Background

JMIR Form Res

formative

JMIR Formative Research

JMIR Form Res

2561-326X

JMIR Publications

Toronto, Canada

v9i1e67047

10.2196/67047

Original Paper

Barriers and Determinants of Referral Adherence in AI-Enabled Diabetic Retinopathy Screening for Older Adults in Northern India During the COVID-19 Pandemic: Mixed Methods Pilot Study

Chauhan

Anshul

PhD1Goyal

Anju

PhD2Masih

Ritika

MPH1Kaur

Gagandeep

B Optometry1Kumar

Lakshay

B Optometry1Neha

B Optometry1Rastogi

Harsh

MSc1Kumar

Sonam

MPH1Singh

Bidhi Lord

PhD1Syal

Preeti

MS2Gupta

Vishali

MS1Vale

Luke

PhD3Duggal

Mona

MD14

Advanced Eye Centre, Post Graduate Institute of Medical Education and Research

Number 139, Kairon Block

Chandigarh

IndiaDepartment of Statistics, Panjab University

Chandigarh

IndiaNewcastle University

Newcastle

United KingdomNational Institute for Research in Digital Health and Data Sciences, Indian Council of Medical Research

New Delhi

India

Mavragani

Amaryllis

Bonilla-Escobar

Francisco

Intharah

Thanapong

Correspondence to Mona Duggal, MD, Advanced Eye Centre, Post Graduate Institute of Medical Education and Research, Number 139, Kairon Block, Chandigarh, 160012, India, 91 9560374388; monaduggal2@gmail.com

2025

3132025

e67047

300920242202202525022025

© Anshul Chauhan, Anju Goyal, Ritika Masih, Gagandeep Kaur, Lakshay Kumar, Neha, Harsh Rastogi, Sonam Kumar, Bidhi Lord Singh, Preeti Syal, Vishali Gupta, Luke Vale, Mona Duggal. Originally published in JMIR Formative Research (https://formative.jmir.org), 31.3.2025.

2025

This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included.

Background

Diabetic retinopathy (DR) is a leading cause of blindness globally. DR has increasingly affected both individuals and health care systems as the population ages.

Objective

This study aims to explore factors and identify barriers associated with nonadherence to referral recommendations among older adult participants after DR screening (DRS) during the COVID-19 pandemic.

Method

This paper presents findings from a pilot study on artificial intelligence–enabled DRS conducted in two districts in Punjab, India (Moga and Mohali) during the COVID-19 pandemic. The screenings were conducted from March to June 2022 at community health center Badhani Kalan in Moga and from March to June 2021 in community settings (homes) in Block Boothgarh, Mohali. Participants were referred to the district hospital for an ophthalmological review based on artificial intelligence–enabled screening. After 1 month, the participants were contacted by telephone to assess adherence to the referral recommendations. Participants who did not adhere to the referral were then interviewed alongside health care providers to understand the barriers explaining their nonadherence.

Results

We aimed to recruit 346 and 600 older adult participants from 2 sites but enrolled 390. Key challenges included health facility closures due to COVID-19, low motivation among health personnel for recruitment, incomplete nonparticipation data, and high participant workloads. Approximately 45% of the participants were male and 55% female. Most participants (62.6%) were between 60 and 69 years old, while 37.4% were 70 or older, with a mean age of 67.2 (SD 6.2) years. In total, 159 participants (40.8%) were referred, while 231 participants (59.2%) were not. Only 23 (14.5%) of those referred followed through and visited a health facility for ophthalmological review, while 136 (85.5%) did not pursue further evaluation. Our analysis revealed no significant differences in the characteristics between adherent and nonadherent participants, suggesting that demographic and health factors alone do not predict adherence behavior in patients with DR. Interviews identified limited knowledge about DR, logistical challenges, financial constraints, and attitudinal barriers as the primary challenges.

Conclusions

This study, conducted during the COVID-19 pandemic, showed suboptimal adherence to referral recommendations among older adult patients due to knowledge gaps, logistical challenges, and health system issues. Quantifying and understanding adherence factors are crucial for targeted interventions addressing barriers to referral recommendations after DRS. Integrating teleophthalmology into and strengthening infrastructure for artificial intelligence–enabled diabetic retinopathy screening to enhance access and outcomes.

diabetic retinopathydiabetesgerontologygeriatricoldagingagedartificial intelligenceretinopathyretinalreferralscreeningoptometryophthalmologyadherencebarriers

Introduction

Diabetic retinopathy (DR) is a prevalent and severe microvascular complication of diabetes mellitus (DM) [1]. In India, 10.9% (7.2%‐16.3%) of individuals aged 65 years and above with diabetes have DR, with 2.3% (1.2%‐4.4%) suffering from vision-threatening diabetic retinopathy (VTDR) [1], characterized by severe retinopathy or macular edema [2]. DR is typically asymptomatic in its early stages, and it can lead to visual impairment or blindness if left untreated [3]. The rate of blindness due to VTDR is expected to rise proportionately to the exponential increase in DM prevalence [4,5]. Visual impairment due to DR increases by 9.6% in individuals over 50 [6], emphasizing advancing age as a significant risk factor [7]. Individuals with VTDR face greater difficulties in vision-related tasks, such as reading, watching television, and driving [8]. Additionally, people with DR have a higher risk of falls, resulting in reduced mobility, social isolation, and increased physical dependence [9].

India is set to become home to the world’s second-largest older adult population, with the number of people over 60 years of age projected to rise from 100 million in 2013 to 198 million by 2030 [10]. The National Programme for Health Care of the Elderly in India advocates for accessible, affordable, high-quality, long-term, comprehensive care services tailored to the needs of an aging population [10]. Significant scientific evidence shows that early screening and timely treatment referral can prevent most visual loss caused by DR [11]. Conventionally, DR screening (DRS) includes fundus (retina) examination by ophthalmologists or color fundus photography using conventional cameras (mydriatic or nonmydriatic) conducted by trained eye technicians or optometrists [12]. However, the sharp rise in diabetes cases, coupled with a shortage of trained retinal specialists and ophthalmologists [13], makes DRS services accessibility challenging, particularly as 80% of India’s older adult population resides in rural areas [10]. The COVID-19 pandemic posed additional significant challenges to health care systems worldwide, inevitably leading to the curtailment of health services accessibility, including DRS [14,15].

Automated computer-based analysis of fundus images could ease the strain on health care systems by streamlining DRS and offering a more efficient management solution [13]. Artificial intelligence (AI) systems have reduced costs, enhanced diagnostic precision, and expanded patient accessibility to DRS services [16] and immediate AI-supported feedback on referral status was linked to higher referral adherence [17].

