Summary of main results

The purpose of the current review was to determine if certain catheterisation technique, strategies (including re‐use) or designs of catheter are better than others in terms of UTI, complications, user satisfaction, preference, ease of use and/or cost‐effectiveness for adults and/or children whose long‐term (with no predicted end‐point) bladder management is by urethral IC. There remains an absence of robust evidence to support any given technique, strategy or design over another with respect to control of clinical symptoms, particularly symptomatic UTI. None of the trials included an economic evaluation.

We are uncertain if aseptic technique compared with clean technique has any effect on the risk of symptomatic UTI because the wide 95% CI is consistent with possible benefit and possible harm (low‐certainty evidence; summary of findings Table 1). No other evidence was available comparing aseptic technique with clean technique for our other prespecified GRADE outcomes.

We are uncertain if single‐use catheter compared with multiple‐use catheter has any effect on the risk of symptomatic UTI because the wide 95% CI is consistent with possible benefit and possible harm (low‐certainty evidence; summary of findings Table 2). Due to the paucity of data, we could not assess the certainty of evidence relating to adverse events (one study reported zero events in both arms). No other evidence was available comparing single‐use catheter with multiple‐use (clean) catheter for our other prespecified GRADE outcomes.

We are uncertain if hydrophilic‐coated catheter compared with uncoated catheter has any effect on the risk of symptomatic UTI because the wide 95% CI is consistent with possible benefit and possible harm (low‐certainty evidence; summary of findings Table 3). There is moderate certainty of evidence that risk of urethral trauma and bleeding from uncoated catheters is similar or lower compared to hydrophilic‐coated catheters (summary of findings Table 3). We are uncertain if hydrophilic‐coated catheter compared with uncoated catheter has any effect on participant satisfaction (very low‐certainty evidence; summary of findings Table 3). Due to the paucity of data, we could not assess the certainty of evidence relating to participant preference (one cross‐over study reported greater preference for hydrophilic‐coated compared with uncoated catheter).

We are uncertain whether the length of the catheter has any effect on the outcomes of interest, including patient‐reported outcomes and complications.

Overall completeness and applicability of evidence

There is insufficient evidence supporting one catheter technique, strategy or design over another. Given the wide variety of catheter designs available and the various factors that affect use of intermittent catheterisation, not all aspects of catheter design or usability were addressed in the available studies.  Moreover, our ability to compare studies was limited by the wide range of participants, varying in age, sex, health conditions and settings, in the included studies. 

A key clinical question remains about the influence of catheter technique, strategy or design on incidence of symptomatic UTI. The difficulty of establishing robust outcome measures of UTI remains problematic. A positive urine culture is not clinically relevant unless accompanied by symptoms but the symptoms themselves may present in vague and imprecise ways, especially in adults with spinal cord injury (IDSA 2009). However, symptomatic UTI remains the most clinically important outcome variable and was the primary focus of this review. 

A further complication are the differing definitions of symptomatic UTI, which can potentially lead to inconsistencies between trials. As UTI definitions are based on consensus statements (IDSA 2009; NIDRR 1992; see full definitions in the Types of outcome measures), Cochrane leaves the interpretation to the review authors. Differences in these UTI definitions posed a potential issue with data interpretation ‐ for example, cloudy/odorous urine and pyuria were included as symptoms in NIDRR 1992 but were excluded in IDSA 2009. In this Cochrane Review, trialists’ definitions ranged from self‐reported symptoms and the need for antibiotics, to the more specific descriptions by NIDRR 1992 and IDSA 2009.

To test whether heterogeneous or homogenous definitions made a difference, we planned to analyse the data in two ways according to the definitions presented in the trial (IDSA 2009; NIDRR 1992). However, sufficient data were not available. It must be noted that the narrower the definition, the fewer the trials that can be included. We remain uncertain if there is any difference in incidence of UTI between catheter technique, strategy or design.

We did not identify any economic evaluations conducted alongside any of the included trials, so we cannot draw any conclusions relating to the prespecified economic outcomes. Evidence from model‐based economic evaluations exists but we did not include this in the review (Bermingham 2013; Clark 2016; Hakansson 2016; Rognoni 2017a; Truzzi 2018; Watanabe 2017; Welk 2018). 

