The exploitation of administrative health databases in the Basque Country has allowed us to describe the prevalence of chronic diseases, level of multimorbidity and use of health care resources in the entire population and to analyze differences by socioeconomic deprivation and sex. The prevalence of the great majority of the observed chronic diseases is concentrated among the most disadvantaged social groups who also have higher rates of multimorbidity. These inequalities in morbidity burden account for most of the differences found in use of health resources and there is only a small degree of horizontal inequity (difference between use of and need for health care), in most cases this being pro-poor.
Analyzing chronic diseases individually, the largest inequalities are found in pathologies related to high-risk behaviors, substance misuse (smoking, alcohol and other substance abuse) and sedentary life-style, and to exposure to environmental and work-related factors. However, the two sexes do not show analogous patterns in all the conditions and, in general, greater differences were observed in women than men.
Although the methodology used in this study is, in several respects, different to that employed by other authors, our results partially agree with studies in the literature. In a previous study on socioeconomic inequality in the Basque Country, Esnaola et al. found that the greatest differences in causes of mortality between socioeconomic groups were in overdoses of illegal drugs, cirrhosis, AIDS and COPD
. Further, a review of data from surveys in eight European countries identified an association between a lower level of education and a higher prevalence of many types of chronic diseases
, while other research has demonstrated associations between unfavorable socioeconomic conditions and, in particular, cardiovascular disorders
, diabetes, alcoholism, mental illness, asthma and lumbago, among others. It might seem surprising that, although we found inequalities in the prevalence of stroke (especially in women), these were not as large as the results published by other authors
; however, this finding is consistent with a comparative study of causes of mortality across 22 European countries, in which the Basque Country was found to have the lowest levels of inequity associated with level of education for cardiovascular diseases in general, and stroke in particular
On the other hand, for some diseases our results differ from previously published findings. One of the most striking cases is that of skin conditions, which in our study were relatively strongly concentrated in more deprived communities, while other authors have found no relationship between poor socioeconomic conditions and this type of health problem
[27, 29]. Nevertheless, such an association has been established for certain dermatological conditions
, so that it seems reasonable to suppose that an analysis separating dermatologic problems into smaller and less heterogeneous groups might have found various types of relationship between social factors and these conditions.
It is well known that the inequalities between social groups in many risk factors vary between the sexes in magnitude and even direction. For example, there are greater inequalities related to level of education for obesity in women, and for smoking
 and alcohol abuse
 in men. In relation to this, it seems logical to suppose that social class plays a different role in each sex with regards to the prevalence of specific chronic diseases. These facts may explain, at least to a certain extent, the observed differences in CIs between men and women. In particular, we found that inequity in diabetes is much greater among women, confirming the results of previous studies based on data from surveys and mortality registries
[3, 26, 27, 32].
We also found inequity in the prevalence of multimorbidity and this was greater the larger the number of diseases diagnosed in same person. In addition, in all cases, the inequality was larger in women. Although for many years it has been known that the number of people with multiple apparently unconnected conditions is higher than would be expected by chance
, the interest in studying multimorbidity, as well as its consequences for patients, their families, health care organizations and society in general, is a more recent phenomenon. In recent years, numerous publications have confirmed the relationship between poor socioeconomic status and multimorbidity, using data from primary care medical records
[17, 34, 35], surveys of the general population
 or users of certain types of health care services
 and interviews with doctors and patients
. Most of these studies have been carried out in developed countries, but there is also some evidence from low and middle-income countries
In the Basque Country, people with poor socioeconomic status use more health resources, especially females. This inequality is mostly explained by the fact that these deprived groups have greater health care needs given their greater morbidity burden. Nevertheless, we still detected a certain degree of pro-poor inequity in the case of specialized outpatient care and emergency services, and to a lesser extent also in primary care and prescriptions, while no inequity was detected in inpatient care. These results are not in agreement with those of other studies that used surveys
[8, 40–42] or administrative databases
 as a source of information. These other studies have repeatedly found a pro-rich bias regarding specialized outpatient care, in many different countries in Europe and in the USA. As for primary care, the findings are not so consistent, the direction of the inequity (pro-rich or pro-poor) varying between countries; nevertheless, in general, the differences are smaller and the use of primary care seems to be more related to need than economic factors.
