Using an International Clinical Registry of Regional Anesthesia to Identify Targets for Quality Improvement

Abstract

Background

Despite the widespread use of regional anesthesia, limited information on clinical performance exists. Institutions, therefore, have little knowledge of how they are performing in regards to both safety and effectiveness. In this study, we demonstrate how a medical institution (or physician/physician group) may use data from a multi-center clinical registry of regional anesthesia to inform quality improvement strategies.

Methods

We analyzed data from the International Registry of Regional Anesthesia that includes prospective data on peripheral regional anesthesia procedures from 19 centers located around the world. Using data from the clinical registry, we present summary statistics of the overall safety and effectiveness of regional anesthesia. Furthermore, we demonstrate, using a variety of performance measures, how these data can be used by hospitals to identify areas for quality improvement. To do so, we compare the performance of one member institution (a United States medical center in New Hampshire) to that of the other 18 member institutions of the clinical registry.

Results

The clinical registry contained information on 23,271 blocks that were performed between June 1, 2011, and May 1, 2014, on 16,725 patients. The overall success rate was 96.7%, immediate complication rate was 2.2%, and the all-cause 60-day rate of neurological sequelae was 8.3 (95% CI, 7.2–9.7) per 10,000. Registry wide major hospital events included 7 wrong site blocks, 3 seizures, 1 complete heart block, 1 retroperitoneal hematoma, and 3 pneumothoraces. For our reference medical center, we identified areas meriting quality improvement. Specifically, after accounting for differences in the age, sex, and health status of patient populations, the reference medical center appeared to rely more heavily on opioids for post procedure management, had higher patient pain scores, and experienced delayed discharge when compared with other member institutions.

Conclusions

To our knowledge, this is the first large-scale effort to use a clinical registry to provide comparative outcome rates representing the safety and effectiveness of regional anesthesia. These results can be used to help inform quality improvement strategies.

INTRODUCTION

Regional anesthesia is an important component of perioperative health care, contributing to improved patient satisfaction and reduced opioid analgesic medication use.¹ Three randomized, placebo-controlled studies conclusively demonstrated significant reductions in patient reported pain after shoulder, foot, and distal upper extremity surgery as a result of peripheral nerve block (PNB).^2–4 However, like many medical and surgical procedures, trade-offs exist in clinical practice. Despite the recognized benefits of regional anesthesia, these procedures can also lead to quality compromising events. Possible morbidity from regional anesthesia ranges from relatively minor issues such as discomfort during block placement to death related to local anesthetic toxicity.^5–7

Despite the popularity, changing nature of delivery, and potential clinical benefits of regional anesthesia, individual medical centers and physician groups have little knowledge of how they are performing from both a safety and effectiveness standpoint. Such data would be helpful for many scenarios such as the planning of quality improvement projects, departmental reporting, and the completion of the personal practice assessment component of the Maintenance of Certification in Anesthesiology (MOCA). Additionally, in the new era of Accountable Care Organizations (ACO), health care reform, and bundled payments, performance data will likely be progressively linked to reimbursement.⁸

Therefore, in an effort to begin to develop clinical benchmarks, we sought to present summary rates regarding the safety and effectiveness of PNB from an international clinical registry. To demonstrate the utility of the clinical registry, we also compared the performance of one member institution (located in New Hampshire, USA) to the rest of the registry and identified potential areas for quality improvement.

METHODS

International Registry of Regional Anesthesia

The International Registry of Regional Anesthesia (IRORA) is a prospective multi-center clinical registry that collects information on contemporary PNB practice. The study period for this current investigation is from June 1, 2011, to May 1, 2014, (35 total months of data) and, as of May 1, 2014, there were 19 contributing institutions. Detailed methodology for the clinical registry has been previously reported including definitions for immediate and delayed complications and follow-up protocol.⁹ Briefly, key features of the clinical registry include (1) a waiver of the entire consent process for data collection so that all eligible patients can be captured minimizing selection bias; (2) a web-based data collection interface providing the ability to gather information from a large number of patients close to the point of care, improving data integrity and reducing the burden of data collection; (3) data collection at the individual patient level; (4) Systematic postoperative contact with all patients; (5) a defined follow-up and investigative pathway; (6) a clear definition and verification of all key outcomes; and (7) a data quality assurance process that includes site visits, interviews, and chart reviews.

