Educational quality of ocular and orbital anatomy videos on YouTube: a student- and patient-centered evaluation

Abstract

Background

The YouTube platform is increasingly used by medical students and patients as a supplementary learning tool. However, the quality and educational value of YouTube videos remain underexplored. This study aimed to evaluate the quality, reliability, and educational utility of YouTube videos on eye and orbit anatomy.

Methods

A systematic YouTube search was conducted using keywords related to eye and orbit anatomy. Out of 270 videos initially identified, only 100 met the inclusion criteria. Two anatomists and two ophthalmologists independently evaluated the videos using validated tools. The anatomists applied the Anatomy Content Score for Ophthalmology (ACS-O), Global Quality Score for Students (GQS-S), modified DISCERN, and JAMA benchmarks, while the ophthalmologists assessed the patient-focused content using the Global Quality Score for Patients (GQS-P). Video popularity and engagement metrics were recorded. Inter-rater reliability was assessed using Kappa coefficient. Non-parametric tests assessed the correlations between instructional quality and YouTube engagement.

Results

The ACS-O revealed that 54% of the videos are useful for eye and orbit anatomy education. Notably, the percentage of useful videos was higher for student-focused (GQS-S) evaluation at 56% compared to patient-focused (GQS-P) evaluations at just 8%. Correlation analysis demonstrated a strong positive relationship between anatomical accuracy (ACS-O) and student education quality (GQS-S), while an inverse correlation between student and patient scores was observed. A positive correlation was identified between video duration and educational quality for students (ACS-O r = 0.576, GQS-S r = 0.525).

Conclusions

YouTube offers valuable supplementary resources for eye and orbit anatomy education, but significant variability in quality highlights the need for systematic evaluation. Also, YouTube consistently falls short of providing adequate, high-quality information for patient education, revealing a significant gap in public health information equity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12909-026-08942-0.

Background

YouTube is considered one of the most widely utilized resources for medical education, with studies indicating that medical students use it as a major/an augmenting source for learning anatomy [1, 2]. Although YouTube is growing in popularity as an educational resource, concerns persist about its content quality, accuracy, and reliability. Comprehensive reviews of YouTube’s use in medical education have identified substantial inconsistencies in content quality, with many videos lacking validation, peer review, and alignment with recognized educational standards [2, 3]. Various studies have similarly raised concerns about the reliability and accuracy of the anatomical information presented on YouTube [4, 5]. The absence of standardized quality control on open-access platforms raises important questions about the accuracy and reliability of these resources for both professional medical education and patient information.

The eye is a complex sensory organ, consisting of three layers: the outer fibrous layer (sclera and cornea), the vascular layer (uvea), and the innermost neuronal layer (retina), which contains photoreceptors that relay visual signals. The eye is located within the orbit, a protective bony structure that houses the eyeball, the optic nerve, the extraocular muscles, and their crucial neurovascular structures. This structural complexity necessitates the use of high-quality visual aids to facilitate effective learning [6,7].Medical students consistently report challenges in mastering ocular anatomy through traditional teaching methods alone, highlighting the potential value of additional digital resources [8]. Moreover, the decline of formal ophthalmology education in medical curricula has led to an increased reliance on self-directed learning through online platforms, making quality assessment of these resources increasingly critical [9]. In addition to its educational value, YouTube can also serve as a valuable health education tool for both healthcare professionals and patients. This dual audience nature adds another layer of complexity to the required quality assessment. While medical students require detailed, accurate anatomical information aligned with curriculum standards, patients and the general audience need accessible, understandable explanations that promote health literacy without compromising scientific accuracy [10, 11]. Interestingly, several studies have also evaluated YouTube videos covering eye surgeries, such as cataract surgery and keratoplasty, aiming to inform patients by providing accessible yet accurate surgical information [12, 13].

Despite the growing availability of ocular and orbital anatomy content on YouTube, there remains a notable gap in the systematic evaluation of the educational value, reliability, and usefulness of these videos. This highlights the critical need for specialized quality assessment frameworks targeted at ocular and orbital anatomy content. The current literature lacks a comprehensive evaluation of YouTube’s ocular anatomy videos using validated assessment tools that consider educational usefulness for both pre-clinical medical students and patient-oriented accessibility. The aim of this study was to fill this gap by systematically evaluating the quality, educational value, and usefulness of YouTube’s eye and orbit anatomy content for diverse learning audiences. It offers evidence-based recommendations for educators, students, and patients. The findings will contribute to advancing digital medical education quality assurance and guiding the future development of online anatomy educational resources.

