The Skinny on High-Protein, Low-Carbohydrate Diets


Wendy Cunningham, PhD, RD, ETT, California State University, Family & Consumer Sciences Department, 6000 J Street, Sacramento, CA 95819-6053 E-mail:


Short-term studies of high-protein, low-carbohydrate diets have shown weight loss and improvements in plasma lipid profiles. Studies of greater than 6 months' duration, however, have failed to show continued benefit of high-protein, low-carbohydrate diets on weight loss and cardiovascular risk factors compared with conventional diets. Without concurrent weight loss and caloric restriction, these diets offer no additional benefit to lipids or body weight over other weight-loss regimens. In fact, high-protein, low-carbohydrate diets may add additional risk to individuals with cardiovascular disease due to their high fat and cholesterol content combined with decreased intake of fruits, vegetables, whole grains, and other nutrients related to cardiovascular risk. In addition, high-protein, low-carbohydrate diets have been implicated in other risks, including impaired renal, bone, and gastrointestinal health.

The clinical use of high-protein, low-carbohydrate diets (HPLCDs) to promote weight loss and lipid lowering has become popular with patients and practitioners. The primary emphasis of these diets is to drastically lower carbohydrate intake to induce a mild ketotic state with the goal of promoting weight loss. In HPLCDs, dietary carbohydrate intake is typically restricted to 20–30 g/d, representing the equivalent of less than two servings of carbohydrate-containing foods such as bread/grains or fruit/vegetables.

Several studies have shown that HPLCDs are successful in achieving weight loss and improving plasma lipid profiles for triglycerides (TGs) and high-density lipoproteins (HDLs). Recent data, however, have called into question the long-term efficacy of these diets and suggest possible adverse health effects associated with sustained restrictions of dietary carbohydrate. In this short review, the results of studies using HPLCDs (Table) will be reviewed in terms of potential nutritional implications related to the risk of coronary artery disease.

Table TABLE.  Weight Loss and Lipid Changes Seen in Studies Investigating High-Protein, Low-Carbohydrate Diets
   Change From Baseline at Completion of Study Period
StudyNDietWeight (kg* [s/ns])TC (mg/dL* [s/ns])LDL (mg/dL* [s/ns])HDL (mg/dL* [s/ns])TG (mg/dL* [s/ns])
12-Week Studies
Landers et al.191HPLC−5.24±2.85 (ns)NRNRNRNR
  LFC−4.44±3.21 (ns)NRNRNRNR
  Zone−5.4±2.75 (ns)NRNRNRNR
Sondike et al.316HPLC−9.9±9.3 (s)−3.7±18 (ns)3.8±13 (s)3.8±7.2 (ns)−48.3±29 (s)
  LFC−4.1±4.9 (s)−17.3±15.8 (ns)−25.1±25.3 (s)1.8±7.7 (ns)−5.9±70 (s)
6-Month Studies
Westman et al.251HPLC−9.0±5.3−11±26−10±2510±8−56±45
Samaha et al.4132HPLC−5.8±8.6 (s)1.1±34% (ns)4.39±23% (ns)0±5% (ns)−20±43% (s)
  LFC−1.9±4.2 (s)−0.5±29% (ns)2.54±18% (ns)−2.43±7% (ns)−4±31% (s)
Yancy et al.5120HPLC±0.3−13.7 (ns)1.6 (ns)5.5 (s)−74.2 (s)
  LFC−1.1−0.21 (ns)−7.4 (ns)−1.6 (s)−27.9 (s)
Brehm et al.653HPLC−8.5±1 (s)−0.90.866.96−3.49
  LFC−3.9±1 (s)−1.6−64.11.75
12-Month Studies
Foster et al.763HPLC−4.4±6.7% (ns)0.1±9.8% (ns)0.31±16.6% (ns)11±19.4% (s)−17±23% (s)
  LFC−2.5±6.3% (ns)−2.9±8.0% (ns)−3.1±12.0% (ns)1.6±11.1% (s)0.7±3.7% (s)
Stern et al.8132HPLC−5.1±8.7 (ns)(ns)(ns)−1±7 (s)−58±158 (s)
  LFC−3.1±8.4 (ns)(ns)(ns)−5±6 (s)4±85 (s)
Fleming918HPLC−13.7%4.3% (ns)6.0% (ns)−5.8% (ns)5.5% (ns)
 38LFC−12.6%−30.4%−38.8%3.6% (ns)−36.9%
 1610% fat−18.4%−39.1%−52.0%9.0%−37.3%
Dansinger et al.1040HPLC−2.1±4.8−4.3±23 (ns)−7.1±24 (ns)3.4±7.1 (s)−1.2±84
 40Zone−3.2±6.0−10.1±35 (ns)−11.8±34 (s)3.3±10.3 (s)2.5±147
 40Weight Watchers**−3.0±4.9−8.2±24 (s)−9.38±27 (s)3.4±9.91 (s)−12.7±61
 40Ornish (<10%fat)−3.3±7.3−10.8±21 (s)−12.6±19 (s)−0.5±6.55.6±36
Data are presented as mean ± SD except where SD not reported; s=significant differences found between diets; ns=no significant difference between diets; TC=total cholesterol; LDL=low-density lipoprotein; HDL=high-density lipoprotein; TG=triglycerides; HPLC=high-protein, low-carbohydrate as described in text; NR=not reported; LFC=30% (unless otherwise indicated) fat, calorie-controlled; *except where % change from baseline given instead of value; **Weight Watchers International, New York, NY


