Article Text
PDF
Abstract
Background Patients presenting to the ED with suspected renal colic are frequently imaged with CT urography (CTU), which rarely alters diagnosis or management. To reduce use of CTU in this population, we instigated a new imaging and management guideline in our ED.
Methods This was a quasi-experimental prospective study, whereby a new guideline was commenced at the intervention site (Monash Medical Centre) and the existing guideline continued at the control site (Dandenong Hospital). The new guideline promotes focused ultrasound for diagnosing renal colic and restricts CT to those with poor response to analgesia or ‘red flags’. A consecutive series of patients with suspected renal colic were prospectively enrolled and outcomes compared between the sites. The primary outcome was CTU utilisation and secondary outcomes were radiation exposure, stone rate on CTU, admission, ED length of stay and rates of urological intervention and returns to ED at 4-week follow-up.
Results Preintervention CTU rates were 76.7% at Monash and 72.1% at Dandenong. 324 patients were enrolled; 148 at Monash and 176 at Dandenong. Median age 47 years vs 49 years, males 76.4% vs 66.5% and medianSex, Timing, Origin, Nausea, Erythrocytes (STONE) score 10 vs 10 for Monash and Dandenong, respectively. CTU was performed in 54.1% vs 75.0% (p<0.001), median radiation exposure 2.8 vs 4.0 mSv (p<0.001) and urological intervention occurred in 16.4% vs 15.7% for Monash and Dandenong, respectively.
Conclusions We found that use of CTU for renal colic was significantly reduced by introduction of a guideline promoting ultrasound and encouraging selective CTU. Although intervention rates were similar between the two sites, further prospective study is needed to ensure other patient-centred outcomes do not differ.
- uro-genital
- ultrasound
- imaging, CT/MRI
- guidelines
- abdomen - uro-genital
Statistics from Altmetric.com
Key messages
What is already known on this subject?
CT use has become widespread in suspected renal colic without associated improvement in patient-centred outcomes.
Associated harms include overdiagnosis, radiation harm and resource waste.
Ultrasound has been demonstrated to be non-inferior to CT in a randomised trial.
Diagnostic pathways incorporating decision tools and/or ultrasound to safely reduce unnecessary CT use need validation.
What this study adds?
Use of CTU for renal colic was significantly reduced by introduction of a guideline promoting ultrasound and encouraging selective CTU.
Introduction
Renal colic is a condition commonly managed in EDs. The acute symptoms can be severe, but the subsequent course is usually benign, with up to 80% of ureteric stones passing spontaneously.1 Traditionally, renal colic was considered a clinical diagnosis, with imaging reserved for specific cases. These included presence of ‘red flag’ features, such as fever, single or transplanted kidney, renal impairment or the presence of ongoing severe pain.2
Over the past decade, the non-contrast CT urogram (CTU) has become almost routine, first-line imaging in many institutions.2 3 The diagnostic sensitivity and specificity both exceed 90%; details of size and site of the stone, the two best predictors of spontaneous stone passage, are provided.1 4 5 However, despite a documented 12-fold increase in CTU utilisation for suspected renal colic in recent years, diagnostic rates and the number requiring operative intervention have not changed.6 Researchers have raised the issue of harm stemming from CTU overuse. This includes a potential small, but significant, increased lifetime risk of malignancy from radiation exposure7–9; and the detection of incidental findings, which are often investigated, despite usually proving to be benign.10 11
Two recent studies have suggested that CTU usage might be reduced for renal colic by either using ultrasound as the initial imaging,12 or calculating the predictive STONE score.13 Best methods of introducing change and the extent of the effect on CTU use remain unknown. At our institution, a prospective study showed that 75% of patients with suspected renal colic underwent CTU, very few had ultrasound and stone detection rates were below 60%.14 This suggested that there was an opportunity to reduce CTU use in our population. We developed a guideline based on focused point of care bedside ultrasound (PoCUS) as the initial imaging, with CTU use and disposition linked to clinical response to analgesia.
