Advertisement

Glycemic Outcomes of Second-Line Diabetes Drug Choice in a Real-World Population

  • Amisha Wallia
    Correspondence
    Correspondence: Address to Amisha Wallia, MD, MS, Division of Endocrinology Metabolism and Molecular Medicine, Northwestern University Feinberg School of Medicine, 645 N Michigan Ave, Ste 530, Chicago, IL 60611.
    Affiliations
    Division of Endocrinology, Metabolism and Molecular Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL

    Institute for Public Health and Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL
    Search for articles by this author
  • Matthew J. O’Brien
    Affiliations
    Division of General Internal Medicine and Geriatrics, Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL

    Institute for Public Health and Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL
    Search for articles by this author
  • David T. Liss
    Affiliations
    Division of General Internal Medicine and Geriatrics, Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL

    Institute for Public Health and Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL
    Search for articles by this author
  • Raymond H. Kang
    Affiliations
    Institute for Public Health and Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL
    Search for articles by this author
  • Andrew J. Cooper
    Affiliations
    Division of General Internal Medicine and Geriatrics, Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL
    Search for articles by this author
  • Amy Gilmer
    Affiliations
    UnitedHealth Group, Minnetonka, MN
    Search for articles by this author
  • Ronald T. Ackermann
    Affiliations
    Division of Endocrinology, Metabolism and Molecular Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL

    Division of General Internal Medicine and Geriatrics, Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL

    Institute for Public Health and Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL
    Search for articles by this author
Open AccessPublished:May 26, 2021DOI:https://doi.org/10.1016/j.mayocpiqo.2021.02.008

      Abstract

      Hypoglycemia and acute metabolic complications (AMCs; ketoacidosis, hyperosmolarity, and coma) are glycemic outcomes that have high cost and high morbidity; these outcomes must be taken into consideration when choosing initial second-line therapy after metformin. We conducted a retrospective cohort study analyzing national administrative data from adults with type 2 diabetes mellitus who started a second-line diabetes medication (sulfonylureas [SFUs], thiazolidinediones [TZDs], glucagon-like peptide 1 [GLP-1] agonists, dipeptidyl peptidase 4 [DPP-4] inhibitors, basal insulin, or sodium-glucose contransporter 2 [SGLT-2] inhibitors) between April 1, 2011 and September 30, 2015 (N=43,288) and compared rates of hypoglycemia and AMCs. Most patients (24,506 [56.6%]) were prescribed sulfonylurea as second-line treatment, followed by DPP-4 inhibitors (7953 [18.4%]), GLP-1 agonists (3854 [8.9%]), basal insulin (2542 [5.9%]), SGLT-2 inhibitors (2537 [5.9%), and TZDs (1896 [4.4%]). Baseline rates of hypoglycemia varied more than 5-fold across initial second-line antidiabetic medication classes, and rates of AMCs varied 7-fold. Compared with patients taking an SFU, lower adjusted rates of hypoglycemia were associated with taking a DPP-4 inhibitor (63% lower rate; incidence rate ratio [IRR], 0.37; 95% CI, 0.25 to 0.57), SGLT-2 inhibitor (54% lower; IRR, 0.46; 95% CI, 0.22 to 0.94), or TZD (79% lower; IRR, 0.21; 95% CI, 0.08 to 0.56) but not a glucagon-like peptide 1 agonist or basal insulin. For AMCs, only initiation of a DPP-4 inhibitor (43% lower rate; IRR, 0.57; 95% CI, 0.41 to 0.81) was associated with a lower adjusted rate compared with SFU. Use of SGLT-2 inhibitors was not associated with a substantially increased rate of acute metabolic complications compared with SFU. Special attention still needs to be paid to glycemic outcomes when choosing a second-line diabetes therapy following metformin.

