Multifactorial determinants of lost to follow-up in antiretroviral therapy: evidence from a case–control study in Mexico

Abstract

Background

Loss to follow-up (LTFU) remains a major challenge in achieving sustained HIV care. Understanding individual and structural factors influencing disengagement is essential to improve retention, particularly in low- and middle-income settings. This study aimed to identify predictors of LTFU among adults receiving antiretroviral therapy (ART) in western Mexico.

Methods

A case–control study was conducted among adults with HIV treated at a tertiary hospital. Cases met the national definition of LTFU (≥ 90 days beyond the expected clinic visit or pharmacy refill), while controls were retained patients during the same period. A total of 919 participants were included (148 LTFU, 771 retained). Multivariable logistic regression identified factors associated with LTFU.

Results

Median age was 42 years (IQR 34, 51) and 88% were male. The multivariable analysis identified that age was associated with lower risk of LTFU (adjusted odds ratio [aOR] per year, 0.94; 95% CI, 0.91–0.96). Secondary ART resistance (aOR, 4.03; 95% CI, 1.59–9.99), hard-drug use (aOR, 2.57; 95% CI, 1.68–3.93), psychiatric disorders (aOR, 3.58; 95% CI, 2.23–5.72), lower educational level (≥ upper secondary vs. no formal education/primary: aOR, 2.30; 95% CI, 1.34–3.94), emergency department visits (aOR, 2.63; 95% CI, 1.72–4.04), and years living with HIV (aOR per year, 1.06; 95% CI, 1.02–1.10) were associated with higher odds of LTFU.

Conclusions

These findings highlight the role of psychosocial and structural determinants of LTFU, underscoring the need for integrated interventions addressing education, mental health, and substance use to improve retention in HIV care in Mexico.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12981-026-00845-w.

Background

Sustained engagement along the HIV care continuum is essential to achieve the UNAIDS 95‑95‑95 targets and prevent onward transmission [1]. Despite the widespread implementation of universal test-and-treat strategies and the availability of better-tolerated antiretroviral therapies (ART), loss to follow-up (LTFU) remains a persistent challenge, with many individuals continuing to disengage and re-engage in care [1]. Strengthening re‑engagement strategies and addressing multi‑level determinants of disengagement have become priorities for programs worldwide [2].

Recent progress in treating HIV infections has enabled patients to reach life expectancies like those of HIV-negative individuals [3]. This is largely due to the development of novel and more effective ART that also have fewer adverse effects [3]. However, to achieve these outcomes, patients must adhere to their prescribed medication and maintain consistent engagement with healthcare services, a concept known as “retention in care“ [4].

Retention in care is defined as a patient’s sustained commitment to attending and remaining engaged with their healthcare center, encompassing consistent contact and treatment adherence from diagnosis through ongoing care [5]. Conversely, LTFU occurs when a patient interrupts this continuity, defined as absence from medical reviews or medication pick-up for a period of 90 days or more [6–9].

Discontinuation of treatment among patients is characterized by a proportion returning to care after varying durations, while others remain untreated, leading to several adverse outcomes [8]. These include a twofold increase in mortality compared to patients retained in care, increased challenges in achieving viral suppression with first-line treatments, and sustained transmission within the community [9]. Patients not retained in care are estimated to infect an average of five individuals due to inadequate viral control [10]. Furthermore, from a health economics standpoint, ensuring patient retention in care can decrease morbidity and mortality, as well as the resources needed for specialized medical care for disease progression [11].

Studies indicate that sociodemographic variables, including older age and male gender, correlate with increased treatment discontinuation risks [6, 12]. Additionally, lower educational attainment, inadequate access to essential household amenities, insufficient social support networks, and specific ethnic backgrounds are associated with a heightened likelihood of LTFU [13]. Adolescents and young adults face elevated risk of missed visits and non-suppression—driven by depression and substance use—while transitions to adult care are a particularly vulnerable period with declining retention [14].

