Article Text
PDF
Abstract
Introduction Very little is known about the influence of socioeconomic status on type 1 diabetes mellitus (T1DM) complications. Our aim was to determine whether socioeconomic level is a risk factor for the development of diabetic retinopathy (DR) in patients with T1DM.
Research design and methods A cohort of 150 patients with T1DM were studied prospectively over 9 years. Socioeconomic status was assessed using a neighborhood-level measure based on an index of deprivation. The contribution of other variables such as hypertension, dyslipidemia, diabetic nephropathy and smoking habit was evaluated. Cox proportional hazards models were used to quantify the associations.
Results The incidence of DR was 21.6 cases per 1000 patient-years. Multivariable analyses showed that for each percentage point increase in glycated hemoglobin (HbA1c), the risk of developing DR increased by 58% (HR 1.58, 95% CI 1.19 to 2.10).
Patients with T1DM onset >18 years of age and resident in areas of lower socioeconomic levels presented with almost triple the risk of developing DR (HR 2.95, 95% CI 1.08 to 8.00) compared with those with onset <18 years of age and resident in less deprived areas. We did not find significant relationships with other variables studied such as hypertension, dyslipidemia, diabetic nephropathy and smoking habit.
Conclusions Low socioeconomic level is a risk factor, independent of glycemic control, in the development of DR in patients with T1DM when the onset of diabetes is in adulthood. This finding indicates that socioeconomic status and age of onset need to be considered in population screening for DR in patients with T1DM.
- diabetes mellitus
- type 1
- diabetic retinopathy
- social support
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Significance of this study
What is already known about this subject?
Although socioeconomic status is related to important chronic diseases in general and type 2 diabetes mellitus in particular, very little is known about its influence on complications in patients with type 1 diabetes mellitus.
What are the new findings?
In our study we observed a clear association between the level of socioeconomic deprivation and the incidence of diabetic retinopathy, independent of the level of the patient’s glycemic control.
How might these results change the focus of research or clinical practice?
It is essential to develop and promote new social policies centered in patients with type 1 diabetes mellitus with the objective of reducing the consequences of social inequalities in groups with low socioeconomic levels.
Introduction
Diabetes mellitus (DM) comprises several clinical entities with different etiologies, but which has hyperglycemia as its principal characteristic.1 The worldwide prevalence of DM in 2017 was estimated at around 425 million persons, and this level is expected to rise by 48% by 2045 to reach a level of 629 million.2 Of the total number of patients with DM, approximately 5%–10% suffer type 1 diabetes mellitus (T1DM) (1 in every 300–500 people <18 years of age).
Diabetic retinopathy (DR) is the principal cause of acquired blindness in developed countries, and represents 5% of all causes of acquired blindness worldwide. One in every three patients with DM have some grade of DR, and is three times more prevalent in patients with T1DM than with type 2 diabetes mellitus (T2DM).3 The duration of the disease and metabolic (glycemic) control have been traditionally considered the principal risk factors associated with the appearance and progression of DR in patients with DM. Other factors such as hypertension, dyslipidemia, diabetic nephropathy, and the smoking habit have been described as having some involvement as well.3–6
Socioeconomic factors influencing the development and progression of some clinical processes have been well documented, such as in cardiovascular diseases.7 8 In DM, data indicate that a low socioeconomic level is associated with a poorer clinical evolution of the complications specific to DM, such as higher morbidity and mortality,9 10 even in countries with universal healthcare systems.
A low socioeconomic status has been associated with delayed diagnosis and poorer outcomes in cases of amblyopia,11 macular edema12 and glaucoma.13 14 Low economic status has also been associated with higher DR prevalence and more advanced forms of the disease. This has been attributed to lower participation in screening programs,15–17 as well as lower likelihood of being routinely followed up.
The principal objective of our study was to analyze the relationship between socioeconomic level and the development of DR in a cohort of patients with T1DM, without DR at the start of the follow-up. Other influencing factors taken into account were the level of metabolic control, hypertension, dyslipidemia, diabetic nephropathy, and the smoking habit.
This project was implemented in southern Spain (Andalusia) within the context of a universal public health service. The comprehensive program (‘Integrated Plan for Diabetes in Andalusia’) includes early detection of DR using digital retinography.18 19
Research design and methods
Study ambit
The study was developed within the catchment area of the Hospital Universitario Puerta del Mar de Cádiz (Andalusia, Spain) which provides healthcare for 247 000 citizens, of whom approximately 1500 are patients with T1DM.
