You are here

Article Text

Article menu
Socioeconomics
Original research
Cerebral aneurysm coiling: a predictive model of hospitalization cost
  1. Kimon Bekelis1,
  2. Symeon Missios2,
  3. Nicos Labropoulos3,4
  1. 1Section of Neurosurgery, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, USA
  2. 2Department of Neurosurgery, Cleveland Clinic, Cleveland, Ohio, USA
  3. 3Section of Vascular Surgery, Stony Brook University Medical Center, Stony Brook, New York, USA
  4. 4Department of Radiology, Stony Brook University Medical Center, Stony Brook, New York, USA
  1. Correspondence to Dr K Bekelis, Dartmouth-Hitchcock Medical Center, One Medical Center Dr, Lebanon, NH 03756, USA; kbekelis{at}gmail.com

Abstract

Background Several initiatives have been put in place to minimize healthcare expenditures. In new and evolving fields such as endovascular aneurysm treatment, there are limited data to support such measures. The objective of the present study was to develop and validate a predictive model of hospitalization cost after cerebral aneurysm coiling (CACo).

Methods We performed a retrospective study involving CACo patients who were registered in the Nationwide Inpatient Sample database from 2005 to 2010. The cohort underwent 1:1 randomization to create derivation and validation subsamples. Regression techniques were used for the creation of a parsimonious predictive model.

Results Of the 10 928 patients undergoing CACo, 6617 (60.5%) presented with unruptured and 4311 (39.5%) with ruptured aneurysms. Median hospitalization cost was US$35 446 (IQR $13 801–$57 091). Common drivers of cost identified in the multivariate analysis included: length of stay; number of admission diagnoses and procedures; hospital size and region; patient income; hydrocephalus; acute renal failure; and seizures. The model was validated in independent cohorts and demonstrated a final R2 value very similar to the initial model. The predicted and observed values in the validation cohort demonstrated good correlation.

Conclusions This national study identified significant drivers of hospitalization cost after CACo. The presented model can be utilized as an adjunct in the cost containment debate and the creation of data driven policies.

https://doi.org/10.1136/neurintsurg-2014-011222

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Several initiatives have been put in place to minimize healthcare cost in the USA, in light of rising expenditures.1 ,2 The Centers for Medicare and Medicaid Services is monitoring a series of measures to prevent over utilization of healthcare resources. In this context, the development of accountable care organizations and the implementation of bundled payment methods are changing the way we define value in healthcare.1 New and innovative procedures, such as cerebral aneurysm coiling (CACo), can be associated with significant cost and will be in the spotlight in this changing landscape. Estimation of the hospitalization cost for each individual CACo patient, and identification of modifiable drivers of cost, could allow physicians to understand the economic aspects of CACo and modify their practice accordingly. Future attempts at cost containment could focus on these factors, rather than follow data from other areas of medicine.

    Several studies have compared the difference in cost or charges of clipping and coiling.3–15 Some have been retrospective analyses of single institution experiences,6–8 demonstrating results with limited generalization, given their inherent selection bias. Other investigations have focused on patients from the International Subarachnoid Aneurysm Trial study14 or other international centers,3 ,10 ,12 ,13 ,15 with restricted applicability to the US healthcare market. Multicenter studies based on US data4 ,5 ,8 ,9 ,11 did not analyze modifiable drivers of cost after CACo or develop a model for cost approximation.

    The National Inpatient Sample (NIS)16 is an all-payer hospital discharge database that represents approximately 20% of all inpatient admissions to non-federal hospitals in the USA. It allows the unrestricted study of the patient population in question. Using this database, several socioeconomic variables, as well as patient and hospital level factors associated with increased cost after CACo, were identified. Based on these data, a predictive model of cost after CACo was developed and validated in an independent cohort.

    Methods

    NIS database

    All patients undergoing CACo, who were registered in the NIS16 database (Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality, Rockville, Maryland, USA) between 2005 and 2010, were included in the analysis. The NIS is an all-payer prospective hospital discharge database that represents approximately 20% of all inpatient admissions to non-federal hospitals in the USA. More information about the NIS is available at http://www.ahcpr.gov/data/hcup/nisintro.htm.

