Abstract
OBJECTIVE
To compare the risk of hospitalization for infection among patients who achieve intensive versus relaxed glycemic control.
RESEARCH DESIGN AND METHODS
This retrospective cohort study included adults age ≥65 years with type 2 diabetes from an integrated health care delivery system. Negative binomial models were used to estimate incidence rates and relative risk (RR) of hospitalization for infections (respiratory; genitourinary; skin, soft tissue, and bone; and sepsis), comparing two levels of relaxed (hemoglobin A1c [HbA1c] 7% to <8% and 8% to <9%) with intensive (HbA1c 6% to <7%) glycemic control from 1 January 2019 to 1 March 2020.
RESULTS
Among 103,242 older patients (48.5% with HbA1c 6% to <7%, 35.3% with HbA1c 7% to <8%, and 16.1% with HbA1c 8% to <9%), the rate of hospitalization for infections was 51.3 per 1,000 person-years. Compared with HbA1c 6% to <7%, unadjusted risk of hospitalization for infections was significantly elevated among patients with HbA1c 8% to <9% (RR 1.25; 95% CI 1.13, 1.39) but not among patients with HbA1c 7% to <8% (RR 0.99; 95% CI 0.91, 1.08), and the difference became nonsignificant after adjustment. Across categories of infections, the adjusted RR of hospitalization was significantly higher among patients with HbA1c 8% to <9% only for skin, soft tissue, and bone infection (RR 1.33; 95% CI 1.05, 1.69).
CONCLUSIONS
Older patients with type 2 diabetes who achieve relaxed glycemic control levels endorsed by clinical guidelines are not at significantly increased risk of hospitalization for most infections, but HbA1c 8% to <9% is associated with an increased risk of hospitalization for skin, soft tissue, and bone infections.
Graphical Abstract
Introduction
Type 2 diabetes (T2D) is a common metabolic disorder affecting ∼25% of older Americans. Chronic hyperglycemia, the hallmark of diabetes, contributes to the development of microvascular and macrovascular complications, including neuropathy, nephropathy, retinopathy, cardiovascular disease, and stroke. In addition, hyperglycemia disrupts the immune response to infection and increases susceptibility to a broad range of infections in people with T2D (1). Professional society guidelines endorse intensive glycemic control, defined by a target hemoglobin A1c (HbA1c) <7%, for young, healthy patients with T2D to reduce the risk of these complications (2). In contrast, geriatric diabetes guidelines recommend relaxed glycemic control (e.g., target HbA1c 7% to <9%) for many older patients with T2D who have multiple comorbidities, poor health, or limited life expectancy (3,4). Relaxed glycemic control is intended to minimize hypoglycemia risk, reduce treatment burden, and avoid futile treatment in people for whom time to benefit may exceed life expectancy (3,4). However, relaxed glycemic control may increase the short-term risk of serious infection in older adults with T2D (5).
Diabetes confers a significant risk for hospitalization for infection (1). While the association of poor glycemic control (HbA1c >9%) in individuals with diabetes and the risk of infection is well established (5–8), few studies have examined the risk of hospitalization for infection associated with levels of glycemic control that are considered relaxed (HbA1c 7% to <9%) but that are generally recommended by clinical guidelines for many older adults (3,4). The few extant studies reported considerable variation in effect sizes and glycemic thresholds at which infection risk increased (9–11).
Accordingly, evidence is needed regarding the extent to which relaxed glycemic control is independently associated with an increased risk of hospitalization for infection. A clinical trial of relaxed glycemic control and infection as a primary outcome is premature given the lack of evidence of harms, but well-designed, observational studies of patients with T2D in usual care settings can offer reliable evidence or inform decisions to conduct a trial (12). Our study evaluated whether relaxed glycemic control (HbA1c 7% to <9% as recommended by diabetes guidelines for many older adults with T2D) compared with intensive glycemic control (HbA1c 6% to <7%) is associated with an increased, short-term (12-month) risk of hospitalization for infection.
Research Design and Methods
Study Setting
Patients were members of a large, integrated health care delivery system, Kaiser Permanente Northern California (KPNC), who were identified in the KPNC Diabetes Registry, a well-characterized population, maintained continuously since 1993 (13). Registry inclusion is based on a validated algorithm incorporating multiple data sources, including pharmacy records, laboratory data, and outpatient, emergency department, and inpatient diagnoses of diabetes (14).
Study Population
Patients at KPNC were included if they were age ≥65 years at baseline (1 January 2019), were identified as having diabetes prior to 1 January 2018, and had an HbA1c result between 1 January 2019 and 31 December 2019. We excluded patients who had a gap of ≥3 months in health plan membership in the 24 months prior to baseline or a gap of ≥3 months in prescription benefits in the 12 months prior to baseline to maximize capture of important covariates. Patients were also excluded if they had type 1 diabetes, end-stage renal disease (ESRD), a heritable hemoglobinopathy, or an HbA1c widely considered harmful for older patients (i.e., <6% [considered overtreatment] or ≥9% [poor glycemic control]). We focused on the range of HbA1c levels that are currently recommended by guidelines to generate clinically informative evidence.
Main Exposure
The exposure of interest was the first HbA1c collected on or after 1 January 2019, sorted into three glycemic categories: HbA1c 6% to <7% (reference group) and HbA1c 7% to <8% and HbA1c 8% to <9% (relaxed glycemic control groups). An HbA1c result was ineligible (not included) if collected 1) during the 2 weeks immediately prior to a hospital admission for infection (due to the potential effect of the infection itself on glycemia), 2) during a 3-month washout period after any hospital admission (due to inpatient procedures or interventions that could affect accuracy of HbA1c), or 3) during a 3-month washout period after a red blood cell transfusion or administration of erythropoietin.
