• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Chronic Conditions (UK); Centre for Clinical Practice at NICE (UK). Tuberculosis: Clinical Diagnosis and Management of Tuberculosis, and Measures for Its Prevention and Control. London: National Institute for Health and Clinical Excellence (UK); 2011 Mar. (NICE Clinical Guidelines, No. 117.)

5Diagnosis

Updated: 2011.

5.1. Diagnosing latent tuberculosis

5.1.1. Clinical introduction

In asymptomatic persons exposure to, and potential infection with, tuberculosis is demonstrated by a positive skin test, or more recently from a positive blood-based immunological (interferon gamma) test. Those with a strongly positive skin test are then regarded as having been infected with tuberculosis. Of these people presumed infected, there is a 10–15% chance of developing clinical disease at some point in their lives. If a co-morbidity develops which reduces the immune system (see section 10.2), that risk is increased. The majority of exposed persons will kill off the inhaled bacteria, and be left only with a positive skin test as a marker of exposure. About half of those who develop the clinical disease will do so within five years of the initial infection. In cases where a long period elapses between infection and development of disease, dormant bacilli are thought to remain in either the lung or other sites, which can ‘reactivate’ in favourable circumstances for the organism.

Until recently, only Mantoux tests were available to give evidence of exposure. The tuberculin tests had the advantage of being cheap and relatively easy to perform, but suffered from a number of problems. The test results have to be interpreted within a certain timescale, and patients who do not return, or delay returning, will have either no result or a possibly inaccurate one. False positive results can occur because of the sensitising effect on the immune system of either prior BCG vaccination or opportunist environmental mycobacteria. False negative results can occur due to anything reducing immunity, particularly co-infection with HIV, but also treatments such as cytotoxics, or immunosuppression. Extensive tuberculosis (pulmonary or miliary) can itself also temporarily depress the immunity, and can lead to a paradoxically negative Mantoux tests. More recently, selective immunological (interferon gamma) tests have been developed using the tuberculosis antigens ‘early secretion antigen target 6’ (ESAT-6) , ‘culture filtrate protein 10’ (CFP-10) and tb7.7, which are not present in BCG, and are found in only a few species of environmental mycobacteria. These can be done on either cells or cell products derived from whole blood tests. These tests aim to be more specific by removing false positive results, and to be better correlated with latent infection or dormant organisms.

5.1.2. Methodological introduction

Because there is now additional evidence available on the use of IGT, the partial update of CG33 sought to make recommendations on the use of IGT for diagnosis of latent TB.

There are currently three interferon gamma immunological tests commercially available for use in the UK: QuantiFERON-TB Gold, QuantiFERON-TB Gold In tube and T-SPOT.TB. QuantiFERON-TB Gold measures the release of interferon-gamma in whole blood in response to stimulation by ESAT-6 and CFP-10 which are not present in BCG vaccine strains or the vast majority of non-TB mycobacteria. The In tube version measures ESAT-6, CFP-10 and tb7.7 In the T-SPOT.TB test, individual activated ESAT-6 and CFP-10 specific T-cells are enumerated using ELISPOT methodology

In order to make appropriate recommendations, review questions were framed according to the following population groups: adults young people and children from high incidence countries, adults, young people and children who had been in contact with individuals with active TB, or immunocompromised individuals. Children were treated as a separate population because they have a less developed immune system than adults, and the mechanism of action of the tests relies on a fully developed immune system.

The key clinical questions considered were:

  1. Which diagnostic strategy is most accurate in diagnosing latent TB in adults, young people and children who are recent arrivals from high prevalence countries?
  2. Which diagnostic strategy is most accurate in diagnosing latent TB in children?
  3. Which diagnostic strategy is most accurate in diagnosing latent TB in adults, young people and children (children considered as a separate population) who have been in close contact with patients with active TB?
  4. Which diagnostic strategy is most accurate in diagnosing latent TB in immunocompromised patients?
  5. What is the effectiveness of screening using IGT for healthcare workers?

The review protocol is included in appendix B.

A search strategy was used which aimed to identify relevant studies for all the review questions. The following databases were searched: Cochrane database of systematic reviews (CDSR), Database of Abstracts of Reviews of Effects (DARE), Health technology assessment (HTA) database, Medline, Embase, Cinahl, NHS Economic Evaluation database (NHS EED). Trial registers such as Cochrane central register of controlled trials (CENTRAL), UKCRN Portfolio database, current controlled trials, clinicaltrials.gov were searched. Websites of relevant organisations such as World Health Organisation and TB alert were also searched. No methodology search filters or publication date filters were used. A total of 5270 studies were identified for the whole review. After sifting by abstract, 467 studies were selected (n = 56, 70, 69 ,153 and 5 for questions 1 to 5 respectively).

Studies were excluded if they:

  • did not compare Mantoux tests with IGT
  • evaluated IGT based on purified protein derivative
  • did not focus on latent TB
  • focused on treatment of TB
  • focused on non commercial IGT or in-house IGT.

The detailed evidence tables for the included studies and list of excluded papers and reasons for exclusion are given in appendices O and J.

There were methodological issues with the included papers. For example, active TB was not always excluded (either through investigation or not reported), there was repeated testing of both Mantoux and IGTs, the threshold for positive Mantoux tests varied, and it was not clear whether the use of cut-offs was always age appropriate. If identified, these issues were used to downgrade the quality of the evidence in the GRADE tables.

Diagnostic accuracy studies considered as high quality are those where the index test(s) are compared with a recognised, validated reference standard. Measures of accuracy, when compared with the reference test, such as sensitivity and specificity can then be determined. The Mantoux test has been the preferred test in clinical practice for several years but it is not an ideal reference standard; for example, the specificity of the Mantoux test is confounded by BCG vaccination. This implies false-positive results could be seen in this group of people because the Mantoux test is not able to distinguish between individuals who actually have the infection, and those who have been vaccinated with BCG. Because of such concerns about the Mantoux test as a reference standard, other measures of effect such as discordance, concordance and odds ratios are used. These measure the association between the results of the test(s) and the risk of having latent TB, but do not give any information on rates of false positives or negatives.

In addition, the GRADE methodology has not been fully developed for diagnostic studies. A modified form of GRADE was used to assess the quality of evidence found. Standard GRADE profiles for interventions use the following criteria to assess quality of evidence: limitations, inconsistency, imprecision and indirectness. In this review the same criteria were applied. Footnotes have been included to define and describe what the criteria mean in the context in which the studies were analysed. It was not possible to measure imprecision so this has been noted as ‘not measurable’ in the tables. This is because guidance has not yet been developed to address thresholds for imprecision for the measures of effect that were determined. These measures of effect did not appropriately describe the effectiveness of the diagnostic tools. Therefore, the GDG were not asked to agree a pre-defined threshold for imprecision. For questions on children and contact tracing it was possible to pool the ratio of odds ratios and to perform a meta analysis. The ratio of odds ratios is a measure of effect which reflects test performance and provides an approach to evaluating tests in the absence of a reference test. The odds ratio (OR) is a function of test sensitivity and specificity and increases as one or both of these measures increase. Statistically OR = [sensitivity/(1-specificity)]/[(1-sensitivity)/specificity].

The spreadsheets used to calculate and determine the risk categories as defined by level of exposure to active TB are given in appendices P and Q.

The main aim of this update was to review diagnosis of latent TB using tests for which there is no ideal reference standard for comparison. One important objective was to identify appropriate measures of effect to assess the diagnostic utility of the tests. Different approaches were taken to address this objective.

  • Discordance and concordance between the IGT and Mantoux tests were measured in some of the papers. There were few prospective studies to identify participants who would either develop active TB following a positive test result or stay healthy following a negative test result. These studies are designed to determine positive and negative predictive values. For diagnosis of latent TB this type of design would give the most accurate prognosis predicting those who will get active TB and those who would not.
  • In other studies the odds of a positive test associated with graded exposure to an active TB case were measured. In these cases a proxy measure of effect, the ratio of odds ratios could be calculated if figures of positive test results of study participants were clearly stated, and where the exposure status of those participants had been identified. The main disadvantage of this proxy measure is that it fails to identify whether the good performance of a test compared with another is because of either better sensitivity, specificity or both. It is impossible therefore to determine the false positive and false negative rates of a particular test.

5.1.3. Partial update health economics introduction

The following sections outline the updated modelling for two populations identified in the scope: adult contacts (including health care workers) and screening people from high prevalence countries. However, because of an absence of evidence, no cost-effectiveness analysis was conducted for all child and young people populations. Because of an absence of information no new distinct analysis was conducted for screening new NHS employees and the immunocompromised population. For children, the almost complete absence of sensitivity and specificity information and quality of life data meant that a useful analysis could not be produced. For the two remaining adult populations the results of the other two analyses will be extrapolated to these situations

A search for cost-effectiveness studies identified five relevant papers that examined the use of IGT in screening people from high prevalence countries with suspected latent TB infection, and one relevant paper that examined the use of IGT in the adult contacts and healthcare workers contacts with suspected latent TB infection. The papers were reviewed with quality checklists to assess their applicability and limitations. A completed checklist is available in annex 6 in appendix L. None of the papers were considered applicable to the decision problem either because they were not based in the UK or did not include consideration of quality of life. However cost-effectiveness papers were used to explore approaches to modelling strategies and to inform the structure of the model.

A decision model based on the previous guideline was used to compare the expected cost effectiveness of four strategies of testing for latent infection in both adult (aged more than 18 years) populations described above. The strategies compared were:

In the model, treatment follows current policy; with appropriate therapy for people diagnosed with active and latent TB. The analysis did not compare different types of skin tests or IGTs because this was outside the scope of this guideline.

The key areas that were updated were the test accuracies and the relevant costs. All costs were updated to current prices and were validated by the GDG. The test accuracies were based on published reviews which calculated sensitivities and specificities again after validation by the GDG.

The assumptions made in the initial guideline were still applicable unless stated otherwise. Whenever possible, input parameters and assumptions were based on empirical evidence, but some key parameters were estimated by the health economist and GDG. The model considers the quality-adjusted life years (QALYs) lost because of infection, adverse events and developing TB. Therefore, the interventions with the smallest QALY loss are the most effective. Throughout the analysis incremental cost-effectiveness ratios (ICERs) will be compared with a common base line (usually no test) and net monetary benefits will be calculated. Net monetary benefit quantifies which treatment option provides the greatest health benefit for a given threshold. A threshold of £20,000 per QALY gained was used in this analysis. Probabilistic sensitivity analysis was considered, however some of the estimates of the means of variables were assumptions and it was therefore considered more instructive to do a series of one way sensitivity analysis rather than a probabilistic sensitivity analysis.

For each population details were given on the source of the new test accuracy data with base-case results and sensitivity analyses.

5.1.4. Diagnosis of latent TB in people who are recent arrivals from countries where TB is highly prevalent

Key clinical question

Which diagnostic strategy is most accurate in diagnosing latent TB in adults and children who are recent arrivals from highly prevalent countries?

Evidence review

Of the ten studies included:

All studies looked at participants from high prevalence countries from places such as sub Saharan Africa, Central and South America, Eastern Europe and Asia (Table 5).

Table 5. Diagnosis of latent TB infection in foreign-born people and in people arriving from high-prevalence countries.

Table 5

Diagnosis of latent TB infection in foreign-born people and in people arriving from high-prevalence countries.

