Secondary Logo

Journal Logo

Implementation and Operational Research: Clinical Science

Validation in Uganda of the New WHO Diagnostic Algorithm for Smear-Negative Pulmonary Tuberculosis in HIV Prevalent Settings

Walley, John Mcomm, FPHM*; Kunutsor, Setor MBChB, MSt*; Evans, Morgan MRCP; Thoulass, Janine MBChB, MRCP, MFPH, MSc; Katabira, Elly FRCP§; Muchuro, Simon MBChB, MPH§; Matovu, Ahmed MBChB, MPH

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: August 15, 2011 - Volume 57 - Issue 5 - p e93-e100
doi: 10.1097/QAI.0b013e3182243a8c



Tuberculosis (TB) is a major public health problem globally with one-third of the world's population latently infected.1 In 2007, there were an estimated 9 million incident cases of TB, of which an estimated 1 million (15%) were HIV positive.2 HIV infection is a risk factor for progression of latent or recent TB infection to active TB disease, and without HIV, global TB incidence would be falling.2 Though the diseases are epidemiologically linked, TB and HIV services commonly are not.

Diagnosis of pulmonary tuberculosis (PTB) in low-income and middle-income countries relies on sputum smear examination for acid-fast bacilli (AFB) which can detect up to 50%-60% of PTB cases.3 Other studies have reported the sensitivity of AFB microscopy to range from 22% to 80% in patients with culture-confirmed PTB.4,5 Diagnosis of sputum smear-negative PTB is based on clinical and radiological findings. Co-infection with HIV results in higher rates of smear-negative and extrapulmonary TB leading to delayed or missed diagnosis. Progressive immune suppression also affects the clinical and radiological presentation of PTB, compounding the difficulty of diagnosis.6 The mortality rate of TB/HIV-coinfected patients is higher, especially for those with sputum smear negative7 and extrapulmonary TB, and this delay in diagnosis may be a significant contributing factor. Guidelines have traditionally focussed on identifying smear-positive PTB, which was justified by public health concerns about its infectivity. However, as smear-positive PTB case finding and holding targets have increasingly been achieved; and with the advent of the HIV epidemic, smear-negative PTB is growing in significance. Originally smear-negative PTB was thought to have a better prognosis compared with smear-positive PTB, but with co-existing HIV infection, the converse is true.8

At present, there is no simple diagnostic tool for the investigation of smear-negative PTB. The 2003 WHO diagnostic algorithm (UgWHO03)9 used in Uganda and most other high TB burden countries entails a lengthy diagnostic pathway. Diagnosis of PTB relies on sputum smear microscopy and if negative, a trial of antibiotics to exclude bacterial infection. If there is no response to antibiotics, clinical review and chest X-Ray (CXR) are performed. However, patients may experience a transient response to empirical antibiotic treatment, even when PTB is present, which delays diagnosis. Evaluation of guidelines that follow a similar diagnostic pathway have found that they have poor sensitivity levels of 50%-59%.6,7 Specificity conversely is relatively high at more than 80%.10,11

A WHO expert group developed an algorithm for the diagnosis of PTB in National Treatment Programs in low-income settings.12 This aimed to improve detection rates and expedite the diagnostic process of sputum smear-negative PTB to effect a reduction in mortality. Changes included an emphasis on the integration of TB and HIV care from first presentation, CXR and clinical review earlier in the diagnostic pathway, use of antibiotics as clinically indicated rather than as a diagnostic aid to exclude PTB, and use of sputum culture (where available) to rule-in rather than rule-out the diagnosis of smear-negative PTB. A further modification was a reduction in the recommended number of sputum smears done in each set from 3 to 2. The resultant diagnostic algorithm has a shorter pathway from presentation to diagnosis both in terms of the number of visits and days duration. It has since been revised to enhance usability and simplicity without altering the original pathway and has been approved for use by the WHO. This algorithm has not yet been validated within the context of low-income and high HIV prevalent settings where smear-negative PTB poses major public health problems.

