Medications That Cause Dry Mouth As an Adverse Effect in Older People

A Systematic Review and Metaanalysis

Edwin C. K. Tan, PhD; Duangjai Lexomboon, PhD; Gunilla Sandborgh-Englund, PhD; Ylva Haasum, PhD; Kristina Johnell, PhD


J Am Geriatr Soc. 2018;66(1):76-84. 

In This Article


Study Selection

The search of electronic databases retrieved 2,591 articles. After removing duplicates, the titles and abstracts of 1,544 studies were reviewed, of which 867 were excluded because they clearly did not meet the inclusion criteria. The full-text versions of 677 articles were obtained and scrutinized, and 624 were excluded after independent review by at least two authors (Figure 1). A total of 52 studies were included in the final review and are summarized below and in Supplementary Table S2 and Table S3.

Figure 1.

PRISMA flow diagram. [Color figure can be viewed at]

Study Characteristics

Of the included studies, 33 were experimental studies, including 25 placebo controlled[9–33] and 8 with active comparators[29,34–40] (1 pooled analysis[29] included both). Six studies were pooled analyses of multiple trials.[15,20,25,28,29,32] The remaining 19 studies were observational and used a cross-sectional study design.[41–59] The majority of studies were conducted in North America (n = 19),[11,18,19,23,26,34–36,39,40,42,43,46,48,52–54,59] followed by Europe (n = 13)[9,13,14,16,22,30,31,41,44,49,51,56,58] and Asia (n = 7).[12,38,45,50,55,57] Thirteen studies were undertaken across multiple continents as part of large, multinational trials.[10,15,17,20,24,25,27–29,32,33,37]

Study sample sizes ranged from 11 to 13,508 subjects. Most studies included older individuals who were generally healthy or without specific comorbidities (n = 19).[9,18,19,41,44–53,55–59] The most common disease states investigated were urological, including urinary incontinence (n = 15);[12,13,15,16,20,22,27,29,31–33,36,40,43] mental health, including major depression (n = 8),[11,17,25,26,34,35,38,39] generalized anxiety disorder (n = 2),[10,24] psychosis (n = 2),[37,42] cognitive impairment (n = 1),[31] and insomnia (n = 2);[21,23] neuralgia (n = 1);[28] and hypertension (n = 2).[14,54]

Drug Exposure

A wide range of drug treatments was assessed across the included studies. Experimental studies tested specific drugs, the most common being those used for urinary frequency and incontinence (Anatomical Therapeutic Chemical code G04BD, n = 16),[12,13,15,16,18–20,22,29–33,36,40,43] antidepressants (N06A, n = 10)[10,11,21,25–27,34,35,38,39] and antipsychotics (N05A, n = 3).[17,24,37] One study each assessed cardiovascular drugs (C03[14] and C07[54]), hypnotics and sedatives (N05C),[23] and antiepileptics for neuropathic pain (N03A).[28]

The cross-sectional studies mainly investigated broad or nonspecific groups of drugs as exposures. Eight studies assessed number of drugs,[45,47,49,51,53,56,58] and two studies assessed use of any medication.[44,52] Six studies assessed different types or classes of drugs,[45,47,50,51,56,58] 3 studies assessed xerogenic or hyposalivation-inducing drug use,[46,53,59] and 2 studies assessed anticholinergic drug use.[41,48] When a duration of drug exposure was reported, this ranged from use within the previous 2 weeks[44,47] to 1 year.[46]


All experimental studies assessed xerostomia as a dichotomous outcome, without further description on how this was assessed. Follow-up duration for the assessment of xerostomia in experimental studies ranged from 1 week[19] to 3 years.[14] Eleven cross-sectional studies had xerostomia as an outcome,[41–51] and eight had salivary gland hypofunction.[52–59] One study differentiated between daytime and nighttime xerostomia,[44] 2 studies assessed severity of xerostomia,[47,50] and 2 studies objectively assessed oral moisture.[45,50] Regarding salivary gland hypofunction outcomes, four studies assessed stimulated salivary flow rate,[52,55–57] with two studies also assessing unstimulated salivary flow rate;[52,57] one study assessed parotid and submandibular gland secretion rate;[54] one study assessed hyposalivation;[58] and one study assessed minor salivary gland output.[53]

