Risk factors for diagnosis and treatment delay among patients with multidrug-resistant tuberculosis in Hunan Province, China

A retrospective cohort study was conducted in Hunan province, China. Hunan is one of the 34 provincial regions in China. It has an estimated total area of 211,800 square kilometres’ (81,800 square miles) and has just over 66 million inhabitants as of 2020. According to the census conducted in 2022, Hunan province has 41 ethnic groups. Of these, 89.79% identified as Han, and the remaining 10.21% as one of the minority groups, including Tujia, Miao, Dong, Yao, Bai, Hui, Zhuang, Uyghurs, and so on [12]. The proportion of TB cases that are MDR-TB in Hunan province ranges from 10.6% to 25.2%, which is significantly higher than the national average of 1.8% [13]. Hunan Chest Hospital is the only chest hospital in Changsha (the province’s capital) that provides diagnosis and treatment services for patients with MDR and Extensively Drug-resistant (XDR)-TB. The hospital commenced MDR- and XDR-TB care in 2011 and serves as a referral hospital for presumptive DR-TB patients.

The study population was all MDR-TB and XDR-TB patients enrolled at Hunan Chest Hospital between 2013 and 2018. Data were obtained from an internet-based TB management system administered by the TB Control Institute of Hunan province. All pulmonary or extrapulmonary MDR- and XDR-TB patients enrolled during the study period were included. The data contained demographic variables (age, gender, year of diagnosis, occupational status, residence, ethnicity, patient source, and detainee status) and clinical factors (initial date of symptoms, date of diagnosis, date of treatment commencement, registration type, diagnosis type, diagnosis institution type, and treatment delay).

Multidrug-resistant tuberculosis (MDR-TB) is defined as a TB patient who is resistant to at least isoniazid and rifampicin. Extensively drug-resistant (XDR)-TB is defined as a TB patient resistant to isoniazid, rifampicin, fluoroquinolones, and either bedaquiline or linezolid (or both) [14]. We defined a total interval as the entire time required from the onset of symptoms to the start of MDR-TB treatment, which was divided into diagnosis and treatment intervals [15]. Diagnosis interval was defined as the time from the onset of symptoms to the date of MDR-TB confirmation [11]. Treatment delay was defined as the time from MDR-TB confirmation to commencement of MDR-TB treatment [16]. Severely ill was defined as a patient with severe comorbidities, having persistent coughing with breathlessness and weakness with extreme weight loss. Hunan ethnic classification recognizes 55 minority groups, including the Han majority.

Hunan Chest Hospital follows WHO-recommended methods for the diagnosis and treatment of MDR-TB. Clinical assessments based on symptoms, microscopic sputum examinations, radiological examinations, and molecular techniques such as Line Probe Assay and Gene Xpert are commonly used to diagnose MDR-TB in the province.

There are 131 counties in Hunan province, and only 32 counties can provide comprehensive diagnosis services, including culture. In Hunan Chest Hospital, drug susceptibility testing (DST) is mainly carried out to diagnose MDR-TB. As a result, sputum specimens from all culture-positive TB patients from all parts of the province are referred to the Hunan Chest Hospital for DST. In the hospital, phenotypic DST based on solid and liquid culture techniques and molecular methods using line probe assays and Xpert® MTB/RIF are performed. The Hunan Chest Hospital performs DST for rifampicin, isoniazid, ethambutol, streptomycin, kanamycin, and ofloxacin. Solid and liquid cultures are used to follow up on patients’ progress and treatment outcomes.

Data in an Excel spreadsheet was translated from Mandarin to English and exported to Stata version 17 software for analysis. Descriptive statistics were conducted and presented as frequencies (percentages) for categorical variables and medians with interquartile ranges (IQR) for continuous variables. The outcome variables (diagnosis and treatment intervals) were calculated in days and categorized as delayed and non-delayed. The number of days between the onset of symptoms and diagnosis confirmation and between diagnosis and commencement of MDR-TB treatment were calculated. A 14-day cut-off point was used to dichotomize diagnosis and treatment delay to be consistent with a previous study (https://​www.​health.​nsw.​gov.​au/​Infectious/​controlguideline​/​Pages/​tuberculosis.​aspx).

