Conference Material > Video (talk)
Ntone R
Epicentre Scientific Day Paris 2021. 2021 June 10
Journal Article > ResearchFull Text
Trop Med Int Health. 2016 March 6; Volume 21 (Issue 5); 603-609.; DOI:10.1111/tmi.12688
Kosack CS, Nick S
Trop Med Int Health. 2016 March 6; Volume 21 (Issue 5); 603-609.; DOI:10.1111/tmi.12688
OBJECTIVE
To evaluate the diagnostic accuracy of the OraQuick HCV rapid antibody test from OraSure and the Multisure HCV antibody assay from MP Biomedicals.
METHODS
Five seropanels from patients, intravenous drug users and blood donors with and without HCV infection were used on the two rapid immunochromatographic tests. Sensitivity, specificity and predictive values were calculated. In addition, seropanels from 10 seroconverters were used to assess early identification of HCV infection. The study was undertaken in a laboratory at Paul Ehrlich Institute in Germany.
RESULTS
Panel 1 contained of 55 positive and 25 negative samples. The OraQuick HCV test had a sensitivity of 100% (95% CI: 93.5-100) and a specificity of 100% (95% CI: 86.3-100). The Multisure HCV test had a sensitivity of 100% (95% CI: 93.5-100) and a specificity of 96% (95% CI: 79.6-99.9). Panel 2 consisted of 193 pre-characterised anti-HCV-positive patient samples. The OraQuick HCV test identified 191 samples correctly and the Multisure HCV 192. The sensitivity was 99.0% (95% CI: 96.3-99.9) for the OraQuick HCV test and 99.5% (95% CI: 97.1-100) for the Multisure HCV test. Panel 3 was composed of seroconversion samples of 10 patients. The OraQuick HCV test detected all of these 10 infections while the Multisure HCV test detected six and was indeterminate on 2. Panel 4 included 53 anti-HCV negative blood samples from blood donors. Both tests correctly identified all 53. Panel 5 consisted of 26 samples of HCV/HIV co-infected patients. The sensitivity of the OraQuick HCV test was 65.2% (95% CI: 42.8-82.8) after 20 min and 73.9% (95% CI: 51.3-88.9) after 40 min of incubation. The Multisure HCV test had a sensitivity of 96.2% (95% CI: 80.4-99.9).
CONCLUSION
This evaluation revealed good sensitivity for both rapid screening assays. The detection of seroconverters, however, was lower in the MutiSure HCV test. Therefore the MultiSure test should be used with hesitation in high incidence settings. The OraQuick gave HCV false-negative results in almost 25% of the HIV-positive sera. Therefore may the OraQuick be less suited in HIV prevalent areas.
To evaluate the diagnostic accuracy of the OraQuick HCV rapid antibody test from OraSure and the Multisure HCV antibody assay from MP Biomedicals.
METHODS
Five seropanels from patients, intravenous drug users and blood donors with and without HCV infection were used on the two rapid immunochromatographic tests. Sensitivity, specificity and predictive values were calculated. In addition, seropanels from 10 seroconverters were used to assess early identification of HCV infection. The study was undertaken in a laboratory at Paul Ehrlich Institute in Germany.
RESULTS
Panel 1 contained of 55 positive and 25 negative samples. The OraQuick HCV test had a sensitivity of 100% (95% CI: 93.5-100) and a specificity of 100% (95% CI: 86.3-100). The Multisure HCV test had a sensitivity of 100% (95% CI: 93.5-100) and a specificity of 96% (95% CI: 79.6-99.9). Panel 2 consisted of 193 pre-characterised anti-HCV-positive patient samples. The OraQuick HCV test identified 191 samples correctly and the Multisure HCV 192. The sensitivity was 99.0% (95% CI: 96.3-99.9) for the OraQuick HCV test and 99.5% (95% CI: 97.1-100) for the Multisure HCV test. Panel 3 was composed of seroconversion samples of 10 patients. The OraQuick HCV test detected all of these 10 infections while the Multisure HCV test detected six and was indeterminate on 2. Panel 4 included 53 anti-HCV negative blood samples from blood donors. Both tests correctly identified all 53. Panel 5 consisted of 26 samples of HCV/HIV co-infected patients. The sensitivity of the OraQuick HCV test was 65.2% (95% CI: 42.8-82.8) after 20 min and 73.9% (95% CI: 51.3-88.9) after 40 min of incubation. The Multisure HCV test had a sensitivity of 96.2% (95% CI: 80.4-99.9).
