Variability in interpretation of chest radiographs among Russian clinicians and implications for screening programmes: observational study

Testing the performance, adequacy, and applicability of an artificial intelligence model for pediatric pneumonia diagnosis

BACKGROUND

Community-acquired Pneumonia (CAP) is a common childhood infectious disease. Deep learning models show promise in X-ray interpretation and diagnosis, but their validation should be extended due to limitations in the current validation workflow. To extend the standard validation workflow we propose doing a pilot test with the next characteristics. First, the assumption of perfect ground truth (100% sensitive and specific) is unrealistic, as high intra and inter-observer variability have been reported. To address this, we propose using Bayesian latent class models (BLCA) to estimate accuracy during the pilot. Additionally, assessing only the performance of a model without considering its applicability and acceptance by physicians is insufficient if we hope to integrate AI systems into day-to-day clinical practice. Therefore, we propose employing explainable artificial intelligence (XAI) methods during the pilot test to involve physicians and evaluate how well a Deep Learning model is accepted and how helpful it is for routine decisions as well as analyze its limitations by assessing the etiology. This study aims to apply the proposed pilot to test a deep Convolutional Neural Network (CNN)-based model for identifying consolidation in pediatric chest-X-ray (CXR) images already validated using the standard workflow.

METHODS

For the standard validation workflow, a total of 5856 public CXRs and 950 private CXRs were used to train and validate the performance of the CNN model. The performance of the model was estimated assuming a perfect ground truth. For the pilot test proposed in this article, a total of 190 pediatric chest-X-ray (CXRs) images were used to test the CNN model support decision tool (SDT). The performance of the model on the pilot test was estimated using extensions of the two-test Bayesian Latent-Class model (BLCA). The sensitivity, specificity, and accuracy of the model were also assessed. The clinical characteristics of the patients were compared according to the model performance. The adequacy and applicability of the SDT was tested using XAI techniques. The adequacy of the SDT was assessed by asking two senior physicians the agreement rate with the SDT. The applicability was tested by asking three medical residents before and after using the SDT and the agreement between experts was calculated using the kappa index.

RESULTS

The CRXs of the pilot test were labeled by the panel of experts into consolidation (124/176, 70.4%) and no-consolidation/other infiltrates (52/176, 29.5%). A total of 31/176 (17.6%) discrepancies were found between the model and the panel of experts with a kappa index of 0.6. The sensitivity and specificity reached a median of 90.9 (95% Credible Interval (CrI), 81.2-99.9) and 77.7 (95% CrI, 63.3-98.1), respectively. The senior physicians reported a high agreement rate (70%) with the system in identifying logical consolidation patterns. The three medical residents reached a higher agreement using SDT than alone with experts (0.66±0.1 vs. 0.75±0.2).

CONCLUSIONS

Through the pilot test, we have successfully verified that the deep learning model was underestimated when a perfect ground truth was considered. Furthermore, by conducting adequacy and applicability tests, we can ensure that the model is able to identify logical patterns within the CXRs and that augmenting clinicians with automated preliminary read assistants could accelerate their workflows and enhance accuracy in identifying consolidation in pediatric CXR images.

Improving chest X-ray report generation by leveraging warm starting

Automatically generating a report from a patient's Chest X-rays (CXRs) is a promising solution to reducing clinical workload and improving patient care. However, current CXR report generators-which are predominantly encoder-to-decoder models-lack the diagnostic accuracy to be deployed in a clinical setting. To improve CXR report generation, we investigate warm starting the encoder and decoder with recent open-source computer vision and natural language processing checkpoints, such as the Vision Transformer (ViT) and PubMedBERT. To this end, each checkpoint is evaluated on the MIMIC-CXR and IU X-ray datasets. Our experimental investigation demonstrates that the Convolutional vision Transformer (CvT) ImageNet-21K and the Distilled Generative Pre-trained Transformer 2 (DistilGPT2) checkpoints are best for warm starting the encoder and decoder, respectively. Compared to the state-of-the-art (M2 Transformer Progressive), CvT2DistilGPT2 attained an improvement of 8.3% for CE F-1, 1.8% for BLEU-4, 1.6% for ROUGE-L, and 1.0% for METEOR. The reports generated by CvT2DistilGPT2 have a higher similarity to radiologist reports than previous approaches. This indicates that leveraging warm starting improves CXR report generation. Code and checkpoints for CvT2DistilGPT2 are available at https://github.com/aehrc/cvt2distilgpt2.

Assessment of non-tuberculosis abnormalities on digital chest x-rays with high CAD4TB scores from a tuberculosis prevalence survey in Zambia and South Africa

BACKGROUND

Chest X-rays (CXRs) have traditionally been used to aid the diagnosis of TB-suggestive abnormalities. Using Computer-Aided Detection (CAD) algorithms, TB risk is quantified to assist with diagnostics. However, CXRs capture all other structural abnormalities. Identification of non-TB abnormalities in individuals with CXRs that have high CAD scores but don't have bacteriologically confirmed TB is unknown. This presents a missed opportunity of extending novel CAD systems' potential to simultaneously provide information on other non-TB abnormalities alongside TB. This study aimed to characterize and estimate the prevalence of non-TB abnormalities on digital CXRs with high CAD4TB scores from a TB prevalence survey in Zambia and South Africa.

METHODOLOGY

This was a cross-sectional analysis of clinical data of participants from the TREATS TB prevalence survey conducted in 21 communities in Zambia and South Africa. The study included individuals aged ≥ 15 years who had high CAD4TB scores (score ≥ 70), but had no bacteriologically confirmed TB in any of the samples submitted, were not on TB treatment, and had no history of TB. Two consultant radiologists reviewed the images for non-TB abnormalities.

RESULTS

Of the 525 CXRs reviewed, 46.7% (245/525) images were reported to have non-TB abnormalities. About 11.43% (28/245) images had multiple non-TB abnormalities, while 88.67% (217/245) had a single non-TB abnormality. The readers had a fair inter-rater agreement (r = 0.40). Based on anatomical location, non-TB abnormalities in the lung parenchyma (19%) were the most prevalent, followed by Pleura (15.4%), then heart & great vessels (6.1%) abnormalities. Pleural effusion/thickening/calcification (8.8%) and cardiomegaly (5%) were the most prevalent non-TB abnormalities. Prevalence of (2.7%) for pneumonia not typical of pulmonary TB and (2.1%) mass/nodules (benign/ malignant) were also reported.

CONCLUSION

A wide range of non-TB abnormalities can be identified on digital CXRs among individuals with high CAD4TB scores but don't have bacteriologically confirmed TB. Adaptation of AI systems like CAD4TB as a tool to simultaneously identify other causes of abnormal CXRs alongside TB can be interesting and useful in non-faculty-based screening programs to better link cases to appropriate care.

Localization supervision of chest x-ray classifiers using label-specific eye-tracking annotation

Convolutional neural networks (CNNs) have been successfully applied to chest x-ray (CXR) images. Moreover, annotated bounding boxes have been shown to improve the interpretability of a CNN in terms of localizing abnormalities. However, only a few relatively small CXR datasets containing bounding boxes are available, and collecting them is very costly. Opportunely, eye-tracking (ET) data can be collected during the clinical workflow of a radiologist. We use ET data recorded from radiologists while dictating CXR reports to train CNNs. We extract snippets from the ET data by associating them with the dictation of keywords and use them to supervise the localization of specific abnormalities. We show that this method can improve a model's interpretability without impacting its image-level classification.

Commonly Missed Findings on Chest Radiographs: Causes and Consequences

Chest radiography (CXR) continues to be the most frequently performed imaging examination worldwide, yet it remains prone to frequent errors in interpretation. These pose potential adverse consequences to patients and are a leading motivation for medical malpractice lawsuits. Commonly missed CXR findings and the principal causes of these errors are reviewed and illustrated. Perceptual errors are the predominant source of these missed findings. The medicolegal implications of such errors are explained. Awareness of commonly missed CXR findings, their causes, and their consequences are important in developing approaches to reduce and mitigate these errors.