Overall, the median diagnosis interval among MDR-TB patients was 27 days (IQR 7–66 days), and the median treatment interval was one day (IQR 0–24 days) (Table 2). The overall prevalence of diagnosis delay among MDR-TB patients was 62.8% (95% CI: 60.1–65.4), and 30.8% (95% CI: 28.3–33.4) of MDR-TB patients experienced treatment delay. The diagnosis and treatment delay distribution are summarized in Supplementary files, Figs. S1 and S2.

Diagnosis delay among MDR-TB patients increased yearly but not linearly. On the other hand, the treatment delay increased over 2013 and 2015. Then, the trend was steady from 2015 to 2016 and showed a slight reduction between 2016 and 2017. The trend of treatment delay increased alarmingly over a period of 2017 and 2018 (Fig. 1).

The odds of experiencing MDR-TB diagnosis delay among patients who came through referral and tracing was 41% lower (AOR = 0.61, 95% CI: 0.47–0.80) relative to patients identified through consultations due to symptoms. Compared to patients younger than 15 years, patients aged ≥ 65 years had 65% lower odds of experiencing diagnosis delay (AOR = 0.35, 95% CI: 0.14–0.91) (Table 3).

In sensitivity analysis (i.e., using the median 27 days as a cut-off point instead of 14 days), age, occupation of respondents, and patient source of diagnosis were statistically significant factors associated with treatment delay, suggesting the results are sensitive to the selection of the threshold defining a delay. The odds of experiencing diagnosis delay among elderly MDR-TB patients (≥ 65 years) was 60% lower (AOR = 0.40, 0.17–0.94) than among children under 15 years old. The odds of developing diagnosis delay among MDR-TB patients identified through referral and tracing was 25% lower (AOR = 0.75, 0.59–0.96) compared with MDR-TB patients who came to seek consultations due to having had symptoms (Supplementary file Table S1).

No significant variables were identified in the sensitivity analysis that stratified MDR-TB patients as new or previously treated (Supplementary file Table S2).

Supplementary file Table S3 shows the negative binomial regression assessment of factors affecting diagnosis delay among MDR-TB patients. The finding showed that gender, diagnosis institution type, age, and treatment category were significantly associated with diagnosis delay.

Current residence, treatment category, ethnicity, and severity of illness were associated with treatment delay. The odds of developing treatment delay among foreign nationalities and people from other provinces were almost double (AOR = 1.99, 95% CI: 1.30–3.04) compared to the local populations. The odds of experiencing treatment delay among severely ill patients were 2.58 times higher (AOR = 2.58, 95% CI: 1.46–4.58) than patients who were not severely ill. Previously treated TB cases had 41% lower odds (AOR = 0.59, 95% CI: 0.42–0.85) of treatment delay than new MDR-TB cases. Similarly, other ethnic minority groups had 67% lower odds (AOR = 0.55, 95% CI: 0.33–0.93) of experiencing treatment delay than the Han majority (Table 4).

A sensitivity analysis was conducted using different thresholds defining a delay, and sensitivity to this threshold was identified. A sensitivity analysis using the third quartile (24 days) showed that only two variables (ethnicity and patient source) remained significant; the odds of experiencing treatment delay among ethnic minority MDR-TB patients were 62% lower (AOR = 0.38, 95% CI: 0.19–0.74) compared to the Han majority. The odds of experiencing treatment delay among patients identified through referral and tracing was 7.39 (4.75–11.49) times higher than patients who came to seek consultations (Supplementary file Table S4).

In a sensitivity analysis based on treatment category (new vs re-treatment), other occupations had 38% higher odds (AOR = 1.38, 95% CI: 1.01–1.90) of experiencing treatment delay than farmers (Supplementary file Table S5).

