Rate and predictors of loss to follow-up in HIV care in a low-resource setting: analyzing critical risk periods

Abstract

Background

Patient loss-to-follow-up (LTFU) in HIV care is a major challenge, especially in low-resource settings. Although the literature has focused on the total rate at which patients disengage from care, it has not sufficiently examined the specific risk periods during which patients are most likely to disengage from care. By addressing this gap, researchers and healthcare providers can develop more targeted interventions to improve patient engagement in HIV care.

Methods

We conducted a retrospective cohort study on newly enrolled adult HIV patients at seven randomly selected high-volume health facilities in Ethiopia from May 2022 to April 2024. Data analysis was performed using SPSS version 26, with a focus on the incidence rate of LTFU during the critical risk periods. Cumulative hazard analysis was used to compare event distributions, whereas a Poisson regression model was used to identify factors predicting LTFU, with statistical significance set at p < 0.05.

Results

The analysis included 737 individuals newly enrolled in HIV care; 165 participants (22.4%, 95% CI: 19.5–25.2) were LTFU by the end of two years, of which 50.1% occurred within the first 6 months, 29.7% within 7–12 months, and 19.4% from 13 to 24 months on ART. The overall incidence rate of LTFU was 18.3 per 1,000 PMO (95% CI: 15.9–20.6), with rates of 167.7 in the first 6 months, 55.4 in 7–12 months, and 18.1 in 13–24 months. Incomplete addresses lacking a phone number or location information (IRR: 1.61; 95% CI: 1.14, 2.27) and poor adherence (IRR: 1.78; 95% CI: 1.28, 2.48) were factors predicting the incidence rate of LTFU.

Conclusion

LTFU peaked in the first 6 months, accounting for approximately half of total losses, remained elevated from months 7–12, and stabilized after the first year of HIV care and treatment. Address information and adherence were predictors of LTFU. To effectively minimize LTFU, efforts should focus on intensive support during the first six months of care, followed by sustained efforts and monitoring in the next six months. Our findings highlight a critical period for targeted interventions to reduce LTFU in HIV care.

Background

Globally, there were approximately 39 million people living with HIV in 2022, with sub-Saharan Africa (SSA) accounting for 76% of infections [1]. By the end of 2022, around 630,000 people had died from AIDS-related illnesses, a decrease of 51% since 2010. SSA remained the hardest-hit region, with 20.8 million people living with HIV in 2022 [1]. Currently, the world is undertaking the movement of achieving the HIV test and treatment goals of 95/95/95% global targets and ending the HIV epidemic by 2030 [2]. However, the major programmatic challenge appears to be retaining the clients already enrolled in ART and sustaining viral load suppression [2].

Patient loss to follow-up (LTFU) in HIV care presents a significant challenge in managing HIV, particularly in low-resource settings [3]. This term refers to patients who interrupted their HIV care or antiretroviral therapy (ART) appointment for more than 28 days from their appointment date [4]. The consequences of LTFU are severe for both the individual and the community, leading to poorer health outcomes, increased mortality, and a greater likelihood of HIV transmission [5] due to unsuppressed viral loads [6].

The magnitude of LTFU in SSA, including Ethiopia, was high, i.e., 23.4% in South Africa [7], 57.4% in Tanzania [8], and 27.2% in Kenya [9]. In Ethiopia, a national program report revealed that, of the estimated 665,723 PLHIV, approximately 24% had not been retained in care [2]. A recent systematic review and meta-analysis conducted in Ethiopia, which reviewed a total of 45 studies with 546,250 study participants, reported that the pooled magnitude of LTFU was 15.17% [10]. Similarly, more recent studies reported a considerable proportion or incidence of LTFU, i.e., the incidence of LTFU was 11.19 per 100 person-years of observation in Gondar [11], 9.1% of HIV-positive individuals experienced loss to follow-up in southern Ethiopia [12], and 9.7 per 100 person-years of observation in northwest Ethiopia [13]. Another study from Northwest Ethiopia reported the highest incidence of 22 per 1000 person-months at the 6th month after ART initiation [14] and 6.7 per 100 person-years after two years in Gondar Ethiopia [15]. Various studies have identified multiple risk factors. Socio-demographic factors include being male, young, unmarried, a rural dweller, and unemployed [14, 16]. Behavioral factors include poor drug adherence, lack of disclosure, missed appointments, drug use, alcohol consumption, and smoking [17]. Clinical and treatment-related factors include poor functional status, low CD4 count, coinfections, advanced clinical stage, and mental illness diagnoses [9, 18–20]. System-level factors include receiving care at higher levels versus primary care and being on an appointment spacing model [21, 22].

