Objective. To estimate rates of lower extremity amputations (LEAs) in persons with peripheral vascular disease, diabetes mellitus, trauma, neoplasm, osteomyelitis, or emphysematous gangrene.
|Lower extremity amputation, diabetes mellitus, peripheral vascular disease, neoplasm, osteomyelitis, emphysematous gangrene, trauma, capture-recapture|
Lower extremity amputation (LEA) is a major health problem in the general population and is associated with significant morbidity, mortality, and disability. An amputation is not merely the loss of a limb; it can also mean joblessness, disability, high insurance payments, and a poor quality of life. In the United States of America and in Sweden, 50% and 32%, respectively, of all non-traumatic amputations occur in people who have diabetes (1).
In 1987, 56 000 non-traumatic LEAs were performed among people with diabetes in the United States (2). Among diabetics, the relative risk of LEA is approximately 40 times higher than in nondiabetics (3).
In the Oklahoma Indian Diabetes Study, the annual incidence of first LEAs among diabetic patients was 18.0 per 1 000 (4).
A review of the literature beginning in the 1970s (Table 1) indicates that current information on the frequency of LEAs is limited everywhere (4-20). Estimates of LEA annual incidence range from 68 to nearly 1 712 per 100 000 in an American Indian population and in Leicester, England, respectively (7, 11). In the nondiabetic population, the incidence ranges from 2 to 220 per 100 000 in Newcastle, England, and among Pima Indians, respectively (5, 12).
The prevalence of peripheral vascular disease (PVD) is higher in diabetic than nondiabetic subjects in population-based and clinic-based studies, ranging from 5.1% to 38.9%, respectively (21). PVD is among the most important reasons for LEA in individuals with and without diabetes (22).
Gangrene and osteomyelitis are two significant indications for amputations in persons with diabetes, as seen in the United Kingdom (23) and in Spain (24). The annual general incidence rate of LEA in the Netherlands has hovered between 18 and 20 per 100 000 over the last 12 years, and oncologic and traumatic reasons together have accounted for 3% of this incidence, which is among the lowest in Western Europe (25).
Examining the differences in LEA rates by groups of people within communities, especially by applying the same methodology, helps to identify and target high-risk groups (26).
The methodology described here makes use of capture-recapture (CR) modeling to estimate the number of LEAs in the municipality of Rio de Janeiro during the period from 1992 to 1994, according to a breakdown into six different etiologic categories: PVD, diabetes mellitus, osteomyelitis, trauma, emphysematous gangrene, and tumor.
For most diseases, ascertainment rates in traditional passive surveillance systems are acknowledged to be low and inconsistent. Registries (i.e., active surveillance systems), by contrast, have much higher degrees of ascertainment. Approaches for adjusting estimates to reflect ascertainment level (or census undercount) are collectively referred to as capture-recapture methods (27).
It is our belief that CR technology provides a formal means for estimating the degree of undercount of a health problem within a population and for cost-effective and timely universal monitoring of all serious disease. With these powerful tools, considerably more accurate incidence and prevalence rates will be available for comparisons between populations. CR technique allows the number of new cases of diseases in a defined population to be accurately estimated using two or more sources (27).
Table 2 shows a comparison between the traditional method and the ascertainment-corrected or CR method for determining disease rates (28-39). The examples presented illustrate how easily ascertainment correction methods can be applied to complement existing surveillance systems. This approach was useful in estimating the underreporting of pulmonary tuberculosis in Spain (28), lupus erythematosus in Denmark (35), Addison's disease in Italy (37), and AIDS cases in France (39).
Another recent study performed by the Eurodiab Ace Study Group (40) examined the onset of type 1 diabetes in 44 European centers covering a population of about 28 million children under 15 years of age.
Through the CR method, multiple sources of ascertainment were used to validate the completeness of LEA case registrations. The results confirm a very wide range of LEA incidence rates within Europe (3.2/100 000/year to 40.2/100 000/year).
