International comparisons of intraocular pressures, as measured by Tono-Pen and Goldmann applanation tonometry, in healthy adults: A meta-analysis

Background:

Investigate intraocular pressure (IOP), as measured by Tono-Pen (TP) and Goldmann applanation tonometry (GAT), in healthy adults. Provide an updated synthesis of multinational, primary studies, reported during the 10-year period 2011 to 2021 and offer an evidence-based benchmark, against which IOP can be evaluated across subject variables and pathologies. Three primary research questions are investigated: Is there a statistically significant difference between IOP measured by TP and GAT? If yes, is the difference clinically significant? Is measurement of IOP affected by the country or setting location, in which the measurements are made?

Methods:

An aggregate meta-analysis was conducted on 22 primary studies, from 15 different countries. IOP measurements were made from each healthy adult subject, with both the TP and GAT. Primary studies were identified and data extracted according to recommended preferred reporting items for systematic reviews and meta-analysis protocol guidelines. Meta-analysis summary results are reported as the point estimate of the raw mean difference of IOP.

Results:

Meta-analysis reveals a statistically significant difference in raw mean differences in IOP, when measured by TP and GAT, in the healthy adult population. Tono-Pen IOP measurements are higher than GAT IOP measurements. The point estimate for the summary effect size = −0.73 mm Hg, P = .03. The prediction interval for the true effect size, in 95% of all comparable populations, is −4.03 to 2.58 mm Hg. There is no clinically significance difference in IOP when measured by TP and GAT. Meta-regression analysis reveals statistically significant differences in measurement of IOP by countries, R² analog = 0.75, P = .001. There is no statistically significant difference in measurement of IOP as a function of measurement location setting, R² analog = −0.17, P = .65.

Conclusions:

IOP measured by TP are marginally higher compared to GAT, in the healthy adult population. However, from a clinical practice perspective, TP and GAT produce similar IOP measurements. There is evidence of significant variabilities in IOP measurements as a function of country. IOP measurements collected in a research laboratory setting are similar to IOP collected in a clinical setting. Results have implications for the primary care physician requiring a portable, inexpensive, reliable, and easily administered instrument to assess IOP.

1. Introduction

Accurate and reliable measurements of intraocular pressure (IOP) are essential in the identification and management of common eye conditions, for example, glaucoma. Given IOP is the single most objective, quantifiable, and measurable factor, allowing for the early identification of increased IOP and its frequently accompanying impairment in vision, for example, preventable irreversible blindness, it is essential primary care physicians have available to them, a cost effective, accurate, reliable, quickly, and easily administered, portable instrument, with which IOP can be made. Unlike conventional Goldmann Applanation Tonometry (GAT), which requires expensive specialized equipment, specialized training, and typically conducted in the office of an ophthalmologist or optometrist, the Tono-Pen (TP) offers a much-needed alternative.

Since the introduction of GAT over 68 years ago in 1954^[1] and the Tono-Pen XL over 35 years ago in 1987,^[2] comparisons have been made between the 2 assessment procedures and resultant IOPs. Over time, investigators have reported comparison studies in which TP reveal IOP greater than, equal to, and less than IOPs measured by the current gold standard GAT.^[3–28]

This variability and inconsistencies in reported studies have resulted in a confused professional literature. The variability has been explained by differences in research study designs,^[3] research methods applied,^[4] statistical approaches used to analyze IOP measurements,^[5,6] eye biomechanics,^[7,8] diurnal variations of IOP,^[9] body positions in which measurements were taken (sitting or supine),^[10,11] age of subjects,^[12–14] ethnicity of subjects,^[15] central corneal thickness (CCT),^{[7,16–21,29]} cornea shape,^[22] level of training and skill of eye care professional making the measurements (ophthalmologist, technician),^[23,24] subjectivity of GAT measurements,^[25,26] and fundamental principles of physics underlying the differences in how IOP are measured by the TP or GAT procedure.^{[7,22,27,28,30]}

Variability and inconsistencies reported by TP and GAT comparison studies, which have examined IOP as a function of ocular pathologies, have further contributed to a confused professional literature. Specifically, differences have been reported in IOP, when measured by TP and GAT in patient populations such as glaucoma with varying degrees of severity,^{[18,24,31–35]} keratoconus,^[22,36,37] ocular hypertension,^[38] corneal edema,^[39,40] as well as patients who have undergone pars plana vitrectomy,^[41] endothelial keratoplasty,^[42] or penetrating keratoplasty.^[39,43] Such studies are useful in understanding IOP as functions of disease processes and surgical procedures; however, these studies fail to provide the necessary normative information required by primary care physicians, charged with initial screenings of potentially debilitating and permanent impairments in vision.