CONCLUSION
This evaluation revealed good sensitivity for both rapid screening assays. The detection of seroconverters, however, was lower in the MutiSure HCV test. Therefore the MultiSure test should be used with hesitation in high incidence settings. The OraQuick gave HCV false-negative results in almost 25% of the HIV-positive sera. Therefore may the OraQuick be less suited in HIV prevalent areas.
Journal Article > ResearchFull Text
PLoS Negl Trop Dis. 2013 August 15; Volume 7 (Issue 8); DOI:10.1371/journal.pntd.0002366
Martinez-Pino I, Luquero FJ, Sakoba K, Sylla S, Haile M, et al.
PLoS Negl Trop Dis. 2013 August 15; Volume 7 (Issue 8); DOI:10.1371/journal.pntd.0002366
During the 2012 cholera outbreak in the Republic of Guinea, the Ministry of Health, supported by Médecins Sans Frontières - Operational Center Geneva, used the oral cholera vaccine Shanchol as a part of the emergency response. The rapid diagnostic test (RDT) Crystal VC, widely used during outbreaks, detects lipopolysaccharide antigens of Vibrio cholerae O1 and O139, both included in Shanchol. In the context of reactive use of a whole-cell cholera vaccine in a region where cholera cases have been reported, it is essential to know what proportion of vaccinated individuals would be reactive to the RDT and for how long after vaccination.
Journal Article > ResearchFull Text
PLOS One. 2015 April 17; Volume 10 (Issue 4); DOI:10.1371/journal.pone.0122433
Kosack CS, de Kieviet W, Bayrak K, Milovic A, Page AL
PLOS One. 2015 April 17; Volume 10 (Issue 4); DOI:10.1371/journal.pone.0122433
Creatinine is a parameter that is required to monitor renal function and is important to follow in patients under treatment with potentially toxic renal drugs, such as the anti-HIV drug Tenofovir. A point of care instrument to measure creatinine would be useful for patients monitoring in resource-limited settings, where more instruments that are sophisticated are not available. The StatSensor Xpress Creatinine (Nova Biomedical Cooperation, Waltham, MA, USA) point of care analyzer was evaluated for its diagnostic performance in indicating drug therapy change. Creatinine was measured in parallel using the Nova StatSensor Xpress Creatinine analyzer and the Vitros 5,1FS (Ortho Clinical Diagnostics, Inc, Rochester, USA), which served as reference standard. The precision (i.e., repeatability and reproducibility) and accuracy of the StatSensor Xpress Creatinine analyzer were calculated using a panel of specimens with normal, low pathological and high pathological values. Two different Nova StatSensor Xpress Creatinine analyzers were used for the assessment of accuracy using repeated measurements. The coefficient of variation of the StatSensor Xpress Creatinine analyzers ranged from 2.3 to 5.9% for repeatability and from 4.2 to 9.0% for between-run reproducibility. The concordance correlation agreement was good except for high values (>600 µmol/L). The Bland-Altman analysis in high pathological specimens suggests that the Nova StatSensor Xpress Creatinine test tends to underestimate high creatinine values (i.e., >600 µmol/L). The Nova StatSensor Xpress Creatinine analyzers showed acceptable to good results in terms of repeatability, inter-device reproducibility and between-run reproducibility over time using quality control reagents. The analyzer was found sufficiently accurate for detecting pathological values in patients (age >10 year) and can be used with a moderate risk of misclassification.
Journal Article > ResearchFull Text
BMC Health Serv Res. 2021 December 6; Volume 21 (Issue 1); 1314.; DOI:10.1186/s12913-021-07323-1
Schausberger B, Mmema N, Dlamini V, Dube L, Aung A, et al.
BMC Health Serv Res. 2021 December 6; Volume 21 (Issue 1); 1314.; DOI:10.1186/s12913-021-07323-1
BACKGROUND
Traditional healing plays an important role in healthcare in Eswatini, and innovative collaborations with traditional healers may enable hard-to-reach men to access HIV and tuberculosis diagnostic services. This study explored attitudes towards integration of traditional healers into the provision of HIV self-testing kits and sputum collection containers.
METHODS
A qualitative study was conducted in 2019-2020 in Shiselweni region, Eswatini. Eight male traditional healers were trained on HIV and tuberculosis care including distribution of HIV self-testing kits and sputum collection containers. Attitudes towards the intervention were elicited through in-depth interviews with the eight traditional healers, ten clients, five healthcare workers and seven focus group discussions with community members. Interviews and group discussions were conducted in SiSwati, audio-recorded, translated and transcribed into English. Data were coded inductively and analysed thematically.