Quality of the evidence

There remains an absence of robust evidence to support any given catheterisation technique, strategy or catheter design over another with respect to control of clinical symptoms, particularly symptomatic UTI. Generally, the risk of bias in terms of randomisation, allocation concealment and blinding of outcome assessment was unclear due to insufficient reporting. Lack of blinding of participants and personnel in a substantial number of the trials was judged to introduce a high risk of bias but it is recognised that this is unavoidable in some circumstances. Around a third of the trials were at high risk of bias due to incomplete outcome data. The certainty of the evidence was low to very low. We downgraded for risk of bias and for imprecision due to low numbers of participants in the trials.

Assessment of user‐reported outcomes: A total of 10 trials had user‐reported outcomes, nine of which used questions that had not undergone standard psychometric testing and validation. Chartier‐Kastler 2013 was the first trial to apply a newly developed and validated 24‐item Intermittent Self‐Catheterisation Questionnaire (ISC‐Q), which evaluates aspects of quality of life specific to the needs of individuals performing ISC (Pinder 2012). The tool has four domains (ease of use, discreetness, convenience and psychological well‐being) and a total score, although Chartier‐Kastler 2013 only reported the total score. In this review, the most frequently reported outcome measures related to ease of use, with nine studies reporting ease of insertion and seven reporting ease of handling. Fewer studies reported outcomes relating to discreetness and convenience. Future studies would benefit from adopting a more consistent approach to the measurement of user‐reported outcomes in assessing the benefit of one intermittent catheterisation product over another.

Reporting standards: standards varied and not all trials followed the Consort guidelines, making it difficult to extract data. In those that followed good reporting standards, adverse events such as haematuria were clearly attributed to one of the trial arms.

Potential biases in the review process

Notwithstanding the comprehensive literature searches, it is possible there is unpublished evidence pertinent to our review question that we have not identified. Furthermore, suboptimal reporting of trial methods limits our ability to make meaningful comparisons using the relevant data. In our analysis of cross‐over studies, we acknowledge that by including data from only the first treatment period there is a potential source of bias in that we have discarded some outcome data that could theoretically contribute to answering our research question. However, we judged this approach to be appropriate in the context of the cross‐over studies that we identified, which did not include a washout period and in which carry‐over could be a problem.

Cross‐over trials: Cross‐over trials are attractive to researchers as they can reduce confounding covariates and have the participant act as their own control. One important point to note is that cross‐over studies can be much smaller than parallel‐group RCTs whilst still being adequately powered. For example, if a cross‐over trial is designed to detect a 20% reduction from three UTIs per year, and assuming within‐participant variance was no more than half that of the between‐participant variance (which is typical), then a trial of only 20 per arm would have 80% power. However, the methods for analysing cross‐over studies for meta‐analysis are complex. In the 11 cross‐over trials reviewed, there seems to have been two approaches considered.

The first method, where all intervention data is compared with all control data, fails to take into account the paired nature of cross‐over trial data and, as such, results lack precision. The second method, which just uses data from the first period only, is now recognised to be biased as first‐period data is more commonly available when there is significant carryover and this underestimates the true treatment effect. Although these methods are suboptimal, they are, however, conservative approaches: both are more likely to yield Type II rather than Type I errors. The Cochrane Handbook recommends a third method, i.e. to use all the data and also take the paired nature of the data into account. This avoids the imprecision and bias which can occur when other methods are used, and we attempted to use this approach when possible. The problem with this method, however, is that it is dependent on sufficient results (e.g. a paired t‐test) being published in the included trials in order to estimate the within‐patient differences. It may be that lack of reporting (or the inability to contact authors) might mean that one of the other methods needs to be employed, but we suggest that at least an attempt at approximating a paired analysis should be undertaken.

Another issue with cross‐over trials is the need for a washout period. The analysis always assumes that there is no carryover effect, and studies should be designed to include a washout period of a suitable duration if carryover is possible, otherwise the treatment effect can be underestimated. We considered that a UTI which begins in the first period might not be detected until the second period. If this is plausible, then a washout period would be required in a well‐designed cross‐over study where UTI is an outcome. A washout period could therefore be justifiable as one of the inclusion criteria for analysis of UTI outcomes. It might be that the first two weeks of data in both arms is not included in the analysis.