    Cohort definition

    In order to establish the cohort of patients, we used the International Classification of Disease-9-Current Modification (ICD-9-CM) codes to identify patients in the registry who underwent coiling (ICD-9-CM code 39.52 (should also have a code 88.41 and no 39.51 during the same hospitalization), 39.72, and 39.79) for ruptured (ICD-9-CM code 430, excluding 094.87 for ruptured syphilitic aneurysm, 437.4 for cerebral arteritis, 747.81 for arteriovenous malformation, 800.0–801.9, 803.0–804.9, 850.0–854.1, and 873.0–873.9 for traumatic hemorrhage, 39.53 or 92.30 for treatment diagnosis for arteriovenous malformation repair or radiosurgery), and unruptured (ICD-9-CM code 437.3) cerebral aneurysms between 2005 and 2010 (figure 1).

    Figure 1
    Figure 1

    Cohort selection for the study. NIS, Nationwide Inpatient Sample.

    Outcome variable

    The primary outcome variable was total hospitalization cost after CACo. Cost data were obtained by conversion of the hospital charges using the group average cost to charge ratio for each hospital in the database. Group average cost to charge ratio and hospital charges are available in the NIS database. All costs were adjusted to their 2010 dollar value using the national consumer price index.

    Exposure variables

    The association of outcome with pertinent exposure variables was examined in a multivariate analysis. Age was a continuous variable. Gender, race (African–American, Hispanic, Asian, or other, with Caucasian being the reference value), insurance (private insurance, self-pay, or Medicaid, with Medicare being the reference value), and income (defined as the median income based on zip code; income was divided into quartiles, with the lowest quartile being the reference value) were categorical variables.

    The patient level (see online supplementary table S1) comorbidities (categorical variables) were presentation with subarachnoid hemorrhage (SAH), diabetes mellitus, tobacco exposure, hypertension, hyperlipidemia, peripheral vascular disease, congestive heart failure, coronary artery disease, history of prior ischemic stroke, obesity, chronic renal failure, history of a transient ischemic attack event, seizure disorder, and coagulopathy. The patient level postoperative variables (categorical variables) were (see online supplementary table S1): treated hydrocephalus, hyponatremia, postoperative complications, deep vein thrombosis, pulmonary embolism, and acute renal failure (ARF). Lastly, hospitalization specific factors (continuous variables) were length of stay (LOS), number of procedures performed during the hospitalization, and number of admission diagnoses.

    The hospital characteristics used in the analysis as categorical variables included hospital region (West, South, or Midwest, with Northeast being the reference value), hospital location (urban teaching and urban non-teaching, with rural being the reference value), and hospital bed size (medium and large, with small being the reference value). More information of the definitions of the various categories of hospital characteristics can be found at http://www.hcup-us.ahrq.gov/db/vars/nis_stratum/nisnote.jsp.

    Statistical analysis

    Continuous variables are presented as mean (SD) or median (IQR) as appropriate, whereas categorical values are presented as number (percentage). Continuous variables were compared using the Student's t test or the Mann–Whitney test, as appropriate, and categorical variables were compared using the χ2 test.

    Initial analysis of the cost data revealed significant positive skewness and kurtosis, and linear regression analysis using cost resulted in a heteroskedastic variance of errors. In order to achieve normality, the data were transformed using the natural logarithm (ln) transformation. Other transformations attempted included square root, cube root, and inverse transformation. These were not eventually used because the ln transformation provided the best fit for the data. The ln transformation significantly improved the skewness and kurtosis of the cost distribution (skewness=0.132, kurtosis=0.392). Normality was also assessed using histograms and Q-Q plots. The distributions of LOS, number of admission diagnoses, and number of procedures performed also demonstrated significant positive skewness and kurtosis, and were also ln transformed before the analysis to achieve normality.