The objective of this study was to estimate the incidence of LEAs in the general population of the municipality of Rio de Janeiro from 1992 to 1994, adjusting for under-ascertainment, and the incidence rates of LEAs performed on account of each of the following causes: peripheral vascular disease, diabetes mellitus, trauma, neoplasm, osteomyelitis, and gangrene.
MATERIALS AND METHODS
This retrospective study was carried out in the city of Rio de Janeiro, Brazil, from January 1, 1992 to December 31, 1994.
Over this period, the mean population of the city was 9 500 000 inhabitants. Patients submitted to a first or subsequent LEA due to a noncommunicable chronic condition (peripheral vascular disease, diabetes mellitus, neoplasm), acute infectious disease (osteomyelitis), and trauma were included in the study. For the purposes of this analysis, these entities were treated as if there were no overlap between them.
Data sources were categorized into three groups: an amputees register (S1), a limb-fitting center (S2), and a rehabilitation center (S3).
The amputee register established by the Rio de Janeiro State Health Secretariat requires that 23 of the city's hospitals, which represent more than 90% of all hospitals performing LEAs in the municipality, submit standardized information about all patients admitted. Patients with their first or subsequent LEA were identified through the amputee register.
Diagnoses were coded according to International Classification of Diseases (ICD-9) of the World Health Organization (WHO), and six groups of causes were included: peripheral vascular disease, diabetes mellitus, trauma, neoplasm, osteomyelitis, and gangrene.
Age, gender, place of residence, cause of the amputation, level of the amputation, and intervention dates were obtained from the records of each patient in each of the three data sources.
Additional information was available for specific sources. In source 1 (amputee register), such information included if the surgery was a "minor" or "major" amputation (see next paragraph), if the LEA was the first or not, and the discharge evaluation, including death. Source 2 (limb-fitting center) provided data on the period elapsed between rehabilitation and the fitting of a prosthesis, and source 3 (rehabilitation center), on the time elapsed between the surgical procedure and the beginning of rehabilitation. Death certificates were reviewed in order to identify those patients listed in source 1 who died within the first 30 days after surgery.
An LEA was defined as "minor" if it was distal to the tarso metatarsal joint, and "major" if it was performed through or proximal to the tarso metatarsal joint. Reviewing operative mortality, defined as death within the first 30 days after the amputation, the study contemplated two situations: the inclusion of all patients listed in source 1, and the inclusion of all patients except those listed as dead postoperatively in the same source.
To estimate the incidence of LEAs using CR methods, three models were employed.
Model 1: Two sources¾the amputee register (S1) and the limb-fitting center (S2) records¾were examined, but 257 patients who were listed in S1 as dead postoperatively were excluded from the analysis.
Model 2: Three sources¾the amputee register (S1), the limb-fitting center (S2), and the rehabilitation center records (S3)¾were examined. All patients listed in S1 were included in the analysis.
Model 3: Three sources¾the amputee register (S1), the limb-fitting center (S2), and the rehabilitation center records (S3)¾were examined, but the 257 patients who were listed in S1 as dead postoperatively were excluded from the analysis. This was done in order to ensure greater accuracy, since the deceased patients in S1 or S2 are not recaptured.
Capture-recapture methods were adjusted for the undercount (41-44). When only two sources of ascertainment were used, Chapman's formula was applied (45).
Cases were cross-classified, whether they were present or absent in each source. The capture-recapture method was applied to estimate the number of cases missing in any of the sources.
A log-linear model was applied to estimate the incidence of LEAs when three sources were examined, with GLIM statistical software (46). Incidence rates were calculated per 100 000 population.
The diabetes-related incidence of LEAs was estimated using as the denominator the estimated diabetic population in Rio de Janeiro (approximately 232 000) in the middle of the study period (1992-1994), according to the Brazilian census (47).
The frequency of specific amputation rates for the types of problems (peripheral vascular disease, diabetes mellitus, trauma, neoplasm, osteomyelitis, and gangrene) was calculated using the EPI INFO program, version 6.02.
