Risk and protective factors of co-morbid depression in patients with type 2 diabetes mellitus: a meta analysis

Abstract

The aim from this paper is to identify the main influencing factors of co-morbid depression among T2DM (Type 2 Diabetes Mellitus) patients and to provide reliable evidence for relative researches. A systematic review and meta-analysis of risk factors for co-morbid depression in T2DM was performed on all retrieved studies through an observational research of network database. Data were analyzed by Review Manager 5.3 from the extracted results, the heterogeneity index of the studies was determined using Chi-squared I² tests and on the basis of heterogeneity, a fixed or random effect model was used to estimates the pooled effect of each influencing factor. Fourteen observational studies containing total of 82,239,298 cases that have been identified. Diabetic complications (OR = 2.91; 95%CI, 1.76–4.82, p < 0.0001), insulin use (OR = 1.71; 95%CI, 1.18–2.48, p = 0.005), education status (OR = 1.91; 95%CI, 1.30–2.81, p = 0.001) were confirmed as risk factors, while regular exercising (OR = 0.51; 95%CI, 0.27–0.96, p = 0.04), gender (OR = 0.56; 95%CI, 0.47–0.65, p < 0.0001), marital status (OR = 0.53; 95%CI, 0.34–0.83, p = 0.005), current social status (OR = 0.64; 95%CI, 0.47–0.88, p = 0.006) were confirmed as protective factors of co-morbid depression in the patients with T2DM. Subgroup analysis claimed age (≥60 years) was a risk factor and smoking was protective factor for co-morbid depression in the patients with T2DM. Being female, have diabetic complications, insulin use, education level less than secondary are risk factors. However, doing regular exercise, being married and on work are protective factors of co-morbid depression in patients with T2DM. As to the other influencing factors should be further studied.

DIABETES AND DEPRESSION are two frequent chronic conditions of public health globally which are increasingly developing and have a large burden on the patients’ lives and society [1]. In 2017, 451 million people (age 18–99 years) were estimated to be suffering from diabetes, majority of the burden being in low and middle income countries. However, these figures were estimated to upped to 693 million by 2045 [2]. Depression is the most common type of mental illnesses, type 2 diabetes mellitus (T2DM) and depressive disorder are often co-morbid [3]. The reasons for this co-morbidity are unclear. Diabetes may increase the risk of depression because of the sense of threat and loss associated with receiving this diagnosis, several other factors such as hormonal, behavioral, psychosocial and the substantial lifestyle changes necessary to avoid developing debilitating complications [4]. It is estimated that 15%–20% of people with diabetes are suffering from depression, more likely moderate to severe form of depression [5].

Depression can conduct to a worse diabetes control, lower adoption to treatment, and increased economic burden of health care costs, besides depressive symptoms could seriously affect the quality of life of T2DM subject [6]. One study indicated that individuals with depressive symptoms are three times more likely to exhibit decreased adherence [7]. Evidence suggested diabetes and depression could mutually deteriorate, with each condition acting as a risk factor in the development of the other [8]. The interactive role of depression and diabetes in aggravation of T2DM can make the depressive patient having diminished self-care and consequent poorer metabolic control, which results in increasing diabetic complications and activation of a vicious cycle [9]. Age is a common risk factor for both depression and diabetes [10]. Thus, with increasing life expectations, these two conditions represent major burdens on the medical care system [11]. All of these, highlight the fact that co-morbid depression must be distinguished in order to discover the ways to prevent and treat it. To our knowledge, few of prior studies have performed meta analysis or systematic reviews related to the association between T2DM and depression, but none of them were based on cross sectional studies or reported the influencing factors for co-morbid depression in patients with T2DM. Therefore we aim to identify the main influencing factors of co-morbid depression among T2DM patients and provide reliable evidence for relative research.

Methods

Inclusion and exclusion criteria

Articles meeting the following criteria were included: observational studies, mainly focus on cross sectional studies; researches involving qualified depression evaluating scale; investigations reporting the association between T2DM and depression; the participants that are diagnosed with T2DM at the baseline; studies including both depression and anxiety symptoms associated with T2DM; studies reporting specific influencing factors of co-morbid depression with T2DM; studies published in English or Chinese. The exclusion criteria were as follows: studies reporting influencing factors neither depression or T2DM; the outcome was not suitable for meta analysis; not reporting full information; it was not possible to obtain full text; repeated articles; reviews.

