Determinants of cryptococcal antigen (CrAg) screening uptake in Kampala, Uganda: An assessment of health center characteristics

Abstract

Cryptococcal antigen (CrAg) screening and pre-emptive antifungal therapy for people with CD4 cell counts <100 cells/μl are recommended by the World Health Organization and several national HIV guidelines. We sought to evaluate CrAg screening program implementation across Uganda, in relation to health center level and distance from the capital. We conducted a cross-sectional study of 22 health centers across southern Uganda from April to June 2019. We reviewed laboratory records regarding number of CD4 cell count tests performed, proportion of outpatients with CD4 counts <200 cells/μl, and number of CrAg screening tests performed. We administered surveys to health center staff to understand barriers to advanced HIV care. We observed no significant difference in health center level and performance of CrAg screening; with each subsequent health center level, there was 1.17-fold (95% CI: 0.92–1.41) higher odds of CrAg screening performed per level. CrAg screening uptake was not associated with distance from the capital city (odds ratio = 0.96, 95% CI: 0.89–1.04). Qualitative data from surveys indicated that limitations to uptake of CrAg screening were secondary to dysfunctional CD4 machines, lack of provider awareness of CrAg screening guidelines, and inadequate/intermittent supply of CrAg tests. There were no significant associations between CrAg screening uptake and level of health center or distance of health center from the capital city. We identified systemic barriers to CrAg screening related to inadequate CD4 testing, insufficient knowledge regarding national screening guidelines, and irregular laboratory testing supplies.

Lay summary

The objective of this study was to evaluate cryptococcal antigen (CrAg) screening program implementation in Uganda, by type of healthcare center and by distance from the capital city. CrAg screening uptake was not associated with distance from the capital city, or the type of healthcare center.

Introduction

Cryptococcal meningitis is an opportunistic infection responsible for 15–20% of AIDS-related mortality in Africa.¹Cryptococcus neoformans is the most common etiology of meningitis among adults in Uganda, and primarily affects people who are immunocompromised with CD4 cell counts of less than 100 cells/μl.² One-year mortality after a diagnosis of cryptococcal meningitis is approximately 70% in resource-limited settings due to inadequate access to antifungal medications, and the complex hospital management required including therapeutic lumbar punctures and electrolyte monitoring with supplementation.¹

Cryptococcal antigen (CrAg) predicts cryptococcal meningitis, and can be detected in the blood on average 21 days prior to the onset of symptoms.³ A randomized clinical trial of 1999 patients with HIV across urban clinics in Tanzania and Zambia demonstrated that CrAg screening and pre-emptive antifungal therapy reduce mortality by 28%, and are cost-effective in resource-limited settings to prevent cryptococcal meningitis.^4,5 Accordingly, the World Health Organization (WHO) and Ugandan national HIV guidelines recommend CrAg screening and pre-emptive antifungal therapy to reduce mortality from cryptococcal meningitis.

The CrAg Lateral Flow Assay (LFA) (IMMY; Norman, Oklahoma) costs approximately $2.50 per test, and is a cost-effective screening tool to prevent cryptococcal meningitis.⁶ The CrAg LFA can be used to diagnose meningitis from the cerebrospinal fluid, and additionally can detect asymptomatic cryptococcal antigenemia in blood. In 2015, CrAg screening was implemented in 1440 outpatients with HIV in Kampala, Uganda.⁷ However, unlike the aforementioned clinical trial setting, this implementation science study of CrAg screening in a real-world setting did not demonstrate the same survival benefit. The authors concluded that many barriers to implementation and scale-up for a national screening program remain. One qualitative study in Uganda in 2017 identified inadequate supply of fluconazole and CrAg kits, and timely follow-up of patients as barriers to CrAg screening implementation.⁸

Ten years after the CrAg LFA was approved and available in Uganda, barriers to universal implementation of CrAg screening persist. The objective of this study was to examine the associations between health center characteristics and the proportion of patients who received appropriate CrAg testing based on CD4 count. Health center characteristics, specifically, included (i) the range of services provided at a health center, quantified by level, and (ii) the distance from the capital city of Kampala. We hypothesized that health centers that were further from the capital city would have reduced performance of CrAg screening due to limited supply of testing, and reduced healthcare worker training capacity. Given that health centers at higher levels are better staffed and equipped to serve a larger population with greater resources, we expected a positive correlation between health center level and performance of CrAg testing.

Methods

We conducted a cross-sectional study of the association between health center characteristics and the proportion of patients who received appropriate CrAg testing based on CD4 count. This study was the result of a collaboration between investigators at the Infectious Diseases Institute of Makerere University, Kampala, Uganda, and the University of Minnesota, Minneapolis, Minnesota, USA. IRB approval was obtained through the Joint Clinical Research Centre in Uganda.

