Abstract
Aims:
Among adults with insulin- and/or secretagogue-treated diabetes in the United States, very little is known about the real-world descriptive epidemiology of iatrogenic severe (level 3) hypoglycaemia. Addressing this gap, we collected primary, longitudinal data to quantify the absolute frequency of events as well as incidence rates and proportions.
Materials and Methods:
iNPHORM is a US-wide, 12-month ambidirectional panel survey (2020-2021). Adults with type 1 diabetes mellitus (T1DM) or insulin- and/or secretagogue-treated type 2 diabetes mellitus (T2DM) were recruited from a probability-based internet panel. Participants completing ≥1 follow-up questionnaire(s) were analysed.
Results:
Among 978 respondents [T1DM 17%; mean age 51 (SD 14.3) years; male: 49.6%], 63% of level 3 events were treated outside the health care system (e.g. by family/friend/colleague), and <5% required hospitalization. Following the 12-month prospective period, one-third of individuals reported ≥1 event(s) [T1DM 44.2% (95% CI 36.8%-51.8%); T2DM 30.8% (95% CI 28.7%-35.1%), p = .0404, α = 0.0007]; and the incidence rate was 5.01 (95% CI 4.15-6.05) events per person-year (EPPY) [T1DM 3.57 (95% CI 2.49-5.11) EPPY; T2DM 5.29 (95% CI 4.26-6.57) EPPY, p = .1352, α = 0.0007]. Level 3 hypoglycaemia requiring non-transport emergency medical services was more common in T2DM than T1DM (p < .0001, α = 0.0016). In total, >90% of events were experienced by <15% of participants.
Conclusions:
iNPHORM is one of the first long-term, prospective US-based investigations on level 3 hypoglycaemia epidemiology. Our results underscore the importance of participant-reported data to ascertain its burden. Events were alarmingly frequent, irrespective of diabetes type, and concentrated in a small subsample.
Keywords: hypoglycaemia, insulin secretagogue, insulin therapy, real-world evidence, type 1 diabetes, type 2 diabetes
1 ∣. INTRODUCTION
Insulin- or secretagogue-induced severe hypoglycaemia (SH) is a major cause of morbidity and mortality in people with diabetes.1 The American Diabetes Association (ADA) defines SH as a ‘level 3’ low blood glucose (BG) concentration, regardless of value, requiring professional or non-professional aid for recovery.2
In the United States, the human and economic burden of iatrogenic level 3 hypoglycaemia is conceivably profound. At present, an estimated 23 million adults with diabetes are on insulin and/or secretagogues, including all those with type 1 diabetes mellitus (T1DM) and roughly 75% with type 2 diabetes mellitus (T2DM).3-5 Data from Canada, Europe, the United Kingdom and Asia suggest that iatrogenic level 3 hypoglycaemia affects about one-third of adults with diabetes annually, and occurs as often as 2.5-5.8 events per person-year (EPPY).6 Nevertheless, its frequency in the United States remains virtually unquantified.
Current SH surveillance relies mainly on claims or health records,7-9 which fail to capture the 90%-95%10,11 of events treated outside the health care system (e.g. by family/friends/colleagues). Moreover, of the few US studies assessing the total spectrum of level 3 hypoglycaemia, most are cross-sectional and poorly generalizable. Enhanced real-world data on level 3 hypoglycaemia occurrence in the United States are needed to optimize clinical event prevention, particularly as therapies conferring little to no hypoglycaemia risk become more accessible. Furthermore, such insight is imperative to guide public health research, planning, monitoring and policies.
In this study, we analyse data from the Investigating Novel Predictions of Hypoglycemia Occurrence Using Real-world Models (iNPHORM) ambidirectional (retrospective/prospective) panel survey to quantify the absolute frequency and crude incidence of iatrogenic level 3 hypoglycaemia in Americans with diabetes. Framing our results within the wider literature, we characterize events by the type of recovery aid required, distinguishing between professional and non-professional aid. We also compare frequency estimates between retrospective and prospective observation periods. The following article complies with the Strengthening the Reporting of Observational Studies in Epidemiology checklist.12
2 ∣. MATERIALS AND METHODS
2.1 ∣. Study design
Complete details on iNPHORM are published in the study protocol.13 From 2020 to 2021, we carried out an ambidirectional panel survey of a general US-wide cohort of adults with T1DM or insulin- and/or secretagogue-treated T2DM. Online questionnaires elicited self-reported data on level 3 hypoglycaemia frequency and participant characteristics, retrospectively at baseline using a 12-month lookback, and prospectively across 12-monthly follow-ups.
Questionnaires are a preferred method for ascertaining level 3 hypoglycaemia epidemiology: (a) most events are not clinically documented; (b) the lack of a standard BG threshold for SH diminishes the relevance of glucometer data; and (c) deliberate patient non-disclosure14,15 and provider under-recognition16,17 occlude accurate event assessment at point-of-care. Research shows that online questionnaires, which demand little personally identifiable data, are more effective than postal or telephone modes at motivating participation and honest reporting.18,19 Identifiable versus anonymous reporting may underestimate SH frequency by 50%-65%.14,20
2.2 ∣. Participants
We recruited participants from a well-established internet panel (N ≈ 68 000 with diabetes) managed by Ipsos Interactive Services (IIS), a global leader in real-world survey conduct.21 Probability-based sampling (e.g. random address-based techniques); multisource recruitment (e.g. via mail, phone and online); quality monitoring; and a patented weighted selection methodology fortified the panel's geodemographic and household representativeness, including of hard-to-reach (e.g. young adults) and minority subgroups.
Internet panellists could enrol in iNPHORM if they were 18-90 years old and living in the United States (≥1 year) with a self-reported diagnosis of T1DM or T2DM (≥1 year) and taking insulin, secretagogues, or both (≥1 year). Panellists were not permitted to enrol if they were in a concurrent trial or pregnant at screening or within the past year. Our target sample size of N = 1250 was determined by the primary objective of iNPHORM.13,22
2.3 ∣. Recruitment and data collection
A generally worded study invitation was emailed to a randomly selected subset of the internet panel. To enrol, panellists had to pass a screening questionnaire, provide consent, submit a baseline questionnaire and register to receive 12-monthly follow-ups. Recruitment continued until we conveniently sampled 1250 participants (i.e. subpanel A). Individuals were hosted and monitored by IIS.
Members of subpanel A not completing the first follow-up were withdrawn and systematically refreshed with new participants (i.e. subpanel B). Those in subpanel A completing the first follow-up, plus all those in subpanel B, comprised the iNPHORM longitudinal panel. Quota sampling ensured that ≥10% of the iNPHORM longitudinal panel was female; ≥10% male; ≥10% had T1DM; and that ≥5% were ≥75 years old. Additionally, insulin without secretagogues, secretagogues without insulin, and insulin plus secretagogues had to each account for ≥10% of the T2DM subgroup.
The iNPHORM longitudinal panel was monitored for 12 months (subpanel A: February 2020-January 2021; subpanel B: April 2020-March 2021). Each month, participants received a unique link to a follow-up questionnaire that was active for 7 days. Responses were stored synchronously on the IIS platform and tethered by a distinct, random ID assigned at enrolment. Personalized precontacts, notifications, reminders and token cash incentives were used to promote retention.
2.4 ∣. Instruments and measures
AR-L, BLR and SBH devised all iNPHORM questionnaires with reference to pre-existing surveys, particularly related to hypoglycaemia event capture, and established survey design principles to reduce respondent fatigue.23 Questionnaires were developed in English and programmed for diverse internet-equipped devices (e.g. computers, phones and tablets); they were piloted before roll-out. Various quality assurance methods were adopted during questionnaire development to reinforce response integrity, including in-built logic checks (e.g. use of single- or multi-response coding) and calibration questions (e.g. to detect straight-lining, verify item comprehension and avert nonsensical free text). To bypass an item, we required that participants type the words ‘OPT OUT’ in a pop-up response box, thus, helping differentiate intentional non-response from inadvertent omissions/straight-lining.
