(Chest. 2002;122:1299-1301.)
© 2002
American College of Chest Physicians
Tuberculin Test Measurement*
Variability Due to the Time of Reading
Dave Singh, MD;
Chris Sutton, PhD and
Ashley Woodcock, MD
* From the North West Lung Research Center (Drs. Singh and Woodcock), South Manchester University Hospital Trust, Wythenshawe, Manchester, UK; and the Statistics Group (Dr. Sutton), Faculty of Science, University of Central Lancashire, Preston, UK.
Correspondence to: Dave Singh, MD, Lung Function Unit, North West Lung Research Center, South Manchester University Hospital Trust, Southmoor Rd, Wythenshawe, Manchester M23 9LT, United Kingdom; e-mail: dave_sd{at}hotmail.com
 |
Abstract
|
|---|
Study objectives: It is recommended that the PPD Mantoux tuberculin test be read at 48 or 72 h. We have compared the measurements at these time points.
Design: A 10-tuberculin unit (TU) PPD Mantoux test was administered to 116 healthy subjects (76.7% with bacillus Calmette-Guérin scars).
Participants: One hundred sixteen healthy adult volunteers were recruited (health service employees, 29 volunteers; general public, 87 volunteers).
Results: The measurements made at 72 h were significantly higher than those made at 48 h (median, 9.4 vs 4.95 mm, respectively; p = 0.017). In those subjects with induration at either or both time points (n = 69), the readings taken at 72 h were on average 1.7 mm (95% confidence interval, 0.4 to 3.0 mm) larger than those at 48 h. Using an induration of > 15 mm diameter to define a positive result, there were more positive test results at 72 h (36) compared to 48 h (28), with the results from 10 subjects (8.5%) changing because of the time of measurement.
Conclusions: This study demonstrates that, in adults, the size of the 10-TU Mantoux reaction is significantly larger at 72 h compared to the reaction at 48 h. In clinical practice, tuberculin tests should be read at 72 h as negative tests at 48 h may be false results.
Key Words: measurement • tuberculin test • variability
|
Introduction
|
|---|
The tuberculin test is widely used to identify latent tuberculosis infection.1
2
The test also is used to investigate patients who are suspected of having active tuberculosis but who have not had a culture positive for Mycobacterium tuberculosis. It is recommended that tests administered by the Mantoux technique be read at 48 or 72 h after injection.1
However, there is little information comparing readings at 48 and 72 h. We therefore set out to quantify the variability of the tuberculin test caused by the time of reading.
|
Materials and Methods
|
|---|
One hundred sixteen healthy volunteers (health service employees, 29 volunteers; general public, 87 volunteers) between the ages of 19 and 60 years (mean age, 35.9 years) participated in this study, which was conducted over an 18-month period at AstraZeneca Pharmaceuticals and the North West Lung Research Center (South Manchester University Hospitals Trust). The study was approved by the AstraZeneca Local Research Ethics Committee and the South Manchester University Hospitals Trust Research Ethics Committee. Volunteers were recruited by means of local advertising, and written informed consent was obtained from all participants. A medical history was obtained from all subjects to ensure that they did not have a medical condition or were taking any medications. The subjects’ general practitioners were contacted to confirm the medical histories. No subjects had a history of tuberculosis, recent exposure to tuberculosis, engaged in IV drug use, or had HIV. Subjects who developed infections (eg, viral infections) during the study were excluded. Subjects were asked whether they had received a bacillus Calmette-Guérin (BCG) vaccination, and the presence of a BCG scar also was checked. The Mantoux technique3
was used to administer 10 U PPD-Weybridge (Evans Medical; Leatherhead, UK) into the volar aspect of the forearm of all the subjects. This tuberculin is bioequivalent to 10 tuberculin units PPDS.4
Induration was measured in all subjects at both 48 and 72 h (or within 1 h of these times) by a physician with extensive experience in reading tuberculin tests (observer A). The margins of induration were defined using the palpation method,1
and two perpendicular diameters were measured using electronic calipers (model PM200; Sylvac; Crissier, Liechtenstein). We "blinded" the calipers by covering the digital display that showed the measurement value, thus ensuring that the observer could not see his measurements. An assistant recorded the results and calculated the mean of the two perpendicular diameter measurements.
In order to confirm the results of the measurements of observer A, two other observers (B and C) also measured the diameter of induration in 29 randomly selected subjects. All three observers performed four readings on each subject (readings 1 to 4). These readings were performed again using blinded calipers with an assistant recording the results to ensure that the observers remained blinded throughout.
The readings at 48 and 72 h were compared using a Wilcoxon rank sum test. The individual differences between the measurements at these two time points were calculated by the method of Bland and Altman.5
This allowed the limits of agreement between the two time points to be assessed. Additionally, the results were categorized as positive or negative using a cutoff point of 15 mm for induration diameter. In clinical practice, although reactions of this size may be due to previous BCG vaccinations, they are generally regarded as evidence of tuberculosis infection.1
2
These categorized results were analyzed using the Cohen 
coefficient, which was estimated using the variance components method, in which adjustment for additional factors was necessary.6
7
8
A coefficient value of 0 to 0.2 is usually taken to indicate slight agreement, 0.21 to 0.4 to indicate fair agreement, 0.41 to 0.6 to indicate moderate agreement, 0.61 to 0.8 to indicate substantial agreement, and 0.81 to 1 to indicate almost perfect agreement. Ages were compared using an unpaired Student t test.
