Discontinuation of Use and Switching of Antidepressants: Influence of Patient-Physician Communication

Context Although current depression treatment guidelines recommend continuing antidepressant therapy for at least 4 to 9 months, many patients discontinue treatment prematurely, within 3 months.

Objectives To investigate the relationship between patient-physician communication and the continuation of treatment with antidepressants and to explore the demographics, adverse effects, therapeutic response, and frequency of follow-up visits.

Design, Setting, and Patients A total of 401 telephone interviews of depressed patients being treated with selective serotonin reuptake inhibitor (SSRI) therapy between December 15, 1999, and May 31, 2000, were conducted and 137 prescribing physicians completed written surveys from Northern California Kaiser Permanente health maintenance organization outpatient clinics.

Main Outcome Measures Patient-physician communication about therapy duration and about adverse effects; therapy discontinuation or medication switching within 3 months after start of SSRI therapy.

Results Ninety-nine physicians (72%) reported that they usually ask patients to continue using antidepressants for at least 6 months, but 137 patients (34%) reported that their physicians asked them to continue using antidepressants for this duration and 228 (56%) reported receiving no instructions. Patients who said they were told to take their medication for less than 6 months were 3 times more likely to discontinue therapy (odds ratio [OR], 3.12; 95% confidence interval [CI], 1.21-8.07) compared with patients who said they were told to continue therapy longer. Patients who discussed adverse effects with their physicians were less likely to discontinue therapy than patients who did not discuss them (OR, 0.49; 95% CI, 0.25-0.95). Patients who reported discussing adverse effects with their physicians were more likely to switch medications (OR, 5.60; 95% CI, 2.31-13.60). Fewer than 3 follow-up visits for depression, adverse effects, and lack of therapeutic response to medication were also associated with patients' discontinuing therapy.

Conclusions Discrepancies exist between instructions that physicians report they communicate to patients and what patients remember being told. Explicit instructions about expected duration of therapy and discussions about medication adverse effects throughout treatment may reduce discontinuation of SSRI use. Our finding that patients with 3 or more follow-up visits were more likely to continue using the initially prescribed antidepressant medication suggests that frequent patient-physician contact may increase the probability that patients will continue therapy.

Guidelines issued by the Agency for Health Care Policy and Research and the American Psychiatric Association recommend continued treatment with antidepressants for at least 4 to 9 months after depressive symptoms resolve to prevent relapse.^1-3 Despite these guidelines, rates of treatment discontinuation within 3 months after the start of treatment can reach 68%, depending on the type of antidepressant prescribed and the population studied.^4-10

Drug-related adverse effects often cause patients to discontinue treatment, and efforts to improve the continuation of antidepressant use have yielded limited success.^11-15 A strong treatment alliance between physicians and patients that includes discussions about adverse effects throughout treatment may alleviate patients' concerns and help them continue treatment. Moreover, intolerance to one antidepressant is not necessarily accompanied by intolerance to another antidepressant, even within the same drug class¹⁶; therefore, patients who respond poorly to one drug or who experience adverse effects may benefit from having their initial antidepressant medication switched.¹ This switching requires good communication between patients and clinicians about treatment experiences.

Good communication has been shown to improve patient adherence to treatment instructions, especially in treatment for such chronic conditions as hypertension.^17-19 With better-tolerated medications available, nonadherence to medication regimens may result from poor patient commitment to treatment rather than from medication adverse effects.²⁰ In addition, for depression care, both depressive symptoms and the attached social stigma present challenges to good patient-physician communication. To date, research on patient-physician communication in depression care is limited. Available data suggest that explicit communication with patients regarding the expected duration of antidepressant therapy may reduce premature discontinuation of medication use.¹¹

We hypothesized that better communication between patients and physicians about antidepressant treatment, both before and during treatment, may promote favorable patterns of medication use. Fewer patients may prematurely discontinue treatment; more patients may have their medications switched because of adverse effects or poor treatment response.

We conducted surveys of patients who recently began taking a selective serotonin reuptake inhibitor (SSRI) for treatment of a new or recurrent episode of depression and surveys of their prescribing physicians. We investigated the relationship between patient-physician communication about antidepressant therapy and discontinuation of use and switching of antidepressants.

The study was conducted at the Northern California Kaiser Permanente Medical Care Program, which serves 3 million members at 16 hospitals and 31 outpatient clinics. Health plan databases were used to identify potential study participants who received an SSRI for new or recurrent depression between December 15, 1999, and May 31, 2000. Patients were eligible for inclusion if they had recently begun taking an SSRI antidepressant (at least 6 months since any previous antidepressant) and were diagnosed as having major depression or depressive disorder (International Classification of Diseases, Ninth Revision, code 296.2 or 311) within 30 days before or after the date the medication was dispensed; were from 18 to 75 years old; and had no concurrent comorbidity of bipolar disorder, bulimia, eating disorder, dementia, chemical dependency, or psychotic disorder and did not have the diagnosis of cancer, acute myocardial infarction, or stroke recorded in the health plan's databases.

