Does This Patient Have Clubbing?

THE RATIONAL CLINICAL EXAMINATION

Kathryn A. Myers, MD, EdM, FRCPC Context The association between digital clubbing and a host of diseases has been rec- Donald R. E. Farquhar, MD, SM, FRCPC ognized since the time of Hippocrates. Although the features of advanced clubbing are familiar to most clinicians, the presence of early clubbing is often a source of debate. CLINICAL SCENARIOS Objective To perform a systematic review of the literature for information on the Case 1 precision and accuracy of clinical examination for clubbing. A respiratory therapist asks you to see her asymptomatic 76-year-old mother in Data Sources The MEDLINE database from January 1966 to April 1999 was searched for English-language articles related to clubbing. Bibliographies of all retrieved articles consultation because she is concerned and of standard textbooks of physical diagnosis were also searched. that her mother has clubbing. The patient has increased curvature of the nails, Study Selection Studies selected for data extraction were those in which quanti- and you wonder whether other physi- tative or qualitative assessment for clubbing was described in a series of patients. Six- teen studies met these criteria and were included in the final analysis. cal examination techniques can help you decide whether clubbing is present. Data Extraction Data were extracted by both authors, who independently re- viewed and appraised the quality of each article. Data extracted included quantitative Case 2 indices for distinguishing clubbed from normal digits, precision of clinical examination for clubbing, and accuracy of clubbing as a marker of selected diseases. While performing a routine physical examination on a 65-year-old female Data Synthesis The profile angle, hyponychial angle, and phalangeal depth ratio can be used as quantitative indices to assist in identifying clubbing. In individuals without smoker with chronic obstructive pul- clubbing, values for these indices do not exceed 176°, 192°, and 1.0, respectively. When monary disease (COPD), you detect clinicians make a global assessment of clubbing at the bedside, interobserver agree- changes in the fingers suggestive of ment is variable, with ␬ values ranging between 0.39 and 0.90. Because of the lack of clubbing. You recall an association be- an objective diagnostic criterion standard, accuracy of physical examination for club- tween clubbing and certain types of pul- bing is difficult to determine. The accuracy of clubbing as a marker of specific underly- monary disease, and you wonder ing disease has been determined for lung cancer (likelihood ratio, 3.9 with phalangeal whether any further diagnostic evalu- depth ratio in excess of 1.0) and for inflammatory bowel disease (likelihood ratio, 2.8 ation of this patient is warranted. and 3.7 for active Crohn disease and ulcerative colitis, respectively, if clubbing is present). Conclusions We recommend use of the profile angle and phalangeal depth ratio as Why Is the Clinical quantitative indices in identifying clubbing. Clinical judgment must be exercised in de- Examination Important? termining the extent of further evaluation for underlying disease when these values exceed 180° and 1.0, respectively. Clubbing is one of those phenomena with JAMA. 2001;286:341-347 www.jama.com which we are all so familiar that we appear 1 to know more about it than we really do. 7,8 Samuel West, 1897 tion, clubbing usually represents the thy. Hypertrophic osteoarthropa- progression of established diseases, such thy, a systemic disorder affecting bone The association of clubbing with a host as cystic fibrosis or uncorrected cya- and joints, is most commonly associ- of infectious, neoplastic, inflamma- notic congenital heart disease. ated with bronchogenic carcinoma, but tory, and vascular diseases has cap- Digital clubbing is characterized by

tured the imagination of clinicians since the enlargement of the terminal seg- Author Affiliations: Department of Medicine, Divi- Hippocrates first described clubbing in ments of the fingers and/or toes that re- sion of General Internal Medicine, Queen’s Univer- a patient with empyema in the fifth cen- sults from the proliferation of the con- sity, Kingston, Ontario. 2 Corresponding Author and Reprints: Kathryn A. tury BC. Although clubbing can be a nective tissue between the nail matrix Myers, MD, EdM, FRCPC, Queen’s University, Hotel benign hereditary condition, the diag- and the distal phalanx. Although most Dieu Hospital, 166 Brock St, Kingston, Ontario, Canada K7L 5G2 (e-mail: [email protected]). nostic implications in an adult are such often symmetrical, clubbing can be uni- The Rational Clinical Examination Section Editors: that its detection should prompt con- lateral or even unidigital.5,6 Clubbing David L. Simel, MD, MHS, Durham Veterans Affairs Medical Center and Duke University Medical Center, sideration of the underlying etiology can occur in isolation or in associa- Durham, NC; Drummond Rennie, MD, Deputy Edi- (TABLE 1).3,4 In the pediatric population with hypertrophic osteoarthropa- tor, JAMA.

