Health Affairs, 25, no. 1 (2006): 148-162 doi: 10.1377/hlthaff.25.1.148 © 2006 by Project HOPE	New Online   House Health Reform Bill   Paying for Reform   Vetting AHIP's Report   HIV/AIDS Costs   Brief: Insurance Reform   HA Blog Top 10

This Article Abstract Figures Only Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (18) Citing Articles via Google Scholar Google Scholar Articles by Kahn, C. N. Articles by Van Gelder, S. Search for Related Content PubMed Articles by Kahn, C. N., III Articles by Van Gelder, S. Related Collections Hospitals Medicare Business Of Health Quality Of Care Health Spending Consumer Issues Evidence-Based Medicine

Quality

Snapshot Of Hospital Quality Reporting And Pay-For-Performance Under Medicare

PROLOGUE: Early in the new millennium, a new buzzword entered the health care lexicon: pay-for-performance, or P4P. The concept gained a strong foothold when the Centers for Medicare and Medicaid Services (CMS) announced in July 2003 a major initiative to test the use of financial incentives "to encourage hospitals to provide high quality inpatient care," according to a press release.

Known as the Premier Hospital Quality Incentive Demonstration, the project uses financial bonuses to reward hospitals that are members of Premier, a nationwide organization of nonprofit hospitals and health systems, for their performance in selected clinical areas. Rewards are based on quality measures validated extensively by the work of the Agency for Healthcare Research and Quality (AHRQ), the Joint Commission on Accreditation of Healthcare Organizations (JCAHO), the National Quality Forum (NQF), and other collaborators. As of late 2005, more than 270 hospitals nationwide were voluntarily taking part. In May 2005, Premier released results from the first four quarters showing a trend toward improved quality among all participants. Fifth-quarter results released in July 2005 show even greater improvement.

In this paper, Charles N. "Chip" Kahn III and his coauthors probe the P4P universe by examining quality performance activities taking place under the aegis of the Hospital Quality Alliance (HQA). They also examine the P4P approaches of the Premier Demonstration and the Medicare Payment Advisory Commission (MedPAC). Kahn (ckahn{at}fah.org) is president of the Federation of American Hospitals (FAH) in Washington, D.C. He served as a top health adviser on Capitol Hill before moving to the industry side as president of the Health Insurance Association of America, then taking over at the FAH in 2001. Thomas Ault is a principal at Health Policy Alternatives Inc. in Washington, D.C. Howard Isenstein is a vice president at the FAH. Lisa Potetz, director, public policy research, at the March of Dimes, in Washington, D.C., contributed as an independent consultant. Susan Van Gelder is a senior vice president at the FAH.

Abstract

 
This paper examines the impact that Medicare pay-for-performance (P4P) might have upon hospital payment. It uses the initial two quarters of a national quality database to model financial gains or losses using the Premier Hospital Quality Incentive Demonstration rules, as well as the P4P approach recommended by the Medicare Payment Advisory Commission (MedPAC). Findings reveal variation among all types of hospitals and across all measures within each of the three conditions studied: heart attack, heart failure, and pneumonia. Initially, hospitals’ financial gains and losses likely will be marginal using the Premier demonstration payment rules and somewhat larger under the MedPAC recommendations as modeled.

The HQA provided data for Hospital Compare, a U.S. Department of Health and Human Services (HHS) database that included, at the time of the analysis, seventeen measures of clinical quality among three conditions: heart attack, heart failure, and pneumonia. The database is maintained by the CMS, which receives the data voluntarily from about 4,200 hospitals, based on valid and reliable measures that have been shown to reflect quality of care.² This analysis uses the first two quarters of available data. The CMS recently released an additional two quarters of data. Preliminary analysis confirms the results in this paper. New measures will continued to be added over time. For example, in late 2006–early 2007, hospitals are scheduled to publish data about patients’ perspectives of care.

