International Journal for Quality in Health Care Advance Access originally published online on October 18, 2007
International Journal for Quality in Health Care 2007 19(6):368-376; doi:10.1093/intqhc/mzm044
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Published by Oxford University Press 2007
Evaluation of regional variation in total, major, and minor amputation rates in a national health-care system
1 Department of Veteran Affairs-New Jersey Health Care System, Center for Healthcare Knowledge Management, East Orange, NJ
2 University of Medicine and Dentistry of New Jersey-New Jersey Medical School, Newark, NJ
3 Bedford VA Medical Center, Center for Health Quality, Outcomes and Economic Research, Bedford, MA
4 Boston University, School of Public Health, Boston, MA
5 Rutgers University, New Brunswick, NJ
6 Birmingham VA Medical Center, Deep South Center on Effectiveness, Birmingham, AL
7 University of Alabama at Birmingham, Birmingham, AL
8 New York University School of Medicine, New York, NY
Address reprint requests to Chin-Lin Tseng, DVA-New Jersey Healthcare System, Center for Healthcare Knowledge Management, 385 Tremont Avenue, #129, East Orange, NJ 07018. Tel: +1-973-676-1000; Fax: +1-973-395-7114; E-mail: chin-lin.tseng{at}va.gov
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Background. Health-care systems need actionable information on amputation rates in order to improve foot-care delivery.
Objective. To evaluate regional variation in total, major, and minor amputation rates using individual-level data.
Methods. This was a retrospective cohort study of Veterans Health Administration users with diabetes who were Medicare enrolled between fiscal years 1998 and 2000 (10/1/1997–9/30/2000). The outcome was outlier status, based upon observed-to-expected ratios, for total, major, and minor amputations of 22 regional networks in fiscal year 2000.
Results. 331,806 patients incurred a total of 4,037 (12.2 per 1000; range 9.3–16.7 across networks) amputations in fiscal year 2000: 2,271 major amputations (6.8 per 1000; 4.7–9.1) and 1,766 minor amputations (5.3 per 1000; 3.9–7.6). All network outliers based upon the total amputation observed-to-expected ratio were also outliers based on major amputation observed-to-expected ratio. However, two of the five non-outliers based on total amputations were outliers based on major amputations. Simultaneous evaluation of major and minor amputation observed-to-expected ratios demonstrated four patterns of dual outlier status among networks: two networks had lower than expected minor and major amputation rates; two had higher than expected minor and major amputation rates; one network was lower than expected by major but higher by minor amputation rate; one was higher than expected by major but lower by minor amputation rate.
Conclusions. Simultaneous evaluation of major and minor amputation rates identifies different patterns of regional outlier status compared to total or major amputation rates alone. This strategy may facilitate targeted evaluations of health-care processes and structures.
Keywords: amputation, diabetes mellitus, Medicare, quality of health care, veterans
Lower extremity amputations seriously impact quality of life [1], life expectancy [2] and health-care costs [3] in people with diabetes. However, the major risk factors for lower extremity complications can be easily and reliably detected in primary-care settings [4, 5], and by intervening with targeted multidisciplinary care, patient outcomes are improved [5, 6]. Therefore, hospitalizations for amputations are considered an ambulatory-care-sensitive condition because effective outpatient care could prevent them [7]. Several US governmental agencies utilize total amputation rates as a key indicator of foot health on a population basis [8, 9], whereas the Organization for Economic Cooperation and Development [10] has recommended major amputation rates for this purpose. In contrast to communities, health-care systems are in a position to make improvements in foot health-care delivery for their population of enrolled individuals, if provided with actionable information.
Many health-care systems are able to obtain individual-level data to evaluate their foot health-care delivery. Individual-level data allow for more accurate definitions of people with diabetes [11] than estimated population prevalence [12], and risk adjustment for foot risk factors to minimize possible biases resulting from differences in disease severity [13]. It also allows to determine the maximal amputation level (major versus minor) per episode when multiple amputations were performed on a patient [14].