    Our cohort was then randomized (1:1 randomization, in order to create two 50% subsamples) to a derivation and a validation cohort. Subsequently, patients with missing values were removed from each cohort using listwise deletion. A parsimonious model was then developed in the derivation cohort by performing a stepwise linear regression, including all of the variables discussed previously. Dummy variables were created for non-binary categorical variables. The level of significance used for retention in the model was 0.05. No colinearity was observed by assessing tolerance and variance inflation factor. The regression diagnostics performed were the coefficient of determination (R2) and analysis of the residuals. Normality among the distribution of residuals was verified with histograms (see online supplementary figures S1 and S2), and P-P plots (see online supplementary figures S3 and S4). Further diagnostics included scatterplots of the standardized predicted values versus the standardized residuals, which revealed a random symmetric distribution of values about zero (see online supplementary figure S5), therefore suggesting a linear fit of the data.

    The model created in the derivation cohort was applied to the validation cohort, and the R2 was calculated and residual analysis was performed. The predicted values for the validation cohort were plotted against the observed values, and goodness of fit was assessed. No heteroskedasticity was observed for our cohort. For reporting purposes, we back transformed the data to demonstrate the percentage of the contribution of each variable to the cost value.

    All probability values are the results of two sided tests, and the level of significance was set at p<0.05. Statistical analyses were performed using SPSS V.20 (IBM, Armonk, New York, USA), XLSTAT V.2013.6.02 (Addinsoft, New York, New York, USA).

    Results

    Patient characteristics

    During the selected study period there were 10 928 patients (median age 56 years, 72% women) undergoing CACo who were registered in NIS. Of these patients, 6617 (60.5%) presented with unruptured and 4311 (39.5%) with ruptured aneurysms (table 1). Following 1:1 randomization and subsequent listwise deletion, derivation and validation cohorts were created. Randomization resulted in no significant differences in exposure factors between these two subgroups.

    View this table:
    Table 1

    Patient and hospital characteristics for patients undergoing cerebral aneurysm coiling

    Primary outcome

    Mean and median hospitalization costs for patients undergoing CACo for cerebral aneurysms were US$53 209 (95% CI $52 226 to $54 193) and $35 446 (IQR $13 801–$57 091), respectively (table 2). Mean and median hospitalization costs for patients undergoing CACo for unruptured aneurysms were $31 264 (95% CI $30 673 to $31 856) and $25 594 (IQR $15 679–$35 509), respectively. Finally, these values were significantly higher for patients with SAH, with mean and median hospitalization costs being $87 001 (95% CI $85 071 to $88 931) and $69 304 (IQR $36 750–$101 858), respectively (table 2).

    View this table:
    Table 2

    Inflation adjusted cost data

    Model derivation

    All statistically significant factors were included in our parsimonious model after stepwise linear regression (table 3). Hospitals in the West (30.1% more, in comparison with the Northeast) and the Midwest (11.3% more, in comparison with the Northeast), large bed size (30.9% more, in comparison with small bed size), Asians (20% more, in comparison with Caucasians), hydrocephalus (18.3% more), seizures (10.3%), ARF (19.7% more), and higher income (5.9% more for the highest income quartile, in comparison with the lowest quartile) were associated with increased hospitalization cost. A 1% increase in LOS, number of procedures, and number of admission diagnoses was associated with a 0.4%, 0.2%, and 0.1% increase in cost, respectively. In contrast, hospitals in the South (9% less, in comparison with hospitals in the Northeast) and urban non-teaching hospitals (8.5% less, in comparison with rural hospitals) were associated with decreased cost. Our model could explain a significant portion of the variance in cost with an R2 of 0.71.

    View this table:
    Table 3

    Per cent change in hospitalization cost after aneurysmal coiling for the variables included in the final predictive model

    Model validation

    The model was validated in a random cohort of patients, and the final R2 did not differ more than 5% from the initial values (0.68). There was very good association of the predicted values with the observed values in the validation cohort (figure 2) (p<0.001).

    Figure 2
    Figure 2

    Scatterplots demonstrating the association of the observed ln cost in the validation cohort and the predicted values of ln cost by the parsimonious model.

    Discussion

    In this study, utilizing the largest all-payer national database, we were able to identify several drivers of hospitalization cost after CACo. Based on these factors, we developed a predictive model of cost after CACo, and validated it in an independent cohort. Recognizing that the initial hospitalization cost is a major component of the overall economic burden of healthcare, several policies have been put in place in an attempt to limit these expenditures.17 Bundled payment methods and benchmarking of several quality and utilization metrics are part of this drive. The applicability of these national performance based policies18 in novel and constantly evolving areas, such as CACo, is still vague, given the limited literature on identifiable targets. Although some comparative cost studies for clipping and coiling have been performed, there has been no particular focus on specific drivers of hospitalization cost, or the prediction of its magnitude after CACo.