Study design
The prospective cohort of patients with T1DM belonged to the Early Diabetic Retinopathy Detection Plan (Plan de Detección Precoz de la Retinopatía Diabética) of Andalusia conducted using digital imaging or retinography, and with a maximum period of follow-up of 9 years.
Study subjects
A total of 150 patients with T1DM, without DR at the start of the follow-up, were recruited into the study. Of these, 78 had early-onset (0–18 years of age) T1DM, and 72 at a later age (>18 years).20 The duration of the follow-up of the cohort was for 9 years, between 2008 and 2017.
DM diagnosis was performed according to American Diabetes Association guideline.21 When the patients had equivocal signs and symptoms of T1DM (such as ketoacidosis) at the start of the study, diagnosis was confirmed by the presence of autoantibodies (antiglutamic decarboxylase and anti-IA2) and decrease in fasting C-peptide (<0.5 ng/dL) and fasting plasma glucose >250 mg/dL or following intravenous glucagon-stimulated C-peptide test.22
Inclusion criteria
Patients with T1DM >14 years of age, without DR at the start of the follow-up and who are within the catchment area of the Puerta del Mar University Hospital (Cádiz).
Exclusion criteria
Retinography with signs of other ocular disease, or non-evaluable retinography, or type of DM other than T1DM.
Variables recorded in the study
Dependent variables
Grade of DR: Defined and classified according to the International Clinical Diabetic Retinopathy Disease Severity Scale. In case of different grades of severity in the two eyes, we accepted the eye with the greater severity as reference.
Duration of the T1DM: Numbers of years between the age of onset of the T1DM and the age when the retinography first showed signs of DR. For patients not developing DR, the duration was considered up to the last retinography performed in the follow-up.
Independent variables
Age at onset of T1DM: Data collected from the electronically stored patient history.
Grade of metabolic control: Determined as the level of glycated hemoglobin (HbA1c) (mmol/L and %). For the statistical analyses we used the baseline HbA1c and mean HbA1c obtained from the annual determinations recorded during the follow-up.
Arterial hypertension: Dichotomous variable. The patient was considered to be hypertensive if this was recorded (>140/90 mm Hg) at DM diagnosis, or patients receiving any antihypertension medication.
Dyslipidemia: Dichotomous variable. Defined as the presence of high low-density lipoprotein cholesterol (LDL cholesterol >100 mg/dL) and/or hypertriglyceridemia (triglycerides >200 mg/dL)23 or patients being in receipt of hypolipidemic medication, recorded at DM diagnosis.
Diabetic nephropathy: Dichotomous variable. Defined as the presence of renal damage by direct method (renal biopsy), albumin excretion ≥30 mg/g creatinine (urinary albumin/creatinine ratio) or ≥30 mg/24 hours confirmed in at least two or three samples in the previous 6 months, or glomerular filtration of <60 mL/min/1.73 m2, calculated with the Cockcroft-Gault equation for conventional (24-hour urine) creatinine clearance, or isotopic methods.
Smoking habit: Dichotomous variable. Defined by current consumption, or consumption in the year prior to the start of the follow-up period.
Socioeconomic status: To measure the socioeconomic status of the cases, an artificial index of deprivation of the census tract of residence was used. The index has been applied previously in epidemiological research on inequalities in mortality and morbidity in Andalusia.24 25
The index is calculated for each census tract using three census variables: (1) the percentage of persons (both genders) with low levels of education (unable to read or write, or <5 years of conventional schooling) in the general population ≥16 years of age; (2) the percentage of unemployed people (unemployed population ≥16 years divided by the actively employed population ≥16 years); and (3) the percentage of unskilled workers (unskilled population ≥16 years divided by the employed population ≥16 years). A principal component analysis was carried out with the standardized values of the three variables to calculate the index. The necessary conditions for its application were verified by the Bartlett sphericity test and the Kaiser-Meyer-Olkin sampling adequacy measure. The census tracts were ranked and categorized into five groups according to quintiles. Level 1 represents the areas with the least deprivation and level 5 represents the sectors with the highest levels of deprivation in the population.
