2National Centre for Diabetes Care, 2^nd Department of Medicine, Szent Imre Hospital, Budapest, Hungary

Professional Paper

Received: October 31, 2000

Key words: health care quality, type 1 diabetes mellitus, type 2 diabetes mellitus

SUMMARY

DiabCare, the monitoring system of the WHO/IDF-Euro, has been developed for continuous quality improvement of diabetes care based on a structured standardized diabetes data set. In the current report, data of 5 centers of DiabCare Hungary with different patient populations were adjusted for demographic parameters and compared with 2403 records of the national data set collected between 1995 and 1997. More than a half of the investigated patients were aged >56 (centers analyzed in detail, 20%-83%), one third were aged 36-55 (17%-52%), and 45% were male (27%-63%). The prevalence of IDDM was 29% (0-80%). Almost one third of patients had diabetes duration of >15 years (14%-58%). Eyes were examined in 79% (28%-98%), and retinopathy was found in 29% (28%-81%, minimal change after standardization). Risk factors were examined in at least 80% in all but one center (c) (16%-94%). Analysis of individual risk factors showed the following prevalence: HbA1c >8%, 43%; triglycerides >2 mmol/l, 33% (lower in center (e), 17%; after standardization, 37%); and blood pressure ³ 160/95 mm Hg, cigarette smoking and alcohol consumption, 15% each. Albuminuria was examined in 55% (25%-89%); microalbuminuria was present in 24% (0-38%). The frequency of self-monitoring and severe hypoglycemia (49% and 5%, respectively) parallelled the relative weight of IDDM patients during standardization. In conclusion, centralized evaluation and anonymous comparison of crude and adjusted data might be helpful in assessing and improving the quality of diabetes care.

INTRODUCTION

The 1989 St Vincent Declaration as the basis of the European Diabetes Action Programme formulated the exact reduction rates of diabetic complications for the 5-year period. It proposed to simultaneously spread the information technology in the field of diabetes care. Soon after the beginning of the program, it was realized that there were no basic data on changes in the rate and severity of late diabetic complications in Europe (1).

Under the auspices of the International Diabetes Federation (IDF) and World Health Organization (WHO) – Regional Office for Europe, various working groups were formed to design guidelines for the treatment and care of diabetic patients. After years of harmonization, the Information Technology Working Group finally compiled the Basic Information Sheet (BIS), a one-page form containing most important and essential questions and results on diabetes, which is suitable for quality control and quality management as well as for recording changes in patient care at every level of diabetes care (including general practitioner (GP) practices). The group led by Klaus Piwernetz prepared a software version of BIS (2-7).

The first pilot study in Hungary proved the applicability and usefulness of BIS and the respective software, also having explored their pitfalls (8). Based on the experience with these initial results, a software for anonymous central evaluation of Hungarian participants was developed. The DiabCare Hungary as a joint project of the Ministry of Welfare and WHO, and part of the National Diabetes Program, was launched at the national level (9,10).

Among the Hungarian participants there are GP practices that mostly care for orally treated NIDDM patients as well as regional and national centers dealing with IDDM patients with many comorbidities. Other centers provide care for special patient populations, e.g., one center specialized in screening, providing care for and following up pregnant diabetic women. Comparison of centers with very different epidemiologic data of their patients may lead to erroneous conclusions if based on raw quality indicators. To eliminate the impact of differences in patient characteristics, a standardization method was used in the present study.

PATIENTS AND METHODS

Since the beginning of the DiabCare Hungary group's work in 1993, 5044 BIS from 23 care centers (5044 diabetic patients) were centrally collected and evaluated. In this investigation, a total of 2403 aggregated records from 20 participating care centers between 1995 and 1997 were analyzed by the above mentioned special software. In addition to data on the distribution of care indicators in Hungary, detailed and standardized results on different patient populations from 5 centers, each center covering at least 150 patients, were compared.