Measures that distinguish levels of maximal amputation for an individual are desirable to provide actionable information for quality improvement. Since major (below and above knee) and minor (toe and mid-foot) amputations have markedly different functional outcomes [1], preserving as much of the lower extremity as possible is an essential goal for a patient in need of limb salvage. To do so may require a minor amputation in an attempt to preserve maximal function; a study indicated that a comprehensive foot-care team decreased major but not minor amputation rates [6]. In contrast, an emphasis on coordination of foot-care surveillance by primary-care teams can result in a reduction in minor foot lesions [5] and perhaps minor amputations [15].
The Veterans Health Administration is an ideal setting to study variations in amputation rates in systems. The Veterans Health Administration is the US largest integrated health-care system [16], with functionally independent regional networks. One in five veterans who use the Veterans Health Administration has diabetes [11] and the prevention of amputations is a critical mission of the Veterans Health Administration [17]. The objective of this study is to determine whether regional variation in foot-care outcomes differed using total, major and minor amputation rates.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Data sources and study population
This study used the Diabetes Epidemiology Cohort study data, a research administrative database from the Veterans Health Administration and the Centre for Medicare and Medicaid Services for all Veterans Health Administration patients with diabetes. Diabetes was determined using a validated approach based on having two or more diabetes-specific International Classification of Diseases, ninth edition (ICD-9-CM) codes (250.xx, 357.2, 362.0, 366.41) from inpatient or outpatient physician visits (Veterans Health Administration and Medicare) over the 24-month period of fiscal years 1998 and 1999 (1 October 1997 to September 1999) [11]. Because many veteran Veterans Health Administration users (
75%) utilize both the Veterans Health Administration and Medicare for health care and risk covariates and amputation outcomes could be substantially underestimated using Veterans Health Administration data only [18], we limited our study cohort to diabetic veteran clinic users who were also enrolled in Medicare fee-for-service in fiscal years 1999 and 2000 to capture totality of data from both health-care systems. We defined the cohort and baseline adjustment variables in fiscal year 1999 (1 October 1998–30 September 1999) and measured the outcomes in fiscal year 2000 (1 October 1999–30 September 2000). This analysis included 331 806 patients who met the definition of diabetes, enrolled in both Veterans Health Administration and fee-for-service Medicare and were alive as of 30 September 1999. The VA New Jersey Health care System Institutional Review Board approved the study.
Outcome measures
We defined amputation hospitalizations as those with the ICD-9-CM procedure code for any lower extremity amputation (84.1x) in any field in either Veterans Health Administration Patient Treatment Files or Medicare Part A files [18, 19]. Minor amputations were defined as toe (84.11), transmetatarsal (84.12–84.13) and transtibial (84.14) amputations. Major amputations were defined as transtibial (84.15, 84.16) and transfemoral (84.17–84.19) amputations. Multiple procedures with the same ICD-9-CM code on the same day were considered to be a single amputation since there are no modifiers to enable identification of bilateral amputations. Different amputation codes during the same hospitalization were assigned as a single procedure at the highest level.
Risk adjustment and independent variables
We used the known risk factors for lower extremity amputations [3] and classified them into demographic variables (age, sex, race), foot risk factors (chronic infections, peripheral vascular diseases, foot deformity, prior amputation) and medical comorbidities (cardiovascular disease, congestive heart failure, stroke, any renal disease).
Demographic variables were obtained from Medicare files, where race was self-reported [11]. All comorbid conditions and foot risk factors were identified using ICD-9-CM codes present in fiscal year 1999 inpatient or outpatient Veterans Health Administration and Medicare claims data. Table 1 lists the corresponding ICD-9-CM codes for foot risk factors considered in the study. Race was excluded from the risk-adjustment model to avoid the possibility of adjusting away differences in outcomes due to differential treatment of veterans based on race. Gender was not entered in the risk-adjustment model due to the small proportion of women in the sample.
|
Multiple logistic regression models were used to model the occurrence of adverse outcomes. In the first model, we evaluated the risk of having any amputation (major or minor). In the second model, we used a multinomial logistic regression model to evaluate the risk of having a major amputation versus no amputation, as well as the risk of having a minor amputation versus no amputation. We validated the predictive performance of the models using a full model containing all independent variables using a bootstrap method [19, 20, 21]. We used the c-statistic to evaluate the discriminative ability of the model to distinguish an amputation event from a non-event.