The primary aim of this study was to compare the proportion of patients with suspected renal colic undergoing CTU on initial presentation, between Monash Medical Centre where the new imaging guideline was introduced, and Dandenong Hospital, where it was not. The purpose was to determine if this approach would reduce CTU utilisation and hence patient radiation exposure, without negatively impacting those patients who ultimately required urological intervention. Data for STONE score calculation was collected at both sites, in order to externally validate the score and to examine its potential for inclusion in future guidelines.
Methods
Study design, setting and period
A quasi-experimental prospective cohort study was conducted on a consecutive series of eligible patients at two Monash Health hospitals, Melbourne, Australia, from 1 November 2015 to 28 February 2016. The study site was Monash Medical Centre (tertiary referral hospital, ED annual census 78 000 patients); the control site was Dandenong Hospital (urban district hospital, ED annual census 57 000 patients).
Participants
Inclusion criteria: patients aged 18 years or more, where the doctor has an initial clinical suspicion of renal colic prior to investigations being requested.
Exclusion criteria: pregnancy; presentation with renal colic in the preceding 4 weeks.
Materials
The long-standing Monash Health guideline pertaining to renal colic outlined the typical clinical features and highlighted the ‘red flags’. Imaging with CTU was at the discretion of the attending doctor, and PoCUS was not mentioned. At Monash Health, the use of bedside ultrasound in the ED for various conditions has been increasing over the past decade, with focused renal ultrasound being a relatively recent addition to the scope of practice. The new guideline was developed by the authors (see online supplementary appendix 1) In brief, unless ‘red flags’ were present, PoCUS was to be the initial imaging for suspected renal colic. If hydronephrosis was present, the diagnosis was supported, and CTU was only to be done if the analgesic response was poor. If hydronephrosis was absent, other diagnoses were to be pursued appropriately. Imaging was not mandated in either guideline.
Supplementary Material
Monash patients. CTU, CT urography.
The ‘red flags’ were defined as: serum creatinine >105 μmol/L (1.2 mg/dL), fever either subjectively at home or temperature above 37.9°C in ED, known single kidney or renal transplant. A good clinical response to analgesia was only broadly defined as pain being tolerable enough for discharge after 4 hours.
Outcome measures
Primary: comparison of CTU usage at each site.
Secondary: comparison between sites of radiation exposure, stone detection rate on CTU, admission, need for urological intervention and time to this, ED length of stay (LOS) and return to ED. The STONE score was calculated for all patients at each site during data analysis, after study completion.
Study procedure
During October 2015, ED doctors at Monash received education on the new imaging guideline via electronic communications and small group workshops. At this site, the study was primarily promoted as pertaining to guideline compliance. At the Dandenong site, the study was primarily promoted as gathering of information on the STONE score, with staff to follow their usual imaging practices. Data collection took place at both sites from 1 November 2015 to 28 February 2016.
At each site, the ED doctors were asked to complete a Case Report Form (CRF) for every eligible patient. These were defined as patients with initially suspected renal colic for whom a CTU was performed, whether or not stone presence was subsequently proven, or for whom CTU was not performed but the discharge diagnosis remained as renal colic. At both sites, the CRF included the clinical information pertaining to the STONE score, but not the scoring system or its interpretation. Hence, the patient’s score was not readily apparent to the doctor. At Monash, the CRF included a separate page for ultrasound findings, clinical response and CTU use/findings. Twice weekly, the investigators obtained reports from the ED information system (Symphony V.1; Ascribe, Bolton, UK) for all adult patients who either had a CTU performed or who had a discharge diagnosis of renal colic without a CTU being done. These reports were matched against completed CRFs; missing forms were requested from the attending doctor in retrospect. The ED information system was checked for reattendance within 4 weeks; CTU, stone presence and need for urological intervention was noted.
PoCUS reports were primarily obtained from the CRF, but were completed from the report entered on Symphony if absent. ED LOS was obtained from Symphony; disposition and urological intervention were obtained from the hospital’s electronic medical record. Radiation dose was obtained as part of the CTU report.
