Hair analysis of formoterol as a novel tool to assess long‐term inhalation medication adherence: Data from the TRICOLON trial Authors: Liz J. A. Cuperus, A. Mireille A. Wessels, Huib A. M. Kerstjens, Daan J. Touw, Erik W. M. A. Bischoff, Job van der Palen, Arnoud Aldenkamp, Astrid van Huisstede, Folkert Brijker, Stephan Dik, Lidwien Graat‐Verboom, Martijn Goosens, Peter Th. W. van Hal, Jolanda C. Kuijvenhoven, Lisette I. Z. Kunz, Renske van der Meer, Mariette H. A. Scholma‐Bronsema, Job F. M. van Boven, Johannes C. C. M. in't Veen, Marjo van der Poel, Marjo van der Poel, Irini Piso, Marijke Amelink, Anke Rol, Jennifer de Graaf, Petra Hirmann, Fleur van Tour, Elly Jordens, Lydia Alfing, Gerda Lenderink, Truus Rietveld, Jasmijn van Campen, Jantine de Bruijn, Janice van ter Burg, Liselore Baarssmit, Walter van Litsenburg, Len Knoops, Margot Eggermont‐Schilt, Manon de Waard‐Heijligers, Ilonka Paalvast‐Schouten Published: 2026 ABSTRACT AimThe aim of this paper is to examine inhaled formoterol concentrations in hair and assess the potential of hair analysis as a tool for measuring long‐term adherence, by validation with objective data from digital inhalers.MethodsHair samples and digital inhaler data were collected from patients with COPD in 10 hospitals in the Netherlands as part of the TRICOLON trial. Hair samples were divided into three 1 cm segments, each representing 1 month of hair growth. Intra‐patient variability was investigated. Correlations between formoterol concentrations and digitally measured inhalations were assessed. Hair colour, dyeing, sex, body mass index (BMI) and smoking status were examined for their association and effect modification.ResultsOf the 96 patients who provided a hair sample, 49 had sufficient quality hair sample and digital data. Low intra‐patient variability was observed (intraclass correlation coefficient .94, 95% CI 0.87–0.97). Over a 3‐month period, a trend was observed in the correlation between the digitally recorded inhalations and hair formoterol concentration (correlation coefficient [r] = .25, p = .087). Regarding individual segments, a significant correlation was only found in the segment closest to the scalp (r = .30, p = .039). Higher formoterol concentrations were measured in dark hair and in those with a lower BMI. No effect modification by colour, dyeing, sex, BMI or smoking was observed.ConclusionLow intra‐patient variability of hair formoterol concentrations was observed in patients with stable adherence. However, inter‐patient variability and external factors may limit the correlation between formoterol concentrations in hair and adherence measured by a digital inhaler and thus the current utility of hair analysis as an adherence monitoring tool. What is already known about this subject Hair analysis can retrospectively assess drug intake over a longer period of time.Hair concentrations of formoterol have been associated with self‐reported adherence, but not with more objective adherence measures such as electronic monitoring by digital inhalers. Hair analysis can retrospectively assess drug intake over a longer period of time. Hair analysis can retrospectively assess drug intake over a longer period of time. Hair concentrations of formoterol have been associated with self‐reported adherence, but not with more objective adherence measures such as electronic monitoring by digital inhalers. Hair concentrations of formoterol have been associated with self‐reported adherence, but not with more objective adherence measures such as electronic monitoring by digital inhalers. What this study adds A trend was observed between digitally recorded inhaler use and formoterol concentration in hair over 3 months.Low intra‐patient variability of hair formoterol concentrations was observed in patients with stable adherence.Inter‐patient variability and external factors may limit the correlation between formoterol concentrations in hair and adherence measured by a digital inhaler. A trend was observed between digitally recorded inhaler use and formoterol concentration in hair over 3 months. A trend was observed between digitally recorded inhaler use and formoterol concentration in hair over 3 months. Low intra‐patient variability of hair formoterol concentrations was observed in patients with stable adherence. Low intra‐patient variability of hair formoterol concentrations was observed in patients with stable adherence. Inter‐patient variability and external factors may limit the correlation between formoterol concentrations in hair and adherence measured by a digital inhaler. Inter‐patient variability and external factors may limit the correlation between formoterol concentrations in hair and adherence measured by a digital inhaler. INTRODUCTION Medication nonadherence among patients with COPD is highly prevalent and associated with poor clinical outcomes and economic burden. 1 Accurately assessing adherence to inhalation medication is a critical step in addressing this issue. While various measurement methods are available, each comes with its own advantages and limitations. Self‐report is straightforward and easily implemented in daily practice; however, this method is subjective and patients are known to overestimate their adherence frequently. 2 Questionnaires such as the Test of Adherence to Inhalers (TAI) assess adherence more structurally, but remain subjective. 3 Pharmacy prescription monitoring is feasible, but does not measure actual intake and often poorly relates to the true level of adherence. 