Frequent attendance to health care facilities is a problem associated with a substantial consumption of resources. In Spain, it is estimated that 14%–15% of paediatric patients are high-frequency (HF) users of health care services and that this subpopulation consumes more than one third of the working time of health professionals.1 For this reason, numerous studies have been investigating the factors that may lead to this excessive use with the aim of intervening and optimising health care quality.2–6 Several factors have been described in relation to high-frequency use, such as age or the presence of certain disease patterns.1–3,7,8

High-frequency use of health care services raises concerns as to the pressure that these users exert on health professionals to prescribe treatments more frequently,9,10 even in the absence of a disease that clearly indicates their use. There is also evidence that pressure from family members and recurrent visits may be associated with an increased use of diagnostic tests (DTs).11 In addition to causing discomfort or potentially being hazardous to patients, these DTs do not appear to reassure users regarding the disease,12 so they could be dispensed with in many instances. There is also evidence that the likelihood of DTs being ordered for any given condition varies depending on the level of care of the facility (hospital/primary care) or the type of professional managing the patient (general practitioner/paediatrician/resident physician).11,13,14

When it comes to pharmacological treatment, antibiotic prescription is particularly important in paediatrics, as these drugs are used frequently in the early years of life. It is well known that their indiscriminate use may pose a public health problem due to the development of antimicrobial resistance.15,16 As is the case with DTs, the use of antibiotics varies widely between countries,15 levels of care14 and types of prescribing professionals.17,18 There is also considerable variation in the antibiotics prescribed and the degree of adherence to clinical practice guidelines.18,19 The current evidence suggests that despite the differences between countries, health care outcomes are the same, which means that a sizable proportion of prescribed interventions could be avoided.15,20

Given the considerable variability evinced in studies on HF users between different types of health professionals and levels of care,11,17,18 we wondered whether patients that seek services in various facilities may be more exposed to antibiotic prescription and undergo more DTs compared to those that routinely visit their usual paediatrician. Thus, patient loyalty (PL) to their main physician could be another significant factor in antibiotic prescription and the ordering of DTs, although we were unable to find any studies on this subject.

Thus, we hypothesised that high-frequency attendance by paediatric patients results in increased ordering of DTs and antibiotic prescription compared to the rest of the paediatric population, especially in patients with that are less loyal to their assigned paediatrician. The aim of this study was to explore this potential association.

We collected data for 2,859 patients, and the final sample included 2,726 patients for which the data were valid and included in the analysis. We excluded 133 patients because they did not meet the inclusion criteria or data were missing in their forms. A total of 212 paediatricians in the network participated in the study. We excluded the 11 network paediatricians from the Region of Murcia because they did not have access to emergency care records.

There is no clear definition of HF use and the criteria used for its definition in different studies vary widely. The mean number of visits in a given period of time plus 1 or 2 standard deviations are commonly used thresholds. Since these are the definition criteria applied most frequently, we chose to create 2 HF groups, which also allowed us to assess whether there were any differences between what we labelled moderate HF users and extreme HF users. This seemed to be the case, as we generally found higher values in the variables under study in extreme HF users compared to moderate HF users. The proportion of HF users was consistent with the proportion reported in previous studies conducted in Spain: approximately 14% when calculated using the mean+1SD1 and 4%–5% when using the mean+2SDs.4,6

We collected data on the variables that we believed may have the most impact on the frequency of use by patients. An important aspect to take into account in the investigation of HF users is whether children that use services often are justified in doing so. In this regard, there appear to be different patterns of illness that have an impact on frequency of use.6 In fact, there is evidence that the number of types of morbidity present in a patient is associated with the extent of health care utilisation.7 For this reason, our study included an analysis of the number of morbidities using the categories defined by Starfield, and our results were consistent with those of the original study. This variable and the presence of chronic illness seemed to have the greatest impact on HF use, and therefore both should be taken into account in future research.

We did not find any studies that analysed the association with patient loyalty. Thus, we believe that the descriptive data presented here may be of interest. We ought to highlight that out of every 3 visits scheduled on patient demand, 1 was not with the assigned primary care paediatrician. Approximately half of the patients exhibited high PL. This was the case in all age groups, while the proportion of patients with chronic disease was greater in the high PL group compared to the low PL group. High-frequency users mainly belonged to the two medium-loyalty groups. The characteristic profile of patients in the low PL group was a child aged 7–14 years without chronic disease who used health care services infrequently.

As for antibiotic prescription, there is recent evidence of the considerable variability that exists between European countries.15 Since health indicators do not differ significantly between these countries, there must be somewhere antibiotic utilisation is excessive in the paediatric population. Although this was not the aim of our study, we ought to mention that the antibiotic prescription rate that we found was similar to those reported for the countries with the highest rates of prescription, which are far above those of the countries with the lowest rates.

Although we found no published evidence on diagnostic tests, it is likely that the same is the case with them, and that many could be avoided.11,12 We ought to highlight that the most frequently ordered DT was the plain radiograph, which is an invasive test.

The association between antibiotic prescription and high frequency utilisation has already been described in the literature.6,9 Our findings are conclusive in this regard, as HF use was strongly and significantly associated with antibiotic prescription. As we expected, this association became stronger the greater the frequency of use, so that antibiotic prescription was double in moderate HF users and triple in extreme HF users compared to patients that were not HF users. The results were similar for the performance of DTs. It may be that the pressure exerted by the family on health care professionals and the desire of paediatricians to alleviate the anxiety of the family are significant factors at play.9–11 Specific strategies for the management of HF users should be developed to redirect this overutilisation of health care resources, which also seems to lead to these patients being overmedicated and undergoing a greater number of tests.

We expected to find a strong association of low PL with antibiotic prescription and performance of DTs. Our rationale was that these patients receive care from different providers that are not acquainted with the family and are unable to follow up the case, which could influence the approach of providers to their management. We did in fact detect an association, but it was weaker than the one observed in relation with HF use and was not statistically significant in most of the groups under study. We mainly observed this association in patients with low-medium and medium-high PL. The low PL group, which constituted a minority of the sample and was characterised by infrequent use of health care, had antibiotic prescription and DT performance rates that were similar and even lower than those observed in the high PL group, which made sense, as patients in the low PL group were older and less likely to have chronic disease.

There is an intrinsic limitation in studies on health care utilisation and antibiotic prescription, which is that it is all but impossible to get an accurate account of all the services and prescriptions generated in settings outside the public health system.10,16 It is also very difficult to accurately track the dispensation in pharmacies of antibiotics without a prescription. However, only 13.2% of patients in our sample were covered by private insurance (and all of them had access to the public health system, too). Therefore, and considering the large size of the sample, we believe that our study provides an adequate representation of the phenomenon under study, as it analysed data that could be retrieved from the public system, which is used by a vast majority of patients.

Another possible limitation is that there may be disease-related factors that have yet to be identified that influence the frequency of use. In any case, we included the factors currently known, and our study is the first on the subject that includes and adjusts results based on patterns of morbidity.

Thus, we conclude that high-frequency use is strongly associated with antibiotic prescription, and that high-frequency users undergo more DTs. In contrast, patient loyalty had a significantly lesser impact on antibiotic prescription and the ordering of DTs.

The authors have no conflicts of interest to declare.