Energy Expenditure Estimation for Forestry Workers Moving on Flat and Inclined Ground

Abstract

Forestry workers endure highly physical workloads. Japanese forestry workers experience additional up-and-down movements due to geographical features. Fatigue is a common cause of injury. This pilot study aimed to determine an appropriate method for estimating energy expenditure while moving across inclined ground to simulate a Japanese forest. Six participants wore a portable indirect calorimeter ($\dot{\mathrm{V}}$O₂), heart rate (HR) monitor (17 g), accelerometer (20 g; vector magnitude; VM), and a global navigation satellite system (GNSS) device. They walked shouldering 20 kg of weight on flat, 15°- and 30°-slopes. The time course of HR was similar to that of $\dot{\mathrm{V}}$O₂, but that of VM and the vertical movement varied from that of $\dot{\mathrm{V}}$O₂. GNSS cannot correctly detect vertical movements. The HR index (HRI), indicating the ratio of activity HR to resting HR, was significantly correlated with the metabolic equivalent of the task (MET) calculated from $\dot{\mathrm{V}}$O₂ (r = 0.932, p < 0.0001), which fit the previously proposed formula for METs (METs = HRI × 6 − 5). However, VM was not correlated with VM (r = 0.354, p = 0.150). We can use HRI to measure the workload of Japanese forestry workers with a small burden in the field.

1. Introduction

Forestry is one of the most hazardous industrial sectors in most countries [1]. According to Japanese statistics, the incidence of absences from work lasting at least four days, including death, was 9.1 times higher in forestry than in all other industries combined in 2021 [2]. The physically and mentally demanding workload in forestry is related to workers’ fatigue, which is considered to be a contributing factor to accidents and injuries [3,4,5]. Japanese forests have different land characteristics from forests in other countries [6]; 37.2% of Japanese forest land has a slope >20°, which is the average among Asian countries, but higher than that in other regions [7]. Steep slopes hamper machinery operation and increase the risk of machine handling [1]. Workers must transfer forestry equipment and tools up and down forest land, in addition to engaging in high-intensity activities. It has been suggested that the physical properties of land establish unique operational systems based on experience and intuition [1,7,8]. Determining the workload of Japanese forestry workers is challenging, particularly when performing a broad range of forestry procedures. A descriptive study to measure the workload in the forestry industry in a temperate climate throughout the year is required to take measures to reduce fatigue and injury.

Heart rate (HR) monitoring is widely used to estimate the workload. Several studies have measured the workload of forestry workers using HR monitors in the field [9,10,11,12,13,14]. They reported a single task or operation, such as afforestation, pruning, felling, or cable work, and used the crude or HR relative to the maximal or submaximal HR. Relative HR is reliable against inter-individual variability, but is inconvenient for the detection of maximal or even submaximal HR in the field. A few reports on the workload of Japanese forestry workers (walking, mowing, or felling) were also found in the database [15,16,17], in which HR monitoring was conducted with a small sample size; oxygen consumption as an indicator of energy expenditure was not assessed.

HR and O₂ had a linear relationship, except at low levels of activity (below the flex HR; HR_flex) [18,19]. The heart rate index (HRI; actual HR/resting HR) is the most reliable predictor of energy expenditure, not only for healthy people, but also for patients with cardiovascular diseases [20]. The HRI method is applicable in real practice in the forestry field because only the resting HR must be known, whereas the maximal or submaximal HR is not required. To determine the latter, an exercise test using a treadmill or cycle ergometry is needed [21]. Although the HRI method is applicable to forestry work, validation with O₂ consumption has only been tested with workers wearing light clothing under laboratory conditions [20,22,23], and is impractical in the field. Other methods have been proposed to estimate energy expenditure in the field. Geographical information is a possible predictor of energy expenditure during simulated mountain climbing. Hagiwara et al. predicted the energy expenditure on a treadmill from the total weight, including carrying packs and the up–down vertical transfer velocity [24]. De Müllenheim et al. compared an accelerometer, heart rate monitor, and global positioning system (GPS) and concluded that GPS was the most accurate method for estimating the energy expenditure in outdoor uphill and downhill walking activities [25].

