Abstract
Background and Objectives
Recruiting blood donors among a population with a high human immunodeficiency virus (HIV) burden requires detailed information on HIV risks. We studied demographic and behavioural risk factors for incident HIV infection among blood donors in South Africa.
Materials and Methods
We conducted a case–control study. Incident HIV was defined as HIV antibody negative and RNA positive, or concordant serology and RNA positive with a limiting antigen avidity assay optical density of <1.5. Cases were matched to infection‐negative controls (ratio 1:3) on race, age and geography. Risk factors in the 6 months before donation were ascertained by audio computer‐assisted self‐interview. Data were fitted using separate multivariable logistic regression models for males and females.
Results
From April 2014 to March 2017, we enrolled 323 people with incident HIV and 877 controls. Among women, incident HIV was associated with sex with a person living with HIV (PLWH) or unknown HIV status, multiple male sex partners, never or occasional condom use, anal preparation before sex, first‐time donor status and referral to donation by a healthcare worker. Among men, incident HIV was associated with being aged 31–40 years, sex with a PLWH or unknown HIV status, multiple sex partners, more than four lifetime male sex partners, gay/bisexual identity, marriage or stable partnership, lower education, penetrative injury, occasional condom use and first‐time or lapsed donor status. Some novel or indirect risks for incident HIV were also observed.
Conclusion
We confirmed the known sexual behaviours asked on the donor screening questionnaire. The findings highlight ongoing challenges in donor disclosure during selection and the importance of donor education.
Keywords: blood donation, HIV, incidence, risk factors
Highlights.
The study confirmed known sexual risks as primary risk factors for incident infections in South African blood donors.
Risk factors were different for men and women, and included some unexpected risk behaviours.
The challenges involved in obtaining donor disclosure during selection and the importance of donor education are evident.
INTRODUCTION
In 2023, approximately 39.9 million people were living with human immunodeficiency virus (HIV)/acquired immune deficiency syndrome (AIDS) (PLWH) worldwide, with an estimated 1.3 million new infections per year, with AIDS‐related deaths reaching 630,000 [1]. About 60% of all PLWH live in sub‐Saharan Africa [1, 2, 3]. The reported routes of HIV acquisition in sub‐Saharan Africa are (i) perinatally, (ii) through heterosexual sex and (iii) through transfusion of blood and blood products, with intravenous drug use and men having sex with men (MSM) being less frequently reported [4, 5, 6]. HIV remains one of the primary public health concerns in South Africa. The mid‐year population prevalence estimate for 2022 was 13.9% of the total population. The total number of PLWH in South Africa is estimated to be 8.45 million [4].
The South African National Blood Service (SANBS), one of two blood collection agencies in South Africa, collects more than 950,000 units of whole blood each year from a pool of close to 550,000 donors. Against the backdrop of one of the world's highest burdens of HIV, South Africa's need for blood to support medical care continues to grow. Several strategies to reduce the risk of HIV in the blood supply have been successfully implemented. First, reliance on 100% voluntary, non‐remunerated donors and exclusion of high‐risk donors identified through the donor history questionnaire (DHQ) contribute to blood safety. Second, each donation is tested in parallel by individual donation nucleic acid testing (ID‐NAT) and serology for HIV, hepatitis B and C viruses (HBV and HCV) and serology for syphilis. Furthermore, a collection triage policy that limits the production of specialized products to donations from highly regular donors is used [5].
Prevalent HIV, although more amenable to study, provides an incomplete assessment of the risk of recently acquired infection and thus of risk to the blood supply. Surveillance of incident HIV provides a more accurate assessment of currently relevant risk factors. In addition, a higher HIV incidence for female first‐time donors is evident in our setting [7]. The blood donor selection process focuses on higher risk behavioural exposures in the 3–6 months before donation. Questions on the DHQ ask about number of sex partners, having a new sex partner as well as related questions like knowing the HIV status of sex partners. Other questions asked are about sexually transmitted diseases, injection drug use, sexual violence and other exposures that may be associated with the risk of HIV or other infections. This study sought to identify the risk factors for incident HIV in prospective donors with recently acquired infection.
MATERIALS AND METHODS
Overview of study design and objectives
We conducted a matched case–control study of risk factors for incident HIV in blood donors at SANBS in South Africa. The South Africa HIV Response and Policy Synthesis report of 2011 [8] identified multiple serial and concurrent partners, inconsistent condom use, transactional sex and unprotected anal intercourse in homosexual and heterosexual contacts as HIV infection risk factors in the generation population. We hypothesized that the risk factors for incident HIV in donors in South Africa would include a recent change in sex partners, older male/younger female sexual partnerships, a greater number of recent sex partners, condomless anal intercourse and lower socioeconomic status. Enrolment and follow‐up occurred over the period from April 2014 to March 2017.
Donation testing
At the time of the study, all donations were routinely screened in parallel by ID‐NAT using the Procleix Ultrio Plus (Elite) assay on the Panther platform (Grifols, Emeryville, CA) (95% limit of detection [LOD] 18 IU/mL for HIV‐1 RNA; specificity 100%) and HIV serology on the PRISM (Abbott, Wiesbaden, Germany) (sensitivity 100% for HIV‐1 and HIV‐2; specificity 99.7). To assess recent infection in donors who had seroconverted, limiting antigen avidity (LAg) testing of concordant HIV RNA and anti‐HIV positive donations was performed in the SANBS Virology Reference Laboratory using the Sedia® HIV‐1 LAg‐Avidity Enzyme Immunoassay test (Sedia Biosciences Corporation, Portland, OR, USA) [9]. At a cutoff threshold of <1.5 normalized optical density (ODn), this assay measures the mean duration of recent infection (MDRI) of 195 days (95% confidence interval [CI]: 168–222) with an estimated false recency rate of 1.3% [9]. To support identifying and recruiting cases as soon after donation as possible, testing was conducted in real time.
