Whole-course quality of tuberculosis (TB) care in rural China: a retrospective study based on chart abstraction


Tuberculosis (TB) is a major cause of ill health and death worldwide.1 Around 10 million people were infected, and 1.2 million people died from TB in 2019 globally. The COVID-19 pandemic has reversed the progress in combating TB, with the first year-on-year increased TB deaths up from 1.2 million in 2019 to 1.3 million in 2020 and sharp drops in case notifications from 7.1 million in 2019 to 5.8 million in 2020.1 TB coinfection with HIV and drug-resistant TB are also threatening efforts in controlling TB. China is among the high TB burden countries. 1 The annual incidence of pulmonary TB was 52.2/100 000 in 2016,2 and is projected to be 35.9/100 000 in 2030.3 Rural areas have a severely higher TB burden than urban areas (twice higher as urban).4 Underdeveloped regions also have a heavier TB burden, with the incidence of TB in the least developed regions approximately seven times higher than those observed in the most developed regions.5

High-quality care is the priority in ending the TB epidemic.6 The treatment course usually involved at least 6 months. After initiating treatment, it is important to perform assessments timely and adjust medication appropriately in the process. Inappropriate treatments or treatment interruption expose patients to a high risk of relapse or deterioration into drug-resistant TB/multidrug-resistant TB.7 Previous studies have mainly focused on a single doctor–patient interaction and overlooked the quality of the whole course.8–11

China has invested considerable efforts in tackling the TB epidemic. The WHO-recommended directly observed treatment short course (DOTS) strategy was implemented in China in 1990.12 In the 1990s, China’s TB vertical control system was led by the Centers for Disease Control and Prevention (CDC), which delivered TB screening, diagnosis and treatment.13 The CDC-led TB model combined clinical services and public health services in one institution. From 1990 to 2010, a national survey showed China’s TB prevalence decreased from 170/100 000 to 59/100 000.4 However, the rising drug-resistant TB and massive internal immigration in China challenged the traditional model. TB comorbid with chronic diseases raised concerns about the quality of care.3 In 2011, China started to build an integrated TB control system by transferring from a CDC-led model to a designated hospital-led model to integrate care and improve quality.13 The diagnosis and treatment are integrated into designated hospitals. CDC mainly provides surveillance and monitoring. Patients with drug-resistant TB/multidrug-resistant TB are treated at prefecture-level or provincial hospitals.3 Primary health facilities are responsible to refer presumptive TB and follow-up.

Although an important goal of the new model was to improve quality, there was insufficient evidence to draw on. We conducted a retrospective cohort study by examining the TB medical records in rural China using chart abstraction. We studied the quality of TB care for the whole course including diagnosis quality, treatment quality and management quality. We also assessed factors that affect completing treatment. This study could add knowledge on the quality of whole-course TB care and provide preliminary evidence for the TB system reform in China.


Study design and data collection

This study was conducted as part of the evaluation for the ‘China-Gates Primary Health Care Project’ that aims to develop an effective model for addressing chronic illnesses including diabetes, hypertension and TB in rural areas.14–16 An observational retrospective cohort study with chart abstraction design was employed, which has been widely used to evaluate the quality of care.17 18 Two provinces in China were selected (Henan and Hubei provinces), both representing underdeveloped rural regions (online supplemental appendix table A1). In 2019, the TB incidence rate was 51.8 and 61.9 per 100 000 population in Hennan province and Hubei province, respectively, nearly double the incidence rate in well-developed regions such as Shanghai (26.7 per 100 000 population) and Beijing (32.2 per 100 000 population).5 One county in each province was selected and all the medical records (outpatient and inpatient) for patients with TB newly registered from 1 July 2020 to 31 December 2020 were reviewed and abstracted using predetermined checklists. Medical records for one person (identified through name, gender and personal ID) were compiled together and abstracted. The inclusion criteria were as follows: (1) active TB; (2) pulmonary TB; and (3) finished whole-course treatment. The exclusion criteria were as follows: (1) not active TB or extrapulmonary TB; (2) tuberculous pleuritis; and (3) haven’t finished whole-course treatment. The characteristics of the two counties and process for data collection were provided in the online supplemental appendix. We followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline of cohort studies.

