Research Report European Addiction Research
Received: April 15, 2013 Accepted: August 13, 2013 Published online: October 31, 2013
Eur Addict Res 2014;20:94–104 DOI: 10.1159/000355171
Effect of Smoking Status on Healthcare Costs and Resource Utilization in Patients with Type 2 Diabetes in Routine Clinical Practice: A Retrospective Nested Case-Control Economic Study Antoni Sicras-Mainar a Javier Rejas-Gutiérrez c Ruth Navarro-Artieda d Jordi Ibánez-Nolla b
a
Dirección de Planificación and b Dirección Médica, Badalona Serveis Assistencials SA, d Documentación Médica, Hospital Germans Trias i Pujol, Badalona, and c Departamento de Farmacoeconomía e Investigación de Resultados en Salud, Pfizer SLU, Alcobendas, Spain
Key Words Type 2 diabetes · Resource utilization · Direct costs · Indirect costs · Case-control study · Smoking
Abstract Aim: To compare healthcare resource utilization and costs according to smoking status in patients with type 2 diabetes in clinical practice. Methods: A retrospective cohort nested case-control study was designed. Cases were current smokers, while 2 types of controls (former smokers and never smokers) were matched (2 controls per case) for age, sex, duration of diabetes and burden of comorbidity using data from medical records. Noninstitutionalized diabetics of both genders, aged >18 years and seen consecutively over a 5-year period before the index date, were enrolled. Analysis compared healthcare resource utilization, loss of productivity due to sick leave and corresponding costs. Results: In total, 2,490 medical records were analyzed, i.e. 498 cases, 996 former smokers and 996 never smokers. Mean age was 63.4 years (64.9% male). Smokers had higher glycosylated hemoglobin levels (7.4 vs. 7.2 and 7.2%, respectively; p = 0.013) and a lower degree of metabolic control (49.2 vs. 54.7 and 55.8%; p = 0.036). Smokers had higher average annual
© 2013 S. Karger AG, Basel 1022–6877/14/0202–0094$39.50/0 E-Mail [email protected] www.karger.com/ear
costs (EUR 3,583) than former smokers (EUR 2,885; p < 0.001) and never smokers (EUR 2,183; p < 0.001). Conclusions: Diabetic smoker patients had lower metabolic control, higher health resource utilization and more sick leave, resulting in higher healthcare costs and lost productivity compared with both former and never smoker diabetics. © 2013 S. Karger AG, Basel
Introduction
Cardiovascular diseases are the leading cause of morbidity and mortality in developed countries [1]. Detection and control of cardiovascular risk factors remains the key preventive strategy [2]. Diabetes mellitus (DM) is one of the diseases with the greatest public health impact, not only due to its high frequency but, above all, the chronic complications it causes and its leading role as a cardiovascular risk factor [3, 4]. DM is associated with a 2- to 3-fold increase in the probability of cardiovascular events (CVEs), and this increase is greater in women than in men. Glucose intolerance is also associated with a 1.5-fold increased risk of developing a CVE [5–8]. Although it varies from country to country, the prevalence Antoni Sicras Mainar Dirección de Planificación y Desarrollo Organizativo Badalona Serveis Assistencials SA Calle Gaietà Soler, 6–8 entlo, ES–08911 Badalona, Barcelona (Spain) E-Mail asicras @ bsa.cat
of known diabetes is around 8% in females and 12% in males [9, 10]. Smoking, a chronic addictive disease, is the main avoidable risk factor for coronary heart disease [11, 12]. Currently, in Spain, the prevalence of smoking is 35.4% in patients with acute myocardial infarction and about 32% in patients with angina pectoris. Overall, it is estimated that 29% of deaths from coronary heart disease are due to smoking, and 1 in 3 deaths caused by smoking is premature [13, 14]. Therefore, smoking increases cardiovascular risk in both diabetics and nondiabetics [15, 16]. However, the risk of complications associated with smoking added to DM is 4 times higher. In addition to this increased cardiovascular risk, nicotine can reduce insulin sensitivity, and therefore smoking may be related to both an increased risk of DM in smokers and an increased risk of microvascular/macrovascular complications in that population [9, 17, 18]. Evidence shows that smoking cessation leads to reduced cardiovascular risk [15, 16, 19]. Studies have shown a reduction in the incidence of CVEs and a decrease in symptoms of arteriosclerotic disease in patients with heart disease who stop smoking [13, 15, 20]. There is abundant evidence that smoking in diabetic patients is associated with a worse prognosis and acceleration of vascular complications and that cessation provides early benefits [20, 