Lifestyle and health factors associated with progressing and remitting trajectories of untreated lower urinary tract symptoms among elderly men.

BACKGROUND: Knowledge of factors associated with the course of lower urinary tract symptoms (LUTS) before treatment is needed to inform preventive interventions. In a prospective study of elderly men untreated for LUTS, we identified factors associated with symptom progression and remission.
METHODS: In community-dwelling US men aged ≥65 years, the American Urological Association Symptom Index (AUA-SI) was repeated four times, once at baseline (2000-2002) and then every 2 years thereafter. Analyses included 1740 men with all four AUA-SI assessments, who remained free from diagnosed prostate cancer, and who reported no treatment for LUTS or BPH during follow-up that averaged 6.9 (±0.4) years. LUTS change was determined with group-based trajectory modelingof the repeated AUA-SI measures. Multivariable logistic regression was then used to determine the baseline factors associated with progressing compared with stable trajectories, and with remitting compared with progressing trajectories. Lifestyle, body mass index (BMI) (kg/m(2)), mobility, mental health (Short-Form 12), medical history and prescription medications were considered for selection. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated for variables in each model.
RESULTS: We identified 10 AUA-SI trajectories: 4 stable (1277 men, 73%), three progressing (345 men, 20%), two remitting (98 men, 6%) and one mixed (20 men, 1%). Men in progressing compared with stable trajectories were more likely to have mobility limitations (OR=2.0, 95% CI: 1.0-3.8), poor mental health (OR=1.9, 95% CI: 1.1-3.4), BMI≥25.0 kg m(-2) (OR=1.7, 95% CI: 1.0-2.8), hypertension (OR=1.5, 95% CI: 1.0-2.4) and back pain (OR=1.5, 95% CI: 1.0-2.4). Men in remitting compared with progressing trajectories more often used central nervous system medications (OR=2.3, 95% CI: 1.1-4.9) and less often had a history of problem drinking (OR=0.4, 95% CI: 0.2-0.9).
CONCLUSIONS: Several non-urological lifestyle and health factors were independently associated with risk of LUTS progression in older men.

Introduction

Male lower urinary tract symptoms (LUTS) represent a cluster of chronic urinary disorders that are highly prevalent worldwide,[1,2] especially among elderly men.[3,4] Multiple etiologies including benign prostatic hyperplasia (BPH) and bladder overactivity manifest as LUTS.[5] LUTS severity is assessed with the validated American Urologic Association Symptom Index (AUA-SI) or International Prostate Symptom Score (IPSS).[6] Moderate and severe LUTS exert a substantial negative effect on public health through diminished quality of life,[7,8] increased risk of falls and mortality,[9,10] and annual treatment costs totaling upwards of $3.9 billion in the US.[11,12] Given that the average life expectancy among US men who reach age 65 years has increased in the past decade,[13] the health burden of male LUTS is unlikely to abate without preventive interventions.

Prevention of LUTS progression requires knowledge of the natural symptom course before treatment is initiated. To date, prospective studies of risk factors for LUTS included a mixture of men with and without treatment.[14-17] However, factors other than symptom severity influence treatment decisions[18] and men with mild symptoms often report treatment.[8,19] Therefore, to distinguish risk factors for natural LUTS progression, additional studies among untreated men are needed.

Symptom progression is just one aspect of LUTS natural history in men.[20-26] Apparently spontaneous symptom remission and symptom stability are also consistently documented.[20-26] Identification of these patterns requires repeated AUA-SI or IPSS assessments, because LUTS fluctuate considerably within men over time.[20] To date, nearly all previous studies of LUTS risk factors assessed symptom change between only two time points. Additionally, a single study has reported factors associated with LUTS remission.[17] Identifying risk factors separately for LUTS progression and for LUTS remission may reveal novel pathways of LUTS etiology, which could enhance the translational potential for prevention and control of this condition.

This study had two objectives. First, to characterize the natural course of untreated LUTS beyond two time points, we applied group based trajectory models[27,28] to four repeated assessments of the AUA-SI in a large cohort of elderly men. Trajectory analysis is a statistical technique designed to identify mutually exclusive groups of subjects who follow a similar longitudinal pattern while accounting for individual heterogeneity in repeated measurements of an outcome variable. This method is gaining recognition in medical research,[29] but it has yet to be applied to LUTS. Second, to elucidate potential targets for LUTS prevention, we determined the independent associations of progressing and remitting LUTS trajectories with a comprehensive set of baseline lifestyle and health factors.

