Relationship between serum lipids and depression: A cross sectional survey among adults in Haryana, India

ABSTRACT

Introduction:

Dyslipidemia and mental illnesses are significant contributors to the global noncommunicable disease burden and studies suggest an association between them.

Aim:

Using data from a noncommunicable disease risk factor survey conducted in Haryana, India, we undertook a secondary data analysis to examine the association between lipids and depressive symptoms.

Methods:

The survey involved 5,078 participants and followed the World Health Organisation STEPwise approach to NCD risk factor surveillance approach. Biochemical assessments were undertaken in a subset of participants. Lipid markers were measured using wet chemistry methods. Depressive symptoms were assessed using the Patient Health Questionnaire–9. Descriptive statistics were presented for all variables; logistic regression was used for association analyses.

Results:

The mean age of the study population was 38 years and 55% of them were females. A majority of the participants belonged to a rural background. The mean total cholesterol was 176 mg/dL and approximately 5% of the participants were found to have moderate to severe depression. The association of total cholesterol (odds ratio [OR] 0.99, P = 0.84), LDL-cholesterol (OR = 1.00, P = 0.19), HDL-cholesterol (OR = 0.99, P = .76), and triglycerides (OR 1.00, P = .12) with depressive symptoms was not significant.

Conclusion:

This study did not find any association between lipids and depressive symptoms. However, further investigations using prospective designs are warranted to understand this relationship and complex interactions with other mediating factors better.

INTRODUCTION

A vast majority of the noncommunicable disease (NCD) deaths can be attributed to cardiovascular diseases (CVDs) that are a consequence of modifiable behavioral risk factors like unhealthy diet, physical inactivity, diabetes, hypertension, hyperlipidemia, tobacco, and alcohol use.[1] Interestingly, these physical health behaviors are commonly clustered in individuals with mental health conditions also[2,3] that have a 7% share in the total global burden of disease.[4] Over the last few decades, there has been growing interest in research particularly involving lipids and mental and neuro-degenerative conditions, across the world. Changes in membrane lipids have been found to be associated with depression,[5] self-harm behavior,[6] and Alzheimer’s[7] but the relationship is not yet fully understood.[8,9] While some studies have shown associations of both increased and decreased lipid levels with mental illnesses, others have reported a lack of any significant association.[10] Recent systematic reviews and meta-analyses have reported a probable risk of depression and suicidal behavior among individuals with low LDL-cholesterol and low serum cholesterol.[11,12] At the same time, a Mendelian randomization analysis published in 2019 had reported an association between raised triglyceride levels and depression.[13] Therefore, inconsistent findings and the high heterogeneity across studies warrant further high quality research.

Systematic reviews and meta-analyses and epidemiological studies in the recent past have reported associations of heart disease and diabetes with mental disorders like depression, anxiety, schizophrenia, bipolar disorder, cognitive impairment, and more.[14–16] This has in-turn led to the creation of a five-by-five approach by the UN member states, focused on the inclusion of mental health in strategies to tackle the NCD burden.[2]

Literature on the relationship between lipids and psychological disorders like depression, anxiety, and suicidal thoughts in the Indian population is rather scanty. A very few observational studies that have been conducted in some parts of the country have reported contradictory results, with some linking depression to low cholesterol[17,18] and others suggesting an association with high cholesterol[19] or no association at all.[20] Hence, existing research is ambiguous and majorly inadequate for drawing reasonable scientific conclusions.

NCDs impede growth, pressurize health systems, and increase healthcare costs[21,22] and the World Health Organization (WHO) recommends risk profiling for estimating population needs, as the first step in NCD screening.[23] To address the lack of good quality data and provide state-level NCD risk factor estimates, a one of its kind large cross-sectional study was undertaken in 2017, wherein extensive data were collected on all the major NCD risk factors among the adult population residing in Haryana, India.[24] Given that India has a high prevalence of both dyslipidemia and mental disorders and a considerable dearth of studies on the same, we took advantage of data from this surveillance study and used the opportunity to examine the relationship of serum lipids with depressive symptoms and suicidal thoughts among adults in Haryana.