Previous studies have demonstrated telemedicine-based DRS programs’ effectiveness, scalability, and sustainability in improving accessibility and reducing disparities [18]. The COVID-19 pandemic further emphasized the need for AI-enabled DRS and teleconsultations to standardize workflows and reduce human interactions to mitigate transmission risks [19,20].

These studies are particularly relevant for designing and implementing effective AI-enabled DRS in India, where challenges of health care accessibility and resource constraints exist [20]. Suboptimal adherence to follow-up recommendations impacts treatment outcomes and can undermine even the most effective treatment options [21]. A multicenter analysis showed a decline in laser procedures and vitrectomies for retinal detachment repair due to reduced clinic visits during the pandemic [22]. Patients faced barriers to follow-up eye exams, including transportation issues, high costs, and long wait times [23]. Programs such as a point-of-care DR examination program highlight the potential of telemedicine in reducing barriers to DRS, though adherence to follow-up recommendations after screening remains a major challenge [24].

However, some factors (individual, health) influencing adherence and nonadherence to referral services remain unknown [25]. During the COVID-19 pandemic, patients were prompted to limit hospital visits, minimize time spent there, and reduce direct contact with health care providers (HCPs). Limited research exists in India on older adults’ adherence to DRS referral recommendations and the barriers influencing their decisions.

This study aims to (1) evaluate the referral adherence rates among older adult participants after AI-enabled DRS and identify the key factors affecting adherence during the COVID-19 pandemic and (2) identify and analyze the barriers to adherence to referral recommendations during the COVID-19 pandemic. The overall study design is described in Figure 1.

Figure 1.

Overall study design. AI: artificial intelligence; DRS: diabetic retinopathy screening.

MethodsStudy Setting and ParticipantsQuantitative

This paper presents findings from 2 districts in Punjab, India: Moga and Mohali. Screenings in Moga occurred at the community health center (CHC) Badhani Kalan from March to June 2022, while in Mohali, they were conducted in community settings (home) in Block Boothgarh from March to June 2021. The study included older adults [10] aged 60 years and above with a history of DM at both sites. Participants without a history of DM, those unwilling to provide informed consent, and those with a history of intraocular surgery, conjunctivitis, red eye, injury, or any eye inflammation were excluded.

A nurse at the medicine clinic at the CHC, Badhani Kalan, in district Moga, identified the elderlyolder adult participants with DM for DRS. In district Mohali, the list of diabeticparticipants with diabetes was obtained from thePa primary Hhealth Ccentere (PHC), in Boothgarh, and approached with the help of the village’s Accredited Social and Health Activists (ASHA) worker. The participants agreeing to undergo DRS received an appointment card, and a reminder telephone call was placed a day before the screening to confirm their attendance at home.

The repeated lockdowns during the COVID-19 pandemic disrupted health care services, significantly increasing the burden on health systems. Consequently, recruitment efforts were hindered, making it difficult to reach the target sample sizes of 348 at CHC Badhani Kalan and 600 at Block Boothgarh. Since the populations shared similar health systems and implementation challenges, the data from both sites were pooled and presented. Consequently, data pooling enhanced sample size and statistical power and effectively addressed research questions [26].

Qualitative

The in-depth interviews were conducted with 2 distinct categories of participants: older adults and HCPs. Older adult participants who did not adhere to the recommended referral instructions were purposely sampled and approached for in-depth interviews to understand their barriers to referral adherence after DRS. The nurse at the ophthalmology clinic invited older adult participants to the CHC for in-depth interviews, while ASHA workers contacted older adult participants in the community to conduct interviews at homes.

The second group of participants, HCPs, were purposely selected according to their roles in diabetes and DR diagnosis, referral, and treatment (Table S1 in Multimedia Appendix 1). The inclusion criteria were having at least 3 years of experience working in public health care settings and a willingness to describe patient experiences and barriers to referral adherence (Table S2 in Multimedia Appendix 1). The HCPs included ASHAs from the community settings [27], Community Health Officers (CHOs) from health and wellness centers [28], optometrists from a CHC [29] , ophthalmologists from the district hospital [30], and retina specialists from the retina unit of a tertiary eye care hospital.

Semistructured interviews were conducted to explore participants perspectives on the barriers to adhering to referral recommendations following AI-enabled DRS. The in-depth interview guides were developed in English, informed by existing literature on barriers to DR screening and referral [31-34], and later translated into Hindi and Punjabi (Table S3 in Multimedia Appendix 1). These guides, featuring open-ended questions and probes, explored participants’ perspectives on referral adherence after DRS. Pilot interviews with 2 people with DM, an ASHA, and a CHO led to revisions for clarity, replacing ambiguous terms with simpler language for better participant understanding.

Ethical Considerations

Written informed consent was obtained from all participants, with both studies approved by the Institutional Ethics Committee of the Post Graduate Institute of Medical Education and Research, Chandigarh, India (PGI/IEC/2020/000741 and PGI/IEC/2020/001342). To ensure patient privacy and confidentiality, all images and associated metadata were de-identified prior to analysis. The studies adhered to the Declaration of Helsinki guidelines, and the study protocol is registered with the Clinical Trials Registry, India (CTRI 2022/10/046283 and 2022/10/046185).

Hardware and Software for DRSHardware

The data were collected from 2 locations across districts in Punjab, each employing a different DRS methodology. Health facility–based screening at CHCs requires desktop nonmydriatic fundus cameras, while community screenings (home) rely on portable cameras for enhanced accessibility.

Fundus Benchtop Camera

The screening at the CHC Badhani Kalan utilized the Forus 3 Nethra classic (benchtop) camera, supported by a laptop running MS Windows 10 or higher, featuring a 64-bit operating system and an i3 10th-generation processor [35,36].

Smartphone-Based Fundus Camera

The lightweight, portable Fundus on Phone Non Mydriatic 10 (Remidio) handheld device, with a power backup, was used for screening in community settings in Block Boothgarh [37].

Software

Both sites used validated AI screening algorithms for DRS. At the CHC, an online DRS algorithm (Revelo) was integrated into the benchtop fundus camera [38]. A camera-integrated offline AI algorithm (Medios) was used for community-based screening [37,39].

Data CollectionFundus Photography ProtocolTraining Protocol

Before patient recruitment, 2 optometrists underwent 15 days of 1-on-1 training at the Advanced Eye Centre of Post Graduate Institute of Medical Education and Research, Chandigarh, and a primary health care center in Block Boothgarh of district Mohali. The training covered recruitment and consenting procedures, history recording, retinal image acquisition, troubleshooting poor-quality images, automated grading with the AI system, and delivering grading reports to patients. Training and supervision continued until staff were comfortable independently collecting data and performing imaging.

