Diabetes Home

The Economics of Diabetes Mellitus: An Annotated Bibliography

Historical

This webpage is archived for historical purposes and is no longer being maintained or updated.

TYPES OF INTERVENTION

Secondary Intervention

Abstracts 3–12

Screening and Diagnosis for Type 2 Diabetes

TITLE: Comparison of glycosylated Hemoglobin and Fasting Plasma Glucose with Two-Hour Post-Load Plasma Glucose in the Detection of Diabetes Mellitus. Simon, D.; Coignet, M.C.; Thibult, N.; Senan, C.; Eschwege, E. American Journal of Epidemiology. 122(4): 589-593. October 1985.

OBJECTIVE: To compare the use of a glycosylated hemoglobin (HbA_1C) test alone with a combination of that test and a plasma glucose measurement in a diabetes screening program.

CATEGORY: Secondary intervention.

Type of Study: Patient screening.
Methodology: Statistical analysis.
Perspective: Health care system.

CONCLUSION: Using the combination of an HbA_1C measurement and a fasting plasma glucose measurement increases the specificity and predictive value of a positive diagnosis of diabetes over that obtained by glycosylated hemoglobin alone.

RECOMMENDATION: Longitudinal surveys should be performed to assess the validity of HbA_1C as a measurement tool for diabetes screening.

ABSTRACT: From September 15, 1981, to April 1, 1984, 333 outpatients were recruited from the screening diabetes center of the Hôtel-Dieu Hospital in Paris. Patients filled out a questionnaire; underwent a physical exam; had blood drawn for fasting plasma glucose, HbA_1C, cholesterol, and triglyceride measurements; and took an oral glucose tolerance test with a 75-g glucose load. With 2-hour plasma glucose values as a reference, sensitivities of the fasting plasma glucose measurement, HbA_1C, and a combination of fasting plasma glucose and glycosylated hemoglobin were, respectively, 52 percent, 60 percent, and 40 percent; the specificities were 98.7 percent, 90.9 percent, and 99.4 percent; and the predictive values for a positive diagnosis were 76.5 percent, 34.9 percent, and 83.3 percent. Taking into account the economic and psychosociologic implications of a diagnosis of diabetes, it is better for a screening test for asymptomatic diabetes to have a high degree of specificity and a high predictive value for a positive diagnosis than to have good sensitivity but poor specificity. As an oral glucose tolerance test is time-consuming for patients and nurses, measurements of HbA_1C by chromatography and of fasting plasma glucose appear to be less expensive (about 5 versus 10 U.S. dollars). Longitudinal surveys are needed to assess the validity of HbA_1C as a tool for diabetes screening. 1 table, 32 references.

TITLE: Comparison of Screening Tests for Non-Insulin-Dependent Diabetes Mellitus. Hanson, R.L.; Nelson, R.G.; McCance, D.R.; Beart, J.A.; Charles, M.A.; Pettitt, D.J.; Knowler, W.C. Archives of Internal Medicine. 153(18): 2133-2140. September 27, 1993.

OBJECTIVE: To compare four screening tests for type 2 diabetes.

CATEGORY: Secondary intervention.

Type of Study: Population screening.
Methodology: Statistical analysis.
Perspective: Health care system.

CONCLUSION: Fasting plasma glucose was the best test for screening, but glycated hemoglobin and quantitative urine glucose also provided high specificity with sensitivities approximately 80 percent or higher.

RECOMMENDATION: The choice of a particular screening test should depend on assay cost, convenience, and availability.