Psychosocial and behavioral determinants are central to disengagement. Meta-analytic data associate depressive symptoms with a lower likelihood of viral suppression—an indicator of suboptimal adherence and care engagement—while clinical guidelines identify untreated substance-use disorders as major obstacles to sustained ART [15–18]. Additionally, stigma has been linked to LTFU, as it continues to act as a cross-cutting barrier that discourages disclosure, clinic attendance, and medication adherence [19].

In addition, economic and clinical determinants significantly impact treatment adherence. Limited financial resources, coupled with transportation and medical expenses, pose substantial obstacles while food insecurity has been linked to non-retention and non-suppression in diverse settings [20–22]. Moreover, advanced stages of the disease, indicated by a diminished CD4⁺ T cell count or the presence of opportunistic infections, contribute to an increased disease burden and a more complex medication regimen, thereby elevating the probability of patients being LTFU [23, 24].

Recent evidence from Mexico further underscores the clinical relevance of disengagement from HIV care [25]. A cohort study of hospitalized adults with advanced HIV disease found that individuals who had discontinued ART experienced significantly higher one-year mortality than treatment-naïve patients (24% vs. 8%), with treatment interruption independently associated with a two-fold increase in the hazard of death [25]. These findings highlight the severe consequences of LTFU and reinforce the need to better understand the determinants in the Mexican context [25].

In Mexico, accurate identification of LTFU is complicated by silent transfers and unrecorded deaths due to independent health registries [26]. To date, there is only one study in the Mexican population linking clinical records only investigated the role of age, sex, education, sexual orientation and CD4⁺T levels on the risk of LFTU and their mortality risk [26]. This previous study did not explore psychosocial, behavioral and other important HIV-related risk factors and their link with LFTU [26]. To address this, the present study aimed at identifying a wider array of risk factors for LFTU among people living with HIV receiving ART in a specialized HIV unit from a tertiary hospital in Mexico.

Methodology

Study design

We conducted a case–control study to identify determinants of LTFU among people living with HIV receiving ART. Cases were defined using the programmatic definition of LTFU: absence from clinic visits or pharmacy refill for ≥ 90 days beyond the last expected appointment date. The date of LTFU was assigned as the day the patient exceeded the 90-day threshold. Patients were censored if they died, or had documented ART interruption due to medical indication. Short gaps in care (< 90 days) were not classified as LTFU. Controls were adults who remained engaged in care, defined as attending scheduled visits or ART refills without exceeding the 90-day gap during the study period.

Study setting

The study was conducted at the HIV unit of the Hospital Civil de Guadalajara “Fray Antonio Alcalde”, a specialized unit for the care of people living with HIV in the city of Guadalajara, Jalisco, Mexico. This is a high-volume care center with a multidisciplinary team, providing care primarily to individuals from the state of Jalisco, most of whom reside within the Guadalajara Metropolitan Area (ZMG).

At the study site, antiretroviral therapy (ART) refills and routine HIV follow-up visits were provided through in-person care at the specialized HIV clinic. During the study period, multi-month dispensing was available on a limited basis and restricted to patients with sustained virologic suppression, allowing ART refills for up to three months in accordance with national guidelines [27].

ART medications were not available through external pharmacies or decentralized distribution systems; therefore, continued engagement with the clinic was required to maintain uninterrupted treatment. As part of routine care, patients had access to a multidisciplinary team, including referral to psychiatric services for the evaluation and management of mental health conditions or substance use disorders, as well as nutritional assessment when clinically indicated.

Ethical considerations

This study adheres to the International Ethical Guidelines for Health-Related Research Involving Humans established by the Council for International Organizations of Medical Sciences (CIOMS) [28]. Due to the retrospective nature of the study, a waiver of informed consent was granted. Alphanumeric codes were used to ensure participant confidentiality. This study was approved by the institutional bioethics committee under approval number CEI242/23.