Statistical analysis and sample size
Descriptive variables are reported as median with IQR, or number and percentage for continuous and categorical variables respectively. Comparisons were made using Χ2, Fisher’s exact or Wilcoxon rank-sum tests as appropriate. Study variables collected at each site were: patient age, sex, ethnic origin, clinical features (pain duration prior to ED attendance, presence of nausea and vomiting and urinalysis results), CTU (yes/no, if yes: radiation dose, stone present yes/no, disposition (home/short stay unit/ward), ED LOS (hours), urological intervention (yes/no), time to urological intervention (days), ED return visit (yes/no). Median radiation dose was averaged over each cohort, not per patient. Additional variables collected at Monash were: presence of ‘red flags’ (yes/no/which), PoCUS (yes/no, if yes: hydronephrosis yes/no), protocol followed (yes/no, if no: apparent reason, pain response by 4 hours (good/poor)).
Sample size was based on unpublished research conducted at our institution in 2014. At this time, for Monash and Dandenong, respectively, CTU rates for suspected renal colic were 76.1% (95% CI 67.2 to 83.6) and 71.2% (95% CI 61.8 to 79.4). The two sites were also shown to be otherwise comparable, with stone detection rates of 80.2% (95% CI 70.2 to 88.0) and 75.9% (95% CI 65.0 to 84.9), admission rates of 11.5% (95% CI 6.3 to 18.9) and 7.2% (95% CI 3.2 to 13.7) and urological intervention rates of 13.3% (95% CI 7.6 to 20.9) and 8.1% (95% CI 3.8 to 14.8).14 We believed that a clinically worthwhile reduction in CTU use would be from 75% to <60%. A sample of 150 patients per site was sufficient to demonstrate this difference (power 0.80, alpha 0.05). The study was not powered to examine secondary outcomes. All analyses were conducted using Stata statistical software (V.12.1; StataCorp, College Station, Texas, USA).
Ethical considerations
Study conduct was approved by the Monash Health Human Research and Ethics Committee as a Quality and Service Improvement Activity as project 15347Q. There were no financial or intellectual conflicts of interest.
Results
A total of 324 eligible patients attended the two EDs during the 120-day study period: 148 (45.7%) at Monash Medical Centre and 176 (54.3%) at Dandenong Hospital. Baseline characteristics were similar: median age 47 years (IQR: 38–57) vs 49 years (IQR: 38–56), males 113/148 (76.4%, 95% CI 68.7 to 82.9) vs 117/176 (66.5%, 95% CI 59.0 to 73.4) and median STONE score 10 (IQR: 9–11) vs 108–11 for Monash and Dandenong, respectively. CTU was performed at the initial visit for 80/148 (54.1%, 95% CI 45.7 to 62.3) at Monash and 132/176 (75.0%, 95% CI 67.9 to 81.2) at Dandenong, p<0.001 (Χ2).
Although the study was not powered for the secondary outcomes, we did find a statistically significant difference in median radiation: 2.8 mSv (IQR: 0–5.2) vs 4.0 mSv (IQR: 0–6.6), p<0.001 (Mann-Whitney U test) for Monash and Dandenong, respectively. ED LOS and frequency of urological intervention appeared similar for patients in the study at both sites. There was no evidence for an increase in return visits to Monash. Stone detection rate on CTU was different: 61/80 (76.3%, 95% CI 65.4 to 85.1) at Monash vs 83/132 (62.9%, 95% CI 54.0 to 71.1) at Dandenong, p=0.04 (Χ2) (table 1).
Study population and outcomes.
Figure 1 details the imaging, interventions and returns of the 148 patients in the Monash cohort. For 23 patients, 25 ‘red flags’ were noted; 16 went on to have a CT and the other 7 did not (figure 2). These comprised 16 with renal impairment, 8 with fever and 1 single kidney. Of the 125 patients with no red flags, 49 had ultrasound performed and 76 did not. Twenty-four of the 49 patients (49.0%) who had ultrasound eventually went on to have a CTU, whereas 40 of the 76 patients (52.6%) with no ultrasound ended up with a CTU (figure 3). Of the 125, 115 (92.0%) had clinical outcome information recorded. CTU rates from highest to lowest: no ultrasound and poor clinical response (6/6, 100%), ultrasound done and poor clinical response (7/9, 77.8%), no ultrasound and good response (30/63, 47.6%) and ultrasound done with good response (14/37, 37.8%).