The main measures of effect used (Table 6) were:

Table 6. Degree of concordance between Mantoux tests and IGT and corresponding threshold for Mantoux test.

Table 6

Degree of concordance between Mantoux tests and IGT and corresponding threshold for Mantoux test.

  • concordance and discordance between tests
  • agreement between the tests as measured by kappa values
  • odds ratios
  • ratio of odds ratios (ROR). In this guideline ROR is mathematically defined as (odds of positive IGT in a high-risk area divided by the odds of a positive test in a low-risk area) divided by (odds of a positive Mantoux test in a high-risk area divided by a positive Mantoux test in a low-risk area)

Evidence statements

Low quality evidence from four studies with 2646 participants showed that there was a higher level of concordance and agreement between IGT and Mantoux test when both tests were used in non-BCG-vaccinated populations than in populations who were BCG vaccinated.

Low quality evidence from three studies with 2351 participants showed that BCG vaccination decreased both concordance and agreement between the assay results of IGT and Mantoux tests.

Low quality evidence from one study showed IGTs were more likely to detect progression to active TB than Mantoux tests over a 2-year period. Positive predictive values were 14.6% and 2.3% respectively.

Low quality evidence from one study following up 339 immigrant contacts for a median of 1.83 years showed that IGT and Mantoux tests were similar in detecting progression to active TB. Positive predictive values were 3.1% and 3.8% for Mantoux test thresholds of 10 mm and 15 mm and 2.8% and 3.3% for QFT and T-SPOT. Negative predictive values for the Mantoux test thresholds of 10 mm and 15 mm, and QFT and TSPOT were 100%, 99.3%, 98% and 98.3% respectively.

Very low quality evidence from four studies with 1636 participants showed very low levels of concordance between the Mantoux and IGTs in BCG-vaccinated populations

Health economics – diagnosing latent TB in adults and children who are recent arrivals from high prevalence countries

The published reviews of test accuracy identified were Pai et al. (2008) and Diel et al. (2010). Both use active TB as a proxy for the calculation of sensitivities and specificities. Because there was no differentiation between IGTs, midpoints were used for the accuracy estimates.

The base-case analysis is shown in table 7. It used a prevalence of 30% for latent TB in the cohort group. These results demonstrate that Mantoux tests/IGT and IGT are associated with ICERs which are just under £30,000 per QALY. These estimates are within a range that means NICE requires further consideration of the various input parameters before a decision can be made.

Table 7. Cost-effectiveness results for new entrants from high prevalence countries.

Table 7

Cost-effectiveness results for new entrants from high prevalence countries.

A number of sensitivity analyses were run and are presented in appendix L. The prevalence of latent TB in this population and the transformation rate of latent TB to active TB are presented in tables 8 and 9 because the GDG considered them to be two of the key parameters in the model. The net monetary results at £20,000 per QALY are presented in table 8.

Table 8. Net monetary benefits at £20,000 per QALY gained for different prevalence rates and test accuracy sources for screening people from high prevalence countries.

Table 8

Net monetary benefits at £20,000 per QALY gained for different prevalence rates and test accuracy sources for screening people from high prevalence countries.

Table 9. Net monetary benefits at £20,000 per QALY gained for different transformation rates and test accuracy sources for screening people from high prevalence countries.

Table 9

Net monetary benefits at £20,000 per QALY gained for different transformation rates and test accuracy sources for screening people from high prevalence countries.

These results suggest that as the prevalence of TB and the conversion rate of TB increase the tests (Mantoux test/IGT and IGT alone) will be cost effective. IGT appears to be the optimum choice based on cost effectiveness. However, the results indicate that relatively small differences in either the prevalence or the transformation rate could result in Mantoux test/IGT being the optimum choice. In addition, the deterministic ICER per QALY gained for Mantoux test/IGT suggests it is a cost-effective option.

Evidence to recommendations

The issue of generalisability of the studies to the UK population was raised as well as how the results could be applied to a UK setting. It was agreed that the studies had similar settings and prevalence figures to the UK. The GDG noted that IGT was being used in certain UK practices. The evidence presented was of low quality but it showed how a previous BCG vaccination would confound the Mantoux test results and not affect the IGT results. The GDG felt that good quality evidence to predict active TB in the future was required.

Evidence to recommendations – health economics (people who have arrived from high-prevalence countries)

Health economic analysis indicated that none of the tests were associated with ICERs of below £20,000 per QALY gained. However the GDG considered that the mean rate of transformation from latent TB to active TB was an underestimate and that the true rate was closer to 16% over 15 years; evidence from Kik et al. (2010) suggested equivalent rates of close to 3% over 2 years. At estimates this high, IGT alone is the most cost-effective option, followed by the Mantoux test/IGT dual strategy. The threashold for screening was reduced from 500/100,000 to 40/100,000 as the GDG considered this to be cost effective and provided the greatest health benefits. The GDG considered that while IGT alone appeared to be the most cost-effective option, the dual strategy should remain as an alternative because there was significant uncertainty in the point estimates, it was a less expensive strategy that would be more effective in low incidence areas and, in particular, there were still issues over the operation of the tests and intersubject variability.

5.1.5. Diagnosis of latent TB in children

Key clinical question

Which diagnostic strategy is most accurate in diagnosing latent TB infection in children?

Evidence review

Of the 11 studies included:

The studies also looked at other factors such as BCG vaccination and country of birth.

Exposure was measured in several ways:

  • duration of contact

    hours/day

    hours/week

  • sleeping proximity

    same or different house

    same or different room

  • type of contact

    household/close

    non-household

    unknown

    school

    casual.

The following measures of effect were used:

  • concordance between tests
  • agreement between tests measured by kappa value
  • risk factors for positive test result
  • odds ratios.

Risk of development of active TB

Meta analysis of the results of a positive test associated with graded exposure to active TB was performed from six studies (Brock et al. 2004; Chun et al. 2008; Hansted et al. 2009; Higuchi et al. 2009; Lighter et al. 2009; Okada et al. 2008) (Table 10 and Figure 2).

Table 10. Diagnosis of latent TB in children.

Table 10

Diagnosis of latent TB in children.

Figure 2. Forest plot of meta-analysis of IGT and Mantoux test results based on high-risk and low-risk exposure.

Figure 2

Forest plot of meta-analysis of IGT and Mantoux test results based on high-risk and low-risk exposure. Both OR and ROR in this context, reflect test performance and provide an approach to evaluating tests in the absence of a reference test. OR is a function (more...)

There were two longitudinal studies (Higuchi et al. 2007; Higuchi et al. 2009) that followed up participants to investigate the development of active TB (Table 11).

Table 11. Diagnosing latent TB in children (predicting development of active TB).

Table 11

Diagnosing latent TB in children (predicting development of active TB).

Five studies (Anon ; Brock et al. 2004; Chun et al. 2008; Connell et al. 2006; Okada et al. 2008) looked at the concordance between IGTs and Mantoux tests (Table 12).

Table 12. Diagnosis of latent TB in children (agreement between tests).

Table 12

Diagnosis of latent TB in children (agreement between tests).

Evidence statements

Moderate quality evidence from six studies with 935 children aged 0–18 years showed that a positive IGT was more strongly associated with increasing TB exposure than a positive Mantoux test (ratio of odds ratio 2.86 [95% CI 1.56 to 5.23]).

Low quality evidence from two studies that followed up 281 children aged 8–16 years who had a negative IGT test found that none had developed active TB within 888.5 person–years. Each child had been followed up for an average of just over 3 years. All the children had tested positive with a Mantoux test but 99% were BCG vaccinated. The studies were from the same group in Japan.

Moderate quality evidence from two studies with 110 children found that there was a low-to-moderate level of concordance between IGTs and Mantoux tests but a high level of concordance between the two commercial IGTs.

Low quality evidence from five studies with 461 children aged 0–18 years showed a wide variation in concordance between IGTs and Mantoux tests (kappa values ranging from 0.19 to 0.866). These studies were conducted in very diverse populations with different rates of BCG vaccinations and wide age ranges.

Evidence to recommendations

Because of their underdeveloped immune system, children would be more likely to develop active and more serious disease if they had latent infection. This risk is greater in children aged under 5 years.This could lead to disability or death depending on the location of the infection. The GDG observed that the evidence presented that determined the negative predictive values of the tests was of very low quality. It also felt that the generalisability of those studies could be an issue especially with regard to the BCG vaccination program in Japan. It was agreed that most paediatricians would choose to treat a high-risk child if they had a positive Mantoux test and negative IGT because there was very limited evidence to suggest that a negative IGT could completely exclude infection. The difficulty of phlebotomy and obtaining enough blood in children was discussed, generally in those under five years of age and especially when they are under two years. Indeterminate IGT results occur more frequently in younger children. The GDG was of the view that IGTs perform less well in younger children. The group also agreed that careful consideration should be given to high-risk young children, especially those aged under 5 years because false-negative results could have substantial implications.

5.1.6. Diagnosis of latent TB in people who have been in close contact with a person with active TB

Key clinical question

Which diagnostic strategy is most accurate in diagnosing latent TB in people who have been in close contact with a person with active TB?

Evidence review

Of the 27 papers selected:

Table 13. Diagnosing latent TB in people who have been in close contact with a person with active TB.

Table 13

Diagnosing latent TB in people who have been in close contact with a person with active TB.

Figure 3. Forest plot of meta-analysis of IGT and tuberculin skin test results based on high-risk and low-risk exposure to active TB.

Figure 3

Forest plot of meta-analysis of IGT and tuberculin skin test results based on high-risk and low-risk exposure to active TB. Both OR and ROR in this context, reflect test performance and provide an approach to evaluating tests in the absence of a reference (more...)

Figure 4. Forest plot of meta-analysis of IGT and tuberculin skin test results based on high-risk and low-risk exposure to active TB stratified by BCG vaccination rates.

Figure 4

Forest plot of meta-analysis of IGT and tuberculin skin test results based on high-risk and low-risk exposure to active TB stratified by BCG vaccination rates. BCG = Bacille Calmette-Guerin. CI = confidence interval. IGT = interferon gamma test. IV = (more...)

Table 14. Diagnosis of latent TB in people who have been in close contact with a person with active TB (concordance between results).

Table 14

Diagnosis of latent TB in people who have been in close contact with a person with active TB (concordance between results).

Evidence Statement

Low quality evidence from 11 studies with 1844 participants showed that positive IGTs were more strongly associated with increasing TB exposure than positive Mantoux tests (ROR = 1.54 [95% CI 1.08 to 2.19]). In those studies with less that 50% BCG-vaccinated patients the ratio of odds ratio was 1.25 (95% CI 0.94 to 1.67), whereas in those with over 50% BCG-vaccinated patients it was 2.07 (95% CI 1.23 to 3.48).

Low quality evidence from 16 studies showed that the degree of concordance between Mantoux test and IGT results, as measured by kappa values, was between 0.11 and 0.85.

Low quality evidence from one study showed IGTs were more likely to detect progression to active TB than Mantoux tests over a 2-year period. Positive predictive values were 14.6% and 2.3% respectively.

Low quality evidence from one study following up 339 immigrant contacts for a median of 1.83 years showed that IGTs and Mantoux tests were similar in detecting progression to active TB. Positive predictive values were 3.1% and 3.8% for Mantoux test thresholds of 10 mm and 15 mm and 2.8% and 3.3% for QFT and T-SPOT. Negative predictive values were 100%, 99.3%, 98% and 98.3% respectively.

