• adherence;
  • compliance;
  • Cox proportional hazard model;
  • logistic regression model;
  • overactive bladder


  1. Top of page
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. References

Objectives:  To investigate persistence and adherence of medication treatment in chronic overactive bladder/urinary incontinence (OAB/UI) patients, and to evaluate OAB/UI-related comorbidity events associated with persistence.

Methods:  Pharmaceutical outcomes research with a health-care provider perspective was conducted on a California Medicaid (Medi-Cal) chronic OAB/UI population. The primary end point was medication possession ratio (MPR), which was used to measure refill adherence. Secondary end points measuring persistence patterns included discontinuation of OAB drug therapy (medication-uncovered interval > 30 days) and time to discontinuation  (period  from  the  index  date  until  the first discontinuation date). Significant factors on nonpersistence were found by using a Cox Proportional Hazards model. Factors contributing to nonadherence (MPR < 0.8) and the relationship between OAB/UI comorbidity events and persistence were examined by logistic regressions.

Results:  Of 2496 eligible patients, 36.9% had only one OAB/UI prescription. The mean MPR was 0.34 (SD 0.21) and the median was 0.3, indicating that on average only about one-third of period of time since medication initiation was covered by the therapy. Only 122 patients exhibited > 80% adherence during the 6-month follow-up-period. Significant predictors of higher persistence included: white ethnicity, previous hospitalization length, starting with tolterodine or oxybutynin extended-release, and previous use of topical drugs or antipsychotics. Nevertheless, previous depression or urinary tract infection (UTI) diagnosis, polypharmacy, significantly increased the odds of early discontinuation. Treatment discontinuation increased the risk of UTI diagnosis by 37% in the post-treatment period (= 0.03; OR 1.37; 95% CI 1.03–1.84), but had no significant effect on other OAB/UI-related comorbidities.

Conclusions:  For chronic OAB/UI patients identified in this study, both persistence and adherence with medication treatment were suboptimal. These results suggest that persistence and treatment discontinuation remains problematic for the OAB/UI population.


  1. Top of page
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. References

Urinary incontinence (incontinence, UI) is a condition referring to any bladder dysfunction of storing and voiding urine [1,2], and can be transient (reversible) or chronic. Transient incontinence is featured with sudden onset, often secondary to iatrogenic ailments, acute illness, or medication use, whereas chronic incontinence lasts over a relatively longer period of time without iatrogenic or acute illness [3]. Overactive bladder (OAB), defined by the International Continence Society (ICS) in 1988, is “characterized by involuntary detrusor contractions during the filling phase, which may be spontaneous or provoked and which the patient cannot completely suppress [4].” Later, the definition of OAB was modified by ICS in 1999, 2000, and 2001. Specifically, OAB is now defined as “denoting urgency, with or without urge incontinence, usually with frequency and nocturia [5].” Hence, the current OAB (i.e., OAB syndrome) definition includes urge syndrome and urgency–frequency syndrome [5].

It is estimated that 12 million individuals suffer from OAB/UI in the United States [6]. This condition is not as prevalent as asthma (20.3 million) [7], but more prevalent than diabetes mellitus among adults (10.8 million [8]), osteoporosis (10 million [9]) and Alzheimer's disease (4 million [10]). The prevalence of OAB/UI, however, varies in the literature and this condition is often underdiagnosed because of the lack of standard or unified definitions being established, embarrassment of patients to report their symptoms, and/or unawareness of the onset of this disease. Consequently, up to 50% of all cases may not be captured in retrospective data [2].

Whether captured or not, OAB/UI exerts a profound effect on a patient's daily life in both medical and psychological aspects. The patients have an increased incidence rate of urinary tract infections (UTIs), pressure ulcers, urosepsis, perineal rashes, and falls and fractures [11]. Psychologically, this condition may bring about anxiety, depression and isolation to affected patients. They face limitations in social activities, may gradually lose independence, and have a lower quality of life. Approximately $11 billion is spent annually to control OAB/UI symptoms and treat OAB-related comorbidities in the United States [12].

Treatment options for OAB/UI include behavioral therapy, bladder retraining programs, pharmacological therapy, and surgery. Because drug therapy is targeted to specific abnormalities, the choice of specific course of drug therapy is governed by the type of OAB/UI diagnosed. Currently, major medications for OAB/UI are composed of flavoxate, oxybutynin, tolterodine, oxybutynin extended-release (ER), and tolterodine ER.