4 Digital or smart inhalers are currently regarded as the most objective and accurate method for measuring adherence to inhalation medication. Depending on the device type, they can also provide insights into inhalation technique. 4 , 5 However, digital inhalers come with certain limitations. They are unable to retrospectively assess medication adherence, which could be important when evaluating potential adjustments to pharmacological treatment. Additionally, their use in routine clinical practice remains limited, partly due to lack of reimbursement. Furthermore, like all electronic devices, digital inhalers may also face connectivity or technical issues, which can be especially challenging for patients with limited digital literacy. 6 Various bioanalytical methods for assessing adherence are available. While fractional exhaled nitric oxide (FeNO) could potentially serve as a monitoring tool for adherence, it is less suitable for patients with COPD, as it can only be used for corticosteroids and external factors such as smoking could influence the results. 4 Drug concentrations could be measured in blood, providing a direct measurement of medication intake. 7 However, given the relatively short half‐life of most inhaled medications, measurement of drug concentrations in blood only provides insight into short‐term adherence. 8 , 9 Furthermore, there is a risk that patients may take their medication only shortly before they know a blood sample will be collected, i.e. it is prone to whitecoat adherence. Therefore, hair analysis of inhaled drug concentrations is emerging as a promising tool for monitoring medication adherence. This objective bioanalytical method has the potential to assess long‐term adherence retrospectively. While already established in other research fields such as forensics, 10 a recent review highlighted that hair analysis currently holds limited value as a standalone method for assessing adherence to inhaled therapies, primarily due to the lack of validation with objective measures and understanding of factors influencing hair concentrations. 11 Therefore, the aim of this study was to investigate drug concentrations of inhaled medication in hair and to validate this technique against the currently most objective adherence measurement method: digital inhaler data. Only formoterol concentrations were measured, since beclomethasone could not be quantified in test hair samples from outside this study population. In addition, the impact of various factors on drug concentrations in hair, such as hair colour, is examined. METHODS Study design In this prospective study, hair samples and digital inhaler data were collected from patients with COPD simultaneously in order to correlate drug concentrations in hair with medication use as measured by a digital inhaler. Data were obtained from the TRICOLON trial, a 1‐year multicentre three‐arm randomized controlled trial investigating adherence in patients with COPD using single‐inhaler triple therapy (beclomethasone/formoterol/glycopyrronium) vs. multi‐inhaler triple therapy (beclomethasone + formoterol/glycopyrronium) with and without the use of a digital inhaler and app. The study protocol of this trial is described in more details elsewhere. 12 Data were collected from April 2023 to November 2024 in 10 hospitals in the Netherlands. Ethical approval for this study was granted by the Medical Research Ethics Committee‐United (NL79938.100.22). The trial is registered on clinicaltrials.gov (NCT05495698). Study population Participants were eligible for the TRICOLON trial if they had a COPD diagnosis and an indication for triple therapy, as determined by the treating physician in line with the Global Initiative for Chronic Obstructive Lung Disease (GOLD) strategy. Each patient used the combination of beclomethasone 100 mcg, formoterol 5 mcg and glycopyrronium 9 mcg in either one or two pressurized metered‐dose inhalers (pMDI) with all patients prescribed the same dosage: two inhalations twice daily. Patients from the TRICOLON study who were able and willing to provide a hair sample could be included. Patients were excluded from the hair analysis procedure if the hair sample was of inadequate quality, such as insufficient hair quantity or inability to identify which end of the sample originated from the scalp side. Patients were excluded from the statistical analysis if digital inhaler data were missing. Data collection During the second study visit of the TRICOLON trial, conducted 4 months after enrolment, patients provided a hair sample and had their digital inhaler data collected. The digital data was collected from the digital inhaler device of FindAir®. 13 This add‐on digital inhaler was attached on the top of the pMDI and registered the date and time of each actuation. The device was connected by Bluetooth to the e‐health app Gezondheidsmeter®, were data was stored. 14 Hair samples were collected for analysis of inhaled drug concentrations in hair. The sampling, preparation and testing procedures followed the guidelines established by the Society of Hair Testing (SoHT). 15 Hair sampling Hair samples were collected based on two inclusion criteria: (i) patient consent and (ii) a sufficient amount of scalp hair, with a minimum length of 3 cm (cm). Hair was taken from the posterior vertex region of the scalp, as this area has the least variation in growth rates, the highest proportion of actively growing hairs, and the fastest overall hair growth. 