Primary outcome

The primary outcome of this study is LTFU, defined as 90 days or more without attending ART prescription refills.

Risk factors

Sociodemographic variables included age (years) and sex (male or female). Educational level was categorized into no formal/primary (0–6 scholar years), lower secondary (7–9 scholar years) and higher secondary (≥ 10 scholar years). Time travel to clinic, recorded in self-reported minutes, was assessed as indicator of health services accessibility.

Clinical variables included the number and presence of opportunistic infections (e.g., tuberculosis, candidiasis, toxoplasmosis, or Pneumocystis jirovecii pneumonia), hospitalization and (ED) emergency department visits during the previous 12 months, ART regimen at initiation, primary or secondary ART resistance and years living with HIV. Data on clinical, behavioral, and psychosocial variables were extracted from the electronic medical record and from standardized psychiatric evaluations documented as part of routine HIV care. Mental health conditions were recorded only when formally diagnosed by a psychiatrist, and included depressive disorders, bipolar disorder, schizophrenia, and anxiety disorders (e.g., generalized anxiety disorder or panic disorder).

Substance use (alcohol, Cannabis , cocaine, methamphetamine, crack, heroin, and hallucinogens) was also documented by the psychiatry service, based on structured clinical interviews conducted at intake or during follow-up visits. Substance use was operationalized as a binary variable indicating current or recent use within the past 6 months, as recorded in the medical file; lifetime or historical use without recent activity was not classified as “use” in this analysis.

Emergency department visits and hospitalizations were obtained directly from electronic medical records and defined as any documented encounter within the 12 months prior to the index date (ART initiation or reference visit).

Eligibility criteria

Eligible participants were adults (≥ 18 years old) living with HIV who were receiving ART that had at least one documented follow-up visit after ART initiation. Patients were excluded if they had missing data about their transfer of care to another facility or experienced ART interruption due to medical indications such as severe adverse drug reactions. Individuals transferred to another facility were included in the analysis as they continued their ART regimen.

Sample size

The sample size was estimated using the pmsampsize package, which accounts for the anticipated proportion of the outcome, the number of predictors, and adjustment for overfitting (< 10% optimism). In the absence of prior R² estimates for this population, we assumed a conservative R² of 0.15 and planned for 19 candidate predictors based on previous literature [29]. Using national estimates from the Antiretroviral Administration, Logistics, and Surveillance System (SALVAR) database in Mexico (2009–2013), which reported an incidence of treatment abandonment of ~ 5.4%, a minimum of 57 events was required to ensure model stability (≈ 2.97 events per predictor) [30]. In our study, we included 919 participants, of whom 148 experienced LTFU, providing sufficient events to support the planned multivariable analysis.

Statistical analysis

Descriptive analyses were used to summarize the sociodemographic, clinical, HIV-related, and psychological characteristics of the study participants. Categorical variables are presented as absolute numbers and percentages, while continuous variables are presented with means and standard deviations for normally distributed data, or medians and interquartile ranges for non-normally distributed data.

For each variable, univariate odds ratios (OR) with 95% confidence intervals (CI) were estimated. Variables with a p-value ≤ 0.25 in univariate analyses were subsequently included in a multivariable logistic regression model to identify independent determinants of LTFU. All analyses were conducted using R software (version 2025.05.0) [31], with statistical significance set at p < 0.05.

