diabetes research and clinical practice 91 (2011) 363–370
Contents lists available at ScienceDirect
Diabetes Research and Clinical Practice journ al h omepage: www .elsevier.co m/lo cate/diabres
Effects of hypoglycemia on health-related quality of life, treatment satisfaction and healthcare resource utilization in patients with type 2 diabetes mellitus Setareh A. Williams a,*, Michael F. Pollack a, Marco DiBonaventura b a b
AstraZeneca LP, Wilmington, DE 19850, United States Kantar Health Princeton, NJ, United States
article info
abstract
Article history:
Aims: To quantify patient-reported rates of hypoglycemia and its association with health-
Received 12 August 2010
related quality of life (HRQL), treatment satisfaction, and healthcare resource utilization.
Received in revised form
Methods: Data were collected from 2006 to 2008 US National Health and Wellness Survey and
30 November 2010
the Ailment Panel of Lightspeed Online Research, an internet-based questionnaire. Adults
Accepted 21 December 2010
(18 years) with type 2 diabetes taking 1 oral antidiabetic agent (OAD), but not insulin, were
Published on line 19 January 2011
included (n = 2074). Multivariate analyses included logistic regression and generalized linear models.
Keywords:
Results: Overall, patients who reported experiencing hypoglycemia symptoms (n = 286;
Oral antidiabetic
13.78%) were significantly more likely to have a lower HRQL on several parameters includ-
Health related quality of life
ing: increased limitations on mobility (b = 0.66, OR = 1.93, p < 0.0001) and usual activities
Treatment satisfaction
(b = 0.58, OR = 1.78, p < 0.0001), increased pain/discomfort (b = 0.69, OR = 2.00, p < 0.0001)
Diabetes mellitus
and anxiety/depression (b = 0.84, OR = 2.31, p < 0.0001). They also had a lower total treatment satisfaction score as measured by the DiabMedSat tool (b =
7.66, p < 0.0001). Self-
reported rates of diabetes-related emergency room (b = 0.98, p = 0.004) and physician visits (b = 0.30, p < 0.0001) were also higher among these patients. Conclusion: Among OAD-treated type 2 diabetes patients, symptoms of hypoglycemia tend to be correlated with significantly lower HRQL, lower treatment satisfaction and higher levels of healthcare resource utilization. # 2010 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Diabetes mellitus is a chronic disorder with increasing incidence and prevalence in the US and worldwide [1]. Diabetes affects almost 24 million people, or 7.8% of the US population. Of these, the Centers for Disease Control and Prevention (CDC) estimates that 90–95% have type 2 diabetes [2]. Treatment for type 2 diabetes can be complex and multidisciplinary. There are three main goals of type 2 diabetes
treatment: prevention of hyperglycemia and the development of associated disease complications, prevention of hypoglycemia and maintenance of the patient’s quality of life [3]. In order to achieve these goals, current American Diabetes Association (ADA) guidelines recommend both lifestyle therapeutic modifications, the administration of oral antidiabetic drugs (OADs) for achievement and maintenance of glycemic control: a hemoglobin A1c (HbA1c) less than 7% [4].
* Corresponding author at: AstraZeneca LP, Health Economics and Outcomes Research, D3C-114, 1800 Concord Pike, PO Box 15437, Wilmington, DE 19850, United States. Tel.: +1 302 885 1239; fax: +1 302 885 3529. E-mail address:
[email protected] (S.A. Williams). 0168-8227/$ – see front matter # 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2010.12.027
364
diabetes research and clinical practice 91 (2011) 363–370
According to the data from the National Health and Nutrition Examination Survey (NHANES), although the proportion of patients with type 2 diabetes who achieve glycemic control has increased from 43.1 to 57.1% during 1999 2006, 40% of patients still failed to achieve a glycemic target of HbA1c < 7% [5]. Barriers to achieving glycemic targets and maintaining appropriate disease management can be multifaceted; balancing the need for tight glycemic control while minimizing the risk of treatment-induced hypoglycemia may be the most challenging. As noted by the recent guidance [6] of the American Association of Clinical Endocrinologists (AACE) and the American College of Endocrinology (ACE), hypoglycemia may have substantial negative clinical effects in terms of morbidity, mortality, and quality of life, and should be avoided when possible. The occurrence of hypoglycemic episodes is often difficult to document, as many events go unreported or even unrecognized by the patient [7–9]. One study utilizing continuous blood glucose monitoring (CBGM) indicates that patients experience hypoglycemic episodes that are frequently unrecognized [10]. However, several studies have demonstrated that a considerable proportion (24.5–62.9%) of type 2 diabetes patients report experiencing symptoms commonly associated with hypoglycemia [11–13]. While hypoglycemia has been shown to have adverse consequences, and may result in decreased health-related quality of life (HRQL) and increases in healthcare resource utilization or costs across populations, [12–15] there are a number of limitations with these studies. Because many cases of mild or moderate hypoglycemia go unreported or unrecognized by the patient or physician, a predominance of the available data is based on more severe episodes, those requiring medical care [15]. In addition, many of these studies were conducted outside of the US, [13,14] in insulin- or mixed-treatment populations [15] that may not be representative of a type 2 diabetes OAD-treated group in the US. Some other studies included relatively small sample sizes, limiting their generalizability [16]. The current study adds to the existing literature by providing patient-reported data of hypoglycemia among a relatively large sample of US type 2 diabetes patients treated with OADs. The objectives are to: (1) identify patient-reported rates of hypoglycemia and (2) evaluate the association of these self-reports on HRQL, treatment satisfaction, and healthcare resource utilization.