Patients’ compliance and persistence during the course of pharmacological therapy is a major issue in treating OAB/UI, however, little is known about the consequences of poor compliance and persistence of OAB/UI treatment. Pharmaceutical outcomes researchers have found that persistence of long-term medication is typically low (about 50%) in a number of chronic disease areas [13–15], and chronic OAB/UI is likely to be one of them. Patients may give up in continuing medications or the physician might decide to stop prescribing a medication if the side effects or complications of the drug therapy outweigh the known benefits. A Canadian study (in Quebec) on incontinence drug utilization patterns revealed low renewal rates of prescriptions (< 40%) and poor persistence after 6 months (< 12%) [16]. A study of treatment of urge incontinence in veterans of the United States showed the majority of patients who received a prescription did not routinely refill medication over the course of the study, and less than 30% of the patients were able to continue on therapy for 1 years [17]. Currently, there are few published studies on the long-term persistence with drug therapy among OAB/UI patients in the United States and the potential benefits of persistence and adherence are less studied [18,19].

The objectives of the present study include three components aimed to benefit health-care providers: 1) to investigate persistence pattern and refill adherence among chronic California Medicaid OAB/UI patients; 2) to build the risk profile of discontinuation and nonadherence for OAB/UI patients; and 3) to evaluate OAB/UI-related comorbidity events associated with persistence.

Data and Methods

  1. Top of page
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. References


The study cohort was extracted from a 20% random sample of the administrative files provided by the California Medicaid program (Medi-Cal) from January 1999 to April 2002. Medi-Cal is a program that finances California residents who have limited resources and income, with a wide range of health-care services. All claims data are captured by a computerized database of linked pharmacy, physician, hospital and nursing home records and encoded to preserve patient and physician anonymity. Each record contains information on demographic characteristics, the specific drug dispensed (NDC codes), dosage, quantity dispensed, number of days’ supply, diagnosis (ICD-9 codes), service date and place, health-care procedures (CPT codes), length of stay, and eligibility profile. In addition, several variables are created based on these raw data, index drug (OAB/UI medication dispensed at the index date; oxybutynin, tolterodine, oxybutynin ER, and other; data on tolterodine ER were not available at the given study period), switch of drug therapy (when a patient filled a different brand of OAB/UI drug from the index drug after index date at the first time), Charlson Comorbidity Index (CCI) [20], outpatient visits, inpatient hospitalizations, and emergency room visits and the following outcome-related variables. In this study, three different CCI's were calculated. Pre-index CCI was computed by using the associated disease diagnosis codes or procedure codes recorded within 6 months before the index date, not including index date, and index date CCI was calculated by related codes only on the index date. In this study, five OAB/UI-related comorbidities are considered: obesity, depression, UTI, skin infection, and fall/fracture. The diagnosis of comorbidities before the first discontinuation was not counted to avoid potential endogeneity and more robust prediction of comorbidities. For UTI incidence, adjusted incidence rate was used to reflect clinical practice pattern. Since symptoms of OAB in early stage, acute or chronic, mimic UTI, many physicians initially diagnose as UTI and later change to OAB if the symptoms continue. To avoid overestimating the incidence of UTI, those UTI diagnoses that occurred within 30 days of the OAB diagnosis were not counted as a UTI.

Patient Identification

Chronic OAB/UI patients receiving drug therapy were identified as adult patients (age ≥ 18 years) who 1) were dispensed any OAB/UI medication during the index period from July 1999 to April 2001 (the date of the first OAB/UI prescription was marked as the index date), and 2a) were diagnosed with OAB/UI at least twice from January 1999 to April 2002 (study period), or 2b) ever diagnosed with OAB/UI after the index date (initial selection of 3665 patients). These selection criteria were relatively conservative for the given Medi-Cal data set, as there were 14,563 patients with at least one OAB/UI diagnosis; 12,528 patients with at least two diagnoses; 10,861 patients with at least one OAB/UI prescription; 7158 patients with at least two prescriptions in the study period. To evaluate the consequences of OAB/UI drug therapies more accurately, those enrollees dispensed an OAB/UI drug 6 months before the index date (washout period/preperiod) were excluded (n = 712). A few patients (n = 16) were excluded because of multiple OAB/UI prescriptions filled on the index date. Finally, 441 subjects who were not continuously enrolled in the whole study period (6-month preperiod and 6-month follow-up period) were excluded, yielding a final study cohort of 2496 subjects.

