Current Medical Research and Opinion
ISSN: 0300-7995 (Print) 1473-4877 (Online) Journal homepage: http://www.tandfonline.com/loi/icmo20
The burden of severe hypoglycemia in type 1 diabetes Jieruo Liu, Rosa Wang, Michael L. Ganz, Yurek Paprocki, Doron Schneider & James Weatherall To cite this article: Jieruo Liu, Rosa Wang, Michael L. Ganz, Yurek Paprocki, Doron Schneider & James Weatherall (2017): The burden of severe hypoglycemia in type 1 diabetes, Current Medical Research and Opinion, DOI: 10.1080/03007995.2017.1391079 To link to this article: http://dx.doi.org/10.1080/03007995.2017.1391079
View supplementary material
Accepted author version posted online: 11 Oct 2017.
Submit your article to this journal
Article views: 1
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=icmo20 Download by: [UNIVERSITY OF ADELAIDE LIBRARIES]
Date: 13 October 2017, At: 07:03
The burden of severe hypoglycemia in type 1 diabetes
Jieruo Liu [1], Rosa Wang [1], Michael L. Ganz [1], Yurek Paprocki [2], Doron Schneider [3], James Weatherall [2] 1
Evidera, 500 Totten Pond Road, Fifth Floor, Waltham, MA, 02451, USA Novo Nordisk Inc., 800 Scudders Mill Rd, Plainsboro, NJ, 08536, USA 3 Abington Adult Medical Associates, 1235 Old York Rd, Ste. 214, Abington, PA, 19001, USA
t
ip
cr
CORRESPONDING AUTHOR: Michael L. Ganz, 500 Totten Pond Road, 5th Floor, Waltham, MA, 02451, USA. Email: [email protected]
us
TRANSPARENCY
an
Declaration of Funding: Novo Nordisk Inc., a pharmaceutical company that manufactures, markets and/or distributes more than a dozen drugs for diabetes in the United States, provided the funding for this study and article.
ce
pt e
d
M
Declaration of Financial/Other Relationships: JL and MLG are employed by Evidera Inc., an independent research company that provides consulting and other research services to the biopharmaceutical industry. In their salaried positions, they work with a variety of client companies and are precluded from receiving payment or honoraria directly from these organizations for services rendered. Evidera received payment from Novo Nordisk Inc. for the conduct of this study and the development of this manuscript. RW was an employee of Evidera Inc. when this manuscript was written. YP is an employee of Novo Nordisk Inc. JW was an employee of Novo Nordisk Inc when this manuscript was written; he is a minor shareholder of Novo Nordisk Inc. DS has served on advisory boards for, and received honoraria from, Novo Nordisk Inc. and Intarcia Therapeutics, Inc. Author Contributions: All the authors listed have participated sufficiently in the work, including participation in the concept, design, analysis, writing, or revision of the manuscript. All the authors reviewed and approved the final version of this article, and all the authors agree to be accountable for all aspects of the work.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
2
Acknowledgements: Not applicable. Previous Presentations: Liu J, Wang R, Ganz ML, Paprocki Y, Weatherall J. The Burden of Severe Hypoglycemia in Type 1 Diabetes. Poster presented at the AMCP Managed Care Specifically Pharmacy Annual Meeting 2017; March 27–30, 2017; Denver, Colorado.
ABSTRACT
us
cr
ip
t
Methods: Using Truven MarketScan claims, we identified adult T1DM patients using basal-bolus insulin regimens who were hospitalized for SHO (inpatient SHO patients) during 2010–2015. We defined two comparison groups: those with outpatient SHO-related encounters only, including emergency department (ED) visits without hospitalization (outpatient SHO patients), and those with no SHO- or acute hyperglycemia-related events (comparison patients). We estimated lengths of stay and SHO-related hospitalization costs and used propensity score and inverse probability weighting methods to adjust for baseline differences across the groups to evaluate longer-term impacts.
pt e
d
M
an
Results: We identified 8,734 patients; 4.2% experienced at least one SHO-related hospitalization. Among those who experienced SHO (i.e., of those in the inpatient and outpatient SHO groups), 31% experienced at least one SHO-related hospitalization, while 9% were treated in the ED without subsequent hospitalization. Approximately 79% of patients were admitted directly to the hospital; the remainder were first assessed or treated in the ED. The inpatient SHO patients stayed in the hospital, including time in the ED, for 1.7 days and incurred $3,551 in costs. About one-third of patients were hospitalized again for SHO. Inpatient SHO patients incurred significantly higher monthly costs after their initial SHO-related hospitalization than patients in the two other groups ($2,084 vs. $1,313 and $1,372), corresponding to 59% or 52% higher monthly costs for inpatient SHO patients.
ce
Limitations: These analyses excluded patients who did not seek ED or hospital care when faced with SHO; events may have been miscoded; and we were not able to account for clinical characteristics associated with SHO, such as insulin dose and duration of diabetes, or unmeasured confounders. Conclusions: The burden associated with SHO is not negligible. About 4% of T1DM patients using basal-bolus insulin regimens are hospitalized at least once due to SHO. Not only do those patients incur the costs of that SHO hospitalization, but also incur at least $712 (52%) more in costs per month after their hospitalization than outpatient SHO or comparison patients. Reducing SHO events can help decrease the burden associated with SHO among patients with T1DM.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Aims: Approximately 1.25 million people in the United States have type 1 diabetes mellitus (T1DM), a chronic metabolic disease that develops from the body’s inability to produce insulin, and requires life-long insulin therapy. Poor insulin adherence may cause severe hypoglycemia (SHO), leading to hospitalization and long-term complications; these, in turn, drive up costs of SHO and T1DM overall. This study’s objective was to estimate the prevalence and costs of SHO-related hospitalizations and their additional longer-term impacts on patients with T1DM using basal-bolus insulin.
KEY WORDS: type 1 diabetes mellitus, burden of illness, severe hypoglycemia, costs, hospitalization, basal-bolus insulin. SHORT TITLE: Severe Hypoglycemia in Type 1 Diabetes
an
us
cr
ip
t
Type 1 diabetes mellitus (T1DM), which affects about 1.25 million people in the United States (US), is a chronic metabolic disease that develops from the body’s inability to produce insulin and requires life-long therapy with basal (long-acting) and bolus (short-acting or mealtime) insulin [1,2,3]. Most basal insulins, which control blood glucose levels between meals, including during sleep, may be administered once or twice daily, depending on the specific formulation. Bolus insulin, which controls postprandial blood glucose levels, generally should be administered prior to eating. Hypoglycemia, or low blood glucose levels, can be caused by an excess of insulin in the body, which can result from an insulin dose given that is too large, too close in time to the previous dose, or the incorrect insulin was administered (usually an accidental interchange of basal and bolus types). Almost 98,000 emergency department (ED) visits between 2007 and 2011 were due to insulinrelated hypoglycemia and errors, with 29% of those ED visits resulting in hospitalization [4]. Several personal and clinical factors are related to patients’ willingness and ability to adhere to their treatment regimens, specifically the frequency and timing of insulin administration, with fear of hypoglycemia (FOH) due to an excessive insulin dose being one of the most important barriers to adherence [5,6,7,8]. FOH has been shown to be associated with negative impacts on disease management, glycemic control, and health outcomes due to missed or reduced insulin doses [6,7].
ce
pt e
d
M
Severe hypoglycemia (SHO) occurs when blood glucose levels become so low that assistance from another person is required. People who experience SHO may require hospitalization and it can result in seizures, coma, and death [9]. SHO can also be quite costly. Despite the clinical and economic importance of SHO, little is known about the immediate healthcare costs of an SHO event, and even less is known about the longer-term impact of SHO. Most existing literature focuses on the costs of SHO among T1DM patients in European countries, while very few studies have assessed the costs of SHO among T1DM patients in the US, and those are often incomplete. For example, Foos et al. (2015) [10] evaluated the direct and indirect costs of severe and non-severe hypoglycemic events, but the costs were evaluated based on a pooled T1DM and T2DM sample in the US; Reviriego et al. (2008) [11] reported direct and indirect costs per hypoglycemic episode in Spain, but only the associated hospitalization costs were reported separately; Cariou et al. (2015) [12] reported prevalence of SHO (13.4%) using a combination retrospective-prospective approach, but their data were limited to 30-day prevalence; Goldstein et al. (2016) [13] used data from an Israeli claims database covering 25% of the population to estimate the use of healthcare services after an SHO event; Majumdar et al. (2013) [14] used linked administrative healthcare and laboratory data from Canada to estimate the long-term risk of morbidity and mortality following SHO among older people, whether or not they had diabetes; and Elliott et al. (2016) [15] compared the hypoglycemia event rates between real world data and clinical data in the US, but did not evaluate the associated costs. Understanding the full extent of the economic burden of SHO requires longitudinal data, and understanding the economic impact of SHO apart from the costs associated with T1DM requires us to also examine the outcomes of patients who did not experience SHO events, which has not been consistently done in the literature on this topic. The objective of this study, therefore, is to address these gaps in the current literature on the economic costs of T1DM by estimating the frequency and
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
INTRODUCTION AND BACKGROUND
costs of SHO-related hospitalizations (plus preceding ED costs if any) and estimating the impact of SHO-related hospitalizations and preceding ED assessment/treatment on longer-term healthcare resource use and costs. 1
METHODS Data Source, Study Design, and Study Sample Selection
cr
ip
t
We analyzed encounters and services that occurred between January 2010 and December 2015 using data from the Truven MarketScan database. This database contains health insurance enrollment and claims information for approximately 30 million commercially-insured working age adults and their dependents, plus three million individuals insured through Medicare supplemental plans. Enrollment data, medical claims, and pharmacy claims are linked to each patient using an encrypted patient identification number.
