A Retrospective Time Trend Study Of Diabetes Overtreatment In Geriatric Patients

Introduction

Diabetes mellitus (DM) is a common health condition in old age, connected with the higher rates of cardiovascular diseases, increasing risk of premature death and negatively effecting patients’ functional abilities and quality of life.^1–5 There is no doubt at the moment that glycemic control targets in older patients with DM should be individualized, according to their functional status and comorbidities.^6–8 The necessity of therapy individualization is extremely important in case of patients with limited ability to identify and manage hypoglycemia, with limited life expectancy, and especially in terminally ill patients. Therefore, in case of patients with multiple chronic illnesses, cognitive impairment, or functional dependence, less stringent glycemic goals are recommended.^7,9

In spite of this recommendation, the problem of DM overtreatment in geriatric patients was described in some studies.^10–12 We also observed a high prevalence of DM overtreatment in our previous cross-sectional study of the population of patients admitted to the geriatric ward in 2009–2010. The study was carried out just before the introduction of the principle of individualization of DM therapeutic goals for older people burdened with multimorbidity, and with short life expectancy.¹³ This principle was literally included in the guidelines of the Diabetes Poland Association (PTD) regarding DM treatment only in 2011.¹⁴ The lower glycemic goal (glycosylated A_1C hemoglobin (HbA_1C) ≤7% [53 mmol/mol]) was recommended for otherwise healthy, with few coexisting chronic illnesses, older patients with DM, whereas the new DM treatment goal, i.e. HbA_1C <8% [64 mmol/mol], was proposed for older patients with long-term DM and vascular complications. This recommendation was retained in issued annual PTD guidelines, up to the last ones.⁸

Our aim was to assess whether the principle of individualization of DM therapy in the PTD guidelines was related to any change in DM overtreatment in patients with type 2 DM admitted to the geriatric ward, after five years of observation. To the authors’ knowledge, this is the first publication addressing the issue of implementation of the principle of DM therapy individualization in older patients in Poland from a long-term perspective.

Materials And Methods

This time-trend assessment comprised two retrospective cross-sectional cohort studies of type 2 DM patients; admitted to the geriatric ward between 1st January 2009 and 31st December 2010 (1st round),¹³ and between 1st September 2014 and 30th April 2015 (2nd round) – a secondary analysis of the study on frailty syndrome in patients of the geriatric ward.¹⁵ The study population included patients with type 2 DM, on antihyperglycemic medications (insulin and/or oral glucose-lowering agents) prior to admission, who were tested for Hb A1c during a stay in the ward. Medical records of all consecutive patients were analyzed.

The geriatric ward of the Hospital of the Ministry of Interior in Bialystok, Poland, is a sub-acute department. Older people with multimorbidity and physical/mental disability are referred to it by general practitioners or other specialists, and are admitted in a planned manner. A mean length of stay is 7 days, and a comprehensive geriatric assessment carried out by a multidisciplinary team, including reviewing and modifying patient’s pharmacotherapy, is one of the goals of hospitalization. Most often, therefore, it was not possible to indicate a single reason for hospitalization.

Classification Of Groups

The subjects were classified into two groups:

1. A “high-risk group” was defined as patients on antihyperglycemic medications prior to admission, who were 80+ years old, diagnosed with dementia, end-stage renal disease or had a history of macrovascular complications.
2. A “low-risk group” was defined as patients who did not meet the abovementioned criteria.

Study Parameters

End-stage stage chronic kidney disease (CKD) was defined as stage 4 and 5 CKD according to the Kidney Disease Outcome Quality Initiative (KDOQI)-glomerular filtration rate (GFR) <30 mL/min/1.73 m². GFR was counted using the Cockcroft-Gault formula.¹⁶ Dementia diagnosis was based on the clinical neuropsychologist’s examination. Macrovascular complications included history of myocardial infarction (MI), coronary artery bypass graft (CABG), percutaneous transluminal coronary angioplasty (PTCA), stroke and transient ischemic attack (TIA). Only confirmed in medical records diagnoses were included.

