The association between dietary total, animal and plant protein and macronutrient substitution; and the risk of all-cause, cardiovascular or cancer mortality in the Golestan cohort study

Abstract

Background

Dietary protein is essential for health, yet its association with mortality risk across various causes remains unclear. This study examines the associations between total, animal-, and plant-based dietary protein intake and macronutrient substitution, as well as all-cause, cardiovascular (CVD), and cancer-related mortality.

Methods

We used data from the Golestan Cohort Study (GCS), a population-based prospective study initiated in 2004 in Golestan Province, Iran, comprising 43,050 adults aged 40–75 years. Dietary intakes were assessed using validated Food Frequency Questionnaires (FFQ). Mortality outcomes were tracked annually over a mean follow-up period of 15.05 years. Cox proportional hazard regression models adjusted for demographic, lifestyle, and clinical factors were employed. Substitution analysis and restricted cubic spline functions explored the association between different ranges of protein intake and substitution with carbohydrates or fats and mortality risks.

Results

During follow-up, 9,309 deaths were documented, including 3,707 from CVD and 1,917 from cancer. Higher total protein intake was associated with increased all-cause (HR for 10 g intake 1.02, 95% CI 1.01–1.03) and CVD mortality (HR for 10 g intake 1.02, 95% CI 1.00–1.03), while animal protein showed similar associations (HR for 10 g intake 1.01, 95% CI 1.01–1.02 for all-cause mortality; HR for 10 g intake 1.02, 95% CI 1.00–1.03 for CVD mortality), plant protein intake was associated with a reduced risk of all-cause mortality in the total population (HR for 10 g intake 0.97, 95% CI 0.95–1.00). The associations were slightly more significant in female and rural participants. Substituting total protein or animal protein with carbohydrates or fat was associated with a lower risk of total and CVD mortality. Protein intake showed a non-linear association with mortality.

Conclusion

Balanced protein intake, comprising an appropriate mix of animal and plant sources, appears to support longevity and reduce mortality risk, particularly from cardiovascular causes.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-27844-w.

Highlights

This study evaluates the associations between animal and vegetable protein intakes and their substitution with other macronutrients and mortality. Higher intakes of total and animal protein were associated with increased risk of all-cause and cardiovascular disease mortality. The association between protein intake and mortality showed a nonlinear pattern. Substituting total or animal protein with carbohydrates or fats was associated with lower all-cause and CVD mortality.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-27844-w.

Introduction

According to the World Health Organization, many premature deaths are preventable through a healthy lifestyle, including better dietary habits [1]. For years, dietary guidelines had emphasized a balance between calorie intake and energy expenditure, as well as reducing fat and consuming a certain amount of protein [2]. International organizations have recently emphasized the importance of macronutrient qualities rather than quantities in their recommendations for the primary prevention of chronic disease [3]. However, the definition of a macronutrient-balanced diet remains an open question and is a high priority in nutrition science [4].

The optimal amount and source of dietary protein required to maintain health remain subjects of considerable scientific debate. Dietary guidelines for optimal protein intake are mainly based on nitrogen balance rather than on long-term health outcomes [5, 6]. Protein consumption has been linked to a range of health indicators; for example, several studies suggest potential benefits for weight management [7]. Conversely, high total or animal-derived protein intake has been associated with an increased risk of type 2 diabetes [8], while findings regarding cardiovascular disease (CVD) remain inconsistent, with reports of protective, neutral, or adverse associations [6, 9, 10]. An umbrella review on the association between total, animal, and plant protein intake and major health outcomes, including CVD, cancer, and bone health, concluded that current evidence is inconclusive [6]. Additionally, it has been proposed that the health effects of protein may follow a U-shaped relationship and depend strongly on the macronutrient it replaces (e.g., carbohydrates or fats), which could substantially influence both the direction and magnitude of associated health risks [6].

Plant-based dietary patterns are recommended for their health and environmental benefits [11, 12]. Plant-based diets have been associated with a lower risk of type 2 diabetes and cancers [8, 13–15]. Consuming more plant foods and less animal foods has been associated with lower morbidity and mortality from cardiovascular diseases in the general population [14]. Emerging evidence favours higher intake of plant-derived proteins; however, the underlying mechanisms and the extent of benefit require further investigation [6].

To accurately evaluate the health effects of dietary protein intake, it is essential to determine which macronutrient is substituted by protein in the diet. Ideally, such causal relationships would be addressed through long-term randomised controlled trials; however, requiring participants to adhere to specific dietary patterns over many years, or even across the lifespan, is neither practical nor ethically feasible. Given these constraints, several studies have proposed alternative analytical strategies to assess the associations between macronutrient intake and mortality risk in observational studies [16, 17]. These methods model the substitution of one macronutrient for another and investigate the differential effects of proteins or fats from animal versus plant sources on the development of chronic diseases and overall mortality. Few studies have evaluated the effect of substituting animal proteins with plant sources on the risk of incidence and mortality of various diseases. However, these data are far from conclusive [4, 18, 19].

The association between macronutrient intake and mortality may differ in Asian countries where fewer animal foods are eaten. The Iranian population consumes relatively low amounts of animal foods, with an average daily intake of about 64 g of meat and 139 g of dairy products [20, 21]. The primary sources of energy and protein are cereals, with Iranians consuming approximately 450 g of cereals each day [20, 21]. Therefore, studying the association between macronutrients and mortality is necessary to determine favourable intake levels and sources for populations in this region. To address this, we aim to evaluate the associations between animal and plant-based protein intake and their substitution with other macronutrients, and all-cause, cardiovascular, and cancer mortality in the Golestan Cohort Study.

Methods

This study utilised data from the Golestan Cohort Study (GCS). It is a prospective population-based cohort study that started in January 2004 in the Golestan province of Iran. Participants were randomly selected from individuals residing in Gonbad City and 326 villages in Golestan Province. A detailed description of the study is provided elsewhere [22]. Participants were selected using systematic cluster random sampling from three eastern districts of Golestan Province. The inclusion criteria for the cohort study were age between 40 and 75 years, residency in the district for at least 10 years, no plans to emigrate in the next 5 years, and no current or prior diagnosis of upper gastrointestinal (UGI) cancer.