Variability and inconsistencies have additionally been reported by several large TP and GAT comparison studies, which have retrospectively examined IOP pressure measurements, collected through Eye Institute clinics, from around the world, for example, Rachel Eye Center (Abuja, Nigeria),^[44] Narayana Nethralaya Eye Hospital (Bangalore, India),^[45] Ivey Eye Institute (London, Canada),^[35] and the Ira G. Ross Eye Institute (Buffalo, NY).^[24]

In order to better understand the specific factor or factors most responsible for the variability in reported differences between IOP measured by the TP and GAT, in patients presenting with suspected or real ocular pathologies, for example, glaucoma, it is useful to better understand the variability in reported differences between IOP measured by the TP and GAT, in the healthy adult population.

Measurements made with any instrument contain 2 principal sources of variability. Variability due to measurement error and variability due to subject variables. Measurement error is explained, in part by differences in how examiners take standardized measurements, for example, IOP over repeated measurement trials and the limits of precision of measurement inherent to the instrument itself. Variability due to subject variables, for example, central cornea thickness, diurnal variations in IOP, patient’s body position such as sitting or supine during measurement, physiological state of the subject, such as subjects holding their breath during measurement, biometric properties of the eye, and more. Variability explained by measurement error can be partitioned from variability explained by subject variables. Data indicate repeated measurements by TP or GAT routinely vary as much as ± 2.0 to ± 5 mm Hg, in both the healthy and unhealthy eye populations, as a result of both measurement error and subject variables.^{[23,25,26,30,35,46,47]} When subject variables are controlled, TP and GAT data reveal measurement error accounts for approximately ± 2.0 mm Hg of the variability routinely observed by repeated measurements and made during a single measurement session by each of the 2 instruments. The observed ± 2.0 mm Hg variability is typically interpreted to have no clinical significance.^[2,23,25,26]

For the purpose of this meta-analysis and the interpretation of summary results, an observed raw mean difference of ± 2.0 mm Hg will be considered within the expected limits of accepted measurement error for TP and GAT and of no clinical significance.

Previous meta-analyses have attempted to identify, quantify, and explain the variability in reported differences between TP and GAT measurements.^[48,49] No previous meta-analysis has focused specifically on variability between studies which have investigated differences and similarities in the IOP measured by the TP and GAT in the same subjects and restricted the analysis to the healthy adult population.

The present meta-analysis investigates differences in IOP, as measured by TP and GAT, in healthy adults.

Three primary research questions are to be answered:

1. Given the body of evidence-based medicine, published during the 10-year period 2011 to 2021, is there a statistically significant difference in IOP, when measured by TP and GAT, in the healthy adult population?

2. If a statistically significance difference is observed, is the raw mean difference greater or less than can be expected by measurement error alone, operationally defined as a raw mean difference greater than or less than 2.0 mm Hg?

3. Is measurement of IOP affected by the country or setting location, in which the measurements are made?

Findings are discussed within the matrix of clinical applications for the primary care physician, especially physicians practicing in developing countries throughout the world, rural communities, and providing primary medical care to disadvantaged and underserved populations.

2. Materials and methods

2.1. Statistical analysis

An aggregate meta-analysis, using a random-effects model, was conducted on 22 primary studies, which employed a repeated measures research design, appearing in the professional, medical, and scientific literature, during the 10-year period 2011 to 2021, and meeting all additional inclusion/exclusion criteria. A priori power analyses were conducted in order to approximate the number of primary studies required, given anticipated small effect sizes and high heterogeneity between studies. Minimum power for this study was set at 0.80. Alpha levels were set at 0.05 for all statistical analyses.

