You don't need to be signed in to read BMJ Blogs, but you can register here to receive updates about other BMJ products and services via our site.

Archive for January, 2016

Primary Care Corner with Geoffrey Modest MD: Zika Virus

29 Jan, 16 | by EBM

By Dr. Geoffrey Modest

The Zika virus has made the headlines of late. Some details:

  • Zika virus is a mosquito-born flavivirus, mostly transmitted by Aedes aegypti mosquitoes (which also seem proficient in transmitting dengue, chikungunya and yellow fever viruses)
  • Zika infections have been documented through intrauterine as well as intrapartum transmission from a viremic mother. RNA from the virus is also detected in breast milk though transmission has not been documented by breastfeeding
  • 80% of people are asymptomatic with the virusAedes_Albopictus
  • When symptomatic, there are usually only mild symptoms, with acute onset of fever, maculopapular rash, arthralgia, nonpurulent conjunctivitis. Lasting several days to 1 week. Fatalities are rare. But Guillan-Barre has been reported [note: there is a background incidence of Guillan-Barre. Not clear that the association with the Zika virus is causal, though a few cases have been found in Zika-infected individuals]
  • In Brazil outbreak, Zika RNA has been identified in brain tissue, placenta and amniotic fluid
  • Though there has been a dramatic increase in the numbers of infants with microcephaly or intracracial calcifications, it is unclear how many are associated with Zika
  • Testing: there are PCR tests for the viral RNA, and both IgM ELISA and PRNT (plaque reduction neutralization test) testing for antibodies.There can be cross-reacting antibodies causing false positives, but the PRNT is more specific to Zika. Contact the State Dept of Public Health for info
  • Since unclear which test is most reliable, CDC recommends both PCR and one of the antibody tests. All testing is done by the CDC or state labs.
  • PCR testing should be within 2 days of birth. Also CSF, if obtained for other studies, and maternal serum

Zika virus has been found largely in Africa and Southeast Asia in the past. In May 2015, the WHO reported lots in the Western Hemisphere, including travelers to the US (though no active transmission found so far). But almost all of Central and South America has active cases, including Puerto Rico, Mexico, and pretty much everywhere else except Costa Rica, Argentina, Chile, Uruguay. Also active transmission reported in Cape Verde (see for virus/travel updates)

See for the CDC guidelines on evaluation and testing of infants with possible congenital Zika infection. In brief:

  • Follow closely new mothers who were potentially exposed to Zika during pregnancy based on travel or residence in areas with Zika transmission
  • Review fetal ultrasounds and maternal testing for Zika
  • Test infants for the virus, if
    • Infant with microcephaly or intracranial calcifications
    • Infants born to mothers with positive or inconclusive test results for Zika virus
    • Also, Zika virus is a nationally notifiable condition
  • Infants with positive or inconclusive Zika tests: they should have ophthalmologic exam including retinal exam, within the first month of life, given reports of abnormal eye findings in those with possible congenital Zika. And repeat hearing screen at 6 months
  • Still look for other possible etiologies of microcephaly or intracranial calcifications if these findings present, and treat appropriately (including consultation with dysmorphologist, a new term to me…., as well as routine testing for syphilis, toxo, rubella, cmv, lymphocytic choriomeningitis virus infection, hsv) [even with appropriate training and interest, I’m not sure I would like the word “dysmorphologist” to be attached to my name]
  • In infants without any findings at birth but born to a Zika-positive mother, test the infant for Zika infection, and if possible infection, do routine exam with comprehensive neuro exam, check for hepatosplenomegaly and rashes, cranial ultrasound (unless normal in 3rd trimester check).
  • Only mothers who report symptoms suggestive of Zika within 2 weeks in an area with ongoing Zika virus transmission should get Zika testing. Then if positive or inconclusive, test the infant.
  • Management of Zika: nothing specific, no vaccines. Mothers should be encouraged to breastfeed, with apparent benefits outweighing potential risks
  • Prevention: just avoid mosquitoes

