CHAPTER 7 Questions about diagnosis
examples of appraisals from different health professions
This chapter is an accompaniment to the previous chapter (Chapter 6) where the steps involved in answering a clinical question about diagnosis were explained. In order to further help you learn how to appraise the evidence for this type of question, this chapter contains a number of worked examples of questions about diagnosis from a range of health professions. The worked examples in this chapter follow the same format as the examples that are in Chapter 5. In addition, as with the worked examples that were written for Chapter 5, the authors of the worked examples in this chapter were asked not to choose a systematic review, but to instead find the next best available level of evidence to answer the clinical question that is in the worked example. This was done for the same reason that was given in Chapter 5—it is easier to learn how to appraise a systematic review of test accuracy studies if you have first learnt how to appraise a study about test accuracy. Chapter 12 will help you to learn how to appraise a systematic review.
Occupational therapy example
Clinical scenario
Clinical question
Among children with motor performance problems, is The Motor Performance Checklist as accurate as the Bruininks-Oseretsky Test of Motor Proficiency for identifying developmental coordination disorder?
Search terms and databases used to find the evidence
Database: PubMed—Clinical Queries (with ‘diagnosis category’ and ‘narrow search’ selected)
Search terms: (The Motor Performance Checklist)
The PubMed clinical queries diagnosis ‘specific’ filter automatically combines this phrase with the term ‘specificity’ in the title or abstract. An alternative search approach would have been to enter this phrase in the CINAHL database and combine it with the term ‘specificity’. This search retrieves four titles. The second title is obviously relevant so you obtain the full text of the article.
Article chosen
Gwynne K, Blick B. Motor performance checklist for 5-year-olds: a tool for identifying children at risk of developmental co-ordination disorder. J Paediatr Child Health 2004; 40:369–373.
Structured abstract
Study design: This study used a cross-sectional design to compare a new measure of motor performance with a ‘gold standard’ test for identifying children with developmental coordination disorder.
Setting: The study was conducted in schools in Sydney, Australia.
Participants: All 5-year-old children in a random sample of seven schools from 59 primary schools in the northern beaches sector of Sydney were invited to participate. Of the total population of children in the participating schools, 141 (60%) participated in the study (mean age 5 years and 5 months; 54% male). The prevalence of developmental coordination disorder in the study population was 4.2%.
Description of test: The Motor Performance Checklist is a 12-item instrument for identifying children at risk of developmental coordination disorder.
Diagnostic standard: The Bruininks-Oseretsky Test of Motor Proficiency Long Form. A composite standard score of 40 (one standard deviation below the mean) was used as the Bruininks-Oseretsky Test of Motor Proficiency Long Form cut-off/failure point to direct children to occupational therapy.
Main results: The checklist was found to have a sensitivity of 83% and a specificity of 98%. Positive predictive validity was found to be 72% and negative predictive validity 99%.
Conclusion: The Motor Performance Checklist has the potential to aid in identifying children who are in need of referral to community occupational therapy services.
Is the evidence likely to be biased?
Yes. The study report states that all children were tested using both measures. The reference standard used for this study was the Bruininks-Oseretsky Test of Motor Proficiency Long Form. It is a well-validated and frequently used measure for assessing motor performance difficulties in children.
No. A nurse was trained to administer the Motor Performance Checklist and an occupational therapist blinded to the Motor Performance Checklist results administered the Bruininks-Oseretsky Test of Motor Proficiency.
Yes. The study included children from a random sample of seven out of 59 primary schools in a district of Sydney, Australia. The article states that the population from which the sample was drawn was fairly homogenous with 11% from non-English-speaking backgrounds. Twenty percent of the population had a tertiary education. There was an approximately even distribution between male (54%) and female (46%) children who participated in the study.
No. Nurses who administered the Motor Performance Checklist were trained in its use for the purpose of this study and a reference to an article that describes the test procedure in detail is provided. However, the actual testing conditions were not clearly described. As the Bruininks-Oseretsky Test of Motor Proficiency is a standardised test it would had to have been carried out as per the standardised instructions.
What are the main results?
In this study, 6 (4.2%) children were identified by the Bruininks-Oseretsky Test of Motor Proficiency as having developmental coordination disorder. This study presents the sensitivity, specificity, predictive values and likelihood ratios for identifying developmental coordination disorder using the Motor Performance Checklist (see Table 7.1) compared with the Bruininks-Oseretsky Test of Motor Proficiency Long Form using a cut-off score of 40 points.
