Sensitivity and specificity of computer vision classification of eyelid photographs for programmatic trachoma assessment.

BACKGROUND/AIMS: Trachoma programs base treatment decisions on the community prevalence of the clinical signs of trachoma, assessed by direct examination of the conjunctiva. Automated assessment could be more standardized and more cost-effective. We tested the hypothesis that an automated algorithm could classify eyelid photographs better than chance. METHODS: A total of 1,656 field-collected conjunctival images were obtained from clinical trial participants in Niger and Ethiopia. Images were scored for trachomatous inflammation-follicular (TF) and trachomatous inflammation-intense (TI) according to the simplified World Health Organization grading system by expert raters. We developed an automated procedure for image enhancement followed by application of a convolutional neural net classifier for TF and separately for TI. One hundred images were selected for testing TF and TI, and these images were not used for training. RESULTS: The agreement score for TF and TI tasks for the automated algorithm relative to expert graders was κ = 0.44 (95% CI: 0.26 to 0.62, P < 0.001) and κ = 0.69 (95% CI: 0.55 to 0.84, P < 0.001), respectively. DISCUSSION: For assessing the clinical signs of trachoma, a convolutional neural net performed well above chance when tested against expert consensus. Further improvements in specificity may render this method suitable for field use.

Community-level chlamydial serology for assessing trachoma elimination in trachoma-endemic Niger.

BACKGROUND: Program decision-making for trachoma elimination currently relies on conjunctival clinical signs. Antibody tests may provide additional information on the epidemiology of trachoma, particularly in regions where it is disappearing or elimination targets have been met. METHODS: A cluster-randomized trial of mass azithromycin distribution strategies for trachoma elimination was conducted over three years in a mesoendemic region of Niger. Dried blood spots were collected from a random sample of children aged 1-5 years in each of 24 study communities at 36 months after initiation of the intervention. A multiplex bead assay was used to test for antibodies to two Chlamydia trachomatis antigens, Pgp3 and CT694. We compared seropositivity to either antigen to clinical signs of active trachoma (trachomatous inflammation-follicular [TF] and trachomatous inflammation-intense [TI]) at the individual and cluster level, and to ocular chlamydia prevalence at the community level. RESULTS: Of 988 children with antibody data, TF prevalence was 7.8% (95% CI 6.1 to 9.5) and TI prevalence was 1.6% (95% CI 0.9 to 2.6). The overall prevalence of antibody positivity to Pgp3 was 27.2% (95% CI 24.5 to 30), and to CT694 was 23.7% (95% CI 21 to 26.2). Ocular chlamydia infection prevalence was 5.2% (95% CI 2.8 to 7.6). Seropositivity to Pgp3 and/or CT694 was significantly associated with TF at the individual and community level and with ocular chlamydia infection and TI at the community level. Older children were more likely to be seropositive than younger children. CONCLUSION: Seropositivity to Pgp3 and CT694 correlates with clinical signs and ocular chlamydia infection in a mesoendemic region of Niger. TRIAL REGISTRATION: ClinicalTrials.gov NCT00792922.

Anthropometry and Malaria among Children in Niger: A Cross-Sectional Study.

The complex relationship between malnutrition and malaria affects morbidity and mortality in children younger than 5 years, particularly in parts of sub-Saharan Africa where these conditions occur together seasonally. Previous research on this relationship has been inconclusive. Here, we examine the association between anthropometric indicators and malaria infection in a population-based sample of children younger than 5 years in Niger. This cross-sectional study is a secondary analysis of a cluster-randomized trial comparing treatment strategies for trachoma in Niger. We included children aged 6-60 months residing in the 48 communities enrolled in the trial who completed anthropometric and malaria infection assessments at the final study visit. We evaluated the association between anthropometric indicators, including height-for-age z-score (HAZ) and weight-for-age z-score (WAZ) and indicators of malaria infection, including malaria parasitemia and clinical malaria. In May 2013, we collected data from 1,649 children. Of these, 780 (47.3%) were positive for malaria parasitemia and 401 (24.3%) had clinical malaria. In models of malaria parasitemia, the adjusted odds ratio (aOR) was 1.05 (95% confidence interval [CI]: 1.00-1.10) for HAZ and 1.07 (95% CI: 0.99, 1.15) for WAZ. In models of clinical malaria, the aOR was 1.07 (95% CI: 1.02-1.11) for HAZ and 1.09 (95% CI: 1.01-1.19) for WAZ. Overall, we did not find evidence of an association between most anthropometric indicators and malaria infection. Greater height may be associated with an increased risk of clinical malaria.

