Study setting and design
This was a prospective cohort study based in Sierra Leone’s tertiary adult referral hospital in Freetown. This acts as the central hub for HIV care in Sierra Leone, as well as a referral site for other patients nationwide. The HIV clinic serves approximately 4000 patients.
Participant selection
Participants were recruited sequentially over 16 weeks from January to April 2018. Participants were recruited from the CD4 laboratory based in the HIV clinic, with all patients aged 18 years or older with a CD4 less than 100 cells/mm3 eligible for inclusion. This included inpatients and outpatients, and both newly diagnosed patients and those having CD4 cell count measured for disease monitoring or investigation of suspected opportunistic infections. Exclusion criteria were those under 18, patients with confirmed previous cryptococcal disease, and those who declined consent.
Screening procedure
All participants with positive cryptococcal antigen were reviewed by a doctor, had a lumbar puncture unless contraindicated, and followed up for survival for a minimum of 8 weeks. Participants with asymptomatic cryptococcal antigenemia were treated with oral fluconazole monotherapy @ a dose of 800mg daily for 2 weeks followed by 400mg daily for 8 weeks and then 200mg daily long-term (for at least 1 year, until CD4>200.) ART initiation was deferred to 2 weeks. Participants with cryptococcal meningitis were treated with fluconazole monotherapy according to national policy; oral/intravenous fluconazole monotherapy @ 1200mg daily for two weeks, followed by an oral dose of 800mg daily for 8 weeks and then continued on 200mg daily for at least 1 year until CD4 count>200/mm3. ART initiation was deferred at 4–6 weeks.
At the time of study initiation, this regimen was listed as an acceptable, though not a preferred regimen in the WHO guidelines, and is the only treatment available for cryptococcal meningitis in Sierra Leone.
Clinical procedure
Demographic and clinical information was collected from participants using a standardised data collection form by a trained research nurse and cross-checked with clinical records. CrAg-positive patients were followed up by medical and nursing members of the research team, but responsibility for patient management was with the admitting medical consultant (inpatients) and HIV clinic staff (outpatients).
Vital status was confirmed by review or discussion with the participant or a relative by telephone (where consent was given). A post-mortem was not available, and it was therefore not possible to definitively confirm the cause of death; this was based on the clinical impression of the medical team.
Laboratory procedures
CD4 was measured using a point-of-care PIMA analyser using whole blood from fingerpick or EDTA samples. Whole blood CrAg was measured using a commercially available Lateral Flow Assay (IMMY, Oklahoma, USA); quantitative measurement was not performed.
As a minimum, all patients had a CSF CrAg test, and measurement of opening pressure using sterile intravenous giving sets, or approximation using CSF drip rates (manometers not available)(10). Other CSF analysis depended on the availability of tests; where formal laboratory testing was not available, some CSF parameters were approximated using urine dipstick testing of CSF(11).
Sample size
The study was intended to follow the duration of a 16-week screening pilot. If the prevalence of cryptococcal antigen is 10% (similar to studies in Nigeria, the nearest geographical data), the sample size for 95% confidence and 5% accuracy was 138 patients. In 16 weeks, it was estimated that 160 patients would attend with CD4<100.
Statistical analysis
Data were entered into Excel and analysed in Stata version 13.0(College Station, Texas, USA: StataCorp LP). Data were summarised using proportions (categorical data) and median and interquartile ranges (non-parametric continuous data). Fisher’s exact test was used to compare categorical variables.
Ethical issues
This was an observational study in the context of the implementation of a WHO-recommended intervention (cryptococcal antigen screening). When patients had the capacity, informed written consent was sought for completion of a questionnaire and examination, and the use of anonymised clinical and demographic data. For all patients, the consent form was explained verbally in their local languages. Where patients were illiterate, the study and the consent form content were explained to them verbally and they indicated consent through a witnessed fingerprint. Patients who declined consent were still eligible for CrAg testing, but their data was not included in the analysis.
Some eligible participants lacked capacity as a result of suspected neurological opportunistic infection. The inclusion of this group was important in understanding the prevalence of cryptococcal infection in the study population. It was not felt ethical to seek consent from a representative as this would have involved the disclosure of HIV status; less than a third of people living with HIV in Sierra Leone have disclosed their status to adult relatives and levels of stigma are extremely high(12). As study procedures posed minimal risk to participants, anonymised data from patients who were unable to consent were therefore also included. Consent was sought retrospectively from participants who regained capacity. All participants were free to withdraw at any time, and data were stored confidentially and anonymously.
All study protocols were approved by the Sierra Leone Ethics and Scientific Review Committee.