Information on patients’ age, gender, place of residence (urban/rural), functional status (basic activities of daily living (ADL) assessed with the Barthel Index¹⁷ [range 0–100 points]; instrumental activities of daily living (IADL) assessed with the 6-item scale derived from the Duke OARS assessment that included activities such as housework, preparing own meals, using the telephone, handling money, shopping and taking medicines [range 0–12 points];¹⁸ risk of pressure sores assessed with the Norton scale [range 5–20 points; the score ≤14 points pointed to the increased risk]),¹⁹ on nutritional health (body mass index (BMI), number of lymphocytes in blood), depression, hypertension, orthostatic hypotension (diagnosed if systolic pressure decreased by 20 mm Hg or diastolic pressure decreased by 10 mmHg in the first or the third minute of the active standing test) and serum creatinine level were also collected. Hypoglycemic medications use (insulin, sulfonylurea, metformin or others) both prior to admission and recommended at discharge was evaluated. HbA_1C measurements were made with the immunoinhibition method using an Olympus AU400 analyzer (Beckman-Coulter, Brea, CA, USA).

Outcome Measures

The primary outcome measure was HbA_1C ≤7.0% (53 mmol/mol). We defined also three different thresholds of HbA_1C:

1. ≤7.0% [53 mmol/mol] and >6.5% [48 mmol/mol];
2. ≤6.5% [48 mmol/mol] and >6.0% [42 mmol/mol];
3. and ≤6.0% [42 mmol/mol]

that reflected increasingly tight glycemic control, and had been confirmed as connected with the increasing risk of hypoglycemia in older diabetic patients.²⁰

Statistical Methods

IBM SPSS Version 18 Software suit (SPSS, Chicago, IL, USA) and STATISTICA 13.3 software package (TIBCO Software, Palo Alto, CA, USA) were used to analyze data collected. Shapiro–Wilk test was used to assess the distribution of variables. Data were presented as means and standard deviation for normally distributed, as medians and interquartile range for not normally distributed continuous variables, and as the number of cases and percentage for categorical variables. Proportions were compared using χ2 tests or Fisher exact test, as appropriate, while the independent samples Student’s t-test and Mann–Whitney U-test were used to compare means and medians. To assess differences between two dependent samples, Wilcoxon signed-rank test was used. Odds ratio for DM overtreatment in the 2nd round of the study, comparing to its 1st round, was calculated. Missing values were omitted and statistics in such cases were calculated for the adequately reduced groups. A P value of less than 0.05 was regarded as significant.

Results

213 patients in the 1st round and 83 in the 2nd round were included. Figure 1 shows patients’ enrolment in the study. Groups did not differ significantly in age, gender and in the majority of health and functional characteristics analyzed. The majority of them were women (165; 77.5% in the 1st round and 58; 69.9% in the 2nd one), aged 80 years or more (105; 49.3% in the 1st round and 49; 59% in the 2nd one). In the 2nd round of the study, the higher percentage of dementia (36.1% versus 18.8%, P=0.002) and of the “high-risk” cases (78.3% versus 61.0%, P=0.05) was observed (Table 1). A relatively small part of patients admitted to the geriatric ward (16% in the 1st round and 21.7% in the 2nd one) had HbA_1C above 8% [64 mmol/mol]. In the majority of cases (69.5% in the 1st round and 56.6% in the 2nd one), HbA_1C level was ≤7% [53 mmol/mol].

Table 1 Characteristics Of Patients In Two Rounds Of The Study

Figure 1 Flow chart of diabetes (DM) patients’ enrollment.

We identified 130 (61%) patients in the 1st round and 65 (78.3%) patients in the 2nd round as the “high-risk” cases. Over the study period, tight glycemic control prevalence in the high-risk group decreased significantly from 73.1% to 58.5%, P=0.04 (odds ratios 0.68, 95% CI 0.47–0.97), and the median value of HbA₁c increased significantly from 6.4%, IQR 5.7–7.3 [46 mmol/mol, IQR 39–56] to 6.7%, IQR 6.1–7.9 [50 mmol/mol, IQR 43–63], P=0.03. But still, very frequently very low values of HbA_1C were observed in the high-risk population (Figure 2).

Figure 2 A distribution of HbA1c values in the high-risk group, in two rounds of the study.