Evidence to recommendations

The population of this group included healthcare workers who were in contact with people with active TB and non healthcare workers, who by way of residence, had been in close contact with a person with active TB. The GDG was presented with evidence showing the meta-analysis of ROR for comparing IGTs with Mantoux tests. This was stratified by percentage BCG vaccination. When adjusted for BCG vaccination, IGTs showed a better ROR than Mantoux tests. The GDG felt that although IGTs seemed better from ROR, the evidence was of poor quality and that recommendations should ideally be based on longitudinal studies that aimed to determine positive and negative predictive values of a person developing active TB.

Evidence to recommendations – health economics (contacts)

The health economic analysis for contacts was extrapolated to this population. This analysis indicated that there was uncertainty over which testing strategy was the optimal choice. Therefore, the GDG considered that both tests should be offered and that depending on operational issues, the most appropriate should be used.

5.1.7. Diagnosis of latent TB in people who are immunocompromised

Key clinical question

Which diagnostic strategy is most accurate in diagnosing latent TB in people who are immunocompromised?

Evidence review

Of the 16 papers selected:

Table 15. Diagnosis of latent TB in patients who are immunocompromised.

Table 15

Diagnosis of latent TB in patients who are immunocompromised.

Table 17. Diagnosis of latent TB in people who are immunocompromised (indeterminate results).

Table 17

Diagnosis of latent TB in people who are immunocompromised (indeterminate results).

Table 16. Diagnosis of latent TB in children who are immunocompromised.

Table 16

Diagnosis of latent TB in children who are immunocompromised.

Table 18. Diagnosis of latent TB in people with rheumatoid arthritis who are immunocompromised.

Table 18

Diagnosis of latent TB in people with rheumatoid arthritis who are immunocompromised.

Table 19. Diagnosis of latent TB in people who are immunocompromised (association between risk factors and positive test).

Table 19

Diagnosis of latent TB in people who are immunocompromised (association between risk factors and positive test).

Table 20. Diagnosis of latent TB in people with haematological conditions who are immunocompromised.

Table 20

Diagnosis of latent TB in people with haematological conditions who are immunocompromised.

Evidence Statement

Low quality evidence from five studies showed that the level of discordance between IGTs and Mantoux tests in 973 adults with HIV ranged from 0% to 29.7% for negative Mantoux tests/positive IGTs and 1.8% to 28.6% for positive Mantoux tests/negative IGTs.

Low quality evidence from one study showed that in 23 children with HIV (mean age of 4 years) the positive Mantoux tests/negative IGTs discordance ranged from 13% to 25% and negative Mantoux tests/positive IGTs discordance ranged from 0 to 39.1% similar overall discordance

Low quality evidence from three studies showed that the rate of indeterminate results from an IGT test in 837 people with HIV ranged from 1.83% to 17.87%. The rate of indeterminate results was significantly higher in those with a CD4 count below 200cells/mm3.

Low quality evidence from seven studies showed that in 1121 individuals with rheumatoid arthritis, the overall discordance between IGTs and Mantoux tests was between 5.9% and 47.5% for positive Mantoux tests/negative IGTs, and 1.6% to 23.7% for negative Mantoux tests/positive IGTs.

Low quality evidence from two studies showed that the level of discordance in patients with diseases including, chronic liver disease, non-Hodgkin's lymphoma, multiple myeloma, acute myeloid leukaemia, and chronic myeloma was between 6.4% and 29.6% for negative Mantoux tests/positive IGTs, and 2.6% and 8.5% for positive Mantoux tests/negative IGTs.

Evidence to recommendations

The GDG pointed out that it was important to differentiate the groups of people who were immunocompromised. The group agreed that the degree and type of immunosuppression was also important. There was general agreement that the evidence was of low quality. There was a lot of discordance between the tests in the immunocompromised population, but in general IGTs may identify more truly positive latent TB infections than Mantoux tests but the value of such tests varies with the nature and the degree of immunosuppression. The group discussed the stratification of some of the HIV studies by CD4 count and agreed on the basis of the evidence presented that a CD4 count below 200cells/mm3 was significantly associated with an indeterminate result. The group also strongly felt that people with HIV who have a CD4 count of 500cells/mm3 or more should be tested in the same way as people who are immunocompetent because the tests would perform in a similar way in these two groups of people. Evidence that looked at the effect specific anti-TNFalpha medications had on the diagnosis of latent TB was not identified.

Evidence to recommendations – Health economics (immunosuppresion)

No health economic modelling was conducted in this patient group. However the modelling for contacts and people from high prevalence countries indicated that high rates of transformation from latent TB to active TB and worse outcomes would all result in improved cost-effectiveness estimates for the testing strategies.

5.1.8. Screening for latent TB in healthcare workers

Key clinical question

What is the effectiveness of screening using IGT for healthcare workers?

Evidence review

Although studies that included healthcare workers had been analysed as part of the contact tracing question (section 5.1.6), the GDG advised that screening in healthcare workers should be specifically looked at. This was because the GDG felt that the scope was open to interpretation with regard to pre-employment screening in the NHS. It was difficult to identify studies that were screening for latent TB in healthcare workers. Good quality studies would have been those which compared participants who had been screened for latent TB and offered treatment as appropriate with those who had not and followed up to determine those who developed active TB. No such studies were identified.

Five studies were selected for critical appraisal. Of these:

The evidence from these screening studies was of very low quality. Most of the issues had already been addressed and analysed in the contact tracing question. Table 21 summarises this evidence.

Table 21. Effectiveness of IGTs for screening healthcare workers.

Table 21

Effectiveness of IGTs for screening healthcare workers.

Evidence statements

Evidence from three low quality papers showed that there was more discordance between positive Mantoux tests/negative IGT results than negative Mantoux tests/positive IGT results in 381 healthcare workers. Negative Mantoux tests/positive IGTs discordance was very low (less than 2%). Some of the healthcare workers were newly employed. Coverage and timing of BCG vaccination was variable. In two other studies discordance figures were not quantified.

Evidence to Recommendations

The GDG agreed that the level of evidence for screening studies was low. It also considered that healthcare workers would fall into the category of people from high prevalence countries or individuals who had had contact with a person with active TB. They made recommendations based on the evidence from those populations. For healthcare workers who were immunocompromised, the recommendations for the immunocompromised group applied.

Health economics – Contact tracing for healthcare workers (this section also relates to the diagnosis of latent TB in people who have been in close contact with a person with active TB)

The economic model used the same structure, costs and health-related quality of life values as those in the model for adults from high prevalence countries. However, the difference is in the estimates of the test accuracy and the prevalence of latent TB infection in this cohort. The test accuracy was based on Girardi et al. (2009) and Diel et al. (2010). The baseline prevalence used was 20%.

The model assumed the treatment regimen was the same as for people from high prevalence countries and that diagnosing and screening for latent TB was done in an outpatient setting.

The base case analysis for this population is shown in table 22.

Table 22. Cost-effectiveness of testing strategies for contacts.

Table 22

Cost-effectiveness of testing strategies for contacts.

These results indicate that Mantoux test/IGT and IGT alone are both cost-effective testing options and that depending on the test accuracies used either option could be the optimum choice.

Table 23 presents sensitivity analysis on the prevalence of latent TB in this conctacts population. The transformation rate did not appear to be a major variable in the model. Results are reported as net monetary benefits at the £20,000 per QALY gained threshold.

Table 23. Net monetary benefits at £20,000 per QALY gained for different prevalence rates and test accuracy sources for contact tracing.

Table 23

Net monetary benefits at £20,000 per QALY gained for different prevalence rates and test accuracy sources for contact tracing.

At £20,000 per QALY gained the prevalence has to be over 10% for testing to be cost effective. At a £30,000 per QALY gained threshold the lowest prevalence rate that testing remains cost effective at is 6%. In the contacts model, the transformation from latent to active TB was implemented by a relative risk (please see 2006 guideline appendix L for more details) the net monetary results at £20,000 per QALY gained are presented in table 24.

Table 24. Net monetary benefits at £20,000 per QALYgained for different transformation rates and test accuracy sources for contact tracing.

Table 24

Net monetary benefits at £20,000 per QALYgained for different transformation rates and test accuracy sources for contact tracing.

These results indicate that if the risk of latent TB becoming active is high then the cost-effectiveness results improve for all the options.

These results also indicate that IGT or Mantoux test/IGT could be the optimum choice but that it is highly dependent on the prevalence of latent TB in the population.

Evidence to recommendations – health economics (healthcare workers and screening)

Testing for healthcare workers who have come into contact with someone with active TB should follow the recommendations for all people who have been in contact with a person with active TB.

No specific health economic modelling was conducted for this population group. However, evidence from the high prevalence country and contacts analysis indicates that the testing strategies may be cost effective because the outcomes of a healthcare worker contracting TB might be more significant than a regular adult contact. Therefore, given the uncertainties in the model and difference in local circumstances both tests should be offered.

Summary evidence to recommendations

IGTs showed little evidence of being affected by prior BCG vaccination, and showed stronger correlation with exposure categories than the Mantoux tests. This was shown in high prevalence groups and in those who have been in contact with a person with active TB. The specificity of IGTs seemed better, and there was less potential for false-positive results. It was not possible to determine, for either Mantoux tests or IGTs, the rate of false-negative results. The GDG felt that some people with false-negative results would develop active TB and therefore reduce the cost effectiveness of vaccination and treatment of latent TB infection.

High quality prospective studies in people with latent TB (as diagnosed by positive IGTs) found at TB contact tracing and new entrant screening, have not yet been performed to find what proportion of such persons went on to develop clinical disease.

Economic modelling was undertaken with various strategies from no action to a two-step strategy with either Mantoux tests followed by interferon gamma testing, or serial IGTs. Of these options, the model provided most support, on grounds of cost-effectiveness, for a two-step approach with an initial Mantoux test, followed by an interferon-gamma test to confirm positivity. The GDG members also supported this because of clinical utility and feasibility.

In the studies evaluated, IGT show a stronger correlation with exposure than Mantoux tests. Much of the discordance between a positive Mantoux test and a negative IGT can be accounted for by prior BCG vaccination. The GDG agreed that in the absence of good quality longitudinal studies the relative benefit of IGT over Mantoux test in determining the need for treatment of latent infection is not certain. However they made recommendations in populations where they considered IGT to be of clear benefit especially in cases where IGT would reduce the uncertain diagnosis of Mantoux tests.

No further evidence was reviewed for other groups such as Prisoners/prison staff and nursing homes. However, the GDG felt that the tests should perform as with any other adults.

