Introduction
Nasal obstruction is a common manifestation of rhinitis.Reference Starling-Schwanz, Peake, Salome, Toelle, Ng and Marks1 This problem is difficult to quantify directly from clinical examination, so objective assessments such as rhinomanometry and acoustic rhinometry are commonly used.Reference Starling-Schwanz, Peake, Salome, Toelle, Ng and Marks1, Reference Doorly, Taylor and Schroter2 However, these methods require complex, expensive equipment and highly trained operators. Subjective assessment of patient symptoms via questionnaires is also useful for diagnostic and research purposes.Reference Starling-Schwanz, Peake, Salome, Toelle, Ng and Marks1 However, subjective and objective measurements of nasal obstruction do not always correlate.Reference Starling-Schwanz, Peake, Salome, Toelle, Ng and Marks1, Reference Holmström, Scadding, Lund and Darby3 Therefore, a simple, objective measurement of nasal airflow would be a very useful tool for assessing nasal patency.Reference Jones, Viani, Phillips and Charters4
The peak nasal inspiratory flow (PNIF) meter is a simple, cheap, noninvasive tool for this purpose.Reference Holmström, Scadding, Lund and Darby3–Reference Fairley, Durham and Ell5 Furthermore, several authors have shown that PNIF is a reproducible measurement useful for the evaluation of nasal airway obstruction.Reference Klossek, Lebreton, Delagranda and Dufour6, Reference Cho, Hauser and Christiani7 Starling-Schwanz et al. have demonstrated a good correlation between PNIF and the subjective sensation of nasal patency in young adults.Reference Starling-Schwanz, Peake, Salome, Toelle, Ng and Marks1
Interpretation of PNIF data relies upon comparison of measured values with predicted values available from published reference equations or normal value tables. However, to our knowledge, reference equations or normal values have only been established in two populations: Italian subjects aged 16–84 years (PNIF reference equation, Ottaviano et al.),Reference Ottaviano, Scadding, Coles and Lund8 and Greek children aged five to 18 years (normal PNIF values, Papachristou et al.).Reference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9 However, neither of these studies provided prospective verification for their study populations, nor proposed a method of interpreting the measured PNIF (e.g. using a lower limit of normal, or a fixed percentage below which PNIF values would be considered abnormal).
The need for reference equations specific to North African populations has been demonstrated for spirometry,Reference Ben Saad, Rouatbi, Raoudha, Tabka, Laouani Kechrid and Hassen10–Reference Trabelsi, Ben Saad, Tabka, Gharbi, Bouchez Buvry and Richalet15 6-minute walk distanceReference Ben Saad, Prefaut, Missaoui, Hadj Mohamed, Tabka and Hayot16, Reference Ben Saad, Prefaut, Tabka, Mtir, Chemit and Hassaoune17 and fat-free mass data.Reference Gaied Chortane, Ben Saad, Ben Ounis, Zouhal, Gazzah and Tabka18 The applicability and reliability of published PNIF reference equations and normal valuesReference Ottaviano, Scadding, Coles and Lund8, Reference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9 should be assessed as regards North African subjects, in order to avoid erroneous clinical interpretation of PNIF data in this population.
Moreover, the American Thoracic Society and European Respiratory Society have encouraged investigators to publish reference equations and reference values for healthy populations of various racial backgrounds, to enable individual subject results to be compared with data from a racially similar population.Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19, Reference Quanjer, Stocks, Polgar, Wise, Karlberg and Borsboom20 Use of the same kind of assessment equipment and procedure is also recommended.
Therefore, we undertook the present study in order to calculate a PNIF reference equation for a North African population, thereby enabling simple, accurate and inexpensive evaluation of nasal airway patency in patients of a similar racial background.
The present study aimed: (1) to test the applicability and reliability of the previously published PNIF reference equationReference Ottaviano, Scadding, Coles and Lund8 and normal valuesReference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9 in North African subjects aged 13–27 years; and (2) if needed, to establish a PNIF reference equation in this population, and to prospectively assess its reliability.
