Results

Material throughout this chapter has been reproduced with permission from Arakelyan et al.28 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial licence (CC-BY-NC) (https://creativecommons.org/licenses/by-nc/4.0/). This allows for the non-commercial reuse of the published paper so long as the published paper is fully attributed. The text below includes minor additions and formatting changes to the original text.

Some material in this chapter has also been reproduced from Jones et al. 202129 © 2021 International Union Against Tuberculosis and Lung Disease. Reprinted with permission.

Material throughout this chapter has also been reproduced with permission from Karat et al.30 This article is available under the Creative Commons CC-BY-NC-ND (https://creativecommons.org/licenses/by-nc-nd/4.0/) licence and permits non-commercial use of the work as published, without adaptation or alteration, provided the work is fully attributed.

Formative research

Scoping review of quantitative literature

The initial search identified 10,801 studies. After removing 869 duplicates, 9932 remained for title and abstract screening, of which 25 met the inclusion criteria (Figure 9). Data on determinants were extracted for 24 studies only, as one reported no ES (see Appendix 2, Table 25).

FIGURE 9 Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram of screening process and included studies. (Figure not reproduced here; see the original article.)

Included studies were published between 1986 and 2019. Most were from the UK and Ireland (n = 7), USA (n = 6), or Spain (n = 5); and a retrospective cohort being the commonest study design (n = 12). Sample sizes ranged from 62 to 73,591 {median 1009 [interquartile range (IQR) 184–2576]}. The mean/median participant age ranged from 28.0 to 52.1 years. The median percentage of males was 64.4% (IQR 56.0–71.0%).

Most studies (n = 22) included all patients starting ATT within the study period. Three studies focused on specific at-risk groups, that is those experiencing homelessness or unstable living arrangements, individuals with MDR TB, or individuals living with HIV and TB. Two studies compared outcomes between groups within a cohort, such as immigrants versus individuals born within a country.

Non-adherence: definitions and assessment

Most study outcomes (n = 15) related to treatment discontinuation (stopping treatment early). Fewer study outcomes (n = 7) appeared to record suboptimal implementation (missed doses during treatment). One study included both a discontinuation and suboptimal implementation outcome. Two studies used a single outcome encapsulating discontinuation and suboptimal implementation.

Discontinuation outcomes were often measured using data from state or national registries/surveillance databases, hospital/lab records or medical notes.

Suboptimal implementation was assessed using various methods, including adherence scale scores, medical records, physician impressions from interviews/assessments, patient self-report, health visitor reports (including pill counts), urine samples (to detect rifampicin), attendance at appointments and prescription requests.

Overall, it appeared that retrospective studies most frequently used surveillance/registry data to determine adherence, whereas prospective studies had much greater variability in adherence measures used (see Appendix 2, Table 25).29 Reported non-adherence ranged from 0.9% to 89% across studies [median 7.0% (IQR 5.2–16.3%)], with two studies not providing a quantified measure.

Demographic determinants of non-adherence

Demographic determinants were assessed in all 24 studies. Specifically, the most studied determinant groups were place of residence and age (Figure 10). The variable with the greatest strength of evidence for a large effect on non-adherence (Categories 1 or 2, i.e. large ESs with either a p-value ≤ 0.05 or a small sample size, see Methods) was place of residence. Within that variable, homelessness and living in an institution or prison (e.g. a ‘confined institution’, a residence hall, or mental hospital) had the strongest evidence, weighted overall, towards non-adherence.

FIGURE 10 Determinants of non-adherence to treatment for TB disease. Note: Bars may include multiple determinant levels assessed within the same study. Darkest grey indicates the strongest effect (i.e. Category 1: a large risk or protective effect at p < 0.05), medium grey indicates a large risk or protective effect at p > 0.05 with a small sample size (Category 2), light grey indicates a small risk or protective effect at p < 0.05 (Category 3) and lightest grey indicates the weakest effect found at p > 0.05 (Category 4). SES, socioeconomic status. Reproduced with permission from Jones et al. 202129 © 2021 International Union Against Tuberculosis and Lung Disease. Reprinted with permission. (Figure not reproduced here; see the original article.)

Age, sex, ethnicity and nationality also showed mixed evidence of effects, as within each variable a similar number of studies found a weak effect with non-adherence as did those with a large effect. Ethnicity and nationality determinants appeared very context-specific, demonstrated by their variation in baseline comparators. Overall, few demographic determinants were classified by strength of evidence in category 2 (i.e. large ES, p > 0.05, but small sample size) or 3 (small ES, p ≤ 0.05). The variables most often found to have a weak effect on adherence (category 4, small ES, p > 0.05) were age, nationality/origin, and ethnicity. No variable had a consistently large effect on non-adherence.

Clinical determinants

Clinical determinants were the second most studied category (23/24 studies). Substance use/misuse was that most frequently assessed; and it also had the most evidence weighted towards a large effect (see Figure 10). Specifically, illicit drug use/addiction had the strongest evidence for this. The evidence for clinical determinants was also mixed, in terms of both strength of evidence and direction. For example, HIV-positive status was found to be a risk factor for non-adherence, yet a diagnosis of acquired immune deficiency syndrome (AIDS) was protective against non-adherence. Again, few determinants fell into strength of evidence categories 2 and 3, and the grouping variables that most often showed a weak effect with adherence were smear and sputum result, substance use/misuse, and HIV infection.

Health systems determinants

Health systems determinants were much less frequently investigated (8/24 studies). Here, route to care was the most studied grouping variable. HCPs’ perception of patient understanding (e.g. lack of awareness of TB severity, understanding of treatment instructions, language barriers) had a consistently large effect on non-adherence, though this determinant was minimally studied. The grouping variables most often found to have a weak effect with adherence were route to care, and those classified as ‘other’.29

Psychosocial determinants

Psychosocial determinants were the least studied (6/46 studies), and the only determinants assessed were mental health issues and having close relationships with others. Of these, having a mental health problem was found to be both a risk for and protective against non-adherence. Strength of evidence for mental health problems was also mixed, with as many studies finding strong and weak effects on adherence.

From 10,801 results, we identified 24 relevant studies from 10 countries. Definitions and methods of assessing adherence were highly variable, as were documented levels of non-adherence (0.9–89%). Demographic factors were assessed in all studies and clinical factors were frequently assessed (23/24).

Determinants commonly associated with non-adherence were homelessness, incarceration, and alcohol or drug misuse. Health system (8/24) and psychosocial factors (6/24) were less commonly evaluated.

Our review identified some key factors associated with non-adherence to anti-TB treatment in high income, low TB incidence settings. However, some potentially modifiable determinants, such as psychosocial factors, were under-evidenced.29

Scoping review of qualitative literature

A second part of the formative work for the development of the intervention was a scoping review of the qualitative literature on the factors influencing experience of anti-TB treatment. Initial database searches retrieved 1013 studies. After duplicates were removed, 566 titles and abstracts were screened, and 36 full-text articles retrieved. Eighteen studies met the inclusion criteria. An additional 10 studies were identified from searching references and citations, resulting in a total of 28 included studies (see Appendix 2, Figure 26).

Bibliometric findings

Over two-thirds of studies were published after 2010. The majority (n = 22; 79%) were conducted in Europe or North America; and most (n = 22; 79%) used qualitative methods (see Appendix 2, Table 26). Six studies used a mixed-methods approach; the rest exclusively used qualitative methods. Studies reported on the perspectives of individuals with TB (n = 15), community members (n = 3), healthcare providers (n = 2), or a combination of different population groups (n = 8).

Framing the concept of adherence

Despite ‘adherence’ being one of our search terms, only three studies offered explicit definitions of adherence, reflecting the broader approach to treatment experience taken by qualitative researchers. Non-adherence to ATT was defined in one study as taking < 80% of the prescribed regimen and loss-to follow-up was described in another as treatment interruption for 60 consecutive days or more, or for two consecutive months, or insufficient treatment duration. One paper used the Morisky Medication Adherence Scale (not specific to TB) to assess levels of adherence.

The framing of non-adherence within the overall experience of taking ATT was closely related to the theoretical framing of that paper. Four approaches were identified: a pragmatic health services approach (n = 11), critical perspectives (n = 7), applied medical anthropology (n = 7), and social theory (n = 3; Appendix 2, Table 27).

Synthesis of results

Treatment-related factors

Half the studies [n = 14 (50%)] reported on treatment-related factors influencing experiences of ATT. These included individuals’ responses to the number of tablets, complexity of regimens, duration of treatment and its side effects, and potential interactions between TB and other medication. Several studies discussed difficulties faced by patients in adhering to lengthy treatment regimens that often required strictly routinised lifestyles.

Both the pill burden and the drug regimen affected patients’ willingness to continue with their treatment. And multiple side effects and adverse reactions were reported, resulting in patients often questioning the physical safety of the treatment and their perceived health costs. This was further compounded by the sacrifices patients felt they had to make, including prolonged abstinence from alcohol.

