Gender Differences in Psychotropic Medication Use

Abstract

Purpose/Background:

Previous studies among psychiatric patients have shown differences between men and women in psychotropic medication prescription patterns. The factors underlying these differences are still poorly understood. This exploratory study aimed to investigate a range of clinical and demographic factors in their contributions to explaining these gender differences.

Methods/Procedures:

In 2476 outpatients with a mental disorder (mean age 46.1 year, 58.2% women) enrolled in pharmacotherapy monitoring program, regression analyses were used to investigate the associations of gender with use of any psychotropic (yes/no), use of each separate psychotropic type (yes/no), number (0 to 4) and dosage of each used psychotropic. Next, clinical and demographic effect modifiers and explanatory/confounding factors of these associations were identified.

Findings/Results:

Frequencies of general psychotropic use, antidepressant use, anxiolytic use, as well as polypharmacy were significantly higher in women than in men across all ages (premenopausal, perimenopausal, and postmenopausal), with equal dosing of psychotropics, except for higher doses of mood stabilizers in men. Longer illness duration and having a partner were effect modifiers of the gender effect, while higher frequencies of an affective/anxiety diagnosis, higher somatic complaints and symptomatic distress, more previous mental health service (MHS) visits and no current occupation were explanatory factors.

Implications/Conclusions:

Psychotropic use and psychotropic polypharmacy were more common in women than in men. Women on psychotropic medications were more often diagnosed with affective/anxiety diagnoses, experienced more symptomatic distress and somatic complaints, and visited more healthcare services than men, reaffirming findings from previous studies. Future longitudinal studies should investigate in further detail whether these outcomes are cause or consequence of gender differences in psychotropic medication use.

In patients with psychiatric disorders, some types of psychotropic medication seem to be prescribed more often to women than to men, across all diagnoses. For common mental disorders, including affective and anxiety disorders, this has been shown for both treatment by general practitioners and in psychiatric care. Summarizing the findings, several conclusions can be drawn. First, for patients with a diagnosis or symptoms of depression, anxiety or an addiction, antidepressants were found to be prescribed more often to women than men,^1–5 in contrast to antipsychotics, which seem to be used in equal frequencies and dosages by men and women.^2–4 Across all diagnoses, benzodiazepines have also been found to be prescribed more often to women than men.^1–4,6 In bipolar disorders, women are more often treated with antidepressants, benzodiazepines and nonlithium mood stabilizers, whereas men are more often treated with lithium.⁵ However, results are contradictory, with some studies that investigated the aforementioned disorders and psychotic disorders not observing any gender differences in the number and/or type of prescribed psychotropic medications.^7,8 Importantly, multiple studies have shown that more frequent psychotropic medication use in women may be related to poorer physical health and more side effects compared with men,^9,10 indicating that more psychotropic use in women might have specific negative health consequences.

The gender differences in use of psychotropic medication may be related to gender differences in sociodemographic factors, clinical characteristics, such as symptom severity, and response to and side effects of these psychotropics, as indicated in the following summary of studies. One explanation for the observation that antidepressants and benzodiazepines are prescribed more frequently to women than to men could be that depression and anxiety disorders are more commonly reported and more severe in women than in men.^11,12 However, previous studies have shown that gender differences in psychotropic use remain when adjusting for prevalence and severity of psychiatric symptoms.^12–16 Another explanation may be that women report more gender-related complaints (including premenstrual dysphoric disorder and chronic neuropathic pain), for which treatment guidelines advise to prescribe antidepressants.^17,18 Third, other sociodemographic factors (age, income level, employment status, education level, marital status) have been found to be differently related to frequency of psychotropic drug use in men versus women.^13–15,19 However, the gender-related associations with sociodemographic factors are complex, and findings have been contradictory.^10,14 A fourth hypothesized explanation for gender differences in psychotropic medication use is that women have a higher tendency to report and seek help for psychiatric problems, leading to higher treatment rates, and consequently, prescription rates among women.^3,11,20 Rather than seeking professional help, men with psychiatric complaints may have a higher tendency to use alcohol or drugs as a form of self-medication.²¹ The latter could also (partly) explain the higher rates of substance use disorder diagnoses among men, which may sometimes obscure underlying depressive or anxiety symptoms and may cause lower prescription rates of psychotropic medication in men.^2,22

Across a wide range of diagnoses, including depression and anxiety, studies have shown that female hormonal levels and their interaction with the stress response system may alter the response of women to psychotropic medication.^23–27 More specifically, we expect the largest differences in treatment patterns between men and women in the premenopausal phase, as female hormones have their largest effect during that phase.^5,27,28 However, some studies were unable to find gender-specific and age-related menopausal effects on prescription frequency,^29–33 so this topic is still under study.