The CDC recommends that “all pregnant women consider postponing travel to areas where Zika virus transmission is ongoing”, and if she goes there, avoid mosquitoes (remember that this mosquito, unlike the dawn/dusk types that transmit malaria, is more around in the daytime. So, use lots of anti-mosquito protection — though I’m not sure that is so great for the infant, the CDC recommends DEET, picaridin and IR3535 as more likely to be safe)

The NY Times reported on possible risk of Zika sexual transmission. See . Basically, there is a “theoretical risk” of sexual transmission per the CDC.  A Tahitian man was exposed to the Zika virus in 2013, and a high level of virus was found in his semen and in his urine. Another was a malaria researcher in 2008, who was collecting mosquitoes for a malaria study in Senegal. He developed rash, fatigue, headaches, bloodshot eyes, and genital pain/likely hematospermia late in the illness. Tests for malaria, dengue and yellow fever were negative. After his return to the US, his wife who had remained in the US, developed several of these symptoms. Frozen serum from both the researcher and wife were subsequently positive for Zika virus, and no other family members were positive, suggesting sexual transmission.​

So, the clear concern here is yet another emerging infectious disease with potential dire consequences to those infected, unclear if there will be new and threatening modes of transmission, and lack of current vaccine or treatment. But overshadowing much of this is the effects of climate change. It turns out that even small degrees of warming lead to major changes in the territory for mosquitoes. New cases of dengue, for example, have migrated north to the US (sporadic new cases reported in Hawaii and Florida), though the range of the vector mosquito has spread considerably into the southern US

Primary Care Corner with Geoffrey Modest MD: Guidelines on Length of Bisphosphonate Therapy

28 Jan, 16 | by EBM

By Dr. Geoffrey Modest

A not infrequent question in primary care is when to stop bisphosphonates (BPs) for women with osteoporosis. There was a recent guideline on this subject from a task force of the American Society for Bone and Mineral Research (see DOI: 10.1002/jbmr.2708, as well as doi: 10.1056/nejmp1202619 for the 2012 response by the FDA).

Details of their recommendations:

  • Background:
    • Osteoporotic fractures are really common (they quote the stark but I suspect inaccurate frequency of 1 fracture every 3 seconds around the world…), with substantial morbidity and mortality
    • ​1 in 3 women and 1 in 5 men experience a fragility fracture after age 50
    • ​3-4 year studies confirm that BPs work in decreasing the risk of vertebral (40-70%), hip (20-50%) and nonvertebral (15-39%) fractures
    • Lots of people take them: from 2005-9, 150 million scripts were dispensed in the US
    • ​But, not so shockingly, these drugs are not benign, with reports of jaw osteonecrosis (esp in those on very high doses for cancer therapy, though unclear that increased duration of BPs lead to increased risk), and atypical femoral fractures often after minimal or no known trauma (still a pretty uncommon event: increasing from 3.2-50/100K person-years in the general population to double that with median duration of 7 years of BP use, with likely relationship with increased length of BP use). In 2010, the FDA added these atypical femoral fractures to the “warnings and precautions” of BPs.
  • The data (from 2 studies with at least 5 years of treatment):
    • FLEX study (see JAMA 2006;296(24):2927)
      • 1099 postmenopausal women (mean age 73, 96% white, mean T-score was -1.9 for hip and -2.2 for femoral neck who had taken alendronate for 4-5 years were randomized to continue alendronate at 5mg/d, 10mg/d, or placebo. 60% had had a clinical fracture after age 45 and 1/3 had had a vertebral fracture. But those with hip bone mineral density (BMD) T-score worse than -3.5 or those with lower BMD than at baseline were excluded.
      • After 5 more years of follow-up, those who took either 5 or 10mg of alendronate had significantly less bone loss than those on placebo (e.g., femoral neck BMD improved 0.46% vs decreased 1.48% on placebo). But non-spine fracture risk was similar and there was not enough statistical power to assess nonvertebral fractures. And, another post hoc analysis found that there were increasing clinical fractures as both the femoral neck and total hip T-score were lower (going from <10% to 30% from the lowest to highest tertiles) in those who stopped the alendronate.
      • A subgroup analysis found that there was a reduction of nonspine fractures in women who did not have a baseline vertebral fracture and had a femoral neck T-score worse than -2.5
      • As an interesting aside in the FLEX trial, there was no evidence that monitoring bone-turnover markers (e.g. serum procollagen type 1 N-terminal peptide, C-terminal cross-linking telopeptide, or bone-specific alkaline phosphatase) predicted bone loss over the 5-year treatment-free period (i.e., there seems to be no basis for monitoring these markers in order to risk stratify women for future fractures).
    • HORIZON study
      • ​1233 postmenopausal women (mean age 75.5, 95% from Western populations, 5% Asian, mean femoral neck T=-2.6. 60% had at least one prevalent vertebral fracture) who had received 3 annual IV infusions of zoledronic acid, then either continued that for 3 more years or got placebo
      • ​Those on placebo had more femoral neck bone loss (-0.80 vs +0.24). those on continued zoledronic acid had fewer morphometic spine fractures (i.e., looking at vertebral dimensions), but no difference in nonspine or hip fractures (again, low statistical power for these outcomes)
  • Some of their other points:
    • If one measures the suppression of markers of bone turnover (as a surrogate for BP efficacy), it seems that suppression persists for 2-3 years after stopping alendronate, and at least 3 years for zolendronic acid (though for ibandronate and risedronate, it lasts only 1-2 years)
    • In lower risk women (e.g. hip T-score better than -2.5), it may be reasonable to stop alendronate after 5 years, but monitor the BMD after an additional 2-3 years, and perhaps sooner after risedronate, since there tends to be a quicker rebound of bone-turnover markers (i.e., though these markers did not help in the FLEX study after 5 years of alendronate, they are predictive in those never on BPs, and without any data on long-term risedronate and stopping the med, I agree that it would be prudent to reassess them sooner)
    • ​Stopping BP treatment: minimal data, but suggestive that risk of atypical femoral fractures seems to decrease one year after stopping BPs [and, given the long-terminal half-life of BPs, the American Dental and oral/maxillofacial surgeon societies do NOT recommend discontinuing BPs for dental procedures, contrary to what I see in clinical practice in Boston. Though it is reasonable to delay initiating BPs if there is impending dental surgery]
    • There are small studies/reports finding that by using teriparatide therapy, there is healing of the atypical femoral fractures and jaw osteonecrosis

