The Role of Psychosocial Interventions in Increasing Adherence to Tuberculosis Treatment in People Belonging to Socially Vulnerable Categories

Abstract

The article analyzes the effects of psychosocial interventions on adherence to tuberculosis (TB) treatment among vulnerable populations in Romania. The study includes 4104 patients from disadvantaged groups (rural, injecting drug users, homeless), beneficiaries of a national multidisciplinary support program. Multivariate analyses conducted on drug-susceptible TB (DS-TB) patients within this cohort identified some predictors of therapeutic success, such as extrapulmonary diagnosis, peer-to-peer educational support, and a higher level of education. At the same time, men, occupationally inactive people and those in the initial phase of treatment at project entry showed lower adherence. The results support the integration of psychosocial interventions in TB management.

1. Introduction

The WHO strategy for tuberculosis foresees a decrease in its incidence by 50% by 2025 compared to the level in 2015, when the global incidence of tuberculosis was 146 cases (IU: 133–160) per 100,000 inhabitants, and the eradication of the disease by 2035 [1]. After 2015, the rate of decline in tuberculosis was constant until 2020, followed by two consecutive years of global increases in TB incidence, which led to the tuberculosis incidence rate returning to the 2019 level in 2022. At the global level, the net relative reduction in the tuberculosis incidence rate from 2015 to 2022 was 8.7%, well behind the milestone of the WHO End TB strategy.

If we look at the WHO Europe region, however, the situation seems much better in terms of the number of cases, but alarm signals come from the area of multidrug-resistant tuberculosis because 20% of the resistance cases in the world are found in the WHO Europe region, and perhaps even more seriously, 37% of the cases have extensive resistance to antituberculosis drugs [1].

In Romania, tuberculosis is one of the priority public health problems, and the National Tuberculosis Control Strategy 2022–2033 is based on the directions and coordinates established by the World Health Organization’s End TB Strategy 2015 [2].

On the other hand, tuberculosis worldwide currently represents the leading cause of death among infectious diseases, but with a real downward trend in global mortality between 2000 and 2021, from 5th to 10th place, according to published data, a reduction of 23% in 2024, compared to the proposed objective in 2025 of 75%, a level that will probably not be reached [3].

The Global Strategy for Tuberculosis has three main pillars. The first involves an integrated approach to patients, from early diagnosis to treatment throughout the course of the disease, irrespective of the type of resistance, correct treatment of comorbidities, and TB prophylaxis in contacts. The second pillar focuses on the support system needed by patients, from public policy to civil society involvement, social protection, and changing the determinants that directly impact the TB patient. Last but not least, the third pillar targets strategies to increase research and innovation for tuberculosis eradication [4].

One of the international recommendations regarding the support of diagnosis and treatment for population groups at risk is active screening for tuberculosis. Beyond the primary objective of early detection of tuberculosis, tuberculosis screening also addresses social and economic aspects. Psychosocial support is an essential part of the care needed for tuberculosis patients, especially if they are part of risk groups.

Early detection is one of the essential elements of the strategy, passive detection of tuberculosis being a partially effective practice, especially in certain categories of the population where active detection plays an important role, not through statistical impact, but by reducing active infectious outbreaks [5]. More and more studies emphasize that active screening for tuberculosis increases the efficiency of case detection and surveillance in population categories exposed to the risk of abandonment, either as a result of the social determinants to which they are exposed or of the gender or age category to which they belong [6]. Gao states that active screening reduces the tuberculosis notification rate by about 7%, compared to 2.5% for regions where there is no active screening [7]. Another essential aspect is that related to detection, which is one day in active screening and an average of 30 days in passive detection [8].

Reducing the catastrophic cost that tuberculosis has on the patient’s family is also essential, its impact being major, especially among the disadvantaged population, where it manifests itself directly through the costs imposed by treatment and indirectly through the consequences imposed by the patient’s disease or social needs. Travel expenses, decreased income, absenteeism from work, stigma, and stress induced by the disease are barriers that, in the absence of appropriate support, decrease patient compliance with treatment and increase the risk of disease transmission in the community, multidrug-resistant tuberculosis, and death [9,10]. Beyond psychosocial determinants, a significant role in the occurrence of tuberculosis is played by known risk factors, which include malnutrition, alcohol consumption, smoking, HIV infection, and diabetes mellitus [11].

2. Materials and Methods

The present analysis aims to identify the effects of the application of psychosocial services provided by multidisciplinary teams on tuberculosis treatment outcomes among patients identified within disadvantaged populations. The results of this analysis may be useful in the development of models of care centered on these categories of patients.

Our analytical approach involved two main components. Firstly, to identify predictors of therapeutic success (defined as ‘cured’ or ‘treatment complete’), multivariate logistic regression models were applied exclusively to new cases of drug-susceptible TB (DS-TB). This focus was chosen to ensure a homogeneous group for the analysis of treatment success factors, mitigating confounding variables associated with the distinct and longer treatment regimens for drug-resistant TB (DR-TB). Secondly, comprehensive descriptive analyses were performed for the entire patient cohort, including both DS-TB and DR-TB patients. These descriptive analyses aimed to characterize all patient groups, detail the psychosocial support provided, and allow for comparisons of baseline characteristics, comorbidities, and observed treatment outcomes between DS-TB and DR-TB patients to understand the full spectrum of challenges within these vulnerable populations.

The study was conducted within the screening project “Organization of early detection (screening), diagnosis, and early treatment programs for tuberculosis, including latent tuberculosis,” carried out between 2018 and 2023. The psychosocial support activity was carried out by eight multidisciplinary teams, each present in a region of Romania (EMD), for patients diagnosed with TB. This program aimed to enhance early TB case detection in vulnerable populations across several regions of Romania.

