The most popular method for collecting data on activity is the activity diary. Activity diaries are often used because they are relatively inexpensive to initially administer and are considered easily understood by the subject. However, diaries are at best a rudimentary form of estimating energy expenditure because the data is subjective, prone to recall bias, coding errors in collation and overall non-compliance. Even ‘benefits’ can be misleading: The low cost of creating diaries can be counteracted by time/ wage cost of the data-entry and collation once they are completed, and data gaps due to skipped or misunderstood questions can also be a significant issue. In short: diaries should only be considered for estimating energy expenditure when there is a low requirement for accuracy or an enormous sample group that significantly limits the cost-per-subject.

Pedometers (or step counters) have been commercially available for around 45 years; originally promoted as interventional aids to encourage activity there are a huge variety of units available. The purpose of a pedometer is to measure steps; in theory a pedometer records vertical acceleration and looks for a pattern to determine if it is a step. There is no scientific background for the 10,000 steps per day recommendation – this was simply a marketing slogan invented for the first pedometers in 1965. The technology can be broadly split in to two groups: the simplest use a spring-suspended lever arm that moves up and down opening and closing a circuit with each step. The second use an accelerometer which typically measures ‘zero crossings’.

Pedometers are typically used by those with large sample groups because they are the most inexpensive digital solution, however they are also considered inaccurate. One study comparing coiled spring-levered pedometers to accelerometers demonstrated that the pedometer detected 23.5% fewer steps (75.4% vs 98.9%, P < 0.05)13 another comparing actual steps to a spring levered pedometer’s estimation found 34% error, and measurements of children or those with slow walking speed, gait disorders, high BMI, or mobility issues may be less accurate still. Fundamentally pedometers only work well at certain walking speeds as walking either too quickly or too slowly can cause artefacts or missed steps. However even working at optimum efficiency “steps” are somewhat of a subjective unit of measurement as they vary greatly from person to person; although some can be individually calibrated, accuracy is complicated by any unit that converts “steps” to a quantifiable measurement unit (e.g.  kJoules).

Pedometers can be useful in specific circumstances with careful placement where heavy steps are taken in a regular pattern at specific times (such as on a treadmill at a regulated speed) or as an interventional aid to encourage activity. However their inaccuracy in free living conditions, where the standardisation of laboratory conditions is difficult or impossible, limits the use to either very large studies or to those that just need a very general impression of daily activity.

Many instruments designed to calculate energy expenditure use accelerometers as the primary device for collecting data. These devices measure acceleration by converting the energy of movement into an electrical signal. Actigraphs then interpret this data using algorithms to calculate the intensity (and sometimes the type) of exercise taken. Such devices will also include low pass filters to attempt to remove “background noise” and only record actual physical movement.

Accelerometers have been growing in popularity in physical activity assessment because they are small and light; as microchips have reduced in both size and price and batteries can last longer, accelerometer-based systems can be relatively inexpensive and much more accurate than traditional pedometers.

Wrist-worn actigraphy can provide useful insight into sleep analysis where one is recording overall movement to establish if the subject is asleep; however wrist worn actigraphs are not considered accurate enough to record detailed energy expenditure in free living conditions. Although the hip (or lower back) is seen as being the best location for accelerometer placement,  there are doubts over its accuracy at estimating free living EE. Central to this issue is the difficulty of using accelerometers alone to detect activity; exercise that is particularly demanding on the upper body (e.g. rowing, boxing or swimming) or strength training and activity that is demanding for the subject but not seen as “vigorous” by the device will not be detected by an accelerometer alone.

The choice of device is ultimately a matter of convenience vs accuracy; use of accelerometers, particularly on the wrist, is simple to administer and less obtrusive than other methods at the expense of overall accuracy. Our own MotionWatch device is a convenient method for monitoring physical activity, but for the reasons stated above we provide an indication of MVPA rather than an measure of absolute Energy Expenditure.

A number of systems rely on heart rate monitoring to estimate energy expenditure. These devices usually monitor Heart Rate Variability (HRV) and use heart rate to estimate EE. Heart rate monitors obviously have none of the issues that accelerometers have with placement; as long as the skin is adequately prepared and the unit is placed close enough to the heart to get a good signal the system will work. More recently there has been a move towards optical plethysmography on the wrist which introduces further issues with movement artefacts. Using heart rate alone tends to overestimate energy expenditure. One explanation for this is that heart rate alone is not necessarily an indication of exercise taken; for example stress, excitement and stimulants are all known to increase heart rate without AEE taking place. The over estimation of EE by heart rate monitors and underestimation by accelerometers has led some to try using both systems to gain a more accurate calculation. This approach uses the benefits of both methods and allows one to counteract the shortcomings of the other. However, in practice synchronising the two data sources accurately can be difficult. These issues led CamNtech Ltd. to develop the Actiheart in 2003.

Click or tap below to view peer-reviewed scientific research publications using CamNtech products:

Energy Expenditure

Physical Activity