The cases of LEA identified by the three sources in Rio de Janeiro were: source 1 (23 hospitals), 1 191 cases per 100 000, or a total of 934 cases, excluding those who died postoperatively; source 2 (limb-fitting center), 157 amputated cases; source 3 (rehabilitation center): 34 cases.
Thirty-nine cases were common to S1 and S2. Applying Chapman's formula, the estimated number of LEAs in Rio de Janeiro over the study period was 3 555 (95% confidence interval [95% CI]: 2 784 to 4 362) (Figure 1).
Figure 1 shows the estimated number of LEAs, excluding the 257 amputees who died postoperatively according to S1, as determined by the capture-recapture method using two sources of ascertainment (S1 and S2). The log-linear modeling approach to evaluate source dependencies is presented in Figures 2 and 3. As shown in Figure 2, the estimated number of LEAs is 5 040 when all 1 191 patients listed in S1 are included in the analysis. When cases are cross-classified between S1, S2, and S3, the estimated number of missing cases was 3 710 (Figure 2).
Figure 3 shows an estimated number of LEA of 3 954 cases, excluding those of the 257 dead patients listed in S1. When cases were cross-classified between S1, S2, and S3, the estimated number of missing cases was 2 881.
Using the estimated number of 3 954 LEAs in the numerator and the total population of the municipality of Rio de Janeiro in the middle of the 1992-1994 period in the denominator, the crude annual incidence rate of LEAs was 13.9 per 100 000 inhabitants.
The incidence rate of LEAs for the diabetic population was 180.6/100 000 diabetic patients. The routine surveillance system revealed an estimated annual incidence rate of LEAs of 5.4 per 100 000, and an estimated annual incidence rate of diabetics who underwent LEAs of 96.9 per 100 000.
Peripheral vascular disease, the most frequently observed condition as a cause of LEA, was present in 58.1% of all LEA cases, followed by diabetes mellitus, with 27.4%.
It may be argued that, in the case of LEAs, virtually 100% ascertainment should be obtained from either hospital or operating room records and that therefore the use of CR methodology is superfluous for estimating the incidence of LEAs. There are, however, several reasons that make this unlikely. They range from the incorrect coding or misplacement of patient records to the lack of access to such records, which can occur for a variety of reasons.
Data published over last three decades describe the incidence of LEAs, especially in North America, Europe and Australia. The present study is, to our knowledge, the first one in Latin America using the capture-recapture technique (48).
Standardization is urgently required in amputation data collection. Population-based numerator data currently do not indicate whether the left or right extremity was involved, or whether the LEA amputation was the first, second, or bilateral. Greater precision in data collection would enhance our understanding of the problem and improve our ability to target interventions to persons and groups at highest risk.
Capture-recapture methods allow for more accurate estimates and for monitoring diseases that are not easily identified through one or two primary sources (42-44).
Our objective in writing this report has also been to provide a standardized approach with CR techniques that will permit clinicians and researchers to estimate the incidence of LEA and to compare it across different regions.
Several methodological differences appear when reviewing the literature. Most studies measured LEA incidence in the general population (5, 11, 13, 14, 19). Only major amputations were included in the study from Sweden (15); two other studies described only the first LEA (11, 12), while two others reported subsequent LEAs as well (15, 18).
The Danish Amputation Register and Nationwide National Patient Register include all LEAs performed (25 767) during the period from 1978 through 1989. Based upon the code numbers in the WHO classification system (ICD-9), various etiologic categories (i.e., peripheral vascular disease, diabetes mellitus, neoplasm, and trauma) were extracted. However, the study failed to give incidence rates for the local population or standardized rates to their national population (18).
Data from Rio de Janeiro, Brazil, included the first or subsequent LEA. The general population of the city was used as the denominator when calculating the general LEA annual incidence rate (13.8 per 100 000). The diabetes-related incidence of LEAs was calculated with the diabetic population as the denominator (1 806 per 100 000). This revealed 13 times the risk of LEA among diabetics as compared to the nondiabetic population (20, 48).
CR methods merit greater use in Latin America within the field of epidemiology. They are useful not only for assessing the representativeness of surveillance systems, but also for identifying their inadequacies and localizing disease outbreaks.