5.1.9. Evidence statements

Test results and TB exposure

In a UK study{17} of healthy adults in a contact tracing clinic, IGT (ESAT-6 ELISPOT assay) results had a strong positive relationship with increasing intensity of contact exposure (OR 9.0 per unit increase in exposure, 95%CI 2.6 to 31.6, p=0.001), whereas Mantoux test results had a weaker relationship with exposure (OR 1.9, 95%CI 1.0 to 3.5, p=0.05). (2)

In a study{11} of students aged 11–15 years in the UK from the same school as an index case, the odds of a test result being positive for each increase across four stratified exposure groups increased by a factor of 2.78 (95%CI 2.22 to 3.48, p<0.0001) for the IGT (ESAT-6/CFP10 ELISPOT assay) and 2.33 (95%CI 1.88 to 2.88, p<0.0001) for the Mantoux test. The IGT correlated significantly better with increasing exposure across the four groups than the Mantoux test (p=0.03). The odds of a positive IGT result increased by a factor of 2.51 (95%CI 1.58 to 3.99, p<0.0001) with each week of direct exposure, which was significantly higher (p=0.007) than that for the Mantoux test (OR 1.30, 95%CI 1.10 to 1.54, p=0.002). (2)

In contacts of index cases in the Gambia,{13} with increasing M. tuberculosis exposure, the percentage of participants who were tuberculin positive and interferon gamma test (ESAT-6/CFP-10 ELISPOT assay) negative increased from 11% of those sleeping in a different house from the index case to 32% of those sleeping in the same room (p<0.001). (3)

In contacts of an index case on an Italian maternity unit,{19} the odds for a test result being positive for each increase across four stratified exposure groups (from no discernible contact to household contacts) increased by 1.93 (95%CI 1.11 to 3.35, p=0.020) for the IGT (ESAT-6/CFP-10 ELISPOT assay) but there was no significant correlation for the Mantoux test. (3)

In Korea where BCG vaccination is mandatory,{15} a study found that the odds of a positive test result per unit increase in exposure across four groups, increased by a factor of 5.31 (95%CI 3.62 to 7.79) for the IGT (QuantiFERON-TB Gold) and by a factor of 1.52 (95%CI 1.2 to 1.91) for the Mantoux test (p<0.001). (2)

Test results and BCG status

Healthy adults in a contact tracing clinic in the UK,{17} had IGT (ESAT-6 ELISPOT assay) results which were not correlated with BCG vaccination status whereas Mantoux test results were significantly more likely to be positive in BCG vaccinated contacts (OR 12.1, 95%CI 1.3 to 115.7, p=0.03). (2)

Students aged 11–15 years from the same school as an index case in the UK{11} had IGT (ESAT-6/CFP-10 ELISPOT assays) which showed no significant relation with BCG vaccination status, however, BCG vaccinated children were significantly more likely to have higher Heaf grades than unvaccinated children (p=0.002). (2)

In a UK study{16} of healthy household contacts and healthy unexposed controls, ESAT-6 peptide-specific interferon gamma-secreting cells were detected in 85% of the healthy household contacts who were tuberculin positive. None of the healthy control subjects without a history of TB exposure, responded to this IGT even though all unexposed control subjects were BCG vaccinated. (3)

Mantoux test negative Australian born medical students (or those born in another low TB prevalence country),{14} with no prior BCG, and no known exposure to TB, were BCG vaccinated and then tested again at five months. ESAT-6 stimulated interferon gamma levels (using ESAT-6 QuantiFERON) were very low or undetectable in all students both before and after BCG vaccination. Of these students, 46% had Mantoux test responses of 0 to 4 mm and 54% had responses of ≥5 mm. Thirteen percent had Mantoux test results of ≥10mm. Under current Australian guidelines, one student with a 16mm result was defined as having a Mantoux tes result suggestive of M. tuberculosis infection. (3)

High school contacts in a TB outbreak in Denmark{9} who had high exposure to an index case and were not BCG vaccinated, had agreement between Mantoux test and IGT (QuantiFERON-TB Gold) results of 93% (95%CI 86 to 100%). This was 95% (95%CI 88 to 102%) in the low exposure group and an overall agreement between the two tests of 94% (95%CI 89 to 99%) in all subjects tested. The kappa value was 0.866, indicating high agreement between the two tests. (3)

In an Italian study{19} of contacts of an index case on a maternity unit, IGT (ESAT-6/CFP-10 ELISPOT assay) results were independent of BCG vaccination status. (3)

IGTs were prescribed by hospital physicians for inpatients or outpatients in an Italian study with no influence from the study investigators.{12} After excluding indeterminate results, the agreement between IGT (QuantiFERON-TB Gold) and Mantoux test results was significantly lower among BCG-vaccinated individuals than in non-vaccinated individuals (41.5% vs. 80.3%, p<0.0001). (3)

In a study of healthcare workers conducted in India{18} (where non-tuberculous mycobacteria are highly prevalent), previous BCG vaccination was not associated with Mantoux test or IGT (QuantiFERON-TB Gold) positivity. (3)

Indeterminate test results

An Italian study{12} found that indeterminate IGT results (QuantiFERON-TB Gold) were significantly over-represented in patients with a negative Mantoux test (28.6% vs. 6.6% in tuberculin positive patients, p<0.001) and were more frequent in patients receiving immunosuppressive therapies than in those who were not receiving such treatments (OR 3.35, 95%CI 1.84 to 6.08, p<0.0001). Immunosuppressive therapy was defined as cancer chemotherapy, systemic steroids, or anti-tumour necrosis factor alfa agents at the time of testing. (3)

5.1.10. Health economics 2006

A decision model was used to compare the expected cost-effectiveness of four strategies of testing for latent infection in the context of a contact tracing programme in England and Wales. The strategies compared were:

  • Mantoux test followed by IGT for patients with a positive Mantoux test
  • no test (inform and advise only).

It was assumed that treatment followed current policy: with appropriate therapy for people diagnosed with active TB or testing positive for latent infection, and BCG when appropriate for others. The analysis did not compare different types of skin tests or different types of IGT.

The model is a decision tree, which does not account for the dynamics of disease transmission within the population. Instead, for simplicity, it was assumed that each primary case of active disease is associated with a fixed number of secondary cases. This is probably a reasonable assumption when comparing tests with similar sensitivity, since the absolute difference in false negatives, and hence in opportunities for transmission within the community, will be small. However, estimates of the relative cost effectiveness of contact tracing per se are less robust and should be treated with caution.

Various assumptions were made about the epidemiology and likely concordance with testing and treatment programmes. However, it should be noted that these factors will vary with the context of contact tracing. There is also considerable uncertainty over the relative accuracy of the Mantoux test and IGT, as well as over some of the other model parameters. Whenever possible input parameters and assumptions were based on empirical evidence, but some key parameters were estimated by the health economist and GDG.

Cost-effectiveness of testing strategies in contact tracing

The basecase economic analysis suggests that the two-stage strategy (Mantoux test /IGT) is within the range usually considered ‘cost-effective’, at around £26,000 per quality-adjusted life-year (QALY) gained. Compared with this, IGT is not cost-effective (over £150,000 per QALY gained). Mantoux test is both less effective and more expensive than all of the other options (it is ‘dominated’).

Variation in optimal strategy with context of contact tracing

The results of the economic analysis were highly dependent on the context of the contact tracing scheme – with a higher-risk cohort of contacts, the expected benefits of early diagnosis of active cases, treatment of latent infection, and vaccination will be greater. Below a prevalence of about 10% none of the testing strategies is cost-effective. At intermediate levels of prevalence (between about 10% and 40%), the two-stage Mantoux test /IGT strategy is cost effective. Above 40% IGT on its own is the most cost-effective option (Table 25).

Table 25. Cost-effectiveness of diagnostic strategies.

Table 25

Cost-effectiveness of diagnostic strategies.

Uncertainty over optimal testing strategy for contact tracing

The results of the economic analysis were subject to a high degree of uncertainty. The results were very sensitive to assumptions about the relative accuracy of the two types of test, the risk of current and future TB in the cohort, the level of transmission to the wider population, and also to the expected net benefit of avoiding each active case of TB.

5.1.11. From evidence to recommendations

IGTs showed little evidence of being affected by prior BCG vaccination, and showed stronger correlation with exposure categories than did Mantoux test. This was shown in low prevalence groups, in household contacts, and in outbreak situations. The specificity of IGTs seemed better, and there was less potential for false positive results. It is not possible to determine, for either a Mantoux test or IGT, the rate of false negative results. Some people with false negative results will go on to develop active TB and thus reduce the cost-effectiveness of vaccination and treatment of latent TB infection.

Prospective studies in people with latent TB (as judged by positive IGTs) found at TB contact tracing and new entrant screening, have not yet been performed to find what proportion of such persons went on to develop clinical disease.

Economic modelling was undertaken with various strategies from no action to a two-step strategy with either a Mantoux test followed by interferon gamma testing, or serial IGTs. Of these options, the model provided most support, on grounds of cost-effectiveness, for a two-step approach with an initial Mantoux test, followed by an IGT to confirm positivity. The GDG members also supported this because of clinical utility and feasibility.

RECOMMENDATIONS – Partial update 2011

These recommendations update and replace recommendation R1 from CG33

Diagnosing latent TB

R1.

Offer Mantoux testing in line with the Green Book8 to diagnose latent TB in people who are:

  • household contacts (aged 5 years and older) of all people with active TB
  • non-household contacts (other close contacts for example, in workplaces and schools).
R2.

Consider interferon gamma testing for people whose Mantoux testing shows positive results, or in people for whom Mantoux testing may be less reliable, for example BCG vaccinated people.

R3.

If Mantoux testing is inconclusive, refer the person to a TB specialist.

New entrants from high-incidence countries

R4.

Offer a Mantoux test to children aged 5–15 years. If positive, follow with an interferon-gamma test.

R5.

Offer either an interferon-gamma test alone or a dual strategy in people aged 16–35 years. For people over 35 years, consider the individual risks and benefits of likely subsequent treatment, before offering testing. (Refer to other sections for other groups e.g. immunocompromised)

R6.

Offer Mantoux testing as the initial diagnostic test for latent TB infection in children younger than 5 years who have recently arrived from a high-incidence country. If the initial test is positive (taking into account the BCG history):

  • refer to a TB specialist to exclude active disease and
  • consider treating latent TB.

Household contacts 2- 5 years

For children younger than 2 years see R83-85

R7.

Offer Mantoux testing as the initial diagnostic test for latent TB infection in child household contacts between the ages of 2 and 5 years. If the initial test is positive taking into account the BCG history:

  • refer to a TB specialist to exclude active disease and
  • consider treating latent TB.
R8.

If the initial Mantoux test is negative but the child is a contact of sputum-smear-positive disease, offer an interferon-gamma test after 6 weeks and repeat the Mantoux test to increase the sensitivity (to reduce false negative results).

Contacts – outbreak situation

R9.

In an outbreak situation when large numbers of individuals may need to be screened, consider a single interferon-gamma test for people aged 5 years and older.

People who are immunocompromised

R10.

If latent TB is suspected in children who are immunocompromised, refer to a TB specialist.

R11.

For people with HIV and CD4 counts less than 200 cells/mm3, offer an interferon-gamma test and a concurrent Mantoux test. If either test is positive:

  • perform a clinical assessment to exclude active TB and
  • consider treating latent TB infection.
R12.

For people with HIV and CD4 counts of 200–500 cells/mm3, offer an interferon-gamma test alone or an interferon-gamma test with a concurrent Mantoux test. If either test is positive:

  • perform a clinical assessment to exclude active TB and
  • consider treating latent TB infection.
R13.

For other people who are immunocompromised, offer an interferon-gamma test alone or an interferon-gamma test with a concurrent Mantoux test. If either test is positive:

  • perform a clinical assessment to exclude active TB and
  • consider treating latent TB.

Healthcare workers

R14.

Offer a Mantoux test to new NHS employees who will be in contact with patients or clinical materials if the employees:

  • are not new entrants from high-incidence countries and
  • have not had BCG vaccination (for example, they are without scar, other documentation or reliable history)9.
R15.

If the Mantoux test is negative, refer to the Green Book for BCG immunisation guidance. If the Mantoux test is positive, offer an interferon-gamma test.