To demonstrate the utility of the clinical registry for informing improvement of care delivery, we selected one member institution, Dartmouth Hitchcock Medical Center (DHMC), to serve as a reference hospital. This institution contributed data between August 1, 2012, and May 1, 2014 (22 total months of data) to the IRORA.

The patient population of the DHMC comprised patients having PNB in the context of a surgical procedure. DHMC is a tertiary care academic medical center in New Hampshire associated with the Geisel School of Medicine at Dartmouth. DHMC offers a wide range of surgical and regional anesthetic procedures primarily for residents of the greater Northern New England area. We analyzed patients having PNBs associated with surgical procedures at DHMC and compared results to that at 18 other institutions associated with the IRORA in aggregate. The other 18 IRORA member institutions are listed in Appendix 1.

Regional Anesthesia at Dartmouth Hitchcock Medical Center

Patients arriving for surgical procedures at DHMC are evaluated by an anesthesia team. If appropriate, patients may elect to have a nerve block placed to either serve as the sole anesthetic or to provide postoperative analgesia. At DHMC, PNB are routinely performed in a block room with light sedation prior to the surgical procedure. Occasionally, PNBs are placed after the patient is under general anesthesia in the operating room. Postoperative care in the Post Anesthesia Care Unit (PACU) includes management of residual pain not covered by the PNB.

Measures of Safety and Effectiveness

Using the IRORA clinical registry, safety was assessed by quantifying immediate, PACU, and delayed complication rates. PNB related complications were defined, tracked, and pooled where appropriate. Immediate complications were complications that occurred during block placement and intraoperative period. PACU complications were complications recognized in the recovery phase of care. Delayed complications were assessed per the registry follow-up protocol and included procedure-related infections and all cause postoperative neurological sequelae present 60 days postoperatively.

We examined 4 different measures of the effectiveness of regional anesthesia. In the PACU, the following outcomes were recorded by nursing staff: 1) pain assessment was reported as the highest verbal analog score (VAS: 0, none; 10 worst pain ever experienced) recorded by nursing staff; 2) need for any supplemental opioid therapy and 3) the ability to be discharged from the PACU within 30 minutes using local discharge criteria. Procedure success 4) was determined by the physician performing the PNB and was defined as the development of anticipated clinical characteristics with evidence of sensory or motor effects.

Statistical Analyses

Simple descriptive statistics were used to examine the overall rates of the various outcomes for the entire IRORA clinical registry. To graphically examine the distributions of center-specific rates (and any potential variation across member institutions), we used turnip plots. In a cross sectional analysis, we used univariate statistics to compare the reference institution (DHMC) to the other 18 IRORA member institutions. Specifically, we used the chi-squared test to compare proportions for dichotomous outcomes (block success, opioid use, VAS score of 5 or higher, rapid PACU discharge) and a non-paired t-test to compare means for continuous measures (VAS scores, complication rates). We used generalized linear models with a logit link function to predict various dichotomous outcomes and to adjust for patient population differences including age, sex, body mass index (BMI), and American Society of Anesthesiologists (ASA) health status. In these analyses, our primary independent variable (included in our models as indicator variables that represented each institution) was the specific IRORA institution with DHMC used as the reference institution. For all analyses we set our critical alpha level to 0.05 (2-sided). We did not make adjustments for multiple comparisons. 95% CIs for the unadjusted major event rates were calculated using the binomial distribution. Statistical analyses were performed in Stata version 12 (College Station, Texas).

In order to gain insight into performance over time for DHMC, we examined selective variables using statistical process control (SPC) charts, p-charts.¹⁰ Statistical process control charts enrich simple aggregate data by examining temporal performance. Using standard SPC methods, we examined for special cause signals, which, if present, would suggest that the process is not in statistical control.¹⁰

RESULTS

Across all 19 IRORA member institutions, from June 1, 2011 to May 1, 2014, 23,271 PNBs were performed in 16,725 patients for 18,271 surgical procedures. Of these, DHMC performed 4,003 PNBs. Characteristics regarding the patient population and PNBs are displayed in Tables 1 and 2. Overall, summary rates regarding the effectiveness and safety of PNB were technical success, 96.7%; immediate complications, 2.2%; PACU complications, 0.53%. Major adverse events reported to the IRORA registry, were three seizures [1.3 (95% CI, 0.3–3.8) per 10,000], one complete heart block [0.4 (95% CI: 0.0– 2.3) per 10,000], one retroperitoneal hematoma [1.8 (95% CI, 0.0–9.9) per 10,000], and three pneumothoraces [6.6 (95% CI, 1.3–19.4) per 10,000], one complete heart block [0.4 (95% CI, 0.0–2.3) per 10,000]. At 60 days follow-up, there were 172 all-cause postoperative neurological sequelae [8.3 (95% CI, 7.2–9.7) per 10,000]. There were no cardiac arrests.

