Prevalence of BPH and lower urinary tract symptoms in West Africans

Abstract

BACKGROUND:

BPH and lower urinary tract symptoms (LUTS) are very common among older men in Western countries. However, the prevalence of these two conditions in the developing countries is less clear.

METHODS:

We assessed the age-standardized prevalence of BPH and/or LUTS among West Africans in a probability sample of 950 men aged 50–74 in Accra, Ghana, with no evidence of biopsy-confirmed prostate cancer after screening with PSA and digital rectal examination (DRE). Information on LUTS was based on self-reports of the International Prostate Symptom Score (IPSS). BPH was estimated using DRE, PSA levels and imputed prostate volume.

RESULTS:

The prevalence of DRE-detected enlarged prostate was 62.3%, while that of PSA⩾1.5 ng ml^–1 (an estimate of prostate volume ⩾ 30 cm³) was 35.3%. The prevalence of moderate-to-severe LUTS (IPSS⩾8) was 19.9%. The prevalence of IPSS⩾8 and an enlarged prostate on DRE was 13.3%. Although there is no universally agreed-upon definition of BPH/LUTS, making comparisons across populations difficult, BPH and/or LUTS appear to be quite common among older Ghanaian men.

CONCLUSIONS:

We found that after age standardization, the prevalence of DRE-detected enlarged prostate in Ghanaian men is higher than previously reported for American men, but the prevalence of LUTS was lower than previously reported for African Americans. Further studies are needed to confirm these findings and identify the risk factors for BPH in both Africans and African Americans.

Introduction

BPH and lower urinary tract symptoms (LUTS) are among the most common conditions affecting older men.¹ The reported prevalence of microscopic, macroscopic and clinical BPH varies substantially. Autopsy studies conducted in Western countries show that 40% of men in their 50s and up to 90% of men in their 80s have histological evidence of BPH.² It is estimated that in the United States, a man’s lifetime risk of undergoing surgical intervention for symptomatic BPH is up to 40%.³

A key challenge in BPH-related research, including estimating prevalence or identifying risk factors, is the absence of a universally accepted case definition, particularly a case definition that can be applied widely in population-based epidemiologic studies.⁴ BPH is recognized as prostatic enlargement, the presence of voiding symptoms and/or histological evidence of hyperplasia on biopsy. The term ‘clinical BPH’ is synonymous with LUTS in the absence of urethral stricture, or prostate enlargement detected on digital rectal examination (DRE). However, different metrics can be used to assess each of these clinical facets of BPH and LUTS. The metric most often used in epidemiologic studies is the self-reported International Prostate Symptom Score (IPSS),⁵ which assesses self-reported severity of urinary symptoms and is usually classified as asymptomatic, mild, moderate or severe. Based on the IPSS, the age-standardized prevalence of moderate-to-severe LUTS in the 50–74 year age group reported in clinical studies ranges from 12 to 44%.^6–20 However, although LUTS can be a consequence of BPH, it may also be due to other conditions, such as infection or inflammation of the prostate.

Recent data suggest racial differences in LUTS prevalence in the United States. In a community-based study, the prevalence of moderate-to-severe LUTS among African-American men in Michigan (41%) was significantly higher than that among Caucasian-American men in Minnesota (34%).⁷ This finding mirrors the racial differences in prostate cancer incidence and mortality observed in the United States,²¹ but contrasts with the findings of studies utilizing surgically treated BPH²² or limited components of the IPSS²³ as the outcome. If there is indeed such a racial disparity in LUTS and/or BPH, it is unclear what factors may be involved. It would be informative to have some knowledge of the prevalence of BPH and LUTS in Western Africa, where most African Americans’ ancestors originated. However, to date, no population-based studies of BPH have been conducted in Western Africa. In this study, we utilized data from a recent population-based screening survey of prostate cancer to assess the prevalence of BPH and LUTS in a random sample of the population in Accra, Ghana.

Materials and methods

Study population

In collaboration with the Ghana Census Bureau, we selected and enrolled a population-based probability sample of ~1000 healthy men aged 50–74 years from Accra between 2004 and 2006. We used a three-stage design, including randomly selecting enumeration areas, randomly selecting households in each enumeration area and finally visiting each household in person to identify men aged 50–74 years for the study. Where there was more than one male aged 50–74 years, the one with the earliest year of birth was selected. Of the 1049 eligible men thus identified and invited to participate, 3 were too sick to participate and 9 refused, yielding 1037 participants (98.8% participation). For this study, we excluded from the analysis (1) 73 men with biopsy-confirmed prostate cancer identified through the screening survey and (2) an additional 14 men with screening PSA levels indicative of occult prostate cancer (⩾20 ng ml^–1), as the malignancy may have interfered with the detection and diagnosis of BPH. In total, 950 men were included in this study.