Fluoxetine hydrochloride or paroxetine was prescribed for 94% of study patients. Prescribing physicians were asked permission to invite their patients to participate. Invitation letters were sent to patients on behalf of their physicians. The letter outlined the study objectives and informed patients that they would be telephoned and asked to be interviewed. A stamped, self-addressed postcard was enclosed for return within 7 days if the patient did not wish to participate. The study protocol was approved by the Northern California Kaiser Permanente Institutional Review Board.

The patient questionnaire was divided into 3 parts. Nearly all patients were interviewed within 75 to 105 days after the start of antidepressant therapy. In part 1, patients confirmed that they had begun taking the antidepressant prescribed to them. To identify "continuers," "discontinuers," and "switchers," patients were asked whether they were still taking the SSRI antidepressant initially prescribed and, if not, whether they were taking a different antidepressant. Patients were also asked what their physician told them before initiating treatment. Questions included, "How long were you told to stay on the antidepressant prescribed?" and "How long were you told that it would take for the [index SSRI] to begin working?" Patients were also asked whether their physician had discussed potential adverse effects with them before and during treatment.

In part 2, patients reported whether they had experienced any of 17 adverse effects associated with SSRIs. For each adverse effect experienced, patients were asked to rate on a 4-point scale how bothersome it had been to them (1, "not bothersome at all"; 2, "somewhat bothersome"; 3, "a lot bothersome"; 4, "extremely bothersome").

In part 3, patients rated on a 5-point scale changes in their depressive symptoms since starting treatment (1, "much better"; 2, "somewhat better"; 3, "about the same"; 4, "somewhat worse"; 5, "much worse"). Patients also completed the 7-item Beck Depression Inventory Fast Screen (BDI-FS)²¹ about the severity of their depression. The patient questionnaire was pilot-tested with 22 depressed patients. The interview lasted 15 to 25 minutes.

The prescribing physicians completed a written questionnaire about what they usually communicated to their patients about depression treatment. Two questions were used to assess how information communicated by physicians affected patients' treatment expectations: "How long do you usually tell your patients they will be taking the antidepressant?" and "How long do you usually tell your patients that it will take for the antidepressant to begin working?"

Physicians not responding to the initial mailed questionnaire within 2 weeks were e-mailed an electronic copy of the survey and given a follow-up telephone call to solicit a response. The physician survey was pilot-tested with 6 psychiatrists and 6 primary care physicians.

To examine the association of patient-physician communication with antidepressant drug use, 2 logistic regression models were specified. The first compared patients who reported discontinuing use of the initial antidepressant with patients who at the interview reported that they were still continuing to take the initial antidepressant. The second compared patients who reported that they had switched antidepressants with patients who reported that they were still taking the antidepressant initially prescribed. We characterized variables selected for the model as communication variables, adverse effect experiences, and clinical factors.

Separate variables were created for patients who reported discussing adverse effects with their physician before treatment and during treatment. Other communication variables that could also influence medication treatment patterns were tested in the models: reporting being told to take the antidepressant for at least 6 months and reporting being told that the medication would require 4 or more weeks to take effect. Six months was selected because it represented the minimum duration of treatment required to complete the short-term recovery and continuation phase.¹

To examine the impact of adverse effects on patients continuing their initial therapy, indicators were created for patients who reported more than 1 adverse effect (a lot or extremely bothersome). We also examined the relation between number of adverse effects experienced (≥3) and frequency of discontinuing use of or switching the medication.

Clinical variables included the BDI-FS assessment of depression severity (minimal, mild, moderate, or severe) at the patient interview, degree to which global depression symptoms improved (dichotomized as at least "somewhat better" vs "about the same" or "worse") since start of medication use, and participation in a behavioral or counseling intervention. The number of depression-related follow-up visits with a physician (dichotomized at ≥3) came from the health plan's databases.

To control for potential confounding factors, the models included patient's age, sex, race, educational level, previous use of antidepressants, and specialty of treating physician (psychiatrists or other specialty). A stepwise procedure was used to construct the final models. Variables significantly related to outcomes at the P<.05 level were included.

Physicians gave permission to contact 765 of the 1050 potentially eligible study participants. Of the 765, 33% of 260 patients were excluded as follows: 40 of those (15%) never started taking the medication, 36 (14%) took an antidepressant within 6 months before the index date, 17 (7%) did not speak English, 97 (37%) could not be reached by telephone within 4 months of starting treatment (with at least 3 attempts), and 31 (12%) did not respond to the survey for reasons such as inability to concentrate. Eligibility could not be determined for 39 (15%) of the patients contacted. Of the remaining 505 eligible patients, 401 (79%) completed the telephone survey. On average, participants were somewhat younger (mean age 45.8 vs 49.8 years) and more likely to be women (71% vs 62%) than nonparticipants.

