CONTRIBUTORS

This textbook was written primarily as a teaching aid for undergraduate optometry students and for practitioners wishing to review their clinical practice. Although it is not intended to be inclusive of all techniques that can be performed in primary eye care examinations, most popular procedures are covered. For each test, brief background information is provided, then a discussion of the function or state assessed, the advantages and disadvantages of the particular test, the measurement procedure, how to record the results, how to interpret them and a list of errors most commonly made by students, with an attempt to put the most common first. The recommended tests and procedures are explicitly based on evidence from the research literature whenever possible, rather than just clinical experience. Chapter 1 includes a review of how clinical tests and procedures are assessed in the research literature and how such reports should be critiqued. It also discusses the theory behind the use of screening tests and their use in primary eye care exams. Test procedures are grouped within the following six chapters. Chapter 2 compares the various formats of an eye examination and introduces the communication skills needed in primary eye care. A discussion of the case history and how it should be performed completes the chapter. Tests are subsequently grouped together in terms of which system they assess: visual function (Chapter 3), refractive correction (Chapter 4), binocular vision (Chapter 5) and ocular health (Chapter 6). This layout was chosen because the organisation of the book is directed towards the assimilation of a problem-oriented approach (Section 2.1.3) that is built upon a systems examination (Section 2.1.2). Grouping the tests in this way, rather than in the order they are typically used in an eye examination, may also help students to better appreciate the relationship between the various tests that assess a particular system. The book is completed in Chapter 7 with an introduction to some physical examination procedures that may be used in primary care eye

examinations. A list of key references and a Bibliography/Further reading section is provided at the end of each chapter. References have been kept to a minimum, with the most recently published references often chosen as these will usually cite earlier published reports.

The third edition differs from the second edition in several ways:

■Video-clips of various clinical procedures

and fundus and slit-lamp photographs are provided on the website .

■Significantly more diagrams and photographs have been included within the textbook, including several colour panels.

■There is one new chapter that introduces physical assessment procedures that could be used in primary eye care, including sphygmomanometry and carotid artery assessment. In addition, there are new sections that introduce communication skills and variations in appearance of the normal, healthy young and elderly eye.

■The book has a new format with numbered sections and subsections and similar headings for subsections to allow easier manoeuvring around the book.

COMMENTS AND SUGGESTIONS FOR FUTURE EDITIONS

The advantages and disadvantages of each procedure are provided and where possible, the measurement procedure is based on evidence from the research literature. However, there is no doubt that tests and test methodologies have been included which may reflect our biases due to our particular training, research and clinical experience. There may also be errors and omissions. We therefore welcome any comments and suggestions that

would improve any further editions of this textbook. Please write to the editor, Professor David Elliott at the Department of Optometry, University of Bradford, Bradford, BD7 1DP, UK (e-mail: d.elliott1@bradford.ac.uk).

INFORMATION RELEVANT TO STUDENTS

The recommendations in this manual are just that and not hard and fast rules. There are many ways of conducting an eye examination and different ways to properly perform various tests or procedures. Any methods or tests that are not included

in this manual are not in any way ‘wrong’. Indeed, the tests and test methodologies included no doubt reflect our biases due to our particular training, research and clinical experience. In particular, in university primary care clinics it is the supervising clinician’s decision as to which techniques or tests should be used in an eye examination. They are taking legal responsibility for the examination. If they indicate that a particular test needs using, use it! Once the patient has left and you are discussing the case with your supervisor, to further your learning, you should ask them about the advantages and disadvantages of their suggested technique.

