Abstract
Background
Poor adherence to tuberculosis (TB) treatment is associated with disease recurrence and death. Little research has been conducted in India to understand TB medication nonadherence.
Methods
We enrolled adult drug-susceptible TB patients, approximately half of whom were people with human immunodeficiency virus (PWH), in Chennai, Vellore, and Mumbai. We conducted a single unannounced home visit to administer a survey assessing reasons for nonadherence and collect a urine sample that was tested for isoniazid content. We described patient-reported reasons for nonadherence and identified factors associated with nonadherence (ie, negative urine test) using multivariable logistic regression. We also assessed the association between nonadherence and treatment outcomes.
Results
Of 650 participants in the cohort, 77 (11.8%) had a negative urine test. Nonadherence was independently associated with daily wage labor (adjusted odds ratio [aOR], 2.7; confidence interval [CI], 1.1–6.5; P = .03), the late continuation treatment phase (aOR, 2.0; CI, 1.1–3.9; P = .03), smear-positive pulmonary disease (aOR, 2.1; CI, 1.1–3.9; P = .03), alcohol use (aOR, 2.5; CI, 1.2–5.2; P = .01), and spending ≥30 minutes collecting medication refills (aOR, 6.6; CI, 1.5–29.5; P = .01). People with HIV reported greater barriers to collecting medications than non-PWH. Among 167 patients reporting missing doses, reported reasons included traveling from home, forgetting, feeling depressed, and running out of pills. The odds of unfavorable treatment outcomes were 4.0 (CI, 2.1–7.6) times higher among patients with nonadherence (P < .0001).
Conclusion
Addressing structural and psychosocial barriers will be critical to improve TB treatment adherence in India. Urine isoniazid testing may help identify nonadherent patients to facilitate early intervention during treatment.
Keywords: adherence, alcohol use, HIV, India, tuberculosis
We evaluated adherence to tuberculosis medications in 650 Indian patients by conducting urine isoniazid tests collected during unannounced home visits. Structural barriers to collecting medication refills and psychosocial barriers emerged as the most important factors contributing to medication nonadherence.
Poor adherence to active tuberculosis (TB) treatment is associated with increased disease recurrence and death [1–3]. A meta-analysis of recent clinical trials suggests that missing more than 10% of doses during drug-susceptible TB treatment is associated with approximately 6-fold increased risk of unfavorable outcomes, including disease recurrence [2]. Studies in programmatic conditions similarly reveal higher disease recurrence for TB patients with poor adherence [3]. A modeling study suggests that reducing nonadherence could have larger epidemiological impact on TB incidence than decreasing loss to follow-up during treatment in high-burden countries [4].
Despite its importance, there has been little research that has evaluated factors contributing to TB medication nonadherence in high-burden settings such as India, which has the world’s largest epidemic [5]. In the few previous studies in India, researchers have identified clinical (eg, medication adverse effects [6–8], symptom improvement [6]), psychosocial (eg, alcohol use [6, 7, 9], stigma [6, 7]), structural (eg, distance from clinic [6–10], migration [6], work-related challenges [6, 8, 10]), and health system (eg, noncooperative staff [6], drug stockouts [9]) barriers that contribute to nonadherence.
These previous studies have methodological limitations including small samples [6–8]. Some studies reportedly assessing adherence actually measured loss to follow up from treatment, rather than missed doses, which is a behavior with potentially different contributing factors and clinical implications [11–14]. None of these studies used rigorous approaches for measuring adherence, such as monitoring with digital adherence technologies (DATs) [15] or drug metabolite testing [16, 17]—strategies that have been used to understand adherence in human immunodeficiency virus (HIV) and other conditions [18].
Moreover, prior studies were conducted in the context of directly observed therapy (DOT), usually using a facility-based approach, in which patients visited clinics where healthcare providers observed dose ingestion. In recent years, concern has been growing regarding the patient and health system burden, ethical limitations, and effectiveness of DOT [15, 19–22]. At the same time, in 2014, India’s National TB Elimination Program (NTEP) began introducing daily medication dosing for all drug-susceptible TB patients, in place of a prior thrice-weekly dosing regimen [23]. This shift led to concerns about the increased burden of having patients visit clinics daily, rather than thrice weekly [23]. As a result, the NTEP has transitioned away from DOT and towards use of self-administered therapy ([SAT] ie, patients taking pills themselves) or monitoring with 99DOTS, a cellphone-based DAT [23]. No recent studies have evaluated causes of TB treatment nonadherence in India with SAT or monitoring with DATs.
In this manuscript, we investigate TB medication nonadherence using data from a cohort study conducted in 3 Indian cities among people with HIV (PWH) and non-PWH. Adherence was assessed by urine testing for isoniazid during unannounced home visits. Although the study’s primary goal, for which results have been reported [16], was to understand 99DOTS’ accuracy for measuring adherence, the data also provide insights into causes of nonadherence. In this study, our aims are as follows: to identify clinical, structural, and psychosocial factors associated with nonadherence; describe patient-reported reasons for missed doses collected using a structured questionnaire; and assess the association between nonadherence and TB treatment outcomes.
METHODS
Ethics Approvals
The study protocol was approved by ethics committees at the National Institute for Research in TB (Chennai, India), Brigham and Women’s Hospital (Boston, MA), and Tufts University (Boston, MA).
Patient Consent Statement
We obtained written informed consent from all participants enrolled in this study.
Study Setting and Care Delivery Approaches
To enroll a geographically diverse cohort, we recruited TB patients from 3 cities with a relatively high TB burden: Mumbai, Chennai, and Vellore [24, 25]. Almost all patients in Mumbai were HIV-negative and recruited from 11 DOT centers with some of the highest patient volumes. Patients in Chennai and Vellore were PWH recruited from the 5 largest HIV antiretroviral therapy (ART) centers in these cities. As such, in our analyses, HIV status not only serves as a marker of this comorbidity, but it also serves as a proxy for differences in geographic sites and care delivery approaches, as described below.
