A set of administrative controls is the first and most important component of any IPC strategy. These key measures comprise specific interventions aimed at reducing exposure and therefore reducing transmission of M. tuberculosis. They include triage and patient separation systems (i.e. management of patient flows to promptly identify and separate presumptive TB cases), prompt initiation of effective treatment and respiratory hygiene.
Evidence and justification
Recent decades have seen the accumulation of a substantial body of evidence on TB treatment and care. However, research in the area of TB IPC has been rather limited – reflected in the number of studies collected to inform this and the other recommendations included here. Looking at the effect of triage on the incidence of LTBI and TB disease among health workers, the systematic search (17) yielded 15 observational studies from secondary and tertiary health care facilities, of which 73% were carried out in low TB burden settings. A total of six studies measuring the effect of triage on the incidence of LTBI alone among health workers in all settings were included in the analysis (18–23) (see Online annexes 4 and 5).
Given the significant heterogeneity among the studies, only crude estimates were assessed by the Guideline Development Group and are described in this document. Estimates of effect showed an absolute risk reduction of 6% for LTBI incidence among health workers in all settings (n=6). When disaggregating by burden of disease, a 3% absolute risk reduction in LTBI was observed among health workers in low TB burden settings (n=5), compared with a 1.7% reduction in high TB burden settings (n=1). Three additional studies – one in low TB burden and two in high TB burden settings – were further assessed to determine the effect of triage on the incidence of TB disease among health workers (24–26). Estimates of reduction of TB incidence in high TB burden settings, calculated from crude pooled data, seemed to indicate very slight or no reduction in TB incidence (crude incidence rate ratio [IRR]: 0.98) among health workers after the implementation of triage within a set of composite IPC measures. The only study in which triage was implemented with no other interventions in a low TB burden setting (26) reported an incidence rate of 78 episodes of TB disease among health workers in 38 331 person-years in the control group (before the intervention was implemented), and 12 episodes in 18 229 person-years after the implementation of triage (crude IRR: 0.32, after versus before) (see Annex 3 for further information).¹
Although there were scant data from which to assess the impact of triage on prevention of TB infection among non-health care staff (i.e. other persons attending health care settings), two studies from low TB burden countries provided information about the effect of this measure in reducing the incidence of TB disease among this population (27, 28). These studies seemed to indicate that there is a 12.6% absolute risk reduction (crude estimate combining data from two studies) in the number of active TB disease cases in persons attending health care settings with the use of triage (in combination with other IPC measures) compared with similar populations in settings where triage was not implemented.
The analysis presented here had some limitations; namely, the scarcity of studies measuring the effect of TB-specific IPC interventions, and issues with the methodology of the studies and the quality of the data. A major challenge encountered during the evidence assessment was that a wide range of IPC activities were typically undertaken as part of composite interventions, making it impossible to determine the effects of individual components. All the included studies apart from one (26) presented a combination of measures that were implemented either simultaneously or sequentially.
Owing to the heterogeneity of studies identified in this systematic review, meta-analysis was not performed, and the analysis was restricted to the use of crude estimates and narrative summaries for some outcomes. In turn, this limited the possibility of exploring the effect of potential confounders or evaluating the impact of targeting specific risk factors in these analyses. The heterogeneous nature of the included studies (e.g. lack of standardization, use of composite interventions and outcome measurements) led the Guideline Development Group to downgrade all evidence by two levels, based on indirectness arising from several sources, and limitations leading to serious inconsistency and risk of bias (see Online annexes 4 and 5).
Another major source of indirectness was the interpretation or definition of the term triage, and whether measures were standardized and implemented systematically throughout a specific setting, or were applicable only to a perceived at-risk population (19–22, 29, 30). For instance, some studies triaged HIVpositive individuals, and people experiencing homelessness who were presenting to health care facilities with pneumonia or evidence of TB, whereas others described triage as a set of measures that included the prioritization of patients with cough for more than 2 weeks (regardless of perceived risk) and the rapid collection of respiratory specimens or the routine screening of all new admissions with chest X-rays. An additional consideration in the quality assessment was that of applicability or generalizability of the results. Three quarters of studies included for the evaluation of triage systems were conducted in low TB burden settings – 60% were carried out in the United States of America (USA) alone.