Subjects and Methods

Setting

We used data collected prospectively in the Osteoporotic Fractures in Men (MrOS) Study, a cohort of community-dwelling men aged ≥ 65 years. Participants were recruited in 2000–2002 from six US regions.[30,31] Men completed baseline questionnaires and in-person research visits. Subsequently, data were updated about every two years (Figure 1). Institutional Review Boards at each institution approved the study. All men gave written informed consent.

Figure 1

Study flow diagram illustrating the selection of the analytic cohort of 1740 men from the Osteoporotic Fractures in Men (MrOS) Study, USA, 2000–2009.

Urinary measures

The AUA-SI, prostate disease history, and medication use were obtained at all four time points. Categories of LUTS severity defined from the AUA-SI were mild (0–7 points), moderate (8–19 points) or severe (20–35 points).[6] Urinary bother was categorized as 0–2, 3, and 4–6.[6] Men reported histories of diagnosed BPH, laser surgery or transurethral resection of the prostate, and medication use for prostate symptoms. Current prescription medications were inventoried at each time point and matched to ingredients using a standardized method[32] as described previously.[16,25] LUTS medications were alpha-blockers, urinary antispasmodics, anticholinergics, and 5-alpha-reductase inhibitors.

Baseline factors

Cigarette smoking was coded into lifetime pack-years and current alcohol consumption into average drinks per week. History of problem drinking was defined as 2–4 positive responses to the CAGE questionnaire.[33,34] Caffeine consumption (mg/day) was obtained from a Block Food Frequency Questionnaire[35] and categorized into quartiles. Physical activity was obtained with the validated Physical Activity Scale for the Elderly (PASE), which assesses amount of leisure and household activities.[36] Self-reported daily walking for exercise was also assessed. Mobility limitation was defined as difficulty walking two to three blocks or difficulty climbing one flight of stairs.[37] Health related quality of life was obtained with the Short Form-12 (SF-12) Physical Component (PCS) and Mental Component (MCS) scores.[38] MCS ≤ 50 is a valid measure of common mental health disorders (depression or anxiety disorders).[39] Medical conditions included reports of physician-diagnosed diabetes, hypertension, angina, myocardial infarction, stroke, prostatitis, and cancers of the prostate, colon/rectum, lung and skin, as well as dizziness, history of falls and back pain in the past year.

Height and weight were classified into standard body mass index (BMI) (kg/m2) categories as <25.0 (normal), 25.0–29.9 (overweight), or ≥ 30.0 (obese). [40] Baseline prescription medications included hypoglycemics (insulin, glucose), diuretics (thiazide, loop, and potassium sparing) and other anti-hypertensive (ACE inhibitors, angiotensin II receptor antagonists, beta blockers, calcium channel blockers), statins (HMG-CoA reductase inhibitors), and central nervous system (CNS) medications (antiepileptics, benzodiazapenes, antidepressants, opioids, sedatives). Alpha-blockers could not be included as anti-hypertensives because use of these medications was an exclusion criterion (described below). Herbal supplements for LUTS were saw palmetto, South African star grass, stinging nettle, rye grass pollen, pumpkin seed, or African plum from self-report or inventory listing. Men with missing medication information were coded as non-users, because results with this coding were similar to results excluding the missing observations.

Analytic cohort

The 3 594 men with no baseline history of prostate cancer, BPH surgery, or medication use for LUTS or BPH were followed through the fourth AUA-SI assessment. Men who died or withdrew (n=456, 12%), had incident prostate cancer (n=213, 6%), missing AUA-SI (n=120, 3%), reported BPH treatment or used prescription LUTS medications (n=946, 26%), were excluded (Figure 1). The analytic cohort of 1 740 had mean (sd) follow-up of 6.9 (0.4) years. Treatment onset, which may occur in men with mild LUTS,[19] was not used as a marker of LUTS progression.

