The present study was carried out in the context of a national health system offering universal and free coverage, and the results confirm that PC utilisation is lower among immigrants compared to the native population. Moreover, our results corroborate the observation that immigrant populations have a lower morbidity burden and reinforce the so-called 'healthy migration effect' previously described [16
]. Thus, differences in PC utilisation patterns decrease when considering the morbidity burden.
Using administrative data to objectively monitor the use of PC services by the immigrant population is increasingly important to host countries. It aids in the assessment of inequalities in the use of PC services and eventual access problems [6
]. Unlike studies based on heath surveys, data collected from PC centres offer information on the true utilisation of PC services by the entire population, thus avoiding sampling or response biases. In this particular study, data were assessed from more than 69,000 individuals.
Contrary to certain stereotypes that immigrants make excessive use of health services [20
], our study provides evidence for lower use of PC services by both child and adult immigrants after adjusting for age and sex. This finding held true for all types of visits except in the case of programmed visits for children. These results confirm the findings of previous studies in which lower use of PC services by immigrants was observed both in Spain [1
] and in other developed countries [22
We chose the ACG System to measure the morbidity burden because of its acknowledged validity and extensive use as a risk adjustment technology and because it can be constructed from commonly available administrative data. Based on these data, it develops a composite measure of morbidity burden estimated from a mixture of conditions experienced over a defined time interval [6
]. According to the data provided by the ACG System, the health status of immigrants in our study was better than that of native citizens. These findings are consistent with the previously described 'healthy migration effect' [16
]. The 'healthy migration effect' is a consequence of a self-selection process that excludes ill and disabled individuals and includes those with high capacities and personal motivations. Migratory processes favour the success of individuals with enhanced skills in terms of education, work and communication [28
]. This healthier status of immigrants somehow justifies the lower PC utilisation seen among them. In this study, the differences in PC utilisation patterns almost disappeared among the adult population (IRR = 1.00, p > 0.05) and decreased considerably among children (IRR = 0.94, p < 0.05) after adjusting for case mix. These results suggest that for a given morbidity burden, there are no differences in PC utilisation patterns between immigrants and the native population. Health care for non-EU foreign nationals in Spain is regulated by Article 12 of Organic Law 4/2000 of 11th
January regarding the rights and freedoms of foreign nationals living in Spain and their social integration [10
], which guarantees health care to foreign nationals regardless of their residency status under the same conditions as native Spaniards. This point may explain our results.
Interestingly, the mean number of routine diagnostic tests was 77% higher among immigrant children compared to native children (IRR = 1.77, p < 0.05). This finding could be due to preventive medical checkups carried out on immigrant children upon arrival in the host country [29
]. In addition, it could indicate that GPs lack self-confidence when clinically evaluating these patients due to communication problems derived from language difficulties.
While the age, sex and case mix adjusted mean number of emergency visits was 30% lower among immigrant children (IRR = 0.68, p < 0.05), the immigrant adult population made greater use of such services (IRR = 1.2, p < 0.05). Other studies have highlighted the association between the use of the PC emergency service and work conditions among immigrants. The PC emergency service is an extension to GPs' normal schedule available until 9pm, Monday through Saturday. Difficulties in making work schedules compatible with the visiting hours of PC health centres have been suggested to result in higher usage of PC emergency services among immigrants [31
]. It could be that the higher usage of PC emergency services makes up for the lower number of demanded visits (IRR = 0.97, p < 0.05) and home visits (IRR = 0.41, p < 0.05) among the immigrant adult population.
As for home visits, there is an evident lower use of such service by both adult (IRR = 0.41, p < 0.05) and child (IRR = 0.29, p < 0.05) immigrants which is a clear consequence of lack of information. Thus, improvements are still needed regarding the provision of information on health services access conditions.
One of the main limitations of the present study derives from the lack of available socioeconomic indicators. Given the absence of this information in administrative databases, we were unable to adjust for this characteristic, which has previously been found to explain a high degree of variability in the use of healthcare services [32
]. Nevertheless, the consequences of this limitation might have been mitigated by the fact that all three PC centres included in this study belonged to areas with a similar socio-economic level.
The fact that we did not carry out an analysis that stratified the immigrant population according to country of origin suggests that we might have missed some interesting findings. There is evidence indicating that variations exist among different ethnic groups in terms of health care utilisation [32
]. Such stratification was not possible in our study given the limited size of the immigrant population but should be considered in the future. Nevertheless, the majority of immigrants come from low-income countries (e.g., Latin America and Eastern Europe) and are categorised as economic immigrants. Immigrants from high-income countries (e.g., Western Europe and North America) represent a minority group (4.3% of adult immigrants and 3.0% of child immigrants). Immigrants from high-income countries have a legal status similar to that of the Spanish-born population, and their socioeconomic status is above the national average [34
Another potential limitation of this study derives from the quality of the diagnostic information registered in the electronic medical records since the completeness and accuracy of data entry relies on the enthusiasm and idiosyncrasies of individual GPs [35
]. Medical record data contains some errors and omissions, but there have been large improvements in its quality in recent years. As a result, this source of information is increasingly being used for epidemiological research in the Spanish context [36
]. Moreover, a sensitivity analysis was carried out using data from two of the three PC health centres that were known to have received training sessions to encourage and improve the quality of morbidity coding among GPs. Neither the distribution of ACG frequency nor the associations observed in our study were significantly modified. In fact, the 'healthy migration effect' shown in the present study was reinforced given that the differences in the morbidity pattern of immigrants versus that of the native population became wider. Furthermore, the case mix of patients according to ACGs was consistent with that reported in other studies [37
Finally, although all immigrants have the right to request a health card identifier regardless of their legal status and provided that they are registered in the local population census, it may occasionally occur that they do not seek such service for fear of deportation (i.e., irregular immigrants) or due to lack of information [39
]. In such cases, it is possible that the same health card identifier might be used by various immigrants who attend PC health centres and use the card of a relative or a friend. The aforementioned problem could result in overestimation of the use of PC services by immigrants compared to the native population, thus increasing the differences found in this study. In other words, we believe that the consideration of this bias would not alter our findings.
Future research should focus on a comprehensive analysis of the use of health services by the immigrant population. This would entail integrating different data sources from different levels of care. In Spain, greater use of hospital emergency services by immigrants has been reported by several authors [40
], although others have ruled out this hypothesis [42
]. The use of specialist care has been shown to be lower among immigrants [43
], and similar trends have been highlighted concerning the level of double (public-private) coverage [44
]. Our results indicate that there seems to be a shift from demanded visits (during PC visiting hours) towards PC emergency visits among the adult immigrant population. Investigating whether this tendency goes beyond the PC level would lead to important findings regarding health care commissioning.
Also, this type of studies need to be reproduced with data from other urban and rural public healthcare settings in such a way that conclusions can be extended to the regional or even national level.
We did not have access to immigrants' dates of arrival in Spain. Although the increase in the immigrant population is a relatively recent phenomenon in Spain, the evolution of their patterns of health services utilisation and morbidity over time is worth studying. Several authors have shown how the health status of immigrants becomes poorer with time due to progressive deterioration of their lifestyles [25
The conclusions related to the 'healthy migration effect' presented in this study are based on information regarding health conditions obtained from GPs' electronic medical records ('attended' morbidity). Research strategies should be designed to study differences in the objective and representative 'non-attended' morbidity burden of both native populations and immigrants.