Abstract
Background:
To investigate whether meteorological factors (temperature, barometric pressure, relative humidity, ultraviolet index [UVI], and seasons) trigger flares in male and female urologic chronic pelvic pain patients.
Methods:
We assessed flare status every two weeks in our case-crossover study of flare triggers in the Multidisciplinary Approach to the Study of Chronic Pelvic Pain one-year longitudinal study. Flare symptoms, flare start date, and exposures in the three days preceding a flare or the date of questionnaire completion were assessed for the first three flares and at three randomly selected non-flare times. We linked these data to daily temperature, barometric pressure, relative humidity, and UVI values by participants’ first 3 zip code digits. Values in the three days before and the day of a flare, as well as changes in these values, were compared to non-flare values by conditional logistic regression. Differences in flare rates by astronomical and growing seasons were investigated by Poisson regression in the full study population.
Results:
574 flare and 792 non-flare assessments (290 participants) were included in the case-crossover analysis, and 966 flare and 5,389 non-flare (409 participants) were included in the full study analysis. Overall, no statistically significant associations were observed for daily weather, no patterns of associations were observed for weather changes, and no differences in flare rates were observed by season.
Conclusions:
We found minimal evidence to suggest that weather triggers flares, although we cannot rule out the possibility that a small subset of patients are susceptible.
Keywords: Chronic prostatitis, chronic pelvic pain syndrome, interstitial cystitis, bladder pain syndrome, flare, trigger
INTRODUCTION
Chronic prostatitis/chronic pelvic pain syndrome (CP/CPPS) and interstitial cystitis/bladder pain syndrome (IC/BPS) are chronic idiopathic conditions collectively referred to as urologic chronic pelvic pain syndrome (UCPPS). Both of these conditions are characterized by persistent bladder or pelvic pain and urologic symptoms, such as urinary urgency and frequency. There is no cure for CP/CPPS or IC/BPS, and both are difficult to diagnose and treat, resulting in a large physical, mental, and economic burden for patients.1,2
The natural history of UCPPS is widely recognized to include periods of remission and exacerbation, which are frequently termed “flares.”3 Although the etiology of these common and often debilitating symptom exacerbations is poorly understood, a number of triggers have been proposed, including several related to weather. In Scandinavia and the US, CP/CPPS patients4–7 and male8 and female IC/BPS patients8–10 have reported that cold, damp, or windy weather; long periods of rain; wintertime; temperature or air pressure drops; and changes in season or weather tend to worsen their symptoms and/or trigger flares. Data from two cross-sectional surveys of Asian men are also supportive of possible roles for colder weather and limited sunlight. The first of these surveys observed a positive association for lower average daily temperature and sunlight duration with the presence of CP/CPPS symptoms,11 and the second observed positive associations for residence at higher altitude (where temperature, barometric pressure, and oxygen levels are lower) with both the presence and intensity of symptoms.12 Finally, in the only longitudinal study to date, fewer breakthrough episodes were observed in the summer than in other seasons in Asian CP/CPPS patients.13
Although findings from this small body of literature are supportive of a role for weather in UCPPS symptom intensity and flares, most of this evidence was derived from patient reports or cross-sectional surveys of men. Very few studies have included women and only one has investigated the possible association between weather and UCPPS flares prospectively. Therefore, we took advantage of data collected in our case-crossover study of flare triggers in the Multidisciplinary Approach to the Study of Chronic Pelvic Pain (MAPP) longitudinal study to investigate possible associations between a range of meteorological factors and flares in male and female UCPPS patients.