      Abbreviations and Acronyms:

      AMC (acute metabolic complication), DM (diabetes mellitus), DPP-4 (dipeptidyl peptidase 4), GLP-1 (glucagon-like peptide 1), ICD-9 (International Classification of Diseases, Ninth Revision), IRR (incidence rate ratio), SFU (sulfonylurea), SGLT-2 (sodium-glucose contransporter 2), TZD (thiazolidinedione)
      Most patients with type 2 diabetes mellitus (DM) eventually require additional medication after initial treatment with metformin, and several of the available choices have the potential to cause adverse glycemic events.
      • Budnitz D.S.
      • Lovegrove M.C.
      • Shehab N.
      • Richards C.L.
      Emergency hospitalizations for adverse drug events in older Americans.
      With at least 7 classes of medications available as initial second-line therapy, clinicians must balance the effectiveness, cost, and availability of these medication choices with the risk for harms. Clinicians must weigh hypoglycemia risk with the risk of acute metabolic complications (AMCs), which include ketoacidosis, hyperosmolarity, and coma. Rates of hypoglycemia have increased and are of particular concern among older adults and those with multiple comorbidities.
      • McCoy R.G.
      • Lipska K.J.
      • Van Houten H.K.
      • Shah N.D.
      Association of cumulative multimorbidity, glycemic control, and medication use with hypoglycemia-related emergency department visits and hospitalizations among adults with diabetes.
      In addition, sodium-glucose cotransporter 2 (SGLT-2) inhibitors specifically have been associated with ketoacidosis.
      • Douros A.
      • Lix L.M.
      • Fralick M.
      • et al.
      Sodium-glucose cotransporter-2 inhibitors and the risk for diabetic ketoacidosis: a multicenter cohort study.
      Because these complications confer high short-term mortality,
      • Mays J.A.
      • Jackson K.L.
      • Derby T.A.
      • et al.
      An evaluation of recurrent diabetic ketoacidosis, fragmentation of care, and mortality across Chicago, Illinois.
      ,
      • Bonds D.E.
      • Miller M.E.
      • Bergenstal R.M.
      • et al.
      The association between symptomatic, severe hypoglycaemia and mortality in type 2 diabetes: retrospective epidemiological analysis of the ACCORD study.
      even modest differences in risk can have important implications for appropriate medication selection in clinical practice. Not surprisingly, there is interest from national stakeholders to develop quality metrics for monitoring hypoglycemia and AMCs as a means to improve patient safety.
      Although clinical trials provide data on hypoglycemia and AMCs under controlled conditions, there is limited information about the magnitude of these risks under real-world conditions and on a population scale. The objective of this study was to compare rates of hypoglycemia and AMCs among adult patients taking metformin who subsequently initiate a second antidiabetic medication (sulfonylureas [SFUs] or meglitinides, thiazolidinediones [TZDs], basal insulin, glucagon-like peptide 1 [GLP-1] agonists, dipeptidyl peptidase 4 (DPP-4) inhibitors, and SGLT-2 inhibitors).

      Patients and Methods

