Study design and participants
This was a study conducted in a prospective cohort of consecutive patients from January to March 2012 at Bagamoyo district in Tanzania. The district is situated north of Dar es Salaam and has an estimated population of 270,000 inhabitants, all of whom are considered to be at risk to develop malaria.
Recruitment was from three peripheral health care dispensaries (Kiwangwa, Yombo, and Fukayosi) located in the rural vicinity (approximately 30 km from the town of Bagamoyo), and from Bagamoyo District Hospital outpatient clinic in the urban centre of Bagamoyo. Subjects from both genders older than one year who were seeking routine care for malaria were enrolled into the study. A patient was enrolled once they were found to meet inclusion and exclusion criteria and had signed a study specific informed consent form.
Before the conduct of the trial, the protocol was reviewed and approved by national (Ref: NIMR/R.8a/Vol.IX/l244) and institutional (Ref: IHI/IRB/No: 34) ethics committees and regulatory authority, TFDA (Ref: No: CEm57/180/04A/54).
Malaria rapid diagnostic tests
The malaria RDT used in the current trial was SD Bioline Malaria Antigen Pf/Pan (Catalogue No. 05FK60, Standard Diagnostics Inc, Hagal-Dong, Korea, from now on referred as “SD RDT”). This is a lateral flow immunochromatographic test that contains a membrane strip encased in a flat plastic cassette. The strip is precoated with two antibodies: one that is specific for P. falciparum HRP2 and one that is pan-specific for pLDH for detection of other Plasmodium species.
The Deki Reader™
This mobile, rugged, battery-operated device (Fio Corporation, Toronto, Canada) performs the following functions (see Figure 1):
-
1.
Automated interpretation of RDTs by means of image analysis software;
-
2.
Digital data capture, by means of a touch-screen and a simple user interface software;
-
3.
Transmission in real time using local mobile phone network of processed RDT image, diagnostic event data collected, geopositioning of the device, and date and time stamp to a central database, which is accessible via internet.
Procedures
Eligible participants were consecutively enrolled in the dispensaries and at the outpatient clinic in Bagamoyo District Hospital. Inclusion criteria were age over one year, a documented fever or a history of fever in the previous 48 hours and consent to participate in the study. No information regarding recent self-treatment with anti-malarials was collected. Clinicians enrolled study participants and collected demographic and clinical information using a structured form.
After enrolment and physical examination by the study clinician, blood was collected from a finger prick once (no follow ups) and about three to four drops were collected in a heparinized microtube and sent to the reference laboratory of Ifakara Health Institute in Bagamoyo for preparation of thick blood smear (TBS) and a filter paper blood spot for polymerase chain reaction (PCR) testing, to be performed as quality control of microscopy. Technicians working at the reference laboratory were blinded to results of both visual and device interpretation of malaria RDTs performed at the site. Results of TBS microscopy examination and PCR (when applicable) were blinded to site technician who performed RDTs. At each of the study sites, the study technician collected blood from a finger prick for the processing of SD RDTs according to manufacturer’s recommendations. The processing of the RDT was guided by a job aid displayed on the screen of the Deki Reader device. The device also assisted technicians at the sites in keeping track of the incubation time for each RDT processed. Once the incubation period was completed, the study technician interpreted the RDT by visual inspection of the strip and captured the result using the touchscreen of the device. The study technician then immediately inserted the cassette into the device to perform the automated interpretation of the test. The results of the device’s interpretation of the RDT were concealed at all times, and were generated in all cases after the study site technician had entered results of her/his visual interpretation of RDT test results. All results of interpretation of RDTs, along with a high-resolution image of the RDT, and patient data collected were encrypted and automatically transmitted to a central cloud database specifically designed for the purpose of the trial.
Patient management was performed according to routine local protocols, involving results of RDTs after visual interpretation by the study technician, and the opinion of the clinician at the site.
Laboratory analyses
TBS microscopy and PCR tests were performed at Ifakara Health Institute Reference Laboratory, Bagamoyo by technicians who were blinded to results of RDT interpretation by both expert study technicians and the Deki Reader.