METHODS
Study design and population
The MAPP Epidemiology and Phenotyping Study was a one-year longitudinal study of UCPPS patients (IC/BPS and CP/CPPS in men, and IC/BPS in women) designed to better understand the “usual-care” natural history of UCPPS and to identify sub-groups with possibly differing pathophysiology, clinical course, and, ultimately, response to therapy. MAPP participants were recruited at six main clinical sites (Los Angeles, Seattle, Ann Arbor, Chicago, Iowa City, and St. Louis [with one participant residing in Atlanta]) and two sub-sites (San Francisco and Miami) from 11/2009 to 12/2012. Participation involved: 1) attending three in-person visits at baseline, 6 months, and 12 months follow-up, consisting of completion of numerous questionnaires and provision of biologic specimens; as well as 2) completion of bi-weekly online assessments consisting of a brief series of questionnaires.14
To better understand flare etiology, we embedded a case-crossover study of flare triggers into the MAPP longitudinal study. As part of this study, participants were asked to report their current flare status (“Are you currently experiencing a flare of your urologic or pelvic pain symptoms. By this we mean, […] symptoms that are much worse than usual”) at each in-person visit and biweekly assessment. Those who responded affirmatively were directed to a second questionnaire, the Brief Flare Risk Factor Questionnaire, to collect additional information about their flare start date, current flare symptoms, exposures in the three days preceding their flare, and suspected triggers for their current flare (diet, physical activity, sexual activity, infections, stress, other, and don’t know/not sure). This questionnaire was administered for the first three flares reported post-baseline, as well as at three randomly selected non-flare times (one per study third).14,15
Ascertainment and definition of meteorological variables
Mean daily temperature, relative humidity, and barometric pressure data were obtained from the National Oceanic and Atmospheric Administration, and ultraviolet index (UVI) data were obtained from the National Aeronautics and Space Administration Ozone Monitoring Instrument. These data were linked to flare and non-flare assessments by the date of flare onset (flares) or the date of questionnaire completion (non-flares), and by the first 3 digits of participants’ residential zip code. The first 3, rather than 5, digits were used to protect participants’ anonymity. Data were linked using the nearest weather station for temperature, relative humidity, and barometric pressure, and using spatial linking by 1° longitude by 1° latitude grid-square for UVI. Missing values for temperature, relative humidity, and barometric pressure (0.1–1.0%) were imputed by substituting values collected on the same day from the nearest weather station. Missing UVI data (30.4–36.3%) were not imputed because whole regions tended to be missing on the same day. This pattern of missing, which was likely due to seasonality or missing underlying satellite modeling components (e.g., surface albedo, solar zenith angle), precluded imputation.
To account for the fact that the “empirical induction period”16 of flares (i.e., the length of time it takes for an exposure to trigger a flare) is unknown, we explored several meteorological values in relation to flare onset. These included values on the day of a flare or non-flare assessment (Day 0), as patients have reported that flares may begin within hours of exposure to weather,5 as well as values in the three days preceding each assessment (Day −1, Day −2, and Day −3). Temperature, relative humidity, and barometric pressure were analyzed in increments of five units, whereas UVI was categorized into four levels adapted from the World Health Organization exposure categories: low (<3), moderate (3–5), high (6–7), and very high/extreme (≥8). Additional weather variables explored were wind chill, below freezing (<0°C), uncomfortable (<18°C),17 and wet-bulb temperature. Differences between mean daily values (i.e., between Days −1 and 0, −2 and −1, and −3 and −2) were also examined to investigate whether changes in weather trigger flares. Finally, the possible influence of limited sunlight and correspondingly lower vitamin D levels was explored by examining associations for one week of continuously lower UVI (<3 and <6, after which serum vitamin D3 levels drop to over half their peak values) and two weeks of continuously lower UVI (after which serum vitamin D3 levels drop to extremely low levels).18
Ascertainment and definition of seasons
Astronomical seasons (winter, spring, summer, and fall) were defined by the equinoxes and solstices, and growing seasons were defined by the beginning and end of measurable photosynthesis in the local vegetation canopy by year (US Geological Survey). We explored growing seasons in addition to astronomical seasons because astronomical seasons can vary in meteorological values across different climatic regions. For instance, even though winter ends officially on the same day in Ann Arbor and St. Louis, colder temperatures last longer in Ann Arbor. Imputation for growing season values was performed as for temperature, relative humidity, and barometric pressure.
Statistical analysis
We investigated associations for mean daily meteorological values and changes in these values with flare onset by conditional logistic regression, clustering flare and non-flare observations by participant. Associations between seasons and flare rates were analyzed by Poisson regression with robust variance estimation. All associations were explored first using indicator variables for each exposure category and then summarized by a linear trend.
To address the hypothesis that UCPPS comprises a heterogeneous group of conditions/clinical phenotypes that might have differing flare triggers, we performed numerous stratified analyses.15 These included analyses stratified by sex; CP/CPPS versus IC/BPS diagnosis; condition duration (<2 versus ≥2 years); baseline pain limited to versus beyond the pelvis, presence of self-reported chronic overlapping pain conditions (fibromyalgia, irritable bowel syndrome, and chronic fatigue syndrome), bladder hypersensitivity (painful bladder filling or urgency), sensory hypersensitivity (self-reported sensitivity to chemicals [e.g., perfumes, gasoline], sound, odor, or bright lights), and climatic region of residence (flare rate analysis only). We also performed analyses stratified by age (< versus ≥ the mean age of 43 years) to compare our findings to previous cross-sectional studies of weather and CP/CPPS symptoms that were limited to younger participants (<50 years of age).11,12 Each of these analyses was performed for all meteorological variables and seasons, unless indicated otherwise.