ABSTRACT: The authors compared the sensitivity of measuring quantitative urine glucose, dipstick urine glucose, fasting plasma glucose, and glycated hemoglobin to screen for diabetes in Pima Indians, a tribe at high risk for type 2 diabetes. Fasting plasma glucose concentrations and glycated hemoglobin (HbA₁ or HbA_1C) were compared in 2,092 fasting participants; glycated hemoglobin, quantitative glycosuria, and dipstick glycosuria were compared in 237 nonfasting participants. Among nonfasting participants, at specificities of 98 or 99 percent, sensitivities for detecting diabetes were 80.6 percent for quantitative glycosuria, 64.3 percent for dipstick glycosuria, and 92.9 percent for HbA₁. For detecting diabetes with severe hyperglycemia, sensitivities for the 3 tests ranged from 85.0 percent (HbA₁) to 96.0 percent (quantitative glycosuria), with similar specificities. Among fasting participants, at a specificity of 98.3 percent the sensitivity of fasting plasma glucose was 88.0 percent; at similar specificities, HbA₁ had a sensitivity of 78.8 percent and HbA_1C, 80.3 percent. For detecting diabetes with severe hyperglycemia, all 3 tests had a sensitivity of 94.6 percent or greater with specificity of 98 percent. The slightly higher sensitivity of fasting plasma glucose (versus glycated hemoglobin) is probably of minimal significance because fasting plasma glucose is a less convenient test. glycosuria assays have limited ability to detect diabetes with moderate hyperglycemia, but they reliably detect severe hyperglycemia. 4 figures, 4 tables, 41 references.

TITLE: Effectiveness of glycosylated Hemoglobin, Fasting Plasma Glucose, and a Single Post Load Plasma Glucose Level in Population Screening for Glucose Intolerance. Modan, M.; Halkin, H.; Karasik, A.; Lusky, A. American Journal of Epidemiology. 119(3): 431-444. March 1984.

OBJECTIVE: To determine which of the following methods is the most effective screening test for glucose intolerance (impaired glucose tolerance and type 2 diabetes): glycosylated hemoglobin, fasting plasma glucose, combination of fasting plasma glucose and glycosylated hemoglobin, plasma glucose 1-hour post oral glucose load, and plasma glucose 2-hour post oral glucose load.

CATEGORY: Secondary intervention.

Type of Study: Population screening.
Methodology: Statistical analysis.
Perspective: Health care system.

CONCLUSION: The most effective screening methods were the 1- and 2-hour post oral glucose loads. The 1-hour post oral glucose load worked better for detecting glucose intolerance, and the 2-hour post oral glucose load was better for detecting diabetes.

RECOMMENDATION: For clinical diagnosis, a two-stage screening using a fasting or random blood glucose test and an oral glucose load may be adequate, but for epidemiologic studies a full oral glucose tolerance test that includes a fasting test and two post-load levels is preferred.

ABSTRACT: The study population included a sample group of 2,040 people, aged 40-70 years, who were participating in the Israel Study of Glucose Intolerance, Obesity and Hypertension. Members of the sample group who were not known to have diabetes underwent a glucose tolerance test based on a fasting test and 1-hour and 2-hour post oral glucose loads. In 1,058 participants, glycosylated hemoglobin was also measured. Results showed that glycosylated hemoglobin alone is inefficient and inferior to a fasting plasma glucose test to determine glucose intolerance. Although glycosylated hemoglobin increased with glucose intolerance, there was considerable overlap in the distributions between newly identified patients with diabetes and patients with normal tolerance. The combination of the glycosylated hemoglobin test with the fasting plasma glucose test did not perform any better than the fasting test alone. However, testing fasting plasma glucose is not a satisfactory screening method because of its low specificities at adequate sensitivity levels compared with the 1- and 2-hour post glucose load tests. Of these tests, the 2-hour post glucose load level is more reliable when screening for diabetes alone, and the 1-hour post glucose load is more effective in screening for impaired glucose tolerance. Adding a fasting test to the 1- and 2-hour post glucose load tests raised the cost less than 5 percent, and the risk analysis showed that the venipunctures needed for the tests were not associated with any problems in patients. If a distinction is desired between impaired glucose tolerance and diabetes, a two-stage screening is indicated. In this method, everyone gets a 1-hour post load and the positive subgroup is retested by a full oral glucose tolerance test. For purposes of epidemiologic screening, a full oral glucose tolerance test is preferred over any "shortcut" method. 1 figure, 6 tables, 28 references.

TITLE: The glycosylated Hemoglobin as a Diagnostic and Monitoring Tool for Diabetes: Evidence from Claims Data (abstract). Altan, A.E.; Carlson, A.M.; Nettles, A. AHSR FHSR Annual Meeting Abstract Book. 1996; 13:11.