The panel also discussed the potential introduction of inconsistency and bias in the pooled results. Variability in results could be expected because the methods used to measure outcomes varied from study to study; for example, in measuring TB infection, it was not always clear whether a single tuberculin skin test (TST) or two-step testing¹ was used. Also, false readings were possible if readers were insufficiently skilled.
The observational studies identified through the systematic search were, by design, single group comparisons, often in a single health facility. The studies were beforeand-after (n=7), during-and-after (n=4) and cross-sectional (n=1). Before-and-after and during-and-after studies are the simplest ways to evaluate the effect of an intervention in a particular population; however, the panel acknowledged that comparing outcomes before-and-after implementation of a particular intervention introduced serious risk of bias, in addition to the effect of unidentified confounders, mainly due to lack of randomization. These designs cannot control for contemporaneous changes in case mix, background changes in the incidence in the general population, referral patterns or other elements of care. Often, such studies are of too short a duration to determine whether the intervention and its apparent effect are sustainable over time and across all settings.
The analysis was further constrained by the limited data from high TB burden settings; only three studies – from Brazil (four general hospitals), Thailand (one referral hospital) and Malawi (40 hospitals) – were identified and included in the systematic search (18, 24, 25).
No data on the use of triage in primary health care facilities were available.
Despite the lack of direct data, the Guideline Development Group advised that rapid triage systems – covering all health workers and other persons attending health care settings – be recommended in all health care facilities, regardless of the level of care. Although this recommendation was based on very low certainty in the evidence, a strong priority was assigned to this intervention given that, if properly and systematically implemented alongside other recommended IPC interventions, it is unlikely that harm would accrue from this intervention.
The effective implementation of this recommendation and the other recommendations in these guidelines relies on the understanding that interventions within the three-level hierarchy of IPC should not be prioritized individually or implemented separately, but must be considered as an integrated package of IPC interventions.
Implementation of any triage system needs to be focused on fast-tracking of presumed TB cases and on minimizing time in the facility.
Consultation and ongoing dialogue with both health care staff and patients should be considered, to provide feedback that could facilitate the implementation of the recommendation without contributing to stigma or alienation of patients. The considerations outlined below should guide the implementation of the recommendations.
Settings and target population
The recommendation on the implementation of triage of people with TB signs and symptoms as a means of reducing M. tuberculosis transmission to health workers or other persons attending relevant facilities applies explicitly to health care settings. Although the scope of the review was limited to such settings, the Guideline Development Group recognized that it is vital to implement triage in other settings with a high risk of M. tuberculosis transmission where persons with presumed TB may congregate (e.g. long-term care and correctional facilities), regardless of the burden of TB disease.
In addition, community health workers are key to promptly identifying presumptive TB cases at the community level and making use of referral systems, to fast-track TB diagnosis and facilitate the implementation of other interventions. Community health workers could help to improve the early detection of TB cases, and reduce the risk of transmission in the community in general.
The effective implementation of triage goes beyond the minimal infrastructure requirements (e.g. conditions for fasttracking of patients with presumed TB, rapid diagnosis, respiratory separation, use of data-recording tools for documentation, and analysis of data for developing or changing evidence-based policies). The implementation needs to prioritize the availability, education, sensitization and continuous training of health care providers and others working in settings with a high risk of M. tuberculosis transmission.
In line with current guidelines on screening for active TB disease, people living with HIV should be systematically screened for active TB at each visit to a health care facility (31, 32). Similarly, routine HIV testing should be offered to all patients with presumptive and diagnosed TB, especially in high HIV burden settings (33).
