Characteristics influencing high-risk sexual behaviours in elderly men

Abstract

With the objective of investigating the characteristics influencing high-risk sexual behaviours in elderly men (60–74 years of age) in Chongqing, China, a total of 1433 healthy elderly men with sexual intercourse frequencies of one to six times/month who were willing to participate in the questionnaires were studied at four hospitals. We measured serum testosterone levels and performed follow-ups every six months, with a total of 1128 elderly men followed up after two years. We also investigated socio-economic and demographic characteristics (age, education, income, location, marital status and number of marriages), types of sexual partners, age differences with fixed sexual partners, frequency of sexual intercourse, combined basic age-related diseases, sexually transmitted infections (STIs) education, elderly self-care ability and high-risk sexual behaviours (frequency of sexual intercourse and number of sexual partners) using questionnaires. We analysed the influencing factors of high-risk sexual behaviours in elderly men using a univariate analysis, multivariate logistic regression analysis, BP neural network prediction and cluster analysis. Finally, we found that serum total testosterone, age, types of sexual partners, age differences with fixed partners and frequency of sexual intercourse are five factors that influence high-risk sexual behaviours in elderly men.

Keywords

Elderly men high-risk sexual intercourse influencing factors sexually transmitted infections

Introduction

As China’s population is ageing rapidly, sexually transmitted infections (STIs) are affecting the elderly, especially elderly men, and are mostly occurring through extramarital sex.^1,2 People in China mainly hold to the concept that sexual behaviours are primarily activities of young adults, while elderly men have few or even no sexual behaviours. However, it is normal for elderly men to have sexual behaviours, and when their partners cannot participate in sexual behaviours with them due to menopause, they may very likely find other sexual partners (especially sex workers), resulting in high-risk sexual behaviours. For the prevention and treatment of STIs in the elderly, most of the literature and reports have assessed combined therapy, STI education and control over female sex workers, while paying less attention to high-risk sexual behaviours among the elderly, especially those of elderly men. Serum testosterone levels are closely related to sexuality and sexual functions in men, including elderly men.^3,4 They peak at approximately 17 years of age in men and then begin to decrease until old age,⁵ when testosterone levels show a gradual decline.⁶ Studies have shown that testosterone levels drop by 1.6% annually in men aged 40–70 years;⁷ that is, serum testosterone levels in elderly men are stable to some extent for years, with the exclusion of exogenous supplementation, which provides a basis for the use of testosterone levels in this study and objective prospective studies on the social characteristics of these men, such as their ages, education levels and incomes.

In this study, we considered serum testosterone levels as the biological variable, and age, education, income, location, marital status, number of marriages, types of sexual partners, age differences with fixed sexual partners, frequency of sexual intercourse, combined basic age-related diseases, STI education and elderly self-care abilities as the factors related to social demographics. This study included the above 13 variables to identify the factors influencing high-risk sexual behaviours and analysed the characteristics of these factors. At the same time, we propose a hypothesis that the occurrence of high-risk sexual behaviours in elderly men is due to an imbalance between the sexual supply of elderly women with the sexual needs of elderly men, which the former represented by the age difference with fixed partners, and the latter represented by the serum total testosterone. Age plays a role in this balance, while changes in the sexual supply and sexual needs occur with age. Meanwhile, the sexual frequency is a reflection on the other side of high-risk sexual behaviours. Through these factors, government can develop more effective measures to facilitate prevention and interventions for high-risk sexual behaviours among elderly men.

Participants and methods

Participants

A total of 1433 healthy elderly men with sexual intercourse frequencies of one to six times/month (with only heterosexual intercourse considered) who were willing to complete the questionnaires were recruited from four hospitals (The Thirteenth People’s Hospital of Chongqing, The People’s Hospital of Chongqing, The Yongchuan Affiliated Hospital of Chongqing Medical University, and The Fourth People’s Hospital of Chongqing) from 1 September 2015 to 31 December 2015. The participants were 60–74 years of age with a median age of 64.4 ± 2.7 years. They were followed up every six months for two years. We had each participant complete an informed consent form that stated that their privacy would be tightly protected. Ethical approval was given by the medical ethics committee of The 13th People's Hospital of Chongqing with reference number 2015(2–2).