With respect to psychotropic drug dosing, although women may actually need lower dosages of psychotropic medication, given slower drug absorption, metabolism and excretion than men, women often receive similar psychotropic dosages as men.^27,34 Treatment guidelines do not yet provide gender-specific dosage recommendations, and given that most clinical research has been performed in men, women may actually be exposed to relative overdosing when compared with men.²³ Specifically in antidepressant use, it has been suggested that women may need lower dosages of antidepressants than men, and that equal dosages may lead to more side effects in women.²⁶ While there appear to be no gender differences in lithium response, despite differences in frequency of use,³⁵ women tend to need lower dosages of antipsychotics than men.^36,37 This could be due to women reaching higher dopamine receptor occupancy at similar serum levels compared with men, since estrogens increase dopamine sensitivity.²⁷ Concerning benzodiazepines, gender differences in drug metabolism have been observed, but consequences for clinical practice are still unclear.^38,39

To summarize, the frequency and dosing of prescribed psychotropic medication appear to differ between men and women, but the explanatory mechanisms underlying these gender differences are still poorly understood, presumably because they are influenced by a complex interplay of sociodemographic, psychological, and biological factors. So far, there have been few dedicated studies to gain more insight into this. Moreover, a substantial part of the available studies has focused on general population-based samples and survey data of individuals who are not by definition being treated in specialized mental health care.^{1–3,6,9,12–14,16} In addition, most previous studies did not investigate potential correlates of observed gender differences, investigated correlates of psychotropic use separately in men and women but did not formally test if predictors modified the gender-psychotropic use association,^8,12 or only evaluated whether they were explanatory factors/confounders of this association but without testing for effect modification.^{2,3,10,14–16} To the best of our knowledge, no studies to date have investigated gender differences in psychopharmacological treatment in a naturalistic, “real-world” cohort of psychiatric patients with broadly defined affective and anxiety disorders, who are treated in secondary mental health care.

Therefore, the primary aim of this study was to investigate gender differences in psychotropic medication use, i.e., frequency, number of used psychotropic classes and dosages of all psychotropics in general, and specifically of antidepressants, antipsychotics, anxiolytics/sedatives, and mood stabilizers, in a large naturalistic sample of patients at secondary outpatient psychiatric clinics. Next, for significant differences in frequency of use of psychotropics, we investigated for several demographic and clinical characteristics, whether they were an effect modifier (ie, differently associated with medication use in men and women) or an explanatory/confounding factor (ie, explaining part of the association between gender and medication use) of the observed differences. Third, given that the hormonal differences between males and females decrease after menopause, we investigated whether gender differences in psychotropic use are smaller in the postmenopausal life phase compared with the premenopausal phase.

METHODS

Procedures

This is a retrospective study with a data set realized from data of standard clinical visits and based on the information already collected in an outpatient mental health service in the Netherlands.⁴⁰ The study was approved by an independent medical ethics committee (protocol number 928). Written informed consent was obtained from all participants. All study procedures were conducted according to the Declaration of Helsinki. The data were obtained through interviews about somatic health and medication use, physical examinations performed by a qualified nurse, through questionnaires partly filled in at home and partly at the location by the nurse, blood tests used in standard care and medication verification. Patients could fill in questionnaires on a digital device using a special, secure computer program that has been developed for use in health care, but at the same time enables researchers to safely use the collected data for research purposes. Blood tests were collected for standard routine outcome monitoring of somatic health.

Participants

At the time of this study, the database contained baseline data of 7476 adult patients (aged ≥18 years), who were referred for secondary psychiatric care. Patient characteristics were reported for the whole sample and for men and women separately. Patients were included in the current study between 2016 and 2021. There was one inclusion criterion: Participants should have a current depressive disorder, bipolar disorder, anxiety disorder, personality disorder, developmental disorder, psychotic disorder, substance-related disorder, and/or somatoform disorder. These diagnoses were established according to The Diagnostic and Statistical Manual of Mental Disorders, 5th edition, criteria by either a psychiatrist or a clinical psychologist.

Exclusion criteria were the following: First, patients with intellectual disability and/or neurocognitive disorders (delirium, dementia, amnestic and other cognitive disorders based on either a Screener for intelligence and learning disabilities)⁴¹ score <20, Wechsler Adult Intelligence Scale score <80, or clinician’s opinion) could not participate. Second, patients who were not able to read/write in Dutch were also excluded. There were no restrictions on medication use.

Demographics

The following information on demographics was collected: age, gender, living situation (with partner, alone), educational level (low (primary school and practical education) versus middle (vocational education) and high (secondary high school and university) and daily activities/occupation (working or studying vs. not). Classification of educational level was based on the Dutch National Institute of Public Health and the Environment criteria. Age was divided into 3 categories for stratified analyses (<45, 45 to 55, and >55 years), corresponding to the menstrual, perimenopausal, and menopausal life phases, respectively.²⁸

Psychotropic Medication

According to the World Health Organization ATC/DDD Index 2020, psychotropics used were classified as: antidepressants, antipsychotics, mood stabilizers, or anxiolytics (including sedatives). For each subject, the total amount of Defined Daily Doses (DDDs) used per day was computed per category and for all psychotropics together. These are referred to as prescribed daily doses (PDD) across the manuscript.

