The precise evaluation of refractive errors is a fundamental aspect of optometry and ophthalmology.1 Although objective techniques have evolved considerably, subjective refraction remains the reference standard for determining the final prescription, as it takes into account individual perceptual preferences and provides the most accurate correction for optimal visual performance. The interactive nature of subjective refraction, in which the patient's responses guide the refinement of the prescription, ensures a level of personalization that objective methods cannot yet achieve.2

Efforts to continuously improve the efficiency and accuracy of subjective refraction have led to the development of new optical systems capable of continuous and dynamic power adjustment.3–7 Tunable liquid lenses (TLLs) and adjustable astigmatic elements, such as Stokes lenses, are among the most promising technologies in this field.8,9 These devices allow for seamless spherical and cylindrical power variation within compact, lightweight setups and align well with the vector representation of refractive errors.10–14

In a previous study, we explored a DIY subjective refraction approach that allowed participants to perform their own refraction using a tunable lens system. This study demonstrated the feasibility and potential accuracy of this patient-driven method, highlighting its potential as a simplified, accessible approach to refractive evaluation.15

However, it remains unclear how the absence of professional guidance might affect the final results. During a traditional subjective refraction, the optometrist plays a key role in interpreting patient feedback, determining when to conclude the refinement process, and guaranteeing the stability and reliability of the measurements. These aspects could lead to differences in accuracy, consistency, and participant confidence compared to a completely self-directed procedure.16

The present study builds upon previous findings by directly comparing the DIY subjective refraction method with a professionally guided approach, both of which use the same optical system. This study addresses whether patient-driven refraction can achieve comparable precision when performed independently or if professional involvement is essential to maintain clinical reliability.

The mean age of the participants was 29 ± 9 years, and the average spherical equivalent was -0.98 ± 2.28 D. Fig. 1 shows the mean values for M (panel a), J0 and J45 (panel b), best VA (panel c), and time taken (panel d) for each method (ORx and DIY). The median values for both methods are similar across all panels, with comparable interquartile ranges (IQR) in panels a and b. However, in panels c and d, the IQR varies between methods, particularly in visual acuity (panel c) and time taken (panel d), indicating greater variability in these measurements for the DIY method.

Fig. 1.

Boxplots showing the distribution of spherical equivalent component (a), astigmatic components (b), visual acuity (c), and measurement time (d) obtained with the optometrist-guided (ORx) and do-it-yourself (DIY) subjective refraction methods. Central lines represent medians; boxes indicate interquartile ranges; whiskers denote data range.

The repeatability coefficient for the M component was ±0.33 D. For the astigmatic components, the CoR reached ±0.19 D for J0 and ±0.17 D for J45. Visual acuity showed a CoR of ±0.05 logMAR, while the coefficient associated with measurement duration was ±72 s.

Values obtained for each technique and each participant are represented in Fig. 2. The scatter plots display the results for each participant, with red dots representing ORx values, blue squares showing DIY values, and grey lines highlighting the differences between the two methods. For M, the values ranged from -9.52 to 3.72 D with the ORx method and from -9.66 to 3.44 D with the DIY method. The maximum and minimum differences found in the M component are 0.52 and 0 D, respectively, where 73% of the cases show less than 0.25 D difference, and 39% are less than 0.10 D. For J0, the ranges were -1.66 to 0.94 D (ORx method) and -1.64 to 0.91 D (DIY method). Similarly, values ranged from -1.16 to 0.67 D (ORx method) and from -1.30 to 0.67 D (DIY method) for the J45 method. The maximum differences found for the astigmatic components are 0.38 and 0.33 D for J0 and J45, respectively, while the minimum for both is 0 D. 79% of the J0 differences and 76 % of the J45 differences are under the value of 0.05 D.

Fig. 2.

Scatter plot showing the mean value of the repeated measurements obtained for each participant with the optometrist-guided (ORx, red dots) and do-it-yourself (DIY, blue squares) methods. Each pair of symbols represents the average data from one subject under both conditions.

Given that refraction is expressed in vector form, the overall difference between methods can also be summarized by the modulus of the power vector, commonly referred to as blur strength or ΔRx,23,24 calculated as the square root of the sum of squares of the three refractive components. Using this metric, the mean difference in blur strength was 0.11 ± 0.20 D, with a median value of 0.08 D.

Visual acuity ranged from -0.18 to 0.03 logMAR with the ORx method and from -0.18 to 0.02 logMAR with the DIY method. The maximum and minimum differences found in VA were 0.06 and -0.09 logMAR, where only one case had a difference over 0.06 logMAR. Finally, measurement time ranged from 80 to 366 seconds for the ORx method and from 81 to 354 seconds for the DIY method, with a maximum difference between techniques of 121 seconds and a minimum of 0 seconds. In 45% of cases, differences were less than 10 seconds.