2.2. Databases

Primary studies were extracted from 6 popular electronically searchable databases; PubMed, EMBASE, SCOPUS, Cochrane Reviews, Google Scholar, and ProQuest Complete (previously named Dissertation Abstracts International). Additional primary studies were identified by inspection of reference lists contained within relevant primary study articles. Archived conference abstracts from the American Academy of Ophthalmology and American Academy of Optometry were searched electronically. Databases searched contained published professional, medical, and scientific literature, unpublished doctoral dissertations, and professional societies’ conference abstracts. No restrictions were placed upon languages in which the primary study appeared.

2.3. Inclusion criteria

All potential studies required measurement of IOP to be made with the Tono-Pen XL (TP) or Tono-Pen AVIA (TP) applanation tonometers and the Goldmann applanation tonometer (GAT), in the same subject, in a single session, with the subject sitting in an upright position, in accordance with standardized measurement instructions provided by each instruments’ manufacturer. All IOP measurements were required to be reported in mm Hg. Summary statistics which contained the means and standard deviations of TP and GAT IOP measurements were required, for each primary study.

All primary studies were required to report the number of individuals participating, the number of eyes included in the statistical summaries, whether only 1 eye or 2 eyes were reported for each subject, the mean age of subjects (all subjects must have been 17 years of age or older), the country in which the study was completed, and the setting in which the IOP measurements were made (hospital, out-patient clinic, clinician’s office, research laboratory).

All primary studies were required to provide a clear statement, that all subjects reported in the “healthy” or “control” groups were free from eye pathologies, for example, glaucoma, suspected glaucoma, ocular hypertension, corneal edema, inflammation, which would potentially effect IOP measurements. Eye health was assessed by patient history in combination with ophthalmological examination. Primary studies were not required to investigate IOP in healthy adult, as the study’s primary research objective; only that IOP measurements and associated required information could be extracted from the primary study.

In the event mean CCT and associated standard deviations from the means were reported in the primary study, in addition to TP and GAT IOP, this information was recorded and included in the database for descriptive purposes and subsequent secondary analyses. CCT measurements were not required for primary study inclusion in the meta-analysis of IOP.

All primary studies were required to provide a clear statement, the study had been preapproved by the facility’s Institutional Review Board or local Ethics Committee and conducted in accordance with the tenets of the World Medical Association Declaration of Helsinki-Ethical Principles for Medical Research Involving Human Subjects.

2.4. Exclusion criteria

Primary studies which investigated IOP measurements comparing TP and GAT measurements, using between-subjects rather than repeated measurement research designs were excluded from the meta-analysis. Primary studies which restricted their investigations to the pediatric population, defined as children <17 years of age or studies which included individuals under the age of 17 years of age and could not be partitioned out, with high confidence from the published data, were excluded from the meta-analysis. Primary studies which restricted comparisons of TP and GAT measurements to subjects presenting with IOP > 22.0 mm Hg or subjects who had received treatments for current or past ocular pathologies, potentially affecting IOP were excluded from the meta-analysis.

2.5. Dependent variable

The single dependent variable, upon which the meta-analysis was conducted, was the raw difference in means of IOP mm Hg, measured by TP and GAT, from each primary study. A random effects model was used for generating primary study weights and calculating the overall mean summary effect.

2.6. Independent variables

Mean subject age, subject gender (if reported), country of study origin, mean CCT (if reported), and the setting location in which the IOP were measured, were extracted from all primary studies, and summarized using descriptive statistics.

2.7. Quality of primary study assessment

All primary studies identified for inclusion in the meta-analysis were assessed and evaluated using a brief 12 item Primary Study Quality Assessment Checklist (PSQAC).^[50] The PSQAC assesses primary elements of study quality on a binary scale. Specific PSQAC items assess whether the research question is clearly stated and answered; is the sample representative and adequate to answer the research question; are potential confounding factors identified and/or controlled; is the research design correct to answer the research question; are criteria measures for independent and dependent variables valid and reliable; are statistical tests applied appropriately; are post hoc tests completed appropriately; are tables and figures clearly labeled; are the discussion and conclusions consistent with the results; are the research questions clearly answered; are limitations of the study provided; and are ethical standards met. Primary studies considered for inclusion in the meta-analysis were reviewed by 2 independent evaluators, which independently completed the PSQAC, for each primary study reviewed. All primary studies included in the final meta-analysis were required to have an a priori PSQAC score equal to or above 22 (high quality). Maximum score on the PSQAC is 24.