RESULTS
81 HIV self-testing kits and 24 sputum collection containers were distributed by the healers to 99 clients, with 14% of participants reporting a reactive HIV self-test result. The distribution of sputum containers did not result in any tuberculosis diagnoses, as samples were refused at health centres. Traditional healers perceived themselves as important healthcare providers, and after training, were willing and able to distribute HIV self-test kits and sputum containers to clients. Many saw themselves as peers who could address barriers to health-seeking among Swazi men that reflected hegemonic masculinities and patriarchal attitudes. Traditional healers were considered to provide services that were private, flexible, efficient and non-judgemental, although some clients and community members expressed concerns over confidentiality breaches. Attitudes among health workers were mixed, with some calling for greater collaboration with traditional healers and others expressing doubts about their potential role in promoting HIV and tuberculosis services. Specifically, many health workers did not accept sputum samples collected outside health facilities.
CONCLUSIONS
Offering HIV self-testing kits and sputum containers through traditional healers led to high HIV yields, but no TB diagnoses. The intervention was appreciated by healers' clients, due to the cultural literacy of traditional healers and practical considerations. Scaling-up this approach could bridge testing gaps if traditional healers are supported, but procedures for receiving sputum samples at health facilities need further strengthening.
Traditional healing plays an important role in healthcare in Eswatini, and innovative collaborations with traditional healers may enable hard-to-reach men to access HIV and tuberculosis diagnostic services. This study explored attitudes towards integration of traditional healers into the provision of HIV self-testing kits and sputum collection containers.
METHODS
A qualitative study was conducted in 2019-2020 in Shiselweni region, Eswatini. Eight male traditional healers were trained on HIV and tuberculosis care including distribution of HIV self-testing kits and sputum collection containers. Attitudes towards the intervention were elicited through in-depth interviews with the eight traditional healers, ten clients, five healthcare workers and seven focus group discussions with community members. Interviews and group discussions were conducted in SiSwati, audio-recorded, translated and transcribed into English. Data were coded inductively and analysed thematically.
RESULTS
81 HIV self-testing kits and 24 sputum collection containers were distributed by the healers to 99 clients, with 14% of participants reporting a reactive HIV self-test result. The distribution of sputum containers did not result in any tuberculosis diagnoses, as samples were refused at health centres. Traditional healers perceived themselves as important healthcare providers, and after training, were willing and able to distribute HIV self-test kits and sputum containers to clients. Many saw themselves as peers who could address barriers to health-seeking among Swazi men that reflected hegemonic masculinities and patriarchal attitudes. Traditional healers were considered to provide services that were private, flexible, efficient and non-judgemental, although some clients and community members expressed concerns over confidentiality breaches. Attitudes among health workers were mixed, with some calling for greater collaboration with traditional healers and others expressing doubts about their potential role in promoting HIV and tuberculosis services. Specifically, many health workers did not accept sputum samples collected outside health facilities.
CONCLUSIONS
Offering HIV self-testing kits and sputum containers through traditional healers led to high HIV yields, but no TB diagnoses. The intervention was appreciated by healers' clients, due to the cultural literacy of traditional healers and practical considerations. Scaling-up this approach could bridge testing gaps if traditional healers are supported, but procedures for receiving sputum samples at health facilities need further strengthening.
Journal Article > ResearchFull Text
Malar J. 2017 October 28; Volume 16 (Issue 1); 449.; DOI:10.1186/s12936-017-2094-3
Oyet C, Roh ME, Kiwanuka G, Orikiriza P, Wade M, et al.
Malar J. 2017 October 28; Volume 16 (Issue 1); 449.; DOI:10.1186/s12936-017-2094-3
BACKGROUND
Early diagnosis of suspected malaria cases with a rapid diagnostic test (RDT) has been shown to be an effective malaria control tool used in many resource-constrained settings. However, poor quality control and quality assurance hinder the accurate reporting of malaria diagnoses. Recent use of a portable, battery operated RDT reader (Deki Reader™, Fio Corporation) has shown to have high agreement with visual inspection across diverse health centre settings, however evidence of its feasibility and usability during cross sectional surveys are limited. This study aimed to evaluate the performance of the Deki Reader™ in a cross-sectional survey of children from southwestern Uganda.
