Original article

S. VARECHOVA1, J. PLEVKOVA1, M. JAVORKA2, J. HANACEK1


RELIABILITY OF THE CAPSAICIN COUGH REFLEX SENSITIVITY
TEST IN HEALTHY CHILDREN


1Institute of Pathological Physiology and 2Institute of Physiology, Jessenius Faculty of Medicine, Comenius University, Martin, Slovakia


  Testing cough reflex sensitivity (CRS) in children requires suitable methodology. A CRS test performed under control of inspiratory flow rate (IFR) shows excellent reliability in children, but it is difficult to perform, especially in younger children. The aim of the present study was to find whether the capsaicin CRS test performed without direct control of constant IFR in healthy children is reliable enough for practical use. The CRS test was performed in 27 healthy children, aged 7-17 yr three times within 8 days. Cough was induced by inhalation of capsaicin aerosol in doubling concentrations (0.61-1250 µmol/l) for 400 ms each. CRS was defined as the lowest capsaicin concentration that evoked 2 or more coughs (C2). Although the intraclass correlation coefficient values showed good to excellent reliability of this test, the within-subject standard deviation values revealed lower reliability of this method compared to the CRS test performed under control of IFR. From the results obtained it is reasonable to conclude that the method using uncontrolled IFR in CRS testing provides acceptable precision only when a bigger sample size is used or more tests are performed. Good to excellent reliability of this method was found in children with higher values of C2 and in those aged 13-17 yr.

Key words: capsaicin, children, cough reflex, cough sensitivity



INTRODUCTION

Cough provocation testing in adults is used for over 50 years. Cough challenges are a safe and easy method for assessing the sensitivity of cough reflex (CRS) and cough intensity. As cough challenges can detect differences in CRS in disease they are used in specialized centres for the diagnosis of chronic cough and for clinical monitoring of disease or treatment progress (1, 2).

The cough response to inhaled tussigenic agents is dependent, among others, on the diameter of inhaled aerosol particles (3, 4) and on changes in inspiratory flow rate (IFR). Both factors may modify the pattern of deposition of an inhaled aerosol in the airways (5, 6). Barros et al (7) has shown a clear relationship between IFR and the cough response to inhaled capsaicin in that a higher IFR was associated with fewer coughs induced. Chang et al (8) has also pointed out that, when IFR is not controlled, the reliability of the test becomes low. Despite these findings, constant IFR has not been controlled in most recent capsaicin cough challenge studies. Methodology of the CRS test is not standardized up to now, so different research and clinical centers use different delivery, inhalation, and evaluation methods of cough provocation testing.

The number of cough challenge studies performed in children is much lower compared with that in adults. As children should be managed differently from adults (9), it is important to find a way of CRS testing that would be suitable for pediatric use. Such a method should be easy to perform and reliable enough to produce relevant results.

Results from a study of a capsaicin CRS test in children performed under controlled constant IFR show excellent reliability (8). The problem with using this method is that younger children are unable to maintain constant IFR for more than 2 s (8), which is not sufficient to complete the test. The question then arises of whether the CRS test using uncontrolled IFR, which is easier to perform in children, could be reliable enough for use in clinical trials.

The term reliability generally refers either to consistency of a measuring procedure or temporal stability of the target of the measurement (10). Test-retest reliability concerns the reproducibility of the observed value when the measurement is repeated in the same individuals. The aim of the present study was to assess the test-retest reliability of the capsaicin CRS test performed without direct control of constancy of IFR in healthy children.


MATERIAL AND METHODS

Study sample, criteria for enrolling into the study

The study was approved by the Ethics Committee of Jessenius Faculty of Medicine, Comenius University in Martin, Slovakia. Formal consent was obtained from the parents of all children participating in the study.

We studied 27 healthy children (15 boys and 12 girls aged 7-17 yr) from the Žilina and Martin region in Slovakia, who had no upper respiratory tract infection in the preceding 2 wk before examination, no history of allergic diseases, and no history of diabetes mellitus and other serious diseases.

Procedures

Subjects performed capsaicin CRS test, as described below, on three occasions within 8 days. Both the investigator and the subject were blinded to the results of a previous test. To avoid possible diurnal variation, children were always examined in the morning (08:00 – 13:00 h). For the capsaicin CRS test we used the method of Chang et al (8), with some modifications elaborated in our laboratory. Each subject inhaled the aerosol of a control solution (0.9% saline), followed by inhalation of 12 capsaicin aerosol concentrations in doubling doses (0.61, 1.22, 2.44, 4.88, 9.76, 19.53, 39.06, 78.12, 156.25, 312.5, 625, and 1250 µmol/l) at 1 min intervals. Inhaled solutions were nebulized by a ProvoJet nebulizer (Ganshorn Medizin Electronik, Niederlauer, Germany) driven by compressed air and connected to a breath-actuated dosimeter set at 400 ms inhalation time.