      We conducted a retrospective cohort study analyzing national administrative data, including health plan enrollment files, pharmacy claims, medical claims, and laboratory claims from a large commercial health insurer. We included adults with type 2 DM who had (1) at least 6 months of continuous enrollment in a Medicare Advantage or commercial health plan, (2) evidence of only metformin pharmacy claims during their insurance enrollment, and (3) evidence of a new start of a second-line DM medication (SFU, TZD, GLP-1 agonist, DPP-4 inhibitor, basal insulin, or SGLT-2 inhibitor), with an index date between April 1, 2011 and September 30, 2015. Patients were considered to have DM on the basis of their prescription claims for both metformin and a new second-line DM medication (and no additional third agent for 6 months), in addition to one or more medical encounters with a DM-related International Classification of Diseases, Ninth Revision (ICD-9) code occurring on or before the index date for the new medication drug start, similar to definitions we have utilized previously.
      • Ackermann R.T.
      • Wallia A.
      • O'Brien M.J.
      • et al.
      Correlates of second-line type 2 diabetes medication selection in the USA.
      ,
      • O'Brien M.J.
      • Karam S.L.
      • Wallia A.
      • et al.
      Association of second-line antidiabetic medications with cardiovascular events among insured adults with type 2 diabetes.
      As done previously, we also excluded patients with ICD-9 codes for type 1 diabetes, pregnancy, or secondary diabetes. We conducted 2 separate analyses of the outcomes of interest (ie, hypoglycemia and AMCs) for the year after starting the index drug. Hypoglycemic events were defined using the following ICD-9 diagnosis codes: 251.0, 251.1, 251.2, and 962.3, adapted from Ginde et al.
      • Ginde A.A.
      • Blanc P.G.
      • Lieberman R.M.
      • Camargo Jr., C.A.
      Validation of ICD-9-CM coding algorithm for improved identification of hypoglycemia visits.
      Acute metabolic complications were defined using ICD-9 codes 250.2X, 250.1X, and 250.3X (250.XX = DM), adapted from our previous work.
      • Mays J.A.
      • Jackson K.L.
      • Derby T.A.
      • et al.
      An evaluation of recurrent diabetic ketoacidosis, fragmentation of care, and mortality across Chicago, Illinois.
      Covariates included age, race, year of drug initiation, hemoglobin A1c levels, geographic region, health care professional type, receipt of DM education, hospitalization in the year prior to the new drug, insurance type, previous occurrence of the outcome of interest, and a modified diabetes complications severity index score,
      • Chang H.-Y.
      • Weiner J.P.
      • Richards T.M.
      • Bleich S.N.
      • Segal J.B.
      Validating the adapted Diabetes Complications Severity Index in claims data.
      which we adapted slightly to remove outcomes of interest to avoid overadjustment (for a full list of codes, see the Supplemental Appendix [available online at http://mcpiqojournal.org]).
      We used χ2 tests to examine bivariate associations between baseline patient characteristics and index medication class. Because of low event rates, multivariable, zero-inflated Poisson regression models were used to assess the association between index medication class and each of the 2 outcomes while adjusting for all covariates listed previously. Sulfonylureas served as the reference group because they are the most commonly prescribed second-line antidiabetic medications.
      • Ackermann R.T.
      • Wallia A.
      • O'Brien M.J.
      • et al.
      Correlates of second-line type 2 diabetes medication selection in the USA.
      Statistical analyses were conducted using SAS statistical software, version 9.4 (SAS Institute). Because the data were nonidentifiable, the Northwestern University Institutional Review Board judged this study to not be human subjects research.

