2.3 Clinical audits of safety of VLCD

2.3.1 Specific audits

Nevertheless from time to time it is suggested in official reports that "there has been a lack of vigilance in gathering such information". There is no evidence whatsoever to support this contention. Indeed quite apart from vigilance by the various companies monitoring adverse reactions reported in the medical and lay media, there have been several extensive audits undertaken during use in the community under health professional care. These include:

1.      An audit of 500 obese women sequentially joining a modular (12 week) obesity management programme (formulation: protein 43g; carbohydrate 38/45g) in the United Kingdom, of whom 85% completed 1 month and 50% completed 3 months or more. No serious untoward events have been encountered during the ten years that the programme has been running-

2.      Also in the United Kingdom, a GP-based programme has been active since 1990. The data on the first four years has been mislaid but it did not contain any significant adverse reactions. Since 1993, some 31,000 patients have been treated and monitored by their own practitioners following this programme (formulation for women: protein 43 g; carbohydrate 38/45g: formulation for men: protein 57g; carbohydrate 33/60g). The average duration of treatment was 16 weeks and the vast majority have been followed for at least one year post diet. Many of the practices have submitted audit results from these clinics to their respective health authorities. Twenty five such audits were assembled in a paper presented at the Medical Research Society in the UK and at the European Obesity meeting in Barcelona (Beeson 1994). Full medical records show that there have been no serious adverse effects over the whole 10 years.

While these audits cover the generality of the adverse effects during the use of VLCD there is a possibility that they only represent studies in which the starting BMI was well above the 30 level. Moreover it might represent those who did not reach the lower level, which it has been suggested is particularly susceptible to adverse effects.

As a result it was decided to examine those papers in more detail in which it had been shown that there was a significant component of subjects in which the starting BMI was below 30.

2.3.2 Investigations of reputed deaths with nutrient complete VLCD

·        Two papers by Wadden (JAMA: January & June 1990) have been quoted frequently as information relating to safety aspects of the use of VLCD, particularly at BMI levels between 25 and 30. The following comments are relevant:

-   In the USA in the early 1980s a rumour, based upon an FDA Talk Paper, circulated that some six deaths might have occurred using the Cambridge Plan International Diet.  Emil Corwin for the U.S. Food and Drug Administration on February 21^st 1996 issued the following statement  “no deaths have ever been attributed to the use of the Cambridge Plan International Diet”… “during the last five years, as many as 7 million persons have used Cambridge’s programs and products” (Newsday, February 22 1986, page 2)

·        In 1989, Connolly, a general practitioner in Ireland reported a sudden death which he alleged was the result of the use of a VLCD.  The heart muscle was examined by a British Home Office pathologist who reported that there was no evidence of the pathology associated with “liquid protein diets” and that there was no evidence that this was a death that should be attributed to the use of VLCD

·        In 1993 Muller & Grossklaus reported “Autopsy studies performed on 16 persons whose death was considered to be related to the use of VLEDs showed that the weight of heart decreased in proportion to the decrease in body weight” This report was taken to represent  recent deaths in Germany.  In fact the reference he gives is to Isner et al (1979) and this refers to the “liquid protein diets”.

·        In various places in the Wadden articles there is a statement about reduced safety at lower BMI levels. The references are always the same – i.e. the work of Forbes and the repetition of the Forbes views by others. I can not find any other evidence for the statements about safety. The accuracy of the Forbes work is examined elsewhere in this discussion paper (Section III)

-         In this respect, we would draw attention to the recent paper by Professor Sir Colin Berry entitled “Bellmanism: the distortion of reason” (Berry, 2000). The views expressed here are very apposite and particularly on the “imperfect understanding and use of science, in particular in the evaluation of data. Studies which have been discredited are cited repeatedly… in relation to the precautionary principle.”

2.3.3 Audits of subjects with BMI under 30

In some of the papers it is possible to establish what proportion of the data represented experience  at a BMI under 30.

In order to make the necessary calculations it would be necessary to have adequate information to determine that not only is there good compliance on average within the cohort, but that the compliance is equally good across the weight range.