Medicare, for its part, has sought to further connect payment with performance, or at least with the reporting of certain HQA measures. As part of the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003, Congress encouraged hospitals to participate in the public reporting of quality information; those that do not report on ten measures of quality receive a 0.4 percent reduction in their annual Medicare payment update for inpatient hospital services. It should be noted, however, that hospitals were signing up to report quality scores voluntarily even before MMA was passed. As of 12 December 2003, 2,338 hospitals had signed up with the CMS to report their quality data.³ In spring 2005 the Medicare Payment Advisory Commission (MedPAC) presented a series of recommendations that would explicitly link part of Medicare’s payment to providers—including inpatient hospitals—to performance.⁴

In a further contribution to the development of pay-for-performance (P4P) for hospitals, the CMS has undertaken the Premier Hospital Quality Incentive Demonstration. This demonstration includes 268 hospitals that are members of Premier Inc., an alliance of nonprofit hospitals. The demonstration is designed to examine the effect on hospital care of financial rewards and penalties linked with performance on a set of common medical conditions. Under this demonstration, the highest-performing hospitals receive bonuses, while the lowest-performing hospitals might be subject to penalties, based on their performance on certain evidence-based quality measures for patients with heart attack, heart failure, pneumonia, coronary artery bypass graft (CABG), and hip and knee replacements. Participation in the demonstration is voluntary and began in 2003.

Notably, public policy changes often have been based on experiments like the Premier demonstration. Diagnosis-related groups (DRGs), payment methods for Medicare managed care, and changes in payment for home health, for example, were first tested in agency-sponsored demonstrations.

This study examines hospital quality and financial performance under two P4P approaches: Premier and the one envisioned within MedPAC’s spring 2005 recommendations.

Study Data And Methods

Top Study Data And Methods Results Limitations Of Analysis And... Conclusions And Implications NOTES

 
Data. The primary source of data is a public use file released in April 2005 by the CMS containing Hospital Compare data reported by each hospital. Overall, data on 4,203 hospitals are available through Hospital Compare. The quality of care provided to patients treated for the reported conditions (heart attack, pneumonia, and heart failure) clearly is important clinically. These conditions also account for 16 percent of Medicare discharges from acute care hospitals as well as 16 percent of Medicare hospital payments.⁵

The Hospital Compare data encompass the original ten quality measures for which reporting hospitals received a full inflation update under MMA as well as the seven additional measures reported voluntarily by many hospitals. The ten measures cover all patients, not solely Medicare patients, treated from January through June 2004. The seven optional measures are from the period April–June 2004. Both the clinical definitions of the quality measures and the methodology for computing composite scores follow the specifications of the Joint Commission on Accreditation of Healthcare Organizations (JCAHO).⁶

Methods. To examine performance by type of hospital, we used hospitals’ Medicare provider numbers to match the Hospital Compare quality data to characteristics contained in the fiscal year 2005 Final Rule Impact File for Medicare’s hospital prospective payment system (PPS).⁷ Data about the number of patients discharged and the level of Medicare payments for each of the three conditions come from the FY 2003 Medicare Provider Analysis and Review (MEDPAR) file.

We calculated composite scores reflecting the collection of measures included for each of the three conditions. To the extent possible, the calculation methodology adheres to the specifications of the Premier demonstration, which in turn follows the JCAHO definitions and rules that define each quality measure precisely. For example, the measure "aspirin at arrival" in the heart attack condition applies to all patients in ten identified International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) principal diagnosis codes who are not contra-indicated for aspirin and who are not described by several exclusion criteria.⁸ A hospital’s score on a measure equals the percentage of patients subject to the measure for which the hospital fulfilled the indicated action: for example, administered aspirin on arrival.

JCAHO also prescribes how scores on the individual measures are converted into composite scores for the three conditions. Because the individual quality measures are given equal weight in determining the composite score, the score is simply the percentage of instances across measures in which the hospital performed the required action (for example, gave aspirin on arrival or communicated discharge instructions) compared with the cumulative number of actions that should have been performed. For example, a composite heart attack score of 90 percent means that the hospital failed to perform 10 percent of the actions that should have been performed for heart attack patients, assuming that all actions that were performed were recorded. As in Premier, composite scores were not calculated for a hospital that did not have at least thirty patients for at least one of the measures included in the condition.⁹

Heart attack had the fewest number of hospitals with a condition score: 2,008 hospitals out of 4,203 hospitals in the Hospital Compare universe (Exhibit 1). Hospitals with a score, however, account for 83 percent of all Medicare discharges and 88 percent of all Medicare inpatient hospital payments.¹⁰ The distribution of hospitals with valid scores differs somewhat from the Hospital Compare universe. Hospitals with scores are more likely to be larger and urban, and for heart attack they also are much more likely to be teaching and tax-exempt.