R16.

Offer an interferon-gamma test to new NHS employees who have recently arrived from high-incidence countries or who have had contact with patients in settings where TB is highly prevalent.

R17.

Healthcare workers who are immunocompromised should be screened in the same way as other people who are immunocompromised.

Hard to reach groups

R18.

Offer people from hard to reach groups a single interferon-gamma test.

8

In this guideline the ‘Green Book’ is the 2006 edition of ‘Immunisation against infectious disease’, published by the Department of Health(available from http://www​.dh.gov.uk) The Green Book contains details of people who may have suppressed responses to tuberculin skin testing.

9

If there is reliable evidence of BCG vaccination, refer to the Green Book.

5.2. Diagnosing active tuberculosis

5.2.1. Clinical introduction

Signs and symptoms of respiratory TB

Primary respiratory tuberculosis is often asymptomatic, but the fact that infection has occurred is shown by the development of a positive tuberculin skin test or interferon gamma blood test. A history of recent contact with a person with TB is the most important factor in making the diagnosis. Occasionally, tuberculin conversion is accompanied by erythema nodosum or phlyctenular conjunctivitis. Mediastinal nodal enlargement as part of the primary complex can sometimes press on or discharge into a bronchus causing collapse of the distal lung or bronchial narrowing leading to wheeze or obstruction with distal over-inflation.{22}

In children with primary TB, weight loss, or weight loss and cough, are symptoms associated with culture confirmed TB. However, about half of all children with primary TB will have no symptoms.

Post-primary tuberculosis may be asymptomatic in the early stages, but symptoms, which can be either constitutional or respiratory, soon develop. Malaise, weight loss, fever and night sweats are the common constitutional symptoms. Cough is the commonest respiratory symptom, which is initially dry and non-productive but may later become productive, with haemoptysis in a small minority of cases. Breathlessness is a late feature, usually only occurring when a substantial amount of lung is destroyed or there is a significant pleural effusion. Chest pain is relatively uncommon, but can be pleuritic if peripheral lesions are present, or of dull ill-localised nature.

A study in Sudan, grading sputum smear positivity with clinical features showed multiple chest symptoms were positively correlated with sputum smear positivity. Also, the longer the duration of symptoms, the more this correlated with sputum smear positivity.{23} A comparison of the ‘classic’ symptoms of tuberculosis in patients with and without tuberculosis{24} is summarised in Table 26.

Table 26. Classic symptoms of tuberculosis.

Table 26

Classic symptoms of tuberculosis.

A multivariate analysis{25} showed that the following features were positively associated with culture proven tuberculosis:

  • the presence of TB risk factors or symptoms (OR 7.9)
  • a positive skin test for tuberculosis (OR 13.2)
  • a high temperature (OR 2.8)
  • upper lobe disease on a chest radiograph (OR 14.6)

and that the following were negatively correlated with tuberculosis:

  • shortness of breath (OR 0.2)
  • crackles on physical examination of chest (OR 0.29).

Signs and symptoms of non-respiratory TB

Tuberculosis can affect nearly every non-respiratory site, sometimes with a combination of respiratory and non-respiratory sites, or single or multiple non-respiratory sites.{22} As with respiratory tuberculosis, there can be systemic and site-specific symptoms. Weight loss is particularly associated with disseminated (including miliary) and gastrointestinal tuberculosis. Fever and night sweats are common in some non-respiratory sites of disease (disseminated, including miliary, and gastrointestinal TB), but are not common in others (peripheral lymph nodes, skin, bone and joint, genitourinary TB). Tuberculosis has to be considered in the differential diagnosis of an unexplained fever, particularly in those born abroad and/or in ethnic minority groups.

Because of the multiplicity of potential sites of non-respiratory TB, suggestive symptoms are considered site by site.

Signs and symptoms of lymph node TB

Nearly half of all non-respiratory TB in England and Wales occurs in peripheral lymph nodes, mainly cervical.{26},{27} The nodal enlargement in TB is usually gradual and painless, but can be painful if rapid. The usual absence of erythema and warmth makes the classical ‘cold abscess’. The nodes originally are discrete and firm, but may later mat together and become fluctuant as necrosis develops, which can discharge through the skin with sinus formation and superficial ulceration. Persistent lymphadenopathy of over four weeks duration in people other than white UK-born should be regarded as TB until proven otherwise and investigated appropriately.

Signs and symptoms of bone and joint TB

Bone and joint TB accounts for some 10–15% of non-respiratory disease, with approximately 50% in the spine, and 50% in a wide range of other bones and joints.{28},{29}

With spinal disease pain is the commonest symptom, and may be accompanied by local tenderness or slight kyphosis. Grosser kyphosis occurs when disease has progressed. Paraspinal abscesses can develop and may present as a loin mass, or as a psoas abscess pointing below the groin or causing psoas spasm with hip flexion. Compression on spinal nerve roots can mimic abdominal pathology. Extradural abscess or spinal collapse and subluxation can lead to sensory and motor symptoms involving the legs and sphincters due to spinal cord compression. Back pain and/or neurological signs should have an infective process in the differential diagnosis, particularly in ethnic minority groups.

A wide range of other joints can be involved. TB should be included in the differential diagnosis of unusual bone and joint lesions, particularly of an isolated lesion or a mono-arthritis in an ethnic minority group.

Signs and symptoms of gastrointestinal TB

This form of disease, as with nearly all other non-respiratory sites, is much commoner in ethnic minority groups. The gastrointestinal tract can be involved anywhere along its length, but peri-anal and upper gastrointestinal sites are uncommon (3% of gastrointestinal TB).{30} Series in both the developing{31} and developed world{32} show approximately one third of cases present acutely simulating abdominal emergencies and two thirds with a more gradual onset. Of the cases with an acute onset, approximately one half have right iliac fossa pain simulating acute appendicitis and the other half acute intestinal obstruction. Of those with a more gradual onset of symptoms, fever and malaise, abdominal pain and weight loss are the commonest described symptoms,{32} being found in 72%, 60% and 58% of cases respectively in another series.{33} Abdominal distension, usually due to ascites, is reported in between 10%{32} and 65%{34} of cases. There may be right iliac fossa tenderness simulating appendicitis, or a right iliac fossa mass simulating appendix abscess or carcinoma. The ileocaecal area is the commonest site of disease. With bowel involvement there may be acute or sub-acute small bowel obstruction with vomiting and abdominal distension; there may also be palpable mass. The colon distal to the caecum is involved in up to 10% of cases{32} and is a cause of gastrointestinal bleeding.{35}

Signs and symptoms of genitourinary TB

Genitourinary TB is one of the commoner sites of non-respiratory TB in white UK-born people. For example, in 1993 it accounted for 17% of non-respiratory cases in the white UK-born ethnic group, compared with 4% in people of Indian (subcontinent) origin.{27} In white cases renal tract lesions predominate but female genital disease predominates in the Indian sub-continent ethnic group.{36}

Renal tuberculosis is often a ‘silent’ disease with insidious progression which can lead to total unilateral renal destruction. Systemic features such as weight loss, fever and night sweats are not common. As disease progresses, dysuria, haematuria, nocturia and pain either in the loin or anteriorly may occur. Renal disease can lead to ureteric and then bladder involvement by tubercle bacilli seeding distally. Bladder involvement initially leads to cystitis symptoms with frequency and dysuria, but as bladder wall inflammation with associated fibrosis worsens, bladder capacity falls and can be greatly reduced, the so-called ‘thimble bladder’ leading to marked frequency and nocturia due to a tiny bladder capacity. The urine with renal and ureteric disease, but particularly with bladder disease, shows proteinuria and haematuria on dipstick testing, and pus cells on microscopy but is sterile on standard culture. The finding of sterile pyuria should lead to the routine sending of three early morning urines for TB culture. A cold perinephric abscess can occur pointing in either the loin or like a psoas abscess in the groin. Prostatic, epididymal and testicular TB are less common. Testicular TB can present as a mass simulating testicular tumour.

Female genital TB is due to either haematogenous spread or direct spread from intra-abdominal disease. As with urological TB, systemic symptoms are uncommon unless there is associated abdominal tuberculosis. Infertility, either primary or secondary, is the commonest presentation of tubal and endometrial TB.{37} Most have no associated symptoms, but menorrhagia is reported in 20–25%, with much lower proportions having amenorrhoea or post menopausal bleeding.{37}

Signs and symptoms of disseminated (including miliary) TB

Disseminated TB occurs when tubercle bacilli are spread acutely though the blood stream. The symptoms are insidious at the onset with malaise, fever, anorexia and weight loss. In addition, headache from associated TB meningitis can occur with disseminated TB.

Signs and symptoms of central nervous system TB

Although only forming 5% of non-respiratory TB,{36} TB of the CNS is of disproportionate importance because of its significant morbidity and mortality. Early symptoms are non-specific with anorexia, malaise, headache, vomiting and altered behaviour. In children these can be poor feeding, irritability, altered behaviour, drowsiness or seizures. The prodromal phase can last from two weeks to two months, then focal neurological signs or decreasing level of consciousness occur. If cranial nerve palsies are present, 3rd and 6th nerve palsies are commoner than 7th and 8th nerve palsies. Internuclear ophthalmoplegia or lateral gaze palsies are less common but more serious because of midbrain or brainstem involvement.{37} Other neurological signs can develop depending on the site of endarteritis or infarction, including cerebellar signs, extrapyramidal movements such as choreoathetosis, hemiparesis or monoparesis.

Signs and symptoms of skin TB

Skin involvement can be due to disease of underlying structures, usually lymph node, bone or urogenital tract, with discharge through the skin, with sinus formation, so-called ‘scrofuloderma’. Lupus vulgaris is a slowly destructive local skin form with dull red or violaceous edges. The tuberculides are forms of skin disease thought to be a manifestation of TB elsewhere in the body. Panniculitis, erythema induratum (Bazin's disease), and papular and papulo-necrotic forms are described and TB is in the differential diagnosis of such lesions, particularly in ethnic minority groups.{38}

Signs and symptoms of pericardial TB

TB can cause either pericardial effusion or constrictive pericarditis, particularly in ethnic minority groups. Fever, malaise, sweats, cough and weight loss can occur. The signs of pericardial effusion are oedema, pulsus paradoxus, a raised venous pressure, and hypotension with a narrow pulse pressure. With constrictive pericarditis, oedema, abdominal distension and breathlessness are the major signs and symptoms. A lymphocytic exudate on pericardial aspirate should be regarded as TB until proven otherwise.

Signs and symptoms of TB at other sites

TB should be considered in the differential diagnosis of adrenal deficiency, liver abscess, pancreatic mass in young adults with fever, and for isolated ‘cold’ abscesses wherever found, particularly in ethnic minority individuals.

Diagnosing active respiratory TB

The diagnosis of TB is suspected from a combination of context, symptoms, clinical signs and investigations. The diagnosis is rarely made from a single piece of evidence, and the sensitivity and specificity of individual tests may not reflect the strength of multiple tests or data. Most of the data on the utility of individual tests comes from studies in patients with proven tuberculosis by positive culture. Certain clinical settings are highly suggestive of tuberculosis in ethnic minority groups or recent TB contacts. These are: a pleural effusion which is a lymphocytic exudate, or isolated mediastinal lymphadenopathy, either supported by a positive skin tuberculin test (or IGT). These scenarios should be regarded as tuberculosis until proved otherwise and investigated accordingly.