older; self-reported diagnosis of type 2 diabetes made by a healthcare provider; currently taking one or more OADs, but not insulin; ability to read and write English; residence in the US at the time of the study; and the provision of informed consent for study participation. Of the 10,374 patients invited, 4143 (39.9% response rate) provided informed consent and completed the screener. Of these 4143 patients, 2074 (20.0% completion rate) participants met all inclusion criteria and completed the survey.
3.
Materials and methods
A self-administered, internet-based survey was developed and administered in September 2008 to a cohort of respondents who met the study criteria. This cross-sectional survey collected data on demographics, health history, disease and treatment characteristics, and healthcare resource utilization. The study protocol and questionnaire were reviewed and approved by Essex IRB (Lebanon, NJ, USA).
3.1.
Study measures
3.1.1.
Demographics, comorbidity, and disease characteristics
2.
Subjects, materials, and methods
Gender, race/ethnicity (white vs. non-white), marital status (married/living with partner vs. not married), income (<$50K vs. $50K), education (college graduate vs. not a college graduate), health insurance (yes vs. no), and employment (fulltime, part-time, or self-employed vs. all else) were assessed as categorical variables; age and body mass index (BMI) were assessed as continuous variables. BMI was calculated from self-reported height and weight for each study participant. Overall health and disease comorbidity was assessed by applying the Charlson Comorbidity Index (CCI) algorithm to self-reported health conditions [17]. Several disease-specific variables were also included to determine potential severity of type 2 diabetes, such as the number of years since diagnosis of type 2 diabetes, the number of diabetes medications, recent initiation of a new OAD, adherence to diabetes medications as measured by the 4-item Morisky Medication Adherence Scale (MMAS), [18] classes of medication (e.g. sulfonylurea, biguanide, thiazolidinedione, dipeptidyl peptidase-4 inhibitor, other), HbA1c testing frequency (3 months or less, 4–6 months, 7–12 months, 12 months or more, or never), HbA1c target level set by physician (below target, at target, above target, or unknown), years of OAD treatment, and knowledge of and values for glucose parameters.
2.1.
Subjects
3.1.2.
A total of 10,374 potential study participants were identified through self-reported diabetes responses collected from 2006 to 2008 US National Health and Wellness Survey (NHWS) (an annual internet-based questionnaire developed by Consumer Health Sciences/Kantar Health), and the Ailment Panel of Lightspeed Online Research (a partner organization that manages an online-registered consumer panel worldwide, using a multi-source recruiting methodology). These potential participants were invited via e-mail to participate in this study. Specific criteria for inclusion were: 18 years of age or
Hypoglycemia measure
The 30-item Diabetes Symptom Measure (DSM), [19] was used to assess the presence and frequency of cognitive and physiological symptoms in the 2 weeks preceding survey administration. Frequency of disease symptoms was assessed using a five-point Likert-type scale, ranging from ‘‘none of the time’’ to ‘‘all of the time’’. The DSM includes two items that were considered to be closely related to the occurrence of hypoglycemia. Respondents who indicated experienced ‘‘symptoms of low blood sugar,’’ or ‘‘low blood sugar in the middle of the night’’ ‘‘some,’’ ‘‘most’’ or ‘‘all of the time’’ were considered to have hypoglycemia (n = 286) for this analysis.
diabetes research and clinical practice 91 (2011) 363–370
3.1.3.