The March 2005 MedPAC recommendations call for 1–2 percent of hospital payments to be set aside to create a pool for rewarding quality performance, with bonuses to be paid both for attaining a specified level of performance and for achieving improved performance over time. According to MedPAC, details of the P4P policy would be established in regulations by the HHS secretary. To model how the MedPAC recommendations might compare with Premier, the paper assumes that payments in the affected ICD-9 codes are reduced 1 percent to establish a pool and that all money in the pool is distributed based upon achieving a high level of performance; since available data are from a single time period, the paper does not model bonuses based on improvement over time.¹³

Bonus payments are modeled so that all pool dollars are expended, and bonus rates for hospitals above the ninetieth percentile are twice the bonus rates paid to hospitals between the eightieth and ninetieth percentiles. Only hospitals in the top 20 percent receive bonus payments. Although MedPAC does not recommend imposing penalties for low performance, hospitals can experience lower Medicare revenues because payments are reduced 1 percent at the outset.

Results

Top Study Data And Methods Results Limitations Of Analysis And... Conclusions And Implications NOTES

 
As measured by the three condition composite scores, performance varied among hospitals overall and by hospital type. Although fewer hospitals had scores for heart attack than for the other two conditions, the mean composite score was higher for this condition than for the others: 90.0 percent, compared with 74.4 percent for heart failure and 76.2 percent for pneumonia (Exhibit 2). In addition, the range of scores was smallest for heart attack and largest for heart failure.

View larger version (24K): [in this window] [in a new window]   EXHIBIT 3 Proportion Of Hospitals In Top 20 Percent And Bottom 20 Percent, Heart Attack Scores, By Hospital Type

Results for heart failure show a similar pattern to heart attack (Exhibit 4). Major teaching, urban, and tax-exempt hospitals stand out among the top performers, while nonteaching, rural, public, and investor-owned hospitals are more likely to be among the bottom 20 percent.

View larger version (24K): [in this window] [in a new window]   EXHIBIT 4 Proportion Of Hospitals In Top 20 Percent And Bottom 20 Percent, Heart Failure Scores, By Hospital Type

View larger version (24K): [in this window] [in a new window]   EXHIBIT 5 Proportion Of Hospitals In Top 20 Percent And Bottom 20 Percent, Pneumonia Scores, By Hospital Type

For heart attack, performance on individual measures was better and varied less than for the other conditions. The beta-blocker at discharge and aspirin at discharge measures were most important. Scores were lowest and differed most on the measure for angiotensin-converting enzyme (ACE) inhibitor for left ventricular systolic dysfunction (LVSD), but this measure involved fewer patients. Poor performance on this measure, which also affects heart failure scores, as discussed above, might be due in part to clinical controversy over the appropriateness of ACE inhibitors for these patients.¹⁵ Performance varied least on the aspirin at arrival measure.

P4P results. Our comparison of the potential impact of the MedPAC and Premier approaches for varying payment using quality performance showed that the MedPAC method, which redistributes a pool of funds from a 1 percent set-aside, has a greater dollar impact than the Premier approach, which applies bonuses and penalties of 1 percent and 2 percent to top and bottom performers.¹⁶ As Exhibits 7 and 8 show, the MedPAC approach would redistribute $140 million in payments for the three conditions. By contrast, as modeled, the Premier approach would pay almost $10 million more in bonuses than it collects in penalties.

Despite relying on the same quality measures, the two approaches would also vary in their effects by hospital type. Notably, rural hospitals would have a better result under the MedPAC approach, while urban hospitals would gain under the Premier approach. Under the budget-neutral MedPAC approach, rural hospitals would contribute a relatively small amount to the pool of funds (about 13 percent), while a relatively strong performance on the pneumonia condition would yield more than that in bonus payments. Urban hospitals, which would contribute the vast majority to the pool, don’t earn enough in bonus payments to make up those contributions. Under the Premier approach, however, bonus payments earned by urban hospitals on the heart attack and heart failure measures would offset the penalties for poor performance on pneumonia.