Sulfonylurea was the most frequently used hypoglycemic agent at admission to the ward in the 1st round of the study (65.3%), but its use at admittance decreased significantly in the study period to 45.8%, P=0.002. Metformin use in admittance increased significantly in the study period from 39.4% to 61.5%, P=0.001, and it was the most frequently used hypoglycemic agent at admittance in the 2nd round of the study. The study rounds did not differ in the frequency of insulin and α-glucosidase inhibitors usage. Other therapeutic options were absent in both rounds of the study. Metformin was used more frequently at discharge than at admission, and this change was significant only in the 1st round of the study, whereas sulfonylurea was used significantly less frequently at discharge than at admission in both rounds of the study (Table 2). Antidiabetic medications used by patients of the high-risk group before hospitalization, by level of glycemic control, are presented in Table 3. In two rounds of the study, tightly treated patients of the high-risk group did not differ in the percentage of patients taking metformin, sulfonylureas or α-Glucosidase inhibitors, but significantly less frequently were on insulin in the 1st round of the study, and the similar trend was observed in the 2nd round of the study.

Table 2 Glucose-Lowering Medications Use At Admittance, And At Discharge From The Geriatric Ward, In Two Rounds Of The Study

Table 3 Antidiabetic Medications At Admittance To The Geriatric Ward In The High-Risk Group In Two Study Rounds By Level Of Glycemic Control

Discussion

Different studies highlighted the importance of recognizing frailty, or limited life expectancy, and supported the individualization of care in older people with DM.²¹ Despite clear guidelines recommending the higher glycemic targets in certain subgroups of older people, too tight glycemic control was noticed in some studies.^10,11 It concerned also a high proportion of patients admitted to our geriatric ward in 2009–2010.^13,22

In 2011, PTD guidelines on the management of diabetic patients in Poland literally took into account the earlier recommendations of international societies.¹⁴ Despite this change in the Polish guidelines, the current study confirmed that tight DM control was still a far more common in the group of older patients admitted to the geriatric ward than insufficient control of the disease. The HbA_1C ≤7% [53 mmol/mol] was observed in almost 70% of the study participants in the 1st round (2009–2010) and in almost 60% of the 2nd round (2014–2015), but HbA_1C below 6.5% [48 mmol/mol] had 53.9% and 45.8% patients in these rounds, respectively. In many studies, such low levels of HbA_1C were associated with the increased risk of hypoglycemic episodes,^23–25 although severe hypoglycemia was common among patients with type 2 DM across all levels of glycemic control in the Diabetes and Aging Study. Risk tended to be higher in patients with either near-normal glycaemia or very poor glycemic control.²⁶ It is well known that not only lower average glucose levels, but also higher glucose variability indicate a greater hypoglycemia risk.²⁷ Additionally, taking into consideration some authors’ observations, indicating that HbA1c levels may increase with age in patients without DM,²⁸ the results of our study – pointing to the overtreatment of DM in geriatric patients – are even more alarming.

The “high-risk group” (“age 80+ years” or “end-stage renal disease”, or “dementia”, or “macrovascular complications”) constituted 67.7% of patients in 1st round and 79.3% in the 2nd round of our study. The abovementioned patients’ profile was connected with the limited life expectancy and the higher hypoglycemia risk.^29–32 For these patients, less stringent glycemic goals were recommended,⁹ but 73.1% of them in the 1st round and 58.5% in the 2nd round had values of HbA_1C ≤7.0% [53 mmol/mol]. It was confirmed in some studies that the prevalence of hypoglycemia increased significantly as HbA_1C decreased.³³ In both rounds of our study, a high percentage of patients had HbA_1C values below 6.0% [42 mmol/mol]. Therefore, we can assume that episodes of hypoglycemia might have occurred frequently in the study groups before admission to the hospital, and might have been one of the reasons for the changes that were introduced in the treatment of DM during the stay of patients in the geriatric ward. Unfortunately, the retrospective nature of the study did not allow to assess this problem. However, it should be noted that over the course of 5 years of follow-up the decrease in the prevalence of too tight DM treatment was statistically significant, confirming that the principle of individualization of DM therapy in the elderly population had been better adhered to.