Supplemental material

Measurements and definitions

The whole-course quality of TB care included diagnostic quality, treatment quality and management quality. Diagnostic quality refers to completing aetiological diagnosis, completing examinations and laboratory assessments, and making a diagnosis with adequate evidence (online supplemental appendix table A2). Treatment quality refers to prescribing standard chemotherapy regimen and its appropriateness, second-line drugs prescription and its appropriateness. Management quality refers to performing midterm assessments, receiving sufficient services and completing treatment (online supplemental appendix table A2). Receiving sufficient services refers to receiving all the recommended examinations and tests during the whole course of treatment, including (1) sputum tests before treatment, at 2 months and 5 months after initiating treatment, and stopping treatment; (2) chest radiograph before treatment, at 5 months after treatment, and stopping treatment; (3) liver and frenal function tests before treatment, at 2 months and 5 months after treatment, and stopping treatment; (4) complete blood count before treatment, at 2 months and 5 months after treatment, and stopping treatment. If patients stopped treatment due to death, loss-of-follow up or other reasons, services before stopping treatment were evaluated.

Data processing and analysis

Descriptive analysis was used to present the basic. Pearson χ2 tests or Fisher-exact tests were used for examining the differences in quality. The results on quality of diagnosis, treatment and management were stratified by subgroups of smear results, outpatient/inpatient experiences and demographic characteristics. Considering regional variation in TB management, we also reported the results by counties to minimise confounding bias. Multivariable logistic regression for rare events data was used to examine the factors that could have a potential impact on completing TB treatment. We chose the above model to increase the efficiency of estimation because 34 patients (12.7%) did not complete the treatment, and the sample size was smaller in different groups of binary independent variables. Independent variables included county, sex, age, health status (TB symptoms, inpatient admission), types of TB and overall management quality. P values below 0.05 were considered statistically significant. Checklists with incomplete information were re-examined by a third person during the chart abstraction process. In data analysis, reasons for missing data were explained below tables, and no imputation was conducted. Stata V.16.0 (Stata Corp LP) was used to perform all the analyses.

Patient and public involvement



Patient characteristics

In total, we reviewed 268 outpatient medical records and 126 inpatient medical records. Outpatients and inpatients were 48.6 and 52.4 years old on average, respectively. 31.0% of outpatients and 30.2% of inpatients were female. 45.5% of outpatients and 52.4% of inpatients were Rifampicin susceptible. 92.9% of outpatients were recorded as a new treatment, and only 12 (4.5%) outpatients had retreatment. For inpatients, 88.9% were recorded as a new treatment, and 7.1% had retreatment (online supplemental appendix table A3).

Quality of diagnosis

In total, 96.3% of both outpatients and inpatients received sputum tests, and the average number of sputum tests was 2.6. 67.3% received a molecular biological diagnosis. 238 outpatients and inpatients (98.3%) completed chest radiographs. For smear-positive outpatients and inpatients, 91.8% received further tests on phenotypic susceptibility. Differential diagnoses (initiating standard chemotherapy regimen for culture-negative TB diagnosis for 2 months and then evaluating the result) and interferon-gamma release assays (IGRAs) were not commonly used for patients, with 18.5% receiving differential diagnoses for diagnosis and 8.0% receiving IGRAs (table 1).

Table 1

Quality of diagnosis for patients with TB, n (%)

Quality of treatment

Rifampicin susceptible/unknown patients should take the standard treatment (four-drug, first-line antituberculosis therapy including isoniazid (H), rifampicin (R), pyrazinamide (Z) and ethambutol (E), also referred to as HRZE regimen) for 2 months and then take isoniazid and rifampicin (HR regimen) for 4 months according to the national TB treatment standards.19 The rate of taking the initial standard regimen was 91.6% for outpatients and 85.5% for inpatients (table 2).

Table 2

Quality of treatment for patients with TB, n (%)

Inpatients had a high rate of prescribing second-line drugs (53.2%), and 85.1% of which were inappropriate. The most common reason for hospitalisation was for further differential diagnosis (41.9%) (online supplemental appendix table A4).

Quality of management

All patients are required to be assessed in midterm (definition of midterm can be found in table 1), but only 47.9% of both outpatients and inpatients received midterm assessment (table 3). Patients with a negative smear (56.2% vs 36.5%, p=0.003), below 60 years old (53.4% vs 37.5%, p=0.015), and without inpatient admission (56.3% vs 38.2%, p=0.003) had significantly more midterm assessments than their counterparts. 87.3% of outpatients and inpatients completed treatment. By the strict definition of ‘receiving sufficient services during the whole course’, only 19.7% of outpatients and inpatients received sufficient services.