Limitations Of Analysis And Future Research

Top Study Data And Methods Results Limitations Of Analysis And... Conclusions And Implications NOTES

 
Given that only two quarters of quality data were available in the initial release, this analysis is inherently preliminary, and implications should be considered provisional. Additionally, because Medicare payment is tied only to hospital reporting but not to actual performance, the results presented might understate how well hospitals would perform within an incentive-based system like Premier or the withhold-and-payback approach recommended by MedPAC.

The limited time frame means that for most measures, the average number of cases upon which composite scores were computed is not large. Hospitals’ performance can be expected to improve over time as they become more accustomed to complying with the measures and reporting procedures. Given the formative stage of reporting data publicly, both performance measures and reporting procedures likely are still evolving and improving, so the extent to which hospitals may be complying with the measures but not properly documenting it is unclear. In addition, during this period certain measures were undergoing refinement, which likely resulted in both lower overall scores for these measures and increased variation in hospital performance.¹⁷ A further limitation is that results reflect only process measures and could change if patient outcomes were evaluated as well.

Future analysis of quality measures will benefit from additional data and more hospital experience with the measures and reporting requirements; it also will provide an opportunity to examine how well hospitals improve performance over time beyond this initial snapshot. Multivariate analysis also would further our understanding of which hospital characteristics contribute most to variation in hospital performance. In particular, in this univariate analysis, variation by region is difficult to interpret or to distinguish from differences by bed size or teaching status. Finally, examination of variation in patient mix and clinical practices could shed light on the strong performance of rural hospitals in caring for pneumonia patients relative to other hospitals and other conditions.

Despite current interest in using payment-based incentives to improve quality of care, considerable additional research and further demonstrations could increase their effectiveness. A demonstration, for example, might study quality improvement over several years using payment-based incentives compared with a variety of reporting-only and management improvement approaches. Some interpret early results from Premier to suggest that marginal payment incentives can be very effective in improving quality performance.¹⁸ The Premier results, however, do not distinguish the effect of payment incentives from improvements spurred by the scrutiny and public relations impact of public reporting. Demonstrations also could explore what level of bonuses and penalties are necessary to stimulate investment in quality improvement.

Conclusions And Implications

Top Study Data And Methods Results Limitations Of Analysis And... Conclusions And Implications NOTES

 
Any pay-for-performance program inherently involves a number of policy choices and judgments. Although the quality measures are evidence-based and supported by clinical science, collapsing the measures into composite scores and specifying bonus and penalty formulas require policy choices for which there is no scientific foundation. Many questions arise: Should a P4P program be redistributive of current Medicare funds, or should it be designed along the Premier model, which makes available new funds to pay bonuses but could still be budget-neutral, depending on the magnitude of hospital penalties? Should the measures be weighted equally in calculating the composite scores? What is the relative role of outcome and process measures, and how should the outcome measures be risk-adjusted? By what formula will the bonus payments or penalties be calculated?

As our simple simulation of Premier and MedPAC recommendations shows, these are not trivial questions because their answers will determine how payments are redistributed and who wins and loses. Ultimately, there is the potential that any P4P program could be seen as arbitrary and not directly linked to, or predictive of, high-quality care. More troubling, rather than creating a culture of quality, P4P could lead to distortions in reporting or misplaced quality improvement efforts driven primarily by bonus payments. Such developments could lead to the unintended consequence of hospitals narrowly focusing their interventions at the expense of broadly improving care.

Although initial hospital performance is highly variable, our analysis reveals patterns among similar hospitals and provides clear benchmarks for improving quality. By focusing on the biggest quality gaps and how they relate to possible future Medicare financial incentives, hospitals can improve care for their patients and potentially benefit financially from P4P at the same time.

Editor's Notes

 
The authors thank Christopher Hogan of Direct Research LLC for his invaluable data assistance, and Steven Speil of the Federation of American Hospitals for his help in the preparation of this paper.