We wish to thank Kathy Dumbleton, Ken Hadley, Natalie Hutchings, Edward Gilmore (University of Waterloo), Matt Cufflin, Elizabeth Richardson, Sem Sem Chin and Annette Parkinson (University of Bradford) for help with the video clips; the University of Bradford, Flinders Medical School and University of Waterloo eye clinics, Ketha Sivasegaran and Graham Mouat for help with the photographs; Anne Weber (University of Waterloo) for several of the figures; Walter Mittelstaedt for recording the Korotkoff sounds; Mary Elliott, Niall Strang (Glasgow Caledonian University) and

Cathy Starling for valuable comments on earlier versions of the third edition; Anya Bykar for collating many useful suggestions regarding additions to the text and new formatting and Mark Hurst and Barry Winn for their contributions to earlier editions of the book and the students and retired volunteer patients of the University of Bradford for sitting as subjects for many of the photographs and video clips. Finally we wish to especially thank our families and partners for their support and their understanding of the time commitment required to produce this textbook.

1.1.Reviewing the research literature 1

1.2.Routine screening 5

1.3.Bibliography and Further reading 8

1.4.References 9

Evidence-based primary eye care means integrating individual clinical expertise with the best currently available evidence from the research literature. What should always be avoided is the use of exam procedures purely because of tradition or habit.

1.1 REVIEWING THE RESEARCH

LITERATURE

The research literature should be regularly reviewed. There may be reports of newly developed techniques or instruments that are superior to the ones you typically use or even studies indicating that old and forgotten tests are actually better than commonly used ones (e.g. Rainey et al. 1998).

1.1.1 What makes a good research report?

Unfortunately, not all research reports necessarily provide accurate information. A study could be flawed for a variety of reasons (Harper & Reeves 1999, Lai et al. 2006), including a lack of exploration or misinterpretation of previous literature, poor study design, a subject group that is too small or biased, poorly or inappropriately performed tests, an unreliable or inappropriate gold standard test, inappropriate or limited statistical analyses and exaggerated conclusions. In theory, all of this should be picked up in the review process. All papers submitted to academic journals are sent to two or three reviewers with expertise in the area. They send comments to the editor, who decides whether to accept the paper, accept it with minor

revision, accept after major revision and a re-review or reject the paper. However, this process cannot always be perfect and it is useful to be able to critique a research paper, rather than just accept its conclusions. Various criteria can be used to assess the methodological quality of research articles (e.g. Harper & Reeves 1999, Lai et al. 2006) and a high quality paper should include the following:

■The paper should be easy to read and understand. Particularly in the area of the assessment of clinical techniques, there should be little that a clinician cannot understand. The rationale behind any complicated statistical analyses should be explained in a simple way. A paper

that is difficult to understand often indicates a poorly written paper rather than any lack of understanding on the part of the reader.

■The introduction of a paper should include the purpose of the study and discuss pertinent previous work.

■The methods section should be clear and precise. Another researcher should be able to replicate the study from the information provided in the methods section. It is usually necessary to randomise the order in which tests are performed to ensure that there are no significant learning or fatigue effects that could affect the data.

■In studies where tests are compared against a gold standard, the clinicians should be blind to the results from the other test.

■The subject sample should be clearly outlined: Why was that particular group of subjects chosen and how were they recruited? A description of the demographics of the group should be provided. A sufficiently large sample and a broad spectrum of subjects should be used to ensure no recruitment bias.

■The authors may highlight the limitations of the study. The majority of research studies have some limitations and it is very helpful to the reader if the authors indicate them. It also suggests that the authors are not exaggerating the findings of their study.

The research literature that describes a clinical test’s usefulness often uses assessments of validity, discriminative ability and/or repeatability and these are discussed in detail below.

