
initiated when two of the specimens are smear-positive which each laboratory systematically and effectively
or one is smear-positive but X-ray and/or clinical monitors its own performance; 2) external quality
Correspondence to: Dr K Kwasi Addo, Department of Bacteriology, Noguchi Memorial Institute for Medical Research, Box
LG 581, Legon, Ghana. Tel: (233) 21-501178-9. Fax: (233) 21-50218-2. e-mail: kaddo@noguchi.mimcom.net
Article submitted 27 September 2005. Final version accepted 7 February 2006.
TB microscopy quality assurance in Ghana 813
assessment (EQA), a process whereby laboratory per- tive action and on-site training were offered where
formance is assessed by external assessors; and 3) necessary.
quality improvement (QI), which involves continuous
monitoring and identification of defects/errors fol- Blinded re-checking
lowed by remedial action.4
A nationwide evaluation of the public laboratories Slide sampling
that were performing smear microscopy was con- Using a national average slide positivity rate of 20%,
ducted previously (Addo K K, Owusu-Darko K, Dan- 15 examined slides per quarter per laboratory were
Dzide M, et al., country report submitted to the Japan randomly selected and blindly re-assessed in our lab-
International Cooperation Agency under the Infec- oratory. This was based on the ‘Lot Quality Assur-
tious Diseases Project, 2003). The quality of the lab- ance System statistical sampling’ method.5 Minimal
oratories was assessed in terms of manpower, techni- sensitivity was set at 80%. The slides were selected
cal skills, logistics and biosafety using a structured using the laboratory register by one of the study per-
questionnaire. The study recommended the establish- sonnel during the quarterly visits. The total number
ment of a nationwide QA system for smear micros- of diagnostic and follow-up slides examined in the
copy to improve TB diagnostic services. However, im- last quarter in each laboratory was divided by 15 to
plementation of a countrywide QA system can be obtain the sample size. For example, if 120 slides
successful only after a pilot system has been initiated. were used in the last quarter, the number was divided
A successful pilot system can then be replicated in by 15 to get 8, which means every eighth slide was
other laboratories nationwide. sampled.
We report here the results of a study evaluating the In some of the laboratories that had low slide pos-
impact of setting up a pilot QA system on the perfor- itivity, the starting position for sampling was changed
mance of peripheral laboratories performing smear whilst maintaining the sampling interval to include
microscopy in the Greater Accra Region of Ghana. some or all positive slides for re-checking.4 When a
slide was missing, the next slide identified in the lab-
oratory register was substituted, irrespective of the re-
METHODS sult. In the private laboratory where fewer than 15
Study area slides had been worked on in two quarters, all avail-
This study covers the Greater Accra Region, where able slides were selected for re-assessment.