The most common reasons for non-performance of ultrasound were: no accredited doctor present (32/73, 43.8%) and accredited doctor present but no time to perform the test (21/73, 28.8%).
Red flag patients. CTU, CT urography.
No ultrasound patients. CTU, CT urography.
In total, 23 patients ultimately underwent a urological intervention, 10 at Monash and 13 at Dandenong. This was done following admission on the first attendance for 6 of 10 Monash patients and 7 of 13 Dandenong patients. In total, 51 patients returned, 18 at Monash and 33 at Dandenong. Urological intervention was arranged at this return visit for 4 of the 18 at Monash, and 6 of the 33 at Dandenong. The four Monash patients returned on days 1, 6, 8 and 9; five of the six Dandenong patients had returned by 2 days, the other on day 8. Admission for intervention occurred on this day in all cases.
For STONE score calculation, complete data were available for 302 of the 324 (93.2%) patients. Within this group, 280 (92.7%) would have been classified as moderate risk or greater (STONE score ≥6) and 196 (64.9%) had CTU performed. When CTU was done, stone detection rates were significantly different between the risk groups, being 27.8% (95% CI 9.7 to 53.5), 52.6% (95% CI 40.8 to 64.2) and 86.3% (95% CI 78.0 to 92.3) for the low-risk, moderate-risk and high-risk categories, respectively (p<0.001, Fisher’s exact test). No difference in the CTU usage or intervention rates was detected, although numbers were very small (table 2).
Imaging and intervention rates by STONE score.
Discussion
This study found that use of a simple assessment guideline for suspected renal colic was associated with a 21% lower CTU rate at the study site, compared against the control site. Consequently, the median radiation exposure for the population at the study site was also significantly lower. To our knowledge, this is the first study to demonstrate that a guideline based on PoCUS as the initial imaging reduces CTU use in everyday practice. Although this has been explored theoretically in the literature,5 15 16 it has remained unclear how this would translate to real-world practice.
Our aim was to discourage CTU, when initial PoCUS demonstrated hydronephrosis and the clinical response to analgesia was good. We considered the safety of this approach to be supported by a recent RCT, which compared ultrasound with CT as the initial imaging for suspected renal colic.12 In that study, Smith-Bindman et al found that only 40% of patients in the initial ultrasound group went on to have CT scans, and that delayed detection of important alternate diagnoses was rare.12 While we did not collect data on alternate diagnoses, we identified no clinically important delays to urological intervention when this was required.
Our finding that 54% of patients progressed to have CTU is less impressive than the theoretical target of 40% in a three-arm study suggested by Smith-Bindman et al.12 That study was conducted in centres with an interest in PoCUS, and there were two ultrasound arms, one for PoCUS and one for radiologist performed ultrasound, in addition to CTU. In our ED, only a minority of emergency physicians perform renal ultrasound, so it only occurred in 39% of eligible patients. Also, the confirmed stone rate was 31%–34% in that study, versus 63%–76% in our study. It is difficult to compare the populations as the stone detection rate was based on stone passage or urological intervention, while in our study, it was detection of stone on a CT. We also found that only 11 of 17 (65%) PoCUS patients with a stone on subsequent CTU had hydronephrosis detected on ultrasound. This is consistent with prior studies reporting a sensitivity of between 65% and 78% of hydronephrosis on PoCUS for stones on CTU.17–20 It is understandable that non-detection of hydronephrosis would lead to diagnostic doubt, which may explain why the CTU rate for good clinical responders was only 12% when hydronephrosis was seen, but 50% when it was not. Increasing the use of PoCUS may not be an easy aim, given that being too busy was a commonly cited reason for non-performance. It will always be quicker for a harried ED doctor to request a radiology-performed test than to perform an ultrasound themselves.