Their recommendations:

  • After 5 years of oral BPs (3 years with IV), assess whether there were hip, spine, or multiple other osteoporotic fractures. If so, consider continuing therapy and reassess every 2-3 years
  • If no intervening fractures, check the hip BMD. If better than T-score of -2.5, consider drug holiday and reassess in 2-3 years
  • If hip T-score worse than -2.5 or patient considered high fracture risk (older than 70-75, other strong fracture risk, or high FRAX score), consider continuing BPs for up to 10 years, but reassess every 2-3 years

So, I basically support these guidelines. My reading of the literature is pretty much in agreement with this (despite the fact that of 15 members of the committee involved in these recommendations, 11 had drug company ties), and rechecking the BMD after 5 years before considering stopping meds. A few points:

  • Since the cutpoint for enrollment in the FLEX study was a T score of better than -3.5 and no deterioration of BMD after being on meds for 5, I have used a T-score of roughly -3 or so as the cutpoint of when I stopped the alendronate, or if the BMD deteriorated. This current recommendation is more stringent, with a cutpoint of -2.5, which I think is related to the post hoc analyses showing that the worse the BMD, the more likely a fracture in those off the alendronate after 5 years.
  • One thing to keep in mind in dealing with osteoporosis is that BMD is a quantitative assessment of the amount of mineralization of the bones and reveals nothing about the internal, qualitative structure of the bone, which clearly is very important in terms of the structural integrity of the bone. So, people can have fragility fractures even with normal BMD, and there are no data I have seen suggesting that a woman with a fracture should stop the BPs at five years, even with a non-osteoporotic BMD.