the national capital, Accra, is located. The region is
made up of five districts, with a population of over Slide evaluation
4 million. The region was chosen based on its prox- On arrival at the NTBRL, the slides were evaluated
imity to the National Tuberculosis Reference Labora- blind by different study personnel who acted as the
tory (NTBRL). The study was conducted in the Accra first assessor. Each slide was evaluated according
and Tema districts of the region, where all 12 of the to the International Union Against Tuberculosis and
public TB diagnostic centres are situated, including Lung Disease/World Health Organization recom-
the Korle-Bu Teaching Hospital (KBTH). The other mended six parameters of smear preparation:2 1)
districts are without diagnostic centres. specimen quality—indicated by the presence of dust
cells/macrophages and/or other white blood cells; a
On-site evaluation or supporting visits smear of good quality should have 25 leucocytes/
Quarterly visits to all participating laboratories were field at a total magnification of 100; 2) staining—to
carried out between October 2000 and September check whether smear is understained (acid-fast bacilli
2002. Data were collected during the visits using a [AFB] in faint red colour) or overstained (red colour
structured questionnaire about the laboratory situa- background); AFB covered with dark colour are also
tion in terms of availability of written standard oper- an indication of excessive counterstaining; 3) clean-
ating procedures (SOPs), level of laboratory training ness—to check whether the smear is free from stain
of personnel, performance of IQA, availability of deposit, dirt (artefacts) and crystals produced by
functional microscopes and other materials needed overheating of the stain, debris, etc; 4) thickness—to
for TB microscopy, adequacy of consumables and re- check whether the whole depth of the smear is fo-
agents, prompt reporting of results, and observation cused sharply in each field; 5) evenness—to check
of biosafety practices. The laboratory register was whether smear is evenly spread in a repeated, small,
also checked to verify that all information had been coil-like pattern on the slide, not too thick and not
recorded accurately and completely. Slide boxes were too thin; 6) size—to check whether the smear is be-
checked to see whether positive and negative slides on tween the appropriate size of 1  2 and 2  3 cm. All
which xylene had been used to remove excess oil were the centres used the Ziehl-Neelsen (ZN) technique:
stored away from direct sunlight and in the same 0.3% carbolfuchsin for 5 min, 20% sulphuric acid as a
order as they were listed in the laboratory register. Suf- decolouriser for 5 min and 0.3% methylene blue as
ficient time was allotted for all the visits and correc- a counterstain for 1 min.
814 The International Journal of Tuberculosis and Lung Disease
Slide reading ratory personnel that they have a crucial role to play
The slides were re-read by the first assessor and re- in TB control. The number of laboratories initially
sults were compared with those of the participating identified as performing smear microscopy was nine.
laboratory. Interpretation of the results was based on Within 3 months of the study, this number increased
the following: 1) overall agreement—consistency in to 12, with the inclusion of a private laboratory. At
smear-negative and smear-positive results; 2) false- the start of the study, none of the participating labo-
positive—the result by the assessor was negative but ratories was practising IQA, but this changed 6 months
misread by the microscopist as positive; 3) high false- into the study. NTP-approved SOPs were pasted in all
positive—a negative smear was misread by the mi- the laboratories for easy reference, staining reagent
croscopist as 1 to 3; 4) low false-positive—a neg- containers were properly labelled with the date of
ative smear was misread by the microscopist as a scanty preparation and expiration, new batches of stains
positive (1–9 AFB/100 fields); 5) false-negative—the prepared by the regional technologist and distributed
result by the assessor was positive but was misread as to the participating laboratories were tested against
negative by the microscopist; 6) high false-negative— their performance with control slides, and carbol-
a high-positive smear (1 to 3) was misread as neg- fuchsin and methylene blue were filtered regularly
ative; 7) low false-negative—a scanty positive smear to remove crystals. Microscopes were maintained by
was misread as negative by the microscopist; and 8) cleaning objective lenses with lens paper or soft cloth
quantification error—a difference of more than one and covered with plastic cases or stored in cleaned,
grade between the assessor and the microscopist in dry and stable lockers at the end of each work day. In
grading a positive smear. laboratories with high humidity, a 15–20 watt bulb
was installed in the lockers to prevent fungal growth.
Discordant slides A rubber blower was used in most facilities to remove
Discrepancies between the results of the first assessor dust on microscope eyepieces, condensers and stages
and the microscopist at the participating laboratory before storage. Objective lenses (100) were replaced
were resolved by a second assessor. Because stained ba- in three centres and a new microscope was given to a
cilli can fade in certain conditions (e.g., exposure to sun- centre where the microscope could not be repaired.
light and high humidity with high temperatures),6 all Biosafety measures, such as proper handling of spu-
false-positive slides and slides with quantification errors tum and cleaning of work benches with phenol-based
were re-stained and re-read by decolourising the slides disinfectants, were employed. At first it was not pos-
with 20% sulphuric acid and re-staining with ZN. sible to obtain data from any of the participating lab-
oratories due to poor documentation practices. How-
Feedback/follow-up visits ever, this has changed, as laboratory registers are now
Prompt feedback was given to the participating labo- filled out completely and accurately.
