J C E M
O N L I N E
B r i e f
R e p o r t — E n d o c r i n e
C a r e
Linear Association Between Household Income and
Metabolic Control in Children With Insulin-Dependent
Diabetes Mellitus Despite Free Access to Health Care
Johnny Deladoëy, Mélanie Henderson, and Louis Geoffroy
Endocrinology and Diabetes Service, Centre Hospitalier Universitaire Sainte-Justine, Department of
Pediatrics, University of Montréal, Montréal, Canada H3T 1C5
Background: In health care systems with a user fee, the impact of socioeconomic factors on pediatric insulin-dependent diabetes mellitus (IDDM) control could be due to the cost of accessing care.
Hypothesis: There is a linear association between household income and the average glycosylated
hemoglobin (HbA1c) of children and adolescents with IDDM despite free access to health care.
Methods: We used a linear regression model to examine the association between normalized
average HbA1c of 1766 diabetic children (diagnosed at our institution from 1980 to 2011 before
17 years of age) and the median household income of their neighborhoods (obtained from Statistics Canada, 2006 Census data).
Results: We found a negative linear association (P ⬍ .001; r ⫽ ⫺0.2) between the level of income
and metabolic control assessed by HbA1c after controlling for sex, age at diagnosis, duration of
diabetes, ethnicity, geographical factors, frequency of visits, current age (as a proxy for change in
practice over time), and change of measurement methods of HbA1c across time. For every increase
of $15 000 in annual income, HbA1c decreased by 0.1%.
Conclusion: We report a linear association of household income with metabolic control of IDDM
in childhood. Given that Canada has a system of free universal access to health care, confounding
by access to care is unlikely. Considering the impact of poorly controlled IDDM in childhood on the
development of long-term complications, our findings suggest that the higher complication rate
found in adults of low socioeconomic status might originate from the poor control that they
experienced in childhood. Support for the care of IDDM children from low-income neighborhoods
should be increased. (J Clin Endocrinol Metab 98: E882–E885, 2013)
he Diabetes Control and Complication Trial demonstrated the importance of lowering glucose and glycosylated hemoglobin (HbA1c) levels to reduce vascular
complications (1, 2). However, socioeconomic factors
negatively affect metabolic control in children with insulin-dependent diabetes mellitus (IDDM) (3–5). These factors are often aggregated into different indices or distinct
income groups (4, 6 –9), making cross-study comparisons
difficult. In contrast, the median household income, usually by neighborhood, is a good proxy of socioeconomic
status (SES) and is thoroughly collected by census agencies
T
in most industrialized countries. Indeed, the global development of children and adolescents is seriously hampered
by the deleterious effect of low-income neighborhoods
(10), and this negative effect seems to continue into adulthood, even if the individuals increase their income (11).
However, there are surprisingly few data on a possible
direct linear relationship of household income by neighborhood with metabolic control of children and adolescents with IDDM because most studies have analyzed the
association of income grouped by categories (4 – 6). Importantly, most available studies (4 –7) have been carried
ISSN Print 0021-972X ISSN Online 1945-7197
Printed in U.S.A.
Copyright © 2013 by The Endocrine Society
Received January 23, 2013. Accepted March 4, 2013.
First Published Online March 28, 2013
Abbreviations: CI, confidence interval; HbA1c, glycosylated hemoglobin; IDDM, insulindependent diabetes mellitus; SES, socioeconomic status.
E882
jcem.endojournals.org
J Clin Endocrinol Metab, May 2013, 98(5):E882–E885
doi: 10.1210/jc.2013-1212
doi: 10.1210/jc.2013-1212
jcem.endojournals.org
out in areas in which a user fee is required for access to a
health care professional. Our hypothesis was that there is
a linear association between household income and the
average HbA1c of children and adolescents with IDDM
despite free access to health care.
Materials and Methods
Data collection
We obtained the HbA1c values of patients diagnosed with
IDDM at our institution between 1980 and 2011 before their
17th birthday. Patients followed up for less than 1 year and with
fewer than 3 available HbA1c values were excluded, as were
patients in the remission phase (average HbA1c of less than 6%
within the first 2 years after diagnosis) (Table 1). A median of 15
values of HbA1c per patient was available to calculate the average HbA1c (range of 3– 66; 29 862 values for 1766 patients).