Table 3

Quality of management for patients with TB, n (%)

Sputum tests, chest radiographs, liver and renal function tests, and complete blood counts were recommended for patients with TB during the treatment course. TB control guidelines require that sputum test, liver and renal function test, and complete blood count should be monitored at the second month, fifth month and the last month of the treatment course (sixth month), and chest radiograph at the second and the end of the treatment course. All four recommended examinations and tests declined during the whole course. Use of sputum tests declined during the whole course, with only 46.6% receiving the test 2 months after starting treatment (figure 1). However, the two project counties had a large difference in delivering whole-course sputum tests. County B had 91.1%, 87.1% and 79.8% of outpatients and inpatients receiving sputum tests at the second month, fifth month and the last month after treatment, compared with 6.5%, 4.4% and 2.2% in another county (online supplemental appendix table A5). When stopping treatment, 61.1% of outpatients and inpatients received a chest radiograph, 49.6% received a liver and renal function test and 49.2% received complete blood counts.

Figure 1
Figure 1

Quality of management for patients with tuberculosis (TB): tests and examinations during the whole course. Notes: chest radiograph is not required for the fifth months after treatment according to the national guideline for TB treatment in China.

Factors affecting TB treatment completion

Patients’ characteristics and previous healthcare utilisation have a significant impact on treatment completion. Patients with TB symptoms were significantly more likely to complete treatment (OR=1.80, p=0.011) than patients with asymptomatic TB (table 4). Patients who once had inpatient admission were less likely to complete treatment (OR=0.58, p=0.019). Patients from the two counties differed significantly in completing treatment.

Table 4

Factors affecting completing TB treatment


This study examined the whole-course quality of TB care in rural China using the chart abstraction method. To our knowledge, this is the first study that assessed the TB care quality of the whole course in rural China. This study confirmed the achievements in the quality of TB care after the new TB control system reform: the diagnosis of TB was based on adequate evidence of appropriate tests and examination, the treatment for outpatients complied with standards. However, challenges remain in whole-course management and inpatient treatment. Only 47.9% received midterm assessments, and the recommended examinations and tests declined along the treatment course. Inpatient doctors overused irrational second-line drugs, with 53.2% of inpatients prescribed second-line drugs, of which 85.1% were inappropriate.

Achievements in the quality of TB care

Our observations of the achievements in quality TB care align with previous studies. In 2015, a study using standardised patients found that 90% ordered a chest radiograph, sputum test or referral for diagnosis.10 Another study in 2016 using chart abstraction found that 85.3% of patients with TB received adequate diagnostic services (including three sputum smears, chest X-ray or CT examinations, TB symptoms, TB skin tests, IGRAs and differential diagnosis).9 For treatment completion, compared with other high-burden countries, most patients with TB were able to complete the treatment in our study, which was higher than that in India (45%),20 and South Africa (53%).21

The achievements in the quality of TB diagnosis, treatment and treatment completion reflect China’s substantial efforts in TB combat. Massive funding has been injected into TB control, increasing from ¥2.5 million before 2000 to ¥1.1 billion in 2020.22 In 2016, the national health blueprint, Healthy China 2030, highlighted the need to ‘build integrated TB control system, enhance detection and surveillance of multi-drug resistant TB and standardize the treatment and management of TB’.23 In 2019, the China National Health Commission issued End TB Action Plan (2019–2022), further outlining the process of enacting the goal.24 Since 2011, the new designated hospital-led TB control model was launched to improve integration, patient centeredness and quality of care. New information technology and equipment, such as artificial intelligence-based microscopic examination, have been introduced to TB care to increase diagnostic accuracy.25 New technology, such as electronic pillbox, has also been introduced to improve adherence and treatment completion.22

Challenges to the quality of TB care

The first challenge is insufficient whole-course management. Sputum tests, chest radiographs, liver and renal function tests, and complete blood counts all decreased in the whole course, especially the sputum tests. The standard 6-month treatment regimen for drug-susceptible patients is a long course, and even longer for patients with drug-resistant TB. Ensuring patient adherence was the primary challenge.26 Our results suggested that patients were able to complete the treatment but had low adherence to the tests and examinations during the whole course. Low adherence to tests and examinations undermines the opportunity to detect treatment failure or drug resistance, which will lead to a substantial failure in controlling the TB epidemic at both household and community levels.27