According to the International Society for Disease Monitoring and Forecasting (45), there is a critical need for broad national and international monitoring of all serious diseases. Undercount and under-ascertainment are common to all disease-monitoring systems. Through procedures such as the capture-recapture technique, one can assess the degree of undercount and adjust disease counts and incidence rates accordingly. The methods employed in performing this study revealed findings similar to those of previous studies, demonstrating the feasibility of using the capture-recapture technique to estimate LEA incidence rates.
These findings suggest a very high LEA incidence rate in Rio de Janeiro as compared to that of other countries, such as Spain. They also identify diabetes mellitus as the second leading cause of LEAs in that city.
Acknowledgements. We would like to thank Yue-Fang Chang and Stephen Fienberg for their help with the statistical information in this paper.
1. Levin ME. Saving the diabetic foot. In: Krall LP, ed. World book of diabetes in practice. Volume 3. New York: Elsevier Science Publishers; 1988. Pp. 272-277.
2. Pecoraro RE, Reiber GE, Burgess EM. Pathways to diabetic limb amputation: basis for prevention, diabetes spectrum. Diabetes Care 1992;5:111-112.
3. Pecoraro RE, Reiber GE, Burgess EM. Pathways to diabetic limb amputation: basis for prevention. Diabetes Care 1990;13:513- 521.
4. Lee JS, Lin M, Lee VS, Russel D, Bahr C, Lee ET. Lower extremity amputations incidence, risk factors, and mortality in the Oklahoma Indian diabetes Study. Diabetes 1993;42: 876-882.
5. Nelson RG, Gohdes DM, Everart JE, Hartner JA, Zwemer FL, Pettitt DJ, et al. Lower extremity amputations in NIDDM 12-yr. follow-up study in Pima Indians. Diabetes Care 1988; 11:8-16.
6. Moss SC, Klein R, Klein BEK. The prevalence and incidence of lower extremity amputations in a diabetic population. Arch Int Med 1992;152:610-616.
7. Valway SE, Linkins RW, Gohdes DM. Epidemiology of lower extremity amputations in the Indian Health Service, 1982-1987. Diabetes Care 1993;16:349-353.
8. Rith-Najarian SJ, Valway SE, Gohdes DM. Diabetes in a northern Minnesota Chippewa tribe: prevalence and incidence of diabetes and of major complications, 1986-1988. Diabetes Care 1993;16:266-270.
9. Farrell MA, Quiggins PA, Eller JD, Owle PA, Miner KM, Wailkingstick ES. Prevalence of diabetes and its complications in the eastern band of Cherokee Indians. Diabetes Care 1993;16:253-256.
10. Lawee D, Csima A. A diabetic-related lower extremity amputation in Ontario: 1987-1988 experience. Can J Public Health 1992;83:298-302.
11. Gujral JS, McNalhy PG, O'Malley BP, Burden AC. Ethnic differences in the incidence of lower extremity amputation secondary to diabetes mellitus. Diabetic Med 1993;10:271-274.
12. Deerochanawong C, Home PD, Alberti KGMM. A survey of lower limb amputation in diabetic patients. Diabetic Med 1992;9: 942-946.
13. Siitonen OI, Niskanen LK, Laakso M, Siitonen JT, Pyorala K. Lower extremity amputations in diabetic and nondiabetic patients: a population-based study in eastern Finland. Diabetes Care 1993;16:16-20.
14. Pohjolainen T, Alaranta H. Lower limb amputations in southern Finland, 1984-1985. Prosthet Orthot Int 1988;12:9-18.
15. Kald A, Carlsson R, Nilsson E. Major amputation in a defined population: incidence, mortality and results of treatments. Br J Surg 1989;76:308-310.
16. Waugh NR. Amputations in diabetic patients: a review of rates, relative risks and resource use. Community Med 1988;10:279-288.
17. Jones LE. Lower limb amputation in three Australian states. Int Disabil Stud 1990;12: 37-40.
18. Ebskov LB. Level of lower limb amputation in relation to etiology: an epidemiological study. Prosthet Orthot Int 1992;16:163-167.
