Prevalence, Median Time and Associated Factors with the Likelihood of Initial Antidepressant Change in Treatment of Major Depressive Disorder: A Cross-Sectional Study in Qatar


 Background: Major Depressive Disorder (MDD) requires therapeutic interventions during the initial month after being diagnosed for better disease outcomes. International guidelines recommend a duration of 4-12 weeks for an initial antidepressant (IAD) trial at an optimized dose to get a response. If depressive symptoms persist after this duration, guidelines recommend switching, augmenting, or combining strategies as the next step. Premature discontinuation of IAD due to ineffectiveness can cause unfavorable consequences. Hence, we aimed to determine the prevalence and the patterns of strategies applied after an IAD was changed because of a suboptimal response as a primary outcome. Secondary outcomes included the median survival time on IAD before any change; and the factors that were associated with IAD change.Methods: This was a retrospective study conducted in Mental Health Services in Qatar. A dataset between January 1, 2018, and December 31, 2019, was extracted from the electronic health records. Inclusion and exclusion criteria were defined and applied. The sample size was calculated to be at least 379 patients. Descriptive statistics were reported as frequencies and percentages, in addition, to mean and standard deviation. The median time of IAD to any change strategy was calculated using survival analysis. Associated factors were examined using three cox regression models.Results: A total of 487 patients met the inclusion criteria of the study, 431 (88%) of them had an occurrence of IAD change to any strategy before end of the study. Almost half of the sample (212 (49%); 95% CI [44% - 53%]) had their IAD changed within less than or equal to 30 days. The median time to IAD change was 43 days with 95% CI [33.2 – 52.7]. Cox regression analysis showed three statistically significant factors were associated with our secondary outcome: age, un-optimization of the dose, and absence of comorbid anxiety.Conclusions: Our findings suggest that clinicians resorted to IAD change within a median time of six weeks. However, this change often occurred without optimization of the IAD dose and might be too late. Future prospective studies with adequate randomization can better investigate the median survival time on IAD and other associated factors.

e cacy, can help prevent the disease from progressing to a treatment-resistant phase. Avoiding irrational practices such as subtherapeutic doses of IAD, premature switching between the ADs, and refraining from unjusti ed polypharmacy can help the disease to go into a remission phase [18].
In this study, we were interested in nding out the prevalence and the patterns of strategies applied after an IAD changed because of a suboptimal response towards alleviating the depressive symptoms as a primary outcome. Additionally, the secondary outcomes studied were the frequency and percentages of different change strategies applied to the IAD after suboptimal response at different time points; the median survival time on IAD before any change; and the possible risk or protective factors that were associated with the likelihood of IAD change.

Methodology
Study design, setting, and participants: This was a retrospective study conducted in the Mental Health Services (MHS), Hamad Medical Corporation (HMC) in Qatar. A dataset was extracted from the electronic health records (EHR) system (Cerner system®).
Participants were eligible for inclusion in the study if they: (i) were patients who visited the Mental Health Service as an outpatient or as an inpatient, and if they were diagnosed with MDD according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria [19]. (ii) had a new MDD episode, whether rst episode, a relapse, or a recurrence. It was di cult in a clinical practice setting to differentiate between relapse and recurrence; hence, we included both groups into a relapse (or retreatment) group for convenience. (iii) started treatment with a single AD (using a exible-dose regimen if required). (iv) had an age of 18 to 70 years old on their index date (the rst date of prescribing the IAD) (v) had an index date between January 1, 2018, and December 31, 2019. Our follow-up period took place from an index date of each patient to the date of IAD change to any strategy (switching, augmentation, or combination).
Patients were excluded if they: (i) had a prescription of AD within 6 months or less prior to the index date of initial diagnosis as MDD. (ii) had a psychotic disorder or an MDD with psychotic features, in whom antipsychotic treatment was initially prescribed, along with the AD, (iii) had a treatment-resistant depression for whom multiple MDD treatments were deemed necessary when used simultaneously in their index date (iv) had bipolar depression, obsessive-compulsive disorder, or post-traumatic stress disorder in whom giving treatment with AD was warranted. (v) Patients who were pregnant MDD on their index date. (vi) Treatment with electroconvulsive therapy, transcranial magnetic stimulation, monoamine oxidase inhibitors.

