Comparing clinical bladder diaries and recalled patient reports for measuring lower urinary tract symptoms in the Symptoms of Lower Urinary Tract Dysfunction Research Network (LURN)

Abstract

Purpose:

Bladder diaries are a key source of information about lower urinary tract symptoms (LUTS); however, many patients do not complete them as instructed. Questionnaire-based patient-reported outcome measures (PROMs) are another option for reporting LUTS but may have recall bias. We assessed the strength of the associations between PROMs and a 3-day bladder diary.

Materials and Methods:

Symptomatic adults from 6 tertiary care sites completed a 3-day paper bladder diary and 3-, 7-, and 30-day electronic PROMs. We assessed the linear associations between mapped pairs of diary variables and responses to PROM items using biserial and polyserial correlation coefficients with 95% confidence intervals.

Results:

Of 290 enrolled participants, 175 (60%) completed the bladder diary as instructed and at least one corresponding PROM. Linear associations were strongest between the diary and 3-day recall of daytime frequency (r=0.75) and nighttime frequency (r=0.69), followed by voids with urgency sensations (r=0.62), and an item reporting any incontinence (r=0.56). Linear associations between bladder diary and specific incontinence variables (e.g., stress, urgency) were low to negligible (ranging from r=0.16–0.39). Linear associations were consistent across the 3-day, 7-day, and 30-day recall periods.

Conclusions:

Missing and unusable bladder diary data were common, highlighting the patient burden associated with this method of data collection. A questionnaire-based PROM is a reasonable alternative to a diary for reporting voiding frequency and may offer an easier option for reporting some symptoms.

INTRODUCTION

To promote high-quality research and clinical care for patients with lower urinary tract symptoms (LUTS), it is important to understand the implications of adopting different ways of measuring LUTS. Urodynamic studies are used as objective measures of bladder volumes and pressures, but they cannot measure symptoms, are prone to artifacts due to nonphysiologic filling rates, and are costly to conduct. Professional society guidelines, including the American Urological Association (AUA)^1,2 and the International Continence Society (ICS),³ recommend obtaining information from patients using a bladder diary or questionnaire-based patient-reported outcome measures (PROMs).

Bladder diaries are cost-effective and are a key source of information for clinical research and practice. A typical bladder diary is presented in a grid format and includes voiding times, fluid intake and urine output volumes, types of beverages consumed, ratings of urgency, and urinary incontinence/pad changes. Numerous bladder diaries are available, many without formal evidence for validity.⁴ The ICS gives the International Consultation on Incontinence Questionnaire (ICIQ) 3-day bladder diary⁵ “Grade A” evidence for use with men and women with LUTS.⁶ The ideal duration of a diary balances the frequency and consistency of symptoms with patient burden. Bladder diaries are designed to be event-triggered, that is, completed in real time, so as to avoid recall bias. Patient burden is the main drawback of diaries; when patients do not complete them as instructed, it limits the accuracy and thus usefulness of the diary as an information source.⁷ In one study of new patients in a urogynecology practice, less than half of patients completed a 24-hour diary as instructed.⁸

Questionnaire-based PROMs are another option for recording LUTS. PROMs use self-reported structured questions to query patients about their symptoms and voiding function. They are important tools for efficiently understanding patients’ experiences and are commonly used in research to evaluate treatment outcomes.⁹ PROMs of LUTS are legion.¹⁰ PROMs typically include a recall period, over which the respondent is asked to think back about their experiences, e.g., “in the past 7 days”. A drawback of PROMs is concern about recall bias; indeed the choice of recall period for any PROM should consider the nature of the disease, including the frequency of symptoms and day-to-day variability in symptom experience.¹¹