Health-related quality of life
The EQ-5D, a standardized, preference-based utility measure, was used to assess respondents’ HRQL and utility values. The EQ-5D Index is computed from five dimensions of health: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Responses for each of the five dimensions range from ‘‘no health problems,’’ to ‘‘moderate health problems,’’ and ‘‘extreme health problems.’’ A composite EQ-5D score is computed from the five dimensions and ranges from 0 to 1; higher scores indicate greater utility. The EQ-5D questionnaire also includes a visual analogue scale, known as EQ-VAS. The respondent self-rates current health status on a thermometer-like scale ranging from 0 (‘‘worst imaginable health state’’) to 100 (‘‘best imaginable health state’’) [20].
3.1.4.
Treatment satisfaction
Treatment satisfaction was evaluated using the Diabetes Medication Satisfaction (DiabMedSat) tool, [19] a 21-item instrument that measures three domains: satisfaction related to burden (e.g., frequency of monitoring blood glucose, frequency of medication dosing), efficacy, and symptoms, each of which is computed as an individual score. In addition, a total satisfaction score is computed. Each item in the scale uses a 5–7 point Likert-type scale; overall and domain scores range from 0 to 100, with higher scores indicating greater levels of treatment satisfaction.
3.1.5.
Resource use
To assess the association between self-reported hypoglycemia and healthcare resource utilization, survey respondents were questioned about the number of emergency room (ER) visits, diabetes-related ER visits, total days hospitalized, and the number of physician visits in the 6 months prior to taking the survey.
3.2.
365
and use of a sulfonylurea. All analyses were conducted with SAS v9.1 (Cary, NC) and a prior cutoff for statistical significance was set at p < 0.05. Where appropriate for ordered logistic regressions beta-coefficients (b) and odds ratios (OR) are provided in the results section; for all other multivariate regressions beta-coefficients are provided.
4.
Results
4.1.
Predictors of hypoglycemia
Approximately 13.78% (286 of 2074) of all type 2 diabetes patients who fulfilled the study criteria reported experiencing symptoms of low blood sugar or hypoglycemia in the 2 weeks prior to the survey administration. Compared to those who did not report hypoglycemia, patients who did were significantly younger (56.40 vs. 60.64, p < 0.0001), less likely to be white (77.97% vs. 88.14%, p = 0.0001), and less likely to have health insurance (85.31% vs. 92.00%, p = 0.0025), and more likely to have an annual household income below $50,000 (60.49% vs. 49.22%, p = 0.0004) (Table 1). Those experiencing low blood sugar were significantly more likely to use a sulfonylurea (64.34% vs. 45.97%, p < 0.0001) than were those not experiencing low blood sugar. Patients with low blood sugar were also significantly less likely to be at their HbA1c treatment goal (10.84% vs. 21.64%, p < 0.0001) and more likely to not know their HbA1c level (27.62% vs. 20.92%, p = 0.018). Both groups had equivalent percentages of patients who began using a new medication in the past 6 months (11.19% vs. 11.97% for low blood sugar and without low blood sugar groups, respectively, p = 0.70) and 12 months (21.68% vs. 22.20%, p = 0.84). Nonadherence (MMAS 2) was significantly higher among those with low blood sugar, compared with those without (19.93% vs. 13.8%, p = 0.007).
Statistical analyses 4.2.
Bivariate differences between patients who experience low blood sugar and those who do not were examined using chisquare tests and t-tests for categorical and continuous dependent variables, respectively. Depending on the nature of the outcome variable, different multivariate techniques were utilized. The EQ-5D subscale scores are rating scales without clear equidistance between response categories. Because of this property, ordered logistic regressions were conducted. The EQ-VAS, utilities, and medication satisfaction subscale scores are fundamentally continuous; therefore, differences between groups were analyzed using multiple linear regressions. Resource use variables were analyzed using generalized