Study population
Cases were defined as blood donors with incident HIV, established through either combined ID‐NAT/serology (RNA positive/serology negative) ‘NAT yield’ or (RNA positive/serology positive) and LAg Avidity recency testing. We enrolled potential controls who were selected at a 3:1 ratio to HIV cases and strata‐matched according to age (within 5 years) and race/ethnicity (Asian/Indian, Black, coloured, White). Study participants were identified from SANBS blood collection sites except for those in the Free State and Northern Cape areas where it was not possible to conduct the study because of the geographical distances the study participants and research staff would have had to travel. Donations from the included areas represent 91.3% of all donations from the SANBS territory.
Exclusion criteria included autologous and directed blood donation, deferral from donation for any reason, age less than 18 years and the inability to provide informed consent. We also excluded donors from whom insufficient volume of blood was collected to complete donation testing, that is, failed phlebotomies.
For incident HIV cases, research nurses contacted potential participants after initial laboratory screening results from their index donation were available. Appointments were scheduled, and the donors were counselled about their results, informed about the study and invited to participate. Controls were recruited either by recall after or on the same day of their blood donation. For the latter group, a small minority with positive HIV, HBV, HCV or syphilis from donation testing were subsequently excluded from the control group. Of the interviewed controls, 3% did not meet the study eligibility requirements and were excluded from analysis. All participants had to be able and willing to provide informed consent and complete a risk questionnaire using an audio computer‐assisted structured interview (ACASI) [10, 11].
Risk questionnaire
Each participant completed a confidential ACASI on motivations for blood donation and behaviours focused on exposures in the 6 months before the index donation. The ACASI risk factor interview was developed using content from previous studies of infection risks in blood donors [10]. Questions included perceived risks, sexual orientation, number of sexual partners, sexual behaviours and alcohol and drug usage. The interview content from previous studies was modified to include reported and putative risk factors in sub‐Saharan Africa for both HIV and HBV infection, including visiting a traditional healer, ritual scarification, adult circumcision and other medical procedures [12]. We assessed risk exposures over two intervals: ever and within the 6 months preceding the index donation.
Data analysis
The ACASI questionnaire responses were linked with donation testing and demographic data from the blood centre's operational databases and then extracted for analysis. The characteristics of the participants were summarized using descriptive statistics. We used Fisher's exact test to assess associations between categorical variables because of sparse observations in some strata. Results of statistical tests are reported against a level of significance p <0.05.
Because of the expected differences in risk profiles in females and males, risk factor analyses were stratified by sex. For multivariable logistic regression analyses, we identified variables with potential relations to HIV from the bivariate analysis and subject area knowledge. We used statistical learning to screen all possible models. The algorithm created all possible 1‐variable, 2‐variable, 3‐variable models, and so forth, up to a single 12‐variable model based on bivariate analysis, for a total of 4095 potential models for females. Following the same process for males, there were 13 variables and 8191 potential models. The misclassification rate for each model was calculated using five‐fold cross‐validation. For females and males separately, we investigated a small subset of all the potential models by focusing on models with the lowest misclassification rates. From this small subset of potential models, we selected a final model for each sex based on consensus among the researchers.
Research ethics
Ethical oversight was provided by the SANBS (2013/013) institutional review board (IRB), with additional approval obtained from the University of California San Francisco (UCSF; approval 14‐13943) and Research Triangle Institute (RTI) International IRBs. Observational study monitoring board review was also conducted during the enrolment period of the study for the US National Heart, Lung and Blood Institute. Participants' expenses were covered when attending study visits, and they received a small stipend for their time and effort in completing the study procedures.
RESULTS
During the study period, 2,501,022 donations were collected of which 5123 tested positive for HIV, and of these 1248 (24.4%) were incident infections. Of the latter, 323 (25.9%) were enrolled into the study along with 877 controls. Cases included 230 females and 93 males, whereas the controls included 439 females and 438 males (Table 1). The median time to HIV case interview was 34 days with an interquartile range of 17–51 days. Most participants were aged 21–30 years and from the Egoli/Vaal (around Johannesburg) and KwaZulu Natal geographic areas, and belonged the Black African population group. Because of matching, the age distribution, geographic region and population group membership were similar between cases and controls. Most of both sexes reported being single and never married, and most participants were employed. While the educational attainment of females was similar in cases and controls, male cases had significantly lower educational attainment than male controls, p = 0.017.
TABLE 1.
Demographic and donation characteristics of cases with recently acquired human immunodeficiency virus infections and matched controls, a by sex, April 2014 through March 2017.