1.1.2 Assessment of validity

A test is valid if it measures what it says it measures. This is often indicated by how closely the test’s results match those from a ‘gold standard’ measurement. For example, the validity of new tonometers has usually been determined by how similar the results are to Goldmann Applanation Tonometry (GAT) readings. This highlights the possible problem with validity measurements in that they are only as good as the gold standard. If the gold standard is flawed, an excellent instrument could be reported to have poor validity. For example, the GAT is known to give high intraocular pressure readings on thick corneas and low readings with thin corneas (Doughty & Zaman 2000). This has tended to be ignored until recently when significant reductions in IOP have been found after refractive surgery (see section 6.13). If a tonometer that was resistant to corneal thickness effects had been compared to GAT, it would have been shown to be variable. The conclusion would have been that the new tonometer was somewhat variable compared to GAT. Inappropriate gold standards can also be chosen. For example, Calvin et al. (1996) used the von Graefe phoria measurement as the gold standard test to assess the usefulness of the cover test and suggested that the cover test was occasionally inaccurate. The gold standard in this area should be the cover test and not the von Graefe. The cover test is the only test that discriminates between strabismus and heterophoria, it is objective and not reliant on subject responses and a subsequent study has shown it to be far more repeatable than the von Graefe, which appears to be unreliable (Rainey et al. 1998). The 1996 study should have used the cover test as the gold standard and they would then have reported the limitations of the von Graefe. The gold

standard test must also be appropriately measured. For example, Salchow et al. (1999) compared autorefraction results after LASIK refractive surgery against the gold standard of subjective refraction. Subjective refraction was an appropriate choice of gold standard, but was inappropriately measured. The authors concluded that autorefraction compared very poorly against subjective refraction post-LASIK. However, inspection of the results clearly indicates that the majority of the subjective refractions (particularly of the hyperopes) provided a result of plano. This suggests that a normal or near normal visual acuity resulted in a ‘brief’ subjective refraction and a result of plano.

The approach of determining the validity of new tests by comparing them to a gold standard can lead to the invincibility of the gold standard. For example, the limitations of GAT have been known for many years (Doughty & Zaman 2000; see section 6.13), but these limitations have tended to be ignored. The use of subjective refraction as a gold standard assessment of refractive error has meant that there has been little or no comparison of the various methods used in subjective refraction. Subjective refraction is the gold standard, so how can we compare the various methods? Previous studies have tended to compare the various tests against each other. For example, West & Somers (1984) compared the various binocular balancing tests and found that they all gave similar results and concluded that they were therefore all equally useful. Johnson et al. (1996) reported a similar finding when comparing subjective tests for astigmatism. These are not surprising findings, and provide little information regarding which test is the most useful. A very good but under-utilised approach is to use some measure of patient satisfaction as the gold standard. If patients are happy with the results of subjective refraction using a particular test, then the test must be providing appropriate results. Hanlon et al. (1987) used this approach in a comparison of techniques used to determine the reading addition. They examined 37 patients that were dissatisfied with the near vision in their new spectacles. From the case history information in the review (recheck) examination, it was determined whether the improper add was too low or too high. For each patient, their reading addition was then determined using the four methods of age, 1⁄2 amplitude of accommodation, NRA/PRA (negative relative accommodation/positive relative accommodation) balance and binocular

cross-cylinder. The percentage of adds for each test that gave the same result as the improper add or worse (higher than an improper add determined too high or lower than an improper add determined as too low) was calculated. The results are discussed in section 4.21. The study would have been even better if they had confirmed that the patients were subsequently satisfied with their changed spectacles (i.e. that it really was the gold standard). This technique of using patient satisfaction as the gold standard test could be usefully employed to compare the various techniques used in distance refraction, particularly those that assess astigmatism and binocular balancing.