Methods

Search strategy and video selection criteria

A systematic search was conducted on YouTube between 20 July 2025 and 30 July 2025, with the objective of retrieving educational videos relevant to the anatomy of the eye. To ensure a comprehensive collection of relevant videos, the search incorporated a range of targeted keywords, including “Eye anatomy,” “Eye dissection,” “Ocular anatomy,” “Anatomy of the eyeball,” “Visual system anatomy,” “Eye muscles,” “Eye structure,” “Orbit anatomy,” and “Neurovascular supply of the eye”. To guarantee alignment with essential curricular content and ensure comprehensiveness and validity of the keywords, a foundational list of keywords was derived from the core chapters of standard anatomy textbooks, including Gray’s Anatomy for Students, Clinical Anatomy by Regions, and Moore’s Clinically Oriented Anatomy [14–16]. Then, the final keyword selection was further validated through consensus among the study’s co-investigators, all of whom hold professorial appointments in anatomy, to ensure scholarly accuracy and relevance to modern anatomical education. Previous studies suggest that the majority of YouTube users tend exclusively to review the initial 30 videos presented in their search results [17]. Consequently, this study incorporated the top 30 videos associated with each selected search keyword. We excluded videos that were not related to the topic, were less than 1 min long, were in languages other than English, were repetitive, or were not relevant to human anatomy. We conducted the YouTube search following the approach outlined by Mistareehi et al. (2025) [5]. Briefly, we conducted a YouTube search using an incognito browser and cleared the cache to avoid biases from personalized history. Accessing the global YouTube site (www.youtube.com), we searched for the selected keywords using the platform’s default settings, without applying any specific filters, to ensure unbiased and accurate results. The metric data of the final retrieved videos were extracted on 10 August 2025.

Evaluation of videos

For the purpose of this study, “students” refers to undergraduate medical students in the pre-clinical years seeking foundational anatomical knowledge. To evaluate the educational value of the videos, we assessed their educational usefulness using the Anatomy Content Score for Ophthalmology (ACS-O). This ACS-O is a modified version of the original Anatomy Content Score (ACS), developed by Azer in 2012 [18], to focus mainly on the anatomical contents and educational usefulness of the uploaded anatomical videos. Its content validity was established via review by an expert panel of five anatomists and two ophthalmologists. Additionally, the inter-rater reliability was assessed after standardized rater training. In summary, there are five major criteria and six minor criteria. Each major criterion is assigned a score of 2 points, while each minor criterion is worth 1 point (Appendix 1, Table S1). A score is given for each criterion that is met, while those that are not met receive a score of zero, with no allowances for partial scores. Videos that achieve a total score of 13 or more are considered useful videos, provided that all major criteria are fulfilled. Additionally, we utilized the Global Quality Score (GQS) scoring criteria, [19] employing one version for students (GQS-S) and another for patients (GQS-P) to evaluate the quality of educational information presented in the videos for students and patients, respectively (Appendix 1, Table S2 and S3). The GQS has a maximum of 5 points, and videos are categorized as useful videos if they receive 4 points or more, while those with 3 points or fewer are categorized as not useful videos [5]. Additionally, the modified DISCERN (mDISCERN) quality assessment score alongside the Journal of the American Medical Association (JAMA) criteria were both utilized to evaluate the reliability and transparency of the information presented [20, 21] (Appendix 1, Table S4 and S5). Two anatomists (I.H. & A.J.M.) independently evaluated the included videos for their educational quality for the students. At the same time, two ophthalmologists (N.Q. & A.M.) independently assessed the same videos to evaluate their educational quality for the patients.

Evaluation of video visibility

Key metric information was collected for each video, including length, view counts, audience feedback (likes and dislikes), upload dates, verification status of the channels, countries of origin, and source of authorship (academic, company, or individual) (Appendix 2). Metrics such as the like ratio and view rate for the analyzed YouTube videos were calculated using established formulas. The like ratio was calculated by dividing the number of likes by the total number of likes and dislikes, then multiplying by 100. We were able to estimate the dislike counts by adding the “Return YouTube Dislike” extension to Google Chrome.

The view rate was derived by dividing the total views by the number of days since the video was uploaded. Furthermore, viewer engagement was quantified using the Interaction Index, which was calculated by the difference between likes and dislikes divided by total views and multiplied by 100 [22].

Statistical analysis

Cohen’s Kappa coefficient was utilized to evaluate the level of agreement between different evaluators. The data were not normally distributed, as examined by the Shapiro-Wilk test. Therefore, the non-parametric Mann–Whitney U test was used to compare numerical data, while the Chi-square test and Fisher-Freeman-Halton Exact tests were employed to analyze categorical data. Additionally, Spearman’s rank correlation coefficient was used to assess the correlation. The level of statistical significance was set at P < 0.05. All analyses were conducted using IBM SPSS software (29.0.0.0, IBM, Armonk, NY, USA).

Results

Out of the initial 270 videos retrieved using nine search keywords related to eye and orbit anatomy, 170 videos (63%) were excluded based on exclusion criteria. The exclusion reasons included non-English language videos (n = 17), videos shorter than one minute (n = 18), subject-irrelevant content (n = 30), repetitive videos (n = 79), non-human content (n = 23), and videos depicting modified settings (n = 3). Following these exclusions, a total of 100 videos were included for the final evaluation (Fig. 1). The inter-observer agreement for all scoring instruments demonstrated an almost perfect level of reliability. The Kappa coefficients were 0.858 for ACS-O, 0.878 for GQS-S, 0.860 for GQS-P, 0.899 for mDISCERN, and 0.946 for JAMA. Any discrepancies in scoring were resolved through a joint review and discussion to re-evaluate the relevant videos. Additional details concerning the videos and scoring are provided in Appendix 2.

Fig. 1.