Somatic Problems

To monitor subjective somatic complaints (including potential medication side effects), a subset of 18 questions of the Somatic mini-Screen (SmS) was used, which is a self-report questionnaire, based on the Liverpool University Neuroleptic Side Effect Rating Scale.⁴² Each item is scored on a 5-point Likert scale, ranging from 0 (“none”) to 4 (“severe”). A total score was calculated (range: 0 to 72), where higher scores reflect more severe subjectively reported somatic complaints. Factor analysis on the current data set confirmed that items constitute one factor (data not shown). Furthermore, the presence of a range of somatic disorders was self-reported by patients in a medical history assessment. The number of reported somatic disorders was added up for each patient and used in the analyses.

Psychopathology Severity

The Dutch version of the Outcome Questionnaire (OQ)-45 was used to assess general psychopathology severity.⁴³ The OQ-45 is a self-report questionnaire of 45 items, with a 5-point Likert response scale, ranging from 0 (“never”) to 4 (“always”). Scores were calculated for 3 subscales: “Symptomatic distress” (25 items), “Interpersonal relations” (11 items), “Social roles” (11 items). The number of (comorbid) psychiatric diagnoses was counted as a measure of psychiatric symptom load (0, 1, ≥2). Presence of a main or comorbid diagnosis of affective or substance use disorder was determined based on DSM-IV or DSM-5 main and comorbid diagnoses (having a diagnosis=1, no diagnosis=0). The duration of illness, duration of the current episode (<1 year=0, >1 year=1) and number of previously visited mental health care organizations (0, 1, ≥2) were recorded.

Lifestyle

Alcohol, smoking, cannabis, and drug use were established using patient reports during the interview with the research nurse. These questions assessed the number of alcoholic beverages per week (never, at least monthly, at least weekly), cigarettes per week (never, 1 to 5/week, >5/week), number of marijuana joints (never, at least monthly, at least weekly) and use of hard drugs (yes/no). Several lifestyle variables that are known to be related to body mass index (BMI) were measured with a set of screening questions that were specifically developed for the monitoring protocol, based on the following items with one point for every criterion met: (1) >4 days/week moving at moderate intensity for >30 minutes, (2) eating vegetables >3 days/week, (3) eating >1 pieces of fruit/day, (4) eating fast food <2 days/week, (5) drinking <4 sugar-containing beverages/day, (6) spending >15 minutes outdoors per day. A lifestyle score was created based on cutoff values established on expert opinion. The total score had a range of 0 to 6.

Statistical Analyses

All analyses were conducted in SPSS 28 (IBM Corp. Released 2021. IBM SPSS Statistics for Windows, Version 28.0. IBM Corp). The following baseline variables were investigated as effect modifier or explanatory/confounding factor of gender differences in psychotropic medication use: living situation (together vs. alone), education level (low vs. middle and high), work/study (yes/no), duration of illness (years), duration of current episode (>1 year yes/no), number of reported somatic disorders in medical history (proxy for tendency to report health complaints), number of psychiatric comorbidities, number of previous treatments at a mental health care organization (proxy for help seeking), affective or anxiety diagnosis (yes/no, main or comorbid unipolar or bipolar depression or anxiety disorder), substance use disorder (yes/no, main or comorbid), alcohol use (frequency), smoking (frequency), using soft drugs (frequency), using hard drugs (yes/no), lifestyle score, Symptomatic Distress subscale score (OQ-45), Interpersonal Functioning subscale score (OQ-45), Social Role subscale score (OQ-45), subjective somatic complaints based on 18 item score (SmS).

All analyses were conducted with subjects who had complete data on the used variables, meaning that analyses were carried out in varying subsamples. The use of any psychotropic (yes/no), antidepressants (yes/no), antipsychotics (yes/no), anxiolytics (yes/no), mood stabilizers (yes/no), and the number of psychotropic classes used (degree of polypharmacy; 0-4) were compared between genders using a χ² test. The PDD was compared using a Mann-Whitney U test because of non-normal distribution of the data. Given the exploratory nature of the study, results were not corrected for multiple comparisons.

In case age was found to moderate the gender-medication use association, gender differences were investigated per age category that corresponded roughly to the menstrual, perimenopausal and postmenopausal life phases (aged <45, 45 to 55, >55 years, respectively), by adding the age categories (as dummy variables) and their 2-way interactions with gender as independent variables to a regression model with the target medication use as outcome. In case of significant interaction effects, analyses of the association between gender and the outcome were stratified by age group to gain more insight into age-specific gender effects.