21]. Antismoking advice should be integrated into medical practice/education, and structured programs should be provided to help diabetics quit smoking, since this is the most cost-effective preventive measure in that population [22]. Data from the INTERHEART Study [23] highlight intensified smoking control policies worldwide. In Spain, smoking control measures are recent and still inadequate, despite the efforts and recommendations made by medical societies and health authorities (e.g. legislation to prevent smoking in public places) for smoking cessation [24, 25]. Medicines and therapies for smoking cessation are not reimbursed by the Spanish public health system, but measures aimed at specific population groups may be partially reimbursed by some regions and/or municipalities according to their financial capacity. However, due to the austerity measures introduced by the Spanish government because of the current economic crisis, more evidence is needed to convince taxpayers that quitting smoking is a cost-effective strategy. The available evidence in Spain on resource utilization and associated costs for diabetics who smoke is limited or nonexistent, and the clinical implications of smoking are great and generate high consumption of health resources. For that reason, this study is very relevant. There are
few studies that have comprehensively studied these variables, and there is a growing need for naturalistic studies representative of real clinical conditions. Thus, the purpose of this study was to evaluate the effect of smoking status on the use of healthcare and non-healthrelated resources and associated costs in patients with type 2 diabetes (T2DM) in routine clinical practice in a Spanish population.
Smoking, Healthcare Costs and Resource Utilization in T2DM Patients
Eur Addict Res 2014;20:94–104 DOI: 10.1159/000355171
Patients and Methods Design and Study Population We performed an observational nested case-control study in a retrospective cohort, with 2 types of controls and 2 controls per case, by reviewing existing medical records of patients followed in primary care and hospital settings. The index date for both cases and controls was the last medical visit/contact. Data were collected for the year before the index date, and patients were selected in the 5 years starting January 7, 2007, and ending June 30, 2012. Cases were defined as active smokers (≥2 cigarettes per day for ≥12 months). Controls were classified as either former smokers (>12 months without smoking) or never smokers. Smoking status was defined according to established definitions (consensus document). Controls were matched with cases for age (±2 years), sex, years since diagnosis of T2DM (±2 years) and resource utilization bands (RUBs; see below). Medical records were reviewed for patients from 6 primary care centers (Apenins-Montigalà, MoreraPomar, Montgat-Tiana, Nova Lloreda, La Riera and Martí-Julià) managed by Badalona Serveis Assistencials SA. Information on health resources was obtained from two reference hospitals, i.e. Hospital Municipal de Badalona and Hospital Germans Trias i Pujol. The population assigned to these centers is mostly urban, with middle to low socioeconomic status and predominantly industrial occupations. The patient enrollment period lasted 5 years before the index date, but resource utilization and associated costs were those incurred during the 12 months prior to the last visit in the 5-year enrollment period. All consecutive patients seen for medical care before the index date (June 30, 2012) who met the following inclusion and exclusion criteria were enrolled. Inclusion criteria were as follows: (1) age ≥18 years, (2) either gender, (3) at least a 2-year history of T2DM (disease progression), (4) guaranteed regular monitoring of resource utilization and costs at different centers (data available for the 12 months preceding the enrollment date) and (5) cover by the prescription program in order to obtain valid records of each treatment administered. Exclusion criteria were as follows: (1) transfer out due to change of residence, (2) transfer out of the area, (3) permanent institutionalization, (4) former smoking with less than 1 year without smoking and (5) occasional smoking of less than 1 cigarette/day. Data confidentiality was maintained pursuant to the Personal Data Protection Act (Law 15 of December 13, 1999), with dissociation of personal data. The study was classified by the Spanish Agency for Medicines and Health Products as a Post-Marketing Study – Other Designs and was subsequently approved by the Clinical Research Ethics Committee of the Hospital Clinic of Barcelona (study DC-FUM-2012, October 11, 2012).
