PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • A detailed description of the study design and methodology has been published previously (Kastien-Hilka et al 2016). Briefly, the study followed an observational longitudinal design including prospective, repeated measures of HRQOL per study participant. Patient population and participant recruitment: Study participants were recruited between November 2014 and May 2015 at six selected primary health care clinics with the highest TB caseloads per month in Cape Town. Four of these facilities are run by the local government, and two by the provincial government, and are located in the Khayelitsha sub-district of the Cape Town Metro district. The sub-district has the highest TB burden in Cape Town, based on latest available caseloads from 2012 provided through the Western Cape Province. The study population comprised of patients diagnosed with active pulmonary TB. TB patients who were 18 years or older and were not diagnosed with TB before were eligible. Patients were excluded if they were diagnosed with multidrug-resistant TB (MDR-TB) and extensively drug-resistant TB (XDR-TB), and/or had HIV co-infection. Patients underwent HIV testing when diagnosed with TB before starting treatment. The TB nurses of each study site provided to the study team access to the HIV test results during the recruitment process to ensure a negative HIV status. HIV testing was not repeated during the treatment and any change in the HIV status was therefore unknown. The eligibility status of each patient was subject to verification by the nurse dedicated to TB patients of each health facility. Based on a patient information document, eligible patients were informed about the nature, purpose, potential risks and benefits of the study. Patients had the opportunity to decline their participation or to withdraw from the study at any time point. Patients who agreed to participate in the study signed a written informed consent. Recruitment of new TB cases without HIV co-infection was difficult as a high proportion of TB patients are also HIV co-infected in South Africa. In order to meet the sample size in the pre-determined time frame, one additional clinic in a neighboring district was included in the study. Eligible participants received a six month standard TB treatment with rifampicin, isoniazid, ethambutol and pyrazinamide. Standard treatment comprised the intensive treatment phase over the first two months where all four antibiotic drugs were applied, and the continuous treatment phase with isoniazid and rifampicin only over at least additional 4 month. This standard treatment was not influenced by the study design. HRQOL was evaluated during the treatment course. The data collection regimen to monitor changes in HRQOL included five different time points over the six-month treatment period: beginning of treatment (baseline) and at follow-up visits after 4, 8, 16 weeks and after six-month treatment. The last data collection point at six months was selected as this time point represents the minimum treatment time for standard TB treatment. Data collection at treatment start and after six-months of treatment allowed the identification of HRQOL impairment caused by TB and any changes in HRQOL due to treatment. Since standard treatment comprised two treatment phases, data were collected at the switch between both phases (after 8 weeks) as well as in the middle of each treatment phase (after 4 weeks and 16 weeks). Data collection started when TB patients visited the study sites for treatment initiation; four HRQOL measures and one socio-demographic questionnaire were applied at baseline (treatment start). These four HRQOL measures were re-applied at all follow up visits. All HRQOL measures were applied in English language and have been validated for English for South Africa. No other languages or translations of these measures were applied. Data were collected based on completion of paper questionnaires during face-to-face interviews conducted by trained field workers. Field workers were trained and quality of data was ensured based on a Standard Operating Procedure, and details were published previously. Questionnaires were scanned by the field worker for any missing responses during the interviews. The rationale for the selection of HRQOL measures has been described previously (Kastien-Hilka et al 2016). Two generic (European Quality of Life 5 Dimensions 5 levels; EQ-5D-5L and Short-Form 12 items; SF-12), one disease-specific (St. George´s Respiratory Questionnaire; SGRQ) and one condition-specific (Hospital Anxiety and Depression Scale; HADS) HRQOL measures were used. All HRQOL measures have been validated in