Psychosocial support consisted of subsidies in the form of cash subsidies, peer-to-peer support, psychological counseling, social assistance services, and support means to increase their adherence to treatment—periodic reassessment to monitor adherence to treatment. As all patients in the targeted risk categories received mandatory psychosocial support services, these services were not included as variables to differentiate outcomes in the prediction models.

The inclusion criteria targeted TB patients from the following subcategories: people living in rural areas with difficult access to health services, people suffering from drug addiction, and homeless people.

The exclusion criteria targeted patients with an undiagnosed TB diagnosis who were not in the NTPNPCT computer system or could not be assessed, as well as persons who refused screening.

Drug-resistant TB is caused by Mycobacterium tuberculosis bacteria that are resistant to at least one of the most effective medicines used to treat TB—isoniazid or rifampicin. Drug susceptibility testing (DST) for tuberculosis (TB) was performed to ensure patients received effective treatment regimens. Initial screening for drug resistance involved genotypic DST using the GeneXpert MTB/RIF assay. This molecular method rapidly detects Mycobacterium tuberculosis (or M. tuberculosis if previously defined) and resistance to rifampin, serving as the initial DST for rapid results. Subsequently, phenotypic DST (growth-based methods) was conducted. This involved culturing M. tuberculosis isolated from patient specimens and then testing these isolates against a panel of first-line and second-line anti-TB drugs using standard methodologies, such as the proportion method on solid media (e.g., Löwenstein–Jensen) and/or automated liquid culture systems. While results from phenotypic DST typically take several weeks due to the slow growth of M. tuberculosis, these tests provide a broader assessment of drug resistance profiles.

In order to estimate the effects that various interventions had on treatment success, we conducted several multivariate analyses using the logistic regression method. Below we present some modeling with which we identified statistically significant predictors and their influence on the likelihood of treatment success—i.e., whether the patient is categorized as ‘cured’ or ‘treatment complete’ when it comes to the dependent variable treatment assessment. The definition of the treatment assessment category was based on the recommendations of the Methodological Guidelines for the Implementation of the National Program for Prevention, Surveillance, and Control of Tuberculosis and Other Mycobacterial Diseases 2023 [12].

For the level of education stratification, the International Standard Classification of Education (ISCED) classification, revised in 2011, according to which there are nine levels of education, from level 0 to level 8, was used [13].

Since almost all patients received psychosocial support services, it was not possible to include as predictor variables such as “patient received service A, B, C” to see whether receiving/not receiving a service makes a difference between treatment success and treatment failure.

The analysis focused exclusively on the new case sample of TB-DS patients. Evaluation criteria excluded patients with MDR-TB, as the duration of treatment for some of them may exceed the period analyzed, so that cases with continuing treatment may only be cases that have not yet completed the full treatment regimen. We note that, at the time of the study, the duration of MDR-TB treatment was, on average, 20 months. The models used included variables related to socio-demographic characteristics (sex, age, education level, employment status), diagnosis (pulmonary or extrapulmonary), treatment phase, and psychosocial support interventions. The models were evaluated using statistical goodness-of-fit measures, including the R² coefficient and the Hosmer–Lemeshow test, to ensure the validity and reliability of the results.

Given the complexity of the phenomenon studied, in the case of logistic regression, the values obtained for the Cox & Snell R² and Nagelkerke R² (0.024, 0.091, 0.064) reflect a moderate explanatory power. However, the values obtained can be considered typical in social and medical research, where the variables included in the models cannot cover the entire complexity of the phenomenon being treated. In the present context, in which the logistic model includes socio-economic factors and psychosocial interventions, these values are to be expected and suggest that there is still significant variability in the success of the treatment that is not explained solely by the variables included in the analysis.

Three models were applied:

• Model 1: Dependent variable: treatment assessment (cured/complete treatment). Independent variables (predictors): sex, age, education level, labor market status.
• Model 2: Dependent variable: Treatment evaluation (cured/complete treatment). Independent variables (predictors): Diagnosis (P—pulmonary; E—extrapulmonary), 1 (T0–T2), 2 (T3–T6), 3 (>T6), 4 (N/A and unconfirmed), medical services.
• Model 3: Dependent variable: treatment assessment (cured/complete treatment). Independent variables (predictors): adherence monitoring, subsidies/packages, peer educator support, psychological support, social assistance, NGO support services.

These values indicate that the models fit the observed data well, with no signs of significant misfit. In particular, for Model 3, the p-value of 0.967 suggests that the model fits the data extremely well, which is a positive sign of its validity.

The results of the Hosmer and Lemeshow test showed a good fit of the models to the observed data (p_value > 0.05), suggesting that there are no significant disproportions between the model predictions and the data reality, which confirms that the models fit the observations adequately and are not affected by major specification errors. Of note is Model 3, for which p_value = 0.967, suggesting that the model fits the data extremely well, which is a positive sign of its validity.

3. Results

A total of 4157 patients were included in the study from December 2019 to December 2022, of which 51 cases were excluded because, for 15 cases, the diagnosis was invalidated, 15 cases were not in the NTPNPCT electronic database, 20 cases were lost to follow-up, and 1 case was transferred to the penitentiary; all of them were in the TB-DS category.

Of the 4101 patients, 3884 (94.7%) are patients diagnosed with susceptible tuberculosis (SD-TB) and 217 (5.3%) patients with resistant tuberculosis (DR-TB). Of the 3884 TB-DS cases, 3148 (81.05%) are new cases, 645 (16.6%) are relapses, 90 (2.25%) are in the failure, drop-out, or chronic category, and 4 (0.1%) represent transfers from other counties. Of the 217 resistant TB cases, 179 (82.5%) represent DR-TB, and the rest XDR-TB. Drug-resistant tuberculosis is resistant to at least one major anti-TB drug, isoniazid or rifampicin.

