Discussion

4.1 The interpretation of our database

The NHIRD is nationwide anonymous eligibility, enrollment information, and claims for visits, procedures, and prescription medications of 99% of the entire estimated 23 million population of Taiwan (Hsiao et al, 2007). Our study used the LHID that contained all outpatient and inpatient medical claims of 1 million beneficiaries randomly sampled from the NHIRD during the period of January 1, 1999 to December 31, 2013. We identify a cohort based on health services, diagnoses, surgical procedures, and drug utilization from this database. As a result, we believe that our data are reliable, and the patients’ characteristics are similar to those of the whole population.

4.2 Summarize main results

To the best of our knowledge, this is the first study to report the incidence of significant VD (1.42%) in women with type 2 diabetes over a 13-year period. Our study involving only women with type 2 diabetes decreases the confounding effect of concurrent benign prostate enlargement, which shares the same VD. However, information regarding the incidence of VD in women with DM is extremely limited. In hospital-based studies, only the prevalence of VD in women with DM reported in UK and China are 38 % and 55.9%, individually (Fayyad et al, 2009; Changxiao H et al, 2014).. In our community-based study, the incidence of significant VD in women with type 2 diabetes is 1.42%. Although the higher prevalence of VD in women with DM was reported in previous studies, our results showed the lower incidence of significant VD in women with type 2 diabetes. This difference may be present because our patients may only have visited the hospitals when their conditions were already serious. In this

study, we found that increased risks of significant VD in women were associated with age of more than 50 years, CCI score of more than 0, and comorbidities of CVA and PD, but comorbidities of SCI and history of hysterectomy were not associated with increased risk of significant VD. No event of significant VD in SCI patients may be a reflection of more serious VD with intermittent catheterization program or Foley catheter insertion. Only three events of significant VD were observed in patients with history of hysterectomy, which may reflect advances in surgical techniques and management options (Kietpeerakool et al, 2019) After adjusting the effects of age, CCI score, CVA, and PD, the risk of significant VD remained higher in the DM cohort than in the non-DM cohort. Our finding showed a 1.79-fold increased risk of significant VD in women with type 2 diabetes after adjusting for these confounding factors. The incidence rates of significant VD in women was 20.0 and 10.5 per 10,000 person-years in the DM and non-DM cohorts, respectively. Although after matching, the incidence of significant VD in women with type 2 diabetes was underestimated and decreased to 1.28%, the relationship between type 2 diabetes and significant VD in women is still established suggesting that it is highly probable that type 2 diabetes is an independent medical condition that causes significant VD in women.

Initial studies suggested that long standing diabetes mellitus causes paralysis of the detrusor muscle leading to VD (Hill et al, 2008). Our cohort study first indicated that

diabetic cystopathy with VD in women was reported in at least 8-9 years after the diagnosis of diabetes mellitus (Kebapci et al, 2007).. This difference may be due to detrusor underactivity, diabetic urethropathy, or poor health-habit counseling among women in Taiwan, as the time to occurrence of significant VD in women with type 2 diabetes is lower in our study than in the urodynamic study (Yang et al, 2007; Lee et al,2009).

4.3 The interpretation of results of subgroup analysis

Studies have shown that elderly women with CVA, PD, SCI, or history of hysterectomy complained of voiding difficulty, such as small caliber or urinary retention (Sveinbjornsdottir S, 2016 ; Quadri et al, 2018; Akkoç et al, 2019;

Kietpeerakool et al, 2019).In the earlier study, detrusor-impaired contractility with VD is common in elderly women older than 70 years old (Abarbanel et al, 2007). Our subgroup analyses showed that women with type 2 diabetes have an age-dependent increasing incidence rate of significant VD, but the risk of significant VD was higher in the patients aged 20-40 years, which may reflect the influence of childbirth, detrusor underactivity, or diabetic urethropathy (Yang et al, 2007; Lee et al,2009; Beaumont T, 2019).. In addition, a possible misclassification bias cannot be avoided because patients with type 1 diabetes cannot totally be excluded by ICD-9-CM codes 250.X1, especially patients aged 20-40 years. The small sample size of women with type 2 diabetes aged 20-40 years with significant VD is another bias with the effects of inflated false discovery rate. Therefore, it is more believable that the occurrence of significant VD in women with type 2 diabetes aged 41-50 years was 2.66-fold higher. The CCI-dependent increasing incidence rate of significant VD in women with type 2 diabetes was also found, but lower CCI score had a greater magnitude of the risk of significant VD and