A multi-output network with U-net enhanced class activation map and robust classification performance for medical imaging analysis

Computer vision in medical diagnosis has achieved a high level of success in diagnosing diseases with high accuracy. However, conventional classifiers that produce an image-to-label result provide insufficient information for medical professionals to judge and raise concerns over the trust and reliability of a model with results that cannot be explained. To gain local insight of cancerous regions, separate tasks such as imaging segmentation needs to be implemented to aid the doctors in treating patients which doubles the training time and costs which renders the diagnosis system inefficient and difficult to be accepted by the public. To tackle this issue and drive the AI-first medical solutions further, this paper proposes a multi-output network which follows a U-Net architecture for image segmentation output and features an additional CNN module for auxiliary classification output. Class Activation Maps or CAMs are a method of providing insight into a convolutional neural network’s feature maps that lead to its classification but in the case of lung diseases, the region of interest is enhanced by U-net assisted Class Activation Mapping (CAM) visualization. Therefore, our proposed model combines image segmentation models and classifiers to crop out only the lung region of a chest X-ray’s class activation map to provide a visualization that improves the explainability and can generate classification results simultaneously which builds trust for AI-led diagnosis system. The proposed U-Net model achieves 97.72% accuracy and a dice coefficient of 0.9691 on a testing data from the COVID-QU-Ex Dataset which includes both diseased and healthy lungs.

REFLACX, a dataset of reports and eye-tracking data for localization of abnormalities in chest x-rays

Deep learning has shown recent success in classifying anomalies in chest x-rays, but datasets are still small compared to natural image datasets. Supervision of abnormality localization has been shown to improve trained models, partially compensating for dataset sizes. However, explicitly labeling these anomalies requires an expert and is very time-consuming. We propose a potentially scalable method for collecting implicit localization data using an eye tracker to capture gaze locations and a microphone to capture a dictation of a report, imitating the setup of a reading room. The resulting REFLACX (Reports and Eye-Tracking Data for Localization of Abnormalities in Chest X-rays) dataset was labeled across five radiologists and contains 3,032 synchronized sets of eye-tracking data and timestamped report transcriptions for 2,616 chest x-rays from the MIMIC-CXR dataset. We also provide auxiliary annotations, including bounding boxes around lungs and heart and validation labels consisting of ellipses localizing abnormalities and image-level labels. Furthermore, a small subset of the data contains readings from all radiologists, allowing for the calculation of inter-rater scores.

Measurement(s)	gaze locations • radiology report • abnormality localizations • chest localization (lung and heart) • abnormality presence
Technology Type(s)	eye tracking device • Microphone Device • User Interface Device
Sample Characteristic - Organism	Homo sapiens

A Deep Modality-Specific Ensemble for Improving Pneumonia Detection in Chest X-rays

Pneumonia is an acute respiratory infectious disease caused by bacteria, fungi, or viruses. Fluid-filled lungs due to the disease result in painful breathing difficulties and reduced oxygen intake. Effective diagnosis is critical for appropriate and timely treatment and improving survival. Chest X-rays (CXRs) are routinely used to screen for the infection. Computer-aided detection methods using conventional deep learning (DL) models for identifying pneumonia-consistent manifestations in CXRs have demonstrated superiority over traditional machine learning approaches. However, their performance is still inadequate to aid in clinical decision-making. This study improves upon the state of the art as follows. Specifically, we train a DL classifier on large collections of CXR images to develop a CXR modality-specific model. Next, we use this model as the classifier backbone in the RetinaNet object detection network. We also initialize this backbone using random weights and ImageNet-pretrained weights. Finally, we construct an ensemble of the best-performing models resulting in improved detection of pneumonia-consistent findings. Experimental results demonstrate that an ensemble of the top-3 performing RetinaNet models outperformed individual models in terms of the mean average precision (mAP) metric (0.3272, 95% CI: (0.3006,0.3538)) toward this task, which is markedly higher than the state of the art (mAP: 0.2547). This performance improvement is attributed to the key modifications in initializing the weights of classifier backbones and constructing model ensembles to reduce prediction variance compared to individual constituent models.

Uncertainty Quantification in Segmenting Tuberculosis-Consistent Findings in Frontal Chest X-rays

Deep learning (DL) methods have demonstrated superior performance in medical image segmentation tasks. However, selecting a loss function that conforms to the data characteristics is critical for optimal performance. Further, the direct use of traditional DL models does not provide a measure of uncertainty in predictions. Even high-quality automated predictions for medical diagnostic applications demand uncertainty quantification to gain user trust. In this study, we aim to investigate the benefits of (i) selecting an appropriate loss function and (ii) quantifying uncertainty in predictions using a VGG16-based-U-Net model with the Monto–Carlo (MCD) Dropout method for segmenting Tuberculosis (TB)-consistent findings in frontal chest X-rays (CXRs). We determine an optimal uncertainty threshold based on several uncertainty-related metrics. This threshold is used to select and refer highly uncertain cases to an expert. Experimental results demonstrate that (i) the model trained with a modified Focal Tversky loss function delivered superior segmentation performance (mean average precision (mAP): 0.5710, 95% confidence interval (CI): (0.4021,0.7399)), (ii) the model with 30 MC forward passes during inference further improved and stabilized performance (mAP: 0.5721, 95% CI: (0.4032,0.7410), and (iii) an uncertainty threshold of 0.7 is observed to be optimal to refer highly uncertain cases.

Deep Learning to Determine the Activity of Pulmonary Tuberculosis on Chest Radiographs

Background Determining the activity of pulmonary tuberculosis on chest radiographs is difficult. Purpose To develop a deep learning model to identify active pulmonary tuberculosis on chest radiographs. Materials and Methods Chest radiographs were retrospectively gathered from a multicenter consecutive cohort with pulmonary tuberculosis who were successfully treated between 2011 and 2017, along with normal radiographs to enrich a negative class. The pretreatment and posttreatment radiographs were labeled as positive and negative classes, respectively. A neural network was trained with those radiographs to calculate the probability of active versus healed tuberculosis. A single-center consecutive cohort (test set 1; 89 patients, 148 radiographs) and data from one multicenter randomized controlled trial (test set 2; 366 patients, 3774 radiographs) were used to test the model. The area under the receiver operating characteristic curve (AUC) was used to evaluate the performance of the model and of the four expert readers. Results In total, 6654 pre- and posttreatment radiographs from 3327 patients (mean age ± standard deviation, 55 years ± 19; 1884 men) with pulmonary tuberculosis and 3182 normal radiographs from as many patients (mean age, 53 years ± 14; 1629 men) were gathered. For test set 1, the model showed a higher AUC (0.83; 95% CI: 0.73, 0.89) than one pulmonologist (0.69; 95% CI: 0.61, 0.76; P < .001) and performed similarly to the other readers (AUC, 0.79-0.80; P = .14-.23). For 200 randomly selected radiographs from test set 2, the model had a higher AUC (0.84) than the pulmonologists (0.71 and 0.74; P < .001 and .01, respectively) and performed similarly to the radiologists (0.79 and 0.80; P = .08 and .06, respectively). The model output increased by 0.30 on average with a higher degree of smear positivity (95% CI: 0.20, 0.39; P < .001) and decreased during treatment (baseline, 3 months, and 6 months: 0.85, 0.51, and 0.26, respectively). Conclusion A deep learning model performed similarly to radiologists for accurately determining the activity of pulmonary tuberculosis on chest radiographs; it also was able to follow posttreatment changes. © RSNA, 2021 Online supplemental material is available for this article.

Chest X-ray analysis with deep learning-based software as a triage test for pulmonary tuberculosis: an individual patient data meta-analysis of diagnostic accuracy

BACKGROUND

Automated radiologic analysis using computer-aided detection software (CAD) could facilitate chest X-ray (CXR) use in tuberculosis diagnosis. There is little to no evidence on the accuracy of commercially-available deep learning-based CAD in different populations, including patients with smear-negative tuberculosis and people living with HIV (PLWH).

METHODS

We collected CXRs and individual patient data (IPD) from studies evaluating CAD in patients self-referring for tuberculosis symptoms with culture or nucleic acid amplification testing as the reference. We re-analyzed CXRs with three CAD (CAD4TB version (v) 6, Lunit v3.1.0.0, and qXR v2). We estimated sensitivity and specificity within each study and pooled using IPD meta-analysis. We used multivariable meta-regression to identify characteristics modifying accuracy.