Literature search strategy

Literature research was conducted from February to March in 2018 according to Systematic Reviews and Meta-analysis (PRISMA) guidelines without restriction of regions, and publication status. Electronic databases of PubMed, Embase, Web of Science, Wanfang Digital Periodicals (WANFANG), WeiPu, China National Knowledge Infrastructure (CNKI), Springer, and Ovid were searched. Detailed PubMed strategy was listed as follows: ((((type 2 diabetes mellitus) OR T2DM) AND depression) AND risk factors). Related articles of the included studies were screened according to the inclusion and exclusion criteria. The title and abstracts of all studies identified by the literature search were evaluated. The full text of any abstract meeting the inclusion criteria was then reviewed. Computer research was complemented with manual researches for all retrieved studies, reviews, and conference abstracts.

Quality assessment and statistical analysis

Quality assessment was carried out for all retrieved studies. Quality in the systematic review referred to the potential of biases during data analysis. The methodological integrity of the study was carried out according to the “Agency for Healthcare Research and Quality (AHRQ) for assessing the quality of non-randomized studies in meta-analysis” [12]. A score ranging from 0 to 11 (presented as “yes”, “no” and “unclear”) was allocated for each cross sectional study, and the studies that obtain more than seven stars were reckoned to be of high quality. Sensitivity analysis will be conducted to identify the robustness of the result by omitting each of the study or excluding low-quality trials.

Meta-analysis was performed with Review Manager (version 5.3). Heterogeneity was assessed using the Cochrane handbook and I² statistics, p = 0.05 using the χ² test for the Q statistic and I², 50% for the I² statistic were interpreted as low-level heterogeneity. The pooled effect was managed with a fixed-effects (FM) model when there was no statistically significant heterogeneity; otherwise, a random-effects (RE) model was employed [13].

Subgroup analysis was achieved for comparison of the results from different regions, ethnicity (as yellow), sample size and country character. The funnel plots were obtained to screen for potential publication bias.

Data extraction and outcome of interest

Enrolled studies for this paper were cross sectional studies thus data for the risk factors were extracted in the form of case numbers (dichotomous) and 95% confidence intervals (CI) for the outcome, for few of risk factors the data collected in the form of mean and SD (standard deviation). Explicit risk factors were standardized to the same reference category and continuous variables to the same units. For example, the gender risk factor was presented as the risk for being female. Where different units were reported for the same variable, those units reported in the majority of studies were used, and the most studies’ results were converted to the same units. Data were meta-analyzed where more than one study reported results for the same risk factor. Heterogeneity was assessed using the I² statistics. Due to the limited clinical applicability and bias of unvaried analysis of risk factors, results from studies where multi-variate adjustment for traditional co-morbid depression in T2DM risk factors were considered further. Core risk and protective factors including age, gender, BMI, diabetic complications (including neuropathy, stroke, sexual dysfunction, hypertension, heart failure, retinopathy, peripheral vascular disease), duration of diabetes, education status, marital status, current social status, exercise (engage in regular exercise or physical activity, for example exercise 1–2 hours per day), HbA1c, insulin use and smoking.

Results

Nine hundred and eighty-seven studies were searched, Fig. 1 shows the screening process. Full text of the 14 studies was retrieved for further extraction, and the titles and abstracts were inspected. Overall a total of 82,239,298 individuals were included in these studies. Table 1 indicates the summary of 14 cross sectional studies contributing data to systematic review. Included basic information: first author, published year, study design, country, mean age of participants, number of participants, proportion of gender, measurement of evaluation for depressive state, related outcomes and quality score. Quality for the included studies was generally high with an AHRQ score >7, the evaluation score was presented in Table 1. Each included studies presented multi-type of influencing factors of co-morbid depression with T2DM. The risk factors presented frequently among included studies were extracted and united, shown in Table 1. Due to each enrolled studies has different way to data assessment, some continuous variable were not united.