The present study aimed to address the following objectives:

1. To determine the association between lipid parameters and the depression status of the study population.

2. To determine the association between lipid parameters and suicidal thoughts among the study population.

We hypothesized that there is an association between the lipid levels of the population and their depression status as measured by the Patient Health Questionnaire–9 (PHQ-9).

METHODS

Study design, setting, and population

The study was a state-level community-based cross-sectional survey conducted as a collaborative effort between the host institute and several medical and research institutes. The survey methodology has been described in detail elsewhere.[24] The study population was adult residents of Haryana aged 18-69 years. The study duration was 15 months and data on various health and lifestyle indicators were collected for 5 months from April to September 2017. The current article is a secondary analysis of the data obtained from the cross-sectional survey.

Study sampling

A probabilistic, multistage, clustered sampling design was used with households as the ultimate sampling units. A Census Enumeration Block and a village was the primary sampling unit (PSU) in the urban and rural areas, respectively. Systematic random sampling was used to select 35 households from each PSU. Kish method was used to select one individual from each household. A total of 5,250 individuals were recruited from a total of 150 PSUs, with 35 respondents from each PSU.

Study tool

The survey questionnaire was adapted from the standardized step-wise approach to surveillance (STEPS) version 3.1, recommended by the WHO for NCD surveillance.[25]

Data collection and study variables

Survey data were collected using an android-based application called mSTEPS that was piloted before the start of the study. Step 1 involved collection of sociodemographic information. Step 2 involved physical and physiological measurements and biochemical assessments were undertaken in Step 3.

Independent variables: Wet chemistry methods were used to assess the levels of total cholesterol, LDL-cholesterol, HDL-cholesterol, and triglycerides.

Dependent variables: Mental health was assessed for the first time in a STEPS survey in India, using the standard PHQ-9[26] scale for depressive symptoms. The participants were categorized into categories of such symptoms using the standard PHQ cut-offs[26] of minimal (0-4), mild (5-9), moderate (10-14), moderately severe (15-19), and severe (20-27). Depression was also presented as a binary variable using a cut-off score of less than 10 and 10 or more. In addition, suicidal thoughts, indicated by question 9 of the PHQ-9, was recoded as a binary variable using a cut-off score of 0 and 1 or more. The binary variables for depressive symptoms and suicidal thoughts were used in the association analyses.

Covariates: Information on tobacco use included their present status as current smokers, past smokers, and nonsmokers. Alcohol use was categorized as currently consuming, past consumers, nonalcoholics, and those who had ever consumed alcohol. Fruit and vegetable consumption was assessed by splitting participants into two groups: consumption for less than 4 days or 4 days or more. Blood pressure was assessed using a calibrated OMRON digital monitor and the average of two measurements taken at a two-minute gap was used for the analysis. A portable digital weighing scale and stadiometer were used to measure participant weight and height, respectively, in accordance with the standardized WHO recommendations for such surveys (SECA, Hamburg, Germany). The body mass index of each participant was calculated as weight in kilograms divided by height in square meters. Blood glucose was assessed using dry chemistry methods.

Sample collection: Each participant was provided with written instructions regarding fasting and date of blood test for biochemical assessment. Blood serum was stored in ice boxes and blood samples were centrifuged by trained phlebotomists. The analysis was undertaken at the central laboratory at the host institute.

Ethical considerations

The study was approved by the Institutional Ethics Committee vide letter no. IEC/2016/3536. The study protocol was approved by the Technical Advisory Committee that also ensured appropriate implementation and execution of the survey. The participants provided a written informed consent for their participation in the study. The confidentiality of the participating individuals was duly maintained throughout the study period and analysis. Those found to have depressive symptoms were encouraged to seek formal psychiatric consultation and the same was facilitated.

