Of 6140 participants in a baseline cohort study conducted among individuals aged between 10 and 90 years-of-age, data from 5052 participants aged 18 or older were analyzed in the present study. The baseline study was carried out in Amol, a densely-populated city in northern Iran. Local health centers, where almost all study participants had health record files, were used to collect the data. Sampling for the cohort study is explained elsewhere[20]. A schematic diagram of study participants and exclusion criteria is shown in Figure 1.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Schematic diagram of study participants and exclusions criteria.

Weight, height, waist circumference, hip circumference, and blood pressure were measured in health centers, where trained healthcare staff members were responsible for data collection. Height was measured while the participants were standing with their heels and buttocks pressed up against a wall. Waist circumference was determined at the midpoint between the lowest costal ridge and the upper border of the iliac crest. The largest circumference between waist and knee was determined as hip circumference. Measurements were performed with a non-stretchable and accurately calibrated scale with 0.5 cm precision.

Blood pressure was measured, after a minimum 5-min rest period in a quiet room, with the use of a fitted cuff while participants were in the sitting position with their back supported and legs uncrossed. Systolic and diastolic blood pressures were determined and recorded as the first appearance and disappearance of Korotkoff sounds, respectively.

A venous blood sample was drawn from each participant following 12-h fasting to assess fasting blood sugar (FBS) and lipid profiles. All tests, including FBS, triglycerides (TGs), high-density lipoprotein (HDL), low-density lipoprotein, and cholesterol were assessed enzymatically using the BS200 Auto analyzer (Mindray, China). Screening tests for hepatitis B and C viral markers, as well as for autoimmune hepatitis, were performed.

Ten percent of the blood samples were evaluated by the Iranian National Reference Laboratory, with the coefficients of variation being between 1.7% and 3.8% of all laboratory values.

NAFLD was determined via evidence of hepatic steatosis in the sonogram and a lack of evidence of other causes of acute or chronic hepatitis. Evidence of secondary hepatic fat accumulation, such as significant alcohol consumption, use of steatogenic medication, or hereditary disorders, was also used to determine NAFLD.

All ultrasound examinations were carried out by a single sonographer who was an expert in the field of radiology. A 3-5 MHz transducer was used to examine the liver parenchyma and thereby provide sagittal, longitudinal, lateral, and intercostal views. Steatosis was confirmed if a marked increase of hepatic echogenicity was diagnosed or if the hepatic vessels and diaphragm appeared abnormal.

Homeostasis model assessment-insulin resistance (HOMA-IR) was calculated based on the following formula:

HOMA-IR = [insulin (mU/mL) × Glucose (mg/dL)]/405

FLI was calculated based on laboratory and anthropometric measures, including TG, GGT, BMI, and WC, by using the following formula:

FLI = [e^{0.953 × ln (TG) + 0.139 × BMI + 0.718 × ln (GGT) + 0.053 × WC - 15.745}/(1 + e^{0.953 × ln (TG) + 0.139 × BMI + 0.718 × ln (GGT) + 0.053 × WC - 15.745})] × 100

The capability of FLI to discriminate between subjects with and without NAFLD was evaluated using receiver operating characteristic (ROC) curve, for which the sensitivity of infinite decision thresholds of FLI was plotted against their false positive rates and thus the related areas under the curves (AUCs) were calculated. The lower boundary line for AUC was considered to be 0.5, with a significantly greater area than 0.5 showing some ability of FLI to discriminate between patients with and without NAFLD. The optimal cutoff point of FLI was also determined using maximal values of Youden’s J statistics [max (J = sensitivity + specificity - 1)]. The value of FLI corresponding to a maximum value of the Youden’s index was considered the optimal cutoff point for FLI.

Multivariable logistic regression analysis was conducted on NAFLD as an outcome variable alongside additional relevant predictor variables. Five potential predictor variables, including age, gender, MAP, HDL, and the HOMA-IR test were entered into the model in addition to FLI. In the multivariable model, Cox and Snell’s R², as well as Nagelkerke’s R², were calculated to determine how much NAFLD variance could be explained by the model. The Hosmer-Lemeshow test was used to evaluate the adequacy or suitability of the model. The odds ratio and related confidence intervals were reported along with P values. The significance level for all analyses was considered to be 0.05. All statistical analyses were conducted using version 21 of SPSS Inc., Chicago statistical software and STATA software, version 12 (StataCorp, Texas, United States).

In the total population, the AUC of FLI in the diagnosis of NAFLD was 0.8656 (95%CI: 0.8548-0.8764), in which no significant difference was detected between men (AUC = 0.8648, 95%CI: 0.8504-0.8791) and women (AUC = 0.8682, 95%CI: 0.8513-0.8851). Analysis by age group showed that the greatest capability was related to the 18-40 age group (AUC = 0.8930, 95%CI: 0.8766-0.9093), while the lowest accuracy was related to the 40-60 age group (AUC = 0.8293, 95%CI: 0.8095-0.8492). The AUC for the ≥ 60 age group was 0.8403 (95%CI: 0.8124- 0.8683). Although the predictive performance of FLI was significantly higher than BMI (AUC of BMI = 0.8258, 95%CI: 0.8139-0.8378), TG (AUC of TG = 0.6840, 95%CI: 0.6676-0.7004), and gamma glutamyl transferase (GGT) (AUC of GGT = 0.6927, 95%CI: 0.6772-0.7081), with P < 0.0001, no significant difference was detected between the performance of FLI and WC (AUC of WC = 0.8533, 95%CI: 0.8419-0.8646). Figure 2 shows the related ROC curves for FLI, WC, BMI, TG, and GGT in both men and women.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Receiver operating characteristic curves of fatty liver index and their related components. A and B were related to men and women, respectively. Red color dash pattern curve was related to WC, yellow color solid pattern curve was related to BMI, blue color dot pattern curve was related to GGT, purple color long dash 3-dot pattern curve was related to TG, and lime color dash dot pattern curve was related to FLI. BMI: Body mass index; FLI: Fatty liver index; GGT: Gamma glutamyl transferase; TG: Triglyceride; WC: Waist circumference.

