Obesity and Chronic Kidney Disease in US Adults With Type 1 and Type 2 Diabetes Mellitus

Abstract

Objective

Obesity is a global public health challenge and strongly associated with type 2 diabetes (T2D), but its burden and effects are not well understood in people with type 1 diabetes (T1D). Particularly, the link between obesity and chronic kidney disease (CKD) in T1D is poorly characterized.

Research Design and Methods

We included all T1D and, for comparison, T2D in the Geisinger Health System from 2004 to 2018. We evaluated trends in obesity (body mass index ≥ 30 kg/m²), low estimated glomerular filtration rate (eGFR) (≤60 mL/min/1.73m²), and albuminuria (urine albumin-to-creatinine ratio ≥ 30 mg/g). We used multivariable logistic regression to evaluate the independent association of obesity with CKD in 2018.

Results

People with T1D were younger than T2D (median age 39 vs 62 years). Obesity increased in T1D over time (32.6% in 2004 to 36.8% in 2018), while obesity in T2D was stable at ~60%. The crude prevalence of low eGFR was higher in T2D than in T1D in all years (eg, 30.6% vs 16.1% in 2018), but after adjusting for age differences, prevalence was higher in T1D than T2D in all years (eg, 16.2% vs 9.3% in 2018). Obesity was associated with increased odds of low eGFR in T1D [adjusted odds ratio (AOR) = 1.52, 95% CI 1.12-2.08] and T2D (AOR = 1.29, 95% CI 1.23-1.35).

Conclusions

Obesity is increasing in people with T1D and is associated with increased risk of CKD. After accounting for age, the burden of CKD in T1D exceeded the burden in T2D, suggesting the need for increased vigilance and assessment of kidney-protective medications in T1D.

Significant medical advances in the treatment of type 1 diabetes mellitus (T1D) have resulted in dramatic improvements in survival (1). Despite this improvement, the life expectancy of individuals with T1D is still 12 years less than the general population, largely attributable to increased risk of chronic disease, especially cardiovascular and kidney disease (2). In type 2 diabetes (T2D), obesity is thought to be a major contributor to the risk of kidney disease both indirectly through increases in blood pressure and hyperglycemia and directly through increased metabolic demands of higher body weight and the endocrine effects of adipose tissue (3-6). Despite these increased risks, screening for kidney disease in people with diabetes is suboptimal (6-8).

The prevalence of obesity has reached epidemic proportions in the US population, affecting approximately 42% of US adults in 2017-2018 (9). People with T1D may be at a higher risk of obesity than the general population because of the anabolic effects of insulin. In the Diabetes Control and Complications Trial (DCCT), intensive insulin treatment reduced mortality and microvascular complications (10) but was also associated with increased weight gain (11). In recent secondary analyses of the DCCT, participants in the intensive insulin arm of the trial who gained excessive amounts of weight were found to have similar rates of metabolic and cardiovascular disease as participants in the control arm (12,13), suggesting excessive weight gain may negate some of the benefits of tight glucose control. A recent global meta-analysis of over 5 million adults showed that increasing body mass index (BMI) was associated with higher risk for decreased estimated glomerular filtration rate (eGFR) (14), and weight loss has been shown to reduce risk of chronic kidney disease (CKD) in adults with T2D (15). However, the association between obesity and CKD in T1D is understudied (16).

The objective of this study was to characterize trends in obesity and CKD prevalence and their cross-sectional associations in a cohort of adults with T1D in a large healthcare system. As a comparison, we also evaluated the prevalence of obesity and CKD in a cohort of adults with T2D in the same healthcare system and within a general population using the National Health and Nutrition Examination Survey (NHANES), a nationally representative US cohort.

Materials and Methods

Study Populations

The study population of adults with T1D and T2D were drawn from the Geisinger Health System; we included all adult patients (≥18 years old) receiving primary care from the Geisinger Health System between January 1, 2004, and December 31, 2018. Geisinger is a healthcare system that serves 40 counties in central and northeastern Pennsylvania (17). The Geisinger electronic health record (EHR) system contains data on inpatient and outpatient visits, including laboratory data, prescription records, billing codes, and vitals. This study was approved by institutional review boards at the Geisinger Health System and Johns Hopkins Bloomberg School of Public Health. All data were deidentified, and consent was waived.