Downloaded From: http://jama.jamanetwork.com/ by a University of Washington Libraries User on 12/18/2012 CLUBBING

Pathophysiology Inspection Table 1. Conditions Associated With Acquired Clubbing Normally, the nail-bed thickness is less General Appearance. Inspection of the Neoplastic intrathoracic disease than 2.0 mm. Clubbed fingers studied fingers for clubbing can reveal abnor- Bronchogenic carcinoma at autopsy show not only a thickness malities in the nail-fold angles, and in Malignant mesothelioma Pleural fibroma greater than 2.0 mm, but also a lower the shape, depth, and width of the ter- Metastatic osteogenic sarcoma density of nail-bed connective tis- minal phalanges. In addition to the obvi- Suppurative intrathoracic disease sue.16 Morphologic findings include the ous changes in the shape of the termi- Lung abcess Bronchiectasis presence of primitive fibroblasts, el- nal phalanges in established clubbing Cystic fibrosis evated numbers of eosinophils and lym- (FIGURE A), close inspection of the Empyema Chronic cavitary mycobacterial or fungal phocytes, and increased caliber and cuticle may reveal a shiny and smooth infection number of blood vessels. Genetic pre- appearance. Lovibond23 described a lilac Diffuse pulmonary disease Idiopathic pulmonary fibrosis disposition, vagally mediated neural hue of the nail fold in clubbing, caused Asbestosis mechanisms, and the direct effect of tis- by increased vascularity in the connec- Pulmonary arteriovenous malformations sue hypoxia or of circulating vasodila- tive tissue. Although the increased nail Cardiovascular disease Cyanotic congenital heart disease tors that elude metabolism in the lung curvature seen in clubbed fingers has Infective endocarditis through right-to-left shunting have all been studied extensively using chord- Arterial graft sepsis* Brachial arteriovenous fistula† been proposed to explain the morphol- arc measurements and unguisom- Hemiplegic stroke† ogy. While there is experimental and eters, it is not easily measured at the bed- Gastrointestinal disease Inflammatory bowel disease clinical evidence to support each of side. Moreover, nail curvature tends to Celiac disease these hypotheses, it has not been pos- become more pronounced with age and Hepatobiliary disease Cirrhosis (particularly biliary and juvenile) sible to formulate a comprehensive can occur in the absence of other signs Metabolic disease theory of pathogenesis applicable to all of clubbing.5,24 Thyroid acropachy clinical circumstances.5,17-19 Nail-fold Angles. Inspection of clubbed *Associated with clubbing distal to graft sepsis. †Associated with unilateral clubbing. fingers reveals a number of abnormali- Symptoms ties in the angles made by the nail as it Clubbing is almost always painless, un- exits from the terminal phalanx. Lovi- it can occur in association with extra- less it is associated with hypertrophic bond23 popularized this as the profile sign pulmonary malignancies as well as osteoarthropathy. Symptoms of hyper- in his 1939 report on the diagnosis of nonmalignant pulmonary diseases.9 trophic osteoarthropathy include peri- clubbed fingers. He observed that in nor- Pachydermoperiostosis is a rare, con- articular pain and swelling, most of- mal fingers, the nail projects from the nail genital form of hypertrophic osteoar- ten in the wrists, ankles, knees, and bed at an angle of about 160°, but that this thropathy. Congenital clubbing, which elbows. Accordingly, the presentation angle approached 180° in clubbed fingers usually has its onset in childhood, may of hypertrophic osteoarthropathy can (Figure, B). Later, the hyponychial angle represent a limited form of pachyder- be confused with such primary rheu- was proposed as a more reliable sign than moperiostosis.5 matological disorders as rheumatoid ar- theprofileangleintheassessmentofclub- Unlike such physical findings as as- thritis.5 Many patients with clubbing ex- bing (Figure, B).11 cites and splenomegaly, the clinical im- press unawareness of any abnormality Phalangeal Depth Ratio. Estimation pression of clubbing cannot be veri- in their fingers. In one series of pa- of the phalangeal depth ratio can be used fied by simple imaging tests. Over the tients with clubbing, only 32 of 116 pa- to identify clubbing (Figure, C).14 In the past century, many investigators have tients were aware of the onset of the normal finger, the distal phalangeal described possible reference stan- changes in their nails, and only 2 re- depth is smaller than the interphalan- dards for diagnosis of clubbing, includ- ported painful fingers or joints.20 geal depth. As connective tissue depo- ing water displacement of the termi- sition expands the pulp in the terminal nal phalanges, measurement of nail Signs phalanx, this ratio becomes reversed. curvature using a device called an un- Identification of advanced clubbing, The phalangeal depth ratio appears to guisometer, and measuring nail angles which is characterized by so-called drum- be independent of age, sex, and ethnic- and ratios using plaster casts or shadow stick fingers poses little difficulty for cli- ity in randomly selected popula- projections of fingers.10-15 None has nicians. By contrast, the subtleties of the tions.14,25 A similar ratio using distal and been accepted as a criterion standard earlier stages of clubbing may lead to ani- interphalangeal width can be deter- of diagnosis, and all are cumbersome mated bedside debate among medical mined, but it has not been studied as ex- and impractical as a method of verify- students, residents, and experienced phy- tensively as the phalangeal depth ratio. ing the clinical impression of club- sicians. The 2 approaches for identify- Although the phalangeal depth ratio bing. Therefore, physicians must rely ing clubbing on physical examination are was originally described using plaster solely on their skills in clinical exami- visual inspection and palpation of the cu- casts and shadowgrams, subsequent nation to detect clubbing. ticle for increased sponginess.16,21-22 studies have reported the use of cali-