Besides ultrasound being the initial recommended imaging, the other new concept in the guideline was that patients with a good response to analgesia by 4 hours were ‘allowed’ to be discharged without CTU. Of 23 ‘red flag’ patients who ought to have undergone CTU according to our guideline, 7 did not; all were discharged after a good clinical response to analgesia; 2 returned to ED but no delayed CTU was performed and no interventions occurred. Out of the 125 non-red flag patients, 41 had no initial imaging and were discharged with a good clinical response. Five patients returned, all had a CTU positive for stones and three underwent intervention. With the recurrent nature of renal colic, one study of patients undergoing urological procedures reported that 30% had exceeded the recommended annual maximum radiation exposure for little demonstrable benefit.21 The difficulties associated with incidental findings on potentially unnecessary CTUs have also been described.10 11 22–25 These pose problems for both patients and the health system, since the vast majority of such findings are ultimately found to be benign.10 11 Gaining acceptance for a new management approach which yields less diagnostic information may not prove easy, but seems worthwhile.
We prospectively validated the STONE score in our population; stone rates for moderate-risk and high-risk groups were almost identical between our study and the original study.13 Our sample size was too small to determine if there was an association between the STONE score, presence of hydronephrosis or a poor clinical response and urological intervention. We did not see a difference in our secondary outcomes of LOS, hospital admission or ED return, although the study was not powered for these outcomes. Smith-Bindman et al did not find an impact of PoCUS on LOS or disposition.12 This is unsurprising, since the rate-limiting step to discharge is probably time to analgesic response, rather than for performance of imaging.
Limitations
A quasi-experimental design was selected as we were concerned about contamination if running a randomised study with two arms at one site. This could result in unmeasured confounders. However, other than the study guideline, no procedural changes pertaining to renal colic took place at either ED, and stone detection rates were the same before and after the intervention at the control hospital.
Good and poor clinical response was only defined from the practical point of view, in that a good response meant pain was tolerable enough to allow discharge. Smith-Bindman et al reported mean discharge pain ratings of 3 (±3), which were little changed at days 3 and 7 follow-ups, so it seems reasonable to have left this loosely defined. Analgesic regimens were also not dictated, and are likely to have varied between practitioners. No grading of hydronephrosis on ultrasound was requested, and inter-rater reliability between our ED doctors is not known. Detection of postdischarge CTU use and diagnostic outcome relied on return to a Monash Health ED, and patients were not otherwise followed-up.
On the minority of occasions when the CRF was not completed at the time, obtaining the clinical information retrospectively introduces the possibility of recall bias. We believe this is not likely to be a significant issue, because the attending doctor was not solely reliant on memory, but able to refer to the medical record, in which most of the information will have been noted.
The study was powered only for the primary outcome, and thus several of the patient-centred outcomes could not be evaluated for statistical significance.
Some patients with renal colic who did not have a CTU are likely to have been missed through either misrecording of the discharge diagnosis, or non-consideration of the diagnosis. Although some such cases will have been detected on representation, it is not possible to know exact numbers, although we suspect they would be small.
Our inclusion criterion of ‘suspected renal colic’ was fairly subjective, as it is in other studies. This leads to differences, such as the studies by Moore et al 13 and Smith-Bindman et al 12 having 50% male participants, compared with our 70%. The median STONE score in the study by Moore et al was 7–8, and doctors in the Smith-Bindman study rated pretest likelihood for a stone at <50%. Our median STONE score was 10, suggesting a higher pretest probability in our population than prior studies.
Conclusion
This study found that CTU use could be significantly reduced by introduction of a simple guideline, with ED PoCUS as the initial imaging and CTU reserved for those with ‘red flags’ or poor analgesic response. CTU use may be reduced further as training and expertise in PoCUS continues to grow. Expansion of the revised guideline to the other two EDs in the network is planned to further realise the benefits demonstrated.
References
Footnotes
Contributors GB: design, data collection, data analysis, result interpretation, manuscript write up, review and submission.RM: design, result interpretation, manuscript write up and review. DEW: design, result interpretation, manuscript write up and review. PM: design, review.
Competing interests None declared.
Ethics approval Monash Health HREC.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No further data are available.