Correlations between bladder diaries and PROMs have been published for women with incontinence,¹² men and women with overactive bladder,¹³ men with LUTS,¹⁴ and men and women with LUTS.^15,16 However, to our knowledge, no study has examined the correspondence of these two types of measures over the same reporting period. The Symptoms of Lower Urinary Tract Dysfunction Research Network (LURN)¹⁷ developed a suite of PROMs: the Comprehensive Assessment of Self-Reported Urinary Symptoms (CASUS),¹⁸ the LURN Symptom Index-29,¹⁹ and the LURN Symptom Index-10.²⁰ The LURN Recall Study compared 7- and 30-day recall bias²¹ and described patient characteristics associated with recall bias.²² The current study is the third pre-specified analysis from the LURN Recall Study. We assessed the associations between overlapping parameters from LURN PROM items and the ICIQ 3-day bladder diary. We hypothesized that the association between the diary and PROM items would be stronger for the 3-day recall comparisons versus the 7-day and 30-day, since the reporting period matched exactly. We explored the potential influence of recency on the 3-day recall responses, hypothesizing that the association between each individual day on the bladder diary and 3-day recall would be strongest for the 3^rd bladder diary day and progressively weaker for the 2^nd and 1^st days. Finally, we hypothesized that greater variability in bladder diary reporting over the 3 study days would be associated with lower correlations between diaries and PROMs.

METHODS

Study Design and Population

Adult participants were recruited at 6 tertiary care sites between May 2017 and April 2018. Participants were screened using the LUTS Tool, and those who endorsed at least one symptom as moderately severe and bothersome in the past 2 weeks and past 3 months were enrolled, after providing informed consent. The overall study design included both a within- and between-subjects design to support multiple research questions; the current study used the within-subjects design. Participants were randomly assigned in a 2:1:1 ratio to a study group (1, 2A, and 2B; Supplementary Figure 1). Groups 2A and 2B completed a 3-day bladder diary followed by 3-, 7-, and 30-day recall questionnaires on the 3^rd, 7^th, and 30^th days of the study period. Participants did not undergo any new treatments during the study month. Participants received $150 as compensation. To ensure that sufficient overlap existed between the questionnaire and diary, participants were excluded from analysis if their questionnaires were completed more than 1 day before or after the target date.

This study was approved by the Institutional Review Boards of each site. The LURN ClinicalTrials.gov Identifier is NCT02485808.

Measures

The baseline assessment included sociodemographic characteristics and all items from the CASUS¹⁸ using a 7-day recall period. The 3-day, 7-day, and 30-day assessments collected a subset of CASUS items (Supplementary Table 1).

A simplified paper version of the ICIQ 3-day diary captured voiding events, bladder sensation at voiding, incontinence episodes, and pad changes (Supplementary Figure 2). For each indicated void, the participant was also asked to rate their sensation of urgency and indicate if they leaked (and to write in the type of leak) in connection with the void. Voided volumes and fluid intake information were not collected. Details regarding quality control of data are in the Supplementary Text.

Statistical Analysis

Diary variables were mapped a priori to PROM items (Supplementary Table 1). Group comparisons were made using chi-square and Wilcoxon rank sum tests. The diaries were summarized by averaging values across the days reported, except for nocturia, where number of nights with at least one void was used. Linear associations between mapped pairs of PROM items and diary variables were assessed using biserial and polyserial correlation coefficients with 95% confidence intervals (CIs). These correlations were calculated separately for 3-day, 7-day, and 30-day recall questionnaire items to investigate differences in linear associations by recall period. Except for daytime frequency items, participants were excluded from the calculation if they answered “Never” on the corresponding PROM item. Details regarding analyses for effects of recency, symptom variability, sex, and study group are in the Supplementary Text.

Within each set of comparisons or correlations, the false discovery rate (FDR)²³ was used to control for Type I errors resulting from multiple comparisons. Measures with correlation coefficients within the absolute values of 0.0 to 0.3 were considered not associated, 0.3 to 0.5 low association, 0.5 to 0.7 moderate association, and 0.7 and above high association. Analysis of correlations between groups were performed in R 3.5.2²⁴ using the cocor²⁵ package. All other analyses were performed using SAS software, Version 9.4 (SAS Institute Inc., 2013, Cary, NC).

RESULTS

Sample Description

Of 290 participants, 56 (19%) did not return a bladder diary, 3 (1%) diaries did not meet data quality standards, and another 56 participants (19%) were dropped due to insufficient overlap between the diary dates and the recalled periods of the 3-day recall and 7-day questionnaire items (Figure 1). Among 175 participants eligible for analysis, 83 (47%) were male and 92 (53%) were female, 87% were white, and the average age was 59 years (Table 1). Only 14% of participants reported prior LUTS surgeries. There were no statistically significant differences in demographics or LUTS reported at enrollment between groups or between participants not returning diaries or meeting other quality standards, and those who did (data not shown).

