Hypothesis / aims of study
Flow meters play an important role in assessing patients with functional urology symptoms. They allow clinicians to perform flow studies and urodynamics, enabling them to offer the most appropriate treatment options. It is therefore essential that new flowmeters have been thoroughly tested before they are used to ensure they are accurate and usable.
Our centre has been commissioned to test a new design of flowmeter. We developed a protocol to test this flowmeter to assess its accuracy, usability, filtering and operation. We describe this protocol in order to enable standardise testing by prospective users of any new flowmeter in the future.
Study design, materials and methods
The flowmeter’s accuracy was tested using a constant flow bottle which produces a flow rate of 14ml/s. Artefactual flow was simulated by moving the bottle’s flow randomly across the flowmeter. A water column with an outlet at the base was used to test the accuracy of the flowmeter. Artefactual flow was simulated by temporarily blocking the flow from the column for two seconds then allowing the flow to restart. Volumes passed were checked using a calibrated weight scale, while maximum flow rate was checked by manually screening the raw flow data. Since the water column generates a smooth decline in flow rate during its steady state, by Torricelli’s theorem, the output was used to check the linearity of the flowmeter. The impulse response of the flowmeter was tested by stopping and starting the flow, enabling calculation of the effective time constant of the flowmeter’s signal filter. The response of the flowmeter to common artefacts (knocking the flow meter, wagging of the flow and removing the jug from the flow meter) was also tested. These tests are summarised in Table 1.
Usability and operation assessment of the flow meter was carried out in a normal flow clinic. Feedback was received via a questionnaire, adapted from a national equipment assessment report, from experienced departmental staff members. Areas assessed are ease of learning, performing a test, data entry, reporting, analysis and data management.
The tests showed that for the new flowmeter, the values for maximum flow rate, Qmax, and voided volume, Vvoid, were all within stated accuracy and ICS recommendations. The processing of data by the proprietary algorithm is shown in Fig 1. ICS guidelines suggest either a 2 second window filter  or a 1 Hz low pass filter . Comparison of the data in Fig 1 from a 2 second window filter, the new flowmeter’s algorithm filter and both filters combined shows that noise is more effectively removed by additional filtering than simply a 2 second window. The flow column test showed some deviations from a smooth decline of flow, so further checks of the linearity test are required. The response of the flowmeter to step changes in flow showed that the signal can rise at a rate of 0.7 ml/s2, giving an effective 3dB bandwidth of the algorithm of 0.13 Hz, well below ICS recommendations.
Interpretation of results
The differences in the traces in Fig 1 imply that the filtering of a flow meter needs testing with artefacts to ensure that the ICS guideline of allowing “the more common artefacts to be represented and recognized”  can be met. Too much filtering may result in a smooth trace, but will remove features that are potentially useful to the clinician. The new system tested, using a new algorithm with a 2 second window, allows these features to be seen.
The results showed the effective 3dB bandwidth was well below ICS recommendations. However, uroflow features faster than this clearly make it through the algorithm, so assessment as a simple filter is inappropriate. Testing of flowmeters using different methods of signal processing therefore need to be by simulation of actual clinical flows, rather than by theoretical cut-off points.