A significant minority of respiratory TB cases however are not bacteriologically confirmed, but are treated on suspicion and regarded as probable cases because of response to specific anti-tuberculosis medication. The guideline aims to advise clinicians on which tests may help if cultures have been, or are subsequently shown to be, negative.

In children, who often have no culture confirmation, scoring systems have been developed to help diagnosis based on context, symptoms, X-ray appearances and other investigations. Some scoring systems are better validated than others.{39}

Diagnosing active non-respiratory TB

Most forms of non-respiratory tuberculosis have a lower bacterial load than for pulmonary disease, being so-called pauci-bacillary forms. A relatively very low proportion of cases have positive microscopy for acid-fast bacilli (AFB), and with the lower bacterial loads, even with rapid culture (see section 5.4) it takes longer to obtain positive cultures. With many of the non-respiratory sites, biopsy histology, or, in the case of lymph node disease, needle aspiration cytology, is available well before bacteriology. The finding of caseating granulomas, or granulomas with Langhan's giant cells on histology or cytology, is very highly suggestive of tuberculosis. A number of other conditions however can cause non-caseating granuloma formation. In the absence of caseation or Langhan's giant cells, additional tests such as a tuberculin skin test or IGT may be needed to assist in diagnosis. Obtaining a sample for culture is important as this confirms the diagnosis and provides the drug susceptibility profile of the organism. One caution is that in children aged under five, particularly if they are of white UK-born origin, granulomatous lymphadenitis is much more likely to be M. avium complex (MAC) than M. tuberculosis. To confirm this, samples are sent for culture, management for M. avium being completely different from M. tuberculosis in this context.{40}

The yield of histology/cytology depends on tissue sample size, which is much smaller with aspiration cytology than biopsy, and on the level of immune response which generates the histological appearances. In HIV-positive individuals the histological response depends on the level of immunosuppression. With levels of CD4 lymphocytes above 200/μl typical TB histology is the rule, but as the CD4 cell count falls, particularly below 100/μl, less and less granuloma formation occurs, and with profound immunosuppression there may be no cellular histological response at all. In these circumstances however there is an increased likelihood of AFB being seen microscopically. The differential diagnosis in such very immunosuppressed individuals is usually between M. tuberculosis and MAC infection. Polymerase chain reaction (PCR) techniques may help in distinguishing between these infections on AFB microscopy-positive samples (see section 5.3). A similar diagnostic problem can occur when patients with a very low CD4 count are started on highly active antiretroviral therapy (HAART). The rapid fall in HIV viral load and rise in CD4 count allows an immune response to be mounted to either of these organisms, which was not previously possible. Enlargement of cervical and intra-abdominal lymph nodes in particular are described in this context, which is known as the immune reconstitution or IRIS syndrome.

In some cases of non-respiratory tuberculosis, the diagnosis of TB is not entertained in the differential diagnosis, and the doctor, usually a surgeon, does not send any material for culture, instead placing the entire sample in formalin. This then completely precludes any attempt at bacterial culture, although if AFB are seen histologically it still allows PCR-based techniques to be used (see section 5.3). The same histological and cytological criteria apply as in Table 27. Tuberculin skin tests or whole blood interferon gamma based tests may be needed to assist with histological appearances that are not fully diagnostic.

Table 27. Suggested site-specific investigations in the diagnosis and assessment of non-respiratory TB.

Table 27

Suggested site-specific investigations in the diagnosis and assessment of non-respiratory TB.

5.2.2. Methodological introduction

Diagnosing active respiratory TB: testing while awaiting culture results

Studies were identified which calculated the sensitivity, specificity or predictive value of plain X-ray, sputum smear microscopy and gastric washings when compared with culture as the gold standard for the diagnosis of respiratory TB. Studies on sputum smear microscopy were excluded from review if they were conducted in non-Organisation for Economic Co-operation and Development countries as it was thought that in terms of background levels of mycobacteria and laboratory standards they might not be representative of the UK.

Eight studies examined the diagnostic accuracy of sputum smear microscopy in comparison with culture. Two US studies were excluded for methodological reasons.{41},{42}

Of the six remaining sputum microscopy studies, five were conducted in the US{4347} and one in Turkey.{48} Three of these studies reported results for HIV-positive patients or those with AIDS.{43},{44},{47}

Four studies were identified which considered the diagnostic accuracy of chest X-ray in predicting culture results. One Danish study included all patients who had a respiratory sample examined for M. tuberculosis during a specified time period,{49} a South African study was of paediatric patients suspected of having TB{50} whilst two US studies{51},{52} considered diagnostic accuracy of chest X-ray in those with AIDS/HIV.

Three studies considered the diagnostic accuracy of gastric washings in children.{5355} Two of the studies were performed more than ten years ago in developing countries in populations with a high proportion of malnourished children, thus their applicability to the UK today is highly questionable. A more recent study performed in Cape Town, South Africa{55} compared gastric lavage and induced sputum samples from children in terms of their diagnostic yield, reporting how many cases were culture positive, smear positive or both.

Methodological considerations include the following:

  • In terms of sputum smear microscopy, serial testing of sputum samples will increase the sensitivity and specificity of the test.
  • Sensitivity and specificity values are calculated in different ways, either on a patient basis or a specimen basis.
  • Methods used for processing the sputum specimen (including the minimum volume of sputum required and whether the specimen is expectorated or induced) or the method of isolating cultures may differ in various settings.

Generally studies were unblinded (mostly because they were retrospective analyses). Blinding, however, is probably not crucial to avoid bias in the assessment of smear microscopy as the same samples are used for smear and culture and are subject to standardised laboratory procedures and definitions. It was notable that none of the studies identified were performed in the UK.

Diagnosing active respiratory TB if culture results are negative

Two studies{56},{57} addressed the issue of what other test results might support a positive diagnosis in those with a negative culture for TB but with suspected respiratory TB. In a South African study a group of black male goldmine employees with small lesions in the lung apices on chest X-ray, and a positive skin test but negative sputum culture, were followed up.{56} A diagnosis of TB was made if the smear became positive, if the culture yielded M. tuberculosis or if a histological diagnosis was made. A Hong Kong study had a subgroup of patients who had TB diagnosed on the basis of chest X-ray but had negative culture results.{57} This group were followed up for future confirmation of TB by culture of M. tuberculosis from sputum, or by radiographic or clinical deterioration.

Methodological issues for consideration are that the gold standard against which diagnostic tests for TB are usually compared is microbiological identification of TB by culture. This is not a perfect gold standard and culture might be negative in TB cases due to ‘pauci-bacillary disease’ (only a small number of M. tuberculosis organisms are present), sampling error or technical problems. In these cases where culture is negative, the standard against which a diagnostic test might be compared could be response to treatment, clinical features or a positive culture in the future. A TB diagnosis in this population would probably be achieved on a case-by-case basis and this has thus not been the subject of many studies.

Diagnosing active non-respiratory TB: testing while awaiting culture results

Studies were searched for which considered the sensitivity and/or specificity of histology from biopsy when compared with culture as the gold standard for the diagnosis of non-respiratory TB. Biopsies could be obtained during surgical procedures or by fine needle aspiration.

Four studies were identified where sensitivity of histology was calculated or it was possible to calculate sensitivity from the results reported. These studies were performed in India,{58} Malawi,{59} the USA{60} and the UK.{61} Two studies reported results in HIV-positive patients.{59},{60}

Due to the recognition that non-respiratory TB can have low positive culture rates, studies often base a firm TB diagnosis on histology or culture. A positive histology result is thus not necessarily considered to be inaccurate in the presence of a negative culture. For this reason, there are few studies which consider the sensitivity of histology from biopsy compared to culture alone as the reference standard. Studies merely report the numbers positive on each test. This is not useful for calculating the sensitivity of histology as it is necessary to know the results for each patient on both tests.

These studies were not blinded, mostly because they were retrospective analyses. The majority of specimens used in these studies were lymph nodes and little information is available concerning whether sensitivity and/or specificity may differ when using specimens from other sites.

Although the diagnostic accuracy of individual tests was considered in isolation, in reality test results would not be considered in isolation but would contribute to the overall evidence on which a diagnosis is made.

Diagnosing active non-respiratory TB if culture results are negative

Studies of patients with suspected non-respiratory TB where the results of histology from biopsy or tuberculin skin test were used to support a positive diagnosis in those with a negative culture for TB were searched for.

As with respiratory TB, culture is not a perfect gold standard and may be negative in TB cases for several reasons. In particular in non-respiratory TB, this may be due to pauci-bacillary disease.

No studies were identified in culture-negative populations where the results of histology from biopsy or tuberculin skin tests were used to support a positive diagnosis.

5.2.3. Evidence statements: diagnosing active respiratory TB while awaiting culture results

Sputum microscopy

In a comparison in the USA{45} of direct and concentrated specimens, results were analysed for the first three sputum specimens received from patients who were culture-positive for M. tuberculosis and from whom three or more specimens were received. The cumulative proportion of positive smears for each of the three smears for concentrated specimens were 74%, 83% and 91% and this was 57%, 76% and 81% for direct smears. (2)

Sensitivity of smears (all smears, not per patient) using more than or equal to 5 ml of sputum volume in a study in the USA{46} was 92%. This was significantly greater than a sensitivity of 72.5% in a previous period when all specimens were processed regardless of volume. In both periods the specificity of acid-fast smear for M. tuberculosis was comparable at 98%. (2)

The rates of smear positivity were calculated for specimens of expectorated sputum, induced sputum and bronchoalveolar lavage (BAL) specimens in a study in the USA.{43} Findings of smears of expectorated sputum specimens showed that 55% were culture positive for M. tuberculosis and were AFB smear positive. Smear positivity rates for induced sputum were 38% and for BAL were 26%. When the predictive value was calculated by including only the first smear-positive specimen from each patient the values were 87% for expectorated sputum, 70% for induced sputum and 71% for BAL. (2)

A Turkish study{48} compared Ziehl-Neelsen (ZN) and fluorescence microscopy (FM) staining of sputum smears. Where only one specimen was submitted the sensitivities of ZN and FM stains were found to be 61% and 83% respectively. When two were submitted the sensitivities were 66% and 83% and where three or more were submitted sensitivities were 80% and 92%. (3)

In a US study{43} of expectorated sputum specimens that were culture positive for TB, 55% of specimens from both patients with and without AIDS (mean 2.4 specimens per patient for both groups) were smear positive. (3)

In a group of non-HIV infected, culture-positive TB patients in the USA,{47} 57% had positive acid-fast smears compared with 60% of the HIV-infected patients with culture-positive TB (all had at least three specimens tested). Among the TB culture-positive HIV-infected patients, no significant differences were found in the frequency of positive acid-fast sputum smears between groups stratified by CD4 cell counts (in those with a CD4 count of <50, 58% had positive smears, with a CD4 count of 50–200, 60% had positive smears and with a count of >200, 56% had positive smears). (3)

In a USA study,{44} 70% of all HIV-infected culture-positive TB patients and 71% of all non-HIV infected culture-positive TB patients had at least one positive smear (up to three were performed). The sensitivity for the diagnosis of TB dropped to 55% and 64% respectively when only the first smear was considered. (3)