Moreover, understanding the critical periods during which patients are most likely to be lost to follow-up and recognizing the factors that predict LTFU are critical for developing targeted interventions to retain patients in care [23]. Studies have reported that the first 12 months after the initiation of HIV care are particularly vulnerable, with distinct risk periods [14, 24]. A significant number of patients discontinued HIV care during this period [21, 24–26], with the highest LTFU rates occurring within the first six months [21, 25–27]. This study aimed to analyze the rate of LTFU at critical risk periods and its predictors among adult patients newly enrolled in HIV care. By addressing this gap, researchers and healthcare providers can develop more effective strategies to reduce LTFU, enhance treatment outcomes, and improve the overall quality of healthcare delivery.

Methods

Study design and settings

This retrospective cohort study was conducted in low-resource urban settings in central Ethiopia, including Addis Ababa and surrounding areas of the Oromia regional state. The HIV prevalence in this urban area is estimated at 3.4% [28], with over 100,289 HIV-positive individuals receiving antiretroviral therapy (ART) services across 144 health facilities [29].

Patient population

The participants were adults living with HIV, aged 15 years or older, who were newly enrolled in antiretroviral therapy (ART) between May 2, 2022, and April 30, 2024, and had attended the clinic at least once since starting ART. For the LTFU outcome, visit data were included only where a valid next scheduled visit date was recorded. Additionally, patients with incomplete information about outcome variables were excluded from this study. Furthermore, patients who were transferred in (TI), transferred out (TO), deceased, or restarted treatment were excluded from the analysis.

Sample size determination and sampling procedures

The sample size was determined according to the sample size calculation for a Poisson regression model [30]. Several factors were considered, including the desired confidence level (95%), desired power (80%), expected effect size, expected event rate, and attrition rate of 0.1. According to a previous study, the expected incidence rate of LTFU in HIV care was 0.022 person months of observation (PMO) ( during the first six months of ART [14] and 0.0056 (PMO) ( after two years [15]. The formula for sample size calculation for Poisson regression [30] was used:

Thus, with a 10% attrition rate, the sample size was estimated as 802. In order to select the calculated sample size, we first identified 21 high-case-load health facilities in central Ethiopia that had enrolled a minimum of 100 new HIV patients between 2022 and 2023 [29]. From this list, we randomly selected seven facilities: three in Addis Ababa—Zewditu Hospital (N = 136), ALERT Hospital (N = 134), and Yekatit 12 Hospital (N = 104)—and four in nearby urban areas of the Oromia regional state—Adama Teaching Hospital (N = 180), Geda Health Center (N = 107), Asella Hospital (N = 101), and Bishoftu Hospital (N = 160). Finally, we included all eligible patients from the selected facilities.

Study variables

Outcome variable

The rate (count) of LTFU in HIV care over 24 months after starting ART.

Predictor variables

factors such as age, sex, address information, marital status, education, religion, WHO stage and functional status at baseline, TB prevention therapy (TPT) status, cervical status, differentiated service delivery (DSD) enrollment categories, index case testing (ICT), regimen, adherence status, nutritional status, and schedule type.

Time Variable

Time in person‒months of observation (PMO) for each patient.