Studies of Less Than 6 Months' Duration

Several reports have addressed the use of HPLCDs for short periods (up to 6 months). Landers et al.1 compared a “ketogenic” HPLCD (<10% carbohydrate) and the “Zone” diet (moderate protein restriction with 40% carbohydrate) to a conventional 50% carbohydrate hypocaloric diet in healthy normal subjects. Weight loss over a 12-week period ranged from −4.4 to −5.4 kg on the diets and was not significantly different between the groups. Change in body composition, as determined by dual-energy x-ray absorptiometry, was also similar between groups. Plasma lipids were not reported in this study. The high attrition rate on the HPLCD in this and other studies is noteworthy, including subject losses of 43%, 60%, and 36% on the HPLCD, Zone, and hypocaloric diet, respectively.

In addition to evaluating weight loss, many investigations have shown an effect of HPLCDs on plasma lipids or other metabolites. An observational trial2 following 51 overweight or obese subjects on an HPLCD for 6 months reported a −10.3±5.9% (−9.0±5.3 kg) weight loss in compliant subjects (drop-out rate was 20%). There were significant reductions in plasma total cholesterol (TC)(−11±26 mg/dL), low-density lipoprotein (LDL) cholesterol (−10 ±5 mg/dL) and TGs (−56±45 mg/dL) and an increase in HDL (10±8 mg/dL). Other metabolic parameters were altered, including increased blood urea nitrogen and an increase in urinary calcium and uric acid excretion, but there was no change in 24-hour urine creatinine clearance or protein excretion at 24 weeks.

Another trial3 showed similar weight loss (−9.9±9.3 kg) but only modest reduction in LDL level (−2.8 ±13 mg/dL) in a 12-week, randomized HPLCD study of 16 overweight adolescents. A control group of overweight adolescents on a 30% fat diet lost less weight (−4.1 ±4.9 kg) but had a significantly greater reduction in LDL level (−25.1 ±25.3 mg/dL) than those on the HPLCD. Plasma TG levels were greatly reduced by the HPLCD (−48.3 ±29.0) compared with −5.9±70.0 mg/dL on the low-fat diet.3

Another 6-month study4 of HPLCDs compared with a “low-fat” diet (30% fat plus 500-cal/d deficit) showed greater weight loss (−5.8 ±8.6 kg) in the group on an HPLCD compared with −1.9±4.2 kg in the low-fat diet group. Although the level of plasma TG was reduced significantly by the HPLCD (−20 ±43%) compared with the low-fat diet (−4±31%), there was no significant difference in TC, LDL-C, or uric acid excretion between the groups. In a diabetic subgroup, however, there was improved insulin sensitivity (6±9% on the HPLCD vs. −3 ±8% on the 30% fat diet). The amount of weight lost and assignment to the HPLCD were reported as independent predictors of the improvements seen in TG levels and insulin sensitivity.4