Outcome Measures


Persistence in the present study was measured by the length of continuous pharmacological treatment. Patients discontinued their treatment if they failed to refill OAB/UI agents within 30 days after the expected end date of the previous prescription. Time to discontinuation was defined as the number of days between the index date and the first discontinuation date. Patients who switched from one brand of OAB/UI drug to another within 30 days were considered persistent on their therapy. The purpose of allowing a refill gap was to avoid misclassified discontinuations when patients delayed the purchase of their medication by a few days. Patients who restarted any brand of OAB/UI drug after their first discontinuation by the end of the postperiod were considered as “restarters” of medication treatment.

Descriptive analysis for persistence pattern was carried out on two levels: no discontinuation versus discontinuation using the entire study population and a subanalysis of discontinuers with no restart of drug therapy versus discontinuers with restart of drug therapy.

Adherence.  Adherence is the degree of prescription-filling in a given time frame. In a large observational study, failure to refill is considered the most reliable objective measure of adherence [21]. A medication possession ratio (MPR) was used in this study as a measure of failure to refill, calculated as the sum of days with supply of any OAB/UI drug prescribed over the follow-up period divided by the total days followed up (181 days for the 6-month follow-up period). If the last prescription filled in the follow-up period continued after the end of the follow-up period, surplus days outside the follow-up period were not counted in the numerator when MPR was computed. An alternative way of computing MPR is to allow a variable window for each patient based on the last prescription fill in the postperiod. For example, if the last prescription fill includes 10 surplus days after the end of the 6-month postperiod, the denominator of the MPR formula becomes 191 days rather than a fixed 181-day window. Accordingly, the maximum numerator can be up to 191 days. The alternative method showed a slightly higher MPR than the one used in this study (0.36 vs. 0.34), however, the implied adherence was not significantly different (= 0.605). Hence, MPR based on 6-month fixed period was used in this study. For patients simultaneously taking two or more agents in the same class, the overlapping medication-covered days were not double counted. Based on each individual patient's MPR, the cohort was categorized into two groups: those adherents with drug therapy (MPR ≥ 0.8) and those not adherent with drug therapy (MPR < 0.8). To examine the variation of adherence rate due to different MPR cutoff, 0.5 through 0.9 in addition to 0.8 from the literature [22–26], were used for a sensitivity analysis on adherence rate and MPR.

Statistical Analysis

Univariate analyses were used to compare persistence groupings (no discontinuation vs. discontinuation, no restarter discontinuation vs. restarter discontinuation). The Wilcoxon ranksum test was used for continuous variables and chi-square test was used for categorical variables.

A survival curve describing the crude persistence pattern of OAB/UI treatment was generated using the unadjusted Kaplan–Meier method. Patients who remained on their initial treatment until the end of the follow-up period were regarded as censored data, and the length of the follow-up period was assigned as their survival time. Cox proportional hazard regression was applied to investigate the effect of explanatory factors on persistence duration. Time to discontinuation was modeled as the dependent variable, and the independent variables included age, sex, ethnicity, index drug, and other baseline characteristics such as comorbidity profile, medication use history, health-care utilization, length of hospitalization, and the number of drug classes prescribed. Partial Hazard assumption was tested by the significance of added time dependent covariates (covariates interacted with logged time to discontinuation). If any time dependent covariate is found to be significant, it will substitute the original covariates to capture time varying effects.

Multivariate logistic regression analysis was implemented to calculate the marginal risk of the aforementioned factors in the Cox model for nonadherence (dependent variable), and to examine the relationship between persistence and clinical outcomes in terms of incidence of OAB/UI-related comorbidities.

All analyses were performed using sas version 9.1 (SAS Institute Inc., Cary, NC, 2003) and statistical significance was determined at the level of P value = 0.05 (significance at 10% level was also provided in the first two tables of subgroup comparisons only for a better explanation purpose). The stepwise selection method was used to include candidate covariates in the model with a P value equal to 0.25 as a cutoff point, as recommended by Hosmer and Lemeshow [27]. Other levels frequently used range from 0.15 to 0.25 [27–32].