In this prospective observational cohort study, we aimed to validate the feasibility and effectiveness of the WHO07 in Uganda, a high HIV prevalent resource-limited setting in comparison to the existing UgWHO03. Findings from this study will guide further development and implementation of strategies to investigate PTB in HIV prevalent populations.


Study Area

This study was conducted at 2 sites in Uganda. The sites chosen were fairly typical urban and rural settings, which differed in facility and client characteristics, resources, and approach to care provision. We aimed to provide evidence on how well the new guidelines performed and to explore implementation issues in these 2 diverse resource-limited settings. This article presents results from the rural site. Kayunga District Hospital is a government run facility with HIV clinic services. The urban site, Reach Out Mbuya Parish HIV/AIDS Initiative, is a nongovernmental community-based AIDS care organization based in a deprived urban area of Kampala. Results from Reach Out will be presented elsewhere.

Kayunga District Hospital is located in a rural setting in central Uganda. It is a 123-bed facility, about an hour and a half north east of Kampala. New clients are referred from hospital wards, outpatient clinics, lower health facilities, and voluntary counselling and testing centres. Outpatient HIV clinic sessions are held for new patients every Friday. These are staffed on average by 1 doctor, 1 clinical officer, 5 nurses, and 2 nursing assistants. This is typical of government facilities, where much of the medical care is devolved to nursing staff. The hospital laboratory facilities perform diagnostic smears for TB. CXRs are performed within the hospital radiology department and interpreted by the duty doctor. As part of our study, CXRs which could not be easily interpreted by the duty doctor were reviewed by a radiologist at the National Referral Hospital.

Study Design and Participants

A prospective observational cohort study design was used with evaluation of the existing UgWHO03 and followed by the new WHO07 under routine operational program conditions. Both algorithms are available as supporting information (see Figures 1 and 2, Supplemental Digital Content 1,, which has the flowcharts for the UgWHO03 and WHO07, respectively). The study was implemented with minimum additional inputs, reflecting real-life settings. The ethics committee of the Makerere University Medical School reviewed and approved the study procedures and data collection instruments. Permission to conduct the study in Kayunga District was sought from the District Health Offices and Medical Superintendent of the Hospital. Written informed consent was obtained from clients willing to participate. For patients 16 years of age and older but younger than 18, consent was sought from the parents or guardian and the patient. Consent forms were signed by those who were literate and marked with a left thumbprint otherwise.


We hypothesized that the introduction of the WHO07 would reduce time to diagnosis for smear-negative PTB, although improving the sensitivity, compared with the existing algorithm in a high HIV prevalent setting.

Sample Size

The primary outcome measure was a reduction of time to diagnosis for smear-negative PTB. A pilot study conducted at the Reach Out study site showed that with the existing diagnostic algorithm, the mean time to diagnosis of smear-negative PTB was 26.3 days with a range of 8-46 days. We expected a significant reduction in this once the new algorithm was introduced. With the introduction of the new guidelines, the majority of patients were expected to complete their assessment for smear-negative PTB within 3 or 4 visits, which should occur within 3 weeks (the first visit will be the HIV test and the subsequent visit will be on the first Friday after this). Therefore, proportion of patients receiving a diagnosis of smear-negative PTB was to be evaluated within 21 days. Using the existing algorithm, one-third of smear-negative PTB cases were detected clinically (ie, not on culture) within 21 days. The majority of confirmatory cultures were still outstanding, and no cases had been detected on culture that had not been detected clinically. With the introduction of the new algorithm, it was expected to at least double the proportion of cases detected within the 21-day time frame. To provide an estimate of the numbers, the new algorithm would be feasibly able to detect clinically within the 4 visits, the literature was searched for sensitivity of diagnosis based on clinical examination and CXR, but limited data was found. Although there were no directly comparable examples in the literature, a study in Nairobi found CXR to have a sensitivity of 80% (specificity 67%) for smear-negative PTB (though this was using a scoring system).13 Another trial which calculated sensitivity of a diagnostic algorithm using antibiotics stated that sensitivity would be 79% if false negatives due to antibiotic trial were excluded from their analysis of sputum smear-negative diagnosis,11 however this was in a country of low HIV prevalence which could lead to overestimation of sensitivity, and antibiotic trial was performed before CXR. As CXR should be performed in the smear-negative group early in the diagnostic process, a conservative estimate of two-thirds of cases being detected within 21 days after implementation of the new algorithm was made. The figure of two-thirds has been used to account for lower sensitivity of CXR for PTB in HIV prevalent populations. Using these figures, power calculations showed that a sample size of 31 in each group was required to give 80% power to detect a significant difference in the proportion of people diagnosed within 3 weeks at a significance level of P < 0.05. In total, a sample size of 31 sputum smear-negative patients in the group using the existing algorithm and 31 in the group using the new algorithm was needed. Based on data already collected from the Reach Out site, it was intended to recruit 200 patients with a history of 2 or more weeks of cough at each site, in each phase, to ensure this number of patients with smear-negative PTB was obtained. Specificity was not calculated as part of sample size as the aim of this should be to minimize loss of specificity rather than increase it.