Findings from the negative binomial model showed that patient source was significantly associated with the risk of treatment delay (Supplementary file Table S6).

The median diagnosis interval found in this study was longer than the nine days reported in Myanmar [17] and five days in Bangladesh [18]. The difference could result from using different diagnosis modalities and algorithms to diagnose MDR-TB. On the other hand, the interval was shorter than previously reported in China, with a median diagnosis of 84 days [19]. The difference could be because of the introduction of Gene-Expert, which reduced laboratory result turnaround time in diagnosing MDR-TB patients [20]. Previous studies also showed that using GeneXpert had significantly reduced treatment delay among DR-TB patients [21, 22]. The study also revealed that 2/3 of MDR-TB patients developed diagnosis delays. However, our finding was higher than the WHO recommendations that every patient should have an early diagnosis of TB, including universal Drug Susceptibility Testing (DST) [2]. The WHO recommends that all DR-TB patients need to commence their treatment as soon as possible to prevent unnecessary side effects, complications, and poor treatment outcomes.

The trend of treatment delay increased alarmingly over a period of 2017 and 2018. A possible justification could be the high case burden of MDR-TB in Hunan province between 2017 and 2018, and patients may be obliged to wait longer to commence their treatment. Moreover, study subjects’ variation in terms of MDR-TB signs and symptoms and poor health-seeking behaviour may contribute to the difference. Also, previous studies suggested that the initiation of MDR-TB treatment mainly depends on baseline laboratory investigations and PMDT panel team decisions on DST test results and availability of treatment options [23, 24]. So, the alarming treatment delay in 2017/2018 could be due to the lack of treatment options. However, further research is needed to point out possible factors contributing to the high burden of treatment delay in 2017 and 2018.

Many MDR-TB patients experienced treatment delays in Hunan province, and it is highly recommended that the initiation of prompt MDR-TB treatment following confirmatory diagnosis is prioritized to achieve the END-TB strategy targets. Early diagnosis and treatment of TB is particularly important to minimize community transmission of the disease, reduce side effects, and improve disease progression treatment outcomes and quality of life of the patients. It can also help minimize the patients’ catastrophic costs [25].

There are few studies previously conducted in China to determine MDR-TB diagnosis and treatment delay. Our study is the first to investigate MDR-TB diagnosis and treatment delays, focusing on Hunan Province (one of the provinces with high MDR-TB burden in China). Our study has incorporated important variables such as ethnic minority and residence that have been missed in the previous studies. The first study conducted in Taizhou, Zhejiang Province, showed that the overall diagnosis and treatment delay was not illustrated. Instead, it primarily determines factors associated with waiting time for DST, pre-attrition, time of waiting for treatment, and treatment outcomes and associated factors. Moreover, it did not report the overall diagnosis delay and associated risk factors. On the other hand, a study conducted in China to determined diagnosis and treatment delay among MDR-TB patients lacked socio-demographic variables such as ethnicity and residence [26]. However, in our study, ethnic minorities and residents outside Hunan Province were identified as risk factors for diagnosis and treatment delay. This will add a piece of literature to the existing knowledge. Moreover, none of the studies conducted a sensitivity analysis, which is valuable for better decision-making and more reliable predictions and highlights areas for improvement. As there is no cut-off point to decide diagnosis and treatment delay, applying a sensitivity analysis to show the result at different points is significant.

Previous studies suggested that elderly patients often struggle to access diagnosis and treatment centres and usually rely on their families to visit health facilities [27]. However, our findings showed that patients ≥ 65 years of age had lower odds of diagnosis delay than patients aged < 15 years. A possible explanation could be that MDR-TB diagnosis in children is impacted by different diagnostic approaches; for example, children may be treated for other respiratory tract infections before diagnosis testing is undertaken, resulting in delayed MDR-TB diagnosis. A previous study revealed that the diagnosis of MDR-TB is bacteriological, and children need a systematic diagnosis approach that leads to a longer time to diagnosis of MDR-TB [28]. Moreover, a lack of awareness that children develop MDR-TB, a perceived inability to diagnose active MDR-TB without TST and CXR, and no international guidance on preventive therapy against MDR-TB could contribute to MDR-TB diagnosis delay in children aged < 15 years [29]. Therefore, improving diagnosis protocols in children could minimize diagnosis delay among MDR-TB patients.