The gratuitous computer version of BIS, accepted on European consensus (ProDiab 1.5), was used for quality control. It was developed under the auspices of the WHO (11,12). The sheet contains the most important characteristics of diabetic patients over the last 12 months. Only essential parameters that are accessible and possible to assess at any health care level in relation to diabetes are included. The following patient data are contained in the sheet: sex, age, type and duration of diabetes. These data are stored and sent to the evaluating center respecting the patient's anonymity. For center characterization, intermediate outcome indicators based on the pathogenetic progression to complications appear to be more appropriate. These include various stages of retinopathy, diabetic foot issues, microalbuminuria, and elevated creatinine levels. This group of indicators can also include metabolic parameters (glycosylated hemoglobin) and risk factors (increased triglycerides, elevated blood pressure, cigarette smoking, alcohol consumption, increased body mass index).

Among others, there are true outcome indicators (end-stage complications) such as blindness, end-stage renal failure, above ankle amputation, myocardial infarction, and stroke. An important goal of the St Vincent Declaration is reduction of the rate of these complications. In BIS, only new cases (occurring over the last year) are recorded, i.e. only the incidence (new cases/patient year) can be calculated from these data. Data are gathered on the treatment for comorbidities, education, pregnancy, episodes of severe hypo- and hyperglycemia, self-monitoring, and hospitalization.

Upon evaluation, each center is provided with feedback information on the incidence, not only on the examination frequencies but also on the frequencies of pathologic results, e.g., data on the proportion of glycosylated hemoglobin measurement as well as on the proportion of patients with elevated levels of this parameter. These parameters yield accurate characterization of each individual center, and are also highly useful for longitudinal follow-up.

In annual cross-sectional summing ups, there is detailed evaluation of each center, with benchmarking charts on the completeness and risk value occurrence of quality indicators for the centers. Using benchmarking, one can compare the management of a particular center with the others, and find similar centers according to the basic data on treated patients. Thus, the quality indicators of these centers can be objectively compared.

The method of indirect standardization (adjusted rates) was used to compare the national averages with selected centers. The national data were divided into groups according to the basic data, and the data of the selected centers were adjusted to these averages for sex, age, type and duration of diabetes (13). Results are shown as percentage.

Demographic data of these centers were very different. The proportion of male patients ranged from 27% to 63%. There were 0-27% of patients aged <35, while 14%-54% of patients had diabetes duration of >15 years. The percentage of insulin dependent diabetes mellitus (IDDM) varied from 0 to 80% (Table 1).

Table 1. Demographic data of centers (a-e) and total national average (H)

RESULTS

Completeness. Data on the date of birth, sex, date of diabetes diagnosis, type and treatment of diabetes were filled out in an average of 93% of sheets. In the selected centers, this parameter varied between 78% and 99%, and did not change considerably after standardization (detailed data not shown).

On assessing the intermediate outcome indicators, eye pathologies were evaluated first. One fifth provided no data on the state of eyes, the percentage varying from 2% to 72%. Poorest results were found for center (c), where these data failed to change upon standardization. There was no retinopathy in 19%-72% of the centers. The order of centers did not change upon standardization (Fig. 1).

Figure 1. Frequency of eye examination and retinopathy before and after standardization (%).

Risk factors. The levels of HbA1c and triglycerides, blood pressure, cigarette smoking and alcohol use were evaluated in more than 80% of patients in all centers except for center (c). The rate of these factors did not change significantly upon standardization. The measurement of microalbuminuria ranged from 20% to 70%. At center (e), the proportion changed considerably upon standardization, i.e. from 73% to 45%. The percentage of patients with high risk values of glycosylated hemoglobin (>8%) was 43% on an average. A considerably lower rate was observed for centers (c) and (e), where it was 2% and 29% (1% and 28% upon standardization), respectively, while center (d) showed a higher rate of 50% (51% upon standardization). It should be noted that a small number of measurements were performed at center (c), so these results should be considered questionable (Table 2).

Table 2. Frequency of risk factor measurement and proportion of patients with elevated risk results according to centers (before and after standardization) and at the national level (data presented in %)

Elevated triglyceride levels (>2 mmol/l) were found in 33% of patients. Results of each center were close to the national average, except for center (e). In this center, the rate of elevated triglyceride levels was only 17%, which turned out to be similar or even greater than the national average (37% vs. 33%) upon standardization.