For comparison to a general US adult population without diagnosed diabetes, we used data from the 1999-2018 NHANES. NHANES is a cross-sectional, nationally representative sample of the civilian, noninstitutional US population conducted by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention. All protocols for NHANES were approved by the research ethics board of the NCHS. The surveys are conducted in 2-year cycles and consist of interviews and standardized physical exams at a mobile examination center, including laboratory tests. Detailed information on the protocols and procedures for NHANES are available elsewhere (18).

Adults With Type 1 Diabetes: Geisinger

Adults with T1D were identified using an algorithm (19) that has been validated in several EHR systems, including Geisinger (20). Briefly, patients with diabetes are identified by the presence of either (1) inpatient record with an International Classification of Diseases (ICD) diagnosis code for diabetes or (2) at least 2 of (a) outpatient visits with an ICD diagnosis code for diabetes, (b) prescriptions for antidiabetes medications, or (c) fasting glucose ≥ 126 mg/dL, nonfasting glucose ≥ 200 mg/dL, or hemoglobin A1c ≥ 6.5% (48 mmol/mol). From this cohort, patients with T1D were identified by having at least 1 prescription for insulin and either (1) ≥50% of ICD codes for diabetes specific for T1D (inclusive of all autoimmune forms of diabetes) and either a prescription of glucagon or no prescriptions for antidiabetic medications other than insulin or metformin, (2) positive results for diabetes autoantibodies (any detectable), or (3) negative results for c-peptide (<0.8 ng/mL) (Fig. 1). Patients could have received antidiabetic medications other than insulin or metformin as long as they met 1 of the other criteria for T1D (eg, detectable diabetes autoantibodies). The total population with T1D was 4060.

Figure 1.

Flow diagram of diabetes algorithm.

Adults With Type 2 Diabetes: Geisinger

Adults meeting criteria for any diabetes per the previously described algorithm but who did not meet criteria for T1D were classified as having T2D (n = 135 458).

General US Population Without Diagnosed Diabetes: NHANES

We included all adult participants (≥18 years old) without a self-reported history of diagnosed diabetes who attended in the physical exam during the 1999-2018 NHANES survey cycles. Participants were excluded if they were missing measurements of BMI, serum creatinine, or albumin-to-creatinine ratio (ACR; n = 47 611).

Period Prevalent Cohorts

For the Geisinger data, we constructed period prevalent cohorts for each calendar year including all eligible individuals with at least 1 measurement of body weight, creatinine, or urine ACR during the year and at least 1 height measurement over the full study period. To avoid misclassification of diabetes status for those diagnosed during the study period, no future data were used to determine diabetes status (eg, the 2004 cohort included all patients who met the previously described criteria using health record data from 1996 to 2004). For the NHANES data, each 2-year survey cycle was considered as a period prevalent cohort.

Measurements: Geisinger

In the Geisinger cohort, all outcome measurements were taken during routine clinical care. BMI, defined as [body weight (kg)]/[height (m)]², was calculated for each patient visit with a measured body weight and median height over all available records. Creatinine was measured at the Geisinger Medical Laboratory by isotope-dilution mass spectrometry traceable Roche enzymatic method (Roche Diagnostics, Indianapolis, IN, USA) according to manufacturer specifications. We calculated eGFR for each visit with a measured serum creatinine using the Chronic Kidney Disease Epidemiology Collaboration equation (21). For patients with multiple visits in a given calendar year, we calculated the mean BMI, eGFR, and ACR over all visits. Low eGFR was defined as mean eGFR < 60 mL/min/1.73 m², albuminuria was defined as mean ACR ≥ 30 mg/g, and obesity was defined as mean BMI ≥ 30 kg/m². Weight measurements for female patients taken at any date within 3 months of an encounter with a pregnancy-related ICD diagnosis code were excluded.

