Extraction table for wider population studies

Susan Baxter; Elizabeth Goyder; Duncan Chambers; Maxine Johnson; Louise Preston; Andrew Booth

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Baxter S, Goyder E, Chambers D, et al. Interventions to improve contact tracing for tuberculosis in specific groups and in wider populations: an evidence synthesis. Southampton (UK): NIHR Journals Library; 2017 Jan. (Health Services and Delivery Research, No. 5.1.)

Interventions to improve contact tracing for tuberculosis in specific groups and in wider populations: an evidence synthesis.

Show details

Contents

< Prev Next >

Appendix 3Extraction table for wider population studies

Detail of study	Methods	Results	Conclusions
Aissa et al., 2008³⁶
Type of document: journal article Study design: cohort study Country: France Population: individuals over 15 living in the environment of a patient with pulmonary TB and who were culture-positive. Mean age 41.5 years; 325 index cases and 2009 contacts (1575 completed screening) Quality notes: provides statistical analysis to support conclusions	Research methods: participants enrolled over an 18-month period for the first cohort and over a 12-month period for the second cohort. Aim to develop and evaluate a model for TB contact screening Staff involved: N/A Measures used: physical examination, TST and chest radiograph	Results/data: the mean number of contacts per case was 6 (1–122). 98% were BCG vaccinated. Overall infection rate for contacts who completed screening was 27%, 1% active and 26% latent infection Risk of TB infection in a contact was significantly related to receiving free health care (p = 0.005; OR 2, 95% CI 1.2 to 3.2) and also the contact being a smoker (p = 0.021; OR 1.6, 95% CI 1.1 to 2.4). High incidence of TB in country of birth was significantly associated with risk of infection (p < 0.0001; OR 2.2, 95% CI 1.5 to 3.2) Risk of infection was strongly associated with number of hours spent with index case, closeness of contact (at night: p = 0.0009; OR 2.1, 95% CI 1.3 to 3.3) and being a first-degree relative (p = 0.001; OR 2.1, 95% CI 1.3 to 3.3). Infectiousness of index case also important	Main conclusions: a number of risk factors are associated with likelihood of infection A significant proportion of infections may not be due to recent exposure
Anger et al., 2012³⁷
Type of document: journal article Study design: cohort Country: USA Population: Contacts (n = 30,561) diagnosed as having active TB who had been identified during previous investigations of 5182 TB cases. Aged > 5 years. 1% HIV infected, 56% non-household exposure. 39% not USA born Quality notes: provides statistical analysis to support conclusions	Research methods: TB registry in New York used to identify contacts during previous investigations over a 6-year period. Followed up, up to 4 years later Staff involved: N/A Measures used: TST, chest radiography for contacts with symptoms or positive TST results	Results/data: 48% completion rate for chemoprophylaxis Latent TB infection was diagnosed in 79% of contacts who initiated chemoprophylaxis, and 61% who later completed treatment, and 39% who did not complete treatment Highlights the limitations of using the TST to diagnose latent TB infection (TST can produce false-positive results for individuals who have received the BCG vaccine). IGRAs may better prioritise chemoprophylaxis for those with greater risk of TB. In addition, uptake and completion of chemoprophylaxis may be higher when latent TB infection is diagnosed with IGRAs The absolute risk reduction afforded by chemoprophylaxis initiation was estimated to be 1.1% (95% CI 0.6% to 1.9%). This equates to approximately 88 contacts needing to be treated in order to prevent one case of TB (95% CI 53 to 164)	Main conclusions: estimate of 102 contacts need to be evaluated per prevalent case diagnosed (95% CI 90 to 115) Contacts who completed chemoprophylaxis had the lowest incidence, although those who initiated and did not complete also had decreased incidence compared with contacts who did not initiate treatment Contact screening is therefore effective, even when completion rates are below ideal levels
Ansari et al., 1998³⁸
Type of document: journal article Study design: before and after comparative Country: UK Population: residents of South Glamorgan (index cases, n = 103 + n = 103; close contacts, n = 611 + n = 732) Quality notes: provides descriptive statistics only to support conclusions	Research methods: compared data from the TB contact tracing clinic, the Public Health Service Mycobacterium Reference Unit and the Consultant in Communicable Diseases Control, South Glamorgan Health Authority, at two time points: before and after a change to the contact-tracing protocol Staff involved: chest clinic staff Measures used: Heaf’s test and chest radiography	Results/data: previously close contacts had been invited for follow-up annual radiological surveillance. In the changed policy close contacts were either discharged or referred to the chest clinic following their initial screening with no annual follow-up Close contacts defined as members of the same household sharing bathroom and kitchen facilities, and very close associates such as boyfriend/girlfriend or frequent visitors to the home of the index case 97% of contacts screened under both protocols, old protocol 23% considered unnecessary, new protocol 14% considered unnecessary. 1% of contacts screened were found to have disease and treated. 3% were given chemoprophylaxis and 22% were vaccinated with BCG Some lapses in adherence to the protocol were found: out of 707 contacts screened, 181 were casual contacts who need not have been screened Compared with the results of the previous protocol, fewer contacts were unnecessarily screened. However, referrals to the chest clinic increased, and the number given chemoprophylaxis	Main conclusions: the revised protocol seemed to be as effective as the previous, more complex protocol Screening of casual contacts and contacts of extra-pulmonary TB cases is not cost-effective
Bailey et al., 2002³⁹
Type of document: journal article Study design: retrospective analysis of data Country: USA Population: Oklahoma; n = 294 contacts for the index patient, a 23-year-old HIV-positive man who had been in prison five times; n = 1019 contacts for secondary cases Quality notes: some statistical analysis	Research methods: reviewed available hospital admission charts, health department records, chest radiographs and prison records. Contact investigation data were taken from paper records. Used network visualisation and metrics to investigate the outbreak Staff involved: TB control staff Measures used: TST, genotyping closeness of relationship – reach, degree and betweenness	Results/data: the strength of each patient–contact relationship was defined by the local TB control staff as close (> 4-hour exposure indoors or in a confined space), casual (exposure other than close) or undetermined (relationship strength not able to be characterised) 42% of contacts had a positive TST. With the exception of hospital, work and school contacts, all categories of contacts had positive TST rates exceeding 40% The network diagram indicated that the index patient was directly linked to 56% and indirectly linked to 18% of secondary cases Reach, degree and betweenness scores were calculated for relationships between the index case and contacts. The highest 20 scores and lowest 5 scores for each metric were used to prioritisation Contacts prioritised using network analysis were more likely to have latent infection than non-prioritised contacts (OR 7.8, 95% CI 1.6 to 36.6)	Main conclusions: network analysis is useful in earlier detection of TB transmission and for prioritisation of contacts to complement standard contact investigation. It can be useful while awaiting genotyping results, which can take many months Data required to perform network analyses are already routinely collected and need to be organised into the proper format for analysis. Although the costs may be beyond some programmes, principles such as pursuing repeatedly named contacts could be adopted Decisions need to be made regarding how frequently network analysis should be used
Banner, 2013⁴⁰
Type of document: journal article Study design: descriptive Country: Australia Population: primary school children/staff (n = 260 contacts) Quality notes: narrative description of the investigation	Research methods: describes the methods used for investigation of one case (a teacher at the school) and outcomes Staff involved: TB co-ordinator at a chest clinic, public health unit director and head of local health service Measures used: TST and chest radiography	Results/data: the children and staff who the teacher had most contact with were screened initially Information sent to parents of these children and an information session was given to staff by the co-ordinator. A second round of screening was carried out for those judged to be at medium risk Media involvement and parental pressure led to screening of a further low-risk group. 260 contacts screened, with 18 students and 7 teachers infected (12% of high-risk group, 5.5% of medium-risk group and 1.5% of low-risk group). None developed active disease	Main conclusions: highlights the importance of holding information session for teachers, sending letters and factsheets to all parents and holding an open parents’ evening In addition, highlights the need for a central point of communication and for media departments to be alerted and updated
Bargman et al., 2013⁴¹
Type of document: report Study design: descriptive Country: USA Population: high school pupils and staff; n = 1249 contacts screened Quality notes: narrative description of the investigation	Research methods: describes the investigation around one index case Staff involved: 81 staff from county and state health departments, two county medical reserve corps members, representatives from two schools of nursing, one school district representative, five nurses from four health departments, a clerk and two people from the Centers for Disease Control and Prevention. More than 885 person-hours for the screening and 890 hours for the treatment Measures used: TST and IGRA blood test	Results/data: the investigation initially conducted in members of the household and teachers/students who shared at least two classes. It was later extended to all students and school personnel. Evaluation of all contacts with IGRA at the local laboratory was not feasible. A combined strategy using IGRA and TST was adopted. Those who were BCG vaccinated or who reported a positive TST were IGRA tested Local news media and internet social media reported the story and false information. Public meetings and meetings with news reporters were held to address concerns and perceptions about TB The index case later disclosed a number of non-school social contacts	Main conclusions: the investigation was complex and labour-intensive and required immediate availability of a large workforce It is important to counter public fears by providing simple, credible, accurate, consistent and timely information about an event, and to let the public know what action they can take
Behr et al., 1998⁴⁷
Type of document: journal article Study design: retrospective analysis of cases Country: USA Population: cases of TB in San Francisco, CA, between 1991 and 1996 with positive cultures who had been previously identified as contacts to active cases (n = 11,211 contacts) Quality notes: detailed description of links between cases and role of DNA fingerprinting	Research methods: used DNA fingerprinting to further examine links between contacts and active cases. Routinely collected data from the TB registry were analysed Staff involved: disease control investigators Measures used: TST, and chest radiography for those with positive TST	Results/data: people who spent an estimated total of at least 40–100 hours with the index cases in the 3 months prior to diagnosis or during the infectious period were considered to be ‘close contacts’ and those who shared the same front door with the index case were considered to be ‘household contacts’. For index cases with positive sputum smears, evaluation of extended family members and contacts at the school, place of work or social setting was initiated immediately, with further expansion dependent on numbers found with positive TSTs For index cases having negative smears but positive cultures, the initial investigation focused on close and household contacts, and the investigation was extended only if there was a higher-than-expected prevalence of positive TSTs in the inner circle. For culture-negative index cases, investigation was limited to close and household contacts. For index cases < 15 years of age, investigation was directed towards finding a possible source The study found that index and contact cases were infected with the same strain of TB in 38 instances (70%, 95% CI 56% to 82%); and 16 pairs (30%) were infected with unrelated strains. Unrelated infections were more common among foreign-born (risk ratio = 5.22; p < 0.001), particularly Asian (risk ratio = 3.89; p = 0.002)	Main conclusions: DNA fingerprinting demonstrated that 30% of contacts with TB developed the disease at nearly the same time as, but not as a result of transmission from the index case. Contacts may have other risk factors for TB so the infection may have come from a source other than the index case. Contact tracing may be a useful way of identifying individuals at risk
Bock et al., 1998⁴³
Type of document: journal article Study design: descriptive Country: USA Population: residents of a rural low-income county (n = 9 cases) Quality notes: narrative description of the investigation	Research methods: describes an investigation surrounding an outbreak in 1996 Staff involved: Measures used: TST, and chest radiography for those with symptoms	Results/data: initially 61 contacts were named by patients. When an outbreak was suspected, patients were reinterviewed by ‘experienced interviewers’, and patients, family and contacts were all interviewed, resulting in an additional 282 contacts; 19% of these had positive TSTs. Reinterview also established an illegal gambling group and other potential sites of infection such as bars and school Some contacts were missed because the normal daily connections between them were not recognised by investigators and social settings that were frequented were missed	Main conclusions: illicit social connections were not identified as contacts. As TB rates were low in the area there was no specialised/experienced team. The follow-up investigation was delayed by 19 months as an outbreak was not initially suspected
Borgen et al., 2008⁴⁴
Type of document: journal article Study design: descriptive Country: the Netherlands Population: supermarket employees and customers, 80 coworkers and estimated 23,700 inhabitants of the area around the supermarket (4.4 km²) Quality notes: narrative description of the investigation	Research methods: describes the methods used for investigation of one case (an employee), and outcomes Staff involved: Measures used: TST – Mantoux method; chest radiography for older adults, those with BCG vaccination, TB or a positive TST	Results/data: contacts were approached via letter to all households, locally distributed flyers, the internet and press releases. Invited to attend a local sports hall. A questionnaire was completed including demographics and frequency of visits 21,326 customers registered for screening; all but 56 of these were tested In total 15 cases of TB disease were identified by the contact investigation (12 of these were thought to be directly from the source case). 359 cases of latent infection were identified (34% of these were thought to be due to recent exposure). 114 individuals needed to be skin tested in order to identify one case	Main conclusions: the investigation could have been improved by limiting testing to TST only and restricting it to frequent (at least once per week) customers Use of IGRA instead of or in addition to TST could have improved positive predictive value of testing and enables TST for those with BCG vaccination The optimal size of a contact investigation is setting specific, and depends not only on resources and logistics but also on background prevalence of latent TB infection