METHODS
A two-stage, stratified cluster sampling survey was conducted between July and October 2014 in three districts of southwestern Uganda, with varying malaria transmission intensities. A total of 566 children aged 6-59 months were included in the analysis. Blood samples were collected and tested for malaria using: the SD Bioline Malaria Ag Pf/Pan RDT and microscopy. Results were compared between visual inspection of the RDT and by the Deki Reader™. Diagnostic performance of both methods were compared to gold-standard microscopy.
RESULTS
The sensitivity and specificity of the Deki Reader™ was 94.1% (95% CI 69.2-99.6%) and 95.6% (95% CI 93.4-97.1%), respectively. The overall percent agreement between the Deki Reader™ and visual RDT inspection was 98.9% (95% CI 93.2-99.8), with kappa statistic of 0.92 (95% CI 0.85-0.98).
CONCLUSIONS
The findings from this study suggest that the Deki Reader™ is comparable to visual inspection and performs well in detecting microscopy-positive Plasmodium falciparum cases in a household survey setting. However, the reader's performance was highly dependent on ensuring adequate battery life and a work environment free of dirt particles.
Early diagnosis of suspected malaria cases with a rapid diagnostic test (RDT) has been shown to be an effective malaria control tool used in many resource-constrained settings. However, poor quality control and quality assurance hinder the accurate reporting of malaria diagnoses. Recent use of a portable, battery operated RDT reader (Deki Reader™, Fio Corporation) has shown to have high agreement with visual inspection across diverse health centre settings, however evidence of its feasibility and usability during cross sectional surveys are limited. This study aimed to evaluate the performance of the Deki Reader™ in a cross-sectional survey of children from southwestern Uganda.
METHODS
A two-stage, stratified cluster sampling survey was conducted between July and October 2014 in three districts of southwestern Uganda, with varying malaria transmission intensities. A total of 566 children aged 6-59 months were included in the analysis. Blood samples were collected and tested for malaria using: the SD Bioline Malaria Ag Pf/Pan RDT and microscopy. Results were compared between visual inspection of the RDT and by the Deki Reader™. Diagnostic performance of both methods were compared to gold-standard microscopy.
RESULTS
The sensitivity and specificity of the Deki Reader™ was 94.1% (95% CI 69.2-99.6%) and 95.6% (95% CI 93.4-97.1%), respectively. The overall percent agreement between the Deki Reader™ and visual RDT inspection was 98.9% (95% CI 93.2-99.8), with kappa statistic of 0.92 (95% CI 0.85-0.98).
CONCLUSIONS
The findings from this study suggest that the Deki Reader™ is comparable to visual inspection and performs well in detecting microscopy-positive Plasmodium falciparum cases in a household survey setting. However, the reader's performance was highly dependent on ensuring adequate battery life and a work environment free of dirt particles.
Journal Article > ResearchFull Text
PLOS One. 2012 March 19; Volume 7 (Issue 3); DOI:10.1371/journal.pone.0033704
Kosack CS, Page AL, Van Hulsteijn LT, Lentjes EG
PLOS One. 2012 March 19; Volume 7 (Issue 3); DOI:10.1371/journal.pone.0033704
Thyroid-stimulating hormone (TSH) promotes expression of thyroid hormones which are essential for metabolism, growth, and development. Second-line drugs to treat tuberculosis (TB) can cause hypothyroidism by suppressing thyroid hormone synthesis. Therefore, TSH levels are routinely measured in TB patients receiving second-line drugs, and thyroxin treatment is initiated where indicated. However, standard TSH tests are technically demanding for many low-resource settings where TB is prevalent; a simple and inexpensive test is urgently needed.
Conference Material > Abstract
Lissouba P, Huerga H, Rucker C
Epicentre Scientific Day Paris 2021. 2021 June 10
BACKGROUND
The novel point-of-care urine-based FujiLAM test is promising for diagnosis of tuberculosis. We assessed the diagnostic yield of FujiLAM in HIV patients and the feasibility of using the test.
METHODS
We conducted a prospective diagnostic study and a mixed-methods feasibility and acceptability study in 4 countries: Uganda, Kenya,
Mozambique and South Africa. The diagnostic study included 2 groups of ambulatory HIV-positive adults: 1) with TB symptoms, 2) with advanced HIV disease and no TB symptoms. Patients received FujiLAM and AlereLAM, Xpert MTB/RIF, culture and chest X-ray. The feasibility study included test’ users, key informants and patients who participated through standard questionnaires, individual interviews and group discussions.