2.8. Publication bias assessment

Possible publication bias was initially assessed by visual inspection of the funnel plot. The funnel plot was constructed with the conventional presentation, with the dependent variable (difference in TP and GAT IOP means) plotted on the × axis and the standard error of the differences in means plotted on the Y axis. Quantitative assessment of potential publication bias was assessed by the classic Rosenthal Fail-Safe N and Egger Test of the Intercept.

2.9. Software

Comprehensive Meta-Analysis, Version 3.0 (Biostat, Englewood, NJ)^[51] was used to conduct the meta-analysis, conduct assessment of publication bias, complete meta-regression analyses, and generate prediction intervals for the summary effect size. Microsoft Excel for Mac Version 16.48 was used to generated additional descriptive summary statistics. Power analysis for the meta-analysis was conducted using open source metapowerR software script.^[52]

3. Results

The primary electronic search conducted, using the specified key words resulted in 11,532 hits. Thirty-two (n = 32) additional records were identified through gray literature searches and other sources. A total of 11,564 records were identified. The secondary electronic search, applying specific inclusion and exclusion criteria and removing duplicate records, reduced the count to 65 hits. Manual inspection of the reference lists in each of the 65 studies, resulted in the identification of 1 additional study, resulting in a total of 66 studies. The list of 66 studies was manually inspected for meeting inclusion and exclusion criteria and completeness for data extraction. Forty-four (n = 44) studies were excluded for failure to satisfy specific inclusion or exclusion criteria or inability to extract the required information. Twenty-two (N = 22) primary studies met all inclusion, exclusion, quality control criteria, and were entered for meta-analysis. See Figure 1, preferred reporting items for systematic reviews and meta-analysis protocol flow diagram.

Figure 1.

PRISMA-P flow diagram. PRISMA-P = preferred reporting items for systematic reviews and meta-analysis protocol.

Descriptive characteristics of the 22 primary studies: Fifteen (n = 15) different countries are represented; Brazil, Canada, China, India (n = 2 studies), Israel, Japan, Korea (n = 2 studies), Lithuania, Nigeria, Saudi Arabia, Sweden, Switzerland, Turkey (n = 5 studies), United Kingdom, United States of America (n = 2 studies). Five (n = 5) different location settings in which IOP were measured; hospital-Department of Ophthalmology (n = 8), out-patient clinic-ophthalmology (n = 6), out-patient clinic-internal medicine (n = 2), clinician’s office (n = 2), university research laboratory (n = 4). Total number of eyes from which IOP were measured = 2114. Mean age of subjects = 44.4 years, SD = 13.0 years. Gender: males = 729, females = 1090, unknown = 116. Mean CCT = 542.65 microns, SD = 9.57 microns. See Table 1, Primary studies descriptive characteristics.

Table 1.

Primary studies descriptive characteristics.