      Results

      We included a total of 43,288 patients in this study. Table 1 summarizes patient characteristics stratified by second-line medication class. Statistically significant differences between groups were noted in every category (P<.0001). See Supplemental Table 1 for adjusted event rates (available online at http://mcpiqojournal.org). Most patients (24,506 [56.6%]) were prescribed SFU as their second-line agent, followed by DPP-4 inhibitors (7953 [18.4%]), basal insulin (2542 [5.9%]), SGLT-2 inhibitors (2537 [5.9%]), and TZDs (1896 [4.4%]). Baseline rates of hypoglycemia varied more than 5-fold across initial second-line antidiabetic medication classes, and rates of AMCs varied 7-fold. Second-line DM drug choice differed by prescriber type, although the most common prescriber specialty for all drug classes was family practice (34.6% [1333 of 3854] for GLP-1 to 53.5% [1015 of 1896] for TZD). Sulfonylureas were the most commonly selected drug class among all prescriber types except endocrinologists, who most often prescribed GLP-1 agonists. Interestingly, encounters for diabetes education were infrequent (2.4% [46 of 1896] for TZD to 6.9% [267 of 3854] for GLP-1 across groups), similar to previous reports.
      • Li R.
      • Shrestha S.
      • Lipman R.
      • Burrows N.
      • Kolb L.
      • Rutledge S.
      Diabetes self-management education and training among privately insured persons with newly diagnosed diabetes— United States, 2011-2012.
      Table 1Preexposure Patient, Prescriber, and Health Plan Characteristics Among the 43,288 Study Patients
      DCSI, Diabetes Complications Severity Index; DPP-4, dipeptidyl peptidase 4 inhibitors; EPO, exclusive provider organization; GLP-1, glucagon-like peptide 1 agonists; HbA1c, hemoglobin A1c; HMO, health maintenance organization; PA, physician assistant; POS, point of service; PPO, preferred provider organization; SFU, sulfonylureas or meglitinides; SGLT-2, sodium-glucose cotransporter 2 inhibitors; TZD, thiazolidinediones.
      ,
      Data are presented as No. (percentage) of patients unless indicated otherwise.
      VariableDPP-4 (n=7953)GLP-1 (n=3854)Basal insulin (n=2542)SGLT-2 (n=2537)SFU (n=24,506)TZD (n=1896)
      Hypoglycemia rate per 1000 person-years
      Event rates observed prior to index date (date of first pharmacy fill of second-line medication).
      6.635.210.811.37.07.8
      Metabolic complication rate per 1000 person-years
      Event rates observed prior to index date (date of first pharmacy fill of second-line medication).
      15.68.444.76.411.414.1
      Sex (%)
      P<.0001.
       Female3296 (41.4)2330 (60.5)1144 (45)1052 (41.5)9626 (39.3)679 (35.8)
       Male4657 (58.6)1524 (39.5)1398 (55)1485 (58.5)14880 (60.7)1217 (64.2)
      Age (y)
      P<.0001.
       18-34130 (1.6)191 (5.0)76 (3.0)81 (3.2)466 (1.9)23 (1.2)
       35-44726 (9.1)625 (16.2)282 (11.0)367 (14.5)2235 (9.1)155 (8.2)
       45-541990 (25.0)1243 (32.2)680 (26.8)867 (34.2)6087 (24.8)445 (23.5)
       55-643053 (38.4)1341 (34.8)899 (35.4)1008 (39.8)8403 (34.3)623 (32.9)
       65-741520 (19.1)394 (10.2)415 (16.3)198 (7.8)4877 (19.9)424 (22.4)
       ≥75534 (6.7)60 (1.6)190 (7.5)16 (0.6)2438 (10.0)226 (11.9)
      Race/ethnicity
      P<.0001.
       Black769 (9.7)369 (9.6)326 (12.8)257 (10.1)2519 (10.3)120 (6.3)
       Hispanic1152 (14.5)429 (11.1)399 (15.7)329 (13.0)4210 (17.2)389 (20.5)
       Unknown968 (12.1)260 (6.8)214 (8.4)189 (7.5)2593 (10.5)237 (12.5)
       White5064 (63.7)2796 (72.6)1603 (63.1)1762 (69.5)15,186 (62.0)1150 (60.7)
      HbA1c (%)
      P<.0001.
       Not available
      Laboratory values are not routinely available in health plan administrative data sources unless submitted by the laboratory vendor as part of their contract with the health payer; because of the potential for a relationship between baseline HbA1c and both hypoglycemia and AMCs, we included an HbA1c result if available, in addition to the DCSI and other surrogate markers for baseline diabetes severity for all patients.