A more recent study5 of overweight and hyperlipidemic subjects reported data in terms of expected outcomes over 6 months. Sixty subjects on a HPLCD demonstrated an expected outcome of 12.9% weight loss (average body weight declined from 98 to 85 kg) compared with 6.7% expected weight loss (from 98 to 93 kg average body weight) in 60 subjects on a “low-fat” diet (<30% of calories from fat with a 500–1000 cal/d deficit). In the study, however, weight loss reported was assuming that the expected outcome occurred. The actual weight change seen in subjects completing the HPLCD was a gain of 0.3 kg and of those completing the low fat diet a weight loss of −1.1 kg based on actual weights provided. Significantly reduced plasma TG and increased HDL in the subjects on the HPLCD were reported.

Weight reduction on HPLCDs is not always accompanied by improved plasma lipids. In one study6 of over 6 months, 42 out of 53 recruited healthy overweight women had an average weight loss of −9.32% (−8.5±1.0 kg) on the HPLCD vs. −4.22% (3.9±1.0 kg) on a 30%-fat, 55%-carbohydrate, calorie-restricted diet. There was no significant difference, however, in serum lipids or bone mineral content between the groups at 3 and 6 months, although both groups had improved lipid profiles during the course of the study (significant decreases in TC, LDL, and TGs at 3 months and significant increases in HDL at 6 months).

Collectively, these studies demonstrate that short-term use of HPLCDs is effective in promoting weight loss and consistently lowers plasma TG levels during the period of weight loss; tends to positively, although less reliably, improve LDL levels; and generally improves HDL profiles. These outcomes are not achieved, however, without a cost. Reported side effects of HPLCDs include constipation (68% incidence), halitosis (38%–63%), headaches (51%–60%), muscle cramps (35%), diarrhea (23%), general weakness (25%), and others.2,5 While downplayed by some investigators as “minor” side effects, sustained constipation and dehydration accompanied by high-protein intake are likely to have adverse effects on gastrointestinal and renal function over a longer period of time, as discussed further below.

Studies of Greater Than 12 Months' Duration

The relative advantage of the HPLCD in weight loss reported in the short term is not sustained in subjects following an HPLCD for longer periods of time. Several recent reports have examined outcomes after HPLCDs are followed for 12 months or more.

One study7 included 33 subjects on an HPLCD vs. 30 subjects on a high-carbohydrate, low-fat, low-calorie “conventional” diet and found no significant weight difference between the two groups (−4.4±6.7% vs. −2.5±6.3%) after 12 months. Both diets significantly lowered diastolic blood pressure (BP) and insulin response to an oral glucose load. There were no significant differences between groups in TC or LDL concentrations after 12 months on the diets. There was a significant change in TG levels (−17±23.0% vs. 0.7±37.7%), however, and an increase of 11±19.4% vs. 1.6±11.1% in HDL level on the HPLCD compared with the conventional diet.7

A subsequent study by a different investigative group8 also found no significant difference in weight loss (−5.1 ±8.7 kg vs. −3.1 ±8.4 kg) between an HPLCD (<30 g carbohydrate/d) vs. a “conventional diet” of <30% calories from fat plus a restriction of 500 cal/d, respectively, over 12 months. There was a significant change in TGs (−58±158 mg/dL vs. +4±85 mg/dL) and less of a decrease in HDL (−1±7 mg/dL vs. −5±6 mg/dL) in patients on the HPLCD, along with an improvement in hemoglobin A1c in a small subgroup of patients with diabetes in this study. However, no significant differences were seen in the change of TC or LDL between the groups. In the HPLCD group, there was a greater increase in blood urea nitrogen. Adverse events reported on the HPLCD included one patient hospitalized for noncardiac chest pain, two deaths related to hyperosmolar coma complications, and one death from severe ischemic cardiomyopathy. Of the 132 enrolled subjects, follow-up at 12 months was available for only 79 subjects.