Tuberculosis care delivery approaches varied between DOT and ART centers. Before 2014, PWH collected TB medications at DOT centers near their homes, similar to non-PWH. With subsequent introduction of daily TB medication dosing using a SAT- or DAT-monitored approach, PWH started to collect TB medication refills at ART centers concurrently with ART refills (in a “single window”), usually on a monthly basis [26]. Although DOT centers are decentralized in primary health centers in India, there are fewer ART centers that are usually located in tertiary hospitals. As such, the single window approach increased convenience by consolidating HIV and TB pharmacy services; however, it usually increased the distances PWH needed to travel to collect TB medications.
Patient Recruitment and Data Collection
From August 2017 to February 2019, we sequentially recruited adult patients (18 years of age and older) from the selected clinics who were taking drug-susceptible TB therapy monitored with 99DOTS. To understand variability in adherence during treatment, we initially enrolled patients visiting clinics not only when they started treatment but also when they collected monthly medication refills, excluding the last treatment month. After exhausting the pool of patients already taking treatment, we continued to enroll those initiating treatment but randomly assigned participants to receive the home visit in the intensive (first 2 months) or the continuation (last 4 months) phase to ensure distribution of home visits across phases.
Informed consent was obtained upon enrollment. Based on a socioecological framework [27], a baseline questionnaire was designed to collect data on demographic, socioeconomic, structural, and psychosocial factors—including the brief alcohol use disorder identification test (AUDIT-C) [28]. Patients became eligible for a home visit 3 weeks after enrollment or after reaching the continuation phase, for those randomized to receive a home visit in that phase. A random number generator was used to select the exact day of the visit.
We tried to minimize changes in adherence behavior in response to study interactions (Hawthorne effect). We conducted a single home visit with every patient under the assumption that patients may modify their behavior with repeated visits. Although this single visit limits understanding of adherence longitudinally for each patient, home visits for the cohort were distributed throughout the treatment course, providing reasonable representation of adherence for the population. Waiting at least 3 weeks for the home visit likely reduced the impact of short-term behavioral changes related to the initial study enrollment interaction. The home visit was also conducted without prior notice to minimize the likelihood of patients taking medications solely in anticipation of the visit. During the visit, we asked patients questions modified from the AIDS Clinical Trials Group adherence questionnaire to assess reasons for nonadherence and collected a urine sample that was tested using the IsoScreen assay [29].
Interpretation of the Urine Test Result
IsoScreen is a validated assay based on the Arkansas (Potts-Cozart) method. Assay reagents change color through reactions with isoniazid or its metabolite acetylisoniazid [30–32]. Yellow, green, and blue/purple test results suggest no detection, low-level detection, and high-level detection, respectively [30, 31]. These 3 color categories can be used to construct 2 different binary adherence outcomes, which we will refer to as “nonadherence” and “suboptimal adherence.”
When interpreting IsoScreen results by comparing any color change (green or blue/purple) versus no color change (yellow), validation studies show that approximately 100%, 83%, and 11% of patients will have positive test results 24, 48, and 72 hours, respectively, after last ingestion of isoniazid [30, 31]. We defined nonadherence as comprising yellow test results (in comparison to green or blue/purple results), indicating high likelihood of not having taken doses for 48 to 72 hours or longer. This approach has <11% chance of misclassifying patients who did not take a dose within the last 72 hours as being adherent.
Previous validation studies also showed that, whereas blue/purple color indicates medication ingestion within the last 24 hours, green indicates the last dose ingestion occurred 24 to 48 hours earlier, suggesting a dose was missed [30, 31]. Therefore, we defined suboptimal adherence as comprising yellow or green test results (in comparison to blue/purple results), indicating high likelihood of not having taken a dose for 24 hours or longer. By including green color as comprising a negative result, we increased the test’s sensitivity for detecting poor adherence, although this approach may misclassify a small proportion of patients (<17%) who are correctly taking their medications as being suboptimally adherent [31]. Given the strengths and deficiencies of each approach to classifying urine test results, and the fact that each approach provides slightly different insights into adherence behavior, we conducted separate analyses using each adherence outcome.
Analyses
We used JMP Pro 15 (SAS Institute, Inc., Cary, NC) to perform univariable and multivariable logistic regression analyses to identify clinical, structural, and psychosocial factors associated with nonadherence and, separately, suboptimal adherence. All demographic and clinical factors were retained in the multivariable model by design. Other factors were retained if significant at P < .2 in the univariable analyses. Modifications to the model to remediate multicollinearity are described in the Supplementary Appendix. We classified the timing of the home visit based on whether it happened in the intensive phase (≤56 days of therapy), early continuation phase (57 to 112 days of therapy), or late continuation phase (>112 days of therapy). We included alcohol use as a binary variable comparing those with moderate or high use (AUDIT-C score of 1 or greater) to no alcohol use (score of 0). For additional insights into nonadherence, we described the proportion of patients reporting various reasons for missing medication doses in the home visit questionnaire.
Finally, we evaluated whether the urine test result was associated with treatment outcomes. We used separate logistic regression analyses to assess whether nonadherence and suboptimal adherence were associated with outcomes of death, loss to follow up, and “unfavorable treatment outcomes”—the latter being a composite outcome including death, loss to follow up, and treatment failure. We did not separately assess the association between nonadherence and treatment failure, because only 3 patients experienced treatment failure.
We obtained outcomes from patients’ treatment cards through study closure in February 2019 and later from Nikshay (the NTEP’s electronic record) in September 2020 (see Supplementary Appendix for details). We analyzed the association between nonadherence or suboptimal adherence and treatment outcomes separately using each data source.