In this study, patients identified through tracing and referral had a lower diagnosis delay than patients identified through symptomatic consultations. Moreover, tracing MDR-TB patients might have a significant contribution to MDR-TB patients visiting the health facility timely due to possible consultation, which is a basic component of support and overcoming major barriers that lead to not disclosing the disease due to the high nature of MDR-TB stigmatization. Also, referral reduces waiting times for primary care and minimizes double investigation as they can rely on their pre-referral work-up. The shorter delay among tracing and referral sub-groups may also be caused by the bias of symptom onset time.

Moreover, our findings agree with previous studies carried out on DS-TB patients. For instance, age and active case findings [30] were significantly associated with diagnosis delay among DS-TB patients. Moreover, other sociodemographic variables (educational status, knowledge of MDR-TB, and personal beliefs and attitudes) [31], using traditional medicines and healers, and limited health service access due to geographical location [32] were significantly associated with diagnosis delay among DS-TB patients.

Regarding treatment delay, Han ethnicity, patients with previous TB treatment history, residence other than Hunan province, and becoming severely ill were significantly associated with MDR-TB treatment delay. This study found that ethnic minority groups in Hunan province had lower odds of experiencing treatment delay than the Han majority. This could be due to improving educational attainment among ethnic minorities, which is highly prioritized in China [33]. This finding is consistent with a study conducted by Gilmour et al. among DS-TB patients in Hunan province; ethnic minority groups had lower odds of experiencing treatment delay than the Han majority. A systematic review and meta-analysis conducted among DS-TB patients in Ethiopia showed that lower educational level was significantly associated with poor health-seeking behaviour [34]. However, further research is recommended to illuminate why the Han majority groups are at higher odds of experiencing treatment delay than ethnic minority groups.

In this study, previously treated MDR-TB cases had lower odds of experiencing MDR-TB treatment delay. This might be because previously treated cases were more knowledgeable about what to do and aware of the severity of the disease and the risk of developing a resistant form of TB, and they already had relationships with clinicians that expedited their care; they were also less likely to experience diagnosis delay than new MDR-TB patients [34, 35].

Patients from locations other than Hunan province had an increased likelihood of experiencing treatment delay. In China, the hukou system, a household registration system, permits a permanent resident for a single address issued by the government of China to each family member given at birth or by employment in formal sectors [36]. The hukou system allows each family member access to essential public services, education, health services, social benefits, and job recruitment by the governmental or private sectors [37]. As a result, patients from Hunan province will have better access to health care and are less likely to experience treatment delays than patients from outside Hunan province. According to previous reports, a previous history of DS-TB treatment [38] and health system factors (staff shortages, cost of services, drug stockout, and poor health infrastructure) [11] are significantly associated with treatment delay.

The study had several limitations. The absence of a WHO-endorsed, international standard cut-off point to determine diagnosis and treatment delay can significantly impact estimates of the burden of delays and cause misclassification bias when investigating factors associated with the delays. Some important variables that might be associated with diagnosis and treatment delay, such as education status, income level, geographical inaccessibility, knowledge of MDR-TB, and presence of comorbidities (e.g., diabetes mellitus, HIV infection, mental ill health), were not assessed in the current study. Future research should prioritize undertaking prospective studies that measure these variables. Only patients attending the designated health facilities were included in the study. This might have caused selection bias, and findings might not be generalizable to all MDR-TB patients in Hunan province. Recall bias might also have been an issue for patients specifying the exact date of symptom onset. Finally, all sputum specimens were transported to Hunan Chest Hospital for DST might introduce turnaround time that will further influence on the diagnosis delay.