Discussion
The use of HCQ early in the Covid-19 pandemic resulted in a substantial increase in suspected ADR reporting associated with this medication. The rise in HCQ use [14][15]–[17], along with the concerns about its administration to Covid patients in spite of the lack of sufficient scientific evidence, and the fact that everybody’s eyes were focused on any aspects related to the pandemic because of the worldwide mass media coverage may have played a part in this increment in suspected ADR reporting. The subsequent decline in reporting is likely to be related to calls to caution [7] as well as to the publication of findings against HCQ [4],[5].
Both the patient profile and HCQ use pattern proved different in our two groups (i. e, Covid and no-Covid patients). In the case of no-Covid patients, the cases most often corresponded to middle-aged women treated with low daily HCQ doses in a majority of cases. These patients did not receive high daily doses (≥ 800 mg/day), even though the total administered dose was considerably higher (median: 21,600), which is consistent with the fact that the treatment length was substantially longer. It should be underscored that, in both groups, a high percentage of case notifications did not include any data on the administered doses. Though important, the completion of the box ‘Administered doses’ in the report form is not obligatory to validate the notifications submitted to SEFV-H. As expected, the main indications were rheumatic conditions, arthropathies, and lupus erythematosus; however, we also found off-label HCQ indications. Nearly half of patients took only HCQ, and reported co-medications were principally corticoids, other immunosuppressants, and painkillers. By active ingredient, methotrexate, prednisone, enoxaparin, and mycophenolate were the most frequently administered to these patients. Most of cases of Covid patients were males, aged approximately a decade over their no-Covid counterparts, who, in a majority of cases, were given medium daily HCQ doses, and a low total dose in 100% of cases, a finding that is consistent with the fact that the length of their treatment was considerably shorter. Virtually all patients were treated with medications other than HCQ, notably antimicrobials, corticoids, and immunosuppressants. By active ingredient, azithromycin and the lopinavir+ritonavir combination were the most frequently combined with HCQ in a high percentage of Covid patients. This finding mirrors the kind of patients who were treated with HCQ as well as the usual HCQ administration guidelines [7],[18],[19]. These differences in patient features, indications, HCQ doses and treatment length and use of medications other than HCQ have by themselves the potential to change the safety profile of the drug, as was the case of the modifications of ADR profile we found in our study.
Firstly, there was a 24% excess of serious cases among Covid patients. Covid patients on HCQ were, by definition, persons who have been admitted to a hospital, and, therefore, it is assumed that they suffer from a more serious disease, which may contribute to explain why Covid patients in our study were in worse health condition. On the other hand, because of the work overload of hospital healthcare professionals in Spain in that time, it is reasonable to assume that only the most severe ADRs were reported to the pharmacovigilance system. The overall profile of ADRs potentially associated with HCQ we found in no-Covid patients approximated to that in earlier studies [7],[20]. In no-Covid patients on HCQ, the prevailing ADRs are cutaneous in nature, followed by eye disorders, general health problems, nervous system disturbances, and gastrointestinal disorders; while in Covid patients the most frequent ADRs are hepatobiliary disorders, followed by gastrointestinal, cardiac, general disorders and intoxications, and skin disorders. Importantly, in the present study, the main differences between the two groups corresponded to hepatobiliary, gastrointestinal, ocular and cardiac disorders, which are usually reported more frequently in Covid cases, on the one hand, and skin and nervous system disorders, the reporting frequency of which was higher in no-Covid cases, on the other (Figure 3).
The percentage of hepatobiliary disorders reported among the Covid cases is a striking finding in our study. According to the reports released by SEFV-H during the Covid-19 pandemic, most of hepatobiliary disorder notifications came from only a single hospital, in which active pharmacovigilance surveillance specifically for hepatotoxicity was conducted [21], therefore hepatobiliary ADRs might be overrepresented. Furthermore, it should be borne in mind that SARS-Cov-2 infection itself is associated with liver damage [22], which suggests that, in our study, this finding may be flawed by an indication bias. Nevertheless, earlier studies have also reported hepatobiliary disorders to rank either first or second in frequency among Covid patients on HCQ, particularly when this drug is combined with other medications [23]–[25]. The disproportionality analysis we conducted showed that there was an association between HCQ treatment and both overall liver and severe liver disturbances only during the pandemic period (Table 2).