ce
pt e
d
M
an
us
We selected adult (≥18 years) patients with evidence of T1DM, who had used basal-bolus insulin regimens for at least six months, and who were enrolled in their health plan for at least six months prior to their index date (defined below). Specifically, patients were included if they had two or more claims during the baseline period (defined below) with a T1DM diagnosis code (ICD-9-CM code 250.x1 or 250.x3) on separate dates. Patients were excluded from the sample if they had any claims with T2DM codes (ICD-9-CM code 250.x0 or 250.x2); used any non-insulin anti-diabetic medications; had any evidence of pregnancy, secondary diabetes (ICD-9-CM codes 249.xx), gestational diabetes (ICD-9-CM code 648.8), polycystic ovary syndrome (ICD-9-CM code 256.4); or used pre-mixed insulin or an insulin pump. We focused on the basal-bolus regimen because it is the most prevalent schedule among patients with T1DM and it is also the standard of care for most patients. We recognize that the basal-bolus regimen may be too restrictive for some patients and inappropriate for those at high risk of hypoglycemia; these, however, were also reasons that we focused on basal-bolus: to provide evidence on the negative impacts of SHO among people using this regimen. We assigned patients to one of three groups based their SHO history and where they were treated. Patients who were hospitalized, whether they were first assessed or treated at an ED or not, with SHO as their primary or secondary diagnosis were assigned to the inpatient SHO group. The claims database we used did not contain information to allow us to identify which diagnoses were present on admission. Therefore, we limited the diagnoses we examined to those reported for the first day of the hospitalization or ED admission. (Approximately 85% of the inpatient SHO patients using basal-bolus insulin regimen were identified based on the primary diagnosis.) Patients who experienced SHO events that were not treated in a hospital, but were treated in a physician’s office or clinic, were assigned to the outpatient SHO group. Patients who never experienced any SHO or acute hyperglycemia (AHR) were assigned to the comparison group (we excluded patients who ever experienced AHR, in addition to SHO, to create a comparison group that was free of problems associated with extremely high or low blood glucose levels, which are often the result of incorrect insulin doses or the timing of those doses). See Appendix A for further details.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
1.1
t
ip
cr
us
1.2.1
Study Measures Healthcare Resource Use and Costs
an
1.2
ce
pt e
d
M
Overall healthcare resource use, and their associated costs (measured by payments), were assessed during the baseline and follow-up periods and during the initial SHO-related hospitalization for patients in the inpatient SHO group. Total (summed) costs were computed for the baseline and follow-up periods and average monthly costs and any use of healthcare services were also computed and analyzed for the follow-up period. Diabetes-related resource use and costs were also separately assessed and were identified by medical claims with a primary or secondary diagnosis of T1DM. Costs for outpatient services, ED visits (whether for treatment or observation), and hospitalizations were operationalized as continuous measures, as was hospital length of stay (LOS). Costs were also computed for insulin and all other drugs, which, according the sample selection criteria, did not include non-insulin antidiabetic medications. Any hospitalizations and ED visits related to SHO or AHR that occurred during the follow-up period accrued to follow-up resource use and costs.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
We defined two study periods for each patient: a six-month baseline period during which no SHO-related hospitalizations or ED visits (for the inpatient SHO group) or other SHO-related events (for the outpatient SHO group) occurred and the follow-up period that began immediately after discharge from the hospital (for the inpatient SHO group) or the index date (for the outpatient SHO group) and continued until the last date of continuous enrollment in medical and pharmacy benefits, December 31, 2015, or death. All patients were assigned an index date, which was used to differentiate the baseline from the follow-up period. The index date for patients in the inpatient SHO group was the date of their first SHO-related hospitalization or ED visit resulting in a hospitalization. Index dates for patients in the two other groups were randomly selecting from a range of dates from six months after the start of continuous health plan enrollment to the last day of continuous enrollment, December 31, 2015, or death (whichever was earliest). We used this method for assigning index dates to patients in the outpatient SHO and comparison groups to guarantee they would have a six-month baseline period. The earliest possible start of a baseline period was, therefore, January 1, 2010, the earliest possible index date was July 1, 2010, and maximum possible duration of each patient’s follow-up period was 5.5 years.
1.2.2
Severe Hypoglycemia Events
Consistent with previously published research, we identified SHO-related events by the presence of diagnosis codes 251.0, 251.1, 251.2, 270.3, or 962.3; or the presence of diagnosis codes 251.8x without codes 259.8, 272.7, 681.xx, 682.xx, 686.9, 707.1x. 707.2x, 707.8, 707.9, 709.3, 730.1x, 730.2x, or 731.8 in the primary and secondary ICD-9-CM diagnosis fields [16,17,18]. The original algorithm for identifying SHO events published by Ginde et al. (2008) included ICD-9-CM codes 775.0 (hypoglycemia in an infant born to a diabetic mother) and 775.6 (neonatal hypoglycemia) [17], which we excluded due to the sample selection criteria. Hospitalizations immediately preceded (on
the same or day before) by an ED visit and/or ambulance claim (and ED visits preceded by ambulance claim) with an SHO diagnosis were considered related to SHO. Demographic and Clinical Characteristics
Statistical Analyses
an
1.3
us
cr
ip
t
Patients’ demographic and clinical characteristics included age; sex; geographic region of residence; type of insurance plan; presence of comorbid conditions, including microvascular and macrovascular complications; presence of mental health conditions; length of insulin use; and total healthcare costs. Demographic characteristics were assessed as of the index date or, if not available on the index date, the date in the baseline period nearest the index date; any presence of comorbid and mental health conditions were assessed during the baseline period. Length of insulin use was operationalized as a categorical measure capturing the use of insulin for less than six months before the index date (insulin-naïve) or for at least six months before the index date (insulin-experienced). A complete list of the comorbid conditions, complications, mental health conditions, and their definitions, is available in Appendix A. In addition to measuring baseline health status using individual health conditions, we also computed the Charlson Comorbidity Index (CCI) and operationalized it as continuous and ordinal categorical (0, 1, 2, 3, ≥ 4) measures [19,20].
pt e
d
M
Continuous measures were summarized by their means and standard deviations, and categorical variables were summarized by their proportions. Baseline demographic and clinical characteristics and follow-up healthcare resource use and costs were stratified by group; the t-test and Pearson 2 test were used to compare the means of continuous variables and distributions of categorical variables, respectively, between two groups.
ce
We used the inverse probability of treatment weighting (IPTW) method to analyze healthcare resource use and costs during the follow-up period. We first estimated a multinomial logit regression for group membership that included terms for sex, age groups, CCI score groups, total baseline healthcare costs, and length of follow-up to derive multiple propensity scores, which are the predicted probabilities of each observation belonging to each group [18]. Each observation was then weighted by the inverse of its predicted probability of actual group membership in subsequent analyses. We did not compute adjusted healthcare resource use and cost measures because the IPTW method balanced the distribution of observable confounders across the three groups (Appendix B displays the characteristics of the weighted sample).
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
1.2.3
2
RESULTS
We identified 8,734 patients (364 inpatient SHO, 802 outpatient SHO, and 7,568 comparison) who used basal-bolus insulin, corresponding to 8,724 weighted inpatient SHO, 8,701 weighted outpatient SHO, and 8,735 weighted comparison patients who satisfied the inclusion and exclusion criteria. Patients in the inpatient SHO group were followed, on average, for 654 (interquartile range [IQR]
270–980) days while patients in the other groups were followed, on average, for 546 (IQR 217–772) and 824 (IQR 366–1,294) days, respectively (See Table 1). 2.1
Patient Demographic and Clinical Characteristics
t
an
us
cr
ip
The most common comorbid conditions were dyslipidemia and hypertension, observed in 27%–32% and 20%–23% of patients, respectively, in the three groups; no other comorbidities nor conditions were observed in more than 8% of the patients. Renal disease, depression, diabetic retinopathy, and coronary artery disease affected 3% to 7% of the patients. Patients in the inpatient SHO group had slightly higher CCI scores than patients in the other groups (2.5 vs. 2.3); 14% of the patients in the inpatient SHO group had a CCI score ≥4 compared with 8%–9% of the patients in the two comparison groups. Less than 3% of the patients were insulin-naïve.
