Digital clubbing in tuberculosis – relationship to HIV infection, extent of disease and hypoalbuminemia

All adult patients with sputum acid-fast bacilli (AFB) smear positive pulmonary TB presenting to Mulago Hospital between December 2002 and March 2003 were eligible for study participation. Patients were ineligible if they had extra-pulmonary TB with negative sputum smears, sputum smear negative pulmonary disease or diseases including empyema, cyanotic congenital heart diseases, bronchogenic carcinoma, and ulcerative colitis, all known to be associated with clubbing. The study protocol was reviewed and approved by the Research and Ethics Committee of Makerere University Medical School. All subjects gave informed consent for study participation and HIV testing and received pre- and post-HIV-test counseling. HIV seropositive individuals were further categorized using the World Health Organization (WHO) Clinical Staging System for HIV Infection and Disease[5].

We examined 115 general medical patients (irrespective of diagnosis), at the Mulago emergency medical ward, and 143 healthy volunteers for presence of finger clubbing, in order to estimate the prevalence ofclubbing in these populations. No further follow-up studies were carried out on these individuals.

All patients with TB underwent medical history and physical examination including determination of their Karnofsky performance scale score[6]. The Karnofsky scale is a well-validated assessment of an individual's health and well-being based on a performance index of physical ability. It is often used in clinical research including HIV research to monitor and record the health of patients. The Karnofsky score uses a scale from 0 to 100, where 100 is when the patient has no complaints or evidence of disease and 50 is when the patient requires considerable assistance and frequent medical care.

Clubbing was assessed by clinical signs and by determination of the phalangeal depth ratio. The fingers were examined for presence of increased nail curvature, nail bed fluctuation, Schamroth's sign[7] (where opposing of fingers of both hands leads to disappearance of the normal 'diamond' window that forms between the nail beds) and definite clubbing and the findings recorded. For purposes of this study, definite clubbing was defined as finger clubbing that was obvious to the examiner's eyes without need for any measuring tools. Then, using vernier calipers (Stanley Tools, New Britain, CT), the distal phalangeal (DPD) and interphalangeal (IPD) diameters of both the right and left index fingers were measured and the phalangeal depth (DPD/IPD) ratio was calculated [8]. The greater of the DPD/IPD ratios for the index fingers was used in determining the presence or absence of clubbing. Clubbing was considered present if the DPD/IPD ratio was greater than 1.0 [8]. The DPD/IPD is a well-validated objective measure of the presence of clubbing with acceptable reproducibility. The mean normal DPD/IPD ratio is approximately 0.895 and is independent of age, race and sex[9, 10]. A cut-off point of greater than 1.0 exceeds the normal range by about 2.5 standard deviations. In this study, the DPD/IPD ratio was regarded as the best (gold standard) test for clubbing for use in comparison with other findings.

Sputum AFB smears were performed by the Ziehl-Neelsen method and graded according to IUATLD guidelines[11]. A standard postero-anterior chest radiograph was obtained for each patient. The radiographic severity of disease was graded as minimal, moderately advanced, or far advanced disease using the United States National TB and Respiratory Disease scheme[12]. The number of lung zones (out of 6) and the presence of specific findings also were recorded. A cavity was defined as a gas-containing lucent space at least 1 cm in diameter within the lung parenchyma surrounded by an infiltrate or fibrotic wall greater than 1 mm thick. HIV serostatus was determined by HIV1/2 enzyme immunoassay using Determine™ HIV 1/2 kit (Abbott Laboratories, Tokyo, Japan). Serum albumin was measured by using the ALBUMIN Bromocresol Green™ kit (Randox Laboratories Ltd, Diamond Road, Crumlin, UK); the normal range being 3.8 to 4.4 g/dL.

History and physical examination and assessment of clubbing were performed prior to HIV testing. Laboratory technologists performing sputum AFB smears, HIV and albumin determinations and radiologists reviewing the chest radiographs were masked to all clinical data and the presence of clubbing.

Estimation of the sample size was based on the estimated prevalence of finger clubbing of 25% and a precision of 5%. Data were analyzed using SPSS version 10.0 (SPSS, Inc., Chicago, IL). Tests of significance using p-values, Odds ratios and Likelihood ratios, were made using 2 by 2 tables. Confidence intervals and likelihood ratios were calculated using Confidence Interval Analysis (CIA) for windows version 2.1.1. Because of our small sample size and the small number of cases in some groups, we did not perform a multivariable analysis.