Statistical analyses were performed with SAS 9.1 software (SAS Institute, Cary NC). Two-sided p-values were estimated.

LUTS trajectory analysis

Group-based trajectory modeling was applied to the repeated AUA-SI scores as the continuous dependent variable. Trajectory modeling applies a semi-parametric mixture model to longitudinal data using the maximum likelihood method.[27] This method assumes that the population contains an unspecified number of underlying groups, each with different probability distribution for the longitudinal sequence of the dependent variable. Modeling started with three trajectories. As the trajectory number was successively increased by one, model fit was assessed with the product two times the change in the Bayesian Information Criterion (2ΔBIC). Values >10 are considered evidence of better fit of the larger trajectory number compared to the next smallest.[27,28] Mean posterior probabilities in each trajectory were computed and values >0.70 indicate high internal reliability.[27] We specified that the sample size in any trajectory must be at least 1% of the analytic cohort. Ultimately, the 10 trajectory model optimized fit, internal reliability, and sample size. Plots of individual AUA-SI scores in each trajectory confirmed that trajectory analysis successfully grouped men with similar longitudinal patterns (see examples in the online supplemental Figure).

Risk factor analyses

We performed risk factor analyses within strata of mild or moderate baseline LUTS. Too few men had severe untreated baseline LUTS for further study. In each stratum, men with stable trajectories formed the referent group to whom men with progressing LUT were compared. Men with remitting LUTS were compared men with progressing LUTS, because factors associated with symptom improvement could also inform LUTS prevention. Baseline variables that differed between the outcome and referent groups with p-values ≤ 0.25 were candidates for selection in forward, stepwise logistic regression modeling.

In separate models for each comparison defined above, candidate variables associated with the outcome at p ≤ 0.15 were retained. We used this larger alpha-level so as not to ignore potentially important associations for variables with low baseline prevalence. When a medical history variable was replaced with an appropriate medication variable, model fit worsened. Therefore, final models contained the medical history variables. BMI categorized as ‘normal’ and ‘overweight/obese’ improved model fit. Odds ratios (OR) and their 95% confidence intervals (CI) are reported for the final multivariable models.

Results

The 1 740 men in the analytic cohort reflected the baseline untreated cohort on nearly all characteristics including mean age, but had slightly lower mean AUA-SI scores (Table 1). In the analytic cohort, mean (sd) change in the AUA-SI score from baseline to the fourth assessment was 1.0 (4.6).

Table 1

Baseline Characteristics among Men with no History of LUTS Treatment and the Analytic Sample Derived from this Initial Cohort, the Osteoporotic Fractures in Men (MrOS) Study, USA, 2000–2009.

Characteristic	Men with no History of Treatment for LUTS 1N=3 594	Analytic SampleN=1 740
	Mean (SD)	Mean (SD)
Age (years)	72.7 (5.6)	71.4 (4.8)
BMI (kg/m2)	27.3 (3.8)	27.3 (3.7)
PASE Score2	152 (69)	158 (66)
SF-12 Physical Component Score	50 .0 (9.6)	51.4 (8.1)
SF-12 Mental Component Score	55.7 (6.8)	56.3 (6.0)
AUA-SI	7.3 (5.7)	6.0 (4.8)
	Percent (%)	Percent (%)
Race/Ethnicity
Caucasian	89%	90%
African American	4%	3%
Asian	3%	3%
Hispanic/Other	3%	3%
High school education or less	24%	23%
Live alone	13%	11%
Cigarette Smoking
≥40 pack years	17%	15%
20–39.9 pack years	17%	19%
<20 pack years	27%	27%
None	38%	39%
Alcohol Consumption
≥14 drinks/week	12%	13%
7–13.9 drinks/week	14%	16%
≤6.9 drinks/week	40%	40%
None	33%	32%
History of Problem Drinking	16%	16%
Walk Daily for Exercise	50%	51%
Mobility Limitation	11%	8%
Benign Prostatic Hyperplasia	29%	25%
Diabetes	11%	9%
Hypertension	38%	36%
Anti-hypertensive use
Diuretic	17%	13%
Non-diuretic	27%	25%
Statins	25%	24%
Central nervous system medication use	10%	8%
Herbal supplements for LUTS/BPH	12%	10%

Men untreated at baseline and with no prostate cancer history.