Interview and blood collection

Informed consent was obtained and approved by the Institutional Review Boards of the United States National Cancer Institute and the University of Ghana. The official language of Ghana is English. Participants were transported to the Korle-Bu Teaching Hospital for in-person interviews and health examination. Trained interviewers used a structured questionnaire to elicit epidemiologic information, including education, smoking, use of alcohol, medical history, screening history, family history of cancer and utilization of the medical care system, as well as information on the 7-item IPSS.⁵ Overnight fasting blood (20 ml) was collected, and a health examination was administered, including DRE, blood pressure measurement and anthropometric measurements (weight, height and waist and hip circumferences).

Detection and exclusion of prostate cancer

We used DRE and serum PSA levels to screen for and exclude prostate cancer in study participants. DRE was performed on each study subject by experienced Ghanaian urologists. Total PSA was measured using either the Hybritech Tandem-R PSA or the Access2 Hybritech PSA assays (Beckman Coulter, Fullerton, CA, USA) at University of California at Los Angeles (D.C.). Men were referred for trans-rectal ultrasound (TRUS)-guided prostate biopsy and pathology consultation at Korle-Bu Teaching Hospital if their prostates felt hard or nodular on DRE, or if their total PSA levels were >2.5 ng ml^–1. DRE findings of BPH did not warrant biopsy and histological review unless there were co-existing abnormalities indicative of malignancy. Biopsy sections (six cores) were reviewed by both Ghanaian and US pathologists (AD, GN, YT) for the presence of pathologic features.

BPH and LUTS definition

Men were considered to have prostate enlargement consistent with BPH (‘enlarged prostate’) if DRE showed the prostate to be symmetrically enlarged (estimated to be 30 cm³ or larger), smooth and rubbery, with a mobile rectal mucosa and normal median sulcus. We used the self-reported IPSS to assess LUTS severity; men with an IPSS score of 0 were considered asymptomatic, those with scores 1–7 were considered mildly symptomatic and those with scores 8–35 were considered moderately to severely symptomatic.⁵ The IPSS score was combined with the DRE results to define symptomatic prostatic enlargement.

Similar to previous studies, we used PSA levels at screening as a surrogate measure of prostate enlargement (PSA ⩾1.50 ng ml^–1 is indicative of prostate volume ⩾30 cm³ (refs. 24,25)). In addition, in an attempt to improve on prostate size estimation based solely on screening PSA,^24,25 we used data from 234 cancer-free men who had a TRUS-guided biopsy (74 of whom had screening PSA <2.5 ng ml^–1) to develop a regression model estimating prostate volume as a function of age, prostate screening results and anthropometric measurements, similar to a previously described approach.²⁴ Prostate volume and PSA levels were natural log transformed to meet the normality assumption, and independent variables that were non-significant at the 0.05 level were excluded using backwards stepwise selection. The final model was subsequently applied to all 950 subjects, including those without TRUS-guided biopsy, to impute prostate volume, which was dichotomized into <30 vs ⩾30 cm³. Those with an imputed prostate volume ⩾30 cm³ were considered to have an enlarged prostate gland.

Statistical analysis

We calculated the unadjusted point prevalence of each BPH/LUTS measure, both overall and by age group (50–59, 60–69, 70–74 years), using SUDAAN to account for the varying sample weights used in the population sampling scheme.²⁶ To allow comparison to other populations, we calculated overall prevalences for each outcome standardized to the 2000 World Health Organization world standard population²⁷ both for the current study as well as previous studies.

Results

Characteristics of the 950 men are shown in Table 1. Approximately 51% of the study subjects were between the ages of 50 and 59. Almost 40% of the men were overweight or obese (body mass index>25 kg m^–2), 77% had at least a middle-school education and 85% were currently married or living with a partner. The prevalences of physician-diagnosed diabetes and hypertension were 8 and 27%, respectively, and 66 and 43% of the population were regular consumers of alcohol and tobacco, respectively.

Table 1.

