This study uses a large department database of male uroflowmetry patients who had their investigations between 1988 and 2021. The unfiltered data comprises 29,076 patient data entries, with up to three uroflowmetry tests per patient. These tests were conducted in a urology outpatient setting, on men who were referred for lower urinary tract symptoms, including both storage and voiding complaints. Patients who opted out of their data being used for research were excluded from the database.
In the uroflowmetry clinic it is routine to perform between one and three flow tests per patient, and a post void residual scan after each void. Uroflowmetry uses a natural method of filling- men are advised to arrive with a comfortably full bladder to perform the first flow, and asked to continue drinking water for a second or third flow. If their first or second void felt representative of their usual void, the test can conclude without a third flow. The data points of interest (Qmax, voided volume and post void residual- PVR) were entered manually by staff at the time of the clinic.
As this data was collected over a 33-year period, equipment was changed. The flowmeters used during this data collection were Dantec, Mediwatch and MMS.
Data screening was conducted to remove any typographical errors. Patients with duplicate data entries were deleted. Patients with more than one appointment remain in the database. Each patient data point has up to 3 sets of uroflowmetry data.
Only one set of uroflowmetry data was used per patient. This was decided based on the largest voided volume <601ml out of the maximum 3 flows. The corresponding Qmax and PVR are also selected.
At this point, we removed any patient who fell outside the ranges: Qmax <71ml/s, voided volume <601ml, as these are the largest criteria which can be plotted on the Liverpool nomogram.
In calculating the relevant parameters, we noted that Siroky uses total volume (voided volume and PVR volume) but Liverpool uses voided volume only.
For each patient the Liverpool percentile, and the Siroky percentile were calculated using the Excel formulae below, which are derived from references (1) and (3).
Liverpool patient percentile= NORM.DIST(SQRT(Qmax), (2.37+0.18*SQRT(Voided volume)-0.014*Age), 0.727, 1)
The Siroky percentile is calculated using:
Siroky mean= 7.327+0.113*vol-0.000224*vol2+2.6*POWER(10,-7)*vol3-1.62*POWER(10,-10)*vol4
Siroky mean-1SD= 6.071+0.0863*vol-0.000192*vol2+2.58*POWER(10,-7)*vol3-1.699*POWER(10,-10)*vol4
The standard deviation for Siroky is the difference in the two above equations. The percentiles were then calculated using the Excel NORM.DIST function.
For aim 1, patients were grouped into increments of 100ml post void residual (PVR) and plotted on the Liverpool nomogram for visual representation. This allowed visual comparison of post void residual groups, and preliminary observations as to whether the spread of percentiles on the nomogram was changed for each group.
For aim 2 first a binary agreement was investigated. If a patient had a percentile <25% on the Liverpool or <-2SD on the Siroky nomograms, they would receive a “1” in the corresponding nomogram column. If a patient had a percentile >25% on the Liverpool or >-2SD on the Siroky nomograms, they would receive a “0” in the corresponding nomogram column.
This was followed by a Cohen’s Kappa agreement statistical test. This investigates the agreement of criteria (Liverpool <25%, Siroky <-2SD) for each patient, as to whether both agree, both disagree, or they disagree.
These two calculations allowed simple investigation in the agreement of patients being classified as Liverpool <25% and Siroky <-2SD. This was conducted on patients within the parameters of both nomograms.
A correlation study was then performed between the Liverpool percentiles and the Siroky percentiles.
Finally, level of agreement was investigated between Liverpool percentiles and the Siroky percentiles using a paired sample t-test.