Flowchart of Video Selection Methodology. The flowchart depicts the process of identifying and screening YouTube videos via nine search keywords related to eye anatomy. Of the 270 videos initially retrieved, 170 were excluded based on pre-defined criteria (e.g., language, duration, relevance), resulting in 100 videos included for final analysis

Videos ranged in duration from 66 to 2745 s, with a mean of 678.86 s. The total views were 39,179,801, with viewers ranging from 13 to more than 4 million (mean view count = 391,798.01). The view ratio, capturing views standardized by video age, ranged from 0 to 1787.2, averaging 157.8. The likes per video varied widely (ranging from 1 to 144,677; mean = 6987.7), while dislikes were generally lower (0 to 2060; mean = 151.85). The like ratio averaged 98.1. The interaction index, a composite engagement metric, ranged from 0.09 to 31.6 with a mean of 2.56. The mean of days since upload was 2164.2 (range from 11 to 6213), showing that these videos had been available online for varied lengths of time (Table 1).

Table 1.

Descriptive Statistics of the Characteristics and Quality Assessment of YouTube Videos on Eye and Orbit Anatomy (n=100)

	Min	Max	Mean (±SD)	Median
Duration (sec)	66	2,745	678.86 (594.93)	481.5
View	13	4,114,196	391,798.01 (750,711.61)	60,152
View Ratio	0	1,787.2	157.8 (287.79)	42.8
Like	1	144,677	6,987.72 (17,152.24)	1,348
Dislike	0	2,060	151.85 (352.47)	20
Like Ratio	83.1	100	98.1 (2.6)	98.8
Interaction Index	0.09	31.63	2.56 (3.90)	1.60
Days Since Upload	11	6,213	2,164.24 (1,198.04)	2,021.5
ACS-O	8	16	13.48 (1.96)	14
GQS-S	2	5	3.78 (1.01)	4
GQS-P	1	5	1.69 (0.97)	1
mDISCERN	1	5	2.32 (0.63)	2
JAMA	1	4	1.58 (0.84)	1

The table summarizes the 100 sampled YouTube videos, presenting descriptive statistics (minimum, maximum, mean ± standard deviation, median) for quantitative metrics (e.g., duration, views, likes) and validated quality assessment scores. The Anatomical Content Score for Ophthalmology (ACS-O), Global Quality Score for Student (GQS-S), Global Quality Score for Patient (GQS-P), Modified DISCERN (mDISCERN), and Journal of the American Medical Association (JAMA) scores indicate the overall educational quality and reliability of the content

Videos were assessed for educational usefulness using various validated assessment scoring systems. The ACS-O averaged 13.48 (SD = ± 1.96) out of a possible 16 points. Quality ratings differed notably between student- and patient-focused evaluations. The GQS-S averaged 3.78 (SD = 1.01), reflecting moderate overall educational value for undergraduate and professional learners. Conversely, the Global Quality Score for Patients (GQS-P) was much lower, with an average of 1.69 (SD = ± 0.97), highlighting a considerable gap in accessibility and clarity for non-expert audiences. Additional tools such as the mDISCERN and JAMA Benchmark scores further supported these findings, with average scores of 2.32 (SD = ± 0.63) and 1.58 (SD = ± 0.84), respectively (Table 1).

The distribution of quality and reliability assessment scores for the analyzed YouTube videos on eye and orbit anatomy (n = 100) is presented in Fig. 2. The ACS-O identified 54 useful videos out of 100 (54%), GQS-S found 56 useful videos (56%), and GQS-P disclosed only 8 videos (8%). The ACS-O (Fig. 2A) revealed a wide range of scores, with 4% of videos scoring ≤ 10 points, 13% of videos scoring 11 points, 25% of videos scoring 12 points, 6% of videos scoring 13 points, 11% of videos scoring 14 points, 21% of videos scoring 15 points, and 20% of videos scoring 16 points, indicating generally moderate anatomical content usefulness. The GQS-S scores (Fig. 2B) ranged from 2 to 5 points, with most videos scoring 3 points (34%) or 5 points (32%), followed by 4 points (24%) and 2 points (10%), reflecting a moderate educational value for student audiences. For the GQS-P (Fig. 2C), scores were predominantly low, as 55% of videos scored 1 point, 31% scored 2 points, 6% scored 3 points, 6% scored 4 points, and 2% scored 5 points, indicating lower quality for patient-oriented content. The overlap analysis (Venn diagram Fig. 2D) demonstrates that only a small subset of videos (n = 4, 4% of the evaluated videos) was consistently classified as “useful” across all three key measures (ACS-O, GQS-S, and GQS-P), while a larger consensus was observed between ACS-O and GQS-S, sharing 54 videos (54% of the evaluated videos) classified as useful by both criteria. The mDISCERN score distribution (Fig. 2E) showed that most videos clustered at 2 points (67%), followed by 3 points (26%), while a small percentage of videos scored 1 point (3%), 4 points (3%), and only 1% attained the maximum score of 5. The JAMA (Fig. 2F) scores were generally low, with 61% receiving 1 point, 24% scored 2 points, 11% scored 3 points, and only 4% achieving the highest score of 4 points, reflecting limited source credibility across the videos.