11 , 15 A hair lock at least 0.5‐cm diameter was secured with a small rope or elastic band, and the sample was cut as close to the scalp as possible. After cutting, the sample was wrapped in aluminium foil and stored in a dry location at room temperature. Hair preparation Prior to analysis, each hair strand was divided into three 1‐cm segments starting at the cutting side. Each segment was weighed to obtain a sample of preferably 20 mg. Given that the average hair growth rate is approximately 1 cm per month, this segmentation enables the measurement of drug concentrations for each month. Beyond the first 3‐cm closest to the scalp, external factors such as UV light exposure start to significantly influence drug levels in the hair, reducing the reliability of the results. 15 Segments were numbered outward from the scalp: Segment 1 (closest), Segment 2 (middle) and Segment 3 (most distal). Considering that at least 1 week of hair growth remains beneath the skin and that it is nearly impossible to cut hair exactly at the scalp, we assumed that the most recent 2 weeks before sampling were not represented in the cut sample. Accordingly, Segment 1 represented medication intake from 14 to 44 days before sampling, Segment 2 from 44 to 74 days and Segment 3 from 74 to 104 days prior to sampling. The digital inhaler data were matched to these corresponding time periods, Figure S1. After segmentation, the hair was washed to remove any external contamination such as hair products, sebum or residual drug. Hair analysis As part of the sample preparation, the hair samples were washed twice with 3.0 mL of dichloromethane for 10 min to remove external contamination like hairstyling products and unintentional presence of the drugs transmitted to the hair by handling the inhaler. The results of the analysis of the second step of this washing procedure are described in the supporting information. To extract the inhaled drugs embedded in the hair, an extraction fluid was added and the hair was pulverized, increasing its surface area and enhancing drug extraction. In this study, we used 1.0‐mL methanol spiked with the isotope labelled internal standard 13C,2H3‐formoterol as the extraction fluid. The hair was pulverized with small metal balls for 60 min in a ball mill at 30 Hz and the obtained suspension was and centrifuged for 5 min. The supernatant was being filtered using a mini‐uniprep filter unit. Five microliters of the filtrate was injected on the liquid chromatography–tandem mass spectrometry (LC‐MSMS) system for quantifying formoterol. More details on the settings and the validation of this analytical method can be found in the supporting information. To determine accuracy and precision when measuring the hair samples for this study, QC samples LLOQ, LOW and MEDIUM were analysed in triplicate in each analytical run as well. The results of the accuracies and precisions are displayed in the supporting information. For concentrations below the lower limit of quantification (BQL), half of the quantification limit (QL/2) was used as a surrogate value for the limit of detection (LOD). 16 Outcomes Intra‐patient stability of formoterol concentrations between Segments 1 and 3 was evaluated. For this analysis, only patients with relatively stable adherence between Segments 1 and 3 were included, defined as a difference of ≤0.5 mean daily inhalations between the corresponding months, as measured by the digital device. In the total cohort, the correlation between inhaled drug concentrations in 1‐cm hair segments and the mean number of inhalations per day in that corresponding month was assessed. In addition, the correlations between the mean drug concentration in the full 3‐cm hair sample and the mean daily number of inhalations over the corresponding 3‐month period were examined. As a sub analysis, correlations were examined separately for Segments 1, 2 and 3. Hair colour, BMI, sex, hair dyeing and smoking status are examined for their association with inhaled drug concentrations in hair, and their potential effect modification of the relationship between number of inhalations and drug concentrations in hair. Association refers to whether a variable directly influences an outcome; for example, whether men and women with equal number of inhalations have different in formoterol concentrations in hair. Effect modification, on the other hand, refers to whether a variable alters the strength or direction of the relationship between another predictor (number of inhalations) and the outcome (formoterol concentrations in hair). Statistical analysis Since there is no standard recommendation for the minimum number of clinical samples needed for hair analysis validation, 17 we used the recommendations for dried blood spot (DBS) analysis by Capieau et al. 18 They recommend a minimum of 25 patients when multiple samples are collected per patient, as in our case, where three 1‐cm segments per patient are created due to the segmentation of each hair sample. Based on the previous study by Salvator et al., which detected formoterol in 44% of the samples, we calculated that a minimum of 57 patients providing hair samples would be needed (25/44 * 100 = 57). 