According to the WHO criteria (³ 160/95 mm Hg), 15% of patients had hypertension. At center (b) there was a very low number of measured hypertension even after standardization (2%). At center (c), hypertension was quite frequent (23%), however, upon standardization it approached the national average (19%).

The bad habit of cigarette smoking and regular alcohol use were recorded in 15% of investigated cases. At center (c) there was a 20% rate of alcohol use and cigarette smoking, which even increased upon standardization (26%).

It is striking that microalbuminuria was measured in only 55%, while all other risk factors were observed in more than 80% of cases. The prevalence of microalbuminuria (30-300 mg/day) was 24% on an average, ranging from 0 to 38% in the selected centers. At center (b), microalbuminuria was measured only in a small proportion of patients (37%, and 34% after standardization), however, there was a high proportion of those at risk (40% vs. 42%). Interestingly enough, in center (e) where microalbuminuria was present in a lower percentage (12% vs. 24%), the proportion of microalbuminuria showed further decline (11%). Macroalbuminuria was found in 5%-7% of patients at different centers, against the national average of 3% (Table 2).

The number of true outcome indicators was very small and did not change considerably. In center (c) there was a very high rate of myocardial infarction and stroke compared with the results of other care centers and national average (Table 3).

Table 3. Frequency of St Vincent targets, hyperglycemia/ketoacidosis, and hospitalization according to centers (before and after standardization) and at the national level (data presented in %)

Evaluation of the frequency of self-monitoring showed an increase upon standardization in centers (c) and (d), and a decrease in centers (b) and (e). In the former two centers (c and d), the number of IDDM patients was lower, and in the latter two (b and e) it was greater than the national average. The proportion of IDDM patients changed in parallel to the frequency of self-monitoring (Fig. 2).

Figure 2. Association between relative severity of IDDM and frequency of self-monitoring, and hypoglycemia before and after standardization (% of cases).

The rate of severe hypoglycemia episodes (requiring third party help) showed a similar pattern of change as self-monitoring, i.e. if the proportion of IDDM patients grew after standardization, the rate of severe hypoglycemia episodes increased (center d), and vice versa (centers b and e) (Fig. 2). Severe hyperglycemia/ketoacidosis occurred in 3% in the country as a whole, while in the selected centers it ranged from 0 to 2%, reflecting a very small number of cases (Table 3).

Hospitalization. Hospitalization was required in 40% of patients (not only for diabetes and its complications but also for other diseases) during the one-year investigation. This parameter showed extreme differences among the selected centers, ranging from 17% to 72%, and remained so upon standardization (17%-60%) (Table 3).

DISCUSSION

These descriptive data speak for themselves, with some notes and additional information provided below. The male to female ratio was found to be comparable to the 1:1.6 ratio recorded in the population based screening for diabetes in Hungary in 1981 – 1982. These data highly coincide with the female predominance found in a representative sample in the United States (14). The rates different from the screening results could be explained by the special care interests (e.g., screening, care for and follow-up of patients with diabetes complicated pregnancy). There were similar shifts in the proportion of IDDM or insulin treated patients.

The data shown above appear to suggest that real comparisons could only be made after adjustment for age, sex, type and duration of diabetes, etc. In regional centers, more severe cases are treated, so the proportion of diabetes complications may be higher even with better care provided there.

The results of Klein et al. show that the prevalence of eye complications ranged from 20% to 100% in different age and diabetes duration groups (15,16), which is consistent with our data (28%-83%) for centers with different patient populations. The differences after standardization are real differences in the quality of care at a given center.

The risk factors were adequately measured in the selected centers as well as on an average of 20 centers. Nevertheless, microalbuminuria showed a lower frequency of examination. It is an early sign of diabetic nephropathy and an important cardiovascular risk factor to be treated. At center (c), the risk factors were rarely investigated. It could be explained by the practice itself, because this center includes patients from different GPs, and measurements are often inaccessible (problems with blood transport) and underfinanced.