Additional analyses performed for meteorological variables only were those restricted to: 1) more bothersome flares (those with worse pain [≥6 out of 10], longer duration [≥5 days], or occurring close in time to when participants sought care for their symptoms); 2) flare and non-flare observations without preceding sexual activity, as we previously observed that sexual activity was associated with flares in this study population;15 and 3) flares for which participants suspected weather as a trigger (n=1 flare) or were unsure about the likely trigger. These additional analyses were only possible for meteorological variables because of additional flare details collected on the Brief Flare Risk Factor Questionnaire for the three flare and non-flare observations included in the case-crossover study. Finally, to address the possible concern that bias may have been introduced by differences in flare and non-flare observation selection (i.e., first three during follow-up for flare observations versus randomly selected within study follow-up third for non-flare observations), we performed analyses adjusted for study follow-up third and restricted to participants with ≤2 flares (and thus whose flares were not constrained to occur earlier in follow-up).
RESULTS
Of the 424 participants who completed the MAPP study, 290 were included in the analyses of temperature, relative humidity, and barometric pressure, after excluding those with negative responses to all flare questions (n=79), those who did not complete both flare and non-flare assessments from visit 3 onwards (n=53), and those who did not provide a start date for any of their flares (n=2). The 290 participants contributed 792 non-flare and 574 flare observations to the analysis. Two hundred and thirty to 253 participants were included in the analyses of UVI (488–532 non-flare to 382–419 flare observations), after excluding those with missing UVI values (n=37–60 participants depending on the exposure timeframe). Finally, 409 participants (5,389 non-flare and 966 flare observations) were included in the season analyses, after excluding participants who did not complete flare and non-flare assessments from visit 3 onwards. Approximately three fifths of participants were female and the large majority were Caucasian, with a median age of 41.5–43.4 years, a median condition duration of 3.6–3.9 years, and median baseline urologic symptom and pain intensities of 5–6 out of 10. Most participants (84.0–88.1%) had bladder hypersensitivity and over one third (37.3–39.5%) had a chronic overlapping pain condition (Table 1).
Table 1:
Baseline demographic and clinical characteristics of urologic chronic pelvic pain syndrome participants in the Multidisciplinary Approach to the Study of Chronic Pelvic Pain Epidemiology and Phenotyping Study, 2009–2013.
| Case-crossover analysis of: |
Flare rate analysis of seasons | ||
|---|---|---|---|
| Temperature, relative humidity, and barometric pressure | Ultraviolet light index | ||
| N | 290 | 2441 | 409 |
| Female (%) | 60.5 | 61.0 | 55.5 |
| Caucasian (%) | 91.5 | 92.5 | 91.4 |
| Age (years, median [range]) | 41.6 (19.4, 81.5) | 41.5 (19.4, 81.5) |
43.4 (18.9, 81.6) |
| Duration of symptoms (years, median [range]) |
3.6 (0.2, 47.6) | 3.6 (0.2, 47.6) | 3.9 (0.0, 54.1) |
| Baseline symptoms (median [range]): | |||
| Pelvic pain, pressure, or discomfort in the past 2 weeks (on a scale of 0–10) | 5 (1, 10) | 5 (1, 9) | 5 (1, 10) |
| Urgency in the past 2 weeks (on a scale of 0–10) | 6 (0, 10) | 5 (0, 10) | 5 (0, 10) |
| Frequency in the past 2 weeks (on a scale of 0–10) | 5 (0, 10) | 5 (0, 10) | 5 (0, 10) |
| Pain severity score (on a scale of 0–28)2 | 16 (0, 28) | 15 (0, 28) | 15 (0, 28) |
| Urinary severity score (on a scale of 0–25)3 | 14 (0, 25) | 14 (0, 25) | 13 (0, 25) |
| Bladder-associated symptoms (painful urgency, painful filling, %) | |||
| None | 13.5 | 11.9 | 16.0 |
| Either | 30.1 | 30.1 | 31.3 |
| Both | 56.4 | 58.0 | 52.7 |
| Presence of chronic overlapping pain conditions (%) | |||
| Irritable bowel syndrome | 23.4 | 22.1 | 23.1 |
| Fibromyalgia | 3.7 | 3.4 | 3.2 |
| Chronic fatigue syndrome | 2.4 | 2.7 | 2.8 |
| Any chronic overlapping pain condition | 39.5 | 38.6 | 37.3 |
Sample size for the analysis of ultraviolet index data on Day 0, the day of flare onset for flare assessments or questionnaire completion for non-flare assessments.
Defined as the sum of the Genitourinary Pain Index pain sub-score and question 4 of the Interstitial Cystitis Symptom Index.
Defined as the sum of the Genitourinary Pain Index urinary sub-score and questions 1 to 3 of the Interstitial Cystitis Symptom Index.