Image Acquisition and Grading

Adequate ventilation was maintained for patient consultations and fundus photography. Optometrists wore N95 masks and followed strict hand hygiene. Patients without masks were required to wear one during interactions and in waiting areas. Optometrists obtained nonmydriatic macula and disc-centered image color fundus photographs at 45° field of view using low-cost cameras (Figure 1) for participants at both sites. The screening at the CHC was carried out in an ophthalmology clinic. On the day of screening, the optometrists were assisted by ASHA workers at the homes of the eligible participants. The AI algorithms provided results at both sites. Two human graders, a certified optometrist and an ophthalmologist, graded all the fundus images. A senior retina specialist resolved any disagreements between the graders. All the classifications of DR were based on the International Classification for Diabetic Retinopathy [40].

Referral and Telephonic Follow-Up

Participants received an AI-generated report detailing their DR status, severity, and referral recommendations. Patients diagnosed with moderate non-proliferative DR and above were referred to as referable DR [41] and, with ungradable images, were advised to visit the district hospital for an ophthalmological review. One month after the screening, participants were followed up with telephone calls to confirm adherence and obtain diagnosis details. Each participant was contacted thrice, on 3 consecutive days, before being classified as nonresponsive. A structured questionnaire was administered to participants via telephone. The questionnaire assessed factors influencing appointment attendance and treatment adherence, with responses recorded as yes, no, or do not know for each question [42,43]. The questionnaire assessed sociodemographic characteristics and perceived barriers to referral (knowledge, attitude, logistics) and included an open-ended question for unlisted reasons behind incomplete referrals.

In-Depth Interviews

Two research fellows (AC and HR) with master’s degrees and over 5 years of qualitative research experience conducted the in-depth interviews in the participant’s preferred languages (Hindi, Punjabi, and English). Participants with DM were interviewed in quiet hospital locations at the CHC or at home in a properly ventilated space while HCPs were interviewed face-to-face or via Zoom, based on availability. Study procedures were explained, and informed consent was obtained, either written for in-person interviews or verbal for Zoom sessions. Each interview was audio recorded and lasted for 35‐40 minutes. The final sample size for the qualitative study was determined based on the principle of data saturation [44,45].

Data Analysis

Using a mixed methods approach, the results from both the quantitative and qualitative components were compared, synthesized, and discussed.

Quantitative

All participant data were entered into Research Electronic Data Capture (REDCap) [46], exported in Excel format, cleaned, and subsequently analyzed using Stata/IC (version 14.2; StataCorp LLC). The analysis involved reporting the descriptive statistics. Histograms were used to assess the normality of continuous variables. Summary statistics include frequencies (percentages) for categorical variables and means (standard deviations) for normally distributed continuous variables; otherwise, medians and interquartile ranges were reported. A chi-square (Χ²) test was conducted to compare the distribution of categorical variables (eg, gender, education, marital status) between the adherent and nonadherent participant groups.

Univariate logistic regression was used to estimate odds ratios (ORs) and 95% CIs for the association between each independent variable (eg, gender, age, education, occupation, marital status, income, health insurance, duration of diabetes, hypertension status, and previous eye facility visits) and the dependent variable (referral adherence status: adherent versus nonadherent). The multivariate logistic regression model included independent variables with a P value <.05 in univariate analysis to adjust for potential confounding factors and identify independent predictors of referral nonadherence, using a significance level of P<.05 for the final model.

Qualitative

The qualitative researchers proficient in Hindi and Punjabi transcribed the recordings verbatim, translated them into English, and removed all personal identifiers before analysis. We analyzed the interview transcripts using thematic analysis, following Braun and Clarke’s 6-phase approach [47]. Although the phases progressed logically, the process was not linear but recursive and iterative, involving movement back and forth between phases as needed [47]. We systematically organized the data, coded it, and grouped similar codes into themes to identify barriers to referral adherence. The research team members (AC: research scholar; RM: research associate; HR: research scholar) performed independent open transcript coding. MD, an experienced public health specialist, reviewed and confirmed the initial coding framework. Coding and categorization under themes were done using Atlas.Ti 23 software.

We then began extracting relevant portions of text from each interview related to the categories, sorting and selecting quotes and placing them under the appropriate themes. Quotes were edited for readability to retain their original meaning; ellipses (...) show removed text. The participants are denoted in italics with a unique ID number and their group designator: P (people with DM), Opt (Ophthalmologists), ASHA (ASHA workers), CHO (Community Health Officer), Optom (Optometrists), and RS (Retina Specialist). Given that both study sites are in Punjab and share similar health system building blocks [48], the in-depth interviews were analyzed and presented collectively.

ResultsQuantitativeParticipant Characteristics

A total of 390 (41%) of 948 older adult participants were enrolled across both sites, with 176 (45%) males and 214 (55%) females, primarily due to COVID-19 restrictions and participant refusals. Participants aged 60‐69 years comprised 62.6244% of the sample, while those aged ≥70 years constituted 37.4%, with a mean age of 67.2 (SD 6.2) years. Around 48.5% had no formal education, 45.9% had education up to the 10th standard, and 5.6% had education at the 12th standard or higher. Occupationally, 69.7% were unemployed or engaged in home duties, 13.8% were retired, and 16.5% were in other categories. Marital status indicated that 74% were married. Additionally, 69% had a monthly household income below 30,000 Indian rupees (INR; approximately US $350), 52.3% had a history of hypertension, and the mean DM duration was 8.03 (SD 6.9) years. Site-wise demographic details are available in Table S4 in Multimedia Appendix 1.

Referral Recommendations and Adherence Rates

Out of 390 screened participants, 231 (59.2%) were not referred, while 159 (40.8%) were referred for further review and management. Among those referred, only 23 (14.5%) followed the advice and visited a health facility (details in Table S5 in Multimedia Appendix 1 ), while 136 (85.5%) did not adhere to the referral recommendations or seek further medical attention.

Differences Between Referral Categories (Adherent and Nonadherent)

Table 1 shows the demographic and medical characteristic differences between adherent and nonadherent participants in the referral category. The comparison of 159 referral participants showed no statistically significant differences in demographics and health characteristics between adherent (n=23) and nonadherent (n=136) groups (all P>.05). This includes gender distribution (P=.23), age groups (P=.48), educational levels (P=.27), occupational status (P=.33), marital status (P=.97), household income (P=.71), health insurance coverage (P=.93), duration of DM (P=.51), presence of hypertension (P=.16), and previous eye facility visits (P=.45).

Table 1.

Difference in demographics and health characteristics between adherent and nonadherent participants in the referral category.