The proportion of patients with high risk values was quite comparable with the proportion found in the pilot study. In 1993, elevated HbA1c was recorded in 34%, elevated triglycerides in 18%, hypertension in 12%, cigarette smoking in 19%, and microalbuminuria in 43% of patients (8). Differences among centers upon standardization were the real differences in the risk state, however, in some cases it could only be evaluated with some reserve. The high prevalence of microalbuminuria in center (b) (38% vs. 24% at the national level) may be consequential to the fact that a small number of examinations were performed in the high risk population. In the EURODIAB IDDM Complications Study, microalbuminuria was found in 18%, and proteinuria in 15% of patients (17).

The great difference among centers in the pathologic values of glycosylated hemoglobin, recorded even after standardization, calls for an explanation. One of the reasons is methodological, i.e. some centers did not give the methods of measurement. Presumably both HbA1 and HbA1c values with different normal and pathologic ranges were in the database.

The true outcome indicators could only be compared with caution, because these parameters are not typical for current care but for complication end stages that can develop in 5-15 years. The differences found among the centers were partly the result of different definitions used for different end-stage complications.

The association of self-monitoring and prevalence of IDDM cases agrees with the guidelines for IDDM treatment, according to which each IDDM patient should regularly perform self-monitoring (18). According to DCCT Research Group, the proportion of severe hypoglycemia was 9.8% in patients on conservative therapy and 26% in those on intensive therapy (19). Similar results (32% of severe hypoglycemia) were also obtained in the EURODIAB IDDM Study (20). In NIDDM patients, hypoglycemic episodes were much less frequently observed (21). Taking into account the proportion of IDDM and NIDDM cases, the estimated percentage of hypoglycemia is consistent with our results.

Anonymous comparisons among different centers, and calculation of national averages could provide a general view of the position of a given care and help quantify insufficiencies of the investigations. The first step in quality assessment is data collection, data filing, and accessibility of data of a given center. Using the results of anonymous central evaluation, benchmarking comparison of all centers in the country according to quality indicators could be performed. On these scales, valid comparison of centers is only possible with similar patient populations. The standardization used in the present investigation proved to be a suitable method for comparison among centers with different or even special patient populations.

A limitation of this method is that one can standardize only large databases (in our evaluation, 150 patients in each center) according to many parameters. According to one parameter (e.g., type of diabetes), small databases can be adjusted to each other, thus this method could also be suitable for comparing GP practices.

With the continuous use of a uniform European quality control procedure and yearly repeated evaluations, there is an opportunity to quantify the improvement or deterioration of the state of quality indicators. In addition to these longitudinal evaluations, there also is an opportunity to take real cross-sectional comparisons using standardization. This method could be easily computerized according to yearly average. Provided the centers send data on the same patients every year, the system would become 'self-controlled', and this appears to be the way to an appropriate quality assurance. The system could help in collecting data for accreditation of groups, hospital departments, etc., provided the participants give up anonymity and show their results to the respective health authorities.

The widespread use of the program in Hungary would provide an opportunity to assess the needs and deficiencies of the care of the disease that affects almost a half million people. It would also help concentrate properly the financial and manpower resources.

The entire diabetic population could be observed if GPs join the program. And the last but not the least, international comparisons would thus become possible with correct and widespread data collection. As a consequence of the European initiative, the American countries have declared their attitude toward the use of information technology in quality improvement (22).

