For this study, a subset of data from CanCOLD was used. The 6MWT was performed at the first follow-up assessment (CanCOLD visit 2, ∼18 months after the baseline visit) and comprises a subset of 1019 participants. Recruitment for visit 2 was not completed for the entire cohort. CanCOLD is a prospective, random sampled, population-based study conducted across nine sites in Canada (ClinicalTrials.gov Identifier: NCT00920348).33 Participants were noninstitutionalized males and females aged ≥40 years recruited by random telephone digit dialling. All participants provided written informed consent before completing study assessments. The research ethics board of each participating institution approved the study protocol.

Details on participant selection criteria for the current study are described in Fig. 1. Because of the limited number of non-Caucasian participants only Caucasian participants were selected for these analyses. Furthermore, participants were selected if they performed pulmonary function tests (PFT) (including post-bronchodilator spirometry and plethysmography), two 6MWTs, and had a peak rate of oxygen uptake (V'O2peak) on a symptom limited incremental cardiopulmonary cycle exercise test (CPET) between the 95% upper (ULN) and LLN values.34 Having an exercise capacity within normal predicted limits indicated that a participant's exercise tolerance was not limited by any ostensible health condition or any health hazards such as exposure to environmental tobacco smoke nor was there an abnormally high cardiorespiratory fitness, for instance resulting from intensive exercise training.

Fig. 1.

Flowchart of included participants.

(0.62MB).

Participants were excluded from the analyses if: 1) their post-bronchodilator spirometry indicated an (reversible) airflow obstruction or abnormal pulmonary function according to Global Lung Function Initiative reference values35-37 (FEV1, forced vital capacity (FVC) or forced vital capacity (FVC) less than LLN, or total lung capacity (TLC) less than 80%-predicted or an increase in FEV1 or FVC >12% and >200 mL from baseline 10–15 min after bronchodilator administration. A full description of both pre- and post-bronchodilator PFT outcomes are described in Table 1); 2) they had a cigarette smoking history >30 pack years (PY) (based on a univariate regression analyses between different categories of PY, and an additional sensitivity analysis see online Table 3 and online Table 4); and/or 3) they reported clinically significant pulmonary symptoms based on Medical Research Council (MRC) scores (MRC score ≥3) or COPD Assessment Test (CAT) total scores (CAT score ≥95th percentile (ULN) based on age and sex).38

The data used in this study was collected during two time points. Both CPET and PFTs were performed during CanCOLD visit 1. During CanCOLD visit 2, conducted 18 months after visit 1, the 6MWTs were performed. At both visits, pulmonary symptoms were assessed using MRC and CAT. To ensure that no significant changes in pulmonary symptoms occurred in the time between CanCOLD visits 1 and 2, participants needed to report a MRC dyspnoea score <3 and CAT total score

At each visit, general participant characteristics were recorded, as well as previous and current health conditions.

Data distribution was assessed with the Shapiro-Wilke test. Between-group differences were assessed using Chi-square or Fisher-exact tests for categorical variables, and T-tests or Mann Whitney U tests as appropriate for continuous variables.

The cut-off value for PY was determined by a univariate regression analysis between different categories of PY. An additional sensitivity analysis was performed to determine the effect of a lower cut-off value for PY on the results of the univariate regression analyses.

To derive reference equations, univariate regression analyses and multivariate stepwise regression analyses were performed after confirming all assumptions were met. Age, height and body mass (as applicable) were used as predictor variables and 6MWD and 6MWORK as outcome variables. All predictors were included in the final model based on their magnitude (β), significance and physiological impact (r2) on the outcomes.15 Separate reference equations were created for males and females.