Cough reflex sensitivity was defined as the lowest capsaicin concentration that evoked 2 coughs (C2) or 5 coughs (C5). The number of coughs in 30 s after actuation of the dosimeter was counted. The end-point of each trial was when 5 coughs were obtained or when the maximum concentration of capsaicin was achieved. As the values of C2 were not normally distributed, we used natural logarithm (ln) transformed data for analysis. Spirometry (Microlab-ML 3500, MicroMedical, Gillingham, Kent, UK) was performed before and after capsaicin CRS test and the best of three blows was recorded and used for evaluation.

Statistical analysis

For calculation of reliability measures from the values of capsaicin CRS test obtained in our study we followed the instructions of Hopkins (11). We calculated three most important reliability measures: within-subject standard deviation (typical measurement error - WSSD), mean of the differences () and intraclass correlation coefficient (ICC).

Repeated measurements on the same subject vary around the true value because of measurement error (12). Assuming that the measurement error is random and normally distributed, the variability of observed values in a group of subjects (2O) is a combination of the variability of the true values in the group (2T) and the variability in the measurement error (2E): 2O = 2T +2E (14).

The standard deviation of measurements repeated in the same subject (WSSD) enables measuring the size of the measurement error (12). For this reason, it is the most important type of the reliability measure. For a reliability study with the use of only two measurements, it can be calculated as SD of differences/2 (13, 15).

The WSSD is a practical index of the measurement error only when it has the same average magnitude for every subject in the study. If WSSD varies between subjects, statisticians say the data display heteroscedascity or non-uniform error. We checked for non-uniform error by plotting subjects’ WSSD against subjects’ mean (16).

The mean of differences between two tests (or change in the mean - d) consists of two components: random change (it is like a randomly selected number added or subtracted from the true value every time the measurement is taken) and systematic change, which represents changes in the subjects’ behavior between trials (learning or training effect) (13).

Intraclass correlation coefficient estimates the average correlation among all possible orderings of pairs (17). This type of measure gives information on how closely the values of one trial correspond to the values of another trial when we move our attention from individual to individual. The interpretation of the ICC is as follows: In the statistical analysis of our results we did separate analyses for consecutive pairs of trials (trials 2-1 and trials 3-2). That way we wanted to see if there are any substantial differences in WSSD and between pairs of trials. Such differences are indicative of learning or practice effects (13). For calculations of reliability measures we used Excel spreadsheet created by Hopkins (11).

We used the plot of subjects’ mean vs. WSSD calculated from all tests to check if WSSD is or is not related to the magnitude of measurement. We also made separate analyses of different subgroups of subjects (according to age, gender, and magnitude of measurement) to check for non-uniform error (heteroscedascity).


RESULTS

Children tolerated the cough challenge very well; all of them completed the test. The feeling of scratching in the throat after inhalation of higher concentrations of capsaicin aerosol was worse tolerated during the first test than during the second and third one. Children were afraid of inhaling the next concentration when they had anticipated scratching in the throat. None of the children coughed with the control solution. The C5 parameter could not evaluated, as half of the children did not cough 5 or more times.

The geometric mean of C2 in the first trial was 61.5 µmol/l (95% CI: 23.6-160.8 µmol/l), in the second trial it was 120.3 µmol/l (95% CI: 53.5-270.4 µmol/l), and in the third trial it was 144.3 µmol/l (95% CI: 59.7-347.2 µmol/l).

The reliability measures calculated from repeated trials in this study are summarized in Table 1. As the 95% confidence interval of between the second and first trials does not cross zero, it is likely that a systematic error exists between these trials. In another analysis, according to children’s age and the magnitude of measurement, greater differences were seen in the mean of trials 2-1 than in that of trials 3-2 (Table 1). It is evident that WSSD varies among subjects and a non-uniform error is present despite logarithmic transformation of the outcome, which usually reduces such variations. Individual differences increased WSSD of all subjects in the study. The analysis of the groups divided according to age also shows that WSSD was greater also in younger children aged 7-12 years than that in older children aged 13-17 years (Table 1).

Table 1. Statistical analysis of reliability calculated from repeated trials. Data are in the ln-transformed state.

