BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from
BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.
When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.
The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.
BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com).
If you have any questions on BMJ Open’s open peer review process please email [email protected] http://bmjopen.bmj.com/ on September 27, 2021 by guest. Protected copyright. BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from
Validation of handheld fundus camera with mydriasis for retinal imaging to increase uptake of diabetic retinopathy screening in China ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2020-040196
Article Type: Original research
Date Submitted by the 07-May-2020 Author:
Complete List of Authors: Liao, Qinghua; Zhongshan Ophthalmic Centre, Sun Yat-sen University Xiao, Baixiang; Zhongshan Ophthalmic Centre, Sun Yat-sen University, The State Key Laboratory of Ophthalmology; Nanchang University Affiliated Eye Hospital, Li, Yanping; Nanchang University affiliated Eye Hospital Weng, Fan; Zhongshan Ophthalmic Centre, Sun Yat-sen University Jin, Ling; Sun Yat-Sen University Zhongshan Ophthalmic Center Wang, Yanfang; Zhongshan Ophthalmic Centre, Sun Yat-sen University Huang, Wenyong; Sun Yat-Sen University Zhongshan Ophthalmic Center Yi, Jinglin; Nanchang University Affiliated Eye Hospital Burton, Matthew J; London School of Hygiene and Tropical Medicine Yip, Jennifer; London School of Hygiene & Tropical Medicine http://bmjopen.bmj.com/ Diabetic retinopathy < DIABETES & ENDOCRINOLOGY, Diabetic Keywords: nephropathy & vascular disease < DIABETES & ENDOCRINOLOGY, OPHTHALMOLOGY
on September 27, 2021 by guest. Protected copyright.
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 23 BMJ Open
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 5 6 7 8 9 I, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined 10 in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors 11 who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance 12 with the terms applicable for US Federal Government officers or employees acting as part of their official 13 duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd (“BMJ”) its 14 licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the 15 Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence. 16 17 The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to 18 the Submitting Author Forunless you peer are acting as review an employee on behalf only of your employer or a postgraduate 19 student of an affiliated institution which is paying any applicable article publishing charge (“APC”) for Open 20 Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and 21 intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative 22 Commons licence – details of these licences and which Creative Commons licence will apply to this Work are set 23 out in our licence referred to above. 24 25 Other than as permitted in any relevant BMJ Author’s Self Archiving Policies, I confirm this Work has not been 26 accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate 27 material already published. I confirm all authors consent to publication of this Work and authorise the granting 28 of this licence. 29 30 31 32 33 34 35 36
37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44
45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 23
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 Validation of handheld fundus camera with mydriasis for 5 6 7 retinal imaging to increase uptake of diabetic retinopathy 8 9 screening in China 10 11 12 1Qinghua Liao MS, 2, 3Baixiang Xiao MD, 3Yanping Li PhD, 4Fan Weng MS, 2Ling Jin MS,
13 2 2 3 6 6 14 Yanfang Wang MB, Wengyong Huang PhD, Jinglin Yi MD, Matthew Burton PhD, Jennifer 15 L Yip PhD 16 17 18 For peer review only 19 1. 20 Zhenjiang District Hospital, Shaoguan City, China 21 2. The State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Centre, Sun Yat-sen 22 University, Guangzhou, China. 23 3. Nanchang University Affiliated Eye Hospital, Nanchang City, China 24 25 4. Yuexiu District Centre of Disease Control, Guangzhou, China 26 5. International Centre for Eye Health, London School of Hygiene & Tropical Medicine, Keppel 27 Street, London WC1E 7HT, United Kingdom 28 29 30 Address for correspondence: Dr. Baixiang Xiao, The State Key Laboratory of Ophthalmology, 31 32 Zhongshan Ophthalmic Centre, Sun Yat-sen University, #54 South Xianlie Road, Guangzhou 33 City, Guangdong Province, China 510060 Tel/Fax: +86 20 8733 4645, Email: 34 35 [email protected] 36
37 Word count: 3993 http://bmjopen.bmj.com/ 38 39 Running head: Validation of handheld fundus camera 40 41 42 Keywords: validation, fundus camera, sensitivity, specificity, diabetic retinopathy, 43 44
45 on September 27, 2021 by guest. Protected copyright. 46 The conflicts of interest: The authors declare that they have no conflicts of interest. 47 48 49 Funding: The study is fully funded by the State Key Laboratory of Ophthalmology, Zhongshan 50 Ophthalmic Centre, Sun Yat-sen University 51 52 Data sharing: Full data is available on the request from the corresponding author. 53 54 55 56 57 58 59 60 1
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 23 BMJ Open
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from Contributorship statement: 4 5 6 Q Liao: preliminary draft of the tables and manuscript, data analysis and critical revision of the 7 manuscript. 8 9 10 B Xiao: study design, training of the study teams, monitoring data collection and quality 11 control, data cleaning, analysis as well as manuscript drafting and revision. 12 13 Y Li: Monitoring of grading and valuable contribution to the revision of the manuscript. 14 15 16 F Weng: data collection and valuable contribution to the study design. 17 18 L Jin: statistical analysisFor and peer helpful revision review of the manuscript. only 19 20 Y Wang: data cleaning and valuable contribution to the manuscript drafting. 21 22 23 J Yi, W Huang: valuable contribution to the study design and revision of the manuscript. 24 25 J Yip, M Burton: Substantial contributions to the conception, designing, and revision of the 26 manuscript as well as analysis and interpretation of the data. 27 28 29 30 31 32 33 34 35 36
37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44
45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 2
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 23
1 2
3 Abstract: BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 5 6 Objectives: To investigate the clinical validity of using a handheld fundus camera to detect 7 diabetic retinopathy (DR) in China. 8 9 10 Design and settings: Prospective comparison of the handheld fundus camera with the 11 standard validated instrument in detection of DR in hospital and a community screening clinic 12 13 in Guangdong Province, China. 14 15 Participants: Participants aged 18 and over with diabetes who were able to provide informed 16 17 consent and agreed to attend the dilated eye examination with handheld tests and gold 18 standard tests. For peer review only 19 20 21 Primary and secondary outcome measures: Primary outcome was the proportion of those 22 with referable DR (R2 and above) identified by the handheld fundus camera (the index test) 23 compared with the standard camera. Secondary outcome was the comparison of proportion of 24 25 gradable images obtained from each test. 26 27 Results: 28 29 30 In this study, we examined 304 people (608 eyes) with each of the two cameras under 31 mydriasis. The handheld camera detected 119 eyes (19.5%) with some level of DR, 81(13.3%) 32 33 of them were referable, while the standard camera detected 132 eyes (21.7%) with some level 34 of DR and 83(13.7%) were referable. It seems that he standard camera found more eyes with 35 36 referable DR, while McNemar’s test detected no significant difference between the two 37 cameras. http://bmjopen.bmj.com/ 38 39 Of the 608 eyes with images obtained by desktop camera, 598 (98.4%) images were of 40 41 sufficient quality for grading, 12 (1.9%) images were not gradable. By the handheld camera, 42 590 (97.0%) were gradable and 20 (3.2%) images were not gradable. 43 44
45 The two cameras reached high agreement on diagnosis of retinopathy and maculopathy at all on September 27, 2021 by guest. Protected copyright. 46 the levels of retinopathy. 47 48 49 Conclusion: 50 Although it could not take the place of standard desk top camera on clinic fundus examination, 51 the handheld fundus camera showed promising role on preliminary DR screening at primary 52 level in China. To ensure quality images, mydriasis is required. 53 54 55 56 57 58 59 60 3
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 23 BMJ Open
1 2
3 Article Summary BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 5 6 Strengths and limitations of this study: 7 8 1. Study subjects recruited for this study where from both outpatient department of hospital 9 10 and community screening with the variety of diabetic retinopathy for the study comparison. 11 2. Two experienced graders from tertiary eye institute graded the images captured by both 12 cameras separately and reached full agreement on grading. 13 14 3. To avoid bias from personnel, technicians were experienced with the standard camera and 15 received full training in using the handheld camera. 16 4. We noticed the high rate of images with poor quality without dilation at the pilot study and 17 18 started to give mydriasisFor for peer all the participants review at the formal only study, so that there were lack of 19 data on non-mydriasis. 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44
45 on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 23
1 2
3 Introduction BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 5 Diabetes mellitus (DM) is a major cause of morbidity and mortality worldwide, responsible for 6 1.5 million deaths in 2012. 1The ageing population, rising levels of obesity and lifestyle 7 8 changes will increase the diabetes burden of disease further. In China, the prevalence of 9 diabetes increased dramatically in the last 30 years, with recent prevalence estimates in adults 10 2 11 of 10.9%. 12 13 Diabetic retinopathy (DR) is a common microvascular complication causing retinal 14 haemorrhage and oedema in people with diabetes mellitus (PwDM). This reduces visual acuity 15 16 (VA) only at the late stage or the time affects macular when the treatment is not optimal 17 without earlier detection on retina. Worldwide, it is one of the most common cause of visual 18 For peer review only 19 impairment, particularly in working age adults, with significant economic impact.3 In China, a 20 systematic review showed that approximately 1 in 5 people diagnosed with diabetes have 21 22 some level of DR, which is similar to that from findings in high income countries. 4, 5 23 24 There are effective strategies to prevent and treat DR. The risk of vision loss can be reduced 25 26 with stable metabolic control, early detection through screening, and appropriate laser, 27 vitreous injection or vitrectomy. 28 29 30 Diabetic retinopathy screening (DRS) and appropriate referral for treatment of retinopathy 31 have been shown to reduce blindness from DR. 6 However, studies have shown that in rural 32 4 33 areas of China, only 10% of those with diabetic eye disease are diagnosed and treated. This 34 indicates a need to ensure accessible DRS services to reduce potential eye health 35 36 inequalities.
37 http://bmjopen.bmj.com/ 38 The primary health care providers in China encourage PwDM to register and receive an 39 annual health check in the rural township clinic and urban community health centre to monitor 40 41 patients’ glucose level and detect potential complications. Primary care doctors are not trained 42 in eye care, and PwDM are asked to travel to county hospitals for diabetic eye examinations at 43 44 their own expense, which results in inequitable access.
45 on September 27, 2021 by guest. Protected copyright. 46 Fundus cameras have now been installed rapidly in most of the county level hospitals since 47 48 the National study on eye service in 2014 found that only 10% of the secondary level hospitals 49 had the capacity to take fundus images (data not been published). Outreach services are 50 51 limited due to the lack of specialist eye-care staff in secondary care, which also leads to 52 variation in screening uptake and consequently treatment of diabetic eye disease. 53 54 55 In a clinical review of diabetic patients in Guangzhou, overall, 43.2% of patients in tertiary and 56 community urban settings, had never received an eye examination. In rural clinics, 68.7% 57 7 58 have never received an eye examination. 59 60 5
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 23 BMJ Open
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from Provision of screening at primary care level can increase uptake8-10, however, the cost of 4 5 providing all primary care clinics in China with a digital camera would be prohibitive, together 6 with implications of additional training of staff and maintenance of equipment. DR services are 7 8 in their infancy in China. 9 10 The Zhongshan Ophthalmic Centre, Sun Yat-sen University (ZOC) is a leading eye institution, 11 12 and well placed to plan and develop a DR service model in China. Currently, there are no 13 sustainable and scalable models for delivery of DR services in rural China. 14 15 16 The aim of this study is to validate retinal images from a handheld portable retinal camera for 17 DRS, using a desktop digital camera as the gold standard. Findings from this study will lead to 18 For peer review only further investigating the role of handheld camera for retinal images in facilitating access and 19 20 increasing uptake of DRS in primary care clinics in China. 21 22 Methods 23 24 25 Ethical approval of this study was obtained from both London School of Hygiene & Tropical 26 Medicine and Zhongshan Ophthalmic Centre, Sun Yat-sen University. All participants provided 27 28 written, informed consent. Patients with referable diabetic or other eye diseases were referred 29 to ZOC for further examination or treatment. The study fulfilled the tenets of the Declaration of 30 31 Helsinki. 32 33 This was a prospective comparison study of the hand-held fundus camera (index test) of 34 Horus Scope DEC 20011 with the desktop digital camera (gold standard test) Canon (model 35 36 CR-2) in detection of referable DR.
