Association Between Hospital Volumes and Clinical Outcomes for Patients With Nontraumatic Subarachnoid Hemorrhage
- Other version(s) of this article
-
You are viewing the most recent version of this article. Previous versions:
-
Abstract
Background
Previous studies of patients with nontraumatic subarachnoid hemorrhage (SAH) suggest better outcomes at hospitals with higher case and procedural volumes, but the shape of the volume‐outcome curve has not been defined. We sought to establish minimum volume criteria for SAH and aneurysm obliteration procedures that could be used for comprehensive stroke center certification.
Methods and Results
Data from 8512 discharges in the National Inpatient Sample (NIS) from 2010 to 2011 were analyzed using logistic regression models to evaluate the association between clinical outcomes (in‐hospital mortality and the NIS‐SAH Outcome Measure [NIS‐SOM]) and measures of hospital annual case volume (nontraumatic SAH discharges, coiling, and clipping procedures). Sensitivity and specificity analyses for the association of desirable outcomes with different volume thresholds were performed. During 8512 SAH hospitalizations, 28.7% of cases underwent clipping and 20.1% underwent coiling with rates of 21.2% for in‐hospital mortality and 38.6% for poor outcome on the NIS‐SOM. The mean (range) of SAH, coiling, and clipping annual case volumes were 30.9 (1–195), 8.7 (0–94), and 6.1 (0–69), respectively. Logistic regression demonstrated improved outcomes with increasing annual case volumes of SAH discharges and procedures for aneurysm obliteration, with attenuation of the benefit beyond 35 SAH cases/year. Analysis of sensitivity and specificity using different volume thresholds confirmed these results. Analysis of previously proposed volume thresholds, including those utilized as minimum standards for comprehensive stroke center certification, showed that hospitals with more than 35 SAH cases annually had consistently superior outcomes compared with hospitals with fewer cases, although some hospitals below this threshold had similar outcomes. The adjusted odds ratio demonstrating lower risk of poor outcomes with SAH annual case volume ≥35 compared with 20 to 34 was 0.82 for the NIS‐SOM (95% CI, 0.71–094; P=0.0054) and 0.80 (95% CI, 0.68–0.93; P=0.0055) for in‐hospital mortality.
Conclusions
Outcomes for patients with SAH improve with increasing hospital case volumes and procedure volumes, with consistently better outcomes for hospitals with more than 35 SAH cases per year.
Nonstandard Abbreviations and Acronyms
ACV | annual case volume |
AHA | American Heart Association |
CSC | comprehensive stroke center |
HCUP | Healthcare Cost and Utilization Project |
JC | Joint Commission |
NIS | Nationwide Inpatient Sample |
NIS‐SOM | NIS‐SAH Outcome Measure |
NIS‐SSS | NIS‐SAH Severity Score |
TAP | Technical Advisory Panel |
Clinical Perspective
What Is New?
Outcomes for patients with subarachnoid hemorrhage improve with increasing case and procedure volumes with consistently better outcomes with more than 35 subarachnoid hemorrhage cases per year though some lower volume centers do have good outcomes.
What Are the Clinical Implications?
Hospitals that are designated as comprehensive stroke centers should treat at least 35 subarachnoid hemorrhage cases per year.
Aneurysmal subarachnoid hemorrhage (SAH) occurs in ≈30 000 people per year in the United States and accounts for ≈5% of strokes, but represents 27% of all stroke‐related years of life lost before age 65.1, 2, 3, 4, 5 The designation of primary stroke centers along with other efforts has contributed to significant improvements in rates of intravenous thrombolysis and other process metrics and in outcomes after ischemic and hemorrhagic stroke.6, 7 In view of this, it was suggested that a population of complex stroke patients, including those with SAH, may benefit from treatment at institutions that provide an even higher level of care. The Brain Attack Coalition proposed the development of comprehensive stroke centers (CSC) to address the need for optimal care of these patients.8 Among other requirements, the Brain Attack Coalition recommended that CSCs should treat ≥20 patients with SAH and perform ≥10 craniotomies for aneurysm clipping per year.8 These volume thresholds were based on expert opinion and on administrative data demonstrating better outcomes in centers treating more aneurysmal patients with SAH, with reported thresholds ranging from 19 to 50 patients with SAH annually.9, 10, 11, 12
The 2012 American Heart Association (AHA) guidelines for aneurysmal SAH recommended that hospitals discharging fewer than 10 aneurysmal SAH annually should consider transfer of patients to hospitals treating more than 35 cases per year. The AHA and the Joint Commission (JC) subsequently initiated a CSC certification program, utilizing an annual requirement of treating ≥20 aneurysmal patients with SAH and a total of ≥15 endovascular coiling or surgical clipping procedures for aneurysms.13 These thresholds have been questioned as either arbitrary or too low. We sought to determine if there was a clear inflection point in the volume‐outcome relationship in a nationally representative data set to justify setting different minimum annual case volumes (ACV) for CSCs. We analyzed a representative sample of SAH discharges from the Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS) database and examined the relationship of outcomes to ACV for SAH and for coiling and clipping procedures.
Methods
The authors declare that all supporting data are available within the article and its online supplementary file.