95
Description of Treatment This was a noninterventional study, in which information and clinical data on previous treatment with antidiabetics or insulin according to the Anatomical Therapeutic Chemical Classification System [26] were obtained for costing purposes only. The assignment of a patient to a specific therapeutic strategy was at the physician’s discretion. Information was recorded for the following medications: insulin (A10AB*, A10AC*, A10AD*, A10AE*), sulfonylureas (A10BB*), meglitinides (A10BX*), dipeptidyl peptidase-4 (A10BH*), metformin (A10BA*), pioglitazone (A10BG*), acarbose (A10BF*), miglitol (A10BF*), drug combinations (A10BD*) and other antidiabetics (A10BX*). Assessments Smoking was analyzed using the following variables: (1) years of smoking (in smokers and former smokers), (2) number of cigarettes smoked per day (current smokers) and (3) duration of cessation (in former smokers). The diagnosis of T2DM was made according to the International Classification of Primary Care, component 7, diseases and health problems (T90) [27], and coding of hospital and emergency room admissions according to the International Classification of Diseases, 9th Revision, Clinical Modification (code 250). The following variables related to T2DM and microvascular complications were assessed: (1) duration of T2DM, (2) diabetic retinopathy, (3) diabetic nephropathy and (4) diabetic neuropathy. The main sociodemographic variables collected were age, sex and disability pension status. Comorbidity variables were obtained from the International Classification of Primary Care, as follows [27]: hypertension (K86, K87); dyslipidemia (T93); obesity (T82); alcoholism (P15, P16); all types of organ failure (heart, liver and kidney); cerebrovascular accident (K90, K91, K93); chronic obstructive pulmonary disease (R95, chronic airflow obstruction); bronchial asthma (R96); dementia or memory disorders (P70, P20); neurological diseases, i.e. Parkinson’s disease (N87), epilepsy (N88), multiple sclerosis (N86) and other neurological diseases (N99); depressive syndrome (P76), and malignancies (all types: A79, B72–75, D74–78, F75, H75, K72, L71, L97, N74–76, R84–86, T71–73, U75–79, W72–73, X75–81, Y77–79). The general comorbidity summary variables used for each patient treated were (1) the Charlson comorbidity index [28], which is used as a proxy for severity of health status, and (2) the individual causality index, obtained from the Adjusted Clinical Groups (ACG), which was used to ascertain the burden of comorbidity. The ACG is a patient classification system based on individual resource use [29]. The ACG® Case-Mix System Grouper algorithm consists of a series of consecutive steps that provide 106 mutually exclusive ACG groups, 1 for each patient treated. The ACG application provides RUBs for each patient, according to general morbidity, in 1 of 5 mutually exclusive categories, i.e. 1 (healthy or very low morbidity users), 2 (low morbidity), 3 (moderate morbidity), 4 (high morbidity) and 5 (very high morbidity). The following biochemical and anthropometric parameters were measured: systolic blood pressure, diastolic blood pressure, body mass index (BMI), fasting glucose, glycosylated hemoglobin (HbA1c), triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, serum creatinine and cardiovascular risk estimated according to the amended criteria of the National Cholesterol Education Program-Adult Treatment Panel III [30]. These parameters were measured using the latest data available. Correct metabolic control was defined as HbA1c 18 years N = 86,520 Seeking care N = 74,408 Type 2 Diabetes N = 7,374 Patients excluded N = 542
Fig. 1. Patient flow chart. Nested case-control observational study in a retrospective cohort, with 2 types of controls for each case, by review of existing medical records (computerized databases with existing dissociated data) of patients followed in primary care and hospitals.
Smokers N = 671 (9.1%)
Former smokers N = 4,211 (57.1%)
Never smokers N = 2,492 (33.8%)
Smokers N = 498
Former smokers N = 996
Never smokers N = 996
CASES
CONTROLS A
CONTROLS B
ers in a study of secondary prevention of cardiovascular diseases. Assuming a random error of 5% and a minimum statistical power for comparisons of 80%, it was necessary to recruit a minimum of about 1,700 patients to enable comparison of costs between cases (340 patients) and controls (former smokers: 680; never smokers: 680). However, all consecutive records extracted from the database were included until the minimum sample size was reached. Prior to analysis, data were thoroughly reviewed, in particular with reference to computerized medical records, observing frequency distributions and searching for possible recording or encoding errors. A descriptive univariate analysis was performed expressing parametric variables as means, medians, SDs and 95% confidence intervals (CIs), and nonparametric variables as medians and interquartile ranges, once normal distribution was verified using the Kolmogorov-Smirnov test. In the bivariate analysis, Student’s t test, analysis of variance, χ2 test, Pearson’s linear correlation and the Mann-Whitney-Wilcoxon nonparametric test were used according to data distribution. A logistic regression analysis was performed to determine the association between smoking (dependent variable: smoking) and comorbidities (covariates: age, sex, years since diagnosis of DM) and RUBs. Resource utilization and associated costs were compared according to the recommendations of Thompson and Barber [33] using a general linear model (analysis of covariance) adjusted for relevant covariates (estimation of marginal means with Bonferroni adjustment for multiple comparisons, when needed). The analysis was performed with SPSS Statistics for Windows, version 17.0. Statistical significance was established as p < 0.05.