Fig. 1

Flow diagram of studies identified, included, and excluded

Table 1 Summary of 14 cross sectional studies contributing data to systematic review

Author (year)	Study design	Region	Mean age at baselines (years)	Number of participants (n)	Male	Female	Measurement of evaluation for depressive	Related outcomes	Quality score
Mikaliukstiene A 2014 [14]	Cross sectional study	Lithuania	59.3	1,022	372	650	HADSa	③④⑤⑥⑦⑨⑩⑪⑫	8
Park CY 2015 [15]	Community-based epidemiological study	Korea	55.48 ± 8.22	753	441	312	BDIb	①④⑤⑥⑦⑩⑪⑫	8
Joshi S 2015 [16]	Cross sectional study	Nepal	54.8 ± 10.6	379	189	190	BDI-IIc	①⑤⑦⑨⑪⑫	9
Long Hongzhu 2015 [17]	Cross sectional study	Beijing, China	27–86 years	636	/	/	HAMDd	③④⑥⑨	7
AlBekairy A 2017 [18]	Cross sectional study	Saudi Arabia	67.2 ± 12.6	158	84	74	HADSe	②③⑤⑥⑦⑧⑨⑩⑪⑫	9
Lee C-M 2017 [19]	Cross sectional study	Taiwan, China	68.2 ± 9.5	696	290	406	CGDS-SFf	⑤⑥⑦	9
Wang L 2015 [20]	Cross sectional study	Shanghai, China	70.13 ± 20.33	865	403	462	ZSDSg	⑧	8
Zhang W 2015 [21]	Cross sectional study	China	59.77 ± 12.48	412	207	205	BDIb	①⑨⑫	9
Habtewold TD 2016 [22]	Cross sectional study	Ethiopia	55.9 ± 10.9	264	124	140	PHQ-9h	⑩	10
Charles C Chima 2017 [23]	Cross sectional study	The US	>18 years	82,232,151	39,201,926	43,020,009	ICD-9-CMi	①⑦	9
Sun N 2016 [24]	Cross sectional study	Xuzhou, China	63.9 ± 10.2	893	370	523	ZSDSg	②	8
El Mahalli AA 2015 [25]	Cross sectional study	Saudi Arabia	49.87 ± 13.2	260	119	141	CES-Sj	①⑦⑩⑫	9
Islam SM 2015 [26]	Cross sectional study	Bangladesh	49.94 ± 10.21	515	227	288	PHQ-9h	①②③④⑤⑦⑧⑩	8
Xu X 2015 [27]	Cross sectional study	China	57.77 ± 9.64	294	146	148	CES-Sj	⑤⑦⑫	8
				SUM = 82,239,298

Note: ① Age, ② BMI, ③ diabetic complications, ④ duration of diabetes, ⑤ education status, ⑥ exercise, ⑦ gender, ⑧ HbA1c, ⑨ insulin use, ⑩ marital status, ⑪ smoking, ⑫ social status. “/”: means was not mentioned in original study.

a: “HADS” The Hospital Anxiety and Depression Scale [13]; b: “BDI” Beck Depression Inventory [14]; c: “BDI-II” Beck Depression Inventory-II [15]; d: “HAMD” Hamilton depression scale [16]; e: “HADS” Arabic version of the Hospital Anxiety and Depression Scale; f: “CGDS-SF” Chinese version of the Geriatric Depression Scale-Short Form [18]; g: “ZSDS” Chinese version of the Zung Self-rating Depression Scale [19]; h: “PHQ-9” nine-item Patient Health Questionnaire [21]; i: “ICD-9-CM”: International Classification of Diseases, Ninth Revision, Clinical Modification [22]; j: “CES-S” Center for Epidemiologic Studies Depression Scale [24];

Data for the extracted risk and protective factors are shown in Table 2 (Fig. 2) and Table 3 (Fig. 3). Table 2 is specially for dichotomous variable with pooled OR. The pooled data offered 10 potential risk and protective factors associated with depression among T2DM patients. The comparison standard for “age” was ≥60 and <60 years old, “diabetic complications” was have and do not have complications, “duration of diabetes” was ≥5 years and <5 years, “education status” was <secondary and ≥secondary, “exercise” was doing regular and irregular exercise, “gender” was male and female, “insulin use” was use and no use, “marital status” was married and none (being single/divorced/widowed), “smoking” was yes and no, “current social status” was work and none (retired, unemployed and others).