At the initiation of the cohort study, participants provided written informed consent and underwent comprehensive assessments. These included a standard general questionnaire, a semi-quantitative food frequency questionnaire (FFQ), anthropometric measurements, and a physical activity questionnaire. At baseline, trained personnel interviewed 50,045 adult participants to obtain demographic and lifestyle data [22]. Questionnaires were checked by supervisors, and missing data were addressed during the next visits or through follow-up calls. Therefore, there was no missing data for the covariates included in the final models.

Socioeconomic status was assessed using a wealth score that has been described in detail elsewhere [23]. Briefly, this score is constructed by integrating information on household assets, such as ownership of a car, motorbike, television, refrigerator, freezer, vacuum cleaner, washing machine, and the home itself, as well as housing characteristics (e.g., house size) and the participant’s occupation.

Data on smoking behaviors were collected at baseline through face-to-face interviews using a structured questionnaire. Participants were asked separately about their use of cigarettes, waterpipe (hookah), and opium. The questionnaire included a history of substance use over a person’s life, dates of starting and stopping, types of drugs used, routes, and self-reported doses. Individuals were classified as “ever tobacco users” if they reported using cigarettes, waterpipes, or nass at least once per week for a minimum duration of six months. Cumulative cigarette exposure (pack-years) was calculated using information on both the duration and intensity of cigarette smoking. The quantity of opium use was validated by urinary codeine and morphine, with 93% sensitivity and 89% specificity [24]. More detailed descriptions of these assessment methods have been provided in previous publications [25, 26].

Alcohol use was very uncommon in this province and was assessed through the general questionnaire at baseline [24]. Body mass index (BMI) was calculated from weight and height measurements taken according to standard protocols at baseline. Physical activity levels were assessed using a standard questionnaire and categorised into three groups based on tertiles of a physical activity score [22]. Blood pressure was measured at baseline by trained health staff using a standardised protocol. A history of CVD, diabetes, and cancer was obtained through interviews conducted by trained personnel and verified using participants’ health records [27, 28].

Dietary intakes were assessed using a validated FFQ, designed for this population, in 49,112 subjects during face-to-face interviews at baseline. A full description of the FFQ and its validation study was reported previously [29]. It is a 116-item FFQ that asks about the frequency of intake on a daily, weekly, and monthly basis, as well as the amount of intake using household portion sizes specific to this region. Energy and nutrient intakes were calculated using the USDA Food Composition Tables [30].

Since 2004, the GCS has maintained active follow-up by contacting participants annually through telephone interviews and conducting home visits when needed to determine their vital status. Nearly all participants (98%) supplied detailed contact information, including their home address, mobile phone number, and telephone numbers for at least two relatives, which allows for reliable tracking over time. Whenever a death is identified during follow-up, a physician visits the household and completes a verbal autopsy using a standardised and validated instrument to document the circumstances and probable cause of death [28]. Additionally, relevant medical data from hospitals or pathology centers is collected to code the cause of death according to ICD-10 [28]. We used the most current updated data in 2024, resulting in a mean follow-up time of 15.05 years. Participants were excluded due to loss of follow-up (n = 1896), energy intake exceeding 2 standard deviations from the population mean (n = 909) or having cardiovascular disease or cancer at baseline (n = 3256) (supplementary graph 1).

Statistical analysis

The follow-up duration was calculated from baseline to the date of death or the last follow-up date, whichever occurred first. All macronutrient intakes were adjusted to energy intake based on the residual method proposed by Willett [31]. Cox proportional hazard regression was used to estimate HR and 95% CI for all-cause mortality, CVD mortality, and cancer mortality based on a 10-g intake of total protein, animal protein, and plant protein in the total population and subgroups, including male/female and urban/rural populations.

HRs were estimated using three models: the model A adjusted for energy intake (continuous, kcal/day), age (continuous, years), and gender (male/female); the model B further adjusted for physical activity (three levels), total opioid dose (continuous, dose/day), cigarette smoking (continuous, total pack-years), alcohol use (ever, never), residence (rural/urban), ethnicity (Turkman/others), education (five levels), wealth score (continuous/estimated score), hypertension at baseline (yes/no), diabetes at baseline (yes/no), and BMI at baseline (continuous, kg/m²). Model C was additionally adjusted for red meat, processed meat, fruit and vegetable intake, as well as daily intake of fiber, monounsaturated fatty acids (MUFA), polyunsaturated fatty acids (PUFA), saturated fatty acids (SFA), and trans-fatty acids (trans-FA), all measured in grams per day. Gender was clearly not included in the models for male/female stratified analysis, and residence was not adjusted for the urban/rural analysis. We assessed interactions between gender or residence and macronutrient intake by performing likelihood ratio tests to compare nested models differing only in their interaction terms.

Substitution analysis was performed using the leave-one-out model, which calculates the effect of substituting one type of macronutrient for another by omitting the specific macronutrient from the model while keeping the others, along with energy intake, constant [32]. The restricted cubic spline (RCS) function was applied to explore and visualise potential non-linear associations between total, animal, and plant protein intake and mortality. Hazard ratios were plotted using two-way graphs to illustrate the effects across varying intake levels. In the RCS functions, we used 5 knots and set the median of the first quartile of intake as the reference point for each protein intake type. Analyses were conducted using STATA version 18.0 (Standard Edition, College Station, Texas, USA). All reported p-values are two-sided, and p-values under 0.05 were considered statistically significant.