Other issues:

  • What about women with really bad BMD (worse than -3.5) and excluded from the studies? It seems reasonable to me to check the BMD after a couple of years (to make sure their BMD is not deteriorating and that they might benefit from more aggressive therapy, e.g. with denosumab, etc.) if BMD is okay, could check again after 5+ years, though my guess is that pretty much all of these women will need prolonged therapy (and, they should be worked-up for secondary causes of osteoporosis…).
  • What to do if patient at high fracture risk and on BPs for 10 yrs? These guidelines stop at 10 years because there are NO data for longer treatment. Given the lack of incremental adverse events when going from 5 to 10 years in the FLEX study, the benefits of continuing BP therapy may well outweigh the likely (though unknown) risks.
  • What if they are on meds that increase fracture risk (e.g. steroids, aromatase inhibitors)? Studies suggest that the increased fracture risk for steroids is partially independent of the steroid effect on BMD. Again, no data, but seems to me that prolonged BP therapy may be reasonable. On stopping steroids, there are data that fracture risk is reduced, but unclear that it returns to baseline risk.
  • What if the patients are at high risk because of prior fractures? These patients remain at high risk of future fractures. Again, no long-term data, but prolonged therapy (i.e., >10 years) may well be appropriate.
  • What is the appropriate dose of BPs after 5 years? One intriguing possibility (not discussed anywhere I have seen) is reverting to a lower dose of alendronate. The FLEX study found that the equivalent of 35mg/week of alendronate had the same results 5 years later than the 70mg/week dose. What about using the lower dose after 5 years?
  • What is the generalizability of this recommendation to non-white women?  For example, Black Americans have significant differences in vitamin D binding protein (see N Engl J Med 2013;369:1991).
  • What about men? Studies are mostly 2-3 years in length. And fractures have not been the primary endpoints. (There is no evidence that there is a reduction of nonvertebral or hip fractures in men). And no data on long-term safety of these meds in men (and no data finding osteonecrosis or atypical femoral fractures). Despite the lack of data, osteoporosis and especially hip fractures are about as common in elderly men as women, so my sense is that it is reasonable to treat men as women for osteoporosis at this point.

So, my bottom line is that I would probably err to over-treating most women at significant risk of fracture and will probably adopt these recommendations with the T -2.5 as the cutpoint. My primary reason is that the 10 year FLEX study did not find any increases in adverse effects of continued alendronate (and there are even some observational data suggesting lower incidence of breast, colon, and gastric cancers, and even cardiovascular disease and mortality); therefore, it seems to me that the risk of fracture, even a vertebral fracture (1/3 of which can cause chronic pain and disability), and likely (though not proven) decreased likelihood of hip fractures in the more distant future (i.e., BMD is higher in the hip with continued alendronate) seem to outweigh the risks from the evidence so far. And, for people at continued high fracture risk at the 10 year mark of BP therapy, I would strongly consider continuing that therapy.

Primary Care Corner with Geoffrey Modest MD: Phobias and Propranolol

28 Jan, 16 | by EBM

By Dr. Geoffrey Modest

There was a recent op-ed in the NY Times by the psychiatrist Richard Friedman on phobias and medical therapy (see ).

His points:

  • 29% of Americans have some anxiety at some point in their lives
  • He cites a pretty remarkable study on using propranolol to block this anxiety, perhaps from blocking norepinephrine action (see article and review below)
  • He also raises the interesting contrary concern: stimulants (e.g. ritalin) can cause release of norepinephrine and could theoretically enhance fear/anxiety, or even PTSD in those exposed to trauma. He notes that soldiers exposed to stimulants did in fact have more

A small study was done looking at the effects of the b-blocker propranolol in inhibiting memory reconsolidation and decreasing the phobia (see Biological Psychiatry 2015; 78:880). The stimulus for the study was that fear memories are now considered not to be indelible memories, but ones which on reexposure to the object of fear, leads to neural protein synthesis and reconsolidation of that memory. Animal studies suggest that b-blockers can disrupt this process of reconsolidation and decrease anxiety. Based on this model, a small study was done of humans with arachnophobia. Wolf_spider_on_white


  • 15 people with arachnophobia (fear of spiders) received a single dose of propranolol, 40mg, vs 15 who received placebo, after a 2-minute exposure to a tarantula. An additional group received propranolol without the arachnid exposure
  • After the above treatment (propranolol or placebo), the patients stood in front of an open-caged tarantula at a distance of 60cm, then were asked to approach and attempt to touch the spider with their bare fingertips. Patients were tested at 16 days post-exposure, 3 months, and again at 1 year


