Medical expulsive therapy in adults with ureteric colic: a multicentre, randomised, placebo-controlled trial

BACKGROUND
Meta-analyses of previous randomised controlled trials concluded that the smooth muscle relaxant drugs tamsulosin and nifedipine assisted stone passage for people managed expectantly for ureteric colic, but emphasised the need for high-quality trials with wide inclusion criteria. We aimed to fulfil this need by testing effectiveness of these drugs in a standard clinical care setting.


METHODS
For this multicentre, randomised, placebo-controlled trial, we recruited adults (aged 18-65 years) undergoing expectant management for a single ureteric stone identified by CT at 24 UK hospitals. Participants were randomly assigned by a remote randomisation system to tamsulosin 400 μg, nifedipine 30 mg, or placebo taken daily for up to 4 weeks, using an algorithm with centre, stone size (≤5 mm or >5 mm), and stone location (upper, mid, or lower ureter) as minimisation covariates. Participants, clinicians, and trial personnel were masked to treatment assignment. The primary outcome was the proportion of participants who did not need further intervention for stone clearance within 4 weeks of randomisation, analysed in a modified intention-to-treat population defined as all eligible patients for whom we had primary outcome data. This trial is registered with the European Clinical Trials Database, EudraCT number 2010-019469-26, and as an International Standard Randomised Controlled Trial, number 69423238.


FINDINGS
Between Jan 11, 2011, and Dec 20, 2013, we randomly assigned 1167 participants, 1136 (97%) of whom were included in the primary analysis (17 were excluded because of ineligibility and 14 participants were lost to follow-up). 303 (80%) of 379 participants in the placebo group did not need further intervention by 4 weeks, compared with 307 (81%) of 378 in the tamsulosin group (adjusted risk difference 1·3% [95% CI -5·7 to 8·3]; p=0·73) and 304 (80%) of 379 in the nifedipine group (0·5% [-5·6 to 6·5]; p=0·88). No difference was noted between active treatment and placebo (p=0·78), or between tamsulosin and nifedipine (p=0·77). Serious adverse events were reported in three participants in the nifedipine group (one had right loin pain, diarrhoea, and vomiting; one had malaise, headache, and chest pain; and one had severe chest pain, difficulty breathing, and left arm pain) and in one participant in the placebo group (headache, dizziness, lightheadedness, and chronic abdominal pain).


INTERPRETATION
Tamsulosin 400 μg and nifedipine 30 mg are not effective at decreasing the need for further treatment to achieve stone clearance in 4 weeks for patients with expectantly managed ureteric colic.


FUNDING
UK National Institute for Health Research Health Technology Assessment Programme.


Introduction
Ureteric colic is defi ned as episodic severe abdominal pain from sustained contraction of ureteric smooth muscle as a kidney stone passes down the ureter into the bladder. 1 It is a common reason for people to seek emergency health care and was associated with 550 000 emergency room visits in the USA in 2009 2 (costing US$3 billion) and 25 000 hospital admissions in England in 2012 3 (costing £11·6 million). After clinical assessment and stone localisation by noncontrast CT of the kidneys, ureters, and bladder (CT KUB), patients can generally be managed at home with analgesia with the expectation of spontaneous stone passage. 1 The likelihood that the stone will pass within 4 weeks ranges between 50% and 95% depending on stone size and location in the ureter. 4 Expectantly managed patients who develop recurrent pain, sepsis, or compromised renal function need drainage if necessary followed by stone clearance using endoscopy or extracorporeal shock wave lithotripsy. 5 Treatments that increase likelihood of stone passage would be expected to benefi t patients with ureteric colic because the need for an interventional procedure will be reduced. The smooth muscle relaxant drugs tamsulosin (an α-adrenoceptor antagonist) and nifedipine (a calcium channel stabiliser) are possible agents, their use being termed medical expulsive therapy (MET). 6 Meta-analyses of data from randomised controlled trials (RCTs) report a statistically signifi cant benefi t for both tamsulosin and nifedipine over controls for the outcome of spontaneous stone passage 7,8 with tamsulosin better than nifedipine. 8 However, the clinical usefulness of MET is uncertain because of the predominance of small, single-centre, low-to-moderate quality trials and variability in trial design with diff ering inclusion criteria and outcome measure ment. 7,8 Despite these cautions and although not licensed for this indication, MET is recommended by clinical guidance 9 and is being adopted as part of routine expectant management. 10 We now report the clinical results of the pragmatic randomised Spontaneous Urinary Stone Passage Enabled by Drugs (SUSPEND) trial, designed to resolve uncertainty about the benefi t of these drugs when used in routine care of people with expectantly managed ureteric colic. We sought to establish whether tamsulosin or nifedipine increased the likelihood of spontaneous stone passage measured by the absence of need for further intervention and, if so, which was the better drug.

