Impact of connecting tuberculosis directly observed therapy short-course with smoking cessation on health-related quality of life

A multi-centered prospective non-randomized controlled trial using quasi-experimental design was conducted in Malaysia. Using Transtheoretical Model approach [27], 120 eligible participants who were current smokers at the time of TB diagnosis were assigned to either of two treatment groups: the usual TB-DOTS plus SCI (SCIDOTS group) or the usual TB-DOTS alone (DOTS group).

The study design can be schematically represented as below:

In Figure 1 above, ~R signifies that there was no randomization of subjects into the treatment groups. Y_b represents the baseline HRQoL outcome measures (before the interventions were initiated), whereas Y_a implies the HRQoL outcomes at different periods during the interventions. X₁ and X₂ represent the integrated intervention and the usual TB care, respectively.

This study was conducted at five chest clinics in Malaysia. The selected clinics were considered as TB referral centers from which all diagnosis or confirmation of TB cases must come from among the population they serve. The research was reviewed and approved by the Medical Research Ethics Committee of the National Institutes of Health, Malaysia in April 2008.

The sample for the present study included all current cigarette smokers newly diagnosed with TB between June 2008 and April 2009. Patients were referred at the point of initiating DOTS regimen by attending physicians to TB DOTS providers who were purposely trained to provide tobacco cessation intervention in TB care settings [13]. Patients were recruited into the study upon fulfillment of eligibility criteria and written informed consent. Power analysis and sample size requirements in this study were predicated on the primary outcome measure, that is the overall TB treatment outcome based on comparison of TB cure rates between the two treatment groups. A statistical power of 80% to detect a moderate effect size (20% difference in cure rates) at a 95% confidence level were applied in the sample size determination. By WHO standard of care, DOTS has a target to achieve a 95% cure rate in previous reports; therefore, we assumed a cure rate of 95% and 75% in the intervention and the control groups, respectively. The minimum sample size was estimated as 48 per treatment arm with a 2-sided alpha level of 0.05. We assumed a dropout rate of 25%, giving rise to 60 participants to be recruited per each treatment arm.

The participants in both the SCIDOTS and the DOTS groups went to the clinics on daily-basis to receive TB DOTS for at least 6 months. Eleven sessions of individualized cognitive behavioral therapy (CBT) with or without nicotine replacement therapy (NRT) were offered to each participant in the SCIDOTS group, while the DOTS group received the same number of sessions of usual counseling for TB. The follow-up clinic appointments for SCI (beginning on the quit date) were as follows: weekly for the first month, fortnightly for the second and third month, and monthly from the fourth to the sixth month. HRQoL outcomes were measured by using EQ-5D and the effects of the novel strategy on these were determined by comparing the two groups.

A summary of the interventions provided and outcomes measurements for each of the two treatment groups is provided in Figure 2.

The EQ-5D questionnaire was used for measuring the quality of life of the study participants. The instrument was subjected to standardized linguistic validation processes for use in Malaysian population by EuroQoL Group, UK. Permission was obtained prior to use in this study. HRQoL outcomes were measured periodically (at baseline, at 3 months, and at 6 months of TB treatment) for all participants using the EQ-5D questionnaire, which consists of two sections: the EQ-5D self-descriptive assessment and the visual analogue scale (EQ-VAS) [28‐30]. The EQ-5D descriptive system comprises of five domains (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), each having three response choices: no problems, moderate problems, and severe problems. Participants were asked to indicate their own health state by ticking in the box against the most appropriate statement. Participants' health state classifications in the descriptive system were converted into composite utility scores using the UK social tariff [31, 32]. In addition, patients also self-rated their health by drawing a line to the EQ-VAS to represent their current health status. EQ-VAS rates the current health status from 0 to 100, where the endpoints are labeled as best imaginable health state (100) and worst imaginable health state (zero).

The data were analyzed by using SPSS version 16.0 (SPSS Inc., Chicago, IL) and the level of statistical significance was set a priori at p < 0.05. Homogeneity of baseline data was determined by using two-sample tests including Student's t-test for normally distributed continuous variables, and Pearson's chi-square (χ²) or Fisher's exact test for categorical variables. In the analysis of HRQoL, each domain of the EQ-5D self-descriptive assessment was categorized into "no problem", "moderate problem", and "severe problem" and the data were presented as frequencies and percentages. χ² or Fisher's exact tests were used to compare between the SCIDOTS and the DOTS groups for each specific domain. The EQ-5D health state classification for each subject was converted into an overall utility score [31, 32]. Means and standard deviations for EQ-5D utility indices and EQ-VAS scores were calculated and tabulated. Independent t-test was employed to examine the effects of group on HRQoL (measured using the EQ-5D utility and EQ-VAS scores) from baseline to the end of the 6-month follow-up. Furthermore, two-way repeated measure ANOVA was used to examine the main effect of group (intervention and control), time (baseline, 3 months, and 6 months), and group x time interaction on EQ-5D utility scores and EQ-VAS. Sphericity-assumed estimates were applied if Mauchly's test indicated no violation of assumptions; otherwise, Greenhouse-Geisser estimates were used.