The screener captured information on diabetes type, medication regimen, US residency and pregnancy status, concurrent trial involvement, as well as sex assigned at birth; responses were retained for consenting individuals. At baseline and over follow-up, we examined numerous anthropometric, sociodemographic and lifestyle/behavioural variables, often using US Census Bureau and Centers for Disease Control and Prevention measures. We also assessed diabetes-, hypoglycaemia- and general health-related factors. Follow-up questionnaires measured changes in clinical variables.
2.5 ∣. Outcome measure
Level 3 hypoglycaemia was defined in concordance with the 2019 ADA guidelines as a low BG event causing altered mental and/or physical status requiring professional or non-professional aid for recovery (Supplement 1 in Data S1).2 Event frequency was assessed at baseline (1-year lookback) and at each follow-up (‘since the last time an iNPHORM survey was completed’).
For all events, we elicited information on the type of recovery aid required (i.e. ‘Treated outside the care system by a non-healthcare provider [HCP]’; ‘Treated by non-transport emergency medical services [EMS]’, ‘Treated in hospital without admission’ and ‘Treated in hospital with admission’). As spontaneous recovery is clinically plausible, we also provided the response option, ‘I was physically unable to treat my severe hypoglycaemia myself but did not receive assistance to recover because there was no one around to help. I recovered on my own’. Alternatively, to pre-empt false responding, individuals could select ‘Other’ or ‘I don't know’.
2.6 ∣. Statistical analysis
We analysed respondents with ≥1 completed follow-up(s). Retention was calculated as the percentage of completed follow-ups. We report sample characteristics as frequencies and percentages, means and SDs, or medians and interquartile ranges (IQRs), as appropriate. Characteristics were compared between those with (a) ≥1 level 3 event(s) at baseline versus study end, and (b) 0 versus ≥1 level 3 event(s) at study end, using the chi-squared test for frequencies, z-test for means and Wilcoxon rank-sum test for medians.
We tallied the number of level 3 hypoglycaemia events overall, by diabetes type and by type of required recovery aid. The population denominator comprised participants with a non-zero risk of iatrogenic hypoglycaemia. Thus, any prospective period where a person reported not using either insulin or secretagogues was excluded from analysis. We also excluded data for prospective periods of ineligibility [i.e. concurrent trial involvement, pregnancy (right censored), and/or non-US residence (right censored)].
We report incidence proportions (IPs) [% with ≥1 level 3 event(s)] alongside Wilson's confidence intervals (CIs), and incidence rates (IRs) (EPPY) alongside negative binomial CIs. IPs at baseline reflect the full previous year; whereas, at study end, they reflect the length of follow-up (<1 year if incomplete). Contrastingly, IRs at baseline and study end were both annualized. Negative binomial regression computed IRs, offset for zero risk, ineligible and unobserved periods. We report incidences overall and by type of recovery aid required. Relative IP and IR differences are expressed as cumulative incidence ratios and IR ratios (IRRs).
Significance tests were based on a two-sided Bonferroni-adjusted α-level for a family-wise error rate of α = 0.05.
2.7 ∣. Ethical considerations
Before recruitment, we obtained ethics approval from the Western University Health Sciences Research Ethics Board (HSREB) (Project ID: 112986; 17 December 2019) and registered iNPHORM with ClinicalTrials.gov (NCT04219514; 7 January 2020). Individuals consented to enrol and could withdraw at any time. Personally identifiable data (e.g. email addresses) were collected strictly to monitor participants over follow-up; all questionnaires were completed anonymously. Only deidentified data were transferred by IIS to Western University.
3 ∣. RESULTS
3.1 ∣. Characterizing the iNPHORM cohort
3.1.1 ∣. iNPHORM longitudinal panel
See Supplement 2 in Data S1 for a flow diagram of recruitment and participation. Of the iNPHORM longitudinal panel (N = 1206), 978 individuals completed ≥1 follow-up(s) (Tables 1 and 2). The retention rate was 86.2% (85.5% completed eight follow-ups with 66.1% completing all 12) and <15% were lost to follow-up (Supplement 3 in Data S1). The mean prospective period was 9.62 (SD 3.15) months [T1DM 10.15 (SD 2.84) months; T2DM 9.51 (SD 3.20) months; p = .02].
TABLE 1.
Baseline characteristics of the iNPHORM longitudinal panel: anthropometric and sociodemographic characteristics
| Characteristics | Overall (n = 978) | T1DM (n = 163) | T2DM (n = 815) |
|---|---|---|---|
| Age, years | |||
| Mean (SD) | 50.97 (14.29) | 44.61 (13.82) | 52.24 (14.05) |
| Median (IQR) | 52 (24) | 44 (20) | 53 (23) |
| Missing/unknown | 0 | 0 | 0 |
| Sex assigned at birth, n (%) | |||
| Male | 485 (49.59) | 56 (34.36) | 429 (52.64) |
| Female | 493 (50.41) | 107 (65.64) | 386 (47.36) |
| Missing/unknown, n (%) | 0 | 0 | 0 |
| BMI, kg/m2 | |||
| Median (IQR) | 30.35 (12.05) | 26.34 (7.13) | 31.45 (12.42) |
| Missing/unknown, n (%) | 6 (0.61) | 0 | 6 (0.74) |
| Marital status, n (%) | |||
| Married | 615 (62.88) | 94 (57.67) | 521 (63.93) |
| Divorced, separated, widowed | 162 (16.56) | 25 (15.34) | 137 (16.81) |
| Never married | 200 (20.45) | 44 (26.99) | 156 (19.14) |
| Missing/unknown, n (%) | 1 (0.10) | 0 | 1 (0.12) |
| Race, n (%) | |||
| White alone | 776 (79.35) | 148 (90.80) | 628 (77.06) |
| Part-White multiracial | 37 (3.78) | 3 (1.84) | 34 (4.17) |
| Non-White | 143 (14.62) | 10 (6.13) | 133 (16.32) |
| Missing/unknown, n (%) | 22 (2.25) | 2 (1.23) | 20 (2.45) |
| Highest level of education, n (%) | |||
| High school, some high school, or Grade 8 | 170 (17.38) | 30 (18.40) | 140 (17.18) |
| College degree or some college | 627 (64.11) | 105 (64.42) | 522 (64.05) |
| Degree beyond first college degree | 181 (18.51) | 28 (17.18) | 153 (18.77) |
| Missing/unknown, n (%) | 0 | 0 | 0 |
| Employment status, n (%) | |||
| Full-time | 427 (43.66) | 72 (44.17) | 355 (43.56) |
| Part-time | 81 (8.28) | 21 (12.88) | 60 (7.36) |
| Unemployed, retired, or a student | 470 (48.06) | 70 (42.94) | 400 (49.08) |
| Missing/unknown, n (%) | 0 | 0 | 0 |
| Gross annual household income, n (%) | |||
| <$25 000 | 167 (17.08) | 22 (13.50) | 145 (17.79) |
| $25 000-$54 999 | 266 (27.20) | 39 (23.93) | 227 (27.85) |
| $55 000-$84 999 | 211 (21.57) | 53 (32.52) | 158 (19.39) |
| $85 000-$114 999 | 149 (15.24) | 24 (14.72) | 125 (15.34) |
| $115 000-$144 999 | 64 (6.54) | 7 (4.29) | 57 (6.99) |
| ≥$145 000 | 112 (11.45) | 14 (8.59) | 98 (12.02) |
| Missing/unknown, n (%) | 9 (0.92) | 4 (2.45) | 5 (0.61) |
| Insurance, n (%) | |||
| Private insurance plan | 420 (42.94) | 88 (53.99) | 332 (40.74) |
| Government-assistance plan | 319 (32.62) | 47 (28.83) | 272 (33.37) |
| Multiple insurance plans and other insurance plans | 221 (22.60) | 23 (14.11) | 198 (24.29) |
| Out-of-pocket (i.e. no insurance coverage) | 18 (1.84) | 5 (3.07) | 13 (1.60) |
| Missing/unknown, n (%) | 0 | 0 | 0 |
Abbreviations: BMI, body mass index; IQR, interquartile range; SD, standard deviation; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
TABLE 2.
