Tele-Ophthalmology for Retinal Health Surveillance in Denmark

Abstract

In this commentary paper, we discuss the potential of population health surveillance of retinal health in Denmark using optometrist-based opportunist retinal examination with tele-ophthalmology pathway. Using 12 months of data from 79 high-street optometrist stores, we imaged 355,818 unique individuals, which corresponds to 5.99% of the entire population of Denmark and equates to an average of 29,652 individuals each month, corresponding to 0.50% of the population. Coverage was highest among those aged 41–80 years, with coverage rates of between 8.17% and 8.68%. We also observed higher participation rates in female individuals (6.77%) than in male individuals (5.19%). Using Wilson score intervals with finite-population correction, we demonstrate that when such a large sample of the entire population is examined, prevalence estimates can be obtained with very narrow 95% confidence intervals. These results suggest that community imaging can provide stable, near real-time indicators of retinal disease at the population level. Selection bias is discussed to understand the limitations of data.

Population Health Surveillance

Population health surveillance is the systematic, continuous collection, analysis, and interpretation of health-related data to identify trends and guide public health actions [1, 2]. Traditionally, population health surveillance has focused on infectious diseases. However, for the last two decades, this concept has also expanded to include chronic and sensory disorders. Early detection and ongoing monitoring can help prevent functional decline in health [3]. Surveillance differs from research in that its purpose is not hypothesis testing but rather early detection of changes at the population level. This knowledge can help in designing prevention, screening policies, and resource allocation [4].

In ophthalmology, population health surveillance is challenging because many sight-threatening diseases may be asymptomatic for long periods. Patients first receive their diagnoses when they seek care. Although Nordic countries provide excellent registries to allow for such surveillance, these registries only capture diagnoses that require attention at tertiary-level health providers [5–7]. In other words, these registries can be used for advanced diagnoses that require hospital-level treatment or specialist attention, such as neovascular age-related macular degeneration, proliferative diabetic retinopathy, or retinal detachment [8–10], but they may underestimate or be unreliable for monitoring early stage of retinal diseases, asymptomatic diseases, or retinal diseases not treated in hospitals. These circumstances create a blind spot for public health and for population health surveillance in the field of retinal diseases.

Recent advancements in tele-ophthalmology and community-based retinal imaging offer a potential solution to this issue. In India, tele-ophthalmology networks have successfully combined opportunistic imaging with centralized grading to reveal insights into diabetic retinopathy [11]. These opportunistic retinal imaging initiatives generate a large amount of data, and the images thus acquired could potentially provide data on retinal health that may reflect health trends in near real time.

From a methodological point of view, surveillance data must be large, stable, and sufficiently representative to identify meaningful changes beyond random variation [2]. Accuracy depends on sample size and the prevalence of conditions of interest. Large tele-ophthalmology networks can easily gather thousands of observations each month, potentially allowing for narrow confidence intervals even for less common conditions. Additionally, continuous data flow, if large enough, can potentially reveal health trends within a short time frame, thereby allowing for quick responses to health changes. This balance of precision and timeliness is also a key consideration employed in infectious disease surveillance [1].

Incorporating such a system into Denmark's public health would align with international suggestions to strengthen non-communicable disease surveillance using digital data streams [3]. The necessary infrastructure is already in place: optometrists are spread out geographically, offer opportunistic retinal examination at no cost or at very low cost [12], and (a number of optometrists) collaborate with experts for tele-ophthalmological consult [12]. With proper organization and oversight, these data could provide population health surveillance in the field of retinal disease across different demographic groups and geographic areas. In this commentary paper, we present numbers and calculations to illustrate that this nationwide tele-ophthalmological system can potentially be used for retinal health surveillance.

Opportunistic Retinal Examination by Danish High-Street Optometrists

In Denmark, Danish high-street opticians increasingly offer retinal examination services with non-mydriatic imaging with fundus photography and, more recently, also optical coherence tomography. This expansion of services has triggered a debate in the Danish eye care sector regarding false–positive referrals, which may burden the publicly financed healthcare system [13]. This debate has in turn paved the way for solutions where positive findings from optometrist-based examination in a major optician chain are referred to tele-ophthalmology service with experienced ophthalmologists. Briefly, the optician evaluates if the retinae are normal; in case of any doubt, the tele-ophthalmology service evaluates the fundus photographs together with any other performed measurements and provides a diagnosis. Referral is only made for those in need of further examination in the public health system. Details on and experiences reported from this organization are published previously in greater detail [14–16].