37 http://bmjopen.bmj.com/ 38 Population: 39 40 41 Eligible participants were all those aged 18 and over with diabetes who were able to provide 42 informed consent and agreed to attend for dilated eye examination with both index and gold 43 44 standard tests. Diabetes was identified by self-report as well as definite medical records in
45 hand, referral by endocrinologist, or registration in the primary health care centres, on September 27, 2021 by guest. Protected copyright. 46 47 Participants were recruited from Zhenjiang District (Shaoguan Prefecture) hospitals, Chenghai 48 49 District Hospital (Shaotou City) and Yuexiu District community (Guangzhou City) health 50 centres in Guangdong Province, China, with a range of DR severity, including patients without 51 52 DR in order to obtain a representative spectrum of patients in this study. 53 54 Sample size 55 56 57 A sample size of 262 patients has 80% power and 5% significance level to detect a 6% 58 difference in proportion of gradable images between the index and the standard test, where 59 60 the gold standard will produce 90% gradable images. 6
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 23
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from Training 4 5 6 The pilot study was conducted in the community screening clinic in Guangzhou City. 7 Experienced technicians for the handheld camera and desk top camera examined thirty cases 8 9 with both un-dilated and dilated pupils. 10 11 In the other study sites, we trained experienced technicians to capture images on the use of 12 13 the handheld camera as well as the standard camera to ensure a standardized process. They 14 each practiced on the handheld camera for 20 pilot cases, closely supervised by experienced 15 16 technician until there were no more questions on using the camera and quality of the images 17 taken were considered acceptable by the senior ophthalmologist. 18 For peer review only 19 Data collection 20 21 22 A data collection form was developed including information on the patient’s age, gender, 23 education, profession, visual acuity (by illuminated Snellen Visual Chart), use of glasses, 24 25 self-assessed visual function, history of eye examinations and fundus photocoagulation, 26 history of diabetes mellitus (DM), complications, treatment of DM and hypertension (HP), 27 28 fasting glucose on the day of fundus photo taken, which camera the participants preferred and 29 why. Treatment options for DM and HP were given in multiple choices, each with four options 30 31 as for DM: insulin, oral medicine, diet, Chinese Traditional Medicine (CTM) and no treatment, 32 for HP as: oral medicine, diet, CTM and no treatment. 33 34 Pilot study 35 36
37 Without dilation, from both cameras, approximately one-third of the images from the first eye http://bmjopen.bmj.com/ 38 (always right eye) were of poor quality (including un-gradable and poor but still gradable), and 39 40 10% were un-gradable. We asked participants to rest in a darkened room for two minutes after 41 photographs were taken from the first eye, and up to fifteen minutes for those with poor quality 42 images, to facilitate pupil dilation, but there was a higher proportion of poor quality pictures 43 from the second eye (40%) with both cameras (Data not presented). 44
45 on September 27, 2021 by guest. Protected copyright. 46 Due to the high rate of poor images and patients’ dissatisfaction from waiting in the dark room 47 for second camera and sometimes even for the second eye, we decided to give mydiasis to 48 49 every participant for image taken in the formal study. 50 51 The formal study 52 53 All participants had the following tests in sequence during one visit in the clinic within 2 hours 54 of the first photograph to ensure photographs from both cameras were obtained under 55 56 maximal dilation: participant’s basic information, visual acuity test, anterior segment 57 examination under slit lamp by an ophthalmologist, intraocular pressure (IOP) test, dilation of 58 59 both eyes, fundus photograph of both eyes by the handheld retinal camera and the desk top 60 digital camera and finally, the questions on preference of and comments on the cameras. We 7
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 23 BMJ Open
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from randomly assigned patients to a different sequence of cameras, with sufficient time in between 4 5 to minimize discomfort. 6 7 Two photos were taken for each eye of all the participants by each of the two cameras. One 8 9 centreed on the macula and the other on the papilla optica. One technician operated on both 10 the handheld and standard cameras for participants in each of the three study sites. The 11 12 operators of the cameras were asked the advantages and disadvantages of both cameras, the 13 ease of use, graders were asked the acceptability of mode of photography. 14 15 16 The fundus photographs from all tests were uploaded on the DR online grading system and 17 graded independently at the grading centre in ZOC where the experienced graders were 18 For peer review only masked to the mode of photograph where possible. 19 20 21 The two graders graded all the study images separately and then together to compare whether 22 there were disagreement on the grading results. They discussed any disagreement until they 23 reached consensus. They had been trained appropriately as the graders in ZOC’s grading 24 25 centre, which serves for over 70 secondary hospitals across the country in programs. They 26 both had been working in this centre for at least 5 years and constantly monitored by 27 ophthalmologist supervisors on quality of their work. 28 29 30 31 Image quality 32 33 Images were considered good quality when features were focused, well illuminated retinal 34 35 field, showing clarity of the fundus vessels and any retinopathy. If the images were only 36 partially focused, illuminated or retinal field showed, they were defined as poor, but still 37 http://bmjopen.bmj.com/ 38 gradable. If any retinopathy was detected, either DR or non-DR, the images were also defined 39 as gradable. If images are blurred without recognition of the retinal vessels or retinopathy 40 41 features, they are defined as not gradable. The patients with not gradable images were then 42 referred to the ophthalmologist for further examination. 43 44
45 Grading system: on September 27, 2021 by guest. Protected copyright. 46 47 To define the fundus pathology and grade the retinal changes, we used grading definitions for 48 referable disease by the British NHS Diabetic Eye Screening Program.12 That is R0 is 49 50 categorised as the absence of any DR feature including microaneurysms. Microaneurysm with 51 or without exudation is categorised as R1 for its only presence without other DR features. 52 53 54 The eye is categorised as R2 if any of the following features are present: venous beading, 55 cotton wool, venous reduplication, multiple blot haemorrhages and intra-retinal microvascular 56 57 abnormality. 58 59 60 8
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 23
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from If there is presenting proliferative retinopathy, that is new blood vessels or haemorrhage within 4 5 retina or in vitreous, or vitreous tract, the eye is categorised as R3a. If there is evidence of 6 retinal laser treatment and stable retina, the eye is categorised as R3s. Presence of any 7 8 diabetic retinopathy features within 2 disc diameters of the centre of the fovea is categorised 9 as M1. 10 11 12 Statistics Analysis: 13 14 Data was presented as mean (standardized deviation, SD) or median (inter quartile range, 15 16 IQR) for continuous variable and frequency (%) for categorical variable. Participants’ age was 17 categorized by interval of 10 years. Age at diagnosis of diabetes and hypertension was 18 For peer review only categorized by interval of 5 years. The cut-offs of 0.05 and 0.3 were used to describe visual 19 20 acuity in better-seeing eye. Data were analysed by eyes. McNemar's test was performed for 21 comparing the standard and portable cameras. The inter-rater reliability between two cameras 22 23 was measured by Cohen's kappa coefficient (95% confidence interval, CI). Sensitivity, 24 specificity, positive predictive value and area under the Receiver-operator curve with 95% CI 25 26 were calculated to indicate the accuracy of diagnosis by two cameras. All statistical analyses 27 were performed using a commercially available software package (Stata 13.1, StataCorp, 28 29 College Station TX, USA). 30 31 Comments on the cameras: 32 33 34 Comments from both staff and the patients were simply listed and similar ones were 35 categorised together until there were no more new comments and repeated ones were 36
37 included together as one comment. http://bmjopen.bmj.com/ 38 39 Results 40 41 Participants’ characteristics 42 Patient recruitment started in June 2018 simultaneously in three places and ended up a total of 43 44 305 diabetes people examined by the end of the year. Mean age of the participants was 61.3
45 (SD ±10.1) and almost half (41.6%) were between 61 and 70 years old. Among them, 165 on September 27, 2021 by guest. Protected copyright. 46 (54.1%) were female (Table 1). Over one-third received high school and above education. 47 48 Over half (53.6%) of the participants were retired and approximately 10% unemployed. Mean 49 age at diagnosis of diabetes was 52.4 (SD±10.5). Among these participants, around one-third 50 had diabetes for less than 5 years, one-third for 6 to 10 years and the rest had over 10 years. 51 52 Only 12 (3.9%) people had diabetes for over 20 years. The median duration of diabetes was 5 53 years (4-12). 54 55 56 On study days, we detected only 108 (35.4%) participants with fasting glucose below 7mmol/L 57 (Table 1), which is clinically considered as well controlled. Insulin was used by 112 (36.7%) 58 59 people and 262 (85.9%) were taking oral medication, 12 (4%) reported no treatment, not even 60 diet. 9
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 23 BMJ Open
1 2
3 BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from Hypertension was detected in 48.2% participants, nephropathy in 7.54%, cardiovascular 4 5 disease in 11.2% and ulcerated arms/legs was in 2.62%. Of these 305 people with DM, 252 6 (82.6%) did not have any diabetic complications. Mean age at diagnosis of hypertension was 7 8 54.4 (SD±11.1). Of the 147 participants with hypertension, eight could not remember when the 9 problem started or being diagnosed, 59 (40.1%) people were diagnosed less than 5 years and 10 11 10 (6.8%) over 20 years ago (Table 1). Majority (85.0%) of these people with HP were taking 12 oral medication and 16 (10.9%) received no treatment. 13 14 15 Of the 305 participants, 276 (90.5%) had presenting visual acuity (PVA) equal to or over 0.3, 3 16 (0.98%) below 0.05, while by self-assessment, only 11 (3.61%) expressed that they had 17 18 excellent VA, 81 For(26.6%) said peer their VA was review good and 81 (26.6%) only felt their VA was poor (Table 19 2). Over half (59.7%) of the participants did not have their eyes examined by medical staff in 20 21 the previous year. 21 (6.89%) people had received laser photocoagulation. 22 23 Agreement by the two graders 24 25 The two graders disagreed on only two images for R1 and reached to consent after 26 27 discussion. 28 29 Image gradeability, referable eyes and agreement analysis 30 31 32 Of the 305 people (610 eyes) examined, 1 patient (2 eyes) had the images duplicated with the 33 previous patient owing to the wrong saving name in the computer, which left 608 eyes for 34 35 grading. From desktop camera, 482 (79.3%) images were of good quality and gradable, 116 36 (19.1%) were poor but still gradable, and 12 (1.9%) images were not gradable (Table 3). From
37 http://bmjopen.bmj.com/ 38 the handheld camera, 479 (78.7%) were of good quality and gradable, 111 (18.2%) were of 39 poor quality, but still gradable and 20 (3.2%) images were not gradable. Between the two 40 41 groups of good and poor quality but still gradable images by the two cameras, McNemar’s test 42 found no significant difference, while the two groups added together, images taken by desktop 43 gained slightly better quality than handheld although the difference was not significant (p 44