Severe Hypoglycemia-related Hospitalizations
pt e
2.2
d
M
Significant differences in baseline healthcare costs were observed. On average, patients in the inpatient SHO group incurred $8,904, whereas patients in the two comparison groups incurred $5,229–$5,386. Differences in total healthcare costs were primarily due to differences in inpatient hospitalization ($2,447 vs. $418–$434) and ED costs ($354 vs. $91–$174). Complete results are presented in Table 1.
ce
Table 2 displays hospitalization and cost information for the inpatient SHO group of patients, onefifth of whom were admitted directly to the hospital, where they stayed for 3.8 days and incurred $12,726; the remainder of patients were first assessed or treated in the ED and subsequently transferred to the hospital where they stayed for 1.2 days and incurred $1,090. On average, the entire inpatient SHO group of patients spent 1.7 days in the hospital and incurred costs of $3,551. Thirty-two percent of patients in the inpatient SHO group were hospitalized again for SHO and 11% were hospitalized for AHR at some point during the follow-up period.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
On average, patients in the inpatient SHO group were slightly older (47 years) than patients in the other groups (45–46 years); more than 60% of the patients in all three groups were male. Overall, the distribution of geographic region was similar across the groups except that inpatient SHO patients were slightly more likely to reside in the Northeast and slightly less likely to reside in the West. Nearly 90% of patients in all three groups were covered by a commercial plan.
2.3
Follow-up Healthcare Resource Use and Costs
Table 3 displays healthcare utilization and costs during the follow-up period. Patients in the inpatient SHO group were more likely to have used most types of healthcare services and incurred higher costs during the follow-up period than patients in the other groups. Patients in the inpatient SHO group were much more likely to have received inpatient (31%) and ED (53%) care during the follow-up period than patients in the outpatient SHO and comparison groups (15% and 35%, and 13% and 29%, respectively); use of outpatient care was similar across the groups (97%–98%).
t ip
3
DISCUSSION
ce
pt e
d
M
an
us
cr
This study contributes to the literature on the burden of SHO-related hospitalization in T1DM patients using basal-bolus insulin therapy. In particular, this work builds on the work of Heller et al. (2016) [21] by focusing on the real-world experiences of patients with T1DM using basal-bolus insulin regimes and on the work of Goldstein et al. (2016) [13], Lipska et al. (2014) [22], and Majumdar et al. (2013) [14], as well as others, by estimating the additional average monthly healthcare costs following the index SHO event using all available follow-up data, rather than data for only the first 30 days. We found that hospitalizations for SHO are quite common among patients with T1DM who are using a basal-bolus insulin regimen, and that patients who have experienced SHO-related hospitalizations are also at high risk of subsequent hospitalization and ED visits and incur substantial costs. Approximately 30% of T1DM patients using basal-bolus insulin who experienced a SHO event were hospitalized; each hospital admission cost $3,551. Monthly costs for inpatient SHO patients during the follow-up period were at least $712 larger than for patients who did not experience SHO- (or AHR-) related hospitalization. We also found that 9% of patients who experienced SHO were treated in the ED without subsequent hospital admission, a proportion roughly equal to that of patients who experienced SHO and were admitted directly to the hospital. Although our focus here was on SHO-related hospitalizations, including patients treated only in the ED would clearly increase the size, and potentially the cost, of patients in the inpatient SHO group. Using claims and other administrative data to study the short- and longer-term implications of SHO is challenging. Although it is unlikely that SHO-related hospitalizations were incorrectly identified as such in the claims data we analyzed, it is possible that SHO events that did not result in a hospitalization were not captured or recorded correctly, particularly if the events did not result in a billable service, did not involve an interaction with a healthcare provider, or an appropriate diagnosis code was not used, which could be quite common [23]. Such miscoded, or underreported, events would likely have been assigned to the comparison group rather than to the appropriate outpatient SHO group, resulting in underestimated monthly costs for the outpatient SHO group and overestimated monthly costs for the comparison group. The degree to which these cost estimates are biased may also vary by patient characteristics if, as Veronese et al. (2016) has shown using Italian data[24], underreporting is related to disease severity. Our definition of the comparison
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Inpatient SHO patients incurred significantly (51% and 47%) higher costs over the follow-up period than patients in the two other groups ($53,353 vs. $35,414 and $36,230, respectively), corresponding to average monthly costs of $2,084 for inpatient SHO patients and $1,313 and $1,372 for patients in the other groups. Differences in total monthly healthcare costs are primarily due to differences in inpatient ($633 vs. $219 and $272, respectively) and outpatient ($845 vs. $500 and $545, respectively) services. Patients in the inpatient SHO group incurred higher diabetes- and nondiabetes-related non-medication costs ($589 and $985) than patients in the other groups ($329 and $447, $272 and $588, respectively). Medication costs were similar across the three groups, whether for insulin or all other types of drugs ($283–$311 and $228–$237, respectively). Complete results are presented in Table 3.
t
ip
cr
us
ce
pt e
d
M
an
Our results, that patients incur higher healthcare costs after being hospitalized for SHO, are important because they imply that the impacts of SHO-related hospitalizations are not limited to the acute period during which patients are hospitalized, but persist for some time as well, and that health services researchers should be aware of these spill-over effects when designing future studies of SHO. Specifically, the longer-term burden of SHO-related hospitalization is primarily due to greater inpatient and outpatient care costs, even after adjusting, through our use of the IPTW method, for baseline differences in comorbid conditions and the use of healthcare services across the groups studied. These findings can also inform economic evaluations of therapies with different SHO profiles. However, more importantly, our findings further imply that interventions designed to prevent or even just delay SHO events requiring hospitalization can be associated with significant economic benefits beyond just reduced SHO hospitalization costs among commercially-insured and Medicare patients.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
group, which excluded patients who experienced SHO and AHR events, may have resulted in artificially low resource use and cost estimates during the follow-up period for comparison patients because those patients were, by definition, less complex. However, we elected to define the comparison group that way to provide an absolute lower bound on estimates of resource use and costs in the absence of SHO. Our data also limited our analyses in other ways. We were unable to differentiate between daytime and nocturnal hypoglycemia nor could we account for several clinical characteristics known to be associated with SHO such as HbA1c, body-mass index, renal function, and lipid levels that are not typically available in claims data. We were also unable to account or adjust for insulin dose or duration of diabetes and our sample exclusion criterion may have resulted in a sample that was healthier than the target population of T1DM patients using a basal-bolus insulin regimen. Furthermore, we were not able to limit index hospitalizations and ED admissions to those cases in which SHO was the verifiable cause of the admission. Therefore, some index admissions may be due to iatrogenic SHO. However, because we used diagnosis data from the first day of the admission, we have limited the extent of this potential bias. Finally, although we used the IPTW method intended to balance the characteristics of the three groups we investigated, we could not account for unmeasured confounders.
4
REFERENCES
1. American Diabetes Association. Economic costs of diabetes in the U.S. in 2012. Diabetes Care. 2013;36:1033-1046.
cr
ip
t
3. Centers for Disease Control and Prevention. National Diabetes Statistics Report: Estimates of Diabetes and Its Burden in the United States, 2014. Atlanta, GA: U.S. Department of Health and Human Services; 2014 [cited 2016 June 24]. Available from: http://www.cdc.gov/diabetes/pubs/statsreport14/national-diabetes-report-web.pdf.
an
us
4. Geller AI, Shehab N, Lovegrove MC, et al. National estimates of insulin-related hypoglycemia and errors leading to emergency department visits and hospitalizations. JAMA Intern Med. 2014;174:678-686.
M
5. Anderbro T, Amsberg S, Adamson U, et al. Fear of hypoglycaemia in adults with Type 1 diabetes. Diabet Med. 2010;27:1151-1158.
pt e
d
6. Fidler C, Elmelund Christensen T, Gillard S. Hypoglycemia: an overview of fear of hypoglycemia, quality-of-life, and impact on costs. J Med Econ. 2011;14:646-655.
ce
7. International Hypoglycaemia Study G. Minimizing Hypoglycemia in Diabetes. Diabetes Care. 2015;38:1583-1591.
8. Wild D, von Maltzahn R, Brohan E, et al. A critical review of the literature on fear of hypoglycemia in diabetes: Implications for diabetes management and patient education. Patient Educ Couns. 2007;68:10-15.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
2. McCoy RG, Zhang Y, Herrin J, et al. Changing trends in type 2 diabetes mellitus treatment intensification, 2002-2010. Am J Manag Care. 2015;21:e288-296.
9. Cryer PE. Hypoglycemia-Associated Autonomic Failure in Diabetes: Maladaptive, Adaptive, or Both? Diabetes. 2015;64:2322-2323.
