Quality of life, tuberculosis and treatment outcome; a case–control and nested cohort study

Background Global tuberculosis policy increasingly emphasises broad tuberculosis impacts and highlights the lack of evidence concerning tuberculosis-related quality of life (QOL). Methods Participants were recruited in 32 Peruvian communities between July 13, 2016 and February 24, 2018 and followed-up until November 8, 2019. Inclusion criteria were age ≥15 years for “patients” (n=1545) starting treatment for tuberculosis disease in health centres; “contacts” (n=3180) who shared a patient's household for ≥6 h·week−1; and randomly selected “controls” (n=277). The EUROHIS-QOL questionnaire quantified satisfaction with QOL, health, energy, activities of daily living (ADL), self, relationships, money and living place. Findings Newly diagnosed tuberculosis was most strongly associated with lower QOL scores (p<0.001). Patients initially had lower QOL than controls for all EUROHIS-QOL questions (p≤0.01), especially concerning health, ADL and self. Lower initial QOL in patients predicted adverse treatment outcomes and scores <13 points had 4.2-fold (95% CI 2.3–7.6) increased risk of death versus those with higher QOL scores (both p<0.001). Patient QOL was re-assessed 6 months later, and for patients with successful treatment QOL became similar to participants who had never had tuberculosis, whereas patients who did not complete treatment continued to have low QOL (p<0.001). Multidrug-resistant tuberculosis was associated with lower QOL before and during treatment (both p<0.001). Contacts had lower QOL if they lived with a patient who had low QOL score (p<0.0001) or were a caregiver for the patient (p<0.001). Conclusions Tuberculosis was associated with impaired psychosocioeconomic QOL which recovered with successful treatment. Low QOL scores predicted adverse treatment outcome. This brief EUROHIS-QOL eight-item questionnaire quantified the holistic needs of tuberculosis-affected people, potentially guiding patient-centred care.


Introduction
Tuberculosis (TB) makes ∼10 million people ill each year globally, and kills 1.5 million of them, more than any other infection [1]. A TB diagnosis has consequences far beyond patients' physical symptoms, including associations with mental health and catastrophic costs [2,3]. The World Health Organization (WHO) End TB Strategy aims to dramatically reduce TB cases and eliminate economic and social burden through integrated patient-centred care and prevention [4,5]. With currently available diagnostics and treatment the End TB Strategy goals will not be met. Thus, there is increasing priority to mitigate TB-related financial costs and provide support to the psychological and social consequences of TB complementing the longstanding focus on biomedical interventions.
The WHO defines health as "a state of complete physical, mental and social wellbeing and not merely the absence of disease or infirmity", yet national TB programmes are almost universally evaluated in terms of the number of people they have treated and cured, never their impact on wellbeing [6]. Economists highlight the importance of populations' wellbeing to improve productivity and economy, and advocate its use as an indicator of societal progress [7,8]. Tools measuring "quality of life" (QOL) are most appropriate for assessing subjective wellbeing, because they assess satisfaction and functioning that respond to life events and are relatively stable compared to substantially different, fleeting feelings of euphoria or happiness [8].
The WHO strategic and technical advisory group (STAG) for TB in 2017 stated that ignoring QOL in TB is "unacceptable and unethical", recommending that the WHO pursue assessment of TB-related QOL [9]. However, there is no TB-specific QOL questionnaire available. Existing tools such as the EuroQOL-5D and the 36-item Short Form Health Survey, which are used for quality-adjusted life years calculations, have been used in TB-affected people [10,11]. However, these tools neglect economic and social domains that are important for the synergistic relationship between TB, poverty and isolation. The WHO QOL group created a 100-item questionnaire that subjectively measures universal aspects of QOL, applicable to different languages and cultural settings. Truncated versions brief enough for operational use while maintaining psychometric properties, like the 24-question WHO-QOL-BREF, and subsequent eight-question EUROHIS-QOL were produced [12]. Both tools assess four dimensions of QOL highlighted to be affected by TB: physical health, psychological health, social interactions and satisfaction with living conditions, including economic QOL [12][13][14][15][16]. The EUROHIS-QOL is brief without losing the psychometric properties of the familiar WHO-QOL-BREF, but has only been used to evaluate QOL in few conditions [12].
We aimed to evaluate the EUROHIS-QOL tool for quantifying QOL in TB-affected people ( patients and their contacts) versus healthy community controls, and to assess whether QOL at the time of diagnosis predicts treatment outcome, including survival.