Beyond individual patient care, tele-ophthalmology offers an opportunity for systematic retinal health surveillance. Public health surveillance involves the continuous, timely, large-scale, systematic collection and analysis of health data to detect changes in disease prevalence, which then allows for timely intervention and prevention [17]. Traditionally, in the Nordic countries, the publicly financed health sectors with their detailed registries have allowed for registry-based monitoring, but in Denmark, this system primarily captures ophthalmic diseases that require diagnosis, work-up, or intervention at a tertiary level. Such an approach challenges public health surveillance of retinal diseases. In the study reported here, we explored the numbers of unique individuals examined in optician stores of a nationwide chain with tele-ophthalmology collaboration service to understand if such a collaboration can also open doors to a meaningful retinal health surveillance program.

Epidemiological Considerations

To assess the potential for retinal health surveillance, we analyzed examination data from 79 stores of the optician company Louis Nielsen (Louis Nielsen A/S, Aalborg, Denmark) over 12 months (August 2022 to July 2023). All individuals in Denmark have a Civil Person Register number, which is a unique 10-digit personal identifier. This number is used for taxes, banks, healthcare, and various all-purpose functions as a unique serial number identifier across interactions. The use of this number for the purposes of this study allowed for easy assessment of the examination data on individuals. All aspects of this study followed the tenets of the Declaration of Helsinki. All patients provided mandatory consent to data processing and company policies prior to data sharing. All data storage and data management aspects comply with the standards of European Union General Data Protection Regulation. All data extracted were fully anonymous and were classified as quality assessment, hence neither ethics approval is needed nor any permission for the use of the data according to The Danish Data Protection Agency. Data were extracted from Louis Nielsen A/S (Aalborg SØ, Denmark) and MitØje Aps (Aarhus, Denmark) and were accessible to author Danson V. Muttuvelu.

The Wilson score interval with finite population correction adjustment was used to understand the 95% confidence intervals (95% CIs) of prevalence estimates given a certain prevalence [18]. Calculations were made annually (greater accuracy, delayed timeliness) or monthly (lower accuracy, better timeliness). Corresponding population statistics were extracted from Statistics Denmark (Ministry of Digital Affairs, Copenhagen, Denmark). These data from Statistics Denmark are publicly accessible without need for ethics approval [19].

During the study period, 355,818 individuals corresponding to 5.99% of the entire population of Denmark underwent retinal examination (Table 1). The highest examination coverage was among those aged 41–80 years (8.17–8.68%). Female individuals had a slightly higher examination coverage rate (6.77%) than male individuals (5.19%). On average, 29,652 unique individuals were examined monthly, corresponding 0.50% of the entire population of Denmark monthly. We then explored the 95% CIs at different prevalence rates:

• 1%: 95% CI 0.97–1.03% for data/year, 0.89% to 1.12% for data/month

• 5%: 95% CI 4.93–5.07% for data/year, 4.75% to 5.25% for data/month

• 10%: 95% CI 9.90–10.10% for data/year, 9.66% to 10.35% for data/month

• 20%: 95% CI 19.87–20.13% for data/year, 19.55% to 20.46% for data/month

• 50%: 95% CI 49.84–50.16% for data/year, 49.43% to 50.67% for data/month

Table 1.

Number of unique individuals examined, population of Denmark per July 2023, and the percentage of the entire population of Denmark examined within the 12-month study period

Age (years)	Individuals examined			Population of Denmark			Percentage of population examined
	Female individuals	Male individuals	Total	Female individuals	Male individuals	Total	Female individuals	Male individuals	Total
0–10	385	263	648	330,195	348,425	678,620	0.12%	0.08%	0.10%
11–20	15,320	9196	24,516	333,377	349,884	683,261	4.60%	2.63%	3.59%
21–30	22,866	14,635	37,501	389,572	404,344	793,916	5.87%	3.62%	4.72%
31–40	19,950	12,620	32,570	361,036	374,127	735,163	5.53%	3.37%	4.43%
41–50	35,367	24,699	60,066	368,029	366,850	734,879	9.61%	6.73%	8.17%
51–60	40,329	33,086	73,415	403,760	405,563	809,323	9.99%	8.16%	9.07%
61–70	32,885	27,961	60,846	342,992	330,612	673,604	9.59%	8.46%	9.03%
71–80	26,238	22,993	49,231	300,017	267,271	567,288	8.75%	8.60%	8.68%
81–90	8433	7464	15,897	134,811	98,048	232,859	6.26%	7.61%	6.83%
91–100	692	432	1124	24,453	10,112	34,565	2.83%	4.27%	3.25%
101+	2	2	4	577	90	667	0.35%	2.22%	0.60%
Total	202,467	153,351	355,818	2,988,819	2,955,326	5,944,145	6.77%	5.19%	5.99%