Baseline characteristics of the iNPHORM longitudinal panel: clinical characteristics
| Characteristics | Overall (n = 978) | T1DM (n = 163) | T2DM (n = 815) |
|---|---|---|---|
| Duration of diabetes; years | |||
| Median (IQR) | 12 (14) | 26 (20) | 11 (13) |
| Missing/unknown, n (%) | 9 (0.92) | 1 (0.61) | 8 (0.98) |
| Medication regimen, n (%) | |||
| Insulin without secretagogues | 475 (48.57) | 163 (100.00) | 312 (38.28) |
| Secretagogues without insulin | 312 (31.90) | 0 | 312 (38.28) |
| Insulin with secretagogues | 191 (19.53) | 0 | 191 (23.44) |
| Missing/unknown, n (%) | 0 | 0 | 0 |
| Type of insulin, n (%) | |||
| Basal (without bolus) | 163 (16.67) | 8 (4.91) | 155 (19.02) |
| Bolus (without basal) | 231 (23.62) | 74 (45.4) | 157 (19.26) |
| Basal and bolus combination (excluding premixed) | 190 (19.43) | 73 (44.79) | 117 (14.36) |
| Basal and bolus combination (including premixed) | 82 (8.38) | 8 (4.91) | 74 (9.08) |
| Other | 3 (0.31) | 0 | 3 (0.37) |
| Not using insulin | 312 (31.90) | 0 | 312 (38.28) |
| Missing/unknown, n (%) | 0 | 0 | 0 |
| Duration of insulin use, yearsa | |||
| Median (IQR) | 6.67 (12.08) | 25.58 (22.08) | 5.00 (7.50) |
| Missing/unknown, n (%) | 3 (0.003) | 1 (0.006) | 2 (0.002) |
| Duration of secretagogue use, yearsa | |||
| Median (IQR) | 4.50 (5.75) | - | 4.50 (5.75) |
| Missing/unknown, n (%) | 10 (0.01) | - | 10 (0.01) |
| Most recent HbA1c value, %; n (%) | |||
| ≤7 | 323 (33.03) | 58 (35.58) | 265 (32.52) |
| 7.1-8 | 337 (34.46) | 60 (36.81) | 277 (33.99) |
| 8.1-9 | 161 (16.46) | 23 (14.11) | 138 (16.93) |
| ≥9.1 | 95 (9.71) | 20 (12.27) | 75 (9.20) |
| Missing/unknown, n (%) | 62 (6.34) | 2 (1.23) | 60 (7.36) |
| Frequency of hypoglycaemia awarenessb, n (%) | |||
| Never or rarely | 66 (6.75) | 6 (3.68) | 60 (7.36) |
| Sometimes or often | 583 (59.61) | 110 (67.48) | 473 (58.04) |
| Always | 226 (23.11) | 47 (28.82) | 179 (21.96) |
| Missing/unknown, n (%) | 103 (10.53) | 0 | 103 (12.64) |
| Number of diabetes complicationsc, n (%) | |||
| 0 | 415 (42.43) | 52 (31.90) | 363 (44.54) |
| 1 | 253 (25.87) | 42 (25.77) | 211 (25.89) |
| 2 | 104 (10.63) | 24 (14.72) | 80 (9.82) |
| 3 | 64 (6.54) | 19 (11.66) | 45 (5.52) |
| 4 | 28 (2.86) | 7 (4.29) | 21 (2.58) |
| ≥5 | 46 (4.70) | 13 (7.98) | 33 (4.05) |
| Missing/unknown, n (%) | 68 (6.95) | 6 (3.68) | 62 (7.61) |
| Chronic kidney disease, n (%) | |||
| No | 862 (88.14) | 146 (89.57) | 716 (87.85) |
| Yes | 107 (10.94) | 17 (10.43) | 90 (11.04) |
| Missing/unknown, n (%) | 9 (0.92) | 0 | 9 (1.10) |
| Number of other comorbidities (excluding chronic kidney disease)d, n (%) | |||
| 0 | 171 (17.48) | 42 (25.77) | 129 (15.83) |
| 1 | 174 (17.79) | 33 (20.25) | 141 (17.30) |
| 2 | 186 (19.02) | 26 (15.95) | 160 (19.63) |
| 3 | 138 (14.11) | 20 (12.27) | 118 (14.48) |
| 4 | 122 (12.47) | 20 (12.27) | 102 (12.52) |
| ≥5 | 132 (13.5) | 15 (9.20) | 117 (14.36) |
| Missing/unknown, n (%) | 55 (5.62) | 7 (4.29) | 48 (5.89) |
| Method of blood glucose monitoring, n (%) | |||
| None | 52 (5.32) | 0 | 52 (6.38) |
| SMBG use | 713 (72.90) | 81 (49.69) | 632 (77.55) |
| rt-C/FGM use | 53 (5.42) | 9 (5.52) | 44 (5.40) |
| SMBG and rt-C/FGM use | 155 (15.85) | 71 (43.56) | 84 (10.31) |
| Missing/unknown, n (%) | 5 (0.51) | 2 (1.23) | 3 (0.37) |
Abbreviations: HbA1c, glycated haemoglobin; IQR, interquartile range; Not applicable; rt-C/FGM, real-time continuous or flash glucose monitoring; SMBG, self-monitoring blood glucose; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Among those treated.
Measured using a modified item from the Clarke method70; participants reported to what extent they can tell by their symptoms that their blood glucose is low (never, rarely, sometimes, often, or always).
Diabetes complications included amputation, diabetic ketoacidosis, foot damage, gastroparesis, hyperosmolar coma, neuropathy and retinopathy.
Other comorbidities included bone, joint, or muscle problems; cancer; cardiovascular disease; chronic liver failure; eating disorders; gastrointestinal disease; HIV/AIDS; hypertension; mental health conditions; neurological disorders; and stroke.
3.2 ∣. Subset reporting ≥1 level 3 hypoglycaemia event(s)
Almost half [461 of 978 (47.14%)] reported ≥1 level 3 event(s) at baseline or study end (Supplement 4 in Data S1). These individuals were 45 (IQR: 22) years old, 49.89% (230 of 461) female, and 76.36% (352 of 461) white; most had some college education [381 of 461 (82.65%)] and/or health insurance [455 of 461 (98.70%)]. Compared with T1DM [21.48% (99 of 461)] participants, those with T2DM [78.52% (362 of 461)] were older [46 (IQR: 23) vs. 42 (IQR: 20) years, respectively] with a shorter diabetes duration [10 (IQR: 12) vs. 27 (IQR: 22) years, respectively]. All T1DM respondents were on insulin without secretagogues; among those with T2DM, 23.48% (85 of 362) were on insulin without secretagogues, 24.31% (88 of 362) on secretagogues without insulin, and 52.21% (189 of 362) on insulin plus secretagogues. Overall, ~7% (34 of 461) indicated they rarely or never could detect symptoms of hypoglycaemia. Supplement 5 in Data S1 compares sample characteristics for participants with ≥ 1 versus 0 level 3 event(s) at study end.