Study number	First author of primary study	Year of publication	Journal of publication	Country of study origin	Location setting of measurements	Number of eyes
1	Kim, R.[12]	2011	Current Eye Research	Korea	Hospital-Department of Ophthalmology	72
2	Eriksson, E.[25]	2011	International Journal of Ophthalmic Practice	Sweden	University research lab- Optometry Unit Department of Clinical Neuroscience	32
3	Schweier, C.[10]	2013	BMC Ophthalmology	Switzerland	Hospital-Department of Ophthalmology	36
4	Yilmaz, A.[27]	2014	Clinical Ophthalmology	Turkey	Out-patient clinic-Ophthalmology	200
5	Song, Y.[9]	2014	BMC Ophthalmology	Korea	Clinician’s office	10
6	Barkana, Y.[11]	2014	Clinical & Experimental Ophthalmology	Israel	Hospital-Department of Ophthalmology	21
7	Tai, T.[29]	2015	Journal of Clinical and Experimental Ophthalmology	USA	Out-patient clinic-Ophthalmology	50
8	Altinkaynak, H.[36]	2015	Eye	Turkey	Hospital-Department of Ophthalmology	49
9	Acar, D.[16]	2016	Gloko-Katarakt	Turkey	Hospital-Department of Ophthalmology	195
10	Galgauskas, S.[17]	2016	Investigative Ophthalmology and Visual Science	Lithuania	Hospital-Department of Ophthalmology	78
11	Berk, T.[4]	2016	Journal of Glaucoma	Canada	Clinician’s office	40
12	Frampton, P.[7]	2017	Dissertation-Aston University, Birmingham, UK	United Kingdom	University research lab- Department of Ophthalmology	91
13	Magela-Vieira, G.[19]	2018	Revista Mexicana de Oftalmologia	Brazil	Out-patient clinic-Ophthalmology	274
14	Wong, B.[18]	2018	Journal of Glaucoma	USA	Out-patient clinic-Ophthalmology	26
15	Dey, A.[8]	2018	Optometry Visual Science	India	Out-patient clinic-Ophthalmology	82
16	Osman, E.[33]	2018	Middle East African Journal of Ophthalmology	Saudi Arabia	Out-patient clinic-Internal Medicine	92
17	Kato, Y.[13]	2018	International Ophthalmology	Japan	Hospital-Department of Ophthalmology	60
18	Yildiz, A.[34]	2018	Medicinski Glasnik	Turkey	University research lab-Department of Ophthalmology	50
19	Prasanthi, M.[38]	2019	Journal of Medical Science and Clinical Research	India	Out-patient clinic-Ophthalmology	112
20	Dervisogullari, M.[20]	2019	Annals of Clinical and Analytical Medicine	Turkey	University research lab-Department of Ophthalmology	255
21	Yeh, S.[39]	2021	Journal of Chinese Medical Association	China	Out-patient clinic-Primary Care	30
22	Okudo, A.[44]	2021	Journal of Dental and Medical Sciences	Nigeria	Out-patient clinic-Ophthalmology	259

A priori power analysis, assuming small effect sizes (D = 0.20), high heterogeneity (I² = 0.80) revealed a minimum of 20 primary studies were required to ensure overall power for a random-effects model to be equal to or greater than 0.80.

Meta-analysis of 22 primary studies, conducted from 15 different countries, measured in 5 different clinical and research settings, during the period 2011 to 2021, comparing the raw mean difference in IOP, as measured by TP and GAT, in healthy adults, and applying a random effects model, revealed a statistically significant difference between the 2 measurement procedures. Tono-Pen IOP measurements were higher than GAT IOP measurements. Point estimate for the summary effect size = −0.73 mm Hg, SE = 0.34, 95% CI lower limits (LL) = −1.39, upper limits (UL) = −0.07, z = −2.17, P = .03. Q = 1243.26, df = 21, P = .001, I² = 98.31, Tau² = 2.40, SE = 1.05, Tau = 1.55. The prediction interval for the true effect size in 95% of all comparable populations is LL = −4.03 to UL = 2.58. See Figure 2, primary studies’ statistics and summary effect size point estimate with forest plot.

Figure 2.

Primary studies’ statistics and summary effect size point estimate with forest plot. Negative difference in mean values indicate TP values greater than GAT. Positive differences in mean values indicate GAT values greater than TP. GAT = Goldmann applanation tonometry, TP = Tono-Pen XL or Tono-Pen AVIA.

Sensitivity analyses revealed no meaningful change in overall summary results, with the exclusion of any single primary study, with values ranging from a raw mean difference point summary estimate equal to −0.54 with 95% CI LL = −1.06, UL = −0.03 to a raw mean difference point summary estimate equal to −0.88 with 95% CI LL = −1.49, UL = −0.26.

Primary studies quality assessment, using the PSQAC, revealed overall high quality for all primary studies included in the meta-analysis. Mean PSQAC score = 24, SD = 0.

Potential publication bias assessment by way of visual inspection of the conventional funnel plot, revealed no evidence of significant potential publication bias. Quantitative assessment of potential publication bias assessed by the classic Rosenthal Fail-Safe N method revealed a Fail-Safe N = 1589, z = −16.77, P = .001, and no evidence of significant publication bias. Additional quantitative assessment of potential publication bias was assessed by the Egger’s Test of the Intercept and revealed no evidence of significant publication bias, intercept = −0.99, 95% CI LL = −8.74, UL = 6.75, t = 0.27, df = 20, P = .40 (1-tailed).