REFERENCES

1. Krans HMJ, Porta M, Keen H, eds. Diabetes care and research in Europe: the St Vincent Declaration Action Programme. Copenhagen: WHO, Regional Office for Europe, 1992.
2. Carson ER, Home PD, Massi-Benedetti M et al. Guidelines to support the optimal implementation and use of information systems in diabetes care. Diab Nutr Metab 1992;5:141-150.
3. Piwernetz K, Antsiferov MB, Home PD et al. Monitoring instruments for quality improvement in diabetes care. In: Krans HMJ, Porta M, Keen H, eds. Diabetes care and research in Europe: the St Vincent Declaration Action Programme. Implementation document. Giorn Ital Diabetol 1992;12 (Suppl 2):30-32.
4. Piwernetz K, Massi-Benedetti M, Vermeij D et al. DiabCare Thinkshop: 'Quality Network in Europe'. Diab Nutr Metab 1993;6:107-122.
5. Piwernetz K, Bruckmeier A, Home PD et al. for the pilot group DiabCare Diabetes Data Set. DiabCare quality network in Europe – supporting instruments. The Saint Vincent Declaration Newletter 1995;Suppl 1:7-8.
6. Piwernetz K, Bruckmeier A, Home PD et al. for the pilot group DiabCare Diabetes Data Set. DiabCare Diabetes Data Set – a key element of the DiabCare Quality Network. The Saint Vincent Declaration Newletter 1995;Suppl 1:18-26.
7. Piwernetz K, Home PD, Johansen KS et al. DiabCare Quality Network in Europe Telematics Project within the St Vincent Declaration – Consensus Meeting. Diab Nutr Metab 1995;8:243-249.
8. Tamas Gy, Kerenyi Zs, Fφvenyi J et al. Quality control of diabetes management using WHO/IDF DiabCare Programme. Results of a pilot multicentre study in Hungary. Diab Nutr Metab 1993;6:329-332.
9. Kerenyi Zs, Tamas Gy, Tabak AGy and the DiabCare Hungary Group. DiabCare Hungary: Implementation of a tool for continuous quality improvement in diabetes care. Diabetologia 1997;40 (Suppl 1):A648.
10. Tamas Gy, Kerenyi Zs, Tabak AGy and the DiabCare Hungary Group. Assessing and comparing quality of diabetes care: a standardized European monitoring system. Diabetes 1998;47 (Suppl 1):A185.
11. Piwernetz K, Bruckmeier A, Staehr-Johansen K, Krans HMJ. DiabCare quality network in Europe. Diab Nutr Metab 1993;6:311-314.
12. Piwernetz K, Home PD, Snorgaard O et al. for the DiabCare Monitoring Group of the Saint Vincent Declaration Steering Committee. Monitoring the targets of the St Vincent Declaration and the implementation of quality management in diabetes care: the DiabCare initiative. Diabet Med 1993;10:371-377.
13. Jekel JF, Elmore JG, Katz DL, eds. Epidemiology, biostatistics and preventive medicine. Philadelphia – London – Toronto – Montreal – Sydney – Tokyo: WB Saunders, 1996.
14. Widom B, Simonson DC. Iatrogenic hypoglycemia. In: Kahn CR, Weir GC, eds. Joslin's diabetes mellitus. 13^th ed. Philadelphia: Lea & Febiger, 1994:489-507.
15. Klein R, Klein BE, Moss SE et al. The Wisconsin epidemiologic study of diabetic retinopathy. II. Prevalence and risk of diabetic retinopathy when age at diagnosis is less than 30 years. Arch Ophthalmol 1984;102:520-526.
16. Klein R, Klein BE, Moss SE et al. The Wisconsin epidemiologic study of diabetic retinopathy. III. Prevalence and risk of diabetic retinopathy when age at diagnosis is 30 or more years. Arch Ophthalmol 1984;102:527-532.
17. Stephenson JM, Fuller JH. Microalbuminuria is not rare before 5 years of IDDM. EURODIAB IDDM Complications Study Group and the WHO Multinational Study of Vascular Disease in Diabetes Study Group. J Diabetes Complications 1994;8:166-173.
18. European IDDM Policy Group. Consensus guidelines for the management of insulin-dependent (type I) diabetes. Bussum: Medicom Europe BV, 1993.
19. DCCT Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993;329:977-986.
20. Kerenyi Zs, Nunes-Correa J, Santeusanio F and the EURODIAB IDDM Complications Study Group. Acute complications in insulin dependent (type 1) diabetes: the EURODIAB Complications Study. Diabetologia 1992;35 (Suppl 1):A183.
21. Sugarman JR. Hypoglycemia associated hospitalizations in a population with a high prevalence of non-insulin dependent diabetes mellitus. Diabetes Res Clin Pract 1991;14:139-147.
22. Declaration of the Americas on Diabetes. Diabetes Care 1997;20:1040-1041.