For each reference equation, the root mean square error (RMSE or standard deviation of the residuals) was calculated and used as an indicator of the data around the regression line. In order to assess the difference between observed and predicted data, the mean absolute error (MAE) was calculated. The lower limit of normal or 5th percentile (LLN), estimated as the predicted value minus 1.645 multiplied by the RMSE, was calculated to determine below which value the outcomes are regarded as being abnormally low. A Pearson correlation coefficient was calculated to assess the association between predicted and measured values. Finally, a comparison between existing24-31 and our new 6MWD reference equations was made by calculating reference values for all included reference equations using the characteristics from the CanCOLD participants used in our analyses. A priori, the level of significance was set at p<0.05. Statistical analyses were performed using SAS 9.4 (SAS institute, Cary, NC, USA).

1019 participants were screened for eligibility for the current analyses. A total of 335 participants (51% male, age: 68.0 ± 9.1 years) were included. Participants’ general characteristics are presented in Table 1 and Online Tables 6 and 7. On average, participants were overweight (body mass index: 27.0 ± 4.5 kg/m2), and had a V'O2peak (95±21%-predicted) and PFT outcomes within normal predicted limits (FEV1: 101±15%-predicted; FVC 105±15%-predicted; FEV1/FVC: 74±6%; TLC: 105±14%-predicted; and DLCO: 98±20%-predicted). On average, participants reported 4.7 ± 8.5 PY. Participant were similar to the general Canadian population ≥40 years with regards to body mass and height (Mean body mass: Canadian population: 86.5 kg (men), 73.7 kg (women)39 vs included participants: 82.4 kg (men), 69.5 kg (women); mean height: Canadian population: 174.4 cm (men), 161.2 cm (women)39 vs included participants 174.3 cm (men), 160.7 cm (women)). Self-reported health conditions were present in ∼95% of the participants. A full description of the participant's self-reported health conditions and medication use is described in Online Tables 7 and 8.

On average, 6MWD and 6MWORK were 541±98 m and 41.3 ± 11.2 kg.m, respectively. All outcomes were significantly greater in males than females (Table 1).

The univariate regression analysis showed significant associations between age, sex and height versus each of 6MWD and 6MWORK (Online Table 9). The association between body mass and 6MWD was not statistically significant. In the multivariate regression analysis, all predictor variables were significant (Table 2). The derived sex-specific reference equations are listed below: Reference equations for males:

• -

  6MWD (m) =489.22–4.33*ageyrs+3.19*heightcm-2.18*body mass kg

• -

  6MWORK (kg.m) =−32,501.0–384.40*ageyrs+605.84*heightcm

Reference equations for females:

• -

  6MWD (m) =498.06–4.80*ageyrs+2.64*heightcm-1.24*body mass kg

• -

  6MWORK (kg.m) =7207.57–460.55*ageyrs+370.41*heightcm

Even though it is recommended to perform two 6MWT's,14 in clinical practice it may not be possible to perform two 6MWT's and subsequently use the best of the two tests as the final outcome measure. This is why the results of the regression analysis and reference equations based on the first 6MWT are included in the online supplement (Online Table 5).

The online supplement also contains a spreadsheet to calculate predicted values.

The explained variance (cumulative r2) of the multiple regression model ranged from 0.24 to 0.35. Pearson correlations (r, presented in Fig. 2) between predicted and observed values for 6MWD and 6MWORK varied between 0.495 and 0.589 and were all statistically significant (p<0.001). Bland Altman plots of observed and predicted values are presented in Online Fig. 1.

Fig. 2.

Correlation between predicted and actual six-minute walk distance (6MWD; Panels A&B) and six-minute walk work (6MWORK; Panels C&D).

(0.33MB).