RESULTS

We included CXRs and IPD of 3727/3967 participants from 4/7 eligible studies. 17% (621/3727) were PLWH. 17% (645/3727) had microbiologically-confirmed tuberculosis. Despite using the same threshold score for classifying CXR in every study, sensitivity and specificity varied from study to study. The software had similar unadjusted accuracy (at 90% pooled sensitivity, pooled specificities were: CAD4TBv6, 56.9% [95%CI:51.7-61.9]; Lunit, 54.1% [44.6-63.3]; qXRv2, 60.5% [51.7-68.6]). Adjusted absolute differences in pooled sensitivity between PLWH and HIV-uninfected participants was: CAD4TBv6, -13.4% [-21.1, -6.9]; Lunit, +2.2% [-3.6, +6.3]; qXRv2: -13.4% [-21.5, -6.6]); between smear-negative and smear-positive tuberculosis was: CAD4TBv6, -12.3% [-19.5, -6.1]; Lunit, -17.2% [-24.6, -10.5]; qXRv2, -16.6% [-24.4, -9.9]. Accuracy was similar to human readers.

CONCLUSIONS

For CAD CXR analysis to be implemented as a high-sensitivity tuberculosis rule-out test, users will need threshold scores identified from their own patient populations, and stratified by HIV- and smear-status.

Deep learning for chest X-ray analysis: A survey

Recent advances in deep learning have led to a promising performance in many medical image analysis tasks. As the most commonly performed radiological exam, chest radiographs are a particularly important modality for which a variety of applications have been researched. The release of multiple, large, publicly available chest X-ray datasets in recent years has encouraged research interest and boosted the number of publications. In this paper, we review all studies using deep learning on chest radiographs published before March 2021, categorizing works by task: image-level prediction (classification and regression), segmentation, localization, image generation and domain adaptation. Detailed descriptions of all publicly available datasets are included and commercial systems in the field are described. A comprehensive discussion of the current state of the art is provided, including caveats on the use of public datasets, the requirements of clinically useful systems and gaps in the current literature.

Contribution and Acceptability of Bacteriological Collection Tools in the Diagnosis of Tuberculosis in Children Infected with HIV

OBJECTIVE

The objective of this study is to evaluate the feasibility and tolerability of new bacteriological samples to diagnose tuberculosis (TB) in HIV-infected children.

METHOD AND PATIENTS

HIV1-infected children with suspicion of TB in Universitary Hospital Sourô Sanon (Burkina Faso) were included in a prospective cohort study. Children underwent three gastric aspirates (GA) if aged <4 years; two GA, one string test (ST) if aged 4-9 years and three sputum, one ST if aged 10-13 years. All children underwent one nasopharyngeal aspirate (NPA) and one stool sample. To assess feasibility and tolerability of procedures, adverse events were identified and pain was rated on different scales. Samples were tested by microscopy, culture, GeneXpert® (Xpert®).

RESULTS

Sixty-three patients were included. Mean age was 8.92 years, 52.38% were females. Ninety-five GA, 67 sputum, 62 NPA, 60 stool and 55 ST had been performed. During sampling, the main adverse events were cough at 68/95 GA and 48/62 NPA; sneeze at 50/95 GA and 38/62 NPA and vomiting at 4/55 ST. On the behavioral scale, the average pain score during collection was 6.38/10 for GA; 7.70/10 for NPA and 1.03/10 for ST. Of the 31 cases of TB, bacteriological confirmation was made in 12 patients.

CONCLUSION

ST, stool is well-tolerated alternatives specimens for diagnosing TB in children. NPA has a poor feasibility and tolerability in children.

Robust classification from noisy labels: Integrating additional knowledge for chest radiography abnormality assessment

Chest radiography is the most common radiographic examination performed in daily clinical practice for the detection of various heart and lung abnormalities. The large amount of data to be read and reported, with more than 100 studies per day for a single radiologist, poses a challenge in consistently maintaining high interpretation accuracy. The introduction of large-scale public datasets has led to a series of novel systems for automated abnormality classification. However, the labels of these datasets were obtained using natural language processed medical reports, yielding a large degree of label noise that can impact the performance. In this study, we propose novel training strategies that handle label noise from such suboptimal data. Prior label probabilities were measured on a subset of training data re-read by 4 board-certified radiologists and were used during training to increase the robustness of the training model to the label noise. Furthermore, we exploit the high comorbidity of abnormalities observed in chest radiography and incorporate this information to further reduce the impact of label noise. Additionally, anatomical knowledge is incorporated by training the system to predict lung and heart segmentation, as well as spatial knowledge labels. To deal with multiple datasets and images derived from various scanners that apply different post-processing techniques, we introduce a novel image normalization strategy. Experiments were performed on an extensive collection of 297,541 chest radiographs from 86,876 patients, leading to a state-of-the-art performance level for 17 abnormalities from 2 datasets. With an average AUC score of 0.880 across all abnormalities, our proposed training strategies can be used to significantly improve performance scores.

Analyzing inter-reader variability affecting deep ensemble learning for COVID-19 detection in chest radiographs

Data-driven deep learning (DL) methods using convolutional neural networks (CNNs) demonstrate promising performance in natural image computer vision tasks. However, their use in medical computer vision tasks faces several limitations, viz., (i) adapting to visual characteristics that are unlike natural images; (ii) modeling random noise during training due to stochastic optimization and backpropagation-based learning strategy; (iii) challenges in explaining DL black-box behavior to support clinical decision-making; and (iv) inter-reader variability in the ground truth (GT) annotations affecting learning and evaluation. This study proposes a systematic approach to address these limitations through application to the pandemic-caused need for Coronavirus disease 2019 (COVID-19) detection using chest X-rays (CXRs). Specifically, our contribution highlights significant benefits obtained through (i) pretraining specific to CXRs in transferring and fine-tuning the learned knowledge toward improving COVID-19 detection performance; (ii) using ensembles of the fine-tuned models to further improve performance over individual constituent models; (iii) performing statistical analyses at various learning stages for validating results; (iv) interpreting learned individual and ensemble model behavior through class-selective relevance mapping (CRM)-based region of interest (ROI) localization; and, (v) analyzing inter-reader variability and ensemble localization performance using Simultaneous Truth and Performance Level Estimation (STAPLE) methods. We find that ensemble approaches markedly improved classification and localization performance, and that inter-reader variability and performance level assessment helps guide algorithm design and parameter optimization. To the best of our knowledge, this is the first study to construct ensembles, perform ensemble-based disease ROI localization, and analyze inter-reader variability and algorithm performance for COVID-19 detection in CXRs.

Effects of time of day on radiological interpretation

Accurate interpretation of radiological images involves a complex visual search that relies on several cognitive processes, including selective attention, working memory, and decision-making. Patient outcomes often depend on the accuracy of image interpretations, and yet research has revealed that conclusions vary significantly from one radiologist to another. A myriad of factors has been shown to contribute to the likelihood of interpretative errors and discrepancies, including the radiologist's level of experience and fatigue, and these factors are well reported elsewhere; however, a potentially important factor that has been given little previous consideration is how radiologists' interpretations might be impacted by the time of day at which the reading takes place, a factor that other disciplines have shown to be a determinant of competency. The available literature shows that while the time of day is known to significantly impact some cognitive functions that likely relate to reading competence, including selective visual attention and visual working memory, little is known about the impact of the time of day on radiology interpretation performance. This review explores the evidence regarding the relationship between time of day and performance, with a particular emphasis on radiological activities.

Predictors of Treatment Outcomes among Multidrug Resistant Tuberculosis Patients in Tanzania

Background

According to World Health Organization (WHO) the final multidrug resistant tuberculosis (MDRTB) treatment outcome is the most important direct measurement of the effectiveness of the MDRTB control program. Literature review has shown marked diversity in predictors of treatment outcomes worldwide even among the same continents. Therefore, findings could also be different in Tanzanian context, where the success rate is still lower than the WHO recommendation. This study sought to determine the predictors of treatment outcomes among MDRTB patients in Tanzania in order to improve the success rate.

Methodology

This was a retrospective cohort study, which was conducted at Kibong'oto Infectious Diseases Hospital (KIDH) in Tanzania. Patients' demographic and clinical parameters were collected from the MDRTB registry and clinical files. Then, a detailed analysis was done to determine the predictors of successful and unsuccessful MDRTB treatment outcomes.