All study sites experienced some degree of seasonal variation, with hotter, high UVI summers (average monthly temperature: 18.7 to 28.4°C across sites, overall range: 5.0 to 40.6°C; UVI: 6.2 to 9.8, range: 0.8 to 12.7) and colder, low UVI winters (average monthly temperature: −0.5 to 17.5°C, range: −26.1 to 27.2; UVI: 1.1 to 5.7, range: 0.1 to 10.3; Appendix Figure 1). The opposite pattern was observed for relative humidity, with generally higher values in the winter (average monthly relative humidity: 65.9 to 76.4%, range: 18.6 to 100%) and lower values in the summer (average monthly relative humidity: 58.3 to 68.3%, range: 18.0 to 100%) for all sites, except for Los Angeles, Miami, and Atlanta (Appendix Figure 2). Barometric pressure was relatively constant throughout the year, with monthly averages ranging from 982.7 to 1018.4 hPa (overall range: 911.3 to 1033.9 hPa).
Daily weather
Overall, no significant trends were observed for increasing daily mean temperature (odds ratios [ORs] ranging from 0.998 to 1.009 per 5°C increase, p-value for linear trend: 0.769–0.978) and relative humidity (ORs: 0.990–1.012 per 5 percentage point increase, p-trends: 0.325–0.881) on the day of a flare or in any of the three preceding days (Table 2). Null results were also observed for below freezing, uncomfortable, and wet-bulb temperature (data not shown). Possibly suggestive protective trends were observed for increasing daily barometric pressure (ORs: 0.942–0.986 per 5 hPA increase, p-trends: 0.185–0.745) and UVI (ORs: 0.918–1.111 per UVI category increase, p-trends: 0.096–0.560) in the three days before a flare, but none of these trends were statistically significant. When changes in each of these meteorological variables were investigated, generally null findings were observed, with the exception of significant protective associations for a ≥1 level increase in UVI from the day before to the day of a flare (OR=0.342, 95% confidence interval [CI]: 0.123–0.949) and from three to two days before a flare (OR=0.192, 95% CI: 0.052–0.709; Table 3). However, no overall patterns in associations were observed. Finally, null results were observed for time since at least one moderate or high UVI day, and no appreciable pattern of findings was observed in stratified and restricted analyses (data not shown).
Table 2:
Associations for mean daily values of temperature, relative humidity, barometric pressure, and ultraviolet index with urologic chronic pelvic pain syndrome flare onset in a case-crossover study in the Multidisciplinary Approach to the Study of Chronic Pelvic Pain Epidemiology and Phenotyping Study, 2009–2013.
| Flares assessments (“cases”) | Non-flare assessments (“controls”) | Matched odds ratios (95% confidence intervals)1 | P-trend | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| N | Mean | N | Mean | |||||||||
| Temperature (°C) | Per 5 °C increase | |||||||||||
| Day 0 | 574 | 13.11 | 792 | 12.97 | 1.003 (0.946, 1.063) | 0.920 | ||||||
| Day −1 | 574 | 13.04 | 791 | 12.98 | 0.998 (0.942, 1.058) | 0.954 | ||||||
| Day −2 | 574 | 13.07 | 792 | 12.96 | 0.999 (0.943, 1.059) | 0.978 | ||||||
| Day −3 | 574 | 13.27 | 791 | 13.04 | 1.009 (0.951, 1.070) | 0.769 | ||||||
| Relative humidity (%) | Per 5 percentage point increase | |||||||||||
| Day 0 | 571 | 66.24 | 790 | 66.29 | 0.990 (0.949, 1.033) | 0.651 | ||||||
| Day −1 | 571 | 66.08 | 790 | 65.58 | 1.012 (0.971, 1.055) | 0.575 | ||||||
| Day −2 | 573 | 66.16 | 787 | 66.64 | 0.978 (0.936, 1.022) | 0.325 | ||||||
| Day −3 | 571 | 66.13 | 790 | 66.28 | 0.997 (0.954, 1.041) | 0.881 | ||||||
| Barometric pressure (hPa) | Per 5 hPa increase | |||||||||||
| Day 0 | 573 | 994.16 | 790 | 994.83 | 0.946 (0.865, 1.035) | 0.227 | ||||||
| Day −1 | 573 | 994.29 | 788 | 994.92 | 0.964 (0.880, 1.056) | 0.429 | ||||||
| Day −2 | 573 | 994.25 | 792 | 994.76 | 0.986 (0.905, 1.074) | 0.745 | ||||||
| Day −3 | 573 | 993.98 | 791 | 994.76 | 0.942 (0.863, 1.029) | 0.185 | ||||||
| Ultraviolet index | N | Low | Moderate | High | Very high/extreme | N | Low | Moderate | High | Very high/extreme | Per level increase | |
| Day 0 | 411 | 37.7 | 28.7 | 16.6 | 17.0 | 511 | 37.2 | 23.5 | 18.8 | 20.6 | 0.918 (0.807, 1.044) | 0.193 |
| Day −1 | 419 | 35.8 | 25.8 | 18.6 | 19.8 | 532 | 37.6 | 25.0 | 17.3 | 20.1 | 1.038 (0.916, 1.175) | 0.560 |
| Day −2 | 382 | 40.8 | 25.1 | 15.5 | 18.6 | 488 | 35.9 | 28.5 | 16.6 | 19.1 | 0.955 (0.838, 1.089) | 0.495 |
| Day −3 | 411 | 35.3 | 25.3 | 16.3 | 23.1 | 512 | 37.7 | 27.5 | 17.8 | 17.0 | 1.111 (0.981, 1.258) | 0.096 |
All values were calculated by conditional logistic regression, clustering by participant. The analysis included 290 participants for all weather variables except for ultraviolet index: n=244 participants for the Day 0 analysis, 253 for the Day −1 analysis, 230 for the Day −2 analysis, and 249 for the Day −3 analysis.