Results related to tobacco cessation outcomes (i.e. quitting smoking) were previously published elsewhere [33]. Abstinence was biochemically validated using exhaled CO and saliva cotinine testing. The study recorded a linear trend over time on 7-day point prevalence abstinence (PPA) and continuous abstinence (CA) in the SCIDOTS group, whereas abstinence rates remained fairly constant in the usual care (DOTS) group. At the sixth month of follow-up, patients who received the integrated intervention had a significantly higher rate of success in quitting smoking when compared with those who received the usual TB care alone (78% vs. 9%, respectively; p < 0.001). Further details on smoking cessation and TB treatment outcomes could be found in the study published by Awaisu et al. (2011).

The key findings in relation to HRQoL are broadly classified under the two subheadings below.

Table 1 presents a comparison of self-description assessment between the intervention and the control groups in five domains. The Majority of the participants in both groups reported no problems with mobility and self-care for all time periods. The trends in changes observed over time in these two dimensions were also similar between the two treatment groups. Moreover, there were no significant differences between participants in the SCIDOTS and the DOTS groups in their self-rating of mobility and self-care at baseline, 3 months and 6 months (Table 1).

Conversely, substantial proportions of the TB patients believed that they had moderate to severe problems in relation to usual activities, pain/discomfort, and anxiety/depression at diagnosis. At baseline, 12 patients (26.1%) in the usual care group rated themselves as having moderate problems with usual activities compared with 22 patients (55.0%) in the integrated intervention group (p = 0.011). Interestingly, at six months, the proportion of patients in the SCIDOTS group who had no problems with this dimension was significantly higher than that of patients in the DOTS group (95.0% vs. 76.1%; p = 0.015).

Furthermore, about two-thirds of the study participants in both groups assessed themselves as having moderate to severe pain or discomfort during TB diagnosis (p = 0.234). This problem waned over time and at the end of 6 months of treatment, all patients (100%) in the integrated intervention group versus 87% in the usual care group admitted that they experienced no pain or discomfort (p = 0.028). A summary of these findings is presented in Table 1. Similarly, during diagnosis, about 52% of the TB patients in both arms considered themselves as moderately anxious or depressed. Having problems in this dimension progressively decreased by about two to four folds during each visit.

The means and standard deviations for EQ-5D utility indices and EQ-VAS scores were computed and summarized in Table 2. There was no significant difference at baseline between the SCIDOTS and the DOTS groups for mean EQ-5D utility score (0.72 ± 0.20 vs. 0.70 ± 0.31, t = 0.37, p = 0.715). The integrated intervention group had a greater, but statistically non-significant increase in EQ-5D utility index when compared with the usual care group at 3 months follow-up (mean = 0.91 ± 0.16 vs. 0.87 ± 0.22, t = 0.98, p = 0.333). However, at 6 months follow-up, the former had significantly greater increase in EQ-5D utility score than the latter (mean = 0.98 ± 0.08 vs. 0.91 ± 0.14, t = 2.85, p = 0.006). Similarly, the mean scores for EQ-VAS showed a closely similar trend as the EQ-5D utility indices, with the scores increasing over the course of TB treatment (Table 2).

Two-way repeated measure ANOVA was used to examine the main effect of group, time, and group x time interaction on EQ-5D utility score and on EQ-VAS score as shown in Table 3. Covariates were not used in the analyses despite significant baseline differences between the SCIDOTS and the DOTS groups in nicotine dependence [Fagerström test for nicotine dependence (FTND) score] and in expired CO level, as there was no significant correlation between the baseline FTND score and the baseline EQ-5D utility index (r = -0.11, p = 0.3); and between the baseline CO level and baseline EQ-5D utility index (r = -0.21, p = 0.06). Mauchly's test of sphericity showed that the assumption of sphericity was violated (χ² = 20.43, p < 0.001); thus, Greenhouse-Geisser estimates were applied. There were no significant main effects for group [F (1, 84) = 1.86, p = 0.176, η² = 0.02] and group x time interaction [F (2, 138) = 0.548, p = 0.545, η² = 0.006], but there was significant main effect for time [F (2, 138) = 49.42, p < 0.001, η² = 0.37].

Similarly, the main effects of group, time, and group x time interaction on EQ-VAS as shown in Table 3 were assessed. Assumption of sphericity was met as indicated by Mauchly's test of sphericity (χ² = 2.5, p = 0.287); thus, sphericity-assumed estimates were applied. There were significant main effects for group [F (1, 84) = 4.91, p = 0.029, η² = 0.06], time [F (2, 168) = 139.50, p < 0.001, η² = 0.62] and group x time interaction [F (2, 168) = 13.89, p < 0.001, η² = 0.14]. In addition, more in-depth analyses using pairwise comparison showed significant mean differences across all levels: level 1 vs. level 2 (∆mean = 16.03, p < 0.001); level 1 vs. level 3 (∆mean = 24.55, p < 0.001) and level 2 vs. level 3 (∆mean = 8.52, p < 0.001).