45 >0.05%). on September 27, 2021 by guest. Protected copyright. 46 47 48 49 50 For the non-gradable eyes, the two cameras agreed on 5 eyes, two of which had vitreous 51 opacity, the other three eyes had dense cataract. There was one eye captured R1 by handheld 52 camera, but the image captured by desk top was non-gradable as handheld camera happened 53 to capture some peripheral microaneurysms although images taken by both cameras were 54 55 poor from this cataractous eye. 56 57 There were cataracts in 8 eyes and ascertained as R1 by desktop camera, but not gradable by 58 59 the handheld camera. 60 10
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 23
1 2
3 There were 14 eyes graded as R1 by the desk top camera, but ungradeable by the handheld BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 5 camera. 6 7 8 In total, there were 132 eyes (21.7%) with evidence of retinopathy (R1+R2+R3) (table 4), from 9 10 images taken by desk top camera and 119 eyes (19.5%) by handheld camera. We detected 11 referable retinopathy at R2 and above in 83 eyes (13.7%) from the desktop camera and 81 12 eyes (13.3%) from the handheld camera, with no evidence of a statistically significant 13 difference. R3 was detected in 28 eyes (4.6%) by both handheld and desktop cameras. Of the 14 15 132 eyes with some level of retinopathy, 79 (59.8%) eyes had macular involvement by 16 standard camera. 17 18 For peer review only 19 The two cameras reached high agreement on diagnosis of retinopathy and maculopathy at the 20 levels of R1 (Kappa coefficient - KC was 0.79), R2 (KC=0.96), R3 (KC=1.0), M1 (KC=0.94) 21 and other lesion (KC=0.82) (Table 5). 22 23 24 We identified 11 more eyes (49 eyes in total) at R1 from imaged obtained by the desk top 25 camera compared to the handheld camera (38 in total), which resulted in a sensitivity of 71.4% 26 (95% CI: 56.7 – 83.4), specificity of 99.4% (95% CI: 98.4 – 99.9) and positive predictive value 27 28 (PPV) of 92.1% (95%CI: 78.6 – 98.3). 29 30 At the level of referable retinopathy at R2, the desk top camera detected 55 eyes and 31 32 handheld detected 53 eyes, resulting in a sensitivity of 94.6% (95% CI: 84.0 – 98.9), specificity 33 of 99.8% (95% CI: 99 – 100), PPV to 98.1% (95% CI: 89.9 -100). While for R3, the two 34 cameras reached 100% agreement. The desk top camera captured 6 more eyes with 35 36 maculopathy while the handheld one did not detect maculopathy in these eyes and a
37 sensitivity of 91.1% (95% CI: 82.6 – 96.4), specificity of 99.6% (95% CI: 98.9 – 100), PPV of http://bmjopen.bmj.com/ 38 98.6% (95% CI: 92.6 – 100). The desk top camera also captured two more eyes with other 39 lesions, which the handheld camera had not. 40 41 42 Camera preference by both patients and technicians: 43 44
45 Approximately half (51.2%) of the participants had no preference for either of the two cameras, on September 27, 2021 by guest. Protected copyright. 46 114 (37.4%) people preferred the desktop camera and 34 (11.1%) said the handheld one was 47 preferred (Table 6). Those preferring the standard camera gave reasons as follows: flash light 48 49 for images taken by the standard camera was not so bright as the handheld one; it looked 50 more complicated and was bigger, so that should be better, as recognized by most of 51 non-medical people of the medical equipment; it was more convenient for height adjusting as it 52 was on an elevator-platform. Those participants who felt handheld was better gave reasons of 53 54 simplicity, looking smart with easy mobilization of the whole machine with or without the 55 foldable stand (frame) for patient’s chin rest. 56 57 58 Comments from the three experienced technicians on the two cameras were: “for the 59 community DR screening, the handheld one is enough”. “It is easy to install and pack up”. 60 11
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 23 BMJ Open
1 2
3 They also pointed out that, with the simple stand for patient’s chin rest, focusing process BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 5 became much easier and quicker than when without. 6 7 Discussion: 8 We compared the quality of fundus images and the referable eyes with images taken by the 9 10 standard Canon CR2 desk top and a simple Forus handheld fundus cameras from the same 11 305 PwDM in three places of Guangdong Province, China and our analysis found no 12 difference on proportion of gradable images, good images and number of eyes with referable 13 14 retinopathy between the two cameras when mydriasis was used. 15 16 Training for those without experience was easy and relatively quick. The handheld camera is 17 18 simply designed, Foreasy to install peer and to pack review up. The light and only foldable stand designed for the 19 handheld camera makes it possible to be carried by hand. Its images cover the same retinal 20 field as the standard one. 21 22 23 In the English national screening program for DR, Scanlon’s study demonstrated that two 24 images centred on the disc and macular with 450 field camera achieved high sensitivities 25 (>87%) and specificities (>86%) with low un-gradable image rate of below 4.4% against the 26 27 gold standard of seven-field stereophotography or an ophthalmologist using slit lamp 28 biomicroscope.13, 14 The two field images method was recommended for DRS and was used 29 as the gold standard in this study. 30 31 32 Overall, there was no significant difference between using the handheld camera with dilation 33 on detecting number of referable eyes with DR compared with the standard camera. The two 34 cameras reached a high level of agreement on grading results of DR (Kappa from 0.79 at R1 35 36 to 1.00 at R3). When authors looked at the images disagreed against findings from slit lamp,
37 we noticed the following factors: For those eyes with cataract at the stage C3N3 to C4N4 15 http://bmjopen.bmj.com/ 38 and those with some level of vitreous opacity, the desk top could still capture images of blood 39 40 vessels with some degree of clarity compared to the handheld camera. 41 42 In China, primary health staff commonly hesitate to give mydriasis for fundus examinations 43 16 44 and are normally not confident to convince patients about the low risk of complications. In
45 this study, the poor image rate reduced generally from over 30% to around 10%. This in the on September 27, 2021 by guest. Protected copyright. 46 meantime reduced the number of people referred to secondary care for further examination 47 and saved resources. From the findings in this study, we would recommend mydriasis for DRS 48 49 at primary level with training for primary health staff to obtain patient consent and manage 50 potential complications in future programs. 51 52 53 In this study, around one-third - 108 (35.4%) of participants had their fasting glucose controlled 54 at the recommended levels. Furthermore, 12 participants did not take any treatment for their 55 56 DM, not even diet. Over half (59.7%) of the participants had not had their eyes checked by 57 medical staff in the previous year. This emphases the importance of health education and 58 59 available services for DM patients in primary health care. 60 12
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 23
1 2
3 Approximately two-thirds of study participants were the walk-in patients with diabetes medical BMJ Open: first published as 10.1136/bmjopen-2020-040196 on 29 October 2020. Downloaded from 4 5 records in hand to the clinics of secondary level hospitals and one-third were community 6 health unit registry, from them, we detected a similar lower rate of DR as other studies in China 7 5, 17, 18compared to Singapore , India and United States,19 while much lower than Handan Eye 8 Study4 in North China and the more likely findings from other walk-in patients in six provinces 9 10 in China20. The study also detected 82.6% of the participants without any diabetic 11 complications, which is mainly attributed to the median duration of diabetes was 5 years (4-12) 12 and only 3.93% of participants had diabetes for over 20 years. 13 14 15 We detected a similar proportion of referable DR, high proportion of poor quality images at 16 unmydriasis to the previous studies in China.21 Findings for the comparison of the two 17 22 23 18 cameras are similarFor to the studiespeer in other review Asia countries like only Sri Lanka and Thailand . 19 20 This study employed experienced eye care staff, who had been performing similar work for at 21 least three years, to take the images with both desk top and the handheld cameras. There 22 23 were also standardized procedures and training for all staff in the different sites, which would 24 reduce measurement error. Graders were masked from patients’ history and source of images, 25 reducing ascertainment bias. Where the two cameras differed in their images we carried out a 26 27 slit lamp examination to determine the cause of the disagreement. 28 29 This was not a population based study and therefore patient characteristics were not 30 31 representative of DR with respect to severity of the diseases and its complications in the 32 context of China, however, we included the full range of DR severity in our study population, 33 mitigating spectrum bias. 34 35 36 Conclusion:
37 A handheld fundus camera using mydriasis may have a role to play in preliminary DR http://bmjopen.bmj.com/ 38 screening at the primary level in China. 39 40 41 Acknowledgement 42 The authors acknowledge the doctors, technicians and nurses in Shaoguan Railway Hospital, 43 44 Chenghai District Hospital and Community Health Centres in Yuexiu District, Guangzhou City
45 for their contributions to patient coordination during the data collection for this study. We also on September 27, 2021 by guest. Protected copyright. 46 appreciate Dr. James La Nauze from Australia for his contribution to English correction and 47 revision of the manuscript. 48 49 50 —————————————— 51 52 53 54 55 56 57 58 59 60 13
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 23 BMJ Open
1 2 3 References: 4 5 6 1. The World Health Organisation, Global report on diabetes (PDF, 2016), Available at: 7 https://www.who.int/publications-detail/global-report-on-diabetes. Accessed on 6th Jan 8 9 2020. 10 2. Yang SH, Dou KF, Song WJ. Prevalence of diabetes among men and women in China. N 11 Engl J Med 2010; 362(25): 2425-6. 12 3. Jonsson B. The economic Impact of Diabetes. Diabetes Care 1998; 21 (Syookenebt 13 14 3):C7-C10. 15 4. Wang FH, Liang YB, Zhang F, Wang JJ, Wei WB, Tao QS, Sun LP, Friedman DS, Wang NL, 16 Wong TY. Prevalence of diabetic retinopathy in rural China: the Handan Eye Study. 17 18 Ophthalmology 2009;For 116 (3):peer 461-7. review only 19 5. Liu L, Wu X, Liu L, Geng J, Yuan Z, Shan Z, Chen L. Prevalence of diabetic retinopathy in 20 mainland China: a meta-analysis. PloS one 2012; 7(9): e45264. 21 22 6. Liew G, Michaelides M, Bunce C. A comparison of the causes of blindness certifications 23 in England and Wales in working age adults (16-64 years), 1999-2000 with 2009-2010. BMJ 24 open 2014; 4(2): e004015. 25 7. Wang D, Ding X, He M, Yan L, Kuang J, Geng Q, Congdon N. Use of eye care services 26 27 among diabetic patients in urban and rural China. Ophthalmology 2010; 117(9): 1755-62. 28 8. Moreton RBR, Stratton IM, Chave SJ, Lipinski H, Scanlon PH. Factors determining uptake 29 of diabetic retinopathy screening in Oxfordshire. Diabet Med 2017; 34(7): 993-9. 30 31 9. Ng JQ, Morlet N. Improving the uptake of screening for diabetic retinopathy. Med J Aust 32 2013; 198(2): 69-70. 33 10. Leese GP, Boyle P, Feng Z, Emslie-Smith A, Ellis JD. Screening uptake in a 34 35 well-established diabetic retinopathy screening program: the role of geographical access and 36 deprivation. Diabetes care 2008; 31(11): 2131-5. 37 11. The website for introduction of the handheld camera Horus:www.miis.com.tw, 38 retrieved December 2019. 39 40 12. NHS Diabetic eye screening program, 41 https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_ 42 data/file/582710/Grading_definitions_for_referrable_disease_2017_new_110117.pdf 43 44 retrieved on 18th Oct 2019. 45 13. Scanlon PH, Malhotra R, Greenwood RH, Aldington SJ, Foy C, Flatman M, Downes S. 46 Comparison of two reference standards in validating two field mydriatic digital photography 47 48 as a method of screening for diabetic retinopathy. Br J Ophthalmol 2003; 87(10): 1258-63. 49 14. Scanlon PH, Malhotra R, Thomas G, Foy C, Kirkpatrick JN, Lewis-Barned N, Harney B, 50 Aldington SJ. The effectiveness of screening for diabetic retinopathy by digital imaging 51 photography and technician ophthalmoscopy. Diabetic medicine : a journal of the British 52 53 Diabetic Association 2003; 20(6): 467-74. 54 15. Karbassi M, Khu PM, Singer DM, Chylack LT, Jr. Evaluation of lens opacities classification 55 system III applied at the slitlamp. Optom Vis Sci 1993; 70(11): 923-8. 56 57 16. Yan X, Liu T, Gruber L, He M, Congdon N. Attitudes of physicians, patients, and village 58 health workers toward glaucoma and diabetic retinopathy in rural China: a focus group 59 study. Arch Ophthalmol 2012; 130(6): 761-70. 60 14
