This was a cross-sectional hospital-based study conducted at the diabetology units of the Buea and Limbe Regional Hospitals Southwest Region of Cameroon.

The study population included adult diabetic male patient (>20 years) attending diabetic clinic at the Buea and Limbe Regional Hospital.

We excluded from the study

Sample size was calculated using.

$n=\frac{Z^{2}p\left(1-p\right)}{d^2}$

where:

n = minimum sample size, z = the standard normal variant at a confidence interval of 95% (standard value of 1.96), p = prevalence of ED among diabetic patients 41.5% [6] , d = the margin of error (0.05).

$n=\frac{{\left(1.96\right)}^2\left(0.41\right)\left(0.59\right)}{{(0.05)}^2}$

Therefore n = 335 which was the minimum sample size and were selected using a convenient sampling method.

Potential study participants were approached consecutively in the diabetology and in-patient units. The study’s purpose, procedure, risks and benefits were explained to each one in a language they best understood. An information notice about the study was given to the potential participant. They were allowed to ask any questions they had, after which they signed a consent form to participate in the study.

During recruitment, the principal investigator (PI) filled in a standardized structured questionnaire for each patient included in the study following an interview. Measures included socio-demographic data and medical history regarding duration of diabetes, type of diabetes, medications taken for diabetes; presence of co morbid conditions was obtained. Moreover, anthropometric measurements were recorded and blood for laboratory tests was taken.

HbA1c was analyzed using a portable HbA1c analysis, with the appropriate corresponding test strips, height was measured using a stadiometer and weight using an electronic scale from which the BMI was calculated using the formula Weight (in kilograms) divided by the square of the height (in meters). Obesity was defined as a BMI greater than or equal to 30. Blood pressure of the participants was measured using an automated blood pressure machine. Hypertension was defined as SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg.

Data obtained was analyzed using Stata and R version 3.5.3. The IIEF questionnaire has 15 questions, and the total score is obtained by the sum of the individual scores of each question. It addresses and quantifies five domains: erection function, orgasmic (ejaculation) function, sexual desire (libido), intercourse satisfaction (ability to sustain intercourse), and overall satisfaction/premature ejaculation domains. It classifies individual sexual function domains into mild, moderate or severe form depending on scores. Prevalence was calculated by dividing the number of patients found to have ED (IIEF score ≤ 21) by the total number of patients sampled.

Multivariate Logistic regression was used to identify and quantify the adjusted odds predictors of ED. These predictors were adjusted for age, education, gender, and some other demographic variables as well as medical data. A 2-tailed p-value of 0.05 was taken as statistically significant.

The ages of the 332 participants ranged between 21 and 76 years, with a mean age of 55 years. Most participants (64.46%) were married and 75.30% of them were Christians. One third were retired (30.12%) while close to half (42.77%) of the participants earned over 100,000 CFA francs a month ( Table 1 ).

About half (50.60%) of the participants actively consumed alcohol, while only 11.45% of them smoked cigarettes. 57.83% did not do regular physical exercise ( Table 2 ). The sexual behavior of 222 participants (66.87%) was low risk (they had only 1 sexual partner and did or did not always use condoms during intercourse). 92 participants (27.71%) recorded moderate risk sexual behavior (had more than one sexual partner and always or sometimes used condoms), while the remaining 18 participants (5.42%) recorded high risk sexual behavior (had more than one sexual partner and never used condoms). The majority (41.57%) of participants’ partner relationships had more conflict than comfort, while about half of them (57.23%) identified ED as a complication of the diabetes they had ( Table 2 ).

Most participants (290 participants, 87.35%) had type 2 DM ( Figure 1 ), with the duration of disease between 2 - 5 years being the most frequent. About a third of participants (31.07%) were on one or more other drugs such as anti-hypertensives, traditional therapy, highly active anti-retroviral drugs (HAART) or others (aspirin, statins, antacids) besides their routine anti-diabetic medications. 54.32% of participants had at least one chronic disease, with retinopathy and hypertension being the most frequent ( Table 3 ).

Most of the participants were overweight (43.90%). 71.69% had normal blood pressure, 68.67% had elevated blood sugar and 64.55% had elevated glycated haemoglobin ( Table 4 ).

Of the 332 participants, 262 participants had ED i.e. reported an IIEF score of 21 or less, which correlates to prevalence of 78.92%, at a 95% confidence interval and within a 6.21% error margin ( Table 5 ). Among those who had ED, the majority had mild to moderate ED, followed by mild, moderate then severe ED ( Figure 2 ).

In this study, determinants were grouped into 3: behavioural, biological and clinical domains. A logistic regression model was used to test for association between these domains and ED. Significance between a variable and ED will be computed with the use of Confidence Interval. Factors whose intervals crossed 0 were considered non-determinants of the outcome (ED), while factors whose intervals did not cross 0 were considered significant determinants of ED. The results show that for each unit of alcohol consume we expect a −0.1467 increase in the expected value of the outcome (ED) in the log odds scale, given that all of the other variables in the model are held constant. The level of partner relationship was investigated for its association with ED showed that neither having more comflict than comfort or equally both was significantly associated with ED. For each increase in conflict level, a −0.3943 increase in ED is observed. High risk sexual behaviour was associated with a 0.3240 increase in ED though not statistically significant ( Table 6 ).

