August 13th, 2012

Universal Screening for Dyslipidemia in Children: A Debate with Equipoise, but Tarnished by Industry Influence

Note: This post is accompanied by a separate post from Larry Husten, and a counterpoint post from Samuel Gidding.

As a strong proponent of heart and vascular disease prevention and a parent of two teenagers, I have watched closely the debate about the new pediatric lipid guideline recommendation for universal lipid screening at ages 9-11 years. The physicians and researchers on both sides of issue are intelligent, thoughtful, and honest advocates for their positions regarding universal screening, a question that truly has scientific equipoise. In this post, I will briefly describe my position, and hopefully sound a clarion about the influence of the medical industry on this debate. My discussion of industry’s role will emphasize evidence from the social science literature about conflicts of interest, so we can move beyond name-calling, moralizing, and emotional responses to this serious issue.

Universal Screening for Dyslipidemia in Children

I oppose universal lipid screening for children, primarily for the reason that Matt Gilman so eloquently described: because “the harms of screening fall disproportionately on the healthy.” To be clear, the potential harms of universal lipid screening in children, just like its potential benefits, have never been proven. So we are operating in a data vacuum and have scientific equipoise. What are the potential harms? First, increased medical costs, which are born by society, and second, the “medicalization” of lipid values that make children at no short- or even intermediate-term risk of cardiovascular disease (CVD) events, in some way, abnormal. Risks of “medicalization” include stigmatization, labeling, and the practical inconvenience of having every 9-11 year child fast for 12 hours, twice, to get poked with a needle, when the most likely outcome is that they’ll be told that they need to eat a healthy diet and exercise. Of more concern is the astute point raised by Newman et al that girls have higher cholesterol levels and are more likely to get labeled and treated, but paradoxically are at lower CVD risk. In regard to dietary interventions like the CHILD-1 diet, it is not clear to me why healthy eating patterns would not be recommended to all children and families, regardless of their lipid status. In regard to the safety of statins in children, I would argue that the evidence base is slim and likely biased.  What is known about their efficacy in children is limited to their effects on surrogate markers.  Although it is likely that long term use of statins will reduce CVD risk in children with hyperlipidemia, we have to be honest and recognize that changes in lipid levels and carotid intima-media thickness are not the same as reductions in CVD event rates. The chain of evidence is present, but the links are weak. Accordingly, we have scientific equipoise, not the scientific basis for a strong public health recommendation.

One good point for universal screening – previously raised by Marilyn Mann in a comment on – is that after kids leave home, they often drop out of medical care for a time, so it is easier to identify them when young, even if they do not get treated. It is a fair point, but in my opinion, does not justify screening all children, to find the small number of kids with familial hypercholesterolemia (FH) who would not have been identified by targeted screening. However, this is a question that could be addressed, in part, by mathematical modeling. The guideline writers would have strengthened their case if they had presented an analysis of the number of new FH cases identified with universal screening,  the number of cases of “dyslipidemia”  identified, the number of true and false positives and negatives, as well as the costs and effectiveness of the screening program that they proposed. It would be a challenging and imperfect analysis, but one that should have been done before the guideline was approved, rather than being left to future researchers.

For the record, I have declined to have my kids screened, even though we are in a purportedly higher risk group (Ashkenazi Jews).

The Influence of Industry on the Universal Screening Debate

The debate about universal screening is a fair and important one, but the influence of industry interferes with our ability – indeed, our obligation – to honestly debate and discuss this important public health issue. It is amazing that every time the specter of conflict of interest is raised, doctors and researchers get pretty edgy. You can literally see the hairs rise on their backs. And of course, everyone is an expert on this issue and no one is biased. That is unfortunate, because a strong social science literature has taught us that everyone is biased. To illustrate this point, let’s focus on some of the guidelines writers’ responses.

The guideline writers’ contention that “medical treatment recommendations are based on multiple randomized trials of statin therapy in children with FH demonstrating acceptable safety and efficacy over periods up to 2 years” rings hollow. Almost all of those studies were sponsored and conducted by pharmaceutical companies, including several that have been investigated and even sanctioned for improper handling and reporting of data, including safety data. The studies had small sample sizes and were of short duration, and therefore tell us very little about long-term safety of statins in children and young adults. I am sure that the limitations of these studies would be easy fodder at any medicine or pediatrics resident journal club. But the guideline panel uses these studies as high level evidence to support their deeply held belief in statin safety and their desire to prevent CVD. Preventing CVD is laudable, but one must ask where their belief in long-term statin safety came from when the power of the evidence base to detect even common side effects in children and young adults, is limited and only of about 2 years duration. It likely came from repeated encounters with the pharmaceutical industry, which has inundated us with statin safety messages, as well as our own belief in what we do. These are unconscious, intellectual biases that we, as humans, all succumb too. They are even more insidious when an external influence – in this case the pharmaceutical industry – suggests them to us.