Methods
Ethics approval was given by the following committees: the Peruvian Ministry of Health (DIRESA Callao, Peru), Asociacion Benefica Prisma (Peru) and Imperial College London (UK).

Participant recruitment
For this case-control study with a nested prospective cohort study, from July 13, 2016 to February 24, 2018 participants were recruited concurrently in 15 desert shantytowns and 17 urban communities in Callao, Peru, selected for their high TB case notification rates, and followed-up until November 8, 2019. Inclusion criteria were age ⩾15 years in people starting treatment for TB disease in community health centres (termed "patients"; n=1545); people who reported sharing a patient's household for ⩾6 h·week −1 during the 2 weeks before a patient of any age in their household commenced TB treatment (termed "contacts"; n=3180); and randomly selected community (termed "controls"; n=277). Controls were selected by allocating a number to each residence in the 32 participating communities, and using a random number generator to select households, who were interviewed unless no adult gave written informed consent (in which case another randomly selected household was invited). We termed contacts who were a patient's parent or spouse "carers", because they were likely to have a caring role for the patient in their household. We termed other contacts "noncarers". Exclusion criteria were inability or refusal to give informed written consent/assent.

Measures
After consent, the EUROHIS-QOL questionnaire (see box 1 for the English version, and supplementary table  S2 for the Spanish translation used) was read immediately to participants who chose their responses. For patients recruited after November 18, 2016, this was repeated after 6 months' treatment. To minimise loss to follow-up, three follow-up visits were attempted for each patient. Details of the EUROHIS-QOL, psychosocial, socioeconomic, demographic, clinical and treatment outcome data collected are shown in box 2.

Analysis
Data were analysed using Stata (version 13; StataCorp, College Station, TX, USA). The EUROHIS-QOL was assessed as recommended by the Medical Outcomes Trust Scientific Advisory Committee [17]. Paired analyses were used when comparing patient QOL scores at baseline versus follow-up. Multivariable regressions to identify factors associated with QOL and ascertain whether QOL predicted treatment outcome were analysed using backward stepwise selection of variables of interest with p<0.1 in univariable analysis, with robust standard errors and without imputation. QOL clustering at the participant, household and community level was adjusted for by a multilevel random intercepts model. This was done because several characteristics were similar within households, healthcare services varied between communities, and some communities received support aiming to improve TB outcomes from the CRESIPT (Community Randomised Evaluation of a Socio-Economic Intervention to Prevent TB) trial [18]; nongovernmental organisations; and the national TB programme.