Borraccino et al., 2014⁴⁵
Type of document: journal article Study design: retrospective data analysis Country: Italy Population: 833 TB cases and 4441 contacts in one region Quality notes: provides numbers investigated but no links with investigation method	Research methods: analysis of registry data over a 6-year period Staff involved: N/A Measures used: TST – Mantoux method Those with symptoms or positive test also had clinical and radiographic examination	Results/data: median number of contacts per case was 3; the highest number was 150. Those living in congregate settings showed a significantly higher number of contacts (risk ratio = 1.38, 95% CI 1.30–1.46). Homeless people and those not born in Italy had fewer contacts. Contacts aged > 35 years were more likely to be evaluated than those aged < 25 years (OR 1.45, 95% CI 1.05 to 1.94). Regular and household contacts were more easily evaluated	Main conclusions: more effort should be focused on younger TB contacts and those with results of sputum smear negative but sputum culture positive
Borrell et al., 2009⁴⁶
Type of document: journal article Study design: retrospective data analysis. Compares conventional contact tracing with molecular epidemiology methods Country: Spain Population: residents of one city n = 892 cases (contact tracing carried out for 613), 5087 contacts Quality notes: some statistical analysis comparing the methods	Research methods: analysed cases reported to the programme over a 2-year period Staff involved: N/A Measures used: smear positivity; IS6110-based RFLP analysis; mycobacterial interspersed repetitive unit 12 typing	Results/data: 30% of contacts were household. Contact tracing was not performed in 31% of cases – 3.5% lack of consent, 4.5% logistic difficulties, 11.6% patient living alone, 11.6% lack of referral to study by GP A household link and individuals under age 15 years were most frequently identified by conventional contact tracing (predominantly mother–son). Molecular epidemiological methods tended to identify non-household links and identified more individuals from precarious economic circumstances and social difficulties (p = 0.002)	Main conclusions: contacts identified via conventional tracing methods may differ from those identified using molecular epidemiological methods. Although household relationships are important, other links such as neighbourhood and leisure settings are relevant. In a sizeable proportion of cases identified the link with the source case was unknown
Canadian Agency for Drugs and Technologies, 2014¹
Type of document: report Study design: review and guidelines Country: Canada Population: staff and patients in hospital Quality notes: systematic review underpinned the guidelines	Research methods: systematic review Staff involved: unclear Measures used: TST	Results/data: in one trial identified – age was the only predictor of latent TB infection (mean age 40 ± 9 vs. mean age 36 ± 9 years; p = 0.036). There was a low contagiousness of the index case to staff members. The review found little evidence relating to contact investigation in hospital settings	Main conclusions: guidelines regarding contact investigation in other settings may be applicable to a hospital setting, although recommendations are based on a low level of evidence
Carbonne et al., 2005⁴⁷
Type of document: journal article Study design: descriptive Country: France Population: health-care workers and patients in Paris hospitals Quality notes: narrative description of the investigation	Research methods: describes the methods used for the investigation of six cases of health-care workers with infectious TB Staff involved: crisis team including members of a hygiene unit, clinical wards, laboratories, occupational health service, risk management staff, regional centre, health authorities, trained staff for helpline Measures used: TST for children, chest radiography and medical observation	Results/data: testing of patients was based on the degree of infectiousness of the worker, patient characteristics (lowered immunity, young children, those not BCG vaccinated), length of exposure and proximity Patients were notified by letter and GP was notified. Telephone helpline was set up. A press release and media campaign were used in some cases Number of contacts varied widely between cases No TB disease was identified; two potential latent cases had unclear links to the source case The methods used for the TST did not enable latent cases to be identified	Main conclusions: the overall response rate was low. The use of a free telephone number was very valuable Different screening practices used made analysis of information difficult
Castilla et al., 2009⁴⁸
Type of document: journal article Study design: descriptive Country: Spain Population: residents of a small village (n = 751) Quality notes: narrative description of the investigation	Research methods: describes the investigation procedures for eight index cases Staff involved: Measures used: TST, medical consultation, and chest radiography for those with positive TST	Results/data: initial contacts – family/friends/workmate or schoolmate. Investigation expanded to others living in the village in same age group (19–23 years). Summoned to medical consultation, interviewed and completed survey form including sociodemographics, health and disease, and public places frequented Close contact defined as exposure for > 6 hours per week, occasional contact as < 6 hours per week Close contact (friends/lived/work together) explained 24.7% of the cases detected. Sporadic contact explained 37.9%. Frequenting the same bars explained 33.3% of infections	Main conclusions: the cases in the cluster appeared to have no close relationship but frequented some of the same bars
Coleman et al., 2014⁴⁹
Type of document: journal article Study design: cost-effectiveness evaluation Country: USA Population: flight-related contacts Quality notes: economic model	Research methods: a return on investment model was used Staff involved: N/A Measures used: treatment costs	Results/data: the costs calculated included that for contact tracing and also for testing/treating TB disease and latent TB infection. The average cost per contact used in the model was US$16.76. Different states have different contact investigation processes and therefore a range of expenditures per contact was developed – US$28, US$47, US$134 and US$164 The model indicated that every US$1 spent on investigations and treatment resulted in more than US$1 of saving at moderate/high rates of infection and disease. Low rates of infection and disease resulted in negative returns	Main conclusions: at moderate/high rates of infection contact investigation and treatment was cost-effective for flight-related contact A modified contact investigation procedure with sputum culture and smear positive or cavitation on chest radiograph (instead of and chest radiograph) was more cost-effective
Collins et al., 2004⁵⁰
Type of document: journal article Study design: descriptive Country: USA Population: staff and patients of a veterans administration facility and local hospitals Quality notes: narrative description of the investigation	Research methods: describes the methods used for investigation for one case Staff involved: infection control professional, three investigators, regional public health service administrator and two additional investigators from the department of public health Measures used: TST	Results/data: used the ‘concentric circles’ approach to contact tracing. Skin testing performed at workplace of co-employees to facilitate compliance. Patients of the facility and ex-patients were sent letters requesting that they report for testing. The circle of contacts was expanded owing to potential contacts being compromised by additional health conditions. Initial testing and 3-month follow-up. Latent infection in 2% of coworkers, 2.4% of fellow patients, 46% of family members, 13% of closest contacts, 5% of contacts in congregate housing facility. Situation complicated by another family member identified with active TB Challenge in identifying symptoms of many contacts due to side effects of treatment for cancer. Changes in staffing rotas complicated identifying contacts	Main conclusions: the extent of cross-facility contact complicated the investigation, and the importance of good communication among services was highlighted TB should be considered as a potential comorbidity in patients being treated for other diseases such as cancer
Cook et al., 2007⁵²
Type of document: journal article Study design: analysis of existing patient data together with a questionnaire Country: USA/Canada Population: patients and contacts in three counties. TB patients, n = 87; contacts, n = 440; mean age 29 years Quality notes: predominantly narrative description of the investigation	Research methods: used social network analysis methods. Supplemented routine investigation procedures with an interview to collect data on places of social aggregation over a 6-month period Staff involved: unclear Measures used: TST, molecular genotyping – spoligotyping, multiple interspersed repetitive units variable number tandem repeats, IS6110-based RFLP analysis	Results/data: interviews with TB patients and contacts elicited 1056 places of social aggregation. TB patients not linked via conventional contact tracing were linked by mutual contacts or places of social aggregation For two of the counties few interconnections or common places were found, or groups connected by social network analysis were found not to be the same strain on genotyping. For the third county no association between TST positivity and densely connected contacts was found; however, places of social aggregation revealed a connected network. For this investigation an association between TST results and being in the denser area of a person–place network was found (p < 0.01)	Main conclusions: network visualisations can provide evidence of the presence or absence of case clustering before genotype results are available in some instances (one of three investigations examined)
Cook et al., 2012⁵¹
Type of document: journal article Study design: descriptive review Country: Canada Population: any Quality notes: narrative review	Research methods: provides an overview of the literature Staff involved: N/A Measures used: N/A	Results/data: strategies for prioritisation have given priority to household contacts and those at greatest risk. Closeness of contact is based on amount of time rather than environmental or social factors. Extension of contact tracing often depends on number of TST positives compared with background rate. The influence of the infectious period and contact risk factors is unknown. Guidelines and recommendations vary. The limitations of contact tracing in high risk or vulnerable groups are highlighted, with importance of casual contacts and locations not always recognised. The completion of treatment remains a significant barrier. There are currently no guidelines for the use of social network analysis, geographic information systems genomics or genotyping. These may be of particular use in high-risk communities	Main conclusions: focus needs to be on development of questionnaires, electronic data management, local capability and expertise, co-ordinated approaches, strategies and evaluation
Davidow et al., 2003⁵³
Type of document: journal article Study design: retrospective review of data Country: USA Population: employees of five workplace study sites (cases, n = 349; contacts, n = 724) Quality notes: narrative description of the investigation	Research methods: analysis of case and contact records, including TB interview records, clinic charts and TB case reports Staff involved: unclear Measures used: TST	Results/data: subset of data from the Reichler et al.⁹⁹ study For smear-positive cases, workplace investigations were carried out alongside household and social contact investigations at all sites. For smear-negative cases, however, two sites conducted⁵³ investigations conditional on the results of household and social investigations Over 30% of the investigations involved ≥ 20 contacts. The median number of contacts identified differed between smear-positive cases (9 contacts) and smear-negative cases (7 contacts; p < 0.04). 68% of cases were fully screened; of these 29% had a positive TST	Main conclusions: the potential for transmission of TB in the workplace needs further recognition. There was inconsistent and limited recording of data collected during the investigations. There were also differences between the locations with regard to who was selected for screening and who was used as the primary source of information Standard guidelines for workplace investigations, written workplace investigation policies and standard data collection practices are needed
Diel et al., 2004⁵⁵
Type of document: journal article Study design: examination of a cluster of cases during the study period Country: Germany Population: customers of a bar close to a red light district, next door to a hostel for homeless people – 38 patients (12 of no fixed abode) with four index cases, 421 contacts Quality notes: narrative description of the investigation	Research methods: further examined the contact tracing investigation using DNA fingerprinting Staff involved: public health staff Measures used: TST and DNA fingerprinting	Results/data: an average of 12.8 contacts per patient; five reported none Ten patients were not originally included in the investigation and were only identified as linked by DNA fingerprinting. There were reports of a fear of social discrimination and intentionally antisocial behaviour due to irritation over the contact-tracing procedure, which was perceived to be intrusive. Some of these cases reported trying to shield drinking partners from ‘bullying methods’ of the ‘health police’ by naming only contacts in more distant bars, or by naming their more transient contacts such as neighbours Persons who were presumed to be in close contact with eight of the patients showed no disease during the initial contact investigation, with some becoming ill after the investigation. Tracing of contacts was relevant for reaching a diagnosis only in two cases 40.1% of close contacts had positive TST results; 1.9% became ill Twelve of 20 cases with confirmed recent transmission could be determined only by DNA fingerprinting	Main conclusions: only 40% of the 20 cases with epidemiologically confirmed recent transmission were included in the contact investigation Conventional contact tracing is insufficient for the detection of chains of transmission in some harder-to-reach communities. DNA fingerprinting can not only provide important information regarding recent infection of one patient by another; it also allows structural weaknesses in an investigation to be identified Contact investigation should examine the location itself and not focus on personal contacts and inflexible radiography screening schedules There should be a focus on informing the persons concerned about symptoms of disease and, if appropriate, conducting rapid screening by chest radiography or sputum analysis