Before measuring the workload of the various tasks undertaken throughout the year by Japanese forestry operations, a simple, accurate, and applicable method for estimating energy expenditure in the field should be clarified. The aim of this pilot study was to investigate which device is most suitable for estimating the energy expenditure during forestry work on inclined ground, as is common in Japan. To achieve this, we selected a physically intensive task in cable operations as the target task, which requires Japanese forestry workers to bring heavy tools and equipment up and down slopes.

2. Materials and Methods

2.1. Participants and Fields

Six male adults, aged 37–63 years, were recruited for this pilot study. They were not forestry workers, but engaged in forestry research and management, and sometimes worked in the field. The protocol was approved by the ethics board of the Japan Organization of Occupational Health and Safety (approval number, 2022-07) in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants. Field measurements were conducted in an unsurfaced test field at the Forestry and Forest Products Research Institute in Tsukuba, Japan, from 17–20 October 2022. We used flat ground and 15°- and 30°-slopes up to a height of 8 m.

The measurement procedure began two hours after breakfast or lunch to minimize the thermal effect of food on the measurement. First, the participants prepared forestry work appurtenances such as work garments, boots, helmets, and gloves (Supplementary Figure S1) and wore portable metabolic analyzers, accelerometers, and heart rate monitors. When the measurements began, they rested on a chair in a large room for 30 min and then moved to the test field at the bottom of the slope. After sitting on a chair until their heart rate and $\dot{\mathrm{V}}$O₂ became steady, they shouldered a frame pack of 20 kg on their backs and started walking on flat ground. The weight-simulated packs included devices and tools for cable yarding. The participants could walk at any speed, but in the first 10 s, the researcher let them hear a metronome at a speed of 67/min, which taught them a speed of approximately 50 m/min. They turned at the pylon 185 m from the starting point, returned at the same speed as the starting point (a total of 370 m), unloaded the frame pack, and sat down on a chair. After resting until the monitored heart rate and $\dot{\mathrm{V}}$O₂ became steady within 5 min, the participants shouldered the frame pack and started climbing on a 15° slope. They reached their peaks, turned around, and descended. They sequentially repeated the up-and-down procedures three times, unloaded, and rested. They then moved to the bottom of the 30° slope and rested. Next, they climbed and descended the 30° slope three times shouldering the frame pack. After the heart rate and VO₂ stabilized, all devices were removed. The entire procedures took 1.5–2 h, which made it possible to accurately measure the air composition within the battery capacity and maintain the internal device temperature of the portable indirect calorimeter.

2.2. Measurements

2.2.1. Portable Indirect Calorimeter

The metabolic rate was measured using a portable indirect calorimeter, K5 (Cosmed, Albano Laziale, Italy). The K5 sensors were calibrated using a canister syringe for the flow volume and standard gas (16.09% O₂ and 5.01% CO₂) prior to each measurement. During the measurements, the participants wore K5 on their front chest with a mask attached to their face. Data were recorded breath-by-breath and monitored in real-time via Bluetooth in Omnia (Version 2.1, Cosmed, Albano Laziale, Italy) installed on a laptop computer. A K5 with attachments weighed 0.9 kg, which was included in the weight of the frame pack of 20 kg. Oxygen consumption standardized as “standard temperature and pressure, dry (STPD)” was recorded via the device.

2.2.2. Heart Rate Monitor

Heart rate was monitored using Verity Sense (Polar Electro Oy, Kempele, Finland), which weighed 17 g. The device was fully charged, and the inner time was adjusted to the Standard Time daily. The participants attached the device to their non-dominant forearm with an elastic band. Heart rate data for every second (expressed as beats/min) were recorded in the device, which was downloaded to the computer using Polar FlowSync (Version 3.0.0.1337, Polar Electro Oy, Kempele, Finland) on each measurement day. The real-time heart rate was monitored using Bluetooth and K5.

2.2.3. Accelerometer

The participants attached 20 g of the wGT3X-BT with three-axis accelerometric sensors (ActiGraph, Pensacola, FL, USA) to their right waist with an elastic band. The sampling rate was 30 Hz, and we used 1 s epoch counts of acceleration to measure the physical activity. The device was completely charged and adjusted daily. Memorized data were downloaded after each measurement day using ActiLife (Version 6.13.4, ActiGraph, Penacola, FL, USA). The three-axis counts were summarized as vector magnitudes (VM; counts/s).

$$\mathrm{VM}=\sqrt{c_x{}^2+c_y{}^2+c_z{}^2}$$

(1)

Acceleration counts are used to predict METs and energy expenditure per body mass, and we used a unit “counts” in this study.