Personal factors

Over two-thirds of studies [n = 23 (82%)] reported that the personal characteristics of individuals shaped their experiences of ATT and adherence behaviour, as well as vice-versa, i.e. how being on ATT affected daily lives. Struggles to preserve autonomy, regain control of life, and a compromised sense of self-efficacy were linked to behaviour in a few studies. For some patients, subscribing to a supervised treatment regimen was experienced as disempowering.

I asked about [self-administering] and the nurse said, ‘No, you are not allowed to’. So, I was annoyed. ‘I am not a child’, I said to her. ‘I am an adult. I take the medication for my own sake and not for you.

Misconceptions, incorrect or lay knowledge, and ethno-medical beliefs regarding TB aetiology, transmission, and treatment outcomes were reported to negatively impact ATT-related behaviour through diagnostic delays, loss-to-follow-up after initial diagnostic tests, and patients not returning for outpatient follow-up visits.

Patients who lacked knowledge had more negative attitudes to TB and worse experiences with ATT than better-informed patients. Lack of understanding about the nature of TB disease often resulted in non-acceptance, or low levels of perceived risk, and, therefore, delays in seeking help. Acceptance of the disease and its consequences, which was often achieved via support from nurses, was reported to facilitate adherence to ATT.

Disease chronicity and a possibility of recurrence, comorbidities (HIV, diabetes), and associated general physical weakness influenced the illness experience of being on ATT, which, in turn, was reported in some studies to negatively influence adherence behaviour. Compromised health and comorbidities worsened the experience of being a TB patient, especially when care was uneven or fragmented.

Drug and alcohol use influenced ATT-related behaviour and adherence. For people with TB who used drugs, drug use (e.g. crack cocaine, heroin) was often prioritised over treatment; one study noted multiple episodes of self-discharge from hospital to source drugs or secure access to methadone. Anxiety and worries about the consequences of having TB as well as being on treatment, self-stigmatisation, and resulting social isolation and loneliness were commonly reported among individuals on treatment, and especially acute in marginalised groups. Self-stigmatisation resulted in patients restricting their social circles to limit exposure to discrimination or social marginalisation.

Social factors

Sixteen of 28 (57%) papers reported on social factors influencing individuals’ experience of TB care, and specifically their ability to access and stay on treatment.

Social roles and daily lives including interactions with family, friends, peers, and other social networks affected (and in turn were affected by) the experience of having TB and being on ATT. Social support – or the lack of it – was an important determinant of knowledge about TB, the experience of having TB, and the ability to access and stay on treatment. Lack of social support was particularly acute in marginalised groups; however, support could fluctuate over the course of treatment. Several studies highlighted the negative impact of TB-related stigma on individual treatment-seeking behaviour; though the experience of stigma, like social support, could differ depending on location and stage of treatment.

Health systems-related factors

Most papers [n = 24 (86%)] reported on health systems factors that impacted on patients’ lived experiences of TB care, including their ability to access and stay on treatment. Here, we refer to patients’ access to and utilisation of health services, the organisation of care and treatment regimens, perceived quality of care, and relationships with HCPs, including communication and support.

A few studies suggested that knowledge and health-seeking related to TB may be worse among migrant groups and linked these to limited availability and accessibility of information, particularly in languages other than English.

A number of studies documented delays in diagnosis or early misdiagnosis by front-line health services, with subsequent delays in treatment and frustration and anger for patients.2 In many settings, the inflexibility of the health system and care delivery (DOT in particular) served to alienate patients. The quality of relationships and communication with healthcare staff impacted experiences of ATT, especially among marginalised groups. HCPs’ and patients’ conflicting beliefs and priorities were reported to have long-term effects on individuals’ views of the health system. Conversely, the provision of comprehensive support tailored to patient’s characteristics, personalised care and holistic counselling were linked to improved adherence.

You get someone sitting next to you and trying to get your mind off things, just to talk to you, ask you about other things apart from the disease (…) At the end of the day your mind, you kind of jump out of the situation you’re in, and it helps you to take the medication, and to look at life with a different perspective as you’re taking the medication.

Two studies described how HCPs not only provided clinical guidance, adherence oversight and ‘expert companionship’, but also helped TB patients with food, transport and co-ordinated referrals to other health and welfare services.

Structural factors

Seventeen (61%) papers reported on structural factors affecting individuals’ access to and experience of ATT. We defined structural factors as those social and political mechanisms that generate and reinforce social class divisions, placing individuals within hierarchies of power, prestige, and access to resources. Three papers reported on the broader policy and financial environment as negatively affecting availability of support services for people taking ATT, particularly for socioeconomically deprived individuals. Six papers, focused on the experiences of migrants, cited concerns around legal status in the host country as having negative effects on individuals’ health and treatment-seeking behaviour. Migrants faced language barriers and barriers to employment that in turn affected health literacy and health-seeking behaviour. Homelessness, loss of employment, and associated financial difficulties affected TB patients’ ability to access treatment and adhere to ATT. TB patients who lost their work due to TB experienced subsequent financial difficulties but lacked confidence to look for employment opportunities because of fears of being stigmatised or marginalised by potential employers or colleagues.

Conversely, for some HCPs, working with individuals who had little power to negotiate the terms of care sometimes facilitated patients’ adherence. A few papers described the impact of past experiences of segregation, violence, torture, or physical or sexual abuse on current health-seeking behaviour, including treatment-related behaviours.

Twenty-eight studies were identified: the majority (86%) reported on health systems factors, followed by personal characteristics (82%), structural influences (61%), social factors (57%) and treatment-related factors (50%).

Included studies highlighted three points that underpin a relational approach to ATT behaviour:

1. individual motivation and capacity to take ATT is dynamic and intertwined with, rather than separate from, social, health systems and structural factors;
2. individuals’ pre-existing experiences of health-seeking influenced their views on treatment and their ability to commit to long-term regular medicine-taking; and
3. social, cultural and political contexts play an important role in mediating how specific factors work to support or hinder ATT adherence behaviour in different settings.

Based on our analysis, person-centred clinical management of TB should:

1. acknowledge the ways in which ATT both disrupts and is managed within the everyday lives of individuals with TB
2. appreciate that individual circumstances plus accessible support and resources may change over the course of treatment; and
3. display sensitivity towards context-specific social and cultural norms affecting individual and collective experiences of being on ATT.

In-depth interviews

The third element of the formative research was interviews with current/ex-patients, the individuals who were caring or who cared for them, and HCPs.

Healthcare professionals: key findings and mechanisms to support adherence

Twenty-seven interviews were conducted with HCPs across the 4 trusts (12 nurses; 9 doctors; 3 outreach workers, 1 advocate, and 2 administrators). To understand the support that was available to people taking ATT, there follows a short section on risk assessment and the subsequent enhanced case management that may be provided to patients.

Risk assessment

In most NHS TB services, a new TB patient requiring treatment for TB disease is seen by a specialist TB nurse soon after diagnosis. A risk assessment process is mandated in national guidelines and is built into the standard paperwork completed for each new patient.8 This determines the level of support that the patient may require to successfully complete treatment. It involves the nurse assessing the patient against a list of risk factors for non-adherence, which include a history of ‘problem’ drug use, alcohol misuse, homelessness, incarceration and comorbidities. The decision made at the end of this assessment is whether the patient needs ‘Enhanced Case Management’ (ECM).

In practice, there is scope for interpretation by the administering HCP (usually a nurse) and, often, enough flexibility for the HCP to take action using their experience and clinical intuition. Indeed, HCPs interviewed reported that they made their judgements in response to the way a patient engaged with them, the patient’s current living or social situation, their experiences and observations of the patient’s home and surroundings, and other factors.

So, we have a process that we go through and we try and assess the level of support that an individual requires getting through to treatment completion. Just a box ticking exercise, really, and this is where the skill of being a TB nurse lies. So, trying to establish how likely a person is to comply with their treatment. And so there’s an art to that. (…) There’s a whole group of reasons why they might need enhanced case management. (…) [Some of these things] are not on the risk assessment form, but they fit our criteria here, and so the TB nurses know that. Or they might not fit any criteria, but the TB nurse might just think: Hmm, not sure they’re going to comply.

Enhanced case management

For most patients requiring ECM, the degree of support offered is often based on the judgement of the case-managing nurse, who may use a combination of formal and informal measures to get that individual through to the end of treatment, plus the local resources available to support the patient, for example, social care teams, access to housing support.

‘Formal’ measures include ‘additional contact’, which may be as simple as a few extra phone calls or clinic visits; more labour- and time-intensive practices such as DOT; and more drastic (and expensive) measures such as voluntary or mandated admission to hospital.

‘Informal’ measures, such as building trust and relationships with the patient and their family, the use of small monetary incentives for things like transport and food, and liaising with housing or benefits agencies, may be applied in isolation or in combination, depending on the nurse’s judgement and the needs and situation of the patient.