Diel et al., 2006⁵⁶
Type of document: journal article Study design: testing of contacts Country: Germany Population: contacts (n = 309) Quality notes: limited relevant data	Research methods: compared the effectiveness of the two tests Staff involved: N/A Measures used: TST and QFT-G	Results/data: QFT-G was unaffected by BCG vaccination status, unlike the TST	Main conclusions: in close contacts who were BCG-vaccinated, the QFT-G assay appeared to be a more specific indicator of latent TB infection than the TST
Diel et al., 2009⁵⁴
Type of document: journal article Study design: testing of close contacts Country: Germany Population: close contacts (n = 182) Quality notes: limited relevant data	Research methods: carried out testing using IGRAs Staff involved: N/A Measures used: QFT-G assay and the T-SPOT.TB test	Results/data use of either IGRA as a replacement for the TST would decrease the number of latent TB infection suspects to be investigated by approximately 70%	Main conclusions: IGRAs are more accurate than TST
Driver et al., 2003⁵⁷
Type of document: journal article Study design: retrospective analysis of investigations Country: USA Population: people in ‘congregate settings’ (schools (37% of investigations), workplaces (45% of investigations), drug treatment centres, single room hostels, other locations) n = 2740 contacts Quality notes: narrative description of the investigation	Research methods: analysed 100 investigations over a 5-year period Staff involved: outreach staff Epidemiologist, screening, education and training staff, and a co-ordinator Measures used: TST and genetic testing; symptomatic contacts referred for chest radiography	Results/data: testing carried out for high-risk groups as soon as possible and again 10–12 weeks later. For other testing was performed once (10–12 weeks after exposure) Decision to perform testing at a congregate setting (not just household) based on infectiousness of source case, size, crowding, windows of setting, characteristics of contacts such as age and immune status, and case clusters Use of written protocols, checklists and site-specific questionnaires. Contacts were notified by letter with plans for testing and educational materials included. Telephone reminders were used by schools. A health educator conducted group sessions on TB at sites Transmission classified as likely, possible, unlikely or unknown Exposure defined as number of hours per week the source case was at the site in the prior 3 months. Close contacts were defined generally as those spending > 8 hours per week with the source case 83% of contacts were tested, 20% were infected and 52% completed treatment Sites with likely infection compared with those unlikely tended to be those where the source case had a longer duration of cough (median 13 vs. 6 weeks; p = 0.01) cavitary lesions (84% vs. 44%; p = 0.01) Transmission defined as likely in 16%, possible in 7% and unlikely in 72% and could not be assessed in 5%	Main conclusions: transmission at congregate sites was uncommon (22% of investigations) and is resource intensive. TSTs after most contacts would have converted should be considered in low-risk groups. Treatment completion rates were poor
Duarte et al., 2012⁵⁸
Type of document: journal article Study design: comparison of data from two time periods (different strategies in use) Country: Portugal Population: residents of a metropolitan area (contacts in first period, n = 809; contacts in second period, n = 683) Quality notes: narrative description of the investigation	Research methods: data on prevalence of infection among contacts compared 2001–3 and 2004–6 Staff involved: public health professionals and family doctors Measures used: TST, evaluation of symptoms, and chest radiography	Results/data: during the first period the investigation interview targeted close contacts; during the second period visits to home and workplace were also included 67% of eligible contacts were screened in first period, 3% with active TB and 27% with latent TB. 83% completed therapy. Estimate 0.75 cases of infection per index patient identified In second period 87% of contacts were screened. Interviews identified 950 contacts; home and workplace visits helped to identify 2629 contacts. Estimate 1.4 cases of infection per index patient Although there was an increase in workload, resources did not change between the study periods	Main conclusions: expanding contact investigations to home and workplace visits increased the number of individuals screened and identified further patients with active and latent TB
Edelson et al., 2011⁵⁹
Type of document: journal article Study design: systematic review Country: Canada Population: travellers on public transport Quality notes: systematicreview	Research methods: systematic review of literature on TB transmission among bus or train travellers Staff involved: N/A Measures used: evidence of infection	Results/data: 12 documents were included. There was support for the possibility of TB transmission from active TB cases to co-travellers. In most reports exposure occurred daily over weeks or months. Ventilation was frequently reported to be poor The reports did not provide evidence regarding the precise risk to co-travellers or identify which may be at greatest risk	Main conclusions: contact-tracing decisions should be based on proximity to index case, duration of exposure and other risk factors such as infectiousness of index case or susceptibility of contact
Erkens et al., 2010⁶⁰
Type of document: journal article Study design: narrative review Country: the Netherlands Population: N/A Quality notes: descriptive overview of the literature	Research methods: descriptive overview of the literature, expert consultation and recommendations Staff involved: N/A Measures used: N/A	Results/data: key importance of establishing infectiousness of source case, the likelihood of infection among contacts and the risk of them developing TB Outdoors transmission is highly improbable. Room size, air circulation and ventilation are important factors in the dispersal of bacteria. Visits to potential transmission locations to estimate risk are recommended. Children < 5 years old are a main target of investigation IGRA tests more sensitive in detecting TB infection than TST. A positive TST should be followed by IGRA. Although neither are able to distinguish latent from active TB, some studies suggest that IGRAs superior to TSTs in predicting latent infection becoming disease, although this finding not consistent. Chest radiographs are usually normal in persons with latent infection The degree of exposure depends on intensity and duration. Contacts are classified into circles of priority groupings When only a TST is used, a cut-off point for positivity must be decided with decisions regarding sensitivity vs. specificity. Likelihood of infection and BCG status should be considered	Main conclusions: a risk assessment approach is needed. Tests to identify latent TB have variable predictive value
Faccini et al., 2015⁶¹
Type of document: journal article Study design: descriptive Country: Italy Population: workers at a call centre (n = 107) Quality notes: narrative description of the investigation	Research methods: describes methods used in an investigation for one case Staff involved: unclear Measures used: TST, clinical examination, chest radiography, pulmonary assessment, interferon gamma release assay if positive TST, genotyping	Results/data: concentric methods approach used. A source case had been identified several years earlier; however, no contact investigation had been performed beyond family members. Perceived stigma had led the case to claim that they were unemployed	Main conclusions: TB-related stigma has major implications on TB control programmes as it can lead to the incomplete identification of contacts. Establishing trust and rapport between public and patients is important, with training in interviewing staff important Genotyping was important to establish linkages
Forssman et al., 2006⁶²
Type of document: journal article Study design: descriptive Country: Australia Population: residents and staff of a nursing home Quality notes: narrative description of the investigation	Research methods: description of an investigation for one case Staff involved: unclear Measures used: TST, chest radiography and clinical assessment if positive TST	Results/data: no further cases of infection were found. The investigation highlighted that there was no policy in place regarding the screening of residents or staff in the district. Many staff were from TB-endemic countries and were TST positive	Main conclusions: highlights the need for TB screening
Fox et al., 2013³²
Type of document: journal article Study design: systematic review and meta-analysis Country: Australia Population: any Quality notes: systematic review including a large number of studies	Research methods: systematic review of studies reporting the prevalence of TB and annual incidence of TB among contacts of patients with TB Staff involved: N/A Measures used: any	Results/data: 108 studies included from high-income countries. There was an average of 5.1 contacts per index case in high-income studies (95% CI 2.3 to 5 contacts) The definitions of household contact and close contact varied considerably In high-income settings the prevalence of TB among contacts was 3.3% (95% CI 27.6% to 42.7%) and of latent infection was 34.8% (95% CI 27.6% to 42.7%) The incidence is highest in the first year after contact, and remains above background incidence for at least 5 years after exposure. Children < 5 years of age and people living with HIV were particularly at risk Foreign-born contacts were significantly more likely to have latent infection than locally born contacts in high-income countries (OR 3.39, 95% CI 3.10 to 3.71; p < 0.0001)	Main conclusions: contacts of TB are at a high risk of developing infection, particularly in the first year, although there is heterogeneity in reported prevalence. Many of the patterns of infection in contacts reflect that of the whole population. There is a need to demonstrate that contact tracing is more effective than case-finding alone
Funayama et al., 2005⁶³
Type of document: journal article Study design: analysis of test results Country: Japan Population: university students (n = 462) Quality notes: limited data of relevance	Research methods: compared QuantiFERON TB-2G (Qiagen Inc., Germantown, MD, USA) with TST Staff involved: N/A Measures used: QuantiFERON TB-2G and TST	Results/data: in the non-close-contact group, the QuantiFERON TB-2G-positive rate was only 0.8%. In the TST group strong tuberculin reactions with erythema of ≥ 30 mm were seen in 18.2% in the (most likely due to the previous history of BCG vaccination)	Main conclusions: QuantiFERON TB-2G is a useful method for diagnosing TB infection, especially among individuals who show tuberculin reactivity due to past BCG vaccination
Funk, 2003⁶⁴
Type of document: journal article Study design: descriptive Country: USA Population: Alaska natives Quality notes: narrative description of the investigation	Research methods: describes the investigation of an outbreak in one region Staff involved: public health nurses Measures used: TST	Results/data: describes the challenges of investigation in remote areas, including the interviewer being unfamiliar with the culture and being a non-native-language speaker Named contacts collected are not always screened. A workshop for health providers may be beneficial to improve investigations	Main conclusions: contact investigations are challenging and labour intensive
Gaber et al., 2005⁶⁵
Type of document: journal article Study design: descriptive Country: UK Population: locals who attended a house in south-west England including children (the public house had a playroom attached); n = 184 contacts Quality notes: narrative description of the investigation	Research methods: description of the investigation Staff involved: consultant in communicable disease control, respiratory physician, microbiologist, TB nurse, communicable disease control nurse, hospital and ward managers Measures used: chest radiography Contacts who had symptoms or an abnormal radiograph were screened by the respiratory unit of the local district general hospital	Results/data: the investigation centred on three avenues for contact tracing – the local public house, close-contacts and inpatient hospital staff and patients (as the index case was currently a patient) Staff and regular visitors to the public house were sent letters to attend screening, and all children were screened. Other potential contacts were also offered screening and letters were made available at the public house. The letter included education about TB symptoms, mode of transmission, the availability of effective treatment and details of a telephone helpline. Local GPs were kept informed. An open-access radiology service carried out the screening At the hospital patients who had > 8 hours’ contact and other close contacts with the index case were screened. The first seven patients who had undergone anaesthesia using the same ventilator after the index patient were also screened In total 15 contacts were treated for active TB and 13 were given chemoprophylaxis. DNA fingerprinting indicated all infections originated from a single source	Main conclusions: despite identifying few conventional close household contacts, a significant number of secondary cases were detected from tracing contacts at a single location Multidisciplinary team working is essential for the effective management of an investigation
Gardy et al., 2011⁶⁶
Type of document: journal article Study design: further analysis of a previous investigation using additional methods Country: Canada Population: residents of a community in British Columbia (n = 41 cases; included two children) Quality notes: narrative description of re-examination of the investigation	Research methods: compares data and conclusions drawn from contact tracing, social network investigation, DNA fingerprinting and whole-genome sequencing approaches Staff involved: nurses and trained interviewers Measures used: social network questionnaire and laboratory-confirmed TB	Results/data: the social network questionnaire was used subsequent to contact-tracing interviews with the cases. The ‘name generator’ questions related to drug and alcohol use; residential and travel history; places of social aggregation; and identification of contacts in the context of high-risk behaviours and locations. A single social network diagram was developed using software to characterise relationships between contacts During the contact investigation there was an emphasis on contacts of a paediatric case, in an effort to find the source of the child’s infection; however, a single source case could not be clearly identified. A social network approach using the social network questionnaire added to the investigation by revealing previously unreported social interactions and several locations frequented by infectious patients (two hotels, meal/community centres and crack houses) RFLP and 24 loci mycobacterial interspersed repetitive units variable number of tandem repeats whole-genome sequencing was carried out. This revealed two distinct TB lineages with 80% of contacts having both lineages. Social transmission networks were constructed to further examine relationships Genome sequencing allowed the social network to be divided into subnetworks associated with specific genetic lineages of the disease. It was also valuable in enabling removal of social relationships that could not have led to transmission according to the genomic data. This greatly reduced the complexity of the network and aided the identification of index patients	Main conclusions: genotyping and contact tracing alone did not capture the true dynamics of the outbreak Social network analysis outperformed contact tracing in identifying a probable source case as well as several locations and persons who could be subsequently targeted for follow-up DNA fingerprinting had suggested that the outbreak had a single TB lineage, whereas more in-depth whole sequence molecular epidemiology revealed two lineages