Even more dangerous to the academic mission and our societal obligation to have an open and honest debate is the effect of financial conflict of interest. Here the guideline writers reacted emotionally, rather than responding intellectually. In response to the concern about financial conflicts of interest, the guideline writers state: “To contest the integrity of panel members and their deliberations without evidence is unfair and uninformed.” But actually, it is not. It is their response that is uninformed. The guideline writers’ response is moralistic and defensive, rather than academic. I understand how people get defensive when the issue of financial conflict of interest is raised, but no one is impugning anyone in this debate or challenging their integrity for having a potential conflict of interest. All of these individuals are good, honest physicians and researchers, doing their best to come up with helpful public health guidelines to prevent CVD. In that regard, we all are on the same team. But when people have financial relationships, the data demonstrate that they engage in unintentional motivated reasoning – i.e., they become biased. They can’t help it. It is a by-product of being a human being. Humans are living, breathing, bias machines. We all try to manage our biases, but sometimes we can’t. The key point is that bias is unconscious and does not imply malfeasance or wrong-doing. We just can’t help it. And disclosure, although laudable, does not solve the problem of bias because the problem is not secrecy – it is power. Indeed, disclosure may even have the “perverse” effect of increasing bias. In this regard, a major problem with the pediatric lipid guidelines is that they did not follow best practices for managing potential conflicts of interest.

Here’s to an honest, intellectual debate on an important public health topic. Although I oppose universal screening, I am willing to discuss and debate it, ask for more evidence, and recognize its potential advantages and weaknesses. But I suggest we leave the pharmaceutical and device/testing industry out of it, to minimize (not eliminate) the biases and to increase the level of civility in our deliberations.

9 Responses to “Universal Screening for Dyslipidemia in Children: A Debate with Equipoise, but Tarnished by Industry Influence”

  1. Steven Greer, MD says:

    Nice article. Thank you. Cardiologists need to take back their specialty that was hijacked by Big Pharma and the hundreds of billions that went into statins.

  2. Uffe Ravnskov, MD, PhD says:

    James Stein´s comment should be read by every cardiologist and pediatrician in the world. I tried myself to warn against this folly twelve years ago in a letter published in The Lancet on the first day of this millennium (Lancet 2000;355:69), but obviously it has made little impact. Here it is:

    Prevention of atherosclerosis in children

    Sir-Claudio Napoli and colleagues’ (Oct 9, p 1234)1 observation that children from hypercholesterolaemic mothers had more fatty streaks than children from normocholesterolaemic 
mothers is interesting, but their interpretation of the finding does not agree with present 
knowledge, or common sense. If the fatty streaks were early atherosclerotic lesions, why 
were they more common in the fetuses than in the children?

    Fatty streaks are found worldwide in almost all children, equally often in countries where 
atherosclerosis is rare, as in countries where it is frequent.2 The development of raised 
lesions later in life in some individuals must therefore be due to factors other than the mere 
presence of fatty streaks. Even if we assume that fatty streaks are the forerunners of raised 
lesions, there is no evidence for blaming a high maternal cholesterol concentration or athero-
genic genes; the difference in streak frequency may be due to any hereditary or environ-
mental factor associated with high cholesterol concentrations.

    To use Napoli and colleagues’ findings as an argument for cholesterol concentration lowering 
in childhood, as did Berenson and Srinivasan in their commentary,3 seems unfounded. As 
Napoli and colleagues emphasise, cholesterol concentration was normal and similar in both
groups, and the fatty streaks cannot therefore have been caused by high concentrations in the 
child. Even if the fatty streaks had been caused by high concentrations, the predictive value of 
cholesterol screening is low because concentrations in childhood cannot be tracked to adult-
hood with any certainty. In Webber and colleagues’ observational study, half the hypercholesterolaemic children had normal values after 12 years.4

    But let us assume that a screening programme could identify children at high risk only and that 
a lowering of cholesterol would reduce that risk; the question remains of what to do, because 
diet is poor as a cholesterol-lowering treatment, particularly in children. Even if diet were efficient as a cholesterol­lowering treatment, there is no evidence that diet prevents cardiovascular
morbidity or mortality. This effect was shown in a systematic review of eight ecological, 41 
cross- sectional, 25 cohort, six case- control studies, and a meta-analysis of nine controlled 
randomised dietary trials.5 Instead of the prevention of cardiovascular disease, dietary manipu-
lation of healthy children may rather create families of unhappy hypochondriacs, obsessed with 
their blood chemistry and the composition of their diet.
    The only way to lower cholesterol concentrations effectively is by drugs. There is no evidence,
however, that a possible benefit from cholesterol lowering from a young age may balance possible 
side effects from long-term drug use, because luckily, such trials have never been done. I doubt 
that any parents, with all the facts and assumptions, would allow their child to be screened.

    Uffe Ravnskov
    1. Napoli C, Glass CK, Witztum JL,Deutsch R, D’Armiento FP, Palinski W. Influence of maternal
hypercholesterolaemia during pregnancy on progression of early atherosclerotic lesions in child-
hood: fate ofearly lesions in children (FELIC) study. Lancet 1999; 354: 1234-41.
    2. Strong JP, Eggen DA, Oalmann MC, Richards ML, Tracy RE. Pathology and epidemiology 
of atherosclerosis. J Am Diet Assoc 1973; 62: 262-68.

