Measuring symptomatic relief in men with lower urinary tract symptomsBMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g6664 (Published 07 November 2014) Cite this as: BMJ 2014;349:g6664
All rapid responses
The critical evaluation provided by Dr Blanker and colleagues of trials evaluating the benefits and side-effects of medication (Nov 15) generates justifiable uncertainty about current guidelines and current practice. They question whether using medications for lower urinary tract symptoms can be considered evidence based - despite significant p values (after "between group" comparisons) that support improvement on medication during controlled trials. They call for further large randomised controlled trials (RCTs) and subsequent meta-analyses, but using a higher "cut-off" symptom score. They argue "improvement" that is not clinically relevant should not be recorded as a "success" as this is the problem when trivial improvements in score are used - enabling group data to establish "benefit" when contrasting medications with placebo. There is a similar need to analyse for clinically relevant improvement scores in trials comparing different pharmaceuticals or when contrasting medications with other management methods.
The resources required to respond to this suggestion will be difficult to mobilise, especially for those pharmaceuticals already in widespread use after becoming available on the basis of the trials using clinically irrelevant "cut-off" values for symptom scores - but not demonstrating sufficient side-effects or negative outcomes to make use of the medication studied inappropriate.
An alternative approach to generate an evidence base for treatment of this condition (and many others), but only having excluded serious complications of treatment in well planned RCTs, is to analyse past data and plan future studies with a much greater focus on the statistical significance of improvement in individual patients (Knapp, 2001).
If patients, when participants in a double-blinded crossover study or treated in clinical practice, keep symptoms scores on a daily, or more frequent, basis, the data collected becomes a "time-series”. This time series from a patient (n=1) can be subjected to a range of analyses that allocate a statistical probability of that patient having observed a significant improvement in symptoms with change to a pre-determined "cut-off" value.
Analysis of time-series of data from individuals should be supported by data on which regimen the patients would wish to continue (eg drug or placebo or device) and with a similar analysis of side-effect scores. The Bayesian methods of analysing sequential time-series data from individual patients after kidney transplantation should be considered, as one statistical method, as well as others, including some that are in common use in engineering and finance. Many methods used when regular arrival of data is consistent have problems with the "missing data" common in clinical trials and practice but strategies to compensate for missing data points were incorporated in the methods developed in Nottingham for medical time series analysis (Knapp, Smith, Trimble, Pownall, and Gordon, 1983).
The focus on obtaining more information from each patient in a study reduces the problem of heterogeneity for patients with lower urinary tract symptoms and may, in retrospect, enable clinical features that identify "responders" and "non-responders" to be established - as these may not be identifiable by current diagnostic classifications.
1. Knapp, MS, Using clinical evidence. Randomised controlled trials are not the only evidence. BMJ 2001; 323: 165
2. Knapp, MS, Smith, AFM, Trimble, IM, Pownall, R, Gordon, K, 1983. Mathematical and statistical aids to evaluate data from renal patients. Kidney International 1983; 24: 474-486
Competing interests: Medical Advisor (with a financial interest) in Malem Medstat providing advice and equipment for enuresis and other disturbances of bladder emptying.
For what it is worth?
I am well in my sixties and for the last few years 'suffered' from having to get up once or twice in the night to empty my bladder. Tamsulosin didn't help much and made me dizzy so I excepted my age related 'problem'. This year another age related problem (OA) caused me to have knee pain at night. In the last few months I occasionally took 2 paracetamol and 1 ibuprofen ( 200mg) before going to bed. This had the strange effect that it gave me a good night 's sleep because I did not have to get up at all (or at the most early morning) to empty my bladder?! I have now taken this combination for the last 3 weeks and the effect is continuing. I leave it to others to find an explanation. (The painkillers reduce the sensation of having a 'full' blader??)
As I guess that there are many collaegues out there in a similar situation I suggest 'an on-line trial' might prove or disprove this finding?
Competing interests: No competing interests