RESULTS
We included 1117 patients in the diagnostic study: 712 with TB symptoms (Group 1) and 405 with advanced HIV disease and no TB
symptoms (Group 2). TB was confirmed in 9.2% (63/685) and 4.1% (16/395) in Group 1 and 2, respectively. FujiLAM diagnostic yield among patients with confirmed TB was 63.2% and 43.8% in Group 1 and 2, respectively. FujiLAM diagnostic yield by CD4 count was: 75.0% in CD4<200, 77.8% in CD4 200-349, 31.3% in CD4≥350 (Group 1) and 46.7% in CD4<200 (Group 2). Most of the test users (including lay health workers) found FujiLAM easy to perform. The main concern was the multiple timed steps involved. Invalid results were obtained if test cartridges were dropped or performed on blood stained or cloudy urine. Most patients viewed urine sampling
positively and easier than sputum provision.
CONCLUSIONS
FujiLAM detects TB in a high proportion of the HIV patients with confirmed TB who have symptoms of TB and low CD4 counts, and in
a considerable proportion of those asymptomatic. The test is easy to perform at point-of-care. Urine sampling is well accepted by patients. These results encourage the future use of the FujiLAM assay.
KEY MESSAGES: The novel urine-based FujiLAM is a useful and easy to use point-of care test for TB diagnosis in HIV-positive patients. Urine sampling is well accepted.
This abstract is not to be quoted for publication.
The novel point-of-care urine-based FujiLAM test is promising for diagnosis of tuberculosis. We assessed the diagnostic yield of FujiLAM in HIV patients and the feasibility of using the test.
METHODS
We conducted a prospective diagnostic study and a mixed-methods feasibility and acceptability study in 4 countries: Uganda, Kenya,
Mozambique and South Africa. The diagnostic study included 2 groups of ambulatory HIV-positive adults: 1) with TB symptoms, 2) with advanced HIV disease and no TB symptoms. Patients received FujiLAM and AlereLAM, Xpert MTB/RIF, culture and chest X-ray. The feasibility study included test’ users, key informants and patients who participated through standard questionnaires, individual interviews and group discussions.
RESULTS
We included 1117 patients in the diagnostic study: 712 with TB symptoms (Group 1) and 405 with advanced HIV disease and no TB
symptoms (Group 2). TB was confirmed in 9.2% (63/685) and 4.1% (16/395) in Group 1 and 2, respectively. FujiLAM diagnostic yield among patients with confirmed TB was 63.2% and 43.8% in Group 1 and 2, respectively. FujiLAM diagnostic yield by CD4 count was: 75.0% in CD4<200, 77.8% in CD4 200-349, 31.3% in CD4≥350 (Group 1) and 46.7% in CD4<200 (Group 2). Most of the test users (including lay health workers) found FujiLAM easy to perform. The main concern was the multiple timed steps involved. Invalid results were obtained if test cartridges were dropped or performed on blood stained or cloudy urine. Most patients viewed urine sampling
positively and easier than sputum provision.
CONCLUSIONS
FujiLAM detects TB in a high proportion of the HIV patients with confirmed TB who have symptoms of TB and low CD4 counts, and in
a considerable proportion of those asymptomatic. The test is easy to perform at point-of-care. Urine sampling is well accepted by patients. These results encourage the future use of the FujiLAM assay.
KEY MESSAGES: The novel urine-based FujiLAM is a useful and easy to use point-of care test for TB diagnosis in HIV-positive patients. Urine sampling is well accepted.
This abstract is not to be quoted for publication.
Journal Article > ResearchFull Text
East Afr Med J. 2017 March 31; Volume 6 (Issue 2); 383.; DOI:10.4102/ajlm.v6i2.383
Orikiriza P, Nyehangane D, Atwine D, Kisakye JJ, Kassaza K, et al.
East Afr Med J. 2017 March 31; Volume 6 (Issue 2); 383.; DOI:10.4102/ajlm.v6i2.383
BACKGROUND
To confirm presence of Mycobacterium tuberculosis complex, some tuberculosis culture laboratories still rely on para-nitrobenzoic acid (PNB), a traditional technique that requires sub-culturing of clinical isolates and two to three weeks to give results. Rapid identification tests have improved turnaround times for mycobacterial culture results. Considering the challenges of the PNB method, we assessed the performance of the SD Bioline TB Ag MPT64 assay by using PNB as gold standard to detect M. tuberculosis complex from acid-fast bacilli (AFB) positive cultures.