Results

Three hundred and thirty-two patients were diagnosed and put on MDRTB treatment during the year 2009 to 2014. Among them, males were 221 (67%), and 317 (95.48%) were above 18 years of age, mean age being 36.9 years. One hundred and sixty-one patients (48.5%) were living in Dar es Salaam. The number of MDRTB patients has increased from 16 in 2009 to 132 in 2014. Majority of patients (75.7%) had successful treatment outcomes. The following predictors were significantly associated with MDRTB cure: presence of cavities in chest X-rays (aOR 1.89, p value 0.002), low BMI (aOR 0.59, p value 0.044), and resistance to streptomycin (aOR 4.67, p value 0.007) and ethambutol (aOR 0.34, p value 0.041). Smoking and presence of cavities in chest X-rays were associated with MDRTB mortality, aOR 2.31, p value 0.043 and aOR 0.55, p value 0.019, respectively.

Conclusion

The study indicated that overall number of MDRTB patients and the proportion of successful treatment outcomes have been increasing over the years. The study recommends improving nutritional status of MDRTB patients, widespread antismoking campaign, and close follow-up of patients with ethambutol resistance.

Pretreatment chest x-ray severity and its relation to bacterial burden in smear positive pulmonary tuberculosis

BACKGROUND

Chest radiographs are used for diagnosis and severity assessment in tuberculosis (TB). The extent of disease as determined by smear grade and cavitation as a binary measure can predict 2-month smear results, but little has been done to determine whether radiological severity reflects the bacterial burden at diagnosis.

METHODS

Pre-treatment chest x-rays from 1837 participants with smear-positive pulmonary TB enrolled into the REMoxTB trial (Gillespie et al., N Engl J Med 371:1577-87, 2014) were retrospectively reviewed. Two clinicians blinded to clinical details using the Ralph scoring system performed separate readings. An independent reader reviewed discrepant results for quality assessment and cavity presence. Cavitation presence was plotted against time to positivity (TTP) of sputum liquid cultures (MGIT 960). The Wilcoxon rank sum test was performed to calculate the difference in average TTP for these groups. The average lung field affected was compared to log ₁₀ TTP by linear regression. Baseline markers of disease severity and patient characteristics were added in univariable regression analysis against radiological severity and a multivariable regression model was created to explore their relationship.

RESULTS

For 1354 participants, the median TTP was 117 h (4.88 days), being 26 h longer (95% CI 16-30, p < 0.001) in patients without cavitation compared to those with cavitation. The median percentage of lung-field affected was 18.1% (IQR 11.3-28.8%). For every 10-fold increase in TTP, the area of lung field affected decreased by 11.4%. Multivariable models showed that serum albumin decreased significantly as the percentage of lung field area increased in both those with and without cavitation. In addition, BMI and logged TTP had a small but significant effect in those with cavitation and the number of severe TB symptoms in the non-cavitation group also had a small effect, whilst other factors found to be significant on univariable analysis lost this effect in the model.

CONCLUSIONS

The radiological severity of disease on chest x-ray prior to treatment in smear positive pulmonary TB patients is weakly associated with the bacterial burden. When compared against other variables at diagnosis, this effect is lost in those without cavitation. Radiological severity does reflect the overall disease severity in smear positive pulmonary TB, but we suggest that clinicians should be cautious in over-interpreting the significance of radiological disease extent at diagnosis.

Evaluation of a chest radiograph reading and recording system for tuberculosis in a HIV-positive cohort

AIM

To assess the impact of introducing a chest radiograph reading and recording system (CRRS) with a short training session, on the accuracy and inter-reader variability of tuberculosis (TB) interpretation of chest radiographs (CXRs) by a group of non-expert readers in a human immunodeficiency virus (HIV)-positive cohort.

MATERIALS AND METHODS

A set of 139 CXRs was reviewed by a group of eight physicians pre- and post-intervention at two clinics in Shan State, Myanmar, providing HIV/TB diagnosis and treatment services. The results were compared against the consensus of expert radiologists for accuracy.

RESULTS

Overall accuracy was similar pre- and post-intervention for most physicians with an average area under the receiver operating characteristic curve difference of 0.02 (95% confidence interval: -0.03, 0.07). The overall agreement among physicians was poor pre- and post-intervention (Fleiss κ=0.35 and κ=0.29 respectively). The assessment of agreement for specific disease patterns associated with active TB in HIV-infected patients showed that for intrinsically subtle findings, the agreement was generally poor but better for the more intrinsically obvious disease patterns: pleural effusion (Cohen's kappa range = 0.37-0.67) and milliary nodular pattern (Cohen's kappa range = 0.25-0.52).

CONCLUSION

This study demonstrated limited impact of the introduction of a CRRS on CXR accuracy and agreement amongst non-expert readers. The role in which CXRs are used for TB diagnosis in a HIV-positive cohort in similar clinical contexts should be reviewed.

Tuberculosis, Airway Obstruction and Mortality in a Finnish Population

There is little long-term follow-up data concerning the association between past pulmonary tuberculosis (TB), airway obstruction and mortality. We aimed to analyse a national health examination survey data from 6701 adult Finns undergoing spirometry between 1978 and 1980 (follow-up through 2013). We identified TB either through a disease history or by a TB-indicative scar on a chest x-ray. We specified obstruction using the lower limit of normal (LLN) and classified severity using the Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages 1-4. After adjusting for smoking and other confounders, past TB associated with obstruction. Compared to non-TB patients, the adjusted odds ratio (OR; 95% CI) of obstruction reached 2.21 (1.52-3.21) among patients with a scar recorded by one radiologist, 2.48 (1.63-3.78) when recorded by both radiologists and 4.59 (2.86-7.37) among patients with a disease history. Among those with neither past TB nor obstruction, with past TB only, with an obstruction only and with both, we found hazard ratios (HRs; 95% CIs) for subsequent mortality of 1.00 (reference), 1.11 (1.03-1.20), 1.62 (1.31-2.00) and 1.77 (1.45-2.16), adjusted for age, gender, smoking, body mass index (BMI), physical activity, education and general health. In conclusion, past TB strongly determines obstruction, although on its own quite weakly predicts premature death. TB and obstruction combined predict an additive mortality pattern.

Chest radiograph reading and recording system: evaluation in frontline clinicians in Zambia

Background

In Zambia the vast majority of chest radiographs (CXR) are read by clinical officers who have limited training and varied interpretation experience, meaning lower inter-rater reliability and limiting the usefulness of CXR as a diagnostic tool. In 2010–11, the Zambian Prison Service and Ministry of Health established TB and HIV screening programs in six prisons; screening included digital radiography for all participants. Using front-line clinicians we evaluated sensitivity, specificity and inter-rater agreement for digital CXR interpretation using the Chest Radiograph Reading and Recording System (CRRS).

Methods

Digital radiographs were selected from HIV-infected and uninfected inmates who participated in a TB and HIV screening program at two Zambian prisons. Two medical officers (MOs) and two clinical officers (COs) independently interpreted all CXRs. We calculated sensitivity and specificity of CXR interpretations compared to culture as the gold standard and evaluated inter-rater reliability using percent agreement and kappa coefficients.

Results

571 CXRs were included in analyses. Sensitivity of the interpretation “any abnormality” ranged from 50–70 % depending on the reader and the patients’ HIV status. In general, MO’s had higher specificities than COs. Kappa coefficients for the ratings of “abnormalities consistent with TB” and “any abnormality” showed good agreement between MOs on HIV-uninfected CXRs and moderate agreement on HIV-infected CXRs whereas the COs demonstrated fair agreement in both categories, regardless of HIV status.

Conclusions

Sensitivity, specificity and inter-rater agreement varied substantially between readers with different experience and training, however the medical officers who underwent formal CRRS training had more consistent interpretations.

Electronic supplementary material

The online version of this article (doi:10.1186/s12879-016-1460-z) contains supplementary material, which is available to authorized users.

Screening for pulmonary tuberculosis in a Tanzanian prison and computer-aided interpretation of chest X-rays

SETTING

Tanzania is a high-burden country for tuberculosis (TB), and prisoners are a high-risk group that should be screened actively, as recommended by the World Health Organization. Screening algorithms, starting with chest X-rays (CXRs), can detect asymptomatic cases, but depend on experienced readers, who are scarce in the penitentiary setting. Recent studies with patients seeking health care for TB-related symptoms showed good diagnostic performance of the computer software CAD4TB.

OBJECTIVE

To assess the potential of computer-assisted screening using CAD4TB in a predominantly asymptomatic prison population.

DESIGN

Cross-sectional study.