ratories and results of blinded re-checking were dis- Slide labelling has also improved, as lead pencils
cussed with the personnel. Discordant slides were re- are used to label frosted slides and diamond pencils to
turned to the original laboratory and re-read by the label non-frosted slides. Examined slides are now
microscopists. This gave them the opportunity to re- cleaned with xylene and properly stored for re-check-
examine the slides and correct earlier errors.7 Other as- ing in all laboratories.
pects of poor performance were also investigated on-site The on-site and two formal training courses con-
by identifying the possible causes and finding remedies. ducted for 24 participating laboratory personnel were
beneficial, as there has been a general improvement in
Training and stakeholder workshops smear preparation techniques and in the reading abil-
During the follow-up visits, laboratory personnel were
trained on-site. Formal training workshops were also
periodically organised at the NTBRL. A stakeholder
workshop was held quarterly to discuss study progress,
including problems encountered during the visits to
laboratories and any remedial actions needed.
Data analysis
All data were analysed using SPSS (SPSS Inc, Chicago,
IL, USA) and Microsoft Excel (Microsoft, Palisade
Corp, Newfield, NY, USA).
RESULTS
The establishment of the QA system in the Greater Figure 1 Assessment of smear preparations. 4Q  fourth
Accra Region has led to the recognition by the labo- quarter; QA  quality assessment; 3Q  third quarter.
TB microscopy quality assurance in Ghana 815
Table High and low positive/negative readings by microscopist per quarter
False positive False negative Total slides selected Total false slides
readings readings (n  180)*
False False
Year High Low High Low Positive Negative Positive Negative 
2001
First quarter  3 5 7 9 95 85 8 16
Second quarter  3 4 5 7 88 92 7 12
Third quarter  2 4 4 5 90 90 6 9
Fourth quarter  2 3 3 5 96 84 5 8
2002
First quarter  1 2 2 3 89 91 3 5
Second quarter  0 1 0 1 94 86 1 1
Third quarter  0 0 0 0 87 93 0 0
* n  total number of slides selected per quarter (15 slides/centre  12).
ity of the microscopists. In the last quarter of 2000 with minimal training. Functional QA of their work is
(Pre-QA), the average scores for specimen quality, therefore essential. QA is a system designed to contin-
staining ability, smear cleanness, thickness, size and uously improve the reliability, efficiency and use of
evenness were 64%, 79%, 69%, 46%, 67% and 60%, microscopy as a diagnostic and monitoring tool. This
respectively; these rose to 81%, 90%, 86%, 79%, is very important for the management of TB cases, as
80% and 74%, respectively, at the third quarter of suspects are not treated unless diagnosed in the labo-
2002, 2 years after the establishment of the QA sys- ratory. In the current study, the number of centres
tem (Post-QA) (Figure 1). Within the same period, the performing smear microscopy, including private lab-
rate of false-positives/negatives decreased from 14.8% oratories, increased by 25% within one quarter. This
and 20.5%, respectively, to 0% for both. Although was a very important development, as the increase in
false readings were observed, most of them were low diagnostic centres led to increases in case detection
false-positive/negative (Table). The overall smear- in the study areas from 43.5% to 53%,8 and the work-
positive/negative agreements and agreement in posi- load of other centres, especially the teaching hospital,
tive grading also increased from 85% to 100% and was reduced. IQA practices, such as testing of new
from 74% to 95%, respectively (Figure 2). stains and regular filtering, contributed to the reduc-
tion in false-positive/negative results. Safety measures
for handling sputum specimens observed in all the
DISCUSSION laboratories, such as: 1) sputum collection in the open
Smear microscopy is the most cost-effective method air instead of in the washroom; 2) frequent hand
of diagnosing PTB suspects reporting to health insti- washing; 3) careful opening of sputum containers to
tutions. It is used to identify infectious cases of dis- avoid splashes, which can create aerosols; and 4) the
ease, assess response to treatment and monitor cure use of the rough end of wooden applicators instead of
rates.1,2 However, in most diagnostic centres in Ghana, wire loops (which helps prevent the formation of aero-
smear microscopy is usually performed by personnel sols that could occur during the flaming of the loop)
to make smears, have aided in ensuring the safety of
laboratory personnel and patients.