Results

We analysed data from 43,050 individuals (24,799 females and 18,251 males; 34,059 rural and 8,991 urban) with a mean follow-up period of 15.05 years. The mean age at baseline was 49.8 years. The majority of participants were married (88.5%) and of Turkman ethnicity (76.3%). The prevalence of cigarette smoking, opium use, and alcohol consumption in the study population was 17.1%, 16.3%, and 3.4%, respectively. Hypertension was present in 37.9% of participants, while diabetes was observed in 6.0% at baseline. General characteristics of study participants by gender (male/female), residence (rural/urban) and total protein intake class are presented in Table 1.

Table 1.

Baseline characteristics of participants

		Total population	Gender		Residence		decile of total protein intake
		(43,050)	Female (24,799)	Male (18,251)	Rural (34,059)	Urban (8,991)	3rd (4,327)	5th (4,326)	7th (4,327)
Gender, N (%)	Female	24,799(57.6)	-	-	19,312(56.7)	5,487(61.0)	2,491(10.0)	2,555(10.3)	2,579(10.4)
Residence, N (%)	Rural	34,059(79.1)	19,312(77.9)	14,747(80.8)	-	-	3,642(10.7)	3,495(10.3)	3,328 (9.8)
Ethnicity, N (%)	Turkman	32,838(76.3)	18,731(75.5)	14,107(77.3)	28,207(82.8)	4,631(51.5)	3,501(10.7)	3,304(10.1)	3,177(9.7)
Marital status, N (%)	Married	38,038(88.5)	20,257(81.9)	17,781(97.5)	30,169(88.7)	7,869(87.7)	3,871 (10.2)	3,808(10.0)	3,788 (10.0)
Educational attainment, N (%)	Illiterate	22,912(69.5)	21,070(84.9)	8,842(48.5)	25,871(75.9)	4.041(44.9)	3,235(10.8)	3,032(10.1)	2,838(9.5)
	1–5 years	7,422(17.2)	2,615(10.5)	4,807(26.3)	5,357(15.7)	2,065(22.9)	675(9.1)	741(10.0)	791(10.7)
	6–8 years	1,949(4.5)	418(1.7)	1,532(8.4)	1,120(3.3)	829(9.2)	142 (7.3)	182 (9.3)	239(12.3)
	9–12 years	2,810(6.5)	574(2.3)	2,236(12.3)	1,365(4.0)	1,445(16.1)	197(7.0)	257(9.2)	325(11.6)
	University degree	957(2.2)	123(0.5)	834(4.6)	346(1.0)	611(6.8)	57(6.0)	92(9.6)	102(10.7)
Physical activity level, N (%)	Low	14,821(34.5)	3,896(15.7)	10,925(60.0)	11,930(35.1)	2,891(32.2)	1,441(9.7)	1,421(9.6)	1,500(10.1)
	Medium	13,657(31.8)	9,768(39.5)	3,889(21.4)	10,527(30.9)	3,130(34.9)	1,328(9.7)	1,369 (10.0)	1,328(9.7)
	High	14,470(33.7)	11,084(44.8)	3,389(18.6)	11,525(33.9)	2,945(32.9)	1,524 (10.5)	1,506(10.4)	1,454(10.1)
Current smoking, N (%)	No	35,707(82,9)	24,462(98.6)	11,245(61.6)	28,318(83.1)	7,389(82.2)	3,565(10.0)	3,638(10.2)	3,601(10.1)
	Current	5,987(13.9)	280(1.1)	5,707(31.3)	4,814(14.1)	1.173(13.1)	617(10.3)	546(9.1)	548 (9.2)
	Former	1,356(3.2)	57(0.2)	1,299(7.1)	927(2.7)	429(4.8)	124(9.1)	120(8.9)	146(10.8)
Cigar smoking, N (%)	Ever	7,343(17.1)	337(1.4)	7,006(38.4)	5,741(16.9)	1,602(17.8)	3,565(10.0)	3,638(10.2)	3,601(10.1)
Cigar smoking, pack-years, median(range)		0(0,45)	0(0,0.6)	0(0,62)	0(0,45)	0(0,45)	0(0,43)	0(0,45)	0(0,44)
Opium use, N (%)	ever	6,996(16.3)	1,876(7.6)	5,120(28.1)	6,151(18.1)	845(9.4)	745(10.7)	640(9.2)	641(9.2)
Total opioid use		0(0,46929)	0(0,42916)	0(0,48328)	0(0,47484)	0(0,33387)	0(0,47194)	0(0,46934)	0(0,47437)
Alcohol use, N (%)	Ever	1.461(3.4)	18(0.1)	1,443(7.9)	709(2.1)	752(8.4)	109(7.7)	135(9.2)	170(11.6)
Hypertension at baseline, N (%)	Yes	16,306(37.9)	10,346(41.7)	5,960(32.7)	12,868(37.8)	3,438(38.2)	1,590(9.8)	1,579(9.7)	1,671(10.3)
Diabetes at baseline, N (%)	Yes	2,584(6.0)	1,790(7.2)	794(4.4)	1,831(5.4)	753(8.38)	127(4.9)	203(7.9)	271(10.5)
Wealth score level, median(range)		−0.04(−0.43,0.51)	−0.5(−0.43,0.51)	0.01(−0.43,0.51)	−0.08(−0.43,0.5)	0.15(−0.37, 0.51)	−0.08(−0.43–0.5)	−0.04(−0.37,0.51	0.3(−0.37,0.51)
Follow-up Duration median(range)		15.05(1.42,18.09)	15.1(1.9,18.1)	15.0(1.2,18.1)	15(1.5,18.1)	16.1(1.3,18.1)	16.0(1.3–18.1)	15.0(1.5–18.1)	15.0(1.6–18.1)
Age at baseline, median(range)		49.8(39.8,74.0)	49.5(39.8,73,4)	50.3(39.8,74.4)	49.7(39.8, 74.1)	50.4(39,8–73.9)	49.4(39.7,73.6)	49.2(39.8,73.8)	49.5(39.9,74,1)
BMI, median(range)		26.2(16.5,41.4)	27.5(16.6,42,9)	24.8(16.4,36.9)	25.8(16.3, 40.9)	27.9 (17.4,42.3)	25.7(16.2,40.6)	26.3(16.6,41.1)	26.6(16.8,41.9)

Energy, macronutrient and some food groups intakes of participants are presented in Table 2. The mean energy intake was 2371.9 kcal, with a standard deviation of 976.8 kcal. Energy intake was slightly higher in males compared to females, and in urban residents compared to rural ones. However, we minimised the effect of variation in energy intake by using residual-adjusted macronutrient intake in Cox models. The mean total protein intake was 71.6 g/day, with a standard deviation of 19.1 g/day. Total carbohydrate, fiber, animal protein, animal fat, fruit and vegetable intake were different in different deciles of protein intake. Total fat, total protein, and animal protein intakes were slightly higher in women compared to men, and in urban participants compared to rural ones.