Sensitivity analyses

To assess the robustness of the findings, several sensitivity analyses were conducted. First, we excluded patients with psychiatric disorders (depression, bipolar disorder, schizophrenia) given their strong correlation with substance use and other social determinants, which could influence the associations between the other studied predictors and LFTU [32, 33]. Second, to assess whether pandemic-related service disruptions influenced LTFU risk, we included a dichotomous variable indicating whether ART initiation occurred during the strict lockdown period (beginning April 19th to May 30th, 2020) [34]. Third, to account for changes in HIV care delivery over time in Mexico and unmeasured heterogeneity related to changes in treatment guidelines, ART availability, and healthcare delivery over time, we evaluated the ART initiation period as a random effect in a mixed-effects logistic regression model. Patients were grouped into four policy-relevant calendar periods: (1) pre-universal ART period (< 2005); (2) universal ART access before integrase strand transfer inhibitors (InSTIs) were available (2006–2011); (3) InSTI availability prior to the COVID-19 pandemic (2012–2019); and (4) the pandemic and post-pandemic period (2020–2023) [35]. Treating calendar period as a random intercept allowed us to account for clustering of patients within initiation periods and to adjust for unobserved temporal factors, while estimating the associations of individual-level predictors as fixed effects. This allowed to assess whether temporal changes in treatment policies and health care delivery influenced the associations in the main model. Lastly, the main model was adjusted for initial ART regimen to explore whether the type of ART regimen (InSTI simplified versus other/NNRTI-based regimens) influenced the associations with LTFU.

Results

Baseline characteristics

Among 971 registered patients, we excluded 21 patients that had did not have information about their transfer date and status and 31 which had missing data from leaving 919 participants for analysis. No patients died before de 90-day definition of LTFU. No significant demographic or clinical differences were observed between excluded and included individuals (Supplementary Table 1). Overall, LTFU patients were younger, with a median age of 39 years (IQR 33–48), compared with 42 years (IQR 34–52) among those retained in care (p = 0.001). Although significant (n = 0.048), most patients were male in both LFTU (88.3) and those retained in care (82.4%). The median time from HIV diagnosis to LTFU was 4.6 years (IQR 2.4–10.2). Those who were LTFU more frequently had prior hospital or emergency visits and prevalence of psychiatric disorders and illicit drug use. Additionally, years living with HIV was similar between groups (Table 1). To further explore the LFTU patterns in the sample, crude LTFU proportions were examined across age categories and by hard drug use (Table 2). Crude LTFU proportions were consistently higher in patients reporting hard drug use across all age categories.

Table 2.

Stratified distribution of LTFU by age category and self-reported hard drugs use

Age category (years)	Hard drugs use	N	LTFU (n)	Crude LFTU proportion (%)
18–34	No	175	23	13.1
18–34	Yes	70	22	31.4
35–45	No	207	24	11.6
35–45	Yes	121	37	30.6
≥ 46	No	256	19	7.4
≥ 46	Yes	90	23	25.6

LTFU, lost to follow-up

Table 1.

Baseline characteristics of people living with HIV by retention status (n = 919)

