Performance of algorithms for tuberculosis active case finding in underserved high-prevalence settings in Cambodia: a cross-sectional study

Figures & data

Box 1. ACF preparation, procedures, and coordination.

ACF preparation and training

● The TB staff from the HCs were invited for a refresher training on smear preparation and treatment registration.

● Laboratory staff working in district hospitals were trained on the smear reading using LED microscopy.

● VHVs and the village leaders were invited for a training organized in the local HC. During the training the purpose of the ACF, as well as the ACF procedures were explained to the local VHVs and the village leaders. After the training the VHVs and the village leaders mapped the number and size of households in the community. They informed the community members about ACF, the planned arrival date of the MXU, and the venue where the MXU would be operating. People with TB symptoms were advised to come for CXR screening.

ACF procedures

● The MXU moved from village to village and stationed at central venues like schools or community center, used regularly for various public activities. In each community the MXU stayed for a few days.

● On the day the mobile team arrived in the community, a door-to-door TB symptom screening was done by the VHVs in collaboration with the TB officers. All individuals with presumptive TB were referred to the MXU for sputum collection and CXR.

● Two sputum samples were collected with an interval of half an hour and stored in a cool box. Clients unable to expectorate sputum were eligible for CXR. Moreover, when the SSM and/or Xpert came back negative, clients were traced to have a CXR as second diagnostic test. From 16/03/2015 onwards procedures changed. A CXR was taken the same day sputum samples were collected, and interpreted on the spot by the MXU physician, during a clinical consultation. Hence, the CXR result was available on the spot, in the community, and earlier than the SSM result.

Flow of samples and feedback-loop

● As different algorithms applied for different target groups, sputum samples were separated in different cool boxes. A cool box with sputum samples for smear microscopy was sent to the local HC where smears were prepared using glass slides. The same day the slides were sent for LED SSM reading at the district hospital’s laboratory. A cool box with sputum samples for Xpert testing was sent to the SHCH or CENAT laboratory.

● An external quality control of results from LED SSM and X-ray reading was conducted monthly.

● Results of the SSM or Xpert were reported by the laboratory staff to the ACF project coordinator and the TB officer using a short message system (SMS).

● When clients were diagnosed with TB, and not yet put on treatment by the MXU team, the TB officer contacted the VHVs or Village leaders to find the patient and refer them to the local HC for TB treatment initiation.

● A standardized first line TB treatment regimen was used at the local HC. Those who were diagnosed with rifampicin-resistant TB were referred to the CENAT to start the MDR TB treatment regimen.

Coordination

● TB officers met every two weeks to report problems and discuss solutions with the ACF project coordinator, who was responsible for the MXU staff, the preparation of ACF in the HC, and the tracking of patients lost to follow-up.

● The ACF Grant Director, the ACF Project Coordinator, met those responsible for monitoring and evaluation and laboratory testing each week to share updates and identify solutions for reported problems.

ACF: active case finding; CENAT: National Center for Tuberculosis and Leprosy Control; CXR: chest X-ray; HC: health center; MXU: Mobil X-ray unit; SHCH: Sihanouk Hospital Center of HOPE; SSM: sputum smear microscopy; TB: tuberculosis; VHV: Village Health Volunteer

Table 1. Algorithms used for active case finding in Cambodia, by target group, between 25 August 2014 and 31 March 2016.

Target group	Period	Algorithm^b	Sequence of CXR and sputum sampling and testing in patients with TB symptoms^c
Presumptive TB with additional risk factors: ‘high-risk group’^a	25/08/2014–15/03/2015	A Xpert/CXR	Xpert testing was the first diagnostic test. Those with a negative Xpert and for those who could not provide sputum, a second visit was done to conduct a CXR. This subgroup has a higher risk of developing TB and an increased risk of resistance to rifampicin.
	16/03/2015 − 14/07/2015	C CXR+Xpert	CXR and sputum collection were done on the same day to reduce the follow-up visits and to initiate more clients on the same day that they were tested in the community.
	15/07/2015–31/03/2016	D CXR+SSM/Xpert	In patients with TB symptoms CXR and sputum collection were done on the same day. When the CXR was abnormal or SSM was negative Xpert testing was done.
Presumptive TB without additional risk factors: ‘moderate-risk group’	25/08/2014–15/03/2015	B SSM/CXR	SSM was the first test. For those with a negative smear and for those who could not provide sputum, a second visit was made to conduct a CXR. To control costs this algorithm used SSM instead of Xpert.
	16/03/2015 and 31/03/2016	D CXR+SSM/Xpert	In patients with TB symptoms CXR and sputum collection were done on the same day. When the CXR was abnormal or SSM was negative Xpert testing was done.