Operational definitions

Follow-up periods: Follow-up intervals of 6, 12, 18, and 24 months were used for HIV patients newly started on ART. The categorization into these specific intervals aligns with national consolidated guidelines for comprehensive HIV prevention, care, and treatment, which emphasize the importance of tailored follow-up on the basis of patient stability and treatment response [31]. Loss to follow-up (LTFU): Patients who interrupted their HIV care or ART appointment for more than 28 days from their appointment date were referred [4]. In this study, patients who missed appointments for more than 28 days were classified as LTFU, including those who dropped out of care or missed appointments for three months or more. Address information: In HIV care, a color-coded system is used to communicate the status of patient address information. Green indicates that the patient has a complete and up-to-date address, including a phone number and a detailed kebele address with a house number. Yellow signifies that the address information is either incomplete or outdated, such as missing a phone number or lacking a complete kebele address [31]. Adherence: Good adherence means taking ≥ 95% of doses (≤ 2 missed per month for 30 doses or ≤ 3 for 60 days). Poor adherence is < 85% (> 5 missed per month for 30 doses or > 9 for 60) (34). WHO stage: WHO stage 3 or 4 indicates advanced disease in adults or adolescents with CD4 < 200 cells/mm3 [31]. Index case testing (ICT): Identifying and testing contacts of HIV-positive individuals, prioritizing eligible patients (newly diagnosed, high viral load, restarting ART) for efficient resource use in prevention and treatment. Differentiated service delivery (DSD): person-centered HIV prevention, testing, and treatment approach [31]. Nutritional status: underweight, normal, or overweight on the basis of body weight relative to height or age [31]. Schedule type: ‘Scheduled’ referred to planned appointments and follow-ups, while ‘unscheduled’ denotes as-needed visits without prior planning, including those who missed their appointment dates by less than 7 days but had not yet met the criteria for being considered LTFU [31].

Data collection and quality control

The data extraction tool was based on the Ethiopian National HIV Care/ART Intake and ART Follow-Up Forms for routinely collected patient chronic care. De-identified patient data were extracted by each health facility’s data manager and supervised by two experienced supervisors. Two-day training was provided for the data collection facilitators by the research team. The principal investigator and research team trained the facilitators and supervisors on the abstraction tool, data management protocol, data extraction, and confidentiality issues. A pretest of the data extraction tool was conducted at an ART health facility different from the study sites. Data collection facilitators and the respective health facility’s data managers were blinded to the outcome variable and extracted de-identified patient data. Before EMR data were extracted, common data quality issues such as duplication, completeness, consistency, and validation were checked and ensured via the data quality assurance features in SmartCare-ART [32].

Data processing and analysis

The data were initially recorded in a spreadsheet format within the Electronic Medical Records (EMR) database and then checked and cleaned before being exported to SPSS version 26 for further cleaning, editing, coding, and analysis. Exploratory data analysis was conducted to identify and address missing values. Descriptive statistics, including medians, percentages, and frequencies, were computed and presented via text, tables, and graphs. The analysis of the proportion and incidence rate of LTFU was conducted by examining critical risk periods. Cumulative hazard analysis was employed to compare event distributions across these risk periods. A Poisson regression model was employed to identify factors associated with the rate of loss to follow-up (LTFU). Initially, a bivariable Poisson regression analysis (p value ≤ 0.25) was conducted to identify candidate variables for inclusion in the multivariable Poisson regression. Model adequacy was assessed via maximum likelihood estimation, with higher values indicating better model fit. The adjusted incidence rate ratio (aIRR), along with the corresponding 95% confidence interval (CI) and p value, was used to evaluate the strength of the associations and their statistical significance. A p value of < 0.05 was considered statistically significant.

Ethics

This study was approved by Addis Ababa University (AAU), College of Health Science (CHS) ethical review board (IRB) (No. 061/23/SPH, September 20, 2023). Informed consent requirements were waived by the CHS Ethics Committee. The confidentiality of the information provided was strictly maintained and used solely for the purposes of this study. All procedures adhered to applicable guidelines and regulations.

Results

Socio-demographic characteristics

During the study period, a total of 922 newly tested HIV-positive individuals started ART at the 7 high-case-load facilities recorded in the EMR database. Of these, 185 did not meet the inclusion criteria (40 were children under 15 years old, 72 were transfers out, 62 were deceased, and 11 were restarts). After excluding ineligible patients, the study analyzed a total of 737 patients, which accounts for 92% of the originally planned 802. Three-fifths of the patients, 442 (60.0%), were female. The mean age of the patients was 37.78 years (± 10.85 SD), with the following age categories: 184 (25.0%) aged 15–29 years, 352 (47.8%) aged 30–44 years, and 201 (27.3%) were aged over 44 years. Of the patients, 317 (43.0%) were married, 177 (24.0%) single, 172 (23.3%) divorced, and 71 (9.6%) widowed. Around two-fifth of the participants, 277 (37.6%) and 200 (27.1%)) had primary and secondary education levels, respectively. Regarding the address information collected during enrollment practices in Ethiopia, 624 (84.7%) patients were coded as ‘green,’ indicating that their full address, including phone number and location information, was complete. In contrast, 113 (15.3%) patients were coded as ‘yellow,’ signifying that some required address information was missing (Table 1).