Risk of Bias Within Studies

The majority of experimental studies (n = 15) received a Jadad score of 3 out of 5,[9–11,13–15,17–37,40] with 6 studies scoring the maximum 5 out of 5.[9,15,17,19,24,31,33] The main limitations of included RCT studies were their failure to report appropriate methods of randomization sequence generation[12–14,16,20,22,23,25–30,32,34,35,37–40] and double blinding[10–13,16,18,20,21,23,25–29,32,34,35,37–39] (n = 21 for each). The majority of cross-sectional studies scored 4 stars out of 10 on the adapted Newcastle-Ottawa Scale.[41,43,46,48,50,55,59] One study scored the maximum 10 stars,[44] and 3 studies scored 3 stars.[52,54,57] The main limitation of the cross-sectional studies was related to selection of the study sample. The majority of studies failed to describe a sample that was truly or somewhat representative of the target population (n = 12),[41,43,45,46,48–50,52,54,55,57,59] include a sample size that was justified and satisfactory (n = 13),[41,43,45,46,48,50,52–55,57–59] or adequately establish the comparability of respondents and nonrespondents and determine whether the response rate was satisfactory (n = 15).[41–43,45,46,48,50–55,57–59] Full reports of quality assessment are presented in Supplementary Table S4 and Table S5 and Supplementary Text S1 and Text S2.

Results of Individual Studies

In the experimental studies, a greater proportion of those undergoing drug treatment reported xerostomia than of those receiving placebo. In the cross-sectional studies, medication use, number of medications, and using a wide range of medication classes were significantly associated with xerostomia, salivary flow rate, and salivary gland hypofunction. In studies investigating number of drugs as the exposure, all except 1[41] reported a significant association between number of drugs and dry mouth in older people. Main findings are summarized in Supplementary Table S1 and Table S2.


Twenty-two placebo-controlled RCTs were deemed eligible for further inclusion in metaanalyses.[10–12,15–17,19–27,29–33]

Of the RCTs, included trials investigated drugs used for urinary frequency and incontinence (ATC code G04BD) (n = 13),[12,15,16,19,20,22,29–33] antidepressants (N06A) (n = 6),[10,11,21,25–27] and psycholeptics (N05) (n = 3).[17,23,24] Statistical heterogeneity was substantial across the trials assessing drugs used for urinary frequency and incontinence (I[2] = 62%). The main drugs assessed in this group included tolterodine (n = 4)[12,22,29,33] and oxybutynin (n = 3).[16,19,30] Of the tolterodine studies, 1 investigated twice-daily administration of the immediate-release preparation,[22] with the remainder assessing extended-release formulations. All oxybutynin studies included in the metaanalysis assessed immediate-release oral preparations. The metaanalysis found that urological medications were significantly associated with xerostomia (OR = 5.91, 95% CI = 4.04–8.63; RD = 0.25, 95% CI = 0.13–0.36) (Figure 2, Supplementary Figure S1a). When examining specific urological drugs, oral immediate-release oxybutynin was associated with the greatest risk of dry mouth (RD = 0.56, 95% CI = 0.43–0.70), whereas the risk with mirabegron was negligible (RD = 0.00, 95% CI = −0.01–0.02). Statistical heterogeneity was low across trials investigating antidepressants (I[2] = 21%). The main drug assessed in this group was duloxetine (n = 4). Antidepressant use was significantly associated with xerostomia (OR = 4.74, 95% CI = 2.69–8.32; RD = 0.09, 95% CI = 0.04–0.14) (Figure 3, Supplementary Figure S1b). Statistical heterogeneity was low across trials assessing psycholeptics (I[2] = 0%). Psycholeptics were more likely than placebo to cause xerostomia (OR = 2.59, 95% CI = 1.79–3.95; RD = 0.09, 95% CI = 0.05–0.12) (Figure 4, Supplementary Figure S1c).

Figure 2.

Forest plots for randomized controlled trials of drugs used for urinary frequency and incontinence (G04BD) vs placebo. [Color figure can be viewed at]

Figure 3.

Forest plots for randomized controlled trials of antidepressants (N06A) vs placebo. [Color figure can be viewed at]

Figure 4.

Forest plots for randomized controlled trials of psycholeptics (N05) vs placebo. [Color figure can be viewed at]