| Characteristic | Females | Males | Total | |||
|---|---|---|---|---|---|---|
| Case | Control | Case | Control | Case | Control | |
| n (%) | n (%) | n (%) | n (%) | n (%) | n (%) | |
| Total | 230 | 439 | 93 | 438 | 323 | 877 |
| Age (years) | ||||||
| <21 | 31 (13.5) | 60 (13.7) | 11 (11.8) | 36 (8.2) | 42 (13.0) | 96 (10.9) |
| 21–30 | 133 (57.8) | 248 (56.5) | 47 (50.5) | 236 (53.9) | 180 (55.7) | 484 (55.2) |
| 31–40 | 42 (18.3) | 89 (20.3) | 24 (25.8) | 99 (22.6) | 66 (20.4) | 188 (21.4) |
| 41+ | 24 (10.4) | 41 (9.3) | 11 (11.8) | 66 (15.1) | 35 (10.8) | 107 (12.2) |
| Geographic zone/province | ||||||
| Eastern Cape | 37 (16.1) | 75 (17.1) | 5 (5.4) | 48 (11.0) | 42 (13.00) | 123 (14.0) |
| Egoli/Vaal | 62 (27.0) | 128 (29.2) | 30 (32.3) | 124 (28.3) | 92 (28.5) | 252 (28.7) |
| KwaZulu‐Natal | 46 (20.0) | 104 (23.7) | 26 (28.0) | 113 (25.8) | 72 (22.3) | 217 (24.8) |
| Mpumalanga | 40 (17.4) | 75 (17.1) | 16 (17.2) | 89 (20.3) | 56 (17.3) | 164 (18.7) |
| Northern | 45 (19.6) | 57 (13.0) | 16 (17.2) | 64 (14.6) | 61 (18.9) | 121 (13.8) |
| Race | ||||||
| Asian/Indian | 3 (0.7) | 2 (2.2) | 4 (1.0) | 2 (0.6) | 7 (0.8) | |
| Black | 218 (94.8) | 400 (91.1) | 86 (92.5) | 403 (92.0) | 304 (94.1) | 803 (91.6) |
| Coloured | 5 (2.2) | 19 (4.3) | 2 (2.2) | 13 (3.0) | 7 (2.2) | 32 (3.7) |
| White | 6 (2.6) | 17 (3.9) | 3 (3.2) | 18 (4.1) | 9 (2.8) | 35 (4.0) |
| Marital status | ||||||
| Divorced/separated/widowed | 11 (4.8) | 20 (4.6) | 2 (2.2%) | 9 (2.1) | 13 (4.0) | 29 (3.3) |
| Living with/married to a single partner | 48 (20.9) | 114 (26.0) | 25 (27.9) | 138 (31.5) | 73 (22.6) | 252 (28.7) |
| Married to more than one partner | 1 (0.2) | 3 (3.2) | 3 (0.7) | 3 (0.9) | 4 (1.2) | |
| Single, never married | 171 (74.0) | 303 (69.0) | 63 (67.7) | 288 (65.8) | 234 (72.5) | 591 (67.4) |
| Employment | ||||||
| No | 88 (38.3) | 199 (45.3) | 35 (37.6) | 145 (33.1) | 123 (38.1) | 344 (39.2) |
| Yes | 142 (61.7) | 240 (54.7) | 58 (62.4) | 293 (66.9) | 200 (61.9) | 533 (60.8) |
| Education b | ||||||
| Tertiary qualification | 95 (41.3) | 199 (45.3) | 29 (31.2) | 203 (46.4) | 124 (38.4) | 402 (47.5) |
| Incomplete high school | 22 (9.6) | 27 (6.2) | 5 (5.4) | 26 (5.9) | 27 (8.4) | 53 (6.3) |
| Incomplete tertiary | 113 (49.1) | 213 (48.5) | 59 (63.4) | 208 (47.5) | 172 (53.3) | 421 (49.7) |
Abbreviations: FET, further education and training; SA, South Africa.
Matched on race, geographic zone and age category.
Education in SA: Foundation phase: Grades R to 3. Intermediate phase: Grades 4 to 6. Senior phase: Grades 7 to 9. FET: Grades 10 to 12. Higher education and training: Universities and colleges offering various undergraduate and postgraduate degrees.
Bivariate analyses
Bivariate analyses for females showed that cases were more likely than controls to be first‐time donors, p = 0.019 (Table 2). There was no difference in whether a donation site was fixed or mobile, between cases and controls. Cases were less likely to have private medical insurance than were controls, p < 0.005. Compared to controls, cases were more likely to report sex with a PLWH and more likely to not know the HIV status of their partner(s), p < 0.001. Reported condom usage for both vaginal and anal sex differed between cases and controls, but many in both groups responded with ‘unknown’ to these questions. Condomless vaginal sex was reported more often than condomless anal sex. Fewer female cases than controls reported a homosexual or bisexual partner, p = 0.005. Cases and controls had similar distributions for the age difference between themselves and their sexual partners. These numbers were insufficient for further analyses. Compared to controls, cases were more likely to report visiting a traditional healer, p = 0.002, having a needlestick injury, p = 0.028, or having been told to donate blood by a healthcare provider, p = 0.002.
TABLE 2.
Bivariate analysis of exposures and risk behaviours reported by cases with recently acquired human immunodeficiency virus infections and matched controls, a by sex, April 2014–March 2017.
| Exposure/risk behaviour | Females | Males | ||||
|---|---|---|---|---|---|---|
| Case | Control | p | Case | Control | ||
| n (%) | n (%) | n (%) | n (%) | |||
| Total | 230 | 439 | 93 | 438 | p | |
| Marital status | ||||||
| Single, never married | 171 (74.3) | 303 (69.0) | 0.540 | 63 (67.7) | 288 (65.8) | 0.167 |
| Living with or married to single partner (including traditional marriage) | 48 (20.9) | 114 (26.0) | 25 (26.9) | 138 (31.5) | ||
| Married to more than one partner (including traditional marriage) | 1 (0.2) | 3 (3.2) | 3 (0.7) | |||
| Divorced/separated/widowed | 11 (4.8) | 20 (4.6) | 2 (2.2) | 9 (2.1) | ||
| Unknown | 1 (0.2) | |||||
| Donation status | ||||||
| First‐time | 64 (27.8) | 85 (19.4) | 0.019 | 22 (23.7) | 35 (8.0) | <0.001 |
| Lapsed | 57 (24.8) | 101 (23.0) | 19 (20.4) | 43 (9.8) | ||