It is abundantly clear from the text in many of the clinical papers that a proportion of those studied have lost such substantial amounts (or proportions of their starting weight) that they have not only moved into the BMI <30 range (if they were not there at the start), but have achieved the normal BMI range. This is obvious if it is appreciated that the compliant VLCD dieter, loses on average between 1 and 2kg/week. For a woman of average current height, one BMI unit is equivalent to some 2.64kg. The published clinical papers indicate that on average over 80% of the dieters are women. For the men one BMI unit weight equivalent is more but the rate of weight loss is greater.

Thus with the average weight loss per week with the majority of the current commercial VLCD, there is a BMI change at the rate of between about one BMI unit each 2 weeks. For those who start at BMI 30, a BMI about 25 (i.e. down to the normal weight level) will be achieved in about 8-10 weeks, at least three-quarters of which will be below BMI 30. At start BMI 35 the average time to normality will be about 18 weeks, with about half that period below BMI 30.

There are 3 papers which provide enough information tobe able to calculate, with reasonable accuracy, the proportion of the VLCD dieting period which represents a BMI level below 30.  The basis of the calculation is such that even for these three studies it is clear that it does not represent hard data. Nevertheless much of the data cross checks and is sufficiently representative to be useful, bearing in mind that there are no serious side effects in the whole series. The information covers several previously published smaller cohorts, but the three papers concerned are those of  Kirschner et al (1988); Kanders et al (1989) and Bode (1999).

These cover observations using VLCD at 300, 420 and 770kcal per day, though it is not possible to establish with certainty the exact proportion at each intake because the formulation changed over time. One merit of all these studies is that they represent data for observations of at least 12 weeks VLCD dieting a long enough period for any major adverse effect to occur.

The information from these studies is summarised in Table 2.2.

Table 2.2 Data on the three papers from which information on the safety of VLCD at lower BMI levels can be gauged.

* excluding those not completing dieting phase

The important points to note are

·        For those with a BMI above 35, dieting for at least 12 weeks is necessary before the BMI falls below 30

·        In a major proportion of the USA literature, with a very high initial BMI level and a short period of VLCD use there is relatively little data at BMI levels <30

·        On the other hand, in Europe, VLCD have been used at lower starting BMI levels and, in consequence there is greater evidence of safety in the range 25 to 30.

On the basis of this evidence a reasonable proportion of the VLCD dieting for these studies was at BMI < 30. Since these represented experience in a substantial number of dieters for a period of at least 12 weeks, there is substantial evidence of use of  VLCD in the BMI range (i.e. 25 to 30) for which further information was required

Taking each of these papers (Kirschner et al 1988; Kanders et al 1989; and Bode 1999)in turn:

·        Kirschner summarises his experience thus: “Complications of obesity i.e. hypertension, type II diabetes mellitus and hyperlipidemias were remarkably improved after weight loss. Complications of the VLCD including cardiac abnormalities, were minimal”

The relevant portion of the paper reads:

“Complications. A list of side-effects and complications is presented in Table 4.

Table 4. Complications observed in patients on supplemented fast.

                                                                                                                Cases

1) Acute gout                                                                                            8

2) Foot drop (temporary)                                                                         2

3) Acute psychosis                                                                                   4

4) Diabetic ketoscidosis                                                                            2

5) Late hair loss                                                                                      10%

6) Cardiac arrhythmias

       Supraventricular tachycardia (hypoglycemia)                                  1

       Multifocal PVCs                                                                                2

The most common problems noted were early postural lightheadedness and tiredness, The most common late complaint was that of mild transient hair loss, occurring in approximately 10 percent of the population. Other complications included eight cases of acute gout, two cases of foot drop, (thought to be due to sciatic nerve compression from leg crossing during or after weight loss), and four cases of acute psychosis occurring in women and generally thought to be a form of sexual panic.

In view of the great concern regarding cardiac arrhythmias in patients on VLCDs we carefully screened for such occurrences. Over the 8-year period, we documented only one patient who developed a supraventricular tachycardia clearly related to hypoglycemia ans corrected with intravenous glucose supplementation. Two patients required hospitalisation for the development of palpitations associated with multifocal PVCs. To date, we have had no unexplained deaths in contrast to the experience with liquid protein diets.

Non-complications. In view of the genuine concern as to the overall safety of VLCDs there were several important non-complications to record, including: (1) coronary pypass surgery without complications in eight patients; (2) major breast surgery without complications in 12 patients; (3) pregnancies occurring while patients were on VLCD (without subsequent complications) in six women in whom the diet was subsequently discontinued.”