      4784 (60.2)2428 (63)1850 (72.8)1270 (50.1)16,814 (68.6)1265 (66.7)
       <81240 (15.6)843 (21.9)139 (5.5)552 (21.7)2402 (9.8)290 (15.3)
       8-101325 (16.6)382 (9.9)181 (7.1)451 (17.8)3193 (13.0)224 (11.8)
       ≥10604 (7.6)201 (5.2)372 (14.6)264 (10.4)2098 (8.6)117 (6.2)
      DCSI score
      P<.0001.
      ,
      Adapted composite score (index severity score 0-3) of 6 complications: retinopathy, nephropathy, neuropathy, cerebrovascular, cardiovascular, peripheral vascular disease.
       04855 (61.1)2555 (66.3)1538 (60.5)1615 (63.6)15,495 (63.2)1212 (64.0)
       11379 (17.3)706 (18.3)418 (16.5)462 (18.2)3997 (16.3)320 (16.8)
       2-31287 (16.2)496 (12.9)435 (17.1)369 (14.5)3789 (15.5)280 (14.8)
       ≥4432 (5.4)98 (2.5)150 (5.9)92 (3.6)1225 (5)84 (4.4)
      Diabetes education
      P<.0001.
      236 (3.0)267 (6.9)104 (4.1)77 (3.0)605 (2.5)46 (2.4)
      Insurance
      P<.0001.
       Commercial7020 (88.3)3700 (96)2151 (84.6)2509 (98.9)19,980 (81.5)1503 (79.3)
       Medicare933 (11.7)154 (4)391 (15.4)28 (1.1)4526 (18.5)393 (20.7)
      Plan type
      P<.0001.
       EPO641 (8.1)359 (9.3)242 (9.5)258 (10.2)1919 (7.83)143 (7.54)
       HMO2161 (27.2)638 (17.0)733 (28.9)329 (13.0)7624 (31.1)643 (33.9)
       Indemnity84 (1.1)15 (0.4)32 (1.3)16 (0.6)297 (1.2)21 (1.1)
       Other359 (4.5)179 (4.6)72 (2.8)78 (3.1)723 (3.0)50 (2.6)
       POS4542 (57.1)2574 (66.8)1409 (55.4)1812 (71.4)13,412 (54.7)987 (52.1)
       PPO166 (2.1)89 (2.3)54 (2.1)44 (1.7)532 (2.2)52 (2.7)
      Prescriber type
      P<.0001.
       Endocrinologist543 (6.8)797 (20.7)152 (6.0)226 (8.9)765 (3.1)53 (2.8)
       Family practice3353 (42.2)1333 (34.6)1122 (44.1)1135 (44.7)11,449 (46.7)1015 (53.5)
       General/internal medicine2761 (34.7)980 (25.4)752 (29.6)680 (26.8)8097 (33.0)527 (27.8)
       Nurse/PA574 (7.2)401 (10.4)195 (7.7)274 (10.8)1632 (6.7)117 (6.2)
       Other/missing722 (9.1)343 (9.0)321 (12.6)222 (8.8)2566 (10.5)184 (9.7)
      a DCSI, Diabetes Complications Severity Index; DPP-4, dipeptidyl peptidase 4 inhibitors; EPO, exclusive provider organization; GLP-1, glucagon-like peptide 1 agonists; HbA1c, hemoglobin A1c; HMO, health maintenance organization; PA, physician assistant; POS, point of service; PPO, preferred provider organization; SFU, sulfonylureas or meglitinides; SGLT-2, sodium-glucose cotransporter 2 inhibitors; TZD, thiazolidinediones.
      b Data are presented as No. (percentage) of patients unless indicated otherwise.
      c Event rates observed prior to index date (date of first pharmacy fill of second-line medication).
      d P<.0001.
      e Laboratory values are not routinely available in health plan administrative data sources unless submitted by the laboratory vendor as part of their contract with the health payer; because of the potential for a relationship between baseline HbA1c and both hypoglycemia and AMCs, we included an HbA1c result if available, in addition to the DCSI and other surrogate markers for baseline diabetes severity for all patients.
      f Adapted composite score (index severity score 0-3) of 6 complications: retinopathy, nephropathy, neuropathy, cerebrovascular, cardiovascular, peripheral vascular disease.
      Table 2 summarizes adjusted incidence rate ratios (IRRs) for both outcomes, hypoglycemia and AMCs. Compared with patients taking an SFU, lower adjusted rates of hypoglycemia were associated with taking a DPP-4 inhibitor (63% lower rate; IRR, 0.37 [95% CI, 0.25 to 0.57]), SGLT-2 inhibitor (54% lower; IRR, 0.46 [95% CI, 0.22 to 0.94]), or TZD (79% lower; IRR, 0.21 [95% CI, 0.08 to 0.56]) but not a GLP-1 agonist or basal insulin. For AMCs, only initiation of a DPP-4 inhibitor (43% lower rate; IRR, 0.57 [95% CI, 0.41 to 0.81]) was associated with a lower adjusted rate compared with SFU. Taking an SGLT-2 inhibitor was not associated with a significantly increased rate of AMCs compared with SFU (P=.35).
