Reduction of Patient-Reported Antidepressant Side Effects, by Type of Collaborative Care

Adequate trials of antidepressant medications (for example, evaluating appropriate dosing for an adequate period) can improve clinical outcomes in major depressive disorder (1). Interventions to optimize antidepressant therapy have demonstrated a resolution of, or substantial decrease in, depressive symptoms, as well as improved health-related quality of life, social and role functioning, and work performance (2–4).

Side effects from antidepressant medications are common and can adversely affect quality of life. Side effects include, but are not limited to, gastrointestinal disturbances (nausea, vomiting, and diarrhea) (5,6), changes in sleep patterns (insomnia and hypersomnia) (6,7), sexual dysfunction (decreased libido and premature ejaculation) (8,9), headache, and anticholinergic changes (dry mouth, dizziness, and vision change) (5,6). Nearly all (91%) patients beginning a selective serotonin reuptake inhibitor (SSRI) report experiencing at least one side effect (10), of whom half report three or more side effects (11) and half report experiencing moderate to severe side effects (11,12). A meta-analysis examining risks and benefits of antidepressants for major depressive disorder found that in efficacy trials of second-generation antidepressants, approximately 63% of patients experienced at least one antidepressant side effect, but the frequency varied among medications (13,14).

Antidepressant side effects can counteract the beneficial impact of antidepressants by reducing patient well-being and functioning and can decrease adherence to antidepressant treatment. Indeed, antidepressant adverse effects have been found to predict nonadherence (11,15,16). Among patients discontinuing or switching SSRIs, 36%−62% reported that the primary reason was side effects (11,17). Not all antidepressants have the same likelihood of causing a side effect, nor do all patients experience the same side effects from a given medication (7). Comorbid conditions can increase the likelihood of developing antidepressant side effects. In the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study, depressed patients with comorbid anxiety, a substance use disorder, or both had a higher number of side effects compared with others, and side effects were more severe and more likely to result in discontinuation (18). The number of medications a patient is taking (for both general medical conditions and psychiatric disorders) may be the strongest predictor of side effects (14). Patients’ beliefs and attitudes toward medications, including concerns about side effects, have also been found to be related to adherence to medications for both general medical and psychiatric conditions (17).

Collaborative care models improve antidepressant initiation and adherence in primary care settings by providing patient education about the potential risks and benefits of antidepressants and by proactively monitoring symptoms, adherence, and side effects in order to adjust medications when side effects or nonresponse is detected. Trials of collaborative care have demonstrated increased initiation of and adherence to antidepressant therapy (19–21). However, to the best of our knowledge, no studies have examined the impact of collaborative care models on the side effects reported by patients.

This study utilized data from a pragmatic, multisite comparative-effectiveness trial comparing telemedicine-based collaborative care (TBCC) with practice-based collaborative care (PBCC) for patients treated for depression in federally qualified health centers (FQHCs) (22). In the parent study, patients randomly assigned to TBCC had significantly more nurse care manager encounters with presence or absence of side effects documented in the medical record during the first six months (incidence rate ratio [IRR]=4.22, 95% confidence interval [CI]=3.32–5.36, p<.001) and the second six months (IRR=4.46, CI=2.75–7.25, p<.001) of the intervention. Side effects detected by the nurse care managers in the TBCC arm were discussed by the off-site depression care team (telepharmacist, telepsychologist, and telepsychiatrist), which provided medication recommendations (such as titrate dosage) to the prescribing primary care provider. In the PBCC arm, the prescribing primary care provider did not receive medication recommendations. In addition to the care manager’s monitoring of side effects, patients were interviewed by blinded research personnel who systematically assessed the occurrence and severity of side effects related to the most recently prescribed antidepressant medication.

In a secondary data analysis, reported here, we hypothesized that patients receiving TBCC would report fewer moderate to severe side effects during the research interview compared with patients enrolled in PBCC.