Main Outcome
The primary outcome of interest was a count of hospitalizations for infection within 12 months after the first eligible HbA1c. Patients could have multiple hospitalizations for infection during the follow-up period. Since infections can co-occur with other diagnoses (e.g., pneumonia with atrial fibrillation), we included principal, primary, and secondary hospital diagnoses of infection, consistent with prior literature (5,7). The infection diagnosis had to be present on admission (15) to exclude nosocomial infections. Follow-up was censored at 12 months after the first eligible HbA1c, death, hospice admission, disenrollment, or 1 March 2020, whichever occurred first. We censored follow-up at 1 March 2020 to exclude the period of the coronavirus disease 2019 pandemic during which utilization patterns were dramatically altered. Infections were defined by ICD-10 codes and groupings of ICD-10 codes based on the Clinical Classifications Software and Hierarchical Condition Category clinical classification systems, which both categorize the thousands of ICD-10 codes into clinically meaningful groups. Using these groupings, four categories of infections that are common among individuals with diabetes (16–19) were created, including 1) respiratory (e.g., pneumonia, bronchitis, influenza, lung abscess); 2) genitourinary and kidney (e.g., cystitis, acute pyelonephritis, renal and perinephric abscess, prostatitis); 3) bone, skin, and soft tissue (e.g., cellulitis, necrotizing fasciitis, Fournier gangrene, osteomyelitis, foot ulcer, infective arthropathies or myositis, diabetic foot infections); and 4) sepsis and bacteremia.
Covariates
Covariates were all assessed at baseline using data closest to 1 January 2019 and included age, sex, BMI, health behaviors (medication adherence [continuous medication gap] [20], exercise [21], smoking, alcohol use [22]), diabetes medications (sulfonylureas, metformin, insulin, and other dispensed within 6 months prior to eligible HbA1c), established microvascular complications, comorbidities (anemia, liver disease, chronic obstructive pulmonary disease [COPD], heart failure, stroke, chronic kidney disease), use of steroids (any injected or oral corticosteroids with any dispensing within 100 days of the eligible HbA1c), limited English proficiency (23), neighborhood deprivation index (24), and medication financial assistance. In sensitivity analyses, we additionally adjusted for race and ethnicity and diabetes duration.
Statistical Analyses
Negative binomial models were used to estimate incidence rates and associations (incidence rate ratios) between each glycemic category and counts of hospitalizations relative to the reference (HbA1c 6 to <7%). Models were based on the first eligible HbA1c (exposure of interest) after 1 January 2019. Counts of hospitalization for infection (outcome of interest), overall, and by the four categories of infection were collected during the 12 months following the eligible HbA1c. Absolute risk differences across glycemic categories were estimated. Log follow-up time was included as an offset to account for differential person-time. Two-sided tests were used with a type I error rate of 5%. To inform specification of adjusted models and address confounding, we constructed a directed acyclic graph (DAG) to visually represent relationships among HbA1c, infections, and our covariates. We used published methods and Daggity 3.0 software to analyze the DAG to determine a minimal sufficient adjustment set needed for an unbiased estimate of the total effect of glycemic category on hospitalization for infection (25).
Results
From 160,923 individuals with diabetes and age ≥65 years at baseline, 8,551 were excluded due to gaps in KPNC membership of ≥3 months or gaps in KPNC prescription benefits of ≥3 months, resulting in a cohort of 152,372 patients. From this cohort, 3,914 were excluded due to having either type 1 diabetes (n = 1,371) or an unknown type of diabetes (n = 2,543), 4,097 were excluded due to having a diagnosis of ESRD, 1,595 were excluded for having a diagnosis of heritable hemoglobinopathy, and 12,144 were excluded because they did not have a baseline HbA1c available. After excluding HbA1c values that were ineligible (due to occurring within 2 weeks prior to inpatient admission for infection [n = 1,664 HbA1c values], within 3 months after any inpatient discharge [n = 10,475 HbA1c values], within 3 months after blood transfusion [n = 1,338 HbA1c values], and within 3 months after dispensing of erythropoietin [n = 249 HbA1c values]), there were 128,792 patients in the cohort. After including only patients with HbA1c values between 6% and <9%, there were 103,242 patients in the final cohort.
Among the 103,242 patients, 48.5% had an HbA1c between 6% and <7%, 35.3% had an HbA1c 7% to <8%, and 16.1% had an HbA1c 8% to <9% (Table 1). As expected, the proportion of patients not taking any diabetes medications differed across HbA1c categories (33.4% for those with HbA1c 6% to <7%, 13.5% for those with HbA1c 7% to <8%, and 7.6% for those with HbA1c 8% to <9%). Similarly, the proportion of patients taking insulin differed across HbA1c categories (9.3% for those with HbA1c 6% to <7%, 24.7% for those with HbA1c 7% to <8%, and 40.5% for those with HbA1c 8% to <9%).
Table 1.