There are 195 health centers accredited to provide antiretroviral therapy (ART) for HIV patients in Uganda.⁹ This study utilized a convenience sample of 22 health centers across Southern and Western Uganda. Health centers in the sample were selected based on availability of funds for travel and the team's established health center contacts from prior studies. The unit of analysis in the study was the health center; individual-level patient laboratory data between April 1, 2019 and June 30, 2019 were aggregated at the time of the authors’ data collection visit to health center laboratories. Health centers were classified based on the level of services provided (see Supplementary Table 1). Data on shortest driving distance from Kampala to the district where health centers were located were obtained via Google Maps (www.maps.google.com).

The definition of low CD4 cell counts was initially defined as < 100 cells/μl until May 31, 2019, and then was changed to < 200 cells/μl as of June 1, 2019, per WHO and Uganda Ministry of Health guidelines.¹⁰

We measured the proportion of patients with a diagnosis of HIV who had low CD4 counts that received CrAg testing, as recommended by the Ugandan national guidelines. The number of persons who received CrAg testing and number of persons with low CD4 counts were manually abstracted and aggregated from the Ministry of Health CrAg registers and health center laboratory records by the first author (DS), and these frequencies were then re-counted by the authors (DS, FT) to reduce potential error. Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Minnesota. Findings from each health center were entered into a REDCap data collection form.¹¹

If laboratory staff were available at the clinic site, they were asked what barriers they faced in regular follow up of low CD4 results with CrAg testing. Where multiple laboratory staff were available, this question was asked of the supervisor. If only one staff person was available, the question was directed at this staff member. Responses were exported from the REDCap forms for analysis. Qualitative responses were coded and counted according to the reason indicated for limitations in testing.

Statistical analysis

Welch's ANOVA was conducted to assess significant differences in characteristics between health center levels because the distributions of these characteristics did not meet the assumption for equality of variance that is required in order to use a standard ANOVA.

Logistic regression models were constructed to examine the association between health center characteristics. Health center level was modeled as an ordinal variable and distance from capital city as a continuous variable, serving as predictor variables, while CrAg screening for low CD4 cell count was the binary outcome variable (1 = CrAg screening was conducted, 0 = CrAg screening was not conducted). Univariate models were fit for the predictor variables separately. A multivariable logistic model was run to allow for mutual adjustment by both predictor variables. In all analyses, two-sided P-values < 0.05 were considered statistically significant. Analyses were conducted using R.¹¹

Results

We visited a convenience sample of 22 health centers across Southern and Western Uganda in July and August 2019. Aggregate laboratory data from April 1 to June 30, 2019 were collected at each site. The centers were distributed across nine districts in Uganda (Figure 1), and included 10 Level III, 9 Level IV and 3 regional referral centers. The median distance from the capital city was 56 km (IQR: 1–150) for level III health centers, 211 km (IQR: 81–270) for level IV health centers, and 150 km (IQR: 71–304), for regional referral hospitals (P = 0.41; Table 1). Across health center levels, there was no significant difference in the median number of CD4 tests performed (P = 0.15), or CrAg tests performed (P = 0.17). Of patients with low CD4 counts, consistent CrAg screening was performed at 41% for Level III centers, 57% for level IV centers, and 60% for regional referral hospitals (P = 0.35; Table 1, Figure 2).

Figure 1.

Locations of 22 Ugandan health centers visited for this study.

Table 1.

Characteristics for 22 Ugandan health centers queried, data from April 1–June 30, 2019.

Characteristic	Level 3 centers (n = 10)	Level 4 centers (n = 9)	Regional centers (n = 3)	P-value
Median distance from Kampala, km (IQR)	56 (1, 150)	211 (81, 270)	150 (71, 304)	0.41
Median number of low CD4 tests (IQR)	9 (2, 31)	8 (6, 11)	42 (31, 63)	0.15
Median number of CrAg tests performed for low CD4 tests (IQR)	3 (1, 17)	5 (4, 7)	31 (22, 39)	0.17
Uptake of CrAg testing (% low CD4 results followed up with CrAg testing)	41% (6%, 62%)	57% (36%, 83%)	60% (58%, 67%)	0.35

IQR = interquartile range. Distance measured from center of Kampala.

Figure 2.

Percentage of patients with low CD4 cell counts that had CrAg testing by health center level. Low CD4 count was defined as < 100 cells/μl in April to May 2019 and < 200 cells/μl in June 2019.

Accordingly, unconditional logistic regression models showed no association between health center classification and odds of CrAg screening performance (odds ratio 1.17, 95% CI: 0.92–1.41; Table 2). There was no association between the distance from the capital city and CrAg screening uptake (odds ratio 0.96, 95% CI: 0.89–1.04; Table 2). From the multivariable model, there was no association between uptake of CrAg screening and health center level after adjusting for distance from the capital city (odds ratio 1.25, 95% CI: 0.98–1.55; Table 2).

Table 2.

Associations between health center level and distance from Kampala city center, and rate of CrAg screening from univariate and multivariable adjusted logistic regression models.