19 If additional hair samples were available at the end of the collection period, all would be included in the analysis. Additional hair samples could be included if, during the preparation phase, hair samples were of insufficient quality for analysis, or if, during the analytical phase, formoterol could be detected in fewer samples than anticipated. Adding extra hair samples had to be done prior to any statistical analysis. If during the analytical phase formoterol could be detected in more samples than anticipated, fewer patient samples would be needed. Descriptive data were presented as percentages, as means with standard deviations (SD), or medians with interquartile ranges [IQR], as appropriate. The intraclass correlation coefficient (ICC) was calculated using a two‐way random effects model to evaluate absolute agreement between Segments 1 and 3 in patients with a stable adherence. A scatterplot was generated to visualize the relationship. A linear mixed model was used to assess the correlation between drug concentration in each hair segment and the mean number of inhalations per day during the corresponding month. This model accounts for repeated measures, as each patient contributed three hair segments. Hair colour, hair dyeing, and sex were included as separate covariates to explore their associations with drug levels. To examine potential effect modification, interaction terms between number of inhalations and each of these variables (hair colour, hair dyeing, or sex) were added to the model. A linear regression was used to analyse correlations separately in the different segments, and to assess the correlation between the mean drug concentration in the entire hair sample and the mean daily number of inhalations over a three‐month period. A Pearson correlation coefficient (r) was determined to test the strength of these correlations. Nomenclature of targets and ligands Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org and are permanently archived in the Concise Guide to PHARMACOLOGY 2021/22. 20 RESULTS Study population During the sample collection period, 163 patients completed the second study visit of the TRICOLON trial. Among them, 63 (38.7%) did not have hair long enough for sampling, and 1 (0.6%) could not provide a sample due to the study visit being conducted by phone, Figure 1. The remaining 99 patients (60.7%) had a sufficient amount of hair. Of these patients, three (3.0%) did not consent to hair sampling, resulting in a total of 96 patients from whom hair samples were collected. Hair samples from 42 of these 96 patients (43.8%) had to be excluded from the hair analysis procedure due to insufficient hair quantity, inability to identify the scalp site of the sample, or improper fixation of the hair strand. Therefore, the hair samples from 54 patients were available for the hair analysis. An additional five patients were excluded from the statistical analysis due to connectivity issues with the digital inhaler, loss of the device or uncertainty regarding which digital inhaler was used with the formoterol medication. Consequently, hair samples from 49 patients were included in the final study cohort. Each patient contributed three hair segments, resulting in a total of 147 segments. Patient selection. Patient characteristics of the final study cohort and those who provided a hair sample are displayed in Table 1. More patients were female in the total study cohort (75.5%) compared to the percentage providing a hair sample (63.5%), but other characteristics were mostly similar. The median age was 68 [IQR 62–71]. Median weight was 75.0 [64.0–90.5] kilograms. The majority of the patients (n = 30, 61.2%) had grey hair, others had blonde (n = 8, 16.3%) or dark (n = 11, 22.4%) hair. Hair dyeing was reported by 19 patients (38.8%), including 18 females and one male. TABLE 1Patient characteristics.CharacteristicsStudy cohort (n = 49)Patients who provided hair sample (n = 96)Female sex, n (%)37 (75.5)61 (63.5)Age, median [IQR]68 [62–71]68 [63–72]Bodyweight in kilograms, median [IQR]75.0 [64.0–90.5]76.0 [64.1–90.0]BMI, median [IQR]25.7 [23.5–29.5]25.8 [23.1–30.8]Smokers, n (%) Active 13 (26.5)25 (26.0) Former 36 (73.5)71 (74.0) Never 0 (0)0 (0)FEV‐1 post % predicted, median [IQR]44.0 [34.5–59.5]44.0 [35.0–59.8]FEV‐1/FVC %, median [IQR]40.6 [33.5–48.0]40.7 [34.7–51.4]Comorbidity, n (%)Cardiovascular disease5 (10.2)14 (14.6)Heart failure1 (2.0)3 (3.1)Diabetes4 (8.2)11 (11.5)Asthma13 (26.5)24 (25.0)Obesity6 (12.2)17 (17.7)Depression4 (8.2)7 (7.3)Hair colour, n (%)Dark a 11 (22.4)25 (26.1)Blonde8 (16.3)12 (12.5)Grey30 (61.2)59 (61.5)Hair dyed (yes), n (%)19 (38.8)31 (32.3) a Dark hair refers to either black or brown hair. Patient characteristics. Dark hair refers to either black or brown hair. Formoterol concentrations in hair Overall, formoterol could be detected in 143 of the 147 collected hair segments (97.3%). In four segments, the analysis was disrupted, and no concentration could be determined. The median [IQR] concentration formoterol in all hair segments was 1.5 [1.0–2.2] pg/mg. Analysing each segment separately, median formoterol concentration was 1.5 [0.8–2.2] in Segment 1 (closest to the scalp), 1.5 [1.0–2.3] in Segment 2 (middle) and 1.7 [1.1–2.3] in Segment 3 (most distal from the scalp). Out of the 143 analysed hair segments, 40 (28.0%) had formoterol concentrations below the lower limit of quantification (LLOQ). Of these 40 segments, 23 had low sample weight (<20 mg), which resulted in a higher LLOQ. A minimum absolute amount of the drug is required for detection. This minimum detectable amount in smaller sample weights result in a higher concentration (= amount of drugs tested/sample weight), thereby increasing the LLOQ. These low‐weight segments were either blonde (n = 10) or grey (n = 23). Formoterol was quantifiable in 100% of dark hair samples (n = 33), 58.3% of blonde samples (n = 14) and 65.1% of grey samples (n = 56). Formoterol was quantifiable in a similar proportion of patients with dyed hair (73.6%) and with nondyed hair (71.1%). Intra‐patient stability of formoterol concentrations Between Segments 1 and 3, 30 patients had a stable adherence as measured by the digital inhaler. The intraclass correlation coefficient (ICC) was .94 (95% CI 