Systematic review of the agreement of tonometers with Goldmann applanation tonometry
Methological quality of the review: Medium confidence
Author: Cook JA, Botello AP, Elders A, Fathi Ali A, Azuara-Blanco A, Fraser C, McCormack K, Margaret Burr J; Surveillance of Ocular Hypertension Study Group
Geographical coverage: United Kingdom (UK)
Sector: Glaucoma
Sub-sector: Diagnosis of intra-ocular pressure
Equity focus: Age group of 16 years or more were included
Review type: Other review
Quantitative synthesis method: Meta-analysis
Qualitative synthesis method: Not applicable
Background: There is a debate about the role and optimal organisation of a monitoring service for patients with ocular hypertension and whether other health professionals (e.g. nurses and optometrists) might be safely involved in measuring intra-ocular pressure (IOP). Therefore, a tonometer needs to be accurate, precise and easy to use.
Objectives: To assess the agreement of tonometers available for clinical practice with the most commonly accepted reference device – Goldmann applanation tonometer (GAT).
Main findings: In this review, 102 studies – including 130 paired comparisons – were included in which eight tonometers (dynamic contour tonometer, non-contact tonometer (NCT), ocular response analyser, Ocuton S, handheld applanation tonometer (HAT), rebound tonometer, transpalpebral tonometer and Tono-Pen) were used. Authors note that poor reporting limited the assessment of the quality of included studies and the synthesis of the evidence.
The agreement (95% limits) was varied across tonometers: 0.2 mm Hg (3.8 to 4.3 mm Hg) for the NCT to 2.7 mm Hg (4.1 to 9.6 mm Hg) for the Ocuton S. The estimated proportion within 2 mm Hg of the GAT ranged from 33% (Ocuton S) to 66% and 59% (NCT and HAT, respectively). Substantial inter- and intra-observer variability was observed for all tonometers. Based on these findings, authors concluded that non-contact tonometer NCT and HAT are the closest measures to the current reference standard GAT. However, there is a need to standardise the reporting of comparative studies of tonometers.
Reviewers also noted that an in-depth exploration of factors which could influence the pressure measurements is needed for the reference standard and candidate tonometers. This could be addressed by a large primary study, but also has the potential to be explored in a meta-analysis of individual patient data.
Methodology: The following databases were searched from 1987 to February 2010: MEDLINE, MEDLINE In-Process, EMBASE, Science Citation Index, BIOSIS and the Cochrane Central Register of Controlled Trials. Also, key journals were screened for relevant in-press publications and current research registers, including Clinical Trials, Current Controlled Trials and WHO International Clinical Trials Registry Platform. The reference list of included studies was also checked.
The following studies were included: studies which directly assessed the agreement of one or more tonometers compared with the reference standard tonometer (GAT) in the same group of people (paired data). Case-control and cohort design and population studies were eligible provided they incorporated paired data for the GAT and at least one other tonometer that could be used in clinical practice.
Screening, quality assessment and data extraction were conducted by two reviewers independently. Quality assessment was conducted using a checklist adapted from QUADAS tool for diagnostic studies and a checklist for agreement studies. Pooled estimates of mean difference and random error were calculated using the DerSimonian and Laird random effects method, and other statistical methods were used for data analysis.
Authors conducted sensitivity analyses which consisted of a fixed-effect analysis or imputation of correlations using minimum correlation coefficients reported from studies assessing the same tonometer. Authors explored between-study heterogeneity by analysing forest plots and calculation of tau and I2 statistics. To explore possible reasons for heterogeneity, authors conducted sub-group analyses. Formal comparison between subgroups were not performed due to the large of level of heterogeneity in the main analyses.
Applicability/external validity: The review does not specifically discuss how generalizable the results are.
Geographic focus: It seems that the geographic focus of this study was the UK. However, outcomes of this study could be applicable to low/middle-income settings depending on the health system.
Summary of quality assessment: Overall, there is medium confidence in the conclusions about the effects of this study. Authors conducted a thorough search of the literature using several databases for published studies, checked references lists and reviewed websites of key journals. However, publication bias should not be excluded from this study as authors restricted the search to studies written in English only and to published studies. The likelihood of bias of the included studies was addressed and the methodological quality of the included studies was appropriately assessed. Authors clearly described the method of analysis and addressed issues of heterogeneity in the review.
