195 Discussion 7 bias was probably diluted and unlikely to affect the results of our studies. Regarding strengths, the use of routinely collected healthcare data for medical research has many advantages, providing relatively easy access to rich, ecologically valid, longitudinal data from large populations (64). It reflects daily practice and combining two different registries at the patient level makes it possible to examine new causal associations. A second methodological consideration is the use of proxy indicators as in chapters 2 and 6. Proxy indicators, such as ASI or size of a primary care practice can be used where it is not possible to extract the desired endpoint variable, in these cases antimicrobial selection pressure and time per patient visit, respectively, from available healthcare registries. Advantages of these proxy indicators are their availability, reproducibility and measurability compared to the desired endpoints. A disadvantage, however, is the somewhat simplified representation of reality. A third methodological consideration is the context in which the studies took place. The main country of research in this thesis was The Netherlands, which differs from other European countries in a variety of ways. For example, the number of antimicrobial prescriptions in The Netherlands is lower compared to most European countries (65), which could be due to the fact that GPs in The Netherlands are both well informed and constrained by restrictive guidelines, leading to prudent antimicrobial prescribing. Consequently, AMR prevalence is lower compared to most other European countries (65). If AMR prevalence in a country is low, GPs already tend to prescribe narrow-spectrum antimicrobials, helping maintain the low prevalence of AMR. GPs in The Netherlands function as gatekeepers in the healthcare system and all inhabitants are registered with only one primary care centre. Both of these contextual factors help lower the number of antimicrobial prescriptions (66). Despite the relatively lower number of antimicrobial prescriptions and low prevalence of AMR in The Netherlands, it is reasonable to generalize the results from chapters 2, 3, 4 and 6 to other countries, as for example the high number of seemingly inappropriate antimicrobial prescriptions for RTIs described in chapter 6 reflects results of many previous studies in other countries (67-70). Our study underlines the fact that inappropriate antimicrobial prescriptions for RTIs may be high, even in a country with a low overall antimicrobial prescription rate. Despite the low overall rate of antimicrobial prescription there is still room for improvement in The Netherlands, which could act as a reference point for other countries. Furthermore, our findings on specific migrant backgrounds may be reproducible in other European countries, although these findings may need to be reconfirmed in their specific context.
RkJQdWJsaXNoZXIy MTk4NDMw