The improvement in slide labelling and storage has
facilitated the random selection of examined slides
for re-assessment.
Prompt reporting of smear results by microscopists
to requesting officers has also facilitated early treat-
ment for newly diagnosed cases and rapid decision
making as to whether to continue or to change the
patient’s drug regimen in follow-up cases.
One major obstacle of this study was the issue of
rotation or transfer of staff, where eight laboratory
personnel trained in smear microscopy were put on
other benches, while two were transferred to other
laboratories outside the study area. This was evident
Figure 2 Assessment of reading ability. 4Q  fourth quarter; in the fourth quarter of 2001 (mid-QA), when the
QA  quality assessment; 3Q  third quarter; Pos  positive; smear size level dropped from the fourth quarter of
neg  negative. 2000 level of 67% to 64% (Figure 2). Investigations
816 The International Journal of Tuberculosis and Lung Disease
during follow-up visits revealed that the smears in References
two of the participating laboratories were being pre-
1 World Health Organization. Laboratory Services in Tuberculo-
pared by two new technicians who had been trans- sis Control. Part II: Microscopy. WHO/TB/98.258. Geneva,
ferred to the laboratories. This indicates a need to Switzerland: WHO, 1998.
train all laboratory personnel in the country in smear 2 Rieder H L, Chonde T M, Myking H, et al. The Public Health
microscopy. Services National Tuberculosis Reference Laboratory and the
There are many different methods for conducting National Laboratory Network, minimum requirements, role
and operation in a low-income country. Paris, France: The Inter-
QA in smear microscopy. None of these methods had national Union Against Tuberculosis and Lung Disease, 1998.
been studied in Ghana to ascertain their feasibility. 3 Addo K K, Caulley P, Yeboah-Manu D, et al. Tuberculosis mi-
The present study was intended to serve as a model croscopy. a laboratory manual for Ghana. Accra, Ghana: Min-
for future nationwide QA implementation. Our find- istry of Health, 2001: pp 1–9.
ings clearly indicate that supporting visits, which act 4 Aziz M A, Ba F, Becx-Bleumink M, et al. External quality assess-
ment for AFB smear microscopy. Washington, DC, USA: Asso-
as motivation for laboratory personnel, on-site and ciation of Public Health Laboratories, 2002: pp 1–54.
formal training, blinded rechecking of examined 5 Lemeshow S, Hosmer D W, Klar J, Lwanga S K. Lot quality
slides and timely feedback lead to improvements in assurance sampling. In: Lemeshow S, Hosmer D W, Klar J,
TB laboratory services. We therefore recommend that Lwanga S K, eds. Adequacy of sample size in health studies.
this system be extended to the rest of the country. Chichester, United Kingdom: John Wiley & Sons (on behalf of
WHO), 1990: pp 24–28.
6 Van Deun A, Chambugoni N, Hye A, Hossain A. Rapid fading of
carbolfuchsin stained AFB smears under extreme conditions of tem-
Acknowledgements perature and humidity. Int J Tuberc Lung Dis 1997; 4: 384–386.
We thank the Infectious Diseases Project of the Noguchi Memorial 7 Van Deun A, Portaels F. Limitations and requirements for qual-
Institute for Medical Research, funded by the Japan International ity control of sputum smear microscopy for acid-fast bacilli. Int
Cooperation Agency and the Department for International Devel- J Tubercle Lung Dis 1998; 2: 756–765.
opment (contract number AG 2071), for their support of this 8 Ghana Health Service/Ministry of Health, National Tuberculo-
study. We also appreciate the support of Dr Zakaria Lawani, the sis Programme Annual Report, 2003. Accra, Ghana: Ghana
GAR TB coordinator, and all technicians included in the study. Health Service/Ministry of Health.