Statistical analysis of regional network outlier status
The Veterans Health Administration was composed of 143 facilities in fiscal year 2000, organized into 22 regional networks (also known as Veteran-Integrated Service Networks) with 6–10 facilities each. Veterans could have received primary care and foot-related care at one facility, but have undergone amputations at another facility with a surgical programme in the same network. Since the totality of foot care was likely to be within the network, and because of the small number of amputations per facility, we ranked only the network amputation rates. Networks were ranked in order of their standardized (risk-adjusted) amputation ratios (observed-to-expected ratios). The observed numbers of total amputations were determined in each network. Expected numbers of amputations in each network were calculated by summing the predicted probability of amputations derived from the risk-adjustment models. With an observed-to-expected ratio of one, the number of amputations observed equals the number expected calculated from the model. Networks were identified as outliers if the 99% confidence interval [22] for the observed-to-expected ratio did not include one. Networks were classified as high outliers if observed-to-expected ratios >1 and as low outliers if observed-to-expected ratios <1. Note that the outlier identification is solely based on statistical significance tests. We compared the outlier status determined by total, major and minor amputations using a tree diagram, and depicted the relationship between the observed-to-expected ratios based on major and minor amputations in a scatter plot. We also report the Spearman's rank correlation coefficients for the different amputation outcomes.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
There were 331 806 patients included in the study. As shown in Table 2, they were largely male (98.5%) and white (79.3%); 31.6% of the study population were aged
75. They had a high prevalence of comorbid conditions: 22.0% with congestive heart failure, 46.7% with ischaemic heart disease, 3.1% with stroke and 12.4% with renal diseases. The prevalences of prior minor and major amputations were 0.5 and 0.6%, respectively. These 331 806 veterans incurred a total of 4037 amputations in either Veterans Health Administration or private sector hospitals in fiscal year 2000 (12.2 per 1000 patients; range by region 9.3–16.7). There were 2271 major amputations (6.8 per 1000 diabetes patients; range 4.7–9.1) and 1766 minor amputations (5.3 per 1000 patients; range 3.9–7.6).
|
Table 3 shows the results for total (versus none), major (versus none) and minor (versus none) amputations. Veterans who were between 50 and 74 years old and who had skin infections, prior minor amputation, peripheral vascular disease conditions, renal disease, congestive heart failure and no prior major amputation were at highest risk for amputations in fiscal year 2000. The model had an R2 of 0.18 and c-statistic of 0.80 for total amputations versus no amputation; for major amputations, the R2 was 0.17 and the c-statistic was 0.81; for minor amputations, the R2 was 0.14 and the c-statistic was 0.77. The internally validated c-statistics averaged 0.75 (SD = 0.005; range 0.74–0.76) for the total amputation model, 0.75 (SD = 0.004; range 0.75–0.77) for the major amputation model and 0.72 (SD = 0.005; range 0.71–0.74) for the minor amputation model.
|
The results of ranking and outlier status based on the risk-adjustment models are included in Table 4. To facilitate interpretation of these results, identification of regional outliers based upon total versus major amputation observed-to-expected ratios is shown in Fig. 1. All the regional outliers based upon total amputation ratios were also outliers based on major amputation ratios. However, among the five non-outliers based on total amputations, one had a higher than expected major amputation rate and one had a lower than expected major amputation rate.
|
|
Of the 11 regions determined as low outliers by major amputations, two (O, V) had less than expected minor amputation ratios, one (A) had a higher than expected minor amputation rate and the rest were non-outliers. Of the eight high outliers by major amputations, two (E, L) had higher than expected minor amputation ratios, one had a lower than expected ratio (T) and the remaining were non-outliers.