10. Foos V, Varol N, Curtis BH, et al. Economic impact of severe and non-severe hypoglycemia in patients with Type 1 and Type 2 diabetes in the United States. J Med Econ. 2015;18:420-432.
11. Reviriego J, Gomis R, Maranes JP, et al. Cost of severe hypoglycaemia in patients with type 1 diabetes in Spain and the cost-effectiveness of insulin lispro compared with regular human insulin in preventing severe hypoglycaemia. Int J Clin Pract. 2008;62:1026-1032.
ip
t
13. Goldstein D, Chodick G, Shalev V, et al. Use of Healthcare Services Following Severe Hypoglycemia in Patients with Diabetes: Analysis of Real-World Data. Diabetes Ther. 2016;7:295-308.
us
cr
14. Majumdar SR, Hemmelgarn BR, Lin M, et al. Hypoglycemia associated with hospitalization and adverse events in older people: population-based cohort study. Diabetes Care. 2013;36:35853590.
M
an
15. Elliott L, Fidler C, Ditchfield A, et al. Hypoglycemia Event Rates: A Comparison Between RealWorld Data and Randomized Controlled Trial Populations in Insulin-Treated Diabetes. Diabetes Ther. 2016;7:45-60.
pt e
d
16. Ganz M, Wintfeld N, Li Q, et al. Severe hypoglycemia rates and associated costs among type 2 diabetics starting basal insulin therapy in the United States. Curr Med Res Opin 2014;30:1991-2000.
ce
17. Ginde AA, Blanc PG, Lieberman RM, et al. Validation of ICD-9-CM coding algorithm for improved identification of hypoglycemia visits. BMC Endocr Disord. 2008;8:4.
18. Simeone JC, Quilliam BJ. Predictors of emergency department and outpatient visits for hypoglycemia in type 2 diabetes: an analysis of a large US administrative claims database. Ann Pharmacother. 2012;46:157-168.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
12. Cariou B, Fontaine P, Eschwege E, et al. Frequency and predictors of confirmed hypoglycaemia in type 1 and insulin-treated type 2 diabetes mellitus patients in a real-life setting: results from the DIALOG study. Diabetes Metab. 2015;41:116-125.
19. D’Hoore W, Bouckaert A, Tilquin C. Practical considerations on the use of the Charlson comorbidity index with administrative data bases. J Clin Epidemiol. 1996;49:1429-1433.
20. Schneeweiss S, Maclure M. Use of comorbidity scores for control of confounding in studies using administrative databases. Int J Epidemiol. 2000;29:891-898.
21. Heller SR, Frier BM, Herslov ML, et al. Severe hypoglycaemia in adults with insulin-treated diabetes: impact on healthcare resources. Diabet Med. 2016;33:471-477.
ip
t
23. Jakubczyk M, Rdzanek E, Niewada M, et al. Economic resources consumption structure in severe hypoglycemia episodes: a systematic review and meta-analysis. Expert Rev Pharmacoecon Outcomes Res. 2015;15:813-822.
ce
pt e
d
M
an
us
cr
24. Veronese G, Marchesini G, Forlani G, et al. Are severe hypoglycemic episodes in diabetes correctly identified by administrative data? Evidence of underreporting from the HYPOTHESIS study. Acta diabetologica. 2016;53:677-680.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
22. Lipska KJ, Ross JS, Wang Y, et al. National trends in US hospital admissions for hyperglycemia and hypoglycemia among Medicare beneficiaries, 1999 to 2011. JAMA Intern Med. 2014;174:11161124.
Table 1. Baseline Demographic and Clinical Characteristics of Patients with Type 1 Diabetes Mellitus using Basal-Bolus Insulin Therapy (N=8,734) Characteristics
Inpatient SHO (N=364)
Mean Duration of Follow-up Period (SD), days
654.3 (480.2)
546.2 (424.1)
824.4 (521.8)
47.0 (17.2)
44.6 (16.7)
45.6 (16.3)
18–34
26.6
32.8
28.7
35–44
17.3
16.8
19.3
45–54
21.7
22.9
21.4
55–64
21.2
15.8
≥65
13.2
11.6
Mean (SD) Age, years
Outpatient SHO (N=802)
Comparison (N=7,568)
62.6
Female
37.4
Geographic Region, %
t ip
19.2 11.4
62.5
62.5
37.5
37.5
18.8
16.6
25.3
28.6
us
Male
cr
Gender, %
22.0
Midwest
25.8
South
38.5
35.9
37.1
West
12.9
19.2
17.3
0.8
0.7
0.4
87.1
90.4
89.7
12.9
9.6
10.3
31.9
32.4
27.3
23.1
23.2
20.4
7.4
4.0
3.6
Congestive heart failure
4.1
1.7
1.7
Malignancy (excl. malignant neoplasm of skin)
3.3
2.2
2.6
Cerebrovascular disease
2.7
1.6
1.4
Peripheral vascular disease
2.7
1.4
1.6
Chronic pulmonary disease
1.9
3.9
3.4
Liver disease
1.9
0.9
0.8
Rheumatic disease
1.6
1.5
0.9
Hemiplegia or paraplegia
0.5
0.1
0.1
HIV/AIDS
0.3
0.0
0.1
Metastatic solid tumor
0.0
0.2
0.3
Depression
6.3
5.6
4.0
Anxiety
2.5
2.4
1.4
Dementia
0.5
0.0
0.1
an
Northeast
M
Unknown Insurance Type, %
d
Commercial Physical Comorbidities, % Dyslipidemia Hypertension
ce
Renal disease
pt e
Medicare
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Age, years, %
Mental Health Comorbidities, %
Inpatient SHO (N=364)
Characteristics
Outpatient SHO (N=802)
Comparison (N=7,568)
Microvascular Complications, % Diabetic retinopathy
7.4
5.4
6.1
Diabetic neuropathy
5.5
6.4
5.1
Diabetic nephropathy
4.4
4.7
3.2
Coronary artery disease
4.4
3.4
4.5
Peripheral arterial disease
1.9
2.1
1.7
Stroke
1.4
0.1
0.4
2.5 (1.0)
2.3 (0.9)
2.3 (0.9)
0.0
0.0
2
78.6
86.8
3
7.1
≥4
14.3
t 0.0
87.5 4.4
8.9
8.1
2.6
3.2
us
4.4
Insulin-naïve, %
2.7
Mean (SD) Follow-up, Days
654.3 (480.2)
546.2 (424.1)
824.4 (521.8)
an
0, 1
cr
Charlson Comorbidity Index Score, %
ip
Mean (SD) Charlson Comorbidity Index Score
2,447 (19,348)
418 (3,416)
434 (5,236)
354 (2,503)
174 (1,045)
91 (648)
Mean (SD) Baseline Healthcare Costs, $ Inpatient hospitalizations
M
Emergency department admissions Outpatient services Medications
d
Total
2,881 (12,409)
1,985 (5,845)
2,151 (10,300)
3,222 (7,293)
2,810 (2,281)
2,553 (2,711)
8,904 (25,768)
5,386 (7,813)
5,229 (13,512)
ce
pt e
Abbreviations: AIDS, acquired immune deficiency syndrome; HIV, Human immunodeficiency virus; SD, standard deviation; SHO, severe hypoglycemia
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Macrovascular Complications, %
Table 2. Severe Hypoglycemia-Related Hospitalizations and Emergency Department Admissions, Costs, and Risk of Subsequent Hospitalization of Patients with Type 1 Diabetes Mellitus using Basal-Bolus Insulin Therapy (N=364) Outcomes Hospitalized for SHO, % Without Prior ED Admission
21.2
With Prior ED Admission
78.8 1.7 (2.9)
Patients without Prior ED Admission
3.8 (5.6)
ip
t
All Patients Patients with Prior ED Admission
1.2 (0.9)
Mean (SD) Cost, $
3,551 (16,740)
cr
All Patients
12,726 (33,389)
Patients with Prior ED Admission Subsequent Hospitalization with and without Prior ED Admission, % None
an
Second SHO
us
Patients without Prior ED Admission
AHR
1,090 (5,542) 57.1 32.1 10.7
ce
pt e
d
M
Abbreviations: AHR, acute hyperglycemia; ED, emergency department; SD, standard deviation; SHO, severe hypoglycemia. Subsequent hospitalizations may not sum to 100% due to rounding.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Mean (SD) Length of Stay, Days
Table 3. Follow-up Healthcare Utilization and Adjusted Monthly Costs of Patients with Type 1 Diabetes Mellitus using Basal-Bolus Insulin Therapy (N=8,734): Inverse Probability of Treatment Weighted Results P-Value Inpatient SHO (N=364)
Outcomes
8,724
8,701
Comparison (N=7,568)
Emergency department admissions
53.0
35.0
29.4
Outpatient services
98.1
97.4
96.7
100.0
99.9
100.0
Medications Mean (SD) Monthly Costs, $
Outpatient services Medications
Other
0.58
1.00
1.00
ISSN: 0300-7995 (Print) 1473-4877 (Online) Journal homepage: http://www.tandfonline.com/loi/icmo20
The burden of severe hypoglycemia in type 1 diabetes Jieruo Liu, Rosa Wang, Michael L. Ganz, Yurek Paprocki, Doron Schneider & James Weatherall To cite this article: Jieruo Liu, Rosa Wang, Michael L. Ganz, Yurek Paprocki, Doron Schneider & James Weatherall (2017): The burden of severe hypoglycemia in type 1 diabetes, Current Medical Research and Opinion, DOI: 10.1080/03007995.2017.1391079 To link to this article: http://dx.doi.org/10.1080/03007995.2017.1391079
View supplementary material
Accepted author version posted online: 11 Oct 2017.