Table 2.

Nutritional intake of participants at baseline

	Total population	Gender		Residence		decile of total protein intake
		Female	male	Rural	Urban	3rd (4,327)	5th (4,326)	7th (4,327)
Energy	2371.9 ± 976.8	2215.2 ± 921.4	2589.0 ± 1011.7	2366.7 ± 998.4	2400.3 ± 897.7	2229.4 ± 837.2	2166.6 ± 792.0	2184.2 ± 812.1
Total protein	71.6 ± 19.1	72.3 ± 18.3	70.8 ± 20.2	70.5 ± 18.7	75.9 ± 20.2	64.8 ± 0.9	67.9 ± 0.9	74.4 ± 1.1
Animal protein	34.8 ± 22.0	35.5 ± 21.1	33.8 ± 23.3	33.5 ± 21.5	39.5 ± 23.4	23.2 ± 6.2	30.0 ± 5.9	37.3 ± 5.8
Plant protein	36.6 ± 7.1	36.5 ± 6.7	36.8 ± 7.6	36.8 ± 7.0	35.9 ± 7.4	38.3 ± 6.0	37.7 ± 5.7	36.8 ± 5.6
Carbohydrate	350.8 ± 47.9	345.9 ± 45.8	357.4 ± 50.2	346.9 ± 49.1	351.9 ± 47.7	365.9 ± 34.1	356.9 ± 34.1	346.8 ± 31.7
Sugare	62.6 ± 36.2	58.8 ± 31.8	67.8 ± 40.9	61.1 ± 35.6	68.4 ± 37.7	63.9 ± 30.3	60.6 ± 27.4	58.5 ± 26.8
Fiber	18.6 ± 3.4	18.8 ± 3.2	18.4 ± 3.6	18.2 ± 3.0	20.1 ± 4.1	18.6 ± 2.8	18.8 ± 2.6	18.9 ± 2.9
Total Fat	54.9 ± 18.1	56.8 ± 17.6	52.4 ± 18.5	54.7 ± 18.1	55.7 ± 18.5	52.9 ± 15.1	54.1 ± 14.5	55.5 ± 14.0
Animal fat	17.8 ± 10.9	17.8 ± 10.3	17.8 ± 11.8	17.2 ± 10.7	20.9 ± 11.5	14.6 ± 7.9	16.8 ± 7.2	19.1 ± 7.6
Plant fat	36.4 ± 15.2	38.3 ± 15.1	33.8 ± 15.0	37.0 ± 15.2	34.0 ± 15.0	37.6 ± 12.0	36.7 ± 11.6	35.7 ± 11.3
Saturated fat	19.2 ± 6.7	19.8 ± 6.4	18.1 ± 7.0	19.0 ± 6.6	20.0 ± 7.0	18.6 ± 5.8	19.1 ± 5.4	19.6 ± 5.4
PUFA	16.1 ± 5.8	16.8 ± 5.8	15.2 ± 5.9	16.3 ± 5.8	15.5 ± 5.9	15.9 ± 4.7	15.9 ± 4.5	16.0 ± 4.4
MUFA	19.3 ± 6.2	19.9 ± 6.1	18.5 ± 6.3	19.2 ± 6.2	19.8 ± 6.5	18.4 ± 5.1	18.9 ± 5.0	19.5 ± 5.0
Trans FA	0.6 ± 0.3	0.6 ± 0.3	0.6 ± 0.3	0.6 ± 0.3	0.6 ± 0.3	0.6 ± 0.2	0.6 ± 0.2	0.6 ± 0.2
Red meat	12.0 ± 16.6	10.9 ± 14.7	13.6 ± 18.7	11.1 ± 15.1	15.5 ± 21.0	3.5 ± 6.1	3.7 ± 6.8	3.6 ± 7.4
Processed meat	3.7 ± 6.8	3.5 ± 6.1	4.1 ± 7.7	3.9 ± 6.9	3.3 ± 6.7	3.6 ± 6.8	4.0 ± 6.7	4.3 ± 7.3
Fruits	156.8 ± 123.2	147.9 ± 107.6	169.1 ± 141.3	142.3 ± 106.2	212.3 ± 162.6	144.9 ± 109.5	152.4 ± 102.5	162.4 ± 113.2
Vegetables	106.5 ± 67.7	10.6.4 ± 66.9	106/7 ± 69.0	151.1 ± 92.8	94.8 ± 53.6	95.3 ± 57.0	101.8 ± 59.4	111.8 ± 64.2

Values are presented in grams per day, adjusted to total energy intake (kcal/day), and are expressed as mean ± standard deviation

During 15.05 years of follow-up, 9309 deaths were recorded, 3707 from CVD and 1917 from cancers. The HR of 10 g intake of total, animal, and plant protein with all-cause, CVD, and cancer death in different groups is illustrated in Table 3. Total protein intake was associated with higher all-cause mortality (HR 1.02, 95% CI 1.01–1.03) and CVD mortality (HR 1.02, 95% CI 1.00–1.03) but not cancer mortality in the total population. There was a significant positive association between total protein intake and all-cause mortality among females (HR 1.03, 95% CI 1.01–1.04) and among rural participants (HR 1.02, 95% CI 1.01–1.03), and these associations remained significant after additional adjustment for other nutritional factors in Model C.