Study Population
The NIS is a database of all‐payer hospital inpatient stays, including data on patient demographics, diagnoses, procedures, discharge status, and hospital characteristics.14 The NIS includes over 7 million records annually, representing more than 35 million hospitalizations.15 The NIS contains only deidentified patient data. This study is a retrospective analysis of deidentified data and was therefore exempt from institutional review board approval (institutional review board exempt status). Starting in 2012, the NIS was redesigned as a sample of discharges rather than a sample of hospitals from which all discharges were retained. Given the relatively low incidence of SAH, we used data from 2010 and 2011 in 20% of all US community‐based inpatient healthcare facilities for this analysis.
Patients were included in the analysis if they had a primary International Classification of Diseases, NinthRevision (ICD‐9) diagnosis code for nontraumatic SAH (430) and were treated at a hospital that performed at least one coiling or clipping procedure during the study period. To remove non‐aneurysmal causes of SAH, additional exclusion criteria were added including a secondary diagnosis or procedure code for arteriovenous malformation; secondary diagnosis codes indicating traumatic SAH; elective admission not originating in the emergency department or as a transfer; discharge to home after a length of stay ≤1 day; or transfer to another acute care facility (to avoid crediting a hospital with a case it did not manage). A secondary cohort was defined as patients with SAH undergoing coiling using procedure codes 39.72, 39.75, and 39.76 or clipping using codes 39.51 and 39.52. ACVs were calculated for each hospital for SAH, coiling, clipping, and total coiling+clipping.
Outcomes were (1) in‐hospital death and (2) poor outcome assessed by the NIS‐SAH Outcome Measure (NIS‐SOM), a standardized measure using variables available in the NIS and validated for use in studies of patients with SAH in the NIS by demonstrating that it correlates well with the modified Rankin score.16 Poor outcome is defined as any of: in‐hospital mortality; discharge to a nursing facility, extended care facility or hospice; placement of a tracheostomy tube; or placement of a gastrostomy tube.
Statistical Analysis
For generating national estimates of numbers of patients and hospitals, NIS sampling weights were used. All other analyses used unweighted sample data.
For summarizing event rates, hospitals were divided into quintiles of ACVs. The association of each ACV with each outcome was assessed using multivariable logistic regression models. In these models, ACVs were considered as continuous variables and were assessed for linearity of their relationship with outcomes using restricted cubic splines17; where non‐linearity was found, a piecewise linear spline was used to approximate the relationship. Models were adjusted for patient characteristics (age, sex, race, Charlson Comorbidity Index,18 weekend versus weekday arrival) and hospital characteristics (geographic region, academic status, bed size). Missing rates were low among adjustment variables (<1% for patient characteristics, except race, missing in 15%; <3% among hospital characteristics); missing values for these variables were imputed using multiple imputation. Twenty‐five imputation data sets were generated with the fully conditional specification method, which takes into account the joint distribution of all variables. Each complete data set was analyzed using standard statistical analyses, and the results from the 25 complete data sets were averaged to generate the final inferential results; this ensures that the final estimates properly reflect variability and uncertainty due to missing values.19 Additional analyses included the NIS‐SAH Severity Score (NIS‐SSS)16 as a covariate. The NIS‐SSS combines variable available in the NIS to generate a measure of SAH severity. The NIS‐SSS includes treatment with mechanical ventilation, presence of hydrocephalus, treatment of hydrocephalus, and presence of coma, stupor, cranial nerve palsies, paralysis, paraparesis, and aphasia. The NIS‐SSS correlates well with the widely used Hunt‐Hess score for grading the severity of SAH clinically. Analyses using the NIS‐SSS were considered secondary analyses because the score includes diagnoses that may develop during hospitalization and may therefore reflect not only the initial severity of the SAH but also the subsequent quality of care at the hospital, and may therefore interfere with ascertainment of ACV‐risk relationships. Odds ratios were generated from logistic regression models for each 10‐case change in ACV. Figures illustrating the relationship of ACVs to outcomes were generated using restricted cubic splines.
To address concerns that specific types of patients or sites with high event risk had a large influence on results, sensitivity analyses were carried out in subgroups of (1) hospitals with SAH ≥20, (2) hospitals with SAH ≥35, and (3) patients who were not transferred from other acute care facilities.
Because several different criteria have been proposed to use as volume standards required for CSCs expected to provide high quality care for cerebrovascular patients including those with aneurysms and other criteria, we compared outcomes at hospitals that met several different ACV standards. Hospitals were grouped according to a set of 6 pre‐specified alternative proposed minimum ACV standards that incorporate ACV for SAH discharges and specific procedures. These alternative thresholds reflect proposals considered by the JC Technical Advisory Panel (TAP) including those submitted by various professional societies (David Baker, MD, unpublished data, 2019). These additional groups were compared using the same logistic regression models as above, substituting each new criteria for the ACV measures one at a time.
The utility of different ACV cutoffs for distinguishing better‐performing hospitals was assessed by calculating sensitivity and specificity of different ACV cutoffs for SAH and total clipping and total coiling procedures (performed on both ruptured and unruptured aneurysms) for different performance categories, defined by event rates. The Youden Index, defined as sensitivity+specificity–1, was calculated for each set of cutoffs; this index ranges from 0 (a useless test) to 1 (a perfect test) and can be interpreted as the probability of making an informed decision rather than a random guess.20 Visual examination of scatter plots was also used to explore cutoffs.
Analyses were conducted using SAS version 9.4 or higher (SAS Institute, Cary NC).
