Our study shows that: (i) management of TIA in daily clinical practice of the Italian Lombardy Region is suboptimal since in the first year after discharge little less than one in ten patients (8.6%) did not use blood pressure- and lipid-lowering agents, antidiabetics, antiplatelet and oral anticoagulants, nearly half of them (49.7%) have not been underwent to diagnostic procedures such as carotid and cardiac doppler ultrasonography, while one in twenty of them (5.0%) did not show any evidence of exposure to drugs, diagnostic procedures nor laboratory tests; (ii) patients with TIA diagnosed in hospital represent a high-risk group since 11.4 and 34.7% of them died or was hospitalized for stroke or AMI within 1 year and 5 years from index discharge respectively; (iii) the onset of a TIA represents a special opportunity for preventive intervention, since patients who used drug therapies and diagnostic procedures commonly recommended by guidelines showed a 59% risk reduction of death or cardiovascular hospitalization with respect to those who did not use any medical device.

A growing body of literature confirms the high rates of death and re-hospitalization during the first year after TIA. Patients recruited into clinical trials after a TIA have an annual risk of important vascular events (death from all vascular causes, non-fatal stroke, or non-fatal myocardial infarction) between 4 and 11% [22–24]. Similar figures have been reported from population-based observational studies being the corresponding rate around 9% per year [25], while the 10-year risk of stroke, myocardial infarction, or vascular death was reported to be 43% [5]. Our study supports the available data showing that patients discharged with a diagnosis of TIA had from 1.5-fold to 2.1-fold higher risk of unfavourable outcomes with respect to individuals of the same age and gender with no signs of cerebrovascular event.

Because medicines, diagnostic and screening investigations play an important role in secondary prevention for TIA patients [26–29], changes to the usual medicine regimen and diagnostic procedures can be expected for them [30]. Consistently, our study showed that users of blood pressure- and lipid-lowering agents, antiplatelet therapy, cardiac and carotid doppler ultrasonography and screening for diabetes and hyperlipidaemia increased after TIA occurrence. However, similarly to other reports [31–33], our findings show that several patients did not receive even one of these drugs and diagnostic procedures within one year after discharge.

Other findings are noteworthy and also warrant discussion. First, adverse outcomes were reduced of more than a 25% among patients who were treated with lipid-lowering, antiplatelet and anticoagulant agents. Although this is an expected finding according to current evidence-based recommendations on the prevention of future stroke among survivors of ischemic stroke or TIA [34], we suspect that the observed protective effect might be in part explained by the fact that drug exposures likely reflect general medical attention. This would explain because a similar protective effect was conferred by so different therapeutic actions. Second, patients who during follow-up used antidiabetic agents, as well as those who underwent a screening with glycosylated haemoglobin (HbA1C), were at higher risk for the considered outcomes likely because diabetes is known to be associated with unfavourable consequences after a cerebrovascular event occurs [35]. Analogously, patients who during follow-up used antiarrhythmic agents, and those who underwent electrocardiography, had worse prognosis likely because these exposures are proxies of high cardiac risk. Third, we observed that during the one-year period before the TIA occurrence, patients at their first episode of TIA (cases) had higher use of drugs for treating hypertension, cardiac arrhythmias and type 2 diabetes mellitus than patients who did not show evidence of cerebrovascular outcomes (controls). This finding supports the important role of hypertension, arrhythmias and diabetes as risk factors for TIA [36–38], confirming the potential benefits of primary prevention through the control of these conditions.

The present study has several elements of strength. First, the investigation was based on a very large unselected population, which was made possible because in Italy a cost-free healthcare system involves virtually all citizens. Second, the drug prescription database provided highly accurate data, because pharmacists are required to report prescriptions in detail in order to obtain reimbursement, and incorrect reports about the dispensed drugs have legal consequences [39]. Third, patients were included if there was no previous evidence of cerebrovascular disease over several years, which makes our results relevant in the setting of primary prevention of stroke. Finally, patients were identified from the point of the first episode of hospitalization for TIA and the complete sequence of following healthcare services supplied by the NHS was known.

Our study has also some limitations. One, because of privacy regulations, hospital records were not available for scrutiny, which means that diagnosis of TIA could not be checked. Owing the lack of evidence from up-to-date studies performed in the healthcare system of the Region we study, misclassification cannot be completely excluded in our setting. For example, if benign conditions (e.g., migraine, syncope and seizure) were mistaken for TIA [40], a portion of patients at low risk had been included in our cohort. Although our main result was confirmed by including patients underwent to neuroimaging at the index hospitalization, the lack of clinical information useful for risk stratification and prognostication, is a main cause of systematic uncertainty of our study.

Two, not all patients who experience a TIA are hospitalized [41], thus implying that our study selectively included more severe patients, and/or those at whom more clinical attention was reserved. Three, drug treatment was derived from drug prescriptions, which requires, however, the assumption that drug prescription corresponds to drug consumption [42]. There is, however, no guarantee that this is always the case, and indeed it is likely that in a number of patients the prescribed drugs are not consumed. This implies, however, that in the real-world, the actual use of medicines may be even still less than the measured prevalence of drug users.

Finally, the more relevant question in interpreting our findings is whether the observed results are due to our inability to fully account for lower adherence with guidelines to those patients at higher risk of adverse outcomes. That is, the reduction in adverse outcome associated with a better adherence to guidelines and a higher use of healthcare services might rather have been generated by a worse clinical profile, e.g., the severity of ischemic lesion and comorbidities. Yet, TIA clinical presentation (i.e., anterior circulation/posterior circulation TIA) could affect the adherence to medicaments, since, as well known, posterior circulation TIA are often misdiagnosed, representing a challenge in vascular neurology [43, 44]. However, because it is well known that patients with worse clinical profile more likely adhere to medical care [45], the observed risk reduction might be larger than that quantified by our study. Thus, our results are likely to reflect the favourable effect offered by healthcare services based on guidelines on cardiovascular protection.