The sample size in this study was based on the composition ratio of high-risk participants recognized in the pre-test of 24.05%, an allowable error of 2.5% and an α of 0.05. The sample size was then calculated using a simple random sampling formula. Through this calculation, the prospective sample size for the final questionnaire after two years of follow-up was found to be ≥1100 participants. A possible loss rate of 20% was considered, so the final sample size for the questionnaires was 1400.

Note that the specific definition of elderly health was (1) age ≥60 years; (2) able to handle self-care or with only mild dependency according to the Assessment Table of Self-Care Ability of the Elderly issued by the Chongqing Municipal Health and Family Planning Committee, in which scores of 0–3 represent self-care ability, scores of 4–8 indicate mild dependency, scores of 9–18 indicate moderate dependency, and scores ≥19 indicate an inability to provide self-care; (3) no chronic age-related diseases, or only diseases for which symptoms can be controlled and daily life remains normal without treatment or with only oral medications; (4) having sex at least once a month.

High-risk sexual behaviours were defined according to the WHO⁸ and generally refer to any sex-related behaviours that increase the risk of adverse outcomes, including multiple sexual partners, unprotected sex and anal sex. In this study, according to the actual situation, it specifically refers to sexual behaviours with sexual partners of the opposite sex who are married or cohabitating and those performed without a condom.

Methods

A 5-mL blood sample was drawn from all subjects via a vein in the elbow at 8–9 am on each day of testing. The blood was placed in a refrigerator for no longer than one week and was checked for serum testosterone levels. All of the samples were handled by the same group of professionals with the same equipment. Questionnaires (including those regarding sexual behaviours) and personal health surveys were produced by the Chongqing Municipal Health and Family Planning Committee. After unified training, the investigators distributed questionnaires on the 12 social characteristics, including age, education, income, location, marital status, number of marriages, types of sexual partners, age differences with fixed partners, combined basic age-related diseases, STI education and elderly self-care ability and then divided these variables into low, medium and high groups according to their levels (defined criteria). All participants refrained from any sex-related hint and exogenous testosterone supplementation and were followed up by phone or onsite visits every six months for two years.

Statistical analysis

All statistical analyses were performed using SAS 9.2 and MATLAB software. In the present study, we performed a multiple logistic regression analysis with the dependent variable of high-risk sexual behaviours and the independent variables of the 13 variables above. We also conducted BP neural network which was trained and tested with the meaningful independent variables as the inputs and the frequency of high-risk sexual intercourse as the output, selected different values with the meaningful independent variables, formed new datasets, predicted the frequency of high-risk sexual intercourse with the trained BP neural network and finally conducted a clustering analysis to explore the basic characteristics of the variables influencing high-risk sexual behaviours in elderly men.

Results

General information

A total of 1433 subjects were willing to participate in the questionnaires. After two years of follow-up, 1128 cases finally completed the questionnaires with a loss rate of 21.28% (305/1433). Relevant information about social demographic characteristics and high-risk sexual behaviours in elderly men is shown in Table 1.

Table 1.

Social demographic characteristics and relevant information about high-risk sexual behaviours in elderly men.