Characteristics of older patients with T2D included in the study, overall and by levels of glycemic control
| HbA1c | |||||
|---|---|---|---|---|---|
| Full cohort (n = 103,242) |
6 to <7% (n = 50,114) | 7 to <8% (n = 36,486) | 8 to <9% (n = 16,642) | P | |
| Age, mean (SD), years | 74.4 (6.8) | 75.0 (7.0) | 74.0 (6.5) | 73.8 (6.5) | <0.0001 |
| Age-group, years | |||||
| 65–69 | 31,388 (30.4) | 14,069 (28.1) | 11,670 (32.0) | 5,649 (33.9) | <0.0001 |
| 70–74 | 30,469 (29.5) | 14,274 (28.5) | 11,261 (30.9) | 4,934 (29.6) | — |
| 75–79 | 19,719 (19.1) | 9,935 (19.8) | 6,793 (18.6) | 2,991 (18.0) | — |
| 80–84 | 12,510 (12.1) | 6,578 (13.1) | 4,047 (11.1) | 1,885 (11.3) | — |
| ≥85 | 9,156 (8.9) | 5,258 (10.5) | 2,715 (7.4) | 1,183 (7.1) | — |
| Female sex | 50,746 (49.2) | 25,497 (50.9) | 17,541 (48.1) | 7,708 (46.3) | <0.0001 |
| Race and ethnicity | |||||
| White | 48,167 (46.7) | 24,201 (48.3) | 16,489 (45.2) | 7,477 (44.9) | <0.0001 |
| Black | 8,574 (8.3) | 4,400 (8.8) | 2,849 (7.8) | 1,325 (8.0) | — |
| Hispanic | 15,471 (15.0) | 6,990 (13.9) | 5,614 (15.4) | 2,867 (17.2) | — |
| Asian | 25,049 (24.3) | 11,668 (23.3) | 9,423 (25.8) | 3,958 (23.8) | — |
| Other | 5,981 (5.8) | 2,855 (5.7) | 2,111 (5.8) | 1,015 (6.1) | — |
| BMI, mean (SD), kg/m2 | 30.1 (6.4) | 29.8 (6.3) | 30.3 (6.4) | 30.7 (6.5) | <0.0001 |
| T2D duration ≥10 years | 54,296 (52.6) | 21,302 (42.5) | 21,660 (59.4) | 11,334 (68.1) | — |
| Preferred language | |||||
| English | 91,395 (88.5) | 44,714 (89.2) | 32,117 (88.0) | 14,564 (87.5) | <0.0001 |
| LEP, Spanish | 5,304 (5.1) | 2,299 (4.6) | 1,959 (5.4) | 1,046 (6.3) | — |
| LEP, other language | 6,532 (6.3) | 3,098 (6.2) | 2,404 (6.6) | 1,030 (6.2) | — |
| Neighborhood deprivation index, quartile | |||||
| 1st (Least deprived) | 30,083 (29.1) | 14,978 (29.9) | 10,576 (29.0) | 4,529 (27.2) | <0.0001 |
| 2nd | 38,237 (37.0) | 18,659 (37.2) | 13,414 (36.8) | 6,164 (37.0) | — |
| 3rd | 19,049 (18.5) | 8,959 (17.9) | 6,909 (18.9) | 3,181 (19.1) | — |
| 4th (Most deprived) | 15,873 (15.4) | 7,518 (15.0) | 5,587 (15.3) | 2,768 (16.6) | — |
| Financial assistance | 8,285 (8.0) | 4,019 (8.0) | 2,858 (7.8) | 1,408 (8.5) | 0.0474 |
| Comorbid conditions | |||||
| Anemia | 29,667 (28.7) | 14,185 (28.3) | 10,432 (28.6) | 5,050 (30.3) | <0.0001 |
| Heart failure | 12,093 (11.7) | 5,983 (11.9) | 4,032 (11.1) | 2,078 (12.5) | <0.0001 |
| Stroke | 4,134 (4.0) | 2,019 (4.0) | 1,391 (3.8) | 724 (4.4) | 0.0125 |
| Liver disease | 14,598 (14.1) | 7,015 (14.0) | 5,157 (14.1) | 2,426 (14.6) | 0.1775 |
| Metastatic cancer | 2,630 (2.5) | 1,364 (2.7) | 850 (2.3) | 416 (2.5) | 0.0013 |
| COPD | 24,104 (23.3) | 11,961 (23.9) | 8,262 (22.6) | 3,881 (23.3) | 0.0001 |
| Thyroid disease | 21,675 (21.0) | 11,002 (22.0) | 7,349 (20.1) | 3,324 (20.0) | <0.0001 |
| Rheumatoid arthritis | 2,255 (2.2) | 1,178 (2.4) | 750 (2.1) | 327 (2.0) | 0.0015 |
| History of amputation | 806 (0.8) | 308 (0.6) | 308 (0.8) | 190 (1.1) | <0.0001 |
| CKD | |||||
| Stage 0, GFR ≥90 mL/min/1.73 m2 without albuminuria | 19,266 (18.8) | 9,570 (19.3) | 6,925 (19.1) | 2,771 (16.8) | <0.0001 |
| Stage 1, GFR ≥90 mL/min/1.73 m2 with albuminuria | 6,157 (6.0) | 2,579 (5.2) | 2,405 (6.6) | 1,173 (7.1) | — |
| Stage 2, GFR 60–89 mL/min/1.73 m2 | 51,851 (50.7) | 25,795 (52.0) | 18,065 (49.9) | 7,991 (48.4) | — |
| Stage 3A, GFR 45–59 mL/min/1.73 m2 | 16,120 (15.8) | 7,593 (15.3) | 5,679 (15.7) | 2,848 (17.3) | — |
| Stage 3B, GFR 30–44 mL/min/1.73 m2 | 7,240 (7.1) | 3,265 (6.6) | 2,560 (7.1) | 1,415 (8.6) | — |
| Stage ≥4, GFR ≤ 29 mL/min/1.73 m2 and not on dialysis | 1,632 (1.6) | 800 (1.6) | 538 (1.5) | 294 (1.8) | — |