Assessing adherence

Nurses, doctors and outreach workers use a range of methods to gain quantitative and qualitative (or ‘proxy’) estimates of a patient’s adherence to treatment for TB. More quantitative estimates include conducting pill counts; urine tests, to examine the colour or test for drug levels; and blood tests, to check for drug levels. To gain more qualitative estimates of adherence, team members may conduct medication checks (gauging an individual’s familiarity with their tablets and medication routine); ask indirect questions around the topic of adherence, without necessarily addressing it directly; make observations of home circumstances, including where medication is stored; monitor clinical progress; take note of periods of reduced communication or engagement, earmarking individuals for increased attention; and log any self-reported missed treatment doses. These methods are employed more regularly and less formally than those used to obtain quantitative estimates.

Any intervention should be adaptable to the individual patient and their circumstances.

The participation of HCPs in the development of any intervention and a sense of ‘ownership’ is likely to improve its acceptability and sustainability.

An intervention should make a distinction between components that address individual patient capacity to adhere to treatment (e.g. knowledge, skills, support) and provider capacity to support the patient on treatment (e.g. communication skills, time, resources).

A treatment timeline or journey is relevant when considering the timing of specific intervention components and by whom they are delivered.

It is important to consider potential unintended (negative) effects of an intervention on other aspects of patients’ lives as well as time/resources HCP’s have available (and therefore on the care received/experiences of other patients).

Patients and caregivers

Interviews were conducted with 18 patients and 4 caregivers.

Major themes

Themes are grouped loosely in accordance with the framework established through the scoping review of the qualitative adherence literature, under individual- social-, structural-, health systems-, and treatment-related factors.

Individual factors

Most participants had little previous knowledge of TB. A common perception was that it was something that affected other, ‘poorer’ countries, or was a disease of the past. Patients were also often very worried about the implications of their own infectivity, and ruminated on how they may have acquired TB in the first place, occasionally alluding to individuals or groups of people they used to associate with. Patients talked, often in florid terms, of the way they imagined TB affecting their bodies, or of their impressions of their own chest radiographs or other imaging they had been shown.

Often it was a visual stimulus, either for them, or for someone around them, that prompted them to seek care in the first place, and visual cues were often used to signify clinical improvement after a period of treatment (e.g. a changing chest radiograph, reduction in size of a lymph node, or weight gain).

Patients with other significant health concerns described complex lives, involving attendance at multiple consultations, managing different sets of drugs, and interpreting sometimes conflicting health messaging from different HCP teams. A lack of clear communication between teams and with patients seemed to have a large impact on patients’ and their families’ ability to cope, and ‘faith in the system’, once lost, seemed difficult to regain. Several patients spoke about the effect of TB and TB treatment on their mental health, often describing the gradual deterioration in their mental state over the course of treatment. This progression, and the effects it had on individual’s ability to continue with their treatment regimens, was striking and was linked to persistent feelings of fragility and vulnerability that lasted long after an individual had been discharged from care.

Individuals who consumed large amounts of drugs or alcohol described lives that were often chaotic, and talked about how their habit took precedence over most other responsibilities or activities. This meant that there were few regular activities around which a medication routine could be built.

Social factors

Participants spoke in depth about how having TB and taking treatment for it had affected their relationships with friends and family, as well as the impacts of these relationships on their ability to keep taking their medication. For participants with close families or partners, these people were often significant sources of support and directly aided their adherence, through helping remind them to take their tablets, accompanying them to clinic visits, and liaising with their healthcare providers. There were also many examples of family providing other forms of support, including financial assistance, taking responsibility for the patient’s nutrition, and helping deal with the emotional and mental health consequences of illness.

Patients also talked about how their illness and – often – side effects of the treatment had put a strain on their existing relationships, creating a stigma-laden barrier between them and their loved ones.

My folks back home didn’t even know that I was diagnosed, ‘cos I didn’t want to upset them… Even when I told them, my mom was like, she was quite upset about it… Both my parents are pretty old, and I didn’t want to, kind of, stress them out.

Other connections, such as with employers, were also tested. Caregivers spoke of the extra burdens they had been forced to shoulder, and the general lack of acknowledgment or support for this from the health service or others in the family. Younger and more educated patients who appeared to be more ‘treatment literate’, reported struggling with the absence of a close relationship, and talked about how coping on their own had taken a significant toll on their energy and mental health.

Structural factors

Concerns about the effects of TB on people’s ability to work and communicating this to employers was regularly discussed by participants, with many talking about the financial implications of being told not to work or longer-term consequences to career and reputation of having TB or taking time off work. Some individuals had been in and out of the social services system and correctional services, and depended on government funding to remain in housing. Caregivers, too, talked about the extra burden placed on them, having to manage finances or take over responsibilities as breadwinners.

It was coming to the end of it that really got me down… You’ve gotta adjust your whole life around taking these tablets… Like, say, if you want to work – you couldn’t go to work… You just put your life on hold… The longest six months of my life so far.

Health systems factors

For many patients, this was their first serious interaction with any health system, and many expressed frustration with how long it had taken for them to get into the TB system, often requiring several visits to the GP and/or the progression of symptoms to the point that emergency assessment was required. A good working relationship with HCPs, especially TB nurses, was important to many patients, and several individuals praised the extra attention they had received, or were grateful for the easy access they had to their nurse. Though some patients did talk about confusion around the instructions they had received, most recalled that they had been informed about how to take the medication and potential side effects.

Treatment-related factors

Patients talked at length about the physical difficulties of taking treatment, from struggling with the numbers of pills, to issues with the size or shape of individual tablets. Side effects experienced were also a frequent topic of discussion: some participants recalled severe, life-threatening reactions that had required them to be hospitalised; others described the daily effects that treatment had on their state of mind. Many patients spoke about the routines they had developed to be able to take their treatment, including changes that had to be made to existing and established schedules to minimise the impact of side effects.

Relating elicited themes to adherence behaviour

Analysis of the interview data suggests that personal, social, structural, and health systems factors influencing an individual’s treatment journey, engagement with health services, and capacity to stay on treatment are closely intertwined. After mapping the relationships across subthemes to illustrate direct (‘proximal’) and indirect or mediating (‘distal’) determinants of adherence, we identified 16 factors that were likely to have a direct effect on adherence behaviour (see Appendix 2, Figure 20). Factors likely to reduce motivation and ability to take TB medication were a ‘chaotic’ lifestyle (e.g. because of homelessness or substance misuse); social isolation; ‘fragmented’ care; perception of the risk as being low; absence or loss of faith in the health system; poor understanding of the need to take medication; treatment side effects; and pre-existing or new mental health issues. Factors likely to directly increase motivation and ability to take TB treatment were a well-established medication routine; at least one close or caring relationship; a role as a breadwinner or carer; a good understanding of the need to take medication; a supportive employer and/or regular employment; a sense of progression towards the end of treatment; a good relationship with and/or trust in the TB nurse or CM; and perception of the risk as high.

Clinic observations

We observed that HCPs delivered (verbally) a large amount of information to patients, though the structure was not always the same between HCPs or sites. Although the standard risk assessment was used consistently, more experienced HCPs often checked for factors that were not listed, effectively delivering a more comprehensive assessment. These included patient attitudes and motivation (perceived or stated) towards treatment, their existing social support, their understanding of TB, and perceived need for treatment. Adherence support was assigned based on the CMs’ judgement of the patient’s social and clinical needs. Finally, we observed that CMs sometimes made additional phone calls to patients, often simply to check in.

Personal, social, structural and health systems factors influencing an individual’s treatment journey, engagement with health services, and capacity to stay on treatment are closely intertwined.

Patients with more ‘chaotic’ lives (e.g. because of homelessness or substance misuse), had few regular activities around which to build a medication routine, making adherence less likely.

Worsening mental health during treatment was also common, usually with negative consequences for adherence and the individual’s general well-being.

The relationship between patient and CM was likely critical to treatment success.

Based on these findings, we recommended to the IDG that the intervention should:

1. Respond to the diversity of patients’ backgrounds and needs
2. Understand the current system’s constraints and their impact on patients’ experiences of illness, treatment and care
3. Enhance current HCP practices in supporting TB patients; and
4. Take patients’ views into account.

Intervention development

Implications of the findings of formative work for intervention development

The formative work suggested that a successful intervention would need to account for the diversity of patient backgrounds and needs, considering the implications of the diagnosis and treatment in the context of the physical, mental, social and structural circumstances of the individual. The intervention should also be embedded in an understanding of the constraints of the current system and their impact on patients’ experiences of TB treatment and care, including the effects of a delay in diagnosis; comorbidities and resulting fragmentation of care; and uncertainty around diagnosis or prognosis. The ideal intervention would also try to standardise the best elements of current practice, and incorporate suggestions made by patients and caregivers, such as the need for more and better information, particularly around side effects; more awareness of the impacts of TB on mental health; better psychological support; better management of expectations; and greater use of visual representations of progression towards treatment completion.

Details and results of the intervention development process (theoretical framework, intervention components, theory of change and logic model) are provided in the Methods.

Pilot cluster-randomised controlled trial and feasibility study

Between October 2020 and September 2021 adults with TB disease were recruited from four London clinics to a pilot cluster RCT of the intervention. Feasibility, acceptability and effectiveness were examined, and data were collected to inform a possible future larger study.

Patient population

One hundred and fifty-five patients had TB disease across the four sites within the recruitment period. Of these, 130 were eligible, contactable and approached for the study; 82/130 (64.1%) consented (Table 3).