A group of children with a higher cough threshold (CT) had smaller WSSD than the children with lower CT. Clear evidence that the measurements become more variable as the magnitude of measurement decreases can be seen when we plot the subjects´ WSSD against the subjects´ means (Fig. 1). The Spearman rank correlation (r= -0.66; P= 0.0007) suggests the presence of a significant correlation between the two.

Fig. 1. Subjects’ WSSD plotted against subjects’ mean from all trials after ln transformation

Fig. 2. Cough reflex sensitivity, defined as the lowest capsaicin concentration that evoked 2 coughs (C2; in ln values) in the first and second trials plotted against each other.

Fig. 3. Cough reflex sensitivity, defined as the lowest capsaicin concentration that evoked 2 coughs (C2; in ln values) in the second and third trials plotted against each other.

From the values of WSSD we can estimate the range within which the subjects´ true value of the C2 parameter lies. To calculate this range, we had to multiply and divide the observed value by (aw)2, where aw is geometric WSSD (antilog of WSSD in natural logarithmic scale) (16). Taking the first trial as a practice trial, we can see the effect of WSSD on the estimation of the range of subjects´ true value, when we used the geometric mean of C2 from the second test and WSSD calculated from trial 3-2 (Table 2).

Table 2. Precision of the capsaicin CRS test as represented by the range of the subjects´ true value of cough threshold (C2) and expressed in µmol/l and by the number of concentrations of capsaicin aerosol (out of 12) within which the true value of C2 lies.

Figure 2 and Figure 3 show the scatter diagrams of consecutive pairs of trials plotted against each other. These data gave an overall ICC of 0.72 (95% CI: 0.48-0.87) for trial 2-1 and 0.76 (95% CI: 0.54-0.89) for trial 3-2. The results showed that reliability of the CRS test, according to the overall ICC, ranges from good to excellent. A separate analysis showed that the ICC of the subgroups ranges from poor reliability in children with lower values of C2 to excellent reliability in children with higher values of C2 and in children aged 13-17 years (Table 1).


DISCUSSION

In this study we calculated three important reliability measures for the capsaicin CRS test in children, performed without direct control of IFR constancy. According to overall ICC for the CRS test, reliability ranged from good, in trial 2-1, to excellent in trial 3-2. These findings are similar to the value of ICC of 0.78 for the CRS test performed under control of constant IFR, presented by Chang et al (8). The retest correlation is clearly a good measure of reliability and has an advantage of being dimensionless, but for most applications WSSD is a better index of measurement error (19, 20) and the two types of tests cannot be compared only on the basis of their retest correlations (13).

Calculations of WSSD and mean of differences revealed two interesting facts: (i) the presence of a significant learning effect between the first and second test trials, and (ii) the presence of a non-uniform error, which makes the CRS test values obtained in older children and children with higher CT more reliable compared with younger children and children with lower CT.

The difference in the means of the first and second CRS test trials could be caused by stress from lack of previous experience with such an examination and by the feeling of scratching in the throat after inhalation of capsaicin. Familiarization with a single-breath inhalation method and capsaicin challenge apparently caused minimal changes between the means of the second and third CRS test trails. Similar effects also were observed by Morice et al (21) who found a significant learning effect demonstrated between the first and second days of challenge for citric acid in adults.

Subjects may modify their IFR during capsaicin aerosol inhalation when they anticipate the noxious stimulus (7). Lack of control of IFR constancy could be responsible for inhalation with lower IFR with the feeling of scratching in the throat and it is likely that this fact played a crucial role in the change between the means of the first and second test trials. This view is supported by the results of Chang et al (8) who did not find any change between these means in their study performed in the condition of controlled constant IFR.

The presence of a non-uniform error made it impossible to interpret the overall WSSD calculated in the present study. The reason is that the WSSD value is too high for some children and too low for the others. An appropriate approach in the presence of a non-uniform error would be to analyze and interpret the reliability of the subgroups separately (13).

The WSSD value for children with higher cough thresholds showed that the true value of C2 lies within a range of 3 concentrations out of the 12 used (Fig. 2). This suggests a very good precision of the measurement of C2 in this group of children, because it makes the scale of 12 concentrations wide enough to notice or measure a change in CT, caused by disease or its treatment. The problem arises that the WSSD value for children with lower CT suggests poor precision of C2 measurement, because the true value lies within the range of 9 concentrations in this group of children.

Why would children with lower CT have a greater variability of capsaicin CRS? It is possible that it could be due to upper respiratory tract infection (URI) that began during the period of testing or was present before testing and went recognized by the observers. The presence of URI may have caused that the CT has not been normalized at the time of testing. The overall WSSD value of the capsaicin CRS test in children published by Chang et al (8) is similar to the WSSD value in children with higher CT obtained in our study.

