The affected factors of loss to follow up (LFU) among HIV patients with antiretroviral therapy (ART) in Dr. Sardjito General Hospital, Yogyakarta, Indonesia

HIV infection is a global issue which is related to the increasing cases of HIV and AIDS in various countries including in Indonesia. Antiretroviral therapy (ART) that addressed for reducing the virus proliferation is not always followed by good medication adherence among patients. The ART loss to follow up (LFU) is a common problem in Indonesia. However, information concerning the affected factors of LFU among HIV patients with ART is limited. This study aimed to investigate the affected factors of LFU among HIV patients in the early two years of ART. A retrospective cohort study was conducted in Dr. Sardjito General Hospital, Yogyakarta among 369 medical records of HIV patients that met the inclusion and exclusion criteria. The inclusion criteria were HIV patients who started ART in January 2008 to December 2012, aged > 17 years with a complete medical records. The data of sociodemography and medical status patients were obtained from medical record, ART, and pharmacy register of the patients. Patient characteristics observed in this study included demographic factors such as age, sex, residence, education level, risk factors, and also medical status factors such as clinical stage, tuberculosis (TB) co-infection, functional status, and CD4 cell counts. The data were analyzed using Kaplan Meier and Cox Proportional Hazard. The results showed that patients with TB co-infection in early therapy and working functional status were significant factors of LFU (p<0.05). Patients without TB co-infection were half as much protected from LFU compared to patients with TB infection (HR=0.50; 96%CI: 0.34-0.75). It can be concluded that TB coinfection in early therapy and working fuctional status are the significant factors that influenced the LFU incidents in the two years of early therapy.


Subject
This study used the retrospective cohort methods. Secondary data were collected from medical record, ART register and pharmacy register. Patients observation was carried out for two years since ART start. The inclusions criteria were adult patients aged over 17 years who start ART from January 2008 to December 2012. The selected subject were observed until incidents of drop out or ART LFU. The LFU patient were defined as patients who did not come to the clinic for three consecutive months. The observation periods were two years. Patient confidentiality was protected by not including the patient's identity and changing the patient's identity using a subject code.

Procedure of study
Data ware collected from medical records, national ART register, and pharmacy register of the patients who started ART on Januari 2008 -Desember 2012. The identities and clinical data of the patients were observed on early two years ART. The observation was conducted for the LFU patients and saw the affected factors, based on demographic characteristic or medical status. The patients criteria for analyzed were adult patients more than 17 years. The sociodemographic data were age, residence, working status, marital status, intravenous drug use (IDU) and non IDU risk factor. The clinical data were WHO clinical stadium, CD4, TB co-infection and functional status.

Statistical analysis
Kaplan Meier was used to analyzing the survival of the time used of LFU in two years observation. Cox Proportional Hazard as the univariate analysis was used to evaluate the correlation between independent and dependent variables. The multivariate analysis was performed to identify the independent and dependent variables together.

Characteristic of subjects
Three hundred and sixty nine eligible subjects were involved in this study. The characteristic of subjects are presented in TABLE 1.

DISCUSSION
The study result from restrospective kohort have been identified at the end of LFU observation were 16.41%. 8,9 Observation total time were 6119.2 person time with death rate 2.61 by 100 peoples by month.
World Health Organization defines LFU as patients who has not been seen at the clinic for at least 90 days (three months) after the last missed appointment. 10 Some factors can influence patient retention in HIV care such as site specific factor, health system weaknesses, financial barriers, patient/caregiver characteristics, and community factors. 11 From the bivariate analysis, we had variables that have significant relation with LFU incident which is TB coinfection and working functional status. Patients who started ART with TB coinfection had more risks to LFU incidents (95% CI: 0.34-075). Patients with working functional status when ART started had 1.73 more risk of LFU than ambulatories patients (95%CI: 1.17-2.56).
From the medical records distribution we found patients who started ART on late WHO clinical stadium, CD4 numbers <50 cell/mm 3 , working functional status and non IDUs risk factor, did not have significant relation with LFU incidents. These were not appropriate with the study in Zambia and Swiss which had different results. 8 The factors related to the LFU incidents were III/IV Clinical Stadium and CD4 cell count numbers < 100 cell/mm 3 . 8 Study in Zambia had different results with Swiss, which LFU incidents were found among patients with high CD4 numbers in the early treatment and early clinical stadium. 8 HIV patients from IDUs risk factor were having stigma as a bad temper and undiscipline peoples. 12 Bivariate analysis showed that IDUs risk factor does not have significant relation with LFU incidents (p > 0.985), while the TB coinfection were have significant relation (p < 0.05) to increase the LFU incidents among HIV patients with ART for two years.
The high death risk among patients with TB coinfection was related to the bad immune repair process on ART. Age over 40 y.o. were having risk of chronic diseases that could cause death. 13,14 Bad response on ART, rapid development of chronic disease and complication disease such as cancer dan cadiovascular diseases were combined effect from aging, HIV infection and ART effects that could increase death risks because of bad immune recovery on ART. 8, 15 These different results showed that there were different patient responses on ART because of the different characteristic and cultural population di every country. 16 The highest number of LFU occured in early 6 months after ART started were 92.9 per 100 peoples-years.
This study has limitations such as the loss of information related to patient data due to incomplete data from the ARV register, medical records, and pharmacy register and differences in numbering of patient's identities in the ARV register, medical records, and pharmacy register.

CONCLUSION
Tuberculosis co-infection in early ART significantly increases the risk of LFU among HIV patients in early two years ART, where the highest incidents occurs on the first sixth months. In this study, working functional status also influences the LFU incidents. Emphasis needs to be placed on improving followup and documentation of patients care on HIV treatments. Community-based approach such as peer support groups (kelompok pendukung sebaya/KDS) and health information system integration such as SMS reminder designed for atrisk population (patients with TB confection and working functional status) can be further studied to reduce LFU incidents.