    In this context, we identified drivers of cost for coiling of cerebral aneurysms. After controlling for patient and hospital level confounders, one of the major contributors to the observed variation in cost was LOS. In addition, we quantified the contribution of LOS on the hospitalization cost after CACo, for the first time. Although LOS is a major target for cost containment, its importance has been questioned previously.19 The focus for improvement should be only on excessively lengthy hospitalizations, not justified by patient comorbidities. The comorbidities associated with increased LOS, in the setting of CACo, have been identified in prior studies,20 and should be taken into account to avoid penalizing the care of sicker patients.

    Location of the hospital was crucial in determining the cost after CACo. The effect of region on healthcare spending is widely recognized across medical specialties.21 ,22 Minimizing regional disparities could contribute to reduced spending.21 ,22 With regards to CACo, it appears that the West and Midwest were associated with significantly higher hospitalization costs in comparison with the Northeast. From a policy perspective, our study does not indicate whether it is possible to reduce spending without affecting patient outcomes. However, if the USA as a whole could safely achieve spending levels comparable with those of the lowest cost regions, significant savings could be achieved. Further research in that direction is needed.

    The contribution of several other factors was quantified. Complicated patients, with numerous admission diagnoses, and multiple procedures were associated with increased cost. In addition, disease specific factors, such as hydrocephalus, seizures, and ARF (important in the setting of iodine contrast use in the setting of coiling) were significant contributors to higher cost. The importance of disease specific drivers of cost underscores the need to study the characteristics of expenditures in subspecialty areas. Failure to recognize this by policy makers, while applying non-subspecialty specific performance measures, can penalize otherwise justifiably costly hospitalizations. Higher income was associated with higher cost, possibly secondary to the fact that this population has insurance that covers most of the charges claimed by the hospital.

    Although SAH was associated with significantly higher hospitalization costs, the observed association was mainly driven by the effect of LOS. SAH patients have lengthy hospitalizations, most times extending over 2 weeks to include monitoring during the vasospasm period. This results in significantly higher hospitalization costs. After controlling for LOS and the number of procedures in our parsimonious model, the significance of the association of SAH with cost was lost, and therefore this variable was not included in the final model.

    The proposed predictive model for hospitalization cost after CACo was created and validated in a statistically rigorous way. Particular attention was paid to normalizing the distribution of the primary outcome and the continuous exposure variables in order to minimize errors in our regression analysis. In addition, residual analysis confirmed the linear fit of the data. The diagnostics demonstrated that in both cohorts a significant portion of the cost variation could be explained by the variables included in our regression model. The model demonstrated good predictive ability in an independent validation cohort, with the predicted and observed values demonstrating good correlation.

    Although our model cannot account for the full extent of cost variation, since it is limited by the data available through NIS, this is a first step in the direction of the trend in healthcare economics at a national level. It quantifies the contribution and relative importance of several drivers of cost after CACo for the first time. The relative importance of these targets will allow the tailoring of national policies to novel procedures, such as CACo. This predictive model can be utilized as an adjunct in the cost containment debate and the creation of data driven policies. Our model can fuel further studies in the field and provide elements for prospective investigations. The latter design will additionally allow the collection of data on variables (such as structure, size, or location of the aneurysms, number of coils used, length of the procedure, and procedure specific complications) that are unavailable through NIS, permitting more accurate prediction of hospitalization cost.