Variables	Overall	With high-risk sexual behaviours, 265 participants	Without high-risk sexual behaviours, 863 participants	Statistics	P
Social demographic characteristics
Age (years)				Z = 1.0862	0.2773
60–64	692 (61.4)	160 (60.4)	532 (61.7)
65–69	290 (25.7)	57 (21.5)	234 (27.1)
≥70	145 (13.9)	48 (18.1)	97 (11.2)
Education level				Z = −2.6171	0.0089
Primary and junior high school	382 (33.9)	107 (40.4)	276 (32.0)
Technical high school and secondary school	571 (50.7)	125 (47.2)	447 (51.8)
College	173 (15.4)	33 (12.5)	140 (16.2)
Economic income				Z = −1.3837	0.1664
Low income	163 (14.5)	42 (15.9)	122 (14.1)
Middle income	541 (48.0)	133 (50.2)	408 (47.3)
High income	423 (37.5)	90 (34.0)	333 (38.6)
Location				$χ^{2}$ =10.0776	0.0065
Rural	170 (15.1)	44 (16.6)	126 (14.6)
Urban–rural integration	236 (20.9)	57 (21.5)	179 (20.7)
Urban	722 (64.0)	164 (61.9)	558 (64.7)
Marital status				–	0.0001
Married	903 (80.1)	192 (72.5)	711 (82.4)
Divorced or widowed	223 (19.8)	72 (27.2)	151 (17.5)
Separated	2 (0.2)	1 (0.3)	1 (0.1)
Number of marriages				Z = −4.766	0.0001
1	1010 (89.6)	258 (97.4)	752 (87.1)
2	107 (9.5)	7 (2.6)	100 (11.6)
3	11 (0.9)	0 (0.0)	11 (1.3)
Number of underlying diseases of the elderly				Z = −1.5877	0.1123
0	123 (10.9)	50 (18.9)	73 (8.5)
1–2	834 (73.9)	166 (62.6)	668 (77.4)
≥3	171 (15.2)	49 (18.5)	122 (14.1)
Elderly self-care ability				Z = −7.5269	0.0001
Has abilities for daily life	421 (37.3)	149 (56.2)	272 (31.5)
Mild dependence	765 (50.1)	96 (36.2)	469 (54.4)
Moderate dependence	142 (12.6)	20 (7.6)	122 (14.1)
Relevant information about high-risk sexual behaviours in elderly men
Number of high-risk sexual behaviours				Z = 33.2068	<0.0001
0	863 (76.5)	0 (0.0)	863 (100.0)
1	100 (8.9)	100 (37.7)	0 (0.0)
2	78 (6.9)	78 (29.4)	0 (0.0)
≥3	87 (7.7)	87 (32.8)	0 (0.0)
Number of high-risk sexual partners				Z = 33.2207	<0.0001
0	863 (76.5)	0 (0.0)	863 (100.0)
1	106 (9.4)	106 (40.0)	0 (0.0)
2	109 (9.7)	109 (41.1)	0 (0.0)
≥3	50 (4.4)	50 (18.9)	0 (0.0)
Types of sexual partners				$χ^{2}$ = 3.9717	0.0463
Spouse	886 (78.6)	189 (71.3)	697 (80.8)
Non-spouse	229 (20.3)	63 (23.8)	166 (19.2)
Non-fixed partner	13 (1.2)	13 (4.9)	0 (0.0)
Age difference with fixed partners (years)				Z = −5.5113	<0.0001
<5	397 (35.6)	121 (47.6)	276 (32.0)
5–10	490 (43.9)	106 (42.1)	384 (44.5)
>10	229 (20.5)	26 (10.3)	203 (23.5)
Times they had received STI-related education				Z = 3.3139	0.0009
0	23 (2.0)	5 (1.9)	18 (2.1)
1–2	466 (11.3)	86 (32.5)	380 (44.0)
≥3	639 (56.7)	174 (65.7)	465 (53.9)
Frequency of sexual intercourse (times/month)				Z = 16.7235	<0.0001
1	202 (17.9)	0 (0.0)	202 (23.4)
2	344 (30.5)	32 (12.1)	312 (36.2)
3	159 (14.1)	7 (2.6)	152 (17.6)
≥4	423 (37.5)	226 (85.3)	197 (22.8)

STI: sexually transmitted infection.

Multiple logistic regression of variables influencing high-risk sexual behaviours in the elderly

We performed multiple logistic regression explaining $R^{2}$ =0.56 with the dependent variable of high-risk sexual behaviours and the independent variables of the above 13 variables (α-input = 0.05, α-output = 0.05). This result is showed in Table 2.

Table 2.

Multiple logistic regression of variables influencing high-risk sexual behaviours in elderly men.

Variable	b	S	$χ^{2}$	P	OR (95% CI)
Total serum testosterone (1 nmol/L)^a	1.1826	0.1076	156.8593	<0.0001	3.263 (2.643–4.029)
Age (years)
60–64^b
65–69	1.6380	0.3844	18.1570	<0.0001	5.145 (2.422–10.929)
≥70	2.7778	0.4363	40.5375	<0.0001	16.083 (6.839–37.821)
Types of sexual partners
Spouse^b
Non-spouse	1.0580	0.3567	8.7969	0.0030	2.880 (1.432–5.795)
Age difference with fixed partners (years)
≥5^b
<5	1.6365	0.3265	25.1193	<0.0001	5.137 (2.709–9.742)
Frequency of sexual intercourse (times/month)
<4^b
≥4	1.9832	0.3317	35.7484	<0.0001	7.266 (3.793–13.919)

^aTotal serum testosterone as a continuous variable without segmentation.

