Tuberculosis detection and the challenges of integrated care in rural China: A cross-sectional standardized patient study

Abstract

Background: Despite recent reductions in prevalence, China still faces a substantial tuberculosis (TB) burden, with future progress dependent on the ability of rural providers to appropriately detect and refer TB patients for further care. This study (a) provides a baseline assessment of the ability of rural providers to correctly manage presumptive TB cases; (b) measures the gap between provider knowledge and practice and; (c) evaluates how ongoing reforms of China's health system-characterized by a movement toward "integrated care" and promotion of initial contact with grassroots providers-will affect the care of TB patients.

Methods/findings: Unannounced standardized patients (SPs) presenting with classic pulmonary TB symptoms were deployed in 3 provinces of China in July 2015. The SPs successfully completed 274 interactions across all 3 tiers of China's rural health system, interacting with providers in 46 village clinics, 207 township health centers, and 21 county hospitals. Interactions between providers and standardized patients were assessed against international and national standards of TB care. Using a lenient definition of correct management as at least a referral, chest X-ray or sputum test, 41% (111 of 274) SPs were correctly managed. Although there were no cases of empirical anti-TB treatment, antibiotics unrelated to the treatment of TB were prescribed in 168 of 274 interactions or 61.3% (95% CI: 55%-67%). Correct management proportions significantly higher at county hospitals compared to township health centers (OR 0.06, 95% CI: 0.01-0.25, p < 0.001) and village clinics (OR 0.02, 95% CI: 0.0-0.17, p < 0.001). Correct management in tests of knowledge administered to the same 274 physicians for the same case was 45 percentage points (95% CI: 37%-53%) higher with 24 percentage points (95% CI: -33% to -15%) fewer antibiotic prescriptions. Relative to the current system, where patients can choose to bypass any level of care, simulations suggest that a system of managed referral with gatekeeping at the level of village clinics would reduce proportions of correct management from 41% to 16%, while gatekeeping at the level of the township hospital would retain correct management close to current levels at 37%. The main limitations of the study are 2-fold. First, we evaluate the management of a one-time new patient presenting with presumptive TB, which may not reflect how providers manage repeat patients or more complicated TB presentations. Second, simulations under alternate policies require behavioral and statistical assumptions that should be addressed in future applications of this method.

Conclusions: There were significant quality deficits among village clinics and township health centers in the management of a classic case of presumptive TB, with higher proportions of correct case management in county hospitals. Poor clinical performance does not arise only from a lack of knowledge, a phenomenon known as the "know-do" gap. Given significant deficits in quality of care, reforms encouraging first contact with lower tiers of the health system can improve efficiency only with concomitant improvements in appropriate management of presumptive TB patients in village clinics and township health centers.

Fig 1. STROBE flowchart.

SPs were randomly assigned to facilities and within each facility, SPs visited the doctor following the normal procedures for any walk-in patient. Given a choice of which doctor to visit, SPs randomly chose a doctor following a pre-determined randomization protocol. In county hospitals, where patients can choose doctors by specialty, SPs visited generalists. Our results, therefore, are designed to approach the care a walk-in patient would receive at each of the sampled facilities.

Fig 2. Main outcomes of interactions with SPs.

Presaging our simulation results, in the 13 of 46 cases (28%, 95% CI 17%–43%) when village providers (verbally) referred SPs, 5 of 13 (38%, 95% CI 18%–64%) referred to the THC and 6 of 13 (46%, 95% CI 23%–71%) referred directly to the county. Of the 36 (18%, 95% CI 13%–24%) township providers who referred SPs to an upper level provider, a small majority referred to the county hospital (18 of 36; 51%, 95% CI 36%–67%), and the rest referred directly to the CCDC (14 of 36; 39%, 95% CI 25%–55%) or a city level provider (4 of 36; 11%, 95% CI 4%–25%). CXR, chest X-ray; ISTC, international standards for tuberculosis care; SP, standardized patient; THC, township health center.

Fig 3. Correlation between provider certification and SP outcomes among village and township providers.

Odds ratios are adjusted for additional facility and doctor characteristics using regressions reported in S4 Table. A “Practicing Physician” certification is the highest of 3 certification levels above “Assistant Practicing Physician” and “Rural Physician” certifications. “Proportion” refers to the proportion in the full sample of village and township providers. * p < 0.10, ** p < 0.05, *** p <0.01. SP, standardized patient.

Fig 4. Know-do gap: Comparison of data from vignettes versus SP interaction among the same providers.

For all items, the prefix "E" indicates examinations and laboratory evaluations; the prefix "Q" indicates history questions. The gap calculation is the result of a t test comparing the average vignette performance with the average SP performance. P values are in brackets. AFB, acid-fast bacillus; SP, standardized patient; TB, tuberculosis.

Fig 5. Estimated patient pathways under status quo (patients freely selecting into tiers).

Percentages at bottom of figure show the percentage of patients selecting into each health system tier based on survey responses (S5 Table). For each referral pathway, figure shows percentage of total patient population following each path calculated using SP results for subsample of complete health systems.