OBJECTIVE: To investigate how often the glycosylated hemoglobin (HbAlc) test is used to diagnose type 2 diabetes and the extent to which it is used to monitor diabetes control.

CATEGORY: Secondary intervention.

Type of Study: Patient management.
Methodology: Claims review.
Perspective: Health care system.

CONCLUSION: The HbAlc test may be misused as a diagnostic tool for type 2 diabetes and underused as a tool to monitor diabetes control.

RECOMMENDATION: None.

ABSTRACT: The investigators examined claims records of 81,039 persons enrolled in three 1993 managed care plans located in large metropolitan areas in the western, midwestern, or eastern United States. The researchers identified individuals with diagnosed diabetes (both insulin-treated and noninsulin-treated); those who had received the oral glucose tolerance test, the recognized diagnostic test for diabetes; and those who had received the HbAlc test. Among enrollees identified as not having diabetes (n = 70,068), 1,141 received an HbAlc test; of these persons, 4 percent (44 enrollees) also were given an oral glucose tolerance test. Of the 10,970 persons identified as having diabetes, 40 percent had had at least one HbAlc test during the year.

TITLE: Immunization to Prevent Insulin-Dependent Diabetes Mellitus? The Economics of Genetic Screening and Vaccination for Diabetes. England, W.L.; Roberts, S.D. Annals of Internal Medicine. 94(3): 395-400. March 1981.

OBJECTIVE: To determine the relative value of several strategies for preventing diabetes through the use of vaccine.

CATEGORY: Secondary intervention.

Type of Study: Epidemiological cohort model.
Methodology: Cost-benefit analysis.
Perspective: Health care system.

CONCLUSION: Vaccinating all children at age 3 was preferable, both economically and for disease prevention, to strategies that involved histocompatibility leukocyte antigen (HLA) screening prior to vaccination or to no vaccination at all.

RECOMMENDATION: Health programs and technologies need to be analyzed (e.g., for cost and risk) before they become available.

ABSTRACT: Numerous studies have shown an association between type 1 diabetes and viral infections, which suggests that vaccination for diabetes may be possible. The principal assumption behind the present paper is that viruses are involved in precipitating diabetes and that a vaccine for them can be developed. The author calculated an incidence rate for new cases of diabetes and program costs for 12 approaches to vaccination, 10 involving screening — the other 2 were vaccinating everyone and vaccinating no one. A decision tree analysis framework was used: the cost analysis considered the direct cost of diabetes, vaccine production and immunization, screening for predisposition to diabetes (measures evaluated included eight types of HLA tests, the Lewis negative erythrocyte phenotype, and the presence of diabetes in a parent or sibling), and side effects. The investigators used data based on published sources if possible; when data were inadequate or not available, a sensitivity analysis was performed. Only direct costs were tabulated. The analysis presupposed that any vaccinations would take place at age 3 (the most cost-effective age for all policies); incidence rates are the projected results after 27 years of vaccinating. Vaccinating everyone would reduce incidence by 29 percent and decrease the nondiscounted annual cost of diabetes by 18 percent; this policy was preferred to the other 11 on the basis of both incidence and cost. These results were, however, heavily influenced by the cost and efficacy of the vaccine, cost and probability of a vaccine reaction, cost and probability of diabetes, and the discount rate. A decision to vaccinate everyone was preferred to other alternatives for a wide range of vaccine characteristics. 7 tables, 51 references.

TITLE: Multiple Biochemical Blood Testing as a Case-Finding Tool in Ambulatory Medical Patients. Ruttimann, S.; Dreifuss, M.; Clemencon, D.; di Gallo, A.; Dubach, U.C. The American Journal of Medicine. 94(2): 141-148. February 1993.

OBJECTIVE: To prospectively assess the yield, disadvantages, and charges associated with routine biomedical testing in a medical outpatient clinic.

CATEGORY: Secondary intervention.

Type of Study: Patient management.
Methodology: Cost-analysis.
Perspective: Health care system.