Meta-regression analysis revealed statistically significant differences between IOP measured by TP and GAT, when IOP differences were compared between countries, Q = 79.42, df = 14, R² analog = 0.75, P = .001. See Figure 3, Differences in raw mean IOP, as measured by TP and GAT, by origin of the country, in which the primary study was conducted.

Figure 3.

Difference in raw mean IOP (mm Hg), as measured by TP and GAT, by the country in which the primary study was conducted. Circles represent single primary studies. Size of circle represent the number of eyes measured in each single primary study. Horizonal line segments represent the raw mean difference of IOP for all primary studies reported from each individual country. Negative values represent TP greater than GAT. Positive values represent GAT greater than TP. GAT = Goldmann applanation tonometry, IOP = intraocular pressure, TP = Tono-Pen XL or Tono-Pen AVIA.

Secondary meta-regression analysis revealed no statistically significant difference between TP and GAT measurements, when IOP raw difference measurements were evaluated as a function of mean age. Twenty-two (n = 22) of the 22 primary studies reported mean ages and associated SDs. Mean age = 44.4 years, SD = 13.0 years, Mean age coefficient = 0.04, SE = 0.03, 95% CI LL = −0.01 UL = 0.09, z = 1.44, R² analog = 0.01, P = .15.

Secondary meta-regression analysis revealed no statistically significant difference between TP and GAT measurements, when IOP raw difference measurements were evaluated as a function of mean CCT. Only 13 of the 22 primary studies reported mean CCT measurements. Mean CCT = 542.65 microns, SD = 9.58 microns, Mean CCT coefficient = −0.04, SE = 0.03, 95% CI LL = −0.10, UL = 0.02, z = −1.32, R² analog = −0.11, P = .19.

Secondary meta-regression analysis revealed no statistically significant difference between TP and GAT measurements, when IOP raw difference measurements were evaluated as a function of setting locations, in which IOP were measured. Q = 2.47, df = 4, R² analog = −0.17, P = .65.

4. Discussion

Meta-analysis of 22 primary studies, originating from 15 different countries, measuring raw mean IOP differences measured by TP and GAT, in the healthy adult population, appearing in the professional and medical literature, during the 10-year period 2011 and 2021, offers an updated presentation and synthesis of primary study findings.^[48]

Meta-analysis answers the first research question. Yes, there is a statistically significant difference between measurement of mean IOP, when measurements are made by the TP and GAT, in the healthy adult population. Intraocular pressures (IOP), when measured by Tono-Pen are marginally higher than IOP measured by GAT, in the healthy adult population.

Meta-analysis answers the second research question. In light of a statistically significant observed difference between raw mean IOP, when measured by TP and GAT, in the healthy adult population, the raw mean difference is less than that which can be expected by measurement error alone. The observed raw difference of −0.73 mm Hg is well within the 2.0 mm Hg difference criterion, eye care professionals apply when interpreting TP or GAT IOP for clinical purposes and attribute to measurement error, with no clinical significance.

These findings are consistent with Minckler et al initial report, now published over 34 years ago, comparing IOP measured by TP and GAT in patient populations, “…with various ocular diseases, whose corneas permitted accurate Goldmann readings…” and when measured IOP were between 6 and 24 mm Hg.^[2] Minckler, et al reported a statistically significant difference in pressure measurements, between TP and GAT of 1.7 mm Hg, with TP values being higher than GAT values.^[2] The authors interpreted the observed IOP difference between TP and GAT as “… not clinically relevant…,” which is the same conclusion we made from our meta-analysis.^[2]

Meta-regression analysis, evaluating the association of raw mean differences in IOP, as a function of country, revealed considerable variability between primary studies. We have been unable to identify with confidence, specific factors most responsible for the identified variability. It might be due to differences in the subtleties in how the IOP measurements were made or differences in biomechanical properties of the eyes measured, as a function of biomechanical differences associated with racial diversity within and between countries.^{[7,15,20–22]}

Meta-regression findings indicate, the setting in which IOP measurements are made, for example, hospital, out-patient clinics, clinician offices, or university-based research laboratories, does not contribute significantly to observed differences in IOP, when measured by competent evaluators and in accordance with manufacturer’s recommended measurement procedures, in the healthy adult population.

