Are global tuberculosis control targets overlooking an essential indicator? Prolonged delays to diagnosis despite high case detection rates in Yunnan, China

Abstract

Delay in treating active tuberculosis (TB) impedes disease control by allowing ongoing transmission, and may explain the unexpectedly modest declines in global TB incidence. Even though China has achieved TB control targets under the global Directly Observed Treatment, Short course (DOTS) strategy, TB prevalence in western provinces, including Yunnan, is not decreasing. This cross-sectional study investigates whether prolonged delay in identifying and correctly treating TB patients, which is not routinely monitored, persists even when there is a well-functioning TB control programme and global targets are being met. Records of adult smear-positive pulmonary TB patients diagnosed with between 2006 and 2013 were extracted from the Yunnan Centre for Disease Control electronic database, which contains information on the entire population of TB patients managed across 129 diagnostic centres. Delay was investigated at three stages: delay to DOTS facility (period between symptom onset and first visit to at a CDC unit providing standardized treatment); delay to TB confirmation (period between reaching a CDC unit and confirmation of smear-positive TB) and delay to treatment (period between confirmation of TB and initiation of treatment). Data from 76 486 patients was analysed. Delay to reaching a DOTS facility was by far the largest contributor to total delay to treatment initiation. The median delay to reaching a DOTS facility, to TB confirmation and to treatment was 57 days (IQR 25–112), 2 days (IQR 1–6) and 1 day (IQR 0–1) respectively. Prolonged delays to reaching a facility providing standardized TB care occurred in a substantial subset of the population despite all TB control targets being met; overall, 32% (24 676) of patients experienced a delay of more than 90 days to reaching a DOTS facility. Policies that focus on reducing delays in accessing appropriate health services, rather than only on increasing overall case-detection rates, may result in greater progress towards reducing TB incidence.

Tuberculosis, China, policy

Key Messages

• Delay to accessing diagnosis and treatment is not routinely assessed by tuberculosis (TB) control programmes, and this could explain the unexpectedly small decrease in TB incidence among several countries that are meeting global targets.

• Our study, which is the largest in any country, showed that a long delay (median 57 days) occurs between symptom onset and first contact with appropriate health services in Yunnan province, China, despite all TB control targets being met.

• Programmatic targets that consider delays to treatment initiation, rather than focusing only on increasing the overall proportion of cases detected, may result in accelerated progress on TB control.

Introduction

There was a surge in political commitment to control tuberculosis (TB) in 1993, when World Health Organization (WHO) declared TB a global emergency. Directly observed treatment, short course (DOTS) was subsequently launched as the key strategy for global TB control, TB-related indicators were included in the Millennium Development Goals and, more recently, the Global Plan to Stop TB 2006–15 was developed and implemented (World Health Organisation and Stop TB Partnership 2010). As part of DOTS strategy, countries work towards two key targets: to detect and cure at least 70 and 85% of new smear-positive TB patients respectively (Dye et al. 1998). These figures were based on epidemiological models that predicted that meeting these targets would result in lower transmission and incidence of TB, which is the ultimate goal of TB control programmes (Lönnroth et al. 2010).

Although attribution of success is difficult, the medium-term goal, linked to Target 8 of Millennium Development Goal 6, ‘to have halted and begun to reverse the incidence of TB’ by 2015, has been achieved. However, although the number of new cases occurring every year is falling, the estimated rate of global decline is <1% per annum, far less than required to meet the long-term TB elimination goal of reducing incidence to under one case per million by 2050 (Kritzinger et al. 2009). Reports from several high-burden countries indicate that transmission and incidence of TB is not decreasing as expected despite countries having active DOTS programmes and meeting global case-detection and treatment success targets (Whalen 2006; Dye et al. 2008).