Triage systems may be part of collaborative activities established to prevent and identify TB across other disease programmes (e.g. diabetes and conditions that increase the risk of LTBI or TB disease).
Evidence and justification
The systematic search yielded 24 observational studies reporting assessment of the use of respiratory separation or isolation of persons with presumed TB, or with patients demonstrated to have infectious TB. As mentioned for Recommendation 1, the Guideline Development Group identified serious limitations with the studies reporting on the impact of administrative measures for TB prevention and control. In the case of respiratory separation measures, sources of indirectness included the use of composite interventions, and variability on how the intervention¹ was implemented. For the latter, for instance, some facilities made use of negative pressure isolation rooms with highefficiency particulate air (HEPA) filtration, whereas others described isolation rooms designed to provide six or more air changes per hour (ACH) or simpler features.
Among the studies identified, about one third (n=7) were excluded from the summary because data were not reported in a format suitable for aggregation (29, 30, 34–38). Of the selected studies (n=17), 15 were included in the summary analysis (crude summaries of findings) of outcomes related to LTBI and TB disease among health workers (11, 18–25, 39–44); only two additional included studies measured the burden of TB disease among non-health workers; that is, other persons attending health care services (27, 28) (see Online annexes 4 and 5).
Results in this systematic review were indicative of an absolute risk reduction of 2% in health workers when persons with presumed TB and confirmed TB patients underwent respiratory separation or isolation. When data were disaggregated by TB burden (low versus high), there was a relatively small reduction in risk of acquiring LTBI when respiratory separation or isolation was implemented, but no significant differences in absolute risk reductions were observed between low and high TB burden settings (1.6% versus 1.9%). In relation to the studies measuring the effect of respiratory separation or isolation on reducing TB incidence among health workers, two studies conducted in secondary and tertiary care facilities in high TB burden settings seemed to show a slight or no reduction in TB incidence among health workers when isolation was implemented. Both of these studies implemented isolation, together with a number of other administrative, environmental and protective infection control measures. In an additional study reporting on the use of isolation (an infection control audit at 121 primary health care facilities in South Africa), the authors reported slightly increased odds of developing smear-positive
TB (unadjusted odds ratio [OR]: 1.09; 95% confidence interval [CI] 0.99–1.19) in health workers for a unit increase in the administrative audit tool score, where a higher score equates to better administrative control measures (11).
In this systematic review, estimates of LTBI incidence could not be captured for other persons (e.g. non-health workers) attending health care facilities. However, data from two studies from low TB burden settings were available (27, 28). Estimates of TB disease in the observed groups in these studies seemed to indicate that the risk of developing active TB disease in persons attending secondary or tertiary level facilities was reduced by 12.6% when presumed or confirmed TB cases underwent respiratory separation (27, 28). However, in both these studies, the sample size and the number of outcomes were small (45/306 TB cases before and 5/237 after the intervention); also, the intervention was implemented in combination with other IPC measures.
Evaluated studies seemed to indicate the positive effect of respiratory separation in reducing the risk of acquiring LTBI or of developing active TB disease, particularly in individuals attending health care settings (e.g. non-health workers). However, in this review, isolation of TB patients seemed to have an inconspicuous effect or no effect on the risk of active TB disease among health workers, as indicated earlier.
The recommendation given here was set as conditional, based on the limitations of the data (small estimates of effect and large variance), and the various factors that national authorities need to take into account to ensure that TB-specific IPC measures are properly implemented. The comprehensive and effective implementation of IPC measures relies on the measures being implemented as a package. Also, health care authorities need to consider the value that patients place on the interventions, especially because of social alienation, stigma and financial impact. The Guideline Development Group argued that although respiratory separation or isolation measures are commonly used in various settings as basic measures in IPC practices, current evidence suggests that such measures alone are insufficient to help reduce the risk of transmission, especially among high-risk populations.
