Sputum pooling for rapid and cost-effective active case-finding for TB in prisons

Dear Editor,

In Brazil, tuberculosis (TB) incidence among persons deprived of liberty (PDL) is estimated at >2,000 per 100,000, or >40 times greater than incidence in the community^1–3. WHO recommends systematically screening PDL for TB at entry, annually, and upon release, as a minimum⁴. However, systematic screening is not widely implemented in prisons in low- and middle-income countries, even those with a high burden of TB, due to resource constraints. Previous studies have found that pooling sputum from multiple individuals and testing by molecular diagnostics maintains high sensitivity while reducing the number of assays and their associated costs⁵. However, the yield and efficiencies of real-world implementation of sputum pooling are less clear.

To address this knowledge gap, we conducted a prospective mass TB screening study in four male prisons in Campo Grande, Mato Grosso do Sul, Brazil from November 2021 to May 2022. All PDL were invited to participate in the study, and those who provided written informed consent were interviewed using a standardized questionnaire to collect demographic and clinical information. Study participants (regardless of their TB symptoms) were asked to produce at least 2 ml of spot sputum, which were then transported to the Central Public Health Laboratory of Mato Grosso do Sul (LACEN/MS) where Xpert^® MTB/RIF Ultra (Cepheid, Sunnyvale, CA, USA, hereinafter Xpert) tests were performed. We combined sputum into pools of eight and tested them using Xpert as previously described⁵. Briefly, we took 1 mL of sputum from eight study participants and mixed the pooled samples with Xpert reagent (1:1 ratio) using vortex mixer for 15 seconds. After incubation for 10 minutes at room temperature, we transferred 2 mL of mixed samples to Xpert cartridges for processing according to the manufacturer’s instructions. The original samples were tested individually for each positive pool to identify which individual(s) were Xpert positive. Individuals with negative pool Xpert results were considered to have negative individual Xpert results. As a check on the sensitivity of this study, we randomly selected 200 sputum samples with negative Xpert pool from the four study prisons to be tested individually.

The primary outcomes of the study were the diagnostic yield (i.e., the number of cases detected divided by the number of participants screened), the number of Xpert cartridges used, and the number of days required to complete screening in each prison. We additionally compared the number of cartridges utilized with a counterfactual of individual screening. The study was approved by the Research Ethics Committee of the Federal University of Mato Grosso do Sul (#5.082.910) and the National Research Ethics Committee of Brazil, Brasilia DF, Brazil (CONEP) (#5.383.839).

During the study period, 4,564 participants were screened, representing >95% of the PDL residing in those four prisons, in 61 working days. The median age was 32 years (interquartile range [IQR] 26 – 39). Nearly one-fifth of the participants (832/4,564; 18.2%) had at least one TB symptom. The proportion of individuals with a previous TB history was 15.3%. A total of 571 Xpert pool tests were performed, and 145 came back positive (Xpert pool positivity=25.3%, 95% confidence interval [CI] 21.8 – 28.9) (Table 1). After confirming the positive pools with individual testing (n=1,160), the total cartridges used was 1,731 (66% less than required for individual testing of all study participants). Overall, 128 study participants (2.8%; 95%CI 2.2 – 3.1) were found to have Xpert-confirmed TB. Prevalence ranged from 1.0% to 4.0% across the prisons. Overall, there was an average 13.5 cartridges used per TB case confirmed (range 11.0–20.2 across the prisons). An average of 75 participants were screened per day (range 59–89 across the prisons). The 200 pool-negative sputum samples selected for individual confirmation were all negative.

Table 1.

Screening characteristics, resource utilization, and results from the implementation of sputum pooling for systematic screening of tuberculosis, Mato Grosso do Sul, Brazil

	Prisons				TOTAL
	A	B	C	D
Total individuals screened	1,467	472	491	2,134	4,564
Pool tests performed	184	59	61	267	571
Positive pools, n (%)	47 (25.1)	5 (8.5)	5 (8.2)	88 (33.0)	145 (25.3)
Individual tests	376	40	40	704	1,160
Positive individual tests (confirmed tuberculosis)	30	5	5	88	128
tuberculosis prevalence, % (95% CI)	2.0 (1.4–2.9)	1.1 (0.4–2.3)	1.0 (0.4–2.2)	4.0 (3.0–4.7)	2.8 (2.2–3.1)
Total cartridges used	560	99	101	971	1,731
Cartridges saved* (%)	61.8 (%)	79.0 (%)	79.4 (%)	55.0 (%)	62.1 (%)
Cartridges used per case detected	18.7	19.8	20.2	11.0	13.5
Screening duration (days)	22	8	7	24	61

^*Cartridges saved were calculated as the percent reduction in cartridges compared to individual testing for all participants. Abbreviations: CI – Confidence interval, n - number

Sputum pooling allowed us to rapidly screen >4,500 PDL for TB in 61 working days, identifying 128 individuals with undiagnosed disease, while reducing Xpert cartridge use and personnel time. We used >60% fewer Xpert cartridges than individually testing all participants, and our previous experience with mass screening with individual testing took approximately 3 times as long, due to the limited number of modules available on Xpert machines locally. Our results showed that pooled sputum testing increases the efficiency of TB-active case-finding programs in this setting. A minority of prisons in Brazil have Xpert machines on site or locally accessible. Those that do typically perform fewer than 10–15 tests per day due to limited numbers of machines, as well as constraints due to the cost of cartridges. The sputum pooling method used in this study was effective in identifying TB cases while allowing us to conserve a substantial proportion of the screening time and cost. Although a recent systematic review reported that combining more than four samples resulted in lower sensitivity (i.e., 63 – 81% using Xpert MTB/RIF, and 80 – 100% using Xpert Ultra)⁶, our prior results found high sensitivity (100%) in pools of eight sputum, compared with individual testing, guiding our decision to use this pool size⁵. We found that the efficiency of screening and the relative efficiency of pooled testing varied with the TB prevalence across the prisons. During the study period, prisons A and D operated beyond their maximum capacity (i.e., occupancy rates of 457% and 332%, respectively), compared to the other two smaller prisons with 82% occupancy; the former pair of prisons had TB prevalence of 2 and 4%, compared with 1% in the latter. In the higher-burden prisons, the number of cartridges needed to identify a case was lower compared with the lower-burden prisons, but the relative efficiency of pooled testing was greater in the latter (79% versus 55–62% cartridges saved). Nevertheless, the results were qualitatively consistent across the prisons, showing acceptable numbers of cartridges needed to identify an individual with TB, and saving substantial resources compared with individual screening. Further studies evaluating the use of sputum pooling in combination with chest radiography, symptom screens (and other low-cost screening tools) may identify even more efficient algorithms.

As limitations of the study, only spot sputum samples were collected, with many being salivary quality, which may have led to an underestimation of the TB burden. Nevertheless, a lack of sputum induction capacity and variable quality sputum are the norm in many settings where active case finding is performed. Further, our prior studies demonstrated that even salivary samples have a substantial positivity in this high-burden populations⁷. Future studies should consider incorporating a cut-off of the last TB episode⁸, to minimize false Xpert positive issues, especially when resources for TB diagnosis are limited.

In conclusion, sputum pooling is an efficient strategy to increase an active case-finding capacity is settings with high TB prevalence, such as prisons. This enables screening to be performed more rapidly and with fewer resources than individual testing.