Sample Size and sampling techniques
Based on a pilot study we did in the same setting one year earlier on 70 patients with MDD, 40% of the sample had a change to their IAD within less than 28 days. Thus, with an estimate for a proportion of 0.4 and a margin of error of 0.05, the needed sample size to achieve power was at least 379 patients or more. 95% con dence interval was used. Two-sided p-value was considered signi cant if it was less than 0.05. The extracted data were validated, de-identi ed, and listed in excel sheets by six healthcare professionals including the pharmacy informatics specialist. Typical purposive sampling was applied by reviewing 800 patient charts for eligibility consideration.

Variables and statistical Analysis
The identi ed IAD with their frequencies/ percentages were tabulated. Index date from IAD to the nearest one of three pre-planned change strategies was calculated. The rst change strategy was switching which was de ned as discontinuation of the IAD and administration of another AD. The second change strategy was augmentation which was de ned as the addition of another drug, that is not an AD, to the IAD; and it was predetermined by the team members to dedicate this category only for antipsychotic use. The third change strategy was combination strategy which was de ned as either a) switch from IAD to another AD with the addition of another drug, which is not an AD or, b) combined administration of IAD with another AD.
Descriptive statistics were reported as relative frequencies in percentages as well as mean ± SD/median ± IQR (interquartile range). The Shapiro-Wilk test was used to test for normality. Median survival time of IAD before any change strategy was plotted using the Kaplan Meier curve.
Censored patients were either lost to follow-up or had no occurrence of IAD change prior to the end of the study. Time (in days) from the IAD index date to any change strategy or censoring was calculated and labeled as survival time on IAD.
The association between the risk of IAD change and only seven independent variables was studied to avoid over tting of the model. The independent variables, namely age, gender, bothersome side effects, substance use, un-optimization of the IAD dose before any change, absence of comorbid anxiety, and rst experience MDD episode were chosen based on literature review and clinical judgment of experts in our hospital.
A univariate cox proportional hazard model used rst to study the association between the risk of IAD change and individual factors to report crude hazard ratio with 95% con dence interval. Then these factors were entered in two subsequent multivariate cox proportional hazard models to get the adjusted hazard ratio (AHR) with a 95% con dence interval. The multivariate regression was done initially using a complete case analysis (CCA) approach. Then a nal multivariate regression model was performed using multiple imputations (MI) to replace all the missing values in each independent variable. For sensitivity analysis, we used all the available data in the MI to create and analyze 20 multiply imputed models for each independent variable with missing data. Data analysis was performed using IBM SPSS Statistics for Windows (Version 25.0).

Ethics
This study was conducted in accordance with the Declaration of Helsinki and the Qatari ethical guidelines for medical and health research involving human subjects. Prior to the initiation of the study, the study protocol was reviewed and approved by the institutional review board (IRB) of the Medical Research Center (MRC) in HMC under protocol number (MRC-01-20-055). Since this was a retrospective chart review, it was exempted by MRC from the requirement of informed consent. The administrative permissions were attained by the research team to access the data used in the research.