Data Collection

Between June 2008 and September 2010, new HIV-positive patients presenting to the HIV clinic were screened for a history of 2 or more weeks of cough. Those aged 16 years and older, ambulant and with no history of previous TB disease or TB treatment were offered enrollment into the study. Separate guidelines were used for seriously ill or nonambulant patients, and they were excluded from the study. Patients were monitored for the first half of the study under the existing diagnostic UgWHO03 algorithm. The WHO07 was subsequently introduced in the second half of the study. Study participants were followed up for 3 months from the time of enrollment. Training was given to all members of staff involved in the clinical management of the patients both before monitoring of existing practice and before the introduction of the new guidelines.

The sputum specimens taken comprised spot-morning-spot for the first half of the study. For the second half, this became spot-morning (as the updated algorithm required 2 samples in each set). At the first consultation, clients were asked to submit an additional sputum sample, which was transported (in a cool box) to the laboratory for culture. Sputum smears were examined for Mycobacteria using Ziehl-Neelsen (Z-N) stain at the hospital laboratory and sputum cultures were performed on Lowenstein-Jensen medium at the National Tuberculosis Reference Laboratory in Kampala. Culture was used as the gold standard and to validate clinical decisions.


HIV prevalent settings are defined by WHO as countries where the adult HIV prevalence rate amongst pregnant women is ≥1% or in which the HIV prevalence amongst TB patients is ≥5%.14 According to the UgWHO03 guidelines, smear-positive PTB was diagnosed if at least 2 of the 3 sputum examinations was ZN positive. Smear-negative PTB was defined as 3 ZN-negative samples and a CXR consistent with active PTB or decision by clinician to treat with a full course of anti-TB treatment or a positive sputum culture. Using the WHO07 guidelines, a HIV-positive patient was considered to have smear-positive PTB if at least 1 of the 2 specimens obtained was ZN positive. Smear-negative PTB was diagnosed in HIV-positive patients on the basis of 2 ZN-negative smears, by a clinical diagnosis, a CXR consistent with active PTB, or later due to a positive sputum culture or by clinical diagnosis.

Statistical Analysis

Data were entered into an MS Access database, and statistical analysis was carried out using Statistical Package for the Social Sciences software version 15.0 for windows (SPSS Inc, Chicago, IL). Descriptive analyses were conducted for sociodemographic and other secondary outcomes. All time-to-events variables for both algorithms with skewed distributions were log transformed to improve their normality. Thus, geometric means of time-to-events variables and their corresponding 95% confidence intervals were presented. Baseline characteristics and outcomes between the 2 guideline groups were compared using χ2 tests or Fisher exact tests for categorical variables and 2-sample t tests or Wilcoxon rank-sum tests for continuous variables. Sensitivity, specificity, positive and negative predictive values of the diagnostic algorithms, and their respective 95% confidence intervals were calculated directly using Epi-Info (Centers for Disease Control and Prevention, Atlanta, GA). A Z test for proportions was used to test for significant differences in sensitivity and specificity values between the 2 algorithms. The P values <0.05 (2-sided) were considered statistically significant.