Figure 1. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) diagram.

The diagram shows participant flow for this analysis and begins in topmost rectangle with total participants consented, shows reasons why bladder diaries did not pass quality control or why data was excluded from analysis, and ends with the analyzable sample used in the bottommost rectangle.

Table 1.

Patient demographics and clinical characteristics, overall and by study group

Variable	Total (n=175)	Group 2A (n=78)	Group 2B (n=97)	p-value
Age	58.7 (14.8)	59 (14.3)	58.4 (15.3)	0.937
Sex (% male)	83 (47%)	36 (46%)	47 (48%)	0.762
Race				0.332
Black/African-American	14 (8%)	8 (10%)	6 (6%)
White	153 (87%)	65 (83%)	88 (91%)
Multi-racial/other	8 (5%)	5 (6%)	3 (3%)
Hispanic/Latino	2 (1%)	2 (3%)	0 (0%)	0.040
Marital status				0.399
Married/civil union	111 (63%)	53 (68%)	58 (60%)
Living with a partner	9 (5%)	3 (4%)	6 (6%)
Separated or divorced	25 (14%)	13 (17%)	12 (12%)
Widowed	10 (6%)	3 (4%)	7 (7%)
Single, never married	20 (11%)	6 (8%)	14 (14%)
Education				0.415
Less than HS diploma/GED	2 (1%)	0 (0%)	2 (2%)
HS diploma/GED	12 (7%)	5 (6%)	7 (7%)
Some college or tech school, no degree	37 (21%)	19 (24%)	18 (19%)
Associate’s degree	15 (9%)	9 (12%)	6 (6%)
Bachelor’s degree	49 (28%)	18 (23%)	31 (32%)
Graduate degree	60 (34%)	27 (35%)	33 (34%)
Employment status				0.997
Employed part-time	13 (7%)	6 (8%)	7 (7%)
Employed full-time	69 (39%)	30 (38%)	39 (40%)
Unemployed (looking for work)	6 (3%)	3 (4%)	3 (3%)
Not employed (not looking for work, includes stay-at-home, retired)	83 (47%)	37 (47%)	46 (47%)
Body mass index	29.7 (7.2)	29.6 (5.7)	29.8 (8.4)	0.419
Previous surgical treatment	24 (14%)	11 (14%)	13 (13%)	0.894
Symptoms endorsed at enrollment
Daytime frequency	59 (34%)	26 (33%)	33 (34%)	0.924
Nocturia	94 (54%)	40 (51%)	54 (56%)	0.563
Incontinence	94 (54%)	43 (55%)	51 (53%)	0.737
Urgency	124 (71%)	56 (72%)	68 (70%)	0.807
Weak stream	77 (44%)	36 (46%)	41 (42%)	0.607
Incomplete emptying	96 (55%)	45 (58%)	51 (53%)	0.499
Post-micturition dribble	99 (57%)	48 (62%)	51 (53%)	0.235

On the bladder diaries, participants reported a mean ± standard deviation of 9 ± 3 daytime voids, 1 ± 1 nighttime voids, 4 ± 3 daytime voids with urgency sensations, and 1 ± 2 daytime incontinence episodes (all types). The majority (>62%) of average number of daytime voids reported on the diary fell within the range of response values on the 3-day recall PROM item for all responses (Figure 2) except for the lowest response option (n=5, 100% were out of range). That is, the very few patients who responded “1–3 times a day” on the PROM item tended to have average daily voids of 5 times a day or more based on diary reports. Within each response category, average daytime voids between the three recall periods were similar (FDR p=0.935). Similar trends were observed for nocturia (Supplementary Figure 3), though it was the PROM response “2–3 times” that appeared most dissimilar to what was reported on the bladder diary, and in this case, it was not due to small sample sizes.

Figure 2. Boxplots of the distribution of average daytime voids on the bladder diary by PROM recall period within each response to the daytime frequency LURN PROM item.