RESULTS

CAD4TB and seven health care professionals reading CXRs in local tuberculosis wards evaluated a set of 511 CXRs from the Ukonga prison in Dar es Salaam. Performance was compared using a radiological reference. Two readers performed significantly better than CAD4TB, three were comparable, and two performed significantly worse (area under the curve 0.75 in receiver operating characteristics analysis). On a superset of 1321 CXRs, CAD4TB successfully interpreted >99%, with a predictably short time to detection, while 160 (12.2%) reports were delayed by over 24 h with conventional CXR reading.

CONCLUSION

CAD4TB reliably evaluates CXRs from a mostly asymptomatic prison population, with a diagnostic performance inferior to that of expert readers but comparable to local readers.

Speed and accuracy of mobile BlackBerry Messenger to transmit chest radiography images from a small community emergency department to a geographically remote referral center

BACKGROUND

Small emergency departments (EDs) may rely on radiologists at remote centers for interpretations of chest radiographs (CXRs). We investigated systematic transmission of CXR images from a small ED to a geographically remote referral center using the mobile BlackBerry Messenger (BBM) application.

METHODS

Investigators obtained de-identified CXR images of consecutive ED patients via mobile phone camera. Each CXR image, along with a brief clinical history, was sent via BBM to an emergency physician located at a remote referral site, and the receiving physician replied via BBM to confirm reception. All communications, image generation, and image analysis was conducted on mobile phones. The primary outcome was the proportion of BBMs received within two minutes of sending; the secondary outcome was the proportion of BBM replies to the sending physician within five minutes. Image accuracy-comparing the radiologist's interpretation with the receiving emergency physician's interpretation-was estimated using predefined criteria.

RESULTS

Of 1281 consecutive ED patients, 231 (18.0 %) had CXRs obtained, 320 CXRs were analyzed and 611 BBMs sent. All BBMs (100.0%, 95% confidence interval (CI) 99.4-00.0) arrived within two minutes; 595 BBMs (97.4%, 95% CI 95.8-98.4) were replied to within five minutes. Of the 58 CXRs with abnormalities requiring intervention, there were 55 concordances (overall agreement 94.2%, 95% CI 85.9-98.3; kappa 0.95, 95% CI 0.89-1.0) CONCLUSION: Systematic transmission of CXR images from a small ED to a remote large center using mobile phones may be a safe and effective strategy to rapidly communicate important patient information.

Robust and Reproducible Quantification of the Extent of Chest Radiographic Abnormalities (And It's Free!)

Rationale

Objective, reproducible quantification of the extent of abnormalities seen on a chest radiograph would improve the user-friendliness of a previously proposed severity scoring system for pulmonary tuberculosis and could be helpful in monitoring response to therapy, including in clinical trials.

Methods

In this study we report the development and evaluation of a simple tool using free image editing software (GIMP) to accurately and reproducibly quantify the area of affected lung on the chest radiograph of tuberculosis patients. As part of a pharmacokinetic study in Lima, Peru, a chest radiograph was performed on patients with pulmonary tuberculosis and this was subsequently photographed using a digital camera. The GIMP software was used by two independent and trained readers to estimate the extent of affected lung (expressed as a percentage of total lung area) in each radiograph and the resulting radiographic SCORE.

Results

56 chest radiographs were included in the reading analysis. The Intraclass correlation coefficient (ICC) between the 2 observers was 0.977 (p<0.001) for the area of lung affected and was 0.955 (p<0.001) for the final score; and the kappa coefficient of Interobserver agreement for both the area of lung affected and the score were 0.9 (p<0.001) and 0.86 (p<0.001) respectively.

Conclusions

This high level of between-observer agreement suggests that this freely available software could constitute a simple and useful tool for robust evaluation of individual and serial chest radiographs.

A novel multiple-instance learning-based approach to computer-aided detection of tuberculosis on chest X-rays

To reach performance levels comparable to human experts, computer-aided detection (CAD) systems are typically optimized following a supervised learning approach that relies on large training databases comprising manually annotated lesions. However, manually outlining those lesions constitutes a difficult and time-consuming process that renders detailedly annotated data difficult to obtain. In this paper, we investigate an alternative approach, namely multiple-instance learning (MIL), that does not require detailed information for optimization. We have applied MIL to a CAD system for tuberculosis detection. Only the case condition (normal or abnormal) was required during training. Based upon the well-known miSVM technique, we propose an improved algorithm that overcomes miSVM's drawbacks related to positive instance underestimation and costly iteration. To show the advantages of our MIL-based approach as compared with a traditional supervised one, experiments with three X-ray databases were conducted. The area under the receiver operating characteristic curve was utilized as a performance measure. With the first database, for which training lesion annotations were available, our MIL-based method was comparable to the supervised system ( 0.86 versus 0.88 ). When evaluating the remaining databases, given their large difference with the previous image set, the most appealing strategy was to retrain the CAD systems. However, since only the case condition was available, only the MIL-based system could be retrained. This scenario, which is common in real-world applications, demonstrates the better adaptation capabilities of the proposed approach. After retraining, our MIL-based system significantly outperformed the supervised one ( 0.86 versus 0.79 and 0.91 versus 0.85 , and p=0.0002 , respectively).

Diagnostic accuracy of computer-aided detection of pulmonary tuberculosis in chest radiographs: a validation study from sub-Saharan Africa

Background

Chest radiography to diagnose and screen for pulmonary tuberculosis has limitations, especially due to inter-reader variability. Automating the interpretation has the potential to overcome this drawback and to deliver objective and reproducible results. The CAD4TB software is a computer-aided detection system that has shown promising preliminary findings. Evaluation studies in different settings are needed to assess diagnostic accuracy and practicability of use.

Methods

CAD4TB was evaluated on chest radiographs of patients with symptoms suggestive of pulmonary tuberculosis enrolled in two cohort studies in Tanzania. All patients were characterized by sputum smear microscopy and culture including subsequent antigen or molecular confirmation of Mycobacterium tuberculosis (M.tb) to determine the reference standard. Chest radiographs were read by the software and two human readers, one expert reader and one clinical officer. The sensitivity and specificity of CAD4TB was depicted using receiver operating characteristic (ROC) curves, the area under the curve calculated and the performance of the software compared to the results of human readers.

Results

Of 861 study participants, 194 (23%) were culture-positive for M.tb. The area under the ROC curve of CAD4TB for the detection of culture-positive pulmonary tuberculosis was 0.84 (95% CI 0.80–0.88). CAD4TB was significantly more accurate for the discrimination of smear-positive cases against non TB patients than for smear-negative cases (p-value<0.01). It differentiated better between TB cases and non TB patients among HIV-negative compared to HIV-positive individuals (p<0.01). CAD4TB significantly outperformed the clinical officer, but did not reach the accuracy of the expert reader (p = 0.02), for a tuberculosis specific reading threshold.

Conclusion

CAD4TB accurately distinguished between the chest radiographs of culture-positive TB cases and controls. Further studies on cost-effectiveness, operational and ethical aspects should determine its place in diagnostic and screening algorithms.

Time-to-detection in culture predicts risk of Mycobacterium tuberculosis transmission: a cohort study

BACKGROUND

Contact screening is an essential component of all tuberculosis control strategies. We hypothesize that time-to-detection (TTD) in liquid culture of spontaneously produced sputum samples may help identify index cases at high risk of transmission.

METHODS

We studied retrospectively a cohort of patients with pulmonary tuberculosis in Birmingham, United Kingdom (January 2010-December 2012). We studied the correlation of TTD with the risk of transmission of infection from index cases to contacts and compared this with sputum microscopy. Chest radiographs (CXRs) were graded from 0 to 6 (0, no radiographic evidence of disease; 5, bilateral cavitation; and 6, miliary disease).

RESULTS

Of the 184 cases of pulmonary tuberculosis reported during the study period, 111 were included in the final study, and these generated 825 contacts. A transmission event (new latent or active tuberculosis) was identified in 165 contacts (transmission rate 0.20). Short TTD (<9 days) was associated with an increased risk of transmission (odds ratio, 2.56; P < .001), and this relationship persisted after adjusting for potential confounders. A 1-point increase in CXR grade correlated with a 3.2-day decrease in TTD (P < .001), and this correlation persisted after adjustment for potential confounders.

CONCLUSIONS

TTD < 9 days identifies patients at high risk of transmitting tuberculosis and is superior to sputum smear. CXR grade at diagnosis predicts patients with short TTD. Our findings have the potential to guide the organization and prioritization of contact investigations in similar settings.