BOX 1 The EUROHIS-QOL tool in English.
Please listen to the questions with regards to quality of life and pick the best option for you. We ask that you think about your life in the past 2 weeks. BOX 2 The psychosocial, socioeconomic, demographic and clinical data collected from study participants at baseline and at 6-months' follow-up.
The EUROHIS-QOL questionnaire constitutes eight questions quantifying satisfaction with health, energy, activities of daily living, self, relationships, money, living place and global quality of life (QOL) (see box 1). Each question was answered using a five-point Likert response scale numerically scored e.g. 0="very dissatisfied", 1="dissatisfied"; 2="neither dissatisfied nor satisfied"; 3="satisfied"; 4="very satisfied". As all questions were equally weighted [14], the total score, termed QOL score, was calculated as the sum of the eight questions, with a possible range of 0-32. Higher scores indicated better QOL, a total score of 0=very dissatisfied; 8=dissatisfied; 16=neither dissatisfied nor satisfied; 24=satisfied; and 32=very satisfied in all eight items. There is no defined threshold for good versus bad QOL, but a total score of ⩽16 suggests on average no satisfaction for any of the eight items, termed "illbeing". This questionnaire has shown good cross-cultural performance and been validated in numerous languages including Spanish [12]. However, focus groups with 16 research nurses before and after a community pilot evaluation prompted refinements in the Spanish wording and addition of pictures of faces representing the scale, which were used only if individuals struggled to understand the options. These refinements in language are demonstrated in supplementary table S2. Baseline data In addition to the EUROHIS-QOL questionnaire, patients and controls were interviewed using the following: Beck depression inventory II [44], excluding from analysis six potential symptoms of tuberculosis (TB) (appetite, weight loss, health concerns, poor ability to work, tiredness and trouble sleeping) for which higher scores from 0 to 45 indicated low affect and depression; SASCAT (World Bank Shortened and Adapted Social Capital and Perceived Safety Assessment Tool) [45] quantifying the number of perceived sources of emotional support; an adapted 15-item Explanatory Model Interview Catalogue questionnaire [46], for which higher scores from 0 to 45 indicated greater perceived stigma; and a locally validated assessment of basic knowledge of TB disease, treatment, prevention and available services. For all participants, socioeconomic position (poverty) was assessed using completion of secondary education as a proxy measure [47]. Patients were asked a locally validated questionnaire to characterise symptoms and treatment, and their sputum was collected at recruitment for microbiological assessment using sputum smear microscopy, GeneXpert MTB/RIF and thin-layer agar solid culture using the MDR/XDR-TB Colour Test [48]. Follow-up data ∼6 months after recruitment and baseline data collection, patients were re-interviewed using the EUROHIS-QOL questionnaire, and asked about treatment and symptom status. Treatment outcome We used Peruvian guidelines to define outcomes. These follow WHO recommendations except for "incomplete treatment" (World Health Organization termed "loss to follow-up") which Peruvian policy defined as no treatment for >1 month usually due to locally termed "abandonment". In addition, we included rare patients with treatment suspension or who never commenced treatment [43,49]. Patients were considered to have adverse treatment outcome if they died during treatment, their treatment failed or they had incomplete treatment. Unknown outcomes (due to missing data or transfer away with unsuccessful follow-up) were excluded from analyses regarding treatment outcome. Recruited patients whose TB diagnosis was later rescinded were excluded in treatment outcome analyses.

Multivariable regression
Crude comparisons between patients (at recruitment and follow-up), contacts and controls were confounded by several participant characteristics (e.g. sex, age, current or previous TB and days of TB treatment) being independently associated with both TB disease (table 1) and QOL (table 2). Therefore, for the following multilevel regression, participants were regrouped according to TB status (supplementary figure S4) and analysis was adjusted for age, sex and if they were taking treatment the days between interview and TB treatment initiation. Contacts with current TB (n=105) were excluded as details of TB disease and treatment were unknown; consequently, 5757 QOL interviews were included in the following analysis (table 3 and figure 3).       figure S5). Low QOL was better able to predict death ( p<0.0001) (figure 5, supplementary figures S5 and S6), which occurred in 2.9% of patients (41 out of 1416, 11 treated for MDR-TB). The area under the receiver operating curve for QOL score predicting death was 0.70 (95% CI 0.61-0.79). Patients with QOL score <13 points at baseline had 4.2-fold (95% CI 2.3-7.6) increased risk of death versus those with higher QOL scores. This cut-off predicted death with sensitivity 71% (95% CI 54-84%) and specificity 65% (95% CI 62-68%). Patient risk of death (compared with those with higher initial QOL) was 2.3-fold (95% CI 0.95-5.4) increased if they had illbeing and 4.9-fold (95% CI 2.5-9.7) increased if average QOL was "very dissatisfied" (QOL score <8).