| Repeat | 109 (47.4) | 253 (57.6) | 52 (55.9) | 359 (82.0) | ||
| Missing | 1 (0.2) | |||||
| Collection site | ||||||
| Fixed | 64 (27.8) | 111 (25.3) | 0.517 | 40 (43.0) | 179 (40.9) | 0.777 |
| Mobile | 166 (72.2) | 328 (74.7) | 53 (57.0) | 258 (58.9) | ||
| Missing | 1 (0.2) | |||||
| Private medical insurance | ||||||
| No | 163 (70.9) | 262 (59.7) | 0.005 | 72 (77.4) | 263 (60.0) | 0.001 |
| Yes | 67 (29.1) | 177 (40.3) | 21 (22.6) | 175 (40.0) | ||
| Number of sex partners in last 6 months | ||||||
| 0 | 79 (34.3) | 173 (39.4) | 0.069 | 31 (33.3) | 148 (33.8) | 0.005 |
| 1 | 134 (58.3) | 251 (57.2) | 43 (46.2) | 254 (58.0) | ||
| 2–3 | 14 (6.1) | 14 (3.2) | 16 (17.2) | 28 (6.4) | ||
| 4+ | 3 (1.3) | 1 (0.2) | 3 (3.2) | 8 (1.8) | ||
| Lifetime number of female sex partners | ||||||
| 0 | 144 (62.6) | 298 (67.9) | 0.447 | 9 (9.7) | 17 (3.9) | 0.133 |
| 1 | 12 (5.2) | 14 (3.2) | 9 (9.7) | 61 (13.9) | ||
| 2–3 | 8 (3.5) | 18 (4.1) | 21 (22.6) | 111 (25.3) | ||
| 4–10 | 5 (2.2) | 6 (1.4) | 35 (37.6) | 234 (53.4) | ||
| 11+ | (0.0) | (0.0) | ||||
| Unknown | 61 (26.5) | 103 (23.5) | 5 (5.4) | 15 (3.4) | ||
| Lifetime number of male sex partners | ||||||
| 0 | 1 (0.4) | 31 (7.1) | <0.001 | 50 (53.8) | 302 (68.9) | <0.001 |
| 1 | 26 (11.3) | 83 (18.9) | 2 (2.2) | 12 (2.7) | ||
| 2–3 | 87 (37.8) | 159 (36.2) | 8 (8.6) | 11 (2.5) | ||
| 4–10 | 99 (43.0) | 146 (33.3) | 16 (17.2) | 3 (0.7) | ||
| Unknown | 17 (7.4) | 20 (4.6) | 17 (18.3) | |||
| Sex with person living with HIV | ||||||
| No | 44 (19.1) | 237 (54.0) | <0.001 | 26 (28.0) | 257 (58.7) | <0.001 |
| Yes | 10 (4.3) | 4 (0.9) | 1 (1.1) | 3 (0.7) | ||
| Unknown | 176 (76.5) | 198 (45.1) | 66 (71.0) | 178 (40.6) | ||
| Condom use with vaginal sex | ||||||
| Always | 15 (6.5) | 68 (15.5) | <0.001 | 5 (5.4) | 85 (19.4) | 0.005 |
| Sometimes | 71 (30.9) | 68 (15.5) | 27 (29.0) | 92 (21.0) | ||
| Never | 32 (13.9) | 63 (14.4) | 12 (12.9) | 62 (14.2) | ||
| Did not have specified sex | 28 (12.2) | 55 (12.5) | 6 (6.5) | 36 (8.2) | ||
| Unknown | 84 (36.5) | 185 (42.1) | 43 (46.2) | 163 (37.2) | ||
| Condom use with anal sex | ||||||
| Always | 5 (2.2) | 9 (2.1) | 0.014 | 2 (2.2) | 12 (2.7) | <0.001 |
| Sometimes | 15 (6.5) | 7 (1.6) | 6 (6.5) | 4 (0.9) | ||
| Never | 7 (3.0) | 9 (2.1) | 5 (5.4) | 17 (3.9) | ||
| Did not have specified sex | 118 (51.3) | 229 (52.2) | 37 (39.8) | 243 (55.5) | ||
| Unknown | 85 (37.0) | 185 (42.1) | 43 (46.2) | 162 (37.0) | ||
| Sexual orientation | ||||||
| Bisexual/gay/homosexual | 2 (0.9) | 22 (5.0) | 0.005 | 22 (23.7) | 9 (2.1) | <0.001 |
| Straight/heterosexual | 227 (98.7) | 416 (94.8) | 71 (76.3) | 427 (97.5) | ||
| Unknown | 1 (0.4) | 1 (0.2) | 2 (0.5) | |||
| Men who have sex with men | ||||||
| No | 49 (52.7) | 276 (63.0) | <0.001 | |||
| Yes | 13 (14.0) | 12 (2.7) | ||||
| Unknown | 31 (33.3) | 150 (34.2) | ||||
| Age difference with sex partners (in years) | ||||||
| Partners more than years 10 younger | 2 (0.9) | 1 (0.2) | 0.284 | 4 (4.3) | 19 (4.3) | <0.001 |
| Partners 10–5 years younger | 2 (0.9) | 3 (0.7) | 18 (19.4) | 63 (14.4) | ||
| Partners 4 years younger to 4 years older | 91 (39.6) | 174 (39.6) | 31 (33.3) | 199 (45.4) | ||
| Partners 10–5 years older | 40 (17.4) | 66 (15.0) | 8 (8.6) | 2 (0.5) | ||
| Partners more than years 10 older | 12 (5.2) | 12 (2.7) | ||||
| Unknown | 83 (36.1) | 183 (41.7) | 32 (34.4) | 155 (35.4) | ||
| Traditional medicine/healer | ||||||
| No | 216 (93.9) | 431 (98.2) | 0.002 | 87 (93.5) | 430 (98.2) | 0.026 |
| Yes | 13 (5.7) | 5 (1.1) | 5 (5.4) | 6 (1.4) | ||
| Unknown | 1 (0.4) | 3 (0.7) | 1 (1.1) | 2 (0.5) | ||
| Needle stick/sharp object injury | ||||||
| No | 224 (97.4) | 433 (98.6) | 0.028 | 90 (96.8) | 435 (99.3) | 0.034 |
| Yes | 5 (2.2) | 1 (0.2) | 2 (2.2) | |||
| Unknown | 1 (0.4) | 5 (1.1) | 1 (1.1) | 3 (0.7) | ||
| Marijuana use (THC) | ||||||
| No | 208 (90.4) | 371 (84.5) | 0.089 | 70 (75.3) | 322 (73.5) | 0.796 |
| Yes | 22 (9.6) | 65 (14.8) | 23 (24.7) | 116 (26.5) | ||
| Unknown | 3 (0.7) | |||||
| Penetrative physical injury | ||||||
| No | 216 (93.9) | 425 (96.8) | 0.147 | 80 (86.0) | 433 (98.9) | <0.001 |
| Yes | 13 (5.7) | 12 (2.7) | 13 (14.0) | 5 (1.1) | ||
| Unknown | 1 (0.4) | 2 (0.5) | ||||
| Blood donation recommend by a healthcare worker | ||||||
| No | 207 (90.0) | 423 (96.4) | 0.002 | 88 (94.6) | 418 (95.4) | 0.663 |
| Yes | 21 (9.1) | 14 (3.2) | 5 (5.4) | 19 (4.3) | ||
| Unknown | 2 (0.9) | 2 (0.5) | 1 (0.2) | |||
| Anal cleansing before sex | ||||||
| Every time | 28 (12.2) | 14 (3.2) | <0.001 | 12 (12.9) | 24 (5.5) | 0.003 |
| Never | 145 (63.0) | 305 (69.5) | 38 (40.9) | 211 (48.2) | ||
| Once | 5 (2.2) | 6 (1.4) | 4 (4.3) | 2 (0.5) | ||
| Sometimes | 11 (4.8) | 9 (2.1) | 3 (3.2) | 5 (1.1) | ||
| Unknown | 8 (3.5) | 28 (6.4) | 8 (8.6) | 43 (9.8) | ||
| Not applicable | 33 (14.3) | 77 (17.5) | 28 (30.1) | 153 (34.9) | ||
Abbreviation: THC, tetrahydrocannabinol.
