Article type
Year
Abstract
Background: bladder pain syndrome (BPS) is a poorly understood clinical condition with patients presenting with varying symptoms. The management of BPS is challenging for both patients and practitioners. At present, there is not a universally accepted diagnosis and diverse causative factors have been proposed. This diversity is reflected in wide-ranging treatment options, used either alone or in combination, with limited evidence. A network meta-analysis (NMA) comparing multiple treatments, even though challenging, may help to determine the best treatment options for BPS.
Objectives: to describe methodological issues related to conducting a NMA for the treatment of BPS.
Methods: we performed a NMA of randomized controlled trials (RCTs) for the treatment of BPS in adults. We identified relevant RCTs from existing Cochrane Reviews and literature searches based on the Cochrane Incontinence Specialised Register (including MEDLINE, CENTRAL, CT.gov, WHO ICTRP: searched April 2019). Primary outcomes were patient-reported cure or improvement of symptoms, pain, urinary frequency and nocturia, all assessed at 12 months. We assessed risk of bias (RoB) of included studies using the Cochrane 'Risk of bias' tool. For each outcome we fitted random-effects NMA models using WinBUGS 1.4. We monitored odds ratios for binary outcomes and mean differences for continuous outcomes with 95% credible intervals (Crls). We compared the results of the NMA with the direct evidence from head-to-head trials.
Results: we included 81 RCTs with a median of 38 participants (range 10 to 369). There were 65 different active treatments with some comparisons only informed by direct evidence from one trial. To simplify, we grouped treatments into 31 treatment categories by mode of action. The NMA approach allowed the evidence for a large number of treatments to be estimated simultaneously. We faced, however, some methodological challenges. There were a large number of trials but most had small sample size. Few studies compared treatments other than control, resulting in an almost star-shaped network for some outcomes. Definitions of treatments varied and some studies included a combination of more than one treatment.
Conclusions: we successfully conducted the NMA but limited numbers of small trials for each treatment category hampered our ability to fully exploit the advantages of the NMA. This was compounded by the fact that most treatments were compared with control so there were often few closed loops within the network. Component NMA is a possible future approach when treatment arms comprise multiple treatments. Larger, more focused trials are needed to improve the current evidence base.
Patient or healthcare consumer involvement: we did not directly involve patients in the study but we contacted patient associations to refine the list of outcomes at the protocol stage.
Objectives: to describe methodological issues related to conducting a NMA for the treatment of BPS.
Methods: we performed a NMA of randomized controlled trials (RCTs) for the treatment of BPS in adults. We identified relevant RCTs from existing Cochrane Reviews and literature searches based on the Cochrane Incontinence Specialised Register (including MEDLINE, CENTRAL, CT.gov, WHO ICTRP: searched April 2019). Primary outcomes were patient-reported cure or improvement of symptoms, pain, urinary frequency and nocturia, all assessed at 12 months. We assessed risk of bias (RoB) of included studies using the Cochrane 'Risk of bias' tool. For each outcome we fitted random-effects NMA models using WinBUGS 1.4. We monitored odds ratios for binary outcomes and mean differences for continuous outcomes with 95% credible intervals (Crls). We compared the results of the NMA with the direct evidence from head-to-head trials.
Results: we included 81 RCTs with a median of 38 participants (range 10 to 369). There were 65 different active treatments with some comparisons only informed by direct evidence from one trial. To simplify, we grouped treatments into 31 treatment categories by mode of action. The NMA approach allowed the evidence for a large number of treatments to be estimated simultaneously. We faced, however, some methodological challenges. There were a large number of trials but most had small sample size. Few studies compared treatments other than control, resulting in an almost star-shaped network for some outcomes. Definitions of treatments varied and some studies included a combination of more than one treatment.
Conclusions: we successfully conducted the NMA but limited numbers of small trials for each treatment category hampered our ability to fully exploit the advantages of the NMA. This was compounded by the fact that most treatments were compared with control so there were often few closed loops within the network. Component NMA is a possible future approach when treatment arms comprise multiple treatments. Larger, more focused trials are needed to improve the current evidence base.
Patient or healthcare consumer involvement: we did not directly involve patients in the study but we contacted patient associations to refine the list of outcomes at the protocol stage.
PDF