Gerald et al., 2002⁶⁸
Type of document: journal article Study design: describes the development and testing of a decision-support tool Country: USA Population: state of Alabama Quality notes: describes development and testing of the tool rather than usage	Research methods: generalised estimating equations and classification and regression trees were used to develop a decision tree for predicting a positive TST result in contacts. The tree was tested in a set of 3162 contacts Staff involved: N/A Measures used: TST	Results/data: the decision tree developed had a 9% sensitivity and 22% specificity. It had a false-negative rate of 7–10%. It was estimated that use of the decision tree could enable around a 20% reduction in number of contacts investigated Priorities for contacts to be investigated are: if the index case has cavitary disease, or the total exposure per month was > 120 hours, or the contact is below age 15 years If none of these criteria are met than a case should only be investigated if: the contact was exposed to a smear positive case in their home, or the contact was exposed in a place where the ventilation was minimal	Main conclusions: decision trees can be developed to assist in prioritising contacts for investigation
Gerald et al., 2003⁶⁷
Type of document: journal article Study design: development of protocols, standardised recording system and evaluation of an intervention for field workers Country: USA Population: TB field workers n = 6x small groups of 8–10 individuals Quality notes: an evaluative study, although very limited data regarding effectiveness	Research methods: examination of existing protocols and development of revised versions via focus groups; also new sheet to record information The intervention was underpinned by social cognitive theory and the health belief model. It consisted of a workshop and computer-based module with individual supervisors allocated. It focused on interviewing skills and behaviour to improve efficiency and effectiveness, including skills work, use of motivational strategies during interviewing, training on new protocols and forms, and case scenarios. Follow-up monthly meetings in addition to workshops. The training took place over a 6-month period The protocols had been piloted in two areas Staff involved: TB field workers (registered nurses and disease intervention specialists who are college graduates with training in TB) Measures used: feedback from managers and staff; review of records	Results/data: existing protocols required the investigation of those with close and prolonged contact; however, there was considerable variance among field workers regarding the meaning of these terms. There was also variance in understanding of methods for eliciting information and the use of ‘concentric circles’ analysis was apparent. There was a need to quantify information and standardise definitions using a new contact exposure and assessment worksheet The quality of the training sessions was rated at mean 4.61 and overall value of training was rated 4.71 (scale 1–5, 5 meaning excellent) Some further training was required when data entry errors and misunderstandings were identified	Main conclusions: attention should be paid to precisely defining terms, protocols should be standardised and resources should be devoted to training to improve adherence to protocols
Greenaway et al., 2003⁶⁹
Type of document: journal article Study design: review Country: Canada Population: divided into four groups (low/high background prevalence of TB) Quality notes: review of available data	Research methods: reviewed published data relating to the likelihood of tuberculin reactions in casual contacts Staff involved: N/A Measures used: TST	Results/data: casual contacts were defined as ‘persons sharing the same air, but having no direct contact with the index cases’. The hourly risk of infection reported among casual contacts ranged from 0.18% to 0.53%, and averaged 0.28% per hour Individuals exposed early in the course of the disease were less likely to have TB conversion than those exposed later in the course of disease Those with lower previous exposure to TB are more likely to be newly infected following a short duration of exposure (5 hours’ exposure for contacts from areas with low prevalence of TB will result in almost 50% likelihood of new infection vs. 200 hours’ exposure for individuals from countries with higher prevalence will result in 40% likelihood of new infection) Casual contacts with a high likelihood of previous exposure should undergo TSTs only if the initial index case was heavily contagious and/or the duration of contact was prolonged Casual contacts with a low likelihood of prior mycobacterial exposure should be tested following as little as 5–10 hours of exposure	Main conclusions: the decision to extend a contact investigation to a group of casual contacts in a workplace or school should be based on evidence of transmission from the index case to closer contacts the number of hours of exposure the likelihood, in the population to be screened, of previous exposure to mycobacteria
Gulati et al., 2005⁷⁰
Type of document: journal article Study design: descriptive Country: USA Population: employees of an industrial company described as having many foreign workers (n = 104). Index case was an immigrant from El Salvador who lived alone Quality notes: narrative description of the investigation	Research methods: examined a contact investigation for one index case to identify levels of infection and factors associated with transmission Staff involved: occupational health medicine team – physician and industrial hygienist Measures used: TST	Results/data: investigation included individual interviews and assessment of buildings and ventilation systems. The contact investigation was part of evaluation of the workplace for solvent and noise All 104 employees were recommended to be screened as the index case had worked in various areas; 97 completed screening. A screening questionnaire was developed and pilot tested that asked for information regarding risk factors for TB and symptoms Possible risk factors for transmission at the workplace were identified and ORs were calculated for an association with having a positive TST for each. Workplace risk factors included spending time in the lunchroom (OR 4.45, 95% CI 1.32 to 23.25; p = 0.004) and carpooling with the case (OR 5.54, 95% CI 1.32 to 23.24; p = 0.004) 37% were TST positive	Main conclusions: the workplace can be an important site for transmission; screening should be considered for workplaces with large number of employees from high-prevalence countries
Guzzetta et al., 2015⁷¹
Type of document: journal article Study design: development of epidemiological model Country: USA Population: residents of one county Quality notes: modelling of data	Research methods: development of a computational model using notification data during a 10-year period Staff involved: N/A Measures used: N/A	Results/data: the model indicated that the contact-tracing programme (using a household contact, then school/work contact, then wider contact approach) significantly reduced TB incidence (by 18.6%) and deaths (23.7%), compared with passive diagnosis only The model indicated that around one-fifth of recently transmitted cases are identified by contact tracing Consideration of smear-positive cases only has a negative impact on effectiveness rather than also including smear-negative cases (reduces incidence avoidance to 10.4% and deaths to 13.2%)	Main conclusions: a key aspect in success of the programme was investigation of contacts of smear-negative cases which nearly doubled the effectiveness. Although these cases have a lower rate of infectiousness, they contribute a substantial share to transmission rates
Higuchi et al., 2007⁷²
Type of document: journal article Study design: cohort study Country: Japan Population: High school students (n = 349) Quality notes: limited data regarding accuracy of testing rather than other elements of an investigation	Research methods: compared results from different testing methods and followed up participants over time Staff involved: N/A Measures used: TST, QFT-G and chest radiography	Results/data: QFT-G appears more specific than TST as contacts with positive TST and negative QFT-G responses were not offered prophylaxis, and none developed TB during 3.5 years of follow-up	Main conclusions: the replacement of TSTs with QFT-G, or combined use of TSTs and QFT-G, may be more useful in diagnosing true infection
Jackson et al., 2008³
Type of document: journal article Study design: qualitative Country: UK Population: Residents of Greater Glasgow (patients, n = 21; next of kin, n = 3); aged 7–73 years Quality notes: provides some qualitative data to underpin conclusions	Research methods: interviews Staff involved: N/A Measures used: N/A	Results/data: patients with TB understood the cause of TB as a pathogen which was spread by person-to-person contact and could be influenced by level of immunity, social and environmental factors. Most patients believed that they had acquired TB from an unknown infected person and from a short period of contact time. Often public places were suggested as infection locations, particularly confined or crowded locations. Known TB contacts or stereotypical patients were always male, with female participants often described poor, ‘tramp’ individuals. Infection was often perceived to be a matter of bad luck or to have occurred when the immune system was weakened (such as when ill)	Main conclusions: most patients understood the concept of airborne transmission, and contact, but prolonged contact was not thought to be required. Modes of transmission described included airborne, sharing utensils, consumption of infected foods/drink and exchange of bodily fluids Aligning contact tracing with these lay beliefs may improve the approach
Jackson et al., 2009²
Type of document: journal article Study design: retrospective examination of surveillance data, interviews with nurses and patients Country: UK Population: most residents of one health board area. Social connections of 64 patients were investigated, 26 patients were interviewed Quality notes: mostly describes number of links. Interview aimed to uncover this information rather than provide qualitative data	Research methods: social network enquiry approach using molecular epidemiology and staff/patient interviews to further analyse social connections in contact investigations over a previous 10-year period Staff involved: nurse specialists Measures used: IS6110 RFLP and spoligotyping	Results/data: 43 epidemiological links between patients were identified, with 14 of these newly uncovered by interviewing patients Associations detected by previous surveillance review were family–friend relationships, whereas over half of associations reported during the new interviews related to friends and socialising in public houses. Sixteen sites of exposure were identified; 54% of patients frequented more than one of these sites Fourteen previously unidentified links were found. Associations were not discernible for 45% of patients	Main conclusions: the use of a standardised interview schedule including social activities prior to diagnosis enables detection of time, place or person characteristics that link individuals
Jereb et al., 2003⁷³
Type of document: journal article Study design: retrospective data analysis Country: USA Population: data from 29 states Quality notes: limited data presented	Research methods: data from the Centers for Disease Control and Prevention Measures used: number of cases, number of contacts, number evaluated and number completed treatment	Results/data: the number of cases, number of contacts and incidence varied widely between areas Evaluation was carried out for 83% of contacts; 44% completed treatment 10% of cases had no contacts listed	Main conclusions: impact on prevention is limited by low number of infected contacts completing treatment
Joint Tuberculosis Committee of the British Thoracic Society, 2000³³
Type of document: journal article Study design: review and guidelines Country: UK Population: any Quality notes: review of literature to underpin recommendations	Research methods: review of literature Staff involved: N/A Measures used: any	Results/data: contact tracing may be a method of assessing and screening a local population with a high incidence of TB Cases of TB occurring as part of an outbreak can be linked using molecular epidemiological or DNA fingerprinting techniques, provided that they are bacteriologically proven Close contacts – people from the same household sharing kitchen facilities and very close associates such as boyfriend/girlfriend or frequent visitors to the home. A contact at work or in a hospital ward may be as close as a household contact. It is important to examine lifestyle to identify locations of contact. Examination of casual contacts is necessary only if the index case is smear positive, there is notable infectiousness (> 10% of household contacts infected) or contacts are unusually susceptible Contacts should be investigated for the period of time during which the patient has had respiratory symptoms (if unknown, for 3 months preceding the first positive sputum smear or culture) For airline passengers the risk to fellow travellers is small. World Health Organization guidance recommends that contact tracing should be taken when: the diagnosis has been confirmed the index case is smear positive and was symptomatic with cough at the time of the flight the duration of exposure was > 8 hours it is < 3 months since the flight For children, if the index case is smear positive (1) all children, irrespective of BCG history, who have been close contacts should be tested and (2) any children who have not received BCG vaccination who have had casual contact If there is evidence of transmission screening of casual contacts with BCG vaccination be initiated	Main conclusions: importance of location and period of exposure
Josaphat et al., 2014⁷⁴
Type of document: journal article Study design: retrospective data analysis Country: Portugal Population: cases and contacts from one disease centre (n = 61 cases) Quality notes: identifies factors associated with being identified, some statistical analysis	Research methods: review of case records over a 1-year period Staff involved: N/A Measures used: individual characteristics, numbers identified as contacts	Results/data: 67% of cases identified all their contacts; 32% did not 23% of contacts were identified by the public health unit and 76% were identified by the index case Being employed and not being a relative or cohabitant were risk factors for not being identified by the index case (OR 4.82, 95% CI 1.71 to 13.54, and OR 0.22, 95% CI 0.10 to 0.47, respectively) Contacts identified by the index case tended to be younger (mean age 33 years vs. mean age 40 years; p = 0.001). Being a drug user was not a risk factor for not being identified	Main conclusions: employed contacts may not be identified as readily as relatives and cohabitants by index cases