Materials and methods
Before the study commenced, approval was obtained from the human research ethics committee of the Farhat Hached Hospital, Sousse, Tunisia. Written, informed consent was obtained from all adults; infants' parents were informed of the study and their consent was obtained.
Subjects
Upon enrolment in the study and prior to testing, subjects (or parents in the case of infants) were asked to complete two questionnaires: a medical questionnaire recommended for epidemiological research (used to assess subject characteristics),Reference Ferris21 and the Sino-Nasal Outcome Test 20.Reference Piccirillo, Merritt and Richards22
Subjects aged between 13 and 27 years were enrolled in the study if they fulfilled the following criteria: (1) no acute disease; (2) no chest malformations, congenital abnormalities or respiratory tract disease; (3) no cardiovascular or neuromuscular system disease or symptoms of allergic rhinitis for the last 12 months; (4) no complaints of nasal blockage, or other nasal symptoms; (5) no smoking history; (6) no previous nasal or paranasal sinus surgery; (7) a score of less than one on the Sino-Nasal Outcome Test 20 questionnaire; and (8) somatometric parameters above the third percentile.
A large number of subjects (i.e. 100 or more) was necessary to ensure that there was no significant difference between the normal values (and reference equation) established for those subjects, versus normal values for the subjects' general community.Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19 Therefore, we included 212 subjects (100 females and 112 males).
In order to verify the reliability of our calculated PNIF reference equation, we also prospectively measured PNIF values in a second group of 28 healthy subjects (14 females and 14 males) who met the study inclusion and exclusion criteria and who had not participated in the first part of the study.
Peak nasal inspiratory flow measurement
Peak nasal inspiratory flow was measured using an In-Check peak flow meter (Clement Clarke International Ltd., Edinburgh Way, Harlow, Essex, CM20 2TT, UK). Rubber anaesthesia face masks of variable sizes were used (Vital Signs, Totowa, New Jersey, USA); masks had to be large enough not to constrict the nose, but small enough to prevent air leakage under the chin.Reference Blomgren, Simola, Hytönen and Pitkäranta23, Reference Youlten24
All subjects were tested while sitting. They were asked to take a deep, quick, forced inspiration through the nose (and mask), keeping the mouth closed, following the end of a full expiration (i.e. the residual volume method). Three measurements were recorded for each subject; the greatest of the three results was used to calculate the PNIF. An additional recording was made in the event that incorrect technique was used. To avoid the introduction of variability due to different technicians, all measurements were performed (at the same time and under standardised conditions) by the same, highly experienced examiner.Reference Blomgren, Simola, Hytönen and Pitkäranta23
Data analysis
Subjects' highest PNIF values were selected for statistical analysis. Preliminary descriptive analysis included assessment of frequencies for categorical variables (e.g. gender), and means and standard deviations (SDs) for continuous variables (e.g. age and height).
The significance level was set at 0.05.
Data were analysed using Statistica software program (Statistica Kernel version 6, Stat Soft, France).
Univariate analysis
The dependent variable (PNIF) was normally distributed according to the Kolmogorov–Smirnov normality test.Reference Jenicek and Cleroux25 The t-test was used to evaluate the association between PNIF and categorical variables. The Pearson product-moment correlation coefficient was used to evaluate the association between PNIF and continuous variables.
Comparison with published reference equation and normal values
This comparison was made using two methods.
Firstly, individually measured PNIF values were compared with PNIF values predicted by the previously published reference equationReference Ottaviano, Scadding, Coles and Lund8 and normal values,Reference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9 for the same age range, using paired t-tests and scatter plots (for reference equation derived values) or non-parametric tests and histograms (for previously published normal values).