Results
NIS Hospital and Patient Characteristics
The NIS data set contained 8,512 patients discharged with nontraumatic SAH in 2010 and 2011 available for review. This random sample of patients represents an estimated 42,390 nontraumatic patients with SAH nationwide during these years. Of the nontraumatic patients with SAH in the NIS, 28.7% underwent clipping and 20.1% underwent coiling. During hospitalization, 21.2% died and 38.6% had a poor outcome as assessed by the NIS‐SOM.
Among the hospitals that treated patients with SAH, mean (range) of SAH ACV, coiling ACV, and clipping ACV were 30.9 (1–195), 8.7 (0–94), and 6.1 (0–69), respectively (Figure 1). (Please see Tables S1 through S4 and Figures S1 through S7 for characteristics of data set at patient and hospital levels).
Table 1 shows rates of outcomes among hospitals grouped into quintiles by SAH ACV, coiling ACV, and clipping ACV. Rates of mortality and poor NIS‐SOM outcomes were higher among those in the lowest than those in the highest quintiles for SAH ACV, with 31.5% in‐hospital mortality and 49.7% poor outcome rates in the lowest volume group (ACV ≤5) compared to 18.7% and 35.8% in the highest volume group (ACV 49–195), respectively.
Quintile of Hospital ACV | ACV (Min–Max) | No. of Patients | NIS‐SOM Poor Outcome | In‐Hospital Death | |
---|---|---|---|---|---|
Outcomes and SAH ACV | |||||
1 | 0–5 | 143 | 49.7% | 31.5% | |
2 | 6–13 | 527 | 50.1% | 29.6% | |
3 | 13.5–24 | 995 | 44.4% | 26.8% | |
4 | 25–48 | 1783 | 39.0% | 22.0% | |
5 | 49–195 | 5064 | 35.8% | 18.7% | |
Adjusted OR (95% CI) | 0.95 (0.91–0.99) | 0.93 (0.90–0.96) | |||
Per 10 case increase up to value shown | for ACV ≤50 | for ACV ≤70 | |||
P=0.007 | P<0.001 | ||||
Outcomes and coiling ACV | |||||
1–2 | 0–0 | 970 | 49.0% | 29.4% | |
3 | 0.5–12 | 914 | 42.9% | 25.8% | |
4 | 13–35 | 2017 | 40.5% | 23.2% | |
5 | 35.5–188 | 4611 | 34.7% | 17.8% | |
Adjusted OR (95% CI) | 0.95 (0.93–0.97) | 0.92 (0.89–0.94) | |||
Per 10 case increase up to value shown | for ACV ≤65 | for ACV ≤65 | |||
P<0.001 | P<0.001 | ||||
Outcomes and clipping ACV | |||||
1 | 0–1 | 668 | 46.4% | 27.4% | |
2 | 1.5–2.5 | 399 | 38.8% | 22.3% | |
3 | 3–6 | 1163 | 41.7% | 24.8% | |
4 | 6.5–17 | 1729 | 40.7% | 24.0% | |
5 | 18–188 | 4553 | 35.9% | 18.3% | |
Adjusted OR (95% CI) | 1.00 (0.98–1.01) | 0.96 (0.95–0.98) | |||
Per 10 case increase | P=0.53 | P<0.001 |
In adjusted logistic regression models, both SAH ACV and coiling ACV showed significant nonlinear relationships with both outcomes. Increasing ACVs were associated with decreasing risk of poor outcomes up to a threshold of an ACV in the range of 50 to 70, depending on the ACV and outcome combination, with no further reductions in the risk of poor outcomes for hospitals with the highest ACV (Table 1, Figure 1). Piecewise linear splines were used to generate meaningful odds ratios for the lower part of each ACV range (Table 1), but the underlying relationship shows a more gradual change with no clear inflection point (Figure 1 and Figures S8, S9, S12, S13). Sensitivity analysis that included only hospitals with SAH ACV ≥20 showed a similar risk relationship (Table S19 and Figures S40, S41), confirming that the effect is not dependent on inclusion of hospitals with the lowest SAH ACV although the relationship becomes attenuated for higher volume hospitals, eventually leveling off. Clipping ACV has a linear relationship with mortality, with lower risk for higher volume sites, but not with NIS‐SOM poor outcome (Table 1 and Figures S24, S25). Among sites with SAH ACV ≥35, the association of coiling and clipping ACVs with mortality remained significant, as was the relationship of coiling ACV with NIS‐SOM poor outcome (Tables S27 through S29 and Figures S57 through S59, S61).
The results were similar when additional analyses were performed that (1) limited the cohort to those undergoing clipping or coiling (Tables S6, S7, and S9 and Figures S10, S11, S14, S15, S18, and S19) and excluded other patients with SAH, (2) when NIS‐SSS was added as an adjustment variable (Table S5), and (3) when ACVs for unruptured cerebral aneurysms (UCA) were considered separately (Tables S10, S11, S15, S16 and Figures S20 through S23, S32 through S35).
Alternative Volume Standards
Several organizations and existing certification programs have suggested combinations of minimum volume thresholds different from those used by the JC CSC Program of annual aneurysmal SAH cases (n≥20) and aneurysm procedures (n≥15 coiling or clipping). We compared a new requirement proposed by a TAP advising the JC and 5 alternatives. We evaluated these and focused the analyses to explore the range of 20 to 35 SAH ACV (Table 2), which has more variability in outcomes than is found when SAH ACV is >35.