Table 1.

Socio-demographic characteristics of HIV patients, Ethiopia, 2022–2024 (n = 737)

Characteristics	Frequency	Percent
Age (years)
15–29	184	25.0
30–44	352	47.8
44+	201	27.3
Gender
Male	295	40.0
Female	442	60.0
Address
Incomplete	113	15.3
Complete	624	84.7
Marital status
Single	177	24.0
Married	317	43.0
Divorced	172	23.3
Widowed	71	9.6
Education
No formal education	138	18.7
Primary (1–8)	277	37.6
Secondary (9–12)	200	27.1
College/above	122	16.6
Religion
Orthodox	497	67.4
Protestant	121	16.4
Muslim	95	12.9
Others*	24	3.3

Others* = Catholic, Wakefeta

Clinical characteristics

The categorization of patients on the basis of their follow-up periods since initiating ART revealed that 115 patients (15.6%) were in the first 6 months of treatment, while 266 patients (36.1%) fell within the 7- to 12-month range, and 356 (48.3%) were between 13 and 24 months. At baseline, 540 (73.3%) patients were in WHO stage 1 or 2, and 649 (88.1%) had working functional status. With respect to the TB prevention therapy (TPT) status of patients, 541 (73.4%) completed prevention therapy, whereas 196 (26.6%) had not started therapy. The majority, 520 (70.6%) patients, were eligible for ICT, meaning that they were prioritized for partner and contact testing, and 644 (87.4%) received the preferred 1st-line regimen. Two-thirds of the patients, 502 (68.1%), had a body weight that was within the normal range, whereas 128 (17.4%) were within the range of underweight, and 107 (14.5%) were overweight. Regarding compliance with their medication, 595 (80.7%) patients had good adherence. Around two-thirds of patients, 493 (66.9%), had unscheduled appointments regarding their follow-up planning, including missing their appointment dates by less than 7 days, but had not yet met the criteria for being considered LTFU (Table 2).

Table 2.

Clinical characteristics of HIV patients, Ethiopia, 2022–2024

Characteristics	Frequency (n)	Percent (%)
Follow-up time intervals
0–6 months	115	15.6
7–12 months	266	36.1
13–24 months	356	48.3
WHO Stage at baseline
Stage 1/2	540	73.3
Stage 3/4	197	26.7
Functional status at baseline
Working	649	88.1
Ambulatory/Bedridden	88	11.9
TPT status
Completed/started prevention therapy	541	73.4
Not started therapy	196	26.6
Cervical Status
Screened	184	25.0
Not eligible (e.g., males)	355	48.2
Not screened	198	26.9
DSD enrollment categories
Not enrolled to any DSD	264	35.8
3MMD	265	36.0
ASM/6MMD	61	8.3
DSD for MCH	35	4.7
ADH	34	4.6
Others**	78	10.6
Index Case Testing
Eligible	520	70.6
Not eligible	217	29.4
Regimen
Preferred 1st line regimen	644	87.4
Alternative 1st line regimen or 2nd line regimen	93	12.6
Adherence status
Poor	142	19.3
Good	595	80.7
Nutritional status
Underweight	128	17.4
Overweight	107	14.5
Normal	502	68.1
Schedule type
Unscheduled	493	66.9
Scheduled	244	33.1

Abbreviations ADH = Advanced Disease Model; ART = Anti-retroviral Therapy; ASM = Appointment Spacing Model; DSD = Differentiated Service Delivery. MMD = Multimonth Dispensing; TPT = TB Prevention Therapy; WHO = World Health Organization; Others** = Other DSD models such as community-based models, key populations, adolescents, and young people; and PMTCT (Prevention of Mother-to-Child Transmission) models

Incidence rate of loss to follow-up (LTFU)

In a study involving 737 individuals newly enrolled in HIV care, 165 participants (22.4%, 95% CI: 19.5–25.2) were lost to follow-up by the end of a two-year period. The timing of loss to follow-up (LTFU) varied significantly, with 50.1% occurring within the first 6 months, 29.7% occurring between 7 and 12 months, and 19.4% occurring from 13 to 24 months on antiretroviral therapy (ART).