Secondly, as proposed by Bland and Altman, measured and predicted PNIF valuesReference Ottaviano, Scadding, Coles and Lund8 were compared using the limits of agreement method, whereby individual differences (i.e. measured values minus predicted values) were plotted against the corresponding mean value; limits of agreement were then calculated.Reference Bland and Altman26
New reference equation
New reference equations for North African PNIF values (for males, females and the total sample population) were established, using age, height and gender as predictors of PNIF, within a stepwise linear regression model. The accuracy of the model was evaluated using the correlation and determination coefficients and the standard error. The 95 per cent confidence interval (CI) was calculated as follows: 95 per cent CI = 1.64 × residual SD.Reference Jenicek and Cleroux25
Measured PNIF values lower than the lower limit of normal (= PNIF predicted value – 1.64 × residual SD) were considered abnormal.
Reliability of new reference equation
The reliability of our new PNIF reference equation, within a North African population, was evaluated using a second group of 28 healthy North African subjects. Prospectively measured PNIF values were compared with the PNIF values predicted from our new reference equation, using the paired t-test, scatter plotting and the Bland–Altman method.Reference Bland and Altman26
Results and analysis
An initial sample of 220 volunteer subjects (106 females and 114 males) was assessed. According to the epidemiological questionnaireReference Ferris21 and the Sino-Nasal Outcome Test 20Reference Piccirillo, Merritt and Richards22, non-inclusion criteria were found in 8 subjects. Therefore, a total of 212 volunteers (100 females and 112 males) were entered into the study.
Anthropometric and peak nasal inspiratory flow data
Males had significantly greater PNIF values compared with females (Table I). Figure 1 shows PNIF values for the total subject population, distributed according to age (Figure 1a) and height (Figure 1b). A significant difference in PNIF was found between subjects measuring 1.67–1.76 m versus 1.77–1.97 m in height (Figure 1b).
Fig. 1 Peak nasal inspiratory flow (PNIF) according to (a) age and (b) height. Central squares = means; upper and lower borders of boxes = standard deviations. Bars indicate statistical significance comparisons (Mann–Whitney U test). For age ranges: n = *79, †44, ‡62, **27. For height ranges: n = *22, †50, ‡82, **58. Total n = 212. NS = not significant
Table I Subject data
Data represent means ± standard deviations. *n=100; †n = 112; ‡n = 212. **p < 0.05, males vs females. yr = years; PNIF = peak nasal inspiratory flow
Univariate analysis
There was a significant correlation between PNIF values and male subjects' age and height (r = 0.27 and p < 0.05, and r = 0.36 and p < 0.05, respectively). There was also a significant correlation between PNIF values and the total sample population's age and height (r = 0.21 and p < 0.05, and r = 0.48 and p < 0.05, respectively). However, there was no significant correlation between PNIF values and female subjects' age or height (r = 0.18 and p > 0.05, and r = 0.18 and p > 0.05, respectively).
Comparison with published reference equation and normal values
Figures 2 and 3 show respectively the scatter plot and Bland–Altman comparison for measured versus predicted PNIF values as determined from the previously published Italian reference equation.Reference Ottaviano, Scadding, Coles and Lund8 We found no significant difference between our measured PNIF values and those predicted from the previously published reference equation (mean difference ± SD = −7 ± 47 l/min; p = 0.18) (Figure 3). However, comparisons of the two male subpopulations and the two female subpopulations identified wide limits of agreement and systematic errors for both males and females, as shown by the wide disparity from the identity line seen in Figure 2.
Fig. 2 Comparison of peak nasal inspiratory flow (PNIF) predicted by Italian reference equation versus PNIF measured in present study, for subjects in the same age range (16–27 years; 133 subjects). Figure shows the identity line (solid line) and the regression line (dashed line; y = −147.0945 + 1.8501 × x). Coefficient of determination = 0.30; regression coefficient = 0.54; p = 0.0.