NIS Sites, | National Estimates, | % NIS‐SOM | % In‐Hospital | |
---|---|---|---|---|
N=261 | N=1274 | Poor Outcome | Mortality | |
Current JC CSC standard (but not TAP) | 29 | 142 | 43.8 | 24.5 |
20–34 SAH ACV | ||||
≥15 coiling or clipping ACV | ||||
Not (≥10 clipping and ≥20 coiling) | ||||
TAP recommendations | 53 | 258 | 36.7 | 18.6 |
≥35 SAH AC | ||||
≥10 clipping ACV | ||||
≥20 coiling ACV | ||||
Alternative I | 19 | 93 | 32.1 | 20.4 |
≥35 SAH ACV | ||||
≥15 coiling or clipping per year | ||||
Not (≥10 clipping and ≥20 coiling) | ||||
Alternative II | 18 | 88 | 31.9 | 20.5 |
≥35 SAH ACV | ||||
≥30 coiling or clipping | ||||
Not (≥10 clipping and ≥20 coiling) | ||||
SAH ACV 20–34 | 49 | 240 | 43.1 | 24.6 |
SAH ACV ≥35 | 73 | 356 | 35.9 | 19.0 |
SAH ACV ≥35 but not TAP | 20 | 98 | 32.4 | 20.8 |
Hospitals within the NIS that meet the proposed TAP standards had better outcomes and lower in‐hospital mortality for patients with SAH than those meeting only the current JC standards (Table 3 and Table S30), but the differences were not significant after adjustment for patient and hospital characteristics.
NIS‐SOM Poor Outcome | In‐Hospital Death | |
---|---|---|
TAP vs current standard (but not TAP) | 0.86 (0.71–1.05) | 0.83 (0.67–1.04) |
0.13 | 0.11 | |
TAP vs alternative I (but not TAP) | 1.35 (1.15–1.59) | 0.97 (0.81–1.16) |
0.001 | 0.75 | |
TAP vs alternative II (but not TAP) | 1.37 (1.17–1.61) | 0.97 (0.81–1.17) |
0.001 | 0.78 | |
TAP vs SAH ACV ≥35 (but not TAP) | 1.35 (1.15–1.58) | 0.97 (0.81–1.16) |
0.001 | 0.71 | |
SAH ACV ≥35 vs SAH ACV 20–34 | 0.82 (0.71–0.94) | 0.80 (0.68–0.93) |
0.017 | 0.006 |
We considered two alternative sets of criteria that had the same SAH ACV requirement as the one proposed by the TAP but less restrictive procedure requirements (Table 2) and a simplified criterion with the same SAH ACV requirement but no procedure requirement. Reducing or eliminating the procedure requirements added 20 hospitals within the NIS sample; 18 of these met the procedure requirement for Alternative I and 19 of them for Alternative II (Table 2), so these alternatives all turned out to lead to nearly identical sets of facilities. Risk of NIS‐SOM poor outcome was actually higher at hospitals meeting the TAP criteria than at those meeting only the alternative standards, but there was no difference for in‐hospital mortality. Of note, the effects on NIS‐SOM outcomes lose significance or nearly do so (P≥0.05) if the NIS‐SSS was also used to adjust risk (Table S30) and may reflect more serious cases being treated at facilities that tend to do more procedures.
Finally, comparison of simplified criteria that compared hospitals with SAH ACV ≥35 to those with SAH ACV between 20 and 34 demonstrated significantly better outcomes and lower in‐hospital mortality for the higher volume hospitals (Table 3).
ACV Cutoffs
The sensitivity and specificity was calculated individually for each ACV category (SAH discharges, coiling, and clipping) for its ability to identify hospitals with lower rates of poor outcomes, defined by a range of event rates (Table 4). While minimum ACV cut‐offs with high specificity for identifying poorly performing sites exist within the range of values that have been proposed as CSC minimum volume criteria, these cut‐offs are associated with low sensitivity for identifying sites with good performance and would exclude many sites with apparently good outcomes but low volumes. The Youden Index, which combines sensitivity and specificity and ranges from 0 (worst) to 1 (best), was no higher than 0.42 for any cutoff for NIS‐SOM or 0.35 for mortality.