    The present study has limitations common to administrative databases. First, indication bias and residual confounding could account for some of the observed associations. The 1 : 1 randomization of the cohorts and the validation of the model in an independent cohort aimed to minimize this bias. Second, several coding inaccuracies can affect our estimates, as in other studies involving the NIS. However, coding for SAH has shown nearly perfect association with medical record review.23 Third, the NIS during the years studied did not include hospitals from all states.16 However, the creation of the 20% sample is done in such a way by the Healthcare Cost and Utilization Project that the hospitals included are still diverse with respect to size, region, and academic status. Fourth, the NIS does not provide any clinical information on the structure, size, or location of the aneurysms, number of coils used, or length of the procedure, which are important factors to be considered in endovascular surgery. In the same context, the NIS lacks the granularity to identify procedure specific complications, such as intraoperative aneurysm rupture. Fifth, we were lacking disease severity in SAH patients. To the extent that this is correlated with the number of procedures and admission diagnoses, we have partially controlled for that confounder. Sixth, some data categories were not available for all patients. To avoid the introduction of further bias, we excluded those patients from any analysis. Seventh, causality is very hard to establish based on ecologic data. Our target was different however, and was focused on the identification of drivers of cost and the creation of a predictive model for it.

    Conclusions

    The NIS is a prospective all-payer hospital discharge database that contains a representative sample of all inpatient admissions to non-federal hospitals in the USA. By using this, several socioeconomic variables, as well as patient and hospital level factors associated with increased cost after CACo, were identified. Based on these data, a predictive model of cost after CACo was developed and validated in an independent cohort. Although generalization of these predictions should be done with caution, the model can be utilized as an adjunct in the cost containment debate and the creation of data driven policies. This can fuel further studies in the field and provide elements for the design of prospective investigations.