Variable	Overall (n = 919)	Retained in care (n = 771)	Lost to follow-up (n = 148)	p-value
Age, median (IQR), years	42 (34, 51)	42 (34, 52)	39 (33, 48)	0.001
Sex, male	803 (87.4%)	681 (88.3%)	122 (82.4%)	0.048
Travel time to clinic, median (IQR), minutes	35 (24, 53)	35 (23, 55)	36 (26, 50)	0.803
Educational level				< 0.001
No formal/primary	240 (25%)	186 (24%)	59 (40%)
Lower secondary	231 (25%)	186 (24%)	45 (30%)
Upper secondary or greater	448 (49%)	404 (52%)	44 (30%)
Sexual orientation				< 0.001
MSM	608 (66%)	531 (69%)	77 (52%)	< 0.001
Transexual	12 (1.3%)	9 (1.2%)	3 (2%)	0.258
Heterosexual	255 (28%)	201 (26%)	54 (36%)	0.010
Bisexual	176 (19%)	144 (19%)	32 (22%)	0.404
Transition	4 (0.4%)	3 (0.4%)	1 (0.7%)	0.505
Prior hospitalizations	293 (32%)	216 (28%)	77 (52%)	< 0.001
Emergency department visits	340 (37%)	249 (32%)	91 (61%)	< 0.001
ART resistance
Primary	40 (4.4%)	36 (4.7%)	4 (2.7%)	0.283
Secondary	30 (3.3%)	16 (2.1%)	14 (9.5%)	< 0.001
Years living with HIV, median (IQR)	8 (5, 14)	8 (5, 14)	10 (6, 16)	0.064
Initial ART regimen				0.004
BIC/FTC/TAF	239 (26%)	211 (27%)	28 (19%)
EFV/FTC/TDF	14 (1.5%)	11 (1.4%)	3 (2%)
ABC/3TC/DTG	344 (37%)	297 (39%)	47 (32%)
DRV/c	15 (1.6%)	12 (1.6%)	3 (2%)
Other	306 (33%)	240 (31%)	66 (45%)
Mental health disorders
Anxiety	63 (6.9%)	51 (6.6%)	12 (8.1%)	0.510
Depression	136 (15%)	87 (11%)	49 (33%)	< 0.001
Bipolar disorder	10 (1.1%)	5 (0.6%)	5 (3.4%)	0.013
Schizophrenia	8 (0.9%)	5 (0.6%)	3 (2%)	0.124
Substance use
Alcohol	689 (75%)	572 (74%)	117 (79%)	0.211
Cannabis	299 (33%)	238 (31%)	61 (41%)	0.014
Hard drug use	281 (31%)	199 (26%)	82 (55%)	< 0.001
Cocaine	193 (21%)	146 (19%)	47 (32%)	< 0.001
Methamphetamine	128 (14%)	73 (9.5%)	55 (37%)	< 0.001
Crack	24 (2.6%)	15 (1.9%)	9 (6.1%)	0.009
Heroin	9 (1%)	3 (0.4%)	6 (4.1%)	< 0.001
Hallucinogens	21 (2.3%)	16 (2.1%)	5 (3.4%)	0.363
Previous opportunistc infection	269 (29%)	202 (26%)	67 (45%)	< 0.001
Opportunistic infections
Candidiasis	52 (5.7%)	29 (3.8%)	23 (16%)	< 0.001
Tuberculosis	100 (11%)	85 (11%)	15 (10%)	0.750
Histoplasmosis	38 (4.1%)	23 (3%)	15 (10%)	< 0.001
Pneumocystis pneumonia (PJP)	28 (3.0%)	7 (0.9%)	21 (14%)	< 0.001
Toxoplasmosis	30 (3.3%)	22 (2.9%)	8 (5.4%)	0.127
Kaposi’s sarcoma	39 (4.2%)	31 (4%)	8 (5.4%)	0.444
Cryptococcus,	27 (2.9%)	21 (2.7%)	6 (4.1%)	0.422
Cryptosporidium	6 (0.7%)	5 (0.6%)	1 (0.7%)	0.999
Lymphoma	13 (1.4%)	11 (1.4%)	2 (1.4%)	> 0.999
Number of previous opportunistic infections				< 0.001
0	650 (71%)	569 (74%)	81 (55%)
1	200 (22%)	157 (20%)	43 (29%)
≥2	69 (7.5%)	45 (5.8%)	24 (16%)

Data are median (interquartile range) and frequency (%). MSM, men who have sex with men; ART, antiretroviral therapy; BIC/FTC/TAF, bictegravir/emtricitabine/tenofovir alafenamide; EFV/FTC/TDF, efavirenz/emtricitabine/tenofovir disoproxil fumarate; ABC/3TC/DTG, abacavir/lamivudine/dolutegravir; DRV/c, darunavir/cobicistat

Univariate analysis

Table 3 presents the results of the univariate analysis. Male sex was tended to be associated with lower risk of LFTU (p = 0.050). Younger age, and lower education levels were associated with higher odds of LTFU. Psychiatric disorders and substance use (notably hard drugs and marijuana) were also associated with greater risk of LTFU. Clinical factors associated with LTFU included previous hospitalizations, ED visits, and secondary ART resistance. In addition, there was an increasing association of LTFU with the number of previous opportunistic infections.