  1. Top of page
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. References

Patient Population Characteristics

For 2496 patients who met inclusion criteria, the baseline demographic and clinical characteristics along with comparisons by persistence groupings are presented in Table 1 (including no discontinuation vs. discontinuation comparison) and Table 2 (including no restarter discontinuation vs. restarter discontinuation comparison). Twenty percent of the entire study population (534 of 2496 patients) was male and more than half of the patients (1276 of 2496 patients) were white. The average age at the time of index prescription was 63.2 years; and most patients began treatment with tolterodine (44%, 1093 of 2496 patients) or oxybutynin (33%, 812 of 2496 patients). Only 524 patients (21%) were dispensed oxybutynin ER. The initiation of tolterodine ER was not observed in this cohort because of the timing of this study in relation to the approval of the drug. The no discontinuation and discontinuation subcohorts had similar baseline characteristics except that the no discontinuation subcohort had more males (25.0% vs. 20.4%; P = 0.0207), more whites (ethnic composition comparison: P < 0.01), more obese patients (= 0.0267), a higher hospitalization rate (= 0.0144), a higher topical drug use rate (< 0.01), and a higher rate of antipsychotics use (< 0.01). The discontinuation subcohort had a lower percentage of patients who initiated oxybutynin ER and a higher percentage of patients who initiated oxybutynin (index drug comparison: P < 0.01). In the comparison of no-restarters versus restarters groups, restarters group had a higher proportion of patients who were eligible for both Medicaid and Medicare (< 0.01), about 4% fewer male patients (= 0.088), and 5% higher antidepressant use (= 0.0139).

Table 1.  Baseline demographics and clinical characteristics of study cohort (n = 2496)
 No discontinuationDiscontinuationP valueOverall
  1. CCI, Charlson Comorbidity Index not including age.

N5531943 2496
Age (mean, SD)   62.95 (16.79) 63.21 (15.97)0.76 63.15 (16.14)
Dual eligibility: Medicare + Medicaid n (%)156 (28.2%)560 (28.8%)0.78716
Male n (%)138 (25.0%)396 (20.4%)0.02534
Ethnicity  <0.01 
 White n (%)328 (59.3%)948 (48.8%) 1276
 African American n (%)44 (8.0%)167 (8.6%) 211
 Hispanic n (%)19 (3.4%)98 (5.0%) 117
 Asian n (%)25 (4.5%)107 (5.5%) 132
 Other ethnicities n (%)137 (24.8%)621 (32.0%) 758
Drug prescribed on the index date  <0.01 
 Tolterodine255 (46.1%)838 (43.1%) 1093
 Oxybutynin extended-release144 (26.0%)380 (19.6%) 524
 Oxybutynin144 (26.0%)668 (34.4%) 812
 Other OAB drugs10 (1.8%)57 (2.9%) 67
CCI for pre-index 6 months (mean, SD)  1.56 (1.68) 1.5 (1.75)0.171.52 (1.73)
CCI on the index date (mean, SD)  0.12 (0.45)  0.09 (0.36)0.100.09 (0.38)
CCI for the whole study period (mean, SD)  1.56 (1.68) 1.5 (1.75)0.171.52 (1.73)
OAB/UI-related comorbidities
 Obesity n (%)22 (4.0%)44 (2.3%)0.0366
 Depression n (%)38 (6.9%)138 (7.1%)0.85176
 Urinary tract infection n (%) 87 (15.7%)375 (19.3%)0.06462
 Skin infection n (%)39 (7.1%)113 (5.8%)0.28152
 Fall/fracture n (%) 59 (10.7%)227 (11.7%)0.51286
Health-care utilization
 Number of drug classes prescribed (mean, SD)  7.28 (3.98)  7.18 (4.08)0.577.20 (4.06)
 Physician office encounters (mean, SD)  14.21 (15.81) 13.2 (14.08)0.1113.43 (14.48)
 Any inpatient visit n (%)133 (24.1%)375 (19.3%)0.01508
 Any emergency room visit n (%)30 (5.4%)127 (6.5%)0.34157
Medication use
 Any UTI drug use n (%)35 (6.3%)168 (8.6%)0.08203
 Any topical drug use n (%)239 (43.2%)677 (34.8%) 916
 Any antidepressant use n (%)165 (29.8%)578 (29.7%) 743
 Any antipsychotics use n (%)150 (27.1%)258 (13.3%) 408
Table 2.  Baseline demographic and clinical characteristics of patients discontinued therapy (n = 1943)
 Discontinuation* and no-restartDiscontinuation and restartP valueAll discontinuation
  • *

    Discontinuation was defined as medication-uncover interval > 30 occurred during the 6-month follow-up period.