Patients who were ineligible were one or more of the following: aged < 18 years, unable to provide informed consent, already on treatment for TB at the time of potential recruitment, not expected to live for the duration of the study (24 weeks from starting treatment).

Patients who were eligible but could not be contacted included those with TB disease who were in shared care with a non-trial site TB service and/or were started on treatment remotely during COVID lockdown or were started on treatment outside of usual care appointment at a study site.

Two patients were later found to not have TB (late exclusions). One additional patient was removed from the study as they requested for their data to be entirely deleted at withdrawal. This left a final total of 79 study participants, with 36 in the intervention arm (24 at Intervention site 1, 12 at Intervention site 2) and 43 in the standard care arm (18 at Control site 1, 25 at Control site 2; Appendix 2, Table 28).

Figure 11 shows the number of eligible patients initially screened for the study, the numbers consenting, and withdrawing. From month 6, reductions in the number of consenting participants are because few people required regimens of more than 6 months.

FIGURE 11 Flow chart depicting the patient flow into and through the pilot study. (Figure not reproduced here; see the original article.)

The characteristics of non-consenters were different from the 79 people who consented based on gender and ethnicity, with more males and Asian/Asian British adults declining to take part in the study (Table 4). Among the 14 remaining withdrawals (who were retained in the analysis up to the time point of their withdrawal; Figure 11), 7 (50.0%) did so for personal reasons, 4 (28.6%) for medical reasons, 2 (14.3%) due to study error and 1 (7.1%) where the reason was not documented. Seven withdrawals were from the intervention arm and seven from the control arm. Within personal reasons for withdrawal, some participants found the number of questionnaires requiring completion problematic and others that the electronic adherence measuring device was too large for them to use comfortably on a day-to-day basis. Withdrawal was highest between months 1 and 2. Demographically, the individuals who did and did not withdraw were similar.

The time of withdrawals at each study site is shown in Table 5.

Among the 79 patients in the analyses, there was an even split male and female (49.4% each), with 1 person registering as ‘other’ (see Table 4). The age group with the most people was 28–37 years (26/79, 32.9%); 39/49 (49.4%) of the population were Asian or Asian British; 43 (54.4%) were recruited from control sites and 36 (45.6%) from intervention sites; and 58/79 (73.4%) had extrapulmonary TB or extrapulmonary involvement. The study arms appeared balanced in their demographic attributes. Appendix 2, Table 29 gives the sites of TB disease for each study site and Appendix 2, Table 30 the breakdown of extrapulmonary disease at each study site.

Adherence data analysis

The percentage of doses over the first 6 months of treatment for which adherence data were available (i.e. while individuals were still under follow-up) was 97.1%. Our analysis sought to determine the impact of the intervention on the proportion of doses taken within the trial population. This was assessed at 24 weeks (168 doses) from the start of treatment. In unadjusted analyses using the primary outcome variable, the proportion of doses taken at 24 weeks was 0.84 (95% CI 0.78 to 0.91) across both arms; 0.81 (0.68 to 0.93) for the intervention arm and 0.88 (0.67 to 1.00) for the control arm (standard care). The difference in proportions was − 0.07 [intervention relative to control; 95% confidence interval (Cl) − 0.15 to + 0.01; p = 0.07; Table 6]. This equates to 11.8 fewer doses being taken in the intervention than the control arm, out of a total 168 possible doses. Adjusted estimates (which were performed despite the distribution of baseline characteristics being similar between the intervention and control arms) were similar. Broadly, results from the sensitivity analyses were similar to those from the main analysis, except when intermittent box users were excluded. Here, adherence increased in both the trial arms and was more equal between them [unadjusted mean difference 0.01 (95% CI − 0.11 to + 0.14; p = 0.69].

Lasagna plots of adherence over the first 168 days of treatment demonstrated the individual patterns of non-adherence among participants (Figure 12). In both study arms, dose-taking was intermittent, with increasing missed doses as treatment progressed (see Figure 12 parts a and b). Doses were generally missed intermittently (suboptimal implementation) rather than patients ceasing treatment early (early discontinuation; see Glossary). A visual inspection of the plots indicated no obvious differences in adherence patterns between the two trial arms, barring the lower proportion of doses taken in the intervention versus the control arm (note from Table 6 that this difference was not found to be statistically significant).

FIGURE 12 Lasagna plots depicting adherence patterns of patients. Note: Each row depicts the adherence data for one patient over the first 168 days of treatment. Each plot is read from left to right, with the first dose on the left and dose number 168 on the right. Plots on the left are for the intervention arm; plots on the right show the standard care comparator arm. Patients ordered from most to least adherent. Dark blue denotes doses which were recorded as taken. Light blue denotes doses which were recorded as not taken. Three sensitivity analyses are presented. The first was an alternative coding of the adherence data (c, d). The second excluded intermittent evriMED box users (e, f). The third excluded individuals known to have been in hospital but where their dates of admission were unknown (g, h). (Figure not reproduced here; see the original article.)

Using a different form of plot, we examined the relative contribution of early discontinuation and suboptimal implementation in the two arms of the trial. Both types of non-adherence accumulated over time, particularly suboptimal implementation up to 2 months and discontinuation thereafter (Figure 13).

FIGURE 13 Relative contribution of discontinuation and suboptimal implementation to non-adherence over time. Note: Non-adherence due to discontinuation (ceasing treatment and not recommencing; dark grey) vs. suboptimal implementation (sporadic missed doses; light grey) over time. Panel (a) intervention arm, (b) control arm. Discontinuation is not the same as programmatically defined loss to follow-up/default. Figure excludes withdrawals from the study. (Figure not reproduced here; see the original article.)

Analyses of adherence and social support offered and accepted

Another measure of the functionality of the intervention was whether the patients identified as needing adherence and social support were offered (and accepted) such measures.

First, we compared the identification of such individuals using the ‘classical risk factor’ variable versus the offer of adherence support (Table 7). In the intervention arm, 6/6 (100.0%) of individuals with a classical risk factor were identified as requiring adherence support within the trial, as well as an additional 15 people without these factors. This amounted to 21/36 (58.3%) of people in this arm. The control arm only identified 1/5 (20.0%) of patients who had a classical risk factor for non-adherence, as well as three additional people without these factors. Overall, the control arm identified and offered 4/43 (9.3%) patients adherence support at baseline.

Next, we examined the offer and acceptance of adherence support measures, including which measures were accepted. Adherence support measures offered and accepted were highest at baseline in both the intervention and control arms, though were recorded as continuing to occur throughout treatment for 10–15% of both populations (Figure 14). Acceptance was substantially lower than offer, with the proportion of offers accepted roughly similar in both arms.

FIGURE 14 Adherence support offered and accepted at different time points by arm. Note: Offer and acceptance of adherence support across the different time points in the trial. Entire bar represents total number of patients who were offered adherence support. Pale colours (lower portion of the bar) show patients who did not accept and dark colours (upper portion of the bar) indicate those who did. Yellow = intervention, green = control. (Figure not reproduced here; see the original article.)

All forms of adherence support were offered, though the evriMED box reminders were not accepted by any patients (see Appendix 2, Figure 27). Dosette was the most popular form of adherence support used, both at baseline and later in treatment.

Offer and acceptance of an additional suite of social support measures was examined (Figure 15). Offers were typically early during treatment and more in the intervention than control arm. No offers were accepted in the control arm.

FIGURE 15 Social support measures offered and accepted at different time points, split by arm. Note: Entire bar represents total number of patients who were offered social support. Pale colours (lower portion of the bar) show patients who did not accept and dark colours (upper portion of the bar) indicate those who did. Yellow = intervention, green = control. (Figure not reproduced here; see the original article.)

The data on which measures were accepted were available for the intervention sites only (where they were routinely recorded as part of the TNA). Six of the eight social support measures were accepted at site 1 and none at site 2 (Figure 16). Acceptance was generally greater earlier in the study, barring Transport for London vouchers, which were more continuously accepted. Referrals to the social care team for financial or housing support were the support measures most accepted.

FIGURE 16 Social support measures accepted at different time points at Intervention sites. Note: Social adherence support measures across the different time points in the trial. Intervention arm only. Site 1 only; other site had no social support measures accepted. (Figure not reproduced here; see the original article.)

Analyses of treatment outcomes

At 12 months after diagnosis, among patients who had not withdrawn from the study, 60/65 (92.3%) of patients had completed treatment (see Appendix 2, Table 31). Two were still on treatment (3.1%) and two (3.1%) had yet to reach 12 months after their diagnosis before the study came to an end, and one (1.5%) was lost to follow-up. The distribution of outcomes was similar between arms.

Consent was obtained from 82/130 (64.1%) individuals approached. After excluding withdrawals, data from 79 participants were used for analysis. The study arms appeared balanced in demographics. At baseline, only 11/79 (13.9%) participants had one or more ‘classic’ risk factors for non-adherence recorded, six at intervention and five at control sites. Most participants [74/79 (93.7%)] were born in the UK.