Submit your article to this journal
Article views: 1
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=icmo20 Download by: [UNIVERSITY OF ADELAIDE LIBRARIES]
Date: 13 October 2017, At: 07:03
The burden of severe hypoglycemia in type 1 diabetes
Jieruo Liu [1], Rosa Wang [1], Michael L. Ganz [1], Yurek Paprocki [2], Doron Schneider [3], James Weatherall [2] 1
Evidera, 500 Totten Pond Road, Fifth Floor, Waltham, MA, 02451, USA Novo Nordisk Inc., 800 Scudders Mill Rd, Plainsboro, NJ, 08536, USA 3 Abington Adult Medical Associates, 1235 Old York Rd, Ste. 214, Abington, PA, 19001, USA
t
ip
cr
CORRESPONDING AUTHOR: Michael L. Ganz, 500 Totten Pond Road, 5th Floor, Waltham, MA, 02451, USA. Email: [email protected]
us
TRANSPARENCY
an
Declaration of Funding: Novo Nordisk Inc., a pharmaceutical company that manufactures, markets and/or distributes more than a dozen drugs for diabetes in the United States, provided the funding for this study and article.
ce
pt e
d
M
Declaration of Financial/Other Relationships: JL and MLG are employed by Evidera Inc., an independent research company that provides consulting and other research services to the biopharmaceutical industry. In their salaried positions, they work with a variety of client companies and are precluded from receiving payment or honoraria directly from these organizations for services rendered. Evidera received payment from Novo Nordisk Inc. for the conduct of this study and the development of this manuscript. RW was an employee of Evidera Inc. when this manuscript was written. YP is an employee of Novo Nordisk Inc. JW was an employee of Novo Nordisk Inc when this manuscript was written; he is a minor shareholder of Novo Nordisk Inc. DS has served on advisory boards for, and received honoraria from, Novo Nordisk Inc. and Intarcia Therapeutics, Inc. Author Contributions: All the authors listed have participated sufficiently in the work, including participation in the concept, design, analysis, writing, or revision of the manuscript. All the authors reviewed and approved the final version of this article, and all the authors agree to be accountable for all aspects of the work.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
2
Acknowledgements: Not applicable. Previous Presentations: Liu J, Wang R, Ganz ML, Paprocki Y, Weatherall J. The Burden of Severe Hypoglycemia in Type 1 Diabetes. Poster presented at the AMCP Managed Care Specifically Pharmacy Annual Meeting 2017; March 27–30, 2017; Denver, Colorado.
ABSTRACT
us
cr
ip
t
Methods: Using Truven MarketScan claims, we identified adult T1DM patients using basal-bolus insulin regimens who were hospitalized for SHO (inpatient SHO patients) during 2010–2015. We defined two comparison groups: those with outpatient SHO-related encounters only, including emergency department (ED) visits without hospitalization (outpatient SHO patients), and those with no SHO- or acute hyperglycemia-related events (comparison patients). We estimated lengths of stay and SHO-related hospitalization costs and used propensity score and inverse probability weighting methods to adjust for baseline differences across the groups to evaluate longer-term impacts.
pt e
d
M
an
Results: We identified 8,734 patients; 4.2% experienced at least one SHO-related hospitalization. Among those who experienced SHO (i.e., of those in the inpatient and outpatient SHO groups), 31% experienced at least one SHO-related hospitalization, while 9% were treated in the ED without subsequent hospitalization. Approximately 79% of patients were admitted directly to the hospital; the remainder were first assessed or treated in the ED. The inpatient SHO patients stayed in the hospital, including time in the ED, for 1.7 days and incurred $3,551 in costs. About one-third of patients were hospitalized again for SHO. Inpatient SHO patients incurred significantly higher monthly costs after their initial SHO-related hospitalization than patients in the two other groups ($2,084 vs. $1,313 and $1,372), corresponding to 59% or 52% higher monthly costs for inpatient SHO patients.
ce
Limitations: These analyses excluded patients who did not seek ED or hospital care when faced with SHO; events may have been miscoded; and we were not able to account for clinical characteristics associated with SHO, such as insulin dose and duration of diabetes, or unmeasured confounders. Conclusions: The burden associated with SHO is not negligible. About 4% of T1DM patients using basal-bolus insulin regimens are hospitalized at least once due to SHO. Not only do those patients incur the costs of that SHO hospitalization, but also incur at least $712 (52%) more in costs per month after their hospitalization than outpatient SHO or comparison patients. Reducing SHO events can help decrease the burden associated with SHO among patients with T1DM.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Aims: Approximately 1.25 million people in the United States have type 1 diabetes mellitus (T1DM), a chronic metabolic disease that develops from the body’s inability to produce insulin, and requires life-long insulin therapy. Poor insulin adherence may cause severe hypoglycemia (SHO), leading to hospitalization and long-term complications; these, in turn, drive up costs of SHO and T1DM overall. This study’s objective was to estimate the prevalence and costs of SHO-related hospitalizations and their additional longer-term impacts on patients with T1DM using basal-bolus insulin.
KEY WORDS: type 1 diabetes mellitus, burden of illness, severe hypoglycemia, costs, hospitalization, basal-bolus insulin. SHORT TITLE: Severe Hypoglycemia in Type 1 Diabetes
an
us
cr
ip
t
Type 1 diabetes mellitus (T1DM), which affects about 1.25 million people in the United States (US), is a chronic metabolic disease that develops from the body’s inability to produce insulin and requires life-long therapy with basal (long-acting) and bolus (short-acting or mealtime) insulin [1,2,3]. Most basal insulins, which control blood glucose levels between meals, including during sleep, may be administered once or twice daily, depending on the specific formulation. Bolus insulin, which controls postprandial blood glucose levels, generally should be administered prior to eating. Hypoglycemia, or low blood glucose levels, can be caused by an excess of insulin in the body, which can result from an insulin dose given that is too large, too close in time to the previous dose, or the incorrect insulin was administered (usually an accidental interchange of basal and bolus types). Almost 98,000 emergency department (ED) visits between 2007 and 2011 were due to insulinrelated hypoglycemia and errors, with 29% of those ED visits resulting in hospitalization [4]. Several personal and clinical factors are related to patients’ willingness and ability to adhere to their treatment regimens, specifically the frequency and timing of insulin administration, with fear of hypoglycemia (FOH) due to an excessive insulin dose being one of the most important barriers to adherence [5,6,7,8]. FOH has been shown to be associated with negative impacts on disease management, glycemic control, and health outcomes due to missed or reduced insulin doses [6,7].
ce
pt e
d
M
Severe hypoglycemia (SHO) occurs when blood glucose levels become so low that assistance from another person is required. People who experience SHO may require hospitalization and it can result in seizures, coma, and death [9]. SHO can also be quite costly. Despite the clinical and economic importance of SHO, little is known about the immediate healthcare costs of an SHO event, and even less is known about the longer-term impact of SHO. Most existing literature focuses on the costs of SHO among T1DM patients in European countries, while very few studies have assessed the costs of SHO among T1DM patients in the US, and those are often incomplete. For example, Foos et al. (2015) [10] evaluated the direct and indirect costs of severe and non-severe hypoglycemic events, but the costs were evaluated based on a pooled T1DM and T2DM sample in the US; Reviriego et al. (2008) [11] reported direct and indirect costs per hypoglycemic episode in Spain, but only the associated hospitalization costs were reported separately; Cariou et al. (2015) [12] reported prevalence of SHO (13.4%) using a combination retrospective-prospective approach, but their data were limited to 30-day prevalence; Goldstein et al. (2016) [13] used data from an Israeli claims database covering 25% of the population to estimate the use of healthcare services after an SHO event; Majumdar et al. (2013) [14] used linked administrative healthcare and laboratory data from Canada to estimate the long-term risk of morbidity and mortality following SHO among older people, whether or not they had diabetes; and Elliott et al. (2016) [15] compared the hypoglycemia event rates between real world data and clinical data in the US, but did not evaluate the associated costs. Understanding the full extent of the economic burden of SHO requires longitudinal data, and understanding the economic impact of SHO apart from the costs associated with T1DM requires us to also examine the outcomes of patients who did not experience SHO events, which has not been consistently done in the literature on this topic. The objective of this study, therefore, is to address these gaps in the current literature on the economic costs of T1DM by estimating the frequency and
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
INTRODUCTION AND BACKGROUND
costs of SHO-related hospitalizations (plus preceding ED costs if any) and estimating the impact of SHO-related hospitalizations and preceding ED assessment/treatment on longer-term healthcare resource use and costs. 1
METHODS Data Source, Study Design, and Study Sample Selection
cr
ip
t
We analyzed encounters and services that occurred between January 2010 and December 2015 using data from the Truven MarketScan database. This database contains health insurance enrollment and claims information for approximately 30 million commercially-insured working age adults and their dependents, plus three million individuals insured through Medicare supplemental plans. Enrollment data, medical claims, and pharmacy claims are linked to each patient using an encrypted patient identification number.