RESULTS
Descriptive Statistics
Of 832 patients meeting eligibility criteria, 84 (10%) did not consent for the study or could not enroll because family members collected their medications. Of 748 patients who enrolled, we could not complete home visits for 98 (13%), despite 3 attempts (Supplementary Appendix, Figure S1). Of 650 patients in the final analysis, 77 (11.8%) were nonadherent and 116 (17.8%) were suboptimally adherent. Among PWH, 51 of 303 (16.8%) were nonadherent compared with 26 of 347 (7.5%) HIV-negative TB patients (P = .0002). Among PWH, 73 of 303 (24.1%) were suboptimally adherent compared with 43 of 347 (12.4%) HIV-negative TB patients (P < .0001).
More than three quarters of patients had a household monthly income of <15 000 Indian rupees ([INR] approximately <200 US dollars [USD]) (Table 1). Approximately half were housewives, students, or unemployed. Most patients were new (formerly category 1) and had smear-positive pulmonary TB. Few were able to walk or bicycle to clinic; most spent 60 minutes or more collecting medication refills.
Table 1.
Descriptive Characteristics for the Overall Cohort and Disaggregated by HIV Status
| Covariates | Overall Cohorta (N = 650) | People With HIVa (N = 303) | HIV Negativea (N = 347) | P Valueb |
|---|---|---|---|---|
| Demographic Factors | ||||
| Gender | ||||
| Female | 271 (41.7) | 102 (33.7) | 169 (48.7) | <.001 |
| Male | 379 (58.3) | 201 (66.3) | 178 (51.3) | |
| Age | ||||
| 18–29 | 226 (34.8) | 30 (9.9) | 196 (56.5) | <.001 |
| 30–44 | 251 (38.6) | 154 (50.8) | 97 (28.0) | |
| ≥45 | 173 (26.6) | 119 (39.3) | 54 (15.6) | |
| Monthly Income | ||||
| INR <7500 | 244 (37.5) | 174 (57.4) | 70 (20.2) | <.001 |
| INR 7500–14 999 | 263 (40.5) | 93 (30.7) | 170 (49.0) | |
| INR ≥15 000 | 143 (22.0) | 36 (11.9) | 107 (30.8) | |
| Occupation | ||||
| Self-employed | 137 (21.1) | 81 (26.7) | 56 (16.1) | <.001 |
| Government or private sector employment | 130 (20.0) | 61 (20.1) | 69 (19.9) | |
| Laborer on daily wages | 84 (12.9) | 46 (15.2) | 38 (11.0) | |
| Housewife, student, or unemployed | 299 (46.0) | 115 (38.0) | 184 (53.0) | |
| Clinical Factors | ||||
| Phase of Therapy | ||||
| Intensive phase | 178 (27.4) | 102 (33.7) | 76 (21.9) | .002 |
| Early continuation phase | 249 (38.3) | 100 (33.0) | 149 (42.9) | |
| Late continuation phase | 223 (34.3) | 101 (33.3) | 122 (35.2) | |
| Category of TB | ||||
| New | 504 (77.5) | 242 (79.9) | 262 (75.5) | .184 |
| Previously treated | 146 (22.5) | 61 (20.1) | 85 (24.5) | |
| Type of TB | ||||
| Extrapulmonary | 202 (31.1) | 72 (23.8) | 130 (37.5) | <.001 |
| Smear-negative pulmonary | 77 (11.9) | 48 (15.8) | 29 (8.36) | |
| Smear-positive pulmonary | 371 (57.1) | 183 (60.4) | 188 (54.2) | |
| Structural Factors | ||||
| Transport Mode to Clinic | ||||
| Walking or bicycle | 234 (36.0) | 12 (4.0) | 222 (64.0) | <.001 |
| Motorcycle or car | 44 (6.8) | 30 (9.9) | 14 (4.0) | |
| Autorickshaw or taxi | 131 (20.2) | 33 (10.9) | 98 (28.2) | |
| Public transportation | 241 (37.1) | 228 (75.3) | 13 (3.8) | |
| Money Spent to Collect Medication Refills | ||||
| INR 0–24 | 233 (35.9) | 21 (6.9) | 212 (61.1) | <.001 |
| INR 25–49 | 111 (17.1) | 32 (10.6) | 79 (22.8) | |
| INR-50–75 | 111 (17.1) | 68 (22.4) | 43 (12.4) | |
| INR >75 | 195 (30.0) | 182 (60.1) | 13 (3.8) | |
| Time Spent to Collect Medication Refills | ||||
| <30 minutes | 110 (16.9) | 4 (1.3) | 106 (30.6) | <.001c |
| 30 to 59 minutes | 191 (29.4) | 7 (2.3) | 184 (53.0) | |
| 60 to 239 minutes | 180 (27.7) | 123 (40.6) | 57 (16.4) | |
| ≥240 minutes | 169 (26.0) | 169 (55.8) | 0 (0.0) | |
| Psychosocial Factors | ||||
| Current Tobacco Use | ||||
| No | 540 (83.1) | 261 (86.1) | 279 (80.4) | .002 |
| Smokeless tobacco only | 51 (7.9) | 12 (4.0) | 39 (11.2) | |
| Cigarette or beedi use | 59 (9.1) | 30 (9.9) | 29 (8.4) | |
| Probable Alcohol Use | ||||
| No alcohol use | 591 (90.9) | 256 (84.5) | 335 (96.5) | <.001 |
| Any alcohol use | 59 (9.1) | 47 (15.1) | 12 (3.5) |
Abbreviations: HIV, human immunodeficiency virus; INR, Indian rupees; TB, tuberculosis.
aRepresents the number of study participants in a category divided by the overall sample or subsample: eg, there are 271 females out of 650participants in the overall cohort; 102 females out of 303 participants with HIV; and 169 females out of 347 participants who are HIV negative.
bχ 2 was used to assess differences in characteristics between people with HIV and HIV-negative TB patients.
cFisher’s exact test was used to assess differences for time spent to collect medication refills, because some categories had fewer than 5 observations.