Table 2 Results for collected risk and protective factors for co-morbid depression events

Variable	Number of studies	Pooled OR	95% Confidence interval	p-value for OR	I2 (%)	Calculated model
Age	6	0.90	0.37–2.20	0.82	98	RE①
Diabetic complications	6	2.91	1.76–4.82	<0.0001	94	RE
Duration of diabetes	4	1.34	0.73–2.49	0.35	91	RE
Education status	8	1.91	1.30–2.81	0.001	84	RE
Exercise	5	0.51	0.27–0.96	0.04	91	RE
Gender	10	0.56	0.47–0.65	<0.0001	61	RE
Insulin use	5	1.71	1.18–2.48	0.005	73	RE
Marital status	7	0.53	0.34–0.83	0.005	75	RE
Smoking	4	0.83	0.50–1.37	0.46	73	RE
Current social status	7	0.64	0.47–0.88	0.006	69	RE

Note: ① RE: random-effect model.

Fig. 2

Forest plots for prior meta analysis

Table 3 Results for routinely collected risk factors for co-morbid depression events

Variable	Number of studies	Mean difference	95% Confidence interval	p-value for OR	I2 (%)	Calculated model
BMI	3	0.47	–0.50–1.45	0.34	73	RE
HbA1c	3	–0.08	–0.22–0.05	0.23	0	FE②

Note: ② FE: fixed-effect model.

Fig. 3

Forest plots for subgroup meta analysis

In the primary meta analysis, significant difference was found in diabetic complications (OR = 2.91; 95%CI, 1.76–4.82, p < 0.0001), education status (OR = 1.91; 95%CI, 1.30–2.81, p = 0.001); regular exercise (OR = 0.51; 95%CI, 0.27–0.96, p = 0.04), gender of being male (OR = 0.56; 95%CI, 0.47–0.65, p < 0.0001), insulin use (OR = 1.71; 95%CI, 1.18–2.48, p = 0.005), marital status (OR = 0.53; 95%CI, 0.34–0.83, p = 0.005), current social status (OR = 0.64; 95%CI, 0.47–0.88, p = 0.006). Due to I² value was generally high, the pooled effect was calculated with random-effect model. Significant difference was not found among age (p = 0.82), duration of diabetes (p = 0.35) and smoking (p = 0.46).

The pooled continuous variable presented in Table 3, there was no significant difference in both BMI (p = 0.34) and HbA1c (p = 0.23) among co-morbid depression patients with T2DM.

A high level of heterogeneity between studies and subgroups was observed. Subgroup analysis was carried out for comparison of original results with each variable as an influencing factor in foreign and Chinese population, different sample size, ethnicity (yellow) and country character. The results were demonstrated in Table 4 and Table 5. Table 4 is related to Table 2, which were both dichotomous variable, as to Table 5 and Table 3 were continuous variable.

Table 4 Subgroup results for collected risk and protective factors for co-morbid depression events

Variable	Number of studies	Classification of subgroup	Pooled OR	p-value for OR	95% Confidence interval	I2 (%)	Calculated model
Gender	6	①②	0.50	<0.0001	0.40–0.59	42	FE
Diabetic complications	4	④	3.87	<0.0001	2.27–6.59	92	RE
Duration of diabetes	2	③	0.83	0.21	0.63–1.11	0	FE
Education status	3	③	1.51	0.0008	1.19–1.91	0	FE
Exercise	2	⑤	0.34	0.12	0.09–1.35	95	RE
Age	2	③	1.63	0.0002	1.26–2.11	0	FE
Insulin use	3	①	2.06	0.001	1.32–3.19	59	FE
Marital status	3	⑥	0.39	<0.0001	0.30–0.49	0	FE
Smoking	3	③	0.72	0.01	0.56–0.92	0	FE
Current social status	3	③	0.62	0.01	0.42–0.90	61	RE

Note: ① foreign country (non-China), ② sample size less than 1,000, ③ ethnicity (yellow), ④ Asia, ⑤ Chinese, ⑥ country character.