Fig. 2.

Distribution of YouTube Video Quality Assessment Scores and Overlap of “Useful” Video Classifications. Charts (A–F) show the distribution of scores for five quality assessment tools applied to a sample of YouTube videos on eye and orbit anatomy (n = 100). (A) Anatomy Content Score for Ophthalmology (ACS-O), (B) Global Quality Score–Student (GQS-S), (C) Global Quality Score–Patient (GQS-P), (D) Useful videos identified by ACS-O, GQS-S, and GQS-P, (E) modified DISCERN (mDISCERN), (F) Journal of the American Medical Association (JAMA) criteria. The data reveal a skew towards moderate-quality scores for student-oriented tools (ACS-O, GQS-S) and low-quality scores for reliability tools (JAMA, mDISCERN) and the patient-oriented score (GQS-P). The diagram (D) demonstrates a small core of videos (n = 4) that were consistently “Useful” across ACS-O, GQS-S, and GQS-P criteria, while strong consensus exists between ACS-O and GQS-S, with an overlap of 54 useful videos by both criteria

Spearman’s correlation analysis was conducted to examine relationships between video quality scores, content metrics, and engagement parameters (Table 2). There was a statistically significant and strong positive correlation between ACS-O and GQS-S (r = 0.836, p < 0.001), indicating videos with greater anatomical accuracy and completeness were also rated higher for educational quality by students. The GQS-P showed weakly significant negative correlations with ACS-O (r = -0.213, p < 0.05) and GQS-S (r = -0.261, p < 0.01), reflecting an inverse relationship between student-focused educational value and patient accessibility. Content reliability indicators such as mDISCERN and JAMA scores showed no significant correlations with ACS-O and GQS-S. However, JAMA showed a significantly weak negative correlation with GQS-P (r = -0.228, p < 0.05). Video duration demonstrated statistically significant and a moderate positive correlation with ACS-O (r = 0.576, p < 0.001) and GQS-S (r = 0.525, p < 0.001), suggesting that longer videos tended to have higher educational value content for student learners, but showed a significant weak negative correlation for patients and the general audience (GQS-P; r = -0.259, p < 0.01 ), indicating that longer videos were less suitable for these viewers. Regarding engagement metrics, quality scores for students (ACS-O and GQS-S) showed weak positive correlations with the interaction index (ACS-O; r = 0.234, p < 0.05, GQS-S; r = 0.254, p < 0.05), indicating that videos with higher educational value attracted more engagement from students and healthcare professionals. However, GQS-P scores showed no significant correlation with the interaction index. View count and likes were strongly positively correlated with each other (r = 0.960, p < 0.001), but views had no significant correlation with ACS-O (r = -0.021) or GQS-S (r = -0.115), suggesting that the popularity was not a reliable indicator of anatomical or educational quality. In contrast, a statistically significant and weak to moderate positive correlation was observed between view counts and GQS-P (r = 0.326, p < 0.001) (Table 2).

Table 2.

Correlation Analysis of Anatomical Video Quality (ACS-O, GQS-S, GQS-P), Reliability (mDISCERN, JAMA), and Video Engagement and Popularity Metrics

		ACS-O	GQS-S	GQS-P	mDISCERN	JAMA	Duration	View	Like	Interaction Index
ACS-O	rs CI	1
GQS-S	rs CI	0.836*** 0.762–0.887	1
GQS-P	rs CI	-0.213* -0.398 - -0.012	-0.261** -0.440 - -0.062	1
mDISCERN	rs CI	0.083 -0.121–0.280	0.158 -0.045–0.349	-0.102 -0.298–0.102	1
JAMA	rs CI	0.137 -0.067–0.330	0.169 -0.035–0.359	-0.228* -0.411 - -0.027	0.144 -0.060–0.366	1
Duration	rs CI	0.576*** 0.423–0.697	0.525*** 0.361–0.657	-0.259** -0.438 - -0.060	-0.102 -0.298–0.102	0.169 -0.034–0.359	1
View	rs CI	-0.021 -0.223–0.181	-0.115 -0.310–0.089	0.326*** 0.132–0.495	-0.104 -0.300–0.100	-0.331*** -0.500 - -0.138	-0.155 -0.346–0.049	1
Like	rs CI	0.028 -0.175–0.229	-0.062 -0.261–0.141	0.335** 0.142–0.503	-0.111 -0.306–0.094	-0.308** -0.480 - -0.113	-0.099 -0.295–0.105	0.960*** 0.941–0.973	1
Interaction Index	rs CI	0.234* 0.034–0.417	0.254* 0.055–0.434	-0.185 -0.373–0.018	-0.086 -0.283–0.119	0.045 -0.159–0.245	0.325*** 0.131–0.495	-0.206* -0.392 - -0.004	-0.003 -0.205–0.200	1

Spearman’s correlation coefficients (rs) and their 95% confidence intervals (CI) evaluating the relationships between standardized quality scales (ACS-O: Anatomy Content Score for Ophthalmology; GQS-S/P: Global Quality Score for Student/Patient; mDISCERN: Modified DISCERN, and JAMA: Journal of the American Medical Association) and YouTube popularity and engagement metrics. Significant positive correlations were observed between the anatomical content (ACS-O) and student-perceived quality (GQS-S), as well as between video duration and both ACS-O and GQS-S. Patient-rated quality (GQS-P) showed a significant negative correlation with duration and a significant positive correlation with popularity metrics (Views, Likes). Significance: *p < 0.05, **p < 0.01, ***p < 0.001