CONCLUSION: Routine biochemical testing in the university teaching setting was useful in identifying new diagnoses, but the number of newly discovered disorders that required new medical management was small.

RECOMMENDATION: Reducing the number of routine biomedical tests to three — cholesterol, glucose, and alanine aminotransferase measurements — to detect disorders that need treatment may be preferable to ordering large biochemical profiles.

ABSTRACT: The authors assessed the utility of routine biochemical tests in finding new disorders that changed medical management. Patients (n = 493) attending the medical outpatient clinic of the University Hospital in Basel, Switzerland, for the first time during 1989 underwent a 23-test biochemical screen. To differentiate clinically indicated from routine tests, resident physicians were instructed to order only tests required to monitor known or suspected medical conditions, even though the complete profile was performed. Patient charges were $40 for the biochemical profile and $17.60 for each additional visit caused exclusively by abnormal routine tests; charges were counted until diagnosis or patient discharge. Ninety percent of the tests were considered routine; 11.4 percent of these routine tests were abnormal. Four hundred thirty of the 493 patients had new abnormalities diagnosed on the routine tests; further tests were ordered for 10.9 percent of these patients, additional visits were required for 1.4 percent, and new diagnoses were reached for 11.1 percent. A change in management occurred in 5.8 percent (n = 25) of these patients; in all but 4 of these cases the newly detected disorder was hypercholesterolemia, alcoholic liver disease, or diabetes mellitus. These 21 cases could have been detected with cholesterol, glucose, and alanine aminotransferase tests alone. Total charges for all tests were $20,938; total charges for tests considered clinically indicated were $8,256. Additional charges per patient for the detection of the new diagnoses averaged $25.72. Using only the tests relevant to the new diagnoses, per-patient charges would have been reduced 30 percent. Biochemical screening in this setting is feasible and results in acceptable amounts of further testing and additional patient laboratory costs. Prospective studies should be developed to test the utility of reducing screening to cholesterol, glucose, and alanine aminotransferase tests only. 4 tables, 41 references.

TITLE: Screening for Diabetes Mellitus in General Practice Using a Reflectance Meter System: The Islington Diabetes Survey. Forrest, R.D.; Jackson, C.A.; Yudkin, J.S. Diabetes Research. 6(3): 119-122. November 1987.

OBJECTIVE: To assess the accuracy of a reflectance meter system in screening for diabetes mellitus in a community-based screening program; to compare the results obtained with the laboratory assessment of blood glucose; and to estimate the costs of screening for diabetes mellitus with this system.

CATEGORY: Secondary intervention.

Type of Study: Population screening.
Methodology: Cost-effectiveness analysis.
Perspective: Health care system.

CONCLUSION: The reflectance meter system accurately estimated the prevalence of diabetes mellitus in the study population. Blood glucose concentrations obtained with the reflectance meter were higher than those of the laboratory assay, and confirmatory glucose tolerance tests were needed in a large number of misclassified patients. However, total costs of screening, even with the additional confirmatory glucose tolerance tests, were still significantly lower for the reflectance meter assay.

RECOMMENDATION: The glucose reflectance meter system, along with confirmatory glucose tolerance tests, can provide an acceptable, lower-cost alternative to the autoanalyzer glucose-oxidase method of blood glucose assay for community-based screening for diabetes mellitus.

ABSTRACT: The authors assessed the accuracy and costs of a reflectance meter system for screening for diabetes mellitus and compared the results obtained with laboratory assessment of blood glucose. As a component of the Islington Diabetes Survey, 1,084 randomly selected persons over age 40 were screened with an oral glucose tolerance test after an overnight fast. Two-hour blood glucose was measured using an automated glucose-oxidase method. For 530 persons, blood glucose was also measured using the reflectance meter system. The reflectance meter system gave an acceptable estimate of the prevalence of diabetes mellitus, identifying 14 of the 15 cases found by the glucose-oxidase system, but it did not provide an accurate estimate of the prevalence of impaired glucose tolerance. (Reflectance meter values were higher than glucose-oxidase values in 82 percent of cases.) Costs associated with screening the 530 individuals were calculated at £1,166.00 for the glucose-oxidase method and £358.75 for the reflectance meter. Confirmatory glucose tolerance tests were required in 29 persons on the basis of glucose-oxidase values and in 47 on the basis of reflectance meter values. Total costs for these confirmatory tests were estimated at £1,460.90 for patients initially screened via glucose-oxidase and £857.95 for those screened by the reflectance meter. Estimated total costs per case of diabetes mellitus identified were £97.39 for the glucose-oxidase method and £ 61.28 for the reflectance meter. The authors conclude that the reflectance meter is an acceptable screening tool for diabetes; it is less costly, it is simpler and easier to use, and it provides results faster than the glucose-oxidase method. 2 figures, 1 table, 16 references.