      Table 2Adjusted Incidence Rate Ratios for Hypoglycemia and Acute Metabolic Complications
      DPP-4, dipeptidyl peptidase 4 inhibitors; EPO, exclusive provider organization; GLP-1, glucagon-like peptide 1 agonists; HMO, health maintenance organization; IRR, incident rate ratio; PA, physician assistant; POS, point of service; PPO, preferred provider organization; SGLT-2, sodium-glucose cotransporter 2 inhibitors; TZD, thiazolidinediones.
      VariableHypoglycemia, IRR (95% CI)Acute metabolic complications, IRR (95% CI)
      Index second-line medication class (referent group, SFU)
       DPP-40.37 (0.25-0.57)
      P≤.0001.
      0.57 (0.41-0.81)
      P≤.01.
       GLP-10.99 (0.70-1.43)0.75 (0.42-1.34)
       Basal insulin1.36 (0.87-2.11)1.23 (0.85-1.77)
       SGLT-20.46 (0.22-0.94)
      P<.05.
      1.36 (0.71-2.60)
       TZD0.21 (0.08-0.56)
      P≤.01.
      0.83 (0.48-1.43)
      Covariates
       History of outcome event1.50 (1.36-1.67)
      P≤.0001.
      1.6 (1.47-1.74)
      P≤.0001.
      Race/ethnicity (referent group, White)
       Black0.87 (0.59-1.29)0.99 (0.69-1.42)
       Hispanic1.16 (0.83-1.62)1.59 (1.36-2.24)
      P≤.01.
       Unknown1.03 (0.70-1.51)0.93 (0.60-1.43)
      DCSI score (referent group, 0)
      Adapted composite score (index severity score 0-3) of 6 complications: retinopathy, nephropathy, neuropathy, cerebrovascular, cardiovascular, peripheral vascular disease.9
       10.83 (0.56-1.22)1.40 (1.00-1.95)
       2-31.61 (1.17-2.21)
      P≤.01.
      2.13 (1.55-2.93)
      P≤.0001.
       ≥41.89 (1.26-2.84)
      P≤.01.
      1.86 (1.26-2.73)
      P≤.01.
       Diabetes education0.68 (0.87-15.11)2.10 (1.11-3.97)
      P<.05.
       Prior hospital admission2.09 (1.53-2.86)
      P≤.0001.
      2.65 (2.00-3.49)
      P≤.0001.
      Insurance plan type (referent group, PPO)
       EPO1.49 (0.66-3.36)1.76 (0.60-5.19)
       HMO1.44 (0.67-3.11)1.83 (0.64-5.21)
       Indemnity1.85 (0.61-5.64)4.49 (1.42-14.18)
      P≤.01.
       Other3.50 (1.56-7.83)
      P≤.01.
      1.28 (0.35-4.61)
       POS2.23 (1.1-4.51)
      P<.05.
      2.27 (0.83-6.22)
      Prescriber type (referent group, Family Practice)
       Endocrinologist1.85 (1.22-2.81)
      P≤.01.
      0.62 (0.32-1.21)
      Internal medicine0.92 (0.70-1.21)1.02 (0.78-1.33)
       Nurse/PA1.40 (0.88-2.21)0.98 (0.55-1.74)
       Other/missing1.69 (1.23-2.32)
      P≤.0001.
      0.74 (0.52-1.07)
      a DPP-4, dipeptidyl peptidase 4 inhibitors; EPO, exclusive provider organization; GLP-1, glucagon-like peptide 1 agonists; HMO, health maintenance organization; IRR, incident rate ratio; PA, physician assistant; POS, point of service; PPO, preferred provider organization; SGLT-2, sodium-glucose cotransporter 2 inhibitors; TZD, thiazolidinediones.
      b P≤.0001.
      c P≤.01.
      d P<.05.
      e Adapted composite score (index severity score 0-3) of 6 complications: retinopathy, nephropathy, neuropathy, cerebrovascular, cardiovascular, peripheral vascular disease.
      • Chang H.-Y.
      • Weiner J.P.
      • Richards T.M.
      • Bleich S.N.
      • Segal J.B.
      Validating the adapted Diabetes Complications Severity Index in claims data.
      History of previous hypoglycemia or AMC, increasing diabetes complications severity index score, or any hospitalization in the previous year were associated with increased rates of both hypoglycemia and AMC events, independent of medication class. Higher rates of hypoglycemia were associated with an endocrinologist or “other/missing” health care professional prescribing the index drug. Higher AMC rates were associated with a history of diabetes education and enrollment in a non–managed care, fee-for-service (indemnity) health plan. Age, prescription fill year, and baseline hemoglobin A1c level were nonsignificant in either model.