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 23
1 2 3 17. Cui Y, Zhang M, Zhang L, Zhang L, Kuang J, Zhang G, Liu Q, Guo H, Meng Q. Prevalence 4 5 and risk factors for diabetic retinopathy in a cross-sectional population-based study from 6 rural southern China: Dongguan Eye Study. BMJ open 2019; 9(9): e023586. 7 18. Pan CW, Wang S, Qian DJ, Xu C, Song E. Prevalence, Awareness, and Risk Factors of 8 9 Diabetic Retinopathy among Adults with Known Type 2 Diabetes Mellitus in an Urban 10 Community in China. Ophthalmic Epidemiol 2017; 24(3): 188-94. 11 19. Nangia V, Jonas JB, George R, Lingam V, Ellwein L, Cicinelli MV, Das A, Flaxman SR, 12 Keeffe JE, Kempen JH, Leasher J, Limburg H, Naidoo K, Pesudovs K, Resnikoff S, Silvester AJ. 13 14 Prevalence and causes of blindness and vision impairment: magnitude, temporal trends and 15 projections in South and Central Asia. Br J Ophthalmol 2019; 103(7): 871-7. 16 20. Liu Y, Song Y, Tao L, Qiu W, Lv H, Jiang X, Zhang M, Li X. Prevalence of diabetic 17 18 retinopathy amongFor 13473 patientspeer with reviewdiabetes mellitus inonly China: a cross-sectional 19 epidemiological survey in six provinces. BMJ open 2017; 7(1): e013199. 20 21. Ding J, Zou Y, Liu N, Jiang L, Ren X, Jia W, Snellingen T, Chongsuvivatwong V, Liu X. 21 22 Strategies of digital fundus photography for screening diabetic retinopathy in a diabetic 23 population in urban China. Ophthalmic Epidemiol 2012; 19(6): 414-9. 24 22. Piyasena M, Yip JLY, MacLeod D, Kim M, Gudlavalleti VSM. Diagnostic test accuracy of 25 diabetic retinopathy screening by physician graders using a hand-held non-mydriatic retinal 26 27 camera at a tertiary level medical clinic. BMC Ophthalmol 2019; 19(1): 89. 28 23. Suansilpong A, Rawdaree P. Accuracy of single-field nonmydriatic digital fundus image 29 in screening for diabetic retinopathy. J Med Assoc Thai 2008; 91(9): 1397-403. 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 15
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 23 BMJ Open
1 2 3 4 5 Table 1: Demographic information of diabetic participants (N=305 subjects) 6 7 Characteristics Statistics 8 9 10 Recruited sites, n (%) 11 12 Community (registered in) For peer review 89 (29.2) only 13 14 Hospital (being referred by physicians) 141(46.2) 15 75(24.6) 16 Hospital (walk in with medical record of diabetes) 17 18 Age, years, n (%) 19 18 - 31 5 (1.64) 20 21 31–40 4 (1.31) 22 41–50 27 (8.85) 23 51–60 97 (31.8) 24 25 61–70 127 (41.6) 26 >70 45 (14.8) 27 Mean (SD) 61.3 (10.1) 28 29 Female sex, n (%) 165 (54.1) 30 31 Educational level, n (%) 32 33 No formal education 68 (22.3) 34 Elementary school 75 (24.6) 35 Junior school 45 (14.8) 36 High school 89 (29.2) 37 38 College or above 28 (9.18) 39 40 41 1 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 18 of 23
1 2 3 4 5 6 Occupation 7 Farmer 15 (4.93) 8 9 Worker 30 (9.87) 10 Officer/clerk 20 (6.58) 11 Technician 9 (2.96) 12 For peer review only 13 Self-employed/freelance professional 26 (8.55) 14 Educational/medical staff 6 (1.97) 15 Retired 164 (53.8) 16 17 Unemployed 35 (11.5) 18 Age at diagnosis of diabetes, years, mean (SD) 52.4 (10.5) 19 20 Duration of diabetes, years, n (%) 21 22 1~ 5 109 (35.7) 23 6 ~ 10 104 (34.1) 24 25 10 ~15 52 (17.1) 26 16~20 28 (9.18) 27 >20 12 (3.93) 28 29 Median, (IQR) 5 (4-12) 30 Fast glucose level on the day of image taken, mmol/L, n (%) 31 32 <=7 108 (35.4) 33 >7 197 (64.6) 34 35 Current treatment of DM, n (%) 36 37 Insulin 112 (36.7) 38 Medicine 262 (85.9) 39 40 41 2 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 19 of 23 BMJ Open
1 2 3 4 5 Diet control 185 (60.7) 6 Chinese Traditional medicine 58 (19.0) 7 No treatment 12 (3.93) 8 9 Hypertension, n (%) 147 (48.2) 10 11 Age at diagnosis of hypertension, year, Mean (SD)† 54.4 (11.1) 12 For peer review only 13 Duration of hypertension, years, n (%) 14 15 1~ 5 59/147 (40.1) 16 6 ~ 10 42/147 (28.6) 17 18 10 ~15 18/147 (12.2) 19 16~20 10/147 (6.8) 20 >20 10/147 (6.8) 21 Not sure when hypertension started 8/147(5.4%) 22 23 Median, (IQR) 7 (3-12) 24 Diabetic complications except hypertension, n (%) 25 26 Nephropathy 23 (7.54) 27 Cardiovascular 34 (11.2) 28 29 Ulcerated arms/legs 8 (2.62) 30 None 252 (82.6) 31 32 Current treatment of Hypertension, n (%) 33 Tablets 125/147 (85.0) 34 35 Diet 77/147 (52.4) 36 Traditional 28/147 (19.0) 37 No treatment 16/147 (10.9) 38 39 40 41 3 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 20 of 23
1 2 3 4 5 6 7 SD: Standard deviation, IQR: Inter quartile range 8 9 †8 could not remember when the hypertension started/diagnosed. 10 11 12 Table 2. Participants’ visual acuity, eyeFor examination peer and treatment reviewsituation (N=305 subjects) only 13 14 n (%) 15 16 Visual acuity of better eye, n (%) 17 18 <0.05 3 (0.98) 19 20 >/=0.05 but < 0.3 26 (8.52) 21 >/= 0.3 276 (90.5) 22 23 Self-assessed visual acuity, n (%) 24 25 Excellent 11 (3.61) 26 Very good 22 (7.21) 27 Good 81 (26.6) 28 29 Fair 110 (36.1) 30 Poor 81 (26.6) 31 32 Frequency of having eye examinations, n (%) 33 34 At least once in the last year 123 (40.3) 35 No eye examination in the last year 182 (59.7) 36 21 (6.89) 37 Whether received photocoagulation, n (%) 38 39 40 41 4 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 21 of 23 BMJ Open
1 2 3 4 5 6 Table 3. Image quality (N=610 eyes) 7 8 9 Items Desktop Portable P-value† 10 11 Number of gradable images* 598 (98.7) 590 (96.9) 0.064 12 For peer review only 13 Number Good Images, n (%)* 482 (79.3) 479 (78.7) 0.745 14 15 16 Number of poor but still gradable, n (%)* 116 (19.1) 111 (18.2) 0.486 17 18 *2 (2/305, 0.66%) eyes had missing data. 19 20 †McNemar's test was used for comparing desktop and portable cameras. 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 5 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 22 of 23
1 2 3 4 5 Table 4. Grading results by the two cameras 6 7 8 By desk top 9 10 Present Absent Total 11 Present 35 3 38 12 R1 AbsentFor peer14 review532 546 only 13 14 Total 49 535 584 15 Present 52 1 53 16 17 R2 Absent 3 528 531 18 Total 55 529 584 19 Present 28 0 28 20 21 By portable R3 Absent 0 556 556 22 Total 28 556 584 23 Present 72 1 73 24 25 M1 Absent 7 504 511 26 Total 79 505 584 27 Present 14 2 16 28 29 OL Absent 4 564 568 30 Total 18 566 584 31 32 33 34 35 36 37 38 39 40 41 6 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 23 of 23 BMJ Open
1 2 3 4 5 Table 5: Accuracy of diagnosis (N=610 eyes)* 6 7 By desk top 8
9 Positive predictive 10 Kappa Sensitivity Specificity AUC value 11 (95% CI) (95% CI) (95% CI) 12 (95% CI) For peer review only (95% CI) 13 14 0.79 71.4% 99.4% 92.1% 0.85 15 R1 16 (0.69, 0.89) (56.7%, 83.4%) (98.4%, 99.9%) (78.6%, 98.3%) (0.79, 0.92) 17 18 0.96 94.6% 99.8% 98.1% 0.97 19 R2 20 (0.92, 0.999) (84.9%, 98.9%) (99%, 100%) (89.9%, 100%) (0.94, 1.00) 21 22 By 1.00 100% 100% 100% 1.00 23 R3 24 portable (1.00, 1.00) (87.7%, 100%) (99.3%, 100%) (87.7%, 100%) (1.00, 1.00) 25 26 0.94 91.1% 99.8% 98.6% 0.95 27 M1 28 (0.90, 0.98) (82.6%, 96.4%) (98.9%, 100%) (92.6%, 100%) (0.92, 0.99) 29 30 0.82 77.8% 99.7% 87.5% 0.89 31 OL 32 (0.68, 0.96) (52.4%, 93.6%) (98.7%, 100%) (61.7%, 98.4%) (0.79, 0.99) 33 34 35 CI=Confidence interval, DR=Diabetic Retinopathy, NPDR=Non- Proliferative Diabetic Retinopathy, PDR=Proliferative Diabetic Retinopathy AUC=Area Under 36 37 the Receiver-operator Curve, OL: other lesion 38 *2 eyes had missing data. 39 40 41 7 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 24 of 23
1 2 3 Table 6 Camera preferred by both patients and technician (N=305 subjects) 4 5 6 n (%) 7 8 Camera preferred by patients 9 10 Standard 114 (37.4) 11 12 Portable 34 (11.1) 13 14 Same 157 (51.5) 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 8
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open
Validation of handheld fundus camera with mydriasis for retinal imaging of diabetic retinopathy screening in China – A prospective comparison study ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2020-040196.R1
Article Type: Original research
Date Submitted by the 05-Aug-2020 Author:
Complete List of Authors: Xiao, Baixiang; Zhongshan Ophthalmic Centre, Sun Yat-sen University, The State Key Laboratory of Ophthalmology Liao, Qinghua; Zhenjiang District Hospital, Shaoguan City Li, Yanping; Nanchang University affiliated Eye Hospital Weng, Fan; Zhongshan Ophthalmic Centre, Sun Yat-sen University Jin, Ling; Sun Yat-Sen University Zhongshan Ophthalmic Center Wang, Yanfang; Zhongshan Ophthalmic Centre, Sun Yat-sen University Huang, Wenyong; Sun Yat-Sen University Zhongshan Ophthalmic Center Yi, Jinglin; Nanchang University Affiliated Eye Hospital Burton, Matthew J; London School of Hygiene and Tropical Medicine Yip, Jennifer; London School of Hygiene & Tropical Medicine
Primary Subject Diabetes and endocrinology Heading:
Secondary Subject Heading: Ophthalmology
Diabetic retinopathy < DIABETES & ENDOCRINOLOGY, Diabetic Keywords: nephropathy & vascular disease < DIABETES & ENDOCRINOLOGY, OPHTHALMOLOGY
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 22 BMJ Open
1 2 3 4 5 6 7 8 9 I, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined 10 in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors 11 who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance 12 with the terms applicable for US Federal Government officers or employees acting as part of their official 13 duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd (“BMJ”) its 14 licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the 15 Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence. 16 17 The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to 18 the Submitting Author Forunless you peer are acting as review an employee on behalf only of your employer or a postgraduate 19 student of an affiliated institution which is paying any applicable article publishing charge (“APC”) for Open 20 Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and 21 intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative 22 Commons licence – details of these licences and which Creative Commons licence will apply to this Work are set 23 out in our licence referred to above. 24 25 Other than as permitted in any relevant BMJ Author’s Self Archiving Policies, I confirm this Work has not been 26 accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate 27 material already published. I confirm all authors consent to publication of this Work and authorise the granting 28 of this licence. 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 22
1 2 3 4 Validation of handheld fundus camera with mydriasis for 5 6 7 retinal imaging of diabetic retinopathy screening in China – A 8 9 prospective comparison study 10 11 12 1,Baixiang Xiao MD, 2Qinghua Liao MS, 3Yanping Li PhD, 4Fan Weng MS, 1Ling Jin MS,
13 1 1 3 5 5 14 Yanfang Wang MB, Wengyong Huang PhD, Jinglin Yi MD, Matthew Burton PhD, Jennifer 15 L Yip PhD 16 17 18 For peer review only 19 1. 20 The State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Centre, Sun Yat-sen 21 University, Guangzhou, China. 22 2. Zhenjiang District Hospital, Shaoguan City, China 23 3. Nanchang University Affiliated Eye Hospital, Nanchang City, China 24 25 4. Yuexiu District Centre of Disease Control, Guangzhou, China 26 5. International Centre for Eye Health, London School of Hygiene & Tropical Medicine, Keppel 27 Street, London WC1E 7HT, United Kingdom 28 29 30 Address for correspondence: Dr. Baixiang Xiao, The State Key Laboratory of Ophthalmology, 31 32 Zhongshan Ophthalmic Centre, Sun Yat-sen University, #54 South Xianlie Road, Guangzhou 33 City, Guangdong Province, China 510060 Tel/Fax: +86 20 8733 4645, Email: 34 35 [email protected] 36 37 Word count: 3993 38 39 Running head: Validation of handheld fundus camera 40 41 42 Keywords: validation, fundus camera, sensitivity, specificity, diabetic retinopathy, 43 44 45 46 The conflicts of interest: The authors declare that they have no conflicts of interest. 47 48 49 Funding: The study is fully funded by the State Key Laboratory of Ophthalmology, Zhongshan 50 Ophthalmic Centre, Sun Yat-sen University 51 52 Data sharing: Full data is available on the request from the corresponding author. 53 54 55 The study was approved by the medical ethics committee from Zhongshan Ophthalmic Center, 56 Sun Yat-sen University. 57 58 The number of the approval is: 2018KYPJ061 59 60 1