For statistical purposes patients were put into two HIV clinical categories. Those in WHO clinical stages 1 and 2 were categorized as early HIV disease whereas those in stages 3 and 4 were classified as late disease. Karnofsky performance scale scores were analyzed by dichotomizing at 70; those with performance scores equal to or greater than 70, where 70 is defined as being able to care for self, but unable to carry on normal activity or to do active work, and those with scores less than 70. A level of significance of p < 0.05 was used to reject the null hypothesis in all analyses.

During the study period 200 patients with smear positive pulmonary TB were approached. All consented to participate in the study. Fifty-nine percent were males. The mean age ± SD was 32.6 ± 11.3 years. The most frequent symptoms reported by the patients were cough for more than three weeks (99%), sputum production (95%), fever (95%), night sweats (90%) and weight loss (86%). Twenty-four percent reported hemoptysis. Twelve percent had received prior TB treatment within the previous year and 82% were HIV-infected. Of the patients with HIV infection, 58% had early HIV infection and 42% had late stage disease.

We also evaluated 258 adult subjects including 115 newly admitted general medical patients (57% males; mean age 36 years) and 143 healthy medical students and hospital staff volunteers (48% males; mean age 22 years) for the presence of clubbing.

Using a DPD/IPD ratio of greater than 1.0 as the gold standard, we found that 60 (30%) of 200 patients with pulmonary TB had digital clubbing. Twenty-one (18%) of 115 general medical patients and 3 of 143 (2%) healthy volunteers had clubbing.

The clinical signs of finger clubbing identified were nail bed fluctuation, increased nail curvature, a positive Schamroth's sign and definite clubbing and were distributed as 29.0%, 26.5%, 20.5 and 18.5% respectively, among the study participants. Twenty-three patients (11.5%) had only one of these clinical signs, 18 (9.0%) two clinical signs, 14 (7.0%) while 22 (11.0%) had four clinical signs. One hundred twenty three patients (61.5%) did not have any sign of clubbing.

We compared the proportion of subjects determined to have clubbing diagnosed by measurement of the DPD/IPD ratio with other clinical methods and signs (Table 1). The two methods differed significantly as shown in table 1. Presence of nail bed fluctuation, increased nail curvature, a positive Schamroth's sign and obvious clubbing were significantly associated with finger clubbing as per the gold standard. Having all four clinical signs was highly associated with clubbing (p < 0.001; likelihood ratio 49.0; 95% CI 6.7 to 356), whereas having none of the clinical signs was significantly associated with absence of clubbing (p < 0.001; likelihood ratio 0.3; 95% CI: 0.2 to 0.5). Presence of only one clinical sign and two clinical signs had a very low sensitivity for clubbing (Table 1). Compared with the other clinical signs, nail-bed fluctuation had a higher sensitivity of 72% (95% CI: 59.2% to 81.5%) and a good specificity of 89.3% (95% CI: 83.1% to 93.4%). The proportion of patients with pulmonary TB who had clubbing did not differ by HIV serostatus. Fifty of 163 (31%) HIV-seropositive patients had clubbing as compared to 10 out of 37 (27%) of those who were HIV-seronegative (Odds ratio 1.2; 95% CI: 0.5 to 2.9; p = 0.66). Presence of clubbing was not associated with age, sex, or severe symptoms of TB such as hemoptysis (data not shown). Twenty-one percent of patients with performance status of 70% and above (23 of 108), had clubbing compared with 40% (37 of 92), of those who had lower performance capacity. Patients with a performance status score below 70% were more likely to have clubbing (OR 2.49, 95% CI: 1.3 to 2.9; p = 0.003) than those with a higher (able to care for self) performance. Clubbing was more frequent in patients who had received prior anti-TB treatment compared to those who had not been treated previously for TB (OR 2.7; 95% CI: 1.0 to 6.9; p = 0.023).

Next, we examined whether the radiographic extent of disease and presence of cavitary disease on chest radiograph were associated with the presence of clubbing. The proportion of patients with clubbing did not differ between patients with more lung zones involved by disease or bilateral disease, those with minimal as compared to radiographically moderate advanced or far advanced disease, and those with any visible cavitary disease (Table 2).

We used serum albumin as a measure of nutritional status to assess the relationship of clubbing and nutrition. Mean serum albumin was 4.0 ± 0.7 g/dL. Seventy-six percent of the patients with TB had a normal serum albumin. Eight percent had levels less than 2.5 g/dL. Forty-eight patients (24%) had serum albumin level below the normal range and, of these, 40% had finger clubbing compared to 27% of those with normal serum albumin levels. Subnormal albumin (less than 2.5 g/dL) was not associated with presence of clubbing (OR 1.8, 95% CI: 0.9 to 3.7; p = 0.097).