3.3 ∣. Quantifying level 3 hypoglycaemia
3.3.1 ∣. Absolute frequency of level 3 hypoglycaemia and type of aid required for recovery
The distribution of level 3 hypoglycaemia was right-skewed (Supplement 6 in Data S1). At baseline, 110 (11%) participants had ≥5 events, or 84% of all reported events (T1DM 13.5% had 89% of events; T2DM 11% had 82% of events). By study end, 128I (13%) had ≥5 events, or 91% of all reported events (T1DM 12% had 78% of events; T2DM 13% had 92% of events).
See Table 3 for a breakdown of events by type of recovery aid. Following 12 months of follow-up, over 60% (2523 of 4007) of level 3 events were treated outside the health care system by a non-HCP [T1DM 309 of 443 (69.8%); T2DM 2214 of 3564 (62.1%), p = .0017, α = 0.0016], while 9.2% (367 of 4007) resulted in hospital services with 4.6% (183 of 4007) requiring admission [T1DM 7 of 443 (1.6%) vs. T2DM 176 of 3564 (4.9%), p = .0014, α = 0.0016] (Table 3). The percentage of events resulting in no external aid was 50% higher in T2DM [215 of 3564 (6.0%)] than T1DM [17 of 443 (3.8%)], but non-significant (p = .0621, α = 0.0016). Type of recovery aid was unspecified for 10% (388 of 4007) of events [T2DM 296 of 3564 (8.31%) vs. T1DM 92 of 443 (20.8%), p < .0001, α = 0.0016].
TABLE 3.
Frequencies of level 3 hypoglycaemia by the type of aid required for recovery (overall and by diabetes type) at baselinea and study endb
| Overall (n = 978) |
T1DM (n = 163) |
T2DM (n = 815) |
T1DM versus T2DM |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Baseline | Study end | p-Valuec | Baseline | Study end | p-Valuec | Baseline | Study end | p-Valuec | Baseline p-Valuec |
Study end p-Valuec |
| Total number of level 3 hypoglycaemia events, n | ||||||||||
| 2354 | 4007 | - | 750 | 443 | - | 1604 | 3564 | - | - | - |
| Treated outside the care system by a non-HCP (e.g. family/friend/colleague), n (%) | ||||||||||
| 1207 (51.27) | 2523 (62.96) | <.0001d | 432 (57.60) | 309 (69.75) | <.0001d | 775 (48.32) | 2214 (62.12) | <.0001d | <.0001d | .0017 |
| Treated by non-transport emergency medical services, n (%) | ||||||||||
| 234 (9.94) | 486 (12.13) | .0078 | 11 (1.47) | 12 (2.71) | .1317 | 223 (13.90) | 474 (13.30) | .5571 | <.0001d | <.0001d |
| Treated in hospital without admission, n (%) | ||||||||||
| 130 (5.52) | 184 (4.59) | .0981 | 2 (0.27) | 5 (1.13) | .0596 | 128 (7.98) | 179 (5.02) | <.0001d | <.0001d | .0002d |
| Treated in hospital with admission, n (%) | ||||||||||
| 61 (2.59) | 183 (4.57) | .0001d | 4 (0.53) | 7 (1.58) | .068 | 57 (3.55) | 176 (4.94) | .0265 | <.0001d | .0014d |
| No external aid (i.e. spontaneously recovered), n (%) | ||||||||||
| 150 (6.37) | 232 (5.79) | .3453 | 28 (3.73) | 17 (3.84) | .9273 | 122 (7.61) | 215 (6.03) | .0340 | .0003 | .0621 |
| Other/unknown, n (%) | ||||||||||
| 572 (24.30) | 388 (9.68) | <.0001d | 273 (36.40) | 92 (20.77) | <.0001d | 299 (18.64) | 296 (8.31) | <.0001d | <.0001d | <.0001d |
Abbreviations: HCP, health care provider; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Based on a 1-year lookback.
Based on a prospective follow-up period of 12 months.
z-tests were used to compare proportions.
Significant based on a Bonferroni-adjusted α = 0.0016, giving a family-wise error rate of α = 0.05.
Among T2DM respondents, the number of events requiring non-transport EMS was 4.9 times that reported by T1DM respondents (p < .0001, α = 0.0016). Participants with T2DM versus T1DM also reported 4.4 times the number of events requiring hospital care (no admission) (p = .0002, α = 0.0016).
3.4 ∣. Crude incidence of level 3 hypoglycaemia
Figures 1 and 2 summarize IPs and IRs, respectively, at baseline and study end; estimates are presented overall, by type of recovery aid, and by diabetes type (see Supplement 7 for a tabular format in Data S1). At study end, 33.8% (95% CI 30.9%-36.9%) of participants had ≥1 event(s). No significant differences in IPs were observed by diabetes type at baseline and study end. Baseline IPs by diabetes type and type of recovery aid were statistically on par with corresponding values at study end.
FIGURE 1.
Incidence proportions of level 3 hypoglycaemia (overall and by diabetes type) at baseline† and study end‡ *Significant based on a Bonferroni-adjusted α = 0.0007, giving a family-wise error rate of α = 0.05. †Based on a 1-year lookback. ‡Based on an annualization of a prospective follow-up period of 12 months. CI, confidence interval; ED, emergency department; EMS, emergency medical services; HCP, health care providers; SH, severe hypoglycaemia; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
FIGURE 2.
Incidence rates of level 3 hypoglycaemia (overall and by diabetes type) at baseline† and study end‡. *Significant based on a Bonferroni-adjusted α = 0.0007, giving a family-wise error rate of α = 0.05. †Based on a 1-year lookback. ‡Based on a prospective follow-up period of 12 months. CI, confidence interval; ED, emergency department; EMS, emergency medical services; HCP, health care providers; SH, severe hypoglycaemia; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
The total IR was 5.01 (95% CI 4.15-6.05) and 2.41 (95% CI 2.01-2.88) EPPY at study end and baseline, respectively [IRR 2.08 (95% CI 1.61-2.70), p <.0001, α = 0.0007]. Overall, IRs at study end versus baseline were statistically significantly higher for events requiring non-HCP care [IRR 2.40 (95% CI 1.74-3.32), p <.0001, α = 0.0007]; non-transport EMS [IRR 2.57 (95% CI 1.60-4.13), p = .0001, α = 0.0007]; and hospitalization [IRR 3.83 (95% CI 1.98-7.43), p <.0001, α = 0.0007]. For T2DM respondents, specifically, IRs were significantly higher at study end versus baseline, overall [IRR 2.69 (95% CI 2.00-3.61), p <.0001, α = 0.0007] and for events requiring non-HCP care [IRR 3.26 (95% CI 2.27-4.68), p <.0001, α = 0.0007]; non-transport EMS [IRR 2.64 (95% CI 1.62-4.30), p <.0001, α = 0.0007]; and hospitalization [IRR 3.93 (95% CI 1.97-7.86), p = .0001, α = 0.0007]. For T1DM, we found no significant IRRs between baseline and study end.
The IR for total level 3 hypoglycaemia was significantly greater in T1DM than T2DM [IRR 2.34 (95% CI 1.47-3.73), p = .0004, α = 0.0007] at baseline, but not significantly lower by study end [IRR 0.68 (95% CI 0.41-1.13), p = .1352, α = 0.0007]. By study end, more people with T2DM than T1DM reported events treated by non-transport EMS [IRR 8.57 (95% CI 2.93-25.00), p <.0001, α = 0.0007].