Table 3:
Associations for changes in temperature, relative humidity, barometric pressure, and ultraviolet index with urologic chronic pelvic pain syndrome flare onset in a case-crossover study in the Multidisciplinary Approach to the Study of Chronic Pelvic Pain Epidemiology and Phenotyping Study, 2009–2013.
| Flare assessments (“cases”) | Non-flare assessments (“controls”) | Matched odds ratios (95% confidence intervals)1 | P-trend | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N | Decrease (%)2 | Increase (%)3 | N | Decrease (%)2 | Increase (%)3 | Decrease2 | Increase3 | Per category increase | ||
|
Temperature | ||||||||||
| Day −1 to 0 | 574 | 4.5 | 5.1 | 791 | 4.8 | 3.2 | 0.989 (0.577, 1.698) | 1.539 (0.854, 2.773) | 1.228 (0.834, 1.809) | 0.297 |
| Day −2 to −1 | 574 | 5.4 | 4.2 | 791 | 4.2 | 3.0 | 1.295 (0.759, 2.210) | 1.413 (0.780, 2.559) | 1.015 (0.694, 1.484) | 0.938 |
| Day −3 to −2 | 574 | 4.5 | 3.0 | 791 | 4.3 | 4.1 | 1.001 (0.583, 1.719) | 0.655 (0.353, 1.215) | 0.829 (0.563, 1.223) | 0.345 |
|
Relative humidity | ||||||||||
| Day −1 to 0 | 568 | 29.8 | 30.3 | 788 | 29.3 | 30.0 | 1.068 (0.813, 1.403) | 0.994 (0.759, 1.300) | 0.965 (0.836, 1.114) | 0.626 |
| Day −2 to −1 | 570 | 29.1 | 28.3 | 785 | 33.9 | 26.0 | 0.775 (0.596, 1.007) | 0.996 (0.759, 1.307) | 1.138 (0.985, 1.314) | 0.080 |
| Day −3 to −2 | 570 | 28.1 | 29.8 | 785 | 31.1 | 29.3 | 0.762 (0.580, 1.001) | 0.922 (0.704, 1.206) | 1.098 (0.950, 1.269) | 0.205 |
|
Barometric pressure | ||||||||||
| Day −1 to 0 | 572 | 13.8 | 12.6 | 786 | 12.0 | 13.1 | 1.196 (0.847, 1.687) | 0.973 (0.697, 1.357) | 0.904 (0.725, 1.127) | 0.369 |
| Day −2 to −1 | 572 | 13.3 | 13.8 | 788 | 13.1 | 13.7 | 1.033 (0.740, 1.444) | 0.962 (0.691, 1.338) | 0.965 (0.780, 1.193) | 0.740 |
| Day −3 to −2 | 572 | 12.4 | 13.6 | 791 | 13.0 | 14.2 | 0.885 (0.633, 1.237) | 0.939 (0.680, 1.297) | 1.027 (0.833, 1.267) | 0.800 |
|
Ultraviolet index | ||||||||||
| Day −1 to 0 | 256 | 3.1 | 2.3 | 276 | 2.9 | 5.1 | 0.715 (0.254, 2.011) | 0.342 (0.123, 0.949) | 0.665 (0.341, 1.296) | 0.231 |
| Day −2 to −1 | 231 | 4.8 | 4.8 | 263 | 3.0 | 4.2 | 1.444 (0.562, 3.713) | 1.112 (0.465, 2.661) | 0.892 (0.482, 1.651) | 0.715 |
| Day −3 to −2 | 228 | 4.0 | 1.3 | 250 | 4.8 | 5.6 | 0.701 (0.273, 1.799) | 0.192 (0.052, 0.709) | 0.637 (0.324, 1.253) | 0.191 |
All values were calculated by conditional logistic regression, clustering by participant. The analysis included 290 participants for all weather variables except for ultraviolet index: n=175 participants for the Day −1 to 0 analysis, 159 for the Day −2 to −1 analysis, and 155 for the Day −3 to −2 analysis.
< −5°C change for temperature, < −5 percentage point change for relative humidity, < −5 hPa change for barometric pressure, and < −3 level change for ultraviolet index. Unchanged weather variation was the reference.