3.1. Socio-Demographic Characteristics

Of the total of 3884 patients with TB-DS, 3858 (99.33%) were classified in the following social risk categories: 3858 (99.33%) rural people, 10 (0.25%) people suffering from drug dependence (DD), and 16 (0.41%) homeless people (H). Seventy-five percent of the TB-DS patients are male, most of them in the age category 45–54 years, with ISCED 2 (International Standard Classification of Education) education level, i.e., secondary school. More than half of the selected patients did not have a job at the date of entry into the project; only 19% are employed, and the difference in these two categories is people living on social assistance, pensions, or disability benefits (

Table 1. Structure of TB-DS patients analyzed according to socio-demographic characteristics and level of education by vulnerable groups (VGs).

TB-DS	RP	DD	H	Total
	3858	10	16	3884
Sex
Male	75.0%	70.0%	87.5%	75.0%
Female	25.0%	30.0%	12.5%	25.0%
Age
15–19 years	1.8%			1.8%
20–24 years	5.2%		18.8%	5.2%
25–34 years	13.1%	70.0%	6.3%	13.2%
35–44 years	19.2%	30.0%	12.5%	19.2%
45–54 years	29.1%		18.8%	29.0%
55–64 years	16.6%		37.5%	16.7%
65–74 years	9.4%		6.3%	9.4%
Over 75 years	5.7%			5.6%
Education level
ISCED 0	2.3%		6.3%	2.3%
ISCED 1	15.4%	40.0%	6.3%	15.4%
ISCED 2	48.1%	50.0%	56.3%	48.1%
ISCED 3	28.4%	10.0%	18.8%	28.3%
ISCED 4	2.6%			2.6%
ISCED 5	1.5%			1.5%
ISCED 6	0.2%			0.2%
ISCED 7	0.1%			0.1%
ISCED 8	0.2%			0.2%
None	0.7%		6.3%	0.7%
N/A	0.6%		6.3%	0.6%
Occupation
Social worker occupation	0.8%			0.8%
Other occupation/personal assistant	4.2%			4.2%
School/student	0.4%			0.4%
Inactive	55.2%	90.0%	93.8%	55.5%
Disabled	0.4%		6.3%	0.5%
Pensioner	17.7%	10.0%		17.7%
Employee	19.0%			18.8%
Unemployed	2.2%			2.2%
N/A	0.1%			0.1%

).

Of the 217 DR-TB patients, 199 (91.71%) were rural people (RP), 12 (5.53%) were people suffering from drug dependence (DD), and 6 (2.76%) were homeless (H). The proportion of males is also preserved in the DR-TB group, with 77% males exhibiting similar characteristics to the DS-TB group in terms of age and level of education. However, it should be mentioned that in the subgroup of injecting drug users with TB, 50% are unemployed, 25% are retired, and 16% are employed (

Table 2. Structure of DR-TB patients analyzed according to socio-demographic characteristics and education level by vulnerable groups.

TB-DR	RP	DD	H	Total
	199	12	6	217
Sex
Male	75.9%	83.3%	100.0%	77.0%
Female	24.1%	16.7%		23.0%
Age
15–19 years	2.0%			1.8%
20–24 years	2.5%			2.3%
25–34 years	13.1%	50.0%	16.7%	15.2%
35–44 years	21.6%	33.3%	16.7%	22.1%
45–54 years	33.7%	16.7%	33.3%	32.7%
55–64 years	20.1%		33.3%	19.4%
65–74 years	5.5%			5.1%
Over 75 years	1.5%			1.4%
Education level
ISCED 0	2.0%			1.8%
ISCED 1	11.1%	25.0%	16.7%	12.0%
ISCED 2	45.2%	50.0%		44.2%
ISCED 3	34.2%	16.7%	83.3%	34.6%
ISCED 4	2.0%			1.8%
ISCED 5	3.5%			3.2%
ISCED 6	0.5%			0.5%
None	1.0%	8.3%		1.4%
N/A (+ blank)	0.5%			0.5%
Occupation
Social worker	6.5%			6.0%
Other occupation/personal assistant	1.0%			0.9%
School/student	48.7%	50.0%	50.0%	48.9%
Disabled		8.3%		0.5%
Pensioner	24.6%	25.0%	16.7%	24.4%
Employee	18.1%	16.7%	33.3%	18.4%
Unemployed	1.0%	0.0%	0.0%	0.9%

).

3.2. Comorbidities

Information about patients’ chronic conditions and addictions should be treated with caution, as it is self-reported, not extracted from their medical records. It is possible that some characteristics may be under-reported due to stigma (e.g., lack of knowledge of one’s own diagnosis). At the same time, there may be cases of over-reporting of mental health disorders without a psychiatric diagnosis or actually hiding a neurological diagnosis.

Smoking is the most common addiction among all groups analyzed, followed by alcohol consumption. Drug use is reported by both injecting drug users and homeless people. Mental health disorders are reported more than diabetes (

Table 3. Distribution of patients by diseases and addictions by vulnerable groups and type of DR-TB and DR-TB.

	RP		DD		H
	TB-DS	TB-DR	TB-DS	TB-DR	TB-DS	TB-DR
	3861	199	10	12	16	6
Chronic diseases
Mental health disorders	9.5%	12.1%	10.0%	8.3%	6.3%	33.3%
Diabetes	2.7%	0.1%				0.0%
Other addictions
Drug use			100.0%	66.7%	6.3%	16.7%
Injecting drug use			100.0%	66.7%
Drug use					6.3%	16.7%
Alcohol use	10.2%	11.1%	0.0%	0.0%	25.0%	50.0%
Problematic alcohol use	2.9%	4.5%		8.3%	25.0%	33.3%
Alcohol use (unspecified)	6.5%	5.5%
Alcohol use (occasional)	0.8%	1.0%				16.7%
Smoking	18.4%	10.1%			12.5%	33.3%
Other (unspecified)	2.7%	1.5%		33.3%

).