Study design and participants
In this randomised, placebo-controlled trial, we recruited patients presenting to 24 UK National Health Service hospitals with ureteric colic. Adults aged 18-65 years with one stone of 10 mm or less (at the largest dimension) in either ureter identifi ed on CT KUB were included. Patients who were ineligible included those needing immediate intervention decided by clinical assessment, those with sepsis, those with an estimated glomerular fi ltration rate of less than 30 mL/min, and those already taking or unable to take an α blocker or calcium channel stabiliser. We excluded people older than 65 years because nifedipine dose titration is recommended for this age group. Full details of all inclusion and exclusion criteria are available in our published protocol. 11 Patients gave written informed consent in line with Good Clinical Practice and the Declaration of Helsinki. The trial was approved by the East of Scotland Research Ethics Service (reference 10/S0501/31) and the UK Medicines and Healthcare products Regulatory Agency (MHRA; reference 2010-019469-26).

Randomisation and masking
Trained site personnel (research nurses and clinicians) enrolled participants at each site. Participants were allocated in a 1:1:1 ratio to either tamsulosin, nifedipine, or placebo by a remote randomisation system hosted at the Centre for Healthcare Randomised Trials (CHaRT) in Aberdeen, UK, using an algorithm with centre, stone size (≤5 mm or >5 mm), and stone location (upper, mid, or lower ureter) as minimisation covariates. Each randomly assigned participant was given 28 capsules of trial medication (over-encapsulated tamsulosin or nifedipine, or placebo) supplied by an independent source (Tayside Pharmaceuticals, Ninewells Hospital, Dundee, UK) who had no further involvement in the trial, ensuring that participants, clinicians, and trial personnel remained unaware of the allocated group.

Research in context
Evidence before this study In planning our trial we identifi ed two contemporary highquality systematic reviews that had appraised and metaanalysed previous randomised controlled trials in this specialty. One was a Cochrane review that included trials of treatment with α blockers retrieved from the Cochrane Renal Group's Specialised Register up to July 9, 2012. Meta-analysis of 32 trials involving 5864 participants showed that use of α blockers increased likelihood of stone passage compared with control, with a relative risk (RR) of 1·48 (95% CI 1·33-1·64). Additionally, from four trials involving 3486 participants, tamsulosin seemed to be better than nifedipine at increasing the likelihood of stone passage (RR 1·19, 95% CI 1·05-1·35). Overall, the included studies were deemed to have a high or unclear risk of bias around masking and a low risk of bias around outcome collection and reporting. The second, a systematic review by Seitz and colleagues, searched Medline, Embase, the Cochrane Central Register of Controlled Trials, and the Cochrane Database of Systematic Reviews up to Dec 31, 2008, for randomised trials of use of both α blockers and calcium channel stabilisers as medical expulsive therapy (MET). Combining all 29 studies involving 2419 participants, they noted that MET increased the likelihood of spontaneous stone passage (RR 1·45, 95% CI 1·34-1·57). Additionally, they found evidence from nine trials involving 686 participants that nifedipine increased likelihood of stone passage compared with controls (RR 1·49, 95% CI 1·33-1·66). Overall, the trials were found to be of low-tomoderate quality on the basis of a validated scoring method.

Added value of this study
Both reviews emphasised uncertainty in estimates of eff ect due to the small size of most component studies, diff erences in inclusion criteria and outcome measurement, and inadequate masking of participants and assessors. We sought to overcome these defi ciencies by designing a large, multicentre trial with robust means of concealment of allocated treatment. We also chose a clinically relevant, attributable, and clear primary outcome measure. Using this method we established that neither the most frequently used α blocker, tamsulosin, or the calcium channel stabiliser nifedipine showed any clinically useful benefi t for increasing stone passage measured by the absence of need for further intervention.

Implications of all the evidence
Results of previous studies showed a positive benefi t on spontaneous stone passage with these agents. However, our methodologically sound and large trial off ers a strong evidence base for the alternative view that they are unlikely to be useful in the routine clinical care of people with ureteric colic.