Patient characteristics varied significantly by HIV status. It is notable that PWH were substantially more likely to be older, have lower income, and use alcohol. People with HIV also faced substantially greater structural barriers to collecting medication refills, with regard to transport mode, money spent, and time spent on this activity.
Factors Associated With Nonadherence and Suboptimal Adherence
In the multivariable model with nonadherence as the outcome, participants who were laborers on daily wages, in the late continuation treatment phase, smear-positive pulmonary TB patients, spending 30 or more minutes collecting medications, or using alcohol were at significantly higher odds of being nonadherent (Table 2). In the multivariable model with suboptimal adherence as the outcome, participants who were laborers on daily wages, spending 30 or more minutes collecting medications, or using alcohol were at significantly higher odds of being suboptimally adherent (Supplementary Appendix, Table S2). Participants in the late continuation phase were also at higher odds of being suboptimally adherent, although this association did not quite achieve statistical significance (P = .06).
Table 2.
Factors Associated With Nonadherence to TB Medications (N = 650)
| Descriptive Statistics | Univariable Findings | Multivariable Findings | |||
|---|---|---|---|---|---|
| Covariates | Proportion With Nonadherencea, n (%) | Odds Ratio (95% Confidence Interval) | P Value | Odds Ratio (95% Confidence Interval) | P Value |
| Demographic Factors | |||||
| Gender | |||||
| Female | 28 (10.3) | Ref | Ref | ||
| Male | 49 (12.9) | 1.3 (0.8–2.1) | .31 | 1.0 (0.5–1.8) | .92 |
| Age | |||||
| 18–29 | 19 (8.4) | Ref | Ref | ||
| 30–44 | 41 (16.3) | 2.1 (1.2–3.8) | .01* | 1.1 (0.6–2.2) | .71 |
| ≥45 | 17 (9.8) | 1.2 (0.6–2.4) | .62 | 0.6 (0.3–1.2) | .15 |
| Monthly Income | |||||
| INR <7500 | 33 (13.5) | Ref | Ref | ||
| INR 7500–14 999 | 33 (12.5) | 0.9 (0.5–1.5) | .74 | 1.4 (0.8–2.7) | .24 |
| INR ≥15 000 | 11 (7.7) | 0.5 (0.3–1.1) | .08 | 0.9 (0.4–2.0) | .74 |
| Occupation | |||||
| Self-employed | 12 (8.8) | Ref | Ref | ||
| Employed in government or private sector | 16 (12.3) | 1.5 (0.7–3.2) | .35 | 1.7 (0.7–3.9) | .23 |
| Laborer on daily wages | 16 (19.0) | 2.5 (1.1–5.5) | .03* | 2.7 (1.1–6.5) | .03* |
| Housewife, student, or unemployed | 33 (11.0) | 1.3 (0.6–2.6) | .47 | 1.6 (0.7–3.6) | .26 |
| Clinical Factors | |||||
| Phase of therapy | |||||
| Intensive phase | 20 (11.2) | Ref | Ref | ||
| Early continuation phase | 21 (8.4) | 0.7 (0.4–1.4) | .33 | 1.1 (0.5–2.1) | .87 |
| Late continuation phase | 36 (16.1) | 1.5 (0.8–2.7) | .16 | 2.0 (1.1–3.9) | .03* |
| Category of TB | |||||
| New | 55 (10.9) | Ref | Ref | ||
| Previously treated | 22 (15.1) | 1.4 (0.9–2.5) | .17 | 1.4 (0.8–2.5) | .24 |
| Type of TB | |||||
| Extrapulmonary | 15 (7.4) | Ref | Ref | ||
| Smear-negative pulmonary | 11 (14.3) | 2.1 (0.9–4.8) | .08 | 1.9 (0.8–4.7) | .15 |
| Smear-positive pulmonary | 51 (13.7) | 2.0 (1.1–3.6) | .03* | 2.1 (1.1–3.9) | .03* |
| People With HIV | |||||
| No | 26 (7.5) | Ref | Ref | ||
| Yes | 51 (16.8) | 2.5 (1.5–4.1) | .0003* | 1.5 (0.6–3.6) | .43 |
| Structural Factors | |||||
| Transport Mode to Clinic | |||||
| Walking or bicycle | 13 (5.6) | Ref | |||
| Motorcycle or car | 5 (11.4) | 2.2 (0.7–6.5) | .16 | ||
| Autorickshaw or taxi | 14 (10.7) | 2.0 (0.9–4.5) | .08 | ||
| Public transportation | 45 (18.7) | 3.9 (2.0–7.4) | <.0001* | ||
| Money Spent to Collect Medication Refills | |||||
| INR 0–24 | 17 (7.3) | Ref | |||
| INR 25–49 | 10 (9.0) | 1.3 (0.6–2.8) | .58 | ||
| INR 50–75 | 17 (15.3) | 2.3 (1.1–4.7) | .02* | ||
| INR >75 | 33 (16.9) | 2.6 (1.4–4.8) | .003* | ||
| Time Spent to Collect Medication Refills | |||||
| <30 minutes | 2 (1.8) | Ref | Ref | ||
| 30 to 59 minutes | 19 (9.9) | 6.0 (1.4–26.1) | .02* | 6.6 (1.5–29.5) | .01* |
| ≥60 minutes | 56 (16.0) | 10.3 (2.5–43.0) | .001* | 9.0 (1.8–44.2) | .007* |
| Psychosocial Factors | |||||
| Current Tobacco Use | |||||
| No | 59 (10.9) | Ref | |||
| Smokeless tobacco only | 4 (7.8) | 0.7 (0.2–2.0) | .50 | ||
| Cigarette or beedi use | 14 (23.7) | 2.5 (1.3–4.9) | .006* | ||
| Probable Alcohol Use | |||||
| No alcohol use | 61 (10.3) | Ref | Ref | ||
| Any alcohol use | 16 (27.1) | 3.2 (1.7–6.1) | .0003* | 2.5 (1.2–5.2) | .01* |
Abbreviations: HIV, human immunodeficiency virus; INR, Indian rupees; Ref, reference group; TB, tuberculosis.
aRepresents the number of study participants with nonadherence in a given category: eg, 28 of 271 females were nonadherent.