Table 5 Subgroup results for routinely collected risk factors for co-morbid depression events

Variable	Number of studies	Classification of subgroup	Mean difference	95% Confidence interval	p-value for OR	I2 (%)	Calculated model
BMI	2	①	0.02	–0.70–0.74	0.96	0	FE②
HbA1c	2	①	–0.14	–0.46–0.17	0.37	0	FE

Note: ① foreign country, ② FE: Fixed-effect model

In subgroup analysis I² generally adjusted into low value. Significant differences were found in gender (OR = 0.5, 95%CI: 0.40–0.59, p < 0.0001), diabetic complications (OR = 3.87, 95%CI: 2.27–6.59, p < 0.0001), education status (OR = 1.51, 95%CI: 1.19–1.91, p = 0.0008), age (OR = 1.63, 95%CI: 1.26–2.11, p = 0.0002), insulin use (OR = 2.06, 95%CI: 1.32–3.19, p = 0.001), marital status (OR = 0.39, 95%CI: 0.30–0.49, p < 0.0001), smoking (OR = 0.72, 95%CI: 0.56–0.92, p = 0.01), current social status (OR = 0.62, 95%CI: 0.42–0.90, p = 0.01). Only duration of diabetes and exercise are without significant statistical difference. As to the continuous variable shown in Table 5 significant difference was not found in both BMI (p = 0.96) and HbA1c (p = 0.37), but with zero heterogeneity.

The enrolled studies were all included for sensitivity analysis. Among the most studies, the heterogeneity results were not apparently changed after sequentially omitting each study, showing that the results were statistically reliable. Publication bias was performed by funnel plot (Supplementary Fig.1, Supplementary Fig.2), most studies lie inside the 95%CI, indicating no obvious publication bias, only social status and diabetic complications shown higher publication bias which handled in subgroup analysis.

Discussion

Diabetes and depression are one of the most important public health problems in both advanced and emergent nations. This meta analysis of 14 observational studies containing 82,239,298 participants that are trying to highlight several important determinants in correlation with co-morbid depression in T2DM. Being female, having diabetic complications, insulin use with a low educational level may lead to an co-morbid depression with T2DM, by contrast being productive and mobile could be the effective factors for co-morbid depression with T2DM.

Heterogeneity is common because of the disparity of the included studies, methodological heterogeneity [28]. In order to find the source of heterogeneity subgroup analysis that was conducted according to different nationality, sample size, ethnicity and country character. However, heterogeneity was still significant in some outcomes, namely in diabetic duration and exercise. In subgroup analysis, age older than 60 years was a risk factor, but smoking was a protective factor for co-morbid depression with T2DM. Doing regular exercise was without significant different in subgroup analysis. While, continuous variable that included in this paper were without significant difference in both prior and subgroup analysis.

Globally, diabetes and depression are both chronic diseases that aging is their influencing factor, which were reported in the enrolled papers for this meta analysis as well. However, age showed no significant difference in the prior analysis. In order to find out the reason of this result a subgroup analysis was performed on the base of ethnicity. As expected, the result of subgroup analysis showed that age was a risk factor for co-morbid depression of T2DM. Therefore, differences of subjects specially the ethnicity might be the reason that prior analysis had no significant differences. Smoking is a common risk factor for multi-disease as known, at the same time smoking also has a role of decompression [29]. The aim of this paper was to explore the influencing factors between depression and T2DM, in this case smoking might be a factor that decrease depression. While smoking either presented as risk factor or without statistical difference among the included studies which mentioned smoking. In that way demographic characteristics of respondents, publication bias might bring such result.

In prior meta-analysis, doing regular exercise narrowly showed differences by p = 0.04, and in subgroup analysis p = 0.12 was higher than standard value (α = 0.05). Besides, the exact schedule of regular exercise was not reported in the related studies, it was difficult to combine each outcome into specific exercise stratification. Therefore more high quality researches are needed to determine whether exercising is an influencing factor for co-morbid depression with T2DM.

Despite higher heterogeneity diabetic complications showed high level OR value in both prior and subgroup analysis. However; this analysis was based solely on the presence or absence of diabetic complications, and the impact of which complications is greater remains to be further analyzed.

The present analysis showed no significant difference in duration of diabetes (p = 0.35 in prior analysis, p = 0.21 in subgroup analysis). This was consistent with two of included studies [15, 26] and a study conducted in Jazan area [30]. However, other studies contradicted to this finding where the prevalence of depression was found significantly related to the duration of diabetes [31]. This could be attributed to the differences in culture or long experience and adaptation gained in dealing with diabetes complications.