Progress through both diagnostic algorithms from presentation till diagnosis of PTB is available as supporting information (see Figures 3 and 4, Supplemental Digital Content 2,, which outlines progress through the UgWHO03 and WHO07, respectively).

Sociodemographic Data and Patient Characteristics

One thousand four hundred and eight new patients were screened for a 2-week history of cough between June 2008 and September 2010. Four hundred seventy-five patients satisfied the 2 or more weeks of cough eligibility criterion; 270 and 205 for the UgWHO03 and WHO07, respectively. Of the 270 and 205 patients, 147 and 166 were enrolled onto the UgWHO03 and WHO07, respectively. Patients excluded were those who did not meet the inclusion criteria or did not return after the first visit or have any investigations done. Table 1 below shows the demographic and patient characteristics for both algorithms. The groups did not differ significantly with respect to demographic or treatment characteristics.

Baseline Characteristics of Participants for Diagnostic Algorithms

Outcomes of Diagnostic Process

Tables 2, 3, 4, and 5 shows summary of outcome variables of the 2 diagnostic pathways from presentation to diagnosis of PTB. Of the total number of patients enrolled onto the UgWHO03 and WHO07, only 3% and 13%, respectively, completed all elements of both algorithms in the correct sequence including smears, CXRs, and cultures mainly due to inability of the patient to produce sputum. Seventeen (94%) of 18 smear positives were detected on the first specimen. A total of 48 (33%) and 41 (25%) patients defaulted within 3 months follow-up for the UgWHO03 and WHO07, respectively, though majority had received a final diagnosis before defaulting [27 (56%) and 32 (78%) for the UgWHO03 and WHO07, respectively] and therefore had data on their progress through the diagnostic process. Five (3%) were transferred out, and 7 (5%) patients were confirmed by the clinic to have died during the course of the 3-month follow-up period for the UgWHO03. These were, respectively, 1 (1%) and 8 (5%) for the WHO07. Those who defaulted potentially included deaths but which could not be confirmed because they could not be traced during follow-up.

Numbers Followed and Outcomes of Diagnostic Process
Diagnosis Made by Algorithms Compared With Sputum Culture Results
Sensitivity, Specificity and Positive and Negative Predictive Values of Algorithms Compared With Sputum Culture Results
Geometric Mean (95% CI) of Time to Events for the Diagnostic Algorithms

Of the 12 culture-positive results for the UgWHO03, 8 (67%) were smear negative; and of the 26 culture-positive results for the WHO07, 19 (73%) were smear negative. For all patients with available culture results and a final diagnosis, only 2 (2.2%) were treated unnecessarily for PTB and 3 (3.3%) with PTB were given a non-PTB diagnosis for the UgWHO03. The numbers were even lower for the WHO07: 2 (1.5%) and 1 (0.8%), respectively.

Diagnostic sensitivity for all PTB increased nonsignificantly from 75% to 96% (exact P = 0.084), and specificity was maintained at 97% and 98% (exact P = 0.999) with the introduction of WHO07. When analysis was restricted to smear-negative PTB, diagnostic sensitivity increased from 63% to 95% (exact P = 0.065) and specificity remained high at 99% and 98% (exact P = 0.999).