4 ∣. DISCUSSION
iNPHORM is the first US prospective epidemiologic investigation to assess guideline-defined level 3 hypoglycaemia in the real world. We show that a small subset of events requires professional health care for recovery, and that incidences in T1DM and T2DM remain high, despite treatment and monitoring advances. Our data also reiterate the right-skewed distribution of SH documented in earlier studies.24-26
Our results draw attention to the burden of community-treated SH that is otherwise uncaptured by records-based surveillance. Aligning with international research,27-32 only 10% of level 3 events resulted in hospital care, of which <5% necessitated admission. Notably, three-quarters of events were treated outside the health care system (>10% by non-transport EMS and >60% by non-professional aid). That health care-related events were significantly more common among T2DM than T1DM respondents, echoing Donnelly et al.,33 emphasizes the need for improved preventive action in this numerically larger disease group.
Of note, 4% and 6% of people with T1DM and T2DM, respectively, experienced an event that resolved spontaneously. Whether the guideline definition of level 3 hypoglycaemia needs to be refined warrants further research: in its current form, severe events where individuals recovered without third-party aid risk being misclassified as non-severe.
Akin to previous research,33-39 approximately one-third of respondents reported ≥ 1 level 3 event(s) at baseline. This IP was equable to our prospective 12-month estimate, corroborating evidence on the durability of annual SH recall of one or multiple events.40 Baseline IPs were significantly higher in T1DM than T2DM but differences were attenuated by study end (p = .0404, α = 0.0007). Our finding challenges the assumption that SH affects fewer people with T2DM (on insulin and/or secretagogues) than T1DM.
The overall baseline IR fell within range of the literature6,35,41,42; however, by study end, it was 5.01 (95% CI 4.15 to 6.05) EPPY (p <.0001, α = 0.0007). Future primary, real-world evidence is needed to qualify the reliability of our results. We identified only one other US prospective study on level 3 hypoglycaemia, and it enrolled a small (N = 344) chiefly male T2DM cohort on stable human insulin.26
Level 3 event rates, as with IPs, are often presumed to be lower in T2DM than in T1DM. Yet, analogous to the Canadian InHypo-DM35 and Hypoglycaemia Assessment Tool (HAT)43 studies, we observed comparable rates by diabetes type over the prospective phase (p = .1352, α = 0.0007). Explanatory factors could include our enrolment of combo insulin-secretagogue users,44 and the relatively long therapy duration reported by T2DM participants (~5 years).45 Moreover, previous research describes lower engagement in recommended hypoglycaemia self-management among people with T2DM versus T1DM,46 a concern potentially exacerbated by COVID-19 period effects.
Following the prospective phase (i.e. pandemic), 13% of T2DM respondents reported 92% of all level 3 events, while about the same percentage with T1DM reported 78% of events. Conversely, baseline event distributions (i.e. pre-pandemic) were steadier (between diabetes types) and less skewed (within diabetes types). Described elsewhere,47 more iNPHORM participants with T2DM than T1DM experienced difficulties testing their BG, monitoring hypoglycaemia and accessing social support to help manage hypoglycaemia because of the pandemic. Such compromised aspects of self-management could explain why, in this specific study, people with T2DM versus T1DM experienced considerably higher level 3 IRs at study end than at baseline.
Most saliently, T2DM rates increased for non-transport EMS, which by study end amounted to an IR nine times that reported in T1DM (p <.0001, α = 0.0007). Interestingly, we found no significant differences between T1DM- and T2DM-specific rates of health care-related SH at baseline. Our results could reaffirm the moderating effects of COVID-19, although some pre-pandemic studies document greater use of SH-related parenteral therapy and longer in-hospital stays among patients with T2DM versus T1DM.33,48,49 Speculated factors include deficient T2DM glucagon dispensation in the United States50-52 and high uptake of secretagogues that, compared with insulin, can aggravate worse cognitive dysfunction and SH prolongation.3,53
Future iNPHORM analyses will focus on pinpointing the exact causes of level 3 hypoglycaemia occurrence and associated variability, helping inform preventive strategies moving forward. Notwithstanding, that overall event rates did not decline from baseline to study end (a roughly 2-year observation period) intimates the need for improved individual- and population-level tactics to reduce SH occurrence, particularly among patients with T2DM and recurrent hypoglycaemia. To this end, practitioners should consider tailoring glycaemic goals; deintensifying, simplifying, or changing therapies; prescribing real-time continuous/flash glucose monitoring or hybrid closed-loop pumps; screening for impaired awareness of hypoglycaemia; integrating structured diabetes education; and engaging significant others. System-level interventions could include risk-based glycaemic targets, electronic health record notifications, and multidisciplinary care approaches.
Multiple strengths shore confidence in iNPHORM. First, we selected participants using community- versus clinic-based sampling to survey the US outpatient diabetes population; we employed this approach to minimize confounding by indication. To reflect the target population at risk of iatrogenic hypoglycaemia, we required participants be on insulin, secretagogues, or both for at least 1 year before enrolling. We also used broad eligibility criteria that encompassed historically overlooked subgroups. Namely, our sample comprised adults of all ages, including those ≥ 65 years; without newly diagnosed diabetes; and with varying histories of hypoglycaemia and/or impaired awareness. Previous epidemiological research on SH mostly excludes these higher-risk populations, which has, according to systematic reviews,28,42 contributed to a putative underestimation of real-world event frequencies.
Several additional strategies functioned to promote the real-world representativeness of the iNPHORM sample and, ultimately, the external validity of reported results. Web-based sampling from a source population purposefully designed to reflect the general public with diabetes facilitated US-wide recruitment. Today, 93% of adults are internet-connected.54 Quota sampling was used as a cost-effective backstop to satisfy our sample size objectives, while limiting coverage bias. To avoid enforcing subgroups unrepresentative of the real world (i.e. researcher bias), we applied prespecified minimum, rather than exact, quotas on a parsimonious set of key demographic/clinical variables.
Second, our retention rates surpassed other mailed and online hypoglycaemia surveys.16,33,45,55 To reduce attrition bias, we employed one-time systematic refreshment and push factors. In addition, fair cash amounts were determined to ensure a revenue-neutral experience for participants: specific values were selected to reasonably compensate individuals for their time and contribution and to assist in overcoming any barriers (e.g. cost of mileage/bus fare to access public WIFI). Incentives were approved by Western HSREB before the study start.
Third, conduct of a panel survey enabled us to glean comprehensive, up to date (including time-varying) data. To decrease reporting and ascertainment bias,18,20,56 we piloted and disseminated standardized, anonymously completed questionnaires that integrated proven design principles, quality assurance methods, and previously validated measures. Individuals had 7 days to reflect on items and, if necessary, review medication lists/containers or BG logs, before responding.
Monthly follow-ups bolstered episodic memory (by continually calling events to participants' minds)57 and decreased availability bias58,59 (e.g. where individuals with T1DM vs. T2DM may differentially recall events). Evidence asserts that the accuracy of past-year SH recall is 90%,40 an estimate that may be even higher in our study as events were elicited up to monthly. To promote respondent comprehension and reduce misclassification, a level 3 case definition (in line with the ADA2 and a recent post factum validation survey60) was carefully crafted by our team of diabetes clinical and methodological experts and included in each questionnaire.
Fourth, by collating responses over 1 year, we were able to reduce the risk of extrapolation bias. The global HAT study (which excluded the United States) is one of few other investigations to assess the frequency of past month SH.41 However, annualized rates were extrapolated from just 4 weeks of data. Previous research shows that Hawthorne effects are strongest for the first 2 months of participant follow-up.61 Thus, compared with iNPHORM, the prospective IRs reported in HAT are liable to underestimate true level 3 event occurrences.