A test’s validity is often described by the correlation coefficient between the results from the test and the gold standard. However, care must be taken when considering such analyses as correlation coefficients are very much affected by the range of values used in the analysis (Bland & Altman 1986, Haegerstrom-Portnoy et al. 2000). If a small range of values is used in calculations (for example, visual acuity values between 6/4 and 6/9), the correlation coefficient is likely to be much smaller than if a larger range (e.g. 6/4 to 6/120) is used. This is highlighted in Figure 1.1, which shows a plot of correlation coefficients between visual acuity and other clinical measures of visual function versus the range of visual acuity of subjects used in the studies. Plotting frequency distributions of the differences between test and standard provides more useful information as well as being independent of the range of values. For example, although a correlation coefficient of 0.97 between the equivalent sphere predicted by an autorefractor and subjective refraction does indicate that the two provide very similar results, knowing that 83% of autorefractor results are within 0.50D of subjective refraction and 96% are within 1.00D is much more meaningful (Elliott & Wilkes 1989). A similar analysis, commonly known as a Bland–Altman plot (Bland & Altman 1986) shows the 95% confidence limits of the difference between the test and the gold standard (Fig. 1.2). Of course, this analysis can be used only if the gold standard and test are measured in the same units.

1.1.3 Discriminative ability

Discriminative ability indicates how well a test discriminates between normal and abnormal eyes. In

Range of x axis (log units)

Fig. 1.1 Correlation coefficients from the literature between high contrast visual acuity and other spatial vision measures are plotted as a function of the range of high-contrast acuities in those studies. The solid line is the regression line and the correlation coefficient for the plotted data points is 0.91. (Redrawn from Haegerstrom-Portnoy et al. 2000, with permission of The American Academy of Optometry).

DR Spherical equivalent (D)

Fig. 1.2 A Bland–Altman plot of test – gold standard differences showing the validity of automated subjective refraction (ASR) using a comparison against optometrist subjective refraction. (Reprinted from Sheedy et al. 2004, with permission of The American Academy of Optometry).

published results of clinical studies, it is often reported that a significant difference was found between a group of patients with an ocular abnormality and a control group. It should be noted that

this only indicates that there is a difference between the averages of the two groups. It does not indicate how well the test predicts whether an individual patient has the abnormality or not. Sensitivity and specificity values and plots of one against the other for a range of cut-off values in receiver operating characteristic (ROC curves, Fig. 1.3) or kappa functions (Gilchrist 1992) are usually presented to represent discriminative ability. Sensitivity is the proportion of abnormal eyes correctly identified, and specificity is the proportion of normal eyes correctly identified. They are discussed in detail in section 1.2.2. It should again be noticed that to determine whether a test can correctly identify ‘abnormal’ or ‘normal’ eyes, a ‘gold standard’ test (or battery of tests) must have been used to classify eyes as abnormal or normal. The assessment of discriminative ability is therefore also dependent on the validity and reliability of the gold standard test. For example, new instruments or techniques that attempt to identify primary open-angle glaucoma are typically assessed against classifications of patients into glaucomatous and control groups by clinical evaluation of optic nerve head assessment, visual fields and tonometry (e.g. Hutchings et al. 2001). This classification will therefore differ depending on the type of visual field assessment and tonometer used and the skill/experience of the clinician.

1.1.4 Repeatability

Repeatability assesses the variability of results between testing occasions. Reliability, also called precision (the ability of a measurement to be consistently produced) or repeatability, has most often been described in terms of correlation coefficients. The limitations of correlation coefficients have already been discussed in section 1.1.2 and it can often be better to assess repeatability in terms of the coefficient of repeatability (COR). This represents the 95% confidence limits of the difference between the test and retest scores (Bland & Altman 1986, Fig. 1.4). For example, correlation coefficients between test and retest visual acuity scores have been found to be 0.95 (unaided) and 0.90 (aided), respectively (Elliott & Sheridan 1988). This suggests that the repeatability of the two tests is similar and may be slightly better for unaided

Fig. 1.3 Receiver-operating characteristic (ROC) curve showing the discriminative ability of visual acuity systems. (Redrawn from Tong et al. 2004, with permission of The American Academy of Optometry)

acuity. However, COR results were 0.07 logMAR (test and retest scores generally differ by less than a line) for aided acuity and 0.21 logMAR (test scores can differ by up to two lines on retest) for unaided acuity, showing a much better repeatability for aided acuity (Elliott & Sheridan 1988). The high