The Guideline Development Group emphasized that the risk of transmission of airborne pathogens can increase as a result of inadequate infrastructure of health care facilities, inconsistent use of personal protective equipment such as respirators by health workers, and staff shortages, coupled with lack of knowledge of basic IPC. The panel expressed concerns about the notion of separation of patients without the implementation of treatment and proper airborne precautions, including airborne precaution protocols. Also, the panel emphasized that success in reducing transmission would depend on how well the interventions were implemented and what standards were followed by those implementing the interventions.
It is critical that national health authorities and public health policy-makers consider these recommendations in the context of the burden of disease; the strengths and weaknesses of health systems; and the availability of financial, human and other essential resources. Additionally, they should be aware that the data assessment and conclusions reached by the Guideline Development Group supported the implementation of respiratory separation in certain circumstances (provided that rapid initiation of effective anti-TB treatment is in place), and other measures to prevent or reduce M. tuberculosis transmission.
Current recommendations on models of care for all TB patients – including the management of cases with DR-TB, and recommendations on patient care and support – have been described elsewhere (45–47). A decentralized¹ model of care is recommended over a centralized model for TB patients (including those on DR-TB treatment). However, this model of patient care may not be appropriate for patients for
whom treatment adherence is of concern, severely ill patients with extremely infectious forms of the disease or serious comorbidities, or cases where there are important barriers to accessing other forms of ambulatory care (e.g. outpatient or community-based care). In such situations, an individual risk assessment should be considered; this assessment should follow a human rights-based approach to TB, balancing the potential risks and benefits of the proposed interventions (i.e. respiratory separation or isolation) to the patient with the potential risks and benefits to health workers and the community in general.
Health care systems must implement available patient care and support measures before resorting to isolation of any person. In situations where isolation is required, this should be decided in consultation with the patient, and carried out in medically appropriate settings.
The Guideline Development Group did not address the use of involuntary hospitalization and incarceration of TB cases.
For the adequate implementation of isolation, it is important that health care authorities and those implementing the interventions consider the rights and freedoms of TB patients, balancing such individual liberties with the advancement of the common good (47).
The use of respiratory isolation or separation measures for TB patients can present several challenges, especially if:
• such measures are not implemented through clear protocols;
• facilities do not meet minimal standards for implementation;
• staff are not trained; and
• the undesirable effects (e.g. perception of alienation) for those affected are not considered.
Appropriate financial resources would be required to provide proper respiratory separation or isolation measures in such a way that the intervention protects the rights of the patient, and does not increase the risk for health workers or other persons attending health care or settings with a high risk of M. tuberculosis transmission. In situations where respiratory isolation is not feasible, health care facilities should consider the use of referral systems, in consultation with the patient.
Patients admitted to isolation have higher rates of anxiety and depression than other hospitalized subjects (48, 49). Therefore, it is essential that patients are informed of the rationale for respiratory separation or isolation measures, and that psychological support is provided to patients who are isolated. In addition, health care staff should be trained in the identification of anxiety and depression in TB patients, and the provision of the necessary support. Mental health risk assessments can be conducted to inform isolation decisions, to discuss supportive measures with the patient and their families, and to provide opportunities for the patient to participate in decision-making, as appropriate.
Although evidence on physical separation at home, including specifications of such, was not evaluated in this systematic review, it is important to emphasize current recommendations on decentralized models of care (46). In situations in which patients are considered to be infectious and care is being provided at decentralized facilities (e.g. patient’s home), patients and family members providing care should receive clear guidance and indications on IPC, particularly if the TB patient is receiving palliative and end-of-life care.
Settings and target population
The use of respiratory isolation or separation measures applies to health care settings, as well as other settings with a high risk of M. tuberculosis transmission (congregate settings where health care services, including hospitalization is provided, such as correctional facilities), regardless of the burden of TB disease in the community.