0.87–0.97, p < .001). Figure 2 shows the correlation between the formoterol concentration in Segment 1 and Segment 3. Correlation between formoterol concentration in Segment 1 (y axis) and Segment 3 (x axis) in patients with a constant adherence (defined as a difference of ≤0.5 mean inhalations per day between the corresponding timeframes of these segments). The dotted diagonal line is the y = x line. Correlation between number of inhalations per day and formoterol concentration in hair In the total cohort including all 1‐cm hair segments (n = 143), the mean number of inhalations per day in the corresponding month was not statistically significantly associated with the formoterol concentration in hair (regression coefficient .00 with 95% CI −0.086 to 0.086, p = .998), Figure 3A. Additional analyses were performed, including only the hair segments in which formoterol was above the lower limit of quantification (n = 103) and only in those of patients with dark hair (n = 33). No statistically significant correlation was observed between the mean number of inhalations and hair formoterol concentration in either subgroup (regression coefficient .022 with 95% CI −0.078 to 0.123, p = .658, and regression coefficient .014 with 95% CI −0.085 to 0.113, p = .757, respectively). Correlation between number of inhalations per day and formoterol concentration in hair. In Figure 3A (left), each dot represents a single hair segment (three per patient; 147 segments in total): dark blue indicates Segment 1 (closest to the scalp), light blue represents Segment 2 (middle) and Grey corresponds to Segment 3 (most distal). The lines represents the regression lines: dark blue line for Segment 1, light blue for Segment 2 and grey line for Segment 3. In Figure 3B (right), each dot represents the entire hair sample, showing the mean formoterol concentration across the full 3‐cm hair segment and the mean number of inhalations per day recorded by the digital inhaler over the corresponding 3 months. The line represents the regression line. Investigating the closest (1), middle (2) and most distal (3) segments separately, a significant correlation was found for Segment 1 between the mean daily number of inhalations and hair formoterol concentration: regression coefficient .24 with 95% CI 0.013–0.47, p = .039, Figure 3A. The Pearson correlation coefficient (r) was .3. In Segments 2 and 3, the associations were not statistically significant: the regression coefficient was .23 with a 95% confidence interval of −0.118 to 0.568 (p = .193), and .28 with a 95% confidence interval of −0.083 to 0.649 (p = .126), respectively. Pearson correlation coefficient was .19 in Segment 2 and 0.23 in Segment 3. When examining the correlation between the formoterol concentration in the full 3‐cm hair sample and the mean number of inhalations per day over the corresponding 3‐month period, no statistically significant correlation was observed, although a trend was apparent: regression coefficient .28 with 95% CI −0.041 to 0.59 (p = .087), Figure 3B. Pearson correlation coefficient was .25. Factors influencing formoterol concentrations in hair Dark hair had statistically significantly higher formoterol concentrations compared to grey hair (regression coefficient 1.46, 95% CI 0.61–2.32, p = .001), Figure 4. There was no statistically significant difference between blonde and grey hair (p = .52). No effect modification by hair colour was observed. BMI showed a statistically significant inverse association with formoterol concentrations in hair (regression coefficient −.067, 95% CI −0.13 to −0.003, p = .041). BMI was not an effect modifier of the relationship between number of inhalations per day and formoterol concentrations in hair (p = .604). Sex was not statistically significantly associated with formoterol concentrations in hair (p = .83) and was not an effect modifier (p = .92). Similar findings were observed for dyed hair: hair dyeing status was not associated with formoterol concentrations (p = .536), and no effect modification was observed (p = .432). Finally, former smokers had no significantly different formoterol concentrations compared to active smokers (regression coefficient −0.57 with 95% CI −1.43 to 0.28, p = .182). Smoking status was no effect modifier (p = .647). Formoterol concentrations in different hair colours boxplot shows formoterol concentration in dark hair segments (n = 33), blonde hair segments (n = 24) and grey hair segments (n = 90). DISCUSSION Over a 3‐month timeframe, a trend in the correlation between the number of inhalations recorded by a digital inhaler and formoterol concentrations in hair was observed. When the segments and the corresponding one‐month period were examined separately, a significant correlation was only found in the segment closest to the scalp, reflecting the most recent medication intake. By contrast, more distal (older) segments showed no significant correlations. Furthermore, the observed correlations were of limited strength. Intra‐patient stability of formoterol concentrations was high among patients with stable adherence. Hair colour showed a strong association with formoterol concentrations, with higher concentrations measured in dark hair. An inverse relationship was observed between BMI and formoterol concentrations, so higher BMI was associated with lower formoterol concentrations in hair. This study demonstrated low intra‐patient variability in hair formoterol concentrations in patients with stable adherence, consistent with findings reported by Sportel et al. 21 However, our findings suggest that although formoterol concentrations