| Stage 5/dialysis, GFR <15 mL/min/1.73 m2 or GFR >15 mL/min/1.73 m2 and on dialysis | 32 (0.0) | 16 (0.0) | 14 (0.0) | 2 (0.0) | — |
| Proliferative diabetic retinopathy | 2,862 (2.8) | 772 (1.5) | 1,188 (3.3) | 902 (5.4) | <0.0001 |
| Peripheral vascular disease (ever) | 14,117 (13.7) | 6,761 (13.5) | 4,850 (13.3) | 2,506 (15.1) | <0.0001 |
| Diabetes therapy | |||||
| Not on any diabetes medications | 22,938 (22.2) | 16,755 (33.4) | 4,916 (13.5) | 1,267 (7.6) | <0.0001 |
| Sulfonylureas | 34,597 (33.5) | 10,855 (21.7) | 15,165 (41.6) | 8,577 (51.5) | <0.0001 |
| TZDs | 2,522 (2.4) | 837 (1.7) | 1,126 (3.1) | 559 (3.4) | <0.0001 |
| GLP-1 receptor agonists | 152 (0.1) | 44 (0.1) | 68 (0.2) | 40 (0.2) | <0.0001 |
| SGLT2 inhibitors | 105 (0.1) | 22 (0.0) | 57 (0.2) | 26 (0.2) | <0.0001 |
| DPP-4 inhibitors | 783 (0.8) | 184 (0.4) | 353 (1.0) | 246 (1.5) | <0.0001 |
| Metformin | 58,143 (56.3) | 23,727 (47.3) | 23,532 (64.5) | 10,884 (65.4) | <0.0001 |
| Insulin | 20,422 (19.8) | 4,659 (9.3) | 9,019 (24.7) | 6,744 (40.5) | <0.0001 |
| Other | 139 (0.1) | 44 (0.1) | 61 (0.2) | 34 (0.2) | 0.0002 |
| Antihypertensive therapy | 89,036 (86.2) | 42,556 (84.9) | 31,933 (87.5) | 14,547 (87.4) | <0.0001 |
| Lipid-lowering therapy | 85,046 (82.4) | 40,300 (80.4) | 30,770 (84.3) | 13,976 (84.0) | <0.0001 |
| Corticosteroids (oral or intravenous) | 5,128 (4.0) | 4,051 (3.9) | 1,417 (3.9) | 713 (4.3) | 0.0304 |
| Smoking | |||||
| Never | 56,095 (56.9) | 26,908 (56.1) | 20,215 (58.0) | 8,972 (56.9) | <0.0001 |
| Former | 38,158 (38.7) | 18,883 (39.4) | 13,189 (37.8) | 6,086 (38.6) | — |
| Current | 4,298 (4.4) | 2,149 (4.5) | 1,448 (4.2) | 701 (4.4) | — |
| Physical activity, min/week | |||||
| 0 | 48,905 (49.9) | 23,735 (49.7) | 17,001 (49.2) | 8,169 (52.1) | <0.0001 |
| 1–149 | 23,408 (23.9) | 11,275 (23.6) | 8,328 (24.1) | 3,805 (24.3) | — |
| ≥150 | 25,674 (26.2) | 12,720 (26.6) | 9,254 (26.8) | 3,700 (23.6) | — |
| Excessive alcohol use | 2,800 (3.2) | 1,583 (3.7) | 903 (2.9) | 314 (2.2) | <0.0001 |
| Medication adherence | <0.0001 | ||||
| Not calculated (none or only single dispensing) | 5,056 (4.9) | 2,966 (5.9) | 1,456 (4.0) | 634 (3.8) | |
| Adherent: 0–20% CMG | 85,796 (83.1) | 41,388 (82.6) | 30,777 (84.4) | 13,631 (81.9) | — |
| Nonadherent: > 20% CMG | 12,390 (12.0) | 5,760 (11.5) | 4,253 (11.7) | 2,377 (14.3) | — |
Data are n (%) unless otherwise indicated. CMG, continuous medication gap; DPP-4, dipeptidyl peptidase 4; GFR, glomerular filtration rate; GLP-1, glucagon-like peptide 1; LEP, limited English proficiency; SGLT2, sodium–glucose cotransporter 2; TZD, thiazolidinedione.
During follow-up (mean [SD] 9.8 [2.8] months), 3.6% of patients were hospitalized for any of the four categories of infection, with 2.2% hospitalized for sepsis, 1.4% for respiratory infection, 1.2% for urinary infection, and 0.8% for bone, skin, and soft tissue infection. The crude rate of hospitalization for the four categories of infection was 51.3 per 1,000 person-years.
The rate of hospitalizations for infection overall and by category was similar among patients with HbA1c 7% to <8% compared with those with HbA1c 6% to <7% (Fig. 1). In contrast, the rate of hospitalization for infection, and for categories of sepsis and bone, skin, and soft tissue infections, was higher among patients with HbA1c 8% to <9% compared with those with HbA1c 6% to <7%.
Figure 1.
Crude rates of hospitalization for infections by all four categories and by each category of infection (sepsis and bacteremia, respiratory, genitourinary and kidney, and skin, soft tissue, and bone).