> 5°C change for temperature, > 5 percentage point change for relative humidity, > 5 hPa change for barometric pressure, and > 3 level change for ultraviolet index.
Seasons
No differences in flare rate were observed by astronomical or growing season in the main (Table 4) and most stratified or restricted analyses (data not shown).
Table 4:
Relative rates and 95% confidence intervals1 for the association between seasons and urologic chronic pelvic pain syndrome flares in the Multidisciplinary Approach to the Study of Chronic Pelvic Pain Epidemiology and Phenotyping Study, 2009–2013.
| N | Astronomical Season2 | Growing season2 | |||||
|---|---|---|---|---|---|---|---|
| Participants | Flare assessments | All assessments | Spring | Summer | Fall | ||
| All regions | 409 | 966 | 6,355 | 1.065 (0.896, 1.266) | 1.062 (0.897, 1.259) | 1.022 (0.845, 1.236) | 1.028 (0.889, 1.188) |
| Individual regions:3 | |||||||
| University of California – Los Angeles (Southern California/Arizona/Nevada) | 63 | 113 | 854 | 1.035 (0.631, 1.700) | 0.866 (0.503, 1.491) | 1.124 (0.686, 1.840) | 1.169 (0.776, 1.763) |
| Stanford University (Central California) | 23 | 44 | 262 | 0.786 (0.294, 2.099) | 1.007 (0.672, 1.508) | 0.904 (0.507, 1.613) | 1.134 (0.595, 2.160) |
| University of Washington (Oregon/Seattle/British Columbia) | 70 | 163 | 1,069 | 1.099 (0.696, 1.735) | 1.248 (0.804, 1.936) | 0.807 (0.501, 1.301) | 0.959 (0.670, 1.373) |
| University of Iowa (Iowa) | 58 | 161 | 1,115 | 1.266 (0.817, 1.962) | 1.520 (0.979, 2.361) | 1.209 (0.744, 1.965) | 1.073 (0.735, 1.564) |
| Northwestern University (Northern Illinois/Indiana) | 60 | 145 | 784 | 1.000 (0.667, 1.498) | 0.781 (0.486, 1.255) | 0.897 (0.574, 1.403) | 0.891 (0.623, 1.274) |
| Washington University (Missouri/Southern Illinois) | 55 | 106 | 866 | 0.671 (0.432, 1.042) | 1.007 (0.648, 1.564) | 1.107 (0.634, 1.932) | 1.005 (0.660, 1.530) |
| University of Michigan (Michigan/Ohio) | 69 | 201 | 1,232 | 1.224 (0.856, 1.751) | 0.937 (0.659, 1.333) | 0.946 (0.593, 1.510) | 1.047 (0.779, 1.409) |
Calculated by Poisson regression with robust variance estimation.
The reference group for astronomical season was winter and for growing season was season without photosynthetic activity.
Excludes Atlanta and Miami because of small numbers.
DISCUSSION
Overall, we found minimal evidence to suggest that weather (temperature, barometric pressure, relative humidity, UVI, and seasons) or changes in weather trigger flares in our large prospective analysis of UCPPS participants. Although we observed suggestive protective trends for higher barometric pressure and higher UVI (or conversely, positive trends for lower barometric pressure and lower UVI), none of these trends were statistically significant.
Our generally null findings differ from those from most previous studies, including patient reports that cold, rainy, or windy weather, and changes in weather worsen their UCPPS symptoms and trigger flares,4–10 and positive cross-sectional associations for lower daily temperature and sunlight, and higher altitude with CP/CPPS symptoms.11,12 Our null findings also differ from those from the only previous longitudinal study, to our knowledge, to have examined weather and UCPPS flares. This study observed fewer breakthrough episodes (or flares) in the summer than in other seasons in CP/CPPS patients.13
Reasons for differences between our findings and those from previous patient surveys and empirical studies are unclear, but could be multi-fold. Although most previous studies focused on men, CP/CPPS, and younger participants, our null findings in both male and female participants, those diagnosed with CP/CPPS and IC/BPS, and younger and older participants suggest that our broader focus on both sexes and conditions, and the full range of adult age does not explain our null findings. In addition, our exclusive focus on North American participants, and consequently North American weather, is unlikely to explain our null findings because weather triggers have previously been reported by participants in both American (including our previous MAPP studies)8–10 and non-American (Scandinavian) studies.4–7
Additional differences between our study and several previous, empirical studies are our outcome definition, study design, and sample size. We used a case-crossover study design nested in a longitudinal study to investigate the association between weather and UCPPS flare onset (with all comparisons made within participants), whereas most previous empirical studies used a cross-sectional study design to examine the association between weather (or altitude) and the presence and intensity of current CP/CPPS symptoms.11,12 Therefore, one possible explanation for previous cross-sectional findings could be the potential influence of weather on symptom reporting – i.e., lower (better) symptom values on longer, sunnier and warmer days than on shorter, darker and colder days, which might be more pronounced for symptom intensity than for flare reporting. This possibly greater susceptibility to weather-related mood might lead to seemingly protective associations for higher temperature, longer sunlight duration, and lower altitude with symptom intensity, but not with flares.