  • ​The effect of the propranolol was striking and longstanding: patients were able to handle the tarantula after propranolol but not placebo, from the 16-day test to that at 1 year, without any falloff over time. In fact, all of the participants in the propranolol group were able to touch the tarantula 16 days later, 3 months later and 1 year later. In the other groups (both those on placebo and those on propranolol but not previously exposed to the tarantula), patients “barely touched the container” and demonstrated fears on approaching the container throughout the follow-up period. So, it was not just giving propranolol alone: taking propranolol without the tarantula exposure had no protective effect
  • In the group exposed to the tarantula, there was no effect of the propranolol in the patients’ self-declared fear of spiders at the 16 day post-exposure test (though, as noted, they were able to physically handle the spider at that time). But at 3 months there was less reported fear of spiders, and this persisted for the 1-year test

So, a few points:

  • Pretty remarkable that a single dose of propranolol can block the phobia for at least one year (though there were small numbers of patients in this study, propranolol certainly seems worth trying, it being a known and pretty innocuous med). Data from cognitive-behavioral therapy and extinction therapy (progressively increasing exposure to the feared object) show effectiveness, though that lasts only a brief time (personal testimony: I have some significant fear of heights. When I need to work on the roof of my house, it is really anxiety-provoking the first or second time up the ladder. But after a few times, I am fine going up and down, without concern — except that I am very careful. But then several months later, I am back to square one….)
  • And it is pretty interesting that the physiologic effect of propranolol is initially distinct from the cognitive effect, in that patients still stated they were still afraid of spiders at the day-16 test.
  • There are also some preliminary evidence that b-blockers decrease physiological responses to re-experiencing trauma in people with PTSD (it might be really interesting to try propranolol just after a person with PTSD has an experience which brings back memories of their trauma…). But, as a conceptual aside,​ I have seen several articles on the use of prazosin for PTSD (e.g., see AnnPharmacother 2007; 41: 1013)​, especially for decreasing the associated nightmares, and I have treated several patients with great, rather unexpected success. The concept is that sleep disorders are common with PTSD (70%), and that prazosin inhibits norepinephrine and perhaps thereby decreases the arousal in response to a stressor. In my experience, even very low dose prazosin has had dramatic results (e.g. 1-3 mg at night), though a recent article (Ther Adv Psychopharmacol 2014; 4: 43) notes that often higher doses are needed for full responses. These researchers also present 2 cases of patients of patients with psych comorbidities, one with PTSD and underlying major depressive disorder, but with a lot of daytime symptoms as well (hyperarousal, flashbacks, and re-experiencing the trauma) who responded pretty dramatically to prazosin 15mg in the am, 10mg at noon, and 20mg at night, and this high dose was very well-tolerated. The second patient with long-standing treatment-resistant major depressive disorder, PTSD and panic disorder and having failed a litany of meds, had lots of flashbacks, hypervigilance, reliving the experience, avoidance, nightmares, insomnia and concentration difficulties. She was titrated up to a dose of prazosin 15mg in am, 5mg at noon, and 10mg at night, and was also put on clomipramine for the depression, with a phenomenal response (PHQ-9 of 0 and asymptomatic PTSD).
  • So, as more studies unfold, the brain seems to be increasingly plastic/reprogrammable (i.e., it is not set for life at age 2, or 20, or ….)

Primary Care Corner with Geoffrey Modest MD: Are Healthcare Metrics Good for Our Health?