Downloaded From: http://jama.jamanetwork.com/ by a University of Washington Libraries User on 12/18/2012 CLUBBING

pers on live fingers. To perform this mea- fessional competence; sensitivity and To expand the search, the titles and surement, the calipers should touch but specificity or sensitivity and specificity; abstracts of all articles retrieved using not compress the tissue at the distal pha- reproducibility of result; observer varia- the MeSH heading hypertrophic osteo- lanx and the interphalangeal joint of the tion; diagnostic tests, routine; decision arthropathy or the textwords clubbing index finger during measurement. support techniques; and Bayes theorem. and Hippocratic fingers were evaluated Baughman et al26 estimated that this tech- This strategy resulted in a limited num- by each author independently. Based on nique takes no longer than 1 minute to ber of articles. this review, relevant publications were perform. Visual estimation for the reversal of the phalangeal depth ratio has been Figure. Appearance on Inspection for Clubbing suggested as a simple bedside technique for clubbing, but the precision of A Appearance this method has not been tested. Normal Clubbed Schamroth Sign. In 1976, Scham- roth27 reported a new clinical sign that incorporated 2 of the clinical features of clubbing (Figure, D). Normal fingers create a diamond-shaped window when the dorsal surfaces of terminal phalanges of similar fingers are opposed. In the clubbed finger, the diamond becomes obliterated because of the loss of the profile angle and the increase in the soft tissue at the cuticle. Since its original description, this tech- B Nail-fold Angles nique has become popular with physi- Normal Clubbed C cians as a quick test to establish the B A B A presence of clubbing. The precision and D C accuracy of this sign, however, have not D been formally tested.28 Palpation. On palpation of the base of the nail bed, the examiner perceives C Phalangeal Depth Ratio that the nail is “floating” within the soft tissue, and in advanced cases may even Normal Clubbed be able to feel the proximal edge of the nail. This sign is best elicited by gently IPD DPD IPD rocking the nail. The examiner grips the DPD sides of the subject’s finger between the thumb and middle finger of each hand. D Schamroth Sign Exerting downward pressure with his/ her own index fingers, the examiner then Normal Clubbed rocks the distal and proximal ends of the subject’s nail, using the nail bed as a fulcrum. METHODS We used the MEDLINE database to search for English-language articles related to the clinical evaluation of clubbing that were published between Janu- A, Normal finger viewed from above and in profile, and the changes occurring in established clubbing, viewed ary 1966 and April 1999. The MeSH from above and in profile. B, The finger on the left demonstrates normal profile (ABC) and normal hyponychial heading hypertrophic osteoarthropa- (ABD) nail-fold angles of 169° and 183°, respectively. The clubbed finger on the right shows increased profile and hyponychial nail-fold angles of 191° and 203°, respectively. C, Distal phalangeal finger depth (DPD)/ thy, followed by the textword club- interphalangeal finger depth (IPD) represents the phalangeal depth ratio. In normal fingers, the IPD is greater bing, were used in the following search than the DPD. In clubbing, this relationship is reversed. D, Schamroth sign: in the absence of clubbing, oppo- strategy: physical examination/ or physi- sition of the index fingers nail-to-nail creates a diamond-shaped window (arrowhead). In clubbed fingers, the loss of the profile angle due to the increase in tissue at the nail bed causes obliteration of this space (arrowhead). cal exam$; medical history taking; pro-