A total of 86,520 patients aged >18 years assigned to the participating centers were originally selected. Of these, 8.5% (95% CI 8.3–8.7%; n = 7,374) had a diagnosis
of T2DM. Of these, 9.1% were current smokers. Finally, 498 patients who met the study selection criteria for inclusion as cases were enrolled (fig. 1). Table 2 describes the baseline characteristics of the three study groups, namely smokers (cases, n = 498), former smokers (n = 996) and never smokers (n = 996). Among all enrolled patients, the mean age was 63.4 years (SD 9.9), 64.9% were male, 62.6% had dyslipidemia, 59.7% had hypertension, 28% were obese and 26% had suffered a CVE. In the logistic regression model, when smokers were compared with controls (former smokers, never smokers), smoking was associated with chronic obstructive pulmonary disease (odds ratio 1.6, 95% CI 1.1–2.2; p = 0.010), while never smoking was associated with obesity (odds ratio 1.5, 95% CI 1.2–1.9; p < 0.001). The distribution of antidiabetic medication by study group is shown in table 3. The distribution of therapeutic active principles was heterogenous in the three groups. The rate of insulin consumption was 23.8% (smokers: 26.2%), thiazolidinediones/biguanides 73.9% (former smokers: 75.4%), insulin secretagogues 39.0% (former smokers: 41.5%) and combinations of active ingredients 20.6% (never smokers: 22.0%). Metformin was the most commonly used drug (91.4%). Use of the metformin/pioglitazone combination was greater in never smokers than in smokers and former smokers (3.9 vs. 2.2 and 2.1%, respectively; p < 0.05). Table 4 shows the distribution of anthropometric and biochemical parameters by study group. Compared with controls (former and never smokers), smokers had a higher HbA1c level (7.4 vs. 7.2 and 7.2%, respectively;
Smoking, Healthcare Costs and Resource Utilization in T2DM Patients
Eur Addict Res 2014;20:94–104 DOI: 10.1159/000355171
Results
97
Table 2. Baseline characteristics of patients by smoking status
Smokers (n = 498)
Former smokers (n = 996)
Never smokers (n = 996)
Total (n = 2,490)
p
Sociodemographic characteristics Mean age, years Males Disability pension
63.3 (10.4) 64.7% 63.3%
63.4 (9.6) 65.0% 63.4%
63.5 (10.0) 64.9% 63.2%
63.4 (9.9) 64.9% 63.3%
General comorbidity Mean number of diagnoses Mean Charlson index score Mean RUB RUB-1 RUB-2 RUB-3 RUB-4 RUB-5
5.0 (2.6) 1.0 (0.9) 2.9 (0.7) 3.6% 16.7% 69.5% 8.2% 2.0%
5.0 (2.3) 1.0 (0.7) 2.9 (0.6) 3.0% 14.6% 74.0% 6.9% 1.5%
5.1 (2.5) 1.0 (0.7) 2.9 (0.6) 2.4% 14.7% 74.4% 7.8% 0.7%
5.0 (2.4) 1.0 (0.7) 2.9 (0.6) 2.9% 15.0% 73.3% 7.6% 1.3%
Comorbidities Hypertension Dyslipidemia Obesity Alcoholism Ischemic heart disease Cerebrovascular accident CVE Organ failure Bronchial asthma COPD Neuropathies Dementia (all types) Depressive syndrome Malignant neoplasias
61.6% 63.5% 22.1% 7.2% 13.9% 18.9% 25.5% 17.9% 4.0% 11.2% 2.0% 2.4% 19.7% 11.8%
60.2% 63.3% 28.6%** 5.9% 14.5% 17.2% 26.7% 17.2% 3.6% 11.3% 2.1% 1.6% 20.1% 9.1%
58.2% 61.4% 30.4%** 5.3% 13.8% 17.3% 25.6% 14.9% 3.5% 4.5% 2.1% 1.4% 20.1% 9.5%
59.7% 62.6% 28.0% 5.9% 14.1% 17.6% 26.0% 16.4% 3.7% 8.6% 2.1% 1.7% 20.0% 9.8%
0.407 0.637 0.003 0.339 0.894 0.684 0.817 0.230 0.885
Received: April 15, 2013 Accepted: August 13, 2013 Published online: October 31, 2013
Eur Addict Res 2014;20:94–104 DOI: 10.1159/000355171
Effect of Smoking Status on Healthcare Costs and Resource Utilization in Patients with Type 2 Diabetes in Routine Clinical Practice: A Retrospective Nested Case-Control Economic Study Antoni Sicras-Mainar a Javier Rejas-Gutiérrez c Ruth Navarro-Artieda d Jordi Ibánez-Nolla b
a
Dirección de Planificación and b Dirección Médica, Badalona Serveis Assistencials SA, d Documentación Médica, Hospital Germans Trias i Pujol, Badalona, and c Departamento de Farmacoeconomía e Investigación de Resultados en Salud, Pfizer SLU, Alcobendas, Spain
Key Words Type 2 diabetes · Resource utilization · Direct costs · Indirect costs · Case-control study · Smoking