Of the four biological determinants, age, FBS and HbA1c were found to be positive determinates of ED ( Table 7 ). The result showed that for every one-year increase in age, we expect a 0.04199 increase in the expected value (occurrence of ED, decrease in IIEF score) of the outcome (ED) in the log odds scale, given that all of the other variables in the model are held constant. For a one-unit (1 mg/dl) change in blood sugar (moving from elevated to normal blood sugar level), we

expect a −0.77085 decrease in the expected value of ED on the log odds scale, given all of the other variables in the model are held constant. Similarly, for HbA1c, we expect a −1.24354 increase in the expected value of ED, for every unit increase of the HbA1c value.

In this study, no medical factors investigated were significantly associated with ED. Though the results showed that DM type 2 was associated with an increased risk of 0.8458 for developing ED, likewise duration of diagnosis which was associated with a 1.12312 increased risk in developing ED and a risk of −0.02319 for developing ED in patients with comorbidity ( Table 8 ). However, none of these findings had a statistically significant association with developing ED in the study population ( Table 8 ).

ED is a diabetic complication that affects sexual, reproductive, physical and mental health and overall quality of care. In this study, we sought to determine its prevalence, determinants and some aspects of care, within the population of diabetic patients who attend the Limbe and Buea Regional hospitals.

Of the 332 participants included in our study, 262 had some form of ED, giving a prevalence of 78.92%. This prevalence is within the reported global range of 30% - 90%. The high value is like the results of other studies carried out in Ethiopia, Nigeria, Iran and other parts of the world [5] [6] [9] [13] . It is however lower than the 95% prevalence recorded by Kemp et al. [8] in South Africa. This difference could be because the latter carried out their study at a tertiary hospital, more likely to have patients with advanced complications of the disease. Also, their population was predominantly older (mean age = 62 years) compared to ours (mean age = 55 years). Our prevalence was higher than the 51% reported by Selvin et al. [14] in USA and Reuben et al. [9] in Tanzania. The US study used the one-item questionnaire which could have underestimated the prevalence while in Tanzania, socioeconomic and methodological differences could account for their lower prevalence.

Concerning the differential types of ED according to severity, our study revealed that most men with ED had mild-moderate and mild disease, while the minority had moderate and severe ED. This corroborates the findings of Seid et al. [6] but is opposite to what Kemp et al. [7] reported. In the latter’s study, most diabetic men with ED had severe ED. The fact that the latter’s study population was dominated by older participants could equally explain this difference.

From this study, we established that, alcohol consumption, cigarette smoking, and physical activity were not associated with occurrence of ED in this study. Same findings were reported by Ahmed et al. [2] in Egypt and Sharifi F et al. [11] in Iran. This however contradicts other studies [15] which recorded significant associations between each or both of these variables, with the occurrence of ED. This can be explained by the fact that the present study did not account on whether those who reported to be smokers were current smokers or had smoke in the past. It equally did not examine the relationship between amount smoked and erectile dysfunction in current smokers.

From scientific literature, risky sexual behaviour increases risk of Sexually Transmitted Infections (STI’s); which in turn increase risk for ED. However, our study failed to identify any association between ED and high-risk sexual behaviour among diabetic patients. This could be because STI’s in men are usually symptomatic, compared to women. Therefore, if they develop STI’s they are more likely to be aware and seek treatment; avoiding the situation of chronic, untreated STI’s which is responsible for ED.

The level of respondent’s relationship with their sexual partner and found that neither “more conflict than comfort” or “equally both” had an effect on the occurrence of ED from our study, suggesting that ED onset among diabetic patients is more organic than psychogenic and that conflicting partner relationships with diabetic patients is more the consequence than the cause of ED.

Regarding the biological factors studied, the presence and severity of ED was significantly higher among patients with elevated FBS and HbA1c, confirming the mediation of hyperglycemia in the pathogenesis of ED. These findings corroborate what is reported in literature [14] [16] [17] . The association between glycemic control and ED explains the high prevalence of ED recorded in our study, as over half of the respondents (64.85%) had poor glyemic control (using HbA1c cut-off of >7%).

We found increase in age to be a significant determinant of both ED occurrence and severity. This is constant across other scientific literature [6] - [8] [11] . The decline in male sex hormones as well as physiological testicular changes that come with age have been reported to be responsible for this association.

The type of DM the respondent had was not found to be a determinant of ED, in this study. Similar results were reported by Roth et al. [15] in Israel. However, Kemp et al. [7] found an association between DM type 2 and ED. This could be due to the difference in age groups affected by each type of DM. Type 1 DM patients are of a younger age group than type 2, and age itself is a determining factor for ED, so these could be why type 2 DM can appear to have an association with ED in contrast to type 1 DM.

The duration of disease was not associated with occurrence of ED in this study. This was not the case with either the studies by Ziaei-Rad et al. [13] in Iran or Zheng et al. [12] . Two reasons can explain these differences. Longer duration of diagnosis reported to be an association of diabetic ED only considered the duration of disease from a doctor’s diagnosis, which can be significantly shorter than the duration from the actual onset of disease; and it is this duration from disease onset that determines the duration of pathology that causes ED. Therefore, it could be that in our study, late diagnosis of DM (which is common in our setting), was masked as shorter duration of disease, which would mean that, both those with long and short disease durations may have had the same duration of pathology and hence the same chances for the occurrence of ED.

Even though a good number of respondents had either hypertension or retinopathy, our study found no significant association between the presence of these diseases and the occurrence of ED. The co-existence of these diseases with diabetes corroborates literature, but the absence of a predictive capacity of the co-morbidities for diabetic ED does not [16] [17] . This could be explained by the fact that the ability of a particular co-morbidity to influence ED is based on the duration and magnitude of its pathological process, determined by disease duration, control and medications amongst others. These factors occur differently across patients with the same co-morbidity and could be why the same co-morbidity may induce ED in one patient and not the other.