The type of agents used to treat DM may contribute to the risk of hypoglycemia. A patient-centered approach is recommended to guide the choice of pharmacologic treatment, and the medication regimen should be reevaluated at regular intervals, and adjusted as needed to incorporate new patient’s factors.^34,35 Since many years now, metformin has been the preferred initial pharmacologic agent for the treatment of type 2 DM, and once initiated it was recommended to continue it as long as it was tolerated, and not contraindicated. The study showed that the frequency of metformin use in patients admitted to the geriatric ward had increased significantly over the years of follow-up, and changes in metformin prescribing during the hospital stay were no longer significant in the 2nd round of the study. The approval for metformin up to GFR of 30 mL/min/1.73 m² that took place between the study rounds certainly could have contributed to this. Contrary to that, although the frequency of sulfonylurea use had decreased significantly over the study period, the sulfonylurea was still often discontinued during patients stay in the geriatric ward. It had been confirmed that sulfonylurea was associated with the high risk of hypoglycemia and alternative hypoglycemic agents should be considered rather, when metformin was not tolerated or contraindicated in the older patient.³⁶ In the last PTD recommendations for the treatment of DM, sulfonylureas were placed at a distant place in the treatment regimen,⁸ but they were one of the foremost groups of drugs in DM in the period covered by our study. Some studies revealed that the highest risk of hypoglycemia is under intensive insulin therapy;³⁷ however, this treatment strategy is not very common in patients of advanced old age.³⁸ Moreover, tightly treated patients less frequently were on insulin, and our previous analysis of the first round of the study had confirmed that insulin intake was a negative predictor of low HbA1C values in our patients.¹³ This could have been the result of a large variation in glycemic values in these patients, but also the consequence of a different level of care (including medical) provided to these patients, although data gathered in our study did not allow us to verify that.

In patients with established atherosclerotic cardiovascular disease, or CKD, sodium–glucose cotransporter 2 (SGLT2) inhibitors, or glucagon-like peptide 1 (GLP1) receptor agonists are safer, and their use brings additional benefits and improves prognosis.^39–41 These drug groups were not completely taken by our patients in both rounds of our study. We can assume that it was largely due to financial constraints, as SGLT2 inhibitors and GLP1 receptor agonists were (and still are) rather expensive, and not refunded in the Polish health care system. For many years now, the PTD has been unsuccessfully made efforts to include these drugs on the reimbursement lists. Inadequate management of DM is more common in patients with poor societal conditions,⁴² so screening for that is recommended.⁴³ Unfortunately, the study design did not allow to analyze social determinants of health in our patients. But we should be aware that each of the new groups of antidiabetic medications also comes with its own unique advantages and safety profile, so a personalized case by case approach should be adopted while establishing treatment strategy for older patients, after weighing the overall risks and benefits of therapy. Therefore, for instance, treatment with SGLT 2 inhibitors in elderly people may be problematic, as it requires a special body hygiene, which is often difficult to keep for patients with physical disability or dementia.

Limitations And Strengths

The strength of this study is the use of the real-world data. It was not based on administrative claims, but we have evaluated the problem of DM treatment in old age from the perspective of everyday clinical practice. Our study was not a randomized clinical trial, but included older patients burdened with disability and multimorbidity, usually excluded from most clinical trials. Furthermore, this is the first study to investigate the problem of implementation of the principle of DM therapy individualization in older patients in Poland from a long-term perspective.

However, some limitations of this study are worth mentioning. First of all, it was performed not in the sample randomly selected from the general population of older people, but in the convenient sample of patients admitted to one geriatric ward, so the results can be generalized for the patients of similar characteristics (in more advanced age, more disabled and with different geriatric syndromes, such as dementia, depression, malnutrition, dependence on others in activities of daily living). The multicenter study would increase the external validity of the results. Some limitations of our study resulted from its retrospective design and secondary analysis of data collected in previously conducted studies. Thus, we were unable to determine, for example, how the dose of hypoglycemic medications was modified, or what the specific reason for the absence of new, safer drug groups in treatment regimens was. The availability of data on the frequency of hypoglycaemia in the last year was also limited, so there was no possibility to analyze that. The consequence of the secondary analysis of earlier studies was also the different number of patients recruited for them.

Conclusion

Despite the principle of individualization of DM therapy that was in force, after five-year observation the problem of DM overtreatment still concerned a large percentage of geriatric patients, although a positive trend was noted in this respect.

This may indicate the need to further broaden the knowledge of doctors about the specificity of DM treatment in older patients, with particular emphasis on the high-risk of hypoglycemia groups. It is also worth considering changing system solutions in the care of these patients, taking also into account the social determinants of health. It seems that there is an urgent need to facilitate older patients access to new hypoglycemic medications associated with lower risk of hypoglycemia, by including them in the reimbursement lists.