Table 3.

The hazard ratio for all-cause, cardiovascular, and cancer mortality in different sources of protein intake^a in the Golestan cohort study

Cause of mortality	Total population	Gender			Residence
		female (24,799)	Male (18,257)	Pb	Rural (34,059)	Urban (8,991)	Pb
Total protein intake
All-cause mortality
No. of death	9309	4363	4946	-	7453	1856	-
Model Ac	1.01(1.00–1.02)	1.03(1.01–1.04)	1.00(0.99–1.01)	< 0.001	1.03(1.02–1.04)	0.99(0.98–1.02)	< 0.001
Model Bd	1.02(1.01–1.03)	1.03(1.01–1.04)	1.01(0.99–1.02)	0.06	1.02(1.01–1.03)	1.00(0.98–1.02)	0.01
Model Ce	1.02(1.01–1.03)	1.03(1.02–1.05)	1.01(0.99–1.03)	0.05	1.02(1.01–1.04)	1.01(0.99–1.03)	0.01
CVD mortality
No. of death	3,707	1800	1907	-	2967	740	-
Model Ac	1.02(1.01–1.04)	1.03(1.00–1.05)	1.02(1.00–1.04)	0.18	1.04(1.02–1.06)	1.01(0.98–1.04)	0.59
Model Bd	1.02(1.00–1.03)	1.01(0.99–1.04)	1.02(1.00–1.04)	0.61	1.02(1.01–1.04)	1.01(0.97–1.04)	0.12
Model Ce	1.02(1.01–1.04)	1.03(1.00–1.06)	1.02(1.00–1.04)	0.001	1.03(1.01–1.05)	1.02(0.98–1.05)	0.08
Cancer mortality
No. of death	1917	821	1096	-	1566	351	-
Model Ac	0.97(0.95–0.99)	0.98(0.94–1.02)	0.96(0.93–0.99)	< 0.001	0.98(0.95–1.01)	0.96(0.91–1.02)	0.73
Model Bd	0.99(0.97–1.01)	0.99(0.96–1.03)	0.99(0.96–1.02)	0.78	0.99(0.97–1.02)	0.98(0.92–1.03)	< 0.001
Model Ce	0.99(0.96–1.01)	0.99(0.96–1.03)	0.99(0.95–1.02)	0.81	0.99(0.96–1.02)	0.98(0.92–1.04)	0.80
Animal protein intake
All-cause mortality
No. of death	9309	4363	4946	-	7453	1856	-
Model Ac	1.01(1.00–1.02)	1.02(1.01–1.03)	0.99(0.99–1.01)	0.08	1.02(1.01–1.03)	0.99(0.97–1.01))	< 0.001
Model Bd	1.01(1.01–1.02)	1.02(1.01–1.04)	1.01(0.99–1.02)	0.05	1.02(1.01–1.03)	0.99(0.98–1.02)	0.13
Model Ce	1.02(1.01–1.03)	1.03(1.02–1.05)	1.01(0.99–1.03)	0.13	1.03(1.01–1.04)	1.01(0.99–1.03)	< 0.01
CVD mortality
No. of death	3,707	1800	1907	-	2967	740	-
Model Ac	1.01(1.00–1.03)	1.01(0.99–1.04)	1.01(0.99–1.03)	0.59	1.03(1.01–1.04)	0.99(0.96–1.02)	< 0.01
Model Bd	1.02(1.00–1.03)	1.01(0.99–1.03)	1.02(1.00–1.04)	0.05	1.02(1.01–1.04)	1.00(0.97–1.03)	0.86
Model Ce	1.02(1.01–1.04)	1.03(1.00–1.05)	1.02(1.00–1.04	0.84	1.03(1.01–1.04)	1.02(0.99–1.05)	0.85
Cancer mortality
No. of death	1917	821	1096	-	1566	351	-
Model Ac	0.98(0.96–0.99)	0.99(0.95–1.02)	0.97(0.94–0.99)	0.01	0.99(0.96–1.01)	0.97(0.92–1.02)	0.84
Model Bd	0.99(0.97–1.01)	0.99(0.97–1.03)	0.99(0.96–1.02)	0.02	0.99(0.98–1.02)	0.98(0.93–1.03)	0.01
Model Ce	0.99(0.96–1.01)	0.99(0.96–1.03)	0.99(0.96–1.02)	0.01	0.99(0.96–1.02)	0.98(0.93–1.03)	< 0.01
Plant protein intake
All-cause mortality
No. of death	9309	4363	4946	-	7453	1856	-
Model Ac	1.02(0.99–1.05)	1.02(0.98–1.07)	1.01(0.98–1.05)	< 0.001	0.98(0.95–1.01)	1.14(1.07–1.21)	0.001
Model Bd	0.97(0.95–1.00)	0.95(0.91–0.99)	0.99(0.95–1.03)	0.27	0.96(0.93–0.99)	1.02(0.96–1.09)	0.81
Model Ce	0.93(0.89–0.97)	0.91(0.84–0.98)	0.95(0.89–1.01)	0.23	0.91(0.86–0.96)	1.01(0.91–1.13)	0.73
CVD mortality
No. of death	3,707	1800	1907	-	2967	740	-
Model Ac	1.05(1.00–1.09)	1.08(1.01–1.16)	1.03(0.97–1.09)	< 0.001	1.01(0.96–1.06)	1.18(1.06–1.31)	< 0.01
Model Bd	0.97(0.93–1.02)	0.97(0.90–1.03)	0.97(0.91–1.03)	0.047	0.96(0.91–1.01)	1.02(0.92–1.14)	0.61
Model Ce	0.92(0.86–0.99)	0.88(0.78–0.99)	0.96(0.87–1.06)	0.29	0.92(0.85–1.00)	0.89(0.76–1.06)	0.11
Cancer mortality
No. of death	1917	821	1096	-	1566	350	-
Model Ac	1.01(0.95–1.08)	1.01(0.91–1.12)	1.02(0.94–1.09)	0.02	0.99(0.93–1.06)	1.07(0.92–1.23)	0.16
Model Bd	0.99(0.94–1.06)	0.97(0.88–1.07)	1.02(0.94–1.11)	0.51	0.99(0.93–1.06)	1.06(0.91–1.23)	0.87
Model Ce	1.02(0.92–1.14)	1.02(0.86–1.22)	1.04(0.91–1.18)	0.52	1.01(0.90–1.14)	1.14(0.89–1.47)	0.30