Abstract Aim: To compare healthcare resource utilization and costs according to smoking status in patients with type 2 diabetes in clinical practice. Methods: A retrospective cohort nested case-control study was designed. Cases were current smokers, while 2 types of controls (former smokers and never smokers) were matched (2 controls per case) for age, sex, duration of diabetes and burden of comorbidity using data from medical records. Noninstitutionalized diabetics of both genders, aged >18 years and seen consecutively over a 5-year period before the index date, were enrolled. Analysis compared healthcare resource utilization, loss of productivity due to sick leave and corresponding costs. Results: In total, 2,490 medical records were analyzed, i.e. 498 cases, 996 former smokers and 996 never smokers. Mean age was 63.4 years (64.9% male). Smokers had higher glycosylated hemoglobin levels (7.4 vs. 7.2 and 7.2%, respectively; p = 0.013) and a lower degree of metabolic control (49.2 vs. 54.7 and 55.8%; p = 0.036). Smokers had higher average annual
© 2013 S. Karger AG, Basel 1022–6877/14/0202–0094$39.50/0 E-Mail [email protected] www.karger.com/ear
costs (EUR 3,583) than former smokers (EUR 2,885; p < 0.001) and never smokers (EUR 2,183; p < 0.001). Conclusions: Diabetic smoker patients had lower metabolic control, higher health resource utilization and more sick leave, resulting in higher healthcare costs and lost productivity compared with both former and never smoker diabetics. © 2013 S. Karger AG, Basel
Introduction
Cardiovascular diseases are the leading cause of morbidity and mortality in developed countries [1]. Detection and control of cardiovascular risk factors remains the key preventive strategy [2]. Diabetes mellitus (DM) is one of the diseases with the greatest public health impact, not only due to its high frequency but, above all, the chronic complications it causes and its leading role as a cardiovascular risk factor [3, 4]. DM is associated with a 2- to 3-fold increase in the probability of cardiovascular events (CVEs), and this increase is greater in women than in men. Glucose intolerance is also associated with a 1.5-fold increased risk of developing a CVE [5–8]. Although it varies from country to country, the prevalence Antoni Sicras Mainar Dirección de Planificación y Desarrollo Organizativo Badalona Serveis Assistencials SA Calle Gaietà Soler, 6–8 entlo, ES–08911 Badalona, Barcelona (Spain) E-Mail asicras @ bsa.cat
of known diabetes is around 8% in females and 12% in males [9, 10]. Smoking, a chronic addictive disease, is the main avoidable risk factor for coronary heart disease [11, 12]. Currently, in Spain, the prevalence of smoking is 35.4% in patients with acute myocardial infarction and about 32% in patients with angina pectoris. Overall, it is estimated that 29% of deaths from coronary heart disease are due to smoking, and 1 in 3 deaths caused by smoking is premature [13, 14]. Therefore, smoking increases cardiovascular risk in both diabetics and nondiabetics [15, 16]. However, the risk of complications associated with smoking added to DM is 4 times higher. In addition to this increased cardiovascular risk, nicotine can reduce insulin sensitivity, and therefore smoking may be related to both an increased risk of DM in smokers and an increased risk of microvascular/macrovascular complications in that population [9, 17, 18]. Evidence shows that smoking cessation leads to reduced cardiovascular risk [15, 16, 19]. Studies have shown a reduction in the incidence of CVEs and a decrease in symptoms of arteriosclerotic disease in patients with heart disease who stop smoking [13, 15, 20]. There is abundant evidence that smoking in diabetic patients is associated with a worse prognosis and acceleration of vascular complications and that cessation provides early benefits [20, 21]. Antismoking advice should be integrated into medical practice/education, and structured programs should be provided to help diabetics quit smoking, since this is the most cost-effective preventive measure in that population [22]. Data from the INTERHEART Study [23] highlight intensified smoking control policies worldwide. In Spain, smoking control measures are recent and still inadequate, despite the efforts and recommendations made by medical societies and health authorities (e.g. legislation to prevent smoking in public