linear models (GLMs) with a log-link function specifying either a Poisson distribution (diabetes-related ER visits) or negative binomial distribution (all others) depending on what best fit the data. Corrections to the standard errors were applied by setting the scale parameter to one. All multivariate differences between groups controlled for gender (with female serving as the reference category), ethnicity (with non-white serving as the reference category), income (with <$50K serving as the reference category), health insurance, age, comorbidities, BMI, HbA1c level (with below target serving as the reference category), number of diabetes medications,
Hypoglycemia and HRQL
Those with low blood sugar had significantly higher mean EQ5D subscale scores (mobility: 1.55 vs. 1.38, p < 0.0001; self-care: 1.12 vs. 1.07, p = 0.015; usual activities: 1.52 vs. 1.35, p < 0.0001; pain/discomfort: 1.90 vs. 1.63, p < 0.0001; anxiety/depression: 1.66 vs. 1.33, p < 0.0001), indicating poorer quality of life (see Table 2). Similarly, those experiencing low blood sugar reported significantly lower mean levels on the VAS (56.02 vs. 64.59, p < 0.0001) and utility scores (0.72 vs. 0.82, p < 0.0001). In multivariate models, adjusting for confounders, patients who experienced low blood sugar were significantly more likely to report increased limitations on mobility (b = 0.66, OR = 1.93, p < 0.0001), and usual activities (b = 0.58, OR = 1.78, p < 0.0001) but were no more likely to report increased selfcare limitations (b = 0.31, OR = 1.37, p = 0.16). Patients who experienced low blood sugar were also significantly more likely to report increased pain/discomfort (b = 0.69, OR = 2.00, p < 0.0001) and anxiety/depression (b = 0.84, OR = 2.31, p < 0.0001), as assessed by the individual EQ-5D domains (see Table 3). Because interpretation of these coefficients can be challenging, the adjusted probabilities for each of these subscales are reported in Table 4. Those who experienced low
366
diabetes research and clinical practice 91 (2011) 363–370
Table 1 – Bivariate differences between patients experiencing hypoglycemia symptoms (n = 286) and patients not experiencing hypoglycemia symptoms (n = 1788)a. Hypoglycemia symptoms (low blood sugar) n = 286
Demographics Age White <$50K $50K Health insurance Medication use Sulfonylurea user Morisky non-adherence (MMAS 2) HbA1c level Below target Do not know Patient characteristics Charlson comorbidity index (mean/SD) BMI (mean/SD) Number of diabetes medications (mean/SD)
No hypoglycemia symptoms (no low blood sugar) n = 1788
p-Value
n/mean
%/SD
n/mean
%/SD
56.40 223 173 97 244
12.47 77.97% 60.49% 33.92% 85.31%
60.64 1576 880 790 1645
10.43 88.14% 49.22% 44.18% 92.00%
<0.0001 0.0001 0.0004 0.0008 0.0025
184 57
64.34% 19.93%
822 247
45.97% 13.81%
<0.0001 0.007
31 79
10.84% 27.62%
387 374
21.64% 20.92%
<0.0001 0.018
2.23 36.32 1.69
1.83 11.15 0.73
1.81 8.86 0.71
0.038 0.036 0.013
1.99 34.87 1.57
a
Note that variables that were not significantly different between groups were removed for brevity. These variables include: gender, decline to answer household income, employment, marital status, education, biguanide user, thiazolidinedione user, dipeptidyl peptidase-4 inhibitor user, other medication user, new medication user (6 months and 12 months), HbA1c testing frequency, HbA1c level above target, HbA1c at target, number of years diagnosed, and the number of days hospitalized.
blood sugar were about 1.5 times more likely to have ‘‘some problems walking about’’ (0.55 vs. 0.37) and ‘‘some problems performing usual activities’’ (0.50 vs. 0.33), compared with those who do not experience low blood sugar. Similarly, those who experienced low blood sugar were more than twice as likely to also experience ‘‘extreme pain or discomfort’’ (0.14 vs. 0.06) and report being ‘‘extremely anxious or depressed’’ (0.10 vs. 0.03). Low blood sugar was associated with significantly lower levels on the VAS (b = 5.53, p = 0.0008) and utility scores
(b = 0.08, p < 0.0001). Additional factors significantly related to VAS and utility scores included ethnicity, income, HbA1c level, age, Charlson comorbidity index, BMI, and number of diabetes medications.
4.3.