Annual median household incomes by neighborhood were available by postal code on the Statistics Canada web site (2006 Census at http://www.statcan.gc.ca/) and were used as a proxy for
family income. Median household income has been used in many
clinical settings as a surrogate for SES (4, 12, 13). The study was
approved by the Head of Medical and Academic Affairs of the
Ste-Justine Hospital. The study was determined to pose minimal
risk to participants, and no informed consent was required.
Cohort description and statistical analysis
Breakdown of study subjects through exclusion criteria and
categorical variables are presented in Table 1. The mean (⫾SD)
age at diagnosis was 8.5 (⫾4.2) years, and the duration of follow-up was 6.9 (⫾3.9) years. Two different techniques of HbA1c
measurement were used over time. Before February 2003,
HbA1c was measured with an immunological method and had
a mean (⫾SD) of 9.0% (⫾1.2). After February 2003, HbA1c was
measured with HPLC and had a mean (⫾SD) of 8.3% (⫾1.0).
E883
Therefore, to allow pooling results from these 2 periods, HbA1c
values were standardized for each period and then normalized on
the current average and SD. We next used linear regression [controlled for sex, age at diagnosis, duration of diabetes, ethnicity,
geographical factor (urban vs suburban/rural), frequency of visits, current age as a proxy for change in practice over time, and
change of methods of HbA1c measurement across time, ie, before and after 2003] to examine the association between the
normalized average HbA1c and neighborhood median household income. We ascertained whether our models met the underlying assumptions of linear regression by plotting residuals
against all continuous independent predictors to ensure that
there was no systematic pattern to the residuals. To assess that
our model was correctly specified, we also used several tests
including the Ramsey RESET test. To account for positive skewness of the income distribution, we also assessed the validity of
our model after exclusion of outliers. Statistical analysis, calculation, and graphic production of effects were performed with
the statistical program R (http://www.r-project.org/) (14).
Results
We found a negative linear association (P ⬍ .001; r ⫽
⫺0.2) between the level of income by neighborhood and
metabolic control assessed by HbA1c (n ⫽ 1766) after
controlling for sex, age at diagnosis, duration of diabetes,
ethnicity, frequency of visits and time period (before and
after 2003, and current age of the patient). For every increase of $15 000 in annual income, HbA1c decreased by
0.1%. Indeed, the effect size between the highest
($233 792) and lowest ($26 020) median income was
more than 1% HbA1c [from 7.20% (95% confidence interval [CI] 6.75–7.65%) to 8.55% (95% CI 8.45–
8.65%)] (Figure 1). The association also remains valid
Table 1. Breakdown of Study Subjects Through
Exclusion Criteria and Categorical Variables
Number
Excluded
Potential pool of study subjects
Exclusion criteria
Followed up for less than a year
and fewer than 3 HbA1cs
Average HbA1c less than 6%
Incomplete postal code
Ethnicity
Caucasian
Black
Asian
Hispanic
Not specified
Sex
Male
Female
Geographical factors
Urban area
Suburban/rural
Number
Remaining
2426
640
1786
11
9
1775
1766
649
27
4
14
1072
950
816
577
1189
Figure 1. Effect of the median household income on the average
HbA1c of children and adolescents with IDDM. The vertical axis is
labeled as Z-score (left) and percentage scale (right) of HbA1c, and a
95% pointwise CI is drawn around the estimated effect. Short lines on
the x-axis represent the individual income data.
E884
Deladoëy et al
Household Income and IDDM Control in Children
(P ⬍ .001) and conserves its strength (r ⫽ ⫺0.2, $15 000
per 0.1% HbA1c) after exclusion of the 10 poorest and the
10 richest households or after exclusion of the 10 highest
and the 10 lowest HbA1c values (P ⬍ .001, r ⫽ ⫺0.2,
$15 000 per 0.07% HbA1c).
Discussion
To our knowledge, this is the first description of a linear
association between household income and metabolic
control in pediatric IDDM despite free access to health
care, which makes confounding by access to care unlikely.