^bAs references.

Prediction of the frequency of high-risk sexual behaviours in the elderly using a BP neural network

No favourable results were obtained using both traditional linear and nonlinear predictions at the beginning of the study. The BP neural network belongs to the category of machine learning and has strong self-learning and generalization abilities. As it can be used for predictions, a BP neural network was ultimately used here. The neural network training and prediction in this study were performed using MATLAB software. We constructed a BP neural network with the inputs of the five variables entered into the logistic regression model (total serum testosterone, age, types of sexual partners, age differences with fixed partners and frequency of sexual intercourse), and the output was the frequency of high-risk sexual intercourse in elderly men. From the 1128 participants, 800 were randomly selected as the training set, and the other 328 participants were used for the test set. A 5–3–1 topological structure was constructed. The number of nodes in the input and output layers were five and one, respectively, while three nodes of hidden layer were chosen following many experiments. The transfer function was tansig in the hidden layer and purelin in the output layer. The maximum number of iterations was 100, the learning rate was 0.1 and the target error was 0.00004. The training was completed on the 24th iteration, and then the simulation test was performed with the test set at an accuracy rate of 80.49%. We chose possible values for the five variables selected above to create a new dataset of 1377 participants. Specific valuations are shown in Table 3. Predictions were made with the new dataset using the trained BP neural network.

Table 3.

Valuation of variables.

Variable	Valuation
Total serum testosterone (nmol/L)	10, 10.5, 11, 11.5, 12, 12.5, 13, 13.5, 14, 14.5, 15, 15.5, 16, 16.5, 17, 17.5, 18
Age	1=60–64 years, 2=65–69 years, 3= ≥70 years
Types of sexual partners	1=spouse, 2=non-spouse, 3=non-fixed partner
Age differences with fixed sexual partners	1 = <5 years, 2 = 5–10 years, 3 = >10 years
Frequency of sexual intercourse	1=1–2 times/month, 2=3–4 times/month, 3=>4 times/month

Clustering analysis of predicted values

Clustering analysis results

We performed a clustering analysis of the predicted results of the frequency of high-risk intercourse in the elderly using a BP neural network with a stepwise clustering method in SAS 9.2 software. The number of types was determined to be three. Specific results are shown in Table 4 and Figure 1 (have been converted to dimensional figure). As can be seen in Table 3, the frequency of high-risk sexual intercourse was low for type 1, medium for type 2 and high for type 3 men.

Figure 1.

Results of the clustering analysis.

Table 4.

Results of the clustering analysis.

Type	Participants	Mean frequency of high-risk sexual intercourse
Type 1	749	0.55
Type 2	496	1.90
Type 3	132	3.46

The analysis of the clustering results

We calculated the percentages of the total serum testosterone levels, ages, types of sexual partners, age differences with fixed partners and frequencies of sexual intercourse in each type, and then conducted a univariate analysis. The results are shown in Table 5. As can be seen, no significant differences were found in the frequency of sexual intercourse among the three types (P = 0.4489). For type 1 men (low frequency), serum testosterone levels were mainly <14 nmol/L (82.91%), the age groups were mainly 60–64 years (38.58%) and 65–69 years (33.78%), the types of sexual partners were mainly spouses (40.45%) and age differences with fixed sexual partners were mainly >5 years (73.66%). For type 2 men (medium frequency), serum testosterone levels mostly exceeded 14 nmol/L (94.56%) and the men were more often aged ≥70 years (36.49%); the types of sexual partners were mainly non-spouses (36.29%) and non-fixed partners (35.08%), and age differences were <5 years (35.48%). For type 3 men (high frequency), serum testosterone levels were only found to be between 16 and 18.8 nmol/L (100.00%), the ages were mainly over 70 years (53.79%), the types of sexual partners were mainly non-fixed partners (56.82%), and age differences were mainly <5 years (48.48%).

Table 5.

Univariate analysis of the clustering results.