3.3. Characteristics Reported at the Time of Inclusion in the Psychosocial Support Program

In an analysis of the relationship between treatment phase at project entry and treatment outcome, it appears that treatment success is more prevalent among patients—with TB-DS or DR-TB—who, at project entry, were at least in the sixth month of treatment (T6). Loss to follow-up is more frequent among DR-TB patients enrolled in the intensive phase of treatment, and deaths are more common among DR-TB patients who entered the project at T3–T6 (

Table 4. Structure of the sample of TB-DR and TB-DS patients by treatment assessment and treatment phase.

TB (1–6 TB-DS, 7–8 TB-DR)			Treatment Phase				Total
			(T0–T2)	(T3–T6)	(>T6)	N/A
N			2434	1468	177	25	4104
TB-DR (N1 = 217)	Treatment evaluation	Lost to follow-up	3.8%	2.7%	2.0%		3.2%
		Continuous treatment	42.7%	35.1%	26.5%		37.8%
		Death	10.7%	16.2%	8.2%		11.1%
		Treatment failed		2.7%			0.5%
		Treatment completed	7.6%	10.8%	8.2%		8.3%
		Cured	35.1%	32.4%	55.1%		39.2%
	N1		131	37	49		217
TB-DS (N2 = 3188)	Treatment evaluation	Lost to follow-up	2.3%	1.1%	1.6%	4.0%	1.9%
		Continuous treatment	28.7%	12.9%	3.9%	28.0%	22.1%
		Death	4.6%	3.5%	2.3%	8.0%	4.1%
		Treatment failed	0.3%	0.6%			0.4%
		Treatment completed	17.0%	17.5%	23.4%	12.0%	17.4%
		Cured	47.1%	64.3%	68.8%	48.0%	54.1%
	N2		2303	1431	128	25	3887

).

Considering patients in the Intensive Treatment Phase (T0–T2), the data show that among those with TB-DS, the proportion of patients declared cured is 21 percentage points higher than among DR-TB patients (64% vs. 43%).

Considering patients in Phase I of treatment continuation (T3–T6), the data show that the proportion of patients declared cured is 39 percentage points higher among those with TB-DS than among those with DR-TB (82% vs. 43%), while for patients in Phase II continuation (>T6), the data show the proportion of patients declared cured is 29 percentage points higher among patients with TB-DS than among those with DR-TB (92% vs. 63%, 4).

In terms of treatment success and belonging to one of the vulnerable groups, it can be observed that the highest cure rate is recorded in rural patients, and the structure is similar to the one described previously [3]. Considering rural patients, the data show that the proportion of cured patients with TB-DS is 22 percentage points higher than that of TB-DR patients (72% vs. 50%). In H, the proportion of cured patients with TB-DS is 69%, while the proportion of cured patients with TB-DR is 33%. DD has the lowest treatment success rate, with the proportion of cured patients with curable TB being 2 percentage points lower than the proportion of cured TB-DR patients (10%, compared to 8%, 4).

While the proportion of rural and homeless participants is approximately the same, the proportion of DD participants who are cured is below 5%. This category also has the highest dropout rate among the three categories of participants—22.7% DD, compared to 13.6% for H and 1.7% for RP.

In the case of patients with newly notified cases of TB-DS (classification code 1) (

Table 5. Sample structure of patients with TB-DS/TB-DR according to GV and treatment evaluation.

TB (1–6 TB-DS, 7–8 TB-DR)			Target Group Placement			Total
			DD	RP	H
N			22	4060	22	4104
TB-DR (N1 = 217)	Treatment evaluation	Lost to follow-up	16.7%	2.5%		3.2%
		Continuous treatment	25.0%	38.2%	50.0%	37.8%
		Death	50.0%	8.5%	16.7%	11.1%
		Treatment failed		0.5%		0.5%
		Treatment completed	8.3%	8.5%		8.3%
		Cured		41.7%	33.3%	39.2%
	N1		12	199	6	217
TB-DS (N2 = 3188)	Treatment evaluation	Lost to follow-up	30.0%	1.7%	18.8%	1.9%
		Continuous treatment	40.0%	22.1%	12.5%	22.1%
		Death	20.0%	4.1%		4.1%
		Treatment failed		0.4%		0.4%
		Treatment completed		17.4%	12.5%	17.4%
		Cured	10.0%	54.2%	56.3%	54.1%
	N2		10	3861	16	3887

), according to treatment assessment and treatment phase, the proportion of patients declared cured is over 65% regardless of the treatment phase in which the patients were at the time of entry into the project (Appendix A,

Table A1. Patient sample structure—new cases notified (classification code 1), by TB type, treatment assessment, and treatment phase.

TB-DS Code 1—New Cases		Treatment Phase				Total
		(T0–T2)	(T3–T6)	(>T6)	N/A
		1944	1108	78	18	3148
Treatment evaluation	Lost to follow-up	2.0%	0.9%		5.6%	1.6%
	Continuous treatment	28.4%	12.5%	1.3%	33.3%	22.2%
	Death	4.1%	2.7%	1.3%	5.6%	3.5%
	Treatment failed	0.4%	0.2%			0.3%
	Treatment completed	17.4%	19.1%	24.4%	16.7%	18.2%
	Cured	47.7%	64.5%	73.1%	38.9%	54.2%

). Depending on the VG to which they belong, while the structure is similar to that at the cumulative level of the three VGs in the case of rural patients, a different structure is observed in the case of the H and DD groups. This may also be due to the low number of patients in each of the two VGs (Appendix A,

Table A2. Structure of the sample of patients from rural areas—new cases notified (classification code 1), by TB type, treatment assessment, and treatment phase.