Comparison of Mass Azithromycin Coverage Targets of Children in Niger: A Cluster-Randomized Trachoma Trial.

Repeated oral azithromycin distribution targeted only to children has proven effective in reducing the ocular Chlamydia that causes trachoma. Here, we assess whether an enhanced coverage target of at least 90% of children is superior to the World Health Organization recommendation of at least 80%. Twenty-four trachoma-endemic communities in Matamèye, Niger, were randomized to a single day of azithromycin distribution aiming for at least 80% coverage or up to 4 days of treatment and > 90% coverage of children under age 12. All distributions were biannual. Children < 5 years of age and adults > 15 years were monitored for ocular Chlamydia infection by polymerase chain reaction every 6 months for 36 months in children and at baseline and 36 months in adults. Ocular Chlamydia prevalence in children decreased from 24.9% (95% confidence interval [CI] 15.9-33.8%) to 4.4% (95% CI 0.6-8.2%, P < 0.001) at 36 months in the standard coverage arm and from 15.6% (95% CI 10.0-21.2%) to 3.3% (95% CI 1.0-5.5%; P < 0.001) in the enhanced coverage arm. Enhanced coverage reduced ocular Chlamydia prevalence in children more quickly over time compared with standard (P = 0.04). There was no difference between arms at 36 months in children (2.4% lower with enhanced coverage, 95% CI 7.7-12.5%; P = 0.60). No infection was detected in adults at 36 months. Increasing antibiotic coverage among children from 80% to 90% may yield only short term improvements for trachoma control programs. Targeting treatment to children alone may be sufficient for trachoma control in this setting.

Annual Versus Biannual Mass Azithromycin Distribution and Malaria Parasitemia During the Peak Transmission Season Among Children in Niger.

BACKGROUND: Azithromycin has modest efficacy against malaria, and previous cluster randomized trials have suggested that mass azithromycin distribution for trachoma control may play a role in malaria control. We evaluated the effect of annual versus biannual mass azithromycin distribution over a 3-year period on malaria prevalence during the peak transmission season in a region with seasonal malaria transmission in Niger. METHODS: Twenty-four communities in Matameye, Niger, were randomized to annual mass azithromycin distribution (3 distributions to the entire community during the peak transmission season) or biannual-targeted azithromycin distribution (6 distributions to children <12 years of age, including 3 in the peak transmission season and 3 in the low transmission season). Malaria indices were evaluated at 36 months during the high transmission season. RESULTS: Parasitemia prevalence was 42.6% (95% confidence interval: 31.7%-53.6%) in the biannual distribution arm compared with 50.6% (95% confidence interval: 40.3%-60.8%) in the annual distribution arm (P = 0.29). There was no difference in parasite density or hemoglobin concentration in the 2 treatment arms. CONCLUSIONS: Additional rounds of mass azithromycin distribution during low transmission may not have a significant impact on malaria parasitemia measured during the peak transmission season.

Effectiveness of expanding annual mass azithromycin distribution treatment coverage for trachoma in Niger: a cluster randomised trial.