Table 5 shows time to events for both diagnostic algorithms. There were significant reductions in the geometric mean days (11.0 vs. 21.2, P < 0.05) and number of health facility visits (1.7 vs. 2.8, P < 0.001) from presentation to diagnosis of all PTB for the new algorithm (WHO07) compared with the UgWHO03, respectively. When analysis was restricted to smear-negative PTB, the geometric mean days (28.5 vs. 34.1, P = 0.489) and number of health facility visits (2.5 vs. 2.6, P = 0.690) from presentation to diagnosis were nonsignificantly reduced for the new algorithm (WHO07) compared with the UgWHO03, respectively. Whereas the geometric mean days from presentation to diagnosis (1.3 vs. 9.3, P < 0.001) and number of health facility visits (1.1 vs. 2.2, P < 0.001) for smear-positive TB were significantly reduced, respectively.

Significant barriers to implementation of both diagnostic algorithms were identified. These are available as supporting information (see Text Document, Supplemental Digital Content 3,, which lists barriers to implementation of both algorithms).


The results show significant reductions in time and number of visits to diagnosis of all PTB with the WHO07. Sensitivity for all PTB diagnosis was 75% for the WHO03 and higher at 96% for the WHO07. The specificity of all PTB remained high at 97% and 98% for the 2 algorithms. Only 4% of patients with culture-positive TB were not diagnosed correctly and approximately 2% of patients with culture-negative results were diagnosed as having PTB by the WHO07. For the diagnosis of smear-negative PTB, sensitivity increased nonstatistically significantly from 63% to 95%. Whereas the specificity remained high at 99% and 98% for the 2 algorithms. The change in sensitivity approached but did not reach statistical significance, due to the relatively small numbers of cases. For the diagnosis of smear-negative PTB, the time and number of visits from presentation to diagnosis were reduced for the WHO07, but not to the level of statistical significance. Operational constraints during implementation of both algorithms (especially the unavailability of an attendant physician for a period during the implementation of the WHO07) may have reduced the impact of the WHO07 on outcomes such as the duration and number of visits from presentation to diagnosis of smear-negative PTB. Our findings, however, do suggest that the new diagnostic pathway is an improvement for HIV prevalent populations in resource-constrained settings in expediting the diagnosis of smear-negative PTB with accuracy, especially if applied rigorously.

Our study was effectiveness research under routine conditions (no additional staff or resources etc.) and aimed to identify operational constraints that should be taken into account during replication elsewhere. Major problems were encountered in the transport and processing of culture samples with relatively high numbers of contaminated samples and smear-positive but culture-negative results. False-negatives may have occurred due to inadequate processing or transport conditions. This may potentially have caused loss of viability. Such barriers are likely to be commonplace in resource-limited settings and may constitute a significant barrier to the full implementation of the new algorithm. Polymerase chain reaction-based diagnostic tests are currently too expensive for low-income settings. Sputum cultures with their high specificity and sensitivity around 99% and 80%, respectively,15 are the most appropriate gold standard test in this situation. Hence, there should be a focus on finding feasible methods of transporting and processing sputum cultures in resource-limited settings that can achieve high levels of performance. There should also be evaluation of feasible and alternative culture techniques since traditional solid media requires up to 8 weeks for the growth of TB bacilli. In 2007, the WHO issued new policy recommendations on the use of liquid culture and rapid species identification.16 Liquid culture systems are more rapid and sensitive but more costly than solid culture and may improve smear-negative PTB diagnosis. They are already in use in developed countries and are currently being evaluated to assess their impact and feasibility in low-income countries with high TB prevalence.16 More recently, WHO has endorsed the nucleic acid amplification test codeveloped by the Foundation for Innovative and New Diagnostics, which can provide accurate diagnosis for many patients in about 100 minutes. Implementation of this test could double the number of HIV-associated TB cases diagnosed in areas with high rates of TB and HIV. It is expensive but will be subsidised in low-income and middle-income TB endemic countries. WHO is currently releasing recommendations and guidance for countries to incorporate this test in their programs.17