ce
pt e
d
M
an
us
We selected adult (≥18 years) patients with evidence of T1DM, who had used basal-bolus insulin regimens for at least six months, and who were enrolled in their health plan for at least six months prior to their index date (defined below). Specifically, patients were included if they had two or more claims during the baseline period (defined below) with a T1DM diagnosis code (ICD-9-CM code 250.x1 or 250.x3) on separate dates. Patients were excluded from the sample if they had any claims with T2DM codes (ICD-9-CM code 250.x0 or 250.x2); used any non-insulin anti-diabetic medications; had any evidence of pregnancy, secondary diabetes (ICD-9-CM codes 249.xx), gestational diabetes (ICD-9-CM code 648.8), polycystic ovary syndrome (ICD-9-CM code 256.4); or used pre-mixed insulin or an insulin pump. We focused on the basal-bolus regimen because it is the most prevalent schedule among patients with T1DM and it is also the standard of care for most patients. We recognize that the basal-bolus regimen may be too restrictive for some patients and inappropriate for those at high risk of hypoglycemia; these, however, were also reasons that we focused on basal-bolus: to provide evidence on the negative impacts of SHO among people using this regimen. We assigned patients to one of three groups based their SHO history and where they were treated. Patients who were hospitalized, whether they were first assessed or treated at an ED or not, with SHO as their primary or secondary diagnosis were assigned to the inpatient SHO group. The claims database we used did not contain information to allow us to identify which diagnoses were present on admission. Therefore, we limited the diagnoses we examined to those reported for the first day of the hospitalization or ED admission. (Approximately 85% of the inpatient SHO patients using basal-bolus insulin regimen were identified based on the primary diagnosis.) Patients who experienced SHO events that were not treated in a hospital, but were treated in a physician’s office or clinic, were assigned to the outpatient SHO group. Patients who never experienced any SHO or acute hyperglycemia (AHR) were assigned to the comparison group (we excluded patients who ever experienced AHR, in addition to SHO, to create a comparison group that was free of problems associated with extremely high or low blood glucose levels, which are often the result of incorrect insulin doses or the timing of those doses). See Appendix A for further details.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
1.1
t
ip
cr
us
1.2.1
Study Measures Healthcare Resource Use and Costs
an
1.2
ce
pt e
d
M
Overall healthcare resource use, and their associated costs (measured by payments), were assessed during the baseline and follow-up periods and during the initial SHO-related hospitalization for patients in the inpatient SHO group. Total (summed) costs were computed for the baseline and follow-up periods and average monthly costs and any use of healthcare services were also computed and analyzed for the follow-up period. Diabetes-related resource use and costs were also separately assessed and were identified by medical claims with a primary or secondary diagnosis of T1DM. Costs for outpatient services, ED visits (whether for treatment or observation), and hospitalizations were operationalized as continuous measures, as was hospital length of stay (LOS). Costs were also computed for insulin and all other drugs, which, according the sample selection criteria, did not include non-insulin antidiabetic medications. Any hospitalizations and ED visits related to SHO or AHR that occurred during the follow-up period accrued to follow-up resource use and costs.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
We defined two study periods for each patient: a six-month baseline period during which no SHO-related hospitalizations or ED visits (for the inpatient SHO group) or other SHO-related events (for the outpatient SHO group) occurred and the follow-up period that began immediately after discharge from the hospital (for the inpatient SHO group) or the index date (for the outpatient SHO group) and continued until the last date of continuous enrollment in medical and pharmacy benefits, December 31, 2015, or death. All patients were assigned an index date, which was used to differentiate the baseline from the follow-up period. The index date for patients in the inpatient SHO group was the date of their first SHO-related hospitalization or ED visit resulting in a hospitalization. Index dates for patients in the two other groups were randomly selecting from a range of dates from six months after the start of continuous health plan enrollment to the last day of continuous enrollment, December 31, 2015, or death (whichever was earliest). We used this method for assigning index dates to patients in the outpatient SHO and comparison groups to guarantee they would have a six-month baseline period. The earliest possible start of a baseline period was, therefore, January 1, 2010, the earliest possible index date was July 1, 2010, and maximum possible duration of each patient’s follow-up period was 5.5 years.
1.2.2
Severe Hypoglycemia Events
Consistent with previously published research, we identified SHO-related events by the presence of diagnosis codes 251.0, 251.1, 251.2, 270.3, or 962.3; or the presence of diagnosis codes 251.8x without codes 259.8, 272.7, 681.xx, 682.xx, 686.9, 707.1x. 707.2x, 707.8, 707.9, 709.3, 730.1x, 730.2x, or 731.8 in the primary and secondary ICD-9-CM diagnosis fields [16,17,18]. The original algorithm for identifying SHO events published by Ginde et al. (2008) included ICD-9-CM codes 775.0 (hypoglycemia in an infant born to a diabetic mother) and 775.6 (neonatal hypoglycemia) [17], which we excluded due to the sample selection criteria. Hospitalizations immediately preceded (on
the same or day before) by an ED visit and/or ambulance claim (and ED visits preceded by ambulance claim) with an SHO diagnosis were considered related to SHO. Demographic and Clinical Characteristics
Statistical Analyses
an
1.3
us
cr
ip
t
Patients’ demographic and clinical characteristics included age; sex; geographic region of residence; type of insurance plan; presence of comorbid conditions, including microvascular and macrovascular complications; presence of mental health conditions; length of insulin use; and total healthcare costs. Demographic characteristics were assessed as of the index date or, if not available on the index date, the date in the baseline period nearest the index date; any presence of comorbid and mental health conditions were assessed during the baseline period. Length of insulin use was operationalized as a categorical measure capturing the use of insulin for less than six months before the index date (insulin-naïve) or for at least six months before the index date (insulin-experienced). A complete list of the comorbid conditions, complications, mental health conditions, and their definitions, is available in Appendix A. In addition to measuring baseline health status using individual health conditions, we also computed the Charlson Comorbidity Index (CCI) and operationalized it as continuous and ordinal categorical (0, 1, 2, 3, ≥ 4) measures [19,20].
pt e
d
M
Continuous measures were summarized by their means and standard deviations, and categorical variables were summarized by their proportions. Baseline demographic and clinical characteristics and follow-up healthcare resource use and costs were stratified by group; the t-test and Pearson 2 test were used to compare the means of continuous variables and distributions of categorical variables, respectively, between two groups.
ce
We used the inverse probability of treatment weighting (IPTW) method to analyze healthcare resource use and costs during the follow-up period. We first estimated a multinomial logit regression for group membership that included terms for sex, age groups, CCI score groups, total baseline healthcare costs, and length of follow-up to derive multiple propensity scores, which are the predicted probabilities of each observation belonging to each group [18]. Each observation was then weighted by the inverse of its predicted probability of actual group membership in subsequent analyses. We did not compute adjusted healthcare resource use and cost measures because the IPTW method balanced the distribution of observable confounders across the three groups (Appendix B displays the characteristics of the weighted sample).
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
1.2.3
2
RESULTS
We identified 8,734 patients (364 inpatient SHO, 802 outpatient SHO, and 7,568 comparison) who used basal-bolus insulin, corresponding to 8,724 weighted inpatient SHO, 8,701 weighted outpatient SHO, and 8,735 weighted comparison patients who satisfied the inclusion and exclusion criteria. Patients in the inpatient SHO group were followed, on average, for 654 (interquartile range [IQR]
270–980) days while patients in the other groups were followed, on average, for 546 (IQR 217–772) and 824 (IQR 366–1,294) days, respectively (See Table 1). 2.1
Patient Demographic and Clinical Characteristics
t
an
us
cr
ip
The most common comorbid conditions were dyslipidemia and hypertension, observed in 27%–32% and 20%–23% of patients, respectively, in the three groups; no other comorbidities nor conditions were observed in more than 8% of the patients. Renal disease, depression, diabetic retinopathy, and coronary artery disease affected 3% to 7% of the patients. Patients in the inpatient SHO group had slightly higher CCI scores than patients in the other groups (2.5 vs. 2.3); 14% of the patients in the inpatient SHO group had a CCI score ≥4 compared with 8%–9% of the patients in the two comparison groups. Less than 3% of the patients were insulin-naïve.