Based on the DAG, the minimal sufficient adjustment set included age, sex, BMI, established microvascular complications, chronic kidney disease (stage ≥3), comorbidities (anemia, liver disease, COPD, heart failure, stroke, and other), diabetes medications (sulfonylurea, metformin, insulin, and other), health behaviors (exercise, alcohol, medication adherence), treatment with steroids, limited English proficiency, neighborhood deprivation index, and medical financial assistance (see the Supplementary Material for the DAG). After accounting for the minimal sufficient adjustment set, the relative risk (RR) of hospitalization for infection was not significantly different among patients with HbA1c 7% to <8% (RR 0.99; 95% CI 0.90, 1.08) or among patients with HbA1c 8% to <9% (RR 1.03; 95% CI 0.92, 1.15) compared with those with HbA1c 6% to <7% (Table 2). When examined separately by category of infection, the same was true for genitourinary infections and sepsis and bacteremia. The risk of respiratory infection was lower among patients with HbA1c 7% to <8% (RR 0.85; 95% CI 0.74, 0.98) compared with those with HbA1c 6% to <7% but not different among those with HbA1c 8% to <9%. For bone, skin, and soft tissue infection, the rate was higher among patients with HbA1c 8% to <9% compared with those with HbA1c 6% to <7% (RR 1.33; 95% CI 1.05, 1.69) but not different from those with HbA1c 7% to <8%.
Table 2.
Rates of hospitalization for infections by HbA1c category, with unadjusted and adjusted RR according to glycemic control level category
| Glycemic control level | Patients, n | Hospitalizations, n | Crude rate (per 1,000 person-years) | Unadjusted RR (95% CI) | Adjusted RR (95% CI) | |
|---|---|---|---|---|---|---|
| All four infections | HbA1c 6% to <7% | 50,114 | 2,007 | 49.28 | Reference | Reference |
| HbA1c 7% to <8% | 36,486 | 1,519 | 49.39 | 0.99 (0.91, 1.08) | 0.99 (0.90, 1.08) | |
| HbA1c 8% to <9% | 16,642 | 863 | 61.54 | 1.25 (1.13, 1.39)* | 1.03 (0.92, 1.15) | |
| Sepsis and bacteremia | HbA1c 6% to <7% | 50,114 | 1,118 | 27.45 | Reference | Reference |
| HbA1c 7% to <8% | 36,486 | 901 | 29.30 | 1.06 (0.96, 1.17) | 1.02 (0.91, 1.15) | |
| HbA1c 8% to <9% | 16,642 | 489 | 34.87 | 1.27 (1.12, 1.44)* | 1.03 (0.89, 1.20) | |
| Respiratory | HbA1c 6% to <7% | 50,114 | 807 | 19.81 | Reference | Reference |
| HbA1c 7% to <8% | 36,486 | 539 | 17.53 | 0.88 (0.78, 0.99)* | 0.85 (0.74, 0.98)* | |
| HbA1c 8% to <9% | 16,642 | 313 | 22.32 | 1.12 (0.96, 1.30) | 0.90 (0.76, 1.07) | |
| Genitourinary and kidney | HbA1c 6% to <7% | 50,114 | 677 | 16.62 | Reference | Reference |
| HbA1c 7% to <8% | 36,486 | 491 | 15.97 | 0.94 (0.82, 1.08) | 1.01 (0.87, 1.18) | |
| HbA1c 8% to <9% | 16,642 | 262 | 18.68 | 1.12 (0.94, 1.32) | 0.93 (0.76, 1.14) | |
| Skin, bone, and soft tissue | HbA1c 6% to <7% | 50,114 | 395 | 9.70 | Reference | Reference |
| HbA1c 7% to <8% | 36,486 | 362 | 11.77 | 1.23 (1.03, 1.47)* | 1.10 (0.90, 1.35) | |
| HbA1c 8% to <9% | 16,642 | 234 | 16.69 | 1.79 (1.45, 2.22)* | 1.33 (1.05, 1.69)* |
*P < 0.05.
In sensitivity analyses, adjusted models additionally included race and ethnicity as well as diabetes duration (<10 years vs. ≥10 years). The results obtained from the adjusted models in the sensitivity analyses were consistent with those obtained with the analysis using DAG-selected models (see Supplementary Material).
There was no significant interaction between glycemic control category and age (65 to <75 years vs. ≥75 years) with risk of infections in the fully adjusted model (P for interaction = 0.50). Likewise, there was no significant interaction between glycemic control category and duration of diabetes (<10 years vs. ≥10 years) with risk of infections in the fully adjusted model from the sensitivity analysis (P for interaction = 0.077).
Conclusions
In this large retrospective study using data from an integrated health care delivery system, older patients with T2D who achieved relaxed glycemic control levels (HbA1c 7% to <8% or HbA1c 8% to <9%), which are endorsed by clinical guidelines, were not at significantly increased risk of hospitalization for most infections compared with older adults who achieved intensive glycemic control levels (HbA1c 6% to <7%). The one exception was the risk of hospitalization for skin, soft tissue, and bone infections, which was 33% higher among patients with HbA1c 8% to <9% compared with HbA1c 6% to <7%. Overall, these results are reassuring with respect to the safety of relaxed glycemic control among older adults with T2D.
Patients with diabetes are at increased risk of infection compared with those without diabetes (7,26,27), and hospitalization rates for infections are nearly four times higher in adults with versus without diabetes (16). In a prior study, among the top causes of admissions among adults with diabetes were sepsis (fourth) and pneumonia (fifth), and these admissions were each more common than admissions for acute myocardial infarction (17). Hyperglycemia, a hallmark of diabetes, disrupts the immune response to pathogens and increases susceptibility to infection (28,29). Hyperglycemia inhibits neutrophil migration, phagocytosis, superoxide production, and opsonization of bacteria and may weaken barrier defenses by causing breakdown of the skin and mucosa due to vascular dysfunction, neuropathy, and increased mucosal permeability (30). Hyperglycemia also provides a conducive environment for some pathogens (e.g., fungi [31,32]). Older adults with T2D are at increased risk of infection due to immunosenescence (33), comorbid conditions and diabetes complications, diminished mucosal barriers, decreased cough reflex, malnutrition, and changes in the urinary tract system (34,35). Thus, understanding the risk of infection associated with relaxed glycemic control levels is highly relevant to the care of older patients with T2D.