We conclude that the capsaicin CRS test performed without direct control of constancy of IFR provides acceptable precision and could be used in comparative studies in children, although the reliability of such a test is lower than that when IFR is controlled. Reliability of the CRS test ranges from good to excellent; the latter takes place in a group of children with higher values of C2 (higher CT) and in children aged 13-17 years. If the measurement error (WSSD) is greater than the smallest clinically important change in capsaicin CRS for the study group, the capsaicin CRS test performed without direct control of IFR constancy will provide an acceptable precision only when a bigger sample size is tested or more CRS tests are performed in the same group of individuals.

Acknowledgments: We thank to Lila Surinova, Lenka Mazurova, Tomas Zatko, and Marta Ilovska for their outstanding technical assistance and help. This work was supported by Science and Technology Assistance Agency under contract No. APVT-20-005304.


REFERENCES
  1. Fuller RW. Cough Provocation Tests: Their Clinical Value. Pulm Pharm Ther 2002; 15: 273-276.
  2. Fuller RW. Cough sensitivity: the use of provocation tests. In Cough: Causes, Mechanisms and Therapy, F Chung, J Widdicombe (eds). Homer Boushey Blackwell Publishing, 2003, pp. 49-55.
  3. Bates DV, Fish BR, Hatch TF, Mercer TT, Morrow PE. Deposition and retention models for internal dosimetry of the human respiratory tract. Task group on lung dynamics. Health Phys 1966; 12: 173-207.
  4. Hansson L, Wollmer P, Dahlback M, Karlsson JA. Regional sensitivity of human airways to capsaicin-induced cough. Am Rev Respir Dis 1992; 145: 1191-1195.
  5. Ryan G, Dolovich, M, Obminski G et al. Standardization of inhalation provocation tests: influence of nebulizer output, particle size and method of inhalation. J Allergy Clin Immunol 1981; 67: 156-161.
  6. Agnew JE. In Aerosols and the Lung. S Clarke, D Pavia (eds). London, Butterworths, 1984, p. 197.
  7. Barros MJ, Zammattio SL, Rees PJ. Effect of changes in inspiratory flow rate on cough responses to inhaled capsaicin. Clin Sci 1991; 81: 539-542.
  8. Chang AB, Phelan PD, Roberts RGD, Robertson CF. Capsaicin cough receptor sensitivity test in children. Eur Respir J 1996; 9: 2220-2223.
  9. Chang AB. Cough: Are children really different to adults? Cough 2005; 1: 7.
  10. Nunnally JC, Bernstein IH. Psychometric Theory. New York, McGrow-Hill Inc, 1994.
  11. Hopkins WG. Reliability from consecutive pairs of trials. Excel spreadsheet. In A New View of Statistics. Sportsci Org Internet society of sport science. Sportsci.org/resource/stats/xrely.xls (2000).
  12. Bland JM, Altman DG. Statistic Notes: Measurement error. BMJ 1996; 313: 744.
  13. Hopkins WG. Measures of reliability in sports medicine and science. Sports Med 2000; 30: 1-15.
  14. Williams AE, Maskarinec G, Franke AA, Stanczyk FZ. The temporal reliability of serum estrogens, progesterone, gonadotropins, SHBG and urinary estrogen and progesterone metabolites in premenopausal women. BMC Women´s Health 2002; 2: 13.
  15. Chinn S. Repeatability and method comparison. Thorax 1991; 46: 454-456.
  16. Bland JM, Altman DG. Measurement error proportional to the mean. BMJ 1996; 313: 106.
  17. Bland JM, Altman DG. Statistic Notes: Measurement error and correlation coefficients. BMJ 1996; 313: 41-42.
  18. Shrout PE, Fleiss JL. Intraclass correlations: Uses in assessing rater reliability. Psychol Bull 1979; 86: 420-428.
  19. Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Med 1998; 26: 217-238.
  20. Hopkins WG, Hawley JA, Burke LM. Design and analysis of research on sport performance enhancement. Med Sci Sports Exerc 1999; 31: 472-485.
  21. Morice AH, Higgins KS, Yeo WW. Adaptation of cough reflex with different types of stimulation. Eur Respir J 1992; 5: 841-847.

Author’s address: S. Varechova, Institute of Pathological Physiology, Jessenius Faculty of Medicine, Comenius University, Sklabinská 26 St., 03753 Martin, Slovakia; phone: +42 1434238213, fax: +42 1434134807.
e-mail: varechova@lefa.sk