Fig. 3 Bland–Altman comparison of measured peak nasal inspiratory flow (PNIF) values versus predicted PNIF values from Italian reference equation, for subjects in the same age range (16–27 years; 133 subjects). Dashed line = mean; solid lines = mean ± 1.96 × standard deviation
Figure 4 compares our measured PNIF values for subjects of different ages, versus normal values reported by the previously published Greek study.Reference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9 In subjects aged between 13 and 16 years, our mean ± SD PNIF values were significantly greater (p < 0.05) than the previously published normal values.
Fig. 4 Comparison of peak nasal inspiratory flow (PNIF) measured in the Greek study versus the present study, for subjects in the same age range (13–18 years; 93 subjects). ■ = mean for Greek study; ▴ = mean for present study; dotted line = standard deviation for Greek study; dashed line = standard deviation for present study. *p < 0.05 (unpaired t-test). Subject numbers: †10, ‡28, **41, §3, #5, ¥6. NS = not significant
New reference equation
We suspected that the previously published reference equationReference Ottaviano, Scadding, Coles and Lund8 and normal valuesReference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9 for PNIF were inaccurate as regards patients of North African origin; therefore, we established a new reference equation for this population, based on multiple regression analyses (Tables II and III). To facilitate simple interpretation of PNIF values, a reference equation should include only easily measured anthropometric data. Therefore, our reference equation contained only three common parameters: height, gender and age. In our total sample population, these parameters were significant independent predictors. The single model for the total sample appeared to explain 30 per cent of PNIF variability. We thus used this simplified model as the reference equation for our population (Tables II and III).
Table II North african PNIF reference equation parameters I
See text for reference equation (RE). *n = 100; †n = 112; ‡n = 212. **Male = 1, female = 0. SC = standardisation coefficient; yr = years
Table III North african PNIF reference equation parameters II
See text for reference equation. *n=100; †n = 112; ‡n = 212. r 2 = coefficient of determination; SE = standard error; LLN = lower limit of normal
Our North African simplified PNIF reference equation is as follows:
where gender = 0 for females and 1 for males.
After calculating the predicted PNIF value for an individual subject using this equation, the lower limit of normal for that subject could be obtained by subtracting 84 l/min.
In the 212 subjects in our study, the mean (95 per cent CI) measured PNIF corresponded to 100 per cent (96–104 per cent) of the predicted PNIF. In addition, no subject showed a PNIF which was less than 50 per cent of the value predicted.
Figures 5 and 6 show respectively the scatter plot and Bland–Altman comparison for measured versus predicted PNIF values determined from our new North African reference equation, for subjects aged 16–27 years. As reported for the Italian reference equation (Figure 3),Reference Ottaviano, Scadding, Coles and Lund8 our measured mean ± SD PNIF (difference between measured and predicted PNIF values = 0 ± 45 l/min) did not differ significantly from our predicted reference value (p = 0.98) (Figure 6). However, compared with the Italian reference equation, our North African reference equation enabled a slight improvement in the determination coefficient (Figures 2 and 5), superposition of the identity line with the regression line (Figure 5), and limits of agreement which were closer to zero (Figures 3 and 6), for the measured age range.
Fig. 5 Comparison of predicted peak nasal inspiratory flow (PNIF) values from North African reference equation versus measured PNIF values, for subjects in the same age range as the Italian study (16–27 years; 133 subjects). Figure shows the identity line (solid line) and the regression line (dashed line; y = −6.6150 + 1.0427 × x). Coefficient of determination = 0.31; correlation coefficient = 0.56; p = 0.0.
Fig. 6 Bland–Altman comparison of measured peak nasal inspiratory flow (PNIF) values versus predicted PNIF values from North African reference equation, for subjects in the same age range as the Italian study (16–27 years; 133 subjects). Dashed line = mean; solid lines = mean ± 1.96 × standard deviation
Reliability of new reference equation
The second group of 28 subjects had the following characteristics: 14 males, 14 females; mean ± SD age = 20 ± 4 years; and mean ± SD height = 1.68 ± 0.10 m. The mean ± SD PNIF prospectively measured in this group was 147 ± 40 l/min. This represented a mean (95 per cent CI) of 100 per cent (93–107 per cent) of the predicted value calculated by our North African reference equation.