*Indicates a statistically significant finding at the 5% level.
Patient-Reported Reasons for Nonadherence
Of 650 participants who answered the home visit survey, 167 (25.7%) reported missing at least 1 medication dose during therapy. Of these, more than one fifth reported traveling or being away from home, forgetting to take medications, feeling depressed, or running out of pills as contributing to nonadherence (Table 3). The Supplementary Appendix provides details on why participants ran out of pills and medication adverse effects that impacted adherence.
Table 3.
Patient-Reported Reasons for Missing Tuberculosis (TB) Medication Doses (N = 167)
| Proportion of Patients Reporting This Problemb | Frequency With Which Patients Were Affected by This Problem | |||
|---|---|---|---|---|
| Reason for Missing Dosesa | n (%) | Rarely n (%) | Sometimes, n (%) | Often, n (%) |
| Traveling or being away from home | 67 (40.1%) | 37 (22.2%) | 19 (11.4%) | 11 (6.6%) |
| Simply forgot | 50 (29.9%) | 28 (16.8%) | 11 (6.6%) | 11 (6.6%) |
| Felt depressed | 39 (23.3%) | 16 (9.6%) | 16 (9.6%) | 7 (4.2%) |
| Ran out of pills | 35 (21.0%) | 29 (17.4%) | 3 (1.8%) | 3 (1.8%) |
| Wanted to avoid medication adverse effects | 29 (17.4%) | 15 (9.0%) | 8 (4.8%) | 6 (3.6%) |
| Reduced motivation because TB symptoms improved | 19 (11.4%) | 7 (4.2%) | 5 (3.0%) | 7 (4.2%) |
| Had too many pills to take for different conditions | 17 (10.2%) | 9 (5.4%) | 4 (2.4%) | 4 (2.4%) |
| Did not want others to notice me taking medication | 16 (9.6%) | 12 (7.2%) | 2 (1.2%) | 2 (1.2%) |
aParticipants could report more than 1 reason for missing doses.
bRepresents the proportion of study participants reporting a given problem over the total number in the cohort who reported having ever missed medication doses: eg, 67 of 167 participants reported traveling or being away from home as a reason for missing doses.
Association Between Medication Nonadherence and Tuberculosis Treatment Outcomes
Using Nikshay data, nonadherence had a statistically significant association with loss to follow up and an association with death that did not quite achieve statistical significance (Table 4). Nonadherent participants had 4.0 (95% confidence interval, 2.1–7.6) higher odds of unfavorable treatment outcomes (P < .0001). Analyses using suboptimal adherence as the outcome, and using data from treatment cards alone, yielded similar findings (Supplementary Appendix, Tables S3–S5).
Table 4.
Association Between Medication Nonadherence and TB Treatment Outcomes Reported in India’s National TB Elimination Program Nikshay Electronic Record System for the Cohort (N = 565)a
| Descriptive Statistics | Univariable Findings | |||
|---|---|---|---|---|
| Treatment Outcome | Proportion of Sample in Given Categoryb, n (%) | Proportion With Medication Nonadherencec, n (%) | Odds Ratio (95% Confidence Interval) | P Value |
| Treatment success (cure or treatment completion) | 513 (90.8) | 48 (9.4) | Ref | |
| Died | 18 (3.2) | 4 (22.2) | 2.8 (0.9–8.7) | .08 |
| Lost to follow up | 34 (6.0) | 11 (32.4) | 4.6 (2.1–10.1) | .0001* |
Abbreviations: Ref, reference group; TB, tuberculosis.
aSample excludes 85 (13.1%) study participants, including 77 for whom the study closed before they finished treatment and outcomes were also not reported in Nikshay, 5 who underwent a change in treatment regimen, and 3 who experienced treatment failure.
bRepresents the number of participants in a category divided by the overall cohort sample with available treatment outcomes of 565: eg, 513 of 565 participants experienced treatment success.
cRepresents the number of participants with nonadherence in a given category: eg, 48 of 513 participants with treatment success were nonadherent.
*Indicates a statistically significant finding at the 5% level.
DISCUSSION
In this cohort study, we used survey questions and urine metabolite testing, a rigorous marker of isoniazid ingestion, to identify factors associated with nonadherence to drug-susceptible TB treatment in India. We found that medication nonadherence is a complex problem that involves structural, psychosocial, and clinical barriers that may be amenable to intervention.
As found with other TB care cascade gaps, structural barriers—particularly distance from the clinic [33, 34]—may be a central challenge leading to nonadherence. Nonadherence increased considerably for participants who spent 30 minutes or more collecting medication refills—a factor that was collinear with other structural barriers (ie, transportation mode and money spent collecting medications). These structural barriers triangulate with the survey finding that more than one fifth of participants reported running out of pills as a reason for missing doses. Similar to findings in the HIV literature, many patients attributed pill shortages to transportation difficulties or other challenges collecting medications [35].
People with HIV disproportionately experienced these structural barriers, likely related to the single window policy in which PWH now collect TB medications concurrently with their HIV medications at ART centers instead of DOT centers. Although “one-stop shopping” seems to be a laudable goal, creation of new barriers through longer commutes to ART centers may attenuate the benefits of consolidated pharmacy services. Our findings may explain results of a study in Karnataka, India, in which outcomes were worse for PWH receiving TB treatment at ART centers via the single window system, compared with PWH receiving treatment at DOT centers [26].
