Study Design
This study combines survey responses regarding HIV knowledge and attitudes with patient data on HIV testing acceptance among patients in an ED. The study was conducted at Livingstone Hospital in the Eastern Cape of South Africa from June 4th to July 15th, 2018 as part of the larger Walter Sisulu Infectious Diseases Screening in the Emergency Department (WISE) study. The aim of the WISE study is to implement HIV testing as per the national guidelines in the Eastern Cape and to quantify the burden of HIV among patients in this setting (23). Primary data collection of the HIV Knowledge and Attitudes Survey (HKAS) was conducted concurrently with the WISE study during a three-week period between June 18th and July 8th, 2018.
Setting
Most hospitals and clinics in the Eastern Cape are overcrowded, understaffed, lacking resources, and poorly managed (26). Livingstone Hospital is a provincial tertiary hospital situated in the Korsten suburb of Port Elizabeth, South Africa and forms part of the Port Elizabeth Hospital Complex. The hospital provides 24-hour emergency care, including trauma services, to the Port Elizabeth area and receives referrals from regional and district hospitals and clinics from its catchment area. Currently, there are no dedicated HIV counselors present in the ED at Livingstone Hospital, requiring medical officers and nurses to take on this responsibility. The ED has 50 beds and 15 doctors managing an average annual volume of 32,000 patients. The hospital serves both walk-in patients and patients arriving by ambulance.
Life Orientation Curriculum
The LO curriculum was introduced into the South African national curriculum in 1997 in an effort to help learners develop life skills and make responsible decisions about their health. The LO curriculum is initiated from grade 4 in the Foundation phase and is compulsory for South African students in grades 10 through 12. Part of the curriculum aims to educate learners about sexually transmitted infections, risky sexual behaviors, and HIV prevention. Two hours per week are dedicated to the curriculum, totaling 80 hours of instruction in each grade level in which the curriculum is implemented (5).
Recruitment
Patients were recruited by HIV Counseling and Testing (HCT) staff from the waiting room of the emergency department and were verbally asked if they would spend 10-15 minutes completing a brief questionnaire about their thoughts around HIV testing. All adult patients aged 18 years or older who were clinically stable and agreed to participate in the study were eligible for enrollment. Patients were excluded from the study if they were minors, unable to give informed consent due to decreased levels of consciousness or critically ill status, or patients returning to the ED who had been enrolled previously. Critically ill patients were defined as those with a South African Triage Scale (SATS) score of ‘emergency’ (27). HCT staff only approached patients who were initially assigned an ‘emergency’ SATS score after their condition was stabilized in the ED.
Data Collection
Data collection occurred through two parallel processes. Information on patient knowledge and attitudes towards HIV, demographics, and exposure to the LO curriculum was collected through the HKAS conducted via convenience sampling of patients enrolled in the WISE study. HCT staff aimed to survey five to ten patients per day. HCT staff briefly introduced the survey and obtained verbal consent before proceeding. The questions and answer options were dictated to the patient by HCT staff in English, Afrikaans, or Xhosa, and responses were recorded on electronic tablets. To capture HIV testing data, HCT staff approached eligible patients presenting to the ED once the triage process was completed, so as not to interfere with patient care. Patients were informed of the ongoing study and offered a rapid, point-of-care (POC) HIV test. Written informed consent was sought for HCT. Data on age, sex, SATS score, chief complaint, past medical history, clinical course, and HIV status, were recorded using case report forms. HCT staff also noted if the patient accepted or refused an HIV test, the test results, and the patient’s reasoning for accepting or refusing the test.
Survey Instrument
The HKAS consisted of forty-two questions assessing patient demographics, education level, exposure to the LO curriculum, and HIV knowledge and attitudes. The HKAS was created on the Qualtrics© survey platform (Qualtrics, Provo, UT). The attitudes questions in the HKAS were derived from a previously validated survey instrument (28). A pooling of 43 HIV attitudes questions were tested in ED populations to develop a validated survey for this context. Surveys were conducted among English or Xhosa-speaking patients in South African EDs, and answers were recorded in English. Exploratory factor analysis was used to determine correlation patterns between individual questions and Cronbach alpha scores were calculated. The fewest number of questions that represented the maximum variation from the original pool were chosen, resulting in an 18-question validated survey. Cognitive assessments were not conducted during the validation process. The survey instrument has previously been used to assess attitudes among English-speaking patients in an ED in East London, in the Eastern Cape of South Africa (7). A breakdown of the survey used in this study can be found in Figure 1. Knowledge and attitudes questions on the survey were recorded using a 5-point Likert scale with 1 being ‘strongly disagree’ and 5 being ‘strongly agree.’ Negatively worded questions were reversed in numeric value, so the number 5 consistently reflected positive attitudes. Responses to the knowledge questions were categorized as ‘correct’ or ‘incorrect’, wherein ‘strongly agree’ or ‘agree’ were grouped as ‘correct’ and ‘no opinion’, ‘disagree’, or ‘strongly disagree’ were grouped as ‘incorrect.’
Data Analysis
The primary outcome measure of this study was the effect of HIV knowledge and attitudes on testing acceptance. Case report forms were scanned and uploaded onto DataFax© (DataFax, Clinical DataFax Systems Inc., Hamilton, Ontario, Canada) to facilitate data validation. HKAS data on Qualtrics were imported into Stata v.15 (StataCorp LLC, Texas) for analysis. Patient survey responses were linked to their corresponding case report form using a unique study identification number. This facilitated the linking of patient acceptance or refusal of an HIV test to their responses on the surveys for further analysis. Patients reporting a known HIV positive status were removed from analysis.
Scored responses for the eighteen attitudes questions were summed to create an overall attitude score, and separate scores were calculated for each of the thematic attitude groups listed in Figure 1. A ‘perfect’ overall attitudes score was defined as a score of 90 (scoring a 5 on every question), while an ‘overall positive’ attitudes score was defined as a score of 72 or higher (scoring an average of 4 or above on every question). HKAS scores were analyzed as binary variables, where a score of 72 for overall attitudes was considered to be a ‘positive’ attitude towards HIV and a score less than 72 was considered to be a ‘negative’ attitude towards HIV. For the thematic attitudes groups the cut-off score for the binary variable was 8, with the exception of four categories; the cut-off score for Openness to HIV knowledge, HIV testing stigma, and ED-based HIV was 12, and the cut-off score for Cost of HIV Testing was 4. The descriptive titles of the binary variables for the thematic attitudes groups, represented in Table 4, are based on the content of the questions within the specific sections of the survey.
Analysis was conducted using chi-square tests to explore individual associations between HIV knowledge indicators, attitudes scores, and testing acceptance. Simple logistic regressions and two multiple logistic regression models were used to examine the crude and adjusted odds of accepting a POC HIV test. The first multiple logistic regression was adjusted for age and gender, while the second multiple logistic regression was adjusted for age, gender, race, and attitude score in each of the seven HKAS attitude domains. One participant was excluded from the attitudes analysis because they refrained from answering all eighteen attitudes questions on the HKAS. However, the patient answered the two knowledge questions and was therefore included in the analysis of the two knowledge questions. A total of 26 (11.66%) survey participants did not answer at least one survey question. In a sensitivity analysis, imputation was used for missing attitude scores for each category and the overall score, and our findings did not change. Due to the convenience sampling approach of the HKAS, an a priori sample size could not be determined.