Second, effect modification of the association between each given outcome and gender by each of the other predictors, ie, whether predictors showed a different association with medication use in men compared with women, was investigated in a similar way, running a series of regression analyses, each time including one predictor and its interaction with gender as independent variables in the regression model. Effect modification of the association between gender and medication use by a predictor was deemed present if its interaction with gender had an uncorrected P-value <0.05. For significant effect modifiers, stratified analyses were conducted to compare the association between gender and medication use across the levels (eg, “no” vs. “yes,” or quartiles) of the effect modifier.

For all predictors that were not effect modifiers, it was evaluated if they explained a relevant portion of the association between gender and each outcome as an explanatory variable/confounder, by running separate regression models with the target outcome as dependent and gender as an independent variable, each time adjusting for the explanatory effect of one predictor variable. Here, adjustment was done using inverse probability weighting (IPW) rather than multivariable adjustment to prevent issues with noncollapsibility when comparing adjusted and unadjusted effects in logistic regression.⁴⁴ To obtain an estimate of the predictor’s explanatory effect, the crude effect (B) can be compared with an estimate (B*) from a univariable logistic regression that is weighted for the predictor in question.^45,46 To this end, weights were created for each predictor using a 3-step approach.⁴⁷ First, a logistic regression analysis was run with gender (0=male, 1=female) as the dependent variable and the predictor (C) as the independent variable. Next, propensity scores were calculated by using the resulting logistic regression model to predict conditional probabilities of female gender (p(gender=1|C)). Finally, weights were calculated based on these propensity scores: for gender=1 (female sex), weights were calculated as 1/p(gender=1|C) and for gender=0 (male sex) as 1/(1-p(gender=1|C)). For each investigated predictor, a weighted univariable regression with the psychotropic use variable of interest as the dependent variable and gender as the independent variable was run, using the calculated weights for the predictor in question. The resulting predictor-weighted gender effect estimate (B*) was compared with the crude estimate (B*-B) and the percentage of the change relative to the crude effect was calculated. Predictors that resulted in a change of >10% of the gender effect on the log-odds scale were identified as potentially important explanatory factors. The number of missing values varied across predictors, so the weighted estimate (B*) was obtained in a different subsample for each predictor. Importantly, for each predictor, the crude effect that was used for comparison (B) was estimated in the same subsample with nonmissing predictor values to enable a direct comparison between the two estimates (B and B*) in the same sample.

The abovementioned effect modification and IPW analyses were only conducted if a significant initial bivariate association between gender and a given outcome was observed. In addition, they were only conducted with the dichotomous (eg, any psychotropic use: yes/no) and ordinal (number of psychotropic types) medication-use outcomes in the full sample and not with PDD, as the latter were only determined for small subsamples of medication users.

RESULTS

Sample Characteristics

Demographic and clinical characteristics are summarized in Table 1. Data were available from 2476 patients (mean age 44.6 years), of whom 58.2% were female. As a primary diagnosis, 39% had a depressive disorder, 35% an anxiety disorder, 13% a bipolar disorder, 7% a substance use disorder and 6% a developmental disorder or psychotic disorder. For all 3 subscales of the OQ-45, the average scores were above the clinical threshold (Symptomatic Distress >31, Interpersonal functioning >12, Social Role >12), indicating that clinical symptoms were present in all 3 symptom domains. Concerning psychotropic medication use, ~55% of patients were without psychotropic medication at intake, and of the remaining group 35% of the patients used antidepressants, 20% used antipsychotics, 20% anxiolytics/sedatives, and 10% mood stabilizers. Percentages of missing values per variable are indicated in Table S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A977.

TABLE 1.

Patient Characteristics of the Study Sample for Males and Females Separately and Together