Of the 167 physicians who treated study participants, 137 (82%) returned the questionnaire. Among them, 15 (11%) were family practice physicians, 75 (55%) were internal medicine physicians, 45 (33%) were psychiatrists, and 2 (1%) had no specialty named. No significant differences were detected between psychiatrists and nonpsychiatrists with respect to information communicated to patients about duration of therapy and time before the medication took effect. Therefore, we grouped both psychiatrists and nonpsychiatrists together in all analyses.

The mean age of study participants was 45.8 years. Two hundred sixty-seven (67%) of the participants were non-Hispanic white and 284 (71%) were women. No statistically significant differences in demographic characteristics were found between patients who continued, discontinued, or switched medication use (Table 1). One hundred fifty-one patients (38%) reported that they had previously taken antidepressants but not within 6 months before study entry. Of the 401 patients, 133 (33%) reported that they were no longer taking the original antidepressant: 54 (13%) had switched to a different antidepressant, and 79 (20%) had discontinued antidepressant treatment. Of those not taking the original antidepressant, 93 (70%) stopped using the medication without consulting their physicians (14 [15%] of whom were subsequently prescribed a different antidepressant), most often due to adverse effects 33 (36%), followed by 22 (24%) feeling better, 19 (20%) for lack of efficacy, and 19 (20%) for other reasons.

Seventy-two percent of 137 physicians reported that they usually told patients to take the medication for at least 6 months; only 29 (21%) reported that they initially do not specify treatment duration (ie, that they do not tell patients precisely how long to take the medication, but they would rather wait to see how they feel). In contrast, 137 patients (34%) reported that they were told by their prescribing physician to take the antidepressants for at least 6 months, and 228 (56%) reported receiving no instructions about expected therapy duration. Nevertheless, an association was found between what patients reported that they were told and what their treating physicians indicated they usually tell their patients. For example, patients whose physicians usually tell their patients to take antidepressants for at least 6 to 11 months reported their mean treatment duration was 6.8 months, but patients whose physicians usually tell their patients to continue antidepressant therapy for 3 to 5 months reported their mean treatment duration was 3.0 months. We observed a similar modest correlation between patients' expectations about how long it would take for the medication to "begin working" and their physicians' reports about when to expect it to take effect (r = 0.11; P = .07).

Table 2 shows the number and percentage of patients who described aspects of patient-physician communication, medication adverse effects, and clinical variables grouped by antidepressant use status. Thirty-six patients (9%) reported that they were told to take medication for less than 6 months, and 192 (48%) reported that they were told by a physician or pharmacist about medication adverse effects. Two hundred eight participants (52%) reported experiencing 3 or more adverse effects, and 219 (55%) reported 1 or more adverse effect that was either "a lot" or "extremely" bothersome.

Patients who reported that they were told to take their medication for less than 6 months had 3.12 times higher odds of discontinuing antidepressant therapy than patients who reported they were told to take the medication for 6 or more months (Table 3). Communication about adverse effects significantly decreased the odds of discontinuing antidepressant therapy, as did visiting a physician 3 or more times after starting a medication regimen and rating depression symptoms as improved. Presence of adverse effects (a lot or extremely bothersome) significantly increased the odds of discontinuing medication use. Patients who were separated, divorced, or widowed were more likely to discontinue antidepressant treatment than patients who were currently married.

Switching medications was significantly associated with presence of adverse effects and discussing adverse effects at both the start of therapy and follow-up (Table 4). Patients who continued to have severe symptoms were 6 times more likely to switch medications.

Successful antidepressant treatment requires patients to continue therapy long enough to provide some benefit and to reduce the likelihood of relapse.^1-3 Our findings show that premature discontinuation depends on multiple factors, including patient-physician communication, family support, clinical response to therapy, and adverse effects. Although explicit communication increased the likelihood of continuing treatment, there appeared to be a communication gap between patients and treating physicians. Whereas 72% of physicians said they usually tell patients to continue using antidepressants for at least 6 months, only 34% of patients reported being told that. Furthermore, more than half of patients reported not being told about expected therapy duration. Patients' failure to remember specific information communicated to them by their physicians may partially explain the gap. If so, we suggest that it may be helpful for physicians to reinforce expected duration of treatment to their patients during follow-up visits. Discussing adverse effects with patients during treatment was associated with less premature discontinuation. This outcome may have resulted from patients receiving information and support from their physicians when adverse effects occurred. Informing patients in advance about adverse effects has been reported to prevent patients from fearing adverse effects and to increase patients' physiologic tolerance to antidepressants.²² We suggest that physicians discuss adverse effects with patients throughout treatment to encourage their continuation of antidepressant therapy.