  • Restart was define as patients filled any brand of OAB/UI drug after their first discontinuation by the end of the 6-month follow-up period.

  • CCI, Charlson Comorbidity Index not including age.

N1380563 1943
Age (mean, SD)62.97 (16.12)63.8 (15.58)0.463.21 (15.97)
Dual eligibility: Medicare + Medicaid n (%)374 (27.1%)186 (33.0%)<0.01560
Male n (%)295 (21.4%)101 (17.9%)0.09396
Ethnicity  0.73 
 White n (%)678 (49.1%)270 (48.0%) 948
 African American n (%)116 (8.4%)51 (9.1%) 167
 Hispanic n (%)70 (5.1%)28 (5.0%) 98
 Asian n (%)70 (5.1%)37 (6.6%) 107
 Other ethnicities n (%)445 (32.3%)176 (31.3%) 621
Drug prescribed on the index date  0.36 
 Tolterodine587 (42.5%)251 (44.6%) 838
 Oxybutynin extended-release274 (19.3%)106 (18.8%) 380
 Oxybutynin473 (34.3%)195 (34.6%) 668
 Other OAB drugs46 (3.3%)11 (2.0%) 57
CCI for pre-index 6 months (mean, SD)1.49 (1.77)1.52 (1.71)0.54 1.5 (1.75)
CCI on the index date (mean, SD)0.09 (0.38)0.07 (0.31)0.30 0.09 (0.36)
CCI for the whole study period (mean, SD)1.49 (1.77)1.52 (1.71)0.55 1.5 (1.75)
OAB/UI-related comorbidities
 Obesity n (%)32 (2.3%)12 (2.1%)0.8044
 Depression n (%)100 (7.3%) 38 (6.75%)0.70138
 Urinary tract infection n (%)257 (18.6%)118 (21.0%)0.24375
 Skin infection n (%)81 (5.9%)32 (5.7%)0.87113
 Fall/fracture n (%)156 (11.3%) 71 (12.6%)0.42227
Health-care utilization
 Number of drug classes prescribed (mean, SD)7.27 (4.16)  6.97 (3.88)0.26 7.18 (4.08)
 Physician office encounters (mean, SD)13.36 (14.39)12.82 (13.28)0.38 13.2 (14.08)
 Any inpatient visit n (%)273 (19.8%)102 (18.1%)0.40375
 Any emergency room visit n (%)92 (6.7%)35 (6.2%)0.72127
Medication use
 Any UTI drug use n (%)120 (8.7%)48 (8.5%)0.90168
 Any topical drug use n (%)467 (33.8%)210 (37.3%)0.15677
 Any antidepressant use n (%)433 (31.4%)145 (25.8%)<0.01578
 Any antipsychotics use n (%)181 (13.1%) 77 (13.7%)0.74258

The mean CCI of patients and OAB/UI-related comorbidities [33] were not significantly different in no discontinuation versus discontinuation comparison nor in no restarter versus restarter comparison. Among OAB/UI-related comorbidities, UTI was most prevalent (18.5%, 462 of 2496), followed by falls/fractures (11.5%, 286 of 2496), depression (7.1%, 176 of 2496), skin infections (6.1%, 152 of 2496), and obesity (2.6%, 66 of 2496).

Health-care resources were highly utilized by patients in the 6-month preperiod. On average, patients filled seven classes of drug and had 13 outpatient visits in this period. About 20% of the patients were hospitalized at least once and 157 patients (6.3%) had at least one emergency room visit within this period.

Topical drugs were most widely used before index  date  among  the  study  population,  and  the use of antidepressants was second most common followed by antipsychotics and UTI drugs. Significant difference was shown in the use of topical drugs and antipsychotics in the both comparisons of no discontinuation versus discontinuation and no restarters versus restarters. Interestingly, the no discontinuation subcohort also had a higher utilization rate of topical drugs and antipsychotics than the discontinuation subcohort. In the discontinuation subcohort, the no restarter group had a higher rate of antidepressants use than the restarter group. Perhaps the high prevalence of skin infection, depression, and UTI can explain the widespread use of corresponding drugs (topical drugs, antidepressants, antipsychotics and UTI drugs) observed during the wash-out period.