Additionally, gastrointestinal disorders are reportedly the most frequent ones upon HCQ treatment initiation. This holds true for both Covid and no-Covid patients [26]. In our study, the gastrointestinal tract is the second most frequently involved system in Covid patients on HCQ, which is in line with the results from a Portuguese prospective study [23]. Indeed, in our study, HCQ treatment in Covid patients is associated with “non-infectious diarrhoea”, “ischaemic colitis” and “gastrointestinal perforation, ulceration, haemorrhage or obstruction” (see Table 2). Of note, the latter HCQ gastrointestinal disturbance had not been reported anywhere before Covid-19 pandemic [7].
Cardiac involvement represented one of the major safety concerns when HCQ was used for Covid treatment, especially given its use in combination with azithromycin.. [27]. In the present study, cardiac derangements have been reported to occur twice more frequently in Covid patients as compared with their no-Covid counterparts. In fact, according to disproportionality analysis results, three cardiac disturbances ranked first, second and third, with cardiac arrhythmias, specifically TdP/QTp, ranking first; and this was the heart problem for which we found the largest number of differences in relation to disproportion between the two patient groups (Table 2). This striking increase in TdP/QTp in these patients may result from the use of higher HCQ doses [26] or, alternatively, may be due to the use of concomitant medications such as azithromycin [27], as shown by the Ω statistic estimation (Table 3). Likewise, it is an outstanding finding that the summaries of product characteristics of some of the drugs currently commercialised in Spain that contain HCQ do not include this risk in spite of mass media information impact and safety statements released by both regulatory authorities and scientific societies [7],[28]. Concerning cardiomyopathy, which was a cardiac ADR reported in our study, we did not find any important differences in ROR for the two groups of patients, despite cardiomyopathy was associated with lengthy treatment with HCQ [26].
When HCQ was first used as a therapy for COVID-19, another cause of concern was the attendant risk of retinopathy. Although this is an infrequent complication, eventually it can cause irreversible blindness. According to current evidence, retinopathy is related to HCQ dose and treatment length. So, the most recent clinical guidelines recommend not to exceed 5mg/kg body weight/day in order to prevent this HCQ therapy complication [19]. In Spain, with a patient mean weight of 70 kg, this recommendation would correspond to a dose of 350 mg/day. In our study, most of doses given to Covid patients exceeded this figure. Nonetheless, we found that retinopathy was more prevalent in no-Covid patients in both frequency and disproportionality. Therefore, it could be assumed that the risk of eye disturbances is more strongly associated with HCQ treatment length than is with the dose, which would be in keeping with results from earlier studies [26].
Skin reactions have been reported most frequently in HCQ treatment for no-Covid patients. They are well established reactions in patients on HCQ, and may become apparent in the form of rashes, itching and/or hyperpigmentation. However, some of these skin adverse reactions may be more worrying, as is the case of Stevens-Johnson syndrome (SJS) or toxic epidermal necrolysis (TEN), and acute generalised exanthematous pustulosis (AGEP). In some cases, they manifest themselves with psoriatic lesions and hair loss or discoloration [7],[29]. Such HCQ-related skin reactions can be classified according to the accumulated dose, the most common being reactions that become apparent following the administration of high accumulated doses [29], which is in keeping with our finding that skin adverse reactions were most often reported in patients undergoing prolonged HCQ therapy, that is, in no-Covid patients. However, when looking at severe skin reactions, including blistering and exfoliative conditions, we noted that these reactions were notified relatively more frequently in Covid patients, and furthermore ROR was higher in the pandemic period (Table 2). HCQ accumulated dose range in which skin reactions appear is very broad. In fact, such reactions may become evident with accumulated doses as low as 3,000 or 4,000 mg, as is the case of some severe skin complications such as AGEP or Drug reaction with eosinophilia and systemic symptoms (DRESS) [29]. Thus, it can be speculated that HCQ-related skin reactions in the setting of COVID-19 therapy either were less frequent but more severe or, alternatively, only the most severe reactions were notified.