appear stable over time, the correlation between digital inhaler data and hair formoterol concentrations is limited. Currently, this correlation was only statistically significant in the first segment, representing approximately the most recent month. Moreover, although statistical significance was observed in the first segment, the strength of the correlation was modest (r = 0.30). The reliability of the measurements may be insufficient to determine adherence on a month‐to‐month basis accurately. Interindividual variability among patients and the hair samples may complicate the interpretation of the results. For instance, differences in hair growth rates between patients could complicate accurate group‐level comparisons when aligning hair segments with the similar corresponding timeframes of the digital data. Furthermore, each hair sample may have been cut slightly differently with respect to distance from the scalp, introducing additional variability. 22 Both factors have greater influence when assessing short timeframes, whereas their impact could be reduced when analysing over a longer period, such as 3 months. Several factors may complicate the measurement of inhaled drugs in hair and the interpretation of the results. Notably, these drugs are incorporated into hair in very small amounts (pg/mg). Multiple factors, including the drug characteristics, influence the efficiency of this incorporation. 11 For example, although our initial plan was to measure both formoterol and beclomethasone, we were unable to quantify beclomethasone in hair samples collected outside this study population. Similarly, Hassall et al. were also unable to quantify beclomethasone. 23 This inability might be due to the relatively high lower limit of quantification of the bioanalytical method employed, highlighting the challenges of measuring certain inhaled drugs in hair. Furthermore, external factors influence drug concentrations in hair, which could result in either a decrease or an increase in the detected levels. Several studies on non‐inhaled drugs have shown that drug concentrations decrease in more distal hair segments, likely due to degradation resulting from external factors such as sunlight exposure or the use of shampoos and other cosmetic products. 24 , 25 , 26 Although this has not been specifically studied for formoterol or other inhaled medications, a similar degradation effect in older segments could be expected. However, it has also been reported that drug concentrations can be detected in segments corresponding to periods without drug intake, and one study observed that even a single drug administration may be detectable across multiple hair segments. 27 , 28 Together, these findings suggest that also elevated concentrations could occur in older hair segments. This may be explained by additional drug incorporation in the more distal segments through mechanisms such as diffusion from sweat or sebum, contamination from exhaled medication in the surrounding air, or axial diffusion between adjacent segments. 28 Furthermore, the hair shaft has a conical morphology that is characterized by progressive thinning in the more distal segments. As a result, the relative proportion of the cortex, the part of the hair where drugs are incorporated, increases compared to the overall hair structure. 29 Consequently, drug concentrations may appear relatively higher in these older segments. To date, three other studies have also investigated formoterol concentrations in hair. In our study, the median concentration was 1.5 [IQR: 1.0–2.2] pg/mg, whereas Salvator et al. found a median of 0.9 [IQR: 0.6–3.9] pg/mg, 19 Hassall et al. reported a wide range of 0.8–500 pg/mg without providing a median, 23 and Dierick et al. observed a mean of 1.5 pg/mg with a range of 1.1–1.9. 30 Similar to the study by Hassall et al., we detected higher formoterol concentrations in dark hair. 23 Furthermore, none of the dark coloured hair segments had formoterol concentrations below the lower limit of quantification (BLQ). This could probably be explained by two factors: first, none of the dark hair segments, but 23 of the blonde or grey segments, had low sample weight (<20 mg), leading to a higher lower limit of quantification; second, if melanin serves as the principal binding site of formoterol, this might explain this difference since dark hair contains more melanin. To the best of our knowledge, this is the first study to correlate inhaled drug concentrations in hair with digital inhaler data; others compared drug concentrations in hair with self‐reported adherence and the prescribed dose. 19 , 23 Digital inhaler data is currently considered the most objective and accurate method for measuring adherence, making it the most favourable approach for clinical validation of this technique. Nonetheless, no perfect gold standard for adherence measurement exists, and digital inhalers have limitations. For example, connectivity issues or loss of the digital device led to the exclusion of some patients from the statistical analysis. Moreover, patients could theoretically actuate the digital inhaler without actually inhaling the medication. Conversely, additional doses of formoterol might have been administered using a device without a digital inhaler. One patient, for instance, had no recorded inhalations on the digital device, yet formoterol was detected in the hair sample. Although all participants were clearly instructed to use only the devices with digital inhalers during the trial, this small risk remains in real‐world settings. Our findings suggest that hair analysis is generally acceptable to patients, as the majority was willing to provide a hair sample. However, several obstacles for implementation in clinical practise still exist. Due to insufficient quantity of hair, more often in male patients, eligibility for sampling was 60.7% in our study. Among these patients sampled, only 56.3% of hair samples was of sufficient quality and quantity for analysis, compared with 62.8% in Hassall et al. 23 Salvator et al. did not report whether any samples were inadequate for analysis, but noted that formoterol was detectable in only 44% of the analysed hair samples. 