Diabetes screening: a pending issue in hypertense/obese patients

Setting

The Valencian Community is a Mediterranean region in eastern Spain with a population (January 2004) of 4,518,126 inhabitants. Primary health care is provided in health centres, has a universal coverage and has no cost for patients. The prevalence of diabetes is high and it is strongly associated with other known risk factors. Moreover, a large proportion of persons are not diagnosed, although recently screening for diabetes has greatly increased (Generalitat Valenciana: Conselleria de Sanitat, 2010a).

In this Community, at the end of 2003, a preventive activities program addressed to all persons aged 40 years or over was started. Each person was invited to participate by post, after which an appointment at their health centre was made by telephone. A preventive examination was made by physicians and nurses, and the person was given a report with the result of the examination and convenient recommendations, a copy of which remained at the health centre. This program followed the guidelines of the Spanish Society of Family and Community Medicine (Villar Alvarez et al., 2003; Sepehri et al., 2014). The program included screening for cardiovascular (hypertension, diabetes, dyslipidaemia, smoking, obesity...) and gynaecologic (cytology, mammogram,…) conditions and a vaccination campaign (flu, tetanus and pneumococcal).

Study population

The study population comprised all persons who attended their health centre. The main characteristics of these persons were: predominantly women, coexistence of CVRF, older age and frequent visitors (Pedrera Carbonell et al., 2005).

Study design and participants

This observational, cross-sectional study analyzed a sample of nondiabetic persons aged ≥40 years who had hypertension or were obese and who attended their health centre during the first six months of the preventive activities program of the Valencian Community and who were willing to collaborate. A patient was considered to have diabetes/hypertension if their physician had previously diagnosed these conditions with codes 250/401 of ICD-9-CM. Obesity was defined as a body mass index (BMI) ≥30 kg/m². Any patient who was diabetic or who did not have either of the two disorders (hypertension or obesity) was excluded.

Variables

The main parameters constructed were:

1. Poor BP control (≥140/90 mmHg) in patients with hypertension (Chobanian et al., 2003).

2. Class II obesity (BMI ≥ 35 kg/m²) in obese patients (WHO, 1997).

3. A SCORE ≥ 5% (total cholesterol (TC) version) (Conroy et al., 2003). The SCORE is a probabilistic equation to estimate the 10-year risk of cardiovascular death in persons aged 40–65 years without prior CVD. The variables included in this model were: age, gender, TC, systolic BP (SBP), and smoking. This parameter was defined for both analyses (hypertension and obesity).

The following variables were also analyzed: gender, FBG group (<7.0 mmol/L, 5.6–7.0 mmol/L, <5.6 mmol/L) (Expert Committee on the Diagnosis and Classification of Diabetes Mellitus, 2003; Genuth et al., 2003), personal history of smoking or CVD (coronary heart disease (CHD) or stroke), SBP and diastolic BP (DBP) (mmHg), TC (mmol/L), age (years), and BMI (kg/m²). Based on the personal history of CVD, the variable secondary cardiovascular prevention was defined as that patient who had already experienced a cardiovascular event (CHD or stroke) prior to starting the study. A patient was considered to be in primary cardiovascular prevention if they had no history of CVD at the time of entering the study.

Measurements

The measurements of FBG and TC were made from first morning blood samples taken after a minimum 8-h fast and using calibrated devices. The personal history of diseases and smoking, the gender, and the age were ascertained at a patient interview and corroborated with the medical history. The BP was measured in accordance with current guidelines using well-calibrated semiautomated anaeroid—mercury—devices under suitable conditions. The BMI was calculated from the weight and height, measured with a calibrated scale and stadiometer after removing any objects that could affect the weight and with no shoes. The SCORE was computed using all the variables required for its calculation (Conroy et al., 2003).