2.2.4. Global Navigation Satellite System Device and Video Camera

A global navigation satellite system (GNSS) device, Trimble TDC150 (Nikon-Trimble Co., Ltd., Tokyo, Japan), was used to trace the geographical movements of the participants. Latitude, longitude, and height were recorded in seconds and downloaded from a CSV file. This device, weighing 0.85 kg, was fixed on a frame pack, which weighed 20 kg in total. Participants’ behaviors were recorded in the video camera memory to define the time points of events, such as starts, turns, and ends of walking.

2.2.5. Body Height and Weight

Before wearing the devices, body height was measured without boots to 0.1 cm using a stadiometer. Weight was measured with fieldwork clothing and boots (0.1 kg using an electronic scale, AD6201 (A&D Co., Ltd., Tokyo, Japan)). Body weight was calculated by subtracting 3 kg from the measured weight.

2.3. Statsitical Analysis

The intensity of the activity was compared between measurements. For standard activity intensity, the metabolic equivalents of tasks (METs) were calculated from oxygen consumption as follows:

$$\frac{O_2 \left(\mathrm{mL}·{\min }^{-1}\right) }{body weight \left(\mathrm{kg}\right) 3.5 \left(\mathrm{mL}·{\mathrm{kg}}^{-1}·{\min }^{-1}\right)}$$

(2)

The heart rate index (HRI) was calculated as the mean heart rate during walking divided by the resting heart rate (HR_rest), which was defined as the lowest heart rate while the preceding rest was on a chair. Heart rate and oxygen consumption did not reach a steady state during short walking on slopes; they increased after the start and gradually decreased during the following resting period. First, we calculated the cumulative amounts of oxygen consumption and HR exclusively during each of the three walking types because the resting heart rates after walking were not consistent for each participant. Next, we calculated the cumulative heart rate index and oxygen consumption as those during walking plus those in the recovery period after walking, minus the resting amounts. For example,

$${\displaystyle \sum }_{\mathit{sec}}^{\mathit{walk}}\dot{V}{O}_{walk}\left(\mathrm{mL}/\sec \right)+{\displaystyle \sum }_{\mathit{sec}}^{\mathit{rest}}\dot{V}{O}_{rest}\left(\mathrm{mL}/\sec \right)-\dot{V}{\mathrm{O}}_{\mathit{rest}}\left(\mathrm{mL}/\sec \right)\times \mathit{Rest}\left(\sec \right).$$

(3)

$${\displaystyle \sum }_{sec}^{walk}HR{I}_{walk}+{\displaystyle \sum }_{sec}^{rest}HR{I}_{rest}-HR{I}_{rest}\times Rest\left(\sec \right).$$

(4)

The walking periods of the accelerometer data were clearly demarcated, differing only by one or two seconds from the video camera recordings, and the VM during rest was zero. The cumulative vector magnitudes divided by each walking period (counts/min) were compared with the METs calculated from $\dot{\mathrm{V}}$O₂.

Each participant received three data points for each activity intensity variable. HRI and VM per minute were examined using Pearson’s correlation coefficients, with oxygen consumption as the MET. The Deming regression method was also used for comparison with METs because of possible measurement errors in the two variables [26]. METs estimated from HRI using a previously proposed formula [20] were compared with METs calculated from $\dot{\mathrm{V}}$O₂ (Equation (5)) using a Bland–Altman plot [27]:

$$\mathrm{METs}=\mathrm{HRI}\times 6-5$$

(5)

The GNSS data from the device were compared with the land formation in the test field. The data were expected to obtain work in physics from the vertical transfer and body and appurtenant weights (frame pack, clothing, shoes, and other equipment). However, the vertical movement was not accurately detected; therefore, we did not analyze the GNSS data further. Graphical and statistical analyses were conducted using R (Version 4.2.1, R Core Team, Vienna, Austria) and RStudio (Version 2022.12.0, Posit Software, Boston, MA, USA). The significance level was set at p = 0.05.

3. Results

The participant characteristics are presented in Supplementary Table S1. On the first day, there was a small amount of precipitation, and the ground was a little dirty until the second day, when three out of six participants were measured. Three participants were measured in the morning and the other three in the afternoon. The ambient temperature during the measurement was 22–25 °C, and the relative humidity was 53%–72%. We missed the GNSS recordings for one participant and video recordings for another participant.