			Male		Female		Total
	Total N	Category	Mean/%	SD/N	Mean/%	SD/N	Mean/%	SD/N
Gender	2476		41.8	1034	58.2	1442
Age (y)	2476		45.6	15.6	43.9	16.7	44.6	16.3
Age group (y)	2476	<40	46.1	477	53.8	776	50.6	1253
		40 - 50	25.0	259	20.9	302	22.7	561
		>50	28.8	298	25.2	364	26.7	662
Living situation	1770	Alone	30.3	223	30.4	314	30.4	537
		Together	52.4	386	52.6	543	52.4	929
		With parents	10.2	75	8.0	83	9	158
		Sheltered	2.7	20	1.9	20	2.2	40
		Other	4.5	33	7.1	73	6	106
Occupation	1763	Household	77.2	571	90	932	84.7	1503
		Work	50	370	40.4	418	44.4	778
		Volunteer work	11.4	84	14.6	151	13.2	235
		Study	11.4	84	12.6	130	12.1	214
		Nothing/other	8.1	60	3.4	35	5.4	95
Educational level	1687	Low level	29	214	27.1	278	27.8	492
		Middle level	39.5	292	39.8	410	37.2	702
		High level	27.4	203	28.2	290	27.8	493
Diagnosis	2099	Depressive disorder	36.7	335	39.8	531	38.6	866
		Bipolar disorder	15.7	143	11.3	150	13.1	293
		Anxiety disorder	28.8	263	39.3	524	35.1	787
		Substance use disorder	10.2	93	4.5	60	6.8	153
Comorbidities	2245	0 comorbid disorders	42.4	387	44.2	589	43.5	976
		1 comorbid disorder	29.8	272	30.0	400	29.9	672
		>1 comorbid disorder	27.7	253	25.8	344	26.6	597
Smoking	1977	Currently smoking	40.9	337	36.9	426	38.6	763
		No. units/day	11.0	8.5	9.6	8.3	8.6	9.9
Alcohol	1975	Never	24.8	204	34.7	400	30.6	604
		1/mo	20.3	167	29.2	337	25.5	504
		2-4/mo	22.7	187	19.1	220	20.6	407
		2-3/wk	17.9	147	10.7	123	13.7	270
		>4/wk	14.2	117	6.3	73	9.6	190
Drugs	1977	Cannabis	15.3	126	8.8	102	11.5	228
		Hard drugs	4.5%	37	3.5	40	3.9	77
Symptoms	1259	OQ-45 symptomatic distress	47.1	18.1	51.1	16.4	49.3	17.3
		OQ-45 Interpersonal functioning	15.8	7.1	15.8	6.7	15.8	6.8
		OQ-45 social role	13.0	5,6	12.2	5.2	12.6	5.4
	1922	SmS side effects	21.4	13.8	23.2	12.9	22.4	13.3
Psychotropics	2444	Antidepressants	29.2	306	39.3	559	35.4	865
		Antipsychotics	19.7	201	20.8	296	20.3	497
		Anxiolytics	20.9	214	26.8	381	24.3	595
		Mood stabilizers	10.4	106	10.5	150	11.2	274

Gender and Psychotropic Use

Percentages of patients using psychotropics and percentages using 0, 1, 2, 3, or 4 classes of psychotropics are shown in Figure 1. The average number of times DDD used (PDD) is shown in Figure S1, Supplemental Digital Content 1, http://links.lww.com/JCP/A977. Univariate analyses comparing users with nonusers are shown in Table 2. There was a significant difference between men and women for use of psychotropics (yes/no), antidepressants (yes/no), anxiolytics (yes/no), and the number of classes of psychotropics, with women having a higher frequency of medication intake for each outcome. There were no differences in average prescribed daily dosages (PDD) between men and women, except for the median PDD of mood stabilizers, which was significantly higher in men (0.9) than in women (0.7) in the relatively small group of mood-stabilizer users (n=273).

FIGURE 1.

Percentages of individuals using psychotropic medication in general and certain types of psychotropics per age group. The lower right corner shows the percentage of individuals using a certain number of psychotropic types, ranging from none (0) to four. Males=gray, females=white.

TABLE 2.

Psychotropic Drug Use for the Whole Group and Men and Women Separately

		Sample	Males	Females
		(n=2.444)	(n=1.022)	(n=1.422)	Gender Comparison
	Medication Group	N (%/median)	N (%/median)	N (%/median)	χ2	P
Frequency	Antidepressants	865 (35.4)	306 (29.9)	559 (39.3)	22.8	<0.001
	Antipsychotics	497 (20.3)	201 (19.7)	296 (20.8)	0.5	0.490
	Anxiolytics	595 (24.3)	214 (20.9)	381 (26.8)	11.1	<0.001
	Mood stabilizers	274 (11.2)	109 (10.7)	165 (11.6)	0.5	0.470
	At least one type	1343 (55.0)	515 (50.4)	828 (58.2)	14.8	<0.001
Number of classes	0	1101 (45.0)	507 (49.6)	594 (41.8)	18.9	<0.001
	1	694 (28.4)	282 (27.6)	412 (29.0)
	2	438 (17.9)	162 (15.9)	276 (19.4)
	3	183 (7.5)	60 (5.9)	123 (8.6)
	4	28 (1.1)	11 (1.1)	17 (1.2)
PDD	Antidepressants	830 (1.0)	294 (1)	536 (1.0)		0.36
	Antipsychotics	404 (0.4)	161 (0.5)	243 (0.4)		0.28
	Anxiolytics	496 (0.5)	189 (0.5)	307 (0.5)		0.79
	Mood stabilizers	273 (0.8)	109 (0.9)	164 (0.7)		<0.001

N, Percentages, and χ² test results are indicated for using vs. not and number of types used, N, median and Mann-Whitney test results for the prescribed daily dose (PDD).

Gender-specific Effects of Age

As can be observed in Figure 1, the frequency of psychotropic drug use increased with age in both men and women. However, there were no significant age group by gender interactions for any of the outcome measures, indicating that the gender difference in medication use did not differ between age groups.

