A cross-sectional study to estimate the point prevalence of painful diabetic neuropathy in Eastern Libya

Study design

This was a cross-sectional study that collected S-LANSS scores and indices of glycaemic control from a sample of patients with diabetes attending the Benghazi Diabetes Centre (BDC), Libya. The BDC has been operational for 48 years and is one of the largest diabetes centres in Libya, servicing the Eastern region of the country. The Centre employs 25 health care professionals, including five physicians dedicated to patient follow-up, with between 300 and 500 consultations each day and a throughput of approximately 100,000 patients per year [20, 21]. Services include reviewing patients’ medication, measuring fasting plasma glucose concentration, checking for complications of diabetes and referral to follow-up clinics for endocrine care, eye care, dental care and diet. There is no follow-up service for painful nor non-painful diabetic neuropathy and limited provision for foot care to prevent gangrene.

Ethical approval to conduct the study was granted from Leeds Beckett University research ethics committee and study-site permission from Benghazi Health Authority. The well-being, privacy, dignity and safety of participants were respected at all times in accordance with the Research Governance Framework and Policy and Practice of Research Ethics of Leeds Beckett University, and Medico-legal principles approved by the Libyan General Medical Council.

Sample population and sample size

The sample size was calculated using an online calculator (Soft Stat software) based on the minimum sample size needed to estimate the true population proportion with the required margin of error of 0.05% and confidence level at 95%. A sample of 400 participants was required to detect a relationship between the duration of diabetes (independent variable) and S-LANSS score (Arabic version, dependent variable). The level of statistical significance was set at 0.05 (two-sided). A target sample of 450 participants was set to account for attrition due, for example, to withdrawal of participants, with equal numbers of men and women matched for age into five subgroups: 18 < 30 years, 30 < 40 years, 40 < 50 years, 50 < 60 years and ≥ 60 years.

Receptionists and health care team members from the BDC advertised the study verbally. Potential participants were contacted in liaison with the BDC and an information form was distributed. Patients expressing interest in the study were given a participant information pack and were contacted 48 h later to be invited to attend one, one-hour study visit. The process of invitation, selection, inclusion/exclusion and enrolment is described in the study flow chart (Additional file 1).

Study visit

All eligible volunteers were enrolled onto the study by providing signed consent. Participants were interviewed by the Principal Investigator (SG) and provided data on anthropometrics, socioeconomics and education. If participants answered ‘yes’ to the question “Have you suffered from any pain in your feet in the last 6 months?” they were invited to complete the Arabic version of the S-LANSS pain scale. A fasting sample of venous blood was collected by a trained member of staff from the BDC.

Data management and analysis

Data was analysed using Statistical Package for Social Science (SPSS) version 22. Data is described as mean and standard deviation unless otherwise stated. Distribution of continuous variables were tested for normality using the kolmogrov-Smirnov statistic. Comparison between two continuous variables (e.g. duration of diabetes of men versus women) were examined using an unpaired t-test if the distribution of the variable was normal, and a Mann-Whitney statistic if the distribution was not normal. Differences between two proportions or two dichotomous variables were tested statistically using Chi-square or Z tests.

Initially, we calculated crude odd ratios of having PDN for groups of participants classified according to; sex, age group, Body Mass Index (BMI), duration of diabetes, fasting plasma glucose concentration and smoking for men only, as all women were non-smokers. The assignment of a reference group in all odds ratio calculations was based on previous literature to allow for meaningful comparisons [22]. We used a logistic regression model to test whether potential risk factors (sex, age, BMI, smoking history for men, duration of diabetes and fasting plasma glucose concentration) predicted probable neuropathic pain (i.e. a total S-LANSS score of ≥ 12). We chose logistic regression because the outcome variable was dichotomous and it would be possible to calculate adjusted odds ratios for all potential predictors. Each independent variable in the model had two categories (e.g. age = ≤ 60 years and > 60 years (old age); BMI = < 30 Kg/m²and ≥ 30 Kg/m² (obese); duration of diabetes = 5–15 years and > 15 years; and fasting plasma glucose concentration ≤ 125 mg/dl and > 125 mg/dl). In addition, the association between potential risk factors (i.e. collinearity) was tested using the Spearman correlation coefficient [23].

Characteristics of the sample

Two hundred and twenty six participants (50.2%) used insulin on its own for diabetes control. One hundred and sixty three participants (36.2%) used oral hypoglycaemic medications on their own and 61 participants (13.6%) used a combination of insulin and oral hypoglycaemic medication. Fig. 1 presents complications associated with diabetes.

One hundred and ninety five participants answered ‘yes’ to having suffered from pain in the foot in the last six months. Thirty seven of these 195 participants (19%) received over the counter medications (OTCs) without specifying names. Thirty three participants (17%) received paracetamol (PCM), 27 participants (13.8%) received PCM and non-steroidal anti-inflammatory drugs (NSAIDs), 22 participants (11.9%) received tramadol, 22 participants (10.7%) received co-codamol, 19 participants (9.7%) received NSAIDs, 13 participants (6.7%) received codeine, eight participants (4.1%) received herbal medicine, 5 participants (2.6%) received cupping, four participants (2%) received gabapentin and three participants (1.5%) received duloxetine. Three participants reported receiving no treatment at all (1.5%).

Prevalence of pain with neuropathic characteristics

Of the 450 participants in our sample, 195 (43.3%) reported pain in the previous 6 months. All of these participants reported that this pain was in their lower limbs. Of the 195 participants reporting pain, 190 scored ≥ 12 on S-LANSS. Thus, 42.2% (190/450) of participants with diabetes were categorised as experiencing pain with neuropathic characteristics. Interestingly, only seven of the 195 (3.6%) participants with pain reported that they had been clinically diagnosed with painful diabetic neuropathy.