places) for smoking cessation [24, 25]. Medicines and therapies for smoking cessation are not reimbursed by the Spanish public health system, but measures aimed at specific population groups may be partially reimbursed by some regions and/or municipalities according to their financial capacity. However, due to the austerity measures introduced by the Spanish government because of the current economic crisis, more evidence is needed to convince taxpayers that quitting smoking is a cost-effective strategy. The available evidence in Spain on resource utilization and associated costs for diabetics who smoke is limited or nonexistent, and the clinical implications of smoking are great and generate high consumption of health resources. For that reason, this study is very relevant. There are
few studies that have comprehensively studied these variables, and there is a growing need for naturalistic studies representative of real clinical conditions. Thus, the purpose of this study was to evaluate the effect of smoking status on the use of healthcare and non-healthrelated resources and associated costs in patients with type 2 diabetes (T2DM) in routine clinical practice in a Spanish population.
Smoking, Healthcare Costs and Resource Utilization in T2DM Patients
Eur Addict Res 2014;20:94–104 DOI: 10.1159/000355171
Patients and Methods Design and Study Population We performed an observational nested case-control study in a retrospective cohort, with 2 types of controls and 2 controls per case, by reviewing existing medical records of patients followed in primary care and hospital settings. The index date for both cases and controls was the last medical visit/contact. Data were collected for the year before the index date, and patients were selected in the 5 years starting January 7, 2007, and ending June 30, 2012. Cases were defined as active smokers (≥2 cigarettes per day for ≥12 months). Controls were classified as either former smokers (>12 months without smoking) or never smokers. Smoking status was defined according to established definitions (consensus document). Controls were matched with cases for age (±2 years), sex, years since diagnosis of T2DM (±2 years) and resource utilization bands (RUBs; see below). Medical records were reviewed for patients from 6 primary care centers (Apenins-Montigalà, MoreraPomar, Montgat-Tiana, Nova Lloreda, La Riera and Martí-Julià) managed by Badalona Serveis Assistencials SA. Information on health resources was obtained from two reference hospitals, i.e. Hospital Municipal de Badalona and Hospital Germans Trias i Pujol. The population assigned to these centers is mostly urban, with middle to low socioeconomic status and predominantly industrial occupations. The patient enrollment period lasted 5 years before the index date, but resource utilization and associated costs were those incurred during the 12 months prior to the last visit in the 5-year enrollment period. All consecutive patients seen for medical care before the index date (June 30, 2012) who met the following inclusion and exclusion criteria were enrolled. Inclusion criteria were as follows: (1) age ≥18 years, (2) either gender, (3) at least a 2-year history of T2DM (disease progression), (4) guaranteed regular monitoring of resource utilization and costs at different centers (data available for the 12 months preceding the enrollment date) and (5) cover by the prescription program in order to obtain valid records of each treatment administered. Exclusion criteria were as follows: (1) transfer out due to change of residence, (2) transfer out of the area, (3) permanent institutionalization, (4) former smoking with less than 1 year without smoking and (5) occasional smoking of less than 1 cigarette/day. Data confidentiality was maintained pursuant to the Personal Data Protection Act (Law 15 of December 13, 1999), with dissociation of personal data. The study was classified by the Spanish Agency for Medicines and Health Products as a Post-Marketing Study – Other Designs and was subsequently approved by the Clinical Research Ethics Committee of the Hospital Clinic of Barcelona (study DC-FUM-2012, October 11, 2012).