While the association of poor glycemic control (HbA1c >9%) and the risk of infection is well established (5–8,27,36–38), few prior studies have focused on risks associated with relaxed glycemic control levels (HbA1c 7% to <9%) recommended by geriatric diabetes guidelines. Moreover, existing studies have reported considerable variation in effect sizes (including no effect) and glycemic thresholds at which infection risk increased (9–11). Some of the conflicting findings may be due to methodological issues, including inadequate control of confounding (e.g., sicker patients are more likely to experience hyperglycemia and infection) or reverse causality (HbA1c measured during infection may reflect infection-related stress hyperglycemia). Our study used causal methodology to identify and adjust for critical confounders (including type of glucose-lowering medications) and excluded HbA1c levels within 2 weeks of admission to avoid some of these methodological problems.
Our findings provide some reassurance with respect to the safety of geriatric diabetes guidelines, which recommend relaxed control (i.e., HbA1c levels between 7% to <9%) for many older adults with T2D, including those who have multiple comorbidities, poor health, or limited life expectancy. The intent of relaxed glycemic control is to minimize the risk of hypoglycemia, reduce treatment burden, and avoid futile treatment in older adults with T2D for whom time to benefit (≥10 years for microvascular disease risk reduction) may exceed life expectancy (4,39). In older adults with limited life expectancy, short-term outcomes, such as hypoglycemia and infection, may be more relevant compared with long-term outcomes, such as development of microvascular complications. Our findings demonstrate the overall safety of achieving relaxed glycemic control levels in these patients, particularly HbA1c levels 7% to <8%, with respect to hospitalization for infection. Notably, in our study, the crude risk of hospitalization for infection tended to be significantly higher in patients with relaxed glycemic control levels, but these associations were greatly attenuated after adjustment for confounders.
The risk of hospitalizations for skin, bone, and soft tissue infections was significantly higher in our study among patients with HbA1c 8% to <9% compared with those who achieved intensive glycemic control. Hyperglycemia may play a more important role in the development of these types of infections compared with the development of sepsis or pneumonia. In addition, prevalent complications of diabetes, such as neuropathy or vascular disease, may be tightly associated with both higher HbA1c levels and skin, bone, and soft infections. In contrast, hospitalizations for respiratory infections may co-occur with cardiac events, with more moderate HbA1c levels potentially protective.
Our study has some limitations that need to be considered. This study used observational data to examine the association of relaxed glycemic control and hospitalization for infection and is subject to confounding. However, we found that our adjustment greatly attenuated and mostly rendered the associations not significant. We used a single HbA1c level to evaluate its association with the outcomes, but HbA1c levels may change over time. A recent analysis found that greater HbA1c variability is more strongly associated with the risk of infection compared with HbA1c average over a period of 5 years (40). Our study considered a relatively short follow-up time (12 months), mitigating the concern of using a single HbA1c. In addition, HbA1c may reflect glycemia differently by health conditions (e.g., ESRD, anemia, and sickle cell trait can affect glycosylation) and by race and ethnicity. We excluded patients with known ESRD or heritable hemoglobinopathies but could not account for all factors that affect HbA1c levels. Our study was designed to characterize hospitalizations for acute infection but did not address chronic and less severe infections that do not lead to hospitalizations and may also influence glycemic goals and achieved HbA1c. Because we required that the diagnosis of infection was present on admission, we may be underestimating the rate of sepsis since patients may progress to sepsis during hospitalization when admitted for one of the other infection types. Finally, because the study focused on hospitalizations for infection, it may be subject to surveillance bias where patients with poor glycemic control levels may be more likely to be hospitalized rather than treated at home for their infection. This is probably less of an issue in our study because we did not include patients with HbA1c levels >9%.
In conclusion, older patients with T2D who achieved relaxed glycemic control levels (HbA1c 7% to <8% or 8% to <9%), which are currently endorsed by most clinical guidelines, were not at significantly increased risk of hospitalization for most infections compared with older adults who achieved intensive glycemic control levels (HbA1c 6% to <7%). However, risk of hospitalization for skin, soft tissue, and bone infections was significantly higher in those with HbA1c levels 8% to <9%. If achieving an HbA1c 7% to <8% is feasible without undue risk of hypoglycemia or treatment burden, this level of relaxed glycemic control may provide an added layer of safety.
This article contains supplementary material online at https://doi.org/10.2337/figshare.27138732.
Article Information
Funding. This work was supported in part by National Institute on Aging grants R56AG074986 and R01AG063391 and National Institute of Diabetes and Digestive and Kidney Diseases grant P30DK092924.
This work was also supported by the Paul Beeson Emerging Leaders Career Development Award (grant K76AG064548-01 to H.Z.), the HHMI Emerging Pathogens Initiative, and the Yale Pepper Center (2P30AG021342-21).
The contents do not represent the views of the National Institute on Aging, National Institute of Diabetes and Digestive and Kidney Diseases, or the U.S. government.