Variable	Referral category (n=159)
	Adherent (n=23)	Nonadherent (n=136)	P value
Gender, n (%)			.23
Male	12 (52.7)	53 (39)
Female	11 (47.8)	83 (61)
Age group (years), n (%)			.48
60‐69	14 (60.8)	72 (53)
≥70	9 (39.1)	64 (47)
Education, n (%)			.27
No formal education	8 (34.8)	72 (53)
Up to 10th standard	14 (60.8)	59 (43.4)
12th standard and above	1 (4.4)	5 (3.6)
Occupation, n (%)			.33
Retired from service	4 (17.4)	15 (11)
Unemployed/home duties	14 (60.9)	103 (75.8)
Other	5 (21.7)	18 (13.2)	.97
Marital status, n (%)
Married	16 (69.5)	94 (69)
Unmarried, divorced, separated, widow, widower	7 (30.4)	42 (31)
Monthly household income (INR)^a, n (%)			.71
<30,000	15 (65.2)	94 (69)
>30,000	8 (34.8)	42 (31)
Health insurance, n (%)			.93
Yes	8 (34.8)	46 (33.8)
No	15 (65.2)	90 (66.2)
Duration of DM^b (years), n (%)			.51
0‐10	17 (73.9)	91 (67)
≥10	6 (26.1)	45 (33)
Hypertension (years), n (%)			.16
Yes	15 (65.2)	67 (49.2)
No	8 (34.8)	69 (50.8)
Previous visit to the eye facility, n (%)			.45
Once	19 (82.6)	120 (88.2)
Never	4 (17.4)	16 (11.8)

^aINR: Indian rupee. 30,000 INR=US $350.

^bDM: diabetes mellitus.

Analysis of Determinants for Nonadherence to Referral Recommendations

Table 2 compares determinant factors between 23 adherent and 136 nonadherent participants using univariate and multivariate logistic regression analyses. None of the variables showed statistically significant differences (all P>.05), indicating no evidence of associations with adherence status across the examined factors. Gender (female) showed nonsignificantly higher odds of nonadherence, for univariate (OR 1.7, 95% CI 0.7-4.2; P=.23) and multivariate (OR 1.07, 95% CI 0.31-3.72; P=.91). Age ≥70 years had univariate and multivariate ORs of 1.4, 95% CI 0.56-3.4 (P=.48) and 1.4, 95% CI 0.5-4.09 (P=.47), respectively, indicating no significant association with adherence. Similarly, no significant differences were found in occupation categories, marital status, household income, health insurance status, duration of DM, hypertension, or previous eye facility visits (all P>.05).

Table 2.

Univariate and multivariate analysis of the determinants for nonadherence to referral recommendations. Reference: Base category used for comparison.

Variable	Adherent (n=23)		Nonadherent (n=136)
	Univariate OR^a (95% CI)	P value	Multivariate OR (95% CI)	P value
Gender, n (%)
Male	Reference		Reference
Female	1.7 (0.7-4.2)	.23	1.07 (0.31-3.72)	.91
Age group (years)
60‐69	Reference		Reference
≥70	1.4 (0.56-3.4)	.48	1.4 (0.5-4.09)	.47
Education, n (%)
No formal education	Reference		Reference
Up to 10th standard	0.47 (0.18-1.19)	.11	0.49 (0.16-1.45)	.20
12th and above	0.5 (0.06-5.36)	.61	0.57 (0.04-7.40)	.67
Occupation, n (%)
Retired	Reference		Reference
Unemployed/home	1.9 (0.56-6.7)	.28	1.98 (0.38-10.14)	.41
Other	0.9 (0.21-4.2)	.95	0.98 (0.17-5.6)	.98
Marital status, n (%)
Married	Reference		Reference
Unmarried, divorced, separated, widow, widower	1.02 (0.39-2.67)	.96	0.68 (0.23-1.97)	.48
Monthly household income^b, n (%)
<30,000	Reference		Reference
>30,000	0.84 (0.33-2.12)	.71	0.94 (0.32-2.77)	.92
Health insurance
Yes	Reference		Reference
No	1.3 (0.55-3.4)	.50	0.71 (0.24-2.03)	.52
Duration of diabetes mellitus (years)
0‐10	Reference		Reference
≥10	1.4 (0.52-3.8)	.51	1.56 (0.52-4.68)	.42
Hypertension (years)
Yes	Reference		Reference
No	1.9 (0.76-4.8)	.10	2.09 (0.79-5.48)	.13
Previous visit to the eye facility
Once	Reference		Reference
Never	0.63 (0.19-2.09)	.46	0.54 (0.14-2.04)	.37

^aOR: odds ratio.

^bINR: Indian rupee. 30,000 INR=US $350.

Telephone Follow-Up: Assessing Referral Adherence Rates and Reasons

Participants with ungradable images or diagnosed with referable DR (n=136) were contacted by phone 1 month after screening. The most commonly reported barriers were categorized as other factors (n=70, 51.5%), logistical challenges (n=31, 22.8%), and attitudinal issues (n=27, 19.9%).

Regarding attitude, 17 (12.5%) participants believed their eyes were fine, while logistical obstacles included lack of family support (n=21, 15.4%) and family problems (n=8, 5.8%). Financial constraints, particularly high treatment costs, affected 4 (2.9%) participants. Additionally, 4 (2.9%) were unaware of the importance of DR treatment. Seasonal harvesting (n=14, 10.3%) and other health issues (n=9, 6.6%) were also contributing factors (Table 3).

Table 3.

Reasons for nonadherence to referral recommendations.

Categories and barriers to referral adherence		Participants (n=136), n (%)
Attitude
	Eyes are fine	17 (12.5)
	Not willing to treatment	3 (2.2)
	Lack of concern about diabetic retinopathy	3 (2.2)
	Not interested in treatment	4 (2.9)
Financial
	High treatment cost	4 (2.9)
Awareness
	Not aware that DR treatment is important	4 (2.9)
Logistical support
	Lack of family support	21 (15.4)
	Travel distance to a health facility	2 (1.6)
	Family problems	8 (5.8)
Other
	Harvesting seasons	14 (10.3)
	Work commitments	4 (2.9)
	Out of station	6 (4.4)
	Phone with family members	3 (2.2)
	Planning to visit a health facility soon	2 (2.2)
	Other health issues	9 (6.6)
	Weather conditions	1 (0.7)
	Technical error (wrong contact numbers, switch-offs, unavailability)	31 (22.8)

QualitativeCharacteristics of the Participants

In total, 28 in-depth interviews were conducted with 9 older adult people with DM and 19 HCPs. Of the 9 people with DM, 4 (44%) were male and 5 (56%) were female. Most (n=7, 78%) were aged 60‐69 years, and 8 (89%) were married. Two-thirds (n=6) lacked formal education, and 6 (67%) were unemployed or engaged in home duties (Tables S6 and S7 in Multimedia Appendix 1). The HCP group included 1 retina specialist (5%), 4 ophthalmologists (21%), 2 optometrists (10%), 3 CHOs (16%), and 9 ASHA workers (47%). Among 19 health care providers, 13 (68%) were female and 6 (32%) were male. Most (n=12, 63%) were aged 31‐40 years.