Statistical analysis

All statistical tests were performed using Stata version 15.1 (StataCorp, College Station, TX, USA). Survey data analysis techniques were used to undertake weighted analyses using the survey (.svy) command in Stata, to account for weights (sampling, population, and nonresponse weights), clustering, and stratification. Descriptive statistics were presented for all study variables, with continuous variables summarized as means and standard errors and categorical variables summarized as frequencies and percentages. The sociodemographic characteristics were presented as unweighted values. Outlying values of lipid parameters were removed from the analysis, based on biological plausibility.

Logistic regression was used to determine the associations between lipid levels (total cholesterol, HDL-cholesterol, LDL-cholesterol, and triglycerides) and depressive symptoms and suicidal thoughts among the participants.

RESULTS

The sample size of the present study was 5,250 while 5,078 participants gave the consent for data collection. The percentage of missing observations ranged from 0%-6%. Almost all the study participants underwent questionnaire, physical, and physiological assessments and biochemical assessments were undertaken for a subsample of around 50% of the total participants. A complete case analysis was therefore not possible and association analyses were conducted on a smaller subset of the total participants. This is described in Figure 1.

Figure 1.

Flowchart depicting the total number of participants included in the present analysis

The sociodemographic profile of the study participants, overall and segregated by gender, is described in Table 1. The mean age of the participants was 38 years and around 70% of them belonged to a rural background. Only 14% of the population had received college education. The mean monthly income of the participants was approximately 24,500 rupees.

Table 1.

Sociodemographic characteristics of the survey participants, STEPS Survey, Haryana, India (n=5,078)

Characteristics	n (%)
Mean age in years (SE)	38.52 (12.99)
Sex
Males	2,294 (45.18)
Females	2,784 (54.82)
Age group
18-44	3,473 (68.39)
45-69	1,605 (31.61)
Residence
Rural	3,368 (68.33)
Urban	1,710 (33.67)
Education
No formal schooling	514 (10.12)
Less than primary	321 (6.32)
Primary school	743 (14.63)
Secondary school	878 (17.29)
High school	1,113 (21.92)
Graduate	575 (11.32)
Post graduate	137 (2.70)
Refused	797 (15.70)
Marital status
Never married	612 (12.05)
Currently married	4,132 (81.37)
Separated	8 (0.16)
Divorced	5 (0.10)
Widowed	271 (5.34)
Refused	50 (0.98)
Occupation
Government job	138 (2.72)
Non-government job	976 (19.22)
Self-employed	857 (16.88)
Non-paid	19 (0.37)
Student	291 (5.73)
Homemaker	2,239 (44.09)
Retired	57 (1.12)
Unemployed (can work)	91 (1.79)
Unemployed (cannot work)	349 (6.87)
Refused	61 (1.20)
Mean monthly income (SE) (n=3,791)	24,504.28 (39,317.11)

Categorical variables expressed as n and percentage. Continuous variables expressed as mean and standard error (SE)

Table 2 describes the distribution of lifestyle and behavioral risk factors of the study participants. More than 70% of the participants were nonsmokers and nondrinkers. The consumption of fruits was not adequate but the participants reported consuming vegetables regularly. The frequency of having processed salty food was also moderate. The level of physical activity and walking in the population was suboptimal and the participants reported sitting for approximately 6.5 hours daily, on an average.

Table 2.

Distribution of lifestyle characteristics among the participants (n=5,078)