Characteristics of Ghanaian men aged 50–74, without prostate cancer^a

Characteristic	Category	N	%
Total		950	100.0
Age (years)	50–59	487	51.3
	60–69	339	35.7
	70–74	124	13.1
Highest education	None	111	11.6
	Primary school	60	6.3
	Middle school	431	45.4
	Secondary school	173	18.3
	Higher education	127	13.4
	Missing	48	5.1
Marital status	Currently married	797	83.9
	Living with a partner	8	0.8
	Widowed	46	4.8
	Divorced/separated	94	9.9
	Never married	4	0.4
	Missing	1	0.1
BMI (kgm−2)	<18.5	62	6.5
	18.5–22.9	315	33.0
	23–24.9	180	19.0
	25–29.9	293	30.9
	30+	81	8.6
	Missing	19	2.0
Ever drank alcohol regularly (at least once a week for 6 months)		628	66.0
Ever smoked regularly (at least one cig per day for 6 months)		413	43.4
Ever told had diabetes		73	7.7
Ever told had hypertension		256	27.0

Abbreviation: BMI, body mass index.

^aExcludes men with histologically confirmed prostate cancer and PSA levels >20 ng ml^–1.

Table 2 shows the prevalence of BPH/LUTS by each case definition, both age standardized and by age group. The prevalence of DRE-detected prostate enlargement was 62.3%, with higher prevalence seen for men in the 60–69 year age group (68.3%) and lowest for men aged 50–59 years (58.9%). In contrast, using the self-reported IPSS, the overall prevalence of moderate-to-severe urinary symptoms (LUTS) was only 19.9%. This estimate increased with age, reaching 34.7% among men aged 70–74 years. Using a more stringent threshold of IPSS ⩾ 15, the overall prevalence of moderate-to-severe symptoms was 4.3%. The distribution of IPSS by age group is shown in Figure 1. Only 13.3% of men overall had both DRE-detected prostate enlargement and moderate-to-severe symptoms (that is, symptomatic prostatic enlargement).

Table 2.

Prevalence of BPH and LUTS (various definitions) in Ghanaian men aged 50–74, without prostate cancer

	All ages			Age standardizeda	Age group (years)
					50–59			60–69			70–74
	N		%b	%b	N		%b	N		%b	N		%b
Total	950		100.0		487		100.0	339		100.0	124		100.0
DRE-detected BPHc	593		62.4	62.3	287		58.9	232		68.3	74		59.6
IPSS
0 (No symptoms)	52		5.5		32		6.6	15		4.5	5		4.0
1–7 (Mild symptoms)	706		74.3		383		78.6	247		72.9	76		61.2
8–19 (Moderate symptoms)	181		19.1		69		14.2	73		21.5	39		31.5
20–35 (Severe symptoms)	11		1.2		3		0.6	4		1.2	4		3.2
⩾8 (Moderate-to-severe symptoms)	192		20.2	19.9	72		14.8	77		22.7	43		34.7
⩾14 (Alternate definition of severe symptoms)	42		4.4	4.3	17		3.4	16		4.7	9		7.3
BPH on DRE (any) and IPSS (⩾8)	128		13.5	13.3	43		8.9	58		17.1	27		21.8
PSA
0–1.49 (ng ml–1)	611		64.3		364		74.7	195		57.5	52		41.9
1.50–9.99 (ng ml–1)	313		32.9		121		24.8	126		37.2	66		53.2
10.00–20.00 (ng ml–1)	26		2.7		2		0.4	18		5.3	6		4.8
⩾1.50 (ng ml–1)	339		35.7	35.3	123		25.3	144		42.5	72		58.1
Median (ng ml–1)		1.05				0.86			1.26			1.81
Mean (ngml–1)		1.91				1.35			2.42			2.74
s.d. (ngml–1)		2.53				1.51			3.25			3.00
Imputed prostate volume ⩾30 cm3d	281		29.6	29.0	71		14.6	138		40.8	72		58.2

Abbreviations: BMI, body mass index; DRE, digital rectal examination; IPSS, International Prostate Symptom Score; LUTS, lower urinary tract symptoms; TRUS, trans-rectal ultrasound; WHR, waist-to-hip ratio.

^aAdjusted to the 2000 World Health Organization World Standard Population ages 50–74.

^bPercentages account for sampling weights used in population sampling scheme.

^cDRE-detected prostatic enlargement consistent with BPH.

^dImputed based on men who underwent biopsy and were negative for prostate cancer, and whose ^[c]PSA was ⩽20 ng ml^–1 (n = 234 men, 74 of whom had PSA <2.5 ng ml^–1). Natural log-transformed prostate volume from TRUS (PV) was regressed on log-transformed screening PSA (PSA), age in years (AGE), DRE-detected enlargement (DRE-BPH), measured weight in kg (WT) and measured height in cm (HT). Final model: ln(PV) = 0.2312 × (ln(PSA))+0.0211 × (AGE)+0.0123 × (WT)+0.0091 × (HT) + 0.1878 × (DRE-BPH)–0.5712. Waist and hip circumferences, as well as WHR and BMI, were considered but were nonsignificant in the regression model. Pearson correlation between imputed PV and actual PV was 0.62. When dichotomizing PV as <30cm³ vs ⩾30 cm³, the k statistic between imputed and actual measures was 0.52, implying moderate agreement. Using the final regression model, PV was imputed for all subjects in the dataset, then dichotomized. Results for this imputed PV are presented here.