Some patients could not produce adequate or any specimens for both smears and culture. Sputum induction by simple chest physiotherapy18 or with nebulized hypertonic saline19 may be alternative solutions in resource-poor areas. This can, however, be hazardous to staff and other patients unless good infection control measures are put in place. Smear-positive disease may be missed if sputum samples are inadequate and indeed we found relatively high numbers of smear negatives culture-positive results for both algorithms. Some studies have shown a 2-fold increase in the detection of AFB in negative smears using sputum liquefaction and concentration through centrifugation,20,21 whereas other studies have reported no benefit of sputum concentration over direct microscopy.22,23 Our results showed that all smear-positive cases were detected on the first specimen for both algorithms which is consistent with reports from other studies.24,25 This supports the reduction of sputum smears in each set from 3 to 2 in the WHO07 and that smear-positive PTB be diagnosed if at least 1 of the 2 specimens obtained was microscopy positive. The nucleic acid amplification test, or similar polymerase chain reaction-based tests, when affordable will greatly improve diagnosis and reduce delays and visits as part of the WHO07 algorithm. Until then, efforts should also focus on improving sputum microscopy. Z-N stained smears, which are generally used in resource-constrained settings, are less sensitive than fluorochrome-stained smears (fluorescence microscopy) which take only 1-2 minutes to read. Limitations of fluorescence microscopy include high costs involved in purchasing a fluorescence microscope and a constant electricity supply.26

There were frequent delays in radiograph interpretation because only 1 physician was allocated to the HIV clinic. Though much of the clinical decision-making is devolved to nursing staff, they lacked CXR interpretation skills, and so diagnosis was delayed awaiting interpretation by the physician. This could be addressed by adequate training of the available staff in x-ray interpretation.

Only 3% and 13% of patients completed all elements of the UgWHO03 and WHO07, respectively, with even fewer completing them on time. In particular, many of these HIV-positive patients could not produce sputum. Many of the respondents were low-income earners and lived long distances from the hospital. They experienced difficulties with travel, which limited the frequency with which they visited the hospital. Some clinicians found it difficult to adhere to the UgWHO03 as it was felt it might delay the diagnostic process unnecessarily. There were situations when CXRs were requested before start of diagnostic work up. This was driven by staff concerns and patient requests, especially where the patient had a relative working in the hospital. This may be indicative of low acceptability of the existing UgWHO03 amongst both staff and patients. An attempt was made to minimize loss to follow-up by tracing patients who failed to turn up at their appointments, but approximately one-third of patients in each algorithm failed to return. The high proportions of patients lost to follow-up for both algorithms are concerning but unsurprising given that this study was conducted under operational conditions without the addition of staff and resources. Similar studies conducted in resource-constrained settings have reported such high lost to follow-up rates.7,8 There may be feasible, sustainable, low-costs ways of improving the number of people who complete all elements of the algorithm in such resource-constrained settings. In situations where financial costs and long distances of travel prevent patients from submitting the required set of smears, community-based workers could be used to collect sputum samples from patients and deliver them at the hospital. A number of nongovernmental community-based health care programs such as our other (urban) study site employ this strategy with good outcomes.

Our study was subject to several important limitations. There were many operational constraints, which are likely to occur in other HIV prevalent resource-constrained settings and should be taken into account when introducing new guidelines for TB diagnosis. The sample size was limited by exclusions, though our study was originally powered to detect a significant reduction of time to diagnosis for smear-negative PTB with the introduction of the new algorithm. Overall default rate was high, but then defaulters had data on their progress through the diagnostic process with majority having received a final diagnosis before defaulting and therefore were included in the analyses. A number of participants had no sputum culture results and final diagnoses available which may have affected our sensitivity and specificity results. We did not perform a sensitivity analysis in this situation since this would mean imputing (ie, making up) outcome data for participants without both culture results and final diagnoses. There were more than 2 different possibilities for each outcome for each participant. The numbers without culture results and final diagnoses are not very high and therefore we believe the findings are robust though caution may still be needed when interpreting the findings. Another limitation was that although sputum culture is 99% specific, it is only 80% sensitive.15 Its use as the gold standard could have underestimated the specificity of both arms of the study. However, though it may have affected the absolute values, it did not affect the ability of the study to elicit the relative performance of both algorithms. It also does not impede comparability to data from other studies as culture is used as gold standard in all literature reviewed in such settings. A further limitation is the prospective observational design of the study as it limits the ability to exclude confounding factors.