Discussion
The EUROHIS-QOL eight-item questionnaire was a valid instrument to measure general QOL in TB-affected people. It was successfully completed by almost all participants and demonstrated good  Quality of life (QOL) adjusted for age, sex, and if taking treatment, the days between interview and treatment initiation. Multilevel multivariable analysis was used to adjust the QOL score to that of a 30-year-old male from a typical household and community. MDR: multidrug-resistant; TB new: baseline data for patients newly diagnosed with tuberculosis disease; TB 6 months treatment: data for patients continuing TB treatment at the 6-month follow-up; TB carer: contacts who were the parent or spouse of a patient in their household, and assumed to have a caring role; TB previously: controls and contacts who were not TB carers and reported having received TB treatment previously, and patients at 6-month follow-up who were no longer receiving TB treatment; TB never: controls and contacts who reported never having received TB treatment and were not currently TB carers. The  reliability, validity and responsiveness as a patient-reported QOL instrument. Patients with TB, especially MDR TB, had lower QOL than community controls, and this was associated not only with TB symptoms, but also with psychological and social dimensions of QOL. Patients with lower QOL at diagnosis were less likely to complete their TB treatment and survive. This study also assessed QOL in patients' household contacts, drawing attention to the negative impact of informal caregiving on QOL.
TB programmes have traditionally focused on treating the infectious TB pathogen and the symptoms it causes. Increasing awareness of the considerable financial consequences of TB disease led to rapid development of metrics to measure TB-related household costs, and the WHO End TB Strategy target to prevent catastrophic costs [5]. In comparison with this recent focus on economic aspects of TB, awareness of the psychosocial burden of TB has neither an agreed metric nor a target. Our findings demonstrate that TB-related QOL can be measured with the simple EUROHIS-QOL questionnaire. It highlights the need to improve TB-related QOL, including the profound dissatisfaction with one's self, relationships and global QOL, potentially worsened by TB-related distress, stigma and isolation. As costing tools are being rolled out to assess the financial burden of TB, we recommend integrating the EUROHIS-QOL eight-item questionnaire to concurrently assess burden on QOL. As recommended by the WHO STAG, this would simultaneously capture all domains of QOL, not just the economic dimension, identifying the more holistic needs of TB-affected households [9].  Patients and carers had less satisfaction with self, with patients expressing high levels of guilt in the BDI-II questionnaire. Fear of exclusion or blame for infecting others may lead to patients feeling guilty, having poor self-esteem, anticipated stigma and worse perceived symptom severity [19,20]. Longitudinal cohort studies reported that during TB treatment, improvements in mental health took much longer than improvements in physical symptoms [10,21]. In particular, satisfaction with social and role functioning were at their lowest 1 month after starting TB treatment [22]. In addition, our findings show that QOL was impaired during treatment, especially MDR treatment, and support the WHO recommendations for social support such as education and counselling to improve adherence and treatment completion [23].
We also recommend supplementary socioeconomic support, which can further increase the likelihood of successful treatment, helping to return QOL to the community average [24,25]. Misinformation and mistreatment from healthcare professionals can contribute to patient dissatisfaction [26]. Within 2 weeks of appropriate treatment the concentration of infectious TB-causing bacteria in sputum usually reduces by 99% and cough frequency more than halves [27,28]. Therefore, increasing the pragmatism of infection control policies for patients confirmed to be receiving appropriate treatment may improve QOL without increasing TB transmission. Meanwhile, providing training in patients' rights and sharing experiences can challenge discrimination from healthcare workers, while improving working conditions and associated stresses further cultivates compassionate care [29][30][31].