TB-DS—Rural Code 1—New Cases Notified		Treatment Phase				Total
		(T0–T2)	(T3–T6)	(>T6)	N/A
N		1934	1107	77	18	3136
Treatment evaluation	Lost to follow-up	1.9%	0.9%		5.6%	1.5%
	Continuous treatment	28.4%	12.6%	1.3%	33.3%	22.2%
	Death	4.1%	2.7%	1.3%	5.6%	3.5%
	Treatment failed	0.4%	0.2%			0.3%
	Treatment completed	17.5%	19.2%	24.7%	16.7%	18.3%
	Cured	47.7%	64.5%	72.7%	38.9%	54.2%

and

Table A3. Sample structure of PAFA and CDI patients—new cases notified (classification code 1), by TB type, treatment assessment, and treatment phase.

TB-DS Code 1—New Cases		PAFA				CDI
		Treatment Phase			Total	Treatment Phase	Total
		(T0–T2)	(T3–T6)	(>T6)		(T0–T2)
N		8	1	1	10	2	2
Treatment evaluation	Lost to follow-up	12.5%			10.0%	50.0%	50.0%
	Treatment continuation	25.0%			20.0%	50.0%	50.0%
	Cured	62.5%	100.0%	100.0%	70.0%

).

Regarding the influence of educational level on the treatment success rate, it can be observed that the highest proportion of patients declared cured is among persons with a PhD or equivalent level (87.5%, 5).

Regarding the influence of the level of education on the treatment success rate, it can be observed that the highest share of patients declared cured is among people with a PhD or equivalent level (87.5%,

Table 6. Structure of the sample according to type of TB and assessment of adherence to treatment according to level of education.

TB (1–6 TB-DS, 7–8 TB-DR)			Education Level											Total
			N/A	NONE	0	1	2	3	4	5	6	7	8
			26	29	94	625	1966	1174	103	65	9	5	8	4104
TB-DR (N1 = 217)	Treatment evaluation	Lost to follow-up					3.1%	5.3%						3.2%
		Continuous treatment	100.0%		100.0%	30.8%	39.6%	38.7%	25.0%	14.3%				37.8%
		Death		33.3%		19.2%	13.5%	5.3%		14.3%				11.1%
		Treatment failed					1.0%							0.5%
		Treatment completed				7.7%	3.1%	12.0%	50.0%	28.6%				8.3%
		Cured		66.7%		42.3%	39.6%	38.7%	25.0%	42.9%	100.0%			39.2%
	N1		1	3	4	26	96	75	4	7	1			217
TB-DS (N2 = 3188)	Treatment evaluation	Lost to follow-up		3.8%	1.1%	2.8%	2.1%	0.8%	4.0%	1.7%				1.9%
		Continuous treatment	36.0%	3.8%	34.4%	24.9%	21.8%	18.9%	30.3%	29.3%	25.0%	60.0%		22.1%
		Death	8.0%	3.8%	5.6%	6.5%	4.0%	3.1%	4.0%				12.5%	4.1%
		Treatment failed		3.8%		0.5%	0.4%	0.5%						0.4%
		Treatment completed	12.0%	19.2%	12.2%	16.9%	16.3%	19.4%	21.2%	22.4%	12.5%		37.5%	17.4%
		Cured	44.0%	65.4%	46.7%	48.4%	55.4%	57.3%	40.4%	46.6%	62.5%	40.0%	50.0%	54.1%
	N2		25	26	90	599	1870	1099	99	58	8	5	8	3887

). Regarding the cases of dropout for TB-DR patients, the share of cases of dropout is 2.3% in the case of secondary-school-educated patients and 2.7% in the case of high-school-educated patients [5], these being the categories with the highest share in the sample (48% secondary-school-educated patients and 29% high-school-educated patients out of the total). In the case of patients with TB-DS, dropout cases are observed even in the category of undergraduate-educated patients [5]. It is noteworthy that patients declared cured have the highest share in the sample, regardless of the type of TB they were diagnosed with (71.5% for patients with TB-DS, 47.5% for patients with TB-DR), followed by those who continue treatment (22.1% for patients with TB-DS, 37.8% for patients with TB-DR, 6).

In the case of patients with TB-DS, for newly notified cases (classification code 1), according to the treatment evaluation and the level of studies, the proportion of patients declared cured is over 50% regardless of the treatment phase in which the patients were at the time of entering the project (Appendix A,

Table A4. Structure of the patient sample—new cases notified (classification code 1), by TB type, treatment assessment, and level of studies.

TB-DS Code 1—New Cases		Education Level (ISCED)											Total
		N/A	None	0	1	2	3	4	5	6	7	8
N		20	17	82	480	1469	923	87	52	7	4	7	3148
Treatment evaluation	Lost to follow-up			1.2%	2.3%	2.0%	0.5%	2.3%	1.9%				1.6%
	Continuous treatment	35.0%		34.1%	25.4%	21.7%	19.5%	28.7%	26.9%	28.6%	50.0%		22.2%
	Death		5.9%	4.9%	5.2%	3.5%	2.6%	4.6%				14.3%	3.5%
	Treatment failed				0.2%	0.4%	0.2%						0.3%
	Treatment completed	15.0%	29.4%	13.4%	17.7%	17.0%	19.9%	21.8%	23.1%	14.3%		42.9%	18.2%
	Cured	50.0%	64.7%	46.3%	49.2%	55.3%	57.2%	42.5%	48.1%	57.1%	50.0%	42.9%	54.2%

). According to the VG to which they belong, while the structure is similar to that at the cumulative level in the case of rural patients, a different structure is observed in the case of DD and H. This may also be due to the small number of patients in each of the two VGs (Appendix A,

Table A5. Structure of the sample of patients from rural areas—new cases notified (classification code 1), according to TB type, treatment assessment, and level of education.