Alternatively, associations may have been easier to detect in at least one of the two previous cross-sectional surveys because of its much larger sample size,11 particularly if only a small proportion of UCPPS patients are susceptible to the potentially deleterious effects of weather (estimated to be as low as 26% in our previous MAPP patient survey8). Therefore, depending on the prevalence of susceptibility, sample sizes much larger than ours may have been necessary to detect associations between weather and flares. For instance, in the migraine/headache field, significant associations have been observed between weather and migraines/headaches, but these were typically observed in studies with considerably larger sample sizes than our study (i.e., larger number of migraine/headache episodes), particularly for associations between non-extreme changes in weather and migraines/ headaches.19–22 and references therein Smaller sample sizes were sufficient to detect associations for more extreme weather changes, such as Chinooks and atmospheric pressure changes during the typhoon season,23,24 as well as associations among participants believed to be sensitive to weather.25 Although we performed analyses restricted to flares more likely to have been caused by weather (i.e., those not suspected by participants to have been caused by other factors), this restriction may still not have been specific enough to detect associations.
Additional factors that may have made it difficult to detect associations in our study include possible misclassification of weather values, as we used the first 3 digits of participants’ residential zip codes to assign weather values for the full duration of the study, which may have introduced some exposure misclassification if they were travelling. This type of misclassification could be reduced in future studies by data collection by app. An additional source of misclassification may have been uncertainty surrounding the empirical induction period, as this time period is unknown. However, this concern should have been reduced by testing multiple possible induction periods (hours to three days). Finally, use of medications after a weather trigger may have prevented weather-induced flares and made it difficult to detect associations. However, the fact that several previous studies have observed associations between weather and migraines/headaches19–21 despite use of similar study design and exposure characteristics as our study suggests these possible limitations may have been less of a concern. Rather, our sample size, which was large relative to many UCPPS studies, may have needed to be even larger with a greater number of flares to detect associations in the possible presence of heterogeneity of flare etiology.
In conclusion, in one of the few prospective studies of weather and UCPPS flares, we observed minimal evidence to suggest that weather triggers flares in most CP/CPPS and IC/BPS participants. Future studies should include a greater number of flares per participant or more targeted sub-analyses to explore the possibility that only a subset of participants are susceptible to weather.
Supplementary Material
ACKNOWLEDGMENTS
We thank the research staff at the MAPP discovery sites and the data coordinating center for implementing the MAPP Epidemiology and Phenotyping Study, and the participants for their generous participation.
The content of this manuscript was presented at the International Continence Society Annual Meeting in Gothenburg, Sweden, in September 2020. This work was supported by the US National Institutes of Health/National Institute of Diabetes and Digestive and Kidney disease (U01 DK082315, U01 DK82316, U01 DK82325, U01 DK82333, U01 DK82342, U01 DK82344, U01 DK82345, and U01 DK82370).
Citations
- 1.Clemens JQ, Markossian T. & Calhoun EA. Comparison of economic impact of chronic prostatitis/chronic pelvic pain syndrome and interstitial cystitis/painful bladder syndrome. Urology 73, 743–746, doi: 10.1016/j.urology.2008.11.007 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Naliboff BD. et al. Widespread Psychosocial Difficulties in Men and Women With Urologic Chronic Pelvic Pain Syndromes: Case-control Findings From the Multidisciplinary Approach to the Study of Chronic Pelvic Pain Research Network. Urology 85, 1319–1327, doi: 10.1016/j.urology.2015.02.047 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sutcliffe S. et al. A longitudinal analysis of urological chronic pelvic pain syndrome flares in the Multidisciplinary Approach to the Study of Chronic Pelvic Pain (MAPP) Research Network. BJU Int 124, 522–531, doi: 10.1111/bju.14783 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Mehik A, Hellstrom P, Lukkarinen O, Sarpola A. & Jarvelin M. Epidemiology of prostatitis in Finnish men: a population-based cross-sectional study. BJU Int 86, 443–448 (2000). [DOI] [PubMed] [Google Scholar]
