Variations in the quality of tuberculosis care in urban India: A cross-sectional, standardized patient study in two cities

Abstract

Background: India has the highest burden of tuberculosis (TB). Although most patients with TB in India seek care from the private sector, there is limited evidence on quality of TB care or its correlates. Following our validation study on the standardized patient (SP) method for TB, we utilized SPs to examine quality of adult TB care among health providers with different qualifications in 2 Indian cities.

Methods and findings: During 2014-2017, pilot programs engaged the private health sector to improve TB management in Mumbai and Patna. Nested within these projects, to obtain representative, baseline measures of quality of TB care at the city level, we recruited 24 adults to be SPs. They were trained to portray 4 TB "case scenarios" representing various stages of disease and diagnostic progression. Between November 2014 and August 2015, the SPs visited representatively sampled private providers stratified by qualification: (1) allopathic providers with Bachelor of Medicine, Bachelor of Surgery (MBBS) degrees or higher and (2) non-MBBS providers with alternative medicine, minimal, or no qualifications. Our main outcome was case-specific correct management benchmarked against the Standards for TB Care in India (STCI). Using ANOVA, we assessed variation in correct management and quality outcomes across (a) cities, (b) qualifications, and (c) case scenarios. Additionally, 2 micro-experiments identified sources of variation: first, quality in the presence of diagnostic test results certainty and second, provider consistency for different patients presenting the same case. A total of 2,652 SP-provider interactions across 1,203 health facilities were analyzed. Based on our sampling strategy and after removing 50 micro-experiment interactions, 2,602 interactions were weighted for city-representative interpretation. After weighting, the 473 Patna providers receiving SPs represent 3,179 eligible providers in Patna; in Mumbai, the 730 providers represent 7,115 eligible providers. Correct management was observed in 959 out of 2,602 interactions (37%; 35% weighted; 95% CI 32%-37%), primarily from referrals and ordering chest X-rays (CXRs). Unnecessary medicines were given to nearly all SPs, and antibiotic use was common. Anti-TB drugs were prescribed in 118 interactions (4.5%; 5% weighted), of which 45 were given in the case in which such treatment is considered correct management. MBBS and more qualified providers had higher odds of correctly managing cases than non-MBBS providers (odds ratio [OR] 2.80; 95% CI 2.05-3.82; p < 0.0001). Mumbai non-MBBS providers had higher odds of correct management than non-MBBS in Patna (OR 1.79; 95% CI 1.06-3.03), and MBBS providers' quality of care did not vary between cities (OR 1.15; 95% CI 0.79-1.68; p = 0.4642). In the micro-experiments, improving diagnostic certainty had a positive effect on correct management but not across all quality dimensions. Also, providers delivered idiosyncratically consistent care, repeating all observed actions, including mistakes, approximately 75% of the time. The SP method has limitations: it cannot account for patient mix or care-management practices reflecting more than one patient-provider interaction.

Conclusions: Quality of TB care is suboptimal and variable in urban India's private health sector. Addressing this is critical for India's plans to end TB by 2025. For the first time, we have rich measures on representative levels of care quality from 2 cities, which can inform private-sector TB interventions and quality-improvement efforts.

Fig 1. City-representative quality of care estimates.

City-level estimates of quality of care for each of our case scenarios. These proportions represent the estimated frequency with which the action would be observed if the standardized case scenario was presented to a provider randomly selected from the sampling frame. These estimates are calculated using inverse probability weights corresponding to the sample frame as detailed in S2 Text for every city–qualification–PPIA–case combination in the data. N = 2,602. AFB, acid-fast bacilli; PPIA, Private Provider Interface Agency; TB, tuberculosis; Xpert MTB/RIF, Xpert Mycobacterium tuberculosis/Rifampicin, also known as GeneXpert.

Fig 2. Management of Case 1 when no correct treatment was given.

Frequency in which Case 1 was managed with possible combinations of steroids, cough syrups, broad-spectrum antibiotics, and FQs, when no correct management was given. There were N = 834 Case 1 interactions that did not meet the criteria for correct management, and 172 interactions resulted in none of these case management behaviors. FQ, fluoroquinolone.

Fig 3. Quality of care differences by provider qualification and location.

Estimated ORs between various groups of providers, for the frequency in which the indicated management action is observed across all case scenarios. Panel A reports differences by MBBS qualification level, pooled across all observations. This regression includes controls for city setting and case scenario (N = 2,602). Panels B and C report similar ORs estimated across cities, stratified by MBBS qualification (N = 1,448 and 1,154, respectively). These regressions include controls for case scenario. AFB, acid-fast bacilli; MBBS, Bachelor of Medicine, Bachelor of Surgery; OR, odds ratio; TB, tuberculosis; Xpert MTB/RIF, Xpert Mycobacterium tuberculosis/Rifampicin, also known as GeneXpert.

Fig 4. Quality of care differences between SP case scenarios.

Estimated ORs between specific case scenarios for the frequency with which the indicated management action is observed. Panel A reports estimated ORs between Case 1 and Case 3, including only those providers who received both cases (N = 759 interactions). Panel B reports estimated ORs between Mumbai MBBS providers who received the experimental version of Case 4 that carried the same sputum report against a comparable sample who received the ordinary Case 4 presentation (as described in S1 Text; N = 101 interactions). AFB, acid-fast bacillus; MBBS, Bachelor of Medicine, Bachelor of Surgery; OR, odds ratio; SP, standardized patient; TB, tuberculosis; Xpert MTB/RIF, Xpert Mycobacterium tuberculosis/Rifampicin, also known as GeneXpert.