Sample size

The sample included 5,347 patients with hypertension (of whom 3,003 fulfilled the criteria for calculation of the SCORE) and 7,833 obese patients (SCORE in 5,827). We aimed to determine whether there existed dependence (Pearson X² tests in a 3 × 2 table) between the FBG groups and the main parameters (control, class II obesity and high SCORE in patients with hypertension or obesity). As the sample was collected prior to calculating the sample size, we calculated the power of the contrast of the X² with the sample (Chow, Wang & Shao, 2007). Thus, we approximated the X² value with the data of 1,500 hypertensive and 1,500 obese patients selected randomly from the final sample. With these values and using a significance level of 5%, we obtained a Pearson X² contrast power >98% in all the analyses done.

Statistical analysis

Absolute and relative frequencies are used to describe the qualitative variables, and means and standard deviations for the quantitative variables. Multivariate logistic regression models were used to estimate the adjusted odds ratio (ORs) in order to analyze the relation between the main parameters (poor control of the BP, class II obesity and high cardiovascular risk) and the other variables. For the analysis of the poor control of the BP and class II obesity, the ORs were adjusted for gender, FBG group, smoking history, SBP, DBP, TC, age, and BMI. In this analysis, we eliminated from the model those parameters that had collinearity with the control of the BP and class II obesity (SBP and DBP for hypertension, BMI for obesity). In the analysis of the high cardiovascular risk, the ORs were adjusted for FBG group and BMI. The other variables were not used, given their collinearity with the SCORE function (Conroy et al., 2003). The likelihood ratio test was used to determine the fit of the models. All the analyses were done with a significance level of 5% and the confidence intervals (CI) were calculated for the most relevant parameters. The analyses were all done using IBM SPSS Statistics 19.

Ethical considerations

This study is based on an institutional agreement between the Conselleria de Sanitat and Miguel Hernández University, Elche (reference number: AVS-UV1.07X) authorized by an Institutional Review Board (Universidad Miguel Hernández de Elche-Conselleria de Sanitat de la Generalitat Valenciana). The data were therefore analyzed in compliance with current legislation on medical ethics. This institution played no role in data collection, analysis or interpretation, nor in the publication of the final manuscript. Additionally, the data were anonymized and encrypted, in accordance with the data protection law.

This population-based, non-interventional study (data from the Valencian Community) used data from medical records and informed consent was not required for included patients. The institutional agreement approved this consent procedure and ensured that information access was restricted, it did not compromise the interests or welfare of any patient, it minimized the risk of injury and its use was in line with current legislation.

Strengths and limitations of the study

The main strength of this study is that it is the first to look at the possible cardiovascular consequences of committing diagnostic inertia in patients with diabetes mellitus who have a diagnosis of hypertension or obesity. The results therefore are novel. In addition, the statistical power of the study is enhanced by the large sample size. This minimizes the random error when drawing conclusions about a population attending their health centre. Moreover, the fact that all the health centres in the Valencian Community participated in the study provides external validity. Furthermore, the contrast power used was over 98%, when studies generally calculate their sample sizes from tables indicating an 80% or 90% power.

The limitations are related to those found in any cross-sectional study measuring the magnitude of a problem; that is, they are unable to produce any longitudinal or causal findings. Concerning bias, the most important is selection bias, given the characteristics of the study sample. This bias concerns the fact that it is the most motivated patients who request a preventive check-up. For obvious reasons this cannot be changed, as each person has a different degree of motivation. However, this does not affect the purpose of the study, which was to assess the lack of detection of high FBG concentrations in patients with no diagnosis of diabetes but who did have a diagnosis of hypertension or obesity. As for measurement bias, all the teams were requested to use reliable devices to measure the variables and to do the clinical interview correctly. Logically, this bias too is accepted in this type of study. Finally, although we used 2003–2004 data, we have to take into account that diagnostic inertia is a prevalent problem which was already detected in the 1980s (Bell & Dippe, 1988); however, it was not named inertia, and even nowadays it is being studied (Palazón-Bru et al., 2014). During these last decades the clinical guidelines have been updated and certain cut-off control parameters have been modified, but diagnostic inertia is still taking place.