The boxplots are grouped by the different recall periods for each response level to the LURN item “During waking hours, how many times did you typically urinate?”. Blue, red, and green boxes represent the distributions for 3-day, 7-day, and 30-day recall LURN items, respectively. Within each box, the horizontal line represents the median, and the circle represents the mean. Lines at the bottom and top of each box represent the 25^th and 75^th percentiles, respectively. The whiskers at the bottom and top of each plot extend to observations closest to 1.5 times the interquartile range (IQR), and dots represent observations outside of this range. Shaded blue boxes cover the range of values represented by each survey response.

Linear Associations

Magnitudes of correlations by LUTS type

For the 3-day recall questionnaire, which most closely overlapped with the diary recording period, linear associations were strongest for recall of daytime frequency (correlation of 0.75) and nighttime frequency (0.69) (Figure 3, Supplementary Table 2). The linear associations for voids with urgency sensations were moderate, with a correlation of 0.62 for 3-day recall. Linear associations between incontinence variables and associated PROM items were moderate for the overall item asking about any incontinence (0.56), low for two of the specific stress incontinence items (“laughing, sneezing, coughing” at 0.35 and “physical activities” at 0.39) and not associated for the other specific stress incontinence items, the urgency incontinence item, and the other/unknown incontinence items (ranging from 0.16–0.28). Note that, for many of these individual incontinence items, there were fewer than 25 participants in the analysis, reflected in the wide CIs.

Figure 3. Forest plot of biserial and polyserial correlation coefficients calculated between LURN PROM items and bladder diary variables by PROM recall period.

LURN items are shown on the left side of the y-axis, and the associated bladder diary variable (averaged over the whole diary) are shown on the right side of the y-axis. Within each pair of LURN item and bladder diary variable, squares, triangles, and circles represent the correlation coefficients between the bladder diary variable and the 3-day, 7-day, and 30-day LURN item, respectively. The horizontal blue bar represents the 95% CI for that correlation. Sample size is shown at the left side of the figure, and vertical red lines are included for reference at 0.0, 0.3, 0.5, and 0.7.

Similarities across different recall periods

Linear associations were largely consistent across the 3-day, 7-day, and 30-day recall periods (Figure 3, Supplementary Table 2, Supplementary Text).

Effects of recency, symptom variability, sex, and study group

Overall, there was no consistent effect of recency (Supplementary Table 3, Supplementary Text). There was an effect of symptom variability: compared to those with high variability, those with low variability had higher correlations between diary and PROM for frequency variables (Figure 4, Supplementary Text). Relationships were generally consistent by sex and across study groups (Supplementary Tables 4 and 5, Supplementary Text).

Figure 4. Forest plot of biserial and polyserial correlation coefficients between 3-day recall LURN PROM items and bladder diary variables by symptom variability groups.

The 3-day recall LURN items are shown on the left side of the y-axis, and the associated bladder diary variables (averaged over the whole diary) are shown on the right side of the y-axis. Within each pair of 3-day recall LURN item and bladder diary variable, squares, triangles, and circles represent the correlation coefficients between the bladder diary variable and the 3-day recall LURN item for the low/no, medium, and high/any variability groups, respectively. The horizontal blue bar represents the 95% CI for that correlation. Sample size is shown at the left side of the figure, and vertical red lines are included for reference at 0.0, 0.3, 0.5, and 0.7. Panel A shows correlations for LURN items about voids, and panel B shows correlations for LURN items about leaks.

DISCUSSION

In this multi-site study of tertiary care-seeking men and women with LUTS, one of the most notable findings concerned the rate of missing and/or unusable bladder diary data—a problem that has been previously documented in clinical care (50% unusable)⁸ and research (50% unusable).¹⁶ Examining this again in a research context in which participants received an incentive for participation, we found that 40% of the participants had missing and/or unusable bladder diary data, including 20% who did not return a useable diary and another 20% who did not complete the diary during the correct time period to correspond with the PROMs. These high rates in both clinical and research settings significantly undermine the informativeness of bladder diaries.