	Odds ratio	95% CI	P-value
Univariate models
Level of health center	1.17	0.92–1.41	0.19
Distance from Kampala center, per 50 km	0.96	0.89–1.04	0.26
Multivariable model	1.25	0.98–1.55	0.07

Out of 22 health centers, laboratory staff were present at seven health centers and were available to answer our question about barriers to CrAg screening. Potential reasons for incomplete CrAg testing are summarized in Table 3 based on survey responses from the laboratory staff in seven health centers. Commonly stated reasons for non-performance of CrAg testing were (a) depleted stock of CrAg kits, (b) dysfunctional CD4 quantification instruments, (c) depleted CD4 stock cartridges, (d) and lack of awareness among physicians regarding criteria for performing CrAg testing. Additionally, many health centers relied on central facilities to process samples for HIV viral-load testing, followed by return referral to the local facility for CD4 and CrAg testing. With transportation challenges to referral health centers, HIV viral load tests were not completed, and hence CD4 testing and CrAg testing were not performed.

Table 3.

Factors affecting CD4 testing and follow-up CrAg screening identified through surveys of laboratory personnel in seven health centers.

	Frequency of reported responses (n = 7)
Barriers to CrAg screening	Level III	Level IV	Regional
Depleted stock of CrAg kits/irregular supply	1
Dysfunctional CD4 instrument		2
Depleted stock of CD4 cartridges/reagent	3	1
Transport to the referral center is a challenge	1
Lack of awareness of CrAg testing guidelines among healthcare providers	1	1

*Clinics could select all barriers that were applicable (more than one).

Discussion

In this cross-sectional study of 22 health centers of southern Uganda, we found no statistically significant difference between CrAg screening uptake by facility level and distance from the city center. Depleted stock of cartridges/reagents and dysfunctional CD4 quantification instruments were among the most frequently identified barriers to CrAg screening by health center laboratory staff. Other impediments to CrAg screening included lack of awareness among healthcare workers of CrAg testing guidelines and depleted stock of CrAg kits. Lack of awareness among healthcare workers and depleted stock of CrAg kits were also identified as impediments to the CrAg screening program by Lofgren et al. (2018), who interviewed healthcare workers in Kampala in 2017. Their study highlighted fluconazole shortages as a main barrier to successful CrAg screening and treatment. Our study shows that shortages of CrAg testing supplies and education of healthcare workers remain a barrier, even if fluconazole supply is adequate. Furthermore, CrAg screening uptake does not vary by type of health facility and distance from the urban center.

One potential intervention would be to ensure that clinical supervisors at rural and semi-rural health centers are trained in national guidelines, and establish a system to monitor healthcare centers’ fidelity to CrAg testing guidelines. Previous studies have found supportive leadership and management to play a key role in improving healthcare worker motivation.¹² Another potential solution includes implementation of point-of-care CD4 tests. The Burnet Institute in Australia has developed the VISITECT CD4 Advanced Disease Lateral Flow Assay, an instrument-free, point-of-care test.¹³ In April 2020, this test was evaluated in three secondary and tertiary healthcare centers across Malawi, Zimbabwe, and the Democratic Republic of Congo and evaluation is underway in Uganda. VISITECT has a high agreement with reference samples and is feasible in resource-limited settings. Implementation of point-of-care CD4 testing has not yet been evaluated in peripheral health centers. If implementation of point-of-care CD4 testing increases expeditious identification of persons with advanced HIV disease, then CrAg screening can be performed at the same clinic visit.

One limitation of our study is the small sample size of health centers, 22 out of 195 accredited centers in Uganda, due to limited financial resources. Another limitation was the need to conduct manual review of laboratory records, which is prone to error. Uganda is in the initial stages of transitioning to an electronic medical record system. As a result, manual counting of persons who received CrAg testing and persons with low CD4 counts may lead to error.

Further supplementation of quantitative data with more structured qualitative data through interviews with clinical and lab staff at similar capacities across health centers would help assess the impact of additional factors that could not be captured through the quantitative data collection. For example, in collecting quantitative laboratory data, we could not capture reasons such as ‘lack of awareness among health center staff’ or ‘lack of functional equipment’. Per the responses to our qualitative survey, we postulate that regular training and continuing education for health center laboratory staff, streamlined supply of CrAg testing kits, and training of local district-level health authorities on the supply and maintenance of CD4 testing could be valuable in improving uptake of CrAg screening programs. While CrAg screening was first introduced 10 years ago, continued efforts toward healthcare worker training and establishment of stable supply chains remain critical for the care of persons with advanced HIV disease.

Funding

Support is received by the National Institutes of Allergy and Infectious Diseases (K23AI138851, T32AI055433, U01AI125003). Research was additionally supported by the Fogarty International Center (FIC) and the National Institute of Neurological Disorders and Stroke (NINDS) under grant #D43TW009345 awarded to the Northern Pacific Global Health Fellows Program.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and the writing of the paper.