The oxygen consumption (VO₂), heart rate (HR), vector magnitude (VM) of the accelerometer, and vertical (height) movement from the GNSS data were visually compared at the same timescale (Figure 1 and Supplementary Figure S2). The start and end of each activity and rest coincided with the inflection points of $\dot{\mathrm{V}}$O₂ and HR. $\dot{\mathrm{V}}$O₂ and HR during activity increased in the order of flat, 15° incline, and 30° incline, but VM was not significantly different among the three walking types. Additionally, VM was stable at zero during rest. The height movement clearly depicted up and down transfers, but the baseline was unstable, as observed in all participants. In the following analysis, we used $\dot{\mathrm{V}}$O₂, HR, and VM, excluding the height from the GNSS device.

The activity intensity as METs calculated from $\dot{\mathrm{V}}$O₂ was plotted against HRI and VM (Figure 2). HRI was significantly correlated with METs (r = 0.932, p < 0.001), and regression coefficients were 6.38 (95% confidence limits, 4.71, 8.06) for slope and −5.49 (−7.97, −3.02) for intercept. In contrast, VM did not correlate with METs (r = 0.354, p = 0.150). The coefficients of its regression model were not significant (0.0012 (−0.0001, 0.0026) for slope, and 0.402 (−4.258, 5.062) for intercept). When involving lag time for $\dot{\mathrm{V}}$O₂ and HR, HRI and VM were significantly correlated with METs, but were lower than those measured exclusively during activity (r = 0.843, p < 0.001; r = 0.506, p = 0.032, respectively).

The activity intensities as METs calculated from the Equation (5) and from $\dot{\mathrm{V}}$O₂ were plotted using the Bland–Altman method (Figure 3). The difference between the two methods was 0.134 METs with 95% confidence limits, −0.543, and 0.811 METs.

4. Discussion

In this pilot study, the HRI method was found to be the best for estimating the energy expenditure rate of forestry workers. We simulated the movement of forestry workers carrying 20 kg on their backs on flat and inclined ground. The activity intensity in METs estimated from HRI using the equation previously proposed by Wicks et al. [20] was significantly correlated with that from the measured oxygen consumption. In addition, most of the estimates were within ±1.0 METs of measured intensity on the Bland–Altman plot.

Heart rate is used to estimate energy expenditure because of its well-known linear relationship, but at a low level of activity, a linear association between heart rate and oxygen uptake O₂ breaks down. The heart rate above which the heart rate is a good predictor of energy expenditure and below which the resting metabolic rate is used as energy expenditure is called the flex heart rate, HR_flex. A study of 167 adults proposed that heart rate monitoring is a useful method for assessing energy expenditure in epidemiological studies [28]. Although most participants (70%) from five experimental studies revealed that the total energy expenditure estimated using the flex heart rate method was within ±10% of the standards [29], the flex heart rate and a linear relationship above the flex heart rate are vulnerable to between-individual variation [19,30].

Wicks et al. evaluated the relationship between heart rate and oxygen consumption in exercise training using three variables: absolute HR (HR_absolute), HR_absolute minus HR_rest (HR_net), and the ratio to resting HR (HRI) [20]. They selected 60 articles involving healthy people and cardiovascular patients whose ages ranged from 27 to 85 years. Among the regression models with three variables, the model with HR_absolute explained 76.9% of the variation; however, it was the lowest among the three methods. The model using the HRI had the highest variance explanation R² value of 95.2%. The HRI method was validated using treadmill tests for professional rugby and soccer football players [22,23]. HRI is a possible indicator of the intensity of physical activity patterns. The results of this study revealed a high correlation between HRI and oxygen consumption during the simulated walk of forestry workers. Unlike previous reports, in which the workload was applied using a treadmill or cycle ergometry in a well-conditioned room [20,22,23], this study simulated forestry work. The HRI method must be applicable to real forestry fields under various conditions.

Factors influencing heart rate should be considered, such as environmental and genetic factors, and medication. The HR_flex method was reviewed by Leonard, who evaluated it under several conditions, including age, sex, ethnicity, season, and lifestyle [31]. Medical condition, particularly using β-blockade, may create problems in field assessment. In a previous report, however, pathophysiological condition and using β-blockade did not affect the relationship between the HRI and the oxygen consumption [20].