China is a prime example. The country has ∼ 1 million new cases of TB every year, more than any country apart from India (WPRO 2016). Between 2000 and 2005, substantial political and financial commitments were made to strengthen China’s TB control programme (Wang et al. 2007). In terms of TB control targets, China is performing well, with 100% reported DOTS coverage since 2005 indicating that the whole population has access to a health facility providing free standardized TB diagnosis and treatment (Wang et al. 2007). Although some surveys indicate that overall prevalence across the country has fallen in recent decades, western China has not experienced a decrease in prevalence (Wang et al., 2014), and survey data indicates that case detection measures have been insufficient in preventing transmission of infection (Hill and Whalen 2015).

Mounting evidence that TB transmission and incidence is not decreasing as expected raises critical questions about whether the global TB control strategy has focused on the appropriate programmatic targets and indicators to achieve disease reduction goals. For case detection, the target centres on the proportion of new cases diagnosed and initiated on treatment. Delay between symptom onset and diagnosis of new patients is rarely monitored as part of the evaluation of TB control strategies. In HIV-negative, smear-positive individuals the duration of infectiousness has been estimated at ∼ 2 years before diagnosis, self-cure or death (Corbett et al. 2004) but once treatment is initiated, patients become non-infectious within weeks (Brindle et al. 1993). The time that it takes for an infectious patient to start treatment and subsequently become non-infectious is therefore also important in reducing transmission than the proportion of cases that are successfully diagnosed and treated. For example, a country diagnosing 70% of TB cases within 60 days of having active disease will likely have lower transmission rates than a country diagnosing 70% of TB cases >90 days after developing active disease (all else being equal) (Golub et al. 2006).

Delay to diagnosis and treatment initiation of infectious pulmonary TB is, therefore, an important indicator of the functioning of TB control programmes as it is directly related to ongoing transmission, disease progression and secondary cases (Lin et al. 2008; Cheng et al. 2013). However, it is an indicator on which systematic, large scale information is rarely available as it is not prioritized as a TB control target. To address this gap in evidence, we comprehensively analyse delays in diagnosis of smear-positive TB patients across all public (DOTS) diagnostic centres across Yunnan province, China. Although over 70 and 90% of smear-positive TB cases have been successfully diagnosed and treated respectively since 2005 in Yunnan, a study conducted in 2005 reported substantial delays from the onset of symptoms to TB treatment, with one-third of patients delaying >90 days before seeking care at a DOTS-implementing health centre for their TB symptoms (Storla et al. 2008).

We hypothesize that implementation of the DOTS strategy in Yunnan, China, while effective in achieving case-detection targets, may not have been effective in reducing delays to diagnosis, thereby limiting impact on the number of new cases infected. Our study addresses the following critical question: do delays in initiating TB treatment—which may allow ongoing transmission of infection—persist despite full implementation of the DOTS strategy, a well-functioning TB control programme and success in meeting case detection targets?

Materials and methods

Study setting

To ensure standardized diagnosis and treatment, all government hospitals or smaller public health facilities in China are refer patients with symptoms consistent with TB or diagnosed with it to the local Centre for Disease Control and Prevention (CDC) TB centre for further evaluation and treatment. Private health facilities are also required to refer patients suspected of having TB to the CDC, but information how closely private healthcare providers comply with referral diagnostic and treatment guidelines is not available. Yunnan province, which lies in southwest China and has a largely rural population of 44 million, is covered by 129 county-level CDC TB centres. In 2004, an electronic Tuberculosis Information Management System (TBIMS) was introduced to allow doctors at CDC units to register and update information on TB diagnosis and treatment in real time (Huang et al. 2014). The TBIMS therefore receives standardized reports on all new TB cases being treated and monitored according to national guidelines (Lin et al. 2008; Zhao et al. 2013). County-level TB doctors register and update TB patient information every day based on standard operational guidelines. The data recorded by doctors includes demographic information, clinical details, date of onset of TB symptoms, date of first contact with county (or higher) TB centre/hospital, date of TB diagnosis, and date treatment is initiated.

Data management and definitions

We retrieved raw data from the Yunnan CDC TBIMS database, containing information on the entire population of TB patients managed in the public (DOTS-affiliated) system from 2005 to 2013—a total of 211 112 records—with patient identifiers removed. The database was translated from Chinese to English and was checked for any duplicate patient entries, which were removed.