In our study, the duration from presentation to diagnosis for smear-negative PTB for the WHO07 was shorter compared with the UgWHO03. Sensitivity was also higher, and the specificity was maintained. The new diagnostic pathway may expedite diagnosis of smear-negative PTB in resource-constrained settings and may also improve diagnostic accuracy, especially if applied rigorously. We have identified multiple barriers to the application of both sets of guidelines. Some of these barriers can be minimized by the use of the WHO07 algorithm with its fewer steps. Where the new WHO07 is implemented, these barriers will need to be addressed as part of implementation, which will ultimately improve overall sensitivity and time to diagnosis.


The Communicable Disease Research Programme Consortium led by the Nuffield Centre at Leeds University which itself is funded by the Department for International Development, United Kingdom, appreciates the financial support received from the WHO which made this study possible. The authors wish to thank Dr. Haileyesus Getahun from Stop TB Department of WHO for the immense technical support offered from the design stage of the study to preparation of the manuscript. We acknowledge the Makerere University, College of Health Sciences, Faculty of Medicine, Department of Medicine, which hosted the project offices and the Ugandan Ministry of Health National Tuberculosis and Leprosy Programme, National Tuberculosis Reference Laboratory, and National AIDS Control Programme for the support and facilitation of this study. We extend our profound gratitude to Dr Harriet Kisembo of the Department of Radiology, Mulago National Referral Hospital, for the invaluable support offered during the course of the study. The study team also acknowledges the enthusiasm and immense contributions of the 2 research assistants, Ernest Kabizwe and Nicholas Ssendage and their coworkers. We thank the management and health workers of Kayunga Hospital where this study was conducted. Special thanks go to all clients, their relatives, and members of the community at Kayunga who willingly participated in the study. Without their input, this study would not have been possible. Last, Communicable Disease Research Programme is grateful to the Malaria Consortium, Uganda, for providing timely operational support to the study.