At forestry work sites, individual resting HR may be difficult to obtain. Japanese forestry workers usually move between their home and the site and take breaks outdoors, but not in field offices. The HR_flex method uses the highest HR for lying or sitting and the lowest HR for activity, and requires individual calibration [18,19]. In contrast, the lowest HR at rest, used to calculate the HRI, may be relatively easy to obtain before work or on breaks. The time course of the HR suggested a shorter resting period before measurement. Calculating HRI, unlike the relative HR, does not require a submaximal heart rate, which may be influenced by age, sex, and fitness.

An accelerometer-based approach for assessing physical motion cannot correctly discriminate activity types in terms of the magnitude proportional to energy expenditure [32,33]. Several trials have been conducted to estimate the energy expenditure from accelerometer data [34]; however, these have not yet been used in the field. Energy expenditure in ascending and descending inclines is not mirror symmetry [24]; oxygen consumption on a −8% descending incline is the least and increases on a more descending incline. Global positioning systems [35], barometric sensors [36], and accelerometry are expected to accurately estimate the energy expenditure. However, in this study, the GNSS receiver could not accurately measure vertical movements of 8 m up and down. This device had no barometric sensors, but the portable indirect calorimeter K5 had them. However, even the barometric altitude at K5 was unstable during each measurement.

A heart rate monitor is a light device that may place a slight burden on the participants. The device used in this study is worn with an arm strap attached to the forearm or upper arm. It may be more applicable in forestry fields than chest-strapped ones because there is no need to take off clothes, and it may not disturb the use of hand tools as much as a wrist-worn monitor. Regarding where the monitor is worn, the accuracy of various forestry tasks is warranted in the future [37]. Some conditions that influence the HR should be considered. First, although it can estimate high workloads in forestry activities, HR during periods of low workload and recess may be influenced by non-work-related factors when considering daily work. We need to determine the energy expenditure specific to hard forestry work. Knowledge of the workload regarding energy expenditure can help us properly allocate jobs to workers, deploy workers for their fitness, and prevent heat illness by predicting thermogenesis from the workload. Second, terrain surface conditions influence the energy cost [38], but this factor may not influence the relationship between HR and oxygen consumption. Third, in some studies, such as felling, pruning, and mowing, upper limb activity is considered the main cause of energy expenditure. In contrast to motion sensors and GNSS devices, HR monitoring, as an estimator of energy expenditure, has advantages for forestry workers.

This study has some limitations regarding the interpretation and adoption of the results in the field. First, the sample consisted of only male participants and the sample size was small. However, male workers were dominant in Japanese forestry, and the age of the participants varied widely. The results of the association between heart rate and metabolic rate were in accordance with a previous formula [20]. Several studies have reported small sample sizes in actual forestry fields (5–13 participants) [10,11,12,13]. A low risk of a statistical type 2 error due to small sample size, which indicates a small measurement error in correlation, may be a strength for implementing the main study. Second, the test field was not the same as the forestry field. Forestry activities are a combination of complicated motions, rather than walking. We did not have information on how and when heart rate monitoring was interrupted during forestry work activities due to loss of contact with the skin. For one participant, the heart rate decreased to zero when the weights were pulled down, and the elastic band of the monitor was soon tightened.

5. Conclusions

The HRI method was the best estimator of METs, similar to the energy expenditure rate in this pilot study. Accelerometry was insufficient to discriminate between activity types, and estimated energy expenditure less efficiently than the HRI method. The GNSS device did not accurately trace the vertical (height) movements; therefore, the work in physics could not be calculated. The HRI method uses a light device to monitor HR, which imposes only a small burden on forestry workers in the field. The arm-worn HR monitor used in this study is more convenient for forestry workers in the field because they do not have to wear a chest strap under their shirt. Simultaneous and repeated measurements using a low-cost device by a team of forestry workers will provide valuable information about various forestry tasks throughout the year. This study bridges the gap between laboratory settings and HRI. The HRI method can be used to estimate the workload in forestry work and can present the workload of each forestry task and activity as a physical activity level (MET) in future studies in the field. MET as a common unit will be practical in the prevention of fatigue and injury, healthy behavior related to energy and nutrition intake, and heat-related illness.