19 In contrast, we were able to detect formoterol in nearly all analysed samples (97.3%) and quantify it in 72.8%. Hassall et al., who also investigated other inhaled drugs besides formoterol, detected inhaled drugs in 72% of samples and was able to quantify them in 42%. 19 It should be taken into account that the patient groups were not entirely comparable: our study included patients with COPD who participated in a clinical trial, whereas Hassall and Salvator included patients with asthma and COPD from a routine clinical setting. Nevertheless, a substantial proportion of patients across all studies were unsuitable for sampling, provided samples of insufficient quality, or provided samples in which formoterol could not be detected. This poses a challenge in clinical settings, especially in elderly populations and men. Improving sampling techniques, accuracy of mass spectrometers, and providing more training and practice for healthcare professionals, could help to address these issues. In conclusion, this study provides further insight into the potential and limitations of hair analysis for assessing long‐term adherence to inhaled medication. These findings represent important initial steps towards applying this technique in respiratory medicine, although it is not yet ready for wide scale clinical implementation. Future work should focus on improving sampling methods, analytical procedures, and the sensitivity of mass spectrometry equipment to enhance accuracy. Further research is required to assess the applicability of this technique in other inhaled medications in both patients with COPD and asthma. Furthermore, it would also be beneficial to compare not only this new method (hair analysis) and the gold standard (digital inhalers) but also the most commonly used method in clinical practise: self‐report. Gaining more insight which patient would benefit the most of each measurement methods, together with taken into accounts the costs of each method, would further help tackling the nonadherence issue. AUTHOR CONTRIBUTIONS J.i.V. initiated this study. The steering committee, including L.C., J.V., H.K., J.v.d.P., E.B., J.v.B., were involved in the study design and protocol. A.W. and D.T. performed the hair analysis. L.C. performed the statistical analysis and wrote the first draft of the manuscript text. All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. CONFLICT OF INTEREST STATEMENT L.C.: the faculty has received an unrestricted grant from Chiesi Pharmaceuticals B.V. A.W.: none. H.K. has received research/educational grants and served on advisory boards for Boehringer Ingelheim, GSK and Novartis and has served on advisory boards for AstraZeneca and Chiesi. D.T.: none. E.B.: none. J.v.d.P.: none. A.A.: none. A.v.H.: none. F.B.: none. S.D.: none. L.G.: none. M.G.: none. P.v.H.: none. J.K.: none. L.K.: none. R.M.: none. M.S.: none. J.v.B. received research grants and/or consultancy fees from Aardex, ALK, AstraZeneca, Chiesi, European Commission, COST (COST Action 19 132), GSK, Novartis, Pfizer, Teva, Trudell Medical, Vertex and the World Health Organization, outside the submitted work and all paid to his institution. J.i.V. received unrestricted grants to faculty form GSK, Astra Zeneca, Teva pharma, Sanofi and Chiesi and received personal fees for speaker bureau and advisory board for Chiesi, Astra Zeneca and Sanofi. TRICOLON STUDY GROUP The TRICOLON Study Group consortium consists of the individuals listed below: Liz J.A. Cuperus1,2,3, Job van der Palen4, Arnoud Aldenkamp5, Astrid van Huisstede6, Erik W.M.A. Bischoff7, Job F.M. van Boven8, Folkert Brijker9, Stephan Dik10, Renske van der Meer11, Martijn Goosens12, Peter Th.W. van Hal13, Jolanda C. Kuijvenhoven14, Lisette I.Z. Kunz15, Huib A.M. Kerstjens 2, Johannes C.C.M. in't Veen1,3, Marjo van der Poel17, Irini Piso17,Marijke Amelink9, Anke Rol18, Jennifer de Graaf18, Petra Hirmann19, Fleur van Tour20, Elly Jordens13, Lydia Alfing13, Gerda Lenderink21, Truus Rietveld21, Jasmijn van Campen15, Jantine de Bruijn15, Janice van ter Burg15, Liselore Baarssmit15, Walter van Litsenburg5, Len Knoops22, Margot Eggermont‐Schilt23, Manon de Waard‐Heijligers23, Ilonka Paalvast‐Schouten23, Mariette Scholma24, Sandra Brookman‐Mozes24, Inge Paas24, Lidwien Verboom25, Karin van Mierlo‐Snoeijs25 Pulmonology Department, Franciscus Gasthuis and Vlietland, Rotterdam, The NetherlandsPulmonology Department, University of Groningen, University Medical Center Groningen, and Groningen Research Institute for Asthma and COPD, Groningen, The NetherlandsPulmonology Department, Erasmus Medical Centre, Rotterdam, The NetherlandsDepartment of Epidemiology, Medisch Spectrum Twente, Enschede, The NetherlandsDepartment of Pulmonary Medicine, Catharina Hospital, Eindhoven, The NetherlandsDepartment of Pulmonology, Northwest Clinics, Alkmaar, The NetherlandsDepartment of Primary and Community Care, Radboud University Medical Center, Nijmegen, The NetherlandsDepartment of Clinical Pharmacy & Pharmacology Groningen Research Institute