Initiation and duration of isolation
The systematic review¹ attempted to estimate the effect of effective treatment on the infectiousness of TB cases, to guide the duration of isolation (see Annex 3). However, the temporal dynamics indicating when effective treatment renders the patient noninfectious could not be ascertained in the present review. Where management policies differ across settings, in some settings individuals with infectious TB are separated at the outset of treatment. However, elsewhere, priority areas of patient care (e.g. treatment supervision, treatment adherence interventions and decentralized models of care) have been recommended. If these measures fail and there is an increased risk of transmission of M. tuberculosis to the community, health care authorities can resort to isolating a patient.
In situations in which patients are isolated, de-isolation should be based on the likely infectivity of the individual case and the availability of other supportive systems (in particular, decentralised models of care).
Patients who are isolated for extended periods of time, regardless of disease, have been shown to experience greater levels of anxiety, depression, anger and feelings of imprisonment; this is difficult for patients and their families.
Health care authorities need to allocate enough resources, based on a needs assessment, to strengthen the implementation of IPC interventions.
Data on the cost of the intervention were not extracted or captured in this systematic review; however, members of the Guideline Development Group discussed the allocation of resources, and noted that this would vary, depending on factors such as existing structures, burden of disease, and respiratory separation or isolation measures (e.g. open-wall concept versus closed-wall isolation rooms).
The Guideline Development Group did not assess any evidence on the implementation of respiratory separation or isolation measures in children.
Evidence and justification
Evidence continues to mount that delays in initiation of effective TB treatment increase the probability of onward transmission of the disease (50, 51).
The current systematic review identified four observational studies evaluating how provision of effective treatment (based on TB drugsusceptibility testing [DST]) for TB patients can have an effect on the burden of LTBI among health workers (19, 20, 40, 42) (see Online annexes 4 and 5). The included studies did not assess the incidence of TB disease among health workers. Evaluations in health workers in health care settings where patients rapidly received effective treatment based on DST indicated an absolute risk reduction of 3.4% compared with settings where effective treatment was delayed. The review also identified one retrospective cohort study that evaluated the protective effect of specific IPC measures in other persons attending a New York City hospital (27). Results suggested a reduction of 6.2% in incidence in active TB disease among HIV-positive individuals admitted to the ward, from 19/216 (8.8%) in the period before the intervention to 5/193 (2.6%) after implementation (P = 0.01).
To better inform some of the recommendations outlined within these guidelines, an additional systematic review¹ was undertaken to determine changes to infectiousness once effective anti-TB therapy has been initiated. Considerable variation was noted in the time taken for patients with proven drug-susceptible pulmonary TB receiving appropriate first-line TB treatment to achieve smear and culture conversion (see Annex 3).
The scope of this systematic review was not to evaluate a particular treatment regimen, but rather to assess the effect on onward transmission of timely administration of effective TB treatment.
TB treatment has a direct effect on survival of TB patients; it also has the potential to indirectly decrease M. tuberculosis transmission, provided the treatment is effective (i.e. treatment is appropriate, based on DST results) and administered in a timely manner. In attempting to assess the former, the Guideline Development Group found insufficient data in the systematic review to determine the real impact or effect that administering effective TB treatment has on health workers and on other at-risk groups; however, the panel considered that the desirable effects (i.e. the potential benefits) of the use of treatment outweighed the potential undesirable effects or harms (e.g. adverse events) that could arise from this medical intervention. The rationale for deciding on a strong recommendation in favour of using effective (and timely) TB treatment – based on very low certainty in the evidence – was further informed by discussion of the paradigmatic situations in which a strong recommendation is warranted, despite low certainty in the estimates of effect (52, 53). Members of the Guideline Development Group referenced the first paradigmatic situation, where low certainty evidence suggests a benefit in a lifethreatening situation, not only for the patient themselves but to the strong benefit of others who may be exposed to sources of infection, including transmission of resistant strains of M. tuberculosis. The Guideline Development Group placed a high value on the benefits of both effective and timely TB treatment at the individual (i.e. patient) level, and the potential reduction in harm (i.e. in transmission) at the community level, given the small incremental cost (or resource use) relative to the benefits. The review, guided by Background question 3 (see Annex 3), attempted to clarify the period after which TB patients are likely to become less infectious once they have started on effective TB treatment. Bacteriological culture conversion signifies a clear reduction of infectiousness, but does not usually occur during the first weeks of treatment. Many experts believe that reduction in infectiousness takes place much earlier than culture or smear conversion; for example, during the first 2 weeks on effective treatment for drugsusceptible patients. To determine the time point at which patients may not be infectious, four eligible studies were reviewed, with experimental data from animal-based models – using guinea-pigs as sensitive air samplers exposed to air exhausted from dedicated isolation rooms in which human patients with TB were treated. All four studies presented data suggesting that patients on TB treatment are less infectious to guinea-pigs than patients not receiving effective TB treatment, but none of the studies had data indicating the time it takes for a patient receiving effective treatment to become non-infectious to guinea-pigs (see Online annexes to access the data analysis report). The included studies did not effectively capture or stratify data by cavitary disease or cough behaviour.