Outcome Measure | NIS‐SOM | In‐Hospital Mortality | ||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
% With Poor Outcome | <45% | <50% | <55% | <20% | <25% | <30% | ||||||||||||
Sens | Spec | YI | Sens | Spec | YI | Sens | Spec | YI | Sens | Spec | YI | Sens | Spec | YI | Sens | Spec | YI | |
SAH ACV cutoff | ||||||||||||||||||
≥20 | 56.3 | 66.4 | 0.23 | 58.0 | 75.9 | 0.34 | 55.1 | 83.9 | 0.39 | 52.8 | 57.4 | 0.10 | 58.4 | 68.8 | 0.27 | 55.2 | 70.1 | 0.25 |
≥35 | 40.4 | 89.1 | 0.29 | 39.7 | 95.4 | 0.35 | 35.6 | 100 | 0.36 | 39.6 | 80.0 | 0.20 | 39.6 | 87.5 | 0.27 | 36.8 | 89.7 | 0.26 |
≥50 | 27.8 | 92.7 | 0.21 | 27.6 | 97.7 | 0.25 | 24.4 | 100 | 0.24 | 28.3 | 87.1 | 0.15 | 28.2 | 92.9 | 0.21 | 26.4 | 95.4 | 0.22 |
≥65 | 19.2 | 98.2 | 0.17 | 17.8 | 100 | 0.18 | 15.1 | 100 | 0.15 | 18.9 | 92.9 | 0.12 | 18.8 | 97.3 | 0.16 | 17.2 | 98.9 | 0.16 |
Coiling ACV cutoff | ||||||||||||||||||
≥20 | 47.7 | 83.6 | 0.31 | 46.6 | 89.7 | 0.36 | 43.4 | 98.2 | 0.42 | 44.3 | 72.3 | 0.17 | 49.7 | 85.7 | 0.35 | 45.4 | 87.4 | 0.33 |
≥35 | 29.8 | 92.7 | 0.23 | 29.3 | 97.7 | 0.27 | 25.9 | 100 | 0.26 | 33.0 | 88.4 | 0.21 | 32.2 | 95.5 | 0.28 | 29.3 | 97.7 | 0.27 |
≥50 | 19.9 | 99.1 | 0.19 | 17.8 | 100 | 0.18 | 15.1 | 100 | 0.15 | 21.7 | 94.8 | 0.17 | 19.5 | 98.2 | 0.18 | 17.2 | 98.9 | 0.16 |
Clipping ACV cutoff | ||||||||||||||||||
≥10 | 41.1 | 82.7 | 0.24 | 42.0 | 90.8 | 0.33 | 37.6 | 92.9 | 0.30 | 38.7 | 74.2 | 0.13 | 42.3 | 83.9 | 0.26 | 39.7 | 86.2 | 0.26 |
≥20 | 25.2 | 93.6 | 0.19 | 25.9 | 100 | 0.26 | 22.0 | 100 | 0.22 | 24.5 | 87.7 | 0.12 | 26.8 | 95.5 | 0.22 | 24.7 | 97.7 | 0.22 |
≥25 | 21.2 | 95.5 | 0.17 | 21.3 | 100 | 0.21 | 18.0 | 100 | 0.18 | 22.6 | 91.6 | 0.14 | 22.1 | 96.4 | 0.19 | 20.1 | 97.7 | 0.18 |
≥35 | 15.2 | 97.3 | 0.13 | 14.9 | 100 | 0.15 | 12.7 | 100 | 0.13 | 17.0 | 94.8 | 0.12 | 15.4 | 97.3 | 0.13 | 14.4 | 98.9 | 0.13 |
To investigate whether greater accuracy and sensitivity could be achieved with standards combining SAH and procedural ACVs, we generated a series of scatterplots (Figures 1 and 2). Hospitals with 20 to 35 SAH cases per year but a total of more than 30 coiling cases and clipping cases performed as well as did those with more than 35 SAH cases per year on either NIS‐SOM or in‐hospital mortality as outcome measures (Figure 2). There is also a group of hospitals with >20 coiling procedures per year but fewer than 35 SAH per year that have results equivalent to hospitals with more than 35 SAH per year (Figure 1). The results of hospitals with >10 clipping and >20 coiling procedures are at most marginally better than those <10 clipping and >20 coiling procedures when the NIS‐SOM is used as the outcome measure (Figure 1A). Very few hospitals in the 20 to 34 SAH ACV range had more than 20 clippings per year, but these did have outcomes similar to hospitals with more than 35 SAH (Figure 1). Hospitals in this SAH volume range with fewer than 20 clippings per year had widely scattered distribution of outcomes (Figure 1). A criterion of a combined coiling and clipping volume >15 per year also identifies hospitals with relatively good outcomes even when the SAH ACV is below 35, but a criterion of a combined volume >30 appears to do so more reliably (Figure 2). Finally, Figures 1 and 2 suggest that the effects of volume on outcome are attenuated at ACV for SAH at the high end of the ACV range for hospitals in our sample.
Finally, we looked at the effects of the rates of transferring patients out to another acute care facility prior to coiling or clipping. Higher volume hospitals tended to have lower transfer rates, but transfer rates did not influence outcomes when hospitals with similar ACV were compared (Figure 3).
Discussion
The aim of our investigations is to provide insights into relationships between hospital ACV and outcomes to provide an evidence‐based framework for CSC designation to improve the care for patients with SAH. While our findings replicate the relationship between better outcomes and higher hospital SAH patient volume and higher procedure volume observed in prior studies, the current work has important practical implications.