      Discussion

      Analyzing data from over 43,000 adults with type 2 diabetes who initiated a second-line DM medication following metformin monotherapy, we found low event rates for adverse glycemic outcomes resulting in health care visits. Substantially different rates of hypoglycemia and/or AMCs were observed when patients initiated second-line therapy with alternative DM medication classes, relative to SFUs, which currently remain the most common initial second-line choice after metformin. Compared with SFUs, the initiation of SGLT-2 inhibitors, DPP-4 inhibitors, or TZDs was associated with lower rates of hypoglycemia, whereas initiation of a GLP-1 agonist or basal insulin resulted in comparable rates of hypoglycemia. Because GLP-1 agonists are generally considered to pose a lower risk for hypoglycemia, it was surprising that event rates in real-world practice were not substantially lower than those for SFUs. Prior studies have found that additional hypoglycemia risk is noted when initiating dual therapy with combined new DM agents (ie, metformin plus a new additional agent), which are not thought to present added risk when prescribed separately from metformin
      • Kamalinia S.
      • Josse R.G.
      • Donio P.J.
      • Leduc L.
      • Shah B.R.
      • Tobe S.W.
      Risk of any hypoglycaemia with newer antihyperglycaemic agents in patients with type 2 diabetes: a systematic review and meta-analysis.
      ; however, our data show clear differences among DM drug classes initiated after metformin.
      Dipeptidyl peptidase 4 inhibitors were associated with fewer AMCs than SFU, whereas prior hospital admission, increased diabetes-related complications, and previous diabetes education were all substantially associated with severe AMCs (ketoacidosis, hyperosmolarity, coma). Interestingly, although the study population comprised patients starting a second-line DM agent, some had underlying comorbidities, including diabetes-related complications, which may have contributed to these poor DM-related outcomes. Sodium-glucose contransporter 2 inhibitors, which in some previous studies have been linked to an increased risk for euglycemic and hyperglycemic diabetic ketoacidosis,
      • Jeon J.Y.
      • Kim S.-K.
      • Kim K.-S.
      • et al.
      Clinical characteristics of diabetic ketoacidosis in users and non-users of SGLT2 inhibitors.
      ,
      • Sharma P.V.
      • Jobanputra Y.B.
      • Lewin K.
      • Bagatell S.
      • Lichtstein D.M.
      Diabetic ketoacidosis in patients with type 2 diabetes on sodium-glucose cotransporter-2 inhibitors - a case series.
      had no difference in overall metabolic complications in our study, which included episodes of diabetic ketoacidosis in the outcome definition. However, because event rates were generally low and the confidence interval for the IRR comparing SGLT-2 inhibitors to SFUs was wide, this finding requires confirmation.
      This was an observational study and has some notable limitations. Even after adjustment for differences in demographic and clinical covariates, it is possible that patients initiating insulin or SFUs may still be at higher risk for AMC events, unrelated to the medication, resulting in residual confounding. For this reason, confirmatory studies are needed for some unanticipated associations, such as the lower rate of AMCs among patients initiating DPP-4 inhibitors. Diabetes education, meant to improve self-efficacy, is designed to minimize AMCs or severe hypoglycemic events, but our analysis revealed low rates of diabetes education and an association with higher rates of adverse glycemic outcomes, suggesting that patients who receive these services are at a higher risk for AMCs, regardless of the second-line medication choice. Diabetes education has been previously reported to be similarly low in other studies.
      • Li R.
      • Shrestha S.
      • Lipman R.
      • Burrows N.
      • Kolb L.
      • Rutledge S.
      Diabetes self-management education and training among privately insured persons with newly diagnosed diabetes— United States, 2011-2012.
      ,
      • Strawbridge L.M.
      • Lloyd J.T.
      • Meadow A.
      • Riley G.F.
      • Howell B.L.
      Use of Medicare's diabetes self-management training benefit.
      However, variable reimbursement for these services and inconsistent use of billing codes (see Supplemental Appendix) may result in estimates of diabetes education services derived from claims data that are lower than actual practice. In addition, there could be residual confounding, specifically confounding by indication, present in our analysis despite adjusting for confounders. This issue is a known limitation of the observational study design, and we have included covariates available in claims data that can account for glycemic control, health care provider, and insurance plan type.

      Conclusion

      Results from this study are important given a paucity of prior real-world evidence on the association of second-line medication choice with subsequent hypoglycemia and AMC events among patients with type 2 DM. Despite recent expansion in the numbers of second-line DM medication alternatives, hyperglycemia and hypoglycemia remain critical issues in DM management. Although infrequent, these episodes are extremely costly and could be deemed partially iatrogenic given that risks for these outcomes are driven partly by choices in medication prescribing. Moreover, because the rates of glycemic complications associated with different second-line agents varied with patient comorbidities in our study, our analysis also underscores the importance of clinical guidelines and quality metrics that allow health care professionals and patients autonomy in selecting the most appropriate medication option based on circumstances unique to each individual.

      Acknowledgments

      We thank Sanya Pasricha, Susan Karam, MD, Cassandra Aikman, MPH, Emily Parker, MPH, PhD and Nicola Lancki, MPH for their thoughtful input on the submitted manuscript.
      Contributions, Data Access, Responsibility, and Analysis: Dr Wallia contributed to the study design, interpreted analytic results, drafted and edited/revised the submitted manuscript, and serves as guarantor; Mr Kang and Mr Cooper contributed to the study design, managed and cleaned raw data received from the sponsor, conducted the analyses, interpreted the findings, and edited/revised the submitted manuscript; Drs O’Brien, Liss, and Ackermann contributed to the study design, interpreted the results, and edited/revised the submitted manuscript; Ms Gilmer contributed to interpretation of results and edited/revised the submitted manuscript; Dr Ackermann 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.
      The funding organization collected and ensured access to the data and supported the evaluation team through a grant to Northwestern University. The funding organization had no role in the design and conduct of the study; management, analysis, and interpretation of the data; preparation of the submitted manuscript; or decision to submit the manuscript for publication.