Procedures
Participants self-administered tamsulosin 400 μg, nifedipine 30 mg, or placebo orally once daily until spontaneous stone passage occurred, the need for intervention was agreed, or until 4 weeks had passed since randomisation, whichever came fi rst. We did not verify adherence to trial medication.
Baseline data were collected before randomisation in hospital. Local research staff obtained follow-up data by use of participant questionnaires, which were completed at home at 4 and 12 weeks, and case report forms, completed during clinic visits or telephone contact at 4 and 12 weeks. No clinical tests were mandatory as part of the trial protocol.
Safety outcomes were reported as and when they happened (via the case report form, patient questionnaires, and patient and clinician report). Suspected serious adverse events were graded at site by the local principal investigator, reported to the trial offi ce to be confi rmed by the chief investigator. Safety events were monitored by the sponsor, research ethics committee, and MHRA. Non-serious adverse events were not collected or reported.

Outcomes
The primary outcome was spontaneous stone passage in 4 weeks, defi ned as the absence of need for additional interventions to assist stone passage at 4 weeks after randomisation. Other outcomes were pain assessed by participant-reported number of days of analgesic use and visual analogue scale at 4 weeks, time to stone passage assessed by the date of imaging showing no stone at up to 4 weeks, health status assessed by the Short Form (SF)-36 questionnaire, and safety assessed by participant report of dis continuation of medication due to adverse eff ects and by serious adverse events monitoring. We also assessed health outcomes with the EQ-5D questionnaire, and heath-care resource use and participant costs (health economic components), the results of which will be reported elsewhere.

Statistical analysis
The trial was powered for the most conservative hypothesis: that the proportion of participants who passed their stone would be 10% higher in the tamsulosin group compared with the nifedipine group (85% vs 75%). We prespecifi ed two comparisons, MET (tamsulosin or nifedipine) against placebo, and tamsulosin against nifedipine. We also made post-hoc comparisons of each agent against placebo. For 90% power and a type I error rate of 5%, we required 1062 participants (354 in each group), which we infl ated to 1200 to allow for 10% loss to follow-up.
We use summary statistics to describe the characteristics of the trial groups at baseline and during follow-up. We used generalised linear models to analyse outcomes with adjustment for minimisation covariates, logistic regression for binary outcomes, and linear regression for continuous outcomes. We present treatment eff ects with 95% CI; for the primary outcome we present both the odds ratio and absolute risk diff erence. We estimated CIs for the risk diff erence using the delta method. We analysed data for the primary outcome from the modifi ed intention-to-treat population, which included all randomly assigned participants apart from those with missing primary outcome data and those who were found to be ineligible after randomisation. For secondary outcomes, we   included participants in a complete case analysis under a missing-at-random assumption.
Sensitivity analyses assessed the eff ect of missing data on pain and health status outcomes. 12 First, we repeated analyses using multiple imputation models with predictions based on all baseline covariates collected with results combined across ten imputed datasets. We then tested the robustness of these results using pattern mixture models, which imputed missing SF-36 scores across a range of potential values from minus half the observed SD to plus half the SD of the outcome being analysed.
All participants were analysed as randomised. Subgroup analyses explored the possible eff ect modifi cation of stone size, location in ureter, and sex, all using treatment by subgroup interaction terms and 99% CI. Further details are available from our published protocol. 11 We did all analyses in Stata 13. An accompanying cost-eff ectiveness analysis will be reported separately.

Role of funding source
The funder (through their peer and funding board review process) approved the study proposal but had no role in the collection, analysis, or interpretation of data, or writing of the report. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

Results
Between Jan 11, 2011, and Dec 20, 2013, 1167 participants were randomly assigned (391 to tamsulosin, 387 to nifedipine, and 389 to placebo; fi gure 1). Of these, 17 were subsequently excluded because of ineligibility and 14 participants were lost to follow-up, and were not included in the primary outcome analysis. We were able to ascertain the primary outcome for 1136 (97%) participants in the fi nal analysis. 719 (62%) of 1150 eligible participants completed the 4-week questionnaire and 564 (49%) of 1150 eligible participants completed the 12-week questionnaire, with no diff erences in the proportion returned between groups (data not shown). Baseline characteristics were similar for the three groups (table 1).
Spontaneous stone passage, defi ned by absence of need for intervention to assist stone passage during the 4 weeks after randomisation, did not diff er between groups (table 2). 307 (81%) of 378 participants in the tamsulosin group needed no further intervention compared with 304 (80%) of 379 in the nifedipine group, and 303 (80%) of 379 in the placebo group. These fi ndings were consistent across the predefi ned subgroups of sex, stone size, and stone location (fi gure 2). We also noted no diff erence in stone passage at up to 12 weeks (data not shown), by which time an additional 27 (7%) participants in the tamsulosin group, 25 (6%) in the nifedipine group, and 28 (7%) in the placebo group had an intervention planned. No diff erences were recorded in the secondary outcomes of days of analgesic use (table 3), time to stone passage  (table 3), and health status between the groups (fi gure 3). Data are mean (SD) and number (%). Data do not include participants who were excluded after randomisation. *Ranging from 0 (no pain) to 10 (worst imaginable pain). †Ranging from 0 (maximum disability) to 100 (no disability).  Sensitivity analyses showed that these estimates were robust to assumptions about missing data for all but implausible scenarios (data not shown). Serious adverse events were reported in three participants allocated to nifedipine (one had right loin pain, diarrhoea, and vomiting; one had malaise, headache, and chest pain; and one had severe chest pain, diffi culty breathing, and left arm pain) and in one participant in the placebo group (headache, dizziness, lightheadedness, and chronic abdominal pain). No deaths were reported.