Previous data used by the Brain Attack Coalition to define CSC requirements showed a 17% decrease in mortality and a 20% decrease in adverse outcome at hospitals treating ≥21 SAH/year versus hospitals treating <21.9 Another study found that 30‐day mortality rates were significantly higher in hospitals with fewer than 10 SAH/year compared to those with more than 35.10 Cowan et al found better outcomes in hospitals with more than about 30 SAH/year compared to those with fewer than 30.21
Since the original Brain Attack Coalition proposal for CSC designation,8 a number of studies have evaluated the SAH volume‐outcome relationship further.16, 22, 23, 24, 25, 26, 27, 28, 29 These studies have shown that centers that meet current CSC volume requirements have better outcomes for surgical clipping than centers that have lower volumes for patients with SAH and for those with unruptured aneurysms.28, 29 This relationship is not dichotomous, and the benefit of increasing volume may extend well beyond the threshold of 20 SAH cases and a total of 15 clipping and coiling cases. Studies have found a 21% decrease in the odds of in‐hospital mortality between high volume (12.9–94.5 SAH/year) and low volume (4–6.6 SAH/year) centers26 and demonstrated a strong relationship between SAH volume and outcome with mortality increasing from 18.7% at hospitals treating 100 SAH/year to 28.4% at hospitals treating 20 SAH/year.27 We confirmed and extended these results by demonstrating significant correlations between ACV of SAH and of coiling and clipping procedures with outcomes for patients with SAH treated at hospitals with a wide range of case volumes. Investigation of different volume cut‐offs demonstrates that outcomes are better when more stringent cut‐offs are used although some low volume centers that would be excluded by this approach appear to have outcomes equivalent to higher volume centers. Our results suggest that hospitals treating more than 35 patients with SAH per year have consistently better outcomes than those treating fewer cases, and that there is a very wide variation in outcomes in hospitals treating fewer than 20 patients per year. Many of these low volume sites have unacceptably high rates of poor outcomes. In the intermediate range of 20 to 35 ACV, features such as larger procedural volumes (more than 20 coiling procedures or more than 20 clipping procedures) appear consistently to identify hospitals with outcomes comparable to the higher volume centers. For hospitals with more than 35 SAH per year, adding procedural volume requirements actually identified hospitals with worse outcomes, but this effect was not significant after correction for hemorrhage severity and may reflect more serious cases being treated at facilities that tend to do more procedures.
Introducing volume thresholds for certification does pose a risk of creating perverse incentives, whereby hospitals may wish to keep patients who might benefit from transfer. In developing standards for CSCs, it may be desirable to allow hospitals that do not have an SAH ACV of 35 to be CSCs if they have an adequate number of coiling and/or clipping cases or meet outcome standards similar to those achieved by higher volume centers. Some of these low volume hospitals may have proceduralists and neuro‐intensivists they share with larger volume centers in view of the similar outcomes reported at low and high volume hospitals affiliated with the same medical school.12
Another option for CSC standards would be to require hospitals to achieve a certain percentage of good outcomes as measured by the NIS‐SOM or by in‐hospital mortality regardless of annual case volumes. Using outcomes directly to set standards avoids the problem of excluding hospitals that have low volumes but good outcomes, but opens the process to concerns about proper risk adjustment, or improving outcomes by transferring out patients expected to do poorly to other acute centers or to nursing or hospice facilities to lower in‐hospital mortality. These issues could be dealt with by including patients transferred out in the statistics of the transferring hospital (in addition to including them in the statistics of receiving acute care hospitals). There is much debate about the proper methods for risk adjustment, and the use of the NIS‐SSS as a risk adjustment variable has not been validated. It may be challenging to identify and administer the most optimal thresholds for defining certification, and often criteria default to volumes as they are easily measured, have high rates of inter‐rater reliability and are difficult to manipulate. It will be important to ensure that low volume hospitals in regions of the country with no alternative access have a method for recognition as an appropriate site for care if they have consistently good outcomes and meet all of the other requirements.
Limitations
Our study has several important limitations. The data are drawn from the NIS which is an administrative data set, and therefore the data available are inherently limited. There are no data assessing initial clinical severity directly, which likely contributes strongly to outcomes. In secondary analyses, we did attempt to correct for severity using the NIS‐SSS which does correlate with the Hunt‐Hess measure of clinical severity but incorporates variables related to subsequent treatment so it may itself be influenced by the quality of treatment. In any case, controlling for the NIS‐SSS did not change the results of the primary models that we investigated although it did reduce the significance of related to some of the alternative volume standards that have been proposed.
The sampling methods in the NIS may also have impacted the findings. Another limitation is that we did not have data available about the volumes of individual surgeons, which is another factor that likely influences outcomes. In particular, one explanation for the finding of some low volume centers with good outcomes may be that they have a single proceduralist who does most or all of the procedures and therefore has a relatively high volume of procedures compared to other hospitals that may have more proceduralists who each do fewer cases. In addition, the diagnostic categories in the NIS did not distinguish between aneurysmal and nonaneurysmal SAH, which are clinically distinct with the former having a worse prognosis and the latter not needing coiling or clipping procedures. In addition, clipping and coiling rates were relatively lower than might be expected at 28.7% and 20.1%, respectively; this likely reflects inclusion in the population we identified of patients who had unrecognized trauma, sinus thrombosis, amyloid‐related hemorrhages, and other nonaneurysmal causes of hemorrhage. In this regard, secondary analyses limited to patients who underwent coiling or clipping did not change the main results significantly. Lastly, while the relationship between volume and outcome in our study extends beyond 35 SAH cases per year, it is attenuated at higher volume levels. There are hospitals in other countries where aneurysm care is more centralized with higher case volumes; our results cannot address the association of outcomes with very high ACV since few US centers had more than 100 SAH cases/year and none had more than 200. In this regard, McNeill et al found a 24% decrease in mortality for patients with SAH with every 100 additional cases per year in a survey of British centers treating between 50 and 367 SAH per year.24 This is an important point to consider in deciding about how centralized the care of aneurysmal patients with SAH should be. As noted above, our sample included few sites with more than 100 patients, so we cannot make definite conclusions about how outcomes may change at very high volumes, though the effects of increased volume that we observed appear to become weaker at the higher end of the volumes that were present in our sample.
In summary, our data suggest that increasing the minimum ACV for SAH to 35 is warranted as a requirement for CSC to maximize the percentage of CSCs likely to have low rates of poor outcomes. It is unclear the extent to which individual versus combined measures of procedural volume add discriminating information to distinguish high performing centers, but all CSCs should be able to safely and effectively perform surgical clipping when necessary despite the growing trend toward endovascular coiling as the preferred method for treatment.