Among those patients with usable bladder diary data, when comparing LUTS recalled over multiple days on a PROM compared to reports over the same time period from real-time paper bladder diaries, concordance varied depending on the symptom. Recall of voiding habits (i.e., daytime and nighttime frequency) was more similar to bladder diary reports than recall of urgency and incontinence symptoms, as has been seen in other studies.¹² Of the LUTS that we studied, recall of daytime frequency on the PROMs was the only one that was consistently correlated >0.7 (“high”) with the corresponding bladder diary variable. That said, there was a striking mismatch for the lowest PROM response option for daytime frequency (“1–3 times per day”) and the reports from the bladder diary. The sample size for this analysis was very small (4, 5, or 6 respondents depending on the recall period, 8 different people in total), yet the pattern was consistent across the different recall periods. One hypothesis is that these people misinterpreted the PROM question, which refers to voids “during waking hours” and did not include voids first thing in the morning or directly before sleeping.

There are several factors that might explain the relative lack of concordance between the bladder diary and PROM data on urgency and incontinence symptoms. First, the two measures differ in terms of the reporting period, in that patients are supposed to record symptoms at the time that they void with the bladder diary, whereas the PROM items require patients to recall and summarize over a multi-day period. Thus, lower concordance could be due to recall error in the PROM. However, it is noteworthy that, in our previous study comparing 7- and 30-day recalled PROM responses to the average of 7 or 30 daily (end-of-day) responses, we found that, for the same urgency and incontinence symptoms assessed using consistent wording, the correlations ranged from 0.73 to 0.95 for both 7- and 30-day recall—much higher correlations than those observed in this study, which ranged from −0.12 to 0.62. This suggests that, while recall error might be responsible for some of the disagreement between methods, other aspects, such as wording, format, and/or context, might be more influential.²⁶ Whereas reporting voiding habits is a relatively straightforward task of counting voids, describing symptoms on the bladder diary involves tying them to a particular void and/or writing in the type of incontinence. This is potentially problematic for the concepts of urgency and incontinence, which are frequently experienced not in relation to a particular void. Reporting urgency symptoms on the PROM is more general (e.g., “how often did you feel a sudden need to urinate?”), while reporting incontinence symptoms on the PROM ties the frequency of the symptom to a particular activity (e.g., “how often did you leak urine or wet a pad while laughing, sneezing, or coughing?”).

Another potential reason for mismatch between diary and PROM is symptom variability. Our study showed that participants with greater symptom variability had smaller linear associations between their averaged responses on the bladder diary and their 3-day recalled response on the PROM. This was statistically significant for daytime frequency, with a high correlation between diary and PROM for those with low variability across the 3-day reporting period but a moderate association for those with high variability. This finding might be because people with relatively stable frequencies can more easily summarize their experience by estimating the near-constant rate of occurrences, whereas this is more difficult for people with more variable frequencies.

To most closely approximate how PROMs and bladder diaries are used in the real world, we chose an electronic PROM and a paper bladder diary. Thus, another potential reason for some of the lack of congruence between the bladder diary and PROM for urgency and incontinence is this difference in administration method. However, there are two reasons why this factor is an unlikely cause of differences. First, if there were a broad effect of mode of administration, one would expect lower agreement for the voiding habits as well as the symptoms, and this was not the case. Second, a growing body of work supports the view that paper versus electronic administration of the same items does not result in meaningful differences.³

A limitation of this study was the sample of generally well-educated, predominantly white, English-speaking, sub-specialty care-seeking participants, which does not represent all people with LUTS. Missing data from the bladder diaries limited our sample size, though we did not see evidence that those with missing/unusable diary data were systematically different from those who provided usable diaries.

CONCLUSIONS

Our findings suggest recommendations for the use of bladder diaries and PROMs in research and clinical practice. The very low correspondence between the two methods, especially for urinary incontinence, suggests PROM and bladder diary responses should not be taken as interchangeable. Bladder diaries are the better option for tracking fluid intake and voiding volumes. For the ~50% of people who will complete the diary correctly, a 3-day diary records urinary frequency, variability, and timing of voids. (A 24-hour diary has less burden but would not show variability.) On the other hand, bladder diaries have high missing/unusable data, and provide less detail regarding urgency and incontinence symptoms. PROMs offer a reasonable alternative for reporting voiding frequencies and may offer an easier option for reporting symptoms.

Data Availability Statement

The data that support the findings of this study are openly available in the NIDDK Central Repository at https://repository.niddk.nih.gov/; please reference the acronym “LURN”.