BACKGROUND/AIMS: The WHO recommends 3-5 years of annual mass azithromycin distribution with at least 80% treatment coverage to districts with active trachoma prevalence over 10% among children. Here, we assess the efficacy of expanding the coverage target to at least 90% for trachoma control in a mesoendemic region of Niger. METHODS: Twenty-four communities were randomised to a single day of azithromycin distribution with a coverage target of 80% of the community or up to 4 days of treatment, aiming for greater than 90% coverage. Distributions were annual and individuals above 6 months of age were treated. Children under 5 years of age were monitored for ocular chlamydia infection and active trachoma. RESULTS: At baseline, ocular chlamydia prevalence was 20.5% (95% CI 9.8% to 31.2%) in the standard coverage arm and 21.9% (95% CI 11.3% to 32.5%) in the enhanced coverage arm, which reduced to 4.6% (95% CI 0% to 9.5%, p=0.008) and 7.1% (95% CI 2.7% to 11.4%, p<0.001) at 36 months, respectively. There was no significant difference in 36-month ocular chlamydia prevalence between the two arms (p=0.21). There was no difference in the rate of decline in ocular chlamydia between the two arms in a repeated measures model (p=0.80). CONCLUSIONS: For annual mass azithromycin distribution programme to an entire community, there may be no additional benefit of increasing antibiotic coverage above the WHO's 80% target. TRIAL REGISTRATION NUMBER: NCT00792922, post-results.

Mass Azithromycin and Malaria Parasitemia in Niger: Results from a Community-Randomized Trial.

Studies designed to determine the effects of mass administration of azithromycin on trachoma have suggested that mass azithromycin distributions may also reduce the prevalence of malaria. These studies have typically examined the impact of a small number of treatments over short durations. In this prespecified substudy of a cluster-randomized trial for trachoma, we compared malaria parasitemia prevalence in 24 communities in Niger randomized to receive either annual or biannual mass azithromycin distributions over 3 years. The 12 communities randomized to annual azithromycin received three treatments during the high-transmission season, and the 12 communities randomized to biannual azithromycin received a total of six treatments: three during the high-transmission season and three during the low-transmission season. Blood samples were taken to assess malariometric indices among children in all study communities at a single time point during the high-transmission season after 3 years of the intervention. No significant differences were identified in malaria parasitemia, parasite density, or hemoglobin concentration between the annual and biannual treatment arms. When compared with annual mass azithromycin alone, additional mass azithromycin distributions given during the low-transmission season did not significantly reduce the subsequent prevalence of malaria parasitemia or parasite density after 3 years, as measured during the high-transmission season.

A Cluster-Randomized Trial to Assess the Efficacy of Targeting Trachoma Treatment to Children.

Background: The World Health Organization recommends annual treatment of entire trachoma-endemic communities, although children typically have a higher load, longer duration, and greater likelihood of infection. Methods: Forty-eight communities in Matameye, Niger, were randomized to annual oral azithromycin treatment of the entire community or biannual treatment of children aged 0-12 years only. Both children and adults were monitored for ocular chlamydial infection by polymerase chain reaction. Results: The prevalence of childhood infection was reduced in the annually treated arm from 21.2% (95% confidence interval [CI], 15.2%-28.0%) at baseline to 5.8% (95% CI, 3.2%-9.0%) at 36 months (P < .001) and in the biannual arm from 20.2% (95% CI, 15.5%-25.3%) to 3.8% (95% CI, 2.2%-6.0%; P < .001). Adult infection in the annual arm was reduced from 1.7% (95% CI, .9%-2.7%) to 0.3% (95% CI, .0%-.7%) and in the biannual arm from 1.2% (95% CI, .5%-2.2%) to 0.0% (95% CI, .0%-.7%; P = .005). The effect of biannual treatment of children compared with annual treatment of the entire community in both children (95% CI, -.04% to .02%) and adults (95% CI, .9%-2.7%) excluded the prespecified noninferiority threshold of 6% (P = .003 and P < .001, respectively). Conclusions: Periodic distribution of antibiotics to children in trachoma-endemic communities reduces chlamydial infection in both children and untreated adults, suggesting a form of herd protection. Biannual treatment of children was comparable to (specifically, noninferior to) annual treatment of the entire community, and may offer lower antibiotic use and other logistical advantages. Clinical Trials Registration: NCT00792922.

Association of Dry Eye Tests With Extraocular Signs Among 3514 Participants in the Sjögren's Syndrome International Registry.