Precision and Timeliness

With a full year of data, the estimated 95% CIs were extremely narrow. For example, at the 20% prevalence rate, the 95% CI was 19.87% to 20.13%. However, even for smaller prevalences such as 1%, the 95% CI was narrow at 0.97–1.03%. Although the 95% CIs became larger when using data from a single month, they still remained quite narrow, such as, for example, from 19.87% to 20.13% to 19.55% to 20.46% for a 20% prevalence, and from 0.97% to 1.03% to 0.89% to 1.12% for a 1% prevalence. Therefore, surveillance using monthly data would appear to be suitable for timely and rapid monitoring of signals, while yearly data would provide very tight precision for summary reporting. A more complex approach could be applied for rare events or for sub-group analyses, such as, for example, municipality-level changes or specific high-risk strata, in which the number of available observations or the total population at risk (or both) are reduced. In such cases, specific considerations can be made to balance between precision and timeliness.

Bias and Representativeness

One major limitation of these data is that those examined are likely to be those in need for refractive products or those with visual complaints. This introduces a selection bias. Specifically, this also means that retinal diseases more often seen in myopes (e.g., myopic degeneration) or hyperopes (e.g., central serous chorioretinopathy) may be more likely to be overestimated. The numbers in the table also highlight that certain demographics are better represented in the data than others. Although optometrist stores in our strategy are available nationwide, it could be argued that certain stores/areas may examine a higher proportion of the population than others for various reasons that can be explained (nationwide differences in advertisement, competition, access to other types of eye health) or are more complex and difficult to explain (nationwide differences in sociocultural aspects towards prioritizing eye health or health-related coping mechanisms). These challenges should be kept in mind when discussing the potential uses of this system. For certain analyses, it could be argued that data weighting which takes differences into account could be one solution to addressing these issues.

Perspectives

In this commentary paper, we present numbers and calculations to illustrate that this nationwide tele-ophthalmological system can potentially be used for retinal health surveillance. If such a system is employed, it could deliver retinal health surveillance with very narrow confidence intervals even when data per month are used. Potentially, this allows for a timely and precise surveillance system. Limitations in population sampling should be kept in mind. The integration of retinal health surveillance into national public health strategies requires further exploration. More research is necessary to understand implementation for population-wide benefits.

Author Contributions

Conceptualization: Marie L.R. Rasmussen, Danson V. Muttuvelu, Yousif Subhi. Methodology: Marie L.R. Rasmussen, Danson V. Muttuvelu, Yousif Subhi. Analysis and interpretation: all authors (Marie L.R. Rasmussen, Danson V. Muttuvelu, Lasse J. Cehofski, Jakob Grauslund, Josef Huemer, Pearse A. Keane, Tor P. Utheim, Tunde Peto, Goran Petrovski, Charles C. Wykoff, Yousif Subhi). Writing—original draft preparation: Marie L.R. Rasmussen, Danson V. Muttuvelu, Yousif Subhi. Writing—review and editing: all authors (Marie L.R. Rasmussen, Danson V. Muttuvelu, Lasse J. Cehofski, Jakob Grauslund, Josef Huemer, Pearse A. Keane, Tor P. Utheim, Tunde Peto, Goran Petrovski, Charles C. Wykoff, Yousif Subhi). Supervision: Yousif Subhi. All authors (Marie L.R. Rasmussen, Danson V. Muttuvelu, Lasse J. Cehofski, Jakob Grauslund, Josef Huemer, Pearse A. Keane, Tor P. Utheim, Tunde Peto, Goran Petrovski, Charles C. Wykoff, Yousif Subhi) read and approved the final manuscript.

Funding

No funding or sponsorship was received for this study or publication of this article.

Data Availability

All data generated or analyzed during this study are included in this published article. Access to individual data points was not part of the consent given by the participants in study therefore individual data points are not publicly available.