Physical Activity Scale for the Elderly (PASE).[36] Higher scores indicate greater activity. Percentages may not add to 100% due to rounding.

Trajectory results

We identified 10 trajectories of AUA-SI scores (Figure 2), illustrated with mean scores at each time point. Four trajectories consistent with LUTS stability (blue) contained 1 277 (73%) men and were observed in the low and high AUA-SI range. Three trajectories consistent with progression (red) contained 345 men (20%), were primarily in the moderate range, and had distinct profiles including abrupt increase late in follow-up. Two trajectories consistent with remission (green) contained 98 (6%) men and were in the moderate-high range. One trajectory had mixed progression and remission (yellow) and contained 20 men (1%).

Figure 2

Trajectory shape as illustrated with plots of mean American Urologic Association-Symptom Index (AUA-SI) scores over time among elderly men never treated for lower urinary tract symptoms (LUTS), the Osteoporotic Fractures in Men (MrOS) Study, USA, 2000–2009.

Supplemental tables S1–S3 provide mean posterior probabilities and distributions of urinary measures in each trajectory. Patterns of urinary bother, which increased in progressing groups and decreased in remitting groups, further support the internal consistency of the trajectory results.

Percentages of men in stable, progressing or remitting trajectories differed by baseline LUTS severity (Figure 3). In men with mild baseline LUTS, 90% were in stable trajectories. Of men with moderate baseline LUTS, 49% were classified into progressing and 17% into remitting trajectories. Of the 28 men had severe baseline LUTS, most were classified into remitting or stable trajectories.

Figure 3

Percentages of men in stable, progressing or remitting trajectories according to baseline LUTS severity, the Osteoporotic Fractures in Men (MrOS) Study, USA, 2000–2009.

Risk factors

In univariable analyses, men in progressing compared to stable trajectories more often had MCS <50, history of hypertension, and back pain, regardless of baseline LUTS category (Table 2). Within strata, several additional factors differed between men in progressing compared to stable trajectories. Among men with moderate baseline LUTS, those in remitting compared to progressing trajectories (Table 3) less often had high BMI, ≥ 40 pack-years of smoking, problem drinking, high caffeine intake, diabetes, hypertension, angina, or anti-hypertensive medication use (especially diuretics), but more often used CNS medications.

Table 2

Comparison of baseline demographic, lifestyle, quality of life and medical factors among elderly men in stable and progressing trajectories stratified by mild or moderate LUTS.1