Mapped on race, geographic zone and age category.
Bivariate analyses for males showed that cases were more likely than controls to be first‐time blood donors and less likely than controls to have private medical insurance (both p < 0.001). Cases were more likely than controls to have more than one sexual partner, and this difference was largely driven by a significantly higher number of male sexual partners in cases, p < 0.001. Also, compared to controls, cases were more likely to have visited a traditional healer, p = 0.026, and more likely to report either a needlestick, p = 0.034, or a penetrative injury, p < 0.001. Circumcision ever or in the 6 months before donation was not different for male cases and controls (data not shown). Reported condom usage for both vaginal and anal sex differed between cases and controls, but many in both groups responded with ‘unknown’ to these questions. Condomless vaginal sex was reported more often than condomless anal sex.
Multivariable analyses
For females, among the potential sexual risk factors associated with incident HIV infection in the bivariate analyses, the multivariable model identified the highest odds of incident infection in women who reported having sex with a PLWH (odds ratio [OR] 27.7, 95% CI: 6.8–112.4) or not knowing the HIV status of sex partners (OR 26.7, 95% CI: 14.0–50.9) (Figure 1 and Table 3). Reporting more than one sex partner (OR 10.6, 95% CI: 1.6–68.6) and increasing numbers of male sex partners were associated with incident HIV. Sometimes (OR 4.8, 95% CI: 2.0–11.5) or never (OR 3.4, 95% CI: 1.3–8.7) using condoms for vaginal sex was associated with incident HIV compared to always using condoms. Reporting anal cleansing every time before sex was associated with incident HIV (OR 3.5, 95% CI: 1.5–8.0). Additional factors associated with incident HIV in females included not having medical insurance (OR 1.8, 95% CI: 1.1–2.8), current employment (OR 2.1, 95% CI: 1.3–3.4), a recommendation from a healthcare provider to be tested for HIV at the blood centre (OR 2.8, 95% CI: 1.0–7.5) and being a first‐time blood donor (OR 1.7, 95% CI: 1.0–2.9).
FIGURE 1.
Multivariable model of factors associated with incident HIV infection among female blood donors (left panel) and male donors (right panel). CI, confidence interval; HCW, health care worker; HIV, human immunodeficiency virus; THC, tetrahydrocannabinol.
TABLE 3.
Multivariable model of factors associated with incident human immunodeficiency virus infection among female blood donors.
| Characteristic | Odds ratio | 95% confidence interval | |
|---|---|---|---|
| Private medical insurance | |||
| Yes | Ref. | ||
| No | 1.77 | 1.11 | 2.83 |
| Age category | |||
| <21 years | 1.89 | 0.93 | 3.84 |
| 21–30 | Ref. | ||
| 31–40 | 0.73 | 0.41 | 1.30 |
| >40 | 1.12 | 0.55 | 2.28 |
| Marijuana use (THC) | |||
| No | Ref. | ||
| Yes | 0.41 | 0.20 | 0.81 |
| Current employment | |||
| No | Ref. | ||
| Yes | 2.10 | 1.28 | 3.43 |
| Sex with person living with HIV | |||
| No | Ref. | ||
| Yes | 27.72 | 6.84 | 112.44 |
| Unknown | 26.71 | 14.02 | 50.90 |
| Number of sex partners in last 6 months | |||
| 0 | Ref. | ||
| 1 | 5.40 | 1.16 | 25.20 |
| >1 | 10.55 | 1.62 | 68.57 |
| Lifetime number of male sex partners | |||
| 0–1 | Ref. | ||
| 2–3 | 2.96 | 1.54 | 5.67 |
| 4–10 | 3.54 | 1.83 | 6.82 |
| Unknown | 6.70 | 2.49 | 18.01 |
| Condom use with vaginal sex | |||
| Always | Ref. | ||
| Sometimes | 4.75 | 1.96 | 11.49 |
| Never | 3.36 | 1.30 | 8.71 |
| Unknown | 1.45 | 0.28 | 7.60 |
| No | 1.71 | 0.64 | 4.56 |
| Anal cleansing | |||
| Never | Ref. | ||
| Once | 1.24 | 0.25 | 6.23 |
| Sometimes | 1.74 | 0.47 | 6.44 |
| Every time | 3.47 | 1.51 | 7.96 |
| Unknown | 0.96 | 0.26 | 3.52 |
| Not applicable | 0.78 | 0.44 | 1.37 |
| Donation recommended by HCW | |||
| No | Ref. | ||
| Yes | 2.76 | 1.02 | 7.46 |
| Donor type | |||
| Repeat | Ref. | ||
| First‐time | 1.72 | 1.02 | 2.88 |
| Lapsed | 0.98 | 0.58 | 1.65 |
Abbreviations: HCW, health care worker; THC, tetrahydrocannabinol.