There was no statistically significant differences in the prevalence of pain with neuropathic characteristics between women and men (odds ratio (OR) = 0.97, 95% CI = 0.67–1.4). Risk factors for the development of PDN were more than 60 years of age (OR = 1.93, 95% CI = 1.22–3.04), longer duration of diabetes (OR = 23.41, 95% CI = 14.26–38.46), had smoked tobacco for more years (OR = 5.60, 95% CI = 3.11–8.60), had higher fasting plasma glucose levels (OR = 3.11, 95% CI = 2.1–4.6), had higher BMI (OR = 4.56, 95% CI = 3.1–6.8) and had higher employment status.

Multivariate (adjusted) odd ratio

Reasons for the differences in estimates between countries

There are variations in the estimates of the prevalence of PDN between and within countries in research literature. Reasons include disparities in study methodologies, sample characteristics, eligibility criteria for co-morbidities, treatments to manage diabetes and operational definitions and measurement techniques to identify neuropathic pain [27–29]. For example, a cross-sectional study conducted in the UK using a postal survey estimated that 64% of individuals with diabetes experienced neuropathic elements but this figure declined to 30% when these respondents were assessed using a neurological examination [27]. There has been a debate about the possibility of imprecise estimates of prevalence due to the use of neuropathic pain screening tools rather than full neurological examination [30–33]. Symptoms-based questionnaires including S-LANSS are widely used to screen for probable neuropathic characteristics by researchers and health care professionals prior to further examination if required. Thus, we decided to use the S-LANSS as it would enable direct comparison with previous studies. We intend to conduct a study using full neurological examination in the future [33].

Risk factors for the development of PDN

The main risk factors for PDN identified in our study were increased duration of diabetes (> 15 years), obesity (BMI ≥ 30 Kg/m²), high fasting plasma glucose concentration (> 125 mg/dl) and smoking (only for men). This is consistent with research findings from other studies in the MENA region [12–15, 24, 34, 35] and worldwide [36]. Other risk factors not measured in our study include hypertension and high cholesterol. Evidence also suggests that there is a higher incidence PDN in individuals with prolonged impaired glucose intolerance before diabetes has been diagnosed [37–39]. Socioeconomic factors including lifestyle, diet, and health care policies such as food labelling and the availability low glycaemic food items facilitating blood sugar level control contribute to higher incidence of diabetes and associated complications and may be one reason for country and regional differences in the prevalence of PDN [29].

Higher levels of education and employment were also associated with higher percentages of people with PDN possibly because they are more likely to have sedentary jobs and higher BMI.

Clinical implications of the findings

In our study, only 3.6% of individuals with PDN reported that they had received a diagnosis of PDN from their clinician and this was from consultations outside of Libya (Tunisia and Egypt). We speculate that clinicians in Libya may not be aware of the presence of PDN in patients or that they consider PDN to be an inevitable consequence of diabetes and not worthy of specific consideration and/or treatment. Clearly, there is a need to raise awareness of the importance of diagnosing and managing PDN including the consequence of inadequate control of diabetes in future policy. Our findings provide evidence that long-term glycaemic control is critical to reduce the incidence of PDN and that there is inadequate management of pain. This is not unique to Libya. For example, Daousi et al. reported that PDN was inadequately managed in more than 40% of individuals in the United Kingdom [29, 40], with similar findings in Canadian populations [41].

Strong evidence from systematic reviews suggest that PDN can be managed using pharmacological interventions including tricyclic antidepressants and gabapentin or pregabalin as first-line treatments and serotonin-norepinephrine reuptake inhibitors or opioids as second-line treatments [3, 12]. This approach has been endorsed by professional bodies, including the International Association for the Study of Pain (IASP) and the National Institute for health and Care Excellence (NICE) and has been adapted for use in the MENA region [12, 42]. However, treatments received by participants in our sample suggest that this approach was not implemented, with only four participants receiving a first-line drug (gabapentin) and 53 a second-line drug (either tramadol, co-codamol or codeine). Of the remaining participants who received treatment for their pain, some received analgesics not suitable for the neuropathic pain or alternative therapies such as herbal medicine or cupping.

Strengths and shortcomings of the study

This is the first attempt to estimate the prevalence of PDN in individuals living in Libya using an appropriately powered study with adequate sample size. The absence of diagnosis based on a full neurological examination is a limiting factor, although we estimated that the use of our validated Arabic version of the S-LANSS pain scale would identify at least 75% of individuals with PDN [43]. We measured fasting plasma glucose concentration rather than haemoglobin A1c (HbA1c) which is a more robust indicator of blood sugar control. We did not measure blood pressure, triglycerides and cholesterol levels and such data would have enabled determination of their role as probable risk factors.

The sample may be subject to selection bias because the study site clinic only services the Eastern region of Libya. However, there are no primary services for individuals with diabetes in Libya so individuals have to attend regional specialised clinics. Thus, the sample is likely to have captured all people with diabetes who were willing and able to attend the clinic. Nevertheless, we express caution in the generalisability of our estimate because of the use of a single-site and the impact of unmeasurable confounders.

Future directions

It is hope that our findings will be used to inform future policy for the diagnosis and management of PDN associated with diabetes in Libya, including the creation of a national prevention program [44]. Future epidemiological research is needed from resource limited countries to provide a more a balanced picture of the problem of PDN globally. There is an increasing focus on epigenetics to identify individuals susceptible to PDN [45], and studies investigating the incidence of microvascular pathology preceding diagnosis of diabetes would provide valuable insights into the pathogenesis of PDN. Microvascular pathology is known to contribute to neuropathy and may have clinical utility as a predictor of PDN.