26 Jan, 16 | by EBM

By Dr. Geoffrey Modest

The NY Times had a recent editorial by Robert Wachter, a professor of medicine at UCSF, suggesting that focusing on specific metrics/measurements may distort good quality care (see ). [Thanks to Karen Henley for bringing this to my attention]. His points:

  • It is important to use metrics as part of the assessment of health care quality purview
  • There certainly is a basis to look at some issues of health care maintenance, or trying to decrease medical errors
  • But,
    • “There are so many different hospital ratings that more than 1,600 medical centers can now lay claim to being included on a ‘top 100’, ‘honor roll’, ‘grade A” or ‘best hospitals’ list”
    • Electronic medical records are supposed to help with the metrics/quality, but a 2013 study showed that they were the leading cause of doctor burnout, and ER doctors click a mouse 4,000 times during a 10-hour shift. [And, I would add, computers can decrease many aspects of quality, such as really listening to the patient and having good, empathic eye contact. See​ ]
    • He argues that “measurement cannot go away”, but that what we measure should be important clinical targets [I.e., not just the easy computer-generated claims-based ones], and that clinicians should be involved in the process of determining those metrics. He contrasts a really good measurement (certain hospital-acquired infections) with probably less useful ones (e.g., did the provider provide discharge instructions on asthma?, which is may be just remembering to check a box and giving the patient some pieces of paper, which they often do not look at later, without even going through the intended comprehensive patient education). And there needs to be a way to make these targets appropriate for different populations
  • But,
    • There are innumerable times that I get healthcare insurers sending me notifications that my nonhypertensive, nonproteinuric diabetic patient is not on the preferred ACE/ARB (and diabetes organizations have stated that these are not appropriate, for at least the last couple of years). These “reminders” add to the dramatically increasing amount of paperwork/emails/computer work that consumes my day…
    • I am put in a conflicting position: pretty commonly, I have patients ​come to me who have not had an A1c for the past 4-6 months, where getting the test is clinically useless. A frequent scenario is that they were out of the country for a few months and ran out of their meds (and Medicaid, a frequent insurer for my patients, will not give more than a 30-day supply of meds and they cannot be picked up early). So, when the patient returns to see me, their blood sugar is high, I reinstate their meds, and plan, appropriately I think, to check their A1c in 2-3 months. BUT if I do not check the A1C in the earlier visit, I get zinged by the insurance company (which I might get anyway, since the A1C will be “uncontrolled”), which affects the income to our health center.
    • Paying attention to some of the metrics takes time away from what may be the most important patient issues (getting a lot of testing may undercut my and the patient’s focus on the real issue: the fact that they are now homeless, do not have any income, are victims of domestic violence….  and that should take precedence). The point is that providing high-quality health care reflects a complex provider-patient relationship, where we (provider and patient) are sorting through an array of potentially complex medical, psychological/psychiatric, and social/behavioral issues, and by being pushed to focus on the mammogram or A1C, or discussing the risks and benefits of low-dose chest CT in smokers, is really a distraction in our way-too limited time together.
    • And, pay-for-performance (PFP, the system by some insurers to financially reward clinicians or their groups for adhering to the metrics) in studies a couple of years ago mostly showed that well-organized and well-funded, mostly middle-class organizations did well (and got extra money), and they did well on these metrics even before there was PFP. Less well-funded institutions did not get the monetary benefit of the PFP money, and really needed more money to help set up better organized systems.
    • So, I am certain there are lots of inefficiencies in the current system and that we as clinicians have missed important chances to provide accepted preventive interventions. And I support the concept of reminders and setting up systems in the practice to improve our performance (though the thrust in “poorly-performing” practices should be to help them financially and organizationally in improving their systems and not effectively decreasing their income). But these reward systems need to be appropriate clinically, derived from the local community needs, vary depending on the community (e.g., in some communities, rewards should be greater for screening for domestic violence, abuse, etc than A1C levels), be flexible enough to allow for clinician/patient discretion in their encounters (and dealing with social issues may happen much more often in some communities/clinics, such as a “healthcare for the homeless” clinic, where holding them to the same % target for some preventive services may be less appropriate), and be focused on helping practices in providing high quality care for their specific communities.

Primary Care Corner with Geoffrey Modest MD: New DM Management Algorithm

25 Jan, 16 | by EBM

By Dr. Geoffrey Modest

The Am Assn of Clinical Endocrinologists and the Am College of Endocrinology just came out with their 2016 updated diabetes management algorithms (see ). Note: this is different from the annual Am Diabetes Assn diabetes care summary, which I cannot access now (website down) but will review at some point.