To put this paper in context it should be appreciated that this represented something in excess of 40,000 weeks dieting experience in 4026 patients at 420kcal/day with some 10% in the BMI range <30. The quoted paragraphs followed several that detailed the beneficial effects of VLCD on medical problems in this same patient cohort.

·        The equivalent paper by Kanders et al (1989) gives a substantial amount of information about beneficial effects. The phrasing of the report suggests that side effects were not a problem but there is no specific reference to side effects throughout the whole paper. It is abundantly clear from the nature of the paper, that, had side effects been a problem, they would have been recorded.

·        The same situation is true of the report by Bode (1999). He describes reasons for the subjects stopping the use of the VLCD but these are primarily social. There are no significant medical problems.

·        We have also obtained access to the detailed information relating to Lipotrim^Ò a monitored VLCD programme (Kreitzman & Beeson, 1996). This provides an indication of the VLCD dieting pattern in the practical commercial area and moreover gives some indication of the post diet period, though not over a prolonged period.

This study by Kreitzman & Beeson (1996) was undertaken with trained independent medical observers. Again, no VLCD related adversa effects were reported.

An audit of 746 case records was made. This covers an overall period of total food replacement (TFR) by VLCD of 13,446 weeks; a total weight loss in the group of 15,657Kg  representing an average weight loss for each dieter of 20.99kg at an average of 1.2kg per week.

Of the 746 dieters 140 started below BMI 30. The average BMI before dieting was 28.54 and that at the end of the TFR period 24.06 (12.16kg) achieved on average in 78.84 days (11.2 weeks) at an average of 1.08kg per week. All these spent their whole TFR at BMI <30 representing 1,568 weeks TRF. As a matter of interest the average follow up for this group after re-feeding was 35 weeks and at that stage 79% had maintained all their lost weight with the average regain under 0.9BMI unit.

The remaining 606 dieters started at average BMI 34.6. They lost 23.03kg weight on average at an average of 1.18kg/week (total weight loss in this group 13,955kg). Their average number of days on TFR was 137.10 days (19.6 weeks), achieving an average BMI at the end of the TFR period of 26.08. Their total TFR period was 11,878 weeks of which 5,465 was under BMI 30.

This means that in this whole group, out of 13,446 weeks VLCD use, no less than 7,033 weeks (something over 50%) represented experience at BMI <30. In no dieter at any stage was there any serious or worrying side effect.

2.3.4 Audits on the prime data

It was also considered important to examine the prime data in more detail to determine whether the information provided covered the broad range of the energy and carbohydrate levels (rather than just high carbohydrate levels providing energy levels close to 800kcals.

The results of this further analysis are shown in Tables 2.3 a and b

Table 2.3 A further split of the original data into Table  a) <400kcal, 400-600kcal, 600-800kcal and b) carbohydrate levels 30-40g, 40-45g, 45-50g and >50g..

It should be noted that these figures differ slightly from those in Table 1 – some categories were not represented and for some a more detailed analysis was not feasible.

Nevertheless, the further analysis demonstrates not only that a major proportion of the data was derived from VLCD with an energy value less than 600kcals, but that over 90% of the subjects were receiving a daily carbohydrate level below 50g.

2.4 Special tests for safety

2.4.1.Laboratory studies

Several of the published clinical studies have investigated laboratory parameters of organ change (eg cardiac, haemopoietic, hepatic, renal) on a regular basis.  Many more have undertaken spot laboratory checks.  These have demonstrated evidence of improved health due to the weight loss but no adverse effects from the use of these diets.

2.4.2. Electrocardiographic evidence of cardiac integrity

It was suggested in some of the earlier reports that even if there appeared to be no overall excessive loss of protein during the use of VLCD, there might be specific extra and dangerous protein loss from the heart or other vital organs. However, as mentioned previously, clinical observations and laboratory investigations over a period in excess of 20 years, on more than 50,000 subjects using a very wide range of products of different composition with an energy content under 800kcal, have not demonstrated that any such problem exists. Nor is there any theoretical reason why damage to specific organs should occur.

Specifically there is extensive electrocardiographic evidence of cardiac integrity.