We also found that switching medication was related to discussions about adverse effects before and during treatment. Switching medications as a result of these discussions can be considered a favorable outcome if use of a different medication is associated with greater treatment continuation and improved symptoms.

Both the number of adverse effects and their perceived bothersomeness indicate the importance of patient-physician discussions about medication adverse effects. The findings that 55% of patients experienced 1 or more adverse effects (a lot or extremely bothersome) and 29% of patients had 1 or more extremely bothersome adverse effects should lead physicians to expect that at least half their patients receiving SSRIs will experience potentially troublesome adverse effects. Physicians should therefore discuss adverse effects with patients during treatment. Physicians who seek to mitigate patients' concerns by only selectively warning them of adverse effects may need to reconsider this practice.

The findings indicate that frequent office visits after start of pharmacotherapy are associated with continuation of initial antidepressant treatment. This finding supports the new Health Plan Employer Data and Information Set measure for antidepressant management developed by the National Committee for Quality Assurance as one of the accreditation criteria for managed care organizations. This measure requires antidepressant management to include at least 3 follow-up contacts during the first 12 weeks of treatment.²³

Our study results also showed a correlation between clinical response to medication and patterns of SSRI medication use. This correlation is consistent with an early study showing that depression itself is a risk factor for nonadherence to medical treatment.²⁴ In our study, patients who remained severely depressed, according to the BDI-FS (applied during the telephone survey), were 6 times more likely to switch to a different medication. We believe that this outcome is positive and indicates that physicians were aware of nonresponse to the treatment regimen selected initially. Close follow-up of these patients is warranted to ensure that they are offered other therapy if appropriate.

When initially prescribing antidepressants, physicians should also consider the greater likelihood of treatment discontinuation in some populations. We found that separated, divorced, and widowed patients were more than twice as likely to discontinue antidepressant use as married patients. Social support may encourage them to continue treatment.

The study captures information communicated about antidepressant treatment from the perspective of both the physician and the patient. The study provides information from participants about their perception of treatment experiences during routine care (ie, not as part of a clinical study protocol). Rates of discontinuation (20%) observed in our study were comparable with rates typically seen throughout the Kaiser Permanente Northern California Region (J. Chan, PharmD, PhD, unpublished data, 1999) but are higher than rates typically seen in randomized controlled clinical trials of SSRIs.¹⁴ Higher rates of treatment discontinuation and switching observed in the naturalistic setting are probably due to both differences in the populations studied and intensity of follow-up care provided to patients.

A limitation of this study is that the data from the physicians did not refer to specific patients but to the usual information physicians routinely communicated to patients. Asking physicians to review the records of each patient and to report what they told that patient months ago would have been impractical, and, furthermore, such information may not routinely be recorded in medical records. This lack of patient-specific information probably makes it more difficult to establish an association between patient-physician responses. However, our data still demonstrated a modest correlation between physicians' and their patients' expectations of duration of treatment and interval before treatment response.

Because patient participation in the study was voluntary and our study patients may be more cooperative than the usual patient population, our findings may overestimate the proportion of patients who communicate well with their physicians. In addition, participants may be more likely to provide socially desirable responses, which could conceal a portion of actual communication observed in this study (eg, if a patient discontinued using medication without consulting his/her physician). To minimize this potential bias, we obtained information on frequency of physician office visits from the health plan computerized registration database rather than patients' self-report. We also assured participants that their responses would be anonymous and would not be passed on to their treating physician.

Some questions asked patients to report information communicated to them for up to as long as 3½ months before the telephone interview. Memory failure, variation in response to treatment, or both may have contributed to the discrepancy observed between physician and patient responses. In addition, the status of medication use at the time of the survey may have influenced what patients reported being told about treatment. For example, patients who switched or discontinued using the originally prescribed SSRI may have been more likely to report certain aspects of information communicated to them. However, querying patients on a more frequent basis to minimize this potential bias may have yielded an elevated adverse effect rate and may have produced the unintended effect of enticing participants to ask particular or more frequent questions than usual of their treating physician. Despite these limitations, we believe that our findings are valid for the population studied and have major implications for depression care.

Discrepancies exist between instructions that physicians report they communicate to patients and what patients remember being told. Explicit instructions about expected duration of therapy and discussions about medication adverse effects throughout treatment may reduce discontinuation of SSRI use. Our finding that patients with 3 or more follow-up visits were more likely to continue taking the initially prescribed antidepressant medication suggests that frequent patient-physician contact may increase the probability that patients will continue therapy.