TB populations: SF-12 [18], EQ-5D-5L [37], SGRQ [38], and HADS [39]. EQ-5D-5L [40] is widely used as a utility index for estimating QALYs in cost-effectiveness studies [40, 41]. It comprises five items/domains (5D) (Mobility, Self Care, Usual Activities, Pain/Discomfort, and Anxiety/Depression) with each domain having five levels (5L): no problems, slight problems, moderate problems, severe problems, and extreme problems; The EQ-5D-5L includes further one vertical visual analogue scale (VAS 20 cm). Index-based values (utilities) are calculated from EQ-5D-5L by applying country-specific valuation algorithms. No South African specific valuation algorithms were available. Therefore, algorithms developed for the UK and for Zimbabwe, the only African country with an available algorithm, were used in this study. The EQ-5D-5L utilities range from 0 to 1, with higher scores indicating better health. A minimally important difference (MID) is only known for the 3 level version of EQ-5D, with a MID of 0.074 (range -0.011–0.140) and a MID of 7 for VAS scores [42]. EQ-5D-5L has an improved sensitivity compared to the 3 level version and its MID is also assumed for the 5 level version. Short-Form 12 (SF-12) is an abbreviated version of SF-36 containing 12 items representing eight domains (physical functioning, role physical, bodily pain, general health, vitality, social functioning, role emotional, and mental health) [41, 43]. Domains are aggregated into composite summary scores, Physical Component Score (PCS-12) and Mental Component Sore (MCS-12). Scoring ranges from 0 to 100 with greater scores representing better HRQOL. A score between 47 and 53 reflects normal scores for both PCS and MCS based on US population norms [44, 45]. No population norms for South Africa or other African countries are available. A MID of at least 3 points has been suggested for the SF-36 [44] and can also be used for SF-12v2 [46]. St. George´s Respiratory Questionnaire (SGRQ) is a disease-specific instrument designed to assess patients with respiratory tract and immune system diseases, especially asthma, pulmonary diseases, and chronic obstructive disease [41, 47]. SGRQ comprises 50 items in three domains (Symptoms, Activity, and Impacts on daily life). Scores are scaled from 0 to 100, with higher scores indicating worse HRQOL. A MID for SGRQ is defined as an improvement of 4 points in the domain scores and the total score [48]. Hospital Anxiety and Depression Scale: Hospital Anxiety and Depression Scale (HADS) is a condition-specific instrument applied in psychology and psychiatry to detect states of anxiety and depression [49]. HADS comprises 14 items in two domains, Anxiety domain and Depression domain. The scores of each subscale range from 0–21 (8–10 mild, 11–14 moderate, 15–21 severe). A MID for HADS is not available from the developer. A MID of 1.5 points has been estimated for chronic obstructive pulmonary disease (COPD), corresponding to a change from baseline of 20% and informed by both anchor- and distribution-based methods [50]. The primary endpoint was defined as change in mean score of PCS-12 between baseline and six-month treatment. As SF-12 and SF-36 are comparable measures, sample size determination was based on the change in mean score of PCS-36 as reported by Walter [51]. We assumed a 4.0 point change in PCS-12 mean score (higher than the MID of 3 points) between baseline and visit 4 (six-months treatment) [45, 52], and a corresponding standard deviation (SD) of 7.0 for the mean score after six-months treatment [45], resulting in a standardized effect size of 0.57. Using the standardized effect size with a two-sided 5% level of significance and a 95% power yielded in an estimated sample size of n = 80 participants. We estimated an attrition rate of 20% resulting in a final sample size of n = 96 participants. The study adhered to the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) guidelines, the Declaration of Helsinki and South African Good Clinical Practice (GCP). The study was approved by the institutional review commission of the Swiss Tropical and Public Health Institute in Basel, Switzerland. Ethical approval was obtained from the ethical commission of North-West and Central Switzerland (EKNZ) and the ethics committee of the University of Cape Town. City Health of the City of Cape Town and the Western Cape Government of South Africa gave institutional approval of the study. Missing interviews were recorded as the absence of the study participant at the appointed interview time frame. Missing data were recorded when the interviewed participant did not complete HRQOL measures, or only insufficiently, hindering calculation of a domain or total