Interventions addressing structural barriers may include providing travel vouchers to help patients reach clinics or delivering medications to patients. For example, recent studies have evaluated the benefits of HIV adherence clubs, in which PWH meet locally to discuss their care and collect ART, facilitating easy access to medications and peer support [36]. Such interventions merit evaluation in Indian TB patients.
Laborers who receive daily wages had higher nonadherence. This finding remained significant after adjusting for income, which suggests that this risk may be related to wage losses when people leave work to collect medications and other job constraints, rather than poverty in general. Home medication delivery may benefit patients with such jobs. Many patients reported traveling away from home as a reason for missing doses, which may also reflect job-related constraints, a barrier also reported in the HIV literature [37]. Future research should examine strategies to support medication access and adherence during travel.
Psychosocial factors also emerged as important barriers to adherence. Alcohol use was significantly associated with nonadherence, consistent with findings from previous Indian studies [38–40]. Although we did not formally measure depression, more than one fifth of patients who reported missing doses described feeling depressed as a reason. In a recent systematic review, researchers found that depression is associated with increased loss to follow up and death—but not medication nonadherence—during TB treatment; however, the studies that were included had suboptimal measures of adherence [41]. Future studies using rigorous measures of both depression and adherence, such as the urine testing we used here, may clarify whether nonadherence mediates the association between depression and treatment outcomes. Some patients reported missing doses because they did not want others to notice them taking medications, which highlights stigma as another barrier [42, 43].
Clinical factors may also contribute to nonadherence. Smear-positive pulmonary TB patients had higher adjusted odds of nonadherence. These patients have poorer outcomes on average compared with smear-negative and extrapulmonary TB patients in the NTEP [44]. Our finding suggest that poorer treatment outcomes in smear-positive patients could be mediated not only by greater disease severity but also by higher nonadherence. A desire to avoid medication adverse effects was also reported as a reason for missing doses, consistent with previous studies in India [45]. Finally, nonadherence was higher in the late continuation phase of therapy, a finding that was consistent with some patients describing symptom improvement as a reason for missing doses. Enhanced counseling may help to address patient concerns about adverse effects while also increasing motivation to adhere later in the treatment, when symptoms have improved.
Many patients said they simply “forgot” to take medications. Although forgetfulness is often perceived as a cognitive problem, it is also shaped by other barriers. Depression and alcohol use can reduce cognitive function, increasing the chance that patients will forget to take pills, whereas structural barriers to collecting medications may result in patients prioritizing other tasks. One rationale for using DATs is that reminders provided by these technologies might reduce forgetfulness; however, these reminders do not address the contextual barriers that contribute to forgetting medication doses.
Despite only conducting a single urine isoniazid test for each patient, a negative test result was strongly associated with unfavorable treatment outcomes. This finding highlights possible benefits of urine testing not only for research, but also for identifying patients at risk for poor outcomes in routine care. Conducting multiple urine tests throughout therapy might facilitate better prediction of treatment outcomes. However, further research is needed to (1) understand whether urine testing might be acceptable to patients in routine care, (2) evaluate whether tests conducted during prescheduled clinic visits are as accurate as tests conducted during unannounced home visits [18], and (3) ensure that testing does not have the unintended consequence of leading healthcare providers to stigmatize nonadherent patients.
A study limitation is that our cohort may not be representative of the overall TB patient population at these clinics, because 53 of 832 (6%) patients did not consent for the home visit and 31 of 832 (4%) could not enroll because family members collected their medications. We were also unable to complete home visits for 98 of 748 (13%) enrolled participants. We suspect these participants may have been more likely to be nonadherent and to have suffered unfavorable treatment outcomes. As such, adherence and treatment outcomes may be higher in our study cohort, because participants who were successfully enrolled and available for a home visit may have been more motivated to adhere and complete treatment. In addition, because many participants in our cohort had their home visit relatively late in treatment (ie, the continuation phase), the predictive value of the urine test may have been reduced, because these participants were far along the pathway to their treatment outcome.
Our understanding of the influence of HIV infection on TB medication adherence was also limited, because HIV status also served as a proxy for differences in geographic sites and care delivery approaches. Although inclusion of PWH and HIV-negative patients facilitated unique insights, such as the importance of time spent collecting medications, combining these populations may have masked meaningful associations with factors that could be more relevant to one of these groups.
A strength of our study is that we present analyses using nonadherence and suboptimal adherence as outcomes, because these outcomes represent 2 practical ways the urine test could be interpreted in clinical practice. Although the general consistency of findings across these preplanned analyses is reassuring, conducting multiple analyses with several predictors and similar outcome measures may increase risk of type I error—that is, that we incorrectly identified significant associations. In addition, we had a modest number of outcomes (ie, patients with negative urine tests), which may have limited statistical power to identify factors associated with nonadherence. Future studies involving a larger sample, with multiple urine tests on each patient, might identify additional factors associated with nonadherence.
CONCLUSIONS
In this study of drug-susceptible TB patients in India, we found that nonadherence is a complex problem shaped by structural, psychosocial, and clinical barriers. Structural barriers—in particular, transportation challenges and time and money spent collecting medications—had the strongest association with nonadherence, particularly for PWH. Psychosocial barriers such as alcohol use, depression, and stigma also emerged as major problems contributing to nonadherence.
Our findings highlight a need to facilitate easier access to medication refills and to develop counseling interventions that can address depression, substance use disorders, and stigma. Drug metabolite testing is a useful approach for measuring adherence in TB patients taking SAT- or DAT-monitored therapy, particularly because nonadherence may predict subsequent outcomes in the TB care cascade [46], including unfavorable treatment outcomes and disease recurrence. Future research should focus on assessing whether urine testing can be used to improve adherence in routine care and on evaluating interventions addressing the diverse barriers contributing to TB medication nonadherence in high-burden countries.