Table 1.

Characteristics of patients^Φ

Mean (SD) or proportion
Characteristic	Other Hospitals n=13,677	Dartmouth n=3,048	p value*
Age (years)	56.8 (19.9)	56.7 (16.1)	NS
Sex (male)	0.519	0.475	<0.001
BMI	28.2 (6.7)	30.1 (7.4)	<0.001
ASA III or higher	0.405	0.345	<0.001
Preoperative neuropathy	0.064	0.077	0.001
Diabetes	0.183	0.121	<0.001
Cancer	0.066	0.052	0.006
Emergency conditons	0.302	0.006	<0.001
Ambulatory surgery	0.08	0.31	<0.001

^ΦValues calculated from unique medical record numbers

^*p value compares Dartmouth with Other Hospitals

NS = not significant

Table 2.

Characteristics of Peripheral Nerve Blocks

	Other Hospitals		Dartmouth
Characteristic	n	%	n	%
Block region
Lower extremity	10,148	52.7	2,208	55.2
Upper extremity	4,954	25.7	1,214	30.3
Trunk	3,026	15.7	515	12.9
Paravertebral/lumar plexus	1,139	5.9	66	1.7
Total	19,268	100	4,003	100
Technology used
Ultrasound alone	13,708	71.1	3,908	97.6
Ultrasound & Nerve Stimulation	3,018	15.7	81	2.0
None	1,560	8.1	14	0.4
Nerve Stimulation alone	982	5.1	0	0
Total	19,268	100	4,003	100
Needle type
Short bevel	15,787	81.9	3,703	92.5
Tuohy	3,008	15.6	289	7.2
Long bevel	473	2.5	11	0.3
Total	19,268	100	4,003	100
Block purposeΦ
Analgesia	7,087	58.4	1,251	82.7
Anesthesia & Analgesia	4,467	36.8	231	15.3
Anesthesia (surgical)	412	3.4	27	1.8
Diagnostic	163	1.3	4	0.3
Total	12,129	100	1,513
Block sedation
Sedated	10,122	52.5	3,370	84.2
Awake	4,121	21.4	171	4.3
General Anesthesia	5,025	26.1	462	11.5
Total	19,268	100	4,003	100
AdditiveΦ
None	11,245	91.8	3,419	85.4
Epinephrine	714	5.8	60	1.5
Dexamethasone	214	1.8	461	11.5
Clonidine	70	0.6	6	0.2
Other	1	0.0	57	1.42
Total	12,244	100	4,003	100
Local Anesthetic
Long action	17,561	91.1	3,980	99.4
Short acting	1,707	8.9	23	0.6
Total	19,268	100	4,003	100
Continuous block
Yes	4,859	25.2	733	18.3
No	14,409	74.8	3,270	81.7
Total	19,268	100	4,003	100
Block performed after hours
No	16,550	85.9	3,991	99.7
Yes	2,718	14.1	12	0.3
Total	19,268	100	4,003	100

^ΦVariable only tracked for limited dates

Assessment of Regional Anesthesia at Dartmouth Hitchcock Medical Center

Regional anesthesia safety

Overall, serious complications were rare at DHMC and comparable to those seen at the other IRORA institutions (Table 3). There were no seizures, cardiac arrests, or block related infections at DHMC. Dartmouth reported 20 cases (0.52%) of all cause postoperative neurological sequelae whereas the remainder of the registry reported 154 cases (0.89%), (P value 0.02). Figure 1 depicts cross-sectional comparisons between hospitals for both immediate and PACU complications.

Table 3.