Fig. 3 shows the predicted 6MWD based of the newly derived 6MWD reference equations, both based on the best test and based on the first test, compared to results from other reference equations,24-32 using data from the CanCOLD dataset (n = 346). The line representing the CanCOLD predicted 6MWDs is within the range of the lines generated from the other prediction equations.24-31 Compared to Enright et al.31 and Enright & Sherrill,30 the newly developed equation's predicted 6MWD values are consistently higher except for the youngest males in the sample. In contrast, the newly developed equation's predicted 6MWD values are consistently lower than those predicted using the equations of Troosters et al.,29 Hill et al.,25 Gibbons et al.28 and Jenkins.26 The equations of Chetta et al.27 are mostly below the newly derived predicted 6MWD, but both lines presenting the predicted 6MWDs cross each other in the older participants. The reference values based on Cazzoletti et al.32 are very close to the CanCOLD reference values for both genders, especially in the men aged 75 years or older and women aged 75 years or younger, while the equations of Beekman et al.24 show a different pattern for males and females. Whereas the line of the predicted values for females is consistently close to the line of the newly derived predicted values, the predicted 6MWD for males crosses the line of the CanCOLD derived predicted 6MWD around the age of 70 years. Younger participants have higher predicted values and older participants have lower predicted values compared to the CanCOLD reference values.

Fig. 3.

Comparison of newly derived reference equations for six-minute walk distance (6MWD) to existing reference equations.

(0.19MB).

A clear strength of this study is the comprehensive assessment performed on CanCOLD participants, which provided us with a unique opportunity to identify a subset of adults with normal pulmonary function, normal self-reported health status, normal breathlessness, and V'O2peak on symptom limited incremental cycle CPET within normal predicted limits, where CPET is widely considered the gold-standard method of assessing exercise capacity.46 Next to this, care was taken in determining a valid cut-off value of ≤30 PY as inclusion criterion. Next to the univariate regression analysis assessing the effect of different categories of PY (Online Table 3), an additional sensitivity analysis was performed in which the multivariate stepwise regression analysis was repeated with data from participants with ≤5 PY. This analysis resulted in similar point estimates, indicating that the seemingly high cut-off value of >30 PY is valid (Online Table 4). Using these selection criteria instead of selecting participants based on the complete absence of comorbidities, has led to a unique and representative sample of the Canadian population of adults, aged ≥40 years. Since many Canadians suffer from comorbidities,47 the prediction equations developed in the current study are likely more relevant for use by healthcare providers in clinical practice.

The 6MWD is susceptible to a learning effect, which reaches a plateau after performing two tests within one week.14 All participants included in our analyses performed two 6MWTs in order to decrease the likelihood of a learning effect and ensure optimal performance.

While interpreting the results, some limitations need to be considered. All 6MWTs were performed on a 20 m course, whereas the ATS guidelines for the 6MWT recommend a 30 m course length.14 Several studies have investigated the effect of course length on 6MWD and the results are inconclusive. Significantly higher distances in 30 m courses compared to 20 m courses were found in healthy adults,48 patients with COPD49,50 and individuals with stroke.51 In addition, Beekman et al.52 found a significant effect of a 10 m versus 30 m course length on 6MWD. In contrast, Veloso-Guedes et al.53 and Sciurba et al.42 found no significant effect of course length on 6MWD in patients with liver cirrhosis and patients with COPD, respectively. In addition, the European Respiratory Society/ATS technical standard report for field walking tests in people with chronic respiratory disease15 indicated that for course lengths >15 m, differences in 6MWD may be small enough such that 6MWTs performed on courses of different lengths can still be used for risk stratification. However, based on the above-mentioned studies it is recommended to use course length-specific reference equations.

Even though all participants included in our analyses had a V'O2peak on symptom limited incremental cycle CPET within normal predicted limits, it is nevertheless possible that comorbidities or other factors (e.g., intermittent claudication (reported by one participant), motivation or weather conditions) may have led to a suboptimal performance during the 6MWT despite V'O2peak being within normal predicted limits.

The CPET and 6MWTs were performed 18 months apart, during which time the health status and functional capacity of our participants might have changed. However, we mitigated the risk of clinically meaningful changes in physical capacity by including only participants with normal self-reported ratings of respiratory health status and activity-related breathlessness at both CanCOLD Visits 1 and 2.

Although the sample is representative of the Canadian population aged ≥40 years, the majority of participants was 50–80 years old (86%). Reference values may be less accurate for adults <50 and >80 years old.15,30