Low mean HbA1c does not increase all-cause and cardiovascular mortality in patients with diabetes: Effect-modifications by anemia and chronic kidney disease stages

Abstract

Background Previous studies reported that low levels of glycated hemoglobin A1c (HbA1c) were associated with increased mortality. We investigated rates and risks of all-cause and cardiovascular mortality in association with mean HbA1c levels with stratification of anemia and chronic kidney disease (CKD) stages, major causes of low HbA1c. Methods 47,145 patients with prescription of antidiabetic agents >6 months in the outpatient visits (2003–2018) were linked to Taiwan’s National Death Registry to identify all-cause and cardiovascular mortality. Poisson assumption was used to estimate the mortality rates, and the Cox proportional hazard regression model was used to evaluate the relative hazards of respective mortality in relation to mean HbA1c in different statuses of anemia and CKD stages. Results All-cause and cardiovascular mortality rates were the lowest in non-anemic stages 1–2 CKD patients, and the highest in anemic stages 3–5 CKD patients. In stages 1–2 CKD, excessive HRs observed in those with mean HbA1c <6.0% (Hazard Ratio [HR]) 1.58; 95% Confidence Interval [CI] 1.18–2.12) became inconsequential after adjustment of medications and laboratory results (HR: 1.26; 95% CI 0.89–1.79). The similar patterns were observed in anemic stages 1–2 CKD, anemic or non-anemic stages 3–5 CKD. Low HbA1c was not related to cardiovascular mortality in any anemia status or CKD staging. Conclusions Higher risks associated with low mean HbA1c and all-cause mortality were attenuated by adjustment of medications and comorbidities. It is imperative for the diabetologists to consider confounding effects of underlying illness before concluding low HbA1c associated higher mortality.

Different levels of hypoglycemia in patients with type 2 diabetes, their achieved mean HbA1c vs. all-cause and cardiovascular mortality

“…We observed that those patients with mean HbA1c <6.0% consisted of minor portions of all levels of hypoglycemia (~4%). HbA1c value might be underestimated in patients with diabetes in the status of advanced chronic kidney disease and anemia, and hypoglycemia was more frequent in those with mean HbA1c 7.0–9.9% rather than in those with mean HbA1c <6.0% in our previous study [ 12 ]. Similar to our findings, in the diabetes and aging study of Kaiser Permanente North California integrated healthcare delivery system, hypoglycemia was common in all levels of glycemic control, and the authors did not find patients with the lowest HbA1c levels were at high risk of hypoglycemia [ 22 ].…”

“…FEMH, a tertiary medical center, has set up its electronic medical database of out-patient and emergency visits since 2001 and 2008, respectively. The details of our out-patient electronic database have been described in previous studies [12][13][14]. In this study, we extended our database to emergency visits electronic records.…”

“…From the out-patient medical records, the information of oral hypoglycemic agents, insulin, antihypertensive and antilipid medications were collected during the study period as stated in our previous studies [12][13][14]. We also explored each study participant's comorbidities and complications as described in S1 Table [15].…”

“…Neither a temporal nor dose–response relationship between hypoglycemia and mortality observed in the ACCORD trial produces the uncertainty about the causal relationship between hypoglycemia and all-cause or cardiovascular mortality. In our previous study [ 12 ], higher risks associated with low mean glycated hemoglobin A1c (HbA1c) and all-cause mortality were attenuated by adjustment of medications and comorbidities, but the relationship between mean HbA1c and hypoglycemia patients has not been clearly elucidated.…”

The relationship of low-density lipoprotein cholesterol and all-cause or cardiovascular mortality in patients with type 2 diabetes: a retrospective study

“…The age of each study subject was determined using the difference between the index date and the date of birth. Similar to the study design of our previous reports ( Ooi et al., 2022 ; Yeh et al., 2022 ), we collected all information on oral antidiabetic agents, insulin, antihypertensive, and antilipid medications (statins and fibrates) between the index date and the end of follow-up, which is either death or censoring. Various cardiovascular risk factors (coronary artery disease (ICD-9-CM: 410, 411, 413, 414 or ICD-10-CM: I20-I25), heart failure (ICD-9-CM: 428 or ICD-10-CM: I50), and cerebrovascular disease (ICD-9-CM: 433-436, 437.0, 437.1 or ICD-10-CM: I63-I66, I67.2, I67.3, I67.6, I67.81, I67.82, I67.9)) retrieved from the outpatient medical records between the index date and the end of follow-up were considered as potential confounders ( Glasheen, Renda & Dong, 2017 ).…”