95
Description of Treatment This was a noninterventional study, in which information and clinical data on previous treatment with antidiabetics or insulin according to the Anatomical Therapeutic Chemical Classification System [26] were obtained for costing purposes only. The assignment of a patient to a specific therapeutic strategy was at the physician’s discretion. Information was recorded for the following medications: insulin (A10AB*, A10AC*, A10AD*, A10AE*), sulfonylureas (A10BB*), meglitinides (A10BX*), dipeptidyl peptidase-4 (A10BH*), metformin (A10BA*), pioglitazone (A10BG*), acarbose (A10BF*), miglitol (A10BF*), drug combinations (A10BD*) and other antidiabetics (A10BX*). Assessments Smoking was analyzed using the following variables: (1) years of smoking (in smokers and former smokers), (2) number of cigarettes smoked per day (current smokers) and (3) duration of cessation (in former smokers). The diagnosis of T2DM was made according to the International Classification of Primary Care, component 7, diseases and health problems (T90) [27], and coding of hospital and emergency room admissions according to the International Classification of Diseases, 9th Revision, Clinical Modification (code 250). The following variables related to T2DM and microvascular complications were assessed: (1) duration of T2DM, (2) diabetic retinopathy, (3) diabetic nephropathy and (4) diabetic neuropathy. The main sociodemographic variables collected were age, sex and disability pension status. Comorbidity variables were obtained from the International Classification of Primary Care, as follows [27]: hypertension (K86, K87); dyslipidemia (T93); obesity (T82); alcoholism (P15, P16); all types of organ failure (heart, liver and kidney); cerebrovascular accident (K90, K91, K93); chronic obstructive pulmonary disease (R95, chronic airflow obstruction); bronchial asthma (R96); dementia or memory disorders (P70, P20); neurological diseases, i.e. Parkinson’s disease (N87), epilepsy (N88), multiple sclerosis (N86) and other neurological diseases (N99); depressive syndrome (P76), and malignancies (all types: A79, B72–75, D74–78, F75, H75, K72, L71, L97, N74–76, R84–86, T71–73, U75–79, W72–73, X75–81, Y77–79). The general comorbidity summary variables used for each patient treated were (1) the Charlson comorbidity index [28], which is used as a proxy for severity of health status, and (2) the individual causality index, obtained from the Adjusted Clinical Groups (ACG), which was used to ascertain the burden of comorbidity. The ACG is a patient classification system based on individual resource use [29]. The ACG® Case-Mix System Grouper algorithm consists of a series of consecutive steps that provide 106 mutually exclusive ACG groups, 1 for each patient treated. The ACG application provides RUBs for each patient, according to general morbidity, in 1 of 5 mutually exclusive categories, i.e. 1 (healthy or very low morbidity users), 2 (low morbidity), 3 (moderate morbidity), 4 (high morbidity) and 5 (very high morbidity). The following biochemical and anthropometric parameters were measured: systolic blood pressure, diastolic blood pressure, body mass index (BMI), fasting glucose, glycosylated hemoglobin (HbA1c), triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, serum creatinine and cardiovascular risk estimated according to the amended criteria of the National Cholesterol Education Program-Adult Treatment Panel III [30]. These parameters were measured using the latest data available. Correct metabolic control was defined as HbA1c 18 years N = 86,520 Seeking care N = 74,408 Type 2 Diabetes N = 7,374 Patients excluded N = 542
Fig. 1. Patient flow chart. Nested case-control observational study in a retrospective cohort, with 2 types of controls for each case, by review of existing medical records (computerized databases with existing dissociated data) of patients followed in primary care and hospitals.
Smokers N = 671 (9.1%)
Former smokers N = 4,211 (57.1%)
Never smokers N = 2,492 (33.8%)
Smokers N = 498
Former smokers N = 996
Never smokers N = 996
CASES
CONTROLS A
CONTROLS B
ers in a study of secondary prevention of cardiovascular diseases. Assuming a random error of 5% and a minimum statistical power for comparisons of 80%, it was necessary to recruit a minimum of about 1,700 patients to enable comparison of costs between cases (340 patients) and controls (former smokers: 680; never smokers: 680). However, all consecutive records extracted from the database were included until the minimum sample size was reached. Prior to analysis, data were thoroughly reviewed, in particular with reference to computerized medical records, observing frequency distributions and searching for possible recording or encoding errors. A descriptive univariate analysis was performed expressing parametric variables as means, medians, SDs and 95% confidence intervals (CIs), and nonparametric variables as medians and interquartile ranges, once normal distribution was verified using the Kolmogorov-Smirnov test. In the bivariate analysis, Student’s t test, analysis of variance, χ2 test, Pearson’s linear correlation and the Mann-Whitney-Wilcoxon nonparametric test were used according to data distribution. A logistic regression analysis was performed to determine the association between smoking (dependent variable: smoking) and comorbidities (covariates: age, sex, years since diagnosis of DM) and RUBs. Resource utilization and associated costs were compared according to the recommendations of Thompson and Barber [33] using a general linear model (analysis of covariance) adjusted for relevant covariates (estimation of marginal means with Bonferroni adjustment for multiple comparisons, when needed). The analysis was performed with SPSS Statistics for Windows, version 17.0. Statistical significance was established as p < 0.05.