There were much more conventional and clinical influencing factors associated with co-morbid depression in T2DM mentioned either in included studies or other studies, such as financial income [15, 16, 18, 20, 21, 23, 25, 27] and sleeping hours [19-21]. Suitable data for meta analysis were not offered in enrolled studies, besides failed to get the original data, therefore these two variables were not included to meta-analysis. It is reported that reduced sleep quality with low levels of slow-wave sleep, as occurs in many individuals, may contribute to increase the risk of T2DM [32]. Accordingly present data about sleep quality and financial income demonstrated that low financial income and poor sleep are might be risk factors to co-morbid depression with T2DM [15, 18, 23, 25, 27]. There was a compelling factor also studied in an enrolled articles [15] which was dietary pattern, mostly vitamin B6 was found to be significant. It was reported that vitamin B6 deficiency is associated with depression and support for adequate vitamin B6 might prevent development of diabetes associated with depression [33, 34]. Therefore, vitamin B6 as a harmless agent could be an ideal protective factor for co-morbid depression in T2DM.

Heterogeneity among enrolled studies limits the interpretation of the results of meta analysis, particularly in observational studies [35, 36]. The ideal method for selecting and combining studies is uncertain, but by limiting the analysis to studies with at least some adjustment for traditional risk factors, aimed to reduce heterogeneity but at the cost of reduced power, via exclusion of some studies’ results of the meta-analysis [37]. Further limitations were included, the conversion of many prognostic factors from continuous to categorical variables, leading to a loss of statistical power and comparison difficulties between studies due to differing thresholds [38, 39]. Some continuous variables were presented as dichotomous in certain studies like age and diabetic duration, to combine such data into meta-analysis is also a difficulty, moreover might cause loss of data and potential bias. The cross-sectional design does not allow for cause-effect inferences, indeed, the included studies were cross–sectional, with different measurement conditions for depression and T2DM associated with a high inter and intra–individual variability, therefore more qualified studies are needed to make an exact conclusion about the aim of this paper.

Conclusion

This meta-analysis revealed risk and protective factors of co-morbid depression in patients with T2DM. Depression may affect both gender but as a female that is less productive and active and that has a very low literacy, retaining diabetic complications, it can be confirmed as a risk factor while being effective, being married and being on work were determined as protective factors of co-morbid depression in patients with T2DM. As to the other influencing factors should be further studied.

Ethical Approval

None sought. Our paper is a systematic review, so it does not involve ethical issues.

Funding

This study was funded by Key Discipline of the 13^th Five-Year Plan in XinJiang Uygur Autonomous Region-Public Health and Preventive Medicine, 99-11091113404; National Natural Science Foundation of China, 81360127;

Competing Interests

None declared.