Persistence and Adherence of OAB/UI Treatment for a 6-Month Follow-up Period

Of 2496 subjects, 920 patients (36.9%) had no refill after the index prescription during the 6-month follow-up period. There were 1671 patients (66.9%) unable to remain on their treatment for longer than 90 days, and 1943 patients (77.8%) for longer than 150 days. The median time to discontinuation was 50 days with 25% and 75% quartile of 30 and 131 days, respectively (Fig. 1).


Figure 1. Kaplan–Meier survival curve of persistence of OAB/UI patients for 12 months.

Download figure to PowerPoint

Approximately, 4.8% (121 of 2496) of patients renewed their prescriptions on time or within the 30-day gap. On average, about one-third of the follow-up period was covered by OAB/UI medications (mean MPR = 0.35, SD 0.21), and 50% of patients received pharmacological treatment for less than 60 days cumulatively during the 6-month follow-up period (median MPR = 0.3, interquartile range = 0.17–0.49).

Sensitivity Analysis on Persistence and Adherence

Considering the fact that more than 20% of patients remained on drug therapy at the end of the 6-month follow-up period, we extended the follow-up period from 6 to 12 months to more carefully examine the degree of variation in persistence.

The comparison between the 6- and 12-month follow-up period is presented in Table 3. Eighty-one patients were additionally excluded in 1-year follow-up study because the eligibility of those patients was changed during the extended 6 months, resulting in 2415 patients remaining in 1-year follow-up study. In this analysis, the rate of discontinuation was increased to 88.6% (2139 of 2415 patients). Only 10% of patients continued their pharmacological treatment for longer than a year. The median time to discontinuation was 50 days and the quartiles of time to discontinuation were 30 days and 133 days, which were similar to the distribution in the 6-month follow-up study. The mean MPR, however, dropped significantly to 0.22 and the adherence rate (MPR > 0.8) was merely 0.7%.

Table 3.  Persistence and adherence patterns at 6- and 12-month of follow-up, respectively
 Following up for 6 months (N1 = 2496)Following up for 12 months (N2 = 2415)*
  • *

    There are a smaller number of patients for the 12-month follow-up because of the additional continuous enrollment required.

  • Rates of switching were determined for those receiving a medication on the index date and starting a different brand of medication later in the follow-up period.

Discontinuation n (%)1943 (77.8%)2139 (88.6%)
Switch n (%) 303 (12.1%) 474 (19.6%)
MPR (mean, SD)0.34 (0.21)0.22 (0.16)
Adherence n (%)122 (4.9%) 17 (0.7%)
Time to discontinuation (median, Q1, Q3)  50 (30,131)  50 (30,133)
Discontinuation at 30 days n (%)1066 (42.7%)1032 (42.7%)
Discontinuation at 90 days n (%)1671 (67.0%)1614 (66.8%)
Discontinuation at 150 days n (%)1943 (77.8%)1874 (77.6%)
Discontinuation at 270 days n (%)Not available2084 (86.3%)

To examine the variation of adherence rate due to different MPR cutoff points, we applied 0.5, 0.6, 0.7, and 0.9, respectively, as a cutoff point in addition to 0.8. Our findings suggest that, during 6-month follow-up period, adherence rates increased from 5% at 0.8 levels to 20% at 0.5 level, rose up to 14% at 0.6 levels and 8% at 0.7 levels, but decreased markedly from 5% at 0.8 levels to 2% at 0.9 levels. This sensitive change of adherence rate by different MPR cutoff points seems an obstacle to compare the results of OAB/UI studies on adherence using different MPR cutoff points.

Determinants of Persistence and Adherence

Figure 2 and Table 4 describe the final Cox regression model selected by the stepwise model selection method and proportional hazard assumption tests. The most notable finding from the Cox regression model is that the index OAB drug significantly affected persistence. Compared with oxybutynin (the reference group), the tolterodine group and the oxybutynin  ER  group  showed  a  favorable  result  for persistence. On the contrary, previous depression diagnosis, UTI diagnosis, and polypharmacy increased the hazard rate. Even though this is a retrospective study, the association between UTI diagnosis and persistence may suggest that a further clinical research is needed.


Figure 2. Cox regression on time to discontinuation during 6 months follow-up (reference index drug = oxybutynin).