Kidney Function Screening Rates: Geisinger

We evaluated the screening rates for kidney disease in T1D and T2D by the proportion of patients who had at least 1 measurement of eGFR, at least 1 measurement of ACR, and at least 1 measurement of both eGFR and ACR. The American Diabetes Association and Kidney Disease Improving Global Outcomes recommend annual screening for people with diabetes, using both ACR and eGFR (22,23). To allow for some variation in the timing of annual screening visits (eg, visits in December 2016 and January 2018), we calculated the screening rates for 2-year intervals.

Trends in Antidiabetes Medications in Adults With T1D and T2D

We evaluated the proportion of participants with T1D and T2D who received at least 1 prescription for each of the major classes of antidiabetes medications [insulin, metformin, sulfonylureas, thiazolidinediones, glucagon-like peptide-1 (GLP-1) receptor agonists, and sodium-glucose transport protein 2 (SGLT2) inhibitors] in each of our period prevalent cohorts.

Measurements: NHANES

All NHANES measurements and laboratory specimens were collected by trained health technicians using standardized procedures (18). We calculated BMI from measured height and weight. Body weight measurements were taken using calibrated digital scales with the participants wearing a disposable gown, underwear, and slippers. Heights were measured using a stadiometer with a fixed backboard and adjustable head piece. We calculated eGFR from serum creatinine using the Chronic Kidney Disease Epidemiology Collaboration equation and calculated the ACR from urine albumin and urine creatinine. We calibrated serum creatinine levels using methodology recommended by the NCHS to account for changes in instrumentation and laboratory methods over the survey cycles (24). Similar to the definitions used in the Geisinger populations, low eGFR was defined as eGFR < 60 mL/min/1.73 m², albuminuria was defined as ACR ≥ 30 mg/g, and obesity was defined as BMI ≥ 30 kg/m². Weight measurements for pregnant female participants were excluded.

Statistical Methods: Geisinger

We described baseline characteristics of the study population at study entry (diagnosis of diabetes or 2004 if diagnosed prior to that year) by diabetes status and type using number (percentage) for categorical variables and mean (SD) or median [interquartile range (IQR)] for continuous variables. We calculated adjusted period prevalence using logistic regression with generalized estimating equations to account for intraindividual correlation and postestimation marginal effects. Models were adjusted for age in the period, sex, and race (white vs nonwhite). We fit separate models for T1D and T2D, and we estimated period prevalence using postestimation marginal effects for each period (calendar year) at age 50, female sex, and white race for both T1D and T2D. We evaluated the distributions of eGFR and ACR in 2018 using enhanced box plots. We used multivariable logistic regression to examine cross-sectional associations between obesity status and CKD. We adjusted for age, sex, race, and hypertension, defined by the presence of a diagnosis code for hypertension, a prescription for anti-hypertensive medications, or at least 2 office-measured blood pressures above 140/90 mmHg within 2 years. In adults with T2D, we additionally adjusted for the use of renoprotective antidiabetes medications (GLP-1 receptor agonists and SGLT2 inhibitors).

Statistical Methods: NHANES

We described baseline characteristics for NHANES participants at the time of the medical examination using number (percentage) for categorical variables and mean (SD) or median (IQR) for continuous variables. We calculated adjusted period prevalence using logistic regression and postestimation marginal effects for each 2-year period cohort at age 50, female sex, and white race. We evaluated the distributions of eGFR and ACR in 2017-2018 using enhanced box plots. We used multivariable logistic regression to examine cross-sectional associations between obesity status and CKD. We adjusted for age, sex, race, and hypertension, defined as either self-reported diagnosis of hypertension, a prescription for anti-hypertensive medications, or the mean of 3 standardized blood pressures above 140/90 mmHg.

All NHANES analyses incorporated weighting procedures recommended by NCHS to account for unequal probabilities of selection, oversampling, and nonresponse using the svy suite of commands in Stata.