Hypoglycemia and treatment satisfaction
Significantly lower levels of treatment satisfaction, as assessed by the DiabMedSat were also observed among those
Table 2 – Bivariate health outcome differences between patients experiencing hypoglycemia symptoms (n = 286) and patients not experiencing hypoglycemia symptoms (n = 1788). Hypoglycemia symptoms (low blood sugar) n = 286
Medication satisfaction Diabetes medication satisfaction: Diabetes medication satisfaction: Diabetes medication satisfaction: Diabetes medication satisfaction: Quality of life Mobility Self-care Usual activities Pain/discomfort Anxiety/depression VAS Utilities Resource use ER visits Diabetes-related ER visits Physician visits
burden efficacy symptoms total
No hypoglycemia symptoms (no low blood sugar) n = 1788
p-Value
Mean
SD
Mean
SD
82.27 62.23 58.20 67.57
15.69 18.62 21.16 14.22
90.66 71.12 71.32 77.7
10.8 19.21 14.17 11.55
<0.0001 <0.0001 <0.0001 <0.0001
1.55 1.12 1.54 1.90 1.66 56.02 0.72
0.51 0.34 0.59 0.65 0.67 28.00 0.21
1.38 1.07 1.35 1.63 1.33 64.59 0.82
0.49 0.27 0.51 0.59 0.53 26.09 0.17
<0.0001 0.015 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001
0.30 0.07 5.22
0.71 0.33 4.88
0.19 0.02 3.83
0.62 0.16 3.42
0.014 0.008 <0.0001
367
diabetes research and clinical practice 91 (2011) 363–370
Table 3 – Adjusted effects of experiencing hypoglycemia symptoms on health outcomes. Quality of life: ordered logit
b
EQ-5D: EQ-5D: EQ-5D: EQ-5D: EQ-5D:
0.66 0.31 0.58 0.69 0.84
mobility self-care usual activities pain/discomfort anxiety/depression
Quality of life: multiple regression
b
EQ-5D: VAS EQ-5D: utilities
5.53 0.08
Resource use: generalized linear models
b
Number Number Number Number
0.28 0.98 0.09 0.30
ER visits of diabetes-related ER visits of days hospitalized of physician visits
Medication satisfaction: multiple regression Diabetes Diabetes Diabetes Diabetes a b
medication medication medication medication
satisfaction: satisfaction: satisfaction: satisfaction:
burden efficacy symptoms total
b 6.42 5.84 10.72 7.66
95% LCL 0.38 0.13 0.31 0.43 0.58 95% LCL 8.76 0.1 95% LCL 0.09 0.32 0.83 0.2 95% LCL 7.86 8.13 12.61 9.08
95% UCL 0.94 0.75 0.84 0.95 1.1
ORa
x2
1.93 1.37 1.78 2.00 2.31
21.15 1.95 17.46 26.42 39.11
95% UCL
t
2.3 0.06 95% UCL 0.65 1.64 0.65 0.41 95% UCL 4.99 3.56 8.83 6.25
3.35 7.4 ERb 1.32 2.66 0.91 1.35
x2 2.25 8.46 0.06 31.44 t 8.78 5.01 11.14 10.63
p <0.0001 0.1627 <0.0001 <0.0001 <0.0001 p 0.0008 <0.0001 p 0.1339 0.0036 0.8129 <0.0001 p <0.0001 <0.0001 <0.0001 <0.0001
OR represents odds ratio (the ratio of odds between those with hypoglycemia versus those without adjusting for covariates). ER represents event ratio (the ratio of means between those with hypoglycemia versus those without adjusting for covariates).
experiencing low blood sugar. Mean satisfaction for individual subscales [burden (82.27 vs. 90.66, p < 0.0001), efficacy (62.23 vs. 71.12, p < 0.0001), and symptoms (58.20 vs. 71.32, p < 0.0001)] and overall treatment satisfaction (67.57 vs. 77.70, p < 0.0001) were all lower in patients experiencing low blood sugar. In multivariate analyses, patients who experienced low blood sugar had a lower total satisfaction score as measured by the DiabMedSat tool (b = 7.66, p < 0.0001). This 7.66 unit
decrease equates to a 9.9% reduction in overall satisfaction scores relative to those without low blood sugar, even after controlling for demographic and patient characteristic variables. These patients also reported significantly lower satisfaction on the burden (b = 6.42, p < 0.0001), efficacy (b = 5.84, p < 0.0001), and symptom (b = 10.72, p < 0.0001) subscales. Gender, ethnicity, income, HbA1c level, age, Charlson comorbidity index, and BMI were also significantly related to satisfaction scores.
Table 4 – Adjusted probabilities of EQ-5D subscale response categories between patients who experience hypoglycemia symptoms and patients who do not.
EQ-5D: mobilit No problems walking about Some problems walking about Confined to bed EQ-5D: self-care No problems with self-care Some problems washing or dressing myself Unable to wash or dress myself EQ-5D: usual activities No problems performing my usual activities Some problems performing my usual activities Unable to perform my usual activities EQ-5D: pain/discomfor No pain or discomfort Moderate pain or discomfort Extreme pain or discomfort EQ-5D: anxiety/depression Not anxious or depressed Moderately anxious or depressed Extremely anxious or depressed
Hypoglycemia symptoms n = 286
No hypoglycemia symptoms n = 1788
0.45 0.55 0.01
0.62 0.37 0.00
0.88 0.12 0.01
0.93 0.07 0.00
0.46 0.50 0.04
0.65 0.33 0.02
0.11 0.75 0.14
0.37 0.57 0.06
0.34 0.56 0.10
0.66 0.30 0.03
368
4.4.