Follow-up
Annually, from the date of inclusion in the study, three images of each eye were made with a non-mydriatic digital retinography (Topcon NW-200). One of the images centered in the macula (central), another in the optic disc (nasal) and another temporal to the macula (temporal). All images were reviewed independently by two DR expert ophthalmologists (authors SJC, PAM).
The first retinography free of DR was considered as the basal level. The detection of any grade of DR requiring ophthalmological treatment, or other alterations that required specialist evaluation, signaled the transfer of the patient out of the follow-up study.
Statistical analyses
The baseline characteristics of the patients included in the study were established and the cumulative incidence and the incidence rate of DR were calculated. Person-years of follow-up were estimated for each case by computing the time between the date of onset and the date of last follow-up, with respect to the development of DR or the completion of the study. The risk factors associated with the development of DR were compared. Initially, the distributions of these factors between those patients who did develop DR were compared with those who did not develop DR using χ2 test for categorical variables and the Mann-Whitney U test for quantitative variables. The incidence rates of subgroups of each of the risk factors were calculated, as were the incidence rate ratios with their 95% CIs. Finally, once the most important covariables were identified, regression models were developed using Cox proportional hazards models to calculate the adjusted HR of each of the variables, using years from diagnosis of T1DM up to the event as the time variable. In the final step, interaction between the included variables was tested and added to the model in case they yielded statistical significance.
Results
Demographic variables
The baseline characteristics of the patients are presented in table 1. Of the total patients in the screening program, 150 (78/150 women) without DR were incorporated in the follow-up program. The mean age at the start of the study was 31.6±11.6 years, with a mean age at onset of T1DM of 19.9±11.3 years, and a mean duration of DM of 17±8.9 years of clinical evolution of DM at the start of the follow-up period. The mean HbA1c was 7.7% (60 mmol/mol), and 68.7% of the patients had HbA1c >7% (53 mmol/mol). Hypertension was present in 15.3%, dyslipidemia in 16.7%, diabetic nephropathy in 6.7% and smoking habit in 24.7%. Nearly 16% of the patients (n=23) were resident in socioeconomic-deprived (grades 4 and 5) census tracts (table 1).
Baseline characteristics of the patients at the start of the study (n=150)
Cumulative incidence and incidence rate of DR
At the end of the follow-up period, of the 150 patients studied (2250 patient-years), 55 had presented with DR, implying a cumulative incidence of 36.6% and an incidence rate of 21.6/1000 patient-years. Of these 55 patients with DR, 30 were slight and 25 were of moderate DR, implying an incidence of 20.0% and 16.6%, respectively. We did not observe any severe case of DR at the end of the follow-up period.
Risk factors associated with DR development
Table 2 summarizes the principal characteristics studied in patients who developed DR, as well as those who did not develop DR over the long-term follow-up. In the bivariate analysis, we observed that the patients with DR were younger at the onset of T1DM (15.7±11.5 vs 22.3±10.5 years; p<0.050), had a longer duration of the disease (15.4±9.5 vs 9.7±7.2 years; p<0.050), higher values of basal HbA1c (7.8%±1.1% (62 mmol/mol) vs 7.6%±1.3% (59 mmol/mol); p<0.05) and mean HbA1c (7.8%±1.1% (62 mmol/mol) vs 7.6%±1.1% (60 mmol/mol); p=0.060) over the period of follow-up.
Distribution of vascular disease risk factors in the population studied
Table 3 summarizes the incidence rate for each socioeconomic category, and the incidence rate ratio of DR of the different variables studied. The patients with HbA1c >7% had an incidence rate of 24.0 cases/1000 patient-years, while the rate was 14.7 cases/1000 patient-years in those with HbA1c <7% (p=0.078). With respect to the socioeconomic variable, the DR incidence rate in those with levels 4 and 5 (higher deprivation) compared with those with levels 1–3 (less deprivation) was 25.7 cases/1000 patient-years vs 21.4 cases/1000 patient-years; p=0.291.
Incidence rate and the incidence rate ratio of DR
The results obtained with the Cox proportional hazards regression (table 4) showed that, in the final model, the variables that explained the most risk of developing DR were: HbA1c, age of onset of DM, and level of deprivation associated with the residence census tract. The proportional hazards model indicates that these three variables have independent influences on the log hazard function describing the risk of DR.