Electronic records of patients meeting our inclusion criteria—those with newly diagnosed (first treatment) smear-positive pulmonary TB and aged 15 or above—were extracted from the database and imported into Stata version 11. For the analysis, we excluded patients diagnosed in 2005 as this was the first year that doctors were learning to use the database and some variables had missing or inconsistent entries. Individual patients’ records meeting the following exclusion criteria were also excluded from the study database: missing data on date of symptom onset or diagnosis; date of symptom onset/diagnosis/treatment earlier than or equal to the date of birth; date of symptom onset later than date of diagnosis or treatment initiation; date of diagnosis later than date of treatment initiation.

As there is no consensus definition on how to define the start and end of the delay period (Storla et al., 2008), we based our approach on relevance for disease control policies through discussions with the Yunnan CDC. Since standardized free diagnosis and treatment under DOTS can only be provided through CDC units, we defined total delay as the period between self-reported appearance of TB symptoms and treatment initiation through a CDC TB centre.

‘Delay to DOTS facility’ was defined as the period between self-reported appearance of TB symptoms and first contact with a doctor in a CDC TB centre. ‘Delay to TB confirmation’ was defined as the period between a patient’s first contact with a doctor at a CDC TB centre and laboratory confirmed diagnosis of smear-positive TB. ‘Delay to treatment’ was defined as the period between confirmed diagnosis of smear-positive TB and treatment initiation. Based on consultations with the Yunnan CDC on what would be considered prolonged or extreme delay in a programmatic context, and consistency with previous studies, the following cut-offs were used to define ‘prolonged delay’: a delay to DOTS facility of >90 days, a delay to TB confirmation of >30 days and a delay to treatment of >15 days (Lin et al. 2008).

Analyses

A descriptive cross-sectional analysis was conducted to investigate the duration of delays, divided into three categories along the continuum of care as defined above, among TB patients in Yunnan. We calculated the median delays along with interquartile ranges (IQRs) and examined annual changes in median delays between 2006 and 2013. We also compared the annual proportion of patients experiencing ‘prolonged’ delays, as defined earlier, between 2006 and 2013.

Based on the relative contribution of each category of delay, as indicated by our cross-sectional analysis, we further analysed delay to reaching a DOTS facility to elucidate factors that could be addressed by policymakers and programme planners, using a mixed methods approach. First, we investigated the quantitative association of prolonged delay to reaching a DOTS facility with five key sociodemographic characteristics recorded in the TBIMS: sex, age, occupation, residential status and ethnicity. Based on information on the patients’ primary occupation recorded in the TBIMS database, we grouped occupation into seven categories: farmers, urban workers (employed in factories etc.), students, retired/resigned (previously employed), housework/unemployed (only work inside the home without pay) or others (which included occupation groups with the lowest numbers of patients: caterers, child-care workers, fishermen, medical personnel, seafarers, long-distance drivers and teachers). Residency status was classified in the TBIMS database as local or non-local (address indicates residence within Yunnan but outside of diagnosing CDC centre catchment area, or outside of Yunnan province). We summarized the proportion of patients with prolonged delay in each age, gender, occupation, residence type and ethnic group, and calculated adjusted odds ratios (ORs) to measure the magnitude of the association of each characteristic with prolonged delay independent of all other sociodemographic characteristics. Statistical significance was assessed by the CI and P value associated with each OR.

We also identified factors that may be contributing to the delay in reaching DOTS facilities through a deductive thematic analysis of 21 in-depth interviews with health professionals working in the main TB hospitals in Kunming city and CDC units in two rural counties in Yunnan province, and 26 in-depth interviews with TB patients (17 men and nine women). These interviews were conducted between August 2014 and May 2015 as part of a larger study investigating barriers to accessing TB and multidrug resistant TB care in Yunnan; full details of the methodology can be found in the forthcoming publication (Hutchison et al. 2017).