Figure 2 plots the risk-adjusted ratios for major and minor amputations by region. We divided the grid into quadrants based on the ratios >1 and <1. Seven regions (C, D, K, M, O, R, V) were in Quadrant 1 (lower than expected major and minor amputations); four (A, B, N, U) were in Quadrant 2 (lower than expected major and higher than expected minor amputations); seven (E, G, I, J, L, P, S) were in Quadrant 3 (higher than expected major and minor amputations) and four (F, H, Q, T) were in Quadrant 4 (higher than expected major and lower than expected minor amputations). Overall, two regions (J, S) were non-outliers by both measures of amputations. One or two regions were outliers for both major and minor amputations in each quadrant; these regions were O and V (Quadrant 1), A (Quadrant 2), E and L (Quadrant 3) and T (Quadrant 4).
|
The Spearman's Rank correlation coefficients were 0.90 (P < 0.001) for total versus major amputations, 0.53 (P = 0.01) for total versus minor amputations and 0.18 (P = 0.43) for major versus minor amputations.
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
We developed and applied measures of total, major and minor amputation rates in 22 regions corresponding to Veterans Health Administration networks. After risk adjustment for patient factors, our results demonstrate that the use of major amputation rate alone accurately identifies regions that are outliers by total amputation rates. However, we also show that regions that are not outliers by total amputation rates can be outliers by major and/or minor amputation rates. Therefore, simultaneously examining major and minor amputation rates provides important additional information that would not be evident if only total amputation rates were evaluated.
Our results support the Organization for Economic Cooperation and Development decision to use major amputations as a public health indicator [10], in contrast to the US Agency for Health care Research and Quality and the Department of Health and Human Services recommendations to use total amputations (8.9). They also provide empirical support for the argument that it is important to distinguish between major and minor amputations when making inferences about the quality of foot care in diabetes [14]. Several recent studies have highlighted the rates of both minor and major amputations [6, 23, 24]. Wrobel et al. [25] recently proposed using a ratio of major to minor amputation rates to assess foot-care programme performance based on the analyses of hospital discharge data.
The regional differences in major and minor amputation rates could reflect the underlying differences in structure, process and patient characteristics. The additional information provided by the simultaneous use of risk-adjusted major and minor amputation rates is potentially useful to clinical managers who evaluate the individual components of foot-care programmes: identification of patients at risk for lower extremity complications through population screening; close surveillance of those at risk with preventive foot care and a multidisciplinary approach to treatment of lower extremity complications in order to either prevent amputation or maintain the highest level of function by avoiding major amputations. The roles and responsibilities of health-care professionals may differ for each component of a foot-care programme; thus, success in one component (e.g. screening of risk factors by primary care) may not correlate with success in another component (e.g. limb salvage through multidisciplinary specialty care) [26, 27].
We present a visual representation of major and amputation rates (as depicted in Fig. 2) that may facilitate comparisons of regional amputation outcomes and can prompt a search for ways to improve these outcomes through modification of foot-care programmes. For example, regions which are low outliers by both minor and major amputation rates (regions O and V in Quadrant 1 of Fig. 2) may warrant study for identification of best practices across all components of foot-care. In contrast, regions that had low rates of major amputations but higher rates of minor amputations (region A in Quadrant 2) could be postulated to have excellent salvage teams but may need to review their surveillance processes. Regions with high rates of major and low rates of minor amputations (region T in Quadrant 4) may need to focus on their multidisciplinary foot care, whereas those with higher than expected rates of both major and minor amputations (regions L and T in Quadrant 3) may need more comprehensive evaluation.
On the basis of an evaluation of 10 Veterans Health Administration facilities in 2000–2001, we suggest that managers and clinicians could evaluate local policies and procedures [27], conduct surveys of the clinician and administrative cohesiveness of the foot-care programmes [15] and evaluate the microsystem of foot care [28]. It may also be necessary to evaluate other elements, such as the volume of surgery performed and the composition and experience of multidisciplinary foot-care teams. This approach is encouraged by the Veterans Health Administration preservation, amputation care and treatment programme at all Veterans Health Administration facilities [17]. Over a 7-year period (1999–2005), the rates of diabetes-related major and minor amputations performed in Veterans Health Administration hospitals have decreased with an increasing proportion of minor amputations [29].