Throughout the study, participants contributed a total of 9,022 person‒months of observation. Notably, 501 of these person-months were contributed by patients during the initial 6 months of ART. The overall incidence rate of LTFU was 18.3 per 1,000 person‒months of observation (PMO) (95% CI: 15.9–20.6). When the incidence rates across different time periods were analyzed, the first 6 months had an incidence rate of 167.7 per 1,000 PMO (95% CI: 147.6–185.7); from 7 to 12 months, the rate was 55.4 per 1,000 PMO (95% CI: 41.8–70.0); and from 13 to 24 months, the rate was 18.1 per 1,000 PMO (95% CI: 12.5–23.8) (Table 3). The cumulative hazard graph of LTFU showed a pronounced peak during the first 6 months, remained elevated during the 7–12 month period, and then stabilized afterwards (Fig. 1).

Table 3.

Incidence rates of LTFU across different time periods after ART initiation among HIV patients, Ethiopia, 2022—2024

Cohort time (Months)	Person time in months	LTFU	Rate (per 1000 PMO)	95%CI
0–6	501	84	167.7	147.6–185.7
7–12	885	49	55.4	41.8–70.0
13–24	1768	32	18.1	12.5–23.8

Abbreviations CI = Confidence Intervals, LTFU = Loss to Follow Up, PMO = Person Months of Observation

Fig. 1.

Cumulative hazard of LTFU over risk periods among HIV patients in Ethiopia, 2022–2024

Factors predicting the rate of LTFU

In an unadjusted Poisson regression analysis, patient age, sex, address information, WHO stages at baseline, patient functional status at baseline, follow-up time interval, TB preventive therapy (TPT) status, adherence, nutritional status, and schedule type were factors predicting rates of LTFU (P value < 0.25). However, after adjusting for potential confounders in the multivariable Poisson regression analysis, patient address information, follow-up time interval, and adherence were factors that significantly predicted rates of LTFU within the first 24 months (P value < 0.05). The incidence rate of LTFU in the first 6 months was 9 times higher than that in the 13–24 month period (IRR: 9.22; 95% CI: 5.97, 14.26). The incidence rate of LTFU between 7 and 12 months was three times higher than that at 13–24 months (IRR: 3.02; 95% CI: 1.92, 4.74). Patients with an incomplete address lacking phone number or location information had a 61% higher incidence of LTFU than did those with complete and up-to-date address information (IRR: 1.61; 95% CI: 1.14, 2.27). Patients with poor adherence to ART had a 78% greater incidence of LTFU than did those with good adherence (IRR: 1.78; 95% CI: 1.28, 2.48) (Table 4).

Table 4.

Poisson regression analysis to identify predictors of LTFU among HIV patients, Ethiopia, 2022–2024

Characteristics	#LTFU	Unadjusted IRR (95% CI)	P valuez	Adjusted IRR (95% CI)	P value
Duration on ART
0–6 months	84	13.40(8.92, 20.13)	< 0.001	9.22(5.97, 14.26)	< 0.001
7–12 months	49	3.38(2.16, 5.28)	< 0.001	3.02(1.92, 4.74)	< 0.001
13–24 months	32	1.00		1.00
Age (year)
15–29	46	1.56(1.01, 2.43)	0.049	1.40(0.87, 2.27)	0.170
30–44	85	1.45(0.97, 2.15)	0.070	1.34(0.89, 2.02)	0.159
44+	34	1.00		1.00
Gender
Male	75	1.28(0.94, 1.74)	0.116	1.20(0.86, 1.68)	0.285
Female	90	1.00		1.00
Address information
Incomplete	50	2.57(1.84, 3.58)	< 0.001	1.61(1.14, 2.27)	0.007
Complete	115	1.00		1.00
WHO stage
Stage 1 or 2	128	1.29(0.89, 1.85)	0.176	1.13(0.75, 1.71)	0.564
Stage 3 or 4	37	1.00		1.00
Functional status
Working	139	0.72(0.48, 1.10)	0.128	0.78(0.48, 1.26)	0.307
Ambulatory/Bedridden	26	1.00		1.00
TPT Status
Not started therapy	71	2.33(1.71, 3.17)	< 0.001	1.39(0.99, 1.94)	0.056
Completed/started prevention therapy	94	1.00		1.00
Adherence
Poor	64	2.81(2.05, 3.84)	< 0.001	1.78(1.28, 2.48)	0.001
Good	101	1.00		1.00
Nutritional status
Underweight	29	0.95(0.63, 1.42)	0.796	0.88(0.58, 1.36)	0.570
Overweight	15	0.55(0.32, 0.94)	0.030	0.75(0.43, 1.29)	0.293
Normal	121	1.00		1.00
Schedule type
Unscheduled	130	1.87(1.29, 2.72)	0.001	1.44(0.98, 2.12)	0.062
Scheduled	35	1.00		1.00