Methods

This study was approved by the University of Arkansas for Medical Sciences Institutional Review Board; a detailed description of the parent study design is available elsewhere (22). Participants were recruited between November 2007 and June 2009 from five Arkansas FQHCs. Patients who screened positive for depression and gave informed consent, stratified by clinic, were randomly assigned to TBCC or PBCC (N=364). For this secondary analysis, we identified participants from the PBCC and TBCC groups who were prescribed antidepressant medication and who reported at any of the three follow-up research interviews adhering to it and having moderate to severe side effects that they attributed to the antidepressant. Adherence was defined as taking the full prescribed dosage every day in the previous month.

Collaborative Care Description

The PBCC intervention involved on-site primary care providers and on-site nurse depression care managers (DCMs). The primary care providers could refer patients to community-based specialty mental health services, but these referral resources were not part of the PBCC team. Patients may have consulted with a pharmacist when they filled their antidepressant prescription at the pharmacy, but these pharmacists were not part of the PBCC team. The TBCC intervention involved five types of providers: an on-site primary care provider, an off-site DCM, and an off-site pharmacist (Pharm.D.), psychologist (Ph.D.), and psychiatrist (M.D.).

The DCMs for both PBCC and TBCC were registered nurses or licensed practical nurses who were trained with the same methods and materials. In both conditions, the DCM protocol specified encounters at baseline, every two weeks during the acute treatment, and every four weeks during continuation treatment. Patients received the intervention for up to 12 months. The PBCC DCM encounters were conducted either face to face or by telephone; all TBCC encounters were conducted via telephone. In both PBCC and TBCC, encounters with the DCM included symptom monitoring with the nine-item Patient Health Questionnaire (PHQ-9), patient education and activation, barrier assessment and resolution, self-management and behavioral activation goal planning, monitoring and promotion of medication adherence, and monitoring and management of side effects. The DCM addressed side effects during the clinical encounter by educating patients about antidepressant side effects, asking the patient if he or she was experiencing antidepressant-related side effects, and helping the patient manage existing side effects (such as advising the patient to take medications with meals to avoid nausea).

In TBCC, the telemedicine team held weekly conference calls to discuss new patients and patients who were not responding to treatment. During team meetings, the telepharmacist was responsible for reviewing the patient’s entire medication regimen and identifying potential problems, including side effects and drug interactions. The telepharmacist worked with the team to select antidepressants that would minimize side effects and drug interactions. If a medication-related problem was identified during a DCM encounter, the DCM discussed it with the team. The team could recommend a new antidepressant or dose, ask the DCM to obtain more information from the patient, or ask the telepharmacist to call the patient and conduct a detailed medication history. In PBCC, the DCM received no supervision from a mental health specialist and worked directly with primary care providers to develop and implement treatment plans.

Research Assessments

Sociodemographic and clinical case-mix factors were collected at baseline, six, 12, and 18 months via blinded telephone interviews. Depression severity was measured with the Hopkins Symptom Checklist (HSCL) (23). Participants were asked several questions related to their beliefs about antidepressant medications and side effects. Using a rating scale from 1 to 5, with 1 indicating strongly agree and 5 indicating strongly disagree, participants rated statements about any side effects they were experiencing with antidepressant medications (24), and they rated the extent to which they felt antidepressant medications were effective in treating depression. Using a scale from 1 to 4, with 1 indicating definitely acceptable and 4 indicating definitely not acceptable, participants rated the acceptability of taking antidepressants for treating depression (19,20).

During the research interviews, participants were asked to gather all pill bottles for medications that had been prescribed for treatment of personal and emotional problems during the past six months. The interviewer asked each participant to read from the pill bottle the medication name, strength, and how many times a day the medication was taken. Beginning with the antidepressant medication the participant identified as “most recently prescribed,” the interviewer stated that people sometimes have side effects from antidepressants and, naming the drug that the patient had identified as most recently prescribed, asked whether the patient had experienced specific side effects from it.