Supplementary Data
Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Acknowledgments
We thank all of the participants who took part in this study. We acknowledge the immense contribution of all the field investigators for their efforts in collecting data. The Mumbai team included Yogesh Nivrutti Salve, Kamble Rakesh Suresh, Shweta Shantaram Bagade, and Meena Atul Kamble. The Tamil Nadu team included C. Jeganathan, S. Yokeshwaran, K. Sathiyamoorthy, E. Michael Raj, P. Brindadevi, K. Jegan, B. Sathyan Raj Kumar, and Mariyamma Paul.
Financial support. This work was primarily funded by grants from the Bill & Melinda Gates Foundation (Grant Numbers OPP1154670 [to B. E. T.] and OPP1154665 [to R. S.]). This work was also funded by a Harvard Catalyst KL2/Catalyst Medical Investigator Training Award (Grant Number KL2 TR001100; to R. S.), the Harvard Center for AIDS Research (Grant Number 5P30AI060354-13; to R. S.), and a Doris Duke Clinical Scientist Development Award (Grant Number 2018095; to R. S.). J. E. H. was supported by a grant from the National Institute of Mental Health (Grant Number K24MH114732).
Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.
References
- 1.Waitt CJ, Squire SB. A systematic review of risk factors for death in adults during and after tuberculosis treatment. Int J Tuberc Lung Dis 2011; 15:871–85. [DOI] [PubMed] [Google Scholar]
- 2.Imperial MZ, Nahid P, Phillips PPJ, et al. . A patient-level pooled analysis of treatment-shortening regimens for drug-susceptible pulmonary tuberculosis. Nat Med 2018; 24:1708–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Thomas A, Gopi PG, Santha T, et al. . Predictors of relapse among pulmonary tuberculosis patients treated in a DOTS programme in South India. Int J Tuberc Lung Dis 2005; 9:556–61. [PubMed] [Google Scholar]
- 4.Arinaminpathy N, Chin DP, Sachdeva KS, et al. . Modelling the potential impact of adherence technologies on tuberculosis in India. Int J Tuberc Lung Dis 2020; 24:526–33. [DOI] [PubMed] [Google Scholar]
- 5.Subbaraman R, Jhaveri T, Nathavitharana RR. Closing gaps in the tuberculosis care cascade: an action-oriented research agenda. J Clin Tuberc Other Mycobact Dis 2020; 19:100144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Zaman FA, Sheikh S, Das KC, et al. . An epidemiological study of newly diagnosed sputum positive tuberculosis patients in Dhubri district, Assam, India and the factors influencing their compliance to treatment. J Nat Sci Biol Med 2014; 5:415–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Shameer M, Manjunath K, Marconi S, et al. . A community based case control study on risk factors for treatment interruptions in people with tuberculosis in Kollam district, Kerala, southern India. Int J Community Med Public Health 2016; 3:962–7. [Google Scholar]
- 8.Kulkarni P, Akarte S, Mankeshwar R, et al. . Non-adherence of new pulmonary tuberculosis patients to anti-tuberculosis treatment. Ann Med Health Sci Res 2013; 3:67–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Bagchi S, Ambe G, Sathiakumar N. Determinants of poor adherence to anti-tuberculosis treatment in Mumbai, India. Int J Prev Med 2010; 1:223–32. [PMC free article] [PubMed] [Google Scholar]
- 10.Gopi PG, Vasantha M, Muniyandi M, et al. . Risk factors for non-adherence to directly observed treatment (DOT) in a rural tuberculosis unit, South India. Indian J Tuberc 2007; 54:66–70. [PubMed] [Google Scholar]
- 11.Sarpal SS, Goel NK, Kumar D, Janmeja AK. Reasons for interruption of anti-tubercular treatment among the retreatment patients in category II of RNTCP in Chandigarh, north India. Indian J Tuberc 2014; 61:121–8. [PubMed] [Google Scholar]
- 12.Masthi NR, Rajanna MS, Parasuramalu BG. A study on the effectiveness of DOTS on tuberculosis patients treated under RNTCP. Indian J Public Health 2006; 50:55–7. [PubMed] [Google Scholar]
- 13.Jaggarajamma K, Sudha G, Chandrasekaran V, et al. . Reasons for non-compliance among patients treated under Revised National Tuberculosis Control Programme (RNTCP), Tiruvallur district, south India. Indian J Tuberc 2007; 54:130–5. [PubMed] [Google Scholar]
- 14.Gupta S, Gupta S, Behera D. Reasons for interruption of anti-tubercular treatment as reported by patients with tuberculosis admitted in a tertiary care institute. Indian J Tuberc 2011; 58:11–7. [PubMed] [Google Scholar]
- 15.Subbaraman R, de Mondesert L, Musiimenta A, et al. . Digital adherence technologies for the management of tuberculosis therapy: mapping the landscape and research priorities. BMJ Glob Health 2018; 3:e001018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Thomas BE, Kumar JV, Chiranjeevi M, et al. . Evaluation of the accuracy of 99DOTS, a novel cellphone-based strategy for monitoring adherence to tuberculosis medications: comparison of DigitalAdherence data with urine isoniazid testing. Clin Infect Dis 2020; 71:e513–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kendall EA, Durovni B, Martinson NA, et al. ; THRio and TEKO teams . Adherence to tuberculosis preventive therapy measured by urine metabolite testing among people with HIV. AIDS 2020; 34:63–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Musinguzi N, Muwonge T, Thomas K, et al. . Does adherence change when no one is looking? Comparing announced and unannounced tenofovir levels in a PrEP trial. AIDS Behav 2016; 20:2639–43. [DOI] [PubMed] [Google Scholar]
- 19.Yellappa V, Lefèvre P, Battaglioli T, et al. . Coping with tuberculosis and directly observed treatment: a qualitative study among patients from South India. BMC Health Serv Res 2016; 16:283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Sagbakken M, Frich JC, Bjune GA, Porter JD. Ethical aspects of directly observed treatment for tuberculosis: a cross-cultural comparison. BMC Med Ethics 2013; 14:25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wynne A, Richter S, Banura L, Kipp W. Challenges in tuberculosis care in Western Uganda: health care worker and patient perspectives. IJANS 2014; 1:6–10. [Google Scholar]
- 22.Karumbi J, Garner P. Directly observed therapy for treating tuberculosis. Cochrane Database Syst Rev 2015; 5:CD003343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cross A, Gupta N, Liu B, et al. . 99DOTS: a low-cost approach to monitoring and improving medication adherence. In: Proceedings of the Tenth International Conference on Information and Communication Technologies and Development. New York, NY: Association for Computing Machinery, 2019: pp 1–12. [Google Scholar]
- 24.Dhanaraj B, Papanna MK, Adinarayanan S, et al. . Prevalence and risk factors for adult pulmonary tuberculosis in a metropolitan city of South India. PLoS One 2015; 10:e0124260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Central TB Division. TB India 2018: Revised National TB Control Programme Annual Status Report. Available at: https://tbcindia.gov.in/showfile.php?lid=3314. Accessed 20 July 2020.