Downloaded From: http://jama.jamanetwork.com/ by a University of Washington Libraries User on 12/18/2012 CLUBBING

retrieved and their bibliographies were RESULTS 192°. The phalangeal depth ratio has evaluated for additional material. We Quality of the Evidence been reported in 359 disease-free sub- also examined standard textbooks of By consensus and using criteria previ- jects, and in only 1 did it exceed unity. physical diagnosis for information on ously developed for this series, we TABLE 3 shows the nail-fold angles the physical examination for club- appraised the quality of the evidence con- and phalangeal depth ratios in pa- bing. We attempted to contact the au- tained in the articles that we retrieved.29 tients with diseases associated with thors of articles in which more than 1 For reasons of selection bias, small sample clubbing. In such chronic diseases as observer made a determination of club- size, and lack of an independent, blind cystic fibrosis and cyanotic congenital bing to obtain additional data about pre- comparison of the physical sign with a heart disease, the nail-bed angles and cision of the examination for club- criterion standard, we classified all of the the phalangeal depth ratios are signifi- bing. Studies selected for data extraction included studies as level 4, leading to cantly higher than those found in dis- were those in which quantitative or grade C recommendations.29 ease-free populations. In case series of qualitative assessment for clubbing was asthma and COPD, phalangeal depth described in a series of patients. Al- Quantitative Indices of Clubbing ratios are slightly higher than normal though our expanded electronic search in Normal and Disease States values. However, it is impossible to ex- identified 567 articles related to club- Using plaster casts, shadowgraphs, and clude the possibilities that these series bing, only 16 studies met the criteria calipers, nail-fold angles and the pha- may have included patients with other for inclusion in our analysis. langeal depth ratio have been mea- pulmonary disorders associated with sured in normal populations and in sub- clubbing or that some patients were se- Study Characteristics jects with diseases associated with lected because they had clubbing. Clubbing differs from other physical clubbing. The precision of these quan- In summary, in disease-free subjects, signs evaluated in the Rational Clini- titative techniques is high. Using the a phalangeal depth ratio above 1 is rare, cal Examination series in that the lack shadowgraph method, Kitis et al30 ex- the profile angle does not exceed 176°, of an accepted objective diagnostic cri- amined the precision of measuring nail- and the hyponychial angle does not ex- terion standard precludes meaningful fold angles. Duplicate measurements of ceed 192°. To facilitate clinical use, we assessment of the accuracy of clinical 51 subjects showed a difference of 0.2° suggest accepting values of less than 180° examination. However, our review of in the mean of both the hyponychial for the profile angle (a straight line) and the literature on clubbing permitted us and profile angles, with SDs of 4.6° and less than 190° for the hyponychial angle to evaluate quantitative indices used to 4.3°, respectively. Although Waring et as describing normality. distinguish clubbed from normal fin- al15 found that the measurement of the gers; precision of physicians’ bedside phalangeal depth ratio with calipers on PRECISION AND ACCURACY clinical examination for clubbing; and live fingers rather than plaster casts re- Precision of the Clinical accuracy of clubbing as a marker of se- sulted in a loss of precision, Baugh- Examination for Clubbing lected diseases. We chose to limit our man et al26 investigated intrarater reli- Four studies35-38 have reported the pre- review of the quantitative indices of ability and found an SD of only 0.0008. cision of physicians’ bedside examina- clubbing to studies of nail-fold angles In the same study, 2 observers inde- tion for clubbing (TABLE 4). Although and the phalangeal depth ratio, be- pendently measured the ratio in 20 sub- several of the case series describing the cause of their potential applicability at jects, and the maximal difference in prevalence of clubbing in various dis- the bedside. phalangeal depth ratio was 0.03. ease states used multiple examiners, Published data pertaining to the mea- none reported interrater reliability. We Data Analysis surement of nail-fold angles and phalan- have excluded from this section re- Pooled weighted averages were calcu- geal depth ratios in disease-free indi- ports of precision that used only casts lated for quantitative measurements of viduals are summarized in TABLE 2. The or shadowgraphs for determination of nail-fold angles and phalangeal depth pooled weighted mean values for the precision, since potentially important ratios from data in studies of normal profile and hyponychial angle are 167.2° clinical information from inspection or and diseased populations. Using data and 179.0°, respectively. The pooled palpation of the live finger was not avail- available in 2 articles on the precision weighted mean phalangeal depth ratio is able to the examiners. of clubbing, we calculated ␬ statistics 0.900. Do these measurements help dis- In an attempt to challenge the pre- using the Stata statistical package (ver- tinguish those with from those without vailing wisdom that clubbing was sion 3.0, Computing Resource Cen- clubbing? The range was available for easily recognized, Pyke35 studied the ter, Santa Monica, Calif). Sensitivities, only 45 of the 161 disease-free subjects precision of physicians’ global assess- specificities, and likelihood ratios of in whom the profile angle was mea- ment for the sign. He enlisted 12 phy- clubbing as a marker of specific under- sured, and none exceeded 176°. In stud- sicians and 4 medical students to ex- lying disease were calculated from origi- ies of hyponychial angles, none of the 171 amine 12 patients for the presence of nal data when possible. disease-free subjects had angles above clubbing. He purposefully chose pa-