Quality and Morbidity Associated with Regional Blocks

	Other Hospitals		Dartmouth		p value
	N = 19,268		N = 4,003
	n	%	n	%
Immediate EventsΨ
Local anesthetic toxicity (mild)	7	0.04	0	0	NS
Local anesthetic toxicity (severe)	3	0.02	0	0	NS
Hematoma (localized)	24	0.12	2	0.05	NS
Arterial puncture	91	0.47	11	0.27	NS
Respiratory depression	3	0.02	0	0	NS
Pneumothorax°	1	0.03	0	0	NS
Paresthesia/dysesthesia	192	1	34	0.85	NS
Wrong site block	7	0.04	0	0	NS
Intraneuronal injection	12	0.06	1	0.02	NS
Horner’s syndrome	39	0.20	11	0.27	NS
Shortness of breath (mild)	15	0.08	5	0.12	NS
Shortness of breath (severe)	4	0.02	0	0	NS
Retroperitoneal bleeding⌘	1	0.02	0	0	NS
Cardiac arrest	0	0	0	0	NS
Death	0	0	0	0	NS
Unintentional removal of catheter♦	17	0.35	4	0.55	NS
Other	32	0.17	2	0.05	NS
Overall	451	2.34	71	1.77	0.03
PACU EventsΩ*
Local anesthetic toxicity (mild)	3	0.03	0	0	NS
Hematoma	3	0.03	1	0.06	NS
Cardiac arrest	0	0	0	0	NS
Death	0	0	0	0	NS
Cardiovascular instability	14	0.14	2	0.13	NS
Shortness of breath (mild)	12	0.12	3	0.19	NS
Shortness of breath (severe)	3	0.03	0	0	NS
Hoarseness	7	0.07	2	0.13	NS
Horner’s syndrome	8	0.08	3	0.19	NS
Pneumothorax°	1	0.03	1	0.09	NS
Overall	51	0.51	12	0.77	NS
Delayed Complications*
Neurological dysfunctionΔ	154	0.89	20	0.52	NS
Block related infection	2	0.01	0	0	NS
Block Quality
Any opioid medication *	3,980	29.76	987	46.38	<0.001
VAS above 5*	2,984	23.2	732	35.1	<0.001
Block success	18,545	96.0	3,969	99.1	<0.001
Rapid discharge (≤ 30 min)*	6,147	47.97	334	15.57	<0.001

^ΨImmediate complications were defined as complications identified around time of block insertion and monitoring

^°Rate based on total blocks that could possibly cause pneumothorax (e.g. interscalene), n= 1,103 for Dartmouth and n= 3,419 for registry

^⌘Rate based on total blocks that could possibly cause retroperitoneal bleeding (e.g. lumbar plexus), n= 624 for Dartmouth and n= 4,951 for registry

^ΩPost Anesthesia Care Unit (PACU) complications occurred or were noted in the recovery room. Rates were based on a limited sample of N=9,809 for Other Hospitals and N=1,559 for Dartmouth.

^♦Based on patients with continuous blocks, n= 733 for Darmtouth and n= 4,859 for other hospitals

^*Rate based on unique surgical procedures, (e.g. a femoral and sciatic block for an ankle arthrodesis was analyzed as a single event for the ankle surgery)

^ΔDefined as all cause neurological sequelae present at 60 days postoperatively

Figure 1.

Immediate (A) and PACU (B) complication rates. Regional anesthesia procedures (number of complications per 100) ranked in order from low to high across International Registry of Regional Anesthesia member institutions. Abbreviations: DHMC, Dartmouth Hitchcock Medical Center; PACU, Post Anesthesia Care Unit. The Immediate complication rates were based on the entire data set of blocks. PACU complication rates were based on unique surgical procedures.

Regional anesthesia effectiveness

The average highest VAS score in the PACU for DHMC was 1.0 higher than for the remainder (3.0 versus 2.0, 95% CI, for difference of 0.9–1.2, P < 0.001). This corresponded to a higher proportion of patients at DHMC (35.1%) reporting a VAS score of 5 or higher compared to 23.2% of the remainder (P < 0.001). These outcomes are summarized in Table 3 and Figure 2. The occurrence of high VAS scores was demonstrated to be in statistical control, with no evidence for special cause variation (Fig. 3). DHMC experienced a low block failure rate of 1% compared to 4% for other institutions (P < 0.001). Patients at DHMC had a very low rate of early discharge from the PACU compared to other institutions (15.6% vs 48.0%, P < 0.001; Table 3). This low rate of rapid discharge is displayed in an analysis of means chart (Appendix 2). Of all 19 hospitals in the registry, DHMC had the highest proportion of patients receiving opioid analgesic medications in the PACU (46.4% vs 29.8%, Figure 2). Opioid administration was also examined using a SPC chart and found to be in statistical control (Fig. 3).

Figure 2.