In another study,9 four diets ranging from HPLCD to a very low-fat, high-carbohydrate diet were compared over 12 months in 100 subjects. There was significant weight loss after following either a very low-fat (<10% fat) diet, moderate-fat calorie-controlled diet (<30% fat), or high-fat diet (HPLCD), but no significant difference between these groups. There were no changes in body fat in the fourth diet group (moderate fat without caloric restriction). Subjects on the moderate-fat, calorie-controlled and very low-fat diets experienced significant reductions in TC of −30.4% and −39.1%, along with significant reductions in LDL of−38.8% and −52.0% on each of these diets, respectively. Although not significant, there was an increase in TC of 4.3%, LDL of 6.0%, and 9.8% increase in the TC/HDL cholesterol ratio on the HPLC diet, with reductions in all lipids on the very low-fat and calorie-restricted diets. It is also noteworthy that this study showed potential adverse effects on clinical markers of coronary artery disease after 1 year on the HPLCD, including increased levels of plasma homocysteine, fibrinogen, and lipoprotein(a). In contrast, with varying degrees of fat and caloric restriction, there were improvements in these markers, although a moderate-fat diet, without caloric restriction, was associated with increased homocysteine in the absence of weight loss.9 Although not all of these outcomes were statistically significant, there was clearly a trend toward worsening of these biomarkers on the HPLCD and improvement on the lower-fat, high-carbohydrate protocol.

A very recent study10 analyzed responses to the Atkins (HPLCD), Zone, Weight Watchers (Weight Watchers International, New York, NY), and Ornish diets in 160 randomly assigned participants. After 1 year, weight loss was −2.1±4.8 kg in 21 of the 40 participants completing the Atkins diet, −3.2±6.0 kg for the 26 of 40 completing the Zone diet, −3.0±4.9 kg for the 26 of 40 completing the Weight Watchers diet, and −3.3±7.3 kg for the 20 of 40 completing the Ornish diet. Each diet significantly reduced the LDL/HDL ratio with no significant effects on BP or glucose after 1 year. The amount of weight loss was associated with self-reported dietary adherence, but not with type of diet. For each diet, decreasing TC/HDL cholesterol, C-reactive protein, and insulin were associated with the amount of weight loss, with no significant differences between diets. Contrary to the studies of short-term duration, these long-term studies suggest that the HPLCDs offer no further advantage in reducing body weight or improving cardiovascular (CV) risk factors compared with other weight-loss methods.


Typical HPLCDs emphasize unlimited consumption of animal products, including all meats, eggs, and cheese, which are significant sources of saturated fat, cholesterol, and total fat in the American diet. As reviewed above, the effects of these diets on short-term weight loss and blood lipid profiles appear promising at the outset. A closer examination of the lipid changes, however, suggests that with longer-term adherence to an HPLCD, the improvement in lipids is limited to serum TG and HDL levels and this improvement is attributable primarily to the subject's concurrent or active weight loss—regardless of the dietary approach used to induce weight loss. Several reviews have indicated that the observed weight loss on the HPLCDs is due to the restriction of energy in the diet, not to the carbohydrate restriction per se.11,12 Additionally, improvements in lipids and other CV risk factors are attributable to the weight loss and not to the dietary composition of the HPLCD.12,13

An overlooked observation is that many studies of HPLCDs have not shown any improvement in LDL levels compared with other weight loss diets and, in fact, have shown a trend toward higher values for LDL compared with other weight loss diets.3–8,10,13 Furthermore, decades of research including intervention and observational studies demonstrate that excessive, sustained intake of fat, particularly saturated fat and cholesterol, leads to long-term risk of heart disease and some cancers.14,15