For males, among the potential sexual risk factors associated with incident HIV infection in the bivariate analyses, the multivariable model identified that the highest odds of incident infection were not knowing the HIV status of sex partners (OR 22.3, 95% CI: 8.4–59.1) (Figure 1 and Table 4). Reporting more than one female or male sex partner was associated with incident HIV (OR 13.6, 95% CI: 1.9–99.5), and specifically reporting 4–10 male sex partners (OR 11.0, 95% CI: 1.1–115.0) was associated with increased risk of infection. Homosexual or bisexual orientation was associated with incident HIV infection (OR 10.0, 95% CI: 1.3–75.4) compared to heterosexual orientation. Sometimes compared to always using condoms for vaginal sex was associated with incident HIV (OR 4.0, 95% CI: 1.1–14.0), and never compared to always using condoms for vaginal sex was borderline associated (OR 3.8, 95% CI: 0.9–15.9). Additional factors associated with incident HIV in males included reporting a penetrative physical injury (OR 30.0, 95% CI: 6.9–130.1), not having medical insurance (OR 4.4, 95% CI: 2.0–9.8), donating at a fixed blood collection site compared to a mobile site (OR 2.8, 95% CI: 1.4–5.8) and being a first‐time blood donor (OR 15.9, 95% CI: 5.9–42.7). Specific categories of some demographic characteristics, such as age group, educational attainment and marital status, were also associated with incident HIV infection.
TABLE 4.
Multivariable model of factors associated with incident human immunodeficiency virus infection among male blood donors.
| Characteristic | Odds ratio | 95% confidence interval | |
|---|---|---|---|
| Private medical insurance | |||
| Yes | Ref. | ||
| No | 4.43 | 2.00 | 9.78 |
| Age category | |||
| <21 years | 1.54 | 0.51 | 4.69 |
| 21–30 | Ref. | ||
| 31–40 | 2.97 | 1.25 | 7.07 |
| >40 | 1.63 | 0.54 | 4.91 |
| Education category | |||
| Graduate | Ref. | ||
| Grade 10 | 1.68 | 0.32 | 8.65 |
| Grade 12 or incomplete degree | 3.05 | 1.47 | 6.36 |
| Marital status | |||
| Single/never married | Ref. | ||
| Divorced/separated/widowed | 1.92 | 0.30 | 12.38 |
| Married/cohabiting | 2.51 | 1.06 | 5.96 |
| Donation recommended by HCW | |||
| No | Ref. | ||
| Yes | 0.38 | 0.14 | 1.03 |
| Bisexual/gay/homosexual | |||
| No | Ref. | ||
| Yes | 10.01 | 1.33 | 75.39 |
| Number of sex partners in last 6 months | |||
| 0 | Ref. | ||
| 1 | 6.44 | 1.06 | 39.22 |
| >1 | 13.55 | 1.85 | 99.53 |
| Sex with person living with HIV | |||
| No | Ref. | ||
| Yes | 2.24 | 0.15 | 33.17 |
| Unknown | 22.33 | 8.44 | 59.10 |
| Lifetime number of male sex partners | |||
| 0 | Ref. | ||
| 1 | 2.23 | 0.24 | 20.38 |
| 2–3 | 4.07 | 0.63 | 26.24 |
| 4–10 | 10.98 | 1.05 | 114.97 |
| Unknown | 1.34 | 0.61 | 2.92 |
| Condom use with vaginal sex | |||
| Always | Ref. | ||
| Sometimes | 3.97 | 1.12 | 14.01 |
| Never | 3.82 | 0.92 | 15.93 |
| Unknown | 1.80 | 0.24 | 13.74 |
| Not applicable | 0.54 | 0.11 | 2.79 |
| Donor type | |||
| Repeat | Ref. | ||
| First‐time | 15.91 | 5.93 | 42.68 |
| Lapsed | 4.16 | 1.57 | 11.03 |
| Clinic type | |||
| Mobile | Ref. | ||
| Fixed | 2.81 | 1.37 | 5.77 |
| Penetrative physical injury | |||
| No | Ref. | ||
| Yes | 29.97 | 6.86 | 130.87 |
Abbreviation: HCW, health care worker.
DISCUSSION
In our case–control study of risk factors for incident HIV infection in blood donors in South Africa, we identified several risk factors which are consistent with the known epidemiology of HIV infection in sub‐Saharan African countries [12]. Heterosexual risk behaviours dominated, although homosexual risk behaviours were evident among male cases. Several other observed associations with incident HIV without an obvious biological basis for causality may indicate contextual risk in certain groups, which is worthy of further investigation. Surprisingly, some of our hypothesized behaviours were not associated with incident HIV infection, including sexual relationships between partners with large age differences. Sex and relationships in the South African sociocultural context influence risk behaviours. There is a complex relationship between sexual, marital status and socioeconomic risk factors that are relevant to the risk of HIV [13].
Among blood donors, we confirmed several known sexual risk factors for HIV in females and males reported from studies done in other settings outside of Africa [13, 14]. A common risk factor for both sexes was not knowing whether one's sex partners were PLWH. For both sexes, increasing number of sex partners, particularly the number of male sex partners, was a risk factor for incident HIV infection. Condomless vaginal sex was a risk factor for both females and males. For about one‐quarter of males with incident HIV infection, MSM behaviour was reported, and gay or bisexual identity was independently associated with incident HIV.
Since May 2014, DHQs as part of the donor assessment processes at SANBS have been gender‐neutral. At that time, questions about sexual orientation were removed. At the time of this study (2014–2018), potential donors were asked, regardless of condom use, whether in the past 6 months they had a new sex partner or more than one sex partner and whether they had additional information about the behaviours of their sex partners. Even with these questions in use, the findings from our study suggest that donor disclosure of recent sexual risk at the time of donation remains suboptimal. In our study, both female and male donors with incident HIV infection, who disclosed on the ACASI that they had more than one sex partner in the past 6 months before donation, were >10 times more likely to have incident HIV. Thus, the rationale for asking about sex partners as part of donor selection aligns with the results of the study. However, these findings indicate that donor disclosure of sexual risks at the time of selection is inadequate [15, 16].