Variable	Type 1	Type 2	Type 3	$χ^{2}$	P
Serum testosterone level (nmol/L)				-(Fisher’s exact test)	<0.0001
10–11.5	324 (43.26)	0 (0.00)	0 (0.00)
12–13.5	297 (39.65)	27 (5.44)	0 (0.00)
14–15.5	115 (15.35)	209 (42.14)	0 (0.00)
16–18	13 (1.74)	260 (52.42)	132 (100.00)
Age (years)				44.9876	<0.0001
60–64	289 (38.58)	148 (29.84)	22 (16.67)
65–69	253 (33.78)	167 (33.67)	39 (29.55)
≥70	207 (27.64)	181 (36.49)	71 (53.79)
Types of sexual partners				65.8098	<0.0001
Spouse	303 (40.45)	142 (28.63)	14 (10.61)
Non-spouse	236 (31.51)	180 (36.29)	43 (32.58)
Non-fixed partner	210 (28.04)	174 (35.08)	75 (56.82)
Age differences with fixed sexual partners (years)				25.2335	<0.0001
<5	219 (29.24)	176 (35.48)	64 (48.48)
5–10	251 (33.51)	166 (33.47)	42 (31.82)
>10	279 (37.25)	154 (31.05)	26 (19.70)
Frequency of sexual intercourse (times/month)				3.6945	0.4489
1–2	245 (32.71)	161 (32.46)	53 (40.15)
3–4	249 (33.24)	167 (33.67)	43 (32.58)
>4	255 (34.05)	168 (33.87)	36 (27.27)

Discussion

In China, on the topic of the prevention and treatment of STIs in the elderly, most of the literature and reports have emphasized combined therapy, STI education (cut off the route of transmission) and control over female sex workers (control the source of infections).^9,10 However, there have been few controls or interventions aimed at elderly men (the susceptible population) with a propensity for high-risk sexual behaviours. Elderly men with a potential propensity for high-risk behaviours are difficult to determine in reality. Given that high-risk sexual behaviours can increase the incidence of STIs,¹¹ identifying and intervening in this susceptible population is the key to the prevention and treatment of STIs in the elderly. The discovery and summary of the most basic characteristics of the susceptible population as well as the further construction of an ideal prediction model are very practical for preventing and treating STIs in the elderly, especially elderly men. However, analyses and summaries require not only sufficient, informative basic data but also scientific, reasonable statistical methods that are able to avoid the omission of necessary basic variables and remove unnecessary intermediate ones.

As the prospective study predominated in this study, we first found five meaningful variables among the 13 variables using a multiple logistic regression analysis, then performed a prediction analysis for the number of high-risk sexual behaviours using a BP neural network and a newly constructed dataset, and ultimately analysed the predicted values using a clustering analysis. Compared with other prediction methods, a BP neural network has a mapping function from inputs to outputs. Moreover, its ability to achieve complex nonlinear mapping has been proven by mathematical theory, enabling it to solve problems with complex internal mechanisms. A BP neural network can automatically extract the output and the ‘reasonable rules’ among the outputs through learning and adaptively store the learning content in the weights of the network, that is, a BP neural network has great self-learning and self-adaptability abilities and can apply the learning achievements to new knowledge.¹² Since neither traditional linear prediction nor non-linear prediction obtained favourable results at the beginning of the study, we ultimately chose a BP neural network for the prediction portion, as it has all of the above advantages and obtains ideal prediction results.

In the combined multivariate logistic regression with cluster analysis, we can see that serum total testosterone, age, types of sexual partners, age differences with fixed partners and frequency of sexual intercourse are several influencing factors affecting high-risk sexual behaviours in elderly men. We found the situation that in cases of higher serum total testosterone, higher ages, greater age differences with fixed sexual partners, higher frequencies of sexual intercourse and sexual partners who are non-partners, these elderly men are more likely to have high-risk sexual behaviours. Among these, the increase in high-risk sexual behaviours with increasing age seems to be a surprise. However, we found that as the age increases, the proportion of the fixed partners who are spouses decreases and the proportion of non-spouses increases. The proportion of non-spouses in the group of 60 year olds accounts for 17.06% of them; in the group of 65 year olds, this was 22.30%, and in the group of those ≥70 years old, this was 34.04%. The difference was statistically significant (P < 0.0001). As ages increase, changes in sexual partners among elderly men due to illness or divorce will appear more and more. On the other hand, a higher age can cause a decline in cellular and humoral immune function in elderly men.¹³ STIs can easily be transmitted in older men who have just one or two episodes of high-risk sexual intercourse with a non-spouse.