Abbreviations ART = Anti-retroviral Therapy; CI = Confidence Interval; IRR = Incidence Rate Ratio; LTFU = Loss to Follow-Up; TPT = TB Prevention Therapy; WHO = World Health Organization

Discussion

The overall incidence rate of LTFU was 18.3 (95% CI: 15.9–20.6) per 1000 person‒months of observation (PMOs). This incidence rate aligns with studies reported from Uganda [33]. However, it was higher than that reported in studies conducted in Cameroon [34] and several previous studies in Ethiopia [14, 35–37]. On the other hand, this rate was lower than that reported in studies conducted in Malawi [21] and Kenya [38]. The variation in incidence rates might be explained by factors such as differences in follow-up periods, socioeconomic disparities, variations in sample sizes, and differences in the level of care provided across settings [21, 22]. For example, in our study, we observed data from high-volume health facilities, which might increase the likelihood of patient LTFU [29].

First 6 months

The incidence rates of LTFU varied significantly over time. This study revealed that 50.1% of LTFUs occurred during the first 6 months of ART, with an incidence rate of 167.7% lost per 1000 PMOs (95% CI: 147.6–185.7). This finding was consistent with rates reported from South Africa, which reported that half of the patients were lost to follow-up within six months of starting ART [24]. This finding was also confirmed through adjusted Poisson regression analysis stratified by the follow-up period, which revealed that the incidence rate of LTFU in the first 6 months was 9 times higher than that in the 13–24 month period. The high risk of LTFU at the early ART phase could suggest that patients were most vulnerable to discontinuing treatment shortly after the beginning of ART. Other studies support this finding as well [21, 24–26]. Studies have shown that patients often face numerous challenges when starting ART. This high rate of LTFU in the initial months after starting ART highlights a critical window where interventions are urgently needed. The possible explanations might be adjustment difficulties, such as struggling with drug side effects and a lack of adequate patient support or follow-up. According to the health belief model, patients might not yet see or feel the benefits of the treatment, reducing their motivation to continue [39].

Between 7 and 12 months

Although there was a context for the decrease in LTFU during the time interval of 7–12 months, it was still a critical period. This study revealed that 29.7% of the total LTFUs occurred between 7 and 12 months on ART, with an incidence rate of 55.4 per 1000 PMOs (95% CI: 41.8–70.0). The findings were consistent with rates reported from South Africa [24]. While the risk of LTFU decreases compared with that in the first 6 months, it remains significant during the 7–12 month period. Other studies also reported that retention improved significantly after the initial months, with retention rates rising from 80.1% at 6 months to 95.7% between 6 and 12 months [40]. The adjusted Poisson regression analysis also confirmed that the incidence rate of LTFU between 7 and 12 months was three times higher than that in the 13–24 month period. Other studies support this finding as well [14, 24, 37]. The possible explanation for the decreasing incidence rate could be that some patients might overcome initial challenges, but others continue to face barriers such as socioeconomic factors, stigma, and logistical issues that affect patients’ ability to stay in care.

After 12 months

After one year of ART, the incidence rate further decreased to 18.1 per 1000 PMO (95% CI: 12.5–23.8). This finding was supported by other studies from sub-Saharan LTFUs at 36 months, which have decreased in recent years, with most losses occurring within the first year of follow-up [38]. Similarly, a study conducted in Tigray, Ethiopia, reported an overall retention rate of 85.1% after one year of ART, with lower rates of loss to follow-up (5.5%) [41]. This decline to a level closer to the overall incidence rate suggests that after the first year, patients may become more stable and engage in their treatment as they progress beyond the initial critical phase of ART. In other words, patients were more likely to have established routines, integrated ART into their daily lives, and experienced significant health improvements, reinforcing their commitment to the therapy.