Table 1 summarizes the linear polynomial adjustment coefficients obtained for each parameter. The largest CI for the slope was shown for the required time (0.35), whereas the M components showed the shortest interval (0.04). The CIs for intercept values only exceeded 0.05 D M component among refractive components and were set to 0.03 logMAR and 76” for the VA agreement and required time, respectively. In all cases, no significant differences were found at the statistical level between the techniques (p > 0.1). Fig. 3 shows the Passing-Bablok regression test for the M (panel a), J0 (panel b), J45 (panel c), VA (panel d) and time (panel e). As it can be seen, the identity line is included in the CIs for the astigmatic components. For the M, VA and time taken, the identity line extends beyond the CI in some areas.

Fig. 3.

Passing–Bablok regression plots for M (a), J0 (b), J45 (c), VA (d), and time (e). The regression line is red, and the identity line is black. The colored area represents the 95% confidence intervals for the regression coefficients, and the pink dotted lines represent the 95% confidence limits.

Fig. 4 shows the boxplots of the differences between methods according to the spherical equivalent of the participant. Spherical equivalent is divided between negative M values and M values positives or equal to zero diopters. In Fig. 4a, the median value difference for the M component for positive refractive errors (M ≥ 0 D) was 0.13 D [IQR: 0 – 0.36 D]. For negative refractive errors (M < 0 D), the median value difference was 0.11 D [IQR: 0 – 0.23 D]. Regarding the astigmatic components (Fig. 4b), the median values of the differences for positive refractive errors were 0.00 D [IQR: -0.05 – 0.04 D] and 0.00 D [IQR: -0.04 – 0.07 D] for J0 and J45, respectively. For negative refractive errors, the median difference was 0.00 D [IQR: -0.06 – 0.05] for J0 and 0.00 D [IQR: -0.04 – 0.04 D] for J45. In Fig. 4c, the median values of the differences for the achieved VA were 0.01 logMAR [IQR: -0.00 – 0.03 logMAR] and 0.00 logMAR [IQR: -0.01 – 0.02 logMAR] for positive and negative refractive errors, respectively. Concerning the required time (Fig. 4d), the median value for the differences were 22 seconds [IQR: -18 – 47seconds] and 19 seconds [IQR: -18 – 42 seconds] for positive and negative refractive errors, respectively.

Fig. 4.

Boxplots showing the distribution of the differences between the optometrist-guided (ORx) and do-it-yourself (DIY) subjective refraction methods of spherical equivalent component (a), astigmatic components (b), visual acuity (c), and measurement time (d) obtained for participants with negative spherical equivalent (M < 0 D) and positive or zero spherical equivalent (M ≥ 0 D).

The present study aimed to evaluate the role of the optometrist during the performance of a novel subjective refraction routine based on a tunable liquid lens (TLL) and a Stokes lens. The comparison between the optometrist-guided refraction (ORx) and the do-it-yourself (DIY) conditions allowed us to assess whether the clinician’s involvement significantly influences the outcomes of this simplified approach.

Regarding repeatability, the M component exhibited the highest CoR, while the astigmatic components presented CoR values below ±0.20 D. These outcomes are in line with those previously reported for the DIY method,15 which showed CoRs of ±0.38 D for M and ±0.21 D for both J0 and J45. Although marginally lower CoR values were observed with ORx, the overall magnitude of the differences between methods was small and remained within clinically acceptable limits.

A comparable trend was found for visual acuity, with CoR values of ±0.06 logMAR for the DIY method and ±0.05 logMAR for ORx. In contrast, repeatability in measurement time differed more noticeably, with a CoR of ±55 s for the DIY approach and ±72 s for ORx, indicating greater between-measurement variability in the ORx method.

Overall, the results showed a high level of agreement between both techniques, with no statistically significant differences in any of the measured components. The slopes of the Passing–Bablok regressions consistently included the value 1 within their confidence intervals except for the time required, confirming proportional agreement between methods for the other variables. However, for both the M component and the required time, the intercept values did not include zero as can be seen in Table 1. The non-zero intercept observed for the M component (-0.15 D) denotes a systematic shift between methods rather than a random discrepancy. In practical terms, this reflects a slightly more positive (less myopic) outcome for the professional-guided refraction, consistent with better accommodative relaxation under clinician supervision. Although the difference is small and within clinically acceptable limits, it reinforces the value of professional control in guiding subjective decisions near the refraction endpoint.25 Both astigmatic components (J0 and J45) exhibited excellent agreement between methods, with slopes and intercepts near unity and zero, respectively. These results suggest that the participants were able to reliably adjust the Stokes lens orientation and magnitude, even without professional supervision, highlighting the intuitive nature of the tunable system for astigmatic refinement. The strong agreement in visual acuity outcomes further supports the validity of the DIY approach. The small deviation of the slope (0.84) indicates slightly lower VA values for the DIY condition, which may be attributed to subtle undercorrection or less precise endpoint perception by the participant. Nonetheless, these differences remained well below thresholds of clinical significance, confirming the robustness of the system for visual optimization.