The sensitivity and specificity of using a computer aided diagnosis program for automatically scoring chest X-rays of presumptive TB patients compared with Xpert MTB/RIF in Lusaka Zambia

Objective

To determine the sensitivity and specificity of a Computer Aided Diagnosis (CAD) program for scoring chest x-rays (CXRs) of presumptive tuberculosis (TB) patients compared to Xpert MTB/RIF (Xpert).

Method

Consecutive presumptive TB patients with a cough of any duration were offered digital CXR, and opt out HIV testing. CXRs were electronically scored as normal (CAD score ≤60) or abnormal (CAD score>60) using a CAD program. All patients regardless of CAD score were requested to submit a spot sputum sample for testing with Xpert and a spot and morning sample for testing with LED Fluorescence Microscopy-(FM).

Results

Of 350 patients with evaluable data, 291 (83.1%) had an abnormal CXR score by CAD. The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of CXR compared to Xpert were 100% (95%CI 96.2–100), 23.2% (95%CI 18.2–28.9), 33.0% (95%CI 27.6–38.7) and 100% (95% 93.9–100), respectively. The area under the receiver operator curve (AUC) for CAD was 0.71 (95%CI 0.66–0.77). CXR abnormality correlated with smear grade (r = 0.30, p<0.0001) and with Xpert C_T(r = 0.37, p<0.0001).

Conclusions

To our knowledge this is the first time that a CAD program for TB has been successfully tested in a real world setting. The study shows that the CAD program had high sensitivity but low specificity and PPV. The use of CAD with digital CXR has the potential to increase the use and availability of chest radiography in screening for TB where trained human resources are scarce.

Laboratory diagnosis of paediatric tuberculosis in the European Union/European Economic Area: analysis of routine laboratory data, 2007 to 2011

Laboratory confirmation of paediatric tuberculosis (TB) is frequently lacking. We reviewed the range of routine laboratory tests and their performance in different biological samples used to diagnose active TB in children. A questionnaire-based survey was conducted among the European Reference Laboratory Network for TB followed by collection of routine laboratory data on 10,549 paediatric samples tested in 2007 to 2011 at six reference laboratories (in Croatia, Germany, Italy, Latvia, Lithuania and the United Kingdom (UK)). The questionnaire showed that all laboratories used rapid assays. Non-respiratory samples were collected more often in Germany (135/275, 49.1%) and the UK (490/2,140, 22.9%) compared with Croatia (138/2,792, 4.9%), Latvia (222/2,401, 9.2%) and Lithuania (76/1,549, 4.9%). Overall laboratory positivity rates (isolation of Mycobacterium tuberculosis complex and/or identification of its nucleic acids in a sample) were higher in lymph node and gastric aspirate samples (14/203 (6.9%) and 43/1,231 (3.5%)) than in sputum samples (89/4,684 (1.9%)). Pooled sensitivity, specificity, positive and negative predictive values and accuracy of molecular assays assessed against solid or liquid culture were 79.2%, 93.6%, 67.1%, 96.5% and 91.6%, respectively. A more intensive approach in obtaining gastric aspirate and non-respiratory samples may increase laboratory confirmation of paediatric TB. Major effort is needed in optimisation and validation of molecular tests in these samples.

Active pulmonary tuberculosis and latent tuberculosis infection among homeless people in Seoul, South Korea: a cross-sectional study

Background

The aim of this study was to determine the prevalence rate of latent TB infection (LTBI) and active TB among homeless in Seoul metropolitan city, South Korea, and to compare the TB burden among homeless people with that of a control group.

Methods

The homeless participants were recruited from five sites between October 30, 2009 and April 12, 2010. LTBI was diagnosed through the QuantiFERON(R) TB Gold In-Tube(QFT-GIT) assay and a tuberculin skin test(TST) and, and active PTB was diagnosed based on chest radiography.

Results

Among 313 participants, the prevalence of LTBI was 75.9% (95% CI, 71.1-80.8%) and 79.8% (95% CI, 74.9-84.7%) based on a QFT-GIT assay and the TST, respectively, and that of active PTB was 5.8% (95% CI, 3.2-8.3%). The prevalence of LTBI among homeless participants was about five times higher than controls. Also, the age-specific prevalence rate ratio of active PTB was as high as 24.86.

Conclusions

The prevalence rate of LTBI as well as active PTB among homeless people was much higher than that of the general population in South Korea. Thus, adequate strategies to reduce the TB burden among homeless people are needed.

The effect of a tuberculosis chest X-ray image reference set on non-expert reader performance

OBJECTIVES

In low-resource settings, limitations in diagnostic accuracy of chest X-rays (CXR) for pulmonary tuberculosis (PTB) relate partly to non-expert interpretation. We piloted a TB CXR Image Reference Set (TIRS) to improve non-expert performance in an operational setting in Malawi.

METHODS

Nineteen doctors and clinical officers read 60 CXR of patients with suspected PTB, at baseline and using TIRS. Two officers also used the CXR Reading and Recording System (CRRS). Correct treatment decisions were assessed against a "gold standard" of mycobacterial culture and expert performance.

RESULTS

TIRS significantly increased overall non-expert sensitivity from 67.6 (SD 14.9) to 75.5 (SD 11.1, P = 0.013), approaching expert values of 84.2 (SD 5.2). Among doctors, correct decisions increased from 60.7 % (SD 7.9) to 67.1 % (SD 8.0, P = 0.054). Clinical officers increased in sensitivity from 68.0 % (SD 15) to 77.4 % (SD 10.7, P = 0.056), but decreased in specificity from 55.0 % (SD 23.9) to 40.8 % (SD 10.4, P = 0.049). Two officers made correct treatment decisions with TIRS in 62.7 %. CRRS training increased this to 67.8 %.

CONCLUSION

Use of a CXR image reference set increased correct decisions by doctors to treat PTB. This tool may provide a low-cost intervention improving non-expert performance, translating into improved clinical care. Further evaluation is warranted.

KEY POINTS

• Tuberculosis treatment decisions are influenced by CXR findings, despite improved laboratory diagnostics. • In low-resource settings, CXR interpretation is performed largely by non-experts. • We piloted the effect of a simple reference training set of CXRs. • Use of the reference set increased the number of correct treatment decisions. This effect was more marked for doctors than clinical officers. • Further evaluation of this simple training tool is warranted.

Identification and interpretation of maxillofacial plain radiographs by junior dental trainees

BACKGROUND

Accurate identification and interpretation of plain maxillofacial radiographs is key to making correct clinical decision.

OBJECTIVE

To assess the ability of junior dental trainees to correctly identify and adequately interpret oral and maxillofacial plain radiographs.

STUDY DESIGN

Comparative analysis Setting: University of Port Harcourt Teaching Hospital, Port Harcourt and University College Hospital, Ibadan.

MATERIALS AND METHODS

Twenty six plain film radiographs were selected and serialized. The films were presented to three specialislts; a radiologist and two oral surgeons for independent identification and interpretation. The level of agreement between them was tested using kappa statistics (k) and intraclass coefficient (ICC). Minor areas of discrepancy were finally reconciled and the correct identification and interpretations of every radiograph confirmed. The same set of radiographs was subsequently presented to twenty junior dentists (House officers and Registrars) for identification and interpretation. Accuracy of performances in identification and interpretation exercises was assessed by a test of agreement using kappa statistics and a mathematical performance rating method respectively.

RESULTS

In terms of identification, the agreement between the three specialists was very strong with ICC of 0.96. Kappa (k)-values of 1.00 suggesting perfect agreement was observed between the two oral surgeons. Agreement between each oral surgeon and the radiologist was very good (k= 0.84). The k-value for agreement in identification between trainees and specialists ranged between 0.23 and 1.00. As for interpretation, the percentage accuracy of the junior dental trainees ranged between 60.5% and 87.2% compared to specialists' range of 89.5% to 95.3%. The common areas of discrepancy in identification and interpretation are highlighted.

CONCLUSION

Based on this study, dental trainees demonstrate varying levels of expertise in identification and interpretation of maxillofacial plain radiographs. Knowledge gaps were identified and modification of teaching method suggested.