NOTES

Top Study Data And Methods Results Limitations Of Analysis And... Conclusions And Implications NOTES

 

1. Other members of the Hospital Quality Alliance include the American Hospital Association, Federation of American Hospitals, American Association of Medical Colleges, Joint Commission on Accreditation of Healthcare Organizations, AARP, and National Quality Forum.
2. This database is accessible at http://www.hospitalcompare.hhs.gov.
3. American Hospital Association, "The Quality Initiative: Update on Participation" (Presentation by AHA representative at Hospital Quality Alliance principals meeting, 18 December 2003).
4. Medicare Payment Advisory Commission, Report to the Congress: Medicare Payment Policy (Washington: MedPAC, March 2005).
5. Analysis of the FY 2003 Medicare Provider Analysis and Review (MEDPAR) file.
6. Joint Commission on Accreditation of Healthcare Organizations, Specification Manual for National Implementation of Hospital Core Measures, Version 2.0, 26 April 2004, http://www.jcaho.org/pms/core+measures/information+on+final+specifications.htm (accessed 8 July 2005). The seventeen measures are as follows. For heart attack: aspirin at arrival, aspirin at discharge, ACE inhibitor for LVSD, beta blocker at discharge, beta blocker at arrival, and smoking cessation counseling. For heart failure: left ventricular assessment, ACE inhibitor for LVSD, discharge instructions, and smoking cessation counseling. For pneumonia: oxygenation assessment, pneumococcal vaccine, antibiotic timing, blood cultures, and smoking cessation counseling.
7. Centers for Medicare and Medicaid Services, "FY 2005 IPPS Final Rule Impact File," from http://www.cms.hhs.gov/providers/hipps/ippspufs.asp (accessed 24 June 2005, no longer available). Hospital characteristics are not available for 498 of the 4,203 hospitals in the quality database because they could not be matched to the Final Rule Impact File. These hospitals are displayed as "unknown" in the tables.
8. The ten principal diagnoses are 410.01, 410.11, 410.21, 410.31, 410.41, 410.51, 410.61, 410.71, 410.81, and 410.91. The exclusion criteria are patients less than eighteen years of age; transferred to another acute care hospital or federal hospital on day of arrival; received in transfer from another hospital, including another emergency department; discharged on day of arrival; expired on day of arrival; left against medical advice (AMA) on day of arrival; or have documentation in the medical record of one or more several listed reasons for not prescribing aspirin. The listed reasons for not prescribing aspirin are active bleeding on arrival or within twenty-four hours after arrival; aspirin allergy; Warfarin/Coumadin as pre-arrival medication; or other reasons documented by physician, nurse practitioner, or physician assistant for not giving aspirin within twenty-four hours before or after hospital arrival.
9. For example, consider the four quality measures summarized by the heart failure composite score. If a hospital reported data on twelve patients for the assessment of LVF, twenty-one patients for the administration of ACE inhibitor for LVSD, seventeen patients for smoking cessation counseling, and thirty patients for provision of discharge instructions, then the hospital would receive a composite score for heart failure. This is so because at least one of the measures (that is, discharge instructions) included at least thirty patients. Given that a total of 80 patient observations are involved over the four measures, the thirty observations for smoking cessation would account for 37.5 percent of the composite score (30 divided by 80); and similarly for the other measures. If the discharge instruction measure had observations for only twenty-nine patients, the hospital would not be given a composite score for heart failure.
10. Many of the hospitals below the thirty-patient threshold are in the group of 498 hospitals with unknown characteristics because they could not be matched to the provider file. Thus, hospital characteristics are unknown for only 6 of the 2,008 hospitals with composite scores for heart attack. Similarly, characteristics were not found for 48 of the 2,963 hospitals with heart failure scores and for 166 of the 3,393 hospitals with pneumonia scores.
11. Hospital characteristics are not available for 498 of the 4,203 hospitals in the quality database because they could not be matched to the FY 2005 IPPS Final Rule Impact File. These hospitals are displayed as "unknown" in the tables.
12. As noted earlier, payment data are from the FY 2003 MEDPAR file. The paper also follows the Premier demonstration’s policy to apply the bonuses and penalties to the base DRG payments after wage-index adjustment but not including indirect teaching, disproportionate-share hospital (DSH), or outlier payments.
13. As recommended by MedPAC, an additional offset of up to 1 percent might be used to create a pool for rewarding improvements in performance.
14. Discussion of these measures and additional analysis of variation in scores among hospitals was recently published. See A. Jha et al., "Care in U.S. Hospitals—The Hospital Quality Alliance Program," New England Journal of Medicine 353, no. 3 (2005): 265–274.[Abstract/Free Full Text]
15. Controversy among physicians regarding the appropriate use of ACE inhibitors versus angiotensin-receptor blockers (ARBs) led to a change in this measure. Beginning in January 2005, the use of either treatment is permitted. At the time these data were collected, many medical staff were resisting use of ACE inhibitors.
16. This is a direct result of the fact that MedPAC would redistribute 1 percent of total payments made for the three conditions to the best performers, whereas Premier applies bonuses and penalties only to the payments made to the top and bottom performers—not all hospitals—in the three conditions.
17. CMS, "Change in ACEI for LVSD measures (HF-3, AMI-3): Incorporation of ARBs," 15 November 2004, http://www.cms.hhs.gov/quality/hospital/SummaryofMeeting1.pdf (accessed 8 July 2005).
18. For example, in a 3 May 2005 press release, CMS administrator Mark McClellan stated: "These early returns demonstrate that using financial incentives to reward better quality patient care works to deliver better care and avoid costly complications for our patients. We are seeing improvements across the board, regardless of a hospital’s initial performance on the quality measures."CMS, "Medicare Pay-for-Performance Demonstration Shows Significant Quality of Care Improvement at Participating Hospitals," Press Release, 3 May 2005, http://www.cms.hhs.gov/media/press/release.asp?Counter=1441 (accessed 26 September 2005).