The reasons for non‐disclosure of risk need further research to better understand what factors are leading to nondisclosure. Fear of being judged by blood centre staff, lack of privacy during donor selection or an overarching desire to help others regardless of one's HIV status may contribute to non‐disclosure of risk as was recently reported by our group for PLWH who were taking antiretroviral medications and donating [16]. Furthermore, potential donors may be making a self‐assessment of perceived risk rather than actual risk and responding to the DHQs through that lens, as has been reported in other studies [15].
Women are more likely to seek healthcare than men [17] for sociocultural and biological reasons [17, 18], so men may be motivated to donate blood at least in part by HIV testing [19]. The association between incident HIV and the reported recommendation by healthcare workers to donate blood is unexpected and concerning, as healthcare workers should be aware that blood centres are not appropriate sites for contracting HIV. Further research into this phenomenon is needed.
We also found several unexpected factors to be associated with incident HIV infection. In females, these included lack of medical insurance, current employment and a protective effect of marijuana use. In males, they included lower educational attainment and a history of penetrative physical injury. We speculate that these are contextual factors and could be potential mediators between risk behaviours and incident infections or represent group characteristics placing individuals at a higher or lower risk of HIV. These types of behaviours or exposures are unlikely to cause HIV infection but may be indicative of particular social strata that have differing risks of incidence [13].
In this study, we assessed several ways in which age may be associated with incident HIV infection, including age of sexual debut, age at the time of donation and age differences between study participants and their sex partners. Age at sexual debut was not associated with incident infection in this study (data not shown). Incident HIV in the general population in South Africa is highest in the age group of 15–24 years [3, 20]. In blood donors in South Africa, incident HIV infection is higher in persons at least 21 years of age or older, relative to 16–19‐year‐old donors [21].
In contrast to our hypotheses, age difference in sexual partnerships was not associated with incident HIV infection. In our bivariate analysis among males, having partners 5–10 years younger was associated with incident HIV, but this was not observed among females, and the finding regarding age difference in males was not maintained in the multivariable model after adjusting for other factors.
This study has limitations. As in other case–control studies, differential recall and reporting bias may have occurred, with cases more likely to recall and report risk exposures. There may have been selection or participation bias because recruitment of controls was done both on the day of donation and by call back, as opposed to all case recruitment done only by call back following completion of testing to define incident infection. In addition, the study findings should be valid for incident HIV infection in persons presenting to donate blood, but the results must be extrapolated with caution to the general population because of potential biases related to donor self‐selection and selection by the blood centre.
Although participant enrolment for this study was conducted between 2014 and 2018, we expect the relevance of the identified risk factors to be applicable and informative for donor selection today because the identified behaviours associated with incident HIV in this study are consistent with the current epidemiology of HIV in sub‐Saharan Africa [22]. However, changes to the SANBS pre‐donation questioning process and exposure period during blood donation selection (from 6 to 3 months) and increasing availability of HIV treatment and prevention therapies mean the risk factors we identified may not be identical to those for incident HIV infection today.
In conclusion, we identified known behavioural and demographic risks associated with HIV acquisition; but more importantly, we also identified some previously unknown risks or contextual factors specific to blood donation. These contextual and specific risk factors could be incorporated into donor screening processes in South Africa, neighbouring countries and the rest of sub‐Saharan Africa, as these settings also face a high incidence of HIV infection. This is especially relevant to resource‐constrained countries that lack access to NAT testing and are more dependent on donor screening by questionnaire to identify those with a higher risk of recent HIV infection. In our setting, the results will help us to re‐evaluate the content of our donor screening questionnaire, donor deferral criteria and testing strategies. The findings also support the need to evaluate potential changes in messaging directed to candidate female and male donors to improve disclosure of recent risk at the time of donation.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
ACKNOWLEDGEMENTS
The REDS‐III South Africa programme was the responsibility of the following principal investigators (PIs), committee chairs and programme staff. Field sites: Edward L. Murphy (PI, UCSF) and Ute Jentsch (PI, South African National Blood Service). Data Coordinating Centre: Donald Brambilla and Marian Sullivan (co‐PI's Research Triangle International). Central Laboratory: Michael P. Busch (PI, Vitalant Research Institute). Steering Committee and Publications Committee chairs: Steven Kleinman and Roger Dodd. National Heart, Lung and Blood Institute (NHLBI) programme staff: Simone Glynn, Kelli Malkin and Benyam Hailu for their oversight and research leadership. The SANBS Virology Reference Laboratory staff were responsible for managing and coordinating laboratory testing for this study and the SANBS Nurses for enrolment and administration of the audio computer‐assisted structured interview (ACASI). The work was supported by research contracts from the National Heart, Lung and Blood Institute (NHLBI) of the National Institutes of Health for the Recipient Epidemiology and Donor Evaluation Study‐III International programme: HHSN268201100009I (to UCSF and the South African National Blood Service), HHSN268201100002I (to RTI International) and HHSN2682011‐00001I (to Vitalant Research Institute). A.S., U.J., K.v.d.B. and E.L.M. were also supported by the NIH Fogarty International Center research training grant 2D43‐TW010345.
A.S. wrote the first draft of the manuscript and reviewed and edited the manuscript, K.v.d.B. designed the research study, oversaw data acquisition and reviewed and edited the manuscript, M.V. and E.L.M. designed the research study and reviewed and edited the manuscript, U.J. reviewed and edited the manuscript, D.C. designed the research study, analysed the data and reviewed and edited the manuscript, J.J.H.‐F. designed the research study and managed data acquisition and B.C. designed the research study, oversaw data analysis and reviewed and edited the manuscript.