PURPOSE: To identify a screening strategy for dry eye patients with a high likelihood of having Sjogren syndrome (SS) through the evaluation of the association of ocular surface tests with the extraocular signs used for the diagnosis of SS. DESIGN: Multicenter cross-sectional study. METHODS: The Sjogren's International Clinical Collaborative Alliance (SICCA) registry enrolled 3514 participants with SS or possible SS from 9 international academic sites. Ocular surface evaluation included Schirmer I testing, tear breakup time (TBUT), and staining of the cornea (0-6 points) and conjunctiva (0-6 points). Multivariate logistic regression analysis was performed to identify predictive factors for (1) histopathologic changes on labial salivary gland (LSG) biopsies (positive = focus score of ≥1 focus/4 mm2) and (2) positive anti-SSA/B serology. RESULTS: The adjusted odds of having a positive LSG biopsy were significantly higher among those with an abnormal Schirmer I test (adjusted OR = 1.26, 95% CI 1.05-1.51, P = .014) and positive conjunctival staining (for each additional unit of staining 1.46; 95% CI 1.39-1.53, P < .001) or corneal staining (for each additional unit of staining 1.14; 95% CI 1.08-1.21, P < .001). The odds of having a positive serology were significantly higher among those with an abnormal Schirmer I test (adjusted OR = 1.3; 95% CI 1.09-1.54, P = .004) and conjunctival staining (adjusted OR = 1.51; 95% CI 1.43-1.58, P < .001). CONCLUSIONS: In addition to corneal staining, which was associated with a higher likelihood of having a positive LSG biopsy, conjunctival staining and abnormal Schirmer I testing are of critical importance to include when screening dry eye patients for possible SS, as they were associated with a higher likelihood of having a positive LSG biopsy and serology.

Nasopharyngeal Pneumococcal Serotypes Before and After Mass Azithromycin Distributions for Trachoma.

Twenty-four Ethiopian communities were randomized to receive either (1) quarterly mass azithromycin distributions for trachoma for 1 year or (2) delayed treatment. Nasopharyngeal swabs collected from separate cross-sectional population-based samples of children were processed for Streptococcus pneumoniae Mass azithromycin did not significantly alter the pneumococcal serotype distribution, and hence it would not be expected to alter vaccine coverage.

Control of Trachoma from Achham District, Nepal: A Cross-Sectional Study from the Nepal National Trachoma Program.

BACKGROUND: The WHO seeks to control trachoma as a public health problem in endemic areas. Achham District in western Nepal was found to have TF (trachoma follicular) above 20% in a 2006 government survey, triggering 3 annual mass drug administrations finishing in 2010. Here we assess the level of control that has been achieved using surveillance for clinical disease, ocular chlamydia trachomatis infection, and serology for antibodies against chlamydia trachomatis protein antigens. METHODS: We conducted a cross-sectional survey of children aged 1-9 years in communities in Achham District in early 2014 including clinical examination validated with photographs, conjunctival samples for Chlamydia trachomatis (Amplicor PCR), and serological testing for antibodies against chlamydia trachomatis protein antigens pgp3 and CT694 using the Luminex platform. FINDINGS: In 24 randomly selected communities, the prevalence of trachoma (TF and/or TI) in 1-9 year olds was 3/1124 (0.3%, 95% CI 0.1 to 0.8%), and the prevalence of ocular chlamydia trachomatis infection was 0/1124 (0%, 95% CI 0 to 0.3%). In 18 communities selected because they had the highest prevalence of trachoma in a previous survey, the prevalence of TF and/or TI was 7/716 (1.0%, 95% CI 0.4 to 2.0%) and the prevalence of ocular chlamydia trachomatis infection was 0/716 (0%, 95% CI 0 to 0.5%). In 3 communities selected for serological testing, the prevalence of trachoma was 0/68 (0%, 95% CI 0 to 5.3%), the prevalence of ocular chlamydia trachomatis infection was 0/68 (0%, 95% CI 0 to 0.5%), the prevalence of antibodies against chlamydia trachomatis protein antigen pgp3 was 1/68 (1.5%, 95% CI 0.04% to 7.9%), and the prevalence of antibodies against chlamydia trachomatis protein antigen CT694 was 0/68 (0%, 95% CI 0 to 5.3%). CONCLUSION/SIGNIFICANCE: This previously highly endemic district in Nepal has little evidence of recent clinical disease, chlamydia trachomatis infection, or serological evidence of trachoma, suggesting that epidemiological control has been achieved.