Figure 1.

Distribution of International Prostate Symptom Score (IPSS) by age, Ghanaian men aged 50–74, without prostate cancer.

The prevalence of PSA ⩾1.5 ng ml^–1 was 35.3%. Using this PSA threshold as a surrogate for enlarged prostate (prostate volume ⩾30 cm³), 25.3, 42.5 and 58.1% of men aged 50–59, 60–69 and 70–74 years had enlarged prostates.

Based on the regression model incorporating TRUS data to impute prostate volume, the prevalence of enlarged prostate (⩾30 cm³) was 29.0% for all age groups and 58.2% for those aged 70–74 years. Age (in years), screening PSA, DRE-detected prostate enlargement, height and weight were significant predictors of TRUS-measured prostate volume; body mass index, waist-to-hip ratio, waist circumference and hip circumference were not. The adjusted r² of the final linear regression model for imputed prostate volume was 0.441. Among the biopsied men with actual TRUS-based prostate volume, concordance of imputed vs actual prostate volume (⩾ 30 cm³ vs <30 cm³) was moderate (k = 0.52).

Discussion

In this population-based study of West African men aged 50–74 years with no evidence of prostate cancer, the prevalence of BPH or LUTS appears to be high. The observed age-standardized prevalence of prostate enlargement by DRE was 62.3%, that of self-reported moderate-to-severe LUTS was 19.9% and those estimated from PSA values alone and a regression model based on age, PSA and DRE were 35.3 and 29.0%, respectively. The prevalence of BPH and/or LUTS increased with age for all case definitions except for DRE-detected BPH.

All of these estimates imply that BPH is a common condition in West African men. However, because of the lack of a standard and universally accepted case definition, and issues of comparability of measures from different studies and across different populations, putting these estimates in context is difficult. The estimate of 19.9% based on self-reported IPSS is comparable and in some cases higher than estimates from mostly white populations from France (12%), Germany (14%) and Canada (21%).^13–15 However, our estimate for West African men was lower than those from studies using this same IPSS measure among African-American men in the United States,^7,10,11,17 which ranged from 26 to 45% (Table 3).

Table 3.

Population-based studies of BPH prevalence in African Americans, vs current study^a

Reference	Population	Year	Community-based	Exclusions at outset		Years of accrual	Definition of BPH
				Prostate cancer	Prostate surgery		IPSS ⩾8
Sarma et al,7	Flint Michigan—369 blacks	2003	Y	Y	Y	1996	50–59: 41.7%
							60–69: 51.1%
							70–79: 40.0%
							Overall: 44.8%b
Glasser et al,11	USA—596 blacks	2007	Y	N	N	2001–2002	Overall: 26.4%c
Markland et al,10	San Antonio—355 blacks	2007	N	Y	N	2001–2006	50–59: 34%
							60–69: 33%
							70–79: 57%
							Overall: 36.3%b
Kupelian et al,17	Boston—1770 blacks aged 30–79	2006	Y	Y	Y	2002–2005	Overall: 19.3%d
Current study	Accra, Ghana—950 West Africans	2009	Y	Y	N	2004–2006	50–59: 14.8%
							60–69: 22.7%
							70–74: 34.7%
							Overall: 19.9%b

Abbreviation: IPSS, International Prostate Symptom Score.

^aIncludes population-based studies reporting IPSS⩾8.

^bOverall estimates standardized to the 2000 World Health Organization world standard population for men aged 50–74 years.

^cOverall estimates are not age standardized; data to standardize not presented.

^dOverall estimates are age standardized to the US 2000 Census for specified age grouping.

While measures of LUTS via the IPSS are the most widely available data for comparing populations, factors other than BPH could influence these estimates. Although the most common cause of LUTS is BPH, these symptoms may also be caused by other conditions, such as urethral stricture, infection and inflammation, whose prevalences can differ between populations. In addition, cultural and social factors can impact perception and reporting of these symptoms, and thus the reported prevalence. For example, regardless of LUTS severity, African Americans reported less urinary ‘bother’ than Caucasian Americans. The extent to which cultural factors specific to West Africans might influence self-reporting of urinary symptoms is unknown.