OBJECTIVES
The aim of this study was to determine the sensitivity, specificity and turnaround time of the SD MPT64 assay for identification of M. tuberculosis complex, in a setting with high prevalence of tuberculosis and HIV.
METHODS
A convenience sample of 690 patients, with tuberculosis symptoms, was enrolled at Epicentre Mbarara Research Centre between April 2010 and June 2011. The samples were decontaminated using NALC-NaOH and re-suspended sediments inoculated in Mycobacterium Growth Indicator Tubes (MGIT) media, then incubated at 37 °C for a maximum of eight weeks. A random sample of 50 known negative cultures and 50 non-tuberculous mycobacteria isolates were tested for specificity, while sensitivity was based on AFB positivity. The time required from positive culture to reporting of results was also assessed with PNB used as the gold standard.
RESULTS
Of the 138 cultures that were AFB-positive, the sensitivity of the SD MPT64 assay was 100.0% [95% CI: 97.3 - 100] and specificity was 100.0% (95% CI, 96.4 - 100). The median time from a specimen receipt to confirmation of strain was 10 days [IQR: 8-12] with SD MPT64 and 24 days [IQR: 22-26] with PNB.
CONCLUSION
The SD MPT64 assay is comparable to PNB for identification of M. tuberculosis complex and reduces the time to detection.
To confirm presence of Mycobacterium tuberculosis complex, some tuberculosis culture laboratories still rely on para-nitrobenzoic acid (PNB), a traditional technique that requires sub-culturing of clinical isolates and two to three weeks to give results. Rapid identification tests have improved turnaround times for mycobacterial culture results. Considering the challenges of the PNB method, we assessed the performance of the SD Bioline TB Ag MPT64 assay by using PNB as gold standard to detect M. tuberculosis complex from acid-fast bacilli (AFB) positive cultures.
OBJECTIVES
The aim of this study was to determine the sensitivity, specificity and turnaround time of the SD MPT64 assay for identification of M. tuberculosis complex, in a setting with high prevalence of tuberculosis and HIV.
METHODS
A convenience sample of 690 patients, with tuberculosis symptoms, was enrolled at Epicentre Mbarara Research Centre between April 2010 and June 2011. The samples were decontaminated using NALC-NaOH and re-suspended sediments inoculated in Mycobacterium Growth Indicator Tubes (MGIT) media, then incubated at 37 °C for a maximum of eight weeks. A random sample of 50 known negative cultures and 50 non-tuberculous mycobacteria isolates were tested for specificity, while sensitivity was based on AFB positivity. The time required from positive culture to reporting of results was also assessed with PNB used as the gold standard.
RESULTS
Of the 138 cultures that were AFB-positive, the sensitivity of the SD MPT64 assay was 100.0% [95% CI: 97.3 - 100] and specificity was 100.0% (95% CI, 96.4 - 100). The median time from a specimen receipt to confirmation of strain was 10 days [IQR: 8-12] with SD MPT64 and 24 days [IQR: 22-26] with PNB.
CONCLUSION
The SD MPT64 assay is comparable to PNB for identification of M. tuberculosis complex and reduces the time to detection.
Journal Article > ResearchFull Text
Nat Commun. 2021 February 19; Volume 12 (Issue 1); DOI:10.1038/s41467-021-21187-3
Pascucci M, Royer G, Adamek J, Al Asmar M, Aristizabal D, et al.
Nat Commun. 2021 February 19; Volume 12 (Issue 1); DOI:10.1038/s41467-021-21187-3
Antimicrobial resistance is a major global health threat and its development is promoted by antibiotic misuse. While disk diffusion antibiotic susceptibility testing (AST, also called antibiogram) is broadly used to test for antibiotic resistance in bacterial infections, it faces strong criticism because of inter-operator variability and the complexity of interpretative reading. Automatic reading systems address these issues, but are not always adapted or available to resource-limited settings. We present an artificial intelligence (AI)-based, offline smartphone application for antibiogram analysis. The application captures images with the phone's camera, and the user is guided throughout the analysis on the same device by a user-friendly graphical interface. An embedded expert system validates the coherence of the antibiogram data and provides interpreted results. The fully automatic measurement procedure of our application's reading system achieves an overall agreement of 90% on susceptibility categorization against a hospital-standard automatic system and 98% against manual measurement (gold standard), with reduced inter-operator variability. The application's performance showed that the automatic reading of antibiotic resistance testing is entirely feasible on a smartphone. Moreover our application is suited for resource-limited settings, and therefore has the potential to significantly increase patients' access to AST worldwide.