In 2015 there were 2.8 million incident tuberculosis (TB) cases in India, comprising 27% of the 10.4 million global estimated TB cases [1, 2]. Reducing TB incidence will require a multi-pronged approach including efforts to decrease time to accessing care. Untreated, TB patients spread infection and increase their own mortality risk [3–9].

Although there have been several studies in India of predictors of delay in seeking care, results have been mixed and are likely setting-specific in such a large heterogeneous country [5, 10–16]. Previous studies have found that longer delay is associated with factors including income, substance abuse, location of first seeking care, and distance to health care facility among others [5, 10–16]. However other studies found no association with income and alcohol use [10, 11, 13, 16].

One potential component of delay is the disconnect between different healthcare systems. Patients can obtain healthcare through public and private clinics or a mix of both. The Revised National TB Control Programme (RNTCP) offers free TB diagnosis and treatment to patients in public clinics [17]. However, a recent study found that more than a quarter of prevalent TB patients are never evaluated at RNTCP clinics [18] and review of drug sales show that this number is likely even higher [1]. Patients often follow a circuitous route to diagnosis including self-medication and consulting private practitioners which often results in delayed diagnosis and treatment initiation [5, 11, 14–16]. A study of standardized patients found that only 21% of the cases were properly managed by private healthcare providers [17, 19]. More work is needed to understand factors associated with delay in the private and public sectors which include the time it takes patients to enter the health system and how long until they are started on TB treatment.

Previous studies have shown that hazardous alcohol use is prevalent in India [20–24]. The World Health Organization (WHO) estimated that 25% of the country’s men have consumed alcohol in the past year [24]. Other studies in India have found risky alcohol use in 40% of all males who consume alcohol [20, 22] and 52% of all male TB patients who consume alcohol [23]. Furthermore, in 2014, the WHO reported that alcohol use in India has increased in the last five years [25]. Alcohol use has been shown to be associated with poor TB treatment outcomes [23, 26] however, the degree of association between delays in seeking care and alcohol use has not been consistently demonstrated, possibly due to varying definitions of high alcohol use and regional differences in alcohol consumption [11–13, 27, 28]. Furthermore, while knowledge about TB has been identified as a reason for delay in qualitative studies [29, 30], to our knowledge, only one study in India has examined this association quantitatively [12]. We aimed to identify predictors of delay to treatment initiation at RNTCP clinics for new TB patients in Puducherry and Tamil Nadu to help inform public health interventions.

This study of newly diagnosed pulmonary TB patients in Puducherry and Tamil Nadu, India found that risky alcohol users and those with higher income levels were more likely to delay seeking care at RNTCP clinics. Identifying and addressing these predictors may improve the time to treatment initiation and reduce transmission.

Previous studies of the association between alcohol use and engagement in care in India report mixed results. Studies of new pulmonary TB patients in Wardha District in western India (n = 53) and Tamil Nadu in the south (n = 531) found that alcoholism (categorized into “never”, “sometimes”, and “regular” consumption, and habitual alcohol consumption respectively) was associated with delay [11, 27]. Other studies in Mandi District in northern India (n = 234) and Tamil Nadu (n = 601) found no association; the former categorized alcohol use as daily use, and the latter study did not differentiate between casual and risky drinkers [12, 13]. Importantly, these studies used several less-standardized measures of alcohol use which could have masked the association with delay. The AUDIT-C used in our study is a validated measure of risky alcohol use, providing a standardized definition that has been applied in a variety of settings [32, 33]. Regional differences in alcohol use also may affect its role in delaying access to care [28, 35].

Because 40.7% of TB patients were risky alcohol users in this cohort, these results demonstrate a need to target risky drinkers in Puducherry and Tamil Nadu to encourage earlier TB diagnosis. We found risky alcohol use was associated with being male, married, having lower income (but being employed), and having less education – all characteristics that have been found to be associated with alcohol use in other studies [20, 28, 35]. Understanding the characteristics of risky alcohol users in this population may help public health providers design interventions to achieve this goal.

In our study, higher income was associated with delay in seeking care. This is in contrast to other studies in India that found that lower income individuals were more likely to delay [12, 15]. This difference could be due to income definitions and analysis techniques. We defined income as household income and use multivariable analysis, while one study used per-capita income and also did not control for any confounders [15] Furthermore, a study in northern India that found low income was associated with delay used a dichotomized household income variable while our study classified income into four levels [12].