diabetes research and clinical practice 91 (2011) 363–370
Hypoglycemia and resource use
Though small in magnitude on a per-patient basis, patients with low blood sugar reported significantly more ER visits (0.30 vs. 0.19, p = 0.014), diabetes-related ER visits (0.07 vs. 0.02, p = 0.008), and physician visits (5.22 vs. 3.83, p < 0.0001) in the past 6 months. There were no differences between the groups with respect to the mean number of days hospitalized (0.53 vs. 0.65, p = 0.49). Though these mean differences appear small on a per patient basis, when projected out and aggregated to a larger sample (e.g. health plan population), the impact of low blood sugar may become readily apparent. For every 1000 patients in each group, these values would project to 110 more ER visits (300 visits for patients with low blood sugar vs. 190 visits for those without), 50 more diabetes-related ER visits (70 vs. 20, respectively), and almost 1400 more physician visits (5220 vs. 3830, respectively) for those who experience low blood sugar. The effects of low blood sugar also had an impact on resource use. Patients who experienced low blood sugar reported significantly more diabetes-related ER visits (b = 0.98, p = 0.004) and physician visits (b = 0.30, p < 0.0001) in the past 6 months. However, no effect was observed for overall ER visits (b = 0.28, p = 0.13) or the number of days hospitalized (b = 0.09, p = 0.81). Income, age, Charlson comorbidity index, and the number of diabetes medications, were also significantly related to resource use.
5.
Discussion
The current study sought to determine patient reported rates of low blood sugar and its association with HRQL, treatment satisfaction, and resource use. According to the findings, a significant proportion of diabetes patients experience symptomatic hypoglycemia regardless of duration of disease or timing of onset of new medication. Counter to conventional belief, the current analysis showed that low blood sugar was more prevalent among younger diabetic patients, though older patients are generally considered to be at greater risk for hypoglycemia [7,8]. This could be a result of underreporting among the older patients who may not recognize these symptoms or may attribute them to other chronic conditions or concomitant medications. Other characteristics of patients reporting low blood sugar were, however, similar to those reported in earlier studies, including lower income, lack of access to health insurance and non-white race [11,13]. When comparing the two groups, medication use was almost identical, with the exception of greater likelihood of taking a sulfonylurea among those reporting low blood sugar. Concerning HbA1c values, lack of knowledge of HbA1c values was more common with patients reporting low blood sugar. We anticipate that the observed relationship with HbA1c responses is a consequence of both personal glucose targets and question phrasing. The results demonstrated a significant relationship between experiencing low blood sugar symptoms and decreased HRQL, as determined by the EQ-5D and the EQ-VAS. Even after adjusting for a number of variables, patients who experienced
symptoms still reported decreased utility and greater difficulties with mobility, daily activities, pain/discomfort, and anxiety/depression, relative to patients who did not report those symptoms. Self-care activities were not significantly different between the groups. This is similar to the results of Marrett et al, who also used the EQ-5D and found that selfreported hypoglycemia was associated with a 0.05 unit decrease in adjusted EQ-5D scores [12]. Though the current study did not assess whether or not the observed difference in EQ-5D was perceived as beneficial to patients (i.e. minimal important difference [MID]), the results do fall within the range of potential MID values reported elsewhere for diabetes and other chronic conditions [21]. Marrett et al. [12] and Guisasola et al. [13], using the generic Treatment Satisfaction Questionnaire for Medication (TSQM) instrument, noted that hypoglycemic episodes were associated with decreased treatment satisfaction. In the present study, when compared to those without low blood sugar, patients with low blood sugar reported significant decreases in treatment satisfaction on all subscales of the disease specific DiabMedSat tool (burden, efficacy and symptom) as well as the total composite score. Not surprisingly, those with low blood sugar had somewhat higher healthcare resource utilization than those without low blood sugar. This is similar to the results of Pelletier who found significant utilization of physician office visits, laboratory tests and outpatient visits [22]. Although patients with low blood sugar reported significantly greater diabetes-related ER visits and clinician visits after adjusting for several covariates, there were no differences observed in overall ER visits or hospitalization. From this, it would appear that patients with low blood sugar do utilize healthcare resources to some extent for issues related to their diabetes, whether it means going to the ER or to the clinician to monitor clinical progress. The difference in resource use