Declarations

Conflict of Interest

Marie L.R. Rasmussen declares to have received speaker fee for lectures from Santen, and to have received advisory board honorarium from Santen. Danson V. Muttuvelu declares to have received consultancy honorarium from Alcon. Lasse J. Cehofski declares to have received speaker fees from AbbVie, Bayer, and Roche, and to have served as an advisory board member for AbbVie, Bayer, Novartis, and Roche. Jakob Grauslund declares to have received speaker fees from Allergan, Bayer, Novartis, and Roche, and to have served as an advisory board member for Allergan, Apellis, Bayer, Novartis, and Roche. Pearse A. Keane declares speaker fees from Heidelberg Engineering, Topcon, Allergan, and Bayer; consultancy honorarium from Google, DeepMind, Roche, Novartis, Apellis, and BitFount; and equity ownership in Big Picture Medical. Tor P. Utheim is co-founder and co-owner of The Norwegian Dry Eye Clinic/Tørreøyneklinikken and the Clinic of Eye Health/Øyehelseklinikken, Oslo, Norway, which delivers talks for and/or receives financial support from the following: ABIGO, Alcon, Allergan, AMWO, Bausch&Lomb, Bayer, European School for Advanced Studies in Ophthalmology, InnZ Medical, Medilens Nordic, Medistim, Novartis, Santen, Specsavers, Shire Pharmaceuticals, and Thea Laboratories. Goran Petrovski declares to have received consultancy honorarium from AbbVie and Roche, and serves on the advisory board of Bulbitech, Profundus, RetinaRISK and Astellas. Charles C. Wykoff declares consultancy work for 4DMT, AbbVie, Adverum Biotechnologies, Aerie, AGTC, Alcon, Alimera, Allergan, Allgenesis, Alnylam, Annexon Biosciences, Apellis, Arrowhead, Ascidian, Bausch + Lomb, Bayer, Bionic Vision Technologies, Boehringer Ingelheim, Cholgene, Clearside Biomedical, Curacle, Eyebiotech, EyePoint Pharmaceuticals, Foresite, Frontera Therapeutics, Genentech, Gyroscope Therapeutics, IACTA, IVERIC bio, Janssen, Kato Pharma, Kiora, Kodiak Sciences, Kriya Therapeutics, Merck, Nanoscope, Neurotech, NGM Biopharmaceuticals, Notal Vision, Novartis, OccuRx, Ocular Therapeutix, Ocuphire, OcuTerra, OliX, ONL, Opthea, Oxular, Palatin Technologies, Perceive Bio, Perfuse, PolyPhotonix, Ray, RecensMedical, Regeneron, REGENXBIO, Resonance, Roche, Sandoz, Sanofi, SciNeuro Pharmaceuticals, Stealth Biotherapeutics, Surrozen, Suzhou Raymon, Takeda, Thea, Therini, TissueGen, Valo, and Verana Health; research funding from 4DMT, Adverum Biotechnologies, AffaMed Therapeutics, Aldeyra, Alexion, Alimera, Alkahest, Allergan, Allgenesis, Amgen, Annexin Pharmaceuticals, Annexon Biosciences, Apellis, AsclepiX Therapeutics, Bayer, Boehringer Ingelheim, Chengdu Kanghong, Clearside Biomedical, Curacle, Eyebiotech, EyePoint Pharmaceuticals, Gemini, Genentech, GlaxoSmithKline, Graybug Vision, Gyroscope Therapeutics, IONIS, iRENIX, IVERIC bio, Janssen, Kodiak Sciences, LMRI, McMaster University, Nanoscope, Neurotech, NGM Biopharmaceuticals, Novartis, Ocular Therapeutix, Ocuphire, OcuTerra, OliX, Ophthotech, Opthea, Oxurion, Oxular, Oyster Point Pharma, Perceive Bio, RecensMedical, Regeneron, REGENXBIO, Rezolute, Roche, Sam Chun Dang Pharm, Sandoz, Senju Pharmaceutical, Shanghai Henlius Biotech, Taiwan Liposome Company, Unity Biotechnology, Verily Life Sciences, and Xbrane Biopharma; on the advisory board of Aeri and Kato Pharma; board member of ASRS, Vit-Buckle Society; and stock options for ONL, PolyPhotonix, RecensMedical, TissueGen, Visgenx, and Vitranu. Yousif Subhi declares to have received speaker fees from Bayer and Roche, and to be the inventor of a patent related to the use of biomarkers of polypoidal choroidal vasculopathy (WO2020007612A1). Yousif Subhi is Section Editor of Ophthalmology and Therapy. Yousif Subhi was not involved in the selection of peer reviewers for the manuscript nor any of the subsequent editorial decisions. Josef Huemer declares to have received speaker fees from Bayer, Roche and Zeiss, and travel support from Roche, Bayer and Alimera. Tunde Peto declares no competing interests.

Ethical Approval

All aspects of this study followed the tenets of the Declaration of Helsinki. All patients provided mandatory consent to data processing and company policies prior to data sharing. All data storage and data management aspects comply with the standards of European Union General Data Protection Regulation. All data extracted were fully anonymous and are classified as quality assessment, hence neither ethics approval is needed nor any permission for the use of the data according to The Danish Data Protection Agency. Data were extracted from Louis Nielsen A/S and MitØje Aps and were accessible to author Danson V. Muttuvelu.