Table 3 summarizes the characteristics of YouTube videos classified as useful or not useful according to the ACS-O, GQS-S, and GQS-P scoring tools. All variables are reported as mean and median values. Our analysis showed that videos considered as useful were significantly longer than not useful videos for both ACS-O and GQS-S (P-value < 0.001), whereas no significant difference was observed for GQS-P (P-value = 0.457). However, no significant differences in total views or view ratio were observed for ACS-O (P-value = 0.585 and 0.547, respectively) or GQS-S (P-value = 0.258 and 0.216, respectively). In contrast, GQS-P useful videos had significantly more views (P-value = 0.011) and a higher view ratio (P-value = 0.010). The number of likes and dislikes did not differ significantly for ACS-O (P-value = 0.934 and 0.359, respectively) or GQS-S (P-value = 0.483 and 0.141, respectively). For GQS-P, useful videos received significantly more likes (P-value = 0.010) and dislikes (P-value = 0.019), although the like ratio did not differ significantly (P-value = 0.251). The interaction index was significantly higher for useful videos than not useful videos in ACS-O (P-value = 0.030) and GQS-S (P-value = 0.038), but not for GQS-P (P-value = 0.939). Additionally, our analysis demonstrated that no significant differences were observed in the number of days since upload between useful and not useful videos across all scoring systems (ACS-O: P-value = 0.691; GQS-S: P-value = 0.789; GQS-P: P-value = 0.139).

Table 3.

Comparison of YouTube Video Metrics Between Useful and Not Useful Content as Classified by the Anatomy Content Score for Ophthalmology (ACS-O) and Global Quality Score (GQS-S, GQS-P)

	ACS-O					GQS-S					GQS-P
	Useful 54		Not Useful 46			Useful 56		Not Useful 44			Useful 8		Not Useful 92
	Mean	Median	Mean	Median	P-value	Mean	Median	Mean	Median	P-value	Mean	Median	Mean	Median	P-value
Video length	954.4	775	355.4	267	< 0.001	946.6	743.5	338.1	259	< 0.001	473.4	512.5	696.7	481.5	0.457
Views	278795.4	57,888	524453.2	66632.5	0.585	268895.9	52,819	548218.8	74,348	0.258	640762.1	442,844	370149.0	45769.5	0.011
View Ratio	101.8	42.8	223.5	57.7	0.547	98.2	41.2	233.6	77.8	0.216	260.3	242.2	148.9	36.6	0.010
Like	7468.4	1308	6423.4	1617	0.934	7202	1262.5	6714	1885.5	0.483	26,443	6114.5	5295.9	1176	0.010
Dislike	94	15.5	219.4	25	0.359	91	13	229.3	29.5	0.141	213.3	214.5	146.5	15.5	0.019
Like Ratio	98.5	99.1	97.5	98.3	0.081	98.6	99.1	97.4	98.3	0.021	97.9	98.3	98.1	98.9	0.251
Interaction Index	2.9	2	2.1	1.4	0.030	2.9	1.9	2.1	1.4	0.038	3.5	1.7	2.5	1.6	0.939
Days since upload	2145.2	2073.5	2186.5	2021.5	0.691	2158.3	2073.5	2171.8	2021.5	0.789	3053.3	3423	2086.9	1977	0.139

Descriptive statistics (mean, median) and p-values comparing YouTube metrics for videos classified “Useful” versus “Not Useful” based on the Anatomy Content Score for Ophthalmology (ACS-O), the student version of the Global Quality Score (GQS-S), and the patient version (GQS-P). Key findings indicate that student-rated useful content (ACS-O & GQS-S) is characterized by significantly greater video length and a higher interaction index. Patient-rated useful content (GQS-P) is associated with significantly greater popularity metrics (views, likes) but also higher dislike counts, suggesting more controversial yet widely viewed content. Statistical significance (p < 0.05)

We evaluated the usefulness of ACS-O, GQS-S, and GQS-P categories across several variables, including uploader country, source of authorship, primary visual modality, timing relative to COVID-19, and channel verification status. The distribution of usefulness across uploader countries showed no statistically significant differences in any group (Fisher-Freeman-Halton Exact tests, ACS: P-value = 0.080; GQS-S: P-value = 0.127; GQS-P: P-value = 0.849). Similarly, the source of authorship (individual, company, or academic) did not significantly affect the perceived usefulness (Fisher-Freeman-Halton Exact tests, ACS-O: P-value = 0.174; GQS-S: P-value = 0.111; GQS-P: P-value = 0.086). Moreover, analysis of the primary visual modality indicated no significant association between type of material (drawings/images, models, living/cadavers) and usefulness ratings in all three categories (Fisher-Freeman-Halton Exact tests, ACS-O: P-value = 0.600; GQS-S: P-value = 0.586; GQS-P: P-value = 1.000). Furthermore, the comparison of video upload date (before and after the onset of COVID-19) indicated non-significant differences in usefulness (exact chi-square, ACS-O: P-value = 0.844; GQS-S: P-value = 0.843; GQS-P: P-value = 0.715). Verification status of the uploader channel similarly did not influence usefulness ratings significantly (exact chi-square, ACS-O: P-value = 0.879; GQS-S: P-value = 0.687; GQS-P: P-value = 0.577) (Table 4).