Kasaie et al., 2014⁷⁵
Type of document: journal article Study design: simulation modelling Country: USA Population: 2000 households Quality notes: development of a epidemiological model	Research methods: developed an agent-based simulation model of a TB epidemic. Compared household contact tracing with active case finding in the community Staff involved: N/A Measures used: incidence	Results/data: the model indicated that the maximum 5-year reduction in TB incidence achievable by household contact tracing was 10–15% (2–3% per year), although impact would be lower with imperfect coverage or reduced sensitivity The model suggested that TB incidence might continue to decline for 2 years and would remain below baseline levels for > 15 years after a 5-year contact tracing intervention. The addition of preventative therapy nearly doubled the estimated impact	Main conclusions: contact tracing can have substantial epidemiologic impact (up to 7% reduction in incidence per year) but only if it achieves relatively complete population coverage, is sustained over time, and includes preventative therapy. Short-term evaluations of contact tracing are likely to underestimate their long-term impact; therefore, contact-tracing evaluation should encompass longer-term evaluation of latently infected contacts
Kawatsu et al., 2015⁷⁶
Type of document: journal article Study design: retrospective review of data Country: Japan Population: residents of Tokyo (patients, n = 8; contacts, n = 376) Quality notes: calculates degree of relationship scores used for network analysis	Research methods: reviews data relating to an investigation for a TB outbreak surrounding one index case. Developed social network analysis matrices Staff involved: primary health centre nurse Measures used: TST	Results/data: relationship score was calculated based on nature of contact [household, work (same room, same floor, same building, shares smoking room)] Two values for degree of contact and one value for betweenness centrality were calculated. The OR was calculated for the association between the likelihood of latent TB being diagnosed and contact relationship score at each percentile rank The OR was not significant for the degree of contact score and the likelihood of latent TB at any percentile rank. There was a significant association for contacts with higher betweenness score and latent TB infection (p = 0.020; OR 2.12, 95% CI 1.14 to 3.96 at the 40th percentile). For contacts with betweenness scores ≥ 90th percentile they were 3.66 times more likely to have latent TB infection diagnosed	Main conclusions: betweenness scores (but not centrality scores) were useful to identify contacts who may be at greater risk of latent TB infection Social network analysis matrices can be useful during contact investigations; however, the complexity and time-consuming nature of the method at present reduces the potential for it to be incorporated into routine contact investigations
Kettunen et al., 2007⁷⁷
Type of document: journal article Study design: descriptive Country: USA Population: residents of one state (n = 87) Quality notes: narrative description of the investigation	Research methods: describes the approach used for investigation of one case Staff involved: public health nurse and infection control practitioner Measures used: TST	Results/data: family members, friends and coworkers were tested. Only people who had spent time in the same residence on a routine basis were infected. The source case was believed to have had TB for some time before diagnosis. A pre-employment TST had been carried out but the patient denied symptoms of TB and a co-existing condition blurred the symptoms	Main conclusions: contact tracing requires diligence and effective communication
Klovdahl et al., 2001⁷⁸
Type of document: journal article Study design: descriptive, further examination of contact investigations Country: USA Population: Houston, TX. Study focuses on 37 patients with active TB associated with an outbreak 1993–6 having identical DNA fingerprints. 70% identified themselves as gay, 10% identified as bisexual, and 70% reported that they had a positive HIV test Quality notes: narrative description of links between contacts	Research methods: describes an initiative to DNA fingerprint all new cases of TB during a 5-year period. Fingerprints were obtained and stored in a database and pattern-matching software was used. Newly diagnosed patients were approached and interviewed using the Houston Myobacteria Active Surveillance Form Staff involved: research staff Measures used: DNA fingerprint	Results/data: contact investigation had identified only 12 links among these 27 cases. The index case could not be linked to any other; half (51%) of cases could not be linked to another case 33 out of 34 of the cases could, however, be linked by location (44 bars/restaurants/cafes) A network diagram was constructed and centrality scores were calculated. About 80% of the patients were linked by other people or places, and individuals were often linked by multiple places providing several opportunities for infection. Based on the centrality scores, 6 of the top 10 most significant people/place elements in the network were locations, in particular bars	Main conclusions: DNA fingerprinting identified the size of the outbreak and recognition of the importance of location (bars) was central to understanding of the outbreak. Contact-naming investigation had been unable to identify links in the transmission network
Kowada, 2013⁷⁹
Type of document: journal article Study design: cost-effectiveness analysis Country: Japan Population: 20-year-old contacts in developed countries Quality notes: economic modelling	Research methods: assessed the cost-effectiveness of the different assessment methods using Markov modelling Staff involved: N/A Measures used: high-resolution computed tomography, chest radiography, QFT-G, TST and cost-effectiveness	Results/data: QFT-G followed by high-resolution computed tomography yielded the greatest benefit at the lowest cost (US$6308.65; 27.56045 quality-adjusted life-years) Cost-effectiveness was sensitive to BCG vaccination rate	Main conclusions: a strategy with QFT-G followed by high-resolution computed tomography strategy yielded the greatest benefits at the lowest cost. High-resolution computed tomography instead of radiography is recommended
Lambregts et al., 2003⁸⁰
Type of document: journal article Study design: investigation of the role of DNA fingerprinting and cluster monitoring to TB control Country: the Netherlands Population: national data Quality notes: outlines links established by using the method	Research methods: retrospective analysis of clusters to establish the impact of establishing cluster links via DNA fingerprinting on contact investigations 1995–2000 Staff involved: project nurse Measures used: number of cases	Results/data: DNA fingerprinting established an epidemiological link in 31% of clustered cases where no link had been assumed or documented. Cluster feedback significantly improved the confirmation of documented epidemiological links (p < 0.001) The additional information regarding cluster links seemed to have limited impact on contact investigations, however, with only 1% extended as a result of receiving cluster feedback	Main conclusions: DNA fingerprinting and cluster monitoring can be useful to confirm suspected epidemiological links and to identify new links for which transmission is not suspected It may be useful where links between cases of TB are vague or with long periods in between It may be regarded as a complementary strategy to contact tracing, with both approaches required
Langenskiold et al., 2008⁸¹
Type of document: journal article Study design: retrospective analysis of data Country: Switzerland Population: patients and contacts from one hospital in Geneva, excluding those HIV-infected or children (n = 3582) Quality notes: describes brief characteristics of the data	Research methods: medical record data examined from a 10-year period Staff involved: N/A Measures used: TST and chest radiography	Results/data: there was an average of 4.3 contacts per index case. Being of foreign origin, level of exposure and contagiousness of the index case were predictive of latent infection. Treatment completion rate was 67%; 0.2% of those screened were found to have active disease and 36% to have latent TB	Main conclusions: contact-tracing effectiveness relies on improving therapy acceptance and completion rates
Lobato et al., 2003⁸²
Type of document: journal article Study design: retrospective review of data Country: USA Population: children aged < 5 years resident in California (n = 164) Quality notes: brief description of characteristics of the data	Research methods: review of patient records from 38 contact investigations Staff involved: TB programme staff Measures used: TST	Results/data: 61% of contacts with a positive TST were started on treatment; no data for completion. Almost half of the investigations found two or more previously undiagnosed TB cases	Main conclusions: improved strategies are required for identifying young children with latent TB
Logan et al., 2003⁸³
Type of document: journal article Study design: development of a tool for self-evaluation of investigations Country: USA Population: any Quality comments: describes the process of development but no evaluative data	Research methods: used discussion and stakeholder engagement to develop tools and logic models based on the Centers for Disease Control and Prevention framework for developing evaluation tools Staff involved: nurse care managers and public health nurses Measures used: tool developed	Results/data: the logic model outlines resources required, activities and outputs detailing the processes and required standards of a contact investigation. A self-evaluation questionnaire was developed from the models. The questionnaire was piloted to refine the questions A transmission risk assessment checklist was also developed to be used at initial visits to the TB case/suspect’s home, work or school, and other places A decision tree was also developed that illustrates the contact investigation process at the patient level, from determining the infectiousness of the TB case/suspect to screening close contacts for TB infection and disease The checklist and decision tree were intended to be piloted, although no suitable investigations were started during the period of the study	Main conclusions: the paper refers to a number of tools that were developed; however, they are not provided within it. The focus of the paper is on using the framework for developing evaluation tools
MacIntyre et al., 1998⁸⁴
Type of document: journal article Study design: retrospective analysis of data Country: Australia Population: Victoria (n = 1142 contacts screened in 1991) Quality notes: narrative description of the investigation	Research methods: further examination of a contact investigation in 1991 Staff involved: unclear Measures used: TST and chest radiography	Results/data: chest radiography was overused and was the sole screening tool for nearly 40% of contacts. 80% of repeat radiography was carried out following a normal initial study. TST was underused and had been carried out for only 60% of contacts. 22% of contacts had received preventative therapy	Main conclusions: the underuse of guidelines led to inefficiencies in the investigation
MacIntyre et al., 2000⁸⁵
Type of document: journal article Study design: cost-effectiveness analysis Country: Australia Population: all Quality notes: used modelling methods to investigate a hypothetical scenario	Research methods: compared contact tracing as it had been carried out, if guidelines had been followed, and a hypothetical model Staff involved: N/A Measures used: costs in terms of cases prevented, cases found and contacts traced	Results/data: the cost for contact investigation during 1991 was estimated at AU$309,065 per case prevented. It was found that during this period prevention was not considered a priority, and few infected contacts identified received preventative therapy. The authors estimated that if guidelines had been correctly followed, the cost would have been AU$58,742 per case prevented. The cost of the hypothetical evidence-based model was estimated at AU$3881 per additional case prevented, although this would be impacted by lower referral rates, lower rates of preventative therapy and lower efficacy of preventative therapy than the rates used during the calculations	Main conclusions: case finding is expensive in all three models Clear programme aims, adherence to guidelines and high rates of preventative therapy are essential in order to achieve cost-effectiveness
MacLellan et al., 2015⁸⁶
Type of document: journal article Study design: qualitative Country: UK Population: contacts referred to a TB screening clinic in north London. Contacts, n = 30; nurses, n = 8 Quality notes: includes qualitative data to underpin conclusions	Research methods: questionnaires to contacts (half who attended half who did not) and interviews with clinic nurses Staff involved: specialist TB nurses and one nurse from the health protection unit Measures used: N/A	Results/data: people reported attending owing to concern regarding the severity of the disease and worry regarding other vulnerable people. TB reportedly kept a secret Good working relationships with GPs were reported as important in prompting non-attenders to attend, although it was recognised that people in hard-to-reach populations may not have GPs Importance of working with a community to allay fears, educate and reduce the stigma The need for contact screening could cause fears of eviction in shared housing. Sending letters to the patient for them to distribute could be helpful Most non-attenders reported that they were unaware of their missed appointment. Reasons for not attending were given as contact details being inaccurate, living in shared houses with muddled incoming post, having limited understanding of the need for screening, being unable to take time off work and having childcare issues. Nurses highlighted the importance of outreach workers in tracing contacts via home visits/telephone calls Walk-in appointments could be inconvenient for people with limited time available; mornings and different days were reported as best. A GP surgery location was more convenient than a hospital. Use of SMS reminders was recommended Need for TB services to raise their profile, for leadership and resources for administration, community outreach, and core nursing staff	Main conclusions: three core areas of awareness, hospital service delivery and leadership in the service were identified
Mandal et al., 2012⁸⁷
Type of document: journal article Study design: retrospective data analysis Country: UK Population: patients of a clinic in Edinburgh aged > 18 years. Index cases, n = 275; contacts, n = 24 Quality notes: limited data	Research methods: analysed data on screening episodes over a 3-year period Staff involved: Measures used: TST and QFT-G	Results/data: the programme screens close contacts, household contacts and casual contacts (mostly workplace) 14.7% of contacts declined screening Contact tracing for non-pulmonary TB is as important as for pulmonary TB. Active TB was identified in 4.3% and latent TB was identified in 21.7% of contacts of non-pulmonary patients The number of casual contacts in the study was small; difficult to conclude if this group should be routinely screened or not. 21.2% of casual contacts were found to have active or latent disease	Main conclusions: contact tracing should be carried out for non-pulmonary as well as pulmonary Screening of casual contacts could be limited to certain circumstances