Standard errors were estimated using Taylor series method. All statistical analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA), Stata version 15 (Stata Corp., College Station, TX, USA), or R version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Characteristics of Adult Patients With T1D and T2D in Geisinger

There were 4060 people with T1D and 135 458 people with T2D who had at least 1 visit within the Geisinger system over the study period (Table 1). People with T1D at study entry were younger than the general US population [median (IQR) age 39 (25-53) vs 43 (30-57) years] and much younger than people with T2D [median (IQR) age 62 (51-73) years]. The age of the US population is increasing, so we also calculated the median age for our 3 populations at the midpoint of the study period (2011 in Geisinger and 2011-2012 in NHANES) and found similar results. Consistent with the racial demographics of central Pennsylvania (the Geisinger Health System region), a higher proportion of our population of adults with T1D and T2D were white as compared to the general US population (91.2% and 93.6% vs 68.6%, respectively). There was a slight male predominance in T1D as compared to T2D and the general US population (51.5% vs 49.4% and 48.3%, respectively). Hypertension was much more common in adults with T1D and T2D as compared to the general US population (66.2% and 86.8% vs 26.9%, respectively). Use of angiotensin-converting enzyme inhibitors among hypertensive patients was higher in T1D and T2D than in the general US population (47.7% and 40.3% vs 25.8%, respectively).

Table 1.

Baseline demographics

	Geisinger Health System		NHANES
	Type 1 diabetes	Type 2 diabetes	Nondiabetic general population
	(n = 4060)	(n = 135 458)	(n = 47 611)
Agea	39 (25-53)	62 (51-73)	43 (30-57)
Female	48.5	50.6	51.7
Race
White	91.2	93.6	68.6
Black	4.7	2.9	10.8
Other	4.1	3.5	21.6
Antidiabetes medications
Insulina	100	12.8	—
Metformina	14.0	36.4	0.6
Sulfonylureaa	2.7	18.5	—
TZDa	1.0	4.3	—
DPP4-inhibitora	1.8	4.4	—
GLP-1 RAa	1.9	1.2	—
SGLT2-inhibitora	1.7	1.0	—
Hypertensionb	66.2	86.8	26.9
Any anti-hypertensive medicationa,c	71.6	71.8	65.4
ACE inhibitora,c	47.7	40.3	25.8
ARBa,c	12.7	13.9	15.3
β-blockera,c	28.2	40.5	22.7
Calcium channel blockera,c	15.8	20.4	13.9
Diureticsa,c	25.9	42.3	29.7

Data are given as median (interquartile range) or %.

Abbreviations: ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; DPP4, dipeptidyl-peptidase 4; GLP-1 RA, glucagon-like peptide-1 receptor agonist; NHANES, National Health and Nutrition Examination Survey; SGLT2, sodium-glucose transport protein 2; TZD, thiazolidinedione.

^aAge and medications for type 1 diabetes and type 2 diabetes participants are at first visit in the Geisinger Health System after diabetes diagnosis or 2004 for participants diagnosed with diabetes before 2004.

^bHypertension is defined as either International Classification of Diseases diagnosis codes, 2 measured blood pressures ≥ 140/90 mmHg or medications in the Geisinger Health System, and average measured blood pressure ≥ 140/90 mmHg or medications in NHANES.

^cAmong participants with hypertension.

Trends in Antidiabetic Medications in T1D and T2D

Metformin was the most prescribed non-insulin antidiabetes medication in adults with T1D, and prescription rates increased from 6.2% in 2004 to 13.8% in 2018 (Fig. 2). Prescriptions for GLP-1 receptor agonists and SGLT2 inhibitors were rare before 2016, but in 2018, 4.6% of T1D patients received at least 1 prescription for a GLP-1 receptor agonist and 3.1% received a prescription for an SGLT2 inhibitor. Prescriptions for other antidiabetic medications were uncommon. In adults with T2D, metformin was also the most prescribed medication, and use increased from 30.0% in 2004 to 41.1% in 2018. The use of sulfonylureas and thiazolidinediones decreased over the study period, while the use of newer classes of cardio- and renoprotective medications (GLP-1 receptor agonists and SGLT2 inhibitors) increased in the later part of the study period (Fig. 3).

Figure 2.

Trends in prescriptions for major classes of antidiabetes medications in adults with type 1 diabetes—Geisinger Health System, 2004-2018.

Figure 3.

Trends in prescriptions for major classes of antidiabetes medications in adults with type 2 diabetes—Geisinger Health System, 2004-2018.