score. Significance testing (paired-samples t-test), multivariable analysis and repeated measures ANOVA were based on observations that contained all data points required for a specific analysis. All HRQOL responses were transferred to an excel-based database and all measures were scored according to each measure´s instructions. EQ-5D-5L utility scores were calculation based on valuation algorithms for the UK and Zimbabwe, applying the EQ-5D calculator [40]. SF-12 component scores PCS-12 and MCS-12 as well as SGRQ domain and total score were calculated applying specific scoring software provided through the measure´s developer. HADS was scored based on an excel worksheet according to the instructions in the HADS manual [53]. Interpretation of EQ-5D-5L domain scores and HADS Anxiety and Depression domains were both based on categories: EQ-5D-5L on its five levels (no problems, slight problems, moderate problems, severe problems, and extreme problems) and both HADS domains on three levels (mild, moderate, severe). The SF-12 component scores and EQ-5D VAS were both interpreted based on a range from 0 (worst health) to 100 (best health), the EQ-5D-5L utility index on a range from 0 (worst health) to 1 (best health). The SGRQ domains and total score were interpreted based on a range between 0 (best health) to 100 (worst health) and compared to scores derived from a population with no history of respiratory disease [47]. Descriptive statistics were applied to socio-demographic data; further to all HRQOL data at baseline and at all follow-up visits to understand the HRQOL impairment of TB patients. Descriptive statistics were conducted including frequencies (N, N missing, %), central tendency (mean, median), and confidence interval (set at 95%). Distribution of data was examined by standard deviation (SD), minimum and maximum values, and frequency plots. Overall changes in HRQOL between baseline and six-month treatment (visit 4) were calculated as frequencies (%). Longitudinal changes were determined by the change in mean scores between all follow-up visits and baseline. The change in mean scores was examined by paired-samples t-test with a statistical significance (2-tailed) set a priori at P < 0.05. Changes in mean scores in the intensive treatment phase (baseline to visit 2) were compared to changes in the continuous treatment phase (visit 2 to visit 4) based on paired-samples t-test with a statistical significance (2-tailed) set a priori at P < 0.05. The change in mean scores at each time point from baseline was also compared to the reported minimally important difference (MID) for each measure to understand if the longitudinal changes in HRQOL were clinically meaningful. Paired-samples t-test was applied to examine the difference between changes in mean scores and MID at a significance level of P< 0.05. Differences in HRQOL mean scores among all HRQOL measures over time (baseline and follow-up visit 1–4) were examined by repeated measures ANOVA. Bonferroni correction was applied to repeated measures ANOVA for multiplicity of tests. Responsiveness over time for each HRQOL was measured as an effect size partial eta squared providing information of the effect of time on changes in HRQOL aspects. The time effects were based on tests of within-subjects effects. When Mauchly´s test of sphericity was not met (significance < 0.05), both partial eta squared and observed power were derived from the Greenhouse-Geisser correction. The impact of socio-demographic factors including age, gender, educational status and work status were elaborated. At baseline univariable analysis was applied to understand which factors might be associated with HRQOL, by using a cut off of P = 0.2. Resulting candidate factors were further assessed in multivariable models to understand the impact of socio-demographic factors over time (change from baseline to six month treatment (visit 4)). The univariable and multivariable analysis included a general linear model and an analysis of variance (ANOVA). The time effect of socio-demographic factors based on an effect size partial eta squared was further included by applying repeated measures ANOVA. Threshold values for the effect size were derived from Cohen [54]: 0.1 was interpreted as small, 0.3 as medium, 0.5 as large and 0.8 as very large. Robustness of the findings was assessed by sensitivity analyses by excluding HRQOL data from visit 3. Further, the assessment of longitudinal changes of HRQOL over time including baseline and all follow-up data as described above was repeated by excluding data from visit 3, which was affected by a very low number of available observations. Results were checked for consistency.
rdf:type