The Guideline Development Group further emphasized that the estimates of effect of included studies presented a challenge because of the composite nature of interventions on IPC, leading to a high level of indirectness, as presented in selected studies. Other factors (e.g. the applicability of the evidence) were also questioned by the panel. All selected studies were conducted in the USA, during the mid-1990s, principally in HIV wards reporting outbreaks of DR-TB (19, 20, 27, 40, 42).
When assessing the evidence, the Guideline Development Group noted that treatment of patients needs to be guided by the use of DST, something that is important for field practitioners and implementers when putting these recommendations into practice. As currently recommended by WHO, universal access¹ to DST for M. tuberculosis should be a standard practice in all settings. DR-TB cases treated with first-line regimens are likely to continue to be infectious and propagate ongoing transmission.
National TB programmes must also consider the implementation of other interventions that facilitate treatment adherence, including strengthening of social protection systems for preventing financial hardship; providing nutritional support, and patient and family health education; and implementing decentralized models of care. Instituting this intervention without sufficient support measures may deter patients from continuing treatment.
Overall, health care infrastructure and available resources will determine access to rapid diagnostics, including DST and, most crucially, sustainable access to anti-TB medication.
Access to effective TB treatment is clearly essential for TB patients to be cured; however, benefits also accrue to the larger community and population in reducing the risk of transmission.
Evidence and justification
Respiratory hygiene (including cough etiquette) to reduce the dispersal of respiratory secretions that may contain infectious particles has been used as an additional measure to prevent M. tuberculosis transmission. Although there is literature on understanding the dynamics of cough aerosols of M. tuberculosis, data for comparing the effectiveness of respiratory hygiene manoeuvres are scarce, especially data on humans. Respiratory hygiene (or hygiene measures) is defined as the practice of covering the mouth and nose during breathing, coughing or sneezing (e.g. wearing a surgical mask or cloth mask, or covering the mouth with tissues, a sleeve, or a flexed elbow or hand, followed by hand hygiene) to reduce the dispersal of airborne respiratory secretions that may contain M. tuberculosis bacilli.
The systematic review identified a total of five relevant studies: four before-and-after studies (18, 24, 25, 28), and one animal model measuring the effect of surgical masks used by MDR-TB patients on transmission to guinea-pigs exposed to ward air (55) (see Online annexes 4 and 5). Meta-analysis was precluded because of significant differences between the interventions that were evaluated and potential differences between study populations, which also made it difficult to calculate crude estimates. All studies, apart from the animal study, reported on the effect of composite interventions (i.e. interventions that combine multiple components).