Effect Modifiers

Investigation of effect modification by predictor variables (Table 3; Table S2, Supplemental Digital Content 1, http://links.lww.com/JCP/A977) showed a significant interaction between gender and duration of illness in regression models that used any psychotropic use, antidepressant use, or the number of psychotropic classes as outcomes. Post-hoc stratified analyses (Table S2, Supplemental Digital Content 1, http://links.lww.com/JCP/A977) in quartile groups of illness duration showed that only in the upper 2 quartile groups with a longer duration of illness (range 14 to 67 years), women used significantly more psychotropics in general, more antidepressants and more classes of psychotropics than men. Having a partner (yes/no) also showed a significant interaction with gender in a regression model that used any psychotropic use as the outcome. Women with a partner used psychotropics more frequently than men with a partner, while there was no difference between men and women without a partner.

TABLE 3.

Interaction-term P-values to Identify Effect Modifiers of Associations Between Gender and Medication Use

	Psychotropics	Antidepressants	Anxiolytics	Polypharmacy
Living situation (together vs. alone)	0.028	0.21	0.73	0.061
Education level (high vs. low)	0.53	0.12	0.14	0.64
Occupation (yes/no)	0.12	0.350	0.73	0.16
Duration of illness (y)	0.002	0.031	0.056	0.001
Duration of current episode (>1 y)	0.31	0.56	0.74	0.28
Somatic complaints in medical history (1-2)	0.19	0.95	0.83	0.38
Somatic complaints in medical history (>2)	0.13	0.38	0.55	0.11
Psychiatric comorbidities (1)	0.92	0.71	0.95	0.76
Psychiatric comorbidities (>1)	0.88	0.82	0.76	0.82
Previously visited MHS (1-3)	0.39	0.056	0.33	0.16
Previously visited MHS (>3)	0.056	0.14	0.82	0.23
Affective or anxiety disorder (yes/no)	0.068	0.230	0.23	0.07
Substance use disorder (yes/no)	0.29	0.58	0.18	0.21
Using alcohol (monthly)	0.82	0.95	0.39	0.91
Using alcohol (weekly)	0.40	0.55	0.56	0.25
Smoking (1-5 cigarettes/day)	0.50	0.18	0.85	0.31
Smoking (>5 cigarettes/day	0.17	0.48	0.37	0.25
Using cannabis (monthly)	0.21	0.090	0.61	0.23
Using cannabis (weekly)	0.14	0.82	0.30	0.24
Using hard drugs (yes/no)	0.65	0.39	0.88	0.62
Lifestyle score (medium)	0.46	0.78	0.081	0.39
Lifestyle score (high)	0.34	0.13	0.055	0.38
OQ-45 Symptomatic distress	0.96	0.050	0.36	0.96
OQ-45 Interpersonal functioning	0.95	0.33	0.16	0.95
OQ-45 social role	0.42	0.13	0.45	0.42
SmS side effects	0.44	0.34	0.83	0.28

Significant values (p < 0.05) are indicated in bold.

Each predictor’s interaction with gender was estimated for each predictor separately.

Explanatory Factors

Investigation of explanatory factors (Table 4; Table S3, Supplemental Digital Content 1, http://links.lww.com/JCP/A977) showed that several predictors explained >10% of the association between gender and psychotropic medication use. Having an affective or anxiety disorder (more frequent in women and users versus nonusers; see Table S3, Supplemental Digital Content 1, http://links.lww.com/JCP/A977, for frequencies or means) explained >10% of the association between gender and each of the outcomes (10.4 to 24.3%). Occupation (less frequent in women and users vs. nonusers) explained >10% of the association of gender with any psychotropic use, anxiolytic use and the number of classes of psychotropics used (16.2-22.4%). SmS score, indicative of subjective somatic complaints (higher in women and users vs. nonusers) explained >10% of the association between gender and the same outcomes (11.2-15.1%) and almost explained >10% of the association between gender and antidepressant use (9.9%). OQ-45 Symptomatic Distress scores (higher in women and antidepressant users vs. nonusers) explained 11.6% of the association between gender and antidepressant use. Finally, the number of previous MHS visits and illness duration explained >10% of the association between gender and anxiolytic use (more visits in women and users vs. non-users).

TABLE 4.