The interviewer asked about 14 side effects: sleepiness during the daytime, trouble falling asleep, nausea or upset stomach, difficulty urinating, dizziness or lightheadedness, difficulty with sexual activity, dry mouth, constipation or diarrhea, blurred vision, headaches, weight gain or weight loss, anxiety or jumpiness, muscle stiffness or tightness, and feeling hot or sweaty. Participants could also use “other” to report other side effects, and any such response was counted as a side effect and rated as described below. If the participant reported having one or more side effects from that antidepressant, the interviewer recorded the severity of the side effect. The interviewer provided the following definitions to the participant: a mild side effect was defined as one that did not interfere with the participant’s normal activities. A moderate side effect was defined as one that interfered somewhat with normal activities. A severe side effect was defined as one that was so bothersome that the participant could not perform normal activities. The participant then rated that specific side effect as mild, moderate, or severe. Following the Nebeker and colleagues (25) approach, we focused on side effects that are common with antidepressants but elicited the patient’s perspective on whether he or she was attributing side effects to the antidepressant medication. Data from these interviews were used to calculate the daily antidepressant dose, which was categorized as starting, usual, or high (26).

Data Analysis

Unadjusted analyses were used to describe differences between the PBCC and TBCC groups. Because we were interested in modeling the likelihood of reporting a side effect that patients felt adversely affected normal activities, we counted the number of moderate and severe side effects that each patient reported. Patients were categorized as reporting one, two, or more than two side effects. This three-level variable was specified as the dependent variable in an ordered, trichotomous logistic–mixed-regression model that included data for all 190 participants and included repeated measures at six, 12, and 18 months. Proc Glimmix was used to calculate the odds ratio (OR) of side effects being reported in the TBCC group at six, 12, and 18 months. The main side effect for the group was used to test the hypothesis that patients randomly assigned to TBCC reported fewer moderate to severe side effects at the six-, 12-, and 18-month follow-ups compared with those assigned to PBCC.

The model also included case-mix factors that, according to our literature review, might be associated with reporting medication side effects. Time-variant, case-mix factors that could vary at the six-, 12-, and 18-month follow-up research assessments included most recent antidepressant prescribed (citalopram, the most commonly prescribed antidepressant, was specified as the reference group), antidepressant dose (starting, usual, or high), number of antidepressants currently prescribed, and number of medications currently prescribed for general medical problems. Time-invariant, case-mix factors measured at baseline included depression severity, comorbid mental disorders, number of chronic general medical conditions, sociodemographic characteristics, prior antidepressant treatment, and beliefs about antidepressants. Being treated for depression prebaseline was included as a covariate because we thought this might influence patients’ experience with medication side effects. Patients’ beliefs included whether they believed antidepressants are an acceptable and effective treatment for depression and their concerns about antidepressant side effects.

Results

As shown in Table 1, 190 patients were included in the analysis; they were about 50 years old, 83% were female, 80% were Caucasian, and 48% were uninsured. The mean±SD baseline HSCL score was 2.0 out of a possible 4.0, with higher scores indicating greater severity. Most patients (84%) were already being treated with antidepressants at study enrollment; the remaining 16% initiated antidepressant therapy after baseline. The mean number of comorbid, chronic general medical illnesses at baseline was 4.9; psychiatric comorbidity was common, especially generalized anxiety disorder (66%). The most commonly prescribed antidepressants at baseline were citalopram (17%) and paroxetine (15%). No patient reported taking tricyclic antidepressants or bupropion. At baseline, most of the patients (62%) were prescribed a starting-level dose, and the mean number of antidepressants prescribed in the six months prior to enrollment was 1.2. There were no statistically significant clinical or demographic differences between participants in the TBCC and PBCC groups.

Based on the data gathered from the blinded research interviews, the mean number of moderate or severe side effects was 2.09±2.75 at six months in the PBCC group, compared with 1.10±1.77 in the TBCC group (p=.008). At 12 months, the PBCC group reported a mean of 1.90±2.15 side effects, compared with .95±1.50 in the TBCC group (p=.002). At 18 months, the PBCC group reported a mean of 2.44±2.80 side effects, whereas the TBCC group reported 1.52±2.50 side effects (p=.048). As reported previously, from baseline to six months, the DCM addressed side effects and did so a mean of 5.18±2.29 times in the TBCC group and 1.24±1.49 times in the PBCC group (p<.001) (22). From six months to 12 months, the DCM addressed side effects a mean of 3.73±2.45 times in the TBCC group, compared with .80±1.19 times in the PBCC group (p<.001) (22).