Obesity itself can produce cardiac changes (Hinkle et al 1969; Eisenstein et al, 1982; Frank et al, 1986; Carella et al 1996; Quaade et al 1996),).  Indeed it has been calculated that an obese cohort subjected to surgery has 40 times the risk of sudden death compared to a normal weight cohort (Drenick & Fisler, 1988).  Because the use of the seriously nutritionally deficient liquid protein diets in the early 1970s (which bear no relationship to the modern VLCD) led to fatal cardiomyopathies (FDA 1979; Sours et al 1981; Isner Sours et al 1981; Van Itallie et al 1984, random electrocardiographic recordings and regular Holter monitoring have been an important feature of studies on the modern very low energy diets in order to confirm that they do not have similar adverse effects. 

The 12 formal studies (summarised in Table 2.4) all confirm that well formulated very low calorie

Table 2.4 Studies which examined the electrocardiograph by Holter monitoring at intervals during the use of VLCDs

diets produce no evidence of adverse effects on cardiac function. Additionally Linet (1983) ans Singer (1981) found no abnormality in food based diets (with added vitamins and minerals) at energy levels below 600Kcals/day.

Indeed calculations demonstrate that the incidence of sudden death in those being treated with VLCD well below the level which would be predicted from their age and weight

Examination of the "liquid protein diet" data on the 17 deaths that were studied (FDA 1979; Sours et al 1981; Van Itallie et aI 1984) show that the shortest period to any cardiac abnormality was several weeks. As Meuller and GrosskIaus wrote ( 1993) :

"Electrocardiographic changes do not become evident before 4-6 weeks of hypocaloric dieting" and "In subjects on VLEDs electrocardiograms are usually negative up to 4-6 weeks" and "a well supplemented VLED has fewer adverse effects on the electrical activity ofthe heart" (In fact examination of the original data suggests that the figure is more like 8 weeks than 4-6 weeks)

Hence the 12 ECG studies quoted here are not as valuable as they would appear at first sight because most cover the period before electrocardiographic changes wouId be expected. It must however be stressed that since 1978 there is no confirmed evidence of electrocardiographic changes at any stage.

This raises the whole question of the suggestion made by some authorities that regular routine electrocardiographic examination should be undertaken (even as frequently as weekly) on all those using VLCD. This suggestion has no validity-.-.

·      The data indicates that the only reliable method for determining changes is by Holter continuous monitoring which is clearly impracticable. Thus in a study by Lantigua et al (1980) 3 out of the 6 subjects deliberately subjected to a hydrolysed collagen diet with added tryptophan showed cardiac conduction abnormalities, of which none were detected by 12-1ead ECG, only by continuous Holter monitoring.

·        In six of the 17 who died in the liquid protein tragedy, normal ECGs had been found while they were taking the liquid protein diet.

·        There is no agreement about the predictive relevance of any electrocardiographic features that might be found.

One recent widely publicised study (Greenway et al 1994) purports to demonstrate changes (particularly in the Q-T segment) related to the energy rather than the protein content but the changes reported are those seen commonly in untreated obese people and have no significance relative to the safety of VLCD (Quaade, 1996). Moreover Hinkle et al ( 1969) pointed out that " disturbance of heart rhythm and conduction occur in apparently healthy people with high frequency" while in the study by Carella et al ( 1996) between 41% and 53% (depending upon the method of determination) of obese people showed a prolongation of the Q-T interval before any dietary intervention.

It has also been postulated that since the protein turnover of myocardial proteins differs from that in skeletal muscle, this may produce a failure of protein conserving mechanisms in the heart and greater risks to that organ. Mueller & Grossklaus (1993) have pointed out that there is no failure of protein-conserving mechanisms in the heart for " the weight of the heart decreased in proportion to the decrease in body weight" and "Myocardial wasting...is not necessarily associated with reduced functional capacity" and "A 24-week of semi-starvation resulted in a weight loss of 15-20kg...but not circulatory insufficiency or heart failure" and "These data suggest a normal myocardial fiber performance in response to semi-starvation". These views expressed by Mueller & Grossklaus are in accord with the earlier observations of Smith (1928) and Amad et al (1965) that the weight loss in the heart during starvation is proportional to that lost in the rest of the body .

The relationship between heart weight and body weight also applied to those who died in the 1977 liquid protein diet tragedy (Sours et al 1981, Isner, Sours et al 1981 and Van Itallie et al 1984) while none of these 17 dieters showed untoward evidence of heart muscle insufficiency.