    3. Berenson GS, Srinivasan SR. Prevention of atherosclerosis in childhood. Lancet 1999;

    4. Webber LS, Srinivasan SR, Wattigney WA, Berenson GS. Tracking of serum lipids and lipo-
proteins from childhood to adulthood. The Bogalusa Heart Study. Am J Epidemiol 1991;

    5. Ravnskov U. The questionable role of saturated and polyunsaturated fatty acids in cardiovascular disease. J Clin Epidemiol 1998; 51:443-60.

  3. Antonio Reis, Ph.D says:

    Wise comments those from Stein and Ravnskov! Anyone knows any study on the effects of statins in children? The focus must be global. Statins as the perfect drug for Big Pharma: millions of lifelong clients!

  4. Even if some children would have been screened positive for FH, I can see no benefit in treating lipid values. This assumption is made on the following facts from “The fallacies of the lipid hypothesis – UFFE RAVNSKOV”: In cohorts of people with familial hypercholesterolaemia, LDL or total cholesterol do not predict future coronary heart disease or peripheral atherosclerosis; those with moderately elevated cholesterol run the same risk as those whose cholesterol is 2-3 times higher than the mean value in normal people. Indeed, in one study those with the highest cholesterol had the lowest risk of heart disease. … Among FH patients (both heterozygous and homozygous), there is considerable variation in the rate of progression of atherosclerosis, despite uniformly elevated LDL levels.

  5. Steven Greer, MD says:

    All of these essays for and against screening of children for elevated cholesterol are too lengthy. It is very simple.

    Do we have any studies in children to support the screening with mortality or CV outcomes? No.

    Do adult screenings of LDL and prescribing of statins in low risk factor adults confer mortality or CV risk benefit? Nope.

    Are statins drugs with serious side effects like muscle damage, weakness, liver toxicity, and forgetfulness? Yes.

    Are statins over prescribed? yes, according to Eric Topol and others.

    Should LDL be abandoned as a surrogate endpoint used to guide statin therapy? Yes, according to Krumholz et al.

    Are the authors of the pediatric screening guidelines receiving significant cash from the statin makers? Yes.

    Follow the money, folks. Follow the money.

  6. Steven Greer, MD says:

    Harlan Krumholz explains in our video interview why LDL should be abandoned as the surrogate market to guide statin therapy and as the endpoint in trials

  7. Dan Hackam, MD PhD says:

    Lipid screening in childhood is warranted, since simple lifestyle changes can dramatically alter lifelong lipoprotein abnormalities and reduce risk in a highly cost-effective manner. A lipid panel could be a wake-up call to a parent to start a child on a vegetarian diet (LDL problem) or a diet restricted in simple and refined carbohydrates (TRIG/HDL problem).

  8. Neville Wilson, M.B., Ch.B., D.Fam. Med., M.Sc., D.Obs (COG) SA., B.A. says:

    The obsession with proposals to practice widespread lipid screening in childhood is driven by the presumption that abnormal lipids are the major risk factor for CHD in adulthood, a premise that is not validated by scientific evidence.

    The Simon Broome Scientific Steering Committee at the Dept. of Public Health in Oxford, UK,followed 526 FH screenees considered to be at high risk for CHD, and reported their mortality outcomes within a 5 year period, at different ages (1).
    For those under 40 years 6 of 214 died within 4 years (3%). Within the 40-59 year old group 8 of 237 died within a 5 year period (3.3%)and for those in the 60-74 year old group 1 of 75 died (1.3%).

    Since many persons with FH live long and uneventful lives, had some of these been included amongst the selected screenees, the mortality figures would have been much lower.

    After following 3000 persons with FH for many years the same group reported a lower mortalty rate than expected, with fewer deaths from cancer than in a comparable group without FH. (2)

    Does the the focus on hyperlipidaemia not ignore the greater relevance of inborn errors of coagulation, which predominate in children with FH, and which presents the risks for CHD rather than their lipid levels ? (3)

    Since raised serum cholesterol levels are not a predictor of CVD mortality, and is unlikely to be modified by low cholesterol dietary intervention, the “wake up call” to dietary changes should not be prompted by blood screening results, but by education regarding the health risks of dietary sugar and refined carbohydrates which are a common replacement in childhood dietary habits for healthy nutrient dense meals, which included saturated fats !

  9. Neville Wilson, M.B., Ch.B., D.Fam. Med., M.Sc., D.Obs (COG) SA., B.A. says:

    1. BMJ 303, 893-896, 1991
    2. Atherosclerosis 179, 293-7, 2005
    3. B Heart Journal 1985, 53:265-8

    Even Brown and Goldstein, despite their proposals that elevated LDL was a causative facor in atherosclerosis, were forced to acknowledge the “considerable variation in the rate of progression of atherosclerosis, despite uniformly elevated LDL levels”(4)
    4. Ann Rev Biochem 52, 223-61, 1983.