This study has several advantages. Using individual-level data, we were able to reliably ascertain the population of people with diabetes, their foot risk predictor variables and amputation outcomes, avoiding likely overestimation of amputation rates associated with the use of aggregated data [30]. At the same time, complete reliance on administrative data results in some limitations. The under-coding of an important risk factor for amputations, diabetic neuropathy, made detection of this condition infeasible using available claims data [14]. Finally, our methods need to be validated in other health-care settings and other time periods.
The simultaneous evaluation of risk-adjusted major and minor amputation rates provides important additional information beyond total or major amputation rates alone and could enhance the ability of clinical and policy decision-makers to improve foot-care programmes.
|
Funding |
|---|
|
TopAbstractMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
This study was funded by a grant from the US Department of Veterans Affairs Health Services Research and Development (IIR04-204 to Dr. Tseng), a grant from US Veterans Health Administration Clinical Services Research and Development (to Dr. Pogach), and a Research Career Development Award (to Dr. Helmer).
|
Acknowledgements |
|---|
|
TopAbstractMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
This study was presented in part at the American Diabetes Association 64th Scientific Meeting, June 2004. We thank Ms. Christina Croft for her editorial assistance. The opinions expressed are solely those of the authors and do not represent the views of the Department of Veterans Affairs.
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
- Peters EJ, Childs MR, Wunderlich RP, et al. Functional status of persons with diabetes-related lower-extremity amputations. Diabetes Care (2001) 24:1799–804.
[Abstract/Free Full Text] - Boyko EJ, Ahroni JH, Smith DG, et al. Increased mortality associated with diabetic foot ulcer. Diabet Med (1996) 13:967–72.[CrossRef][Web of Science][Medline]
- Reiber GE, Boyko EJ, Smith DG. Lower extremity foot ulcers and amputations in diabetes. In: National Diabetes DataGroup (eds). In: Diabetes in America (1995) 2nd edn. Washington, DC: National Institutes of Health, NIDDK. Publ. no.: NIH 95-1468.
- Edelman D, Sanders L, Pogach L. Reproducibility and accuracy among primary care providers of a screening examination of foot ulcer risk in diabetic patients. Prev Med (1998) 27:274–8.[CrossRef][Web of Science][Medline]
- Litzelman DK, Slemenda CW, Langefeld CD, et al. Reduction of lower extremity clinical abnormalities in patients with non-insulin dependent diabetes mellitus: a randomized, controlled trial. Ann Intern Med (1993) 119:36–41.
[Abstract/Free Full Text] - McCabe CJ, Stevenson RC, Dolan AM. Evaluation of a diabetic foot screening and protection programme. Diabet Med (1998) 15:80–4.[CrossRef][Web of Science][Medline]
- Billings J, Zeitel L, Lukomnik J, et al. Impact of socioeconomic status on hospital use in New York City. Health Aff (Millwood) (1993) 12:162–73.[Abstract]
- Office of Disease Prevention and Health Promotion, U.S. Department of Health and Human Services. In: Healthy People 2010. Chapter 5, Diabetes [cited 2007 July 12]. http://www.healthypeople.gov/data/midcourse/html/focusareas/FA05TOC.htm.
- Agency for Healthcare Research and Quality. 2005 National Healthcare Quality Report. (NHQR) [cited 2007 July 12]. http://www.ahrq.gov/qual/nhqr05/nhqr05.htm.
- Nicolucci A, Greenfield S, Mattke S. Selecting indicators for the quality of diabetes care at the health systems level in OECD countries. Int J Qual Health Care (2006) 18:26–30.
[Abstract/Free Full Text] - Miller DR, Safford MM, Pogach LM. Who has diabetes? Best estimates of diabetes prevalence in the veterans health administration based on computerized patient data. Diabetes Care (2004) 27:B10–B21.