Table 3.

Univariate analysis of factors associated with loss to follow-up among people living with HIV (n = 919)

Variable	OR (95% CI)	p-value
Age, per year increase	0.97 (0.95–0.98)	< 0.001
Sex
Female	Ref.
Male	0.62 (0.39–1.02)	0.050
Years living with HIV per year	1.02 (0.99–1.04)	0.174
Marital status
Single	Ref.
Married	0.61 (0.28–1.18)	0.172
Widowed	1.16 (0.26–3.68)	0.815
Cohabiting (common-law union)	1.03 (0.62–1.64)	0.917
Educational level
Upper secondary or greater	Ref.
Lower secondary	2.22 (1.42–3.49)	0.001
No formal/primary	2.99 (1.96–4.61)	< 0.001
Residence
Outside metropolitan area	Ref.
Living in Metropolitan area (ZMG)	1.07 (0.72–1.64)	0.746
Self-reported travel time to clinic	1.00 (0.99–1.00)	0.098
ART resistance
Primary	0.57 (0.17–1.44)	0.280
Secondary	4.93 (2.32–10.36)	< 0.001
Substance use
Alcohol	1.31 (0.87–2.04)	0.212
Cannabis	1.57 (1.09–2.25)	0.014
Hard drugs	3.57 (2.49–5.14)	< 0.001
Mental health
Anxiety	1.25 (0.62–2.32)	0.511
Psychiatric disorder (depression, bipolar disorder, schizophrenia)	4.07 (2.72–6.06)	< 0.001
Sexual orientation
Heterosexual	Ref.
Sexual and gender minorities	1.24 (0.81–1.87)	0.312
Prior hospitalization	2.79 (1.95–3.99)	< 0.001
Emergency department visit	3.35 (2.33–4.84)	< 0.001
Previous oportunistic infection (any)*	2.33 (1.62–3.34)	< 0.001
Number of previous opportunistic infections
0	Ref.
1	1.92 (1.27–2.89)	0.002
≥2	3.75 (2.14–6.43)	< 0.001

Odds ratio (OR) and 95% confidence intervals (CI) estimated with univariate logistic regression. *Opportunistic infections included candidiasis, cryptococcosis, tuberculosis, toxoplasmosis, Kaposi’s sarcoma, non-Hodgkin lymphoma, histoplasmosis, Pneumocystis jirovecii pneumonia (PJP). Sexual and gender minorities included men who have sex with men, bisexual, transexual and in transition. Use of hard drugs was defined as current or recent use within the past 6 months of cocaine, methamphetamine, crack, heroin, or hallucinogens. ZMG, Guadalajara metropolitan area.

Multivariable analysis

Risk factors associated with LTFU in the multivariable model are presented in Fig. 1. Older age remained associated with LFTU, with each additional year associated with 6% lower odds. Other risk factors included secondary ART resistance, hard drug use, psychiatric disorders, and emergency department visits. Lastly, using upper secondary education or greater as the reference category, there was an increasing risk of LTFU with lower education level, with the highest risk among patients with lower or no education.

Fig. 1.

Forest plot of risk factors associated with lost to follow-up. Odds ratios are adjusted for the variables in the plot (aOR)

Sensitivity analyses

Sensitivity analyses showed that the main findings were robust. The exclusion of participants with incomplete data did not materially alter the study findings (Supplementary Table 1). Excluding patients with psychiatric disorders did not substantially alter the associations, with age, education, hard drug use, emergency visits, and years living with HIV remaining significant predictors (Supplemental Table 2). Adjustment for the strict COVID-19 lockdown period yielded consistent results, with the lockdown variable itself not associated with LTFU (Supplemental Table 3). The distribution of LTFU was similar among the different periods of ART therapy initiation (Supplementary Table 4). In addition, the mixed-effects logistic regression model including ART initiation period as a random effect showed no residual variation attributable to treatment era, reinforcing temporal stability of the associations (Supplementary Table 5). Finally, adjusting for initial ART regimen did not materially change the estimates, indicating that the observed determinants were independent of regimen type (Supplemental Table 6).