Characteristics	n (%)
Tobacco Use
Never smoked	3,837 (72.94)
Past smokers	86 (1.66)
Current smokers	1,155 (25.40)
Alcohol consumption
Never consumed	4,195 (79.01)
Quit alcohol	27 (0.59)
Currently consuming	710 (17.26)
Ever consumed	146 (3.14)
Days of fruit consumption (n=5077)
<4 days	3,429 (64.26)
>=4 days	1,648 (35.74)
Days of vegetable consumption
<4 days	1,816 (35.15)
>=4 days	3,262 (64.85)
Adding salt to cooked food
Always	110 (2.03)
Often	263 (5.17)
Sometimes	1,627 (31.82)
Rarely	615 (12.99)
Never	2,444 (47.69)
Don’t know	19 (0.30)
Frequency of eating processed salty food
Always	25 (0.59)
Often	169 (4.27)
Sometimes	1,756 (36.46)
Rarely	1,233 (25.21)
Never	1,878 (33.22)
Don’t know	17 (0.24)
Vigorous physical activity at work	1,312 (24.31)
Moderate physical activity at work	2,944 (54.28)
Involved in vigorous intensity sports	1,195 (25.64)
Involved in moderate intensity sports	728 (13.43)
Walk/cycle for at least 10 min/day	2,835 (55.52)
Mean sitting time/day in min (SE)	380.05 (10.95)

Categorical variables expressed as n and percentage. Continuous variables expressed as mean and standard error (SE)

Table 3 describes the disease status and family history of the study participants. About 20% and 40% of the participants had a family history of chronic diseases, that is, diabetes and hypertension and about 8% reported having gotten their cholesterol checked. The mean values of total cholesterol and LDL-cholesterol were 176 mg/dL and 106 mg/dL, respectively. Approximately 95% of the participants had a PHQ-9 score of less than 10 and about 5% were found to have moderate to severe symptoms of depression as per the PHQ-9 scale. About 35% of the participants reported having thoughts of suicide/self-harm on several or almost all days.

Table 3.

Distribution of health status and family disease history of the participants (n=5,078)

Characteristics	n (%)
History of raised cholesterol (n=427)	120 (22.48)
Taking oral cholesterol medicine (n=124)	41 (30.46)
History of heart disease	886 (15.86)
Taking statin for heart disease	15 (0.33)
Family history of diabetes	1,009 (20.78)
Family history of high blood pressure	2,058 (42.64)
Family history of stroke	157 (3.10)
Family history of cancer	248 (3.74)
Family history of high cholesterol	156 (3.60)
Mean BMI (kg/m2) (SE) (n=5,041)	23.64 (0.15)
Mean SBP (mmHg) (SE) (n=5,063)	123.70 (0.35)
Mean DBP (mmHg) (SE) (n=5,063)	82.59 (0.24)
Mean fasting blood glucose* (mg/dl) (SE) (n=2,490)	97.56 (1.07)
Mean total cholesterol* (mg/dl) (SE) (n=1,996)	175.84 (2.68)
Mean triglycerides* (mg/dl) (SE) (n=2,016)	138.87 (3.58)
Mean HDL-C* (mg/dl) (SE) (n=2,362)	52.60 (2.32)
Mean LDL-C* (mg/dl) (SE) (n=1,942)	106.05 (3.04)
Level of depression
Minimal/none (0-4)	3,130 (64.14)
Mild (5-9)	1,633 (30.52)
Moderate (10-14)	0 (0.00)
Moderately severe (15-19)	236 (4.01)
Severe (20-27)	79 (1.32)
Depression status
No depression (<10)	4,763 (94.67)
Depression (>=10)	315 (5.33)
Suicidal thoughts
Not at all	3,153 (64.92)
Several days and above	1,897 (35.08)

Categorical variables expressed as n and percentage. Continuous variables expressed as mean and standard error (SE). *Biochemical measurements undertaken for only 50% of the total participants

We ascertained the association of lipid parameters with symptoms of depression and suicidal thoughts in the participants. No significant association was found between lipid levels (total cholesterol, triglycerides, HDL-cholesterol, and LDL-cholesterol) and the PHQ-9 score, in both the crude model and models adjusted for age, sex, family income, lifestyle, body mass index, blood pressure, and blood glucose. Similarly, we could not find a significant association between lipid levels (total cholesterol, triglycerides, and LDL-cholesterol) and suicidal thoughts in both crude and adjusted models. There was very weak evidence for a borderline association between HDL-cholesterol and suicidal thoughts in the adjusted model [Table 4].

Table 4.