Duality of Interest. K.J.L. receives support from the National Institutes of Health, Patient-Centered Outcomes Research Institute, and the Department of Veterans Affairs to conduct research; royalties from UpToDate to write and edit content; and other support from the Centers for Medicare & Medicaid Services to develop and evaluate publicly reported quality measures. A.J.K. receives grant support from the National Institute of Aging, National Institute of Diabetes and Digestive and Kidney Diseases, The Leona M. and Harry B. Helmsley Charitable Trust, Patient-Centered Outcomes Research Institute, and Dexcom. H.M.M. and M.M.P. receive grant support from the National Institutes of Health, The Leona M. and Harry B. Helmsley Charitable Trust, and Dexcom. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. K.J.L. wrote the manuscript. K.J.L. and A.J.K. conceived the study. C.L., J.Y.L., M.M.P., and A.J.K. designed the statistical plan. J.Y.L. performed the analysis. J.Y.L., H.H.M., M.M.P., and A.J.K. collected all the data for the study. All authors made critical revisions to and approved the manuscript. V.X.L. and H.Z. contributed to study design and interpretation of findings. A.J.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Handling Editors. The journal editors responsible for overseeing the review of the manuscript were John B. Buse and Stephanie L. Fitzpatrick.
Funding Statement
This work was supported in part by National Institute on Aging grants R56AG074986 and R01AG063391 and National Institute of Diabetes and Digestive and Kidney Diseases grant P30DK092924.
References
- 1. Holt RIG, Cockram CS, Ma RCW, Luk AOY.. Diabetes and infection: review of the epidemiology, mechanisms and principles of treatment. Diabetologia 2024;67:1168–1180 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. American Diabetes Association . 6. Glycemic goals and hypoglycemia: Standards of Care in Diabetes—2024. Diabetes Care 2024;47(Suppl. 1):S111–S125 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. American Diabetes Association . 13. Older adults: Standards of Care in Diabetes—2024. Diabetes Care 2024;47(Suppl. 1):S244–S257 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. LeRoith D, Biessels GJ, Braithwaite SS, et al. Treatment of diabetes in older adults: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2019;104:1520–1574 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Critchley JA, Carey IM, Harris T, DeWilde S, Hosking FJ, Cook DG.. Glycemic control and risk of infections among people with type 1 or type 2 diabetes in a large primary care cohort study. Diabetes Care 2018;41:2127–2135 [DOI] [PubMed] [Google Scholar]
- 6. Boyko EJ, Fihn SD, Scholes D, Abraham L, Monsey B.. Risk of urinary tract infection and asymptomatic bacteriuria among diabetic and nondiabetic postmenopausal women. Am J Epidemiol 2005;161:557–564 [DOI] [PubMed] [Google Scholar]
- 7. Mor A, Dekkers OM, Nielsen JS, Beck-Nielsen H, Sørensen HT, Thomsen RW.. Impact of glycemic control on risk of infections in patients with type 2 diabetes: a population-based cohort study. Am J Epidemiol 2017;186:227–236 [DOI] [PubMed] [Google Scholar]
- 8. Zacay G, Hershkowitz Sikron F, Heymann AD.. Glycemic control and risk of cellulitis. Diabetes Care 2021;44:367–372 [DOI] [PubMed] [Google Scholar]
- 9. Bartelink ML, Hoek L, Freriks JP, Rutten GE.. Infections in patients with type 2 diabetes in general practice. Diabetes Res Clin Pract 1998;40:15–19 [DOI] [PubMed] [Google Scholar]
- 10. Hammar N, Farahmand B, Gran M, Joelson S, Andersson SW.. Incidence of urinary tract infection in patients with type 2 diabetes. Experience from adverse event reporting in clinical trials. Pharmacoepidemiol Drug Saf 2010;19:1287–1292 [DOI] [PubMed] [Google Scholar]
- 11. Boyko EJ, Fihn SD, Scholes D, Chen C-L, Normand EH, Yarbro P.. Diabetes and the risk of acute urinary tract infection among postmenopausal women. Diabetes Care 2002;25:1778–1783 [DOI] [PubMed] [Google Scholar]
- 12. Hernan M. Antihyperglycemic Therapy and Cardiovascular Risk: Design and Emulation of a Target Trial Using Healthcare Databases Washington, DC: Patient-Centered Outcomes Research Institute, 2019 [Google Scholar]
- 13. Karter AJ, Parker MM, Moffet HH, et al. Missed appointments and poor glycemic control: an opportunity to identify high-risk diabetic patients. Med Care 2004;42:110–115 [DOI] [PubMed] [Google Scholar]
- 14. Karter AJ, Schillinger D, Adams AS, et al. Elevated rates of diabetes in Pacific Islanders and Asian subgroups: the Diabetes Study of Northern California (DISTANCE). Diabetes Care 2012;36:574–579 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Department of Health and Human Services; Centers for Medicare & Medicaid Services. . Hospital-acquired conditions and present on admission indicator reporting provision. Accessed 10 October 2022. Available from https://www.cms.gov/files/document/wpoafactsheetpdf
- 16. Harding JL, Benoit SR, Gregg EW, Pavkov ME, Perreault L.. Trends in rates of infections requiring hospitalization among adults with versus without diabetes in the U.S., 2000-2015. Diabetes Care 2020;43:106–116 [DOI] [PubMed] [Google Scholar]