Results

We assessed records from 80 551 eligible patients diagnosed between 2006 and 2013. Of these, 4065 (5%) had missing or incorrect values for dates of symptom onset, diagnosis or treatment initiation recorded in the TBIMS and were excluded. Records from 76 486 newly diagnosed smear-positive pulmonary TB patients were included in the cross-sectional analysis (Figure 1). As summarized in Table 1, we found that delay to a DOTS facility was by far the largest contributor to total delay between symptom onset and treatment initiation. The median delay to DOTS facility, delay to TB confirmation and delay to treatment was 57 days (IQR 25–112), 2 days (IQR 1–6) and 1 day (IQR 0–1), respectively. Our analysis of changes in delay to DOTS facility between 2006 and 2013 indicates that there was a decrease in between 2008 and 2010. However, reductions in delay to DOTS facility appear to plateau after 2010, remaining between 46 and 49 days between 2010 and 2013 despite 100% DOTS implementation since 2004. There was no change in median delay to TB confirmation and delay to treatment over the study period.

Table 1.

Median patient, health system and treatment delays in Yunnan province, China between 2006 and 2013

Registration year	Smear-positive TB patients	Median patient delay (IQR)	Median health system delay (IQR)	Median treatment delay (IQR)
2006	9101	65 (31–139)	2 (1–11)	1 (0–2)
2007	10 232	64 (31–139)	2 (1–10)	1 (0–2)
2008	11 153	64 (31–130)	2 (1–10)	1 (0–1)
2009	10 809	54 (23–106)	1 (0–5)	1 (0–1)
2010	11 194	49 (20–101)	2 (0–6)	1 (0–1)
2011	11 323	47 (20–95)	2 (0–6)	1 (0–1)
2012	7602	46 (21–98)	2 (0–7)	1 (0–1)
2013	5072	48 (21–95)	2 (0–7)	0 (0–1)
Total	76 486	57 (25–112)	2 (1–6)	1 (0–1)

Figure 1.

Inclusion and exclusion of patient records for the analysis

Our results indicate that prolonged delays to reaching a DOTS facility, >90 days between symptom onset and accessing a CDC TB diagnostic centre, occurred in a substantial proportion of the population despite implementation of the DOTS strategy across the entire province; 32% (24 676) and 15% (11 636) of patients had a delay to DOTS facility of >90 and 180 days respectively. A reduction in the proportion of TB patients experiencing prolonged delay to DOTS facility from 38 to 31% was observed between 2006 and 2009; however, by 2013, 27% of patients were still encountering prolonged delays (Table 2). Overall 6162 (8%) of patients experienced a prolonged delay to TB confirmation of >30 days; no reduction in prolonged delay to TB confirmation was observed since 2009 (Table 2). Prolonged delay to treatment (>15 days after confirmed diagnosis) was only experienced by 870 (1%) of newly diagnosed patients over the study period; after confirmation of a smear-positive TB diagnosis, treatment was initiated within 3 days for 96% of patients, indicating that treatment initiation was prompt.

Table 2.

Proportion of patients experiencing ‘prolonged’ patient, health system and treatment delays in Yunnan province, China between 2006 and 2013

Registration year	Smear-positive TB patients	Patient delay > 90 days (%)	Health system delay > 30 days (%)	Treatment delay > 10 days (%)
2006	9101	3476 (38)	691 (8)	174 (2)
2007	10 232	3860 (38)	1013 (10)	194 (2)
2008	11 153	4151 (37)	1093 (10)	204 (2)
2009	10 809	3353 (31)	718 (7)	121 (1)
2010	11 194	3255 (29)	818 (7)	63 (1)
2011	11 323	3087 (27)	848 (7)	57 (1)
2012	7602	2114 (28)	585 (8)	47 (1)
2013	5072	1380 (27)	396 (8)	10 (0.2)
Total	76 486	24 676 (32)	6162 (8)	870 (1)