R É S U M É
CONTEXTE : Région Greater-Accra du Ghana. la coloration, la propreté, l’épaisseur, la dimension et la
OBJECTIF : Mettre en place un système pilote d’assu- régularité des frottis étaient respectivement de 64%,
rance qualité (AQ) en microscopie des frottis de crachat 79%, 69%, 46%, 67% et 60%, mais ces points ont aug-
et évaluer son impact. menté respectivement à 81%, 90%, 86%, 79%, 80% et
SCHÉMA : Des visites d’assistance trimestrielles ont été 74%, 24 mois après la mise en œuvre du système AQ.
effectuées aux laboratoires participants entre 2000 et Pendant la même période, les taux des faux positifs et
2002. Quinze lames déjà examinées ont été sélectionnées faux négatifs ont diminué de 14,8% et 20,5% respec-
au hasard dans chaque laboratoire au cours des visites et tivement à 0%, et la concordance du degré de positivité
ont été évaluées à l’aveugle. Les résultats ont été envoyés a augmenté de 74% à 95%. La performance des labora-
immédiatement aux laboratoires. Des ateliers de forma- toires participants sur le plan de la documentation ad-
tion et des partenaires ont été organisés selon les besoins. ministrative a aussi connu une amélioration, puisque les
RÉSULTATS : Une amélioration générale dans la prépa- registres des laboratoires ont été remplis complètement
ration et la coloration des frottis ainsi que la capacité de et avec précision.
lecture du personnel du laboratoire inclus dans l’étude a CONCLUSION : Il y a un besoin d’expansion du système
été remarquée. Ainsi, au dernier trimestre de 2000, les AQ pour couvrir le reste du pays.
points moyens pour la qualité du spécimen, la qualité de
R E S U M E N
MARCO DE REFERENCIA : La región del Gran Accra en reevaluación se transmitió prontamente a los diversos
Ghana. laboratorios y se organizaron talleres de adiestramiento
OBJETIVO : Establecer un sistema experimental de con- para el personal de laboratorio y de discusión con las
trol de la calidad (AQ) para el examen microscópico del partes interesadas, cuando se consideró necesario.
esputo y evaluar su repercusión práctica. RESULTADOS : Se observó una mejoría global en la
MÉTODOS : Se realizaron visitas trimestrales de apoyo a preparación de los frotis y las coloraciones y en la ca-
los laboratorios participantes entre 2000 y 2002. En pacidad de lectura del personal de los laboratorios par-
cada visita, se escogieron aleatoriamente 15 frotis exa- ticipantes. En el último trimestre de 2000 la calificación
minados por cada laboratorio y se llevó a cabo una se- promedio fue del 64% para la calidad de la muestra, del
gunda lectura en forma anónima. El resultado de la 79% para la coloración, del 69% para la limpieza del
TB microscopy quality assurance in Ghana 817
frotis, del 46% para su espesor, del 67% para su tamaño al 0% y la concordancia en el grado de positividad au-
y del 60% para su homogeneidad. Las calificaciones me- mentó del 74% al 95%. Asimismo, se observó una me-
joraron respectivamente al 81%, 90%, 86%, 79%, 80% joría en la documentación de los laboratorios, pues los
y al 74%, 24 meses después de la instauración del sistema registros se llenaron integralmente y de manera exacta.
de AQ. En el mismo lapso, las tasas de falsos positivos CONCLUSIÓN : Es necesario extender el sistema de AQ
del 14,8% y de falsos negativos del 20,5% disminuyeron de la baciloscopia al resto del país.