A reduction in the incidence of LTBI was observed in two of the included studies (a reduction in TST conversions in the intervention group compared with the control group). One study showed a reduction of between 4.1 and 12.4 TST conversions per 1000 person-months among health workers (18); the second study indicated that the use of surgical masks by people with presumed or confirmed TB was associated with 14.8% risk reduction in incident TB infection among health workers (24). Estimates from the two studies in which TB disease was measured showed a slight or no reduction in TB incidence in health workers after surgical masks were used by patients; the assessment of the effect of respiratory hygiene on the development of active TB disease in health workers showed a reduction in incident TB of 0.29 cases per 100 person-years in one study (24) and of 0.5% in another (25).
Two additional studies assessed the impact of surgical masks used by patients on the burden of TB infection and disease among other persons attending health care settings (28, 55). A prospective cohort study using an animal model evaluated the role of respiratory hygiene in reducing transmission of M. tuberculosis in settings with a high risk of M. tuberculosis transmission. A retrospective study assessed the effect of implementing IPC measures on transmission during an MDR-TB outbreak, in an HIV ward in Italy (28). The prospective cohort study quantified the effect of surgical masks (worn by MDR-TB patients) on incident infection among pathogen-free guinea-pigs exposed to ward air (55). The study found that 76.6% of animals exposed to air from patients not wearing surgical masks (the control group) became infected with M. tuberculosis. In contrast, only 40% of animals exposed to exhaust air from patients wearing masks (the intervention group) acquired infection.¹ The effect of the intervention in animals was extrapolated to a representative control population derived from nine studies where outcomes were measured in humans. Based on this calculation, the intervention would be expected to reduce the incidence of infection from 6.5% in the control group to 3.4% with the intervention – an expected absolute risk reduction of 3.1%. This represents a relative risk reduction of 47.8%. The retrospective outbreak investigation found that no patients developed MDR-TB after IPC measures were fully implemented.
Despite the low certainty in the evidence, the recommendation given here was set as a strong recommendation. As per the five paradigmatic situations offered by the GRADE methodology, this discordance was justified given the potential for preventing a lifethreatening or catastrophic situation that could occur in the event that health care or nonhealth care individuals develop TB infection and progress to active disease. The Guideline Development Group stressed that, despite limited evidence on the impact of respiratory hygiene (e.g. surgical masks worn by infectious TB patients, and cough etiquette) in settings of interest, the use of this measure as part of a composite intervention can help to reduce transmission of M. tuberculosis. Such an effect was pronounced in the animal model included in this systematic review, which allowed a more direct assessment of the intervention.
In evaluating the evidence, Guideline Development Group members were concerned that the study using the animal model had been considered of low quality, owing to serious concerns about the indirectness of the data. Panel members argued that animal studies could provide a valid indication of the effectiveness of an intervention. In particular, guinea-pigs are more susceptible to acquiring TB infection than other models, and may show progression of the disease that displays many features of TB in humans. Members of the panel argued that the guineapig model of M. tuberculosis infection has been used as a valuable tool to understand and describe TB disease mechanisms, as well as its role in determining the effect of specific interventions; hence, if well-conducted, this model can generate high-quality evidence. However, because of the failure to randomize animals to a particular group, the panel argued that the certainty of the evidence should be downgraded by one level due to indirectness (see Online annexes 4 and 5). A growing body of evidence now suggests that failure to randomize and to employ blind outcome assessment contributes to exaggerated effect sizes in animal studies across a wide range of disease areas; it also fails to provide the foundations for extrapolating animal research findings to humans (56, 57).
The Guideline Development Group noted the limits and limitations of existing data; in particular, the group recognized the lack of data evaluating the effectiveness of other face covers (e.g. covering the mouth and nose with a cloth mask, tissues, a sleeve or a flexed elbow).
Overall, the Guideline Development Group reflected on the reasonable assumption that coughing is an important driving force for transmission of M. tuberculosis, and therefore supported a strong recommendation in favour of the use of respiratory hygiene to reduce the release of infectious airborne particles into the environment. The group also emphasized the feasibility of wearing surgical masks.