was not as large as expected, based on other published studies. This could be attributed to recall bias and the respondents’ failure to remember details of health care visits over the previous 6-month period. Moreover, as noted earlier, many previous studies examining resource use due to hypoglycemia rely on administrative claims data or examine resource use over a longer time period [15,22]. The current results should be considered within the context of several limitations. NHWS is a patient-reported, cross-sectional, web-based survey. Data were not verified against clinician diagnoses or chart reviews, nor were reports of low blood sugar confirmed by blood glucose monitoring. Additionally, the cross-sectional nature of the analysis rules out a causal inference interpretation of the results. Although results suggest that low blood sugar might lead to decreased HRQL and treatment satisfaction, and increased healthcare resource use, unmeasured variables related to home or personal life and detailed history of healthcare utilization may further explain the observed associations. Further, this study considered patients as having hypoglycemia if they indicated they experienced ‘‘symptoms of low blood sugar,’’ or ‘‘low blood sugar in the middle of the night’’ ‘‘some,’’ ‘‘most’’ or ‘‘all of the time.’’ This definition of hypoglycemia was very specific and may be considered limited, as it does not include other symptoms frequently associated with hypoglycemic episodes. The presentation of
diabetes research and clinical practice 91 (2011) 363–370
symptoms may vary widely among patients and can depend upon each individual’s level of sensitivity. Symptoms may include dizziness, sweating, irritability, nervousness, hunger and headaches, just to name a few [8,9] As a consequence of using a narrower yet specific definition, the prevalence of hypoglycemia may be underreported in this study. Earlier analyses conducted on this data using a broader, though potentially less specific spectrum of signs and symptoms of hypoglycemia, showed that more than 57% of survey respondents experienced at least one sign or symptom that is repeatedly associated with hypoglycemia [23,24]. Nonetheless, the finding that more than 13% of this population reported low blood sugar falls within the range of values reported elsewhere and could be considered a conservative estimate [12]. Lastly, while the Diabetes Symptom Questionnaire does ask the frequency of events, the evaluation of impact on outcomes by symptom frequency was beyond the scope of the current study and was not considered. Future research should evaluate results across the strata of response levels. Finally, the web-based sample may not be representative of the total US diabetic population. Although the overall type 2 diabetes prevalence of NHWS is similar to that of National Health Interview Survey (NHIS) (10.7% vs. 8.4%, respectively), respondents in the current study may have differed from the diabetes population in certain respects, such as income, severity of condition, etc. Caution should be used when generalizing these results to other populations. Nevertheless, even among this web-based sample (which may be slightly more educated with potentially more access to resources than the overall diabetes population), there was a substantial impact of hypoglycemia symptoms on health outcomes. In summary, a sizeable percentage of patients with selfreported type 2 diabetes reported having low blood sugar that could be indicative of experiencing hypoglycemic episodes. Among this sample of established OAD-treated type 2 diabetes patients, low blood sugar was found to be significantly associated with reduced HRQL, decreased medication satisfaction, and increased diabetes-related healthcare resource utilization. These observations highlight the influential effect that hypoglycemia can have on patients’ daily lives, treatment satisfaction, and resource utilization. According to the recent AACE/ACE guidelines, the risk of hypoglycemia may warrant specific choices in therapy and reevaluation of therapy goals to minimize the impact of hypoglycemic episodes on patient lives, morbidity, and mortality [6]. Reducing the frequency or severity of hypoglycemic episodes through individualized care and utilizing therapies with lower the risk of hypoglycemia may, in turn, improve glycemic control by increasing treatment satisfaction and quality of life. Future research should prospectively examine the association between risk factors for hypoglycemia and its impact on subsequent outcomes among an OAD-treated population. Answers to these questions may assist clinicians in their selection of appropriate treatment options for their patients. Treatment strategies that provide effective glycemic control, with a lower potential for inducing hypoglycemia should be considered in the management of type 2 diabetes.
369
Conflict of interest The authors have a competing interest to declare. Grant support: The National Health and Wellness Survey (NHWS) is conducted by Consumer Health Sciences/KantarHealth, Princeton, NJ. AstraZeneca, Wilmington, DE licensed access to NHWS and funded the analysis and preparation of this paper.