The reasons for insufficient whole-course management are complex. Some patients tend to discontinue their therapy once there are no cultured bacilli found in their sputum after 2 months of therapy. However, it’s recommended to continue therapy for additional 4 months to avoid relapse.26 DOTS strategy is useful in improving adherence but faces many challenges. For example, village doctors are the main providers of direct observation but are not incentivised enough for such time-consuming work.12 Care in county-level hospitals is mainly episodic. Doctors have no incentives to urge patients to finish whole-course management.

The fragmentation between CDC and hospitals also contributes to suboptimal whole-course management. Under the traditional CDC-led TB control model, all diagnosis, treatment and management services were centralised within CDC. This centralised model provided CDC a stronger incentive to actively engage patients with TB, encouraging them to receive essential sputum tests and chest radiographs to improve TB control performance. However, in the new designated hospital model, the responsibility for diagnosis and treatment has shifted to hospitals, while CDC primarily assumes a supervisory and monitoring role over hospital activities, including patient management. Yet CDC, especially at the county level, often has challenges to supervise hospitals effectively. CDC’s authority over departments within hospitals is often limited. Compounded by the fact that most health workers in CDC do not have medical license, their capacity to effectively monitor and address quality of care regarding hospital physicians is further constrained.28

The second challenge in quality is the high irrationality of second-line drugs in inpatients. Lin et al also found a substantial proportion of patients using second-line drugs (24.4%) in Yunnan province, a resource-poor area in China.29 Huang et al reported that 54.9% of the inpatients used second-line drugs in hospitals in 2012.28 The reasons behind are twofold. First, outpatient and inpatient TB care in designated hospitals are usually provided by different teams. There are no TB departments in inpatient settings, so inpatients are primarily admitted to pneumology departments. While outpatient doctors are full-time based in outpatient TB departments, many doctors from pneumology departments do not receive sufficient training on treatment and management of complex patients with TB.13 28 Second, first-line drugs, smear tests and chest radiographs are fully subsidised in China and are provided for free of charge. County-level public hospitals largely depend on service revenues, but TB departments are unable to generate large revenues. Second-line drugs can generate profits, and doctors’ incomes are closely associated with hospitals’ revenue, so inpatient doctors may have incentives to prescribe second-line drugs.28 30

Patients exhibiting symptoms are more likely to complete treatment. Patients with prior inpatient admissions showed lower rates of treatment completion. The extent of receiving services throughout the whole course did not have a significant impact on treatment completion. In China, ‘completing treatment’, a mandatory benchmark on the TB watchlist used to evaluate local governments’ progress in combating TB, simply means patients obtaining drugs from a dispensary for the whole course.19 ‘Receiving sufficient services’, however, is only ‘recommended’ to local governments and is currently not included on the evaluation watchlist. Local governments are more incentivised to fulfil mandatory benchmarks rather than those that are merely recommended.

Chart abstraction was criticised for recording bias.17 Results measured by chart abstraction tend to underestimate the quality of care and can serve as a lower-bound estimate.31 Detailed discussion on methodologies can be found in the online supplemental appendix.


This study was subject to several limitations. First, the results may be subject to bias because only active pulmonary TB were included. Some important populations such as patients with drug-resistant TB were not included because they were admitted by prefecture-level or provincial hospitals. Second, the generalisation of the findings should be cautious because the project counties are both resource-poor areas. Third, we are unable to assess the quality of TB detection and the delay between symptoms onset and diagnosis. The quality of TB detection can be evaluated by standardised patients.10 11 Chart abstraction could be combined with other methods like standardised patients in the future to assess different aspects of quality.


The whole-course quality of TB care has presented achievements and challenges. TB diagnosis was based on adequate evidence, and treatment for outpatients complied with standards. Nevertheless, the quality of TB care faced pressing concerns of insufficient whole-course management and irrational use of second-line drugs in inpatients. Outpatient and inpatient care should be integrated to close the fragmentation along the course of care for inpatients. Integration between CDC and hospitals should be enhanced to improve whole-course management.

This post was originally published on https://bmjopen.bmj.com