Agreements and disagreements with other studies or reviews

Four systematic reviews with meta‐analysis have been published on the occurrence of UTI in IC users (Bermingham 2013; Health Quality Ontario 2019; Li 2013; Rognoni 2017b). 

Bermingham 2013 reviewed trials to determine the most clinically effective and cost‐effective approach for patients performing self IC with either hydrophilic‐coated, single‐use uncoated, gel reservoir or clean uncoated catheters. Data were obtained from eight trials (Cardenas 2009; De Ridder 2005; Duffy 1995; Giannantoni 2001; King 1992; Pachler 1999; Sutherland 1996; Vapnek 2003). Data from two trials indicate IC users were significantly less likely to report one or more UTIs compared with sterile non‐coated catheters (De Ridder 2005; Giannantoni 2001). Their conclusions that were related to gel reservoir catheters were based on one cross‐over study (Giannantoni 2001). In our review, this study was ineligible as we classified gel reservoir catheters as uncoated. They found no differences in mean monthly or total annual UTI between hydrophilic‐coated and single‐use uncoated and little difference between clean versus sterile IC (P = 0.06). Overall, multiple use was reported to be the most cost‐effective type of IC. Of note is their point that there are limitations and gaps in the evidence base and non‐coated PVC catheters are designated as single‐use devices in the UK. Therefore, they recommend a precautionary principle should be adopted and that patients should be offered a choice between hydrophilic and gel reservoir catheters. We agree with Bermingham 2013 that there is inadequate evidence to state that incidence of symptomatic UTI is affected by any one catheter design.

Health Quality Ontario 2019 reviewed 14 RCTs: single‐use vs multi‐use (Chick 2013; Duffy 1995; Kiddoo 2015; Pachler 1999; Prieto‐Fingerhut 1997; Vapnek 2003); hydrophilic single‐use versus non‐coated single‐use (Cardenas 2009; Cardenas 2011; DeFoor 2017; De Ridder 2005; Sarica 2010) or gel reservoir single‐use versus non‐coated single‐use (Giannantoni 2001; Quigley 1993; Sarica 2010). In line with our review, they concluded that given the overall low certainty of evidence in available studies, there is uncertainty on whether any specific type of IC (coated or non‐coated, single‐ or multiple‐use) significantly reduces symptomatic UTI, haematuria, or other serious adverse clinical events, or whether a specific type improves patient satisfaction. They additionally concluded that, in the absence of any certainty, the lowest‐cost catheter is likely the most cost‐effective.

Rognoni 2017b conducted a systematic review of seven papers exploring the same questions as Bermingham 2013 and reviewed essentially the same trials regarding coated versus uncoated and uncoated, re‐used or single‐use catheters (Cardenas 2009; Cardenas 2011; De Ridder 2005; Pachler 1999; Sarica 2010; Sutherland 1996). UTI and haematuria were the primary outcomes. However, the authors concluded that hydrophilic‐coated catheters are associated with a reduced risk of UTI among patients using IC. Rognoni 2017b’s findings concur with our own that hydrophilic catheters may affect the incidence of UTI but that the certainty of the evidence is weak and that further studies are crucial to provide more direct evidence of the comparisons. Similar to our review, the authors noted that the ability to draw meaningful conclusions is compromised by methodological limitations, including heterogeneity of outcomes and definitions, absence of high‐quality trials and a higher dropout rate in the arms related to hydrophilic catheters.

Li 2013 also sought to examine the benefit of one catheter design (hydrophilic versus non‐hydrophilic) in the occurrence of UTI and haematuria. Reviewing five papers (Cardenas 2009; Cardenas 2011; De Ridder 2005; Sutherland 1996; Vapnek 2003), the authors concluded that both UTI and haematuria occurred less frequently with the use of hydrophilic‐coated catheters. The review findings were based on two errors: mistaking proportions for raw data in the Cardenas 2011 trial where raw data were not reported; and errors in reporting attrition. These two errors skewed interpretation of the data in favour of hydrophilic‐coated. We do not agree with the conclusion based on the analysis.