Barriers Associated With Low Referral AdherenceOverview

The data revealed three key themes that best explained the barriers to referral adherence: (1) awareness and knowledge-related obstacles, (2) logistical support challenges, and (3) health care system limitations. The results of the in-depth interviews with the stakeholders complemented the quantitative findings of nonadherent participants and shed further light on their findings (Figure 2). The thematic analysis suggested a range of factors that impact adherence to referral instructions and put low adherence or nonadherence into perspective. The quotes under the categories are available in Table S8 in Multimedia Appendix 1.

Figure 2.

Barriers to referral nonadherence. DR: diabetic retinopathy; HF: health facility; VEGF: vascular endothelial growth factor.

Awareness and Knowledge-Related Barriers

The HCPs observed that patients with DR often delayed treatment due to a lack of awareness about the condition and its complications, frequently mistaking vision changes for aging, cataracts, or simply needing new glasses. These misconceptions often lead patients to request remedies such as eye drops or new glasses, mistakenly believing these will resolve their vision problems. One patient recounted an interaction with fellow patients who had undergone extensive treatments and incurred substantial expenses without experiencing any improvement in vision. This perception highlights a common gap in patient knowledge. Individuals often rely excessively on peer anecdotal evidence, hindering adherence to evidence-based medical treatments. This issue is further complicated by a limited awareness of the crucial role regular follow-up appointments play in managing DR, resulting in poor compliance with treatment protocols. Additionally, some patients tend to avoid health care facilities under the false assumption that their ocular condition has stabilized while engaging in self-treatment, which introduces significant risks and compromises effective disease management.

Logistical Support BarriersMobility and Transportation

The HCPs said that the lack of public transport restricts participants from accessing local transport options for treatment, making it hard for patients to reach the facility for their appointments. One patient with a 2-wheeler reported difficulty driving alone because of their low vision. Longer distances to treatment facilities also decreased the frequency of visits, resulting in higher dropout rates despite initial engagement in treatment. Sometimes, multiple health issues can limit mobility and hinder access to eye care services, leading individuals to prioritize managing other health conditions over maintaining regular DR treatment appointments. This preference reflects a practical decision based on immediate health needs and logistical challenges posed by mobility restrictions rather than a lack of recognition of the importance of DR care.

Financial

Cost presents a formidable barrier, restricting many from affording regular travel to treatment centers. Initially, patients may adhere to treatment plans diligently, but as time progresses, dropout rates escalate. A patient articulated their predicament: “I was referred to another hospital for injections, which are financially out of reach due to the distance from home.” The ophthalmologist and retina specialist underscored the substantial financial burden of DR treatment, specifically anti-VEGF (vascular endothelial growth factor) injections, which significantly impacts adherence. Financially disadvantaged patients frequently prioritize other expenses over necessary visits to eye care facilities, thereby exacerbating the challenge of managing DR effectively and consistently. This underscores the critical need for accessible and affordable eye care solutions to mitigate the economic barriers hindering patient care and treatment adherence.

Health Care System BarriersBurdened Health System

The tertiary eye care hospital experiences high demand, resulting in delayed appointments that led many patients to discontinue treatment after 1-2 visits. Long queues also discourage patients from seeking care at larger eye care centers, underscoring the systemic challenges that affect access to DR treatment services. The large number of patients and extended waiting times pose significant challenges for individuals trying to attend appointments and receive timely treatment.

Technical Challenges

Telemedicine facilities are beneficial for extending eye care access [49], but they face implementation challenges. The limited availability of specialized doctors and unreliable internet in peripheral areas hinder teleconsultations. Village residents encountered prolonged wait times at telemedicine centres and expressed reluctance to wait for consultations with doctors at these facilities.

DiscussionPrincipal Findings

This study investigated barriers to referral adherence after AI-enabled DRS at primary (community) and secondary (CHC) sites among those who did not follow referrals to district hospitals for further management. We implemented AI-enabled DRS programs for older adults in public health settings in Punjab, India, during the COVID-19 pandemic. AI has been promoted as a transformative tool in health care to help attain health and health-related Sustainable Development Goals among older adults [50,51]. However, the impact of implementing AI on patient outcomes depends on many factors beyond diagnostic accuracy, including care access, referral adherence, and the implementation of treatment and management recommendations [52]. In our study, only 23 of 159 (14.5%) patients screened adhered to referral recommendations. Of these, only 9 visited the district hospital, while the others chose different facilities, indicating various factors influencing their decision-making regarding referral locations [53]. This finding contrasts with a study conducted in South India examining the factors influencing the utilization of referral services from secondary to tertiary eye care, which reported noncompliance rates of 31.65% and 24.2%, respectively [33].

Telephone interviews identified key barriers in our study, including low awareness about DR (n=4, 2.9%), attitudinal issues (n=27, 19.9%), and a lack of family support (n=21, 15.4%). Economic 27 (16.4%) and attitudinal barriers (44.2%) to referral service access were consistent with findings by Padhy et al [33] Our analysis showed no differences in the characteristics of adherent and nonadherent participants. However, nonstatistically significant differences between adherent and nonadherent groups do not imply that these factors are irrelevant or that there are no relationships. This indicates that we did not detect significant associations with the current sample size and variability. However, the wide confidence intervals suggest additional quantitative data are needed to improve estimate precision.

Nevertheless, we have no evidence that demographic and health variables alone are sufficient to understand or predict adherence behavior in patients with DR. Technical errors (Table 3) prevented us from contacting 31 of 136 (22.8%) participants, potentially impacting our quantitative analysis, including the univariate and multivariate analysis results. The unprecedented COVID-19 pandemic may have also negatively affected health-seeking behavior and treatment adherence [19,22].

We prioritized incorporating insights from key stakeholders through qualitative, in-depth interviews to guide informational power. Our study found that older adult individuals lack adequate awareness about DR complications, resulting in misconceptions about the causes of vision changes. Lack of awareness discourages service use, enabling unchecked harmful conditions and potentially leading to adverse outcomes due to delayed care-seeking. Further, the participants encountered difficulties affording transportation and treatment expenses and taking time off work for medical appointments, driven by concerns about their daily income [54]. Additionally, attitudes such as believing their eyes are healthy and logistical challenges like lack of family support impeded referral adherence [23]. Reliance on anecdotal evidence (do not go for eye treatment) from peers discouraged adherence to referral recommendations. Local and regional factors, such as the peak harvesting season (Table 3), hindered older adult adherence to referral recommendations (n=14, 10.3%). In-depth stakeholder interviews complemented the telephone follow-up findings, revealing barriers to referral recommendations. Similar programs, like the point-of-care DR examination program initiative, reported suboptimal follow-up rates despite targeted interventions, highlighting the need for more personalized strategies to enhance adherence [24]. Community-based programs have proven effective in enhancing access to eye care and improving patient adherence to continuity of care recommendations [24,55]. Bonilla-Escobar et al [24] showed higher adherence to referral recommendations when personalized approaches, such as phone calls, voicemails, and result letters with appointment details, were used. Transportation interventions, including bus passes, taxi vouchers, reimbursed travel, and tailored services, enhance health outcomes in chronic diseases, particularly for older adults and women [56].