Distributions of rates of selected outcomes for all International Registry of Regional Anesthesia member institutions. Any complication comprises immediate, Post Anesthesia Care Unit (PACU), and delayed. Abbreviations: DHMC, Dartmouth Hitchcock Medical Center; VAS; visual analog scale; PACU, Post Anesthesia Care Unit.

Figure 3.

Statistical process control charts (P-charts) for DHMC specific performance for percentage of patients experiencing high VAS scores (A) and percentage of patients receiving opioid analgesic medications (B). For both charts, the solid centerline represents the grand mean of the means from the individual institutions and the dashed lines represents the control limits based on approximately three standard deviations above and below the mean of the means. Abbreviations: DHMC, Dartmouth Hitchcock Medical Center; VAS; visual analog scale.

Comparisons adjusted for differences in patient populations

After adjusting for difference in patient population age, sex, BMI, and ASA score, differences remained between DHMC and other hospitals with respect to VAS scores, use of opioid medications, and early PACU discharge. Adjusted VAS scores revealed a 0.8 higher value on a 0 to 10 numeric scale compared to other hospitals (95% CI, 0.7–1.0). Compared with other institutions, patients at DHMC were approximately 80% less likely to be rapidly discharged from the PACU (OR 0.20, 95% CI, 0.18–0.23) and approximately twice as likely to receive opioids in the PACU (OR 1.90, 95% CI, 1.7–2.1). For complications, DHMC was associated with a 25% reduction in the likelihood of having at least one immediate complication (OR 0.75, 95% CI, 0.58–0.97). DHMC had a higher adjusted likelihood of experiencing at least one PACU complication, but this did not reach statistical significance (OR 1.49, 95% CI, 0.78–2.84). DHMC had a 45% reduction in the likelihood of sustaining all cause postoperative neurological sequelae (OR 0.54, 95% CI, 0.34–0.88). Figure 4 reveals the adjusted OR for opioid use, high VAS score, and delayed discharge across the IRORA member institutions.

Figure 4.

Adjusted odds ratios for the association between International Registry of Regional Anesthesia member institutions and various clinical outcomes. DHMC is the reference institution and odds ratios are adjusted for differences in patient population age, sex, body mass index, and American Society of Anesthesiologists health status. abbreviations: DHMC, Dartmouth Hitchcock Medical Center.

DISCUSSION

We have characterized the baseline performance data, regarding both safety and effectiveness, from an international clinical registry of PNB procedures, IRORA and used it to compare and contrast the performance of one reference institution, DHMC. To our knowledge, this is the first study to demonstrate the utility of a clinical registry for informing care improvement related to regional anesthesia. The overall registry performance in regards to all cause postoperative nerve injury falls within the 95% CI of previously reported literature from both DHMC and this registry.^9,11 The current analysis reports three cases of LAST in more than 20,000 blocks representing a reduced incidence than previously reported.⁹

Our reference institution, DHMC, had lower rates of immediate and all cause neurologic sequelae compared with the average of other institutions in the clinical registry. Arguably, many of the tracked complications were minor, such as venous puncture and Horner’s syndrome. When considering serious complications such as cardiac arrest, death, or severe local anesthetic toxicity (seizure), DHMC did not have any events in 4,003 PNBs. With respect to block quality, Dartmouth reported a low failure rate of <1%. Despite this low failure rate, nearly half of the patients having regional anesthesia received supplemental opioid therapy. This high utilization rate of opioids is consistent with the finding that 38% of patients reported having a VAS score of 5 or higher (out of 10). Similarly, a minority of patients were discharged from the PACU within 30 minutes of arrival.