An impact of SES on health outcomes has not been consistently found for all pathologies in children. For example, in Ontario, where a system of universal health care
also exists, SES was not associated with risk of death in
children treated for Hodgkin and non-Hodgkin lymphomas (15). The treatment of lymphomas is intensive but
hospital-based and limited in time; once free from disease,
a child does not need daily monitoring and treatment,
which may explain the lack of SES effect on clinical outcome. In contrast, IDDM is a chronic disease, which requires glucose monitoring and insulin injections several
times a day, and throughout life. While a negative association between household income and metabolic control
has previously been reported in youth with IDDM, they
were observed mostly in areas with a user fee for access to
health care professionals and incomes were grouped in
categories (4 – 6). Only two studies from jurisdictions with
an almost free access to care (Germany and Australia)
showed association between metabolic control and SES
reported by categories; however, a linear association was
not reported (8, 9). Herein, the strength of the linear association we found is comparable to that of low-level lead
exposure and intelligence quotient in children (16, 17).
Our study has several limitations. First, it is retrospective. Hence, there may be a number of unknown and unmeasured confounders such as changes in the types of insulin, use of insulin pumps, lack of familial resources (eg,
transportation), inability of the parents to take time off
from work to bring their children to the visit or lower
parental educational levels. Second, data about ethnicity
are limited because they were collected only after 2003.
However, the regression was also controlled for geographical factors (urban vs suburban/rural area); suburban and
rural areas are essentially populated by Caucasians (95%)
whereas urban areas are multiethnic (9% Asians, 8%
Black African, 3.5% Hispanics, 6% other nonspecified
ethnicity) (18). Finally, we used the Canadian 2006 Census and we assumed that differences of income between
J Clin Endocrinol Metab, May 2013, 98(5):E882–E885
neighborhoods had remained stable over our study
period.
Considering the impact of poorly controlled IDDM in
childhood on the development of long-term complications
(19, 20), our findings support the idea that the higher
complication rate found in adults of low SES with IDDM
might originate from the poorer control that they experienced in their childhood (2, 21). Indeed and based on the
Diabetes Control and Complication Trial results (1), our
observed difference of more than 1% HbA1c between the
lowest (8.55%) and highest (7.20%) incomes corresponds
to a doubling of the risk of retinopathy. Support for the
care of IDDM children from low-income neighborhoods
should be increased, such as increased implication of social workers and increased time allocated for these patients during the follow-up visits, but studies are required
to assess the efficacy of these approaches. Nevertheless,
our results together with other studies (4 – 6, 8 –10) raise
more global policy questions and may suggest that the
most powerful intervention would be to raise the income
of the poorest families with IDDM children. It remains
unclear whether this financial intervention will be effective
and ultimately will save money to the society through the
expected decrease of diabetic complications (22). With the
widening income gap observed over the last years in North
America (11), the future well-being of disadvantaged
IDDM children is a policy challenge (23).
Acknowledgments
We thank Dr Jocelyne Cousineau (Clinical Biochemistry, Centre
Hospitalier Universitaire Sainte-Justine, University of Montréal,
Montréal, Canada) for her collaboration and Dr Guy Van Vliet
for reviewing the manuscript.
Address all correspondence and requests for reprints to:
Johnny Deladoëy, MD, PhD, Centre Hospitalier Universitaire
Sainte-Justine, 3175 Côte Sainte-Catherine, Montréal, Québec,
Canada H3T 1C5. E-mail: [email protected].
J.D. is supported by the Girafonds/Fondation du Centre Hospitalier Universitaire Sainte-Justine.
Disclosure Summary: All authors have nothing to disclose.
References
1. The Diabetes Control and Complications Trial Research Group.
The effect of intensive treatment of diabetes on the development and
progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med. 1993;329:977–986.
2. The Diabetes Control and Complications Trial Research Group.
Effect of intensive diabetes treatment on the development and progression of long-term complications in adolescents with insulin-dependent diabetes mellitus: Diabetes Control and Complications
doi: 10.1210/jc.2013-1212
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
Trial. Diabetes Control and Complications Trial Research Group.
J Pediatr. 1994;125:177–188.
Limenis E, Shulman R, Daneman D. Is the frequency of ketoacidosis
at onset of type 1 diabetes a child health indicator that is related to
income inequality? Diabetes Care. 2012;35:e5.