Prevalence of Obesity in T1D and T2D Patients at Geisinger and General US Adult Population Without Diagnosed Diabetes in NHANES

The prevalence of obesity in people with T1D increased over time (32.6% in 2004 to 36.8% in 2018, P-trend = 0.0091). The crude and adjusted prevalence of obesity were similar in T1D and the general US population; trends were also similar (Fig. 4A and 4D). In contrast, the prevalence of obesity was high in adults with T2D at Geisinger and remained high over the full study period: crude prevalence of 58.1% in 2004 and 61.6% in 2018.

Figure 4.

Unadjusted and adjusted prevalence of obesity (A and D), reduced estimated glomerular filtration rate (B and E), and elevated albumin-to-creatinine ratio (C and F) in adults with type 1 diabetes, type 2 diabetes, and a nondiabetic general population.

Prevalence of CKD in T1D and T2D Patients at Geisinger and General Population Without Diagnosed Diabetes in NHANES

The crude prevalence of low eGFR was relatively stable over time in the population with T1D (17.5% in 2004 to 16.1% in 2018) in Geisinger. The crude prevalence of low eGFR in T1D was higher than the general US population but lower than the T2D population (T1D 17.5% vs US adults 5.7% vs T2D 26.6% in 2004), but after adjustment for differences in age, sex, and race, the prevalence was highest in adults with T1D (Fig. 4B and 4E). Prevalence of albuminuria followed the same pattern, with highest or equal prevalence in the T2D population from Geisinger in crude analysis but highest prevalence in the T1D population from Geisinger after adjusting for age, sex, and race (Fig. 4C and 4F). Prevalence of both low eGFR and albuminuria in the general US population were low and remained stable over time.

Distribution of Kidney Measures: 2018

The median eGFR in adults with T1D in Geisinger was similar to the median eGFR in the general US population (89.5 mL/min/1.73 m² vs 89.1 mL/min/1.73 m²), but the distribution of eGFR in adults with T1D was skewed toward lower values. The entire distribution of eGFR was lower (shifted downward) in adults with T2D than the nondiabetic general US population (Fig. 5A). The median ACR was higher in adults with T1D (median 10.5 mg/g) as well as those with T2D (median 13.0 mg/g) compared to the nondiabetic general US population (median 7.0 mg/g). The distributions of ACR were skewed toward higher values in adults with both types of diabetes (Fig. 5B). These distributions were not adjusted for age.

Figure 5.

Distributions of estimated glomerular filtration rate (A) and albumin-to-creatinine ratio (B) in adults with type 1 diabetes, type 2 diabetes, and a nondiabetic general population.

Screening Rates for Kidney Disease in Geisinger

The screening rates for kidney disease were moderate, but eGFR was checked much more frequently than ACR, particularly in people with T2D. In 2003-2004, only 60% of people with T1D and 40% of people with T2D had both eGFR and ACR measured. Screening rates improved slightly in people with T2D, to around 50% for 2006-2014, but dropped after that (Fig. 6A). In 2003-2004, 85% of people with T1D had at least 1 measurement of eGFR. The rate dropped to approximately 80% in 2009-2012 but rose almost 90% in 2017-2018. The screening rate with eGFR in people with T2D remained constant around 90% over the study period (Fig. 6B). Screening rates for albuminuria were under 75% for both T1D and T2D in all years, but rates were substantially higher in people with T1D than in people with T2D (Fig. 6C).

Figure 6.

Biennial kidney function screening rates for people with type 1 diabetes and type 2 diabetes—estimated glomerular filtration rate (eGFR) and albumin-to-creatinine ratio (ACR) (A), eGFR (B), and ACR (C).