The Motor Performance Checklist has high specificity, which means that there would be very few false positives. The sensitivity of 83% is also reasonably high, meaning not many children who had developmental coordination disorder would be missed (few false negatives). The positive predictive value looks at the data in a different way: how to interpret the results for a given client whose true diagnosis we do not know when we have the test result. A positive predictive value of 72% means that we know the chance of a child having developmental coordination disorder after their score on the Motor Performance Checklist is positive is 72%. Similarly, based on the negative predictive value, the chance of their not having it after a negative test is better, at 99%. In other words, a negative test seems better at telling us the true diagnosis than a positive one. As you saw in Chapter 6, two things contribute to the predictive values: the quality of the test (how well it performs as described by the sensitivity and specificity) and the prevalence of the disorder. In this example only approximately 4% of children had the condition. This means that we can only generalise the predictive values to other populations that have similar condition prevalences.
Another way to deal with this is to use likelihood ratios, which use a clever algebraic approach enabling us not to have to rely on prevalence to describe the usefulness of a test, yet also employ both sensitivity and specificity. Thus the positive likelihood ratio is the likelihood of a positive test result in a child with the condition compared with the same likelihood in one without the condition. In this study the positive likelihood ratio is 41.5 [calculated as sensitivity ÷ (100 – specificity)]. Using the approximate guide values that were presented in Chapter 6, a positive likelihood ratio over 10 indicates that the test is extremely good for ruling in the presence of developmental coordination disorder if it is present. The negative likelihood ratio was 0.17 [calculated as (100 – sensitivity) ÷ specificity] which, again using the values presented in Chapter 6, indicates that it is a test that can also help rule out the presence of developmental coordination disorder.
How might we use this evidence to inform practice?
Although this study may be prone to some types of bias that are common in cross-sectional studies it was otherwise well-designed and you are reasonably confident about the results. There are three factors about this study to think about, though. First, the ability of The Motor Performance Checklist to identify children with developmental coordination disorder was restricted in this study to children who were 5 years old. Testing this measure with children from 4 to 10 years is needed as this is the age range that this assessment was designed to be used with. Second, the study reports a low prevalence of developmental coordination disorder and the authors state this is lower than reported in the literature. This means that, in populations with a higher prevalence of developmental coordination disorder, the positive predictive value (or the chance of the test being correct) will be greater than reported in this study. Finally, the brevity of this measure is appealing and the article also reports on the concurrent validity and reliability of this measure, which are other psychometric test properties that must be considered when considering using an assessment with clients. You think back to your original dilemma. Can you use The Motor Performance Checklist for identifying children with developmental coordination disorder? The results of this study are limited to children 5 years of age so, until further research is done that involves children of other ages, it may have limited, yet useful, value to your clinical practice.
Physiotherapy example
Clinical scenario
Clinical question
In athletes with shoulder pain, is there a manual test for diagnosing posteroinferior lesions of the labrum and what is its diagnostic utility?
Search terms and databases used to find the evidence
Database: PubMed—Clinical Queries (with ‘diagnosis category’ and ‘narrow search’ selected)
Search terms: posteroinferior (labral OR labrum)
The search returns four records. One of these articles seems very promising: a comparison of the ability of two manual tests to diagnose posteroinferior labral tears, with surgical observation as the gold standard. It looks highly relevant but you are concerned that your search is too narrow. You repeat the search, selecting the ‘broad, sensitive search’ option. This returns 17 records, but none of these is as relevant as the original article that you found.
Article chosen
Kim S, Park J, Jeong W et al. The Kim test: a novel test for posteroinferior labral lesion of the shoulder—a comparison to the jerk test. Am J Sports Med 2005; 33:1188–1192.
Structured abstract
Setting: Department of Orthopaedic Surgery at a hospital in Korea.
Participants: 172 adults awaiting arthroscopic examination for undiagnosed shoulder pain. Exclusion criteria were septic arthritis, fracture of the greater tuberosity, arthroscopic capsular release due to frozen shoulder, frozen shoulder and previous surgery.
Description of tests: For the Kim test, the client sits with the trunk against a backrest and the arm abducted to 90°. The examiner applies axial force along the humerus at the elbow to compress the glenohumeral joint and elevates the arm by 45°. With the other hand, the examiner applies downward and backward force to the upper arm. Sudden onset of pain indicates a positive test. For the jerk test, the client sits with the arm abducted to 90° and internally rotated 90°. The examiner stands behind and supports the scapula with one hand. With the other hand, axial force is applied at the elbow and maintained while the arm is horizontally adducted. Sharp pain indicates a positive test. Each test was performed by two independent examiners.
Diagnostic standard: Arthroscopic examination of the glenohumeral joint and subacromial space.
Main results: Thirty (17%) of the 172 participants had a posteroinferior labral lesion. The Kim test had sensitivity of 80% and specificity of 94%. The positive predictive value of the Kim test was 0.73 and the negative predictive value was 0.96. The jerk test had sensitivity of 73% and specificity of 98%. The positive predictive value of the jerk test was 0.88 and the negative predictive value was 0.95. The sensitivity in detecting a posteroinferior lesion increased to 97% when the two tests were combined.