Interestingly, time was the only parameter for which the slope did not include the value 1 within its confidence interval. The slope below unity (0.76) and positive intercept (34”) indicate that while the DIY approach generally required less time for longer refractions, shorter measurements tended to take slightly longer compared to the professional-guided condition. This pattern suggests that participants became more efficient as they gained familiarity with the procedure, whereas the clinician maintained a more constant pace across cases. Such a trend aligns with the notion of a learning effect intrinsic to self-driven methods.

In order to evaluate whether there was a learning effect or not, some checking was made for M and the time required. Median differences for M were 0.37 D [IQR: 0 – 0.39 D] and 0.20 D [IQR: 0 – 0.39 D] for the first and the third measurement. Regarding time, median differences were 20 seconds [IQR: -173 – -34 seconds] and 20 second [IQR: -65 – -7 seconds] for the first and the third measurement. Although the inter-method differences in task duration did not reach statistical significance (Wilcoxon signed-rank test, p-value M = 0.55; p-value time = 0.88), the interquartile range for the time markedly decreased from 90 seconds in the first repetition to 44 seconds in the third, indicating greater consistency between methods after repeated testing. This reduction in variability suggests a potential learning or adaptation effect, even in the absence of statistically significant mean changes.

In general, the DIY condition resulted in shorter measurement times. This is likely due to the absence of verbal feedback and decision-making exchanges between the participant and the clinician, which naturally reduces the total duration. Despite these time differences (which decrease after repeated testing), overall accuracy was comparable between the two procedures, suggesting that the DIY approach could be a valid alternative when professional supervision is not available. In terms of time, other subjective refraction approaches such as new methodologies or hybrid technologies combining objective and subjective refraction had reduced the required time to perform the measurements.4,26–28

Moreover, when the value of the spherical equivalent was considered, this trend was more pronounced in hyperopic and near-emmetropic participants, whose M values under professional guidance tended to be even more positive than those obtained in the DIY condition as can be seen in Fig. 4a. This pattern suggests that optometrist supervision can help control accommodation and guide the endpoint, particularly in participants prone to overaccommodation. Similarly, Rodríguez-López et al.28 found a standard deviation up to ±0.65 D when the blur-detection task was performed in an unsupervised way. From a clinical perspective, the median differences observed (0.12 D) are small compared with the variability observed between independent optometrists measuring the same population,25,26 indicating that the practical impact of professional guidance in this context is limited. Nevertheless, even such minimal systematic biases could have perceptual implications for certain individuals, especially hyperopic participants or those with limited accommodative flexibility. In contrast, no relevant differences were observed in the astigmatic components (J0 and J45) or in visual acuity, which supports the robustness of the DIY measurements for these parameters.

The narrower interquartile ranges observed in the ORx condition for measurement time further underline the standardizing role of the practitioner. The clinician likely provides a more consistent pacing and decision structure during the procedure, leading to reduced variability between individuals. However, both methods showed an evident learning effect, with shorter measurement times across consecutive repetitions.15 This trend suggests that familiarity with the system and procedure contributes to improved efficiency, regardless of whether the refraction is performed independently or under supervision.

From a practical standpoint, these results suggest that DIY-based subjective refraction could serve as a preliminary or screening tool, reducing examination time and professional workload, while professional-guided procedures remain preferable for final prescriptions or for patients with complex refractive profiles. The comparable refractive outcomes between the two approaches highlight the potential of self-directed refraction systems to increase accessibility in vision testing, particularly in tele-optometry or low-resource contexts.

Several limitations should be acknowledged. First, the level of accommodation could not be fully controlled and may have varied among participants, especially in the DIY condition. Second, the participants’ understanding of the instructions and their subjective criteria for "best clarity" may have influenced the outcomes. Third, since the study was conducted in a controlled laboratory setting, further research is needed to evaluate the approach's performance in real-world clinical or remote screening environments. It should be noted that the age of participants could influence their ability to perform the DIY approach. For instance, children would probably be unable to follow the procedure independently, and older adults might face physical or cognitive limitations that could affect their performance. In any case, successful DIY refraction would likely require the participant’s cooperation, understanding of the procedure, and sufficient motivation to carry it out effectively.

Recent developments in subjective refraction technology have already introduced hybrid devices that combine automated optimization algorithms with limited clinician supervision.26,29,30 These systems have shown promising results in maintaining refractive accuracy, reducing chair time and examiner dependency. Integrating such semi-automated elements into patient-driven refraction protocols could represent a natural next step toward scalable, accessible eye-care solutions. Future studies could explore hybrid approaches that combine automated or software-based feedback with clinician supervision to further optimize the balance between accuracy, time efficiency, and accessibility.

In summary, the results suggest that while the optometrist’s guidance introduces subtle differences in the subjective refraction—particularly in the control of accommodation and consistency of timing. These findings support the potential of simplified, patient-driven subjective refraction systems as complementary tools in clinical optometry, especially in scenarios where professional resources are limited or where remote refraction may be beneficial.

Ministerio de UniversidadesFPU20/05624 and EST24/00228.

Nothing to declare.

The data underlying the results presented in the study are available from Zenodo (https://doi.org/10.5281/zenodo.17349090).