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 22 BMJ Open
1 2 3 Contributorship statement: 4 5 6 B Xiao: study design, training of the study teams, monitoring data collection and quality 7 control, data cleaning, analysis as well as manuscript drafting and revision. 8 9 10 Q Liao: preliminary draft of the tables and manuscript, data analysis and critical revision of the 11 manuscript. 12 13 Y Li: Monitoring of grading and valuable contribution to the revision of the manuscript. 14 15 16 F Weng: data collection and valuable contribution to the study design. 17 18 L Jin: statistical analysisFor and peer helpful revision review of the manuscript. only 19 20 Y Wang: data cleaning and valuable contribution to the manuscript drafting. 21 22 23 J Yi, W Huang: valuable contribution to the study design and revision of the manuscript. 24 25 J Yip, M Burton: Substantial contributions to the conception, designing, and revision of the 26 manuscript as well as analysis and interpretation of the data. 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 2
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 22
1 2 3 Abstract: 4 5 6 Objectives: To investigate the clinical validity of using a handheld fundus camera to detect 7 diabetic retinopathy (DR) in China. 8 9 10 Design and settings: Prospective comparison study of the handheld fundus camera with a 11 standard validated instrument in detection of DR in hospital and a community screening clinic 12 13 in Guangdong Province, China. 14 15 Participants: Participants aged 18 and over with diabetes who were able to provide informed 16 17 consent and agreed to attend the dilated eye examination with handheld tests and a standard 18 desktop camera.For peer review only 19 20 21 Primary and secondary outcome measures: Primary outcome was the proportion of those 22 with referable DR (R2 and above) identified by the handheld fundus camera (the index test) 23 compared with the standard camera. Secondary outcome was the comparison of proportion of 24 25 gradable images obtained from each test. 26 27 Results: 28 29 30 In this study, we examined 304 people (608 eyes) with each of the two cameras under 31 mydriasis. The handheld camera detected 119 eyes (19.5%) with some level of DR, 81(13.3%) 32 33 of them were referable, while the standard camera detected 132 eyes (21.7%) with some level 34 of DR and 83(13.7%) were referable. It seems that the standard camera found more eyes with 35 36 referable DR, although McNemar’s test detected no significant difference between the two 37 cameras. 38 39 Of the 608 eyes with images obtained by desktop camera, 598 (98.4%) images were of 40 41 sufficient quality for grading, 12 (1.9%) images were not gradable. By the handheld camera, 42 590 (97.0%) were gradable and 20 (3.2%) images were not gradable. 43 44 45 The two cameras reached high agreement on diagnosis of retinopathy and maculopathy at all 46 the levels of retinopathy. 47 48 49 Conclusion: 50 Although it could not take the place of standard desktop camera on clinic fundus examination, 51 the handheld fundus camera showed promising role on preliminary DR screening at primary 52 level in China. To ensure quality images, mydriasis is required. 53 54 55 56 57 58 59 60 3
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 22 BMJ Open
1 2 3 Article Summary 4 5 6 Strengths and limitations of this study: 7 8 1. Study subjects recruited for this study where from both outpatient department of hospital 9 10 and community screening with the variety of diabetic retinopathy for the study comparison. 11 2. Two experienced graders from tertiary eye institute graded the images captured by both 12 cameras separately and reached full agreement on grading. 13 14 3. To avoid bias from personnel, technicians were experienced with the standard camera and 15 received full training in using both the handheld and standard cameras. 16 4. We noticed high rate of images with poor quality without dilation at the pilot study and 17 18 started to give mydriasisFor for peer all the participants review at the formal only study, so that there were lack of 19 data on non-mydriasis. 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 4
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 22
1 2 3 Introduction 4 5 Diabetes mellitus (DM) is a major cause of morbidity and mortality worldwide, responsible for 6 1.5 million deaths in 2012. 1The ageing population, rising levels of obesity and lifestyle 7 8 changes will increase this figure of DM further. In China, the prevalence of diabetes increased 9 dramatically in the last 30 years, with recent prevalence estimates in adults of 10.9%.2 10 11 Diabetic retinopathy (DR) is a common microvascular complication causing retinal 12 13 haemorrhage and oedema in people with diabetes mellitus (PwDM). This reduces visual acuity 14 (VA) at a late stage in the condition/disease or when the macular is affected when the 15 16 treatment is not optimal. Worldwide, it is one of the most common causes of visual impairment, 17 particularly in working age adults, with significant economic impact.3 In China, a systematic 18 For peer review only 19 review showed that approximately 1 in 5 people diagnosed with diabetes has some level of 20 DR, which is similar to other high income countries. 4, 5 21 22 23 There are effective strategies to prevent and treat DR. The risk of vision loss can be reduced 24 with metabolic management, early detection through screening, and appropriate laser, 25 26 intra-vitreous injection or vitrectomy. 27 28 Diabetic retinopathy screening (DRS) and appropriate referral for treatment have been shown 29 6 30 to reduce blindness from DR. However, studies have shown that in rural areas of China, only 31 10% of those with DR are diagnosed and treated. 4 This indicates a need to ensure accessible 32 33 DRS services to reduce potential eye health inequalities. 34 35 The primary health care providers in China encourage PwDM to register and receive an 36 annual health check in the rural township clinic and urban community health centre to monitor 37 38 patients’ glucose levels and detect potential complications. Primary care doctors are not 39 trained in eye care, and PwDM are asked to travel to county hospitals for eye examinations at 40 41 their own expense, which results in inequitable access. 42 43 Fundus cameras have now been installed rapidly in most of the county level hospitals since 44 45 the National study on eye service in 2014 found that only 10% of the secondary level hospitals 46 had the capacity to take fundus images (data not been published). Outreach services are 47 48 limited due to the lack of specialist eye-care staff in secondary care, which also leads to 49 variation in screening uptake and consequently treatment of DR. 50 51 52 In a clinical review of PwDM in Guangzhou, 43.2% of people in tertiary and community urban 53 setting had never received an eye examination. In rural clinics, 68.7% had never received an 54 7 55 eye examination. 56 57 Provision of screening at primary care level can increase uptake8-10, however, the cost of 58 providing all primary care clinics in China with a digital camera would be prohibitive, together 59 60 5
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 22 BMJ Open
1 2 3 with implications of additional training of staff and maintenance of equipment. DR services are 4 5 in their infancy in China. 6 7 The Zhongshan Ophthalmic Centre, Sun Yat-sen University (ZOC) is a leading eye institution, 8 9 and well placed to plan and develop a DR service model in China. Currently, there are no 10 sustainable and scalable models for delivery of DR services in rural China. 11 12 13 The aim of this study is to validate retinal images from a handheld portable retinal camera for 14 DRS, using a desktop digital camera as the comparison. Findings from this study will lead to 15 16 further investigation of the role of handheld cameras for the acquisition of retinal images and 17 improving access and increasing uptake of DRS in primary care clinics in China. 18 For peer review only 19 Methods 20 21 22 Ethical approval of this study was obtained from both London School of Hygiene & Tropical 23 Medicine and Zhongshan Ophthalmic Centre, Sun Yat-sen University. All participants provided 24 25 written, informed consent. Patients with referable diabetic or other eye diseases were referred 26 to ZOC for further examination or treatment. The study fulfilled the tenets of the Declaration of 27 28 Helsinki. 29 30 Patient and public involvement statement: We talked with participants about the purpose of 31 the study, how it could be done and what support needed from them before and during the 32 33 study. These information were also disseminated by primary health workers before consent 34 obtained from each of the participant 35 36 37 This was a prospective comparison study of the hand-held fundus camera (index test) of 38 Horus Scope DEC 20011 with the desktop digital camera (standard test) Canon (model CR-2) 39 40 in detection of referable DR. 41 42 Population: 43 44 Eligible participants were all those aged 18 and over with diabetes who were able to provide 45 46 informed consent and agreed to attend for dilated eye examination with both index and 47 standard tests. Diabetes was identified by self-report as well as definite medical records in 48 49 hand, referral by endocrinologist, or registration in the primary health care centres. 50 51 Participants were recruited from Zhenjiang District (Shaoguan Prefecture) hospitals, Chenghai 52 53 District Hospital (Shaotou City) and Yuexiu District community (Guangzhou City) health 54 centres in consecutive series in Guangdong Province, China, with a range of DR severity, 55 56 including patients without DR in order to obtain a representative spectrum of patients in this 57 study. 58 59 60 Sample size 6
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 22
1 2 3 A sample size of 262 patients has 80% power and 5% significance level to detect a 6% 4 5 difference in proportion of gradable images between the index and the standard test, where 6 the standard test will produce 90% gradable images. 7 8 9 Training 10 11 The pilot study was conducted in the community screening clinic in Guangzhou City. 12 13 Technicians with at least one year experiences of operation for the handheld camera and desk 14 top camera examined thirty cases with both un-dilated and dilated pupils according to the 15 16 assessment on the quality of images and operation on the camera by an ophthalmologist . 17 18 In the other studyFor sites, we peertrained technicians review with experiences only of fundus cmeras to capture 19 images on the use of the handheld camera as well as the standard camera to ensure a 20 21 standardized process, including “install/uninstall”, how to capture images with both camera 22 and fill the data collection form. They each practiced on the handheld camera for 23 24 approximately 20 pilot cases, closely supervised by experienced technician until there were no 25 more questions on using the camera and quality of the images taken were considered 26 27 acceptable by the senior ophthalmologist. The training took 2 hours and the trainer observed 28 for the first whole day before technicians operated independently. 29 30 31 Data collection 32 33 A data collection form was developed including information on the patient’s age, gender, 34 education, profession, visual acuity (by illuminated Snellen Visual Chart), use of glasses, 35 36 self-assessed visual function, history of eye examinations and fundus photocoagulation, 37 history of diabetes mellitus (DM), complications, treatment of DM and hypertension (HP), 38 39 fasting glucose on the day of fundus photo taken, which camera the participants preferred and 40 why. Treatment options for DM and HP were given in multiple choices, each with four options 41 42 as for DM: insulin, oral medicine, diet, Chinese Traditional Medicine (CTM) and no treatment, 43 for HP as: oral medicine, diet, CTM and no treatment. 44 45 46 Pilot study 47 48 Without dilation, from both cameras, approximately one-third of the images from the first eye 49 50 (always right eye) were of poor quality (including un-gradable and poor but still gradable), and 51 10% were un-gradable. We asked participants to rest in a darkened room for two minutes after 52 photographs were taken from the first eye, and up to fifteen minutes for those with poor quality 53 images, to facilitate pupil dilation, but there was a higher proportion of poor quality pictures 54 55 from the second eye (40%) with both cameras (Data not presented). This took one day in 56 Guangzhou and 30 PwDM were examined with an ophthalmologist onsite for the assessment. 57 58 59 60 7