      Supplemental Online Material

      References

        • Budnitz D.S.
        • Lovegrove M.C.
        • Shehab N.
        • Richards C.L.
        Emergency hospitalizations for adverse drug events in older Americans.
        N Engl J Med. 2011; 365: 2002-2012
        • McCoy R.G.
        • Lipska K.J.
        • Van Houten H.K.
        • Shah N.D.
        Association of cumulative multimorbidity, glycemic control, and medication use with hypoglycemia-related emergency department visits and hospitalizations among adults with diabetes.
        JAMA Netw Open. 2020; 3: e1919099
        • Douros A.
        • Lix L.M.
        • Fralick M.
        • et al.
        Sodium-glucose cotransporter-2 inhibitors and the risk for diabetic ketoacidosis: a multicenter cohort study.
        Ann Intern Med. 2020; 173: 417-425
        • Mays J.A.
        • Jackson K.L.
        • Derby T.A.
        • et al.
        An evaluation of recurrent diabetic ketoacidosis, fragmentation of care, and mortality across Chicago, Illinois.
        Diabetes Care. 2016; 39: 1671-1676
        • Bonds D.E.
        • Miller M.E.
        • Bergenstal R.M.
        • et al.
        The association between symptomatic, severe hypoglycaemia and mortality in type 2 diabetes: retrospective epidemiological analysis of the ACCORD study.
        BMJ. 2010; 340: b4909
        • Ackermann R.T.
        • Wallia A.
        • O'Brien M.J.
        • et al.
        Correlates of second-line type 2 diabetes medication selection in the USA.
        BMJ Open Diabetes Res Care. 2017; 5: e000421
        • O'Brien M.J.
        • Karam S.L.
        • Wallia A.
        • et al.
        Association of second-line antidiabetic medications with cardiovascular events among insured adults with type 2 diabetes.
        JAMA Netw Open. 2018; 1: e186125
        • Ginde A.A.
        • Blanc P.G.
        • Lieberman R.M.
        • Camargo Jr., C.A.
        Validation of ICD-9-CM coding algorithm for improved identification of hypoglycemia visits.
        BMC Endocr Disord. 2008; 8: 4
        • Chang H.-Y.
        • Weiner J.P.
        • Richards T.M.
        • Bleich S.N.
        • Segal J.B.
        Validating the adapted Diabetes Complications Severity Index in claims data.
        Am J Manag Care. 2012; 18: 721-726
        • Li R.
        • Shrestha S.
        • Lipman R.
        • Burrows N.
        • Kolb L.
        • Rutledge S.
        Diabetes self-management education and training among privately insured persons with newly diagnosed diabetes— United States, 2011-2012.
        MMWR Morb Mortal Wkly Rep. 2014; 63: 1045-1049
        • Kamalinia S.
        • Josse R.G.
        • Donio P.J.
        • Leduc L.
        • Shah B.R.
        • Tobe S.W.
        Risk of any hypoglycaemia with newer antihyperglycaemic agents in patients with type 2 diabetes: a systematic review and meta-analysis.
        Endocrinol Diabetes Metab. 2019; 3: e00100
        • Jeon J.Y.
        • Kim S.-K.
        • Kim K.-S.
        • et al.
        Clinical characteristics of diabetic ketoacidosis in users and non-users of SGLT2 inhibitors.
        Diabetes Metab. 2019; 45: 453-457
        • Sharma P.V.
        • Jobanputra Y.B.
        • Lewin K.
        • Bagatell S.
        • Lichtstein D.M.
        Diabetic ketoacidosis in patients with type 2 diabetes on sodium-glucose cotransporter-2 inhibitors - a case series.
        Rev Recent Clin Trials. 2018; 13: 156-160
        • Strawbridge L.M.
        • Lloyd J.T.
        • Meadow A.
        • Riley G.F.
        • Howell B.L.
        Use of Medicare's diabetes self-management training benefit.
        Health Educ Behav. 2015; 42: 530-538