^aHazard ration per 10 g of protein intake is reported

^bP for interaction, based on the likelihood ratio test between models with and without the interaction term

^cAdjusted for energy intake (continues, kcal/day), age (continuous, year), and gender (male/female)

^dFurther adjusted for physical activity (three-level), total opioid dose (continues, dose/d), cigar smoking (continuous, total packyear), alcohol use (ever, never), residence (rural/urban), ethnicity (Turkman/others), education (five-level), wealth score (continuous/an estimated score), hypertension at baseline (yes, no), diabetes at baseline(yes, no), and BMI at baseline(continuous/kg/m²). In the stratified analyses by residence and sex, these variables were excluded from the corresponding models

^eAdditionally adjusted for red meat, processed meat, fruit and vegetable intake, as well as daily intake of fiber, monounsaturated fatty acids (MUFA), polyunsaturated fatty acids (PUFA), saturated fatty acids (SFA), and trans-fatty acids (trans-FA), all measured in grams per day

The association between animal protein intake and all-cause, CVD, and cancer mortality was similar to that of total protein intake in the total population and subgroups. Animal protein intake increased the risk of all-cause mortality in the total population (HR 1.01, 95% CI 1.01–1.02), females (HR 1.02, 95% CI 1.01–1.04), and rural participants (HR 1.02, 95% CI 1.01–1.03). It also increased the risk of CVD mortality in the total population (HR 1.02, 95% CI 1.00–1.03), males (HR 1.02, 95% CI 1.00–1.04), and rural participants (HR 1.02, 95% CI 1.01–1.04).

Plant protein intake was associated with decreased risk of all-cause mortality in the total population (HR 0.97, 95% CI 0.95–1.00). The association was significant in females (HR 0.95, 95% CI 0.91–0.99) and rural participants (HR 0.96, 95% CI 0.93–0.99). The association between plant protein intake and risk of cancer or CVD mortality were not significant in any group. Moreover, the differences in risk between males and females, and rural and urban participants, were not significant for protein intakes from different sources.

The association between substituting 10 g of protein intake from different sources with 10 g of other macronutrients is shown in Table 4. Substitution of total protein with carbohydrate (HR 0.99, 95% CI 0.98–0.99) and fat (HR 0.97, 95% CI 0.95–0.99) was associated with a lower risk of all-cause death. The association was similar for CVD death for carbohydrate (HR 0.98, 95% CI 0.97–0.99) and fat (HR 0.97, 95% CI 0.93–1.00). The association between substituting animal protein with carbohydrates (HR for all-cause mortality 0.99, 95% CI 0.98–0.99 and CVD mortality 0.98, 95% CI 0.97–0.99) and fat (HR for all-cause mortality 0.97, 95% CI 0.94–0.99 and CVD mortality 0.96, 95% CI 0.93–1.00) was similar to that of total protein. Substituting animal protein, carbohydrate or fat for plant protein did not change the risk of death significantly.

Table 4.

The hazard of substituting 10 g of protein intake with other macronutrients and the risk of death

Total protein
Substituted with		Carbohydrate	Fat
All-cause mortality
Crude	-	0.96(0.95–0.97)	0.88(0.86–0.90)
Model A	-	0.98(0.97–0.99)	0.95(0.93–0.97)
Model B	-	0.99 (0.98–0.99)	0.97(0.95–0.99)
CVD mortality
Crude	-	0.95(0.94–0.96)	0.86(0.82–0.89)
Model A	-	0.97(0.96–0.99)	0.92(0.89–0.96)
Model B	-	0.98(0.97–0.99)	0.97(0.93–1.00)
Cancer mortality
Crude	-	1.00(0.98–1.03)	0.98(0.92–1.04)
Model A	-	1.02(0.99–1.05)	1.04(0.98–1.10)
Model B	-	1.01(0.99–1.03)	1.02(0.96–1.07)

Plant protein
Substituted with	Animal protein	Carbohydrate	Fat
All-cause mortality
Crude	1.08(1.05–1.12)	1.04(1.00–1.07)	1.02(0.96–1.09)
Model A	0.97(0.94–0.99)	0.95(0.92–0.98)	0.89(0.83–0.95)
Model B	1.03(0.99–1.06)	1.01(0.98–1.05)	1.02(0.96–1.09)
CVD mortality
Crude	1.04(0.99–1.09)	0.99(0.94–1.04)	0.92(0.83–1.03)
Model A	0.93(0.88–0.98)	0.91(0.86–0.96)	0.80(0.72–0.89)
Model B	1.01(0.96–1.07)	0.99(0.94–1.05)	0.99(0.89–1.09)
Cancer mortality
Crude	1.08(1.01–1.16)	1.08(1.01–1.16)	1.13(0.98–1.29)
Model A	0.98(0.91–1.05)	0.99(0.93–1.07)	0.99(0.87–1.15)
Model B	1.02(0.95–1.09)	1.03(0.97–1.10)	1.05(0.91–1.21)