Risk of Kidney Disease by Obesity Status

Obesity was associated with 50% higher odds of low eGFR [odds ratio (OR) 1.52, 95% CI 1.12, 2.08] in analyses adjusted for age, sex, and race in the population with T1D; the association was attenuated when we adjusted for hypertension (OR 1.34, 95% CI 0.98, 1.83). The odds of albuminuria were also higher with obesity but not statistically significant (OR 1.35, 95% CI 0.99, 1.81) in adults with T1D, and this was also attenuated when we adjusted for hypertension (OR 1.16, 95% CI 0.84, 1.61). In adults with T2D, obesity was associated with higher odds of both low eGFR (OR 1.29, 95% CI 1.23, 1.35) and albuminuria (OR 1.13, 95% CI 1.06, 1.21). These associations were slightly attenuated but remained statistically significant after adjusting for hypertension but were not appreciably affected by adjustment for the use of renoprotective antidiabetes medications. Obesity was also associated with higher odds of low eGFR in the general US population, but the magnitude of effect was smaller than that seen in adults with either T1D or T2D (Table 2).

Table 2.

Adjusted odds ratios for chronic kidney disease associated with obesity status (body mass index ≥ 30 kg/m ² )

	Geisinger Health System		NHANES
	Type 1 diabetes, 2018	Type 2 diabetes, 2018	Nondiabetic general population
eGFR < 60 mL/min/1.73m2	n = 1257	n = 51 468	n = 42 948
Model 1	1.52 (1.12, 2.08)	1.29 (1.23, 1.35)	1.22 (1.09, 1.36)
Model 2	1.34 (0.98, 1.83)	1.26 (1.21, 1.32)	1.14 (1.02, 1.12)
Model 3	—	1.27 (1.22, 1.33)	—
ACR > 30 mg/g	n = 805	n = 19 942	n = 44 824
Model 1	1.35 (0.99, 1.81)	1.13 (1.06, 1.21)	1.24 (1.14, 1.36)
Model 2	1.16 (0.84, 1.61)	1.09 (1.02, 1.16)	1.12 (1.02, 1.22)
Model 3	—	1.08 (1.03, 1.16)	—

Model 1 adjusts for age, sex, race. Model 2 adjusts for age, sex, race, and hypertension. Model 3 adjusts for age, sex, race, hypertension, and use of glucagon-like peptide-1 receptor agonist or sodium-glucose cotransporter 2 inhibitor medications.

Abbreviations: ACR, albumin-to-creatinine ratio; eGFR, estimated glomerular filtration rate; NHANES, National Health and Nutrition Examination Survey.

Conclusions

Epidemiological studies of adults with T1D are difficult. Even with rising incidence in recent years and increased survival, T1D is still a relatively uncommon disease, estimated to account for only about 5% of total diabetes cases (25). Therefore, traditional community-based epidemiologic cohorts have too few participants with T1D to make valid estimates. Most evidence on long-term health outcomes in T1D come from small, regionally based cohorts recruited in the 1980s (26,27) and from the DCCT, a clinical trial that recruited patients < 40 years old with no or only mild diabetes-related complications (28). The present study adds to the current literature by adding data from a contemporary T1D population and using EHRs to investigate real-world screening of kidney disease and long-terms trends in the prevalence of obesity and CKD, as well as its associations.

Historically, people with T1D have been thought to have lower rates of obesity than the general population. In this study, we showed that the prevalence of obesity in a population of adults with type 1 diabetes was similar to prevalence in the nondiabetic general US population; this mirrors trends seen in several international T1D cohorts (29,30). The trends for increasing rates of obesity may signal a setback in the progress of improving health in this vulnerable population.