Table 4.

Comparison of YouTube Video Metadata Between Useful and Not Useful Videos

		ACS-O			GQS-S			GQS-P
	Total	Useful 54	Not Useful 46	P-value	Useful 56	Not Useful 44	P-value	Useful 8	Not Useful 92	P-value
Uploader Country				0.080			0.127			0.849
US	43	21	22		22	21		3	40
UK	12	10	2		10	2		1	11
India	11	7	4		7	4		1	10
Germany	8	2	6		2	6		0	8
Australia	5	4	1		4	1		1	4
Canada	3	1	2		1	2		0	3
Egypt	2	2	0		2	0		0	2
Norway	1	1	0		1	0		0	1
Undefined	15	6	9		7	8		2	13
Source of Authorship				0.174			0.111			0.086
Individual	55	27	28		28	27		7	48
Company	14	6	8		6	8		1	13
Academic	31	21	10		22	9		0	31
Primary Visual Modality				0.600			0.586			1.000
Drawings/images	90	50	40		52	38		8	82
Models	6	2	4		2	4		0	6
Living/cadavers	4	2	2		2	2		0	4
COVID				0.844			0.843			0.715
Before	51	27	24		28	23		5	46
After	49	27	22		28	21		3	46
Verified Channel				0.879			0.687			0.577
Yes	47	25	22		25	22		3	44
No	53	29	24		31	22		5	48

Descriptive statistics of YouTube video characteristics (uploader country, source of authorship, Primary Visual Modality, timing relative to the COVID-19 pandemic, and channel verification status) between videos classified as “Useful” versus “Not Useful” based on the Anatomy Content Score for Ophthalmology (ACS-O), the Global Quality Score for Student (GQS-S), and the Global Quality Score for Patient (GQS-P). Fisher-Freeman-Halton Exact tests and exact chi-square tests revealed no statistically significant associations (p > 0.05) for any variable across all three rating scales

Discussion

This study presents a comprehensive systematic evaluation of YouTube’s video content related to the anatomy of the eye and orbit using validated assessment tools adapted for anatomical education, including Anatomy Content Score for Ophthalmology (ACS-O), Global Quality Score-Student (GQS-S), and Global Quality Score-Patient (GQS-P). It also utilizes two of the most widely recognized tools (Modified-DISCERN (mDISCERN) and Journal of the American Medical Association (JAMA)) to measure the reliability, transparency, and authorship credibility of the YouTube health content [20, 21]. Using a comprehensive search for YouTube videos by search keywords that cover the anatomy of the eye and orbit (Bony components, vascular supply, innervations, layers of the eyeball, and extraocular muscles), our findings revealed that the videos considered as useful for educational purposes represent 54% and 56% of the total rated sample as evaluated by ACS-O and GQS-S scales, respectively. These findings revealed a moderate level of overall usefulness, supporting the role of YouTube as a valuable supplementary educational tool. Interestingly, we found significant differences between the educational value of these videos for different target audiences (students vs. patients), as evaluated by GQS-S and GQS-P, highlighting important considerations for both medical educators and patients seeking anatomical information online.

YouTube has emerged as a crucial, accessible resource in modern medical education, including anatomy, offering dynamic visualizations such as 3D models and dissection videos that enhance traditional anatomy learning. The on-demand availability of YouTube videos provides supreme supplementary support for students and practitioners seeking to strengthen their understanding of complex anatomical structures. However, various studies indicated that YouTube should not be considered a sufficient standalone source for learning anatomy [1, 18, 23, 24]. Although the open and accessible framework of YouTube empowers content creators, it also leads to significant variability in the quality, accuracy, and depth of content [25, 26]. Our study on eye anatomy reflects this variability, revealing that only 54% of videos based on ACS-O and 56% based on GQS-S were classified as educationally useful. Despite YouTube’s widespread use for anatomy learning, a notable gap exists in large-scale analyses of its anatomical educational content [23, 27]. This research gap underscores the need for systematic evaluation using validated instruments to help learners identify high-quality resources.

The use of ACS-O and GQS-S criteria in the current study yielded remarkably similar results, suggesting that ACS-O is effective for evaluating eye and orbit anatomy videos [5, 18, 28]. This consistency strengthens the validity of our usefulness ratings. The fact that both assessment criteria identified the same videos as useful, and that the same pattern of longer videos being of higher quality, reinforces the importance of using validated assessment tools rather than relying solely on YouTube’s metrics. This approach aligns with recommendations by Curran et al. (2020), who emphasized the necessity of using multiple quality assessment criteria to properly evaluate medical educational content [2].