Animal protein
Substituted with	Vegetable protein	Carbohydrate	Fat
All-cause mortality
Crude	0.87(0.84–0.91)	0.97(0.96–0.98)	0.87(0.84–0.89)
Model A	1.01(0.97–1.06)	0.98(0.97–0.99)	0.95(0.93–0.97)
Model B	0.97(0.93–1.01)	0.99(0.98–0.99)	0.97(0.94–0.99)
CVD mortality
Crude	0.93(0.87–0.99)	0.95(0.94–0.97)	0.85(0.82–0.88)
Model A	1.08(1.01–1.16)	0.97(0.95–0.98)	0.93(0.89–0.97)
Model B	0.98(0.92–1.05)	0.98(0.97–0.99)	0.96(0.93–1.00)
Cancer mortality
Crude	0.84(0.77–0.91)	1.01(0.99–1.03)	0.95(0.89–1.01)
Model A	0.96(0.88–1.05)	1.02(1.00–1.05)	1.03(0.97–1.09)
Model B	0.97(0.89–1.07)	1.01(0.99–1.03)	1.01(0.95–1.07)

^aAdjusted for age (continuous, year), and gender (male/female)

^bFurther adjusted for physical activity (three-level), total opioid dose (continues, dose/d), cigar smoking (continuous, total packyear), alcohol use (ever, never), residence (rural/urban), ethnicity (Turkman/others), education (five-level), wealth score (continuous/an estimated score), hypertension at baseline (yes, no), diabetes at baseline(yes, no), and BMI at baseline(continuous/kg/m²)

The associations between the range of protein intake from different sources and the risks of all-cause, CVD, and cancer mortality are presented in Figs. 1, 2 and 3. A statistically significant nonlinear association was observed for both all-cause and cardiovascular mortality across levels of total protein intake. In contrast, the association with cancer mortality was neither clear nor statistically significant. Overall, the pattern across the three outcomes suggests that a daily intake of approximately 70 g of total protein, comprising roughly 40 g/day of plant protein and 30 g/day of animal protein, may represent a reasonable and balanced intake based on the findings of this study.

Fig. 1.

Hazard for all-cause mortality across different sources of protein intake in the Golestan cohort study. ¹The presented p-value corresponds to the test of the significance of the nonlinear association

Fig. 2.

Hazard for cardiovascular mortality in different sources of protein intake in the Golestan cohort study. ¹The presented p-value corresponds to the test of the significance of the nonlinear association

Fig. 3.

Hazard for cancer mortality across different sources of protein intake in the Golestan Cohort Study. ¹The presented p-value corresponds to the test of the significance of the nonlinear association

Discussion

In our study of 43,050 individuals with a 15-year follow-up, higher intakes of total and animal protein were associated with an increased risk of all-cause and CVD mortality. Conversely, plant protein intake was associated with a reduced risk of all-cause mortality, although this association was not significant for CVD and cancer mortality. Substituting total protein intake with carbohydrates or fats was associated with a lower risk of all-cause and CVD mortality. Similar associations were observed when substituting animal protein. In this study, total protein intake around 70 g per day, comprising 40 g of plant protein and approximately 30 g of animal protein, was associated with lower mortality. More studies are needed to confirm this result.

This study found that higher protein intake, particularly from animal sources, was associated with an increased risk of mortality. Although the overall nonlinear dose–response association was statistically significant, the association was not apparent at lower intake levels. The absence of a significant association at lower intake levels may be due to the fact that protein intake in this population exceeded dietary guideline recommendations, even within the lowest decile of consumption. While several studies have shown that low protein intake in older adults, particularly those with chronic diseases, is associated with increased mortality [33–36], evidence regarding this association in healthy, middle-aged populations remains limited and inconclusive [37, 38]. Part of the scarcity of data on low protein intake is attributable to the fact that, under normal conditions, protein consumption among healthy individuals generally meets or exceeds dietary requirements, especially in regions where most epidemiologic studies have been conducted [39]. Protein plays a crucial role in maintaining muscle mass, supporting immune function, and ensuring overall metabolic health. Insufficient protein consumption has been associated with sarcopenia, a weakened immune response, and impaired wound healing, which could increase mortality [37, 40–42].

The association between high protein intake, the substitution of animal protein with plant protein, and mortality have been studied in multiple studies. A cohort study in the US population with a 15-year follow-up found that higher plant protein intake and substituting animal proteins with plant proteins could reduce all-cause and CVD mortality in both men and women [43]. A cohort study conducted in Japan reported comparable findings for plant protein intake and the substitution of animal protein with plant protein; however, intake of animal protein was not associated with all-cause or cardiovascular disease mortality [44]. The Rotterdam cohort study showed that higher intake of total and animal protein is associated with a higher risk of all-cause and CVD mortality, whereas plant protein intake could reduce these risks [45]. The Isfahan cohort study demonstrated a positive association between animal protein intake and a negative association with plant protein intake with all-cause mortality. Total protein intake was not associated with mortality in their study [46].

A meta-analysis of 32 cohort studies by Nagshi et al. demonstrated that plant protein intake reduces all-cause and CVD mortality, but not cancer mortality. Total protein intake was associated with a lower risk of all-cause mortality. However, animal protein and total protein intake were not associated with cancer or CVD mortality. The study also found heterogeneity between studies and a scarcity of data from low- and middle-income countries [47].

The association between higher protein intake from different sources and the risk of all-cause mortality is primarily explained by CVD, with cancer risk contributing to a lesser extent. The link between protein intake and CVD risk can be attributed to its effects on weight management, blood pressure, and lipid control [6, 48]. The association between protein intake from various sources and cancer risk is mainly related to obesity and insulin-like growth factor 1 (IGF-1) [36]. Additionally, the other components of animal and plant protein foods can influence the risk of CVD and cancer. Several studies have demonstrated that lower intake of animal protein and a higher intake of plant protein can favorably alter the gut microbiome. These changes are associated with a reduced risk of cancers, particularly gastrointestinal cancers, and have beneficial effects on blood sugar and cholesterol levels, weight, and body composition [49, 50].

There are other components in animal and plant foods, besides protein, that should be considered when explaining the association between protein intake and mortality risk. Plant foods have lower levels of SFA but higher levels of MUFA and PUFA, fiber, and other bioactive compounds, which could reduce the risk of cardiovascular diseases and cancers [51–53]. Therefore, the observed associations should be interpreted with caution, as they may reflect the influence of other components present in animal- and plant-based foods rather than protein alone. However, after further adjustment for red and processed meat intake, fruit and vegetable consumption, and daily intake of fiber, MUFA, PUFA, SFA, and trans fatty acids in Model C in this study, the associations remained significant. This suggests that the protective effect of plant protein and the adverse effect of animal protein are not fully explained by these dietary components.