Adherence support was offered at baseline to 21/36 (58.3%) intervention participants and 4/43 (9.3%) control participants. Dosette boxes were the most popular overall (17 acceptances), whereas the evriMED box reminders were not accepted by anyone. Blisterpacks and VOT were accepted by one participant each (both intervention) and DOT by two participants (one per arm).

Additional social support was offered at intervention sites. Transport vouchers (n = 9) and referrals to the social care team for financial (n = 11) or housing (n = 9) support were the most commonly accepted measures.

Adherence data were available for 97.1% of doses in the first 6 months of treatment. In unadjusted analyses, 81% of doses were taken at 24 weeks in the intervention arm and 88% in the control arm (difference in proportions −0.07 [95% Cl − 0.15 to + 0.01]; p = 0.07). This equates to 11.8 fewer doses being taken in the intervention than the control arm, out of a total 168 possible doses. Adjusted estimates were broadly similar. Lasagna plots showed no obvious differences in non-adherence patterns between arms.

At 12 months after diagnosis, 60/65 (92.3%) of participants who had not withdrawn had completed treatment (93.1% intervention, 91.7% control). Two (3.1%) individuals were still on treatment (one each arm), two (3.1%) had not yet reached 12 months (both control arm), and one (1.5%) was lost to follow-up (intervention arm).

Adherence data were available for 97.1% of doses in the first 6 months of treatment. In unadjusted analyses, 81% of doses were taken at 24 weeks in the intervention arm and 88% in the control arm. The difference in proportions was −0.07, 95% Cl − 0.15 to + 0.01, p-value 0.07. This equates to 11.8 fewer doses being taken in the intervention than the control arm, out of a total 168 possible doses.

Psychological assessments

Study participants were asked to complete several questionnaires during their treatment for TB. These explored their beliefs about illness and TB, its treatment and the practicalities of taking treatment.

Beliefs about tuberculosis

The BIPQ total score measures an overall level of perceived threat from TB (Table 8). This decreased over time in both intervention and control, with no statistically significant differences across arms (p > 0.05).

Beliefs about tuberculosis medicines

The intervention had a significant, albeit temporary, effect on patient’s perceived need for TB treatment (Necessity beliefs). Between 2 weeks and 3 months, Necessity beliefs increased in the intervention arm and decreased in the control arm [adjusted mean difference (aMD): 0.49, t = 13.01, p = 0.05, 95% CI 0.11 to 0.97] (see Table 8). However, this difference was not maintained after this.

Concerns about treatment for TB decreased over time in both arms, with no significant differences across arms (see Table 8).

Practical barriers to taking treatment for TB

Perceived practical barriers to taking anti-TB treatment increased over time in both groups. No significant differences between the trial arms were identified throughout treatment (p > 0.05; Table 8).

Adverse effects

Lastly, patients’ side effect experiences attributed to their TB medicines were compared between the trial arms. Again, there were no statistically significant differences in change over time at any time point (p > 0.05; Table 8).

Participants’ perceived need for TB treatment increased in the intervention arm and decreased in the control arm between 2 weeks and 3 months after enrolment [adjusted mean difference: 0.49, t = 13.01, (95% CI 0.11 to 0.97); p = 0.05] but between 3 and 6 months there was no significant difference between arms.

Concerns about treatment for TB decreased over time in both arms, with no difference detected between arms (p > 0.05 for all time periods).

Similarly, no differences between arms were detected for changes in perceived practical barriers to taking ATT (increased over time in both groups; p > 0.05 for all time periods) or in side effects experienced (reduced over time in both groups; p > 0.05 for all time periods).

Participants’ overall ‘level of perceived threat’ (measured by the BIPQ) reduced over the course of the study, with no difference detected between arms (p > 0.05 for all time periods).

Health economics

We calculated the costs of delivering the TNA and the standard needs assessment in the English NHS setting. We also determined the feasibility of calculating utility scores and costs by investigating the completeness of quality of life and resource use data collected within the study, with the aim of informing the design of a future full economic evaluation alongside a full clinical trial.

Data completeness

Overall, participants’ completion of EQ-5D-5L and modified CSRI was good. Across time points, response rates were 71–99% for EQ-5D-5L and 71–97% for modified CSRI. Table 9 presents the response rates for EQ-5D-5L and modified CSRI at baseline, 2, 4 and 6 months. Time spent delivering needs assessment was recorded for 25–86% of patients across time points. We were able to calculate costs for TB-related medications in 78–99% of patients across time points. The breakdown of data completeness for variables collected from EQ-5D-5L, modified CSRI, medication records are given in Appendix 2 (Tables 32–34, respectively).

Study intervention

The study process evaluation recorded the length of TB nurse sessions in both the intervention and control arms. Table 10 shows the proportion of total planned clinic appointments for which the data were available and could be used for further cost analysis. In both arms, after the week 2 appointment this fell below 50%.

Table 11 presents the mean duration of timed sessions with a TB nurse at each study time point by randomisation arm. Mean session length was longest at baseline and shortest at 6 months in both arms. In the intervention arm (middle data column), mean (SD) session length was 65 (20) minutes at baseline and 19 (7) minutes at 6 months. In the control arm, mean (SD) session length was 47 (25) minutes at baseline and 10 (5) minutes at 6 months. We further calculated the mean session length for patients in the intervention arm who were recorded as having received the TNA during their sessions with a TB nurse (93% of the expected total) (left hand data column). Mean (SD) session length was 65 (20) minutes at baseline and 15 (5) minutes at 6 months.

We calculated the mean costs for sessions with a TB nurse in each arm and TNA recipients. In line with the mean session length, mean cost appeared to be highest at baseline and lowest at 6 months, although we did not formally test for differences and the SD were large (Table 12). At baseline, mean (SD) session cost was £55 (£17) in the intervention arm and £40 (£21) in the control arm. At 6 months, mean cost was £16 (£6) in the intervention arm and £9 (£4) in the control arm. The mean intervention cost for TNA recipients was £55 (£17) at baseline and £13 (£4) at 6 months.

Quality of life

Table 13 presents the mean EQ-5D-5L utility score for the total patient group and by randomisation arm [where a higher score indicates better health-related quality of life (HRQoL)]. Within the overall patient group, mean (SD) EQ-5D-5L utility score was 0.75 (0.27) at baseline and 0.85 (0.22) at 6 months. For the intervention arm, mean (SD) EQ-5D-5L utility score was 0.77 (0.28) at baseline and 0.90 (0.11) at 6 months. For the control arm, mean (SD) EQ-5D-5L utility score was 0.73 (0.27) at baseline and 0.81 (0.27) at 6 months.

Table 14 shows the mean (SD) EQ-5D-5L score for each of the five health states across time points. Note that here, a higher score indicates a greater perceived problem by the respondent. At both baseline and 6 months, the dimension of pain/discomfort appeared to have the highest score and self-care appeared to have the lowest score. At baseline, mean (SD) EQ-5D-5L score was 2.18 (1.11) for pain/discomfort and 1.40 (0.84) for self-care. At 6 months, mean (SD) EQ-5D-5L score was 1.70 (0.87) for pain/discomfort and 1.26 (0.81) for self-care.

Table 15 presents the mean and SD, and median and IQR, EQ-VAS scores for the total group and by randomisation arm (with higher scores indicating perceived better health). Mean (SD) EQ-VAS was 66.1 (20.9) at baseline and 79.7 (18.2) at 6 months in the overall patient group. Mean (SD) EQ-VAS score was 68.0 (21.0) at baseline and 83.0 (17.5) at 6 months in the intervention arm, and 64.5 (20.9) at baseline and 77.5 (18.6) at 6 months in the control arm.

Resource use

Modified CSRI

Table 16 shows the mean total health and social care costs based on resource use data collected from modified CSRI at baseline, 2, 4 and 6 months. Out-of-pocket expenses (OOPE) incurred by the patients were excluded from the calculation of mean total costs on the modified CSRI as the analysis was conducted from the perspective of NHS and PSS. Within the overall patient group, mean (SD) was £1859 (£3067) at baseline and £165 (286) at 6 months. At baseline, patients in the intervention arm incurred a mean (SD) total cost of £1761 (£3064), and those in the control arm a mean (SD) total cost of £1941 (£3104). At 6 months, the mean (SD) total cost for the intervention £125 (£234) and £197 (£322) in the control arm. Breakdowns of these costs are available in Appendix 2, Tables 35–38.

Tuberculosis-related medications

Table 17 shows the mean costs for TB-related medications in the whole group and by randomisation arm. For the total group, mean cost was £158 (113) at baseline, £59 (58) at 3 months and £376 (1788) at 5 months. The increase at 4 and 5 months was driven by the prescription of a high-cost drug, bedaquiline to patients in both intervention and control arms. Mean (SD) cost was £128 (£70) at baseline and £288 (£1094) at 5 months in the intervention arm and £184 (£136) at baseline and £434 (£2141) at 5 months in the control arm.

In both arms, mean session length was longest at baseline (65 minutes intervention, 47 minutes control) and shortest at 6 months (19 minutes intervention, 10 minutes control).