Agreement between novice and experienced trachoma graders improves after a single day of didactic training.

BACKGROUND/AIMS: Prevalence estimates and treatment decisions for trachoma are based entirely on ocular clinical examination. The aim of the current study is to demonstrate that ophthalmic assistants can be trained and certified to provide trachoma grading within a single day. METHODS: Conjunctival photographs from an area with endemic trachoma were randomised into two sets of 60 cases. Photographs were graded for trachomatous inflammation-follicular (TF) and trachomatous inflammation-intense (TI) by three experienced graders. Inter-rater reliability of eight ophthalmic assistants and three experienced graders were compared before and after training. RESULTS: The mean κ agreement between the ophthalmic assistants and the consensus grades of the experienced graders for TF was 0.38 (95% CI 0.18 to 0.58) before training, and increased to 0.60 (95% CI 0.42 to 0.78) after training (p=0.07). The mean κ agreement for TI was 0.16 (95% CI 0.02 to 0.30) before training, and increased to 0.39 (95% CI 0.20 to 0.58) after training (p=0.02). CONCLUSION: A single day of training improves agreement between prospective and experienced trachoma graders, and provides the basis for certification of workers who are able to accurately grade trachoma and generate reliable prevalence estimates.

The Effect of Mass Azithromycin Distribution on Childhood Mortality: Beliefs and Estimates of Efficacy.

A cluster-randomized trial demonstrated that mass oral azithromycin distribution reduced childhood mortality 49.6% (Trachoma Amelioration in Northern Amhara [TANA]). The relative risk of childhood mortality was then estimated using two approaches: an expert survey and a Bayesian analysis. The survey asked public health experts to estimate the true effect of mass azithromycin distribution on childhood mortality. The Bayesian estimation used the TANA study's results and prior estimates of the efficacy of other effective population-level interventions. The experts believed mass azithromycin reduces childhood mortality (relative risk = 0.83, 95% credible intervals [CrI] = 0.70-1.00). The Bayesian analysis estimated a relative risk of 0.71 (95% CrI = 0.39-0.93). Both estimates suggest that azithromycin may have a true mortality benefit, though of a smaller magnitude than found in the single available trial. Prior information about nonantibiotic, population-level interventions may have informed the expert's opinions. Additional trials are needed to confirm a mortality benefit from mass azithromycin.

Inter-grader Agreement of the Ocular Staining Score in the Sjögren's International Clinical Collaborative Alliance (SICCA) Registry.

PURPOSE: To determine the intra-observer and inter-observer reliability of a novel ocular staining score among trained ophthalmologists. DESIGN: Reliability analysis within a prospective, observational, multicenter cohort study. METHODS: Those enrolled in the National Institutes of Health-funded Sjögren's International Collaborative Clinical Alliance (SICCA) who presented for follow-up at the University of California San Francisco, Aravind Eye Hospital, Johns Hopkins University, and the University of Pennsylvania were included. Study participants were graded using the ocular staining score by at least 2 masked SICCA-trained ophthalmologists. The primary outcome for this study was the intraclass correlation coefficient (ICC) for the total ocular staining score. ICCs were also calculated for tear break-up time (TBUT) and conjunctival and corneal staining. RESULTS: Total ocular staining score had an ICC of 0.91 for the right eye (95% confidence interval [CI] 0.85-0.96) and 0.90 for the left eye (95% CI 0.83-0.97). Corneal staining (right eye 0.86, 95% CI 0.76-0.93, left eye 0.90, 95% CI 0.81-0.95) and conjunctival staining (right eye 0.87, 95% CI 0.80-0.93, left eye 0.85, 95% CI 0.75-0.93) demonstrated excellent agreement. The ICC for TBUT was slightly lower (right eye 0.77, 95% CI 0.64-0.89; left eye 0.81, 95% CI 0.68-0.90). CONCLUSIONS: Previous studies have shown that the ocular staining score is correlated with other diagnostic components of Sjögren syndrome. In this study, we demonstrate high reliability in grading among trained ophthalmologists, completing the validation of this test.