Beyond the well-documented health benefits of reducing animal protein intake in favor of plant protein, this dietary shift has gained increasing attention due to its environmental implications [11, 12]. Compared with plant protein production, animal protein production is associated with substantially higher carbon emissions, water use, and land requirements [11, 12, 54]. On average, an animal-based diet may exert up to 1.5 times the environmental impact of a plant-based diet [54]. Additionally, ethical concerns raised by animal rights advocates often oppose the consumption of animal protein due to issues related to animal welfare [11]. Considering these factors collectively, replacing animal protein with plant protein represents a meaningful step toward achieving a more sustainable diet, as recommended by the World Health Organization [11, 55]. However, both the environmental footprint and, possibly, the associated health effects differ markedly among various animal-source foods [56–58]. Future research should assess the health impacts of various animal- and plant-derived proteins with different environmental footprints.

This study has several strengths, including a large sample size, the use of a validated food frequency questionnaire, and both substitution and dose–response analyses. In addition, the high follow-up rate (> 96%) reduces the potential for bias. The similarity between the protein and macronutrient intake observed in our study and those reported in the national food survey conducted during the same period further supports the representativeness and reliability of our data [59]. This underscores that our study population is highly representative of the general population. Importantly, this is the first study to investigate the relationship between protein intake from animal and plant sources, substitution patterns, and mortality risk in middle-income countries and the Middle East, while also accounting for potential nonlinear associations at both low and high intake levels.

However, this study also has certain limitations. Although we used a validated FFQ with demonstrated accuracy when compared with 24-h dietary recalls and biological markers [29], the potential for recall bias and misclassification inherent to FFQs cannot be fully eliminated. These sources of error are well recognised, but they typically result in non-differential misclassification, which tends to bias estimates toward the null rather than inflate associations. Therefore, such errors are unlikely to significantly affect the observed association [60].

Moreover, FFQs provide a method for classifying individuals according to their habitual intake rather than precisely quantifying absolute consumption [61]. Therefore, the proposed optimal level of protein intake identified in this study should be interpreted as preliminary and viewed primarily as a hypothesis-generating finding for future research. It should be applied with caution and not considered a definitive basis for dietary guidelines.

Moreover, this FFQ has been validated for assessing protein intake (based on 24-h dietary recalls and urinary nitrogen), as well as carbohydrate, fat, and other macro- and micronutrient intakes [29], and it is also valid for estimating intakes from various food groups of both animal and plant origin [62]. Based on these validations, we considered it appropriate to use the FFQ for assessing plant- and animal-derived protein intake; however, these were not evaluated separately.

Another limitation is that dietary intake was assessed only once at baseline, without accounting for potential changes during the extended follow-up period. However, evidence suggests that in the absence of significant life events or interventions, individuals' dietary patterns tend to remain relatively stable over time [63].

In this study, we examined the associations between substituting total, animal, and plant protein with total fat and total carbohydrates. However, our analyses did not account for the specific dietary sources or subtypes of these macronutrients, for example, distinctions between saturated and unsaturated fats, whole-grain versus refined carbohydrates, or protein derived from different plant or animal foods. As a result, the findings reflect broad macronutrient categories rather than the potentially heterogeneous health effects of their individual components.

We should account for the confounding effects of various health-related factors, including socioeconomic status, particularly education, and physical activity levels. Numerous studies, both in Iran and other countries, have shown that higher socioeconomic status is associated with increased intake of protein, especially animal proteins [21, 64–67]. Individuals who adhere to healthier diets are also more likely to engage in higher levels of physical activity and have reduced exposure to life-threatening factors [68–71]. Although we adjusted for these factors, we cannot be sure of not having any residual effects.

In our study, higher intakes of total and animal protein were associated with a higher risk of all-cause and CVD mortality. In contrast, plant protein intake demonstrated a protective association with reduced all-cause mortality. Furthermore, substituting total protein or animal protein intake with carbohydrates or fats was associated with a decreased risk of all-cause and CVD mortality. Consuming a moderate amount of protein composed of a balanced amount of animal and plant-based protein appears to be associated with increased longevity.

Abbreviations

BMI

Body mass index

CI

Confidence Interval

CVD

Cardiovascular Diseases

GCS

Golestan Cohort Study

HR

Hazard Ratio

FFQ

Food Frequency Questionnaire

MUFA

Monounsaturated Fatty Acids

PUFA

Polyunsaturated Fatty Acids

SFA

Saturated Fatty Acids

trans-FA

Trans-Fatty Acids

Authors’ contributions

HP, AE, FK, CCA, SD, and RM were PIs of the original study. FT and AH designed this study on existing data. FT designed the recent study and performed the statistical analysis in consultation with BS and MS. FT, with the contribution of PB, wrote the draft. AH was primarily responsible for the final content; all authors reviewed and approved the final manuscripts.

Funding

This study was partially supported by a grant from the Italian Association for Cancer Research (AIRC, grant No. 24706 IG). The original study was supported by a fund from the Digestive Diseases Research Institute (No I.R. TUMS 82–603).

Data availability

Study data are available on reasonable request from RM.

Declarations

Ethics approval and consent to participate

The study protocol of the Golestan Cohort study was approved by the Ethics Committee of the Digestive Diseases Research Institute of Tehran University of Medical Sciences (no I.R. TUMS 82–603), the International Agency for Research on Cancer (IARC) from the World Health Organization, the US National Cancer Institute, and the Harvard T.H. Chan School of Public Health. All procedures adhered to the ethical principles outlined in the Declaration of Helsinki. All participants provided written informed consent after receiving complete information about the study procedures. No identifiable participant data were published.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.