TB–DS—Rural Code 1—New Cases		Education Level (ISCED)											Total
		N/A	None	0	1	2	3	4	5	6	7	8
		19	17	81	478	1464	920	87	52	7	4	7	3148
Treatment evaluation	Lost to follow-up			1.2%	2.1%	2.0%	0.5%	2.3%	1.9%				1.5%
	Continuous treatment	36.8%		33.3%	25.3%	21.7%	19.6%	28.7%	26.9%	28.6%	50.0%		22.2%
	Death		5.9%	4.9%	5.2%	3.6%	2.6%	4.6%				14.3%	3.5%
	Treatment failed				0.2%	0.4%	0.2%						0.3%
	Treatment completed	15.8%	29.4%	13.6%	17.8%	17.1%	20.0%	21.8%	23.1%	14.3%	0.0%	42.9%	18.3%
	Cured	47.4%	64.7%	46.9%	49.4%	55.3%	57.1%	42.5%	48.1%	57.1%	50.0%	42.9%	54.2%

and

Table A6. Structure of the PAFA and CDI patient sample—new cases notified (classification code 1), by TB type, treatment assessment, and education level.

TB-DS Code 1—New Cases		PAFA						CDI
		Education Level (ISCED)						Education Level (ISCED)
		N/A	0	1	2	3	Total	1	2	Total
N		1	1	1	4	3	10	1	1	2
Treatment evaluation	Lost to follow-up			100%			10%		100%	50%
	Continuous treatment		100%	0	25%		20%
	Cured	100%			75%	100%	70%	100%		50%

).

Considering three categories that can characterize the patient’s status in the labor market: employed, inactive, and unemployed, the data show that among those with TB-DS, the share of employed patients declared cured is 19.7 percentage points higher than among employed TB-DR patients (76.3% vs. 56.5%). The largest difference, by TB type, for patients declared cured is in the inactive category (70.8.1% for patients with TB-DS and 44.8% for TB-DR patients,

Table 7. Sample structure according to TB type, treatment evaluation, and labor market status (LM).

TB (1–6 TB-DS, 7–8 TB-DR)			Labor Market Status				Total
			N/A	Employee	Inactive	Unemployed
N			26	884	3011	183	4104
TB-DR (N1 = 217)	Treatment evaluation	Lost to follow-up		4.3%	3.1%		3.2%
		Continuous treatment	100.0%	37.0%	37.4%	42.9%	37.8%
		Death		2.2%	14.1%		11.1%
		Treatment failed			0.6%		0.5%
		Treatment completed		2.2%	10.4%		8.3%
		Cured		54.3%	34.4%	57.1%	39.2%
	N1		1	46	163	7	217
TB-DS (N2 = 3887)	Treatment evaluation	Lost to follow-up		0.8%	2.0%	4.0%	1.9%
		Continuous treatment	36.0%	20.2%	22.0%	30.7%	22.1%
		Death	8.0%	2.4%	4.7%	2.8%	4.1%
		Treatment failed		0.4%	0.5%	0.6%	0.4%
		Treatment completed	12.0%	17.7%	17.6%	13.6%	17.4%
		Cured	44.0%	58.6%	53.3%	48.3%	54.1%
	N2		25	838	2848	176	3887

).

In the case of patients with newly notified cases of TB (classification code 1), according to treatment evaluation and labor market status, the proportion of patients declared cured is over 60% regardless of labor market status (Appendix A,

Table A7. Structure of the patient sample—new notified cases (classification code 1), by TB type, treatment assessment, and labor market status.

TB-DS Code 1—New Cases		Labor Place Status				Total
			Employee	Inactive	Unemployed
N		20	714	2286	128	3418
Treatment evaluation	Lost to follow-up		0.8%	1.7%	3.1%	1.6%
	Continuous treatment	35.0%	20.2%	22.3%	30.5%	22.2%
	Death		1.8%	4.2%	2.3%	3.5%
	Treatment failed		0.3%	0.3%	0.8%	0.3%
	Treatment completed	15.0%	18.6%	18.4%	12.5%	18.2%
	Cured	50.0%	58.3%	53.1%	50.8%	54.2%

). While in the case of rural patients, the structure is similar to that of the three GVs, in the case of patients from the other two GVs (PAFA and CDA) it can be observed that none of them is active in the labor market (Appendix A,

Table A8. Structure of the sample of patients from rural areas—new cases notified (classification code 1), according to TB type, treatment assessment, and labor market status.

TB–DS—Rural Code 1—New Cases		Labor Place Status				Total
			Employee	Inactive	Unemployed
N		20	714	2286	128	3418
Treatment evaluation	Lost to follow-up		0.8%	1.7%	3.1%	1.5%
	Continuous treatment	36.8%	20.2%	22.2%	30.5%	22.2%
	Death		1.8%	4.2%	2.3%	3.5%
	Treatment failed		0.3%	0.3%	0.8%	0.3%
	Treatment completed	15.8%	18.6%	18.5%	12.5%	18.3%
	Cured	47.4%	58.3%	53.1%	50.8%	54.2%

and

Table A9. Structure of the PAFA and CDI patient sample—new cases notified (classification code 1), by TB type, treatment assessment, and labor market status.

TB–DS Code 1—New Cases		PAFA			CDI
		Labor Place Status		Total	Labor Place Status	Total
		N/A	Inactive		Inactive
N		1	9	10	2	2
Treatment evaluation	Lost to follow-up		11.1%	10.0%	50.0%	50.0%
	Continuous treatment		22.2%	20.0%	50.0%	50.0%
	Cured	100.0%	66.7%	70.0%

).

4. Therapeutic Success Is a Condition of Psychosocial Support

Of the 4104 patients, 99.78% were monitored to assess treatment adherence, with 98.95% benefiting from financial support. Psychological and social assistance was offered to 90.64% and 91.01% of patients, respectively. Peer-to-peer support activities were provided to 14.38% of patients, and other community support activities to 3.24% (

Table 8. Proportion of support services provided to tuberculosis patients in the study.

Support Services Provided		(% From N = 4104)
Adherence monitoring	4095	99.78%
Financial support	4061	98.95%
Peer-to-peer support	590	14.38%
Psychological support	3720	90.64%
Social assistance	3735	91.01%
Community support services	133	3.24%

).

The applied logistic regression models revealed that the patient’s age has a positive effect on the chances of the patient being declared cured at the end of the analyzed period, keeping under control all other factors included in the analysis. In addition, the chances increase by almost 300% if the patient is diagnosed with extrapulmonary TB. The chances of treatment success decrease if the patient is male, unemployed, or inactive in the labor market. The overall performance of the models was assessed at approximately 70–73% accuracy, indicating a decent ability to predict treatment success. These classification percentages are relatively good, considering that most of the patients in the study are from vulnerable groups, and treatment success depends not only on the variables included in the models, but also on factors that were not taken into account (comorbidities, personal adherence to treatment, individual psychological factors). The results suggest that the models have a reasonable performance considering the diversity of variables involved and the complexity of the patients studied and can also be used to make predictions about the success of anti-TB treatment in the populations studied. Exp (B) coefficients are used to assess the influence of each independent variable on the probability of the patient being declared “cured” or having “complete treatment”, indicating how changes in an independent variable affect the chances of treatment success.

The chance of therapeutic success is higher if inclusion in the psychosocial support program occurs in the first 2 months of treatment, compared to outpatient therapy. The lowest risk of dropout occurs in the T0–T2 phase of treatment or outpatient treatment if the patients received subsidies or psychosocial support. Exp (B) coefficients indicate the probability of a change in the dependent variable resulting from a one-unit change in the independent variable. When Exp (B) has a sub-unit value, a one-unit increase in the independent variable leads to a decrease in the probability that the dependent event (treatment success) will occur. Coefficients marked with * are statistically significant, for p = 0.05(

Table 9. Predictors of success in anti-TB treatment. Coefficients marked with * are statistically significant, for p = 0.05.

Predictors of Success in Anti-TB Treatment (Patients with Susceptible TB Declared Cured or with Complete Treatment)	Exp (B)
Male patient	0.843 *
Age at time of project registration	1.007 *
Labor market ISCED 2	0.555 *
Labor market ISCED 3	0.691 *
Diagnosis of extrapulmonary TB	3.136 *
Treatment in phase T0–T2 at time of project registration	0.42 *
Outpatient treatment at time of project registration	0.821 *
Received financial support	0.182 *
Received peer-to-peer support	3.742 *
Received support in special situation	0.807 *

).

Although the analytical models employed indicate that R² and p-values suggest moderate explanatory power and a good fit with the observed data, the results provide a robust foundation for interpreting the key factors influencing treatment success—such as the type of diagnosis (pulmonary or extrapulmonary), patient age, labor market status, and the implementation of psychosocial support interventions. The logistic regression models applied yield valuable insights into the determinants of antituberculosis (TB) treatment outcomes among vulnerable populations. These findings further substantiate the importance of psychosocial support and integrated treatment management, highlighting their potential to contribute significantly to the development of more effective and personalized public health policies targeting disadvantaged groups.

Moreover, the results underscore the critical role of psychosocial interventions in the management of anti-TB treatment, particularly for vulnerable subpopulations, and point to the necessity of public policies that incorporate tailored approaches for groups such as individuals experiencing homelessness or people who inject drugs (

Table A10. Model 1—Dependent variable: treatment evaluation (cured/continuing treatment)—1 = yes; 0 = no. Independent variables: gender, age, education level, labor market status.

Model Summary
Step				−2 Log Likelihood				Cox & Snell R-Squared				Nagelkerke R-Squared
1				4928.013 a				0.017				0.024
a. Estimation terminated at iteration number 5 because parameter estimates changed by less than 0.001.
Hosmer and Lemeshow Test
Step						Chi-Squared				df					Sig.
1						13.082				8					0.109
Classification Table
Observed								Predicted
								Treatment evaluation (cured/continued treatment)									Percentage Correct
								1						0
Step 1			Treatment evaluation (cured/continued treatment)					1	2874					8			99.7
								0	1211					11			0.9
			Overall Percentage														70.3
a. The cut value is 0.500
Variables in the Equation
		B			S.E.		Wald	df	Sig.		Exp(B)		95% C.I. for EXP(B)
													Lower			Upper
Step 1a	Gender	−0.171			0.082		4.324	1	0.038		0.843		0.718			0.990
	Age	0.007			0.002		9.189	1	0.002		1.007		1.003			1.012
	Level of study						33.597	10	0.000
	Level of study (1)	1.275			1.153		1.223	1	0.269		3.579		0.373			34.308
	Level of study (2)	0.251			1.180		0.045	1	0.831		1.286		0.127			12.984
	Level of study (3)	1.548			1.092		2.009	1	0.156		4.704		0.553			40.011
	Level of study (4)	1.135			1.076		1.114	1	0.291		3.112		0.378			25.629
	Level of study (5)	0.919			1.074		0.733	1	0.392		2.507		0.306			20.557
	Level of study (6)	0.741			1.074		0.475	1	0.491		2.097		0.255			17.223
	Level of study (7)	1.393			1.091		1.629	1	0.202		4.027		0.474			34.198
	Level of study (8)	1.184			1.105		1.148	1	0.284		3.269		0.374			28.534
	Level of study (9)	0.745			1.340		0.309	1	0.578		2.107		0.152			29.154
	Level of study (10)	2.337			1.416		2.721	1	0.099		10.347		0.644			166.150
	Labor Market Status						12.675	2	0.002
	Labor Market Status (2)	−0.589			0.173		11.540	1	0.001		0.555		0.395			0.779
	Labor Market Status (3)	−0.370			0.159		5.382	1	0.020		0.691		0.506			0.944
	Constant	−1.703			1.088		2.450	1	0.118		0.182
a. Variables entered in step 1: Gender, Age, Level of study, Labor Market Satatus

).

5. Discussion

A patient with tuberculosis must face not only the difficulties posed by symptoms and treatment but also the social challenges brought on by stigma, lack of knowledge about the disease, and the economic burden felt by both the patient and their family. As a result, managing tuberculosis requires a multidisciplinary approach, involving teams with specialists from various fields [14]. A 2023 meta-analysis [15] evaluates the psychosocial support provided across multiple published studies, based on the WHO’s definition of such support as a combination of psychological assistance (e.g., counseling sessions, peer support, and health education) and/or material aid (e.g., cash transfers, transport vouchers, food coupons, food packages, or supplements) [16].

• Financial Support in Tuberculosis Patient Care

The meta-analysis revealed that financial support for tuberculosis (TB) patients provided low-quality evidence of a beneficial effect on treatment success (OR 2.11, 95% CI 1.45–3.06; 1933 participants; four NRSIs, I² = 54%), alongside a serious risk of bias. These findings somewhat contradict the observations from our study, which highlight a reduced risk of treatment abandonment among recipients of financial support. The discrepancy may stem from differences in how support was delivered, particularly when combined effectively with other psychosocial interventions. In Romania, financial aid was provided through food vouchers, restricting use to food purchases and thus limiting fund misuse. Another critical aspect was the conditionality of the financial support, which depended on the number of treatment days completed. We believe this conditional approach may play a key role in motivating patients to adhere to treatment.

• Tuberculosis Patient Education

Published studies offer low- to very-low-quality evidence regarding interventions that solely provide health education, showing no significant beneficial effect on tuberculosis treatment irrespective of the patient’s progress [17]. Moreover, our study emphasizes practical interventions centered on concrete psychological and social strategies, as education alone is not suitable in a condition that generally follows a favorable and limited course when treatment is fully completed.

A study from Peru assessed financial support combined with psychosocial interventions, reporting higher treatment success among individuals from “poorer households” compared to those from “less poor” households [18]. Similarly, our findings highlight that adherence to treatment among disadvantaged groups increased significantly when supported by this comprehensive approach—with 98.95% benefiting from financial assistance, 90.64% receiving psychological interventions, and 91.01% engaging in social support.

• Supporting TB Patients with Mental Health Conditions

When tuberculosis (TB) occurs alongside mental health disorders, improving treatment outcomes often requires a multifaceted approach that includes health education, counseling, and peer support. These psychosocial interventions can be especially valuable and transformative for patients with drug-resistant TB (DR-TB) and mental health comorbidities [19,20].

In our patient group, self-reported mental disorders were generally similar across different categories: approximately 9.5–10% in TB drug-susceptible (TB-DS) and 8.3–12.1% in DR-TB patients. However, the prevalence was notably higher at 33.3% among homeless DR-TB patients, a category where alcohol use was reported in 50% of cases and smoking in 33.3%.

Non-adherence to TB treatment is a major contributor to the development of multidrug-resistant TB. When mental health issues are present, there is a strong correlation between psychological problems and treatment non-compliance, emphasizing the need for timely, coordinated interventions that address both conditions with appropriate treatment strategies [21,22].

• Treatment Adherence in Vulnerable Groups

Most patients included in the study originate from rural areas, where the overall cure rate is higher. However, subgroups such as injecting drug users and homeless individuals continue to experience increased dropout rates and mortality. Depression and substance abuse have been identified as key factors contributing to treatment discontinuation and higher mortality among these groups, though a standardized approach remains elusive [21].

This indicates that standard psychosocial support alone is insufficient for these categories, and tailored interventions that address the complexity of comorbidities—such as mental health disorders, addictions, and homelessness—are necessary [23]. Implementing psychological assessments at the outset of antituberculosis treatment may prove beneficial in identifying these risks early and guiding targeted support [24].

• TB Treatment Adherence in Relation to Education Level and Professional Status

Education level and professional status are important predictors of treatment success, aligning with findings from similar studies [24]. Patients who are professionally active and have medium to high education levels tend to adhere better to treatment, highlighting the importance of public policies aimed at reducing social and educational inequalities to enhance overall public health [15].

The study’s limitations include the lack of a control group (patients without psychosocial support), reliance on self-reported data for comorbidities (which may be subject to under- or over-reporting), and the inability to compare different intervention types directly. Despite these limitations, the findings underscore the necessity of integrating psychosocial support into comprehensive tuberculosis control strategies.

While our statistical associations do not definitively prove causation or allow us to rank interventions in terms of ‘most helpful’ in an absolute sense, they do suggest that within our cohort and program structure, interventions like peer-to-peer support were particularly noteworthy correlates of treatment success. This provides valuable indications for potential areas of emphasis in future TB management programs, complementing the overall benefit of the integrated psychosocial approach that our study generally supports.

A limitation of this study is the reliance on self-reported data for patient comorbidities, including mental health conditions, smoking, alcohol, and drug use. This method is prone to under-reporting in vulnerable populations due to factors such as stigma, recall bias, or lack of awareness of a formal diagnosis. The percentages presented in Table 3 for these conditions may therefore be considerable underestimates of the true prevalence within our cohort. Future studies should endeavor to use verified data from medical records where possible to assess comorbidities in such populations.

6. Conclusions

The involvement of multidisciplinary teams and the provision of psychosocial support contribute significantly to the success of tuberculosis treatment among patients from vulnerable groups, especially for cases of susceptible TB. Predictors that are positively associated with cure—peer-to-peer educational support, extrapulmonary TB, and higher education—confirm the importance of integrated medical–psychosocial approaches. The results highlight the need to develop public policies focused on the social needs of patients, with an emphasis on personalizing interventions for subgroups such as homeless people and injecting drug users, and the need to train specialists dedicated to this field.