Limitations

This study has some limitations: 1) the cross-sectional study design limits conclusions about causality; 2) results from the Finnish population may not be directly generalisable to other ethnicities, 3) although the response rate was high, approximately one third of those contacted did not participate in the study; 4) impact of self-reporting on validity of the results remains unclear, and 5) identifying every

5.7.1 Study design and study population

This dissertation is based on the baseline data (cross-sectional) of the FINNO Study.

The cross-sectional study design limits conclusions about causality. It is difficult, if not impossible, to separate cause from effect, as measurements of exposures and outcomes were conducted at the same time (except when their occurrence times could be established retrospectively). Furthermore, subject´s exposure status at the time of data collection may have had little to do with exposure status at the time the outcome first occurred. In future, the first survey of the FINNO Study will be the baseline of a prospective cohort study. However, as LUTS (including nocturia) are dynamic, symptoms fluctuate over time (Moller et al. 2000a, Häkkinen et al. 2006, Wennberg et al. 2009). This is likely to increase misclassification and bias any associations towards unity. Defining an incident case is a major methodological challenge for prospective studies of LUTS.

Descriptive results from the Finnish population may not be directly generalisable to other ethnicities. There may (Fitzgerald et al. 2007, Kupelian et al. 2009) or may not be (Liew et al. 2006) ethnic differences in the prevalence of nocturia. However, the potential differences may be largely explained by differences in socioeconomic status (Kupelian et al. 2009), while the direct impact of ethnicity remains unknown.

5.7.2 Self-reporting

We used postal questionnaires to explore the prevalence, impact and risk factors of urinary symptoms (especially nocturia and urinary urgency). Both nocturia and OAB are symptom-defined conditions requiring self-report (discussion regarding questionnaires and FVCs later). The validity of self-reported information is largely unclear. The agreement between questionnaire data and medical records has been shown to be good for major chronic diseases in Finland (Haapanen 1997). Patients´

self-reports have been shown to be reliable when estimating recent use of cardiovascular and diabetic drugs (Glintborg et al. 2007). Mailed questionnaires reflect urodynamics better than interview-assisted questionnaire responses (Khan et al. 2004).

This may be due to feeling and not feeling embarrassed in interview-assisted and

self-reliable information than telephone surveys in several aspects, including higher participation (Armstrong et al. 1992).

5.7.3 Limitations related to potential risk or confounding factors

Anthropometric factors. BMI was used as an indicator of obesity. Although the correlation between BMI and body fat adjusted for age is high (r = 0.82–0.91) (Spiegelman et al. 1992), it is not possible to distinguish lean body mass from fat on the basis of BMI. For example, the proportion of body fat is higher among women than among men with a similar BMI. In addition, body fatness has been shown to increase with age; that is, similar BMIs may correspond to a greater body fat content in older subjects compared to younger subjects (Ross et al. 1994, Gallagher et al. 1996).

Despite these limitations, BMI is a simple and useful measure of obesity in adults. A BMI of 30 is a widely recognised cutoff point for obesity, and the cutoff point for overweight (BMI 25) is recommended by the World Health Organization (World Health Organization 1995). However, waist circumference may be a better indicator of obesity. However, to the best of our knowledge, there is still a paucity of studies on this topic in general, and especially regarding LUTS (including nocturia).

Lifestyle factors. Our study was limited by fact that 14% of men and 24% of women did not report alcohol consumption. However, the prevalence of nocturia among subjects who provided alcohol consumption information did not differ from those who did not. Furthermore, reported alcohol consumption (daily mean consumption of absolute alcohol: 12 grams for men and 6 grams for women) corresponded very well with the statistics of the National Public Health Institute (Laatikainen et al. 2003). We did not have information on physical activity. In earlier reports, more physically active subjects had lower frequency of LUTS (Platz et al. 1998, Orsini et al. 2006, Litman et al. 2007), but not specifically nocturia (Schatzl et al. 2000). However, in a small uncontrolled study, significant improvement in nocturia (mean of nocturia decreased from 3.3 voids to 1.9 voids per night, p < 0.001) was achieved by a 8-week walking

training programme (including pelvic floor muscle training) among incontinent, elderly women (Johnson et al. 2005a).

Reproductive factors. There are also potential limitations in the assessment of reproductive factors (Study V). We could not distinguish between delivery modes. This may be a minor limitation, given the little convincing evidence of its importance. No studies have examined the association between delivery mode and nocturia. Four earlier studies demonstrated no effect of delivery mode on prevalence of urgency, urgency incontinence or OAB (Parazzini et al. 2003, Handa et al. 2004, Zhang et al.

2006, van Brummen et al. 2007). By contrast, a study among customers of the Kaiser-Permanente health insurance plan found that those delivering vaginally were more likely to have OAB than those having Caesarean section (adjusted for parity, obesity and age) (Lukacz et al. 2006). The study was somewhat limited because of low participation (37%) and age differences between the groups (women delivering vaginally were older than those with Caesarean section). Hence, delivery mode seems not to be a risk factor of OAB in the earlier literature generally.

Another potential limitation is the lack of information on different surgical hysterectomy procedures and the indication for hysterectomy. This may not be an important limitation, as RCTs found no difference in prevalence of nocturia or urgency post-operatively between subtotal and total hysterectomies for benign indication with one (Thakar et al. 2002, Gimbel et al. 2005) or two-year (Learman et al. 2003) follow-up. Furthermore, reliability (Horwitz & Yu 1985) and validity (Brett & Madans 1994) of hysterectomy self-reporting has been established only for hysterectomy itself, not for the various techniques separately. Since post-operative nocturia and urgency RCTs comparing vaginal hysterectomy to other techniques are not available, current evidence suggests that surgical technique is unlikely to influence the effect (Roovers et al. 2001, Thakar & Sultan 2005). Additionally, no information was obtained in this study regarding type of SUI surgery. A Cochrane review found no difference in de novo detrusor overactivity after laparoscopic colposuspension compared to tension-free vaginal tape (Dean et al. 2006). RCTs reported no difference in de novo urgency incontinence after fascial sling compared to Burch colposuspension (Albo et al. 2007), or in de novo urgency after suburethral sling procedure by retropubic compared to

We had no laboratory (serum levels of follicle stimulating hormone) or other diagnostic studies to confirm menopausal status, which was based on reported menstruation during the past year. Those reporting hysterectomy or MHT were regarded as separate groups as their natural menopausal status remains unclear. We used age 40 years as a cut-off point for menopause, as amenorrhea is due to other reasons than menopause in the vast majority of women aged under 40 years (Nippita & Baber 2007). In our study, among women aged under 40 years, 46 (4%) reported no menstruation in the past year.

Overall, we had abundant information on female reproductive factors compared to the earlier literature. However, further information on some aspects would have strengthened the study.