Swarts A, van den Berg K, Vermeulen M, Jentsch U, Creel D, Swanevelder R, et al. Risk factors for incident human immunodeficiency virus infection in South African blood donors. Vox Sang. 2025;120:892–903.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
REFERENCES
- 1. UNAIDS . Epidemiological Estimates. 2024. UNAIDS financial estimates. Available from: https://hivfinancial.unaids.org/hivfinancialdashboards.html. Last accessed 23 Apr 2025.
- 2. Rapodile T, Mitchel J, Swanevelder R, Murphy EL, van den Berg K. Re‐engineering the medical assessment of blood donors in South Africa: the balance between supply and safety. Transfusion. 2021;61:3361–3371. [DOI] [PubMed] [Google Scholar]
- 3. National Department of Health . The national HIV and syphilis prevalence survey, South Africa. Report. 2007. Available from: https://www.gov.za/xh/documents/other/national‐hiv‐and‐syphilis‐prevalence‐survey‐south‐africa‐2007‐29‐aug‐2008. Last accessed 1 Jul 2025.
- 4. StatsSA . Mid‐year population estimates for 2022. 2022. Available from: https://www.statssa.gov.za/publications/P0302/P03022022.pdf. Last accessed 5 Jan 2025.
- 5. Vermeulen M, Lelie N, Sykes W, Crookes R, Swanevelder J, Gaggia L, et al. Impact of individual‐donation nucleic acid testing on risk of human immunodeficiency virus, hepatitis B virus, and hepatitis C virus transmission by blood transfusion in South Africa. Transfusion. 2009;49:1115–1125. [DOI] [PubMed] [Google Scholar]
- 6. Kharsany AB, Karim QA. HIV infection and AIDS in sub‐Saharan Africa: current status, challenges and opportunities. Open AIDS J. 2016;10:34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Vermeulen M, Chowdhury D, Swanevelder R, Grebe E, Brambilla D, Jentsch U, et al. HIV incidence in South African blood donors from 2012 to 2016: a comparison of estimation methods. Vox Sang. 2021;116:71–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. South Africa . HIV epidemic, response and policy synthesis getting to success: improving HIV prevention efforts in South Africa. 2011. Available from: https://www.hst.org.za/publications/NonHST%20Publications/South%20Africa%20national%20synthesis%20report%20rev1%20with%20covers.pdf Last accessed 5 Jan 2025.
- 9. Sedia Biosciences Corporation . Sedia® HIV‑1 LAg‑avidity EIA (for serum or plasma specimens) [technical report] Portland, OR: Sedia Biosciences Corporation; 2014. [Google Scholar]
- 10. de Almeida‐Neto C, Goncalez TT, Birch RJ, de Carvalho SMF, Capuani L, Leão SC, et al. NHLBI Retrovirus Epidemiology Donor Study‐II (REDS‐II) International Component, Brazil. Risk factors for human immunodeficiency virus infection among Brazilian blood donors: a multicentre case‐control study using audio computer‐assisted structured interviews. Vox Sang. 2013;105:91–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Seighali F, Divkolaye NSH, Rezaei N, Kangarloo M. Risk factors associated with human immunodeficiency virus infection in blood donors in Iran: a case‐control study. Asian J Transfus Sci. 2021;15:183–188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Jentsch U, Vermeulen M, van den Berg K, Swanevelder R, Creel D, Jacobs G, et al. A case‐control study of risk factors for incident hepatitis B virus infection in South African blood donors. Int J Infect Dis. 2024;141:106958. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Zuma K, Simbayi L, Zungu N, Moyo S, Marinda E, Jooste S, et al. The HIV epidemic in South Africa: key findings from 2017 National Population‐Based Survey. Int J Environ Res Public Health. 2022;19:8125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Braunstein SL, van de Wijgert JH, Nash D. HIV incidence in sub‐Saharan Africa: a review of available data with implications for surveillance and prevention planning. AIDS Rev. 2009;11:140–156. [PubMed] [Google Scholar]
- 15. Willson S, Miller K, Seem D, Kuehnert MJ. Cognitive evaluation of the AABB uniform donor history questionnaire. Transfusion. 2016;56:1662–1667. [DOI] [PubMed] [Google Scholar]
- 16. van den Berg K, Murphy EL, Louw VJ, Maartens G, Hughes SD. Motivations for blood donation by HIV‐positive individuals on antiretrovirals in South Africa: a qualitative study. Transfus Med. 2023;33:277–286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Pearson S, Makadzange P. Help‐seeking behaviour for sexual‐health concerns: a qualitative study of men in Zimbabwe. Cult Health Sex. 2008;10:361–376. [DOI] [PubMed] [Google Scholar]
- 18. Parsons J. Not mission impossible. Aust Fam Physician. 2009;38:85. [PubMed] [Google Scholar]
- 19. Duong YT, Qiu M, De AK, Jackson K, Dobbs T, Kim AA, et al. Detection of recent HIV‐1 infection using a new limiting‐antigen avidity assay: potential for HIV‐1 incidence estimates and avidity maturation studies. PLoS One. 2012;7:e33328. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Leung Soo C, Pant Pai N, Bartlett SJ, Esmail A, Dheda K, Bhatnagar S. Socioeconomic factors impact the risk of HIV acquisition in the township population of South Africa: a Bayesian analysis. PLOS Glob Public Health. 2023;3:e0001502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Vermeulen M, Swanevelder R, Chowdhury D, Ingram C, Reddy R, Bloch EM, et al. NHLBI recipient epidemiology and donor evaluation study‐III (REDS‐III) international component. Use of blood donor screening to monitor prevalence of HIV and hepatitis B and C viruses, South Africa. Emerg Infect Dis. 2017;23:1560–1563. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Let our actions count South Africa's national strategic plan for HIV, TB, and STIs 2017–2022 . Available from: https://www.gov.za/sites/default/files/gcis_document/201705/nsp-hiv-tb-stia.pdf. Last accessed 5 Jan 2025.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.