Sources of Funding
The American Heart Association funded this study. It receives some of the fees paid to the JC for PSC and CSC certification.
Disclosures
Dr Leifer has served on the AHA Hospital Accreditation Science Committee (HASC) and is on the New York State Department of Health Stroke Physician Advisory Panel. Dr Fonarow receives research funding from the Patient Centered Outcome Research Institute, serves on the AHA HASC, and is employed by the University of California Regents which have a patent on endovascular devices. Drs Baker and Williams are employed by the JC, Mr Schoeberl by the AHA and Dr Suter formerly by the AHA. Dr Schwamm serves as chair of the AHA HASC and Quality Oversight Committees and as a member of the JC Technical Advisory Panel for CSCs. The remaining authors have no disclosures to report.
Footnotes
The results of this work were presented in preliminary form at the International Stroke Conference, February 18‐22, 2020, Los Angeles, CA.
Supplementary Material for this article is available at https://www.ahajournals.org/doi/suppl/10.1161/JAHA.120.018373
For Sources of Funding and Disclosures, see page 12.
References
- 1 Johnston SC, Selvin S, Gress DR. The burden, trends, and demographics of mortality from subarachnoid hemorrhage. Neurology. 1998; 50:1413–1418. DOI: 10.1212/WNL.50.5.1413.CrossrefMedlineGoogle Scholar
- 2 Stepanova M, Venkatesan C, Altaweel L, Mishra A, Younossi ZM. Recent trends in inpatient mortality and resource utilization for patients with stroke in the United States: 2005–2009. J Stroke Cerebrovasc Dis. 2013; 22:491–499. DOI: 10.1016/j.jstrokecerebrovasdis.2013.03.005.CrossrefMedlineGoogle Scholar
- 3 Connolly ES, Rabinstein AA, Carhuapoma JR, Derdeyn CP, Dion J, Higashida RT, Hoh BL, Kirkness CJ, Naidech AM, Ogilvy CS, et al. Guidelines for the management of aneurysmal subarachnoid hemorrhage: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2012; 43:1711–1737. DOI: 10.1161/STR.0b013e3182587839.LinkGoogle Scholar
- 4 Nieuwkamp DJ, Setz LE, Algra A, Linn FH, de Rooij NK, Rinkel GJ. Changes in case fatality of aneurysmal subarachnoid haemorrhage over time, according to age, sex, and region: a meta‐analysis. Lancet Neurol. 2009; 8:635–642. DOI: 10.1016/S1474‐4422(09)70126‐7.CrossrefMedlineGoogle Scholar
- 5 D'Souza S. Aneurysmal subarachnoid hemorrhage. J Neurosurg Anesthesiol. 2015; 27:222–240. DOI: 10.1097/ANA.0000000000000130.CrossrefMedlineGoogle Scholar
- 6 Xian Y, Holloway RG, Chan PS, Noyes K, Shah MN, Ting HH, Chappel AR, Peterson ED, Friedman B. Association between stroke center hospitalization for acute ischemic stroke and mortality. JAMA. 2011; 305:373–380. DOI: 10.1001/jama.2011.22.CrossrefMedlineGoogle Scholar
- 7 Lichtman JH, Jones SB, Leifheit‐Limson EC, Wang Y, Goldstein LB. 30‐day mortality and readmission after hemorrhagic stroke among medicare beneficiaries in Joint Commission primary stroke center‐certified and noncertified hospitals. Stroke. 2011; 42:3387–3391. DOI: 10.1161/STROKEAHA.111.622613.LinkGoogle Scholar
- 8 Alberts MJ, Latchaw RE, Selman WR, Shephard T, Hadley MN, Brass LM, Koroshetz W, Marler JR, Booss J, Zorowitz RD, et al. Recommendations for comprehensive stroke centers: a consensus statement from the Brain Attack Coalition. Stroke. 2005; 36:1597–1616. DOI: 10.1161/01.STR.0000170622.07210.b4.LinkGoogle Scholar
- 9 Bardach NS, Zhao S, Gress DR, Lawton MT, Johnston SC. Association between subarachnoid hemorrhage outcomes and number of cases treated at California hospitals. Stroke. 2002; 33:1851–1856. DOI: 10.1161/01.STR.0000019126.43079.7B.LinkGoogle Scholar
- 10 Cross DT, Tirschwell DL, Clark MA, Tuden D, Derdeyn CP, Moran CJ, Dacey RG. Mortality rates after subarachnoid hemorrhage: variations according to hospital case volume in 18 states. J Neurosurg. 2003; 99:810–817. DOI: 10.3171/jns.2003.99.5.0810.CrossrefMedlineGoogle Scholar
- 11 Solomon RA, Mayer SA, Tarmey JJ. Relationship between the volume of craniotomies for cerebral aneurysm performed at New York state hospitals and in‐hospital mortality. Stroke. 1996; 27:13–17. DOI: 10.1161/01.STR.27.1.13.LinkGoogle Scholar
- 12 Chang TR, Kowalski RG, Carhuapoma JR, Tamargo RJ, Naval NS. Impact of case volume on aneurysmal subarachnoid hemorrhage outcomes. J Crit Care. 2015; 30:469–472. DOI: 10.1016/j.jcrc.2015.01.007.CrossrefMedlineGoogle Scholar