Cox proportional hazards model for the development of DR in patients with T1DM
Following adjustment for covariates in the model, HbA1c was significantly associated with developing DR (hazard rate: 1.6; 95% CI 1.19 to 2.10), that is, for every point increase in the mean value of HbA1c there is an increase of 60% in risk of DR. Also, the adjusted model identified a significant interaction between the level of deprivation and the age of onset of DM. Compared with the reference level (onset of DM <18 years of age and resident in less socioeconomically deprived area), the group with later onset (>18 years of age) and resident in the area of highest socioeconomic deprivation had triple the risk of developing DR (HR 2.95; 95% CI 1.08 to 8.00). No significant associations of the appearance of DR were observed with the rest of the combinations of age of onset and socioeconomic status.
Additionally, to rule out a potential role of HbA1c in mediating the effect of socioeconomic status on the development of DR, we calculated the mean and SD of HbA1c in each of the six subgroups; very similar results were observed (onset <18 years and less deprived=7.94±1.16; onset <18 years and medium deprivation=7.74±0.98; onset <18 years and deprived=7.98±1.47; onset >18 years and less deprived=7.40±0.61; onset >18 years and medium deprivation=7.63±1.42; onset >18 years and deprived=7.85±1.10).
Discussion and conclusions
In our study we observed a clear association between the level of socioeconomic deprivation and the incidence of DR, independent of the level of the patient’s glycemic control. This relationship is evident when the results are analyzed as a function of onset of DM. That is, patients who had an onset at >18 years of age and from a lower socioeconomic area present with an almost tripled level of risk of developing DR than those who developed DM in infancy or adolescence and from areas of less deprivation. However, in the rest of the combinations of age of onset and socioeconomic levels, we did not find any significant associations.
The finding of an association between low socioeconomic level and the development of DR concurs with other studies published to date by various research groups, and in different geographic areas. The relationship between education level and the development of DR was described by Klein et al26 who observed a higher risk of DR in women with a low education level, coexisting with early onset of DM. Two subsequent studies conducted in the UK also described associations between low education level and a high prevalence of DR.27 28 Our results are similar to those obtained in a study conducted in China in which a higher education level and higher net income were associated with reduced DR incidence.29
A study conducted in France analyzed the correlation between the index of individual deprivation (‘EPICES scores’) and glycemic control with the associated complications of DM. The authors found that the prevalence of DR was higher in patients with lower socioeconomic level (mean 3.66; 95% CI 1.39 to 9.64; p=0.009) and that this association persisted following adjustment for glycemic control.30 Several studies in the UK correlated low socioeconomic level with DR. One of these studies observed an association between low socioeconomic level (measured as the ‘Index of Multiple Deprivation’) and the presence of DR susceptible to treatment.15 Another study showed similar outcome for T1DM17 but no significant association for T2DM, and another showed the presence of this association with proliferative DR.16 Such associations detected were shown to be independent of the duration of the disease, the values of HbA1c, the concentrations of lipoproteins, and presence of hypertension.
Only in the study published by Klein et al26 was there a difference observed between the onset of DM and the influence of socioeconomic factors in the development of DR. In this case, a greater risk was observed in patients with onset <18 years of age, which is contrary to that observed in our study in which the risk is increased in those patients who had onset >18 years of age.
Several hypotheses have been proposed to explain the relationship between socioeconomic deprivation and the increased risk of DR. Different factors have been invoked such as poor access to healthcare services because of the existence of economic barriers, or even factors associated with a clear deterioration of the quality of life. One postulation is that environmental factors in deprived areas can unlink immune responses in individuals genetically predisposed to developing T1DM, which can accelerate the appearance, or progression, of complications of the disease.31 Another postulation is that of a lower participation in primary prevention screening programs in more deprived areas15 which could increase the risk of more advanced forms of DR being identified in later diagnosis. Further, it is important to consider the impact of the complications of DM on the education level and working practices, especially when DR complications are involved. In our case, the high healthcare cover in the population (especially pediatric services) could explain the lower influence of socioeconomic status of the patients with early onset. The education level could be another factor explaining the interaction between onset of DM and the socioeconomic level since there would not be as much influence in the pediatric population, as there may be in adulthood, that is, in our universal healthcare system, the neonates are followed up with more attention than subsequent adolescence or adulthood when the uptake of healthcare facilities is more a function of the individual’s choice.