references
[1] Mainous 3rd AG, Baker R, Koopman RJ, Saxena S, Diaz VA, Everett CJ, et al. Impact of the population at risk of diabetes on projections of diabetes burden in the United States: an epidemic on the way. Diabetologia 2007;50(5):934–40. [2] Centers for disease control and prevention, National Diabetes Factsheet 2007, http://www.cdc.gov/diabetes/ pubs/pdf/ndfs_2007.pdf; 2007 [accessed 9.10.09]. [3] Pouwer F, Hermanns N. Insulin therapy and quality of life: a review. Diabet Metab Res Rev 2009;25(Suppl. 1):S4–10. [4] American Diabetes Association. Standards of medical care in diabetes – 2009. Diabet Care 2009;32(Suppl. 1):S13–61. [5] Cheung BM, Ong KL, Cherny SS, Sham PC, Tso AW, Lam KS. Diabetes prevalence and therapeutic target achievement in the United States, 1999–2006. Am J Med 2009;122(5):443–53. [6] Rodbard HW, Jellinger PS, Davidson JA, Einhorn D, Garber AJ, Grunberger G, et al. Statement by an American Association of Clinical Endocrinologists/American College of Endocrinology consensus panel on type 2 diabetes mellitus: an algorithm for glycemic control. Endocr Pract 2009;15(6):540–59. [7] Amiel SA, Dixon T, Mann R, Jameson K. Hypoglycaemia in type 2 diabetes. Diabet Med 2008;25(3):245–54. [8] Gabriely I, Shamoon H. Hypoglycemia in diabetes: common, often unrecognized. Cleve Clin J Med 2004;71(4):335–42. [9] Childs BP, Clark NG, Cox DJ, Cryer PE, Davis SN, DiNardo MM, et al. Defining and reporting hypoglycemia in diabetes – a report from the American Diabetes Association Workgroup on Hypoglycemia. Diabe Care 2005;28(5):1245–9. [10] Hay L. Unrecognized hypo and hyperglycemia in wellcontrolled patients with type 2 diabetes mellitus: the results of continuous glucose monitoring. Diab Technol Ther 2003;5(1):19–26. [11] Miller CD, Phillips LS, Ziemer DC, Gallina DL, Cook CB, El-Kebbi IM. Hypoglycemia in patients with type 2 diabetes mellitus. Arch Intern Med 2001;161(13):1653–9. [12] Marrett E, Stargardt T, Mavros P, Alexander CM. Patientreported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain. Diabet Obes Metab 2009;11(12):1138–44. [13] Guisasola FA, Povedano ST, Krishnarajah G, Lyu R, Mavros P, Yin D. Hypoglycaemic symptoms, treatment satisfaction, adherence and their associations with glycaemic goal in patients with type 2 diabetes mellitus: findings from the Real-Life Effectiveness and Care Patterns of Diabetes Management (RECAP-DM) Study. Diabet Obes Metab 2008;10(Suppl. 1):25–32. [14] Davis RE, Morrissey M, Peters JR, Wittrup-Jensen K, Kennedy-Martin T, Currie CJ. Impact of hypoglycaemia on quality of life and productivity in type 1 and type 2 diabetes. Curr Med Res Opin 2005;21(9):1477–83. [15] Bullano MF, Al-Zakwani IS, Fisher MD, Menditto L, Willey VJ. Differences in hypoglycemia event rates and associated
370
[16]
[17]
[18]
[19]
[20]
diabetes research and clinical practice 91 (2011) 363–370
cost-consequence in patients initiated on long-acting and intermediate-acting insulin products. Curr Med Res Opin 2005;21(2):291–8. Gazmararian JA, Ziemer DC, Barnes C. Perception of barriers to self-care management among diabetic patients. Diabet Educ 2009;35(5):778–88. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new methods of classifying prognostic comorbidity in longitudinal studies: develop and validation. J Chron Dis 1987;40(5):373–83. Morisky DE, Green LW, Levine DM. Concurrent and predictive validity of a self-reported measure of medication adherence. Med Care 1986;24(1):67–74. Brod M, Skovlund SE, Wittrup-Jensen KU. Measuring the impact of diabetes through patient report of treatment satisfaction, productivity and symptom experience. Qual Life Res 2006;15(3):481–91. EuroQoL Group. EuroQol – a new facility for the measurement of health-related quality of life. Health Policy 1990;16(3):199–208.
[21] Walters SJ, Brazier JE. Comparison of the minimally important difference for two health state utility measures: EQ-5D and SF-6D. Qual Life Res 2005;14(6): 1523–32. [22] Pelletier EM, Smith PJ, Boye KS, Misurski DA, Tunis SL, Minshall ME. Direct medical costs for type 2 diabetes mellitus complications in the US commercial payer setting: a resource for economic research. Appl Health Econ Health Policy 2008;6(2–3):103–12. [23] Pollack M, Waterman F, Bolge SC, Williams SA. Satisfaction with diabetes treatments: impacts on patient healthrelated quality of life and productivity. In: Poster presented at International Society for Pharmacoeconomics and Outcomes Research; 2009. [24] Pollack M, Bolge SC, Williams SA, Williams SA. The association between patient-reported diabetes symptoms and tolerability issues of oral antidiabetic agents on work and life productivity. In: Poster presented at International Society for Pharmacoeconomics and Outcomes Research; 2009.