Discussion

This study aimed to identify the risk factors associated with LFTU among people living with HIV receiving ART in a tertiary hospital in Mexico. We found that lower educational attainment, hard drug use, presence of psychiatric disorders including depressive, bipolar, and schizophrenic disorders, and a history of ED visits were associated with LTFU, while older age was associated with lower LFTU risk factor. These associations remained robust across sensitivity analyses. Specifically, the risk factors in the main model remained significant, although attenuated, in the absence of these psychiatric disorders. Furthermore, the period of strict COVID-19 confinement and the type of ART initiated did not significantly influence the risk of LTFU, suggesting that structural and individual-level factors outweighed temporal and regimen-related variables.

Our findings are consistent with prior studies showing the impact of mental health disorders and substance use on ART adherence and retention [36–38]. In our cohort, psychiatric disorders tripled the odds of disengagement, a pattern consistent with evidence that depression impairs appointment keeping, adherence, and viral suppression [39]. Hard-drug use was also strongly associated with LTFU, aligning with data showing reduced retention among individuals with substance-use disorders [40]. Methamphetamine use has been linked to persistent viremia and care disruptions, especially among MSM [41]. These results reinforce current clinical recommendations emphasizing the need for systematic screening and treatment of substance-use disorders within HIV care, including evidence-based interventions such as contingency management and medications for opioid use disorder [2].

The association with lower education is consistent with evidence that limited health literacy reduces understanding of treatment benefits and impedes engagement with healthcare services [42]. Similar findings were reported in Latin American cohorts, where limited educational attainment and delayed presentation to care were linked to advanced HIV disease at ART initiation and higher mortality rates [43]. Educational attainment not only reflects individual literacy, but also represents a broader indicator of socioeconomic status, access to health-related information, and health system navigation skills. Individuals with low or no formal education may have limited understanding of disease mechanisms, the importance of treatment adherence, or the potential consequences of treatment interruption [44]. Additionally, lower education is often associated with informal employment, housing instability, and limited social support — all of which can compromise a patient’s ability to attend follow-up appointments or complete its ART regimen [45].

Older age was associated with a lower risk of LTFU (approximately 6% lower odds per additional year of age), where older individuals often demonstrate greater stability in healthcare engagement [46], and it also aligns with contemporary literature showing that younger adults—especially adolescents and young adults—are at highest risk for attrition from the HIV care continuum [14]. A recent registry-linked cohort from Mexico similarly found younger age with higher LTFU after correcting for silent transfers and deaths, underscoring external validity in our setting [26]. Studies from sub-Saharan Africa and Tanzania further document elevated disengagement among youth during the first year on ART, aligning with our age effect and with the notion that developmental stage, mobility, and competing priorities challenge care continuity early in adulthood [47].

Secondary (acquired) drug resistance was strongly associated with LTFU. While resistance is biologically distinct from adherence behavior, the two are tightly coupled: poor or intermittent adherence drives virologic failure and the emergence of resistance, which in turn can decrease patient confidence and precipitate care avoidance [48]. Thus, resistance likely reflects previous suboptimal adherence, which may reinforce disengagement due to perceived treatment futility [49].

Our observation of previous history of ED visits as a predictor of abandonment coincides with prior evidence linking fragmented healthcare utilization with LTFU [24, 50]. Our findings suggest that emergency department encounters identify individuals at increased risk of disengagement, representing critical opportunities for targeted linkage and retention interventions [51].

Interestingly, the lack of effect of ART initiation period or regimen type suggests that, in this setting, systemic barriers and psychosocial determinants outweigh pharmacological simplification or temporal programmatic changes. This observation contrasts with studies in high-income countries where regimen simplification has been linked to improved persistence [43], underscoring contextual differences.