Teleophthalmology, for instance, has accelerated health care access in rural and remote areas [57], overcoming significant travel constraints, time constraints, and economic barriers through community-based program initiatives [58,59]. However, the HCPs noted that technical challenges, such as underdeveloped internet infrastructure in rural areas, pose challenges to implementing teleophthalmology programs [60]. Hence, it is critical to identify methods for establishing DRS programs tailored to available resources and health care infrastructure [24,61]. In our study, older adult participants highlighted financial constraints as barriers to accessing referral services, a sentiment corroborated by the HCPs’ in-depth interviews. The findings align with existing literature on economic challenges among individuals with visual impairment in developing countries [62]. The high costs of DR treatment often compel financially disadvantaged patients to prioritize other essential expenses over necessary eye care. Patients and HCPs should be informed about the comprehensive coverage offered by the Ayushman Bharat Pradhan Mantri Jan Arogya Yojana public health insurance program in India, including DRS and treatment. This ensures adequate access to necessary services through insurance benefits [63].

Our study findings revealed issues in the care process in rural areas, such as patients self-referring and bypassing the referral system hierarchy [64]. Systemic challenges such as delayed appointments and long queues discourage patients from continuing treatment beyond 1 or 2 visits due to access issues [23]. The World Report on Vision has recommended an integrated care model incorporating primary, secondary, and tertiary care, ensuring clear roles and appropriate referrals at each health care system level [65].

This study’s strength lies in triangulating quantitative findings with qualitative insights to address research questions regarding referral adherence. Additionally, conducting the study across primary (community) and secondary (CHC) health care systems provides comprehensive information on referral adherence within a network of public health systems.

A limitation of this study is its focus on a public health network in North India within one state, potentially limiting the generalizability of its findings to broader settings. Potential biases from self-reported data during telephone follow-ups may lead to underreporting or overreporting adherence based on perceived expectations and limited generalizability due to the study’s regional focus. The study was conducted during the COVID-19 pandemic, marked by significant disruptions to health care systems and changes in patient behavior. These unique circumstances, including the impact of COVID-19 on decision-making regarding nonemergency medical care, may have introduced factors related to health care access that are specific to this time. Furthermore, nearly 22.8% of screened patients (Table 3) were not successfully contacted despite multiple attempts, which would increase the imprecision of estimates yielded by the regression analysis. For example, only 23 adherent participants contributed data to the multivariate model and therefore the analyses may be hindered due to issues related to statistical power and imbalanced group sizes [66]. This study did not extensively collect detailed information on health care system factors affecting adherence, such as hospital waiting times, hospital accessibility, and staff attitudes [33]. This highlights the necessity of exploring additional factors that influence adherence, including psychological, social, and systemic barriers [23,67].

In this pilot study, approximately 390 of 948 (41%) participants were recruited due to COVID-19 pandemic restrictions. Additionally, the limited sample size in the nonadherent group and high variability in the collected data resulted in insufficient statistical power to accurately identify determinants of nonadherence, leading to large confidence intervals and reduced precision in the estimates.

Conclusion and Policy Implications

This study, conducted during the COVID-19 pandemic, has demonstrated suboptimal adherence among older adult patients to referral recommendations. Participants considered barriers such as knowledge gaps, logistical challenges, and health system–related issues to have hindered referral adherence. Given the unique context of the pandemic, the results should be generalized with caution. Future studies should consider pragmatic trial study designs incorporating implementation science approaches using mixed methods to examine barriers to referral and adherence across diverse contexts. This could include exploring community beliefs and other social determinants influencing service uptake. Policy-level changes are essential to integrating teleophthalmology into public health programs and improving infrastructure for AI-enabled DRS, enhancing access and outcomes. An effective, well-coordinated referral system with defined protocols, streamlined communication, reliable transportation, trained personnel, integrated systems, and collaboration may address the challenges in the referral systems [68,69].

The authors sincerely thank all participants for their invaluable contributions to this research and the Department of Health and Family Welfare, Punjab, India, for granting permission to conduct the study. All authors declared that they had insufficient funding to support open access publication of this manuscript, including from affiliated organizations or institutions, funding agencies, or other organizations. JMIR Publications provided article processing fee (APF) support for the publication of this article.

None declared.

Abbreviations

ASHA

Accredited Social and Health Activists

CHC

community health center

CHO

community health officer

diabetes mellitus

diabetic retinopathy

DRS

DR screening

HCP

health care provider

VTDR

vision-threatening diabetic retinopathy

References1

Raman

Vasconcelos

Rajalakshmi

Prevalence of diabetic retinopathy in India stratified by known and undiagnosed diabetes, urban-rural locations, and socioeconomic indices: results from the SMART India population-based cross-sectional screening study

Lancet Glob Health2022121012e1764e1773

10.1016/S2214-109X(22)00411-9

36327997

Ting

DSW

Cheung

GCM

Wong

Diabetic retinopathy: global prevalence, major risk factors, screening practices and public health challenges: a review

Clin Exp Ophthalmol201605444260277

10.1111/ceo.12696

26716602

Vujosevic

Aldington

Silva

Screening for diabetic retinopathy: new perspectives and challenges

Lancet Diabetes Endocrinol20200484337347

10.1016/S2213-8587(19)30411-5

32113513

IDF Diabetes Atlas2025-03-26

www.diabetesatlas.org

Teo

Tham

Global prevalence of diabetic retinopathy and projection of burden through 2045: systematic review and meta-analysis

Ophthalmology2021111281115801591

10.1016/j.ophtha.2021.04.027

33940045

Flaxman

Bourne

RRA

Resnikoff

Global causes of blindness and distance vision impairment 1990-2020: a systematic review and meta-analysis

Lancet Glob Health201712512e1221e1234

10.1016/S2214-109X(17)30393-5

29032195

Leley

Ciulla

Bhatwadekar

Diabetic retinopathy in the aging population: a perspective of pathogenesis and treatment