Our analysis of factors contributing to the delay in first reaching a DOTS facility—which is the largest component of total delay—is summarized in Tables 3 and 4. The investigation of sociodemographic characteristics associated with a prolonged delay indicated that certain ethnic (minority) groups are more likely to experience a delay of >90 days before reaching a DOTS facility. For example, compared with the most common ethnic group, Han, the following minority groups were twice as likely (or more) to experience prolonged delay to reaching a DOTS facility: Nu (OR = 1.99; 95% CI = 1.77–2.23), Tibetan (OR = 2.04; 95% CI = 1.83–2.27), Tujia (OR = 2.30; 95% CI = 1.75–3.02). This was in line with findings from the qualitative analysis, which suggested that patients from ethnic minority groups faced greater language barriers at local health facilities where staff were unable to speak their dialects. Similarly, we found that other vulnerable groups that are reliant on family members to assist them in accessing care—such as elderly patients and women—often faced challenges in reaching appropriate health facilities. Both the qualitative and quantitative analysis indicated that farmers (people working in rural areas) experienced greater delays, which the interviews suggest may be due to the direct/indirect costs associated with traveling to distant CDC units from rural areas, as well as poor awareness of TB symptoms among rural populations resulting in self-medication. Overall, patient interviews overwhelmingly pointed towards financial factors—including costs of transport, loss of income while seeking care, and the potential of job loss if the illness is discovered by an employer—being key drivers of delay to accessing care for TB.

Table 3.

Sociodemographic characteristics of TB patients and their association with prolonged delay (>90 days) in accessing a DOTS-affiliated health facility (aOR, adjusted odds ratio; 95% CI, 95% confidence interval).

Variable	Patients	Delay > 90 days (%)	aOR	95% CI	P value
Gender
Male	52 832	16 755 (32)	1
Female	23 654	7921 (33)	1.12	1.08–1.16	<0.0001
Age
15–24	14 336	3992 (29)	1
25–34	15 738	4927 (31)	1.13	1.07–1.19	<0.0001
35–44	15 141	5206 (34)	1.32	1.25–1.39	<0.0001
45–54	11 947	4131 (35)	1.34	1.27–1.37	<0.0001
55–64	10 728	3624 (34)	1.29	1.21–1.37	<0.0001
65+	8596	2796 (33)	1.24	1.17–1.32	<0.0001
Occupation
Farm workers	62 805	20 976 (33)	1
Housework	1571	448 (29)	0.83	0.74–0.93	<0.0001
Others	4091	1124 (27)	0.76	0.70–0.82	<0.0001
Retired	1141	329 (29)	0.85	0.74–0.97	<0.0001
Students	2862	657 (23)	0.69	0.62–0.76	<0.0001
Unknown	436	96 (22)	0.57	0.45–0.72	<0.0001
Urban workers	3580	1046 (29)	0.84	0.78–0.91	<0.0001
Household type
Local	72 577	23511 (32)	1
Non-Local	3909	1165 (30)	1.17	1.08–1.26	<0.0001
Ethnic group
Han	56 933	18417 (32)	1
Bai	6097	1615 (26)	0.94	0.89–1.01	0.077
Dai	3715	1253 (34)	1.37	1.27–1.48	<0.0001
Hmong	125	27 (22)	0.78	0.51–1.19	0.251
Jingpo	1374	549 (40)	1.76	1.57–1.97	<0.0001
Jinuo	1336	331 (25)	0.85	0.74–0.96	0.011
Lahu	748	218 (29)	1.10	0.94–1.29	0.236
Muslim	280	59 (21)	0.62	0.46–0.82	0.001
Nu	1238	524 (42)	1.99	1.77–2.23	<0.0001
Tibetan	1477	648 (44)	2.04	1.83–2.27	<0.0001
Tujia	211	96 (45)	2.30	1.75–3.02	<0.0001
Wa	1031	317 (31)	1.16	1.02–1.33	0.028
Yao	235	58 (25)	0.87	0.65–1.17	0.363
Yi	728	251 (34)	1.31	1.12–1.53	0.001
Zhang	521	158 (30)	1.19	0.99–1.44	0.071
Other	437	155 (35)	1.51	1.23–1.84	<0.0001
Total	76 486	24,676 (32)

Table 4.