Apart from the risk periods, a patient’s address information (i.e., a phone number and a detailed kebele address with a house number) is crucial for ensuring effective follow-up and continuity of care. During the identification of probable LTFU patients, electronic medical record (EMR) databases, complete addresses, and other data sources are useful for obtaining information on patients; phone calls and home visits are the most common strategies for contacting these patients [42]. In this study, patients with incomplete address information, including phone number and location information had a 61% higher incidence of LTFU compared to those with complete addresses. Similarly, a study from Uganda reported that the lack of a telephone significantly increased the risk of LTFU [16, 33]. There might be concerns about patients being lost due to incomplete address information, making it difficult for case managers to reach them for follow-up appointments, medication deliveries, or interventions when adherence issues arise. This might be the case for clients such as homeless individuals, daily laborers, or those from rural areas, making it challenging to obtain all necessary details.

Finally, this study revealed that adherence status was significantly predictive of LTFU. Patients with poor adherence to HIV care had a 78% higher incidence of LTFU than did those with good adherence. This finding was supported by several other studies [22, 25, 35, 36]. A possible explanation could be that poor adherence leads to suboptimal health outcomes, increasing the likelihood of discontinuing treatment. Factors contributing to poor adherence, such as side effects, stigma, and socioeconomic challenges, may also contribute to LTFU. Additionally, poor adherence could be indicative of underlying behavioral patterns that affect consistency in healthcare engagement.

Limitations of the study

The primary limitation of this study was the presence of missing data, particularly regarding viral load tests, which constrained our analysis and discussion of viral load suppression among patients in care for over six months. Additionally, due to reliance on secondary data, important socioeconomic and behavioral factors were omitted, which could have influenced the findings. Lastly, while deceased patients and transfers were excluded, undocumented deaths and self-transfers may still have impacted the outcome, representing another limitation.

Conclusion

The findings of this study indicated that the first six months represent the critical risk period for LTFU in HIV care, with half of the total losses occurring during this time. Although the rate of LTFU decreased after the initial six months, it remained significant during the 7–12-month period. By the 13–24-month period, the incidence rate was much lower and nearly stable as time progressed. Incomplete patient address information and poor adherence status were factors predicting the rate of LTFU. To effectively reduce the future risk of LTFU, health facility HIV care and treatment program managers, along with healthcare providers, should prioritize providing early and intensive support during the initial six months of care, followed by sustained efforts and monitoring in the next six months.

Implications of the study

This study underscores the critical time periods for implementing targeted interventions to reduce LTFU in HIV care. In addition, we recommend that future research evaluate the effectiveness of current patient retention strategies to identify best practices for HIV care and treatment programs.

Abbreviations

ADH

Advanced Disease Model

AIDS

Acquired Immunodeficiency Syndrome

aIRR

Adjusted Incidence Rate Ratio

ART

Anti-Retroviral Therapy

ASM

Appointment Spacing Model

DSD

Differentiated Service Delivery

EMRs

Electronic Medical Records

HIV

Human Immunodeficiency Virus

IR

Incidence Rate

LTFU

Loss to Follow-Up

MMD

Multi-Month Dispensing

PMO

Person Months of Observation

TPT

TB Prevention Therapy

WHO

World Health Organization

Author contributions

TE: conceptualizing, designing, facilitating data collection, analyzing data, interpreting the results, and drafting of the manuscript; WD: designing and supervising the overall research process, including the data collection process, data analysis, result interpretation, and critical revision of the manuscript; GT: designing, supervising the data collection process, data analysis, result interpretation, and critical revision of the manuscript.

Funding

This work would not be possible without the financial support of the Doris Duke Charitable Foundation (DDCF) under grant number 2017187. The mission of the DDCF is to improve the quality of people’s lives through grants supporting performing arts, environmental conservation, medical research, and child well-being and through the preservation of the cultural and environmental legacy of Doris Duke’s properties. The funder had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The data are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by Addis Ababa University (AAU), College of Health Science (CHS) ethical review committee (IRC) (No. 061/23/SPH, September 20, 2023). Informed consent requirements were waived by the CHS Ethics Committee.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.