There was a significant correlation between the prospectively measured and predicted PNIF values (r = 0.76; p < 0.01) (Figure 7).
Fig. 7 Comparison of predicted peak nasal inspiratory flow (PNIF) values from North African reference equation versus measured PNIF values, in 28 subjects assessed prospectively. Figure shows the identity line (solid line) and the regression line (dashed line; y = −4.0809 + 1.0262 × x). Coefficient of determination = 0.58; regression coefficient = 0.76; p = 0.0.
Figure 8 shows the Bland–Altman comparison for prospectively measured PNIF values versus predicted values determined from the North African reference equation. There was no difference between these two sets of values (mean ± SD difference = 0 ± 26 l/min; p = 0.98).
Fig. 8 Bland–Altman comparison of measured peak nasal inspiratory flow (PNIF) values versus predicted PNIF values from North African reference equation, in 28 subjects assessed prospectively. Dashed line = mean; solid lines = mean ± 1.96 × standard deviation
Discussion
This study prospectively measured PNIF in a large group of healthy North African subjects aged between 13 and 27 years. We found that a previously published reference equation and normal values for PNIF did not reliably predict PNIF in our subject population.
Accurate interpretation of naso-respiratory examination results requires reference to standards derived from a reference population.Reference Jenicek and Cleroux25 Comparison with such standard measurements assists clinicians to detect aberrations from normal, to confirm diagnosis and to monitor treatment efficacy.
Thus, by using height, gender and age as independent predictors, we established a new single reference equation which explained 30 per cent of the PNIF variability found in our subject population. Finally, we compared our new reference equation with PNIF values prospectively assessed in a second group of healthy subjects; the new reference equation generated satisfactory predictions, in contrast with a previously published reference equation and normal values.
Study procedure and subject group
When establishing a reference equation for physiological data, assessment of a relatively large number of subjects (suggested n = 100) ensures that there will be no significant difference between the published reference equation (or listed normal values) and the actual values measured in the general subject community.Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19 The sample size and methodological precautions used in the present study allowed us to obtain reliable results comparable with previous findings (Table IV).
Table IV Published PNIF reference data
In reference equations (ref eqns), peak nasal inspiratory flow (PNIF) is expressed in l/min, age in years (yr) and height (ht) in m. *Where male =1 & female = 0. M = male; F = female; PFM = peak flow meter; r 2=coefficient of determination; SE = standard error; LLN = lower limit of normal; CI = confidence interval
Peak nasal inspiratory flow measurements
Many researchers have used PNIF to evaluate nasal patency, establishing the reliability of this method.Reference Youlten24, Reference Pedersen, Hansen and Fuglsang27 Rhinomanometry has also been used as an acceptable, safe method of assessing nasal airway obstruction, with only minimal error;Reference Doorly, Taylor and Schroter2 however, it is time-consuming, expensive and not easily transportable, and also requires experience for accurate usage.Reference Wihl and Malm28 Therefore, a simple, reliable, inexpensive method of assessing nasal airway obstruction would be of value.
In the current study, procedural factors affecting PNIF variability were controlled as much as possible, by application of exclusion criteria, examination at the same time and by the same examiner, and use of standardised assessment conditions (e.g. normal room temperature and humidity).
Study limitations
Questionnaire results indicated that our subjects were free from acute and chronic cardiorespiratory disease. However, the use of a simple questionnaire to detect such conditions is debatable. The use of spirometry, chest X-ray and electrocardiography would have supplied objective data, enabling more accurate exclusion of unsuitable subjects.
Another debatable element of our assessment protocol was the lack of weight assessment. Indeed, in paediatric studies PNIF appears to follow a linear incremental pattern when plotted against weight.Reference Prescott and Prescott29
New reference equation
In this study, PNIF values were found to be influenced by height, gender (Table I) and age.