Clin Interv Aging20211613671378

10.2147/CIA.S297494

34290499

Lamoureux

Tai

Thumboo

Impact of diabetic retinopathy on vision-specific function

Ophthalmology2010041174757765

10.1016/j.ophtha.2009.09.035

20122736

Fenwick

Rees

Pesudovs

Social and emotional impact of diabetic retinopathy: a review

Clin Exp Ophthalmol20124012738

10.1111/j.1442-9071.2011.02599.x

21575125

National Programme for Health Care of the Elderly

Ministry of Health and Family Welfare2025-03-26

https://mohfw.gov.in/?q=major-programmes/Non-Communicable-Diseases/Non-Communicable-Diseases-1

Liew

Michaelides

Bunce

A comparison of the causes of blindness certifications in England and Wales in working age adults (16-64 years), 1999-2000 with 2009-2010

BMJ Open2014021242e004015

10.1136/bmjopen-2013-004015

24525390

Ryan

Rajalakshmi

Prathiba

Comparison Among Methods of Retinopathy Assessment (CAMRA) study: smartphone, nonmydriatic, and mydriatic photography

Ophthalmology2015101221020382043

10.1016/j.ophtha.2015.06.011

26189190

Padhy

Takkar

Chawla

Kumar

Artificial intelligence in diabetic retinopathy: a natural step to the future

Indian J Ophthalmol20190767710041009

10.4103/ijo.IJO_1989_18

31238395

Abbasi

The scandals of covid-19

BMJ202004369m1434

10.1136/bmj.m1434

Shih

Kwong

ASK

Wang

JHL

Diabetic retinopathy screening during the coronavirus disease 2019 pandemic

Eye (Lond)20200734712461247

10.1038/s41433-020-0928-7

32367007

Grzybowski

Brona

Lim

Artificial intelligence for diabetic retinopathy screening: a review

Eye (Lond)202003343451460

10.1038/s41433-019-0566-0

31488886

Mathenge

Whitestone

Nkurikiye

Impact of artificial intelligence assessment of diabetic retinopathy on referral service uptake in a low-resource setting: the RAIDERS randomized trial

Ophthalmol Sci20221224100168

10.1016/j.xops.2022.100168

36531575

Bonilla-Escobar

Ghobrial

Gallagher

Eller

Waxman

Comprehensive insights into a decade-long journey: the evolution, impact, and human factors of an asynchronous telemedicine program for diabetic retinopathy screening in Pennsylvania, United States

PLoS One2024197e0305586

10.1371/journal.pone.0305586

38995899

Ahmed

Liu

TYA

The impact of COVID-19 on diabetic retinopathy monitoring and treatment

Curr Diab Rep2021098211040

10.1007/s11892-021-01411-6

34495377

Kumar

The transformation of the Indian healthcare system

Cureus202305155e39079

10.7759/cureus.39079

37378105

Burkhart

Sabaté

Adherence to long-term therapies: evidence for action

J Nurs Scholarsh2003353207

14562485

Breazzano

Nair

Arevalo

Frequency of urgent or emergent vitreoretinal surgical procedures in the United States during the COVID-19 pandemic

JAMA Ophthalmol20210411394456463

10.1001/jamaophthalmol.2021.0036

33662093

Piyasena

Murthy

GVS

Yip

JLY

Systematic review on barriers and enablers for access to diabetic retinopathy screening services in different income settings

PLoS One2019144e0198979

10.1371/journal.pone.0198979

31013274

Bonilla-Escobar

Eibel

Gallagher

Waxman

Follow-up in a point-of-care diabetic retinopathy program in Pittsburgh: a non-concurrent retrospective cohort study

BMC Ophthalmol20240820241356

10.1186/s12886-024-03581-9

39164678

Khanna

Kim

Giridhar

Mettla

Marmamula

Rao

Barriers to uptake of referral services from secondary care to tertiary care and its associated factors in L V Prasad Eye Institute network in Southern India: a cross-sectional study

BMJ Open201807787e020687

10.1136/bmjopen-2017-020687

29982204

Adhikari

Patten

Patel

Data harmonization and data pooling from cohort studies: a practical approach for data management

Int J Popul Data Sci2021611680

10.23889/ijpds.v6i1.1680

34888420

ASHA training modules

Ministry of Health and Family Welfare2025-03-26

https://nhm.gov.in/index1.php?lang=1&level=3&sublinkid=184&lid=257

Ayushman bharat comprehensive primary health care through health and wellness centres operational guidelines

Ministry of Health and Family Welfare2023-10-13

http://www.ab-hwc.mohfw.gov.in/

Indian public health standards. Community health centre 2022

National Health Systems Resource Centre2023-10-08

https://nhsrcindia.org/sites/default/files/CHC%20IPHS%202022%20Guidelines%20pdf.pdf

Indian public health standards, sub district and district health hospital

Ministry of Health & Family Welfare2025-03-26

https://nhsrcindia.org

Mukamel

Bresnick

Wang

Dickey

Barriers to compliance with screening guidelines for diabetic retinopathy

Ophthalmic Epidemiol199903616172

10.1076/opep.6.1.61.1563

10384685

Watson

MJG

McCluskey

Grigg

Kanagasingam

Daire

Estai

Barriers and facilitators to diabetic retinopathy screening within Australian primary care

BMC Fam Pract20211130221239

10.1186/s12875-021-01586-7

34847874

Padhy

Pyda

Marmamula

Khanna

Barriers to uptake of referral services from secondary eye care to tertiary eye care and its associated determinants in L V Prasad Eye Institute network in Southern India: a cross-sectional study-report II

PLoS ONE2024195e0303401

10.1371/journal.pone.0303401

38743737

Kumar

Velu

Patient and provider perspectives on barriers to screening for diabetic retinopathy: an exploratory study from southern India

BMJ Open202012101012e037277

10.1136/bmjopen-2020-037277

33303431

Shah

Mishra

Shanmugam

Doshi

Jayaraj

Ramanjulu

Validation of deep convolutional neural network-based algorithm for detection of diabetic retinopathy - artificial intelligence versus clinician for screening

Indian J Ophthalmol202002682398405

10.4103/ijo.IJO_966_19

31957737

Darwish

Patel

Gao

Diagnostic accuracy and reliability of retinal pathology using the Forus 3nethra fundus camera compared to ultra wide-field imaging

Eye (Lond)201905335856857

10.1038/s41433-019-0339-9

30679873

Sosale

Murthy

Sengupta

Naveenam

Medios- an offline, smartphone-based artificial intelligence algorithm for the diagnosis of diabetic retinopathy

Indian J Ophthalmol202002682391395

10.4103/ijo.IJO_1203_19

31957735

Duggal

Responsible adoption of cloud-based artificial intelligence in health care: A validation case study of multiple artificial intelligence algorithms for diabetic retinopathy screening in public health settings (forthcoming)

Rajalakshmi

Subashini

Anjana

Mohan

Automated diabetic retinopathy detection in smartphone-based fundus photography using artificial intelligence