	AUA-SI 1–7 points (mild)			AUA-SI 8–19 points (moderate)
Trajectory type	Progressing	Stable	P	Progressing	Stable	P
Number in group	101	1 103		242	156
Age Group			0.73			0.16
65–69 years	42%	45%		41%	32%
70–74 years	32%	31%		31%	34%
≥75 years	27%	24%		28%	34%
White race	87%	90%	0.37	93%	89%	0.13
High school education or less	19%	22%	0.41	22%	24%	0.60
Live alone	17%	11%	0.11	9%	15%	0.06
BMI≥25.0 kg/m2	81%	72%	0.05	76%	72%	0.34
Cigarette Smoking			0.94			0.91
≥40 pack years	15%	15%		19%	19%
20–39.9 pack years	18%	19%		18%	18%
<20 pack years	25%	27%		27%	24%
None	42%	40%		36%	39%
Alcohol Consumption			0.43			0.66
≥14 drinks/week	12%	13%		13%	14%
7–13.9 drinks/week	19%	15%		18%	13%
≤6.9 drinks/week	35%	42%		36%	39%
None	35%	30%		33%	34%
History of Problem Drinking	19%	14%	0.15	23%	22%	0.87
Caffeine Intake			0.56			0.82
Quartile 1	27%	24%		20%	23%
Quartile 2	25%	24%		25%	24%
Quartile 3	24%	24%		25%	23%
Quartile 4	25%	27%		30%	30%
Physical Activity Score2			0.04			0.45
0–99 points	27%	16%		20%	22%
100–149 points	25%	31%		31%	33%
150–199 points	24%	28%		30%	23%
≥200 points	24%	26%		19%	22%
Walk Daily for Exercise	42%	53%	0.03	48%	51%	0.51
Mobility Limitation	14%	6%	0.002	8%	8%	0.98
SF-12 Physical Component Score			0.29			0.38
<50 points	28%	24%		33%	34%
50–54 points	27%	22%		26%	20%
≥55 points	46%	54%		41%	46%
SF-12 Mental Component Score			0.02			0.01
<50 points	18%	10%		19%	9%
50–54 points	13%	10%		14%	12%
≥55 points	69%	80%		67%	79%
Medical History
Diabetes	11%	8%	0.26	9%	13%	0.17
Hypertension	44%	34%	0.06	43%	33%	0.05
Angina	8%	11%	0.33	16%	12%	0.33
Myocardial infarction	10%	9%	0.88	12%	16%	0.31
Stroke	5%	3%	0.19	5%	3%	0.51
Cancer (other than prostate)	23%	23%	0.04	21%	20%	0.97
Trouble with dizziness	25%	16%	0.02	27%	23%	0.40
Back pain in past year	68%	59%	0.08	74%	64%	0.04
Prostatitis	9%	5%	0.11	12%	10%	0.54
Medications or Supplements
Hypoglycemic	11%	6%	0.05	6%	10%	0.15
Anti-hypertensive			0.71			0.25
Diuretic	15%	12%		18%	12%
Non-diuretic	27%	26%		26%	29%
Statin	19%	24%	0.25	29%	25%	0.38
Antidepressant	8%	3%	0.02	4%	3%	0.63
Central nervous system	9%	6%	0.35	10%	8%	0.37
Herbal use for LUTS/BPH	7%	7%	0.98	20%	15%	0.26

Variables with p≤0.25 were considered for selection in logistic regression.

Physical Activity Scale for the Elderly (PASE).[36] Higher scores indicate greater activity.

Table 3

Comparison of baseline demographic, lifestyle, quality of life and medical factors among elderly men in remitting compared to progressing trajectories.1

	AUA-SI 8–19 points (moderate)
Trajectory type	Remitting	Progressing	P
Number in group	82	242
Age Group			0.49
65–69 years	35%	41%
70–74 years	30%	31%
≥75 years	34%	28%
White race	91%	93%	0.56
High school education or less	29%	21%	0.15
Live alone	7%	9%	0.62
BMI≥25.0 kg/m2	68%	76%	0.15
Cigarette Smoking			0.17
≥40 pack years	10%	19%
20–39.9 pack years	15%	18%
<20 pack years	34%	27%
None	41%	36%
Alcohol Consumption			0.67
≥14 drinks/week	10%	13%
7–13.9 drinks/week	15%	18%
≤6.9 drinks/week	40%	36%
None	35%	33%
History of Problem Drinking	12%	23%	0.03
Caffeine Intake			0.25
Quartile 1	26%	20%
Quartile 2	26%	25%
Quartile 3	29%	25%
Quartile 4	20%	30%
Physical Activity Score2			0.84
0–99 points	24%	20%
100–149 points	30%	31%
150–199 points	27%	30%
≥200 points	18%	19%
Walk Daily for Exercise	56%	45%	0.28
Mobility Limitation	11%	8%	0.46
SF-12 Physical Component Score			0.31
<50 points	43%	33%
50–54 points	21%	26%
≥55 points	37%	41%
SF-12 Mental Component Score			0.43
<50 points	13%	19%
50–54 points	12%	14%
≥55 points	74%	67%
Medical History
Diabetes	5%	9%	0.22
Hypertension	30%	43%	0.05
Angina	7%	16%	0.06
Myocardial infarction	12%	12%	0.96
Stroke	4%	5%	1.00
Cancer (other than prostate)	20%	21%	0.82
Trouble with dizziness	26%	27%	0.82
Back pain in past year	72%	74%	0.72
Prostatitis	15%	12%	0.47
Medications or Supplements
Hypoglycemic	4%	6%	0.39
Anti-hypertensive			0.06
Diuretic	9%	17%
Non-diuretic	21%	26%
Statin	24%	29%	0.47
Antidepressant	7%	4%	0.25
Central nervous system	17%	10%	0.10
Herbal use for LUTS/BPH	15%	20%	0.29

Variables with p≤0.25 were considered for selection in logistic regression.