Discussion
The results of our trial, done in a routine care setting with masking of treatment allocation, showed that use of tamsulosin and nifedipine did not aff ect the proportion of patients needing further intervention to clear their stone during 4 weeks. This fi nding, suggesting similar rates of spontaneous stone passage across the trial groups, was consistent when possible modifi ers of the likelihood of spontaneous stone passage were taken into account, and was unchanged at 12 weeks after randomisation. We found no evidence that the drugs reduced pain, hastened time to stone passage, or improved health state. The precision of trial estimates of treatment eff ect is suffi cient to rule out any clinically useful benefi t of these drugs to assist stone passage in this patient group at the dose and duration examined.
We used a pragmatic trial design in a multicentre setting to ensure that the drugs were tested in a manner aligned with present clinical practice recommendations, allowing direct translation of the results into standard clinical care. As recommended by a Cochrane review, 7 the key entry criterion was identifi cation of one ureteric stone by CT KUB, ensuring that only patients with a symptomatic ureteric stone were included. The trial cohort was representative of the target population for MET with baseline characteristics such as stone size and location consistent with recent case series. 13,14 We chose to include treatment with tamsulosin as the α-adrenoceptor antagonist most often used for MET, and nifedipine since, at the time of trial design, both agents had been put forward as having effi cacy at the dose and duration used in our trial to increase rates of spontaneous stone passage. 7,8 Inclusion of a placebo group, independent   control of randomisation sequence, and over-encapsulated trial medication supply ensured masking of participants, clinicians, and outcome assessors to allocation. We were able to collect data for the primary outcome for 97% of the large trial cohort, which surpassed the prespecifi ed sample size. The design and completion to target of this trial therefore allows strong and clinically useful conclusions to be drawn concerning the lack of eff ectiveness of MET. Our trial design diff ered from other trials in two ways that seekers of evidence regarding the use of MET will have to consider. First, we chose to have wide eligibility criteria, including patients with symptomatic stones of 10 mm or smaller (at the largest dimension) located at any site in the ureter. We chose these criteria because we could fi nd no evidence to suggest that use of MET is more eff ective if targeted at specifi c patient groups categorised by stone size or site. Our choice is supported by present guidance that recommends MET but does not suggest any preliminary patient categorisation, 9,15 and although data are limited, this recommendation seems to have led to increasing use of MET. 10,16,17 Second, we focused on the absence of need for further intervention as the measure of spontaneous stone passage. We chose this endpoint as our primary outcome because it was consistent with the pragmatic study design assessing clinical eff ectiveness of the drugs. Having to undergo an intervention is a key outcome of interest for patients, signifying unresolved ill health with resultant costs. For clinicians it signifi es the need to upscale care, and for health-care providers it drives requirement for additional capacity and funding. Other trials designed to have a low risk of bias have placed an additional lower limit on stone size, typically 3-5 mm, with or without a further restriction to patients with stones in the lower (distal) ureter at diagnosis. 8,18 The investigators of these trials reasoned that this patient group, with an expected proportion of patients with spontaneous passage of about 70%, were most likely to benefi t from MET. 4 In planning our trial and considering the evidence, we believed that such subcategorisation was unwarranted and furthermore was unlikely to be feasible in routine emergency practice since it requires accurate stone measurement. Our study results, including the absence of any interaction between stone size or location and the primary outcome, gives us no cause to change this belief, although the trial was not formally powered for such subgroup analysis. We considered having imaging evidence of stone passage as a trial outcome, in common with other studies, but rejected it for two main reasons. Modalities with minimal radiation risk such as ultrasound and plain abdominal radiograph have inadequate diagnostic accuracy for the presence or absence of a stone, 19 whereas the defi nitive modality, CT KUB, has a radiation dose and cost that is suffi cient to result in it not being recommended for this purpose by present clinical    guidance. 19,20 In our trial population, as in routine clinical care, further imaging was used when clinically indicated by continued pain, development of infection, renal dysfunction, or radiographic evidence of obstruction on diagnosis. 1 These fi ndings were the clinical criteria against which the need for trial participants to undergo further intervention was judged. The main reason for ineligibility noted during screening was use of an imaging method other than CT KUB for stone identifi cation. The proportion of exclusions accounted for by this criterion fell from 43% to 25% during the recruitment period, suggesting an increase in compliance with diagnostic imaging guidance. 