CXR: chest X-ray; SSM: sputum smear microscopy; TB: tuberculosis; Xpert: Xpert MTB/RIF assay, HC: health center; MXU: Mobil X-ray unit

^aClients with presumptive TB were allocated in the ‘high-risk group’ if they had an additional ‘high-risk factor’, which included presumptive HIV infection (identification of opportunistic infections or self-reported previous HIV diagnosis), history of TB, and household contact with a TB patient. Clients with presumptive TB but without an additional risk-factor were grouped in the ‘moderate-risk group’.

^bTB diagnosis was based on bacteriological confirmation (SSM or Xpert) or clinical signs (physical examination, CXR)

^cUnder algorithm C and D, clients with TB signs on the CXR received a first dose of treatment from the MXU physician and were referred to the local HC to continue TB treatment. In such patients the sputum testing served to classify TB in BC PTB and CLIN PTB.

Table 2. Characteristics of clients tested for TB after a positive symptom screen in the community, between 25 August 2014 and 31 March 2016, in Cambodia.

	High-risk group^a		Moderate-risk group		Total
	N	%	N	%	N	%
Total tested	12,337	100.0	28,804	100.0	41,141	100.0
Male	4645	37.7	10,725	37.2	15,370	37.4
Age group:
15–24	615	5.0	1336	4.6	1951	4.7
25–49	3876	31.4	8705	30.2	12,581	30.6
≥50	7846	63.6	18,763	65.1	26,609	64.7
Symptoms or signs:
2 weeks fever	10,751	87.1	24,515	85.1	35,266	85.7
Shortness of breath	10,364	84.0	24,194	84.0	34,558	84.0
2 weeks cough	10,085	81.7	23,212	80.6	33,297	80.9
2 weeks night sweats	7382	59.8	16,090	55.9	23,472	57.1
2 weeks weight loss	3419	27.7	5791	20.1	9210	22.4
Lymph nodes	180	1.5	229	0.8	409	1.0
Presumptive HV	125	1.0	0	0.0	125	0.3
TB contact	8341	67.6	0	0.0	8341	20.3
TB history	5465	44.3	0	0.0	5465	13.3
Diabetes Mellitus	258	2.1	663	2.3	921	2.2
Malnutrition	2996	24.3	5213	18.1	8209	20.0
COPD	1033	8.4	861	3.0	1894	4.6
Algorithms: A: Xpert/CXR	2812	22.8	0	0.0	2812	6.8
B: SSM/CXR	0	0.0	4561	15.8	4561	11.1
C: CXR+Xpert	1824	14.8	0	0.0	1824	4.4
D: CXR+SSM/Xpert	7701	62.4	24,243	84.2	31,944	77.6

N: number; CXR: chest X-ray; SSM: sputum smear microscopy; TB: tuberculosis; COPD: chronic obstructive pulmonary disease

^aClients with presumptive TB were allocated in the ‘high-risk group’ if they had an additional ‘high-risk factor’, which included presumptive HIV infection (identification of opportunistic infections or self-reported previous HIV diagnosis), history of TB, and household contact with a TB patient. Clients with presumptive TB but without an additional risk-factor were grouped in the ‘moderate-risk group’.

Figure 1. Flowchart showing uptake of diagnostic algorithms during TB active case finding, between 25 August 2014 and 31 March 2016, in Cambodia.

TB: tuberculosis.

^a Clients with presumptive TB were allocated in the ‘high-risk group’ if they had an additional ‘high-risk factor’, which included presumptive HIV infection (identification of opportunistic infections or self-reported previous HIV diagnosis), history of TB, and household contact with a TB patient. Clients with presumptive TB but without an additional risk-factor were grouped in the ‘moderate-risk group’.

Table 3. Number needed to test (after a positive symptom screen) to diagnose one TB case, by algorithm, between 25 August 2014 and 31 March 2016, in Cambodia.