Overall, results indicate IOP measurements can reasonably be made with confidence, in the healthy adult population, using the portable, cost effective, accurate, reliable, quickly, and easily administered TP procedure and as an alternative to the conventional GAT measurement procedure.

These findings have important implications for the nonophthalmologist physician, particularly primary care physicians, choosing to screen IOP in their patients with confidence, especially in rural areas and underserved populations, without the need for expensive, office-based equipment, and the highly specialized training of ophthalmologists or optometrists. Findings provide additional evidence TP IOP measurements, much preferred by children, elderly, movement disordered, and uncooperative patients, might reasonably be used as an alternative measurement procedure to conventional GAT measurements for early detection and screening purposes.

Findings have further implications, given TP measurements can potentially be made in many patient populations in which GAT measurements are impossible, for example, due to lack of necessary equipment such as a slit lamp or nonavailability of highly trained eye care professionals, or patients presenting with specific conditions, for example, corneal edema, corneal infections, corneal epithelial defects, ocular chemical burns, scarred cornea, or patients unable to participate in GAT measurement due to physical disabilities or lack of cooperation during the examination.

Given the TP was never intended to replace the GAT, perhaps it is time to acknowledge advancements in technology and the many benefits of making a low cost, portable, safely administered, reliable, digital, objective, instrument (TP), which can be used to measure IOP in healthy adults, across a wide range of clinical settings, without expecting the TP to generate IOP measurements which mirror subjective GAT IOP measurements. Perhaps it is time to move forward, embrace newer technologies, and make IOP measurements available to the many primary care physicians and other health care providers, working in rural areas and serving the many underserved populations throughout the world.

4.1. Limitations

By design, this meta-analysis restricted itself to published and gray literature appearing in the 10 years, 2011 and 2021. The study only considered primary studies from which required data could be extracted with confidence and restricted to only the healthy adult population. Findings were not interpreted within the matrix of any of the popular current models offered to explain differences in IOP, rather the data is presented as foundation upon which future primary studies, which do evaluate specific casual factors, might be compared.

The study employed an aggregate data meta-analysis procedure. While this procedure accounts for over 95% of published meta-analyses, the statistical procedure is not without limitations.^[53,54] Whenever aggregated study data is analyzed, information contained within individual participants data points, from each primary study is lost. Conducting an individual participant data meta-analysis could have provided additional information and perhaps reduce the probabilities of making inferences about any single individual, based on aggregated group data.

CCT data was available from only 13 of the 22 primary studies, submitted to meta-analysis. The finding of no statistically significant linear association between CCT and IOP, in the healthy adult population, should be interpreted with caution, even though the finding is consistent with findings reported in a meta-analysis by others.^[55] The relationship between CCT and IOP has been well studied in patient populations and indicate CCT is an important moderator variable, whenever applanation tonometry is used. In brief, in multiple patient populations, as CCT increase from normal reference values IOP gradually increase, whereas as CCT decrease from normal reference values, IOP gradually decrease.^[35,56,57] The relationship between CCT and IOP is much less clear in the healthy adult population. Additional investigations should help clarify the relationship in the healthy adult populations.

5. Conclusions

Meta-analysis reveals IOP measured by TP to be marginally higher when compared to IOP measured by GAT, in the healthy adult population. The difference in IOP is statistically significant, however, from a clinical practice perspective, TP and GAT produce similar IOP measurements, in the healthy adult population. Significant variability exists in IOP measurements made across countries, when restricted to healthy adults. IOP measurements completed in a research laboratory are similar to measurements made in clinical settings. Findings have implications for primary care physicians and others requiring a portable, inexpensive, reliable, and easily administered instrument to assess IOP, particularly in developing countries, where access to vision care is limited or simply unavailable.

Author contributions

Conceptualization: William J. Keller.

Data curation: William J. Keller.

Formal analysis: William J. Keller.

Investigation: William J. Keller.

Methodology: William J. Keller.

Project administration: William J. Keller.

Supervision: William J. Keller.

Validation: William J. Keller.

Writing – original draft: William J. Keller.

Writing – review & editing: William J. Keller.