Cost patterns matched resource use patterns. At baseline, mean session cost was £55 at intervention sites and £40 at control sites, and at 6 months was £16 and £9, respectively.

Overall, the mean (SD) EQ-5D-5L utility score was 0.75 (0.27) at baseline and 0.85 (0.22) at 6 months, with no significant differences between arms.

Mean (SD) EQ-VAS scores for the total group and by arm were 66.1 (20.9) at baseline and 79.7 (18.2) at 6 months, with no significant differences between arms.

Overall mean (SD) health and social care costs were £1859 (£3067) at baseline and £165 (£286) at 6 months (intervention £1761 and £125; control £1941 and £197, respectively).

Main cost drivers included inpatient admissions and visits to A&E at baseline and two and 4 months. At 6 months, use of TB nurse and support service was the largest cost.

Mean (SD) costs for TB-related medications in the whole group were £158 (£113) at baseline, £59 (£58) at 3 months and £376 (£1788) at 5 months. The increase at 4 months and 5 months was driven by the prescription of a single expensive drug (bedaquiline) to patients in both intervention and control arms.

Process evaluation

Demographics

Patients

A total 31 patients consented to take part in the process evaluation study (intervention n = 15, standard care n = 16). Interviews were conducted with 25 patients and observations with 23 patients (Table 18).

Two baseline consultations were observed at intervention sites and five at standard care sites. Seven follow-up consultations were observed at intervention sites and eight at standard care sites. Follow-up observations at intervention sites were conducted at month 1 (n = 1), month 2 (n = 3), month 3 (n = 1), month 4 (n = 1) and month 6 (n = 1). Follow-up observations were completed at standard care sites at week 2 (n = 1), month 2 (n = 1), month 3 (n = 3) and month 4 (n = 3).

Demographics of participants who were patients (patients who participated in the process evaluation are referred to in this section as ‘patients’, to differentiate them from participants who were CMs) are shown in Table 19. Most individuals were from Asian/Asian British (n = 14) or White ethnic groups (n = 10) and were born outside of the UK (73.3%). Most had extra-pulmonary disease (n = 18). Only two individuals had previous active disease (n = 1) or latent TB infection (n = 1). Demographics of patients recruited to the process evaluation were similar to the full study population (n = 79).

Case managers

Five interviews were completed with CMs from the two study intervention sites. CMs had a range of experience as a specialist TB nurse, from 1.5 to 14 years [mean 8.1 (SD 5.5) years]. Appendix 2, Table 39 shows the number of study patients for whom each interviewed CM was responsible.

Key results

The process evaluation results are presented based on the ‘key functions’ described in the MRC process evaluation guidance,79 which include intervention context, implementation processes (feasibility, fidelity, dose, adaptations and reach), and mechanisms of impact (including responses to and interactions with the intervention, mediators, and unanticipated pathways and consequences).

Intervention context

Theme 1: The impact of COVID

Data from patient and CM interviews suggested that the COVID pandemic had an effect upon TB care. Several patients described how the pandemic affected their pathway into care by delaying their TB diagnosis, either due to increased hesitancy to seek care or their perceived effects of COVID upon the healthcare system:

I was coughing for a long time…I resisted going for some time as it was last March and I didn’t want to be diagnosed with COVID.

I had a delay with the treatment as the GP refused to accept it wasn’t COVID.

Case managers stated that COVID changed patients’ perceptions of TB and treatment. One CM felt that public awareness of infectious respiratory disease increased, which they described as helpful in educating patients about the comparisons between transmission of COVID and TB. Two CMs felt that patients had a better understanding of terminology used in TB because of the pandemic, such as ‘isolating’ and ‘contact tracing’. Finally, they believed that the stigma of infectiousness and mask-wearing may also have reduced as a result of the pandemic and mask mandates. One patient also found that the pandemic assisted their interaction with medication – which they described as being easier to take during lockdown and when having to work from home.

Implementation

Theme 2: Case manager perspectives on Intervention training

Most CMs felt prepared to deliver the IMPACT intervention with the training provided. Several CMs noted that any initial uncertainties in delivering the intervention reduced with practice. At one intervention site, CMs said that having the research nurse present during clinics was useful as they could ask for help if this were needed. One CM felt unprepared to deliver the intervention; however, they said this was due to the time which lapsed between their training and the first intervention patient – itself affected by delays in recruitment due to COVID restrictions.

Theme 3: The feasibility of delivering the intervention components

Case managers had mixed views regarding the feasibility of delivering the manualised intervention. This appeared to relate mainly to the time needed to deliver its components (as shown earlier this was 65 minutes for baseline and 22 minutes for follow-up for the intervention arm vs. 47 and 15 minutes for follow-up with standard care). The two baseline consultations observed in the intervention arm took 50 and 57 minutes to complete. Standard care baseline consultations observed took between 21 and 53 minutes. Follow-up observations took between nine and 40 minutes to be completed in the intervention arm, and between two and 40 minutes in the standard care arm.

Case managers noted that one of the issues affecting the feasibility of delivering the TNA component was how long it took to deliver. CMs reported that the TNA required 5–20 minutes to complete [mean 10.4 (SD 6.8) minutes], which was similar to the range recorded in observed consultations (baseline 10 minutes; follow-up 6–16 minutes). In comparison, the observed time for completing the usual care risk assessment was 1–4 minutes. Observations showed that the longer times recorded for the TNA were often due to interruptions or tangential conversations that resulted from questions in the TNA. Some CMs felt the time needed to administer the TNA was ‘fine’ while others thought it was too long and impacted on the time available to manage other patients within the clinic. All agreed that the complexity of the needs of the patient receiving the TNA determined the time taken to administer it.

The length of time required for other intervention components such as the booklet, videos and the weeks 1 and 6 phone calls was also an issue for some nurses. CMs reported that providing the booklet and completing the associated activities took between 5 and 10 minutes [mean 7.2 (SD 2.6) minutes]. Showing the videos and answering questions generated by these took 8–15 minutes [mean 12 (SD 3.6) minutes]. Other factors that CMs perceived as affecting TNA delivery included having to use a recommended session structure that felt rigid and sometimes disjointed.

Theme 4: Dose of the intervention components delivered

Most CMs delivered the main components of the intervention to most of their patients. Most CMs interviewed said they usually completed the TNA at every appointment. This was noted during observed consultations, though recorded TNA completion (using clinical report forms filled out at the time of the patient review suggested that this was somewhat lower – see Table 20).

All patients at intervention sites who completed a late interview recalled having had a repeat assessment of their needs across treatment. No repeat risk assessments were observed in the standard care arm, yet nearly all of the patients interviewed [6/7 (85.7%)] recalled having a repeated assessment of their needs. It was therefore unclear what this may have referred to.

In baseline observations, the booklet and videos were shown to both patients observed. However, in interviews, all patients interviewed recalled receiving the booklet though not all recalled seeing the videos in their baseline session (8/12, 66.7% patients). Nurse checklists indicated that while all patients (n = 34) were provided with the booklet, 32/34 (94.1%) patients were shown the videos.

Several CMs [3/4 (75.0%)] said they provided intervention patients with leaflets from a national TB charity (TB Alert) with information about TB medicines and side effects. In observed sessions, these leaflets were both provided to patients and used for medicines and side effect counselling. These leaflets were also observed as being given to patients at standard care sites.

The frequency of delivery of the minor intervention components was similar across both study arms. More intervention patients recalled receiving phone calls from their CM during treatment [11/12 (91.6%) intervention patients vs. 8/12 (66.7%) standard care patients]. Patients in both study arms recalled the phone calls having similar content, such as checking on treatment progress, raising issues of concern, and informing them of test results. Disease-related imaging such as chest radiographs shown to patients was similar between arms [8/12 (66.7%) intervention participants; 7/13 (53.8%) standard care participants]. All CMs interviewed said they did not show imaging more frequently to intervention participants than they would in usual care.

Theme 5: Fidelity and adaptations in delivering the intervention components

While the main intervention components were delivered to most participants, the methods used to achieve this varied. For example, some patients were asked to read through the TNA themselves and comment rather than being asked the questions by the TB nurse.

Deviations were also observed in terms of the assignment of adherence support. For example, one CM offered adherence support before completing the TNA and identifying any risk factors. Some CMs also appeared to rely on their personal judgement to assign adherence support, as was noted in a standard care clinic.

Adaptations were also made to the delivery of the booklet. Two CMs said they were able to deliver the booklet as manualised (completing the medication activities). However, the others said they were only able to provide the booklet. CMs noted that some patients preferred to complete the activities at home, although they were unsure if this had actually happened. In observations, most CMs appeared to discuss a medication plan with the patient, even if this were not completed in the booklet, and encouraged patients to use the medication and symptom diary. In follow-up observations, the medication list and action plan were only reviewed and changed by one of two relevant CMs. Nearly all CMs reviewed the medication and symptom diary at follow-up if patients brought their booklet.