Marienau et al., 2014⁸⁸
Same study as Coleman et al. 2014⁴⁹ Type of document: journal article Study design: cost-effectiveness analysis Country: USA Population: flight-related contacts (n = 9284 contacts) Quality notes: cost-effectiveness analysis	Research methods: economic analysis comparing a previous to a modified protocol for contact investigation for in-flight exposure. Data are from 2007–9 Staff involved: N/A Measures used: transmission risk and cost	Results/data: the modified protocol changes the criteria for a contact investigation from within 6 months of the flight to within 3 months of the flight. Also, the criteria are changed to require sputum positive for TB by culture or nucleic acid amplification test and sputum smear positive for acid-fast bacilli and cavitation on chest radiography Applying the new protocol, it is estimated that 81 fewer potential cases and 409 fewer contacts would be investigated (half of the investigations), of whom 115 might test positive (three with active disease). The risk for new protocol is 1.4–19% and for old protocol is 1.1–24% Estimated cost under old protocol is US$222,000–1,300,000 and under new protocol is US$99,449–584,824	Main conclusions: the new protocol is more cost-efficient while retaining an acceptable level of public health risk
Marks et al., 2000⁸⁹
Type of document: journal article Study design: retrospective data analysis Country: USA Population: data from metropolitan areas of more than 5000,000 residents. Sample of n = 1080 index cases and n = 6225 close contacts Quality notes: describes characteristics of the data	Research methods: data reported to the Centers for Disease Control and Prevention Staff involved: public health nurses, outreach workers, TB programme personnel and health department staff Measures used: TST	Results/data: median of 6 days to patient interview (mean of 22 days). Procedures differed between sites, including who supervised workers and what screening contacts received. All sites defined household contacts as close. 68% of patients identified were household contacts, 24% were non-household relatives, 21% were leisure contacts, 5% were coworkers and 17% were other types of close contacts. One-third of patients identified household contacts only The number and type of data on contacts varied. Risk factors for disease were often not recorded. There was a median of 4 (mean of 6) close contacts per patient. A home visit led to average two additional contacts (likely to be young children). Fewer close contacts were identified by homeless people, men and Asian/Pacific islanders. No contacts were identified for 8% (88). Homelessness was significantly correlated with having no identified contacts (RR 1.3, 95% CI 1 to 1.5) High TST conversion rates among foreign-born contacts may be the result of prior infection or boosting rather than of recent infection. This should be considered before expanding investigations	Main conclusions: contact investigation could be improved by consistently defining a close contact and ensuring that patients list non-household as well as household close contacts Provision of targeted TB screening and access to care is needed for high-risk contacts
Marra et al., 2008⁹⁰
Type of document: journal article Study design: cost-effectiveness analysis Country: Canada Population: drew data from a provincial population-based database and from published literature Quality notes: cost-effectiveness analysis	Research methods: economic modelling using Markov model Staff involved: N/A Measures used: cost-effectiveness	Results/data: the most cost-effective strategy was to administer QFT-G in BCG-vaccinated only contacts, and to reserve TSTs for all others Incremental net monetary benefit was CA$3.70 per contact for BCG vaccinated only. The least cost-effective strategy was QFT-G for all contacts (incremental net monetary benefit of CA$11.50 per contact)	Main conclusions: QFT-G should be used in a targeted fashion
Andre et al., 2007⁹¹
Type of document: journal article Study design: examination of records relating to a cluster of cases, further interviewing of patients Country: USA Population: contacts from community, prison, hospital, school (n = 251). Total of 1039 contacts visualised Quality notes: calculated scores for relationship used in network analysis	Research methods: describes use of network analysis to produce visualisations and calculate measures of importance in the transmission network Staff involved: TB control staff Measures used: TST, clinical observation	Results/data: examined ‘reach’ ‘degree’ and ‘betweenness’ between the contacts in the network. 42% of contacts had a positive TST The diagram helped to link secondary cases with TB who were not named by the index patient. The majority of contacts could be linked to the index case; those who could not were investigated further The metrics calculated enabled contacts with higher scores to be prioritised. Three contacts with high ranking ‘betweenness’ scores were links to the overall network	Main conclusions: network analysis provides a means to identify linkages among cases, to quantify the magnitude of an outbreak and to begin control measures before genotyping results are available. It also can assist prioritisation of contacts for screening
Mohr et al., 2013⁹²
Type of document: journal article Study design: Delphi method Country: Germany Population: 23 topic experts involved Quality notes: describes development but not testing of the instrument	Research methods: describes development of a decision-making instrument for contact tracing in TB and meningococcal disease after contact on public transport Staff involved: N/A Measures used: N/A	Results/data: experts were asked to rate elements on a scale of low indication for contact tracing/neutral position/high indication for contact tracing. Nine elements for TB were identified: Symptoms of index case Infectiousness of index case Drug resistance pattern of index case Evidence of transmission to other contact person Quality of contact between index case and contact person (face to face/social interaction) Proximity of contact to case during exposure (more/less than 1 m) Duration of exposure (more/less than 8 hours) Susceptibility of contact (< 5 years of age/HIV/substance abuse/other disease) Environmental factors (external ventilation present or not/with/without circulation)	Main conclusions: the tool may help rapid decision-making
Muecke et al., 2006⁹³
Type of document: journal article Study design: descriptive Country: Canada Population: University students (n = 1144) Quality notes: narrative description of the investigation	Research methods: describes the investigation around one index case and calculates the risk factors for a positive TST Staff involved: public health unit, occupational health Measures used: TST, observation of symptoms, chest radiography for those with positive TST, genotyping	Results/data: used the ‘concentric circles’ approach. Began with family and close social contacts. Close social contact was defined as every day or every other day. Regular social contact was defined as twice a week. Investigation expanded as high number of positive TSTs to other students and university staff. Local media were used to contact people at three rave parties Duration of exposure based on number of hours of lectures attended with index case Size, type of ventilation and air changes per hour calculated for lecture rooms. Divided into more then or less than 300 m³ 27.5% of students had positive TST results; three had active TB. 69% of close social contacts and family had positive TST results; three had active TB Majority of students shared only one course with index case. Those who were exposed for 3 hours per week over 12 weeks had an approximately fourfold greater risk of infection than those with less exposure. Main risk factors were > 35 hours’ exposure and smaller classroom (OR 6.6, 95% CI 1 to 44.9 and OR 5, 95% CI 1.4 to 10)	Main conclusions: the hourly risk of infection together with ventilation measurements can be a useful element of contact investigations
Mulder et al., 2012⁹⁴
Type of document: journal article Study design: retrospective analysis of data Country: the Netherlands Population: TB patients from the nationwide surveillance register 2006–7 (n = 904 patients); more than two-thirds were immigrant cases (half of these were asylum seekers or illegal residents) Quality notes: describes characteristics of the data	Research methods: the study analysed factors associated with the likelihood of having contacts of an infected patient traced Staff involved: not reported Measures used: frequency of investigation, characteristics of index patient	Results/data: contacts were investigated for 78% of cases Contacts were significantly less often investigated around immigrant index cases (OR 0.60, 95% CI 0.40 to 0.92) than around Dutch index cases. Contacts were significantly more often investigated for smear positive index cases (OR 3.52, 95% CI 2.23 to 5.55) and culture positive index cases (OR 2.71, 95% CI 1.76 to 4.16) than for smear negative and culture negative index cases, respectively. Contacts were significantly less often investigated around actively found index cases when compared with passively found index cases (OR 0.38, 95% CI 0.26 to 0.57) and around index cases who belonged to a risk group compared with index cases who did not (OR 0.44, 95% CI 0.30 to 0.65)	Main conclusions: contacts of immigrant index cases were significantly less often investigated than contacts of Dutch index cases. By not investigating the contacts of immigrant patients, there is a risk of missing a significant number of infected and diseased contacts
Mulder et al., 2012⁹⁵
Type of document: journal article Study design: qualitative Country: the Netherlands Population: public health nurses (n = 14) from different regions Quality notes: provides qualitative data to underpin conclusions	Research methods: interviews; each participant was interviewed around the time a contact investigation had commenced, and then again after the investigation had been completed Staff involved: public health nurses Measures used: N/A	Results/data: participants described identification as challenging, as index cases were not always able to recall or willing to share information Nurses tended to identify more contacts than just those at substantial risk of infection. There was a perception that as many as possible gave a more comprehensive view of the level of infectiousness of the index case. There were concerns regarding missing contacts and further transmission; those who were anxious or expected to cause conflicts were often classified as contacts even though the risk was low. Immunocompromised individuals were often not prioritised Assessing the level of infectiousness of immigrants was described as challenging owing to the high risk of infection in the country of origin. Native Dutch contacts were often included despite their limited exposure Half of nurses used the available prioritisation table. It was described as difficult due to limited ability to capture exposure locations, or the appearance of prioritising different groups could be problematic Terms such as daily, frequent or intensive were often used for level of exposure rather than terms in the guidelines. Nurses who were more experienced tended to use the table less Six investigations were appropriately scaled up and one was appropriately not scaled up. Seven were incorrectly scaled up to casual contacts. Guidelines regarding background prevalence in the community were ambiguous and, therefore, were often not used	Main conclusions: staff did not always adhere to guidelines and tended to identify more individuals as contacts than recommended, and there was evidence of scaling up to casual contacts which was not required The criteria for classification could be perceived as unhelpful/ambiguous and there was difficulty interpreting background population prevalence data. The usefulness of the ‘stone in the pond’ principle requires accurate data regarding prevalence in specific communities (such as immigrants)
Munk et al., 2008⁹⁶
Type of document: journal article Study design: descriptive Country: USA Population: residents and coworkers in Maryland (n = 287 contacts) Quality notes: narrative description of the investigation	Research methods: description of the investigation surrounding one index case Staff involved: TB control staff Measures used: TST, chest radiography, sputum acid-fast bacilli smear for those with symptoms	Results/data: household contacts, close social contacts, close workplace contacts and coworkers who travelled in a van with the index case were given high priority and evaluated within 7 days Family and social contacts with less duration of exposure and in more open environments were categorised as medium priority. Coworkers at sites of employment were classified as low priority High rate of infection in high priority cases (39%) led to expansion to low priority contacts. 15% of these had positive TSTs No cases of active TB were found. 71% of the medium or high priority cases with latent infection agreed to begin treatment, and 33% of low priority	Main conclusions: the workplace can be an important location for contact investigations
National Tuberculosis Controllers Association, 2005³⁴
Type of document: report Study design: guidelines Country: USA Population: all Quality notes: guidelines developed by expert consensus	Research methods: outlines guidelines for contact investigation developed by an expert working group Staff involved: varying by area Measures used: any	Results/data: although guidelines and standards are provided, unusually close exposure (prolonged exposure in a small, poorly ventilated space or a congregate setting) or exposure among particularly vulnerable populations at risk for TB disease such as children or immunocompromised individuals could justify starting an investigation that would normally not be conducted If contacts are likely to become unavailable then the investigation should receive a higher priority. Investigations may be affected by exaggerated concern regarding TB in a community and an investigation being demanded Outlines factors influencing prioritisation, including type of TB, level of infectiousness, age and HIV status Having written policies and procedures for investigations improves the efficiency and uniformity of investigations Establishing trust and consistent rapport between public health workers and patients is critical to gain full information and long-term co-operation during treatment. Workers should be trained in interview methods and tutored. Patients should be interviewed by persons who are fluent in their primary language or interpreters provided Provides recommendations for the content of interviews, a minimum of two of which is recommended Sites visits are required as they enable additional contacts to be identified and are the most reliable source of information regarding transmission settings. Details such as room sizes, ventilation systems and airflow patterns should be considered together with how often and how long the index patient was in each setting Data from the investigation should be recorded on standardised forms Priorities should be assigned to contacts and resources should be allocated to complete all investigative steps for high- and medium-priority contacts. Priorities are based on the likelihood of infection and the potential hazards to the individual contact if infected considering characteristics of the index patient, characteristics of the contact, and intensity, frequency and duration of exposure Provides detail on the contact interview content, decision-making for expanding investigations, involvement of the media and considerations for specific population settings	Main conclusions: provides a comprehensive overview of contact-tracing procedures and decision-making considerations