Summary of factors contributing to delays in accessing DOTS-affiliated health facilities identified through a qualitative analysis

Barrier	Illustrative quote or extract from notes
Economic impact of seeking care including transport costs and loss of income; both related to distance to health facilities	“It takes nine hours to get from here to the hospital in Kunming and it costs about 150 Yuan (24 USD) for a single journey. We also have to stay in a guesthouse”
Fear of stigma within community and consequences on employment	“People with jobs are scared of getting found out by their employers”
Reliance of vulnerable groups (elderly, those with language barriers) on family members to accompany them to health facilities	Sometimes children will not take elders to hospital owing to costs or inconvenience “Language is also a problem; that is why our son has come with us”
Language barriers faced by minority ethnic groups	Some ethnic groups living in isolated rural communities have never been to a town or had formal state education; can be illiterate, face access barriers related to language “Many from Tibet cannot speak Mandarin. We have 26 ethnic groups”
Lack of awareness of TB and its symptoms (among patients and healthcare providers)	“Particularly people living in rural and remote areas, have less knowledge about TB. Also older people, over 60, know even less and they commonly assume it is just respiratory problems”
Self-medication or use of alternative (non-CDC) healthcare providers	“I didn’t know anything (about TB) until I was first diagnosed at 15, 35 years ago. I took drugs myself when I had cough or severe symptoms”
Less priority given to health of women	There are gender differences: Sister helps brother. Wife stays in hospital with husband. But would less likely happen the other way around.
Migrant worker need to keep working; cannot take time off for healthcare	“They (migrant workers) cannot rest, need to keep on working”

Discussion

This study, which to our knowledge is one of the largest on delays to accessing TB diagnosis and treatment in any country, demonstrates that prolonged delay before initiation of appropriate treatment occur in a substantial proportion of the population in Yunnan even when global case detection targets are being exceeded. Our comparison of delays at three stages between symptom onset and treatment initiation revealed that the first stage, delay to reaching a CDC diagnostic centre administering DOTS, is substantially larger than delays experienced after patients enter the DOTS system.

Analysis of patient delays between 2006 and 2013 demonstrated that prolonged delays (>90 days between symptom onset and first presenting a CDC unit) occurred in more than a quarter of patients even though China (and Yunnan province specifically) reported 100% DOTS coverage, defined as the proportion of population living in administrative areas with access ‘free’ TB services under the DOTS strategy, since 2004. Prolonged delay to a DOTS facility and likely consequent transmission of TB may explain why bacteriologically confirmed TB prevalence (per 100 000 people) in western China declined only minimally from 262 (95% CI = 228–302) in 2000 to 212 (180–249) in 2010 despite substantial investments in TB control and 100% DOTS implementation (Wang et al. 2014).

In line with our findings, studies outside of China have also indicated that implementation of the DOTS strategy—which relies on symptomatic patients presenting at DOTS-affiliated health facilities—may be more effective in shortening delays experienced after rather than before reaching appropriate health facilities (Takarinda et al. 2015). In our study, median health system and treatment initiation delays were 2 and 1 day, respectively, much shorter than median patient delay of 57 days. Our mixed methods analysis of factors contributing to prolonged delays in access to DOTS-affiliated facilities identified patient groups that could be better supported by the health system. Among these were specific ethnic minority groups, elderly patients and farmers/rural residents who were at a disadvantage owing to language and information barriers; greater distance and costs involved in accessing health facilities; and loss of income (their own of for the person accompanying them). The qualitative analysis also suggested that these factors can cause TB patients to first self-medicate or seek care at (non-DOTS) health facilities in their local area, resulting in a delay to accessing a DOTS-affiliated facility.