In contrast to the results for IPSS, the prevalence of prostate enlargement based on DRE in West African men (62.3%) was substantially higher than that reported for predominantly white populations in the United States (36.3 and 42.0%).^9,28 This would seem to suggest that self-report of urinary symptoms is not a reliable measure of BPH in this population for comparison with others. However, DRE-based estimates of prostate enlargement could also be affected by comparability issues, such as differences in physician training. Furthermore, the interrater reliability of this measure is unknown in Ghana. As noted earlier, DRE-based prostate enlargement was the only observed measure that did not increase with age in this study. While this could be a factor in the current study, it must be noted that a similar non-linearity with age was observed for IPSS distribution among African Americans but not Caucasian Americans in a previous study.⁷ Accordingly, variability in DRE-based prostate enlargement is unlikely to explain all of the difference in prevalence estimates we observed for DRE-based vs IPSS-based measures of BPH.

PSA, less affected by self-reporting and measurement errors, is another measure that has been used to compare the prevalence of BPH in different populations, since it correlates positively with prostate size, and has been used as a marker for prostate volume.^24,25 However, currently it is unclear whether PSA⩾1.5 ng ml^–1 is an optimal cutoff for all populations. For example, in our study, 211 subjects had TRUS and had PSA ⩾1.5 ng ml^–1. Of these, only 62.5% had a prostate volume ⩾30 cm³, suggesting that other factors may affect prostate volume and should be taken into account in estimating prostate size. In our study, the k coefficients comparing TRUS-measured prostate volumes ⩾30 cm³ to prostate volume defined by PSA ⩾1.5 ng ml^–1 and imputed prostate volume were 0.32 and 0.52, respectively, suggesting that the definition of an enlarged prostate based on the imputed values from the regression model may reflect more closely the true prostate volume than using the screening PSA alone. Nevertheless, the prevalence derived from both methods is comparable in men older than 60. In the model, based on improvement in the adjusted r², screening PSA was the strongest independent predictor of prostate volume, followed by age, DRE-detected benign prostate enlargement, weight and height. The inclusion of variables beyond PSA improved the model fit considerably (adjusted r² increased from 0.25 for PSA alone to 0.44 in the final model). Interestingly, body mass index, a function of both weight and height, was not a significant independent predictor. It should be noted that the imputation of prostate volume was applied to all subjects based on models derived from the subset of subjects who had undergone TRUS (n = 234), and required several statistical assumptions that may in turn affect the validity of the imputation.

We excluded 87 men with either histologically confirmed prostate cancer or PSA levels ⩾20 ng ml^–1, as the presence of confirmed or occult prostate cancer may have impacted the detection, and therefore the reported prevalence, of BPH or LUTS. Had these subjects been included, the reported age-standardized prevalence of IPSS ⩾ 8, PSA ⩾ 1.5 ng ml^–1 and imputed prostate volume ⩾ 30 cm³ would have been slightly different (61.6, 39.4 and 32.9%, respectively).

This study has several strengths that lend credibility to the reported BPH and LUTS prevalence in West Africa. These include the use of probability sample of the population, the high response rate (98.8%), the availability of screening PSA levels and DRE results for all subjects, the use of the IPSS questionnaire and the quality control of the PSA and pathology components. When comparing BPH and LUTS prevalence across populations, it is important to keep in mind that several factors substantially impact BPH detection and reporting of symptoms, and hence on the prevalence of BPH and LUTS. Because urinary symptoms are among the most common ailments affecting older men, these are likely to be the underlying reason for scheduling of many routine office visits. Therefore, studies of men coming in for routine office visits may over-report the prevalence of BPH and LUTS. Furthermore, such studies may not represent populations with poorer access to medical care. The current study’s population-based design avoids such criticisms.

Conclusions

We found that the prevalence of moderate-to-severe self-reported LUTS in Ghana was comparable to those in previous studies. However, it was lower than those of previous studies limited to African Americans, suggesting that racial disparities in LUTS prevalence observed in a previous US study⁷ may be due more to lifestyle rather than genetic factors. In addition, the prevalence of DRE-detected benign prostate enlargement was quite high in Ghana compared with Caucasian-American and African-American men, while prostate enlargement defined as PSA levels ⩾1.5 ng ml^–1 showed a prevalence comparable to that in the only other population-based study.⁹

In summary, while precise differences in the prevalence of BPH/LUTS between West Africans and African Americans may be difficult to quantify, our study shows that prevalence estimates of BPH and LUTS are comparable to those of other populations, indicating that these conditions are major health concerns among older men in West Africa. More research is needed to confirm the prevalence of BPH in Africans and clarify risk factors for BPH.