Certain shortcomings warrant consideration. First, iNPHORM participants opted into the internet panel, introducing potential non-response and volunteer bias. For example, past hypoglycaemia occurrence may have swayed individuals' desires to enrol or not enrol and, accordingly, biased estimates of level 3 hypoglycaemia frequency. Coverage error (because of English-language restriction or limited web literacy/access) as well as survivorship bias (because of unmeasured lifetime event exposure) may also have affected our results.
Second, mirroring other large US diabetes cohorts,41,62-66 iNPHORM participants were mostly white, educated, financially secure, insured, and in relatively good glycaemic control. Conversely, our T2DM participants were slightly younger, and T1DM participants slightly older, perhaps reflecting our inclusion of secretagogue (without insulin) users, and exclusion of paediatric populations, respectively. We furthermore report lower prevalences of impaired awareness of hypoglycaemia67,68 and chronic kidney disease.69
The under- or overrepresentation of certain groups can impose extreme effects on apparent frequencies, given the right-skewed nature of SH (i.e. a minority of people have the majority of events).24-26 In our study, only 11% of participants experienced 84% of events. Nonetheless, fair scrutiny of iNPHORM's generalizability is challenged by absent US census data confirming the true distribution of adults with insulin- and/or secretagogue-treated diabetes. We therefore report, in detail, the anthropometric, sociodemographic and clinical composition of the iNPHORM cohort to facilitate full research trans-parency and critical appraisal.
Third, reporting and misclassification bias cannot be discounted. While self-report was necessary to investigate level 3 hypoglycaemia, we did not verify events using, for example, claim records or third-party testimony. Likewise, we did not corroborate self-reported characteristics with objective clinical data (e.g. laboratory C-peptide or glycated haemoglobin values). Finally, IRs allowed us to factor time to event, differences in follow-up duration, and changes in exposure or eligibility status; however, because IPs do not account for zero-risk, ineligible, or unobserved periods, our estimates may diverge from actual population parameters.
These limitations (i.e. sampling, selection, reporting and misclassification biases) may have compromised data generalizability. Further work is needed to corroborate our results in similar real-world settings. Future adjusted analyses to determine the risk factors (e.g. age, socio-economic status, medication regimen, impaired awareness of hypoglycaemia) of level 3 hypoglycaemia occurrence, which is beyond the scope of this study, will also need to be conducted.
5 ∣. CONCLUSION
By methodically striving to emulate the real-world US adult population with insulin- and/or secretagogue-treated diabetes, iNPHORM stands to provide the truest estimation of level 3 hypoglycaemia occurrence, to date. Our results showed the importance of frequent prospectively collected self-report to ascertain SH epidemiology, while also underscoring the limitations of routine records surveillance. In both T1DM and T2DM, we uncovered remarkably high incidences of level 3 hypoglycaemia, challenging conventional understandings, with a minority of people reporting the majority of events. Improved risk tailored management and increased investment in preventive public health initiatives are urgently needed.
Supplementary Material
ACKNOWLEDGMENTS
The iNPHORM study was funded through an investigator-initiated grant from Sanofi Global. Sanofi Global was not involved in the study design; collection, analysis, and interpretation of data; writing of the report; nor in the decision to submit the article for publication. All authors confirm their independence from funders and that they had full access to the study data (including statistical reports and tables). They take responsibility for the integrity of the data and the accuracy of the data analysis.
Footnotes
CONFLICT OF INTEREST STATEMENT
Alexandria Ratzki-Leewing: Sanofi global: grant, member advisory board, fees paid for presentations; Sanofi Canada: fees paid for presentations; Abbott: consultant, fees paid for presentations; Eli Lilly: consultant, fees paid for presentations; Dexcom: consultant; Novo Nordisk: consultant. Jason E. Black, Anna R. Kahkoska, Bridget L. Ryan, Guangyong Zou, Neil Klar, and Kristina Timcevska have nothing to disclose. Stewart B. Harris: Sanofi: grant, member advisory board, fees paid for presentations, consultant; Eli Lilly: grant, member advisory board, consultant, clinical studies; Novo Nordisk: grant, member advisory board, consultant, clinical studies; Janssen: grant, member advisory board, consultant; AstraZeneca: grant, member advisory board, consultant, clinical studies; Abbott: grant, membery advisory board, consultant; Dexcom: consultant; Boehringer Ingelheim: grant, member advisory board, consultant, clinical studies.
SUPPORTING INFORMATION
Additional supporting information can be found online in the Supporting Information section at the end of this article.
DATA AVAILABILITY STATEMENT
The datasets generated during and/or analyzed during the current study are not publicly available due to limitations in data sharing of participant information.
REFERENCES
- 1.Frier BM. Hypoglycaemia in diabetes mellitus: epidemiology and clinical implications. Nat Rev Endocrinol. 2014;10(12):711–722. [DOI] [PubMed] [Google Scholar]
- 2.American Diabetes Association. 6. Glycemic targets: standards of medical care in diabetes-2019. Diabetes Care. 2019;42:S61–S70. [DOI] [PubMed] [Google Scholar]
- 3.Salutini E, Bianchi C, Santini M, et al. Access to emergency room for hypoglycaemia in people with diabetes. Diabetes Metab Res Rev. 2015;31(7):745–751. [DOI] [PubMed] [Google Scholar]
- 4.Saydah SH. Medication use and self-care practices in persons with diabetes. In: Cowie CC, Casagrande SS, Menke A, et al. , eds. Diabetes in America [Internet]. 3rd ed. National Institute of Diabetes and Digestive and Kidney Diseases (US); 2018. [Accessed 27 July 2022]. Available from: http://www.ncbi.nlm.nih.gov/books/NBK567996/ [PubMed] [Google Scholar]
- 5.Wirtz VJ, Knox R, Cao C, et al. Insulin market profile [Internet]. Health Action International; 2016:104 Available from: https://haiweb.org/wp-content/uploads/2016/04/ACCISS_Insulin-Market-Profile_FINAL.pdf [Google Scholar]
- 6.Pedersen-Bjergaard U, Alsifri S, Aronson R, et al. Comparison of the HAT study, the largest global hypoglycaemia study to date, with similar large real-world studies. Diabetes Obes Metab. 2019;21(4):844–853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ginde AA, Blanc PG, Lieberman RM, Camargo CA. Validation of ICD-9-CM coding algorithm for improved identification of hypoglycemia visits. BMC Endocr Disord. 2008;8:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.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(5):678–686. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.McCoy RG, Herrin J, Galindo RJ, et al. Rates of hypoglycemic and hyperglycemic emergencies among US adults with diabetes, 2011-2020. Diabetes Care. 2023;46(2):e69–e71. doi: 10.2337/dc22-1673 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ratzki-Leewing A, Harris SB, Zou G, Ryan BL. Real-world estimates of severe hypoglycaemia and associated healthcare utilisation in the US: baseline results of the iNPHORM study. Diabetologia. 2020;63(Suppl.1): 750P S363. [Google Scholar]