Download figure to PowerPoint

Table 4.  Cox regression on persistence for 6-month follow-up period
 Hazard ratio (HR)P value95% CI of HR
  • *

    Reference index drug is oxybutynin.

  • UTI diagnosis in 30 days before OAB/UI diagnosis was considered as OAB/UI diagnosis and was excluded from this calculation to reflect clinical practice pattern.

  • CI, confidence interval; CCI, Charlson Comorbidity Index not including age.

Any previous OAB diagnosis1.100.140.97–1.26
Previous hospitalization length (in days)1.00<0.011.00–1.00
Index drug: tolterodine*0.74<0.010.67–0.81
Any previous topical drug use0.79<0.010.70–0.89
Multiple drugs used (polypharmacy)1.26<0.011.09–1.46
Any previous obesity diagnosis0.840.260.62–1.14
Any previous depression diagnosis1.200.051.00–1.44
Any previous UTI diagnosis1.160.011.04–1.30
CCI on index date0.900.110.80–1.02
Any previous OAB-related hospitalization × time to discontinuation1.030.580.92–1.17
Index drug: oxybutynin extended-release × time to discontinuation0.88<0.010.85–0.90
Index drug: other OAB drugs × time to discontinuation0.990.820.93–1.06
Any previous antipsychotics use × time to discontinuation0.85<0.010.83–0.88

Figure 3 and Table 5 present the result of logistic regression on adherence, where the stepwise model selection method was used to select covariates. Compared with oxybutynin, tolterodine and oxybutynin ER showed a higher rate of prescription renewal whereas the other OAB group was not different from oxybutynin. A previous OAB diagnosis showed a negative effect on prescription renewal, which may be resulted by an ineffective OAB drug therapy.


Figure 3. Logistic regression on adherence during 6 months of follow-up (reference index drug = oxybutynin).

Download figure to PowerPoint

Table 5.  logistic regression on adherence for 6-month follow-up period
 Odds ratio (OR)P value95% CI of OR
  • *

    Reference index drug is oxybutynin.

  • CI, confidence interval; CCI, Charlson Comorbidity Index not including age.

Any previous OAB diagnosis0.630.050.40–1.00
Previous all hospitalization length (in days)–1.01
Index drug: tolterodine*1.750.021.10–2.78
Index drug: oxybutynin extended-release*2.25<0.011.36–3.75
Any previous UTI drug use0.350.040.13–0.95
Any previous topical drug use1.370.090.95–1.99
CCI on index date1.260.240.85–1.86

Comorbidities Associated with Persistence

Urinary tract infection incidence was the only comorbidity that showed a significant relationship (< 0.5) with discontinuation of OAB drug therapy (Table 6; age square term was also included in the model to capture nonlinearity of age effect on UTI incidence). Considering other variables selected in the model by the stepwise method, discontinuation of OAB drug therapy increased the incidence of UTI by 37%. Hence, persistence played a positive role to reduce the incidence of UTI. Compared with oxybutynin, tolterodine showed a 24% lower incidence of UTI, whereas, oxybutynin ER and the other OAB drug group did not differ from oxybutynin.

Table 6.  The effect of OAB medication discontinuation on UTI incidence for 6-month follow-up period
 Odds ratio (OR)P-value95% CI of OR
  • *

    Reference index drug is oxybutynin.

  • More than one kind of medication used in follow-up period.

  • UTI diagnosis in 30 days before OAB/UI diagnosis was considered to be a misdiagnosis of OAB/UI diagnosis and was excluded from this calculation.

  • CCI, Charlson Comorbidity Index not including age.

Age squared1.000.921.00–1.00
Dual eligibility (Medicare + Medicaid)0.59<0.010.43–0.80
Any previous emergency room visit1.65<0.011.12–2.44
Any previous OAB-related outpatient visit0.830.150.64–1.07
Any previous OAB diagosis0.74<0.010.53–1.04
Index drug: tolterodine*0.760.020.60–0.96
Any previous antipsychotic use0.700.050.49–1.00
Multiple drugs used (polypharmacy)–1.69
Any previous UTI diagnosis5.70<0.014.45–7.31
Any previous UTI drug use1.360.110.93–1.99
CCI on index date1.16<0.011.09–1.24
Discontinuation of OAB drug therapy1.370.031.03–1.84


  1. Top of page
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. References

In the current study assessing long-term persistence and adherence of OAB/UI drug therapy, we observed a wide spectrum of persistence patterns and poor utilization of OAB/UI drugs. In general, there was a high discontinuation rate, although more than 10% of patients were persistent over 12 months, and a low adherence rate after the initiation of therapy. The present study documents persistence and adherence patterns among patients with chronic OAB/UI in an 1-year time frame, showing the negative impact of early discontinuation on comorbidities (namely UTI), and suggesting that, in the population defined for the analysis, long-term pharmacotherapy of OAB/UI has limited treatment success because of patients’ suboptimal medication filling behavior.