The association between higher income and delay could also be related to the type of clinic where a patient first sought care. Previous studies have shown that patients visiting a government facility first are likely to have longer “patient delay” (onset of symptoms to first seeking health care) while visiting a private facility first is associated with longer “health system delay” (time from first seeking care until diagnosis or treatment initiation) [5, 11–14, 34]. Because our study assessed total delay (patient and health system delays combined), we may not have been able to detect an association between location of first care and delay. Higher income participants were more likely than lower income participants to visit a private facility first, which could have led to longer health system delay thus increasing their total delay. The association between higher income and delay remained, however, when controlling for location of first care (i.e., even when controlling for whether a patient first went to a private clinic, higher income individuals had longer delay).

There could be other facets of the relationship between health-seeking behavior, delay, and income that our questionnaires did not capture – such as the number of facilities visited – which could contribute to longer delays. Higher income could also be associated with delay if these individuals do not suspect that they have TB or, because they perceive that TB is a disease of poverty [36, 37]. Higher income individuals may also delay seeking care at public facilities because they fear the stigma associated with TB which has been reported to affect health-seeking behavior [38–40]. The relationship between higher income, care at private clinics, stigma, and delay should be investigated further to understand the connections between these factors and how best to encourage timely TB diagnoses.

Knowledge of TB transmission in India was found to be 55.5% in a country-wide survey and 65% in Puducherry [41, 42]. Our study did not find a significant association between delay and knowledge about TB, defined by curability or mode of transmission. This finding was surprising, as in previous studies patients cited poor knowledge as a reason for delayed care [11, 29, 30]. Furthermore, improved TB knowledge seems to play a role in engagement in care as following an educational intervention in Delhi, the number of individuals self-reporting with TB symptoms to a treatment center increased [43]. It is possible that we did not find an association between knowledge and delay because we only assessed knowledge of transmission and cure; some studies used numerous questions to create a knowledge score, allowing for finer categorization of knowledge levels [3, 12]. It also could be that in this area, delayed care is dominated by other factors, notably alcohol use or that patients’ knowledge of TB does not affect engagement in care because people seek care for symptoms not because they suspect they have a certain disease. We found that risky alcohol use was associated with poor TB knowledge suggesting that educational programs may be warranted for this group.

Our study had the following limitations. It is difficult to correctly ascertain household income due to inaccurate recall, which could introduce bias into the results about income. Similarly, the use of cough duration to quantify delay hinges upon a subject’s ability to recall symptom duration and there is no standard threshold that indicates delay, which could lead to ascertainment bias. However because there are no other well-defined measures of delay, cough duration with a cutoff at the median or at one month has been used by numerous other studies [4, 5, 34]. The median cough duration for our sample was four weeks therefore defining delay by the median fulfilled both of the common practices. We attempted to minimize inaccurate recall by interviewing subjects within one week of treatment initiation. Furthermore, although we asked participants where they first sought care, we were not able to ascertain the time spent getting evaluated at other clinics. Ultimately, we were still able to quantify the clinically relevant delay between symptom onset and treatment initiation, because the study excludes any individuals who previously received ≥3 doses of treatment (including at another clinic).

Our analysis demonstrated that public health measures to hasten engagement in care in Puducherry and Tamil Nadu should focus on risky drinkers. As of now, RNTCP training materials recognize the need to screen TB patients for alcohol use and encourage abstaining; we suggest targeting risky alcohol users, including those in alcohol treatment programs to enhance TB screening [44]. Additionally, higher income individuals delayed seeking care and studies are needed to determine the reasons for this delay, as it is not solely due to seeking care in private clinics. Social programs may need to target higher income groups with education and/or stigma-reduction campaigns to improve their engagement in care.

As TB prevalence continues to rise in India, it is increasingly important to reduce the time that TB patients are infectious. In order to achieve this goal, the factors that contribute to delay in accessing care need to be thoroughly understood so that populations at risk for delay can be targeted. As stated by the WHO, a key component in eradicating TB worldwide will be the integration of TB control programs with other public health programs, such as for alcoholics in Southern India [2].