Inter-Rater Agreement between Trachoma Graders: Comparison of Grades Given in Field Conditions versus Grades from Photographic Review.

PURPOSE: Trachoma surveillance is most commonly performed by direct observation, usually by non-ophthalmologists using the World Health Organization (WHO) simplified grading system. However, conjunctival photographs may offer several benefits over direct clinical observation, including the potential for greater inter-rater agreement. This study assesses whether inter-rater agreement of trachoma grading differs when trained graders review conjunctival photographs compared to when they perform conjunctival examinations in the field. METHODS: Three trained trachoma graders each performed an independent examination of the everted right tarsal conjunctiva of 269 children aged 0-9 years, and then reviewed photographs of these same conjunctivae in a random order. For each eye, the grader documented the presence or absence of follicular trachoma (TF) and intense trachomatous inflammation (TI) according to the WHO simplified grading system. RESULTS: Inter-rater agreement for the grade of TF was significantly higher in the field (kappa coefficient, κ, 0.73, 95% confidence interval, CI 0.67-0.80) than by photographic review (κ = 0.55, 95% CI 0.49-0.63; difference in κ between field grading and photo grading 0.18, 95% CI 0.09-0.26). When field and photographic grades were each assessed as the consensus grade from the three graders, agreement between in-field and photographic graders was high for TF (κ = 0.75, 95% CI 0.68-0.84). CONCLUSIONS: In an area with hyperendemic trachoma, inter-rater agreement was lower for photographic assessment of trachoma than for in-field assessment. However, the trachoma grade reached by a consensus of photographic graders agreed well with the grade given by a consensus of in-field graders.

Evidence for clonal expansion after antibiotic selection pressure: pneumococcal multilocus sequence types before and after mass azithromycin treatments.

BACKGROUND: A clinical trial of mass azithromycin distributions for trachoma created a convenient experiment to test the hypothesis that antibiotic use selects for clonal expansion of preexisting resistant bacterial strains. METHODS: Twelve communities in Ethiopia received mass azithromycin distributions every 3 months for 1 year. A random sample of 10 children aged 0-9 years from each community was monitored by means of nasopharyngeal swab sampling before mass azithromycin distribution and after 4 mass treatments. Swab specimens were tested for Streptococcus pneumoniae, and isolates underwent multilocus sequence typing. RESULTS: Of 82 pneumococcal isolates identified before treatment, 4 (5%) exhibited azithromycin resistance, representing 3 different sequence types (STs): 177, 6449, and 6494. The proportion of isolates that were classified as one of these 3 STs and were resistant to azithromycin increased after 4 mass azithromycin treatments (14 of 96 isolates [15%]; P = .04). Using a classification index, we found evidence for a relationship between ST and macrolide resistance after mass treatments (P < .0001). The diversity of STs-as calculated by the unbiased Simpson index-decreased significantly after mass azithromycin treatment (P = .045). CONCLUSIONS: Resistant clones present before mass azithromycin treatments increased in frequency after treatment, consistent with the theory that antibiotic selection pressure results in clonal expansion of existing resistant strains.

Does mass azithromycin distribution impact child growth and nutrition in Niger? A cluster-randomized trial.

BACKGROUND: Antibiotic use on animals demonstrates improved growth regardless of whether or not there is clinical evidence of infectious disease. Antibiotics used for trachoma control may play an unintended benefit of improving child growth. METHODOLOGY: In this sub-study of a larger randomized controlled trial, we assess anthropometry of pre-school children in a community-randomized trial of mass oral azithromycin distributions for trachoma in Niger. We measured height, weight, and mid-upper arm circumference (MUAC) in 12 communities randomized to receive annual mass azithromycin treatment of everyone versus 12 communities randomized to receive biannual mass azithromycin treatments for children, 3 years after the initial mass treatment. We collected measurements in 1,034 children aged 6-60 months of age. PRINCIPAL FINDINGS: We found no difference in the prevalence of wasting among children in the 12 annually treated communities that received three mass azithromycin distributions compared to the 12 biannually treated communities that received six mass azithromycin distributions (odds ratio = 0.88, 95% confidence interval = 0.53 to 1.49). CONCLUSIONS/SIGNIFICANCE: We were unable to demonstrate a statistically significant difference in stunting, underweight, and low MUAC of pre-school children in communities randomized to annual mass azithromycin treatment or biannual mass azithromycin treatment. The role of antibiotics on child growth and nutrition remains unclear, but larger studies and longitudinal trials may help determine any association.