Downloaded From: http://jama.jamanetwork.com/ by a University of Washington Libraries User on 12/18/2012 CLUBBING

tients who exhibited the full range of tients. Use of cases of more advanced bing. However, the index was evalu- findings from normal to advanced club- clubbing may have led to an overesti- ated in patients with cyanotic congeni- bing. Overall agreement was fair mation of precision. tal heart disease, whose clubbing was so (␬=0.39). From the reported data, it marked that it was “obvious by simple was impossible to determine the effect Accuracy of Clubbing inspection.”40 Only 1 study36 measured of training on the examiners’ preci- as a Marker of Disease States the accuracy of clinicians’ bedside ex- sion, but it was clear that the examin- Determination of the accuracy of clini- amination for clubbing against a priori ers used different criteria to identify cal examination techniques to detect diagnostic criteria derived from quanti- clubbing. After completing their assess- clubbing has been confounded by incor- tative indices in disease-free popula- ments, Pyke asked the examiners to de- poration bias that results when the clini- tions and those with disease. Unfortu- fine clubbing, and he received a wide cal examination itself forms part or all nately, data were not given in sufficient variety of answers. of the diagnostic criterion standard. One detail to allow calculation of the sensi- Rice and Rowlands36 used several example of such confounding is illus- tivity and specificity of the clinical ex- quantitative indices, including phalan- trated by the digital index of Vasquez et amination. Hence, data on the accuracy geal depth ratios, to assemble 11 pa- al.40 This index, the sum of the ratios of of clinical examination compared with tients who exhibited a range of find- the distal phalangeal finger depth and in- the quantitative indices to detect club- ings from normal to advanced clubbing. terphalangeal depth circumferences in all bing are limited. Nineteen clinicians, all internal medi- 10 fingers, has been reported to have a An alternative approach is to con- cine staff or resident physicians, exam- high sensitivity and specificity for club- sider the accuracy of the presence of club- ined the patients for clubbing. Club- bing was judged to be present in 103 of the 209 subject examinations. As Table 2. Reported Values for Profile Angle, Hyponychial Angle, and Phalangeal Depth Ratio with Pyke’s findings, observer agree- in Disease-Free Subjects No. of ment was only fair (␬=0.36). Technique Population Subjects Mean (SD) Precision of physical examination for Profile angle a variety of signs of pulmonary disease, Bentley et al,13 Shadowgraph Healthy subjects from a 25 168.3° (3.7°) 1976 surgical clinic (age including clubbing, was evaluated in a not specified) study in which 24 experienced physi- Kitis et al,30 1979 Shadowgraph Healthy hospital 116 166.3° (4.3°) cians examined 4 patients each.37 The employees precision of the examination for club- Sinniah and Omar,31 Shadowgraph Healthy children (source 20 171.4° (5.5°) bing was moderate (␬=0.45). While sev- 1979 population not specified) eral signs showed marginally greater pre- Pooled weighted 161 167.2° (4.4°) cision (eg, wheezes, ␬=0.51), most signs mean had significantly lower precision (eg, dis- Hyponychial angle placed trachea, ␬=0.01; whispering pec- Regan et al,12 Plaster casts, Healthy manual workers 10 186.1° (1.97°) 1967 planimeter toriloquy, ␬=0.11). 13 38 Bentley et al, Shadowgraph Healthy manual workers 25 180.1° (4.2°) A 1965 study contrasted other 197613 reports of the precision of the physical Kitis et al,30 1979 Shadowgraph Healthy manual workers 116 177.9° (4.6°) examination for clubbing. Of 21 pulmo- Sinniah and Omar,31 Shadowgraph Healthy manual workers 20 180.7° (5.2°) nary signs, clubbing exhibited the high- 1979 est rate of interobserver agreement Pooled weighted 171 179.0° (4.5°) mean among 9 experienced physicians exam- 39 Phalangeal depth ratio ining 20 patients (␬=0.90). This high Waring et al,15 Plaster casts, Children and adults 160 0.895 (0.041) level of precision may reflect either the 1971 micrometer (source population not specified) experience of the examiners or a selec- Sly et al,25 1973 Plaster casts, Adults (medical center 60 0.903 (0.043) tion bias, since the degree of clubbing in micrometer personnel and affected patients was not described. relatives of patients attending pediatric In summary, the precision of the allergy clinic) clinical examination for clubbing has Paton et al,32 1991 Plaster casts, Children and adults 85 0.890 (0.040) been found to be fair to moderate, with micrometer (random sample from people playing 1 study showing very high precision. in nearby park) Although precision was higher in the Baughman et al,26 Live fingers, Adults (medical center 54 0.920 (0.050) 2 studies that used more experienced 1998 calipers personnel) examiners, neither of these studies re- Pooled weighted 359 0.900 (0.042) ported their selection criteria for pa- mean

Downloaded From: http://jama.jamanetwork.com/ by a University of Washington Libraries User on 12/18/2012 CLUBBING