Our findings are supported by other studies in China which report that patients encounter several barriers to accessing ‘free’ TB health services, including costs of transport and accommodation, distance to health facility, loss of income, first visit to health facilities not affiliated with DOTS and lack of awareness about TB (Lin et al. 2008; Li et al. 2013; Qiu et al. 2015). Policy responses to reduce the costs incurred in accessing care and barriers faced by patients who are less able to travel outside of their local area could include more widespread availability of TB diagnosis and treatment in peripheral health centres and improvements in health insurance for TB as well as other forms of financial support for presumptive TB patients; linked evaluations of the effectiveness of such measures on reducing delay would provide importance evidence to inform policy decisions in other contexts . This study further illustrates that any definition of universal health coverage or DOTS coverage must not only consider nominal physical access to services, but encompasses the health system’s capacity to deliver quality, people-centred care without the risk of financial hardship. Future studies estimating the impact of reductions in delay to diagnosis on TB incidence and prevalence, comparing reductions in delay at different periods such as from 45 to 30 days post-symptom onset vs 90–75 days, would be useful in informing TB control strategies.

Our study fills an important gap in evidence as we include data from all 129 county-level public TB diagnostic centres across Yunnan from 2006 to 2013, thereby reducing selection bias affecting previous studies. A recent systematic review identified 29 studies on delay to TB diagnosis in China analysed data from 23 917 patients (Li et al. 2013). Authors highlighted that confounders and variations in definitions used across studies limited the conclusions they could draw. In contrast, we analysed data from 76 486 adult smear-positive TB patients using uniform definitions of delay and patient inclusion criteria, thereby generating robust evidence from considerably more patients than included in the systematic review. An important strength of our study, compared with previous studies investigating delays, is that we were able to comprehensively include all smear-positive adult pulmonary TB patients diagnosed and treated in the public sector in Yunnan province, China, rather than selecting a group of patients from certain health facilities. This limited selection bias and substantially increases the validity of our results.

As with the majority of studies on delay, we relied on patients’ recall of symptom onset, which may be inaccurate as symptoms may be non-specific, but does allow comparison with other studies (Storla et al. 2008). We highlight that this study analysed delay to first presentation at a CDC TB unit, and we did not have information on prior care seeking at other health facilities including government and non-government healthcare providers. However, since TB control programmes aim for all patients to access free, quality-controlled, standardized TB care through DOTS-affiliated centres, delay to presentation and treatment at a CDC unit is the most relevant time period that needs to be monitored and reduced.

Conclusions

This study provides insights into the unexpectedly modest decline in global incidence rates, by demonstrating that prolonged delays in diagnosis of infectious, symptomatic smear-positive TB cases occur even when there is full implementation of the global TB control DOTS strategy, an excellent electronic reporting system is in place and case-finding and treatment success targets are being exceeded. The evidence generated by this study suggests that the structure of case detection targets in the global TB control strategy may require modification. Since TB transmission is directly related to the period that patients remain infectious (Lin et al. 2008) we propose that delay to initiation of treatment is monitored and included as a key indicator in evaluating the success of TB programmes. Programmatic targets that focus on reducing delays to treatment initiation, rather than only on increasing the overall proportion of cases detected, may result in a greater progress towards the ambitious Post-2015 Global TB Strategy goal, which aims for the annual decline in global incidence to accelerate from 2% per year in 2015 to 10% per year by 2025 (WHO 2015).

Ethics

This study is part of a multidisciplinary investigation of TB in Yunnan province, which was approved by the Yunnan CDC ethics committee, the FHI 360 Protection of Human Subjects Research Committee and the LSHTM Research Ethics Committee.

Funding

This research was funded by the US Agency for International Development’s Control and Prevention—Tuberculosis Project. The funder had no involvement in the design, conduct or analysis of the research and did not influence the decision to publish.

Acknowledgements

We appreciate efforts of the Yunnan CDC and county TB centre doctors for maintaining the electronic patient records and the FHI 360 team for their support throughout data collection and analysis.

Conflict of interest statement. None declared.