Conclusion: The two tests, particularly in combination, have worthwhile clinical utility in the diagnosis of posteroinferior labral lesions.
Is the evidence likely to be biased?
Yes. All participants received the diagnostic tests of interest and the reference standard.
No. The results of the two manual tests (the tests of interest) could not have been influenced by the result of the reference standard (arthroscopic examination) because the manual tests were performed before the arthroscopy. The results of the arthroscopy could not have been influenced by the results of the manual tests because the surgeon was blinded.
Yes. The description of the study population is reassuring: eligible clients were recruited consecutively from an orthopaedic surgical clinic. All participants entered the study with preliminary clinical diagnoses that were potentially consistent but not definitive of the diagnosis of interest.
What are the main results?
Thirty (17%) of the 172 participants had a posteroinferior labral lesion. The article presents the sensitivity, specificity, positive predictive value and negative predictive value of the two tests, which are shown in Table 7.2. You also use these results to calculate the positive and negative likelihood ratios (see Table 7.2).
The positive predictive value of the Kim test indicates that, on average, 73% of people with a positive test result actually have a posteroinferior labral lesion. The positive predictive value of the jerk test indicates that, on average, 88% of people with a positive result actually have a posteroinferior labral lesion. Using the approximate guide values that were presented in Chapter 6, as both of the positive likelihood ratios are over 10, this indicates that these tests are very helpful for ruling in a posteroinferior labral lesion if it is present. Also, as the negative likelihood ratios are <0.5, this suggests that these tests are moderately helpful in ruling out a posteroinferior labral lesion.
How might we use this evidence to inform practice?
You are satisfied that the study is valid and the results are clinically useful. The study population was reasonably similar to the client to whom you are considering applying the tests, although in your practice not all clients are severe enough to be scheduled for arthroscopy, so the prevalence of posteroinferior labral lesions is likely to be lower. The tests can be performed quickly and without requiring specialised equipment. However, before applying these tests, you should consider other diagnoses. The client’s description of the location of the pain suggests that the posteroinferior labrum is involved. Furthermore, the following tests are negative: the apprehension test (for detecting anterior shoulder instability), the impingement test (for detecting rotator cuff inflammation or impingement), the horizontal adduction test (for assessing acromioclavicular joint impingement) and tests for superior labral lesions as mentioned in the clinical scenario. In doing this, you confirm that a posteroinferior labral lesion is the diagnosis that you primarily suspect. You decide to apply the Kim test and the jerk test to your client and use the results of these tests to guide the diagnosis and subsequent management of your client.
Podiatry example
Clinical scenario
You are a podiatrist working in a community health centre and you have just seen a 28-year-old male plumber with a diffuse scaling rash which covers most of his instep area, extending towards the digits of his right foot. It has been present for about 8 weeks, is itchy and there are some vesicles in the instep area. Several weeks ago he tried a topical steroid cream bought from the pharmacy. According to the client, the rash did not really improve and he stopped applying the cream. You suspect a fungal infection, suggest an antifungal cream that is available over the counter, and discuss foot hygiene with him. You wonder how accurate the diagnosis of tinea pedis (a fungal foot infection) is, based on clinical presentation.
Clinical question
In a person with suspected tinea pedis is clinical presentation as accurate as microbiological confirmation for establishing the diagnosis?
Search terms and databases used to find the evidence
Database: PubMed—Clinical Queries (with ‘diagnosis category’ and ‘narrow search’ selected)
Search terms: Tinea OR ‘athletes foot’
You click on ‘Details’ to check how PubMed has processed your search and notice that it has automatically searched the Tinea MeSH term. This search retrieves 33 titles, so you start looking through them. Several studies look promising but, when you read through their abstracts, you find that they are either comparing different laboratory methods for diagnosing mycoses or are case-control studies. You know that case-control studies can overestimate the accuracy of diagnostic tests, so you reject these. There is one study that, on the basis of the abstract, appears to compare the clinical diagnosis of tinea pedis with laboratory methods but the operating characteristics of the tests are not reported. You are really only then left with one possible study. This study investigated the diagnostic value of signs and symptoms compared to culture for diagnosing dermatomycosis in general practice. You consider whether this study, which appears to have looked at the diagnosis of fungal infection on any hairless part of the body, could be relevant to your clinical situation. As there do not appear to be any studies that have looked specifically at the clinical diagnosis of tinea pedis, you obtain the full text of this study.

Stay updated, free articles. Join our Telegram channel

Full access? Get Clinical Tree