				Among those diagnosed: type of TB
	Total tested	Diagnosed with TB		BC PTB		CLIN PTB		EPTB
	N	N	(%)	N	(%)	N	(%)	N	(%)	NNT	95%CI
High-risk group^a
A: Xpert/CXR	2812	230	8.2	72	31.3	144	62.6	14	6.1	12.2	(10.8–13.9)
C: CXR+Xpert	1824	163	8.9	34	20.9	129	79.1	0	0.0	11.2	(9.6–13.0)
D: CXR+SSM/Xpert	7701	445	5.8	148	33.3	293	65.8	4	0.9	17.3	(15.9–18.9)
Moderate-risk group
B: SSM/CXR	4561	213	4.7	60	28.2	138	64.8	15	7.0	21.4	(18.9–24.4)
D: CXR+SSM/Xpert	24,243	846	4.8	367	31.5	784	67.2	13	1.3	20.8	(19.6–22.2)
Total	41,141	2217	5.4	681	30.7	1488	67.1	46	2.2	18.6	(17.9–19.2)

BC PTB: bacteriologically confirmed pulmonary TB; CLIN PTB: clinically diagnosed PTB; EPTB extra-pulmonary TB; Xpert: Xpert MTB/RIF assay; CXR: chest X-ray; SSM: sputum-smear microscopy; N: number; TB: tuberculosis; NNT: number needed to test using an algorithm to diagnose one case; CI: confidence interval

A forward slash ‘/’ shows that test are done sequentially, whereby the next test is done in those who test negative on the previous test. A plus sign “+“ shows that tests are done simultaneously.

^aClients with presumptive TB were allocated in the ‘high-risk group’ if they had an additional ‘high-risk factor’, which included presumptive HIV infection (identification of opportunistic infections or self-reported previous HIV diagnosis), history of TB, and household contact with a TB patient. Clients with presumptive TB but without an additional risk-factor were grouped in the ‘moderate-risk group’.

Table 4. Predictors of TB diagnosis, during active case finding in the community, between 25 August 2014 and 31 March 2016, in Cambodia.

	High-risk^a		Moderate-risk^a
	aOR	[95% CI]	aOR^§	[95% CI]
Sex
Female	1		1
Male	1.4***	[1.2,1.6]	1.7***	[1.6,2.0]
Age groups
15–24	0.8	[0.5,1.2]	0.7	[0.5,1.1]
25–49	1		1
≥50	1.6***	[1.4,1.9]	2.3***	[2.0,2.6]
Algorithms: A: Xpert/CXR	1.4***	[1.2,1.7]
B: SSM/CXR			0.9	[0.8,1.1]
C: CXR+Xpert	1.6***	[1.3,1.9]
D: CXR+SSM/Xpert	1		1

^aOR: adjusted odds ratio; CXR: chest X-ray; CI: confidence interval

* p < 0.05 **p < 0.01 *** p < 0.001

^§Adjusted for potential confounders, including gender, age, and TB symptoms (aOR for each of the TB symptoms not shown in the table)

^aClients with presumptive TB were allocated in the ‘high-risk group’ if they had an additional ‘high-risk factor’, which included presumptive HIV infection (identification of opportunistic infections or self-reported previous HIV diagnosis), history of TB, and household contact with a TB patient. Clients with presumptive TB but without an additional risk-factor were grouped in the ‘moderate-risk group’.

Table 5. Same-day treatment initiation and diagnostic lost to follow-up during active case finding in the community, between 25 August 2014 and 31 March 2016, in Cambodia.

	Total diagnosed	Same-day initiation		Diagnostic LTFU
	N	N	% (95% CI)	N	% [95%CI]
High-risk group^a
A: Xpert/CXR	230	10	4.3 (2.1–7.9)	10	4.3 [2.1–7.9]
C: CXR+Xpert	163	37	22.7 (16.5–29.9)	0	0.0 [0.0–2.2]
D: CXR+SSM/Xpert	445	144	32,4 (28.0–36.9)	8	1.8 [0.8–3.5]
Moderate-risk group
B: SSM/CXR	213	28	13.1 (8.9–18.4)	8	3.8 [1.6–7.3]
D: CXR+SSM/Xpert	1166	656	42.0 (39.2–44.9)	20	1.7 [1.1–2.6]
Total	2217	711	31.9 (30.0–33.9)	46	2.1 [1.5–2.8]

N: number; CXR: chest X-ray; SSM: sputum smear microscopy; TB: tuberculosis; LTFU: Loss to follow-up; CI: confidence interval

^aClients with presumptive TB were allocated in the ‘high-risk group’ if they had an additional ‘high-risk factor’, which included presumptive HIV infection (identification of opportunistic infections or self-reported previous HIV diagnosis), history of TB, and household contact with a TB patient. Clients with presumptive TB but without an additional risk-factor were grouped in the ‘moderate-risk group’.