- 5.Hedelin H. & Jonsson K. Chronic abacterial prostatitis and cold exposure: an explorative study. Scand J Urol Nephrol 41, 430–435, doi: 10.1080/00365590701365123 (2007). [DOI] [PubMed] [Google Scholar]
- 6.Hedelin H. & Jonsson K. Chronic prostatitis/chronic pelvic pain syndrome: symptoms are aggravated by cold and become less distressing with age and time. Scand J Urol Nephrol 41, 516–520, doi: 10.1080/00365590701428517 (2007). [DOI] [PubMed] [Google Scholar]
- 7.Hedelin H, Jonsson K. & Lundh D. Pain associated with the chronic pelvic pain syndrome is strongly related to the ambient temperature. Scand J Urol Nephrol 46, 279–283, doi: 10.3109/00365599.2012.669404 (2012). [DOI] [PubMed] [Google Scholar]
- 8.Lai HH. et al. Management of Symptom Flares and Patient-reported Flare Triggers in Interstitial Cystitis/Bladder Pain Syndrome (IC/BPS)-Findings From One Site of the MAPP Research Network. Urology 126, 24–33, doi: 10.1016/j.urology.2019.01.012 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sutcliffe S. et al. Urological chronic pelvic pain syndrome flares and their impact: qualitative analysis in the MAPP network. Int Urogynecol J 26, 1047–1060, doi: 10.1007/s00192-015-2652-6 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Webster DC. & Brennan T. Self-care strategies used for acute attack of interstitial cystitis. Urol Nurs 15, 86–93 (1995). [PubMed] [Google Scholar]
- 11.Ku JH, Kim ME, Lee NK. & Park YH. Influence of environmental factors on chronic prostatitis-like symptoms in young men: results of a community-based survey. Urology 58, 853–858, doi: 10.1016/s0090-4295(01)01424-8 (2001). [DOI] [PubMed] [Google Scholar]
- 12.Lan T. et al. Influence of environmental factors on prevalence, symptoms, and pathologic process of chronic prostatitis/chronic pelvic pain syndrome in northwest China. Urology 78, 1142–1149, doi: 10.1016/j.urology.2011.04.066 (2011). [DOI] [PubMed] [Google Scholar]
- 13.Shin JH. & Lee G. Seasonal changes in symptoms in patients with chronic prostatitis/chronic pelvic pain syndrome: a seasonal follow-up study. Scand J Urol 48, 533–537, doi: 10.3109/21681805.2014.936496 (2014). [DOI] [PubMed] [Google Scholar]
- 14.Landis JR. et al. The MAPP research network: design, patient characterization and operations. BMC Urol 14, 58, doi: 10.1186/1471-2490-14-58 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Sutcliffe S. et al. A Case-Crossover Study of Urological Chronic Pelvic Pain Syndrome Flare Triggers in the MAPP Research Network. J Urol 199, 1245–1251, doi: 10.1016/j.juro.2017.12.050 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Rothman KJ. Induction and latent periods. Am J Epidemiol 114, 253–259 (1981). [DOI] [PubMed] [Google Scholar]
- 17.World Health Organization. Health impact of low indoor temperatures: Report on a WHO meeting, Copenhagen, 11–14 November 1985 Copenhagen: WHO; 1987. [Google Scholar]
- 18.Matsuoka LY, Ide L, Wortsman J, MacLaughlin JA. & Holick MF. Sunscreens suppress cutaneous vitamin D3 synthesis. J Clin Endocrinol Metab 64, 1165–1168, doi: 10.1210/jcem-64-6-1165 (1987). [DOI] [PubMed] [Google Scholar]
- 19.Martin GV, Houle T, Nicholson R, Peterlin A. & Martin VT. Lightning and its association with the frequency of headache in migraineurs: an observational cohort study. Cephalalgia 33, 375–383, doi: 10.1177/0333102412474502 (2013). [DOI] [PubMed] [Google Scholar]
- 20.Mukamal KJ, Wellenius GA, Suh HH. & Mittleman MA. Weather and air pollution as triggers of severe headaches. Neurology 72, 922–927, doi: 10.1212/01.wnl.0000344152.56020.94 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zebenholzer K. et al. Migraine and weather: a prospective diary-based analysis. Cephalalgia 31, 391–400, doi: 10.1177/0333102410385580 (2011). [DOI] [PubMed] [Google Scholar]
- 22.Yilmaz M, Gurger M, Atescelik M, Yildiz M. & Gurbuz S. Meteorologic parameters and migraine headache: ED study. Am J Emerg Med 33, 409–413, doi: 10.1016/j.ajem.2014.12.056 (2015). [DOI] [PubMed] [Google Scholar]
- 23.Okuma H, Okuma Y. & Kitagawa Y. Examination of fluctuations in atmospheric pressure related to migraine. Springerplus 4, 790, doi: 10.1186/s40064-015-1592-4 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Piorecky J, Becker WJ. & Rose MS. Effect of Chinook winds on the probability of migraine headache occurrence. Headache 37, 153–158, doi: 10.1046/j.1526-4610.1997.3703153.x (1997). [DOI] [PubMed] [Google Scholar]
- 25.Bekkelund SI, Hindberg K, Bashari H, Godtliebsen F. & Alstadhaug KB. Sun-induced migraine attacks in an Arctic population. Cephalalgia 31, 992–998, doi: 10.1177/0333102411409071 (2011). [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.