Decaroli MC et al.¹⁴ found that serum total testosterone in the elderly is inversely related to comorbidities and high prevalence rates of chronic diseases in the general population. Rochira et al.¹⁵ found that low testosterone is associated with poor health status in men with HIV, which is the same as the findings in our research. At first, we thought that elderly self-care ability and the number of underlying diseases, which were used to represent the health status, might enter the model, but they do not. The reasons are as follows: the Spearman correlation coefficients for the frequency of sexual intercourse with the number of underlying elderly diseases and self-care ability in this study were –0.15242 and 0.41711, respectively. In another subsequent sub-study, we used the score from the health status scale to assess health status, and its correlation coefficient with the frequency of sexual intercourse was 0.79219. A good health status may favour having sex more often and with several partners. Elderly men with poor health statuses are less predisposed to sexual activity. The analysis showed a high collinearity between the frequency of sexual intercourse and health status, so the frequency of sexual intercourse entered the model, while the number of underlying diseases and self-care ability did not.

Based on more than 20 years of clinical experience and these results, we propose the hypotheses that the serum total testosterone of elderly men can represent the sexual needs of elderly men, that the types of sexual partners and the age differences with fixed partners can represent the sexual supply of elderly women and that the sexual supply and sexual needs change with increasing ages. The frequency of sexual intercourse is another reaction to high-risk sexual behaviours. The occurrence of high-risk sexual behaviours in elderly men is due to the imbalance between the sexual supply and sexual needs. It can be seen from Table 5 that the serum total testosterone of type 2 (medium frequency) and type 3 (high frequency) men are usually 14 nmol/L or more, relative age differences with fixed sexual partners are mostly <5 years and sexual partners are mainly non-spouses or not fixed sexual partners. According to Morley JE et al.’s research,¹⁶ when the male reproductive system ages, the decrease in sex hormone secretion is constant but slow; however, when the female reproductive system ages, the secretion of sex hormones declines very rapidly, manifesting as menopause, with a series of menopausal symptoms. It is not difficult to understand that among elderly couples with small age differences, the sexual supply of elderly women is obviously insufficient, but the sexual needs of elderly men are not significantly reduced (serum testosterone levels decline by 1.6% per year in elderly men⁷). With increasing ages, the serum total testosterone in elderly men can change. For instance, serum total testosterone declines with time (not obviously) and age differences with fixed sexual partners and types of sexual partners may also change. For example, an elderly man may change sexual partners due to a divorce or a spouse’s illness, and the age difference with the fixed partner will also change. In more than 20 years of clinical work, we found that the main cause of high-risk sexual behaviours in elderly male STI patients is disharmonious sexual lives among elderly couples. The surface phenomenon shows that the age difference between men and women is not large, generally not exceeding five years. However, among the elderly couples who are on their second or more marriages, it is uncommon for males to have high-risk sexual behaviours. Through conversations, it was found that the sexual lives of these elderly men and women were clearly harmonious, and the age differences (male>female) were often greater than 10 and even 20 years. Therefore, we can explain that the occurrence of high-risk sexual behaviours among elderly men is related to the dynamic changes in sexual needs and the sexual supply. However, if this hypothesis can be proven, it requires a subsequent study with larger sample sizes and additional indicators that can better represent the sex supply of women.

There were also some shortcomings in this study. First, this study did not consider the impacts of psychosocial factors, such as whether adverse childhood events, mental illness, self-esteem and social networks were involved. Psychosocial factors may be an important factor in risky sexual behaviours. We have planned follow-up studies to explore it as a specialized research point. Second, this study only targeted heterosexual couples and did not consider homosexual behaviour at all. Homosexual behaviour is also a cause of STIs, but based on China's national conditions, homosexuality is very rare among elderly men. It is difficult and costly to find subjects in this population, so it has not been considered. Finally, oral sex was not considered. Oral sex is also an important intermediary for STIs. Even if the type of sexual intercourse among the elderly in China was generally vaginal intercourse, the impact of oral sex on high-risk sexual behaviours should not be ignored. We will include it in subsequent studies.