Furthermore, recent evidence from Mexico highlights that treatment discontinuation not only increases the risk of LTFU but is also independently associated with significantly higher mortality in hospitalized individuals with advanced HIV (hazard ratio 2.08; 95% CI, 1.14–3.78) [25]. These findings highlight that LTFU is linked to a complex interplay of psychosocial and structural determinants, rather than to isolated clinical factors, underscoring the need for integrated approaches to retention in care. Although our study did not assess clinical outcomes directly, these findings, combined with our results, underscore that disengagement from care reflects a complex interaction of psychosocial and structural factors that precede adverse clinical trajectories.

Our findings highlight the importance of addressing educational barriers, mental health conditions, and substance use to improve retention in HIV care. Similar determinants of LTFU have been reported globally, and interventions such as differentiated service delivery models, peer support, and integration of mental health and substance use services within HIV care have shown benefits for retention, particularly among younger and socially vulnerable populations [52, 53]. In addition, patient-centered educational strategies aimed at improving health literacy have been associated with better engagement in care among individuals with lower educational attainment [54, 55].

A major strength of this study is the use of a cohort from a specialized HIV clinic, which increases the comprehensive dataset allowed adjustment for a wide range of demographic, clinical, behavioral, and healthcare utilization variables. Multiple sensitivity analyses—excluding psychiatric cases, stratifying by confinement period, and adjusting for ART regimen—strengthen the robustness of the results.

Several limitations should be acknowledged. First, this was a retrospective observational study, which precludes causal inference, and potential unmeasured predictors —such as social support, stigma, or income stability—may partly account for the associations observed. Second, most participants were men, limiting the ability to examine sex-based differences in risk factors for LTFU. Fourth, the study was conducted in a single tertiary center, which may restrict generalizability to rural or decentralized HIV care settings or to other regions of Mexico. Finally, the operational definition of LTFU relied on clinical records and may not fully distinguish true disengagement from silent transfer to another facility.

Conclusions

This study analyzed factors associated with LTFU among people living with HIV receiving ART in a tertiary hospital in Mexico. Younger age, secondary ART resistance, hard drug use, lower educational attainment, psychiatric disorders, and emergency department visits were independently associated with disengagement from care, while older age waslinked to decreased LTFU risk. Strengthening mental health and substance use screening and ensuring continuous follow-up for patients with repeated emergency visits, may reduce disengagement. Integrating these approaches within multidisciplinary HIV care could potentially improve long-term treatment continuity and optimize ART outcomes in Mexico.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

aOR

Adjusted odds ratio

ART

Antiretroviral therapy

CI

Confidence interval

HIV

Human immunodeficiency virus

LTFU

Loss to follow-up

MSM

Men who have sex with men

NNRTI

Non-nucleoside reverse transcriptase inhibitor

NRTI

Nucleoside reverse transcriptase inhibitor

INSTI

Integrase strand transfer inhibitor

PI

Protease inhibitor

PJP

Pneumocystis carinii Pneumonia (now Pneumocystis jirovecii Pneumonia)

TB

Tuberculosis

OR

Odds ratio

SD

Standard deviation

IQR

Interquartile range

ZMG

Metropolitan area of Guadalajara

ED

Emergency deparment

Author contributions

LEDMT and PMA conceived the study, designed the research protocol, and supervised data collection and analysis. LEDMT led the manuscript drafting and coordinated the overall writing process. GAAS performed the statistical analyses and contributed to data interpretation. LAGH, FAAL, SVZQ, VVRH, AV, KSR, MAZ, and JFA contributed to data acquisition, clinical validation, and critical review of the manuscript for intellectual content. All authors read and approved the final version of the manuscript.

Funding

This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.

Data availability

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Ethical approval and waiver of informed consent were granted by the Ethics Committee of the Hospital Civil de Guadalajara “Fray Antonio Alcalde” due to the retrospective nature of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.