Reference

• 1   Noroozi  Z,  Hamidian  S,  Khajeddin  N,  Mehrabi Zadeh Honarmand  M,  Zargar  Y, et al. (2017) Improving depression, and quality of life in patients with type 2 diabetes: using group cognitive behavior therapy. Iran J Psychiatry 12: 281–286.
• 2   Cho  NH,  Shaw  JE,  Karuranqa  S,  Huang  Y,  da Rocha Fernandes  JD, et al. (2018) IDF Diabetes Atlas: global estimates of diabetes prevalence for 2017 and projections for 2045. Diabetes Res Clin Pract 138: 271–281.
• 3   Clarke  TK,  Obsteter  J,  Hall  LS,  Hayward  C,  Thomson  PA, et al. (2017) Investigating shared aetiology between type 2 diabetes and major depressive disorder in a population based cohort. Am J Med Genet B Neuropsychiatr Genet 174: 227–234.
• 4   Wang  Y,  Yang  H,  Meng  P,  Han  Y (2017) Association between low serum 25-hydroxyvitamin D and depression in a large sample of Chinese patients with type 2 diabetes mellitus. J Affect Disord 224: 56.
• 5   Khan  ZD,  Lutale  J,  Moledina  SM (2019) Prevalence of depression and associated factors among diabetic patients in an outpatient diabetes clinic. Psychiatry J 2019: 2083196.
• 6   Che  T,  Teng  X,  Huang  Q,  Mu  Y,  Tang  X, et al. (2018) Agomelatine versus fluoxetine in glycemic control and treating depressive and anxiety symptoms in type 2 diabetes mellitus subjects: a single-blind randomized controlled trial. Neuropsychiatr Dis Treat 14: 1527–1533.
• 7   Xu  X,  Zhao  X,  Qian  D,  Dong  Q,  Gu  Z (2015) Investigating factors associated with depression of type 2 diabetic retinopathy patients in China. PLoS ONE 10: e0132616.
• 8  Gunzler D, Sajatovic M, McCormick R, Perszynski A, Thomas C, et al. (2016) Psychosocial features of clinically relevant patient subgroups with serious mental illness and comorbid diabetes. Psychiatr Serv 68: 96–99.
• 9  Hajimonfared M, Ostovar M, Heidery M (2016) Diabetes and depression: from Iranian traditional medicine to modern literature. Iran J Med Sci 41(3 Suppl): S28.
• 10   Tabak  AG,  Akbaraly  TN,  Batty  GD,  Kivimaki  M (2014) Depression and type 2 diabetes: a causal association? Lancet Diabetes Endocrinol 2: 236–245.
• 11  Fung ACH, Tse G, Cheng HL, Lau ESH, Luk A, et al. (2018) Depressive symptoms, co-morbidities, and glycemic control in Hong Kong Chinese elderly patients with type 2 diabetes mellitus. Front Endocrinol (Lausanne) 9.
• 12   Glance  LG,  Li  Y,  Osler  TM,  Mukamel  DB,  Dick  AW (2008) Impact of date stamping on patient safety measurement in patients undergoing CABG: experience with the AHRQ Patient Safety Indicators. BMC Health Serv Res 8: 176.
• 13   Shi  FH,  Li  H,  Cui  M,  Zhang  ZL,  Gu  ZC, et al. (2018) Efficacy and safety of once-weekly semaglutide for the treatment of type 2 diabetes: protocol for a systematic review and meta-analysis. Medicine (Baltimore) 97: e0420.
• 14   Mikaliūkštienė  A,  Žagminas  K,  Juozulynas  A,  Narkauskaitė  L,  Sąlyga  J, et al. (2014) Prevalence and determinants of anxiety and depression symptoms in patients with type 2 diabetes in Lithuania. Med Sci Monit 20: 182–190.
• 15   Park  CY,  Kim  SY,  Gil  JW,  Park  MH,  Park  JH, et al. (2015) Depression among Korean adults with type 2 diabetes mellitus: ansan-community-based epidemiological study. Osong Public Health Res Perspect 6: 224–232.
• 16   Joshi  S,  Dhungana  RR,  Subba  UK (2015) Illness perception and depressive symptoms among persons with type 2 diabetes mellitus: an analytical cross-sectional study in clinical settings in Nepal. J Diabetes Res 2015: 908374.
• 17   Long  H,  Sun  H,  Wu  S (2015) The clinical risk factors analysis of depressionin type-2 diabetes mellitus. World Chinese Med 10: 1607–1610 (In Chinese).
• 18   AlBekairy  A,  AbuRuz  S,  Alsabani  B,  Alshehri  A,  Aldebasi  T, et al. (2018) Exploring factors associated with depression and anxiety among hospitalized patients with type 2 diabetes mellitus. Med Princ Pract 26: 547–553.
• 19    Lee  CM,  Chang  CF,  Pan  MY,  Hsu  TH,  Chen  MY (2017) Depression and its associated factors among rural diabetic residents. J Nurs Res 25: 31–40.