Footnotes

Authors’ contributions

Gui-Yuan Xiao was responsible for the statistical processing and the writing of the manuscript; Bin Peng was involved in the study design and contributed to the entire process of this project; collection, collation and analysis of the data were performed by Bin Peng, Yu Gong, Hong Yi, Feng-Lan Zhao, Deng-Zhi Lv, Shan Geng, Rong-Li Bai, Yu Jia, Shan-Chuan Lei, Wei-Kang Zhou; coordination of various departments and the allocation of funds were performed by Rong-Li Bai; Yu Jia, Shan-Chuan Lei, Wei-Kang Zhou developed the study; Ying Hu, Dou Qu, Xiao-Ni Zhong were responsible for the statistical analysis; Xiao-Ni Zhong and Shan-Li Wang provided assistance with the design of the study and the revision of the manuscript; Xia Jing was responsible for testosterone testing; blood collection, sample delivery and assistance with testosterone testing were performed by others.

Acknowledgements

We thank all of the people who participated in our project.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funds for this study were provided by the Chongqing Municipal Health and Family Planning Committee for the medical scientific research project (Effects of Biological and Social Characteristics on High-risk Sexual Behaviours in Elderly Men, Project No. 2015ZDXM025) (100,000) and The Thirteenth People’s Hospital of Chongqing (100,000).

Ethical approval

Ethical approval was given by the medical ethics committee of The 13th People's Hospital of Chongqing with reference number 2015(2–2).

Patient consent

Patient consent was obtained.

References

Pen

Lei

S-C

Gong

Clinical analysis of 48 elderly male sexually transmitted diseases mixed infections in the area of Chongqing railway. Lab Med Clin 2013; 10: 1950–1952.

Yun

Zhang

Z-K

Wang

F-C

et al . Epidemiological analysis of sexually transmitted diseases among the elderly in Tangshan City from 2012–2015. Occup Health 2017; 33: 1097–1101.

Travison

Morley

Araujo

et al . The relationship between libido and testosterone levels in aging men. J Clin Endocrinol Metab 2006; 91: 2509–2513.

Araujo

Wittert

GA.

Endocrinology of the aging male. Best Pract Res Clin Endocrinol Metab 2011; 25: 303–319.

Elmlinger

Kühnel

Wormstall

et al . Reference intervals for testosterone, androstenedione and SHBG levels in healthy females and males from birth until old age. Clin Lab 2005; 51: 625632

Guo

Y-L

H-J.

Male climacteric syndrome. 1st ed. Beijing: China Medical Science Press, 2005, p.58.

Feldman

Longcope

Derby

et al . Age trends in the level of serum testosterone and other hormones in middle-aged men: longitudinal results from the Massachusetts male aging study. J Clin Endocrinol Metab 2002; 87: 589–598.

World Health Organization. Sexual and reproductive health, www.who.int/reproductivehealth/topics/sexualhealth/sh_definitions/en/ (accessed ■).

Liu

Calzavara

Mendelsohn

et al . Impact evaluation of a community-based intervention to reduce risky sexual behaviour among female sex workers in Shanghai, China. BMC Public Health 2015; 15: 14–147.

10.

Wang

Yin

et al . The effect of a structural intervention for syphilis control among 3597 female sex workers: a demonstration study in South China. J Infect Dis 2012; 206: 907–914.

11.

Goesling

Colman

Trenholm

et al . Programs to reduce teen pregnancy, sexually transmitted infections, and associated sexual risk behavior: a systematic review. J Adolesc Health 2014; 54: 499–507.

12.

The parallelism and application in data mining of BP algorithm. Chongqing: Chongqing University, 2003.

13.

Gruver

Hudson

Sempowski

Immunosenescence of ageing. J Pathol 2007; 11: 144–156.

14.

Decaroli

Rochira

Aging and sex hormones in males. Virulence 2017; 8: 545–570.

15.

Rochira

Diazzi

Santi

et al . Low testosterone is associated with poor health status in men with human immunodeficiency virus infection: a retrospective study. Andrology 2015; 3: 298–308.

16.

Morley

Kaiser

Sih

et al . Testosterone and frailty. Clin Geriatr Med 1997; 13: 685–695.