New York City Department of Health and Mental Hygiene, 2008³⁵
Type of document: guidance Study design: guidance Country: USA Population: city population Quality notes: details on basis for guidance not provided	Research methods: description Staff involved: Measures used: N/A	Results/data: provides definitions of terms Priorities for contact investigation based on both the characteristics of the known or suspected TB index patient and the characteristics of the contact Provides tables and flow charts for decisions to conduct or continue contact investigation. Provides a table outlining contacts most likely to be infected and contacts at high risk of developing TB once infected	Main conclusions: provides flow charts to underpin decision-making during investigations
Pettit et al., 2002⁹⁷
Type of document: journal article Study design: descriptive Country: UK Population: customers of a public house (cases, n = 12; contacts, n = 122) Quality notes: narrative description of the investigation	Research methods: describes the investigation carried out Staff involved: unclear Measures used: TST, DNA fingerprinting	Results/data: initial contact investigation focused on close contacts and revealed no further cases of infection. However, cases were later identified who all were customers of a public house where the presumed index case was a regular customer, and the investigation was extended to all staff/customers. 85 of 122 contacts were screened. The identification of regular patrons was extremely difficult. Two cases came to light as a result of health education/awareness raising One further case was identified by the investigation and two children were given preventative therapy (1.18% case detection rate)	Main conclusions: transmission can occur between customers of a public house; therefore, extended contact screening beyond close contacts must be considered
Pisu et al., 2009⁹⁸
Type of document: journal article Study design: cost-effectiveness analysis comparing conventional contact tracing with contact priority models Country: USA Population: residents of Alabama Quality notes: cost-effectiveness analysis	Research methods: a cost-effectiveness analysis using a decision-analytic model comparing traditional ‘concentric circles’ contact tracing with a contact priority model Staff involved: N/A Measures used: TST and costs	Results/data: the contact priority model uses a decision rule (exposure hours, home, poorly ventilated environment) to explicitly categorise contacts as high risk requiring testing or low risk not requiring testing The analysis used the total cost each model divided by outcomes (number of active TB cases, number of life-years attained) to calculate incremental cost-effectiveness ratios. A decision tree approach was used to model elements such as the likelihood of a TST being performed and Markov models were used to represent costs and outcomes Estimated cost of investigating a contact was US$250 Conventional ‘concentric circles’ contact tracing was found to be more effective but more costly than a contact priority model. Savings would be made on cost of TSTs; however, there would be higher costs from active disease in a contact priority model. The ‘concentric circles’ approach was estimated to prevent one additional case of active disease for a cost of US$92,934 and one additional life-year US$185,920. Estimated cost per 1000 contacts was US$339,896 for ‘concentric circles’ and US$294,596 for contact priority. There would be estimated one or two additional cases of active disease per 3000 contacts with a contact priority model	Main conclusions: conventional ‘concentric circles’ contact tracing was found to be more effective but more costly than a contact priority model
Rea and Rivest, 2014⁶
Type of document: guidance (grey literature). Standards for contact follow-up and outbreak management in TB control (chapter 12 of TB standards) Study design: guidance referencing underpinning literature Country: Canada Population: any Quality notes: no details on process for developing guidance	Research methods: descriptive Staff involved: public health/TB control authorities in collaboration with treating clinicians and other providers Measures used: N/A	Results/data: only respiratory TB is infectiousness in most cases. Contact investigation should be carried out for both smear-negative and smear-positive cases. Source case investigation is recommended for children < 5 years old with a diagnosis of active TB Contact tracing should identify and treat any secondary cases and identify individuals with latent infection in order to offer preventative treatment Interviews should include questions about locations/activities of exposure and also specific named contacts. Prioritisation of contacts should be based on infectiousness of the source case, extent of exposure and vulnerability of those exposed Most effort should be put into contacts who are most at risk of being infected and/or most at risk of developing active TB disease if infected High priority – household contacts plus non-household who are immunologically vulnerable Medium priority – close non-household contacts with daily or almost daily exposure, including those at school or work Low priority – casual contacts with lower amounts of exposure Initial contact investigation should be carried out for high- and medium-priority contacts for index cases with smear-positive/cavitary/laryngeal TB. For smear-negative non-cavitary pulmonary TB the initial contact should be for high-priority contacts only. Investigation should be expanded if initial testing indicates that transmission has occurred TSTs and symptom assessment 8 weeks after exposure is recommended for non-household contacts. Household and high-priority contacts should be tested initially and again at 8 weeks. Two-step TSTs are not recommended. A history of BCG vaccination does not alter the interpretation of the TST result. A positive test is ≥ 5 mm, or an increase of at least 6 mm from previous test Elderly people in long-term care should not be tested with TST In homeless people and those with drug addictions, non-judgemental and supportive staff and use of incentives may help rates of participation. Interventions on site for a single session are likely to have more success. The primary focus should be on detection of secondary cases rather than assessment and treatment of latent TB	Main conclusions: the concentric circles approach does not take into account contacts who are vulnerable but may have had less exposure, and can be difficult to apply in congregate settings. Level of priority should be considered DNA fingerprinting can be useful to confirm or disprove linkages. It can be useful in populations in whom contact investigation is challenging (such as the homeless) All cases should be asked about locations where they spend time. Social network analysis may be helpful School, workplace and other congregate setting investigations including homeless and other marginalised populations are best carried out on site Need for good organisation, adequate staffing and resources, clearly defined roles and responsibilities
Reichler et al., 2002⁹⁹
Type of document: journal article Study design: retrospective review of records Country: USA Population: five areas of the USA with programmes perceived as the best, and best organised data. Patients aged over 14 years (n = 360) and contacts (n = 3824) Quality notes: describes characteristics of the data	Research methods: data were analysed from contact investigations carried out in 1996 Staff involved: N/A Measures used: TST converters, numbers screened	Results/data: close contact was defined differently in different areas. Closeness was not recorded in records for many contacts. For analysis, ‘close’ was defined as members, visitors or workers in the index case household or those who were friends or relatives Number of contacts was higher for patients who had both a positive smear and cavitary disease (median 8; p < 0.001). 13% of patients had no contacts identified; an additional 11% had no close contacts identified. Patients with no contacts were more likely to live in a homeless shelter (13% vs. 2%; p < 0.001). Only half of those patients with no contacts were recorded as being interviewed. Less than two-thirds of contacts completed screening	Main conclusions: effective investigations require standard definitions of close contact and contact, definition of what constitutes exposure (duration, time, frequency, location), standard criteria for expanding investigations, development of effective data management systems and definition the extent of investigation needed in different settings
Rodriguez et al., 1996¹⁰⁰
Type of document: journal article Study design: descriptive Country: USA Population: high school students (close contacts, n = 122; non-close contacts, n = 1804) Quality notes: narrative description of the investigation	Research methods: describes the investigation surrounding one case Staff involved: public health staff Measures used: TST, cost	Results/data: positive TSTs were found among 2.5% of close contacts and 1.9% of non-close contacts. No other active TB case was identified The cost of TST screening was estimated to be US$36,507	Main conclusions: screening was costly and diverted staff from other duties. Existing guidance regarding expanding investigations should be followed
Ruben and Lynch, 1996¹⁰¹
Type of document: journal article Study design: descriptive Country: USA Population: residents of Pittsburgh, PA Quality notes: narrative description of the investigations	Research methods: describes the investigations carried out over a 1-year period Staff involved: NR Measures used: description only	Results/data: unable to source the full text; the abstract refers to the challenges of carrying out contact-tracing investigations, although provides no further details	Main conclusions: the authors highlight the challenges in carrying out contact investigations
Rubilar et al., 1995¹⁰²
Type of document: journal article Study design: retrospective review of data Country: UK Population: residents of Edinburgh (cases of TB notified, n = 632; contacts, n = 3688) Quality notes: describes characteristics of the data	Research methods: reviewed records of cases of TB notified 1982–91 (20% of records had been mislaid) Staff involved: NR Measures used: NR	Results/data: 7.9% of those cases notified had been identified by contact tracing 1.4% of contacts screened had active TB and 1% had latent infection 54% of contacts with TB were under age 14 years, 34% were aged 15–44 years, 8% were aged 45–64 years and 4%) were aged over 65 years 84% of the cases among contacts were in contacts of patients with sputum smear-positive respiratory disease, and 98% of TB in contacts was detected within the first 3 months of screening Infection that may be detected later by ongoing radiography may be the result of infection from another source or reactivation of infection acquired from another source	Main conclusions: screening of close contacts of all index cases other than smear-positive respiratory index cases was not recommended Screening of young people should be prioritised and carried out within the first 3 months
Sanderson et al., 2015¹⁰³
Type of document: journal article Study design: descriptive Country: USA Population: patients and staff on a maternity ward (n = 285) Quality notes: narrative description of the investigation	Research methods: describes the contact investigation methods using electronic health data (electronic medical records and an immunisation register) Staff involved: hospital staff, Department of Health and mental hygiene staff Measures used: none	Results/data: the linking of data from different electronic health systems reduced the resource burden required for the investigation. The identification, notification and evaluation of contacts was aided by the systems, and documented evidence of exposure assisted the focusing of those at greatest risk 100% of those potentially exposed were contacted The systems also identified current health-care providers who were contacted to distribute letters and guidelines	Main conclusions: electronic health data are useful for enhancing contact investigations
Saunders et al., 2014¹⁰⁴
Type of document: journal article Study design: retrospective analysis of data Country: UK Population: data collected at a Birmingham hospital (n = 7365) Quality notes: describes characteristics of the data	Research methods: analysed trends in data over a 20-year period (1990–2010) Staff involved: TB nursing service Measures used: screening completion rates, screening outcomes, number of contacts	Results/data: 40.9% of contacts failed to complete screening. There was no evidence of a trend over the study period The number of contacts screened for each positive screening outcome was 15 for pulmonary TB and 45 for extrapulmonary TB Contacts were less likely to complete screening if they were of working age, male, black or from the Indian subcontinent Contacts tested using IGRAs were more likely to complete screening	Main conclusions: work is required to increase screening completion rates and to increase screening for working age, black and Indian subcontinent populations
Shah et al., 2014¹⁰⁵
Type of document: journal article Study design: systematic review and meta-analysis Country: USA Population: patients with drug-resistant TB Quality notes: systematic review	Research methods: calculated the reported yield of contact investigations Staff involved: N/A Measures used: number of household contacts	Results/data: analysed data from 25 studies. The pooled yield was 7.8% for active TB and 47.2% for latent infection (there was significant heterogeneity; p < 0.0001). The majority of cases were identified in the first year	Main conclusions: there is a high yield of active and latent infection in household contacts of patients with drug-resistant TB
Shrestha-Kuwahara et al., 2003¹⁰⁶
Type of document: journal article Study design: qualitative Country: USA Population: pulmonary TB patients from the USA/Mexico (n = 54), programme staff (n = 18) Quality notes: provides qualitative data to underpin conclusions	Research methods: focus groups Staff involved: contact investigation staff Measures used: N/A	Results/data: patients reported providing contacts easily and willingly. On occasions, although they had given a large number of contacts only a few appeared in the records. Homeless people could only give street names of their friends and acknowledged that finding them was difficult when they had no place of residence Patients reported preferring to inform contacts themselves prior to the health department doing so Trust and good rapport were described as important, as well as good communication. Using interpreters or interpreters not being available could be challenges Differences in understanding of the term ‘contact’ and ‘at-risk contact’; term could be vague or understood incorrectly. Some patients were unclear about the purpose of needing names; it was not clear that the focus was not on getting the names but was on finding the people who may have been infected Fears regarding stigma, loss of employment or housing, and alienation or abandonment were described The timing of the investigation was when patients could be feeling ill and afraid and confused after diagnosis. Reinterviewing patients at a less stressful time could be helpful Staff reported need for training in counselling, improved systems of co-ordination between service providers and training in social and economic issues	Main conclusions: programmes may be enhanced by staff training in listening, culturally appropriate interviewing techniques and reinterviewing patients after initial fears and confusion have reduced There seemed a mismatch between most patients reportedly giving contacts willingly and the small number of contacts recorded. There may be misconceptions regarding understanding of what a contact means; effective communication is needed