19. Duran MA, Martínez MA, Rubio JA, Maranes JP, Calle-Pascual AL. Amputaciones de miem- bros inferiores en personas con y sin diabetes mellitus en Madrid (1989-1993). Av Diabetol 1995;10:41-46.
20. Spichler D, Spichler ERS, Martins CSF, Franco LJ, Lessa I. Diabetes mellitus and lower extremity amputations (1990-1996) in Rio de Janeiro, Brazil. Diabetologia 1998;41 (Suppl.1): I-VI to 279.
21. Palumbo PJ, Melton III LJ. Peripheral vascular disease and diabetes. In: Harris MI, Cowie CC, Stern MP, Boyko EJ, Reiber GE, Bennett PH, eds. Diabetes in America, 2nd ed. Washington, D.C.: Department of Health and Human Services; 1995. (NIH Pub no. 95-1468). Pp. 401-408
22. Reiber GE, Boyko EJ, Smith DG. Lower extremity foot ulcers and amputations in diabetes. In: Harris MI, Cowie CC, Stern MP, Boyko EJ, Reiber GE, Bennett PH, eds. Diabetes in America, 2nd ed. Washington, D.C.: Department of Health and Human Services; 1995. (NIH Pub No. 95-1468). Pp. 409-428.
23. Morris AD, McAlpine R, Steinke D, Boyle DI, Ebrahim AR, Vasudev N, et al. Diabetes and lower amputations in the community: a retrospective cohort study. Diabetes Care 1998; 21(5):738-743.
24. Balsells M, Viade J, Millan M, Garcia Pascual L, DelPozo C, Anglada J. Prevalence of osteomyelitis in non-healing diabetes foot ulcers: usefulness of radiologic and scintigraphic findings. Diabetes Res Clin Pract 1997;38(2):123¾127.
25. Rommers GM, Vos LD, Groothoff JW, Schuiling CH, Eisma WH. Epidemiology of lower limb amputees in the north of the Netherlands: aetiology, discharge destination and prosthetic use. Prosthet Orthot Int 1997; 21(2): 92-99.
26. Unwin N. for LEAs Study Group. Comparing the incidence of lower extremity amputations across the world: the global lower extremity amputation study. Diabetic Med 1995;12: 14-18.
27. McCarty D, Tull ES, May CS, Kwoh CK, LaPorte RE. Ascertainment-corrected rates: applications of capture-recaptures methods. Int J Epidemiol 1993;22(3);559-565.
28. Evangelista FD, Diez BF, Hoyos SP, Adell IR, Carrascosa C, Monllor FJ. Incidence of pulmonary tuberculosis application of the capture-recapture methods. Gac Sanit 1997;11(3): 115-121.
29. Deparis X, Pascal B, Baudon D. Evalution of the completeness of the epidemiological surveillance system for malaria by the capture-recapture system in the French armies in 1994. Trop Med Int Health 1997;2(5):433-439.
30. Razzak JF, Lurby SP. Estimating death and injuries due to road traffic accident in Karachi, Pakistan, through the capture-recapture method. Int J Epidemiol 1998;Oct 27(5): 866-870.
31. Jorda Cd, Axre AJM, Navarro MF. The evaluation of epidemiological services related to brucellosis in the Autonomous Community of the Basque Country. Rev Esp Salud Publica 1997;71(2):181-187.
32. Berger B, Stenstrom A, Chang YF, Sundkvist G. The prevalence of diabetes in a Swedish population of 280 411 inhabitants: a report from the Skaraborg Diabetes Registry. Diabetes Care 1998;21(4):546-548.
33. Dockerty JD, Becroft DM, Lewis ME, Williams SM. The accuracy and completeness of childhood cancer registration in New Zealand. Cancer Causes Control 1997;8(6):857-864.
34. Muir LE, Kay BH. Aedes aegypti survival and dispersal estimated by mark-release-recapture in northern Australia. Am J Trop Med Hyg 1998;58(3):277-282.