Discussion
To our knowledge, this is the rst study to assess the prevalence and patterns of strategies applied after an IAD changed because of a suboptimal response in patients with MDD in Qatar and the Middle East.
In this retrospective study, we found that the IAD was an SSRIs in 80% of the cases. Almost half of the patients (49% [44% − 53%]) had their IAD changed within less than one month, whereas the proportions of patients whose IAD changed in 31 to 90 days, 91 to 180 days, or more than 180 days were lower (20%, 19%, and 11% respectively). The most common IAD change strategy used was switching, mostly to an SSRI or to an atypical AD (40% and 37% respectively), followed by combination strategy (21%), then augmentation with antipsychotics (11%). The median time to IAD change was 43 days [33. 2-52.7]. The factors which were associated with a higher hazard to change in IAD were younger age, un-optimization of the IAD dose before any change, and comorbid anxiety.
ADs are indicated for the treatment of depression, generalized anxiety disorders, obsessive-compulsive disorder, and post-traumatic stress disorder. There are 13 ADs in HMC formulary representing the different AD classes. ADs take considerable time to induce either response or remission and for many patients, the response is considered suboptimal [7]. This lag in the AD response may lead to negative MDD outcomes including increased risk of suicidal behavior and other deliberate self-harm, psychological distress, occupational and functional limitations, and lack of adherence to medications. The results of this study showed that almost half of the patients underwent a change in their IAD treatment due to a perceived lack of response in the rst 30 days of the treatment. The International guidelines such as National Institute for Health and Care Excellence (NICE-2018) and the Canadian Network for Mood and Anxiety Treatments (CANMAT-2016) generally specify a timeframe of 2-4 weeks to switch from the IAD at an adequate dose if no response was observed [6]. The rate of switching considerably varied throughout different studies ranging from 8-40% [20,21].
We found that a quicker change in the IAD strategy was associated with younger age. There has been some evidence that older patients might exhibit a slightly slower response to AD medication even though other studies did not show any link between age and speed of response [21,22]. The possibly slower response in older patients might explain why the time to change IAD in our study was longer in older individuals. In addition, since polypharmacy is much more common in older patients, it is understandable that clinicians could be more reluctant to prescribe yet another drug, and it seems wise to wait longer before having to combine ADs or augment the AD medication with an antipsychotic drug [23,24]. This goes in line with international guidelines recommending monotherapy in older patients with MDD [25].
Patients with unoptimized IAD doses also had a higher hazard to undergo a change in their IAD. This might be due to some clinicians resorted to switch, combine, or augment IAD without optimizing the IAD dose rst. Since we basically relied on documentation of the physicians to report the reasons behind IAD change, we found some patients' les with unclear justi cations whether those patients underwent a change to their IAD due to side effects or suboptimal response. To reduce bias or imprecision, we decided to perform adjustment for "bothersome side effects" variable in the list of independent factors included in the regression models. However, the nal MI multivariate model showed statistically insigni cant association with the likelihood of IAD change.
Even though the conventional approach remains to optimize the dose of the AD medication before switching or adding another psychotropic medication, there has been some evidence that "early switching strategies" might actually yield better MDD outcomes [16,26,27]. A meta-analysis showed that the lack of early improvement (often de ned by less than 20% reduction in a depression scale score) at two weeks may indicate that changes in depression management should be considered earlier than conventionally thought. [16]. Meanwhile, a double-blind, randomized study showed that the time to regaining normal functionality might be shorter when adopting the early compared to the conventional switching strategy [26]. Another recent meta-analysis of nine studies showed signi cant associations between early improvement, response, and remission.
Nevertheless, the treatment scenario associated with the best remission rate was switching at four weeks rather than at two or six weeks [28].
Change in the IAD strategy was slower in patients with comorbid anxiety. This was consistent with previous studies showing that patients with a comorbid anxiety disorder or with anxiety symptoms concurrently with the depressive episode might have a slower onset of action of AD medication [22]. In level one of STAR*D, MDD patients with anxiety symptoms exhibited signi cantly lower rates of remission as well a signi cantly slower remission than in patients with MDD without anxiety [29]. Hence, it seems reasonable to wait longer in patients with comorbid anxiety before changing the IAD strategy.
The ndings of this study can have direct clinical guidance for health care professionals since an optimized, evidence-based use of AD medication can improve the clinical outcomes of patients with MDD; and to identify high-risk factors that could worsen the survival time on IAD such as young age and comorbid anxiety [30]. We included most of the presently available AD medications commonly used in a clinical practice setting, and we compared median IAD change time and percentages in an actual clinical situation in both inpatient and outpatient settings.
Despite the merits, several limitations in our study need to be acknowledged: the retrospective design and the reliance on medical records led to missing data. Nevertheless, most evidence comes from retrospective studies (with their inherent limitations), or from trials where "the most complex" patients are excluded (including patients with multiple comorbidities, with a neurocognitive disorder, or with acute suicidality). This may partly explain the gap that exists between the data of the literature and the clinical practice. In addition, we did not include the "IAD pharmacological group" variable in the Cox regression model because of the number of patients who had their IAD as SSRIs were incomparable to non-SSRIs users. Our sample was also recruited from the main psychiatric hospital in the state; however, patients with MDD treated in primary healthcare centers or in the private sector might have different severity of illness; hence, our ndings can probably be extrapolated to the entire population of patients treated with AD medications with caution. Furthermore, the use of psychometric scales to assess the e cacy of the AD medication could have provided a much more objective and quanti able assessment of the response; however, in this setting, measurement-based care (MBC) is not being routinely used. Finally, adherence was not assessed, and it was possible that adherence problems could have resulted from other residual confounders.

Conclusion
Our ndings suggest that in a real setting, clinicians resorted to a change in the IAD strategy (switching, combination, or augmentation) after a median time of about 6 weeks, which might be reasonable in certain cases especially in older patients or those with comorbid anxiety. However, these changes often occurred without a prior optimization of the IAD dose and might be too late. Future prospective studies with adequate randomization can better investigate the median survival time on IAD and other associated factors.
Abbreviations AD: Antidepressant