Sprinson et al., 2003¹⁰⁷
Type of document: journal article Study design: retrospective analysis of data Country: USA Population: California (n = 15,582 contacts) Quality notes: describes the characteristics of the data	Research methods: analysis of programme management reports over 1 year to estimate effectiveness Staff involved: NR Measures used: TST, costs, contacts identified	Results/data: 11% of cases had no contacts elicited. Mean number of contacts per case was 10.5 (range 0–170); 88.6% of contacts were evaluated, disease was found in 0.6% and latent infection was found in 24.1%. Of the 42 areas, 13 met the performance target of evaluating 95% of contacts 66% of contacts started treatment and 64.2% completed. Around half chose to stop, 17% moved and for 17% the reason was unknown. 31% of areas met the target of 85% treatment completion Investigations were estimated to have detected 81% of TB cases which might have been identifiable and estimated to have prevented 34.6% of future TB cases that might have occurred in the following 2 years Costs were estimated at US$4.8M	Main conclusions: performance did not meet national objectives. Further data are required to evaluate performance
Stoddardt and Noah, 1997¹⁰⁸
Type of document: journal article Study design: survey of consultants in communicable disease control and medical officers of environmental health Country: UK Population: n = 732 contacts Quality notes: limited relevant data	Research methods: survey sent to 155 districts. It requested information on the number of new cases of TB found in the preceding 3 years in which > 100 contacts had been screened Staff involved: NR Measures used: TST	Results/data: 46% of districts reported at least one investigation which had screened > 100 contacts. Forty-four cases of TB were found in 18 of the 56 incidents, giving a detection rate of 0.375%	Main conclusions: the authors question the continued use of large-scale screening. Close contacts should be carefully defined, and only they should be screened
Tian et al., 2013¹⁰⁹
Type of document: journal article Study design: developed mathematical model of TB dynamics Country: Canada Population: province of Saskatchewan Quality notes: mathematical modelling	Research methods: comparison of scenarios with and without contact tracing Staff involved: N/A Measures used: risk of infection	Results/data: comparison of a scenario involving contact tracing and no contact tracing indicated a higher average prevalence of TB infection with no contact tracing. The benefit of tracing the first 45% of contacts was greater than tracing the second 45%, indicating a diminishing return Prioritising contacts on the basis of number of times that they have been named had adverse outcomes Increasing the speed of the investigation (90% of contacts are tested within 30 days of diagnosis) did not lead to projected significant improvement in active cases or prevalence of infection Reducing loss to follow-up to a 10% level could lead to significant benefits in infection rates (5.4% prevented; p = 0.02) Targeting investigations by prioritising by age (children aged < 9 years are traced first) and ethnicity (first nation individuals traced first) could improve the effectiveness compared with non-prioritisation (preventing 11% of cases over 20 years; p < 0.0001)	Main conclusions: contact tracing leads to positive outcomes. Reducing loss to follow-up and targeting investigations may increase effectiveness
Trieu et al., 2013¹¹⁰
Type of document: journal article Study design: descriptive Country: USA Population: HIV-infected index case, hotel resident and colleague contacts (n = 31) Quality notes: narrative description of the investigations	Research methods: describes two investigations using IGRAs Staff involved: unclear Measures used: QFT-G IGRA	Results/data: the study indicated that IGRAs can be used in congregate settings The drawbacks are increased cost compared with TSTs (16 times more) and requirement for staff trained in taking blood samples. Specimens also need to be received at a laboratory within 16 hours of collection Positive aspects of IGRA use are only a single encounter with a contact was needed and there were fewer false-positive results due to BCG or other bacteria present meaning fewer people were given prophylaxis treatment	Main conclusions: use of IGRAs in the field is feasible It may be particularly preferred to TSTs in people hard to follow up and contacts who are BCG vaccinated
Underwood et al., 2003¹¹¹
Type of document: journal article Study design: retrospective data analysis to examine screening vs. contact-tracing approaches Country: UK Population: described as living in a socioeconomically deprived area (Tower Hamlets, London) (cases, n = 227; contacts, n = 643; new UK entrants screened, n = 332) Quality notes: compares data from two methods of screening rather than evaluating effectiveness of contact tracing	Research methods: analysis of patient records and a TB database 1997–9. Compares new entry screening vs. contact tracing Staff involved: specialist TB nurses Measures used: number of cases detected, number screened	Results/data: contact tracing of patients with both smear-negative pulmonary TB and non-pulmonary TB identified individuals with active and latent TB with combined prevalence similar in each group (7%) Contact tracing in cases of non-pulmonary disease is not recommended, but the results suggested that it may be at least as productive as the screening of new arrivals to the UK from high-incidence countries	Main conclusions: a contact-tracing strategy is more effective than new entrant screening Contact tracing (even in cases of non-infectious TB) in high incidence areas could be seen as a way of screening extended families or communities at particularly high risk
Verdier et al., 2012⁸
Type of document: journal article Study design: retrospective analysis of data Country: the Netherlands Population: Rotterdam (n = 21,540 contacts) Quality notes: provides limited data	Research methods: analysed data from contact investigations between 2001 and 2006 to identify risk factors for finding TB latent or active infection Staff involved: public health nurse Measures used: TST	Results/data: ‘stone in the pond’ principle followed – close contacts, then regular contacts, then community contacts Residential or family relationships resulted in highest risk of infection, while work or education contact resulted in lower chance of transmission. Risk factors included older age of the patient and older age of the contact. Greater infection risk in community contacts was present if a large number of close contacts were infected	Main conclusions: the risk factors match current guidelines – diagnosis of index patient, closeness of contact relationship, the age of patients and contacts and the number of infections of close contacts
Ward et al., 2004¹¹²
Type of document: journal article Study design: retrospective analysis of data Country: USA Population: schools in New York state (n = 6990 contacts) Quality notes: describes characteristics of the data	Research methods: analysed data from a TB registry for children aged 5–19 years 1997–2001 Staff involved: N/A Measures used: sputum smear, TST	Results/data: each county used their own cut-off points in measuring millimetres of induration on the TST for determining if a contact was positive TST positivity among close and not-close contacts increased with increasing age of index cases (4.6% aged 5–9 years; 5.5% aged 15–19 years). The number of contacts tested increased with increasing children’s age. In very young children infection suggests recent transmission so the emphasis is on finding the index case. In older children transmission between children becomes more likely so there is greater focus on locating contacts In the first round of testing the mean number of close contacts per index case tested was 81 (range 0–725), and was significantly more than not-close contacts (p < 0.0001) Lower SES status of contacts was associated with greater TB risk (assessed by median income). Pupils at smaller schools were also at greater risk of being TST positive In one of the four regions not-close contacts unexpectedly had a higher rate of TST positivity than close contacts across the study as a whole (7.3% vs. 5.1%). In two regions close contacts had a higher rate of TST positivity than not-close (20% and 15.2% vs. 5.1%). The final region data did not differentiate between levels of contact 47.1% completed treatment (higher for this investigation than for the state as a whole)	Main conclusions: many school investigations test more contacts than might be expected due to parental concern and public pressure (no data presented regarding this) Contact investigations may label close/not close after the investigation has been completed; there may be limited documentation regarding labelling It was often not recorded whether TST results for contacts were ‘negative’ or ‘not read’
Wilce et al., 2002¹¹³
Type of document: journal article Study design: qualitative and document analysis Country: USA Population: staff from 11 urban areas Quality notes: provides a range of descriptive data to underpin conclusions	Research methods: interviews with staff and examination of policies and procedures Staff involved: various Measures used: N/A	Results/data: all areas had policies in the form of documents or checklists. Policies varied widely in their content and comprehensiveness. Most described the main steps involved in a contact investigation, but specific actions within these were often not specified Review of medical records was only briefly mentioned in some policies; the time after diagnosis for an interview was often unclear; follow-up interviews were recommended in seven areas. The content of the interview was typically left to the discretion of the interviewer. Field visits were not required in four areas; policies typically provided few details on the procedure for visits. Checklists for assessing risk of transmission were typically available, although without accompanying instructions or space for recording findings Sites did not consistently define a ‘close’ contact. Information regarding risk factors such as HIV could be hard to obtain, although most sites had policies regarding immunocompromised persons Five sites referred to the ‘concentric circles’ approach (but provided little guidance on how to apply it). At six sites decisions regarding expansion were made by the worker responsible, at four sites they were made by a supervisor and at one site they were made during a meeting. Two sites routinely excluded casual contacts owing to limited resources Staffing models and training varied between sites. Data recording and data managements systems varied; also monitoring procedures Challenges reported related to communication barriers, structural barriers and patient-level barriers	Main conclusions: there was generally inconsistency in guidelines, staffing and training across the different services. Comprehensive policies and support are required to improve the standard of contact-tracing investigations
Yeo et al., 2006¹¹⁴
Type of document: journal article Study design: retrospective review of data Country: Canada Population: residents of Montreal under the age of 18 years (n = 66) Quality notes: describes characteristics of the data	Research methods: examined public health data 1996–2000 and carried out additional genotyping Staff involved: N/A Measures used: number of cases and contacts, IS6110-based genotyping, and spoligotyping	Results/data: 19 children were diagnosed after contact investigations of known adult cases No contact investigation had been carried out for eight children. For the remaining 39 children, a total of 616 contacts were identified. The median number of contacts per child was 9 (interquartile range 6–10 contacts) Four probable source cases were identified, all involving parents or other relatives. Genotyping by the research team identified up to 14 possible additional index cases. From the records available it was possible to identify a link to the children for only one of these additional cases	Main conclusions: the contact investigations were extensive and had mostly been able to identify latent TB infection, but less successfully identified the source cases Genotyping indicated a substantial number of further sources of potential transmission
Zangger et al., 2001¹¹⁵
Type of document: journal article Study design: descriptive Country: Switzerland Population: residents of Lausanne including family and school pupils and staff (n = 53) Quality notes: narrative description of the investigation	Research methods: describes investigation around one source case (a 15-year-old girl) Staff involved: doctor, health officer, a nurse from the TB service Measures used: TST, chest radiography for those with positive TST, cost	Results/data: three proximity groups: people living under same roof and close friends classmates and teachers occasional contacts, other schoolmates Passing from one group to next reduced the rate of infection by 4 times. The compliance rate for treatment was 64% The cost of the investigation was over CHF24,000 The index case had arrived from Africa and had a positive test on arrival but no treatment	Main conclusions: the distribution of cases confirms the importance of duration of contact and proximity of contact with the index case

: N/A, not applicable; NR, not reported; SES, socioeconomic status; SMS, short message service.

Copyright © Queen’s Printer and Controller of HMSO 2017. This work was produced by Baxter et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Included under terms of UK Non-commercial Government License.

Bookshelf ID: NBK409253

Contents

< Prev Next >

PubReader
Print View
Cite this Page
Baxter S, Goyder E, Chambers D, et al. Interventions to improve contact tracing for tuberculosis in specific groups and in wider populations: an evidence synthesis. Southampton (UK): NIHR Journals Library; 2017 Jan. (Health Services and Delivery Research, No. 5.1.) Appendix 3, Extraction table for wider population studies.
PDF version of this title (879K)
Disable Glossary Links

Other titles in this collection

Health Services and Delivery Research

Recent Activity

Clear Turn Off Turn On

Extraction table for wider population studies - Interventions to improve contact...
Extraction table for wider population studies - Interventions to improve contact tracing for tuberculosis in specific groups and in wider populations: an evidence synthesis

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

Bookshelf

Interventions to improve contact tracing for tuberculosis in specific groups and in wider populations: an evidence synthesis.

Appendix 3Extraction table for wider population studies

Views

Other titles in this collection

Recent Activity