|
Results
|
|---|
BCG vaccination scars were present in 89 of the 116 subjects (76.7%), while 17 subjects (14.7%) indicated a history of BCG vaccination but had no scars. Figure 1
shows the distribution of the tuberculin test readings. The median size of induration at 72 h was significantly larger than that at 48 h (9.4 vs 4.95 mm, respectively; p = 0.017). In those subjects with induration at either or both time points (n = 69), the readings at 72 h were on average 1.7 mm larger than those made at 48 h (95% confidence interval, 0.4 to 3.0 mm; SD, 5.4 mm). Using the Bland-Altman method,5
the 95% limits of agreement for measurements at these two time points were 12.5 to -9.1 mm. This means that for 5% of subjects, readings at 72 h were at least 12.5 mm greater or 9.1 mm smaller than those made at 48 h. There were also more positive test results at 72 h than at 48 h (Table 1
). While there was substantial agreement in the classification of readings at 48 and 72 h ( = 0.79), the results from 10 subjects (8.5%) changed with the time of measurement. Age did not influence tuberculin reactivity, as the age of subjects with induration of > 15 mm at either time point was not different from those with induration of < 15 mm (mean, 37.2 vs 35.3 years, respectively; p = 0.29).
All three observers measured more positive results at 72 h than at 48 h (Table 2
). In these 29 subjects, while some results that were negative at 48 h were positive at 72 h (observers A and B, four changes; observer C, three changes), none of the observers recorded a positive result at 48 h that became negative at 72 h. However, there was interobserver variation as all three observers recorded different numbers of positive results at reading 1. For all the readings taken (at both 48 and 72 h), there was moderate agreement between the three observers ( = 0.46). The repeated measurements (readings 2 to 4) of each of the three observers were on average < 1 mm different from that in reading 1.
|
Discussion
|
|---|
We have shown that the size of the tuberculin reaction may change with the time of reading. Measurements at 72 h were on average 1.7 mm larger than those at 48 h. However, some reactions changed in size by > 10 mm because of the time of reading. Using a cutoff point for induration of 15 mm diameter, which is often regarded as evidence of tuberculosis infection,1
2
this variability caused 8.6% of reactions to be reclassified, with more positive results being recorded at 72 h. While Sokol9
also noted that some tuberculin reactions do not peak until after 48 h, we could not find any published studies comparing results at 48h with those at 72 h. Our study indicates that measurements at 48 h may underestimate the reaction size and give a false-negative result.
All three observers measured more positive results at 72 h than at 48 h, confirming that the main finding of this study (ie, the tuberculin reaction size is larger at 72 h) was not based on the readings of a single observer. In particular, while all three observers found that some negative results at 48 h became positive at 72 h, none of the observers recorded a positive result at 48 h that became negative at 72 h. However, there was interobserver variation in the number of positive results. Interobserver variability is well-recognized,10
11
and our findings reemphasize that the measurement of tuberculin reactions is prone to subjective error.
The dose of tuberculin (bioequivalent to PPD-S) used in the United Kingdom and Australia is 10 U,2
12
while in the United States 5 U is used.1
It has been shown that the higher dose causes reactions that are on average 1.5 mm larger, which results in the reclassification of < 1% of tests, using a 15 mm cutoff point to define a positive result.12
Therefore, using the lower dose in this study is unlikely to have changed the results.
In summary, the current clinical guidelines for tuberculin testing1
2
have not standardized the time of measurement. Variability due to the time of measurement may lead to a different clinical management strategy. Our study indicates that tests in adults should be read at 72 h. Readings at 48 h may underestimate the reaction size.
|
Footnotes
|
|---|
Abbreviation: BCG = bacillus Calmette-Guérin
This work was performed at the North West Lung Research Center and Astrazeneca Pharmaceuticals.
Received for publication October 31, 2001.
Accepted for publication May 1, 2002.
|
References
|
|---|
- . American Thoracic Society (2000) Diagnostic standards and classification of tuberculosis in adults and children. Am J Respir Crit Care Med 161,1376-1395[Free Full Text]
- BritishThoracicSociety. JointTuberculosisCommittee. Control and prevention of tuberculosis in the United Kingdom: code of practice 2000. Thorax 2000;55,887-901[Abstract/Free Full Text]
- American Thoracic Society. The tuberculin skin test. Am Rev Respir Dis 1981;124,356-363
- Seibert, F, DuFour, E Comparison between the international standard tuberculins, PPD-S and old tuberculin. Am Rev Tubercul 1954;69,585-594
- Bland, J, Altman, D Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1,307-310[CrossRef][ISI][Medline]
- Fleiss, JL Statistical methods for rates and proportions 2nd ed. 1981 John Wiley New York, NY.
- Cohen, JA A coefficient of agreement for nominal scales. Educ Psychol Meas 1960;20,37-46[CrossRef][ISI]
- Fleiss, JL, Cohen, J The equivalence of weighted kappa and the intra-class correlation coefficient as measures of reliability. Educ Psychol Meas 1973;33,613-619[CrossRef][ISI]
- Sokol, JE Measurement of delayed skin test responses. N Engl J Med 1975;293,501-502[ISI][Medline]
- Frederick, J, Erdtmann, J, Dixon, K, et al Skin testing for tuberculosis; antigen and observer variability. JAMA 1974;228,479-481[CrossRef][ISI][Medline]
- Kendig, EL, Jr, Kirkpatrick, BV, Carter, H, et al Underreading of the tuberculin skin test reaction. Chest 1998;113,1175-1177[Abstract/Free Full Text]
- Stuart, R, Bennett, N, Forbes, A, et al A paired comparison of tuberculin skin test results in health care workers using 5 TU and 10 TU tuberculin. Thorax 2000;55,693-695[Abstract/Free Full Text]
TOP
Abstract
Introduction