Association of lipid parameters with depression status and suicidal thoughts among the participants

Depression
	n	Odds ratio [95% CI]	Linearized standard error	P
Total Cholesterol
Model1	1,996	0.99 [0.994-1.005]	0.003	0.87
Model2	1,996	0.99 [0.995-1.004]	0.002	0.76
Model3	1,436	0.99 [0.992-1.007]	0.004	0.84
Triglycerides
Model1	2,016	1.00 [0.998-1.003]	0.001	0.46
Model2	2,016	1.00 [0.998-1.004]	0.001	0.36
Model3	1,438	1.00 [0.999-1.008]	0.002	0.12
HDL-C
Model1	2,362	0.99 [0.992-1.007]	0.004	0.87
Model2	2,362	0.99 [0.991-1.006]	0.004	0.78
Model3	1,685	0.99 [0.985-1.011]	0.006	0.76
LDL-C
Model1	1,942	1.00 [0.997-1.010]	0.003	0.21
Model2	1,942	1.00 [0.998-1.009]	0.003	0.11
Model3	1,750	1.00 [0.999-1.012]	0.004	0.19
Suicidal Thoughts
	n	Odds ratio [95% CI]	Linearized standard error	P
Total Cholesterol
Model1	1,983	1.00 [0.997-1.003]	0.002	0.78
Model2	1,983	1.00 [0.996-1.003]	0.002	0.95
Model3	1,426	0.99 [0.994-1.003]	0.002	0.67
Triglycerides
Model1	2,006	0.99 [0.997-1.001]	0.001	0.47
Model2	2,006	0.99 [0.998-1.001]	0.001	0.74
Model3	1,429	0.99 [0.996-1.001]	0.001	0.41
HDL-C
Model1	2,348	0.99 [0.992-1.001]	0.003	0.20
Model2	2,348	0.99 [0.991-1.001]	0.003	0.09
Model3	1,674	0.99 [0.991-0.999]	0.002	0.043
LDL-C
Model1	1,931	1.002 [0.997-1.007]	0.002	0.37
Model2	1,931	1.001 [0.996-1.006]	0.002	0.57
Model3	1,365	1.00 [0.994-1.006]	0.003	0.89

Model¹ - Crude model. Model² - Adjusted for age and sex. Model³ - Adjusted for age, sex, income, tobacco use, alcohol, physical activity, diet, BMI, SBP, DBP, and fasting sugar. Level of significance P≤0.05

We also explored the association of both depression and suicidal thoughts with lipids, stratified by gender. However, there was no evidence for an association in the stratified analyses as well.

In addition, we explored the crude and age and sex-adjusted association of NCD risk factors (fasting blood glucose, blood pressure, alcohol, and tobacco use) with depression but did not find any significant relationship among these variables in our study (findings will be made available upon reasonable request).

DISCUSSION

Since impaired lipids are a major precursor to CVD and depression is one of the most prevalent mental illnesses both worldwide and in India, we used data from our Haryana STEPS survey on NCD risk factor profiling, to determine if an association exists between lipid levels and depressive symptoms in the community. We did not find evidence to suggest an association between lipid parameters and symptoms of depression and suicidal thoughts in our adjusted analyses.

These findings are consistent with results from a 2005 hospital-based study from Taiwan that was conducted to ascertain lipid levels among 109 individuals (mean age 31 years) of various depression subtypes and compare them with 59 healthy controls. The serum lipid profile of the participants was normal and no significant association was seen between lipids and depression.[27] A few other cross-sectional studies conducted in the past have also reported no association between hyperlipidemia and depression, albeit in the elderly.[28–30]

On the other hand, many studies have reported significant associations of both hypolipidemia (low total cholesterol,[17,18,31,32] low LDL-cholesterol,[12,18,32,33] low triglycerides,[18] and low HDL-cholesterol[8,32,34]) and hyperlipidemia (high total cholesterol,[8,34–36] high LDL-cholesterol,[8,37] and high triglycerides[8]) with an increased risk of depression in adults. The evidence has therefore been rather inconclusive and conflicting and it remains uncertain whether lower lipid levels or higher lipid levels could be a biological marker in depressive individuals.