The specialized nature of eye and orbit anatomy presents unique challenges for YouTube-based education. The complexity of ocular structures and the precision required in ophthalmological practice demand high-quality visual presentations. In 2020, Schmuter et al. emphasized the potential of well-designed ocular content on social media to enhance medical student learning [29]. Interestingly, in our study, from an initial pool of 270 screened videos, 170 videos (63%) were excluded, leaving 100 videos (37%) for the final analysis. The high number of exclusions aligns with findings from other studies, which reported that a significant number of the screened videos were non-instructional material and irrelevant to the search topic [4, 5, 28, 30, 31]. This highlights the challenge of searching YouTube for educational resources and suggests that the platform is saturated with nonspecific or low-quality educational materials.

The considerable contrast between student-focused (GQS-S: 56% of videos classified as useful) and patient-focused (GQS-P: 8% of videos classified as useful) quality scores highlight a core challenge in online medical education content that serves both professional learners and general audiences. This aligns with prior research noting the highly variable content quality of YouTube videos, suggesting that the videos tend to be optimized for either educational depth or accessibility, but rarely both [2, 27, 32, 33]. Our data suggested that ocular anatomy YouTube videos predominantly target healthcare learners (e.g., medical students), potentially leaving patients with inadequate resources. This gap is likely driven by the search behavior of the general audience, which could significantly influence the content on YouTube and can create a bias toward content related to clinical pathologies affecting the eye rather than instructional anatomy. This is because the general audience are typically interested in specific medical conditions and tend to use search keywords related to the name of the condition, symptoms, or clinical diagnosis of the diseases (e.g., farsightedness over the cornea and lens) [34]. This interest pattern may promote the content creator to generate videos that meet the popular demand, which often focus on diseases and treatment to boost engagement metrics of the produced videos (e.g., views and likes) leading to an underrepresentation of anatomical content for the general public. On the other hand, videos targeting students mainly focus on academic completeness over popularity and engagement metrics, making less appealing content to the general audiences who seek quick clinical answers.

The positive correlation between video duration and educational value for students (r = 0.576 for ACS-O; r = 0.525 for GQS-S) supports established educational principles emphasizing comprehensive content delivery in medical education. Longer videos demonstrated more comprehensive anatomical content with stronger clinical integration. This finding aligns with observations regarding YouTube content about the anatomy topics,^4,5,28,31 where comprehensive coverage of the material, as well as longer videos, was associated with higher educational value. However, the negative correlation between duration and patient-oriented quality (r = -0.259 for GQS-P) suggests that detailed presentations may overwhelm non-expert audiences, supporting recommendations for targeted content creation. The lack of correlation between video popularity metrics (views, likes) and educational quality represents a concerning finding. Our results demonstrated that highly viewed videos are not necessarily more educationally valuable, consistent with systematic reviews [3, 26]. This mismatch between popularity and quality highlights the risk of students and patients being engaged with potentially less accurate or comprehensive content. The phenomenon has been documented across various medical specialties, suggesting a systemic challenge in online health information [2, 3, 35]. However, a notable finding emerged in the interaction index analysis, where useful videos demonstrated significantly higher interaction indices compared to non-useful ones according to ACS-O and GQS-S criteria. Many studies on YouTube anatomical content have reported similar findings in their analyses, where interaction metrics better reflect content quality than raw popularity measures [5, 28, 31]. The moderate engagement levels observed in our study (mean interaction index: 2.56) indicate that ocular anatomy videos, which primarily target expert viewers such as students, generate reasonable interaction. In contrast, content aimed patients or the general audience tends not to achieve a similar level of interaction, possibly reflecting differences in audience interest and expertise. This finding suggests that the specialized, detail-oriented nature of anatomical instruction attracts a dedicated learner but is less likely to produce the broad engagement metrics (views, likes, shares).

Interestingly, our analysis revealed no significant associations between useful videos and factors such as uploader credentials (individual, company, or academic), geographic origin, or account verification status. This finding aligns with previous studies showing wide variability in YouTube video quality across uploader credentials, with academic institutions not consistently producing higher-quality content than individual or commercial creators [2, 3, 27, 28, 35, 36]. This observation underscores the importance of evaluating content based on educational quality rather than source authority alone. However, we found that the useful videos based on GQS-P were uploaded exclusively by individual creators, with no contribution from academic entities. This pattern suggests that academic and corporate creators primarily target students and professionals, creating content that aligns with formal curricula, while individual creators focus on producing accessible content that addresses the lay public’s desire for condition-specific knowledge [35]. Regarding COVID-19’s impact on video, we found no significant differences in usefulness between pre- and post-pandemic content. This suggests that while the pandemic accelerated online learning adoption, [37] it did not necessarily improve content educational value [5]. The consistency in educational value across time periods may reflect established patterns of content creation and the inherent challenges of producing useful educational materials regardless of external circumstances.

The implications of the current study extend beyond ocular anatomy to broader considerations of digital medical education. Our dual evaluation approach (anatomist and ophthalmologist perspectives) provides unique insights into content quality from different professional viewpoints, enhancing the strength of findings. The dual-audience challenge identified in our research reflects a fundamental gap in online health communication. While healthcare professionals and academic learners require detailed curriculum-aligned information, patients and the general audience need accessible explanations that promote health learning without sacrificing scientific accuracy [10]. Few videos in our sample (4%) successfully balanced these demands, suggesting opportunities for targeted content development.