- 26.Thekkur P, Kumar AN, Chinnakali P, et al. . Outcomes and implementation challenges of using daily treatment regimens with an innovative adherence support tool among HIV-infected tuberculosis patients in Karnataka, India: a mixed-methods study. Glob Health Action 2019; 12:1568826. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bronfenbrenner U. Ecological systems theory. Ann Child Dev 1989; 6:187–249. [Google Scholar]
- 28.Nayak MB, Bond JC, Cherpitel C, et al. . Detecting alcohol-related problems in developing countries: a comparison of 2 screening measures in India. Alcohol Clin Exp Res 2009; 33:2057–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.ACTG Adherence Baseline Questionnaire. Available at: https://www.ghdonline.org/uploads/ACTG_Adherence_Baseline_Questionnaire.pdf. Accessed 2 November 2020.
- 30.Elizaga J, Friedland JS. Monitoring compliance with antituberculous treatment by detection of isoniazid in urine. Lancet 1997; 350:1225–6. [DOI] [PubMed] [Google Scholar]
- 31.Soobratty MR, Whitfield R, Subramaniam K, et al. . Point-of-care urine test for assessing adherence to isoniazid treatment for tuberculosis. Eur Respir J 2014; 43:1519–22. [DOI] [PubMed] [Google Scholar]
- 32.Schraufnagel DE, Stoner R, Whiting E, et al. . Testing for isoniazid. An evaluation of the Arkansas method. Chest 1990; 98:314–6. [DOI] [PubMed] [Google Scholar]
- 33.Thomas BE, Subbaraman R, Sellappan S, et al. . Pretreatment loss to follow-up of tuberculosis patients in Chennai, India: a cohort study with implications for health systems strengthening. BMC Infect Dis 2018; 18:142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Subbaraman R, Thomas BE, Sellappan S, et al. . Tuberculosis patients in an Indian mega-city: where do they live and where are they diagnosed? PLoS One 2017; 12:e0183240. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Tuller DM, Bangsberg DR, Senkungu J, et al. . Transportation costs impede sustained adherence and access to HAART in a clinic population in southwestern Uganda: a qualitative study. AIDS Behav 2010; 14: 778–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Fox MP, Pascoe S, Huber AN, et al. . Adherence clubs and decentralized medication delivery to support patient retention and sustained viral suppression in care: Results from a cluster-randomized evaluation of differentiated ART delivery models in South Africa. PLoS Med 2019; 16:e1002874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Taylor BS, Reyes E, Levine EA, et al. . Patterns of geographic mobility predict barriers to engagement in HIV care and antiretroviral treatment adherence. AIDS Patient Care STDS 2014; 28:284–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Thomas BE, Thiruvengadam K, S R, et al. ; CTRIUMPH-RePORT India Study . Smoking, alcohol use disorder and tuberculosis treatment outcomes: a dual co-morbidity burden that cannot be ignored. PLoS One 2019; 14: e0220507. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Zhou TJ, Lakshminarayanan S, Sarkar S, et al. . Predictors of loss to follow-up among men with tuberculosis in puducherry and Tamil Nadu, India. Am J Trop Med Hyg 2020; 103:1050–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Thomas BE, Suresh C, Lavanya J, et al. . Understanding pretreatment loss to follow-up of tuberculosis patients: an explanatory qualitative study in Chennai, India. BMJ Glob Health 2020; 5:e001974. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ruiz-Grosso P, Cachay R, de la Flor A, et al. . Association between tuberculosis and depression on negative outcomes of tuberculosis treatment: a systematic review and meta-analysis. PLoS One 2020; 15:e0227472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Thomas BE, Kumar JV, Onongaya C, et al. . Explaining differences in the acceptability of 99DOTS, a cell phone-based strategy for monitoring adherence to tuberculosis medications: qualitative study of patients and health care providers. JMIR Mhealth Uhealth 2020; 8:e16634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Thomas BE, Kumar JV, Periyasamy M, et al. . Evaluation of the acceptability of the medical event reminder monitor for promoting adherence to multidrug-resistant tuberculosis therapy in two indian cities: qualitative study of patients and healthcare providers. J Med Internet Res 2021. In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Subbaraman R, Nathavitharana RR, Satyanarayana S, et al. . The tuberculosis cascade of care in India’s public sector: a systematic review and meta-analysis. PLoS Med 2016; 13:e1002149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Phukte G, Patil S, Jain Y. Adverse drug reactions during intensive phase of first line anti-tuberculosis treatment: a prospective cohort study from rural central India. Int J Tub Lung Dis 2019; 23:S318. [Google Scholar]
- 46.Subbaraman R, Nathavitharana RR, Mayer KH, et al. . Constructing care cascades for active tuberculosis: a strategy for program monitoring and identifying gaps in quality of care. PLoS Med 2019; 16:e1002754. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.