Importance of including borderline cases in trachoma grader certification.

We assessed trachoma grading agreement among field graders using photographs that included the complete spectrum of disease and compared it with cases where there was consensus among experienced graders. Trained photographers took photographs of children's conjunctiva during a clinical trial in Ethiopia. We calculated κ-agreement statistics using a complete set of 60 cases and then recalculated the κ using a consensus set where cases were limited to those cases with agreement among experienced graders. When the complete set of 60 cases was used, agreement was moderate (κ = 0.61, 95% confidence interval [95% CI] = 0.56-0.67). When the consensus set was used, agreement improved significantly (κ = 0.75, 95% CI = 0.68-0.80). The κ of the consensus set was higher than the complete set by 0.14 (95% CI = 0.12-0.16) (P < 0.001). If testing sets remove difficult-to-grade cases, agreement in trachoma grading may be higher than actually seen in population-based trachoma surveys.

Reliability of trachoma clinical grading--assessing grading of marginal cases.

BACKGROUND: Clinical examination of trachoma is used to justify intervention in trachoma-endemic regions. Currently, field graders are certified by determining their concordance with experienced graders using the kappa statistic. Unfortunately, trachoma grading can be highly variable and there are cases where even expert graders disagree (borderline/marginal cases). Prior work has shown that inclusion of borderline cases tends to reduce apparent agreement, as measured by kappa. Here, we confirm those results and assess performance of trainees on these borderline cases by calculating their reliability error, a measure derived from the decomposition of the Brier score. METHODS AND FINDINGS: We trained 18 field graders using 200 conjunctival photographs from a community-randomized trial in Niger and assessed inter-grader agreement using kappa as well as reliability error. Three experienced graders scored each case for the presence or absence of trachomatous inflammation-follicular (TF) and trachomatous inflammation-intense (TI). A consensus grade for each case was defined as the one given by a majority of experienced graders. We classified cases into a unanimous subset if all 3 experienced graders gave the same grade. For both TF and TI grades, the mean kappa for trainees was higher on the unanimous subset; inclusion of borderline cases reduced apparent agreement by 15.7% for TF and 12.4% for TI. When we assessed the breakdown of the reliability error, we found that our trainees tended to over-call TF grades and under-call TI grades, especially in borderline cases. CONCLUSIONS: The kappa statistic is widely used for certifying trachoma field graders. Exclusion of borderline cases, which even experienced graders disagree on, increases apparent agreement with the kappa statistic. Graders may agree less when exposed to the full spectrum of disease. Reliability error allows for the assessment of these borderline cases and can be used to refine an individual trainee's grading.

Impact of mass azithromycin distribution on malaria parasitemia during the low-transmission season in Niger: a cluster-randomized trial.

We assessed the effect of mass azithromycin treatment on malaria parasitemia in a trachoma trial in Niger. Twenty-four study communities received treatment during the wet, high-transmission season. Twelve of the 24 communities were randomized to receive an additional treatment during the dry, low-transmission season. Outcome measurements were conducted at the community-level in children < 1-72 months of age in May-June 2011. Parasitemia was higher in the 12 once-treated communities (29.8%, 95% confidence interval [CI] = 21.5-40.0%) than in the 12 twice-treated communities (19.5%, 95% CI = 13.0-26.5%, P = 0.03). Parasite density was higher in once-treated communities (354 parasites/µL, 95% CI = 117-528 parasites/µL) than in twice-treated communities (74 parasites/µL, 95% CI = 41-202 parasites/µL, P = 0.03). Mass distribution of azithromycin reduced malaria parasitemia 4-5 months after the intervention. The results suggest that drugs with antimalaria activity can have long-lasting impacts on malaria during periods of low transmission.