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 22 BMJ Open
1 2 3 Due to the high rate of poor images and patients’ dissatisfaction from waiting in the dark room 4 5 for second camera and sometimes even for the second eye, we decided to give mydiasis to 6 every participant for image taken in the formal study. 7 8 The formal study 9 10 All participants had the following tests in sequence during one visit in the clinic within 2 hours 11 of the first photograph to ensure photographs from both cameras were obtained under 12 13 maximal dilation: participant’s basic information, visual acuity test, anterior segment 14 examination under slit lamp by an ophthalmologist, intraocular pressure (IOP) test, dilation of 15 16 both eyes, fundus photograph of both eyes by the handheld retinal camera and the desk top 17 digital camera and finally, the questions on preference of and comments on the cameras. We 18 For peer review only 19 randomly assigned patients to a different sequence of cameras, with sufficient time in between 20 to minimize discomfort. 21 22 23 Two photos were taken for each eye of all the participants by each of the two cameras. One 24 centred on the macula and the other on the papilla optica. One technician operated both the 25 26 handheld and standard cameras for participants in each of the three study sites (total 3). The 27 operators of the cameras were asked the advantages and disadvantages of both cameras, the 28 29 ease of use, graders were asked the acceptability of mode of photography. 30 31 The fundus photographs from all tests were uploaded on the DR online grading system and 32 graded independently at the grading centre in ZOC where the experienced graders were 33 34 masked to the mode of photograph where possible. 35 36 The two graders graded all the study images separately and then together to compare whether 37 there were disagreement on the grading results, an ophthalmologist was used as an arbitration 38 39 grader. They discussed any disagreement until they reached consensus. They had been 40 trained appropriately as the graders in ZOC’s grading centre, which serves for over 70 41 secondary hospitals across the country in programs. They both had been working in this 42 43 centre for at least 5 years and constantly monitored by ophthalmologist supervisors on quality 44 of their work. 45 46 47 48 Image quality 49 50 Images were considered good quality when features were focused, well illuminated retinal 51 52 field, showing clarity of the fundus vessels and any retinopathy. If the images were only 53 partially; focused, illuminated or retinal field showed, they were defined as poor, but still 54 55 gradable. If any retinopathy was detected, either DR or non-DR, the images were also defined 56 as gradable. If images are blurred without recognition of the retinal vessels or retinopathy 57 58 features, they are defined as not gradable. The patients with not gradable images were then 59 referred to the ophthalmologist for further examination. 60 8
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 22
1 2 3 Grading system: 4 5 6 To define the fundus pathology and grade the retinal changes, we used grading definitions for 7 referable disease by the English NHS Diabetic Eye Screening Program.12 That is R0 is 8 9 categorised as the absence of any DR feature including microaneurysms. Microaneurysm with 10 or without exudation is categorised as R1 for its only presence without other DR features. 11 12 13 The eye is categorised as R2 if any of the following features are present: venous beading, 14 cotton wool spots, venous reduplication, multiple blot haemorrhages and intra-retinal 15 16 microvascular abnormality. 17 18 If there is presentingFor proliferative peer retinopathy, review that is new blood only vessels or haemorrhage within 19 retina or in vitreous, or vitreous traction, the eye is categorised as R3a. If there is evidence of 20 21 retinal laser treatment and DR features are stable, the eye is categorised as R3s. Presence of 22 microaneurysms, haemorrhage or exudes within 2 disc diameters of the centre of the fovea is 23 24 categorised as M1. 25 26 Statistics Analysis: 27 28 Data was presented as mean (standardized deviation, SD) or median (inter quartile range, 29 30 IQR) for continuous variable and frequency (%) for categorical variable. Participants’ age was 31 categorized by interval of 10 years. Age at diagnosis of diabetes and hypertension was 32 33 categorized by interval of 5 years. The cut-offs of 0.05 and 0.3 were used to describe visual 34 acuity in better-seeing eye. Data were analysed by eyes. McNemar's test was performed for 35 36 comparing the standard and portable cameras. The inter-rater reliability between two cameras 37 was measured by Cohen's kappa coefficient (95% confidence interval, CI). Sensitivity, 38 39 specificity, positive predictive value and area under the Receiver-operator curve with 95% CI 40 were calculated to indicate the accuracy of diagnosis by two cameras. All statistical analyses 41 42 were performed using a commercially available software package (Stata 13.1, StataCorp, 43 College Station TX, USA). 44 45 46 Comments on the cameras: 47 48 Comments from both staff and the patients were simply listed and similar ones were 49 50 categorised together until there were no more new comments and repeated ones were 51 included together as one comment. 52 53 Results 54 55 56 Participants’ characteristics 57 Patient recruitment started in June 2018 simultaneously in three places and ended up a total of 58 305 diabetes people examined by the end of the year. Mean age of the participants was 61.3 59 60 (SD ±10.1) and almost half (41.6%) were between 61 and 70 years old. Among them, 165 9
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 22 BMJ Open
1 2 3 (54.1%) were female (Table 1). Over one-third received high school and above education. 4 5 Over half (53.6%) of the participants were retired and approximately 10% unemployed. Mean 6 age at diagnosis of diabetes was 52.4 (SD±10.5). Among these participants, around one-third 7 had diabetes for less than 5 years, one-third for 6 to 10 years and the rest had over 10 years. 8 Only 12 (3.9%) people had diabetes for over 20 years. The median duration of diabetes was 5 9 10 years (4-12). 11 12 Table 1: Demographic information of participants with diabetes (N=305 subjects) 13 14 15 Characteristics Statistics 16 17 Recruited sites, n (%) 18 For peer review only 19 20 Community (registered in) 89 (29.2) 21 22 Hospital (being referred by physicians) 141(46.2) 23 Hospital (walk in with medical record of diabetes) 75(24.6) 24 25 Age, years, n (%) 26 27 18 - 31 5 (1.64) 28 31–40 4 (1.31) 29 30 41–50 27 (8.85) 31 51–60 97 (31.8) 32 61–70 127 (41.6) 33 34 >70 45 (14.8) 35 Mean (SD) 61.3 (10.1) 36 37 Female sex, n (%) 38 165 (54.1) 39 Educational level, n (%) 40 41 No formal education 68 (22.3) 42 Elementary school 75 (24.6) 43 44 Junior school 45 (14.8) 45 High school 89 (29.2) 46 College or above 28 (9.18) 47 48 49 Occupation 50 Farmer 15 (4.93) 51 52 Worker 30 (9.87) 53 Officer/clerk 20 (6.58) 54 Technician 9 (2.96) 55 Self-employed/freelance professional 26 (8.55) 56 57 Educational/medical staff 6 (1.97) 58 Retired 164 (53.8) 59 Unemployed 35 (11.5) 60 10
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 22
1 2 3 4 Age at diagnosis of diabetes, years, mean (SD) 52.4 (10.5) 5 6 Duration of diabetes, years, n (%) 7 1- 5 109 (35.7) 8 9 6 - 10 104 (34.1) 10 10 -15 52 (17.1) 11 16-20 28 (9.18) 12 13 >20 12 (3.93) 14 Median, (IQR) 5 (4-12) 15 16 Fasting glucose level on the day of image taken, mmol/L, n (%) 17 <=7 108 (35.4) 18 For peer review only 19 >7 197 (64.6) 20 Current treatment of DM, n (%) 21 22 Insulin 112 (36.7) 23 Medicine 262 (85.9) 24 25 Diet control 185 (60.7) 26 Chinese Traditional medicine 58 (19.0) 27 No treatment 12 (3.93) 28 29 Hypertension, n (%) 147 (48.2) 30 31 Age at diagnosis of hypertension, year, Mean (SD)† 54.4 (11.1) 32 33 Duration of hypertension, years, n (%) 34 35 1- 5 59/147 (40.1) 36 6 - 10 42/147 (28.6) 37 10 -15 18/147 (12.2) 38 39 16-20 10/147 (6.8) 40 >20 10/147 (6.8) 41 Not sure when hypertension started 8/147(5.4%) 42 43 Median, (IQR) 7 (3-12) 44 Diabetic complications except hypertension, n (%) 45 46 Nephropathy 23 (7.54) 47 Cardiovascular 34 (11.2) 48 49 Ulcerated arms/legs 8 (2.62) 50 None 252 (82.6) 51 52 Current treatment of Hypertension, n (%) 53 Tablets 125/147 (85.0) 54 55 Diet 77/147 (52.4) 56 Traditional 28/147 (19.0) 57 No treatment 16/147 (10.9) 58 59 60 11
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 22 BMJ Open
1 2 3 4 5 SD: Standard deviation, IQR: Inter quartile range 6 †8 could not remember when the hypertension started/diagnosed. 7 8 9 On study days, we detected only 108 (35.4%) participants with fasting glucose below 7mmol/L 10 (Table 1), which is clinically considered as good control. Insulin was used by 112 (36.7%) 11 12 people and 262 (85.9%) were taking oral medication, 12 (4%) reported no treatment, not even 13 diet. 14 15 Hypertension was detected in 48.2% participants, nephropathy in 7.5%, cardiovascular 16 17 disease in 11.2%. Of these 305 PwDM, 252 (82.6%) did not have any complications of 18 diabetes. Mean ageFor at diagnosis peer of hypertension review was 54.4 (SD±11.1).only Of the 147 participants 19 20 with hypertension, eight could not remember when the problem started or being diagnosed, 59 21 (40.1%) people were diagnosed less than 5 years and 10 (6.8%) over 20 years ago (Table 1). 22 23 Majority (85.0%) of these people with HP were taking oral medication and 16 (10.9%) received 24 no treatment. 25 26 27 Of the 305 participants, 276 (90.5%) had presenting visual acuity (PVA) equal to or over 0.3, 3 28 (0.98%) below 0.05, while by self-assessment, only 11 (3.61%) expressed that they had 29 30 excellent VA, 81 (26.6%) said their VA was good and 81 (26.6%) felt their VA was poor (Table 31 2). Over half (59.7%) of the participants did not have their eyes examined by medical staff in 32 33 the previous year. 21 (6.89%) people had received laser photocoagulation. 34 35 Table 2. Participants’ visual acuity, eye examination and treatment situation (N=305 36 subjects) 37 38 n (%) 39 40 Visual acuity of better eye, n (%) 41 42 <0.05 3 (0.98) 43 44 >/=0.05 but < 0.3 26 (8.52) 45 >/= 0.3 276 (90.5) 46 47 Self-assessed visual acuity, n (%) 48 49 Excellent 11 (3.61) 50 Very good 22 (7.21) 51 Good 81 (26.6) 52 53 Fair 110 (36.1) 54 Poor 81 (26.6) 55 56 Frequency of having eye examinations, n (%) 57 58 At least once in the last year 123 (40.3) 59 No eye examination in the last year 182 (59.7) 60 12
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 22
1 2 3 21 (6.89) 4 Whether received photocoagulation, n (%) 5 6 7 8 9 Agreement by the two graders 10 11 Grading results from the two graders reached agreement on 606 eyes and disagreed on only 12 two eyes for R1. They reached to consensus after discussion. 13 14 15 Image gradeability, referable eyes and agreement analysis 16 17 Of the 305 people (610 eyes) examined, 1 patient (2 eyes) had the images duplicated with the 18 For peer review only 19 previous patient owing to the wrong saving name in the computer, which left 608 eyes for 20 grading. From desktop camera, 482 (79.3%) images were of good quality and gradable, 116 21 22 (19.1%) were poor but still gradable, and 12 (1.9%) images were not gradable (Table 3). From 23 the handheld camera, 479 (78.7%) were of good quality and gradable, 111 (18.2%) were of 24 25 poor quality, but still gradable and 20 (3.2%) images were not gradable. Between the two 26 groups of good and poor quality but still gradable images by the two cameras, McNemar’s test 27 28 found no significant difference, while the two groups added together, images taken by desktop 29 gained slightly better quality than handheld although the difference was not significant (p 30 31 >0.05%). 32 33 Table 3. The quality of images (N=610 eyes) 34 35 Desktop Portable P-value† 36 Items 37 38 Number of gradable images* 598 (98.7) 590 (96.9) 0.064 39 40 Number Good Images, n (%)* 482 (79.3) 479 (78.7) 0.745 41 42 Number of poor but still gradable, n (%)* 116 (19.1) 111 (18.2) 0.486 43 44 45 *2 (2/305, 0.66%) eyes had missing data. 46 47 †McNemar's test was used for comparing desktop and portable cameras. 48 49 For the non-gradable eyes, the two cameras agreed on 5 eyes, two of which had vitreous 50 opacity, the other three eyes had dense cataract. There was one eye captured R1 by handheld 51 52 camera, but the image captured by desk top was non-gradable as handheld camera happened 53 to capture some peripheral microaneurysms although images taken by both cameras were 54 poor from this cataractous eye. 55 56 57 There were cataracts in 8 eyes and ascertained as R1 by desktop camera, but not gradable by 58 the handheld camera. 59 60 13