    References

    1. ↵
      1. Fisher ES,
      2. McClellan MB,
      3. Safran DG
      . Building the path to accountable care. N Engl J Med 2011;365:2445–47. doi:10.1056/NEJMp1112442
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Reschovsky JD,
      2. Hadley J,
      3. Saiontz-Martinez CB,
      4. et al
      . Following the money: factors associated with the cost of treating high-cost Medicare beneficiaries. Health Serv Res 2011;46:997–1021. doi:10.1111/j.1475-6773.2011.01242.x
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Bairstow P,
      2. Dodgson A,
      3. Linto J,
      4. et al
      . Comparison of cost and outcome of endovascular and neurosurgical procedures in the treatment of ruptured intracranial aneurysms. Australas Radiol 2002;46:249–51. doi:10.1046/j.1440-1673.2002.01053.x
      OpenUrlCrossRefPubMed
    4. ↵
      1. Brinjikji W,
      2. Kallmes DF,
      3. Lanzino G,
      4. et al
      . Hospitalization costs for endovascular and surgical treatment of ruptured aneurysms in the United States are substantially higher than Medicare payments. AJNR Am J Neuroradiol 2012;33:1037–40. doi:10.3174/ajnr.A2938
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Brinjikji W,
      2. Kallmes DF,
      3. Lanzino G,
      4. et al
      . Hospitalization costs for endovascular and surgical treatment of unruptured cerebral aneurysms in the United States are substantially higher than medicare payments. AJNR Am J Neuroradiol 2012;33:49–51. doi:10.3174/ajnr.A2739
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Frontera JA,
      2. Moatti J,
      3. de Los Reyes KM,
      4. et al
      . Safety and cost of stent-assisted coiling of unruptured intracranial aneurysms compared with coiling or clipping. J Neurointerv Surg 2014;6:65–71. doi:10.1136/neurintsurg-2012-010544
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Halkes PH,
      2. Wermer MJ,
      3. Rinkel GJ,
      4. et al
      . Direct costs of surgical clipping and endovascular coiling of unruptured intracranial aneurysms. Cerebrovasc Dis 2006;22:40–5. doi:10.1159/000092336
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Hoh BL,
      2. Chi YY,
      3. Dermott MA,
      4. et al
      . The effect of coiling versus clipping of ruptured and unruptured cerebral aneurysms on length of stay, hospital cost, hospital reimbursement, and surgeon reimbursement at the university of Florida. Neurosurgery 2009;64:614–19. doi:10.1227/01.NEU.0000340784.75352.A4
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Hoh BL,
      2. Chi YY,
      3. Lawson MF,
      4. et al
      . Length of stay and total hospital charges of clipping versus coiling for ruptured and unruptured adult cerebral aneurysms in the Nationwide Inpatient Sample database 2002 to 2006. Stroke 2010; 41:337–42. doi:10.1161/STROKEAHA.109.569269
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Javadpour M,
      2. Jain H,
      3. Wallace MC,
      4. et al
      . Analysis of cost related to clinical and angiographic outcomes of aneurysm patients enrolled in the international subarachnoid aneurysm trial in a North American setting. Neurosurgery 2005;56:886–94.
      OpenUrlPubMedWeb of Science
    11. ↵
      1. Lawson MF,
      2. Hoh BL
      . Clipping versus coiling: the total hospital cost of aneurysm treatment. World Neurosurg 2010;73:430–31. doi:10.1016/j.wneu.2010.05.024
      OpenUrlPubMedWeb of Science
    12. ↵
      1. Maud A,
      2. Lakshminarayan K,
      3. Suri MF,
      4. et al
      . Cost-effectiveness analysis of endovascular versus neurosurgical treatment for ruptured intracranial aneurysms in the United States. J Neurosurg 2009;110:880–86. doi:10.3171/2008.8.JNS0858
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Niskanen M,
      2. Koivisto T,
      3. Ronkainen A,
      4. et al
      . Resource use after subarachnoid hemorrhage: comparison between endovascular and surgical treatment. Neurosurgery 2004;54:1081–86. doi:10.1227/01.NEU.0000119350.80122.43
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Wolstenholme J,
      2. Rivero-Arias O,
      3. Gray A,
      4. et al
      . Treatment pathways, resource use, and costs of endovascular coiling versus surgical clipping after aSAH. Stroke 2008;39(1):111–19. doi:10.1161/STROKEAHA.107.482570
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Zubair Tahir M,
      2. Enam SA,
      3. Pervez Ali R,
      4. et al
      . Cost-effectiveness of clipping vs coiling of intracranial aneurysms after subarachnoid hemorrhage in a developing country—a prospective study. Surg Neurol 2009;72:355–60. doi:10.1016/j.surneu.2008.11.003
      OpenUrlCrossRefPubMed
    16. ↵
      1. Steiner C,
      2. Elixhauser A,
      3. Schnaier J
      . The healthcare cost and utilization project: an overview. Eff Clin Pract 2002;5:143–51.
      OpenUrlPubMed
    17. ↵
    18. ↵
      Centers for Medicare and Medicaid Services (CMS), HHS. Medicare and Medicaid programs: hospital outpatient prospective payment and ambulatory surgical center payment systems and quality reporting programs; Hospital Value-Based Purchasing Program; organ procurement organizations; quality improvement organizations; Electronic Health Records (EHR) Incentive Program; provider reimbursement determinations and appeals. Final rule with comment period and final rules. Fed Regist 2013;78:74825–5200.
      OpenUrlPubMed
    19. ↵
      1. Taheri PA,
      2. Butz DA,
      3. Greenfield LJ
      . Length of stay has minimal impact on the cost of hospital admission. J Am Coll Surg 2000;191:123–30. doi:10.1016/S1072-7515(00)00352-5
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Bekelis K,
      2. Missios S,
      3. Mackenzie TA,
      4. et al
      . A predictive model of outcomes during cerebral aneurysm coiling. J Neurointerv Surg 2014;6:342–8.
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Fisher ES,
      2. Wennberg DE,
      3. Stukel TA,
      4. et al
      . The implications of regional variations in Medicare spending. Part 1: the content, quality, and accessibility of care. Ann Intern Med 2003;138:273–87. doi:10.7326/0003-4819-138-4-200302180-00006
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Fisher ES,
      2. Wennberg DE,
      3. Stukel TA,
      4. et al
      . The implications of regional variations in Medicare spending. Part 2: health outcomes and satisfaction with care. Ann Intern Med 2003;138:288–98. doi:10.7326/0003-4819-138-4-200302180-00007
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Kokotailo RA,
      2. Hill MD
      . Coding of stroke and stroke risk factors using international classification of diseases, revisions 9 and 10. Stroke 2005;36:1776–17781. doi:10.1161/01.STR.0000174293.17959.a1
      OpenUrlAbstract/FREE Full Text

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • KB and SM contributed equally and are co-primary authors.

    • Contributors KB: conception and design, data analysis, manuscript preparation, and study supervision. SM: data collection and analysis, and manuscript review. NL: data interpretation, conception, study supervision, and manuscript review.

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.