Limitations and future directions

Our research has certain limitations that need consideration. The cross-sectional design of the study captures a snapshot of the available content on YouTube, but cannot account for the dynamic nature of YouTube’s platform, where videos are continuously added, removed, or updated. In addition, the focus on English-language content may limit generalizability to non-English-speaking populations. Moreover, our assessment tools, while validated, may not capture all aspects of educational effectiveness, such as long-term retention or practical application of learned concepts. Future research should consider longitudinal studies examining how YouTube’s anatomy content evolves over time, comparative analyses across different anatomical systems, and investigation of learning outcomes associated with different video characteristics. Development of standardized quality metrics specifically designed for anatomical education could enhance consistency across studies and provide clearer guidance for content creators. In addition, the variable baseline knowledge among patients represents a key subjective confounding factor. A binary assessment (useful vs. not useful) cannot determine whether a video judged useful for the general patient population is equally useful for patients with greater prior knowledge. To strengthen the conclusiveness and representativeness of the findings, future study designs should account for these audience characteristics and stratify analyses by patient health literacy and learner level.

Recommendations for content creators and educators

Based on our findings, we recommend that content creators developing eye and orbit anatomy videos prioritize comprehensive coverage over brevity to ensure adequate time for detailed anatomical explanations and visual demonstrations. Specifically, creators should adopt a dual strategy of producing in-depth, clinically correlated videos for medical students, and separate concise videos that use plain language for patients. Additionally, creators should focus on producing content that encourages viewer interaction through clear explanations, engaging presentations, critical thinking, and opportunities for questions and discussion. For medical educators, we recommend implementing systematic quality assessment protocols when selecting YouTube content for educational purposes, rather than relying solely on popularity metrics. The use of validated assessment tools such as ACS-O and GQS-S can help ensure that selected content meets educational standards and learning objectives. Educators should also train students in digital literacy, empowering them to critically appraise the credibility and bias of online health information. Furthermore, academic and professional institutions are uniquely positioned to lead by creating and disseminating high-quality, peer-reviewed video resources to bridge the current quality and accessibility gaps in digital anatomy education.

Conclusions

This study contributes to the existing body of knowledge in several key ways. First, it introduces and content-validates a novel, ophthalmology-specific anatomy scoring tool (ACS-O), which was employed alongside established evaluation instruments to assess YouTube videos on ocular and orbital anatomy in a reproducible manner. Second, the study highlights a clear divergence in educational suitability depending on the target audience by demonstrating a significant mismatch between the needs of students and patients. While many videos are useful for medical students, few are suitable for patient education, and videos that teach effectively for students often fail to address patient needs. Third, it reveals that public engagement metrics (views, likes, and comments) do not reliably indicate educational value, and emphasizes that video selection must prioritize content quality rather than popularity signals. Fourth, the study offers a practical guidance to individual learners, encouraging them to focus on content instead of superficial popularity. Finally, it advocates for proactive involvement by educators and clinicians to take the initiative in recommending high-quality educational videos tailored to the needs of students and patients, rather than leaving learners to navigate the vast and inconsistent landscape of YouTube content unaided.

Nevertheless, it remains the responsibility of individual learners to select educational materials that align with their learning styles and informational needs. Given the open nature of online platforms, the quality and accuracy of uploaded content cannot be consistently regulated. As such/Accordingly, learners should make deliberate, audience-appropriate choices, ideally supported by (1) content-based evaluation and (2) informed recommendations from educators and clinicians.

Authors’ contributions

Conceptualization: Hoja, Mistareehi, AllouhData curation: Hoja, MistareehiFormal analysis: HojaFunding acquisition: AllouhInvestigation: Hoja, Mistareehi, Alqudah, Mohammad, AlraddadiMethodology: Hoja, Mistareehi, Alqudah, Mohammad, Alraddadi, GhozlanProject administration: Hoja, AllouhResources: Hoja, Katselis, AllouhSoftware: Hoja, Katselis, MustafaSupervision: Hoja, AllouhValidation: Mistareehi, Alqudah, Mohammad, Alraddadi, Ghozlan, MustafaVisualization: Hoja, AllouhWriting – original draft: Hoja, MistareehiWriting – review & editing: Hoja, Mistareehi, Alqudah, Mohammad, Alraddadi, Ghozlan, Katselis, Mustafa, AllouhAll authors have read and approved the final submitted version of this manuscript and agreed to hold accountability for all aspects of the work.

Funding

This work was supported by a UPAR grant to M. Z. Allouh from the Office of Research at the United Arab Emirates University (grant code: G00004977, Fund no.: 12M219).

Data availability

All data generated and analysed during this study are included in this published article and its supplementary information file (Appendix 2).

Declarations

Ethics approval and consent to participate

This research exclusively analyzed publicly accessible YouTube videos and did not involve the participation of human subjects, patient information, or animal testing. Consequently, it did not require approval from an institutional ethics review board, in accordance with current guidelines regarding the use of publicly available, non-identifiable data. All analyzed materials were in the public domain, with no collection or use of personally identifiable information at any point in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.