At the reference laboratory, blood smears were stained using 10% Giemsa solution for 30 minutes and examined under high-power objective. In positive smears, asexual and sexual parasites were counted against 200 and 500 white blood cells (WBCs), respectively. Parasite density was obtained by multiplying the parasite counts by 40 for asexual and 16 for sexual parasites (assuming each μL of blood contained 8,000 WBCs). A smear was declared negative after examining 200 high-power fields.
PCR was performed in all cases of discrepant results between expert visual interpretation of RDT, device’s interpretation of RDT and expert microscopy results (tiebreaker); and in a randomly selected subset of negative and positive samples (quality control of TBS microscopy). The PCR technique has high analytical sensitivity, allowing for qualitative and quantitative demonstration of parasites in clinical samples.
Molecular analysis involved extraction of parasite DNA from dried blood spots on filter papers. The DNA was used to determine the presence of P. falciparum beta tubulin target gene. DNA was extracted from dried blood spots by using QIAamp DNA mini kit (Qiagen), and PCR amplification was performed using Invitrogen QPCR SuperMix-UDG kit (Invitrogen). Real-time PCR amplification was done using MxPro3005 QPCR System (Agilent Technologies Inc, USA); forward (5’-TGA TGT GCG CAA GTG ATC C-3’) and reverse (5’-TCC TTT GTG GAC ATT CTT CCT C-3’) primers hybridize to the pathogen DNA. A probe (5’-TA GCA CAT GCC GTT AAA TAT CTT CCA TGT CT-3’) is included in the same reaction mixture. During PCR amplification, if the DNA of interest was present and amplified, the probe was cleaved and the reporter dye and quencher were separated leading to production of fluorescence. At the end of each PCR cycle, the resulting increase in fluorescence was detected by qPCR system as fluorescence units. As amplification continued, the cycle number at which detection of DNA products (threshold cycle value) was reached, and this information together with the standards (3D7 200,000 p/μl; 20,000 p/μl; 2,000 p/μl; 200 p/μl; 20 p/μl) gave the value of parasitaemia.
Statistical analysis
Data were obtained from source documents at each of the participant sites and compared to information collected through devices and transmitted to the Fio cloud database. Data collected were monitored according to FDA regulations and GCP guidelines. There were three outcomes of interest for the purpose of the analysis in the current trial: a) visual interpretation of RDT: performed by laboratory technician at the site following visual inspection of the SD RDT at the appropriate time. This information was entered by the technician into the device; b) device interpretation of RDT: performed by the device, which was collected automatically by the device and transmitted to the database; and, c) microscopy results: performed on TBS by laboratory technician at reference laboratory and entered into the dataset manually. For the purpose of the analysis the result of TBS microscopy was considered the “gold standard”.
Data collected at the Fio database were compared to data collected in the internal memory card of each Deki Reader device to ensure all data points were correctly transmitted to the cloud database. Data analysis was performed using JMP Version 8.0.2 and R Version 2.12.1 (SAS Institute Inc, NC, USA).
The primary analysis was conducted using TBS microscopy as gold standard to determine diagnostic performance characteristics (sensitivity, specificity, negative (NPV) and positive (PPV) predictive values, and overall diagnostic accuracy) of RDT interpretation by the Deki Reader and by visual interpretation. Correspondent 95% confidence intervals were constructed using the Wilson score method.
The percentage of agreement in the interpretation of SD RDT between the device and visual (human eye) done by the expert laboratory technicians at the sites was calculated as percentage of negative agreement, percentage of positive agreement and overall percentage of agreement. Two-sided 95% confidence intervals were constructed using the Wilson score method.
Sample size estimate
Although the RDT used in the present study is able to detect malaria infections caused by all Plasmodium species (categorized as “P.f” and “Pan”), the statistical analysis was performed only in results of P. falciparum test line, due to the small number of infections expected by parasites other than P. falciparum.
The prevalence of malaria among febrile patients presenting to a health clinic in the study area was estimated to be 10 to 15%, with greater than 95% of malaria cases caused by P. falciparum.
Based on this prevalence of P. falciparum infection in the study area, the sample size was estimated to be between 1,200 to 1,500 subjects, under the assumption that it would provide approximately 120-225 positive results for P. falciparum. Assuming that the sensitivity and specificity of the interpretation of SD RDT is the same as the sensitivity and specificity of the human-eye interpretation this sample size would provide 95% confidence intervals within 5% of the point estimate.