Chest X-ray

According to X-ray category in a Danish study,{49} positive predictive values and sensitivity for TB were 61% and 67% respectively with X-ray changes thought to be due to TB. These values were 20% and 19% with X-ray changes compatible with TB; 14% and 9% with previous TB and radiographically active TB; 2% and 3% with previous TB but not radiographically active TB and 1% and 2% with X-ray changes thought to be due to other disease. None of the patients with normal chest X-rays were culture positive. (1)

In a South African study{50} of the diagnostic accuracy of X-ray in children, the results yield a sensitivity of 38.8% and a specificity of 74.4% compared to culture for the diagnosis of pulmonary TB using standard radiographs. (3)

In a group of culture-positive adult AIDS patients a US study{51} found 36% of patients had a primary M. tuberculosis pattern, 28% had a post-primary M. tuberculosis pattern, 14% had normal radiographs, 13% had atypical infiltrates, 5% had minimal radiographic changes and 3% had a miliary pattern. Normal chest radiographs were seen for 10 (21%) of 48 patients with less than 200 T-cells per microlitre and one (5%) of 20 patients with more than 200 T-cells per microlitre (p<0.05). (2)

In a US study{52} of TB culture-positive adults, 78% of HIV-negative patients' radiographs were consistent with post-primary pattern TB versus 26% of patients who were HIV positive (p<0.001). Only 11% of 18 significantly immunosuppressed HIV-positive patients (CD4 counts <200) had X-rays consistent with post-primary pattern TB, while all four patients with CD4 counts >200 had typical post-primary pattern chest radiographs (p<0.005). Of the 16 significantly immunosuppressed HIV positive patients the predominant chest X-ray finding was diffuse or multilobar infiltrates without an upper lobe predominance (N=8) followed by normal chest X-ray (N=3). (3)

Gastric washings

In a study of Haitian children{54} the sensitivity, specificity and predictive value of positive fluorescence microscopy of gastric washings compared with culture were 58%, 95% and 81% respectively from 536 specimens (median three specimens per patient). Among 49 children with at least one positive fluorescence microscopy of gastric washings, pulmonary TB was bacteriologically confirmed in 85%. Specimens were more frequently positive in far-advanced and miliary disease (82%) than in less severe disease (32%) (p<0.001). (3)

Culture was grown in 16 gastric washings samples in a study of Indian children{53} and smears for AFB were positive in only three samples, thus sensitivity was 3/16 or 19% (most children had only one sample taken). (3)

A South African study{55} of children with suspected TB found that sensitivity of gastric lavage compared with culture was 39%, specificity was 99%, positive predictive value was 88% and negative predictive value was 90% (based on three gastric lavage samples). Similar results were found for induced sputum specimens, however the yield of culture positive cases from each method was 88% from induced sputum and 66% from gastric lavage. (2)

5.2.4. Evidence statements: diagnosing active respiratory TB if culture results are negative

In South African black male goldmine employees with small lesions in the lung apices on chest X-ray and positive skin tests but negative sputum culture, TB was subsequently diagnosed in 88 (58%) of the 152 men. A diagnosis of TB was made if the smear became positive or the culture yielded M. tuberculosis or if a histological diagnosis was made. Active TB developed in these men from three to 58 months after entering the study, with a mean of 19.8 months.{56} (2)

A study performed in Hong Kong of patients with TB diagnosed on the basis of chest X-ray, but with negative culture results, obtained eventual confirmation of active disease requiring treatment in 99 (57%) of 173 patients. During the first 12 months 43% had a confirmed diagnosis. Confirmation of TB was by culture of M. tuberculosis from sputum, or by radiographic or clinical deterioration. There was bacteriological confirmation in 41%. (3)

5.2.5. Evidence statements: diagnosing active non-respiratory TB while awaiting culture results

In patients who presented with lymphadenopathy in one or more extra-inguinal sites in Malawi{59} and who did not respond to general antibiotics, it could be calculated that the sensitivity of histology compared to culture was 70%, the specificity was 59%, the positive predictive value was 52% and the negative predictive value was 67%. (2)

In a US study{60} of lymph node specimens where the cytology report was compared with culture results, the sensitivity of cytology was calculated to be 72%. (2)

The sensitivity of histology (using a variety of specimens although most frequently lymph nodes) compared with culture in an East London population was 97% with a positive predictive value of 69%.{61} (2)

Where culture was the gold standard, an Indian study,{58} calculated that in clinically suspected cases of tuberculous lymphadenitis, sensitivity, specificity and positive predictive values for cytology were 78.5%, 73% and 76.7% respectively. (1)

HIV-positive

In a study in Malawi{59} in HIV-negative patients with TB lymphadenitis (diagnosed on the basis of a positive culture or histology result), 100% had positive histology results and 83% had positive culture results. These figures were 78% and 56% for those who were HIV positive. Thus the HIV status of the TB lymphadenitis patients suggests a negative influence of HIV infection on the possibility of both histology and culture being indicative of TB (OR 0.10, 95%CI 0 to 1.17, p=0.06). (2)

In a US study{60} of lymph node specimens where the cytology report was compared with culture results the sensitivity of cytology in those who were HIV negative was 76% and it was 69% in those who were HIV positive. (2)

5.2.6. From evidence to recommendations

The Chief Medical Officer's TB Action Plan{2} calls for primary and community care staff to be aware of ‘the signs and symptoms of the disease, local TB services and local arrangements for referring patients with suspected TB’. As this guideline is aimed at generalist clinicians as well as those working regularly with people with tuberculosis, recommendations include signs, symptoms and potentially helpful imaging techniques. NICE guidelines generally do not include service guidance (although exceptions have been made elsewhere in this guideline), and so recommendations for local referral are not given.

The GDG were aware of the General Medical Council's advice{62} on gaining consent for testing for ‘serious communicable diseases’, but noted that this advice was reprinted from prior guidance specific to HIV and did not feel that routine clinical practice supported it in TB, and that it was at variance with the Public Health Act.{63}

Testing for active respiratory TB while awaiting culture results

The yield of positive sputum microscopy is improved by an adequate sputum sample (5 ml or more), concentration of sputum, analysing multiple samples, and by fluorescence microscopy as the screening tool. Smear positive rates are higher for spontaneously induced sputum than for either induced sputum or BAL samples. The positive predictive value of positive sputum microscopy is 92% for spontaneously produced sputum, and 71% for both BAL and induced sputum. There appeared to be little difference in the results between HIV-positive and HIV-negative patients in terms of bacteriological results and sputum smear positivity. Microscopy on gastric washings has some utility in children, but a recent comparative study in children showed a single induced sputum (by hypertonic saline) to be superior to three gastric washings. Gastric washings are less likely to provide useful material in adults, because of acidic inhibition. Chest X-ray changes are less specific in children and HIV-positive individuals, particularly if the CD4 count is under 200 cells/μl.

Testing for active respiratory TB if culture results are negative

The evidence does not assess the adequacy of the respiratory samples sent for culture; a negative culture result can reflect no growth at that time, while a positive result may be obtained later. Chest X-ray appearances consistent with TB were noted to show progression to culture-proven disease in over 50% of subjects in the studies analysed from South Africa and Hong Kong. The decision whether to start TB treatment will be a clinical one based on experience, context and appraisal of all the individual's results. Further culture samples are sometimes needed after treatment has begun, and will remain viable for a few days, though growth may be slower; the GDG agreed a threshold of one week in this regard.

IGTs may also have a role in ruling out infection with M. tuberculosis; this area is developing rapidly and may need to be updated ahead of the rest of the guideline in 2008.

Testing for active non-respiratory TB while awaiting culture results

Microscopy can be strongly suggestive of TB with certain patterns, and this is often confirmed by a positive culture if material has been sent. Although the data were entirely for peripheral lymph nodes, the GDG thought that this was likely also to apply to other non-respiratory sites.

The decision to biopsy should not be influenced by concerns about sinus formation, as there is no evidence to support this with modern chemotherapy.

Patient preferences are an important consideration in choosing biopsy or needle aspiration.

Posterioranterior chest X-rays in people with suspected non-respiratory disease are helpful through detecting any coexisting respiratory disease, which will aid or confirm the diagnosis, and be another potential source of bacteriological confirmation. The GDG also agreed a range of other potential tests and imaging techniques.

Testing for active non-respiratory TB if culture results are negative

Although there was no evidence in this area, the GDG noted that continuous enhanced surveillance by the Health Protection Agency (HPA) shows that only some 55% of cases of TB are culture confirmed, and that this is often because no samples have been obtained, with the diagnosis being entirely histological. (However, other reasons include failures in the reporting system and limitations of the matching between Enhanced Tuberculosis Surveillance and MycobNet systems.) To raise the proportion of TB cases diagnosed, particularly at non-respiratory sites, more samples from common TB sites should be sent for TB bacteriology, which requires the education of those sending samples such as general, ENT and orthopaedic surgeons and radiologists performing biopsies.

IGTs may also have a role in ruling out infection with M. tuberculosis; this area is developing rapidly and may need to be updated ahead of the rest of the guideline in 2008.

5.2.7. RECOMMENDATIONS

R19.

To diagnose active respiratory TB:

  • a posterioranterior chest X-ray should be taken; chest X-ray appearances suggestive of TB should lead to further diagnostic investigation C(DS)
  • multiple sputum samples (at least three, with one early morning sample) should be sent for TB microscopy and culture for suspected respiratory TB before starting treatment if possible or, failing that, within seven days of starting C(DS)
  • spontaneously produced sputum should be obtained if possible; otherwise induction of sputum or bronchoscopy and lavage should be used B(DS)
  • in children unable to expectorate sputum, induction of sputum should be considered if it can be done safely, with gastric washings considered as third line B(DS)
  • if there are clinical signs and symptoms consistent with a diagnosis of TB, treatment should be started without waiting for culture results (see section 6.1 for details) D(GPP)
  • the standard recommended regimen should be continued in patients whose subsequent culture results are negative D(GPP)
  • samples should be sent for TB culture from autopsy samples if respiratory TB is a possibility. D(GPP)
R20.

To diagnose active non-respiratory TB:

  • advantages and disadvantages of both biopsy and needle aspiration should be discussed with the patient, with the aim of obtaining adequate material for diagnosis B(DS)
  • if non-respiratory TB is a possibility, part or all of any of the following samples should be placed in a dry pot (and not all placed in formalin) and sent for TB culture: D(GPP)

    lymph node biopsy

    pus aspirated from lymph nodes

    pleural biopsy

    any surgical sample sent for routine culture

    any radiological sample sent for routine culture

    histology sample

    aspiration sample

    autopsy sample

  • microbiology staff should routinely perform TB culture on the above samples (even if it is not requested) D(GPP)
  • the appropriate treatment regimen should be started without waiting for culture results if the histology and clinical picture are consistent with a diagnosis of TB (see chapters 6 and 7) C(DS)
  • all patients with non-respiratory TB should have a chest X-ray to exclude or confirm coexisting respiratory TB; in addition, tests as described in Table 27 should be considered D(GPP)
  • the appropriate drug regimen (see chapters 6, 7 and 9) should be continued even if subsequent culture results are negative. D(GPP)

Cross-referring:

5.3. Rapid diagnostic tests: molecular methods

5.3.1. Clinical introduction

Molecular probes for diagnosis

A number of methods have been developed which target and amplify specific regions of mycobacterial DNA, thus allowing a rapid result. However, such tests can result in false negative and false positive findings. Although rare, false positive results may occur due to contamination of the sample with environmental mycobacteria causing non-specific binding to the probe. More commonly, false negative results may occur due to low organism numbers or, in some sample types, for example CSF, to the presence of inhibitors. The specificity and sensitivity of the tests has been compared with culture proven disease. However, since 20–30% of pulmonary cases, and a higher proportion of non-pulmonary cases are not culture proven, the performance of molecular tests in these settings is difficult to assess.

