Point-of-care diagnostics of childhood central nervous system infections, with a focus on usability in low-resource settings

Abstract

Background: The inaccessibility of laboratory services sustains the high burden of paediatric infectious diseases, such as central nervous system (CNS) infections, in low-resource settings. New contextually fit and well-implemented point-of-care tests (POCTs) could relieve such a burden and narrow the diagnostic divide between rich and poor. Yet, current disengagement between product developers, end-users, and implementors of POCTs impedes their clinical use and utility in low-resource settings. Also, the lack of evidence gathered through field evaluations of many diagnostic instruments in low-resource settings raises questions of their clinical utility there.

Objectives: The main aim of this thesis was to provide clinical and contextual guidance for developers of new POCTs for CNS infection diagnosis with high utility, especially in low-resource settings; and to implementors of POCT services towards their optimized clinical benefit. This was addressed through a multidisciplinary combination of qualitative, laboratory, and clinical studies.

Methods: Qualitative focus group discussions were conducted with health care workers (HCW) in Mbarara, Uganda (Paper I), and in Stockholm, Sweden (Paper III). Discussions were audio recorded and transcribed verbatim. Qualitative content analysis with an inductive approach was pursued in for data analysis. Comparisons between the two settings were discussed. In Paper II, a vertical flow DNA microarray printed on paper was developed for the detection of Neisseria meningitidis – a major aetiology of paediatric bacterial CNS infection worldwide. The analytical performance of the microarray was laboratory evaluated on DNA extracted from the bacteria, through the detection of the ctrA gene sequence specific to N. meningitidis. In Paper IV, a commercially available polymerase chain reaction (PCR) instrument with the capability of multiplex single-sample cerebrospinal fluid (CSF) microbiology was prospectively field-evaluated for the diagnosis of paediatric CNS infection in Mbarara, Uganda. Clinical turnaround time (cTAT) was defined as time spent from lumbar puncture until reporting of microbiology analyses to clinicians. The PCR instrument’s influence on clinical and patient-centered outcomes (yield, cTAT, duration of hospitalization and antibiotic exposure, patient outcome) was compared to that of bacterial culture.

Results: Fifty and 24 HCWs of different professions participated in the qualitative studies in Mbarara and Stockholm, respectively, expressing greater similarities than differences in perspectives of POCT use. POCTs were routinely used at both sites and credited for facilitating differential diagnostics and clinical decision-making. While the Ugandan setting with low laboratory accessibility was highly dependent on POCTs for sample analyses, the Swedish setting credited their use for having clinical and social value. Contrary to the described beneficial aspects, current POCTs were deemed contextually unfit in Mbarara, and their use to cause clinical distraction in Stockholm. Deficient implementation of POCT services was exposed in both places. Requests for ideal POCTs were aligned with those stipulated by the ‘ASSURED’ criteria of the World Health Organization. Specific POCTs for infectious diseases, including CNS infections, were requested. The laboratory study demonstrated an analytical sensitivity of 38 copies of ctrA per assay, with high specificity. The clinical study enrolled 212 children aged 0-12 years who were suspected of having CNS infection, with 193 of them being evaluated using the commercially available PCR instrument. A vast majority of children had been pre-administered antibiotics prior to lumbar puncture. Bacterial yield for the instrument was 12 % vs. 1.5 % for culture, with the addition of the instrument’s detection of viruses in 23 samples. Median cTAT for the instrument was 4.2 hours vs. 2 days for culture. Use of the instrument was associated with a statistically significant shorter antibiotic exposure of bacteria-negative vs. positive patients of five days, measured as from the time of reporting of laboratory results to the responsible clinicians. Similarly, its use was associated with a significantly shorter hospitalization for all-negative patients (five days) compared to those with any microorganism detected by it. No statistically significant differences in patient outcome were found due to its use, nor by its detection of any microorganisms.

Conclusion: Point-of-care tests provide laboratory means to settings without laboratory capacity and to situations in need of timely results, and we could show how rapid molecular methods for CSF analysis could benefit paediatric children with suspected CNS infection. Yet, without any observed benefits in patient outcome, and at a cost not financially bearable in most lowresource settings. Contextually fit POCTs for paediatric CNS infections are needed in lowresource settings. Yet, there are design flaws in current POCTs and in implementations for their use, limiting their clinical benefits. Collaborative engagement of product developers, clinicians, laboratory professionals, and health policymakers would better serve low-resource settings with contextually fit POCTs and allow for their optimized implementation. The ‘POCTEST' framework for such an engagement is proposed in this thesis. Finally, as we provided proof of concept for a newly developed paper-printed molecular method, we will pursue its development towards contextual clinical utility in low-resource settings. Should we succeed, we hope to contribute to a decrease in preventable childhood mortality in such settings.