Theme 7: Reach of the intervention

In the pilot trial overall, 130 eligible patients were approached, and 82 (63%) of these consented to take part. Intervention site 2 had the highest refusal rate, with 48% (n = 16) of those approached declining. This was followed by Standard care site 1 [n = 12 (35%) declined], Intervention site 1 [n = 11 (28%) declined], and Standard care site 2 [n = 9 (23%) declined], respectively. Fifteen (31%) of those who did not wish to take part were female, and 33 (69%) were male. The study participants’ age range was representative of the age group most affected by TB in the UK (25–44 years).1 Balanced gender distribution was observed between the trial arms and was similar to the TB population in the UK.1 Most patients were Asian/Asian British, followed by White, Black African/Black Caribbean, Other, and Black British ethnicities.

Identified social risk factors were lower than reported in the wider UK TB population. In 2020, 4.5% of people with TB reported having a current alcohol problem,1 though this was 1.3% in our study sample. Similarly, nationally, 4.4% of people with TB stated having a history of drug misuse compared to 1.3% in our sample. This was also the case for both homelessness (present in 4.3% of a general UK TB population, and 1.3% in our sample) and imprisonment (4.0% of UK TB population, none within our sample). In our data, set we recorded immigrant/migrant status in 3.8% of our participants. Similar data for England in 2022 records this at 5.2%.81

The proportion of our study population with HIV coinfection was 1.3%, which was somewhat lower than national figures (at around 3.6%). So too was clinically relevant DR, which was present in 5.1% of participants (compared to 11.9% in England in 2022 with resistance to any of isoniazid, rifampicin, ethambutol or pyrazinamide).81

Mechanisms of Action

Theme 8: Patient engagement with intervention components

Patient engagement with the intervention components varied, both between patients and across the components. During observations patients appeared mostly engaged and forthcoming with information when answering the TNA. Two of the CMs said they felt that the repetitiveness of the items in the follow-up TNA could cause patients who did not have any concerns to ‘switch off’. All intervention patients interviewed said they felt comfortable discussing the topics in the TNA, however two patients (in the control arm) reported feeling uncomfortable about answering the items on the usual care risk assessment.

Patient interviews suggested that they were most likely to engage with the booklet out of the take-home intervention resources offered to them. All these patients reported using the booklet during treatment, and three-quarters read the entire booklet. Some said the booklet was particularly helpful for providing information early in treatment. Most patients reported engaging with the interactive medication resources to record adherence and side effects and to write down questions to subsequently ask their CM [9/12 (75%) patients]:

I really like the checklist every day to take the medication. There are symptoms column which I really like as I can put all of my feelings there that disturbed me and I can tell my nurse at my next appointment. It contains a lot of information about the video, the websites I need and working related website for my rights if I decided to stay home and not work. It’s great.

About half of the patients observed brought their booklet to follow-up appointments [3/7 (42.9%)]. CMs felt that most patients were engaged when going through the booklet and activities [4/5 (80.0%)], although this did appear to be higher at diagnosis and then drop off unless patients were using the diary to record adherence or side effects. Several patients reported sharing the booklet with others, including their partner/spouse, family member, and/or coworkers. When this did not happen, the main reason given was that they felt others may not find it useful or relevant. One patient did not want to share the booklet outside of their immediate family in case this caused stigma and fear regarding contagiousness.

Observations and CM interviews suggested that patients were also generally engaged when viewing the videos at baseline, although this was patient-dependent. CMs felt that some patients’ interest level was initially high but this then declined during viewing of the 3-minute videos. One patient reported watching the videos again at home, and they did so five times; though only 30% of patients reported watching the videos at home in their own time in the intervention feedback which they provided at the end of the study.

Three patients shared the videos with others, including a partner/spouse, family member, friend and/or manager/coworker. One CM felt that the videos encouraged a more questioning approach and attitude – something confirmed in a patient interview:

…The video gives a lot more information about what it is and what you should do. It was not overwhelming and it was easy to understand and it made me think about questions to ask.

Theme 9: Intervention components changing perceptions and practicalities

These was some evidence that the booklet and video had an impact on patients’ perceptions of TB and treatment, and their ability to adhere to their medication. The booklet appeared to influence patients’ perceptions of TB and treatment, both improving their understanding of TB and reducing their concerns:

It tells you all about TB and its effects and how you can get it and the symptoms and that there is hope at the end of the tunnel.

It tells you it’s curable and it is reassuring.

The booklet appeared to support patients to take their treatment. Patients identified nearly all sections of the booklet as making a difference to the way they took their medication. Patients commonly spoke about how the drug therapy resources in the booklet supported their adherence, and allowed them to monitor side effects and organise their medicines:

All the medicines were very daunting so it was helpful to see it all laid out in a booklet.

It was useful as I used to write the dates and side effects in there… useful to look back and see when they finished.

The video appeared to increase some patients’ beliefs about the necessity of adhering to treatment for TB:

Without the video I would be less encouraged to take my medication and less motivated as I would not know my goal. In the video it says specifically that the medication is working even though I get symptoms and side effects. Helped to keep me optimistic.

There was also evidence of the video changing patients’ perceptions about the cause, transmissibility and curability of TB:

[I]t explained that TB can be in our body for a long time before it becomes symptomatic and that made sense.

Before it was scary and I thought I would be contagious and not be able to work. The video was a relief as my TB was not contagious. I knew what to expect after I watched it. It’s a very good job.

As with the booklet, a small number of patients felt that the videos were interesting, although they did not feel the videos made a difference to their adherence as they would have taken treatment anyway.

Similar numbers of patients in both arms reported seeking extra information about TB. A couple of intervention patients noted that the extra information they found was already covered in their booklet. Generally, other information sources were viewed more negatively by patients, with some noting that other information could be overwhelming or unreliable. A few patients at control sites wanted to be provided with more information about TB, and suggested formats similar to what was provided in the intervention:

Didn’t read leaflets and information, a video would be more helpful. I would like a video from the NHS that you know is accurate that you could watch quickly, and had simple animations or something. It would be much easier to take in.

The minor intervention components (e.g. showing patients their radiological imaging, phone calls during treatment) which were used in both standard care and intervention arms (though positively promoted only in the latter) appeared to be of similar benefit to influence patient’s motivation and ability. Patients in both arms felt that being shown their imaging affected their motivation or understanding of disease:

I wanted to see that shadow on my lungs go away.

[It helped] explain the need for treatment and why my chest needs to recover.

Similarly, the phone calls appeared to impact patient’s perceptions of support in both study arms:

…I really appreciate it…I feel I can text the nurse whenever. Made me feel really comfortable. She even replied to me on a Sunday.

I was supported and felt like I can rely on someone. It feels really hard to experience this alone and all the support matters.

Theme 10: The intervention helping to shape care delivery

A similar number of patients in both arms reported feeling that either the TNA or usual care risk assessment fully addressed their needs [5/5 (100.0%) intervention patients, 5/6 (83.3%) standard care patients]. When asked if there were extra support that patients needed but did not receive, two intervention arm patients expressed a greater need for mental health support. Two standard care arm patients indicated that extra help with a wider range of issues including financial and immigration concerns, mental health, and transport was required.

Case managers had differing opinions regarding whether the TNA identified risk factors for non-adherence better than the usual care risk assessment. Two CMs felt that the TNA performed better, two that the TNA and usual care risk assessment did the same job, and one felt the usual care risk assessment was better. Clinic observations revealed that the usual care risk assessment was completed much faster but gathered less information on a relatively limited number of risk factors.

It didn’t make much of a difference to straightforward patients or the severely chaotic, but very good for vulnerable patients in the middle who were a little bit shy or afraid to speak up and had some concerns.

[T]his way we directly ask rather than assume they don’t have doubts. Also if we ask the question it makes it okay for them to talk to you about it… it opens the floor to talk about these things.

Case managers also had mixed views regarding whether the repeated assessment of risk factors (i.e. at each consultation) in the intervention arm made a difference to patients’ care. One CM felt that repeating the TNA was the part of the intervention that most made a difference to patient’s treatment.

The TNA and repeating the TNA on a regular basis does make a difference… showing patients that we want them to be open about these issues and that we continue asking those questions. It’s important to keep up rapport and support. If circumstances change and they don’t mention it, they will be asked anyway.

Case managers also differed in their opinions on whether the TNA influenced how they assigned additional adherence support. Some felt that this did not make a difference because they would have done this anyway as part of usual care. One CM said they rely on their personal judgement and years of experience and that this was disrupted by the TNA. Another CM felt that the TNA complicated their delivery of care by, for example, needing to prepare dosette boxes (for adherence support) during the appointment. In contrast, one CM felt that the TNA meant they offered support more regularly.

It did a better job at assigning adherence support because we always ask about it. Might not always offer support for all those issues unless the patient is sitting here overwhelmed.

Case managers reported that the adherence support option they assigned most often was a dosette box. One CM attributed this to more complex patients being less likely to be recruited into research. Two patients who received dosette boxes felt that these helped them to organise and adhere to their treatment:

If they were in different boxes every day you would be counting and asking how many days have I been taking it and I’d be worried about missing treatment. With a box I know I have to take what is in the box.

The box made it really easy to remember to take the doses and I didn’t need to think about it or try to organise it all myself. It was a way to monitor things. I think I would have lost count. I think as it’s important that you don’t skip pills it was a lifesaver.