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 22 BMJ Open
1 2 3 There were 14 eyes graded as R1 by the desk top camera, but ungradeable by the handheld 4 5 camera. 6 7 8 In total, there were 132 eyes (21.7%) with evidence of retinopathy (R1+R2+R3) (table 4), from 9 10 images taken by desk top camera and 119 eyes (19.5%) by handheld camera. We detected 11 referable retinopathy at R2 and above in 83 eyes (13.7%) from the desktop camera and 81 12 eyes (13.3%) from the handheld camera, with no evidence of a statistically significant 13 difference. R3 was detected in 28 eyes (4.6%) by both handheld and desktop cameras. Of the 14 15 132 eyes with some level of retinopathy, 79 (59.8%) eyes had macular involvement by 16 standard camera. 17 18 For peer review only 19 Table 4. Grading results by the two cameras 20 21 22 By desk top 23 Present Absent Total 24 25 Present 35 3 38 26 R1 Absent 14 532 546 27 28 Total 49 535 584 29 Present 52 1 53 30 R2 Absent 3 528 531 31 32 Total 55 529 584 33 Present 28 0 28 34 By portable R3 Absent 0 556 556 35 36 Total 28 556 584 37 Present 72 1 73 38 M1 Absent 7 504 511 39 40 Total 79 505 584 41 Present 14 2 16 42 OL Absent 4 564 568 43 44 Total 18 566 584 45 46 The two cameras reached high agreement on diagnosis of retinopathy and maculopathy at the 47 48 levels of R1 (Kappa coefficient - KC was 0.79), R2 (KC=0.96), R3 (KC=1.0), M1 (KC=0.94) 49 and other lesion (KC=0.82) (Table 5). 50 51 52 We identified 49 eyes at R1 by the desk top camera and 38 eyes by the handheld camera , 53 which resulted in a sensitivity of 71.4% (95% CI: 56.7 – 83.4), specificity of 99.4% (95% CI: 54 98.4 – 99.9) and positive predictive value (PPV) of 92.1% (95%CI: 78.6 – 98.3). 55 56 57 58 59 60 14
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 22
1 2 3 Table 5: Accuracy of diagnosis (N=610 eyes)* 4 5 By desk top 6 7 8 Positive Negative Kappa Sensitivity Specificity AUC 9 predictive value predictive value 10 (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) 11 12 For peer review only 13 0.79 71.4% 99.4% 92.1% 97.4% 0.85 14 R1 15 (0.69, 0.89) (56.7%, 83.4%) (98.4%, 99.9%) (78.6%, 98.3%) (95.7%, 98.6%) (0.79, 0.92) 16 17 0.96 94.6% 99.8% 98.1% 99.4% 0.97 18 R2 19 (0.92, 0.999) (84.9%, 98.9%) (99%, 100%) (89.9%, 100%) (98.4%, 99.9%) (0.94, 1.00) 20 21 By 1.00 100% 100% 100% 100% 1.00 22 R3 23 portable (1.00, 1.00) (87.7%, 100%) (99.3%, 100%) (87.7%, 100%) (99.3%, 100%) (1.00, 1.00) 24 25 0.94 91.1% 99.8% 98.6% 98.6% 0.95 26 M1 27 (0.90, 0.98) (82.6%, 96.4%) (98.9%, 100%) (92.6%, 100%) (97.2%, 99.4%) (0.92, 0.99) 28 29 0.82 77.8% 99.7% 87.5% 99.3% 0.89 30 OL 31 (0.68, 0.96) (52.4%, 93.6%) (98.7%, 100%) (61.7%, 98.4%) (98.2%, 99.8%) (0.79, 0.99) 32 33 34 CI=Confidence interval, DR=Diabetic Retinopathy, NPDR=Non- Proliferative Diabetic Retinopathy, PDR=Proliferative Diabetic Retinopathy AUC=Area Under the 35 Receiver-operator Curve, OL: other lesion 36 37 *2 eyes had missing data. 38 39 40 41 15 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 17 of 22 BMJ Open
1 2 3 At the level of referable retinopathy at R2, the desk top camera detected 55 eyes and 4 5 handheld detected 53 eyes, resulting in a sensitivity of 94.6% (95% CI: 84.0 – 98.9), specificity 6 of 99.8% (95% CI: 99 – 100), PPV to 98.1% (95% CI: 89.9 -100). While for R3, the two 7 cameras reached 100% agreement. The desk top camera captured 6 more eyes with 8 maculopathy while the handheld one did not detect maculopathy in these eyes and a 9 10 sensitivity of 91.1% (95% CI: 82.6 – 96.4), specificity of 99.6% (95% CI: 98.9 – 100), PPV of 11 98.6% (95% CI: 92.6 – 100). The desk top camera also captured two more eyes with other 12 lesions, which the handheld camera had not. 13 14 15 Camera preference by both patients and technicians: 16 17 18 Approximately halfFor (51.2%) peer of the participants review had no preference only for either of the two cameras, 19 114 (37.4%) people preferred the desktop camera and 34 (11.1%) said the handheld one was 20 preferred (Table 6). Those preferring the standard camera gave reasons as follows: flash light 21 for images taken by the standard camera was not so bright as the handheld one; it looked 22 23 more complicated and was bigger, so that should be better, as recognized by most of 24 non-medical people of the medical equipment; it was more convenient for height adjusting as it 25 was on an elevator-platform. Those participants who felt handheld was better gave reasons of 26 27 simplicity, looking smart with easy mobilization of the whole machine with or without the 28 foldable stand (frame) for patient’s chin rest. 29 30 31 Table 6 Camera preferred by both patients and technician (N=305 subjects) 32 33 n (%) 34 35 Camera preferred by patients 36 37 114 (37.4) 38 Standard 39 34 (11.1) 40 Portable 41 157 (51.5) 42 Same 43 44 45 46 Comments from the three experienced technicians on the two cameras were: “for the 47 community DR screening, the handheld one is enough”. “It is easy to install and pack up”. 48 49 They also pointed out that, with the simple stand for patient’s chin rest, focusing process 50 became much easier and quicker than when without. 51 52 Discussion: 53 54 We compared the quality of fundus images and the referable eyes with images taken by the 55 standard Canon CR2 desk top and a simple Forus handheld fundus cameras from the same 56 305 PwDM in three places of Guangdong Province, China and our analysis found no 57 58 difference on proportion of gradable images, good images and number of eyes with referable 59 retinopathy between the two cameras when mydriasis was used. 60 16
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 22
1 2 3 Training for those without experience was easy and relatively quick. The handheld camera is 4 5 simply designed, easy to install and to pack up. The light and foldable stand designed for the 6 handheld camera makes it possible to be carried by hand. Its images cover the same retinal 7 field as the standard one. 8 9 10 In the English national screening program for DR, Scanlon et al demonstrated that two images 11 centred on the disc and macular with 450 field camera achieved high sensitivities (>87%) and 12 specificities (>86%) with low un-gradable image rate of below 4.4% against the reference 13 14 standard of seven-field stereophotography or an ophthalmologist using slit lamp 15 biomicroscope.13, 14 The two field images method was recommended for DRS and was used 16 as the gold standard in this study. 17 18 For peer review only 19 Overall, there was no significant difference between using the handheld camera with dilation 20 on detecting number of referable eyes with DR compared with the standard camera. The two 21 cameras reached a high level of agreement on grading results of DR (Kappa from 0.79 at R1 22 23 to 1.00 at R3). When authors looked at the images disagreed against findings from slit lamp, 24 we noticed the following factors: For those eyes with cataract at the stage C3N3 to C4N4 15 25 and those with some level of vitreous opacity, the desk top could still capture images of blood 26 27 vessels with some degree of clarity compared to the handheld camera. 28 29 In China, primary health staff commonly hesitate to give mydriasis for fundus examinations 30 16 31 and are normally not confident to convince patients about the low risk of complications. In 32 this study, the poor image rate reduced generally from over 30% to around 10%. This in the 33 meantime reduced the number of people referred to secondary care for further examination 34 and saved resources. From the findings in this study, we would recommend mydriasis for DRS 35 36 at primary level with training for primary health staff to obtain patient consent and manage 37 potential complications in future programs. 38 39 40 In this study, around one-third - 108 (35.4%) of participants had their fasting glucose controlled 41 at the recommended levels. Furthermore, 12 participants did not take any treatment for their 42 43 DM, not even diet. Over half (59.7%) of the participants had not had their eyes checked by 44 medical staff in the previous year. This emphases the importance of health education and 45 46 available services for DM patients in primary health care. 47 48 Approximately two-thirds of study participants were the walk-in patients with diabetes medical 49 records in hand to the clinics of secondary level hospitals and one-third were community 50 51 health unit registry, from them, we detected a similar lower rate of DR as other studies in China 52 5, 17, 18compared to Singapore , India and United States,19 while much lower than Handan Eye 53 Study4 in North China and the more likely findings from other walk-in patients in six provinces 54 in China20. The study also detected 82.6% of the participants without any diabetic 55 56 complications, which is mainly attributed to the median duration of diabetes was 5 years (4-12) 57 and only 3.93% of participants had diabetes for over 20 years. 58 59 60 17
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 22 BMJ Open
1 2 3 We detected a similar proportion of referable DR, high proportion of poor quality images at 4 21 5 unmydriasis to the previous studies in China. Findings for the comparison of the two 6 cameras are similar to the studies in other Asia countries like Sri Lanka22 and Thailand23. 7 8 This study employed experienced eye care staff, who had been performing similar work for at 9 10 least three years, to take the images with both desk top and the handheld cameras. There 11 were also standardized procedures and training for all staff in the different sites, which would 12 reduce measurement error. Graders were masked from patients’ history and source of images, 13 14 reducing ascertainment bias. Where the two cameras differed in their images we carried out a 15 slit lamp examination to determine the cause of the disagreement. 16 17 18 This was not a populationFor basedpeer study andreview therefore patient only characteristics were not 19 representative of DR with respect to severity of the diseases and its complications in the 20 context of China, however, we included the full range of DR severity in our study population, 21 mitigating spectrum bias. 22 23 24 Conclusion: 25 A handheld fundus camera using mydriasis may have a role to play in preliminary DR 26 27 screening at the primary level in China and other settings worldwide where desktop camera 28 are not prevalent or easily accessible or where screening programs are not operational. 29 30 31 Acknowledgement 32 The authors acknowledge the doctors, technicians and nurses in Shaoguan Railway Hospital, 33 Chenghai District Hospital and Community Health Centres in Yuexiu District, Guangzhou City 34 for their contributions to patient coordination during the data collection for this study. We also 35 36 appreciate Dr. James La Nauze from Australia for his contribution to English correction and 37 revision of the manuscript. 38 39 40 —————————————— 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 18
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 22