Table 3. Reported Values for Quantitative Measures of Clubbing in Disease States* No. of Subjects Technique Quantitative Measure Mean (SD) Asthma Waring et al,15 1971 45 Plaster casts, micrometer DPD/IPD ratio 39/45 Ͻ1.0† Sly et al,25 1973 119 Plaster casts, micrometer DPD/IPD ratio 0.910 (0.050) Bentley et al,13 1976 25 Shadowgraph Profile angle; hyponychial 170.9° (4.1°); 185.4° (6.4°) angle Paton et al,32 1991 20 Plaster casts, micrometer DPD/IPD ratio 0.911 (0.046) Chronic obstructive pulmonary disease Baughman et al,26 1998 54 Live fingers, calipers DPD/IPD ratio† 0.94 (0.06) Bronchogenic carcinoma Baughman et al,26 1998 109 Live fingers, calipers DPD/IPD ratio‡ 0.975 (0.099) Cystic fibrosis Waring et al,15 1971 45 Plaster casts, micrometer DPD/IPD ratio 38/45 Ͼ1.0† Bentley et al,13 1976 50 Shadowgraph Profile angle; hyponychial 179.0° (6.2°); 194.8° (8.3°) angle Lemen et al,33 1978 18 Plaster casts DPD/IPD ratio§ 1.010 (0.016) Pitts-Tucker et al,34 1986 73 Shadowgraph Hyponychial angle 192° Paton et al,32 1991 44 Plaster casts, micrometer DPD/IPD ratio 1.033 (0.079) Cyanotic congenital heart disease Waring et al,15 1971 27 Plaster casts, micrometer DPD/IPD ratio 18/27 Ͼ1.0† Bentley et al,13 1976 25 Shadowgraph Profile angle; hyponychial 179.7° (4.8°); 195.5° (2.5°) angle Asbestos exposure Regan et al,12 1967 50 Plaster casts, planimeter Hyponychial angle 195.0° (9.6°) Crohn disease Kitis et al,30 1979 200 Shadowgraph Hyponychial angle 183.5° (7.8°) *DPD/IPD indicates distal phalangeal depth/interphalangeal depth. †Individual values not reported; proportion of patients with DPD:IPD of greater than 1.0 reported. ‡Value reported in table is for right index finger only. §Pooled weighted average for right index finger only.

raphy at study entry, which was sub- Table 4. Interobserver Agreement of Clinical Examination for Clubbing sequently diagnosed as adenocarcinoma Source, y No. of Observers Observer’s Level of Experience ␬ of the lung. Pyke,35 1954 16 4 Medical students 0.39 4 Medical registrars These data confirm, as expected, that 4 Surgical registrars while a normal phalangeal depth ratio 4 Senior physicians does not rule out lung cancer, an abnor- Rice and Rowlands,36 1961 19 Residents 0.36 Fellows mal ratio implies an increased probabil- Staff physicians ity (likelihood ratio, 3.9; 95% CI, 1.6- Smyllie et al,38 1965 9 5 Medical registrars 0.90 9.4) of underlying lung cancer. Only 3 4 Consultant physicians of the patients with COPD had a pha- Spiteri et al,37 1988 24 2 Senior house officers 0.45 14 Medical registrars langeal depth ratio greater than 1.05, and 8 Consultant physicians none had a ratio greater than 1.1. Among those with lung cancer, there was no sig- bing as a marker of underlying disease. tients with COPD, and 54 control sub- nificant difference in the prevalence of Because many patients with clubbing jects. Of the 54 control subjects, none clubbing (as defined by distal phalan- have pulmonary disease, a relevant clini- had a phalangeal depth ratio in excess geal finger depth/interphalangeal fin- cal question is whether clubbing sepa- of 1. In those patients who had a pha- ger depth ratio Ͼ1) among the differ- rates those with COPD from those who langeal depth ratio greater than 1, 40 ent histologic subtypes of lung cancer. have clubbing associated with pulmo- had lung cancer and 5 had COPD alone Kitis et al30 investigated the associa- nary malignancy. In this way, 1 study26 (likelihood ratio, 3.9 [95% confidence tion of clubbing with the activity of in- assessed the usefulness of the phalan- interval {CI}, 1.6-9.4]). Seventy pa- flammatory bowel disease in 327 pa- geal depth ratio in distinguishing patients who had a phalangeal depth ra- tients. Clubbing was defined as a tients with documented lung cancer tio of 1 or less had lung cancer, and 49 shadowgraph-measured hyponychial from control subjects and those with with the same depth ratio had COPD angle greater than 186°, which corre- COPD. Using calipers, Baughman et al26 alone (likelihood ratio, 0.7 [95% CI, sponded to 1.65 SDs above the mean measured the phalangeal depth ratio in 0.6-0.8]). We reclassified 1 subject in value found in a group of 116 healthy both right and left index fingers in 109 the COPD group who had a pulmo- controls. Disease activity was deter- patients with known lung cancer, 55 pa- nary nodule detected on chest radiog- mined using an index incorporating the

