Study design and population
We conducted a nationwide, multicenter, cross-sectional study which recruited HIV-infected patients receiving medical care in 14 centers for HIV comprehensive care from 13 Colombian cities (Bogotá, Medellín, Cali, Barranquilla, Montería, Cartagena, Bucaramanga, Cúcuta, Pereira, Villavicencio, Manizales, Florencia, and Pasto). From July 2017 to March 2018, treatment-naïve HIV-positive patients aged 18 years or older were consecutively included in the study, regardless of the clinical stage of HIV infection, lymphocyte (LT) CD4+ cell count or HIV viral load, time of diagnosis. We exclude patients on antiretroviral treatment or had been previously treated to avoid selection bias. All patients had been previously confirmed for HIV infection employing immunoenzymatic assays, HIV rapid tests, or Western blot tests, as defined and recommended by the clinical practice guidelines of the Colombian Ministry of Health and Social Protection and the Colombian Association of Infectious Diseases (ACIN).3
The sample size was calculated using the proportion formula for finite populations based on an expected HLA-B*57:01 prevalence of 3% +/- 1%, an estimated population of 6000 people, a 95% confidence level, and an additional 2% to prevent patient loss. The minimum number of patients needed to estimate the prevalence of the allele was calculated as 961 patients. The cities and the corresponding sample size were defined for this study according to the regional prevalence known for Colombia for the year 2016. These cities have a population of diverse racial background and accounted for around 80% of the HIV population according to data from the Colombian non-governmental organization Cuenta de Alto Costo (CAC).28 Colombians are descendant from three racial groups (Caucasians, Africans, Amerindians). The ethnic composition of the Colombian population is an admixture estimated to result in mestizo (50%), Caucasian (25%), mulatto and Zambo (20%), afro-Colombian (4%), and indigenous (1%).29
Data collectors were the professionals in each institution in charge of recruiting the participants. They were trained to fill out a standardized form containing sociodemographic and clinical data, including sex, age, race, the department of origin, time since HIV diagnosis, baseline LT CD4 cell count, baseline HIV viral load, and CDC clinical stage at diagnosis. Each patient had a unique identification code used for both the case report form, and the label for a 5-mL blood sample collected after obtaining written informed consent.
Patient race was classified into four groups, Caucasians, mestizos, Afro-Colombian, and indigenous, considering the characteristics recommended in public health in Colombia, which included: characteristics of the hair, facial features, skin color, self-recognition and place of birth. The departments of origin were classified in the following zones: the Northern zone with predominance of Afro-Colombians and mestizos (departments of Atlántico, Bolívar, Sucre, Magdalena, Córdoba, Cesar, and Norte de Santander); the Central zone with predominance of Caucasians and mestizos departments of Antioquia, Caldas, Risaralda, and Quindío); The Western zone, predominance of Afro-Colombians and mestizos (departments of Chocó, Cauca, Nariño, and Valle); the Eastern (departments of Arauca, Caquetá, Casanare, Guaviare, Meta) and the Andean zones (Bogotá Distrito Especial, and departments of Cundinamarca, Huila, Tolima, Boyacá, and Santander) both with predominance of mestizos.
In two molecular biology laboratories (Primed Laboratory in Barranquilla and Laboratorio de Genética y Biología Molecular in Bogotá), blood samples were screened for HLA-B*57:01 carriage by using allele and group-specific polymerase chain reaction-sequence-specific primers (PCR-SSP) typing. This technique has been previously validated to be a reliable method of distinguishing between HLA-B*57:01 and other commonly occurring -B*57:02 and -B*57:03 alleles, with a sensitivity of 99,4 and 100% in two different studies, and specificity of 100% in both.30-32
Data describing clinical and demographic patient characteristics were summarized using medians with interquartile ranges (IQR) for continuous variables and frequencies and proportions for categorical variables. Data were entered into EpiInfo 6.04 (Centers for Disease Control and Prevention, Atlanta, USA) and then exported to Stata version 12.0 (StataCorp, College Station, TX, USA) for analysis statistical comparisons of patient characteristics by HLA-B*57:01 were made with logistic regression to evaluate the significance of differences in allelic frequencies between sex, ethnicity, and places of residence.