CKD is thought to be higher in adults with T2D than adults with T1D (31,32), but most studies have not adjusted for age. Older age is a major risk factor for CKD (33,34), and the population of adults with T2D is much older than the population of adults with T1D (25,35). This substantial difference in population age distribution is important to consider when comparing these 2 groups. In the present study, we showed that in a well characterized population of adults with both types of diabetes that age-adjusted prevalence of low eGFR was much higher in adults with T1D than in adults with T2D. The distribution of eGFR in our population of adults with T1D was also skewed toward lower values, indicating relatively high prevalence of early kidney dysfunction, despite the young age of the population. Current clinical guidelines recommend screening for kidney disease annually in patients with T1D and T2D, but there is evidence that these guidelines are not followed consistently in the real world (8). In our study, fewer than half of people with T2D had both markers measured, even over a 2-year period. eGFR was checked more frequently overall, and although ACR was checked less frequently, it was more commonly tested in people with T1D than T2D. Overall, CKD awareness in the United States is low, even among those at high risk or advanced stages of CKD (6-8). We also showed that obesity is associated with CKD in both T1D and T2D, with higher magnitude of effect in T1D. This may be due to the fact that people with T2D have higher prevalence of other risk factors for CKD (eg, older age), so obesity may have relatively less effect on the risk for CKD. Obesity was also associated with CKD in the general US population without diagnosed diabetes, but the magnitude of effect was smaller than in persons with diabetes. The associations between obesity and CKD in T1D were significantly attenuated when we adjusted for the presence of hypertension, but the relationship between CKD and hypertension is complex (3), and hypertension may be a mediator in the relationship between obesity and CKD. Regardless, the increase in obesity among T1D is concerning for the future kidney health of these patients. Our results highlight the importance of routine screening for kidney disease, particularly in adults with T1D and perhaps especially in those with obesity.

The higher age-adjusted prevalence of reduced kidney function for T1D vs T2D has implications for long-term health. With increased survival in the population of patients with T1D, it is likely that many patients will progress to end-stage kidney disease and will need kidney replacement therapy during their lifetime. Several kidney-protective medications, including SGLT2 inhibitors, show efficacy for preventing or slowing progression of kidney disease in adults with and without T2D (36-38), but adults with T1D have been excluded from most of these trials. Some small studies of SGLT1 and SGLT2 inhibitors in adults with T1D have shown modest improvements in glucose control, but none of these trials have evaluated changes in kidney function (39,40). These trials did show a higher risk of complications, specifically diabetic ketoacidosis, but without longer and more comprehensive trials, the full range of risks and benefits are not known. Despite this lack of evidence on safety and effectiveness, we found that the use of GLP-1 receptor agonists and SGLT2 inhibitors in adults with T1D has increased in the past few years.

Our study had several limitations, primarily related to the natures of EHR and national survey data. For our analyses of people with T1D and T2D, we were limited to information on patients who engaged in care within the Geisinger Health System, and we only had data for measurements taken at the discretion of the healthcare provider for patient care. Nevertheless, this data source has been used extensively for research (14,17), and it provides rich clinical data from an integrated healthcare system with a stable population. People with T1D are generally highly engaged with the healthcare system, due to their absolute need for insulin. The population of central Pennsylvania is primarily of non-Hispanic white race, which may limit the generalizability of our study. However, this study represents one of the largest cohorts of adults with T1D in the United States, and we have detailed laboratory and pharmacy records and longitudinal follow-up that are not available in other T1D registry systems. Our method of identifying diabetes type primarily using billing records can be subject to misclassification, but the algorithm we used has been previously validated in many healthcare systems, including Geisinger. We were also able to compare characteristics of populations of people with T1D and T2D to people without diabetes in nationally representative data that is highly standardized and comprehensive.

Aging adults with T1D are a growing population in the United States, and they suffer from a high chronic disease burden. Obesity, once thought to be rare in people with T1D, has reached general population prevalence. Screening for kidney disease is not universal, yet the age-adjusted prevalence of CKD is higher in T1D than even T2D. Strong evidence on the risks and benefits of treatments aimed at reducing obesity and protecting kidneys in this population are needed.

Funding

A.S.W. was supported by National Institutes of Health (NIH)/National Heart, Lung, and Blood Institute (NHLBI) grant T32 HL007024. J-I.S. was supported by NIH/National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) grant K01 DK121825. E.S. was supported by NIH/NIDDK grant K24 DK106414. M.E.G. was supported by NIH/NIDDK grant R01 DK115534.

Author Contributions

A.S.W., J.-I. S., M.E.G., and E.S. designed the study. A.S.W. conducted the statistical analysis and drafted the manuscript. J.-I.S., M.E.G., and E.S. guided the statistical analysis. All authors provided critical revisions to the manuscript and approved the final manuscript. A.S.W. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Disclosures

The authors have no conflicts of interest relevant to this article to disclose.

Prior Publications

An early draft of this work was presented in abstract form at the 80th Scientific Sessions of the American Diabetes Association.

Data Availability

Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.