Downloaded From: http://jama.jamanetwork.com/ by a University of Washington Libraries User on 12/18/2012 CLUBBING

results of various laboratory investiga- ratios than do disease-free subjects, it 11. Stavem P. Instrument for estimation of clubbing. Lancet. 1959;2:7-8. tions. The likelihood ratios for club- is unusual for the ratio to exceed 1.05. 12. Regan GM, Tagg B, Thompson ML. Subjective as- bing as a marker of active Crohn dis- A value in excess of this in a patient with sessment and objective measurement of finger clubbing. Lancet. 1967;1:530-532. ease were 2.8 (95% CI, 1.8-4.1) and 3.7 COPD should prompt a search for bron- 13. Bentley D, Moore A, Swachman H. Finger club- (95% CI, 1.4-9.4) for ulcerative coli- chogenic carcinoma. Because most cli- bing: a quantitative survey by analysis of the shad- tis. The sensitivity and specificity val- nicians do not have calipers, visual es- owgraph. Lancet. 1976;2:164-167. 14. Mellins RB, Fishman AP. Digital casts for the study ues were 0.58 and 0.79 for Crohn dis- timation of reversal of the phalangeal of clubbing of the fingers. Circulation. 1966;33:143-145. ease vs 0.30 and 0.92 for ulcerative depth ratio should be assessed. 15. Waring WW, Wilkinson RW, Wiebe RA, Faol BC, Hilman BC. Quantitation of digital clubbing in chil- colitis, respectively. • Although the accuracy of clinicians’ dren. Am Rev Respir Dis. 1971;104:166-174. bedside estimation of nail-fold angles 16. Bigler FC. The morphology of clubbing. Am J SCENARIO RESOLUTION has not been studied, the normal val- Pathol. 1958;34:237-261. 17. Dickinson CJ. The aetiology of clubbing and hy- In the first case, you find that the patient ues for these angles have been estab- pertrophic osteoarthropathy. Eur J Clin Invest. 1993; appears to have increased nail curva- lished. A profile angle that approaches 23:330-338. 18. Racoceanu SN, Mendlowitz M, Suck AF. Digital ture. You use calipers to estimate a pha- a straight line (180°) is rare in disease- capillary blood flow in clubbing. Ann Intern Med. 1971; langeal depth ratio of 0.90, and on inspec- free subjects, and in our opinion is eas- 75:933-935. 19. Martinez-Lavin M. Digital clubbing and hypertro- tion you estimate a profile angle of about ily identifiable at the bedside. Al- phic osteoarthropathy. J Rheumatol. 1987;14:6-8. 160°. Based on your knowledge of these though the normal range of the 20. Onadeko BO, Kolawolw TM. The clinical and ae- values in disease-free subjects, you inform hyponychial angle has also been de- tiological pattern of finger clubbing and hypertrophic osteoarthropathy in Nigerians. Trop Geogr Med. the respiratory therapist that her mother fined, this angle is more difficult to es- 1979;31:191-199. does not have clubbing. On the other timate at the bedside. 21. Hansen-Flaschen J, Nordberg J. Clubbing and hypertrophic osteoarthropathy. Clin Chest Med. 1987; hand, you find that the second patient • No published evidence exists as to the 8:287-298. has a phalangeal depth ratio of 1.1 and a diagnostic yield or the optimal strat- 22. Rabin CB. New or neglected physical signs in diagnosis of chest diseases. JAMA. 1965;194:546-550. profile angle of 180°, findings that are egy for investigating a patient with club- 23. Lovibond JL. Diagnosis of clubbed fingers. Lan- quite unusual for disease-free subjects or bing. Therefore, after completion of a cet. 1938;1:363-364. patients with COPD alone. You con- thorough medical history and physi- 24. Carroll DG. Curvature of the nails, clubbing of the fingers and hypertrophic osteoarthropathy. Trans Am clude that a search for bronchogenic car- cal examination, clinical judgment must Clin Climatol Assoc. 1972;83:198-208. cinoma (or other causes of clubbing) guide the choice of investigations. 25. Sly RM, Ghazanshahi S, Buranakul B, et al. Objec- should be undertaken. tive assessment for digital clubbing in Caucasian, Ne- Author Contributions: Study concept and design, ac- gro and Oriental subjects. Chest. 1973;64:687-689. quisition of data, analysis and interpretation of data, 26. Baughman RP, Gunther KL, Buchsbaum JA, Lower THE BOTTOM LINE drafting of the manuscript, critical revision of the EE. Prevalence of digital clubbing in bronchogenic car- manuscript for important intellectual content, statis- cinoma by a new digital index. Clin Exp Rheumatol. For generations, medical students and tical expertise: Myers, Farquhar. 