TITLE: Screening for Retinopathy: A Relative Cost-Effectiveness Analysis of Alternative Modalities and Strategies. Sculpher, M.J.; Buxton, M.J.; Ferguson, B.A.; Spiegelhalter, D.J.; Kirby, A.J. Health Economics. 1(1): 39-51. April 1992.

OBJECTIVE: To assess the relative cost-effectiveness of various screening strategies for retinopathy associated with diabetes.

CATEGORY: Secondary intervention.

Type of Study: Patient screening.
Methodology: Cost-effectiveness analysis.
Perspective: Health care system.

CONCLUSION: Systematic screening carried out in the general practitioner's office at the same time as routine diabetes care checks can save considerable costs over screening that requires additional visits to a hospital or an optician.

RECOMMENDATION: None.

ABSTRACT: The authors used data from a 1985-1988 English study of the cost-effectiveness of screening for retinopathy to compare various screening options, including single modalities (either ophthalmoscopy or fundus photography), combined approaches, and selective screening. The base study was conducted at three community-based health centers in England; 3,423 patients with diabetes were screened. The authors defined cost-effectiveness as cost per true-positive case detected. Total cost per patient for ophthalmoscopy at the general practitioner's office was £ 20.66; at an optician's office, £ 19.31; for fundus photography, this cost was £ 25.46 at the hospital and £ 14.87 at the general practitioner's office (costs at the general practitioner's office were less if screening was part of a general checkup). The sensitivity of these approaches (in the order just listed) was 0.53, 0.48, 0.40, and 0.58; specificity was 0.91, 0.94, 0.96, and 0.97. Combined approaches (e.g., both ophthalmoscopy and fundus photography at the general practitioner's office) had higher sensitivity but lower specificity. Expected cost per true-positive case detected was £ 784 for ophthalmoscopy (same for general practitioner and optician), £ 1,178 for hospital-based fundus photography, £ 497 for photography at the general practitioner's office, £ 734 for both ophthalmoscopy and photography at the general practitioner's office, and £ 968 for ophthalmoscopy by an optician combined with photography at a general practice. If screening at the general practitioner's office was combined with regular follow-up, these costs dropped to £ 273 for ophthalmoscopy, £ 434 for fundus photography, £ 419 for ophthalmoscopy plus photography, and £ 914 for photography at the office combined with ophthalmoscopy on the optician's premises. Among the selective screening options, directly referring high-risk patients and not screening those who were low-risk had the lowest cost per true-positive case ( £ 168) but only 0.25 sensitivity. Directly referring high-risk patents and providing both ophthalmoscopy and photography at the general practitioner's office had a sensitivity of 0.85 and a cost per true positive of £ 679 ( £ 407 if part of a general check-up). 1 figure, 5 tables, 42 references.

TITLE: Targeted Screening for Diabetes in Community Chiropody Clinics. Gill, G.V.; Lishman, L.; Kaczmarczyk, E.; Tesfaye, S. Quarterly Journal of Medicine. 89:229-232. 1996.

OBJECTIVE: To assess the cost-effectiveness of screening for diabetes in adults attending community podiatry clinics.

CATEGORY: Secondary intervention.

Type of Study: Patient screening.
Methodology: Cost analysis.
Perspective: Health care system.

CONCLUSION: Screening was simple and very cost effective, but the diagnostic yield was moderate.

RECOMMENDATION: Screening of a podiatry clinic's population for diabetes cannot be recommended without knowing what percentage of that population is already known to have diabetes.