Settings and target population
The use of respiratory hygiene measures applies to individuals with confirmed or presumed TB in all health care settings, as well as to such individuals in other settings with a high risk of M. tuberculosis transmission (including households and non-health care congregate settings such as correctional facilities, and refugee and asylum centres), independent of the burden of TB disease in the community and the level of care of the facility (i.e. primary, secondary or tertiary).
Respiratory hygiene must be implemented at all times. The use of surgical masks, in particular, is of utmost importance in waiting rooms, during patient transport and in any situation which can lead to temporary exposure to M. tuberculosis (e.g. in physician offices).
TB continues to be highly stigmatized, with TB patients and their families experiencing considerable discrimination. In some settings, the use of surgical masks by patients may perpetuate social stigma and local misconceptions about TB (58). Thus, the Guideline Development Group emphasized the need to:
• consider health education to key stakeholders – including patients’ families, community members and health workers – to better understand the prevailing causes of discrimination and to implement targeted health education programmes;
• institute effective health counselling for patients as part of a comprehensive package of interventions within social protection systems;
• institute respiratory hygiene (including cough etiquette) as a standard practice for coughing patients; and
• provide “how to” information on wearing of surgical masks, during sensitization and educational activities with both patients and health workers.
Surgical masks are part of the standard medical supplies procured by health care facilities. Hence, the provision of surgical masks to patients and a related education programme may incur minimal additional costs, in health care as well as in non-health care settings (e.g. correctional facilities, and refugee and asylum centres).
National authorities will need to consider the additional costs of providing surgical masks to inpatients as well as to those eligible for home isolation, and those under palliative and end-of-life care.
Childhood TB is often paucibacillary, and likely to contribute little to transmission (59–62). The Guideline Development Group acknowledged that use of surgical masks in paediatric TB patients can have a negative psychosocial impact on children and families (63); nevertheless, children should be provided with masks until they are initiated in effective treatment to ensure that they are non-infectious.
The use of surgical masks may be poorly tolerated in severely ill patients. Therefore, health care authorities need to ensure the proper implementation of interventions within the hierarchy of controls for preventing M. tuberculosis transmission.
¹ Substantial heterogeneity was observed in these studies.
¹ In some persons who are infected with M. tuberculosis, the ability to react to tuberculin may wane over time. When given years after infection, the TST may have a false-negative reaction. However, false-positive reactions may also result due to recent vaccination with bacille Calmette-Guérin (BCG) or a boosted reaction to subsequent Mantoux skin tests.
¹ Some studies reported the use of specific respiratory measures when separating sources of infection whereas others referred to strict isolation procedures specific to acid-fast bacilli (AFB).
¹ “Decentralized care” was defined as care provided in the local community where the patient lives, by nonspecialized or peripheral health centres, by community health workers or nurses, non-specialized doctors, community volunteers or treatment supporters. Care could occur at local venues or at the patient’s home or workplace. “Centralized care” was defined as inpatient treatment and care provided solely by specialized DR-TB centres or teams for the duration of the intensive phase of therapy, or until culture or smear conversion (46).
¹ A systematic review of the literature was conducted to determine how the infectiousness of TB patients (ability to excrete viable bacteria and sustain transmission) changes after starting on effective TB treatment.
¹ As mentioned above, a systematic review of the literature was conducted to determine how the infectiousness of TB patients (ability to excrete viable bacteria and sustain transmission) changes after starting on effective TB treatment.
¹ WHO defines universal access to DST as rapid DST for at least rifampicin, and further DST for at least fluoroquinolones and second-line injectable agents in all TB patients with rifampicin resistance (54).
¹ The methods used to infect guinea-pigs result in high levels of exposure, compared with typical exposure in human populations. Consequently, the absolute proportion of animals with infection is expected to be higher in experimental animal studies than in human studies. To compare the findings in animals with those in humans, the absolute risk difference in a human population was estimated by applying the relative risk in animals to a typical population (based on the average infection incidence in nine studies). Therefore, both the expected absolute risk difference in humans and the relative risk in guinea-pigs are presented for animal studies.