Analysis of Explanatory factors using IPW weighted Regression Analyses to Adjust Gender Associations With Medication Use for different Covariates

		Using psychotropics			Antidepressants			Anxiolytics			Polypharmacy
Covariate	N	Unw. B	W. B	% change	Unw. B	W. B	% change	Unw.	W. B	% change	Unw. B	W. B	% change
Living situation	1.653				0.489	0.483	−1.2	0.306	0.299	−2.3	0.371	0.366	−1.4
Education level	1.674	0.339	0.339	0.0	0.486	0.486	0.0	0.315	0.314	−0.3	0.359	0.360	0.3
Occupation	1.750	0.370	0.31	−16.2	0.492	0.454	−7.7	0.308	0.239	−22.4	0.375	0.310	−17.3
Duration of illness	1.515							0.291	0.261	−10.3
Duration of current episode	1.511	0.247	0.246	−0.4	0.404	0.404	0.0	0.287	0.287	0.0	0.267	0.266	−0.4
Somatic complaints in medical history	1.807	0.321	0.307	−4.4	0.416	0.408	−1.9	0.324	0.315	−2.8	0.329	0.316	−4.0
Psychiatric comorbidities	2.216	0.268	0.266	−0.7	0.391	0.393	0.5	0.317	0.321	1.3	0.287	0.286	−0.4
Previously visited MHS	1.524	0.250	0.229	−8.4	0.416	0.393	−5.5	0.279	0.251	−10.0	0.271	0.251	−7.4
Affective or anxiety disorder	2.216	0.268	0.203	−24.3	0.391	0.334	−14.6	0.317	0.284	−10.4	0.287	0.227	−20.9
Substance use disorder	2.216	0.268	0.274	2.2	0.391	0.4	2.3	0.317	0.338	6.6	0.287	0.299	4.2
Alcohol use	1.947	0.299	0.292	−2.3	0.451	0.438	−2.9	0.282	0.28	−0.7	0.306	0.283	−7,5
Smoking	1.892	0.261	0.271	3.8	0.417	0.42	0.7	0.258	0.273	5.8	0.278	0.288	3,6
Cannabis use	1.949	0.294	0.271	−7.8	0.451	0.425	−5.8	0.275	0.254	−7.6	0.303	0.280	−7,6
Hard drug use (yes/no)	1.949	0.294	0.29	−1.4	0.451	0.45	−0.2	0.275	0.269	−2.2	0.303	0.300	−1.0
Lifestyle score	2.430	0.317	0.312	−1.6	0.411	0.408	−0.7	0.325	0.311	−4.3	0.325	0.318	−2.2
OQ-45 Symptomatic distress	1.249	0.453	0.424	−6.4	0.580	0.513	−11.6	0.406	0.374	−7.9	0.456	0.433	−5.0
OQ-45 Interpersonal functioning	1.249	0.453	0.453	0.0	0.580	0.58	0.0	0.406	0.407	0.2	0.456	0.456	0.0
OQ-45 Social Role	1.249	0.453	0.454	0.2	0.580	0.597	2.9	0.406	0.409	0.7	0.456	0.455	−0.2
SmS side effects	1.894	0.294	0.261	−11.2	0.453	0.408	−9.9	0.291	0.247	−15.1	0.300	0.265	−11.7

Values with a > 10% change are indicated in bold.

Unw. B indicates unweighted beta, W.B, weighted beta, % change is percentual change of beta between weighted and unweighted analyses.

DISCUSSION

This study investigated differences between men and women in psychotropic medication use (versus nonuse), average prescribed daily dose, and the number of used psychotropic classes in a large natural cohort of psychiatric outpatients. In addition, we investigated for a broad set of predictors whether they modified or explained any of the observed associations between gender and psychotropic medication use. To summarize the results, compared with men, more women used psychotropic medication, more specifically antidepressants and anxiolytics. Women also used more classes of psychotropics simultaneously at all age ranges. Daily dosages did not differ between genders, except for a slightly higher PDD for mood stabilizers in men compared with women. The frequency of psychotropic use was higher in women than in men with longer illness duration, while there were no differences for a shorter duration of illness. In addition, women with a partner used psychotropics more frequently than men with a partner, while there was no difference between men and women without a partner. Several explanatory factors for gender differences were identified with weighted regression analyses: Females on psychotropic medication were more often diagnosed with affective/anxiety diagnoses, experienced more side effects, and visited more health care services than males, reaffirming findings from previous studies. Below, the interpretations of these findings are discussed in more detail.

Our findings of more frequent use of psychotropic medication by women, specifically antidepressants and anxiolytics/sedatives, and similar frequency of use of antipsychotics, have been consistently shown by other studies as well.^1–3,10,48 We did not observe gender differences in the frequency of use of mood stabilizers, while previous studies have reported contradictory findings on this topic.^5,36 Between-study differences may be the result of the relatively low number of patients with bipolar disorders in our study and, as a consequence, a lack of power to find meaningful differences. Interestingly, our findings showed that more individuals used psychotropic medication with increasing age across both genders. However, gender differences in the frequency of psychotropic use did not change significantly with age. This indicates that in the current study, changes in female hormonal status (ie, premenopause vs. postmenopause) did not influence medication use in men versus women. This is in line with a large body of previous studies that also failed to show an effect.^29–33