- 17. McCoy RG, Lipska KJ, Herrin J, Jeffery MM, Krumholz HM, Shah ND.. Hospital readmissions among commercially insured and Medicare advantage beneficiaries with diabetes and the impact of severe hypoglycemic and hyperglycemic events. J Gen Intern Med 2017;32:1097–1105 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Dobrzynski DM, Ndi DN, Zhu Y, Markus T, Schaffner W, Talbot HK.. Hospital readmissions after laboratory-confirmed influenza hospitalization. J Infect Dis 2020;222:583–589 [DOI] [PubMed] [Google Scholar]
- 19. Jensen AV, Faurholt-Jepsen D, Egelund GB, et al.; German Community-Acquired Pneumonia Competence Network (CAPNETZ) . Undiagnosed diabetes mellitus in community-acquired pneumonia: a prospective cohort study. Clin Infect Dis 2017;65:2091–2098 [DOI] [PubMed] [Google Scholar]
- 20. Steiner JF, Prochazka AV.. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol 1997;50:105–116 [DOI] [PubMed] [Google Scholar]
- 21. Grant RW, Schmittdiel JA, Neugebauer RS, Uratsu CS, Sternfeld B.. Exercise as a vital sign: a quasi-experimental analysis of a health system intervention to collect patient-reported exercise levels. J Gen Intern Med 2014;29:341–348 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Bryson CL, Au DH, Sun H, Williams EC, Kivlahan DR, Bradley KA.. Alcohol screening scores and medication nonadherence. Ann Intern Med 2008;149:795–804 [DOI] [PubMed] [Google Scholar]
- 23. Chew LD, Bradley KA, Boyko EJ.. Brief questions to identify patients with inadequate health literacy. Fam Med 2004;36:588–594 [PubMed] [Google Scholar]
- 24. Messer LC, Laraia BA, Kaufman JS, et al. The development of a standardized neighborhood deprivation index. J Urban Health 2006;83:1041–1062 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Textor J, van der Zander B, Gilthorpe MS, Liskiewicz M, Ellison GT.. Robust causal inference using directed acyclic graphs: the R package ‘dagitty’. Int J Epidemiol 2016;45:1887–1894 [DOI] [PubMed] [Google Scholar]
- 26. Carey IM, Critchley JA, DeWilde S, Harris T, Hosking FJ, Cook DG.. Risk of infection in type 1 and type 2 diabetes compared with the general population: a matched cohort study. Diabetes Care 2018;41:513–521 [DOI] [PubMed] [Google Scholar]
- 27. Breitling LP. Evidence of non-linearity in the association of glycemic control with influenza/pneumonia mortality: a study of 19 000 adults from the US general population. Diabetes Metab Res Rev 2016;32:111–120 [DOI] [PubMed] [Google Scholar]
- 28. Casqueiro J, Casqueiro J, Alves C.. Infections in patients with diabetes mellitus: a review of pathogenesis. Indian J Endocrinol Metab 2012;16(Suppl. 1):S27–S36 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Peleg AY, Weerarathna T, McCarthy JS, Davis TME.. Common infections in diabetes: pathogenesis, management and relationship to glycaemic control. Diabetes Metab Res Rev 2007;23:3–13 [DOI] [PubMed] [Google Scholar]
- 30. Thaiss CA, Levy M, Grosheva I, et al. Hyperglycemia drives intestinal barrier dysfunction and risk for enteric infection. Science 2018;359:1376–1383 [DOI] [PubMed] [Google Scholar]
- 31. Tsai S-T, Lin F-Y, Chen P-S, Chiang H-Y, Kuo C-C.. Three-year mortality in cryptococcal meningitis: hyperglycemia predict unfavorable outcome. PLoS One 2021;16:e0251749. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Donders GGG, Prenen H, Verbeke G, Reybrouck R.. Impaired tolerance for glucose in women with recurrent vaginal candidiasis. Am J Obstet Gynecol 2002;187:989–993 [DOI] [PubMed] [Google Scholar]
- 33. Fülöp T, Dupuis G, Witkowski JM, Larbi A.. The role of immunosenescence in the development of age-related diseases. Rev Invest Clin 2016;68:84–91 [PubMed] [Google Scholar]
- 34. Gavazzi G, Krause K-H.. Ageing and infection. Lancet Infect Dis 2002;2:659–666 [DOI] [PubMed] [Google Scholar]
- 35. Rajagopalan S. Serious infections in elderly patients with diabetes mellitus. Clin Infect Dis 2005;40:990–996 [DOI] [PubMed] [Google Scholar]
- 36. Pearson-Stuttard J, Blundell S, Harris T, Cook DG, Critchley J.. Diabetes and infection: assessing the association with glycaemic control in population-based studies. Lancet Diabetes Endocrinol 2016;4:148–158 [DOI] [PubMed] [Google Scholar]
- 37. Hine JL, de Lusignan S, Burleigh D, et al. Association between glycaemic control and common infections in people with type 2 diabetes: a cohort study. Diabet Med 2017;34:551–557 [DOI] [PubMed] [Google Scholar]
- 38. Ansari AS, de Lusignan S, Hinton W, Munro N, McGovern A.. The association between diabetes, level of glycaemic control and eye infection: cohort database study. Prim Care Diabetes 2017;11:421–429 [DOI] [PubMed] [Google Scholar]
- 39. American Geriatrics Society Expert Panel on Care of Older Adults with Diabetes Mellitus . Guidelines abstracted from the American Geriatrics Society guidelines for improving the care of older adults with diabetes mellitus: 2013 update. J Am Geriatr Soc 2013;61:2020–2026 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Carey IM, Critchley JA, Chaudhry UAR, et al. Effects of long-term HbA1c variability on serious infection risks in patients with type 2 diabetes and the influence of age, sex and ethnicity: a cohort study of primary care data. Diabetes Res Clin Pract 2024;211:111641. [DOI] [PubMed] [Google Scholar]