the occurrence of significant VD in women with type 2 diabetes with a CCI score of 0 was 2.52-fold higher. This suggests that the urological and medical care of women with type 2 diabetes in Taiwan should be more aggressive. It is highly probable that type 2 diabetes is an independent medical condition that causes significant VD in women, which could impact the comprehensive care of urinary symptoms of patients with type 2 diabetes through effective diabetes mellitus therapy (Tai et al, 2016). However, because the severity of type 2 diabetes cannot be determined through our national databases, this hypothesis could not be validated. We recommend that additional prospective studies are conducted to verify the age- and CCI-related elevated risk of significant VD in women with type 2 diabetes. Owing to the effect of type 2 diabetes, the risk of significant VD in patients with CVA was 1.80-fold higher, which was similar to that in patients without CVA. On the contrary, the risk of significant VD in women with type 2 diabetes who have PD was not significantly different. In short, women with PD are highly associated with significant VD compared to those with CVA.

4.4 Summarize the trend of new users of

Current drugs, such as bethanechol or tamsulosin, have been used for the treatment of VD in women since 2000. Bethanechol, a cholinergic agent produces the effects of the parasympathetic nervous system stimulation. Bethanechol is usually prescribed in elderly women for detrusor atony, urinary retention, or incomplete bladder emptying (Gaitonde S, 2018). Tamsulosin, an alpha-1-adrenergic receptor (a1-AR) blocker is effectively used to improve voiding symptoms in those with benign prostatic hyperplasia (BPH) (Roehrborn et al, 2004).Tamsulosin is also used for the treatment of VD in women (Chang et al, 2008).. In Taiwan, bethanechol has been prescribed to treat

VD in women for a long time, and tamsulosin for treating VD in women has been the most commonly used alpha blocker. Other alpha blockers such as doxazosin or terazosin may be used for treating VD in women or hypertension. To eliminate the confounding effects, we only used the prescription of bethanechol or tamsulosin as the occurrence of significant VD. Owing to the occurrence of significant VD with seeking treatment, we believe that the incidence of significant VD among women with type 2 diabetes in our study is more reliable than those reported in a previous study by questionnaire or urodynamic studies (Fayyad et al, 2009; Changxiao H et al, 2014).

Given that doctors may use bethanechol to treat constipation (Poetter et al, 2013) or tamsulosin to shorten the passage time of smaller ureteral stones (Abdel-Meguid et al, 2010), we exclude patients receiving bethanechol or tamsulosin therapy within 3 months before the index date. In this study, we found that the number of new users of bethanechol and tamsulosin increased by 2.04 and 8.25 times, respectively, between 2005 and 2013. The number of new users of combination therapy of bethanechol and tamsulosin increased by 3.0 times between 2009 and 2013. We believe that many doctors know that tamsulosin can be used to treat women VD. We also find a gradually increasing trend of the use of combination therapy of bethanechol and tamsulosin since 2007.

4.5 Limitations

This study had limitations inherited from the NHIRD. First, ICD-9-CM codes were used to identify cohorts in the NHIRD. All insurance claims in the NHIRD were made by medical doctors according to the standard criteria, but this study may not have the same quality as that of a prospective well-designed study. Second, the NHIRD did not contain personal information regarding laboratory data, record of alcohol and cigarette

use, and exercise that may be confounding variables influencing voiding function.

Therefore, we used a propensity score to match age, CCI score, and comorbidities to minimize this potential influence. Third, the diagnoses of diabetes mellitus and comorbidities were completely dependent on the ICD-9-CM codes, especially diabetes mellitus. We adopted the diabetes mellitus diagnosis as at least two outpatient visits or one hospitalization that was already validated (Lin et al,2017 ). Fourth, type 1 diabetes was present in less than 1% of the diabetic population in Taiwan (Jiang et al, 2012). We excluded the patients diagnosed with type 1 diabetes (ICD-9-CM codes 250.X1), but a possible misclassification bias cannot totally be avoided. Fifth, we only analyzed tamsulosin use for significant VD in women with type 2 diabetes, rather than including all alpha blockers. Therefore, the incidence of significant VD in women with T2DM might be mildly underestimated.