There could be a few reasons to explain the findings of our study. First, the mean age of our study population was 37 years. Thus, our sample comprised of relatively younger individuals and it is possible that the association is more pronounced in slightly older populations who may be at a greater risk of both dyslipidemia and CVDs. In addition, about 5% of our study population was found to be moderate to severely depressive as assessed by the PHQ-9, whereas 95% of them were either normal or mildly depressive. Studies in the past that have reported associations between depression and mood disorders and impaired lipid profiles have majorly been conducted in either populations that were already suffering from a mental health condition[8,17,18,31,35,36] or among individuals with diabetes or heart failure.[33,34] Differences in the overall health status and a lack of a substantial number of participants who either had impaired lipids or were diagnosed with depressive symptoms in our study could be a reason for no association between the two variables.

The difference can also be attributed to the fact that different studies used different scales for the assessment of depressive symptoms which was not necessarily accompanied by clinical judgment. Third, there is literature on the possibility of a change in lipid levels as a consequence of depression[38] and depression being a risk factor for nonadherence to lipid-lowering drugs, which could in turn lead to dyslipidemia.[39] A cross-sectional study cannot ascertain the direction of causality and could thus affect the overall findings. Fourth, conflicting results could also be reflective of methodological differences among studies. Meta-analyses published in this domain[11,12] have reported the presence of a high level of heterogeneity among individual studies, with respect to the type of population, study designs, and differences in the assessment measures employed across studies.

The relationship between cardiovascular conditions and mental health disorders is rather complex and could be confounded by various other environmental and genetic factors. They share some common biological mechanisms. However, both the exact nature of the association and its direction still remain to be elucidated. Research has even suggested a probable effect of Vitamin D levels on serum cholesterol[40] which in turn might be responsible for mediating the relationship between lipids and mental health but the evidence remains equivocal. Studies have shown that depressed individuals have at least a 2-fold greater risk of heart disease compared to those who are not depressed.[41,42] Conversely, cardiac events and morbidity have also been shown to be determinants of acute and chronic depressive episodes in individuals.[43,44] This ambiguity regarding the cause and effect aspect of this relationship has further raised the question of whether cardiovascular and mental health conditions are different complications of a common underlying pathophysiology altogether.

Limitations

Ours is the first state-level survey on profiling of NCD risk factors in Haryana, India and the estimates are representative of the population of the state. Also, in spite of having undertaken biochemical analyses for a subset of the population due to resource constraints, we had a reasonably good sample size for the same.

However, the study has a few limitations. Although we used the PHQ-9 scale, which is a thoroughly validated and robust tool for the assessment of depressive symptoms worldwide,[26] the responses captured by this screening tool are self-reported and maybe subjective, which may have introduced a bias in our estimates. We could not validate the responses with a more objective clinical examination. Second, our survey employed a cross-sectional study design. Since the exposure and outcome are measured at the same time, it is not possible to establish the temporality of the association being studied. It is possible that the occurrence of depressive symptoms among the participants preceded a rise or fall in their lipid values. Finally, we did not evaluate participants for ongoing antidepressant medications, which could have influenced the prevalence of such symptoms.

Implications for future research

The present study did not find evidence for an association between serum lipids and presence of symptoms of depression or suicidal thoughts. However, we plan to study this population cohort prospectively and may repeat the survey after 5 years to obtain deeper insights into the relationship between lipids and mental health in the state of Haryana.

The need of the hour is to design longitudinal studies with long follow-up durations to obtain a more holistic and detailed understanding of the interplay between the various NCD risk markers. Several factors that could mediate and explain this relationship better also offer an opportunity for future research. This would provide insights to clinicians for making well-informed decisions for each individual and reducing the overall mortality and morbidity attributed to these NCDs. It would also help policy makers devise treatment and prevention strategies to cater to populations with varying levels of predisposition to disease risk.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.