Physical Activity Scale for the Elderly (PASE).[36] Higher scores indicate greater activity.

In multivariable analyses among men with mild baseline LUTS (Table 4), men with MCS <50, history of non-prostate cancer, mobility limitations, overweight, dizziness, and no daily walking for exercise were 1.5- to 2-fold more likely to have progressing compared to stable LUTS. When PASE score replaced the walking variable, the OR was elevated for the lowest level of physical activity (0–99 points) compared to the highest (≥200 points) (1.6, 95% CI:0.9–2.9) but were null for 100–149 (0.8, 95% CI: 0.5–1.5) and 150–199 points (0.9, 95% CI: 0.5–1.5).

Table 4

Factors independently associated with progressing or remitting LUTS trajectory according to baseline AUA-SI score.1

Baseline AUA-SI Score 0–7 points :	Progressing vs. Stable
Factor	Referent level	OR (95% CI)	p
SF-12 Mental Component Score
<50 points[2]	≥55 points	1.9 (1.1 – 3.4)	0.03
50–54 points	≥55 points	1.5 (0.8 – 2.8)	0.21
History of cancer (not prostate)	No cancer	1.7 (1.0 – 2.9)	0.03
Mobility limitation	No mobility limitation	2.0 (1.0 – 3.8)	0.04
Overweight or obese (BMI ≥25.0 kg/m2)	Normal/Underweight (BMI<25.0 kg/m2)	1.7 (1.0–2.8)	0.06
Trouble with dizziness	No dizziness	1.6 (0.9 – 2.6)	0.08
No daily walking for exercise	Daily walking for exercise	1.4 (0.9 – 2.2)	0.10

Factors evaluated during model building were those from univariable analyses with p≤0.25 and retained in the stepwise selection procedure at p≤0.15 as described in Methods.

Among men with moderate baseline LUTS, those with progressing compared to stable LUTS were 1.5–2.5-fold more likely to have MCS <50, hypertension, and back pain, and were less likely to have diabetes. Men with remitting compared to progressing LUTS were 2.3-fold more likely to use CNS medications at baseline, but were less likely to have histories of problem drinking, hypertension, or angina.

Discussion

Several distinct AUA-SI trajectories were identified among 1740 elderly men untreated for LUTS and trajectory types differed by baseline LUTS severity. Most men with mild baseline LUTS followed stable trajectories, whereas half of men with moderate baseline LUTS experienced progression and a fifth experienced remission. These data may allow clinicians to advise older men that prospects for worsening (or improving) symptoms are based on their current symptom level. Similarly, the baseline lifestyle and health factors associated with LUTS progression differed somewhat for progression from mild or from moderate baseline symptoms. Clinical or public health interventions that target these factors within different levels of LUTS severity may promote the prevention of symptom progression in older men.

In our study, poor mental health was a strong risk factor for LUTS progression. LUTS remission relative to progression also was associated with factors that could influence mental well-being, such as use of CNS medications and problem drinking.[41] In other studies, depressive symptoms were associated with LUTS progression,[17] but anti-depressant use was associated with higher likelihood of transition from mild to moderate LUTS.[15] Pharmacological modulation of CNS neurotransmitters, such as serotonin and GABA, may inhibit bladder overactivity and/or improve bladder capacity.[42] It is therefore notable that use of benzodiazepines, which enhance GABA actions, was more common among men in remitting than in progressing trajectories in our study. Although use of certain CNS medications could worsen LUTS,[15] their therapeutic potential warrants a more complete understanding of neurological contributions to lower urinary tract function.