19 Scrutiny of baseline characteristics gives no suggestion that requirement for CT KUB diagnosis resulted in systematic diff erences between trial groups, although overall a greater proportion of women were excluded by this criterion than were men, possibly because of radiation concerns. The site initiation and monitoring protocol used in our trial gives confi dence that the need for further intervention was accurately recorded by site staff masked to allocation. However, some participants might have had persistent asymptomatic stones that did not trigger further intervention by 4 weeks. Such participants would be expected to develop symptoms by 12 weeks and at this timepoint a further 80 participants in our trial had intervention planned, but a secondary analysis including these events did not change our fi nding of no diff erence between trial groups. The proportions of patients not needing intervention recorded in our trial populations-80% at 4 weeks and 73% at 12 weeks-is consistent with proportions reported in other cohorts. 4,21 The precision of our fi ndings related to the large sample size gives confi dence that the fi nding of no benefi t for MET was not aff ected by the overall high proportion with spontaneous stone passage.
Collection of measurements in our trial for participantreported secondary outcomes of pain control, time to stone passage, health status, and early discontinuation of trial medication, was incomplete. We found no evidence that the proportion of patients who completed measure ments diff ered between treatment groups, although younger people were less likely to return questionnaires than older people (data not shown), in common with previous community-based trials. 22 Low response rates increase uncertainty around the fi nding of no eff ect for these secondary outcomes because of possible bias from missing data, but sensitivity analyses using imputation did not change the results. The amount of pain suff ered and associated disturbance to social and working life are important characteristics of the disorder, but from an eff ectiveness trial design perspective are diffi cult to measure outside hospital in a suffi ciently large sample for accurate estimation of any treatment eff ect. We also chose not to monitor adherence to trial medication to maintain simplicity and low participant burden. Despite these uncertainties we noted no diff erences between study groups and particularly no evidence of any diff erential use of analgesics that might act as ureteric relaxants.
The aim of this trial was to provide a precise, unbiased estimate of benefi t for MET in reducing need for further intervention to assist stone passage. We minimised allocation bias using a robust randomisation process and ascertainment bias by maintaining masking of all participants involved in the trial until data analysis, and recruited to a large target sample size with complete attribution of the primary outcome. These characteristics mean that results from our trial provide the highest quality evidence about clinical eff ectiveness of MET, since systematic reviews underpinning clinical guidance recommendations have all commented on the generally unclear risk of bias and low methodological quality of previous trials. 7,8,23 Diff erences in inclusion criteria, trial design, and outcome measurement limit the opportunity for qualitative or meta-analytical comparison of component trials reported in these reviews with our trial results. Concurrent with our trial, an alternative α-adrenoceptor antagonist, silodosin, has been studied with initial small trials suggesting benefi t. 24,25 A subsequent multicentre trial reported that silodosin did not increase stone passage compared with placebo, although the prestated sample size was not attained, limiting certainty of the result. 18 Seekers of evidence often have to decide whether to base treatment decisions on a meta-analysis of several small low-quality trials typically showing larger diff erences, or on one large high-quality trial with a smaller eff ect size or a fi nding of no eff ect. 26 A recent review of this dilemma advises that judgments should involve careful consideration of qualitative and quantitative properties in each specifi c circumstance with further sensitivity analyses when possible. 27 Our judgment is that the results of our trial provide conclusive evidence that the eff ect of both tamsulosin and nifedipine in increasing the likelihood of stone passage as measured by the need for intervention is close to zero. Our trial results suggest that these drugs, with a 30-day cost of about US$20 (£13; €18), should not be off ered to patients with ureteric colic managed expectantly, giving providers of health care an opportunity to reallocate resources elsewhere. The precision of our result, ruling out any clinically meaningful benefi t, suggests that further trials involving these agents for increasing spontaneous stone passage rates will be futile. Additionally, subgroup analyses did not suggest any patient or stone characteristics predictive of benefi t from MET.
Ureteric colic is a common, painful disorder and simple treatments that would make spontaneous stone passage more likely and quicker are still needed. Although α-adrenoceptor antagonists (tamsulosin) and calcium channel stabilisers (nifedipine) are ineff ective, alternative classes of agents should continue to be identifi ed and trialled. 28