Advocates of high-protein diets have focused primarily on lipid end points and insulin sensitivity as risk factors for heart disease, but have failed to acknowledge other factors important in the etiology of atherosclerosis. Data are particularly lacking on these and other outcomes in subjects following these diets for longer than 1 year. It is well documented that intake of dietary fat prolongs and exaggerates postprandial lipemia, up-regulates inflammatory and thrombogenic markers, induces transient impairment of brachial artery dilatation,16 and impairs endothelial function.17 In many cases, these derangements can be induced acutely by a single high-fat meal.18 This suggests a potential mechanism by which HPLCDs may be proatherogenic, as their high fat content may have an adverse consequence that outweighs potential beneficial effects on weight and/ or serum lipid profiles. For example, recent evidence from a longer-term study of HPLCDs demonstrated an increase in lipoprotein(a) and fibrinogen levels,9 both potential factors in the development of coronary artery disease.

Investigations of the effect of HPLCDs on inflammatory markers are emerging. Very short-term studies of these diets (6 weeks) showed improvement in several inflammatory parameters, including high-sensitivity tumor necrosis factors, interleukin-6, C-reactive protein, and soluble intercellular adhesion molecules.13 The same parameters were similarly improved, however, on a low-fat diet—suggesting that weight loss itself was the primary determinant of improved inflammatory status, irrespective of the dietary approach. Furthermore, the aforementioned long-term study by Dansinger et al.10 comparing four weight-loss approaches, including HPLCD, also reported that weight loss was the driving force underlying improved inflammatory status in their study.

Studies investigating ketogenic, high-fat and high-protein, and low-carbohydrate diets raise the potential issue of adverse effects on CV risk factors when these diets are followed in the absence of weight loss. Kwiterovich et al.19 and Phinney et al.20 reported dramatic increases in plasma lipids in children and adults on ketogenic diets. Furthermore, platelet aggregation21 and potential for arrhythmias or other cardiac abnormalities have been reported22 in association with ketogenic diets.

Another potential risk factor for heart disease is elevated blood levels of homocysteine.23 High-protein diets provide high levels of amino acids, the substrate for homocysteine. Fleming reported an increase in plasma homocysteine on the HPLCD and a nonsignificant decrease in homocysteine in subjects on low-fat, high-carbohydrate diets.9 HPLCDs are deficient in certain B vitamins such as folate,24 which are essential for lowering homocysteine by metabolizing it to alternate byproducts unassociated with disease risk. Folate is found in dark green leafy vegetables, oranges, and fortified grain products, all of which are limited or eliminated on an HPLCD. Thus, a low-carbohydrate, high-protein diet not only increases the substrate for producing homocysteine, but also decreases the body's ability to metabolize homocysteine due to the low folate content. This provides a possible explanation for increased homocysteine seen in these diets because amino acids— especially those in meats—provide methionine, the substrate for homocysteine. Fruits, vegetables, and whole grains provide the necessary coenzymes for metabolizing homocysteine to cysteine and other nonharmful products.

Elimination of fruits and vegetables from the diet has implications for nutrients that may affect other risk factors for heart disease, including hypertension. The K+/Na+ ratio on these diets is well below recommended levels and would be expected to have adverse effects on BP, as seen in the Dietary Approaches to Stop Hypertension (DASH) study. Several studies of HPLCDs have shown an improvement in BP, as would be predicted given the degree of weight loss. Long-term implications of continued high-Na+, low-K+ intake and long-term effects on BP have not, however, been investigated.

It is not surprising that HPLCDs are low in many nutrients that function as antioxidants, including vitamin C, vitamin E, beta carotene, and selenium.24 The severe limitation on intake of plant-based foods lowers the intake of phytochemicals, many of which have important biologic effects, including antioxidative capacity. The importance of oxidative processes in the etiology of CV disease has been well studied; diets low in antioxidants are associated with increased oxidative markers, although the effects on the actual CV event rate are less clear. In fact, it should not be overlooked that studies of HPLCDs have focused exclusively on CV risk factors, and there is no research on the actual incidence of heart disease or the actual impact on event rate itself. Thus, there is little firm evidence to support the use of these diets to reduce CV disease per se. In contrast, there have been several studies showing a decrease in body weight, heart disease risk factors, and actual event rate in subjects following plant-based, low-fat, high-carbohydrate diets.25–30 Further research is needed to support the use of HPLCDs in cardiology patients for both the treatment of obesity and their potential to lower morbidity and mortality in this population.