The current results agree with our earlier report that LUTS progression is positively associated with overweight and inversely associated with physical activity.[16] However, others showed no associations of BMI with LUTS progression[14,15] or of physical activity with either LUTS progression or remission.[17] In older men, overweight and low physical activity may contribute to lower urinary tract dysfunction through pathways involving microvascular disease,[43,44] metabolic derangements,[45] or autonomic nervous system overactivity.[46] Consistent with these mechanisms, our results also show associations of hypertension and dizziness (a marker of orthostatic control) with LUTS progression. Our results also document that mobility and back pain may contribute to LUTS progression. Men with mobility limitations or back pain may perceive their symptoms as becoming more severe over time, if difficulty with ambulation alone, or because of pain, interferes with their ability get to or use a toilet. Alternatively, degenerative spinal conditions such as disc herniation or lumbar stenosis could contribute to both back pain and urologic dysfunction by impinging on the spinal cord or nerve roots.[47-49]

Risk factors for LUTS progression and remission identified in this study differ from those reported previously for three key reasons. First, we used trajectory modeling to account for LUTS fluctuation within men. Most earlier studies focused on change of a certain magnitude from a single previous time point, such as transition from mild (AUA-SI 0–7 points) to moderate LUTS (AUA-SI ≥ 8 points)[15,16] or 2–3 point difference in AUA-SI voiding or storage subscores.[17] These definitions may introduce misclassification if men who progress are combined with men whose symptoms are randomly fluctuating, or if men with stable and remitting symptoms are combined in the referent group. Misclassification would tend to bias associations with risk factors toward the null, which may explain why we but not others[14,15,17] observed associations with BMI and physical activity. Second, we studied men with untreated LUTS. Studies that included a mix of men with and without treatment for LUTS may have identified factors associated with treatment decisions or treatment effects.[14-17] Third, we studied older men whose risk factors for LUTS progression or remission may differ from those in younger men.

There are limitations to this research. First, we could not assess the reasons that men did not undergo treatment for LUTS. However, about 88% of men remained untreated at each AUA-SI assessment period, a proportion similar to that observed in other community-dwelling cohorts,[8,26] suggesting that the MrOS cohort is not unusual with regard to LUTS treatment initiation. Second, we did not have specific urological metrics. However, such measures would not have necessarily informed this analysis because our aim was to study long-term changes in urinary symptoms which are well-represented by the AUA-SI. Third, some of the factors studied such as CNS medication use had low baseline prevalence, which resulted in wide confidence intervals for OR estimates. Finally, the analytic cohort consisted of men ages 65 or older who survived an average of 6.9 years and results may not apply to all men at risk for LUTS progression.

This study has multiple strengths. First, MrOS was specifically designed to study LUTS prospectively in elderly men.[30] Second, the large sample size and excellent follow-up allowed us to evaluate multiple trajectory solutions and optimally characterize long-term LUTS changes. The small overall mean change in the AUA-SI during follow-up observed by us and others, [20,22-25] belies the dynamic nature of untreated LUTS among elderly men. Trajectory analysis revealed rare patterns that have not been described previously including persistently severe symptoms and mixed progression and remission. Finally, the comprehensive data available in MrOS allowed a comprehensive investigation of risk factors for LUTS change.

Conclusion

Several lifestyle and factors were associated with progressing and remitting LUTS trajectories. Back pain and CNS medication use may represent novel etiologies of LUTS that could be explored in future research. Intervening on lifestyle and health factors, especially mental health, has the potential to reduce the burden of LUTS in older men.

Supplemental Figure. Longitudinal sequence of American Urological Association Symptom Index Scores (AUA-SI) for each man in the trajectory. Examples of each trajectory type were selected for illustration, including the rarely occurring mixed pattern. Color coding corresponds to that show in main manuscript Figure 2. The black line in each panel represents the mean AUA-SI score at each time point in the trajectory.

Panel A (blue): Stable trajectory (n=781) with mean (sd) baseline AUA-SI score of 2.9 (2.1). Panel B (pink): Progressing trajectory (n=90) with mean (sd) baseline AUA-SI score of 12.7 (3.7).

Panel C (green): Remitting trajectory (n=63) with mean (sd) baseline AUA-SI score of 14.3 (2.3).

Panel D (yellow): Mixed trajectory (n=20) with mean (sd) baseline AUA-SI score of 5.9 (3.4).

Table S1. Trajectory information and urinary measures at each assessment in the four stable trajectories shown in Figure 2.1

Table S2. Trajectory information and urinary measures at each assessment in the three progressing trajectories shown in Figure 2.1

Table S3. Trajectory information and urinary measures at each assessment in the two remitting trajectories and one mixed trajectory shown in Figure 2.1