We observed an incidence rate of 21.6 cases/1000 patient-years and a cumulative incidence of various grades of DR of 36.6%. To the best of our knowledge, there has been only one study of incidence density of DR in cohorts of patients with T1DM. The study had observed 97.7 cases/1000 person-years.32 This is a finding that is much higher than ours. In the UK, disparate cumulative incidences between 47% and 97% have been published by Wisconsin Epidemiologic Study of Diabetic Retinopathy.33–36 Other studies, as well, have observed incidence levels much lower than ours, of between 20% and 35%.37–39 To compare these studies we must take into account two fundamental factors jointly: the duration of the T1DM at the start of the follow-up, and the grade of glycemic control. DR is time dependent and several studies show that with a sufficiently long follow-up time, most patients with T1DM will develop DR, so that the time of evolution of DM can be considered the main risk factor for DR.36 40
The other fundamental factor influencing the incidence of DR is glycemic control. Patients with poor metabolic control present with higher incidences33 34 41 than those with better control.37 42 In our cohort of patients, those who developed DR presented with a higher baseline HbA1c than those who did not develop DR, and in multivariable analyses, the mean HbA1c levels were shown to be an independent risk factor for the development of DR; with no significant differences noted with respect to sex. These data are concordant with those published in other studies that showed HR between 1.237and 4.2.43
We did not observe any significant association in the development of DR with respect to the other risk factors such as hypertension, dyslipidemia, diabetic nephropathy, and the smoking habit. In similar studies, an association was observed between hypertension and DR especially between diastolic pressure and proliferative DR.35 41 With respect to dyslipidemia, Romero-Aroca et al observed significant association between the ApoB/ApoA1 ratio and the LDL to high-density lipoprotein cholesterol ratio and DR.40 42 Similarly, significant associations between DR and diabetic nephropathy6 and smoking44 have been communicated. This disparity in the findings could be due to several causes including the heterogeneity of the study sample, and the duration of DM follow-up. The population studied was characterized by its youth, as well as the absence, almost, of comorbidities.
Of importance is our novel finding that there is a significant relationship between economic deprivation and incidence of DR in patients with T1DM. Our results suggest that socioeconomic levels need to be taken into account as a risk factor when designing screening strategies for DR. Also, this factor affects the adherence to outpatient clinic consultations and screening programs; especially in those individuals who were diagnosed with T1DM after 18 years of age. As such, it is essential to develop and promote new social policies centered on this risk group with the objective of reducing the consequences of social inequalities in groups with low socioeconomic levels. Despite there being universal healthcare provision in our environment with free access to healthcare, the socioeconomic level still appears to play a significant role.
As limitations of the study, we need to highlight that the results obtained refer to a specific hospital catchment area and, as such, is not translatable to other situations with different characteristics. Also, comparisons with other studies need to be undertaken with caution because of the different measures of socioeconomic status. Moreover, our study requires further specific details on variables associated with service utilization, for example, the number of visits or adherence to treatment which may associate with socioeconomic status and, as such, could shed more light on the links between deprivation and DR.
Hence, we consider it necessary to apply this study model to other studies under different socioeconomic conditions in order to confirm the impact of the socioeconomic deprivation, together with other risk factors of known importance, in the development of DR in patients with T1DM.
Acknowledgments
We thank all the patients and healthcare professionals who participated in this protracted project. Editorial assistance was by Dr Peter R Turner.
References
Footnotes
Contributors SJC, JACD, and MAD designed the study. PAR, SJC and PAM collected the data. PAR, SJC, PAM, JACD and MAD interpreted the data. PAR, JACD and MAD drafted the manuscript. All authors have read and approved the final version of the manuscript. PAR and MAD take responsibility for the integrity of the data, and for submitting the manuscript for publication.
Funding The study was funded by the Instituto de Salud Carlos III (grant number: PI11/02924) and by the Andalusia Ministry of Health ITI-FEDER (grant number: PI-0029-2017).
Competing interests None declared.
Patient consent for publication Obtained.
Ethics approval This study was approved by the Ethics Committee of the Puerta del Mar University Hospital (ID: 80/2011).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request. The deidentified data have been obtained from digital medical records (demographic, clinical and analytical variables); the assessment of the socioeconomic status was assessed using a neighborhood-level measure based on an index of deprivation. Data are available to BMJ held by the authors.