Diagnostic accuracy of notified cases as pulmonary tuberculosis in private sectors of Korea

The diagnostic accuracy of the data reported in the Korean tuberculosis surveillance system (KTBS) has not been adequately investigated. We reviewed the clinical data of pulmonary tuberculosis (PTB) cases notified from private medical facilities through KTBS between January and June, 2004. PTB cases were classified into definite (culture-proven), probable (based on smear, polymerase chain reaction, histology, bronchoscopic finding, computed tomography, or both chest radiograph and symptoms) or possible (based only on chest radiograph) tuberculosis. Of the 1126 PTB cases, sputum AFB smear and culture were requested in 79% and 51% of the cases, respectively. Positive results of sputum smear and culture were obtained in 43% and 29% of all the patients, respectively. A total of 73.2% of the notified PTB cases could be classified as definite or probable and 81.7% as definite, probable, or possible. However, where infection was not confirmed bacteriologically or histologically, only 60.1% of the patients were definite, probable, or possible cases. More than 70% of PTB notified from private sectors in Korea can be regarded as real TB. The results may also suggest the possibility of over-estimation of TB burden in the use of the notification-based TB data.

Inter-rater agreement in the assessment of abnormal chest X-ray findings for tuberculosis between two Asian countries

Background

Inter-rater agreement in the interpretation of chest X-ray (CXR) films is crucial for clinical and epidemiological studies of tuberculosis. We compared the readings of CXR films used for a survey of tuberculosis between raters from two Asian countries.

Methods

Of the 11,624 people enrolled in a prevalence survey in Hanoi, Viet Nam, in 2003, we studied 258 individuals whose CXR films did not exclude the possibility of active tuberculosis. Follow-up films obtained from accessible individuals in 2006 were also analyzed. Two Japanese and two Vietnamese raters read the CXR films based on a coding system proposed by Den Boon et al. and another system newly developed in this study. Inter-rater agreement was evaluated by kappa statistics. Marginal homogeneity was evaluated by the generalized estimating equation (GEE).

Results

CXR findings suspected of tuberculosis differed between the four raters. The frequencies of infiltrates and fibrosis/scarring detected on the films significantly differed between the raters from the two countries (P < 0.0001 and P = 0.0082, respectively, by GEE). The definition of findings such as primary cavity, used in the coding systems also affected the degree of agreement.

Conclusions

CXR findings were inconsistent between the raters with different backgrounds. High inter-rater agreement is a component necessary for an optimal CXR coding system, particularly in international studies. An analysis of reading results and a thorough discussion to achieve a consensus would be necessary to achieve further consistency and high quality of reading.

A comparative examination of tuberculosis immigration medical screening programs from selected countries with high immigration and low tuberculosis incidence rates

BACKGROUND

Tuberculosis (TB) in migrants is an ongoing challenge in several low TB incidence countries since a large proportion of TB in these countries occurs in migrants from high incidence countries. To meet these challenges, several countries utilize TB screening programs. The programs attempt to identify and treat those with active and/or infectious stages of the disease. In addition, screening is used to identify and manage those with latent or inactive disease after arrival. Between nations, considerable variation exists in the methods used in migration-associated TB screening. The present study aimed to compare the TB immigration medical examination requirements in selected countries of high immigration and low TB incidence rates.

METHODS

Descriptive study of immigration TB screening programs.

RESULTS

16 out of 18 eligible countries responded to the written standardized survey and phone interview. Comparisons in specific areas of TB immigration screening programs included authorities responsible for TB screening, the primary objectives of the TB screening program, the yield of detection of active TB disease, screening details and aspects of follow up for inactive pulmonary TB. No two countries had the same approach to TB screening among migrants. Important differences, common practices, common problems, evidence or lack of evidence for program specifics were noted.

CONCLUSIONS

In spite of common goals, there is great diversity in the processes and practices designed to mitigate the impact of migration-associated TB among nations that screen migrants for the disease. The long-term goal in decreasing migration-related introduction of TB from high to low incidence countries remains diminishing the prevalence of the disease in those high incidence locations. In the meantime, existing or planned migration screening programs for TB can be made more efficient and evidenced based. Cooperation among countries doing research in the areas outlined in this study should facilitate the development of improved screening programs.

Effect of improving the quality of radiographic interpretation on the ability to predict pulmonary tuberculosis relapse

RATIONALE AND OBJECTIVES

Chest radiographic findings are important for diagnosis and management of tuberculosis. The reliability of these findings is therefore of interest. We sought to describe interobserver reliability of chest radiographic findings in pulmonary tuberculosis, and to understand how the reliability of these findings might affect the utility of radiographic findings in predicting tuberculosis relapse.

MATERIALS AND METHODS

Three blinded readers independently reviewed chest radiographs from a randomly selected group of 10% of HIV-seronegative subjects participating in a tuberculosis treatment trial. The three readers then arrived at a fourth, consensus radiographic interpretation.

RESULTS

A total of 241 films obtained from 99 patients were reviewed. Agreement among the independent readers was very good for the findings of bilateral disease (kappa = 0.71-0.86 among readers) and cavitation (kappa = 0.66-0.73). The original interpretation was reasonably sensitive and specific (compared to the consensus interpretation) for bilateral disease, but the sensitivity for cavity decreased from 81% for the 2-month film to 47% at end of treatment (P = 0.013). Substituting the consensus interpretation for the original interpretation increased the odds ratio for the association between cavitation on early chest radiograph and subsequent tuberculosis relapse from 4.97 to 8.97.

CONCLUSION

Radiographic findings were reasonably reliable between independent reviewers and the original interpretations. The original investigators, who knew the patient's clinical course, were less likely to identify cavitation on the end of treatment chest radiograph. Improving the reliability of these findings could improve the utility of chest radiographs for predicting tuberculosis relapse.

Seasonality in pulmonary tuberculosis among migrant workers entering Kuwait

BACKGROUND

There is paucity of data on seasonal variation in pulmonary tuberculosis (TB) in developing countries contrary to recognized seasonality in the TB notification in western societies. This study examined the seasonal pattern in TB diagnosis among migrant workers from developing countries entering Kuwait.

METHODS

Monthly aggregates of TB diagnosis results for consecutive migrants tested between January I, 1997 and December 31, 2006 were analyzed. We assessed the amplitude (alpha) of the sinusoidal oscillation and the time at which maximum (theta degrees ) TB cases were detected using Edwards' test. The adequacy of the hypothesized sinusoidal curve was assessed by chi2 goodness-of-fit test.

RESULTS

During the 10 year study period, the proportion (per 100,000) of pulmonary TB cases among the migrants was 198 (4608/2328582), (95% confidence interval: 192 - 204). The adjusted mean monthly number of pulmonary TB cases was 384. Based on the observed seasonal pattern in the data, the maximum number of TB cases was expected during the last week of April (theta degrees = 112 degrees ; P < 0.001). The amplitude (+/- se) (alpha = 0.204 +/- 0.04) of simple harmonic curve showed 20.4% difference from the mean to maximum TB cases. The peak to low ratio of adjusted number of TB cases was 1.51 (95% CI: 1.39 - 1.65). The chi2 goodness-of-test revealed that there was no significant (P > 0.1) departure of observed frequencies from the fitted simple harmonic curve. Seasonal component explained 55% of the total variation in the proportions of TB cases (100,000) among the migrants.

CONCLUSION

This regularity of peak seasonality in TB case detection may prove useful to institute measures that warrant a better attendance of migrants. Public health authorities may consider re-allocation of resources in the period of peak seasonality to minimize the risk of Mycobacterium tuberculosis infection to close contacts in this and comparable settings in the region having similar influx of immigrants from high TB burden countries. Epidemiological surveillance for the TB risk in the migrants in subsequent years and required chemotherapy of detected cases may contribute in global efforts to control this public health menace.

Prevalence of abnormal radiological findings in health care workers with latent tuberculosis infection and correlations with T cell immune response

Background

More than half of all health care workers (HCWs) in high TB-incidence, low and middle income countries are latently infected with tuberculosis (TB). We determined radiological lesions in a cohort of HCWs with latent TB infection (LTBI) in India, and determined their association with demographic, occupational and T-cell immune response variables.

Methodology

We obtained chest radiographs of HCWs who had undergone tuberculin skin test (TST) and QuantiFERON-TB Gold In Tube (QFT), an interferon-γ release assay, in a previous cross-sectional study, and were diagnosed to have LTBI because they were positive by either TST or QFT, but had no evidence of clinical disease. Two observers independently interpreted these radiographs using a standardized data form and any discordance between them resolved by a third observer. The radiological diagnostic categories (normal, suggestive of inactive TB, and suggestive of active TB) were compared with results of TST, QFT assay, demographic, and occupational covariates.