A total of 86,520 patients aged >18 years assigned to the participating centers were originally selected. Of these, 8.5% (95% CI 8.3–8.7%; n = 7,374) had a diagnosis
of T2DM. Of these, 9.1% were current smokers. Finally, 498 patients who met the study selection criteria for inclusion as cases were enrolled (fig. 1). Table 2 describes the baseline characteristics of the three study groups, namely smokers (cases, n = 498), former smokers (n = 996) and never smokers (n = 996). Among all enrolled patients, the mean age was 63.4 years (SD 9.9), 64.9% were male, 62.6% had dyslipidemia, 59.7% had hypertension, 28% were obese and 26% had suffered a CVE. In the logistic regression model, when smokers were compared with controls (former smokers, never smokers), smoking was associated with chronic obstructive pulmonary disease (odds ratio 1.6, 95% CI 1.1–2.2; p = 0.010), while never smoking was associated with obesity (odds ratio 1.5, 95% CI 1.2–1.9; p < 0.001). The distribution of antidiabetic medication by study group is shown in table 3. The distribution of therapeutic active principles was heterogenous in the three groups. The rate of insulin consumption was 23.8% (smokers: 26.2%), thiazolidinediones/biguanides 73.9% (former smokers: 75.4%), insulin secretagogues 39.0% (former smokers: 41.5%) and combinations of active ingredients 20.6% (never smokers: 22.0%). Metformin was the most commonly used drug (91.4%). Use of the metformin/pioglitazone combination was greater in never smokers than in smokers and former smokers (3.9 vs. 2.2 and 2.1%, respectively; p < 0.05). Table 4 shows the distribution of anthropometric and biochemical parameters by study group. Compared with controls (former and never smokers), smokers had a higher HbA1c level (7.4 vs. 7.2 and 7.2%, respectively;
Smoking, Healthcare Costs and Resource Utilization in T2DM Patients
Eur Addict Res 2014;20:94–104 DOI: 10.1159/000355171
Results
97
Table 2. Baseline characteristics of patients by smoking status
Smokers (n = 498)
Former smokers (n = 996)
Never smokers (n = 996)
Total (n = 2,490)
p
Sociodemographic characteristics Mean age, years Males Disability pension
63.3 (10.4) 64.7% 63.3%
63.4 (9.6) 65.0% 63.4%
63.5 (10.0) 64.9% 63.2%
63.4 (9.9) 64.9% 63.3%
General comorbidity Mean number of diagnoses Mean Charlson index score Mean RUB RUB-1 RUB-2 RUB-3 RUB-4 RUB-5
5.0 (2.6) 1.0 (0.9) 2.9 (0.7) 3.6% 16.7% 69.5% 8.2% 2.0%
5.0 (2.3) 1.0 (0.7) 2.9 (0.6) 3.0% 14.6% 74.0% 6.9% 1.5%
5.1 (2.5) 1.0 (0.7) 2.9 (0.6) 2.4% 14.7% 74.4% 7.8% 0.7%
5.0 (2.4) 1.0 (0.7) 2.9 (0.6) 2.9% 15.0% 73.3% 7.6% 1.3%
Comorbidities Hypertension Dyslipidemia Obesity Alcoholism Ischemic heart disease Cerebrovascular accident CVE Organ failure Bronchial asthma COPD Neuropathies Dementia (all types) Depressive syndrome Malignant neoplasias
61.6% 63.5% 22.1% 7.2% 13.9% 18.9% 25.5% 17.9% 4.0% 11.2% 2.0% 2.4% 19.7% 11.8%
60.2% 63.3% 28.6%** 5.9% 14.5% 17.2% 26.7% 17.2% 3.6% 11.3% 2.1% 1.6% 20.1% 9.1%
58.2% 61.4% 30.4%** 5.3% 13.8% 17.3% 25.6% 14.9% 3.5% 4.5% 2.1% 1.4% 20.1% 9.5%
59.7% 62.6% 28.0% 5.9% 14.1% 17.6% 26.0% 16.4% 3.7% 8.6% 2.1% 1.7% 20.0% 9.8%
0.407 0.637 0.003 0.339 0.894 0.684 0.817 0.230 0.885