Severe Hypoglycemia-related Hospitalizations
pt e
2.2
d
M
Significant differences in baseline healthcare costs were observed. On average, patients in the inpatient SHO group incurred $8,904, whereas patients in the two comparison groups incurred $5,229–$5,386. Differences in total healthcare costs were primarily due to differences in inpatient hospitalization ($2,447 vs. $418–$434) and ED costs ($354 vs. $91–$174). Complete results are presented in Table 1.
ce
Table 2 displays hospitalization and cost information for the inpatient SHO group of patients, onefifth of whom were admitted directly to the hospital, where they stayed for 3.8 days and incurred $12,726; the remainder of patients were first assessed or treated in the ED and subsequently transferred to the hospital where they stayed for 1.2 days and incurred $1,090. On average, the entire inpatient SHO group of patients spent 1.7 days in the hospital and incurred costs of $3,551. Thirty-two percent of patients in the inpatient SHO group were hospitalized again for SHO and 11% were hospitalized for AHR at some point during the follow-up period.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
On average, patients in the inpatient SHO group were slightly older (47 years) than patients in the other groups (45–46 years); more than 60% of the patients in all three groups were male. Overall, the distribution of geographic region was similar across the groups except that inpatient SHO patients were slightly more likely to reside in the Northeast and slightly less likely to reside in the West. Nearly 90% of patients in all three groups were covered by a commercial plan.
2.3
Follow-up Healthcare Resource Use and Costs
Table 3 displays healthcare utilization and costs during the follow-up period. Patients in the inpatient SHO group were more likely to have used most types of healthcare services and incurred higher costs during the follow-up period than patients in the other groups. Patients in the inpatient SHO group were much more likely to have received inpatient (31%) and ED (53%) care during the follow-up period than patients in the outpatient SHO and comparison groups (15% and 35%, and 13% and 29%, respectively); use of outpatient care was similar across the groups (97%–98%).
t ip
3
DISCUSSION
ce
pt e
d
M
an
us
cr
This study contributes to the literature on the burden of SHO-related hospitalization in T1DM patients using basal-bolus insulin therapy. In particular, this work builds on the work of Heller et al. (2016) [21] by focusing on the real-world experiences of patients with T1DM using basal-bolus insulin regimes and on the work of Goldstein et al. (2016) [13], Lipska et al. (2014) [22], and Majumdar et al. (2013) [14], as well as others, by estimating the additional average monthly healthcare costs following the index SHO event using all available follow-up data, rather than data for only the first 30 days. We found that hospitalizations for SHO are quite common among patients with T1DM who are using a basal-bolus insulin regimen, and that patients who have experienced SHO-related hospitalizations are also at high risk of subsequent hospitalization and ED visits and incur substantial costs. Approximately 30% of T1DM patients using basal-bolus insulin who experienced a SHO event were hospitalized; each hospital admission cost $3,551. Monthly costs for inpatient SHO patients during the follow-up period were at least $712 larger than for patients who did not experience SHO- (or AHR-) related hospitalization. We also found that 9% of patients who experienced SHO were treated in the ED without subsequent hospital admission, a proportion roughly equal to that of patients who experienced SHO and were admitted directly to the hospital. Although our focus here was on SHO-related hospitalizations, including patients treated only in the ED would clearly increase the size, and potentially the cost, of patients in the inpatient SHO group. Using claims and other administrative data to study the short- and longer-term implications of SHO is challenging. Although it is unlikely that SHO-related hospitalizations were incorrectly identified as such in the claims data we analyzed, it is possible that SHO events that did not result in a hospitalization were not captured or recorded correctly, particularly if the events did not result in a billable service, did not involve an interaction with a healthcare provider, or an appropriate diagnosis code was not used, which could be quite common [23]. Such miscoded, or underreported, events would likely have been assigned to the comparison group rather than to the appropriate outpatient SHO group, resulting in underestimated monthly costs for the outpatient SHO group and overestimated monthly costs for the comparison group. The degree to which these cost estimates are biased may also vary by patient characteristics if, as Veronese et al. (2016) has shown using Italian data[24], underreporting is related to disease severity. Our definition of the comparison
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Inpatient SHO patients incurred significantly (51% and 47%) higher costs over the follow-up period than patients in the two other groups ($53,353 vs. $35,414 and $36,230, respectively), corresponding to average monthly costs of $2,084 for inpatient SHO patients and $1,313 and $1,372 for patients in the other groups. Differences in total monthly healthcare costs are primarily due to differences in inpatient ($633 vs. $219 and $272, respectively) and outpatient ($845 vs. $500 and $545, respectively) services. Patients in the inpatient SHO group incurred higher diabetes- and nondiabetes-related non-medication costs ($589 and $985) than patients in the other groups ($329 and $447, $272 and $588, respectively). Medication costs were similar across the three groups, whether for insulin or all other types of drugs ($283–$311 and $228–$237, respectively). Complete results are presented in Table 3.
t
ip
cr
us
ce
pt e
d
M
an
Our results, that patients incur higher healthcare costs after being hospitalized for SHO, are important because they imply that the impacts of SHO-related hospitalizations are not limited to the acute period during which patients are hospitalized, but persist for some time as well, and that health services researchers should be aware of these spill-over effects when designing future studies of SHO. Specifically, the longer-term burden of SHO-related hospitalization is primarily due to greater inpatient and outpatient care costs, even after adjusting, through our use of the IPTW method, for baseline differences in comorbid conditions and the use of healthcare services across the groups studied. These findings can also inform economic evaluations of therapies with different SHO profiles. However, more importantly, our findings further imply that interventions designed to prevent or even just delay SHO events requiring hospitalization can be associated with significant economic benefits beyond just reduced SHO hospitalization costs among commercially-insured and Medicare patients.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
group, which excluded patients who experienced SHO and AHR events, may have resulted in artificially low resource use and cost estimates during the follow-up period for comparison patients because those patients were, by definition, less complex. However, we elected to define the comparison group that way to provide an absolute lower bound on estimates of resource use and costs in the absence of SHO. Our data also limited our analyses in other ways. We were unable to differentiate between daytime and nocturnal hypoglycemia nor could we account for several clinical characteristics known to be associated with SHO such as HbA1c, body-mass index, renal function, and lipid levels that are not typically available in claims data. We were also unable to account or adjust for insulin dose or duration of diabetes and our sample exclusion criterion may have resulted in a sample that was healthier than the target population of T1DM patients using a basal-bolus insulin regimen. Furthermore, we were not able to limit index hospitalizations and ED admissions to those cases in which SHO was the verifiable cause of the admission. Therefore, some index admissions may be due to iatrogenic SHO. However, because we used diagnosis data from the first day of the admission, we have limited the extent of this potential bias. Finally, although we used the IPTW method intended to balance the characteristics of the three groups we investigated, we could not account for unmeasured confounders.
4
REFERENCES
1. American Diabetes Association. Economic costs of diabetes in the U.S. in 2012. Diabetes Care. 2013;36:1033-1046.
cr
ip
t
3. Centers for Disease Control and Prevention. National Diabetes Statistics Report: Estimates of Diabetes and Its Burden in the United States, 2014. Atlanta, GA: U.S. Department of Health and Human Services; 2014 [cited 2016 June 24]. Available from: http://www.cdc.gov/diabetes/pubs/statsreport14/national-diabetes-report-web.pdf.
an
us
4. Geller AI, Shehab N, Lovegrove MC, et al. National estimates of insulin-related hypoglycemia and errors leading to emergency department visits and hospitalizations. JAMA Intern Med. 2014;174:678-686.
M
5. Anderbro T, Amsberg S, Adamson U, et al. Fear of hypoglycaemia in adults with Type 1 diabetes. Diabet Med. 2010;27:1151-1158.
pt e
d
6. Fidler C, Elmelund Christensen T, Gillard S. Hypoglycemia: an overview of fear of hypoglycemia, quality-of-life, and impact on costs. J Med Econ. 2011;14:646-655.
ce
7. International Hypoglycaemia Study G. Minimizing Hypoglycemia in Diabetes. Diabetes Care. 2015;38:1583-1591.
8. Wild D, von Maltzahn R, Brohan E, et al. A critical review of the literature on fear of hypoglycemia in diabetes: Implications for diabetes management and patient education. Patient Educ Couns. 2007;68:10-15.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
2. McCoy RG, Zhang Y, Herrin J, et al. Changing trends in type 2 diabetes mellitus treatment intensification, 2002-2010. Am J Manag Care. 2015;21:e288-296.
9. Cryer PE. Hypoglycemia-Associated Autonomic Failure in Diabetes: Maladaptive, Adaptive, or Both? Diabetes. 2015;64:2322-2323.