[Abstract/Free Full Text] - Shwartz M, Pekoz EA, Ash AS, et al. Do variations in disease prevalence limit the usefulness of population-based hospitalization rates for studying variations in hospital admissions? Med Care (2005) 43:4–11.[Web of Science][Medline]
- Iezzoni LI. Dimensions of risk. In: Risk Adjustment for Measuring Healthcare Outcomes—Iezzoni LI, ed. (1997) 2nd edn. Chicago, IL: Health Administration Press. 43–167.
- Jeffcoate WJ, van Houtum WH. Amputation as a marker of the quality of foot care in diabetes. Diabetologia (2004) 47:2051–8.[CrossRef][Web of Science][Medline]
- Wrobel JS, Charns MP, Diehr P, et al. The relationship between provider coordination and diabetes-related foot outcomes. Diabetes Care (2003) 26:3042–7.
[Abstract/Free Full Text] - Kizer KW, Demakis JG, Feussner JR. Reinventing VA health care: systematizing quality improvement and quality innovation. Med Care (2000) 38:I7–16.[CrossRef][Medline]
- Department of Veterans Affairs. Preservation-Amputation Care and Treatment. (Veterans Health Administration Directive 2006–050): [cited 2007 August 9]. www1.va.gov/vhapublications/ViewPublication.asp?pub_ID=1483.
- Tseng CL, Greenberg JD, Helmer D, et al. Dual system utilization affects regional variation in prevention quality indicators: the case of amputations among veterans with diabetes. Am J Manag Care (2004) 10:886–92.[Web of Science][Medline]
- Tseng CL, Rajan M, Miller DR, et al. Use of administrative data to risk adjust amputation rates in a national cohort of medicare-enrolled veterans with diabetes. Med Care (2005) 43:88–92.[Web of Science][Medline]
- Efron B. Estimating the error rate of a prediction rule: improvement on cross-validation. J Am Stat Assoc (1983) 78:316–31.[CrossRef][Web of Science]
- Steyerberg EW, Harrell FE Jr, Borsboom GJ, et al. Internal validation of predictive models: efficiency of some procedures for logistic regression analysis. J Clin Epidemiol (2001) 54:774–81.[CrossRef][Web of Science][Medline]
- Breslow NE, Day NE. Statistical Methods in Cancer Research: Volume II: The Design and Analysis of Cohort Studies (1987) Lyon: International Agency for Research on Cancer.
- Rayman G, Krishnan ST, Baker NR, et al. Are we underestimating diabetes-related lower-extremity amputation rates? Results and benefits of the first prospective study. Diabetes Care (2004) 27:1892–6.
[Abstract/Free Full Text] - Trautner C, Haastert B, Spraul M, et al. Unchanged incidence of lower-limb amputations in a German City, 1990–1998. Diabetes Care (2001) 24:855–9.
[Abstract/Free Full Text] - Wrobel JS, Robbins J, Armstrong DG. The high-low amputation ratio: a deeper insight into diabetic foot care? J Foot Ankle Surg (2006) 45:375–9.[CrossRef][Medline]
- Mayfield J, Reiber G, Sanders L, et al. American diabetes association preventive foot care in people with diabetes mellitus (technical review). Diabetes Care (1998) 21:2161–78.[Abstract]
- Pogach L, Charns MP, Wrobel JS, et al. Impact of policies and performance measurement upon the development of organizational coordinating strategies for chronic care delivery. Am J Manag Care (2004) 10:171–80.[Web of Science][Medline]
- Wrobel JS, Charns MP, Bonacker KM, et al. Diabetes-related foot care at ten veterans affairs medical centers: what predicts a successful microsystem? Jt Comm J Qual Patient Saf (2006) 32:206–13.[Medline]
- Diabetes Mellitus Interagency Coordinating Committee. National Institutes of Digestive, Diabetes and Kidney Disease: [cited 2007 July 12]. Annual Report Fiscal Year 2005 http://www.niddk.nih.gov/federal/dmicc/dmicc2005/summaries.html#vha.
- Department of Health and Human Services, Center for Disease Control and Prevention. Hospital discharge rates for nontraumatic lower extremity amputation by diabetes status—United States, 1997. MMWR Morb Mortal Wkly Rep (2001) 50:954–8.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||