Springer D, Dziura J, Tamborlane WV, Steffen AT, Ahern JH, Vincent M, Weinzimer SA. Optimal control of type 1 diabetes mellitus
in youth receiving intensive treatment. J Pediatr. 2006;149:227–
232.
Gallegos-Macias AR, Macias SR, Kaufman E, Skipper B, Kalishman
N. Relationship between glycemic control, ethnicity and socioeconomic status in Hispanic and white non-Hispanic youths with type
1 diabetes mellitus. Pediatr Diabetes. 2003;4:19 –23.
Cutfield SW, Derraik JG, Reed PW, Hofman PL, Jefferies C, Cutfield
WS. Early markers of glycaemic control in children with type 1
diabetes mellitus. PLoS One. 2011;6:e25251.
Tahirovic H, Toromanovic A. Glycemic control in diabetic children:
role of mother’s knowledge and socioeconomic status. Eur J Pediatr.
2010;169:961–964.
Galler A, Lindau M, Ernert A, Thalemann R, Raile K. Associations
between media consumption habits, physical activity, socioeconomic status, and glycemic control in children, adolescents, and
young adults with type 1 diabetes. Diabetes Care. 2011;34:2356 –
2359.
Hatherly K, Smith L, Overland J, et al. Glycemic control and type 1
diabetes: the differential impact of model of care and income. Pediatr
Diabetes. 2011;12:115–119.
Brooks-Gunn J, Duncan GJ, Klebanov PK, Sealand N. Do neighborhoods influence child and adolescent development? Am J Sociol.
1993;99:353–395.
Ludwig J, Duncan GJ, Gennetian LA, et al. Neighborhood effects on
the long-term well-being of low-income adults. Science. 2012;337:
1505–1510.
Gould JB, Davey B, Stafford RS. Socioeconomic differences in rates
of cesarean section. N Engl J Med. 1989;321:233–239.
jcem.endojournals.org
E885
13. Pappas G, Hadden WC, Kozak LJ, Fisher GF. Potentially avoidable
hospitalizations: inequalities in rates between US socioeconomic
groups. Am J Public Health. 1997;87:811– 816.
14. Fox J. Effect displays in R for generalised linear models. J Stat Softw.
2003;8:1–9.
15. Darmawikarta D, Pole JD, Gupta S, Nathan PC, Greenberg M. The
association between socioeconomic status and survival among children with Hodgkin and non-Hodgkin lymphomas in a universal
health care system [published online November 5, 2012]. Pediatr
Blood Cancer. doi:10.1002/pbc.24386.
16. Needleman HL, Gatsonis CA. Low-level lead exposure and the IQ
of children. A meta-analysis of modern studies. JAMA. 1990;263:
673– 678.
17. Grosse SD, Matte TD, Schwartz J, Jackson RJ. Economic gains
resulting from the reduction in children’s exposure to lead in the
United States. Environ Health Perspect. 2002;110:563–569.
18. Stoppa-Vaucher S, Van Vliet G, Deladoey J. Variation by ethnicity
in the prevalence of congenital hypothyroidism due to thyroid dysgenesis. Thyroid. 2011;21:13–18.
19. Nathan DM, Cleary PA, Backlund JY, et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes.
N Engl J Med. 2005;353:2643–2653.
20. White NH, Cleary PA, Dahms W, Goldstein D, Malone J, Tamborlane WV. Beneficial effects of intensive therapy of diabetes during
adolescence: outcomes after the conclusion of the Diabetes Control
and Complications Trial (DCCT). J Pediatr. 2001;139:804 – 812.
21. Conway BN, May ME, Signorello LB, Blot WJ. Mortality experience of a low-income population with young-onset diabetes. Diabetes Care. 2012;35:542–548.
22. Anderson LM, Charles JS, Fullilove MT, Scrimshaw SC, Fielding JE,
Normand J. Providing affordable family housing and reducing residential segregation by income. A systematic review. Am J Prev Med.
2003;24:47– 67.
23. Sampson RJ. Sociology. Moving and the neighborhood glass ceiling.
Science. 2012;337:1464 –1465.