The evidence on the risk of suicidal/self-injury behaviour has been conflicting, and, up to date, there has been insufficient evidence concerning the relationship of these psychiatric conditions to HCQ use [30]. However, during the Covid-19 pandemic, it was one of the causes of concern that led to change the summary of product characteristics of HCQ. [31],[32]. While few cases had been reported during the pre-pandemic period, this signal could have been detected at that time. (Table 2).
Aside from the aforementioned psychiatric, hepatic and gastrointestinal disorders and cardiac arrhythmias, we found further ADRs associated with the use of HCQ in Covid patients that are usually overlooked or not are observed so distinctly when HCQ is used for other indications, such as dyslipidaemias, shock, ischaemic colitis, kidney disturbances, noninfectious encephalopathy, and acute renal failure (Table 2). During the pre-pandemic period, it had been reported malignant adverse reactions, which may be explained by HCQ immunomodulatory effects. However, these malignant ADRs should be interpreted cautiously. Likewise, cytopaenias, including agranulocytosis, are well known HCQ adverse reactions. Their disproportionality is lower in Covid patients, which may be due to a shorter exposure time to the drug in these patients. Concerning dyslipidaemias, it is remarkable that, in our study, in all cases HCQ was given concomitantly with lopinavir + ritonavir, medications for which dyslipidaemias are well established ADRs. Therefore, it is reasonable to assume that, in the present study, the use of lopinavir + ritonavir was a confounding factor. Alternatively, these ADRs might be due to a true interaction, a possibility suggested by the results we found with the Ω statistic.
We obtained significant estimators for interactions that, curiously enough, only were found when HCQ was used to treat Covid patients. Some of these interactions are known, such as the risk of TdP/QTp with the concomitant use of azithromycin and, to a lesser extent, lopinavir and ceftriaxone. On the contrary, other interactions are less known or even unknown, like severe hepatic derangements and dyslipidaemias.
Pharmacovigilance database-based studies are flawed with obvious limitations. The major drawback is probably under-reporting, which, tough difficult to accurately estimate, has been reported to be as high as 90% [33]. This, along with the lack of data on the comsuption of the drug in question, prevent from accurately estimating the actual incidence of the reported ADRs. Furthermore, the number of cases notified to the database largely relies on a series of factors, such as the involved medication (commercialisation time, use in the clinical setting, current knowledge on the drug and so forth), and the reporting person’s profile (available time, knowledge, expertise, and degree of commitment with routine pharmacovigilance activities, etc.). Additionally, the relevance and impact of all these limiting factors may vary with time or other circumstances, as it was the case with Covid-19 pandemic. No doubt, public’s interest and concern, along with the mass media focus, have markedly influence on ADR reporting during the pandemic. On the other hand, it should be kept in mind that the Covid-19 pandemic had a major impact on the Spanish healthcare system running. However, it is not possible to precisely understand the impact this factor had on the ADR spontaneous reporting system and the pharmacovigilance databases. Finally, the lacking clinical data in notification forms is another constraint to be considered.
Still, spontaneous reporting systems presents several advantages: they cover all types of authorised drugs; it is a simple, quick and economical method enabling to generate hypotheses and identify new potential safety concerns involving drugs, notably rare, infrequent or unexpected events. Several studies showed that, despite their shortcomings, spontaneous reporting systems have the potential to identify early and efficiently emergent risks associated with the use of medications [34]–[37].