This exercise to review our performance has allowed us to establish a baseline for on-going monthly surveillance and real-time monitoring. We feel that the overall complication rate is acceptable and current performance does not mandate quality improvement activities. However, performance around the pain experience, use of opioid analgesic medications, and discharge efficiency suggests the need to investigate further our processes of perioperative care. Additionally, given that our SPC charts (Fig. 3), demonstrate no special cause signals, such as points above the upper control limits, we can be confident that future performance will likely be similar unless changes are made. Given that PNB success is high, we suspect that our lower performance around pain control may reflect characteristics unique to our nursing and PACU processes independent of the quality of the nerve block. Additionally, we suspect the discrepancy may be partially cultural in origin. Within the registry, DHMC is only one of two hospitals in the US, which leads the world in opioid consumption.¹² We plan on conducting detailed process mapping around the microsystems of both the block room and the recovery room. The goal would be to conduct several Plan Do Study Act cycles with the aim of reducing the proportion of patients who experience high pain scores.¹³ One would logically assume that a reduction in the use of opioid analgesic medications would translate into faster patient discharge. Our preliminary investigation initiated by the results of this study suggest that ambulatory PACU nurses often administer surgeon prescribed opioid analgesic medications for patients even with no pain in an effort to provide analgesia for when the block resolves. This may be a practice to change in order to reduce side effects from the unnecessary opioid administration. The benefit of including the statistical process charts (Fig. 3) is that quality improvement interventions can be converted into a time-series analysis. For example, if an intervention improves performance with a corresponding special cause signal, one can feel confident it is the result of the intervention rather than from chance.¹⁴

The apparent discrepancy between the high reported block success rate and the high proportion of VAS scores above 5 for DHMC may reflect several scenarios such as pain outside of the block coverage (eg, sciatic distribution after femoral nerve block for total knee replacement), blocks resolving in the recovery room, over-reporting of success rates by individual operators, or even cultural differences in the reporting of pain. The reason to use multiple overlapping outcome metrics is to ensure an accurate assessment of effectiveness and safety.

Limitations

There are several limitations to our investigation. A clinical registry collects data that is limited in scope and uses defined metrics that are standardized and arguable simplistic in the context of our perioperative care pathways. This facilitates capturing information from a large number of patients. Opioid analgesic medications may be given liberally, independent of the effectiveness of the regional anesthesia; therefore, opioid use is limited as an objective measure of PNB effectiveness. Regardless, we feel that an in depth study of the perioperative pain management processes may result in improvements in clinical care. A further limitation is that some centers or individuals may have potentially under-reported complications or poor outcomes. Therefore, our results may under-estimate the true complication and poor performance rates. To minimize such under reporting IRORA has periodic site visits and data quality assessments and it should be noted that the majority of collaborating hospitals have a culture of reporting incidents.

Future direction

We plan to explore the apparent improved performance at DHMC compared with other hospitals. Examples of improved performance at DHMC are a zero rate of preventable complication such as wrong-site PNB and a reduced rate of postoperative neurological sequelae. Differences in the rate of neurological sequelae may represent unmeasured confounders, such as surgical technique or patient issues and further analysis of this outcome is beyond the scope of this study. Further improvements in the safety of regional anesthesia may be achieved by identifying both processes and structures of care that could be shared by member institutions. Often high-performing centers in clinical registries allow site visits and information sharing to promote the dissemination of high value interventions. Site visits and information sharing among members represents a more granular analysis informed by the aggregated registry data and would potentially highlight reasons for differences in clinical performance.

Conclusions

In summary, we have gained insight into the relative performance of our regional anesthesia practice by using prospective observational data collected by an international clinical registry. We have used both aggregate and time ordered data displayed through traditional and statistical process control charts. With this exercise, we have identified targets for quality improvement and further investigation. Future quality improvement activities will likely identify systems issues and culture characteristics that may not directly relate to the success or failure of the procedure.

Appendix 1: IRORA Hospitals

Hospital Name	Bed number	Teaching Status
Ballarat Hospital	220	Yes
Christchurch Hospital	650	Yes
Dartmouth Hitchcock Medical Center	388	Yes
Geelong Hospital	406	Yes
Gold Coast Hospital	500	Yes
John Flynn Hospital	326	Yes
Lismore Base Hospital	250	No
Logan Hospital	316	Yes
Mater Adult Hospital	126	Yes
Princess Alexandra Hospital	650	Yes
Royal Brisbane and Women's Hospital	929	Yes
Royal Darwin Hospital	363	Yes
St. Vincent's Private Hospital, Melbourne	500	Yes
St. Vincent’s Hospital, Melbourne	350	Yes
Tulane University Medical Center	365	Yes
University Malaya Medical Centre	1109	Yes
Waikato Hospital	600	Yes
Wanganui Hospital	250	No
Wellington Hospital	400	Yes

Appendix 2

Analysis of Means chart for the entire International Registry of Regional Anesthesia clinical registry for rapid discharge following regional anesthesia. The solid line depicts the collective grand mean of all 19-member institutions and bars indicate how much above or below the mean each institution is performing. The control limits are calculated based on the numerator and denominator and represent an alpha of 1%.