- 13 Quick Guide: Comprehensive Stroke Center (CSC) Certification. Oakbrook Terrace, IL: The Joint Commission. https://www.jointcommission.org/‐/media/tjc/documents/accred‐and‐cert/certification/certification‐by‐setting/stroke/comp‐stroke‐quick‐guide‐2018‐print.pdf. Accessed April 5, 2021.Google Scholar
- 14 Healthcare Cost and Utilization Project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality. Available at: https://www.ahrq.gov/data/hcup/index.html. Accessed April 5, 2021.Google Scholar
- 15 HCUP NIS Overview. Rockville, MD: Agency for Healthcare Research and Quality. https://hcup‐us.ahrq.gov/nisoverview.jsp. Accessed April 5, 2021.Google Scholar
- 16 Washington CW, Derdeyn CP, Dacey RG, Dhar R, Zipfel GJ. Analysis of subarachnoid hemorrhage using the Nationwide Inpatient Sample: the NIS‐SAH Severity Score and Outcome Measure. J Neurosurg. 2014; 121:482–489. DOI: 10.3171/2014.4.JNS131100.CrossrefMedlineGoogle Scholar
- 17 Harrell FE. Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York: Springer‐Verlag New York, Inc; 2001.CrossrefGoogle Scholar
- 18 Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987; 40:373–383. DOI: 10.1016/0021‐9681(87)90171‐8.CrossrefMedlineGoogle Scholar
- 19 Liu Y, De A. Multiple imputation by fully conditional specification for dealing with missing data in a large epidemiologic study. Int J Stat Med Res. 2015; 4:287–295. DOI: 10.6000/1929‐6029.2015.04.03.7.CrossrefMedlineGoogle Scholar
- 20 Youden WJ. Index for rating diagnostic tests. Cancer. 1950; 3:32–35. DOI: 10.1002/1097‐0142(1950)3:1<32:AID‐CNCR2820030106>3.0.CO;2‐3.CrossrefMedlineGoogle Scholar
- 21 Cowan JA, Dimick JB, Wainess RM, Upchurch GR, Thompson BG. Outcomes after cerebral aneurysm clip occlusion in the United States: the need for evidence‐based hospital referral. J Neurosurg. 2003; 99:947–952. DOI: 10.3171/jns.2003.99.6.0947.CrossrefMedlineGoogle Scholar
- 22 Grigoryan M, Chaudhry SA, Hassan AE, Suri FK, Qureshi AI. Neurointerventional procedural volume per hospital in United States: implications for comprehensive stroke center designation. Stroke. 2012; 43:1309–1314. DOI: 10.1161/STROKEAHA.111.636076.LinkGoogle Scholar
- 23 Nuno M, Patil CG, Lyden P, Drazin D. The effect of transfer and hospital volume in subarachnoid hemorrhage patients. Neurocrit Care. 2012; 17:312–323. DOI: 10.1007/s12028‐012‐9740‐y.CrossrefMedlineGoogle Scholar
- 24 McNeill L, English SW, Borg N, Matta BF, Menon DK. Effects of institutional caseload of subarachnoid hemorrhage on mortality: a secondary analysis of administrative data. Stroke. 2013; 44:647–652. DOI: 10.1161/STROKEAHA.112.681254.LinkGoogle Scholar
- 25 Boogaarts HD, van Amerongen MJ, de Vries J, Westert GP, Verbeek AL, Grotenhuis JA, Bartels RH. Caseload as a factor for outcome in aneurysmal subarachnoid hemorrhage: a systematic review and meta‐analysis. J Neurosurg. 2014; 120:605–611. DOI: 10.3171/2013.9.JNS13640.CrossrefMedlineGoogle Scholar
- 26 Prabhakaran S, Fonarow GC, Smith EE, Liang L, Xian Y, Neely M, Peterson ED, Schwamm LH. Hospital case volume is associated with mortality in patients hospitalized with subarachnoid hemorrhage. Neurosurgery. 2014; 75:500–508. DOI: 10.1227/NEU.0000000000000475.CrossrefMedlineGoogle Scholar
- 27 Pandey AS, Gemmete JJ, Wilson TJ, Chaudhary N, Thompson BG, Morgenstern LB, Burke JF. High subarachnoid hemorrhage patient volume associated with lower mortality and better outcomes. Neurosurgery. 2015; 77:462–470; discussion 470. DOI: 10.1227/NEU.0000000000000850.CrossrefMedlineGoogle Scholar
- 28 Washington CW, Taylor IL, Dambrino RJ, Clarl PR, Zipfel GJ. Relationship between patient safety indicator events and comprehensive stroke center volume status in the treatment of unruptured cerebral aneurysms. J Neurosurg. 2018; 129:471–479. DOI: 10.3171/2017.5.JNS162778.CrossrefMedlineGoogle Scholar
- 29 Kurogi R, Kada A, Ogasawara K, Kitazono T, Sakai N, Hashimoto Y, Shiokawa Y, Miyachi S, Matsumaru Y, Iwama T, et al. Effects of case volume and comprehensive stroke center capabilities on patient outcomes of clipping and coiling for subarachnoid hemorrhage. J Neurosurg. 2021; 134:929–939. DOI: 10.3171/2019.12.JNS192584.CrossrefGoogle Scholar
eLetters(0)
eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.
Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.