Our subjects' PNIF values tended to increase in proportion with their height (r = 0.48; p < 0.05). This agrees with previous findings indicating the significant effect of this variable.Reference Ottaviano, Scadding, Coles and Lund8, Reference Prescott and Prescott29
In our study, boys had significantly higher PNIF values compared with girls, for all age groups (respectively, 154 ± 45 I/min vs. 119 ± 31 I/min (13–15 years age group), 184 ± 56 I/min vs. 120 ± 37 I/min (16–20 years age group), 191 ± 57 I/min vs. 134 ± 33 I/min (21–23 years age group) and 181 ± 54 I/min vs. 140 ± 28 I/min (24–27 years age group)). In our population, this difference could be explained by subjects' height difference at the same age, since boys and girls of comparable height showed no difference in PNIF. These results allowed us to consider our whole subject population when establishing our reference equation, in line with other studies (Table IV).Reference Ottaviano, Scadding, Coles and Lund8, Reference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9
In agreement with previous findings,Reference Ottaviano, Scadding, Coles and Lund8 we found that subject age appeared to have an independent influence on total sample PNIF (r = 0.21; p < 0.05). However, Blomgren et al. found no appreciable association between age and PNIF.Reference Blomgren, Simola, Hytönen and Pitkäranta23 The gradual increase in height which occurs with growth was probably responsible for the higher PNIF values seen in our older subjects, as also noted by Ottaviano et al. and Papachristou et al. Reference Ottaviano, Scadding, Coles and Lund8, Reference Papachristou, Bourli, Aivazi, Futzila, Papastavrou and Konstandinidis9
Interpretation of peak nasal inspiratory flow results
As recommended by the American Thoracic Society and European Respiratory Society,Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19, Reference Quanjer, Stocks, Polgar, Wise, Karlberg and Borsboom20 publications on reference equations should include explicit definitions of the upper and lower limits of the normal range, or provide information which allows the reader to calculate upper and lower limits of the normal range. For each physiological data (e.g. PNIF), values below the fifth percentile of the frequency distribution of values measured in the reference population are considered to be below the expected ‘normal range’.Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19 If the reference data have a normal distribution, the lower fifth percentile can be estimated as the 5 per cent CI, using Gaussian statistics.Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19 If the distribution is skewed, the lower limit should be estimated using a non-parametric technique, such as the 95th percentile.Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19
In the present study, and as PNIF have a normal distribution, we have proposed to consider as lower PNIF values each measured PNIF lower than the lower limit of normal range [PNIF predicted value −5 per cent CI].Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19, Reference Crapo30 As recommended, the practice of using the 80 per cent predicted threshold as a fixed value for the lower limit of normal can lead to significant errors in PNIF interpretation.Reference Pellegrino, Viegi, Brusasco, Crapo, Burgos and Casaburi19 Indeed, when applied to our healthy subject population, use of the 80 per cent predicted threshold resulted in an erroneous diagnosis of nasal airflow obstruction in 50 (24 per cent) and four (14 per cent) subjects from our initial and subsequent prospective samples, respectively. If a fixed threshold is to be used, we recommend use of the 50 per cent predicted threshold.
• This study assessed peak nasal inspiratory flow (PNIF) in 212 normal North African subjects aged 13–27 years
• Recorded values were used to calculate a reference equation to enable interpretation of PNIF in patients of North African origin
When using our new reference equation, one should avoid extrapolation beyond the upper limit of height and age of our assessed subjects.Reference Stocks and Quanjer31 If a patient's age or height is outside the limits of our sample population, the interpretation should indicate that an extrapolation has been made.
Conclusion
We have established a reference equation for PNIF values in healthy North African subjects aged between 13 and 27 years. This PNIF reference equation represents a valuable addition to the world archive of reference equations. Clinicians and researchers should choose from this archive the appropriate reference equation for their patients' regional and ethnic origin.
Acknowledgements
The authors express sincere gratitude to all the subjects who participated in the present study. Dr Yassine Trabelsi is also acknowledged for his great support and technical assistance.