In the current study, the prescribed dosages of psychotropics did not differ significantly between women and men, as shown by one previous study for antipsychotics and benzodiazepines,⁸ except for the dosage of mood stabilizers, which was higher in men. While some studies showed equal frequency of use of mood stabilizers,^4,5 we are not aware of other studies investigating gender differences in dosing of mood stabilizers. We could speculate that, given the strict monitoring of blood levels of especially lithium, compared with other drug classes, men may eventually be prescribed higher doses in this case. However, this phenomenon needs further investigation. The equal dosing for the other psychotropic classes is interesting, because one can hypothesize that women may actually need lower dosing of psychotropics compared with men,^36,37 given gender differences in body weight, adipose tissue distribution, and interactions between hormonal state, and drug receptor binding, as well as drug metabolism and turnover.^49,50 Moreover, psychotropic blood levels could thus be influenced by gender-related differences in pharmacokinetics, drug-drug interaction, and also the use of other substances (especially drugs of abuse).^26,49,50 Unfortunately, because blood levels of psychotropic medication were not available in our study, we could not investigate whether equal dosages for both genders actually resulted in higher drug plasma levels in women compared with men. Currently, gender-related recommendations are not part of Dutch pharmacotherapeutic guidelines of affective and most other psychiatric disorders, and clinicians are generally not aware of potential differences between men and women in psychotropic prescription.^27,51

With respect to potential explanatory factors for the more frequent use of psychotropics in women, we found that higher frequencies of an affective/anxiety disorder and more reported somatic complaints in women vs. men may partially explain the more frequent use of any psychotropic, antidepressants, anxiolytics and a higher number of psychotropic classes in women compared with men. Further, higher symptomatic distress scores in women were found to explain part of the association between gender and antidepressant use, and a higher number of previously visited MHS and longer illness duration in women explained part of the association between gender and anxiolytic use. These exploratory findings fit well with our hypotheses and previous literature.^{1,3,5,9–13,19,20,23,48,52,53} More specially, although the cross-sectional nature of our study does not allow for conclusions about causal relationships, it has been shown that women with a diagnosis of depression have generally higher rates of comorbid diagnoses, experience more subdromal, atypical and somatic symptoms, experience more episodes than men, which may partly explain the higher frequency of psychotropic use and of mental health services use.^1,5,48 In addition, higher reported scores of experienced psychiatric and somatic complaints may be caused by women's higher awareness of psychological and somatic complaints and their stronger tendency for help seeking compared with men, resulting in clinicians responding by increased medication prescription to treat these symptoms, ultimately leading to more experienced side effects.^2,3,11,20

Concerning effect modifiers, we found that in those with a longer illness duration, more women used psychotropics in general, and women used more antidepressants than men, while there were no gender differences in individuals with a short duration of illness (for anxiolytics this was an explanatory factor, as described in the previous paragraph). To the best of our knowledge, this factor has not been described as related to psychotropic use differences in other studies and might be in line with gender differences in help-seeking behavior as described in the previous paragraph. Intriguingly, females with a partner used psychotropics more frequently than males with a partner, while there was no difference between individuals without a partner. Finally, occupational status was found to be an explanatory factor for the association between gender and medication use; women had less often an occupation than men, which may indicate that not having meaningful or structured daytime activities may be related to increased psychotropic intake in females. This seems to contradict previous studies showing that unemployment or uncertainty in employment and being without a partner were related to increased medication intake in men, but not in women.^10,12,13,19 Maybe, this results from cultural differences; our female patients may be more often employed than in other studies, which may contribute more to exacerbating and maintaining psychological complaints than in previous studies.

This study had several strengths and limitations. The strengths include the large, naturalistic, real-world data set with a broad set of clinical and sociodemographic characteristics. However, several limitations should also be kept in mind. First, the data were cross-sectional, originated from outpatient services in a predominantly rural region of the Netherlands, and therefore generalizability to the total outpatient population is limited. Second, although we could investigate a sizable range of variables, we lacked data on other potentially important explanatory factors, such as actual medication compliance (eg, by assessing blood levels of medication), prescriber characteristics and personality or coping styles of patients. Third, for some variables, a substantial percentage of data was missing. The main reason for missingness was that these data were not collected throughout the whole study period. We therefore assumed that missingness was not systematically related to our study outcomes. Fourth, part of the psychotropics investigated may be for other indications than a psychiatric diagnosis and therefore prescribed at lower doses. However, as we observed that doses gradually increased and doses advised for other indications overlap with doses used in psychiatric treatment, we believe that we could not define clear dose cutoffs to exclude (off-label) psychotropic medication use for other purposes. As a final limitation, the study was exploratory in nature. We did not correct for multiple comparisons and used an arbitrary cutoff (10% on the log-odds scale) to determine whether a variable was of exploratory importance.

In conclusion, this study showed that women more often receive psychotropic medication, including antidepressants and anxiolytics, than men, but equal dosages apart from mood stabilizers. Female psychotropic users had more often an affective/anxiety diagnosis, experienced more somatic complaints and had used more care than males. Future longitudinal studies could investigate whether these factors are cause or consequence of gender differences in psychotropic medication use.