One factor that may partly explain the different use of specialized care observed in our data is that some of the Spanish population, in general those with higher incomes, is covered by private medical insurance as well as the Public Health System
[44, 45]. Specifically, in 2011, 17.6% of the population of the Basque Country were exclusively covered by private health insurance or had private cover in addition to the public provision
. People who are doubly insured often go to private specialized care clinics, in order to avoid the waiting lists for appointments in the Public Health Service. In any case, a previous study carried out in Spain, that included people who have only public health insurance, found pro-rich inequity
 and, in another one analyzing separately the utilization of public and private health care services, there appeared to be an equitable use of public specialized care, and certainly no pro-poor tendency
Our study has the following strengths. First, it includes almost the entire population of the geographical area studied, thus avoiding selection bias. Second, it exploits a database containing information from primary, specialized outpatient and inpatient care, as well as prescriptions; as other authors have established, the use of a single source can produce inaccurate estimates
[49, 50], while the complementary use of various sources contributes to a better description of people’s health problems
. Furthermore, by cross-checking data on prescriptions and diagnoses it is possible to differentiate between active and non-active chronic diseases and, to achieve this, we adapted a methodology that has already been used by other authors
. Thirdly, thanks to this comprehensive data set, our study describes a wide range of specific diseases rather than being limited to self-report general health indicators (which is often the case in survey-based research). It also includes pathologies that are the cause of considerable suffering and disability but not fatal, and so are not considered in studies analyzing only mortality data. Finally, in order to control for morbidity and estimate health care needs we opted to employ the ACG case-mix system, a well-known instrument, the usefulness of which has been confirmed in various different countries.
Our study also has certain limitations. Firstly, administrative databases only contain information about problems for which people seek medical attention. Therefore, the prevalence of diseases can only reflect known cases and excludes the presence of diseases that are present but not known of by the patients or their doctors. The distribution of undiagnosed cases might be influenced by factors including accessibility to health care services and help seeking behavior of patients. Secondly, our database contains information from primary and specialized care, emergency departments and hospital admissions, but not psychiatric hospitals; even though patients admitted in such hospitals are also usually cared for by primary care doctors, given their special characteristics, it is feasible that their health records were not as complete as for the rest of the population. Moreover, as previously noted, we do not have access to information from the health services in the private sector as they are not under the management of the Basque Health Service. As a consequence, in this research we were not able to explore the effect of the existence of double insurance. Thirdly, regarding the cost of care per patient, our health service has no direct data and, hence, costs were calculated from the standard pricing of the services provided and in some specific cases the costs were not available (mainly admissions to mental health hospitals, home hospitalization and some components of day care services). Finally, being based on socioeconomic indicators at the level of area of residence, our study has the limitations common to ecological studies. Although there is known to be a relationship between socioeconomic deprivation and poor health, the contributions of individual and context-related factors have not been clearly established. In any case, the area-based measures, reflecting community-wide characteristics, turn out to be indicators with their own characteristics and do not behave solely as proxy measures for individual socioeconomic variables
. The deprivation index employed in this study is a compound index, built from five indicators corresponding to rates of employment and educational attainment
. In their first models, the developers of this index included a further nine variables related to housing, sociodemographic characteristics (single parent households, old age, immigrants from low-income countries and recently arrived immigrants) and environmental factors (pollution, noise levels, delinquency). On the basis of the results of multivariate analysis, however, they decided not to include these variables, and to construct the index using the five aforementioned indicators which saturated in the first dimension by the identification of a single factor structure applying principal component analysis. In any case, the information in this index could be insufficiently exhaustive to reflect all the factors of interest.
Although the relationship between low socioeconomic status and poorer health status is complex, and the underlying mechanisms have not been clearly established, the existence of health inequalities is well recognized. Most people in the world aspire to greater solidarity in the field of health and wish their governments to introduce policies that guarantee health protection and promote equity
. To face this challenge, it is necessary to develop a broad range of interventions and, as stated by the WHO Commission on Social Determinants of Health
, we need to develop instruments that measure the magnitude of the problem, facilitate its analysis, and evaluate the effects of said interventions. We need new indicators, different from those currently employed, which only provide average values of health of the general population as a whole
. In relation to this, research such as the present study, based on the use of administrative health databases, provides useful information to quantify inequalities and inequities in health and health care provision
 and, given the accessibility and constant updating of such databases, this approach may help to monitor changes in these inequalities and inequities over time.