- 11.Karter AJ, Moffet HH, Liu JY, Lipska KJ. Surveillance of hypoglycemia-limitations of emergency department and hospital utilization data. JAMA Intern Med. 2018;178(7):987–988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Von Elm E, Altman DG, Egger M, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–349. [DOI] [PubMed] [Google Scholar]
- 13.Ratzki-Leewing A, Ryan BL, Zou GY, et al. Predicting real-world hypoglycemia risk in American adults with type 1 or 2 diabetes mellitus prescribed insulin and/or secretagogues: protocol for a prospective, 12-wave internet-based panel survey with email support (the iNPHORM [investigating novel predictions of hypoglycemia occurrence using real-world models] study). J Med Internet Res. 2022;11(2):e33726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ratzki-Leewing A, Black JE, Mequanint S, et al. Severe hypoglycemia rates are highest among those with suboptimal reporting behaviour—results from the InHypo-DM study. Diabetes. 2018;67(Supplement_1):399. [Google Scholar]
- 15.Mojdami D, Mitchell BD, Spaepen E, et al. Conversations and reactions around severe hypoglycemia study: results of hypoglycemia experiences in Canadian adults with insulin-treated diabetes and their caregivers. Can J Diabetes. 2021;45(3):236–242. [DOI] [PubMed] [Google Scholar]
- 16.Östenson CG, Geelhoed-Duijvestijn P, Lahtela J, Weitgasser R, Markert Jensen M, Pedersen-Bjergaard U. Self-reported non-severe hypoglycaemic events in Europe. Diabet Med. 2014;31(1):92–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Peene B, D'Hooge D, Vandebrouck T, Mathieu C. Patient-reported frequency, awareness and patient-physician communication of hypoglycaemia in Belgium. Acta Clin Belg. 2014;69:439–445. [DOI] [PubMed] [Google Scholar]
- 18.Gnambs T, Kaspar K. Disclosure of sensitive behaviors across self-administered survey modes: a meta-analysis. Behav Res Methods. 2015;47(4):1237–1259. [DOI] [PubMed] [Google Scholar]
- 19.Warner CH, Appenzeller GN, Grieger T, et al. Importance of anonymity to encourage honest reporting in mental health screening after combat deployment. Arch Gen Psychiatry. 2011;68(10):1065–1071. [DOI] [PubMed] [Google Scholar]
- 20.Pedersen-Bjergaard U, Færch L, Allingbjerg ML, Agesen R, Thorsteinsson B. The influence of new European Union driver's license legislation on reporting of severe hypoglycemia by patients with type 1 diabetes. Diabetes Care. 2015;38(1):29–33. [DOI] [PubMed] [Google Scholar]
- 21.Ipsos. Medical Devices & Diagnostics Centre of Expertise: 2020 Capabilities [Internet]. [Accessed 9 June 2023]. Available from: https://www.ipsos.com/sites/default/files/ipsos-mdd-global-capabilities.pdf [Google Scholar]
- 22.Ratzki-Leewing A, Black J, Ryan B, et al. Development and validation of a real-world model to predict one-year, recurrent level 3 (severe) hypoglycaemia risk in adults with diabetes (the iNPHORM study, United States). Diabetes Obes Metab. 2023;25(10):2910–2927. [DOI] [PubMed] [Google Scholar]
- 23.Stopher P. Collecting, Managing, and Assessing Data Using Sample Surveys. Cambridge University Press; 2012. [Google Scholar]
- 24.Pedersen-Bjergaard U, Pramming S, Heller SR, et al. Severe hypoglycaemia in 1076 adult patients with type 1 diabetes: influence of risk markers and selection. Diabetes Metab Res Rev. 2004;20(6):479–486. [DOI] [PubMed] [Google Scholar]
- 25.Henderson JN, Allen KV, Deary IJ, Frier BM. Hypoglycaemia in insulin-treated type 2 diabetes: frequency, symptoms and impaired awareness. Diabet Med. 2003;20(12):1016–1021. [DOI] [PubMed] [Google Scholar]
- 26.Murata GH, Duckworth WC, Shah JH, Wendel CS, Mohler MJ, Hoffman RM. Hypoglycemia in stable, insulin-treated veterans with type 2 diabetes: a prospective study of 1662 episodes. J Diabetes Complications. 2005;19(1):10–17. [DOI] [PubMed] [Google Scholar]
- 27.Tattersall R. Frequency, causes and treatment of hypoglycaemia. In: Frier B, Fisher B, eds. Hypoglycaemia in Clinical Diabetes. John Wiley and Sons; 1999:55–87. [Google Scholar]
- 28.Pedersen-Bjergaard U, Thorsteinsson B. Reporting severe hypoglycemia in type 1 diabetes: facts and pitfalls. Curr Diab Rep. 2017;17(12):131. [DOI] [PubMed] [Google Scholar]
- 29.Bjork E, Palmer M, Schvarcz E, Berne C. Incidence of severe hypoglycaemia in an unselected population of patients with insulin-treated diabetes mellitus, with special reference to autonomic neuropathy. Diabetes Nutr Metab. 1990;4:303–309. [Google Scholar]
- 30.Leese GP, Wang J, Broomhall J, et al. Frequency of severe hypoglycemia requiring emergency treatment in type 1 and type 2 diabetes: a population-based study of health service resource use. Diabetes Care. 2003;26(4):1176–1180. [DOI] [PubMed] [Google Scholar]
- 31.Heinemann L, Freckmann G, Ehrmann D, et al. Real-time continuous glucose monitoring in adults with type 1 diabetes and impaired hypoglycaemia awareness or severe hypoglycaemia treated with multiple daily insulin injections (HypoDE): a multicentre, randomised controlled trial. Lancet. 2018;391(10128):1367–1377. [DOI] [PubMed] [Google Scholar]
- 32.Lammert M, Hammer M, Frier BM. Management of severe hypoglycaemia: cultural similarities, differences and resource consumption in three European countries. J Med Econ. 2009;12(4):269–280. [DOI] [PubMed] [Google Scholar]
- 33.Donnelly LA, Morris AD, Frier BM, et al. Frequency and predictors of hypoglycaemia in type 1 and insulin-treated type 2 diabetes: a population-based study. Diabet Med. 2005;22(6):749–755. [DOI] [PubMed] [Google Scholar]
- 34.Chen Y, Liu L, Gu L, Babineaux S, Colclough H, Curtis B. Glycemic control in Chinese patients with type 2 diabetes mellitus receiving oral antihyperglycemic medication-only or insulin-only treatment: a cross-sectional survey. Diabetes Ther. 2015;6(2):197–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ratzki-Leewing A, Harris SB, Mequanint S, et al. Real-world crude incidence of hypoglycemia in adults with diabetes: results of the InHypo-DM study, Canada. BMJ Open Diabetes Res Care. 2018;6(1):e000503. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Leckie AM, Graham MK, Grant JB, Ritchie PJ, Frier BM. Frequency, severity, and morbidity of hypoglycemia occurring in the workplace in people with insulin-treated diabetes. Diabetes Care. 2005;28(6):1333–1338. [DOI] [PubMed] [Google Scholar]
- 37.Færch L, Pedersen-Bjergaard U, Thorsteinsson B. High serum ACE activity predicts severe hypoglycaemia over time in patients with type 1 diabetes. Scand J Clin Lab Invest. 2011;71(7):620–624. [DOI] [PubMed] [Google Scholar]
- 38.MacLeod KM, Hepburn DA, Frier BM. Frequency and morbidity of severe hypoglycaemia in insulin-treated diabetic patients. Diabet Med. 1993;10(3):238–245. [DOI] [PubMed] [Google Scholar]
- 39.ter Braak EW, Appelman AM, van de Laak M, Stolk RP, van Haeften TW, Erkelens DW. Clinical characteristics of type 1 diabetic patients with and without severe hypoglycemia. Diabetes Care. 2000;23(10):1467–1471. [DOI] [PubMed] [Google Scholar]
- 40.Pedersen-Bjergaard U, Pramming S, Thorsteinsson B. Recall of severe hypoglycaemia and self-estimated state of awareness in type 1 diabetes. Diabetes Metab Res Rev. 2003;19(3):232–240. [DOI] [PubMed] [Google Scholar]