Eye (Lond)20180632611381144

10.1038/s41433-018-0064-9

29520050

Wong

Sun

Kawasaki

Guidelines on diabetic eye care: the International Council of Ophthalmology recommendations for screening, follow-up, referral, and treatment based on resource settings

Ophthalmology2018101251016081622

10.1016/j.ophtha.2018.04.007

29776671

Raman

Ramasamy

Rajalakshmi

Sivaprasad

Natarajan

Diabetic retinopathy screening guidelines in India: All India Ophthalmological Society diabetic retinopathy task force and Vitreoretinal Society of India Consensus Statement

Indian J Ophthalmol202103693678688

10.4103/ijo.IJO_667_20

33269742

Lee

Sathyan

John

Singh

Robin

Predictors of and barriers associated with poor follow-up in patients with glaucoma in South India

Arch Ophthalmol2008101261014481454

10.1001/archopht.126.10.1448

18852425

Vengadesan

Ahmad

Sindal

Sengupta

Delayed follow-up in patients with diabetic retinopathy in South India: social factors and impact on disease progression

Indian J Ophthalmol201705655376384

10.4103/ijo.IJO_620_16

28573993

Mason

Sample size and saturation in PhD studies using qualitative interviews

Forum Qual Soc Res2010113

10.17169/fqs-11.3.1428

Ritchie

Lewis

Qualitative Research Practice: A Guide for Social Science Students and Researchers2003

2025-03-27

Sage Publications

https://mthoyibi.files.wordpress.com/2011/10/qualitative-research-practice_a-guide-for-social-science-students-and-researchers_jane-ritchie-and-jane-lewis-eds_20031.pdf

0761971092

Harris

Taylor

Thielke

Payne

Gonzalez

Conde

Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support

J Biomed Inform200904422377381

10.1016/j.jbi.2008.08.010

18929686

Byrne

A worked example of Braun and Clarke’s approach to reflexive thematic analysis

Qual Quant20220656313911412

10.1007/s11135-021-01182-y

Monitoring the building blocks of health systems: a handbook of indicators and their measurement strategies

World Health Organization2025-03-26

https://apps.who.int/iris/bitstream/handle/10665/258734/9789241564052-eng.pdf

Kalogeropoulos

Stefaniotou

Neofytou

The role of tele-ophthalmology in diabetic retinopathy screening

J Optom2020134262268

10.1016/j.optom.2019.12.004

31948924

Sajwani

Artificial intelligence for sustainability development in healthcare

BUiD Doctoral Research Conference 20232024264272

10.1007/978-3-031-56121-4_26

Ageing, older persons and the 2030 agenda for sustainable development acknowledgements

2017

2025-03-27

UNDP

https://www.undp.org/publications/ageing-older-persons-and-2030-agenda-sustainable-development

Haynes

Can it work? Does it work? Is it worth it? The testing of healthcareinterventions is evolving

BMJ199909113197211652653

10.1136/bmj.319.7211.652

10480802

Piyasena

Murthy

GVS

Yip

JLY

Systematic review and meta-analysis of diagnostic accuracy of detection of any level of diabetic retinopathy using digital retinal imaging

Syst Rev201811771182

10.1186/s13643-018-0846-y

30404665

Taber

Leyva

Persoskie

Why do people avoid medical care? A qualitative study using national data

J Gen Intern Med201503303290297

10.1007/s11606-014-3089-1

25387439

Solomon

Shoge

Ervin

Improving access to eye care: a systematic review of the literature

Ophthalmology20221012910e114e126

10.1016/j.ophtha.2022.07.012

36058739

Starbird

DiMaina

Sun

Han

A systematic review of interventions to minimize transportation barriers among people with chronic diseases

J Community Health201904442400411

10.1007/s10900-018-0572-3

30206755

Chabouis

Berdugo

Meas

Benefits of Ophdiat, a telemedical network to screen for diabetic retinopathy: a retrospective study in five reference hospital centres

Diabetes Metab200906353228232

10.1016/j.diabet.2008.12.001

19349202

Hark

Katz

Myers

Philadelphia telemedicine glaucoma detection and follow-up study: methods and screening results

Am J Ophthalmol201709181114124

10.1016/j.ajo.2017.06.024

28673747

Wang

Mitzel

Zhang

Bartlett

Handheld shape discrimination hyperacuity test on a mobile device for remote monitoring of visual function in maculopathy

Invest Ophthalmol Vis Sci2013081354854975505

10.1167/iovs.13-12037

23860761

Yim

Chandra

Sondh

Thottarath

Sivaprasad

Barriers in establishing systematic diabetic retinopathy screening through telemedicine in low- and middle-income countries

Indian J Ophthalmol202111691129872992

10.4103/ijo.IJO_1411_21

34708733

Wong

Sabanayagam

Strategies to tackle the global burden of diabetic retinopathy: from epidemiology to artificial intelligence

Ophthalmologica20202431920

10.1159/000502387

31408872

Sengo

Marraca

Muaprato

García-Sanjuan

Caballero

López-Izquierdo

Barriers to accessing eye health services in suburban communities in Nampula, Mozambique

Int J Environ Res Public Health202203251977

10.3390/ijerph19073916

35409600

Government of india ministry of health and family welfare, AB-PMJAY health benefit package master 2022

National Health Authority2024-07-14

https://pmjay.gov.in/sites/default/files/2022-04/HBP%202022%20..pdf

Eskandari

Abbaszadeh

Borhani

Barriers of referral system to health care provision in rural societies in Iran

J Caring Sci20130923229236

10.5681/jcs.2013.028

25276731

Rao

Khanna

Athota

Rajshekar

Rani

Integrated model of primary and secondary eye care for underserved rural areas: the L V Prasad Eye Institute experience

Indian J Ophthalmol2012605396400

10.4103/0301-4738.100533

22944748

Peduzzi

Concato

Kemper

Holford

Feinstein

A simulation study of the number of events per variable in logistic regression analysis

J Clin Epidemiol199612491213731379

10.1016/s0895-4356(96)00236-3

8970487

Piyasena

Murthy

GVS

Yip

JLY

A qualitative study on barriers and enablers to uptake of diabetic retinopathy screening by people with diabetes in the Western Province of Sri Lanka

Trop Med Health20194734

10.1186/s41182-019-0160-y

31139011

Kruk

Gage

Arsenault

High-quality health systems in the Sustainable Development Goals era: time for a revolution

Lancet Glob Health201811611e1196e1252

10.1016/S2214-109X(18)30386-3

30196093

Kamau

Osuga

Njuguna

Challenges facing implementation of referral system for quality health care services in Kiambu County, Kenya

Health Syst Policy Res201704101

10.21767/2254-9137.100067

Multimedia Appendix 1

Supplementary materials.