1. WHO. WHO Tuberculosis factsheet, 2006. Available at: Accessed August 10, 2010.
2. WHO. WHO Report 2009, Global Tuberculosis Control—epidemiology, strategy, financing (WHO/HTM/TB/2009.411). Available at: Accessed August 17, 2010.
3. Aber VR, Allen BW, Mitchison DA, et al. Quality control in tuberculosis bacteriology: 1, laboratory studies on isolated positive cultures and the efficiency of direct smear examination. Tubercle. 1980;61:123-133.
4. Editorial. Smear-negative pulmonary tuberculosis. Tubercle. 1980;61:113-115.
5. Kim TC, Blackman RS, Heatwole KM, et al. Acid-fast bacilli in sputum smears of patients with pulmonary tuberculosis. AM Rev Respir. 1984;129:264-268.
6. Siddiqi K, Lambert ML, Walley J. Clinical diagnosis of smear-negative tuberculosis in low-income countries: the current evidence. Lancet Infect Dis. 2003;3:288-296.
7. Apers L, Wijarajah C, Mutsvangwa J, et al. Accuracy of routine diagnosis of pulmonary tuberculosis in an area of high HIV prevalence. Int J Tuberc Lung Dis. 2008;8:945-951.
8. Hargreaves NJ, Kadzakumanja O, Whitty CJM, et al. ‘Smear-negative’ pulmonary tuberculosis in a DOTS programme: poor outcomes in an area of high HIV seroprevalence. Int J Tuberc Lung Dis. 2001;5:847-854.
9. WHO. Treatment of Tuberculosis: Guidelines for National Programmes. Geneva, Switzerland: World Health Organization; 2003.
10. Wilkinson D, De Cock KM, Sturm AW. Diagnosing tuberculosis in a resource-poor setting: the value of a trial of antibiotics. Trans R Soc Trop Med Hyg. 1997;91:422-424.
11. Siddiqi K, Walley J, Khan MA, et al. Clinical guidelines to diagnose smear-negative pulmonary tuberculosis in Pakistan, a country with low HIV prevalence. Trop Med Int Health. 2006;11:323-331.
12. WHO. Improving the diagnosis and treatment of smear-negative pulmonary and extra-pulmonary tuberculosis among adults and adolescents: recommendations for HIV-prevalent and resource-constrained settings. March 13, 2007 (WHO/HTM/HIV/2007.01). Available at: Accessed August 17, 2010.
13. van Cleeff MR, Kivihya-Ndugga L, Githui W, et al. A comprehensive study of the efficiency of the routine pulmonary tuberculosis diagnostic process in Nairobi. Int J Tuberc Lung Dis. 2003;7:186-189.
14. WHO STRATEGIC AND TECHNICAL ADVISORY GROUP FOR TUBERCULOSIS (STAG-TB). Recommendations to improve the diagnosis of smear negative pulmonary and extrapulmonary TB amongst adults in resource constrained settings. Draft for discussion by strategic and Technical Advisory Group of Stop TB Department of WHO. June 2006. Available at: Accessed August 23, 2010.
15. Magee JG, Freeman R, Barret A. Enhanced speed and sensitivity in the cultural diagnosis of pulmonary tuberculosis with a continuous automated mycobacterial liquid culture (CAMLiC) system. J Med Microbiol. 1998;47:547-553.
16. WHO. New technologies for TB control: a Framework for their adoption, introduction and implementation. (WHO/HTM/STB/ 2007. 40). Available at:
17. WHO endorses new rapid tuberculosis test. A major milestone for global TB diagnosis and care. WHO, December, 2010. Available at:
18. Parry CM. Sputum smear negative pulmonary tuberculosis. Trop Doct. 1993;23:145-146.
19. Anderson C, Inhaber N, Menzies D. Comparison of sputum induction with fiber-optic bronchoscopy in the diagnosis of tuberculosis. Am J Respir Crit Care Med. 1995;152:1570-1574.
20. Habeenzu C, Lubasi D, Fleming AF. Improved sensitivity of direct microscopy for detection of acid-fast bacilli in sputum in developing countries. Trans R Soc Trop Med Hyg. 1998;92:415-416.
21. Gebre N, Karlsson U, Jonsson G, et al. Improved microscopical diagnosis of pulmonary tuberculosis in developing countries. Trans R Soc Trop Med Hyg. 1995;89:191-193.
22. Wilkinson D, Sturm AW. Diagnosing tuberculosis in a resource-poor setting: The value of sputum concentration. Trans R Soc Trop Med Hyg. 1997;91:420-421.
23. Toyota M. Validity of “the risk index” to predict the infectiousness of tuberculosis patients [in Japanese]. Kekkaku. 1994;69:375-377.
24. Ipuge YA, Rieder HL, Enarson DA. The yield of acid-fast bacilli from serial smears in routine microscopy laboratories in rural Tanzania. Trans Roy Soc Trop Med Hyg. 1996;90:258-261.
25. Nelson SM, Deike MA, Cartwright CP. Value of examining multiple sputum specimens in the diagnosis of pulmonary tuberculosis. J Clin Microbiol. 1998;36:467-469.
26. Rieder HL, Chonde TM, Myking H, et al. The Public Health Service National Tuberculosis Reference Laboratory and the National Laboratory Network: Minimum Requirements, Role and Operation in a Low-Income Country. Paris, France: International Union Against Tuberculosis and Lung Disease; 1998.

algorithm; diagnostic; HIV/AIDS; pulmonary tuberculosis; smear-negative

Supplemental Digital Content

© 2011 Lippincott Williams & Wilkins, Inc.