1998;16:21-26. residents have been quizzed at the bed- Acknowledgment: We appreciate the expert advice 27. Schamroth L. Personal experience. S Afr Med J. offered by Joseph Govert, MD, and John Whited, MD, 1976;50:297-300. side about the diagnostic features of both of Duke University, during the preparation of this 28. Lampe RM, Kagan A. Detection of clubbing— clubbing. Confident though their in- article. Schamroth’s sign. Clin Pediatr (Phila). 1983;22:125. 29. Holleman DR, Simel DL. Does the clinical examina- quisitors may be in their own ability to tion predict airflow limitation?JAMA. 1995;273:313-319. detect clubbing, the literature shows REFERENCES 30. Kitis G, Thompson H, Allan RN. Finger clubbing in that interobserver agreement is only fair 1. West S. Two cases of clubbing of the fingers de- inflammatory bowel disease. BMJ. 1979;2:825-828. veloping within a fortnight and four weeks respec- 31. Sinniah D, Omar A. Quantitation of digital clubbing to moderate, and that the accuracy of tively. Trans Clin Soc London. 1897;30:60. by shadowgram technique. AJDC. 1979;54:145-146. techniques to detect clubbing has not 2. Adams F, trans. The Genuine Works of Hip- 32. Paton JY, Bautista DB, Stabile MW, et al. Digital club- been well established. Nevertheless, pocrates, 1:206. New York, NY: Wm Wood & Co; 1891. bing and pulmonary function abnormalities in children 3. Horsfall FL. Congenital familial clubbing of the fin- with lung disease. Pediatr Pulmonol. 1991;10:25-29. since nonhereditary clubbing is al- gers and toes. CMAJ. 1936;34:145-149. 33. Lemen RJ, Gates AJ, Mathe AA, et al. Relationship most always a portent of serious dis- 4. Buchman D, Hrowat EA. Idiopathic clubbing and among digital clubbing, disease severity and serum pros- hypertrophic osteoarthropathy. Arch Intern Med. 1955; taglandin F2 and E concentrations in cystic fibrosis pa- ease, clinicians need to be as certain as 97:355-358. tients. Am Rev Respir Dis. 1978;117:639-646. possible about its presence. 5. Mendlowitz M. Clubbing and hypertrophic osteo- 34. Pitts-Tucker TJ, Miller MG, Littlewood JM. Finger arthropathy. Medicine (Baltimore). 1942;21:269- clubbing in cystic fibrosis. AJDC. 1986;61:576-579. Recognizing the limitations of the 306. 35. Pyke DA. Finger clubbing. Lancet. 1954;2:352-354. studies we have appraised, we recom- 6. Alvarez AS, McNair D, Wildman J, Hewson JW. Uni- 36. Rice RE, Rowlands PW. A Quantitative Method lateral clubbing of the fingernails in patients with for the Estimation of Clubbing [thesis]. New Or- mend the following: hemiplegia. Gerontologia Clinica. 1975;17:1-6. leans, La: Tulane University Medical School; 1961. • In cases of diagnostic uncertainty, the 7. Marie P. De l’oste´ o-arthropathie hypertrophiante 37. Spiteri MA, Cook DG, Clarke SW. Reliability of phalangeal depth ratio may be help- pneumique. Rev Med. 1890;10:1. eliciting physical signs in examination of the chest. Lan- 8. Bamberger E. Über knochevnera¨ nderungen bei cet. 1988;1:873-875. ful. This ratio can be measured using chronischen lungen und herzkrankheiten. Atschr Klin 38. Smyllie HC, Blendis LM, Armitage P. Observer dis- calipers at the bedside, and in disease- Med. 1891;18:193. agreement in physical signs of the respiratory sys- 9. Coury C. Hippocratic fingers and hypertrophic os- tem. Lancet. 1965;2:412-413. free populations rarely exceeds 1.0. An teoarthropathy: a study of 350 cases. Br J Chest. 1960; 39. Armitage P, Blendis LM, Smyllie HC. The mea- elevated ratio should prompt a search 54:202-209. surement of observer disagreement in the recording 10. Cudowitz L, Wraith DG. An evaluation of the clini- of signs. J R Stat Soc A. 1966;129:98-109. for underlying disease. Although pa- cal significance of clubbing in common lung disor- 40. Vazquez-Abad D, Pineda C, Martinez-Lavin M. tients with COPD have slightly higher ders. Br J Tuberc Dis Chest. 1957;51:14-21. Digital clubbing. J Rheumatol. 1989;16:518-520.

Downloaded From: http://jama.jamanetwork.com/ by a University of Washington Libraries User on 12/18/2012