Gender-based optimal cutoff points of FLI were also obtained for FLI in the diagnosis of NAFLD. The optimal cutoff points of FLI were 46.9 in men (sensitivity = 0.8242, specificity = 0.7687, Youden’s index = 0.5929) and 53.8 in women (sensitivity = 0.8233, specificity = 0.7655, Youden’s index = 0.5888).

According to these cutoff points, the prevalence of a high FLI was 0.4809 (0.4610-0.5007) in men and 0.5021 (0.4782-0.5260) in women. No significant difference was detected between the two sexes.

Univariate logistic regression analysis was performed on NAFLD by entering predictor variables, including age, sex, systolic blood pressure, diastolic blood pressure, HOMA-IR test, HDL, and FLI. The results are reported in Table 2.

Before conducting the multivariate model, its suitability and adequacy were evaluated using relevant specific tests. In the multi-collinearity diagnostic test, SBP and DBP were located in a single common dimension, while a variance proportion of 0.97 was related to SBP. As a result, we replaced SBP and DBP with mean arterial pressure (MAP) [MAP = DRP + 1/3 (SBP|DBP)] multivariate model. Collinearity tests were rechecked before conducting multivariable logistic regression, with no variance proportion ≥ 0.9 related to predictor variables. Tolerances varied from 0.599 to 0.951, with each predictor variable being located separately in a related independent dimension. In the multivariate model, Cox & Snell’s R² and Nagelkerke’s R² that can be explained by the model. On the other hand, Hosmer-Lemeshow test (χ² = 14.476, df = 8 and P = 0.07) indicated that a significant difference could not be established between observed and expected frequencies, thereby confirming the adequacy and suitability of our proposed model. After removing the effects of other predictors in the multivariable model, age (P < 0.001), gender (P = 0.002), MAP (P = 0.002), and FLI (P < 0.001) were significantly associated with NAFLD. FLI was highly associated with NAFLD, to the point that even a one-unit increase in FLI increased the chance of NAFLD occurrence by 5.8%. The odds ratios of the other predictor variables are shown in Table 2.

Table 3 shows the prevalence of NAFLD and high FLI, with the latter being calculated based on our study cutoff points; the second and third columns of the table are related to the prevalence of NAFLD and high FLI. Although an increasing trend of NAFLD prevalence among women was detected according to age, among men the prevalence was highest in the 40-59 age group. We observed a similar pattern for high FLI.

Non-alcoholic fatty liver disease (NAFLD) is a chronic condition. The prevalence of this condition varied from 20% to 30% in different countries. The most common methods used in the diagnosis of fatty liver are histological and imaging procedures, but these have their own limitations. Nevertheless, this disease can lead to a wide range of clinical conditions and, as a result, make its diagnosis difficult. In this context, some indices were recently introduced to diagnose NAFLD, with fatty liver index (FLI) being one example that has had a good prediction value in previous reports.

It is estimated that the prevalence of NAFLD among the Asian and Middle Eastern population is more than 20%. However, some cohort studies in Iran have shown a rate of up to 40%. Therefore, usage of a simple diagnostic modality with high predictive value for the detection of NAFLD is an important issue in this region.

As is the case with other countries, the prevalence of NAFLD in Iran is undergoing an increasing trend that is predicted to cause major health problem in the near future. Therefore, early diagnosis of NAFLD has become more important. In this study, we revealed that FLI has a high discriminatory power in the diagnosis of NAFLD in our population. The performance of FLI was not found to be different between two genders. Furthermore, the highest performance of FLI was seen in the 18-39 age group. The optimal cutoff points of FLI were 46.9 in men and 53.8 in women.

Based on the present study, we can suggest that WC has almost the same value as FLI in practice. In fact, WC could be considered an easy and economic modality with a high discrimination value for the detection of NAFLD.

NAFLD is a chronic condition with vast clinical presentation which is characterized by the accumulation of fat in the liver in the absence of other causes of steatosis, including excess consumption of alcohol or drugs. FLI was calculated based on laboratory and anthropometric measures, including TG, GGT, BMI, and WC by using a specific formula.

There is not enough data regarding diagnosis and discrimination of NAFLD, particularly in the Middle Eastern region. This study presents the discriminatory power of FLI and its cutoff point for NAFLD. The results are notable, with FLI showing no superiority over WC in the discrimination of NAFLD. Furthermore, although the performance of FLI was not different between the two sexes, a higher cutoff point of FLI was obtained for women than for men.