- 41.Khunti K, Alsifri S, Aronson R, et al. Rates and predictors of hypoglycaemia in 27 585 people from 24 countries with insulin-treated type 1 and type 2 diabetes: the global HAT study. Diabetes Obes Metab. 2016;18(9):907–915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Elliott L, Fidler C, Ditchfield A, Stissing T. Hypoglycemia event rates: a comparison between real-world data and randomized controlled trial populations in insulin-treated diabetes. Diabetes Ther. 2016;7(1):45–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Aronson R, Goldenberg R, Boras D, Skovgaard R, Bajaj H. The Canadian hypoglycemia assessment tool program: insights into rates and implications of hypoglycemia from an observational study. Can J Diabetes. 2018;42:11–17. [DOI] [PubMed] [Google Scholar]
- 44.Mogensen UM, Andersson C, Fosbøl EL, et al. Sulfonylurea in combination with insulin is associated with increased mortality compared with a combination of insulin and metformin in a retrospective Danish nationwide study. Diabetologia. 2015;58(1):50–58. [DOI] [PubMed] [Google Scholar]
- 45.UK Hypoglycaemia Study Group. Risk of hypoglycaemia in types 1 and 2 diabetes: effects of treatment modalities and their duration. Diabetologia. 2007;50(6):1140–1147. [DOI] [PubMed] [Google Scholar]
- 46.Harris S, Black JE, Ryan B, et al. Hypoglycemia self-management among adults with diabetes: gaps and leverage points (iNPHORM study, United States). Diabetologia. 2023; In press. [Google Scholar]
- 47.Ratzki-Leewing AA, Ryan BL, Buchenberger JD, Dickens JW, Black JE, Harris SB. COVID-19 hinterland: surveilling the self-reported impacts of the pandemic on diabetes management in the USA (cross-sectional results of the iNPHORM study). BMJ Open. 2021;11(9):e049782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Zhong VW, Juhaeri J, Cole SR, et al. Incidence and trends in hypoglycemia hospitalization in adults with type 1 and type 2 diabetes in England, 1998-2013: a retrospective cohort study. Diabetes Care. 2017;40(12):1651–1660. [DOI] [PubMed] [Google Scholar]
- 49.Fadini GP, Rigato M, Tiengo A, Avogaro A. Characteristics and mortality of type 2 diabetic patients hospitalized for severe iatrogenic hypoglycemia. Diabetes Res Clin Pract. 2009;84(3):267–272. [DOI] [PubMed] [Google Scholar]
- 50.Kahn PA, Wagner NE, Gabbay RA. Underutilization of glucagon in the prehospital setting. Ann Intern Med. 2018;168(8):603–604. [DOI] [PubMed] [Google Scholar]
- 51.Mitchell BD, He X, Sturdy IM, Cagle AP, Settles JA. Glucagon prescription patterns in patients with either type 1 or 2 diabetes with newly prescribed insulin. Endocr Pract. 2016;22(2):123–135. [DOI] [PubMed] [Google Scholar]
- 52.Snoek FJ, Spaepen E, Mojdami D, et al. The multinational conversations and reactions around severe hypoglycemia (CRASH) study: impact of health care provider communications and recommendations on people with diabetes. J Clin Transl Endocrinol. 2022;27:100295. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Monami M, Dicembrini I, Kundisova L, Zannoni S, Nreu B, Mannucci E. A meta-analysis of the hypoglycaemic risk in randomized controlled trials with sulphonylureas in patients with type 2 diabetes. Diabetes Obes Metab. 2014;16(9):833–840. [DOI] [PubMed] [Google Scholar]
- 54.Pew Research Center. Demographics of Internet and Home Broadband Usage in the United States [Internet]. Pew Research Center: Internet, Science & Tech; 2023. Available from: https://www.pewresearch.org/internet/fact-sheet/internet-broadband/ [Accessed 9 June 2023] [Google Scholar]
- 55.Marrett E, Radican L, Davies MJ, Zhang Q. Assessment of severity and frequency of self-reported hypoglycemia on quality of life in patients with type 2 diabetes treated with oral antihyperglycemic agents: a survey study. BMC Res Notes. 2011;4:251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Chang L, Krosnick JA. National surveys via rdd telephone interviewing versus the internet: comparing sample representativeness and response quality. Public Opin Q. 2009;73(4):641–678. [Google Scholar]
- 57.Linton M. Transformations of memory in everyday life. Memory Observed: Remembering in Natural Contexts. Freeman; 1982. [Google Scholar]
- 58.Willén R. Recollection of Repeated Events: Difficulties and Possibilities. University of Gothenburg; 2015. [Google Scholar]
- 59.Schwarz N, Bless H, Strack F, Klumpp G, Rittenauer-Schatka H, Simons A. Ease of retrieval as information: another look at the availability heuristic. J Pers Soc Psychol. 1991;61(2):195–202. [Google Scholar]
- 60.Madar H, Wu Z, Bandini A, et al. Influence of severe hypoglycemia definition wording on reported prevalence in adults and adolescents with type 1 diabetes: a cross-sectional analysis from the BETTER patient-engagement registry analysis. Acta Diabetol. 2023;60(1):93–100. [DOI] [PubMed] [Google Scholar]
- 61.Clark R, Sugrue B. Research on instructional media. In: Anglin G, ed. Instructional Technology: Past, Present and Future. Libraries Unlimited; 1991:327–343. [Google Scholar]
- 62.Wang L, Li X, Wang Z, et al. Trends in prevalence of diabetes and control of risk factors in diabetes among US adults, 1999-2018. Jama. 2021;326:704–716. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Beck RW, Tamborlane WV, Bergenstal R, et al. The T1D exchange clinic registry. J Clin Endocrinol Metab. 2012;97:4383–4389. [DOI] [PubMed] [Google Scholar]
- 64.DeSalvo DJ, Noor N, Xie C, et al. Patient demographics and clinical outcomes among type 1 diabetes patients using continuous glucose monitors: data from T1D exchange real-world observational study. J Diabetes Sci Technol. 2023;17:322–328. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Fishbein HA, Birch RJ, Mathew S, et al. The longitudinal epidemiologic assessment of diabetes risk (LEADR): unique 1.4 M patient electronic health record cohort. Healthcare. 2020;8:100458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Fang M, Wang D, Coresh J, Selvin E. Trends in diabetes treatment and control in US adults, 1999-2018. N Engl J Med. 2021;384(23):2219–2228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.van Meijel LA, de Vegt F, Abbink EJ, et al. High prevalence of impaired awareness of hypoglycemia and severe hypoglycemia among people with insulin-treated type 2 diabetes: the Dutch diabetes pearl cohort. BMJ Open Diabetes Res Care. 2020;8:e000935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 68.Geddes J, Schopman J, Zammit N, Frier B. Prevalence of impaired awareness of hypoglycaemia in adults with type 1 diabetes. Diabet Med. 2008;25:501–504. [DOI] [PubMed] [Google Scholar]
- 69.Centers for Disease Control and Prevention, National Center for Health Statistics. National Health and Nutrition Examination Survey Data. 2001. https://nccdcdc.gov/CKD/detail.aspx?QNum=Q702
- 70.Clarke WL, Cox DJ, Gonder-Frederick LA, Julian D, Schlundt D, Polonsky W. Reduced awareness of hypoglycemia in adults with IDDM. A prospective study of hypoglycemic frequency and associated symptoms. Diabetes Care. 1995. Apr;18(4):517–522. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated during and/or analyzed during the current study are not publicly available due to limitations in data sharing of participant information.