In addition to the nutrient concerns indicated above, HPLCDs are also very low in fiber, carry a high nitrogen load, and overall are nutritionally inadequate. The potential implications of prolonged low fiber intake are well documented15: constipation, hemorrhoids, diverticulosis, polyps, and colon cancer. The high incidence of gastrointestinal-related side effects in the HPLCD studies, including constipation and hemorrhoids, reflect negative effects of these diets on gastrointestinal function. Most major health organizations and national guidelines emphasize the inclusion of dietary fiber to promote improved gastrointestinal health, and reduce the risk of CV disease and cancer. These recommendations are based on outcomes of numerous clinical trials.15 Since dietary fiber is provided by whole grain products, legumes, fruits, and vegetables, all of which are restricted on HPLCD, it is not surprising that important health organizations such as the American Heart Association, American Kidney Fund, American Cancer Society, and American Diabetes Association have issued precautionary statements regarding the use of HPLCDs.

The high nitrogen content of HPLCDs has been implicated in potential adverse effects on joint, renal, and bone health.24 High-protein diets have been correlated with increased blood uric acid concentration,2,24 which is associated with a marked acid load and incidence of gout, kidney stone formation, and possibly osteoporosis. Urinary excretion of calcium and acids is associated with high intake of animal protein, but negatively correlated with plant protein intake.25 A study of HPLCDs demonstrated marked renal acid load over 2–4 weeks, which increases the risk of stone formation, decreases estimated calcium balance, and may increase the risk of bone loss.26 Although a few short-term HPLCD studies discussed above have not shown adverse effects on renal and bone health, the evidence is mixed, underscoring the need for additional long-term investigations to fully understand the impact of these diets on additional health parameters. Until more evidence is available on the effects of sustained high nitrogen intake, it is prudent to use caution, particularly in susceptible groups and those at risk for renal disease, including patients with diabetes, kidney stones, gout, and osteoporosis.


The rising rate of obesity in the United States has caused both the lay public and clinicians alike to seek ways to lose weight quickly and effectively. HPLCDs are advocated on the premise that obesity is related to Americans' high consumption of carbohydrate-containing foods and that simply eliminating these foods from the diet will correct the problem. Although the American diet often does contain a high percentage of calories from carbohydrates, it is simplistic to categorize all dietary carbohydrate as the basis of the problem. Elimination of all carbohydrate-containing foods results in a dietary imbalance with a number of potentially negative consequences, including deficiencies of multiple essential food elements, including and beyond those reviewed above. The inclusion of guidelines for supplementation with multiple vitamin and minerals in most HPLCD promotions is testimony to the nutritional inadequacy of the diet. In addition, the long-term benefit of these diets on chronic disease conditions associated with obesity is not clear. The growing rate of obesity in America is most certainly multifactorial and not likely to be the result of any single factor. The search for the optimal diet continues, as investigators experiment with various combinations and sources of protein, fat, and carbohydrate as well as the impact of these diets on healthy and diseased populations.31–36

One potential positive outcome of HPLCD popularity is that it provides an opportunity to educate patients about dietary approaches. The public needs to be taught that the carbohydrate concern of health professionals is related to the low intake of complex carbohydrates (vegetables, fruits, and whole grains), coupled with the excessive intake of simple carbohydrates such as sodas or simple carbohydrate foods that are also high in fat, including donuts, cakes, pastries, and french fries, and other foods high in fat and calories. Shifting the focus to reducing the intake of simple carbohydrates, along with other low-nutrient, calorically dense foods, should become the realistic focus of “low-carbohydrate” diets. The population following the HPLCD has demonstrated their readiness to alter food intake and their motivation for weight loss, which health care professionals can act on to reinforce and encourage a healthy lifestyle for long-term weight loss and health.