• 20   Wang  L,  Song  R,  Chen  Z,  Wang  J,  Ling  F (2015) Prevalence of depressive symptoms and factors associated with it in type 2 diabetic patients: a cross-sectional study in China. BMC Public Health 15: 188.
• 21   Zhang  W,  Xu  H,  Zhao  S,  Yin  S,  Wang  X, et al. (2015) Prevalence and influencing factors of co-morbid depression in patients with type 2 diabetes mellitus: a General Hospital based study. Diabetol Metab Syndr 7: 60.
• 22   Habtewold  TD,  Alemu  SM,  Haile  YG (2016) Sociodemographic, clinical, and psychosocial factors associated with depression among type 2 diabetic outpatients in Black Lion General Specialized Hospital, Addis Ababa, Ethiopia: a cross-sectional study. BMC Psychiatry 16: 103.
• 23   Chima  CC,  Salemi  JL,  Wang  M,  Mejia de Grubb  MC,  Gonzalez  SJ, et al. (2017) Multimorbidity is associated with increased rates of depression in patients hospitalized with diabetes mellitus in the United States. J Diabetes Complications 31: 1571–1579.
• 24   Sun  N,  Lou  P,  Shang  Y,  Zhang  P,  Wang  J, et al. (2016) Prevalence and determinants of depressive and anxiety symptoms in adults with type2 diabetes in China: across-sectionalstudy. BMJ Open 6: e012540.
• 25   El Mahalli  AA (2015) Prevalence and predictors of depression among type 2 diabetes mellitus outpatients in Eastern province, Saudi Arabia. Int J Health Sci (Qassim) 9: 119–126.
• 26   Islam  SM,  Rawal  LB,  Niessen  LW (2015) Prevalence of depression and its associated factors in patients with type 2 diabetes: a cross-sectional study in Dhaka, Bangladesh. Asian J Psychiatry 17: 36–41.
• 27   Xu  X,  Zhao  X,  Qian  D,  Dong  Q,  Gu  Z (2015) Investigating factors associated with depression of type 2 diabetic retinopathy patients in China. PLoS One 10: e0132616.
• 28   Higgins  JP,  Thompson  SG (2002) Quantifying heterogeneity in a meta analysis. Stat Med 21: 1539–1558.
• 29   Jiang  H,  Li  S,  Yang  J (2018) Work stress and depressive symptoms in Fishermen with a smoking habit: a mediator role of nicotine dependence and possible moderator role of expressive suppression and cognitive reappraisal. Front Psychol 9: 386.
• 30   Albasheer  OB,  Mahfouz  MS,  Solan  Y,  Khan  DA,  Muqri  MA, et al. (2018) Depression and related risk factors among patients with type 2 diabetes mellitus, Jazan area, KSA: a cross-sectional study. Diabetes Metab Syndr 12: 117–121.
• 31    Bensbaa  S,  Agerd  L,  Boujraf  S,  Araab  C,  Aalouane  R, et al. (2014) Clinical assessment of depression and type 2 diabetes in Morocco: economical and social components. J Neurosci Rural Pract 5: 250–253.
• 32   Narisawa  H,  Komada  Y,  Miwa  T,  Shikuma  J,  Sakurai  M, et al. (2017) Prevalence, symptomatic features, and factors associated with sleep disturbance/insomnia in Japanese patients with type-2 diabetes. Neuropsychiatr Dis Treat 13: 1873–1880.
• 33  Oxenkrug G, Ratner R, Summergrad P (2013) Kynurenines and vitamin B6: link between diabetes and depression. J Bioinform Diabetes Sep 14: 1.
• 34   Mesripour  A,  Hajhashemi  V,  Kuchak  A (2017) Effect of concomitant administration of three different antidepressants with vitamin B6 on depression and obsessive compulsive disorder in mice models. Res Pharm Sci 12: 46–52.
• 35   Sutton  AJ,  Higgins  JP (2008) Recent developments in meta-analysis. Stat Med 27: 625–650.
• 36   Shrier  I,  Boivin  JF,  Steele  RJ,  Platt  RW,  Furlan  A, et al. (2007) Should meta-analyses of interventions include observational studies in addition to randomized controlled trials? A critical examination of underlying principles. Am J Epidemiol 166: 1203–1209.
• 37   Major  RW,  Cheng  MRI,  Grant  RA,  Shantikumar  S,  Xu  G, et al. (2018) Cardiovascular disease risk factors in chronic kidney disease: a systematic review and meta-analysis. PLoS One 13: e0192895.
• 38   Altman  DG,  Royston  P (2006) The cost of dichotomising continuous variables. BMJ 332: 1080.
• 39   Turner  EL,  Dobson  JE,  Pocock  SJ (2010) Categorisation of continuous risk factors in epidemiological publications: a survey of current practice. Epidemiol Perspect Innov 7: 9.