10. Foos V, Varol N, Curtis BH, et al. Economic impact of severe and non-severe hypoglycemia in patients with Type 1 and Type 2 diabetes in the United States. J Med Econ. 2015;18:420-432.
11. Reviriego J, Gomis R, Maranes JP, et al. Cost of severe hypoglycaemia in patients with type 1 diabetes in Spain and the cost-effectiveness of insulin lispro compared with regular human insulin in preventing severe hypoglycaemia. Int J Clin Pract. 2008;62:1026-1032.
ip
t
13. Goldstein D, Chodick G, Shalev V, et al. Use of Healthcare Services Following Severe Hypoglycemia in Patients with Diabetes: Analysis of Real-World Data. Diabetes Ther. 2016;7:295-308.
us
cr
14. Majumdar SR, Hemmelgarn BR, Lin M, et al. Hypoglycemia associated with hospitalization and adverse events in older people: population-based cohort study. Diabetes Care. 2013;36:35853590.
M
an
15. Elliott L, Fidler C, Ditchfield A, et al. Hypoglycemia Event Rates: A Comparison Between RealWorld Data and Randomized Controlled Trial Populations in Insulin-Treated Diabetes. Diabetes Ther. 2016;7:45-60.
pt e
d
16. Ganz M, Wintfeld N, Li Q, et al. Severe hypoglycemia rates and associated costs among type 2 diabetics starting basal insulin therapy in the United States. Curr Med Res Opin 2014;30:1991-2000.
ce
17. Ginde AA, Blanc PG, Lieberman RM, et al. Validation of ICD-9-CM coding algorithm for improved identification of hypoglycemia visits. BMC Endocr Disord. 2008;8:4.
18. Simeone JC, Quilliam BJ. Predictors of emergency department and outpatient visits for hypoglycemia in type 2 diabetes: an analysis of a large US administrative claims database. Ann Pharmacother. 2012;46:157-168.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
12. Cariou B, Fontaine P, Eschwege E, et al. Frequency and predictors of confirmed hypoglycaemia in type 1 and insulin-treated type 2 diabetes mellitus patients in a real-life setting: results from the DIALOG study. Diabetes Metab. 2015;41:116-125.
19. D’Hoore W, Bouckaert A, Tilquin C. Practical considerations on the use of the Charlson comorbidity index with administrative data bases. J Clin Epidemiol. 1996;49:1429-1433.
20. Schneeweiss S, Maclure M. Use of comorbidity scores for control of confounding in studies using administrative databases. Int J Epidemiol. 2000;29:891-898.
21. Heller SR, Frier BM, Herslov ML, et al. Severe hypoglycaemia in adults with insulin-treated diabetes: impact on healthcare resources. Diabet Med. 2016;33:471-477.
ip
t
23. Jakubczyk M, Rdzanek E, Niewada M, et al. Economic resources consumption structure in severe hypoglycemia episodes: a systematic review and meta-analysis. Expert Rev Pharmacoecon Outcomes Res. 2015;15:813-822.
ce
pt e
d
M
an
us
cr
24. Veronese G, Marchesini G, Forlani G, et al. Are severe hypoglycemic episodes in diabetes correctly identified by administrative data? Evidence of underreporting from the HYPOTHESIS study. Acta diabetologica. 2016;53:677-680.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
22. Lipska KJ, Ross JS, Wang Y, et al. National trends in US hospital admissions for hyperglycemia and hypoglycemia among Medicare beneficiaries, 1999 to 2011. JAMA Intern Med. 2014;174:11161124.
Table 1. Baseline Demographic and Clinical Characteristics of Patients with Type 1 Diabetes Mellitus using Basal-Bolus Insulin Therapy (N=8,734) Characteristics
Inpatient SHO (N=364)
Mean Duration of Follow-up Period (SD), days
654.3 (480.2)
546.2 (424.1)
824.4 (521.8)
47.0 (17.2)
44.6 (16.7)
45.6 (16.3)
18–34
26.6
32.8
28.7
35–44
17.3
16.8
19.3
45–54
21.7
22.9
21.4
55–64
21.2
15.8
≥65
13.2
11.6
Mean (SD) Age, years
Outpatient SHO (N=802)
Comparison (N=7,568)
62.6
Female
37.4
Geographic Region, %
t ip
19.2 11.4
62.5
62.5
37.5
37.5
18.8
16.6
25.3
28.6
us
Male
cr
Gender, %
22.0
Midwest
25.8
South
38.5
35.9
37.1
West
12.9
19.2
17.3
0.8
0.7
0.4
87.1
90.4
89.7
12.9
9.6
10.3
31.9
32.4
27.3
23.1
23.2
20.4
7.4
4.0
3.6
Congestive heart failure
4.1
1.7
1.7
Malignancy (excl. malignant neoplasm of skin)
3.3
2.2
2.6
Cerebrovascular disease
2.7
1.6
1.4
Peripheral vascular disease
2.7
1.4
1.6
Chronic pulmonary disease
1.9
3.9
3.4
Liver disease
1.9
0.9
0.8
Rheumatic disease
1.6
1.5
0.9
Hemiplegia or paraplegia
0.5
0.1
0.1
HIV/AIDS
0.3
0.0
0.1
Metastatic solid tumor
0.0
0.2
0.3
Depression
6.3
5.6
4.0
Anxiety
2.5
2.4
1.4
Dementia
0.5
0.0
0.1
an
Northeast
M
Unknown Insurance Type, %
d
Commercial Physical Comorbidities, % Dyslipidemia Hypertension
ce
Renal disease
pt e
Medicare
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Age, years, %
Mental Health Comorbidities, %
Inpatient SHO (N=364)
Characteristics
Outpatient SHO (N=802)
Comparison (N=7,568)
Microvascular Complications, % Diabetic retinopathy
7.4
5.4
6.1
Diabetic neuropathy
5.5
6.4
5.1
Diabetic nephropathy
4.4
4.7
3.2
Coronary artery disease
4.4
3.4
4.5
Peripheral arterial disease
1.9
2.1
1.7
Stroke
1.4
0.1
0.4
2.5 (1.0)
2.3 (0.9)
2.3 (0.9)
0.0
0.0
2
78.6
86.8
3
7.1
≥4
14.3
t 0.0
87.5 4.4
8.9
8.1
2.6
3.2
us
4.4
Insulin-naïve, %
2.7
Mean (SD) Follow-up, Days
654.3 (480.2)
546.2 (424.1)
824.4 (521.8)
an
0, 1
cr
Charlson Comorbidity Index Score, %
ip
Mean (SD) Charlson Comorbidity Index Score
2,447 (19,348)
418 (3,416)
434 (5,236)
354 (2,503)
174 (1,045)
91 (648)
Mean (SD) Baseline Healthcare Costs, $ Inpatient hospitalizations
M
Emergency department admissions Outpatient services Medications
d
Total
2,881 (12,409)
1,985 (5,845)
2,151 (10,300)
3,222 (7,293)
2,810 (2,281)
2,553 (2,711)
8,904 (25,768)
5,386 (7,813)
5,229 (13,512)
ce
pt e
Abbreviations: AIDS, acquired immune deficiency syndrome; HIV, Human immunodeficiency virus; SD, standard deviation; SHO, severe hypoglycemia
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Macrovascular Complications, %
Table 2. Severe Hypoglycemia-Related Hospitalizations and Emergency Department Admissions, Costs, and Risk of Subsequent Hospitalization of Patients with Type 1 Diabetes Mellitus using Basal-Bolus Insulin Therapy (N=364) Outcomes Hospitalized for SHO, % Without Prior ED Admission
21.2
With Prior ED Admission
78.8 1.7 (2.9)
Patients without Prior ED Admission
3.8 (5.6)
ip
t
All Patients Patients with Prior ED Admission
1.2 (0.9)
Mean (SD) Cost, $
3,551 (16,740)
cr
All Patients
12,726 (33,389)
Patients with Prior ED Admission Subsequent Hospitalization with and without Prior ED Admission, % None
an
Second SHO
us
Patients without Prior ED Admission
AHR
1,090 (5,542) 57.1 32.1 10.7
ce
pt e
d
M
Abbreviations: AHR, acute hyperglycemia; ED, emergency department; SD, standard deviation; SHO, severe hypoglycemia. Subsequent hospitalizations may not sum to 100% due to rounding.
Ac
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:04 13 October 2017
Mean (SD) Length of Stay, Days
Table 3. Follow-up Healthcare Utilization and Adjusted Monthly Costs of Patients with Type 1 Diabetes Mellitus using Basal-Bolus Insulin Therapy (N=8,734): Inverse Probability of Treatment Weighted Results P-Value Inpatient SHO (N=364)
Outcomes
8,724
8,701
Comparison (N=7,568)
Emergency department admissions
53.0
35.0
29.4
Outpatient services
98.1
97.4
96.7
100.0
99.9
100.0
Medications Mean (SD) Monthly Costs, $
Outpatient services Medications
Other
0.58
1.00
1.00