                      What's this?

This article has been cited by other articles:

				C. P. Roth, Y.-W. Lim, J. M. Pevnick, S. M. Asch, and E. A. McGlynn The Challenge of Measuring Quality of Care From the Electronic Health Record American Journal of Medical Quality, September 1, 2009; 24(5): 385 - 394. [Abstract] [PDF]

				A. K. Jha, E. J. Orav, A. Dobson, R. A. Book, and A. M. Epstein Measuring Efficiency: The Association Of Hospital Costs And Quality Of Care Health Aff., May 1, 2009; 28(3): 897 - 906. [Abstract] [Full Text] [PDF]

				J. B. Christianson, S. Leatherman, and K. Sutherland Lessons From Evaluations of Purchaser Pay-for-Performance Programs: A Review of the Evidence Med Care Res Rev, December 1, 2008; 65(6_suppl): 5S - 35S. [Abstract] [PDF]

				A. K. Jha, E. J. Orav, J. Zheng, and A. M. Epstein Patients' Perception of Hospital Care in the United States N. Engl. J. Med., October 30, 2008; 359(18): 1921 - 1931. [Abstract] [Full Text] [PDF]

				E. T. Kurtzman, E. M. Dawson, and J. E. Johnson The Current State of Nursing Performance Measurement, Public Reporting, and Value-Based Purchasing Policy Politics Nursing Practice, August 1, 2008; 9(3): 181 - 191. [Abstract] [PDF]

				J. B. Dimick and G. R. Upchurch Jr Measuring and Improving the Quality of Care for Abdominal Aortic Aneurysm Surgery Circulation, May 13, 2008; 117(19): 2534 - 2541. [Full Text] [PDF]

				A. K. Jha, E. J. Orav, J. Zheng, and A. M. Epstein The Characteristics And Performance Of Hospitals That Care For Elderly Hispanic Americans Health Aff., March 1, 2008; 27(2): 528 - 537. [Abstract] [Full Text] [PDF]

				J. C. Mitchiner and S. L. Hutto The Effect of Selected Hospital Characteristics on the Timeliness of Antibiotic Administration for Pneumonia American Journal of Medical Quality, August 1, 2007; 22(4): 259 - 264. [Abstract] [PDF]

				A. K. Jha, E. J. Orav, Z. Li, and A. M. Epstein The Inverse Relationship Between Mortality Rates And Performance In The Hospital Quality Alliance Measures Health Aff., July 1, 2007; 26(4): 1104 - 1110. [Abstract] [Full Text] [PDF]

				A. K. Jha, E. J. Orav, Z. Li, and A. M. Epstein Concentration and Quality of Hospitals That Care for Elderly Black Patients Arch Intern Med, June 11, 2007; 167(11): 1177 - 1182. [Abstract] [Full Text] [PDF]