for Asthma and COPD (GRIAC), University Medical Center Groningen, University of Groningen, Groningen, The NetherlandsDepartment of Pulmonary Medicine, Spaarne Gasthuis, Haarlem, The NetherlandsDepartment of Pulmonary Medicine, Albert Schweitzer Ziekenhuis, Dordrecht, The NetherlandsDepartment of Pulmonary Medicine, HagaZiekenhuis, The Hague, The NetherlandsDepartment of Pulmonary Medicine, Gelre Ziekenhuizen, Zutphen, The NetherlandsDepartment of Respiratory Medicine, Van Weel‐Bethesda Hospital, Dirksland, The NetherlandsDepartment of Respiratory Medicine, Medical Centre Leeuwarden, Leeuwarden, The NetherlandsDepartment of Pulmonology, Haaglanden Medical Centre, The Hague, The NetherlandsDepartment of Clinical Pharmacy, Franciscus Gasthuis, Rotterdam, The NetherlandsSpaarne Gasthuis, Haarlem, The NetherlandsNorthwest Clinics, Alkmaar, The NetherlandsMedical Centre Leeuwarden, Leeuwarden, The NetherlandsAlbert Schweitzer ziekenhuis, Dordrecht, The NetherlandsGelre Ziekenhuizen, Zutphen, The NetherlandsCatharina Hospital, Eindhoven, The NetherlandsHagaZiekenhuis, The Hague, The NetherlandsDepartment of Pulmonary Medicine, Wilhelmina Ziekenhuis Assen, Assen, The NetherlandsDepartment of Pulmonary Medicine, Maxima Medical Center, Veldhoven, The Netherlands Pulmonology Department, Franciscus Gasthuis and Vlietland, Rotterdam, The Netherlands Pulmonology Department, Franciscus Gasthuis and Vlietland, Rotterdam, The Netherlands Pulmonology Department, University of Groningen, University Medical Center Groningen, and Groningen Research Institute for Asthma and COPD, Groningen, The Netherlands Pulmonology Department, University of Groningen, University Medical Center Groningen, and Groningen Research Institute for Asthma and COPD, Groningen, The Netherlands Pulmonology Department, Erasmus Medical Centre, Rotterdam, The Netherlands Pulmonology Department, Erasmus Medical Centre, Rotterdam, The Netherlands Department of Epidemiology, Medisch Spectrum Twente, Enschede, The Netherlands Department of Epidemiology, Medisch Spectrum Twente, Enschede, The Netherlands Department of Pulmonary Medicine, Catharina Hospital, Eindhoven, The Netherlands Department of Pulmonary Medicine, Catharina Hospital, Eindhoven, The Netherlands Department of Pulmonology, Northwest Clinics, Alkmaar, The Netherlands Department of Pulmonology, Northwest Clinics, Alkmaar, The Netherlands Department of Primary and Community Care, Radboud University Medical Center, Nijmegen, The Netherlands Department of Primary and Community Care, Radboud University Medical Center, Nijmegen, The Netherlands Department of Clinical Pharmacy & Pharmacology Groningen Research Institute for Asthma and COPD (GRIAC), University Medical Center Groningen, University of Groningen, Groningen, The Netherlands Department of Clinical Pharmacy & Pharmacology Groningen Research Institute for Asthma and COPD (GRIAC), University Medical Center Groningen, University of Groningen, Groningen, The Netherlands Department of Pulmonary Medicine, Spaarne Gasthuis, Haarlem, The Netherlands Department of Pulmonary Medicine, Spaarne Gasthuis, Haarlem, The Netherlands Department of Pulmonary Medicine, Albert Schweitzer Ziekenhuis, Dordrecht, The Netherlands Department of Pulmonary Medicine, Albert Schweitzer Ziekenhuis, Dordrecht, The Netherlands Department of Pulmonary Medicine, HagaZiekenhuis, The Hague, The Netherlands Department of Pulmonary Medicine, HagaZiekenhuis, The Hague, The Netherlands Department of Pulmonary Medicine, Gelre Ziekenhuizen, Zutphen, The Netherlands Department of Pulmonary Medicine, Gelre Ziekenhuizen, Zutphen, The Netherlands Department of Respiratory Medicine, Van Weel‐Bethesda Hospital, Dirksland, The Netherlands Department of Respiratory Medicine, Van Weel‐Bethesda Hospital, Dirksland, The Netherlands Department of Respiratory Medicine, Medical Centre Leeuwarden, Leeuwarden, The Netherlands Department of Respiratory Medicine, Medical Centre Leeuwarden, Leeuwarden, The Netherlands Department of Pulmonology, Haaglanden Medical Centre, The Hague, The Netherlands Department of Pulmonology, Haaglanden Medical Centre, The Hague, The Netherlands Department of Clinical Pharmacy, Franciscus Gasthuis, Rotterdam, The Netherlands Department of Clinical Pharmacy, Franciscus Gasthuis, Rotterdam, The Netherlands Spaarne Gasthuis, Haarlem, The Netherlands Spaarne Gasthuis, Haarlem, The Netherlands Northwest Clinics, Alkmaar, The Netherlands Northwest Clinics, Alkmaar, The Netherlands Medical Centre Leeuwarden, Leeuwarden, The Netherlands Medical Centre Leeuwarden, Leeuwarden, The Netherlands Albert Schweitzer ziekenhuis, Dordrecht, The Netherlands Albert Schweitzer ziekenhuis, Dordrecht, The Netherlands Gelre Ziekenhuizen, Zutphen, The Netherlands Gelre Ziekenhuizen, Zutphen, The Netherlands Catharina Hospital, Eindhoven, The Netherlands Catharina Hospital, Eindhoven, The Netherlands HagaZiekenhuis, The Hague, The Netherlands HagaZiekenhuis, The Hague, The Netherlands Department of Pulmonary Medicine, Wilhelmina Ziekenhuis Assen, Assen, The Netherlands Department of Pulmonary Medicine, Wilhelmina Ziekenhuis Assen, Assen, The Netherlands Department of Pulmonary Medicine, Maxima Medical Center, Veldhoven, The Netherlands Department of Pulmonary Medicine, Maxima Medical Center, Veldhoven, The Netherlands Supporting information Figure S1. Schematic illustration of the different hair segments and its corresponding digital inhaler data. Table S1. Results of accuracy and precision and autosampler stability of formoterol QC samples. Table S2. Results of accuracy and precision of QC samples in study samples runs.