35. Voss A, Green A, Junker P. Systemic lupus erythematosus in Denmark: clinical and epidemiological characterization of a county-based cohort. Scand J Rheumatol 1998;27(2): 98-105.
36. Corrao G, Usai P, Galatola G, Ansaldi M, Muni A, Pelli MA, et al. Estimating the incidence of coeliac disease with capture-recapture methods within four geographic areas in Italy. J Epidemiol Community Health 1996; 50(3):299-305.
37. Laureti S, Vecchi L, Santensanio F, Falormi A. Is the prevalence of Addison's disease underestimated? J Clin Endocrinol Metab 1999;84:5.
38. Craig ME, Howard NJ, Silink M, Chan A. The rising incidence of childhood type 1 diabetes in New South Wales, Australia. J Pediatr Endocrinol Metab 2000;13(4):363-372.
39. Bernillon P, Lievre L, Pillonel J, LaPorte A, Costagliola D. Record-linkage between two anonymous databases for a capture-recapture estimation of underreporting of AIDS cases: France 1990-1993. The Clinical Epidemiology Group from Centres d'Information et de Soins de l'Immunodéficience Humaine. Int J Epidemiol 2000;29(1):168-174.
40. EURODIAB ACE Study Group. Variation and trends in incidence of childhood diabetes in Europe. Lancet 2000; 355(9207):873-876.
41. LaPorte RE. Needed: universal monitoring of all serious diseases of global importance. Am J Public Health 1993;83:941-943.
42. LaPorte RE, McCarty DJ, Tull ES, Tajima N. Counting birds, bees and NCDs. Lancet 1992; 339(8 791):18-19.
43. Bruno G, LaPorte RE, Merletti F, Biggeri A, McCarty D, Pagano G. National diabetes programs: application of capture-recapture to count diabetes? Diabetes Care 1994;17(6): 548-556.
44. Needed: universal monitoring of all serious diseases of global importance. Am J Public Health 1993; 83(7):941-943.
45. International working group for disease monitoring and forecasting. Capture-recapture and multiple record system estimation I: history and theoretical development. Am J Epidemiol 1995;142(10):1047-1058.
46. GLIM Statistical Package, 3.777 Update 2, London: Royal Statistical Society; 1985.
47. Malerbi DA, Franco LJ, Brazilian Cooperative Group on the Study of Diabetes Prevalence. Multicenter study of the prevalence of diabetes mellitus and impaired glucose tolerance in the urban Brazilian population aged 30-69 yr. Diabetes Care 1992;15:1509-1516.
48. Spichler ERS, Spichler D, Franco LJ, Chang YF, LaPorte RE. Monitoring the incidence of lower extremity amputations: application of capture-recapture technology. Diabetologia 1997;40(Suppl.1);I-VI to 209.
Manuscript received 23 May 2001. Revised version accepted 7 September 2001.
Método de captura-recaptura para estimar las tasas de amputación del miembro inferior en Río de Janeiro, Brasil
Objetivos. Estimar las tasas de amputación del miembro inferior (AMI) en individuos con vasculopatías periféricas, diabetes sacarina, traumatismos, neoplasias, osteomielitis o gangrena enfisematosa.
1 Ministério da Saúde, Núcleo Estadual do Rio de Janeiro, Rio de Janeiro, Brazil. Mailing address: Ethel Rejane Stambovsky Spichler, Rua Barão de Icaraí 33/1306 Flamengo, Rio de Janeiro, 22250-110 Brazil. Phone & fax: 55-21-2267-8499; e-mail: firstname.lastname@example.org
2 Universidade Federal da Bahia, Instituto de Medicina Social, Salvador, Bahia, Brazil.
3 Secretaria Estadual de Saúde, Centro Integrado de Diabetes e Hipertensão de Ceará, Fortaleza, Ceará, Brazil.
4 Universidade de São Paulo, Faculdade de Medicina de Ribeirão Preto, Ribeirão Preto, São Paulo, Brazil.
5 Pittsburgh University, Department of Epidemiology, Pittsburgh, Pennsylvania, United States of America.