Low QOL at initiation of TB treatment, and especially scores indicating that a patient was very dissatisfied, was associated with death, adverse treatment outcome and treatment noncompletion. Patients in this study had both PCR-and culture-based drug susceptibility testing and started MDR treatment when drug-resistant TB was suspected or confirmed. Yet, 41 patients died, which the EUROHIS-QOL score predicted. Currently there are no policies in place regarding adjuncts for individuals who are taking seemingly appropriate treatment, but likely to have fatal disease, even though the End TB Strategy aims to have zero TB deaths by 2030 [5]. Impairment of QOL associated with TB defines important multisystem ill health. The EUROHIS-QOL tool was responsive to changes in clinical state and suggested that QOL completely recovered after successfully completing treatment. This contrasts with a study in which patients who had completed TB treatment had lower QOL than healthy controls [32]. Lung destruction due to TB is an important contributor to poor QOL, and pulmonary rehabilitation programmes have been shown to improve lung function and QOL in TB-affected people [33,34]. Thus, the EUROHIS-QOL may have a role in helping to identify who should be offered enhanced care aiming to reduce the risk of mortality.
Patients' spouses and parents, who were assumed to have a caregiving role, had significantly lower QOL than other contacts, independent of confounding differences in age, sex and clustering of data within households. Additionally, lower patient QOL was associated with lower QOL in contacts, which may be explained by socioeconomic risks shared by patients and their caregivers, but also the socioeconomic stress of caring for someone with TB. Furthermore, caregivers may "absorb" some of the stigma, emotional and financial costs of TB [35][36][37]. Stress during informal caregiving has been associated with increased all-cause mortality, possibly due to allostatic load and cellular ageing [38][39][40]. However, these studies were in the context of geriatrics, oncology and paediatrics and reviews from these specialities have shown that involving caregivers in patients' management plans, educational activities and psychosocial support can alleviate caregiver burden [41]. This may be feasible and appropriate for TB programmes.
A limitation of this study is that all QOL questionnaires including the EUROHIS-QOL are inevitably subjective, so prone to response bias including social desirability. This bias was reduced by ensuring privacy, confidentiality and data collection by researchers who were independent of the healthcare system. There is no gold standard method to objectively assess general QOL. However, good correlation of the EUROHIS-QOL responses with other objective determinants of QOL implies that response bias did not undermine our findings. A strength of our study was that we were able to disaggregate patient QOL scores depending on TB treatment scheme and outcome. However, this was limited by the reduced number of patients eligible for follow-up who completed the 6-month interview, partly due to the impossibility of assessing follow-up QOL for people who had died, and difficulty locating people who had incomplete TB treatment. The generalisability of our findings is supported by the diversity of our study setting including 15 peri-urban shantytowns and 17 urban communities; and supports previous studies of the psychosocial costs of TB in other continents and social contexts [10,16,42]. Another limitation is that we did not study preventative treatment, which could have influenced QOL. However, this could only have affected a tiny proportion of our study population because chemoprophylaxis was probably started after the recruitment interview when we assessed QOL (at the time the index patient commenced treatment); we only assessed QOL in people aged ⩾15 years, and in this age group in Peru, chemoprophylaxis was only recommended for people with HIV infection and 15-19-year-olds who were known to be tuberculin skin test-positive close contacts of patients with isoniazid-susceptible TB [43]; and even in this group, there is a very low uptake of chemoprophylaxis [24]. A further limitation is that treatment for MDR-TB in Peru took >2 years at the time of the study, so QOL after MDR-TB treatment could not be evaluated in this study. The number of control households was opportunistic, not determined by sample size calculations, but the significant differences between controls and other participant groups suggests that this sample was generally sufficient.
In conclusion, the brief EUROHIS-QOL eight-item questionnaire meaningfully assessed QOL in TB-affected people and may be used to assess general QOL associated with TB disease in future research and practice. We recommend research to evaluate the EUROHIS-QOL with TB-affected households in diverse settings and trials to assess whether integrating the EUROHIS-QOL as a part of TB care can improve the quality of holistic care for members of TB-affected households. Our data show the profound psychosocioeconomic burden of TB for patients and their caregivers, and that low patient QOL predicts adverse treatment outcomes, especially death. These findings support the use of the EUROHIS-QOL to identify and quantify the holistic needs of people living with TB, potentially guiding patient-centred care addressing these multidisciplinary needs.