ABSTRACT: Patients who have known foot problems may have an increased likelihood of having diabetes. In this study, which was based in Liverpool, England, all patients aged 40 to 75 years attending community chiropody (podiatry) clinics who were not previously known to have diabetes were offered postprandial screening for glycosuria. Patients who reported positive results on the self-test were brought to the hospital for a glucose tolerance test. Of 1,058 patients who accepted screening, 11 (1.0 percent) reported positive results; of this group, 4 had diabetes, 2 had impaired glucose tolerance, and 5 had normal glucose tolerance. The cost for each person screened was 11 pence; for each person with a positive urine test, £ 2.06; and for each newly diagnosed patient with diabetes, £ 34.46. No cost was included for staff time in the coordinating hospital chiropody department to respond to the participants. The authors indicate that their slightly disappointing result (0.4 percent new diabetes patients) is probably due to the very high proportion (17.3 percent) of Liverpool chiropody clinic patients already known to have type 2 diabetes. According to the authors, screening of a podiatry clinic's population cannot be recommended without knowing its proportion of already identified diabetes patients. They also observe that a major difficulty with screening for type 2 diabetes is the lack of a sufficiently sensitive and specific test. The authors also point out that handing out rather than mailing the test strips was cost-saving, as was having the patients telephone in their results. 1 figure, 2 tables, 15 references.

TITLE: Value of Serum Glucose Assay as Part of the Biochemical Profile in Screening for Diabetes. Northam, B.E.; Smith, J.H.; Fitzgerald, M.G.; Nattrass, M.; Wright, A.D. Annals of Clinical Biochemistry. 19(6): 412-415. November 1982.

OBJECTIVE: To evaluate a system of identifying patients with previously undetected diabetes that begins with a screening assay for serum glucose.

CATEGORY: Secondary intervention.

Type of Study: Patient screening.
Methodology: Cost analysis.
Perspective: Health care system.

CONCLUSION: Addition of serum glucose testing for inpatients undergoing other laboratory evaluations is a useful and cost-effective method of screening for diabetes.

RECOMMENDATION: None.

ABSTRACT: The authors report the effectiveness and cost of diabetes screening for adult inpatients at General Hospital in Birmingham, England, during 3 months in 1980. A serum glucose test was added to the routine biochemical profiles of 2,050 patients; 453 had values of 10 mmol/L or higher. After elimination of those known to have diabetes or receiving intravenous glucose, 152 patients underwent capillary blood glucose tests; 71 had elevated values. Ten of these patients were referred to the diabetes clinic, 42 underwent oral glucose tolerance testing, and 19 were lost because of illness, death, or discharge. Fifteen patients were found to have normal glucose tolerance, 11 had impaired glucose tolerance, and for 16, the glucose tolerance test was in the diabetes range (3 of these 16 patients were referred to the diabetes clinic). Retesting as outpatients 3 months later resulted in referral of two of the patients with impaired glucose tolerance and one with apparent diabetes to the diabetes clinic. In all, 16 patients with previously undetected diabetes were identified. Additional costs on an annual basis for direct screening were £ 500 for serum glucose testing (reagent only), £ 320 for 2 blood glucose tests (including labor for sample collection and analysis), and £ 440 for oral glucose tolerance testing, for a total of £ 1,260, or £ 20 per case of diabetes identified (64 cases in a year). If time for staff (other than the clinical staff) to eliminate patients with diabetes and on intravenous glucose was added, the total per case would be doubled. The prevalence of diabetes (0.8 percent of inpatients) may have been underestimated because of losses to follow-up (30 patients), size of dose of glucose for the glucose tolerance testing (50 g rather than 75 g), arbitrary selection of serum glucose concentration required for follow-up, and elimination of patients receiving intravenous glucose. The true prevalence of diabetes was probably 0.8 percent to 3.5 percent. This screening method is cost effective, based on the additional screening cost of only £ 20 per case of diabetes identified. 1 figure, 2 tables, 8 references.Disease Management

Back to Economics of Diabetes Mellitus Indexes

Top of Page