1 2 3 References: 4 5 6 1. The World Health Organisation, Global report on diabetes (PDF, 2016), Available at: 7 https://www.who.int/publications-detail/global-report-on-diabetes. Accessed on 6th Jan 8 9 2020. 10 2. Yang SH, Dou KF, Song WJ. Prevalence of diabetes among men and women in China. N 11 Engl J Med 2010; 362(25): 2425-6. 12 3. Jonsson B. The economic Impact of Diabetes. Diabetes Care 1998; 21 (Syookenebt 13 14 3):C7-C10. 15 4. Wang FH, Liang YB, Zhang F, Wang JJ, Wei WB, Tao QS, Sun LP, Friedman DS, Wang NL, 16 Wong TY. Prevalence of diabetic retinopathy in rural China: the Handan Eye Study. 17 18 Ophthalmology 2009;For 116 (3):peer 461-7. review only 19 5. Liu L, Wu X, Liu L, Geng J, Yuan Z, Shan Z, Chen L. Prevalence of diabetic retinopathy in 20 mainland China: a meta-analysis. PloS one 2012; 7(9): e45264. 21 22 6. Liew G, Michaelides M, Bunce C. A comparison of the causes of blindness certifications 23 in England and Wales in working age adults (16-64 years), 1999-2000 with 2009-2010. BMJ 24 open 2014; 4(2): e004015. 25 7. Wang D, Ding X, He M, Yan L, Kuang J, Geng Q, Congdon N. Use of eye care services 26 27 among diabetic patients in urban and rural China. Ophthalmology 2010; 117(9): 1755-62. 28 8. Moreton RBR, Stratton IM, Chave SJ, Lipinski H, Scanlon PH. Factors determining uptake 29 of diabetic retinopathy screening in Oxfordshire. Diabet Med 2017; 34(7): 993-9. 30 31 9. Ng JQ, Morlet N. Improving the uptake of screening for diabetic retinopathy. Med J Aust 32 2013; 198(2): 69-70. 33 10. Leese GP, Boyle P, Feng Z, Emslie-Smith A, Ellis JD. Screening uptake in a 34 35 well-established diabetic retinopathy screening program: the role of geographical access and 36 deprivation. Diabetes care 2008; 31(11): 2131-5. 37 11. The website for introduction of the handheld camera Horus:www.miis.com.tw, 38 retrieved December 2019. 39 40 12. NHS Diabetic eye screening program, 41 https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_ 42 data/file/582710/Grading_definitions_for_referrable_disease_2017_new_110117.pdf 43 44 retrieved on 18th Oct 2019. 45 13. Scanlon PH, Malhotra R, Greenwood RH, Aldington SJ, Foy C, Flatman M, Downes S. 46 Comparison of two reference standards in validating two field mydriatic digital photography 47 48 as a method of screening for diabetic retinopathy. Br J Ophthalmol 2003; 87(10): 1258-63. 49 14. Scanlon PH, Malhotra R, Thomas G, Foy C, Kirkpatrick JN, Lewis-Barned N, Harney B, 50 Aldington SJ. The effectiveness of screening for diabetic retinopathy by digital imaging 51 photography and technician ophthalmoscopy. Diabetic medicine : a journal of the British 52 53 Diabetic Association 2003; 20(6): 467-74. 54 15. Karbassi M, Khu PM, Singer DM, Chylack LT, Jr. Evaluation of lens opacities classification 55 system III applied at the slitlamp. Optom Vis Sci 1993; 70(11): 923-8. 56 57 16. Yan X, Liu T, Gruber L, He M, Congdon N. Attitudes of physicians, patients, and village 58 health workers toward glaucoma and diabetic retinopathy in rural China: a focus group 59 study. Arch Ophthalmol 2012; 130(6): 761-70. 60 19
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 22 BMJ Open
1 2 3 17. Cui Y, Zhang M, Zhang L, Zhang L, Kuang J, Zhang G, Liu Q, Guo H, Meng Q. Prevalence 4 5 and risk factors for diabetic retinopathy in a cross-sectional population-based study from 6 rural southern China: Dongguan Eye Study. BMJ open 2019; 9(9): e023586. 7 18. Pan CW, Wang S, Qian DJ, Xu C, Song E. Prevalence, Awareness, and Risk Factors of 8 9 Diabetic Retinopathy among Adults with Known Type 2 Diabetes Mellitus in an Urban 10 Community in China. Ophthalmic Epidemiol 2017; 24(3): 188-94. 11 19. Nangia V, Jonas JB, George R, Lingam V, Ellwein L, Cicinelli MV, Das A, Flaxman SR, 12 Keeffe JE, Kempen JH, Leasher J, Limburg H, Naidoo K, Pesudovs K, Resnikoff S, Silvester AJ. 13 14 Prevalence and causes of blindness and vision impairment: magnitude, temporal trends and 15 projections in South and Central Asia. Br J Ophthalmol 2019; 103(7): 871-7. 16 20. Liu Y, Song Y, Tao L, Qiu W, Lv H, Jiang X, Zhang M, Li X. Prevalence of diabetic 17 18 retinopathy amongFor 13473 patientspeer with reviewdiabetes mellitus inonly China: a cross-sectional 19 epidemiological survey in six provinces. BMJ open 2017; 7(1): e013199. 20 21. Ding J, Zou Y, Liu N, Jiang L, Ren X, Jia W, Snellingen T, Chongsuvivatwong V, Liu X. 21 22 Strategies of digital fundus photography for screening diabetic retinopathy in a diabetic 23 population in urban China. Ophthalmic Epidemiol 2012; 19(6): 414-9. 24 22. Piyasena M, Yip JLY, MacLeod D, Kim M, Gudlavalleti VSM. Diagnostic test accuracy of 25 diabetic retinopathy screening by physician graders using a hand-held non-mydriatic retinal 26 27 camera at a tertiary level medical clinic. BMC Ophthalmol 2019; 19(1): 89. 28 23. Suansilpong A, Rawdaree P. Accuracy of single-field nonmydriatic digital fundus image 29 in screening for diabetic retinopathy. J Med Assoc Thai 2008; 91(9): 1397-403. 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 20
For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 22
Reported on page # 1 Section & Topic No Item 2 TITLE OR ABSTRACT 3 1 Identification as a study of diagnostic accuracy using at least one measure of accuracy Page 1 4 (such as sensitivity, specificity, predictive values, or AUC) 5 ABSTRACT 6 7 2 Structured summary of study design, methods, results, and conclusions Page 3 8 (for specific guidance, see STARD for Abstracts) 9 INTRODUCTION 10 3 Scientific and clinical background, including the intended use and clinical role of the index test Page 4 11 4 Study objectives and hypotheses Page 5 12 METHODS 13 Study design 5 Whether data collection was planned before the index test and reference standard Page 5-6 14 were performed (prospective study) or after (retrospective study) 15 Participants 6 Eligibility criteria Page 6 16 7 On whatFor basis potentially peer eligible participants review were identified only Page 6 17 18 (such as symptoms, results from previous tests, inclusion in registry) 19 8 Where and when potentially eligible participants were identified (setting, location and dates) Page 6 20 9 Whether participants formed a consecutive, random or convenience series Page 6 21 Test methods 10a Index test, in sufficient detail to allow replication Page 6-8 22 10b Reference standard, in sufficient detail to allow replication Page 6-8 23 11 Rationale for choosing the reference standard (if alternatives exist) Page 6 24 12a Definition of and rationale for test positivity cut-offs or result categories Page 9 25 of the index test, distinguishing pre-specified from exploratory 26 12b Definition of and rationale for test positivity cut-offs or result categories Page 8-9 27 of the reference standard, distinguishing pre-specified from exploratory 28 29 13a Whether clinical information and reference standard results were available Page 8-9 30 to the performers/readers of the index test 31 13b Whether clinical information and index test results were available Page 8-9 32 to the assessors of the reference standard 33 Analysis 14 Methods for estimating or comparing measures of diagnostic accuracy Page 8-9 34 15 How indeterminate index test or reference standard results were handled Page 8-9 35 16 How missing data on the index test and reference standard were handled 36 17 Any analyses of variability in diagnostic accuracy, distinguishing pre-specified from exploratory Page 9 37 18 Intended sample size and how it was determined Page 7 38 RESULTS 39 40 Participants 19 Flow of participants, using a diagram 41 20 Baseline demographic and clinical characteristics of participants Page 10-11 42 21a Distribution of severity of disease in those with the target condition Page 1 2 43 21b Distribution of alternative diagnoses in those without the target condition Page 12 44 22 Time interval and any clinical interventions between index test and reference standard Page 8 45 Test results 23 Cross tabulation of the index test results (or their distribution) Page 13-14 46 by the results of the reference standard 47 24 Estimates of diagnostic accuracy and their precision (such as 95% confidence intervals) Page 14-15 48 25 Any adverse events from performing the index test or the reference standard 49 50 DISCUSSION 51 26 Study limitations, including sources of potential bias, statistical uncertainty, and generalisability Page 18 52 27 Implications for practice, including the intended use and clinical role of the index test Page 18 53 OTHER 54 INFORMATION 55 28 Registration number and name of registry 56 29 Where the full study protocol can be accessed 57 30 Sources of funding and other support; role of funders 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 22 BMJ Open
1 STARD 2015 2 3 4 AIM 5 STARD stands for “Standards for Reporting Diagnostic accuracy studies”. This list of items was developed to contribute to the 6 completeness and transparency of reporting of diagnostic accuracy studies. Authors can use the list to write informative 7 8 study reports. Editors and peer-reviewers can use it to evaluate whether the information has been included in manuscripts 9 submitted for publication. 10 11 EXPLANATION 12 13 A diagnostic accuracy study evaluates the ability of one or more medical tests to correctly classify study participants as 14 having a target condition. This can be a disease, a disease stage, response or benefit from therapy, or an event or condition 15 in the future. A medical test can be an imaging procedure, a laboratory test, elements from history and physical examination, 16 a combination of these, or anyFor other method peer for collecting review information about onlythe current health status of a patient. 17 18 The test whose accuracy is evaluated is called index test. A study can evaluate the accuracy of one or more index tests. 19 20 Evaluating the ability of a medical test to correctly classify patients is typically done by comparing the distribution of the 21 index test results with those of the reference standard. The reference standard is the best available method for establishing 22 the presence or absence of the target condition. An accuracy study can rely on one or more reference standards. 23 24 If test results are categorized as either positive or negative, the cross tabulation of the index test results against those of the 25 reference standard can be used to estimate the sensitivity of the index test (the proportion of participants with the target 26 condition who have a positive index test), and its specificity (the proportion without the target condition who have a negative 27 index test). From this cross tabulation (sometimes referred to as the contingency or “2x2” table), several other accuracy 28 29 statistics can be estimated, such as the positive and negative predictive values of the test. Confidence intervals around 30 estimates of accuracy can then be calculated to quantify the statistical precision of the measurements. 31 32 If the index test results can take more than two values, categorization of test results as positive or negative requires a test 33 positivity cut-off. When multiple such cut-offs can be defined, authors can report a receiver operating characteristic (ROC) 34 curve which graphically represents the combination of sensitivity and specificity for each possible test positivity cut-off. The 35 area under the ROC curve informs in a single numerical value about the overall diagnostic accuracy of the index test. 36 37 The intended use of a medical test can be diagnosis, screening, staging, monitoring, surveillance, prediction or prognosis. The 38 clinical role of a test explains its position relative to existing tests in the clinical pathway. A replacement test, for example, 39 40 replaces an existing test. A triage test is used before an existing test; an add-on test is used after an existing test. 41 42 Besides diagnostic accuracy, several other outcomes and statistics may be relevant in the evaluation of medical tests. Medical 43 tests can also be used to classify patients for purposes other than diagnosis, such as staging or prognosis. The STARD list was 44 not explicitly developed for these other outcomes, statistics, and study types, although most STARD items would still apply. 45 46 47 DEVELOPMENT 48 This STARD list was released in 2015. The 30 items were identified by an international expert group of methodologists, 49 researchers, and editors. The guiding principle in the development of STARD was to select items that, when reported, would 50 51 help readers to judge the potential for bias in the study, to appraise the applicability of the study findings and the validity of 52 conclusions and recommendations. The list represents an update of the first version, which was published in 2003. 53 54 More information can be found on http://www.equator-network.org/reporting-guidelines/stard. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml