What if more information about tuberculosis treatment is linked to weaker adherence-related skills, while motivation points in the opposite direction? This Indonesian study found exactly that pattern—but only explained a small part of the behavior.

Tuberculosis remains an unresolved public health issue at the global, national, and local levels. It is framed not only as an infectious disease, but also as a disorder that can reduce productivity, quality of life, and social functioning. Tuberculosis also remains one of the leading causes of death from infectious diseases globally, according to the block’s description of the epidemiological burden.

The local problem is especially visible in Serdang Bedagai Regency, North Sumatra, where the region had 3,447 tuberculosis patients from 2016 to 2020. In that period, 48.87% of patients were reported to take medication regularly for less than six months.

Medication-supervisor availability was only 60.82%, so support for medication supervision was not yet fully optimal. Previous studies have often focused on individual determinants such as knowledge, attitudes, medication side effects, or social support, while the integrated pathway linking Information, Motivation, and Behavioral Skills remained insufficiently examined.

The IMB model provides a theoretical explanation of how Information and Motivation influence Behavioral Skills, but empirical testing among tuberculosis patients in primary healthcare settings remained limited. The study therefore tested structural relationships between Information, Motivation, and Behavioral Skills using locally derived indicators among tuberculosis patients in Serdang Bedagai Regency.

The contribution lies in the context-specific operationalization of IMB: ten retained indicators were organized into Information, Motivation, and Behavioral Skills, then evaluated as a measurement and structural model. The study focused on category-I tuberculosis patients in primary healthcare settings across Serdang Bedagai Regency, Indonesia.

It developed and evaluated locally derived indicators based on the treatment context of patients, families, healthcare workers, and local stakeholders. PLS-SEM 3 was used to examine structural relationships between Information, Motivation, and Behavioral Skills rather than only associations between individual factors and adherence outcomes.

The study used an explanatory cross-sectional quantitative survey in the working area of the Serdang Bedagai District Health Office, North Sumatra. The population consisted of 3,447 tuberculosis patients, and the final sample consisted of 230 tuberculosis patients.

Community health centers were selected across high, medium, and low tuberculosis case-burden categories, while patients within selected centers were recruited using cluster random sampling. Responses used a five-point Likert scale, and the final PLS-SEM 3 measurement model retained ten indicators: three for Information, three for Motivation, and four for Skills.

PLS-SEM 3 analysis was conducted in two sequential stages: first the measurement model was estimated, and then the structural model was assessed. For the measurement model, convergent validity used standardized outer loadings and AVE, with AVE of at least 0.50 indicating adequate convergent validity.

Internal consistency used Cronbach’s alpha and composite reliability, with values of at least 0.70 considered acceptable. For the structural model, the analysis examined collinearity, explanatory power, predictive relevance, and path significance; hypotheses were tested by bootstrapping with two-tailed criteria of t greater than 1.96 and p less than 0.05.

Among respondents, the largest age group was 51 to 60 years old, with 67 people, or 29.1 percent. Male respondents numbered 166, or 72.2 percent, while female respondents numbered 64, or 27.8 percent. The majority were self-employed, including farmers, fishermen, livestock breeders, and labourers, totaling 144 people, or 62.6 percent.

Table one summarizes the two hundred thirty respondents’ demographic profile and treatment category. The largest age group is fifty-one to sixty years, with sixty-seven respondents, or twenty-nine point one percent; males account for one hundred sixty-six, or seventy-two point two percent.

Occupation is concentrated among entrepreneurs, farmers, fishermen, and laborers, totaling one hundred forty-four respondents, or sixty-two point six percent. All respondents are listed in treatment Category I, which helps define the study population for the adherence analysis.

Internal consistency and convergent validity were acceptable for all constructs. Cronbach’s alpha was 0.734 for Behavioral Skills, 0.701 for Information, and 0.717 for Motivation. Composite reliability was 0.812, 0.826, and 0.811, respectively, while AVE was 0.523 for Behavioral Skill, 0.614 for Information, and 0.594 for Motivation, all above the 0.50 criterion.

Table three reports the measurement quality for three constructs: Behavioral Skills, Information, and Motivation. Cronbach’s alpha values are zero point seven three four, zero point seven zero one, and zero point seven one seven, while composite reliability values are zero point eight one two, zero point eight two six, and zero point eight one one, respectively.

The AVE values all exceed the stated zero point five zero criterion, supporting acceptable internal consistency and convergent validity. Information and Motivation explained 8.2% of the variance in Behavioral Skills, with an R-squared value of 0.082. That means the two predictors contributed only a limited proportion of variation in patients’ adherence-related behavioral skills.

Other factors outside the model may account for additional differences in behavioral performance during tuberculosis treatment. The path coefficients were tested using bootstrapping with 5,000 bootstrap resamples. Bootstrapping supported both hypothesized structural relationships at the five-percent level, but their directions differed.

Information had a statistically significant negative standardized association with Behavioral Skills: beta was negative 0.226, t was 3.326, and p was 0.001. Motivation had a statistically significant positive association with Behavioral Skills: beta was 0.151, t was 2.049, and p was 0.041.

Although the absolute Information coefficient was larger, it should not be described as a stronger positive influence because its sign was negative; the sign must be interpreted together with the scoring direction of the information items. Table nine reports standardized structural paths to Behavioral Skills within the adapted Information–Motivation–Behavioral Skills model.

Information shows a significant inverse association, with beta equal to negative zero point two two six, while Motivation shows a significant positive association, with beta equal to zero point one five one; both paths have p-values below zero point zero five. The authors therefore describe partial support for the proposed mechanism, while noting that information items included both accurate and misconception-oriented statements.

The study should be interpreted as an exploratory, context-specific examination of relationships among adapted IMB constructs rather than as a comprehensive explanation of tuberculosis treatment adherence. The model tested whether Information and Motivation were associated with Behavioral Skills representing adherence-related actions during tuberculosis treatment.

Because treatment adherence is influenced by multiple individual, social, healthcare, and structural factors, these findings represent one component of the behavioral mechanisms underlying treatment practices. Table 8 reports PLS-SEM model-fit diagnostics for the saturated and estimated models, with identical values in both columns: SRMR is zero point one zero four, d ULS is zero point five nine zero, d G is zero point two two one, chi-square is two hundred ninety-seven point seven zero three, and NFI is zero point four six seven.

The authors emphasize that these diagnostics do not indicate strong global fit: SRMR exceeds the commonly used zero point zero eight benchmark, while d ULS and d G require bootstrap reference quantiles for fuller interpretation. The cross-sectional design does not establish temporal causality, and the study was conducted in one district with only category-I patients.

Behavioral Skills, rather than a separate objective adherence outcome, served as the endogenous construct. The low R-squared value indicates that Information and Motivation explain only a small proportion of variance in Behavioral Skills, so this IMB model should not be treated as a complete explanation of treatment-related behaviors.

Healthcare accessibility, socioeconomic conditions, treatment experience, and healthcare provider support may contribute to patients’ behavioral skills. The adapted IMB model found a significant negative association for Information and a significant positive association for Motivation, but explained only 8.2% of Behavioral Skills.

The result is exploratory, not a complete explanation of adherence.