In the trichotomous logistic regression model, participants randomly assigned to the TBCC group were significantly less likely to report side effects at six, 12, and 18 months compared with the PBCC group (OR=.31, p<.001). The likelihood of reporting side effects did not vary over time, as indicated by the lack of statistical significance at months 12 and 18 compared with month 6 (Table 2). There were no statistically significant differences in reported side effects between citalopram (the most commonly prescribed antidepressant) and other antidepressants. However, the number of antidepressants prescribed at follow-up was associated with an increased likelihood of reporting medication side effects (OR=2.66, p=.02). There were no other statistically significant differences in other covariates tested in the model.

Discussion

The study participants recruited for this pragmatic, randomized comparative-effectiveness trial had mostly treatment-resistant depression (that is, they were being treated for depression at enrollment and had PHQ-9 scores ≥10, indicating clinically significant depression), with multiple general medical and psychiatric comorbidities (22). We previously reported that, compared with the PBCC group, the TBCC participants at follow-up were not more likely to be prescribed antidepressants, were not prescribed more antidepressants (by switching or augmenting), were not prescribed higher dosages, and were not more adherent. Nevertheless, compared with participants in the PBCC arm, participants randomly assigned to TBCC were significantly less likely to report having moderate to severe side effects from their most recently prescribed antidepressant at the six-, 12-, and 18-month follow-ups. This is likely a result of the increased side effect monitoring and management by the DCM in the TBCC group, such that patients did not report side effects during the blinded research interview. The higher number of encounters in the TBCC group, where the DCM discussed side effects by educating the patient about side effects and helping the patient to manage them (for example, advising that medications be taken with meals to avoid nausea or at bedtime to minimize daytime drowsiness), is likely the mechanism of action of the intervention.

Theoretically, recommendations from the telepsychiatrist to switch medications or titrate dosages when severe side effects were reported or side effect management by the telepharmacist could have led to fewer reported side effects in the TBCC group. However, patients in the TBCC group were not more likely to be prescribed more medications (switching or augmenting), and the telepharmacist had direct contact with <6% of participants in the TBCC group. These results indicate that care managers working with a telemedicine-based clinical team can reduce the number of side effects that patients attribute to antidepressant medications.

Our model included risk factors that, theoretically, would increase the likelihood for antidepressant side effects. However, only the number of antidepressants prescribed significantly influenced the likelihood of reporting antidepressant side effects. Unlike findings from the STAR*D trial, comorbid psychiatric conditions did not influence the likelihood of reporting side effects (18).

This study had several notable limitations. Because patients were not randomly assigned to antidepressant regimens, we could not definitively test for differences in side effects between specific antidepressants or dosage levels. Patients often have a difficult time determining which medication (for a general medical problem versus an antidepressant) is causing a particular side effect, and they may have inaccurately reported that side effects were caused by their most recently prescribed antidepressant. However, patient perceptions about which medication is causing the side effects, regardless of accuracy, are likely to be a primary driver of antidepressant nonadherence. It is also possible that some patients experienced a side effect but perceived it as a symptom unrelated to their antidepressant medication and therefore did not report it during the interview. This potential problem with measuring side effects from patient self-report would have been equally balanced across both groups.

A further limitation was that at the time of the study, several antidepressants were still very expensive and were used by none or very few patients in the study. These included bupropion, the controlled release form of paroxetine, desvenlafaxine, and duloxetine. Other studies using a wider range of antidepressants may have different findings. Finally, most of the patients in this study were female, Caucasian, and relatively young, thus limiting the generalizability of study findings to other populations.

Conclusions

The results of this secondary analysis of data from a pragmatic comparative-effectiveness trial indicate that patients who received more intensive TBCC reported fewer side effects that they attributed to antidepressant medications compared with patients receiving less intensive PBCC. This result is likely due to increased clinical attention to identifying and managing side effects among participants in the TBCC group. The decrease in side effects may have contributed to improved quality of life. Other factors in past studies that have been shown to be associated with predicting side effect reporting were not statistically significant in this population and clinical context.