Molecular probes for species confirmation

Species identification may sometimes be possible directly from the specimen using the techniques referred to above. Most usually, this will be possible only for M. tuberculosis complex organisms (M. tuberculosis, M. bovis, M. africanum). However, these methods may allow early differentiation between these organisms and environmental mycobacteria. These tests are most effective when applied to samples in which mycobacteria have been detected microscopically. Their use is currently recommended, to confirm true tuberculosis (ie transmissible disease) before a large contact tracing exercise, for example in a school or hospital, is carried out.{6}

When a sample yields a positive culture, rapid identification of several commonly encountered species may be possible. This may be done by the application of an expanded range of DNA amplification-based assays or by the use of non-amplified hybridisation probes. Both of these approaches are effective since the high numbers of organisms present in a positive culture overcome the problems associated with low bacterial counts and inhibition in the primary sample. The Mycobacterium Reference Service of the HPA now routinely confirms to clinicians whether a positive culture received is from the M. tuberculosis complex or not.

Molecular probes for rifampicin resistance

The incidence of multi-drug resistant strains of M. tuberculosis (MDR TB) in the UK is low (~1%) (see Appendix G). However, in some areas of the country and in some population groups the incidence is much higher. Whilst it should be noted that mono-resistance to rifampicin is found in approximately 5% of rifampicin-resistant strains, a high proportion of rifampicin resistance is associated with concurrent resistance to isoniazid (~95%). Thus the detection of resistance to rifampicin can be used as a marker for MDR TB with a high level of accuracy.

Rifampicin resistance is commonly due to one or more of several possible mutations of the rpoB gene and these can be detected using a PCR-based technique. A positive result from such a test should lead to the implementation of infection control measures and drug treatment for MDR TB until the results of standard drug susceptibility tests are available. Risk factors for MDR TB, which should lead to such tests for rifampicin resistance, are listed in section 9.1. Clinicians should be aware that there is a small (<5%) false negative rate for these tests as a few mutations conferring rifampicin resistance are not at the rpoB gene tested for.{64},{65}

Molecular typing of M. tuberculosis isolates

In the past the typing of M. tuberculosis strains has been principally to detect previous events. This was largely due to the comparatively slow techniques available (for example, restriction fragment length polymorphisms). Newer methods based on the detection of variable numbers of tandem repeat sequences within the M. tuberculosis genome (variable number of tandem repeats (VNTR)/mycobacterial interspersed repetitive unit (MIRU) typing) are amenable to automation. As a result rapid, high-throughput typing systems have become available. These systems also have the advantage of digitised data which allow much easier computerised storage and analysis than previous typing methods. If this rapidity of method is used to type strains as they are isolated, then potential links between patients may be detected early enough to interrupt the disease transmission process. Thus an epidemiological tool may make an impact on diagnosis and transmission.

5.3.2. Methodological introduction

In consideration of the use of molecular methods for rapid diagnosis of TB, the review being developed by the NHS Heath Technology Assessment Programme{66} has been adopted. This aims to conduct a systematic review of the effectiveness of available diagnostic tests to identify mycobacteria. The review is not yet published.

The draft review of nucleic acid amplification tests (NAAT) found 163 studies which compared NAAT with a reference standard. There were 105 comparisons in respiratory specimens and 67 in non-respiratory specimens. In these studies 77 of the tests used were commercially produced (the amplified Mycobacterium tuberculosis direct (AMTD) test, the Amplicor, the Ligase Chain Reaction and Ampicis Myco B) and 86 were produced in-house (insertion element IS6110 or other targets).

Methodological issues concern the complexity of pooling data from diagnostic studies in particular due to variation in diagnostic thresholds. Furthermore, studies report pairs of related summary statistics (sensitivity and specificity) rather than a single statistic, requiring alternative statistical methods for pooling results. This review presents diagnostic odds ratios (DOR) in addition to sensitivity and specificity data. This is a single summary of diagnostic performance which although not easy to apply in clinical practice (it describes the ratio of the odds of a positive test result in a patient with disease compared to a patient without disease) is convenient to use when combining studies as it is often fairly constant regardless of diagnostic threshold. The DOR can be calculated from sensitivity and specificity data and where a test provides no diagnostic evidence the DOR is 1. It has been suggested{67} that a DOR of 25 or more in a test may provide convincing diagnostic evidence.

5.3.3. Evidence statements

The health technology appraisal (HTA) on rapid diagnostic tests{66} is not yet published. The GDG considered interim results, reporting the DOR statistic calculated by comparing NAAT vs. a reference standard. All evidence is graded at level 2.

5.3.4. From evidence to recommendations

Molecular probes for diagnosis

The HTA of rapid tests showed that their sensitivity was equivalent to culture in microscopy negative pulmonary samples, but there was an increased false negative rate in non-respiratory samples, particularly in pleural fluid and CSF. Significant false negative rates in these settings limit their utility, and could lead to failure to diagnose and treat TB.

Molecular probes for species confirmation

The GDG did not look into the HTA's interim results for molecular probes, but noted their role in rapid confirmation. They were not felt to be more reliable or useful than culture confirmation, and use was therefore limited to occasions when a rapid decision is needed on treatment or infection control measures. A further role was in preventing large scale contact tracing exercises from starting unnecessarily. Molecular tests are less feasible on poorer samples, and the recommendations given below advise on their use on biopsy material.

Molecular probes for rifampicin resistance

Again, the GDG recognised the advantages of rapid results for drug resistance, but noted that MDR TB risk factors should be used to determine infection control measures at the earliest opportunity.

Molecular typing of M. tuberculosis isolates

Although this has not been subject to formal HTA appraisal, these methods have been considered by the HPA and a unified strategy using a 15 locus VNTR/MIRU system agreed. Such a strategy was recommended in the TB Action Plan.{2}

5.3.5. RECOMMENDATIONS

R21.

Rapid diagnostic tests for M. tuberculosis complex (M. tuberculosis, M. bovis, M. africanum) on primary specimens should be used only if: D(GPP)

  • rapid confirmation of a TB diagnosis in a sputum smear-positive person would alter their care, or
  • before conducting a large contact-tracing initiative.
R22.

Clinicians should still consider a diagnosis of non-respiratory TB if rapid diagnostic tests are negative, for example in pleural fluid, CSF and urine. B(DS)

R23.

Clinical signs and other laboratory findings consistent with TB meningitis should lead to treatment (see section 7.1), even if a rapid diagnostic test is negative, because the potential consequences for the patient are severe. D(GPP)

R24.

Before conducting a large contact-tracing initiative (for example, in a school or hospital), the species of Mycobacterium should be confirmed to be M. tuberculosis complex by rapid diagnostic tests on microscopy- or culture-positive material. Clinical judgement should be used if tests are inconclusive or delayed. D(GPP)

R25.

If a risk assessment suggests a patient has MDR TB (see section 7.1): D(GPP)

  • rapid diagnostic tests should be conducted for rifampicin resistance
  • infection control measures and treatment for MDR TB should be started as described in chapter 9, pending the result of the tests.
R26.

Rapid diagnostic tests for M. tuberculosis complex identification should be conducted on biopsy material only if: D(GPP)

  • all the sample has been inappropriately placed in formalin, and
  • AFB are visible on microscopy.

Cross-referring:

  • For details of managing drug-susceptible TB, see chapters 6 and 7.
  • For details of managing drug-resistant TB, see chapter 9.

5.4. Rapid diagnostic tests: automated liquid culture

5.4.1. Clinical introduction

Clinicians have been advised to obtain culture confirmation of tuberculosis whenever possible.{68} This not only confirms the diagnosis, but crucially also obtains material for drug susceptibility testing, which is important because of the current levels of drug resistance in England and Wales. The finding of isoniazid resistance (currently 6% of isolates) requires modification of treatment (see section 9.4), and that of MDR TB (currently about 1% of isolates) different infection control procedures (see section 9.3) and individualised treatment regimens based on the drug susceptibility data.

Until recently, culture for mycobacteria was done mainly on solid media, the Lowenstein-Jensen slope, or in broth media. These methods were slow, with cultures from microscopy positive material taking from 2–4 weeks, and for microscopy negative material 4–8 weeks. More recently rapid culture methods have been developed, with the potential advantages of more rapid growth and hence earlier drug susceptibility data, and also possibly increased sensitivity.

The national TB Action Plan has as one of its aims the use of rapid culture methods for diagnosis of all cases of tuberculosis.{2}

5.4.2. Methodological introduction

The reduced turnaround time of automated liquid culture in comparison with solid media is uncontested. In addition to time to detection of mycobacteria, study outcomes in comparisons between solid and liquid media also report increases recovery rates for mycobacteria.{66} Sensitivity and or specificity cannot be reported in these studies as there is no reference standard.

There were no studies identified which directly addressed the issue of when (ie in what circumstances) automated liquid culture methods for the diagnosis of TB are most useful.

The HTA on rapid diagnostic techniques{66} is not yet published. The GDG considered interim findings on liquid culture techniques.

5.4.3. From evidence to recommendations

Given the evidence base and the self-evident speed of automated liquid culture, the GDG recommended their universal use.

Liquid culture methods require batches of samples to be processed. Their use becomes more costly per test if fewer samples are processed at any one time by a laboratory. The batching of samples sent to regional laboratories may not reflect future service organisation as this technology becomes more widely used over the lifetime of this guideline, but the recommendations allude to the effect of throughput on efficiency, quality control and cost-effectiveness. The NICE guideline, in the absence of clinical evidence, is unable to recommend service configurations to address this, though the GDG considered a ‘hub and spoke’ arrangement of regional laboratories.

5.4.4. RECOMMENDATIONS

R27.

Clinical samples should ideally be sent for culture by automated liquid methods, bearing in mind that laboratories need a certain level of throughput to maintain quality control. D(GPP)

In this guideline the ‘Green Book’ is the 2006 edition of ‘Immunisation against infectious disease’, published by the Department of Health(available from http://www​.dh.gov.uk) The Green Book contains details of people who may have suppressed responses to tuberculin skin testing.

If there is reliable evidence of BCG vaccination, refer to the Green Book.

Section 5.1, Copyright © 2011, National Institute for Health and Clinical Excellence. Sections 5.2 to 5.4, Copyright © 2006, Royal College of Physicians of London.

For 2006 original guideline text, no part of the content may be reproduced in any form (including photocopying or storing it in any medium by electronic means and whether or not transiently or incidentally to some other issue of this publication) without the written permission of the Royal College of Physicians of London. Applications for the Royal College of Physicians of London's written permission to reproduce any part of this publication should be addressed to the publisher.

For 2011 updated text, the material may be freely reproduced for educational and not-for-profit purposes. No reproduction by or for commercial organisations, or for commercial purposes, is allowed without the express written permission of NICE.

Cover of Tuberculosis
Tuberculosis: Clinical Diagnosis and Management of Tuberculosis, and Measures for Its Prevention and Control.
NICE Clinical Guidelines, No. 117.
National Collaborating Centre for Chronic Conditions (UK); Centre for Clinical Practice at NICE (UK).

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...