The intervention did not appear to influence adverse effect and medicine counselling provided to patients, which was observed as thorough in both study arms. CMs provided detailed information about TB medicines, their side effects and reasons why these occur as well as helping patients to develop a plan to take their medication. Although side effects and adherence were addressed during observations with both study arms, it appeared that CMs in the intervention arm more directly assessed adherence, which may have been a result of the TNA. In comparison standard care sites, were more likely to assess adherence by asking patients about their treatment routine or checking their knowledge of the treatment regimen (e.g. how many of each medication they took each day). Observations also showed that intervention CMs often used the intervention booklet to assess and address adherence and side effects and encouraged patients to use the diary even if they did not bring their booklet to the appointment.

Most CMs did not think that the IMPACT intervention influenced how they delivered care to patients not in the study. However, one CM felt that they may have more often offered adherence support to patients as a result of the IMPACT intervention. Most CMs also did not feel that the intervention improved continuity of care compared to usual care. Two CMs highlighted ways in which they felt the intervention had negatively interfered with their standard care procedures for intervention patients. One noted that a patient wrote their side effects in the diary, rather than telling the CM about them, as the booklet had made the patient assume that these adverse effects were to be expected. Another CM felt that delivering the intervention components interfered with their ability to establish rapport with patients.

It doesn’t enhance a TB nurse who has experience, it handicaps them.

Case managers who had provided the non-English language videos to patients had mixed views on whether these improved information provision for these patients. One felt that the in-service advocates they could access could provide this information. The two CMs who had given non-English versions of the booklet to patients felt that it offered similar or better information than what they would otherwise have used.

Case managers also differed in their views on whether the intervention influenced patients’ engagement with care: one felt that engagement was increased because of the intervention components. Most, however, did not feel that the booklet made a difference to how patients communicated with them, although one CM said this was dependent on whether or not the patient actually used the booklet (i.e. the medication and symptom diary).

There was no clear evidence that the intervention improved the relationship between CM and patient. CMs felt that the intervention improved their relationships with patients, and patients in both study arms described positive relationships with their CMs whom they viewed as an available source of support during treatment. Standard care observations showed that all CMs offered a lot of support and reassurance to patients, for example, informing patients that they can contact them any time with questions or concerns. Standard care CMs also provided similar practical support to that described in the intervention, such as helping patients create a medication routine and providing clear instructions about their medication regimen.

Theme 11: Participants responses to the intervention content

Patient feedback suggested that overall, there were positive responses to the type of information and resources provided in the intervention. Patients appreciated the format and information provided in the videos:

It took me step by step and I really liked how it explained everything to me.

Colourful, animation is great, very easy to digest. Fun and educational.

Patients described the information in the booklet as reassuring for their care:

It was good to have something that confirmed the discussion you had in the TB clinic that you could refer back to. Particularly when you are on your own, you are not sure if you heard things the doctors and nurses said correctly or if you misremembered. It consolidated what they said.

It is useful to have it with you when you go to appointments and makes you feel more in control.

A few patients suggested that the booklet could be improved by providing extra space to record details in the medication and symptom diary, and more pages for those on extended treatment. Patients did not have suggestions for the videos.

When patients were asked to rate if the intervention changed the way they thought about their treatment for TB on a scale of 0–100 (100 = yes), most patients felt the intervention did do this [mean 70.3 (SD 27.5)]. On a similar scale, patients reported that the intervention material made it easier for them to take their treatment [mean 67.2 (SD 34.7)], and they would recommend it to others [mean 74.2 (SD 28.4)]. When questioned about the individual components, all aspects scored highly (Figure 17).

FIGURE 17 Participant responses to question about individual components of the Intervention: ‘Overall how helpful did you find the’. Note: ‘Helpfulness’ is shown on the horizontal axis and increases from left to right. 0 (extreme left) = ‘not at all helpful’; 100 (extreme right) = ‘extremely helpful’. (Figure not reproduced here; see the original article.)

Theme 12: Case manager views on incorporating the intervention into standard care

Case managers felt that all intervention components needed some adaptations if they were for to continue using these as part of standard care. Two said they would not want to continue using the TNA after the study, and three said they would like to but only with some changes. These reflected a desire for more flexibility about whether to deliver the follow-up TNA to all patients, and for the TNA to be part of a single document that would include all other necessary paperwork and be completed electronically.

The videos were well regarded, with all but one CM saying they would like to use these after the study with some adaptations. Again, this was generally a need for greater flexibility, to either decide which patients should watch the videos (rather than providing these to all) or for patients to watch these outside of the appointment (e.g. at home or in the waiting room). The CM who did not want to continue using these felt that the content needed revising, namely (1) reducing the amount of information provided (which they felt was repetitive), and (2) changing language that they considered patronising to patients (e.g. the phrase ‘say goodbye to TB’).

The booklet was most popular with CMs: two felt that the booklet was the part of the intervention which had the biggest impact on treatment. All CMs wanted to continue using the booklet after the study (one in its current form – including completing the interactive activities with the patients during clinic), while others wanted to provide the booklet but ask the patient to complete these at home. Two CMs also suggested including the medication pictures from the TB Alert pamphlet in the booklet.

Case managers had mixed views about the minor components: they were uncertain whether the phone calls made a difference to patients’ care and whether they would like to continue doing these as routine. The main reason for this was the time that the calls would take. CMs who wanted to continue phone calls expressed a need for greater flexibility in deciding when to deliver them. One CM suggested that they might be better incorporated into usual care procedures if the calls were booked as appointments.

Theme 13: Unanticipated pathways

There was an unexpected pathway of the intervention and of the wider study procedures which may have influenced the mechanisms of action. The unanticipated effect of the intervention was that many patients discussed the booklet as a source of support for them during their treatment, appearing to find the booklet reassuring:

I know it’s there for me and I can look at it any time.

It’s like my little pal and I feel very attached to it.

It is useful to have it with you when you go to appointments and makes you feel more in control.

A second unanticipated effect was the influence of the MMB, used primarily to assess the trial’s adherence outcome, upon patient’s motivation and ability to take their treatment:

[I]t helped me to take it and keep to a routine and more determined to take on time.

It helps you keep a focus every day.

Overall, more patients in the standard care arm reported that the box made a difference to how they took their treatment (10/13, 76.9%) compared to intervention patients (4/11, 36.4%). A few standard care patients mentioned that the monitoring aspect of the box was what influenced their adherence. Patients also discussed how they thought the sound the box made when retrieving their medication had an anchoring effect on their adherence:

It makes me remember and you get accustomed to the beeping sound every morning, and I know I need to hear the beeping sound to remember.

I like it, I like the sound of it opening and closing. Helps me remember to take it.

Patients also used their box to organise their medicines and keep to a routine. However, for two patients who had a dosette box, the interviewer noted that it was difficult to identify whether views towards the ‘medication box’ incorporated the dosette as well. Some patients also used their box to help organise other aspects of their treatment:

I found it useful to store all my papers and documents and hospital letters etc to have everything to do with my TB care in one place.

Not just a box for medication – I keep all my clinic letter and appointment letter in it and it helps to keep it all in one place.

While some CMs felt the box may have made patients more aware of their treatment or more likely to bring their medicines to their appointment, others were unsure of any effects this may have had upon their treatment.

During observations, patients appeared mostly engaged and forthcoming with information when answering the TNA. However, two CMs said the repetitiveness of the follow-up TNA may cause patients to ‘switch off’. The booklet was the component that patients most engaged with, with many sharing it with family or coworkers. There was no clear evidence that our intervention improved the relationship between CM and patient.

Some patients interviewed reported that the animated videos helped changed their perceptions of TB and treatment, improving understanding and reducing concerns. Almost all patients interviewed reported that the booklet was helpful and allowed them to monitor side effects and organise their medications.

In general, CMs felt adequately equipped to deliver the intervention; initial uncertainties resolved over time and with practice. Most CMs delivered the main components of the intervention to most of their patients. At intervention sites, the TNA was completed for 97% of baseline consultations and between 57% and 74% of 1- to 5-month consultations.

CMs were less sure about the overall feasibility of the intervention, citing primarily the length of time required to administer the three main components. Adaptations were sometimes made by CMs, including asking patients to read through the TNA themselves, asking patients to read the booklet at home (instead of in clinic), or relying on their own judgement to assign adherence support (instead of following the TNA’s recommendation’s). The additional activities mandated by the study protocol were also sometimes a problem.

In general, patients were appreciative of the additional information and resources provide, describing the video as ‘easy to digest’ and the booklet as making them feel ‘more in control’. Less experienced CMs found the intervention helpful, as it provided structure and a standardised way to inform their patients. More experienced CMs said it was less helpful, as they felt they already had an effective approach. One CM felt that repeating the TNA was the step that made the most difference to patients.

Finally, the COVID pandemic caused considerable disruption to the study and is likely to have altered the effects of the intervention. Possible mechanisms include changes in perceptions of respiratory illness, changes in adherence behaviour (because people were working from home), and disruptions to service delivery (because of the pressure on the health system).