Results

A total of 330 HCWs with positive TST or QFT underwent standard chest radiography. Of these 330, 113 radiographs (34.2%) were finally classified as normal, 206 (62.4%) had lesions suggestive of inactive TB, and 11 (3.4%) had features suggestive of active TB. The mean TST indurations and interferon-γ levels in the HCWs in these three categories were not significantly different. None of the demographic or occupational covariates was associated with prevalence of inactive TB lesions on chest radiography.

Conclusion/Significance

In a high TB incidence setting, nearly two-thirds of HCWs with latent TB infection had abnormal radiographic findings, and these findings had no clear correlation with T cell immune responses. Further studies are needed to verify these findings and to identify the causes and prognosis of radiologic abnormalities in health care workers.

Are you reading what we are reading? The effect of who interprets medical images on estimates of diagnostic test accuracy in systematic reviews

Observer variation and error in the interpretation of medical images is substantial and has been described as Radiology's Achilles' heel. The enormous development in imaging technologies has brought with it an increase in the complexity and volume of images produced. There is also increased diversity as to who interprets medical images. Whilst the influence of the observer on diagnostic test performance is frequently ignored, there is evidence that this influences estimates of accuracy. Characteristics of observers that should be considered when designing systematic reviews of diagnostic test accuracy are: allocation of images to be read by observers; number, experience and training of observers; profession of observers; and assessment of observer variability and examination of its effect on test accuracy. This information could be used to inform study appraisal, data synthesis, and the investigation of sources of heterogeneity. Establishing the effect of the role of the observer on estimates of accuracy and explaining heterogeneity is important for informing the delivery of these potentially expensive and resource-intensive imaging technologies and the continuing debate about who should read the images.

Inter-observer agreement in interpreting chest X-rays on children with acute lower respiratory tract infections and concurrent wheezing

CONTEXT AND OBJECTIVE

Many children with acute lower respiratory tract infections (ALRI) present to the emergency ward with concurrent wheezing. A chest x-ray is often requested to rule out pneumonia. We assessed inter-observer agreement in interpreting x-rays on such children. DESIGNS AND SETTING: Prospective consecutive case study at Instituto de Salud del Niño, Lima, Peru.

METHODS

Chest x-rays were obtained from eligible children younger than two years old with ALRI and concurrent wheezing who were seen in the emergency ward of a nationwide pediatric referral hospital. The x-rays were read independently by three different pediatric residents who were aware only that the children had a respiratory infection. All the children had received inhaled beta-adrenergic agonists before undergoing chest x-rays. Lobar and complicated pneumonia cases were excluded from the study.

RESULTS

Two hundred x-rays were read. The overall kappa index was 0.2. The highest individual kappa values for specific x-ray findings ranged from 0.26 to 0.34 for rib horizontalization and from 0.14 to 0.31 for alveolar infiltrate. Inter-observer variation was intermediate for alveolar infiltrate (kappa 0.14 to 0.21) and for air bronchogram (kappa 0.13 to 0.23). Reinforcement of the bronchovascular network (kappa 0.10 to 0.16) and air trapping (kappa 0.05 to 0.20) had the lowest agreement.

CONCLUSIONS

There was poor inter-observer agreement for chest x-ray interpretation on children with ALRI and concurrent wheezing seen at the emergency ward. This may preclude reliable diagnosing of pneumonia in settings where residents make management decisions regarding sick children. The effects of training on inter-observer variation need further studies.

Risk factors for developing tuberculosis in remand prisons in St. Petersburg, Russia - a case-control study

BACKGROUND

Detainees have a substantial risk to develop tuberculosis (TB) due to a higher incidence of TB in remand prison compared to the civil community. They develop TB during incarceration not only due to poor living conditions in remand prison, but also due to some factors affecting their life before imprisonment. Prevention measures against TB spread from penitentiary institutions to society include study of factors, which contribute to TB development. Current study aims at identification of important risk factors of TB development in remand prison in St. Petersburg, Russia.

METHODS

A retrospective matched case-control study was performed from May 2002 to May 2003 in two remand prisons in St. Petersburg. One hundred and fourteen prisoners (57 cases, 57 controls) were interviewed by using standardised questionnaire. Logistic regression analysis was performed to identify risk factors.

RESULTS

Six factors were significantly linked to the risk of developing TB: narcotic drug use (odds ratio (OR): 2.6, 95% confidence interval (CI): 1.0-6.9), low income (OR: 3.2, CI: 1.2-8.6), high ratio of prisoners per available bed (OR: 4.0, CI: 1.1-15.0), not having own bed clothes (OR: 13.0, CI: 2.7-61.6), and little time outdoors (OR: 3.3, CI: 1.3-8.5). However, good housing before imprisonment (OR: 4.2, CI: 1.1-15.7) was a separate risk factor for TB.

CONCLUSIONS

Three of the risk factors (high number of prisoners per bed, not having own bed clothes, and little time outdoors) are certainly possible to approach by improvement of conditions in remand prisons. The remaining three factors (narcotic drug use, good housing before imprisonment, and low income) provide knowledge about study population, but cannot be intervened by prison's medical staff.

Reform of tuberculosis control and DOTS within Russian public health systems: an ecological study

OBJECTIVES

To investigate the association between clinical need and hospital bed supply and utilization in Russia; and, to investigate these associations in areas where traditional Russian tuberculosis health care systems exist and where the directly observed therapy-short course (DOTS) strategy has been implemented.

DESIGN

Ecological study using 2002 routine data.

MAIN OUTCOME MEASURES

Hospital bed utilization and hospital admissions for patients with tuberculosis in regions that adhere to the traditional Russian method of managing tuberculosis and those where the DOTS strategy has been implemented.

RESULTS

The ratio of beds per newly notified case was 0.86. The mean duration of hospital stay per admission was 86 days for non-DOTS regions and 90 days for regions where the DOTS strategy had been implemented. The number of admissions in each region correlated closely with the number of newly registered cases and hospital beds were, on average, occupied for 325 days. In the regions where the DOTS strategy had been implemented bed occupancy was 324 days.

CONCLUSIONS

Under the Russian tuberculosis control system, hospital utilization is predominantly determined by supply-side factors, namely the number of tuberculosis dedicated hospital beds, and this system extends across all regions. Implementation of the DOTS strategy in Russia has not led to fundamental structural changes in tuberculosis control systems.

The Directly Observed Therapy Short-Course (DOTS) strategy in Samara Oblast, Russian Federation

BACKGROUND

The World Health Organisation (WHO) defines Russia as one of the 22 highest-burden countries for tuberculosis (TB). The WHO Directly Observed Treatment Short Course (DOTS) strategy employing a standardised treatment for 6 months produces the highest cure rates for drug sensitive TB. The Russian TB service traditionally employed individualised treatment. The purpose of this study was to implement a DOTS programme in the civilian and prison sectors of Samara Region of Russia, describe the clinical features and outcomes of recruited patients, determine the proportion of individuals in the cohorts who were infected with drug resistant TB, the degree to which resistance was attributed to the Beijing TB strain family and establish risk factors for drug resistance.

METHODS

Prospective study.

RESULTS

2,099 patients were recruited overall. Treatment outcomes were analysed for patients recruited up to the third quarter of 2003 (n = 920). 75.3% of patients were successfully treated. Unsuccessful outcomes occurred in 7.3% of cases; 3.6% of patients died during treatment, with a significantly higher proportion of smear-positive cases dying compared to smear-negative cases. 14.0% were lost and transferred out. A high proportion of new cases (948 sequential culture-proven TB cases) had tuberculosis that was resistant to first-line drugs; (24.9% isoniazid resistant; 20.3% rifampicin resistant; 17.3% multidrug resistant tuberculosis). Molecular epidemiological analysis demonstrated that half of all isolated strains (50.7%; 375/740) belonged to the Beijing family. Drug resistance including MDR TB was strongly associated with infection with the Beijing strain (for MDR TB, 35.2% in Beijing strains versus 9.5% in non-Beijing strains, OR-5.2. Risk factors for multidrug resistant tuberculosis were: being a prisoner (OR 4.4), having a relapse of tuberculosis (OR 3.5), being infected with a Beijing family TB strain (OR 6.5) and having an unsuccessful outcome from treatment (OR 5.0).

CONCLUSION

The implementation of DOTS in Samara, Russia, was feasible and successful. Drug resistant tuberculosis rates in new cases were high and challenge successful outcomes from a conventional DOTS programme alone.