Overactive bladder/urinary incontinence is dramatically underdiagnosed by about 50% [2] mainly because patients reluctantly discuss their symptoms with physicians unless it becomes very severe. Hence, based on our patient identification criteria, our study cohort was focused on patients suffering from a chronic OAB/UI condition with above-average severity. The low persistence and adherence observed in the present study may suggest that compliance among less severe OAB/UI patients is even worse. The discontinuation rate of our data is higher than the rates reported in clinical trials of anti-OAB/UI drugs, in which an elderly population reported drop-out rates of about 8% because of adverse effects for tolterodine and oxybutynin ER [34–37], and one reported a withdrawal rate for tolterodine of 15% [38]; whereas the drop-out rate due to adverse effects for oxybutynin could be as high as 17% [34].

The study of persistence and adherence would benefit from patient-reported side effect data. Nevertheless, this was not possible using retrospective data. There is evidence suggesting that OAB patients who continue their medication have greater improvement in relevant symptoms and utilize fewer resources [18]. Nevertheless, patients are often unable to be compliant and persistent with their therapy because of side effects of OAB/UI drugs [22,39–41], Therefore, better management of adverse effect of OAB/UI drugs will help to improve patient's compliance with treatment.

Few studies have assessed the clinical benefit from persistence with OAB/UI drug therapy. The present study demonstrated the adverse effect of therapy discontinuation in that the risk of UTI was 37% higher in those that discontinued drug therapy. This further supports continuous therapy among chronic OAB/UI patients. Nevertheless, because Medi-Cal enrollees are not completely representative of the general population, additional studies in more general population are necessary to fully assess the impact of OAB/UI persistence on UTI incidence.

The gap between prescription fills was used to determine discontinuation. In our study, we used 30 days as the permissible gap between prescriptions without classifying a subject as discontinuing drug therapy. Some other persistence studies used different gap definitions, for example, 45 days [17,22,23]. There is one study defining the gap as a half of days’ supply of the previous prescription and changing the gap to 7 days if the prescription was for 14 days or fewer [16]. Another study defined the gap as 1.5 times the duration of the index dispensation, or 7 days, whichever was longer [42]. Nevertheless, there is no definitive evidence suggesting which gap length is most appropriate to assess persistence among OAB/UI patients.

The value ≥ 0.8 for MPR was used to decide if the patient was adherent or not in this study. Nevertheless, no study can demonstrate that 0.8 is a panacea. Because of the nature of clinical differences among various therapeutic classes, applying the same threshold value to determine patient adherence for different diseases or conditions may result in an over- or underestimation. Therefore, there is a clear need for further research to develop disease-specific MPR threshold values to achieve more accurate estimation of adherence rates.

Overactive bladder/urinary incontinence patients who discontinue treatment prematurely are unable to obtain the full benefit of therapy. Discontinuing beneficial medication can incur preventable morbidity and impose a considerable clinical and economic burden on the health-care system [43]. Establishing effective and efficient intervention programs to improve patient persistence and adherence has been placed on the agenda of improving health outcomes of treatment and patients’ quality of life. Patient education is significant to ensure maintenance of therapy; however, better patient monitoring by physicians and pharmacists to identify potential noncompliant populations is equally important. A comprehensive approach will be required for care of patients including monitoring of filled prescription, patient and clinician education, and the management of patient risk factor profiles.

The authors greatly appreciate Drs. Eric S. Rovner and David Ginsberg for sharing their clinical opinions on OAB/UI with the authors, however, all the remaining errors belong to the authors.

Source of financial support: No funding was received for this study and the authors have no conflicts of interest that are relevant to the contents of this article.


  1. Top of page
  3. Introduction
  4. Data and Methods
  5. Results
  6. Discussion
  7. References