  • Lifestyle optimization is the cornerstone of diabetes therapy:
    • Optimal weight, regular activity (at least 150 min/week of moderate-intensity activity such as brisk walking plus strength training), sufficient amount of sleep (6-9 hours/night is associated with improved cardiometabolic risk factors), tobacco avoidance, behavioral support including groups for emotional support, motivation, and specific support around issues such as weight loss and exercise.
    • They also support the 8 FDA-approved drugs for weight loss in overweight or obese patients, and bariatric surgery if BMI>35.
  • The importance to individualize A1C targets: though an A1c<=6.5 is optimal “if it can be achieved in a safe and affordable manner”. But “minimizing both severe and non-severe hypoglycemia is a priority”.
  • Glycemic control targets include fasting and postprandial blood sugars, as per self-monitoring of blood glucose (SMBG)
  • Minimizing weight gain is a priority
  • The choice of meds should be individualized, looking at the array of risk/benefits (see MEDS below)
  • The drug algorithm stratifies drug choices by initial A1C levels
  • Comprehensive therapy includes lipid and blood pressure therapy [though I would add smoking cessation, and decreasing alcohol when applicable]
  • Until patient is stable, then should be seen at least every 3 months to review SMBG, check A1c, check comorbidites (g. blood pressure, lipids), psychosocial issues [again, would add smoking, alcohol]. When stable, can have less frequent monitoring
  • Prediabetes: primary goal is weight loss when indicated, and they emphasize [appropriately] that these patients have a higher atherosclerotic risk and should have more aggressive risk factor management. There may also be a role for metformin, acarbose, TZDs, and GLP-1 agonists (espliraglutide3mg) in preventing diabetes, as shown in clinical trials


  • Low risk of hypoglycemia, promotes weight loss, “has good antihyperglycemic efficacy at doses of 2,000 to 2,500 mg/d”, and has “robust cardiovascular safety relative to sulfonylureas”. They do upgrade the creatinine/GFR limitations, commenting that some authorities recommend stopping metformin only when the GFR<30 (see blog: And they comment on vitamin b12 deficiency in up to 16% of users, and advocating monitoring levels and treating [see below on my comments on metformin dosing]
  • GLP-1 agonists. Low risk of hypoglycemia, robust A1C lowering, reduce blood sugar fluctuations in both fasting and postprandial states. Do not use exenatide if creat clearance <30, hx of pancreatitis (though no studies showing it can cause pancreatitis), and theycan cause delayed gastric emptying [which may be part of the weight loss, but can be a problem in patients with gastroparesis]
  • Sodium glucose cotransporter 2 (SGLT-2) inhibitors: cause glycosuria, decrease A1c, decrease weight. They recommend, despite the increased risk of mycotic genital infections, severe DKA though infrequent, increases in bone fractures in some studies. limited efficacy if GFR <45
  • Dipeptidyl peptidase 4 (DPP-4) inhibitors: enhances GLP-1 by blocking its breakdown, modest A1C effect, weight neutral, low risk of hypoglycemia,
  • Thiazolidinediones (TZD): directly reduce insulin resistance, relatively potent A1C effects, low risk of hypoglycemia. And pioglitazone may confer cardiovasc disease benefit. Adverse effects include weight gain, increased bone fracture risk, and increased risk of edema/heart failure. Decreased adverse effects if use <=30mg/d of pioglitazone. [Appropriately, they do not even mention rosiglitazone]
  • Alpha-glucosidase inhibitors (AGIs): modest A1C lowering, low risk of hypoglycemia, and some data finding lower cardiovasc Adverse effects of bloating, flatulence, diarrhea.
  • Sulfonylureas (SFUs): relatively potent hypoglycemic effect (though lower durability than others), risk of weight gain and hypoglycemia. Glinides have lower effect on A1C, have shorter half-life, and lower risk of hypoglycemia
  • Colesevelam: bile-acid sequestrant which lowers A1C modestly, no hypoglycemia, decreases LDL levels, though 10% have GI intolerance.  can also increase triglycerides
  • Bromocriptine: slight A1C lowering, no hypoglycemia, and may be associated with decreased cardiovascular events
  • Insulin: most potent glucose-lowering agent. Consider especially in patient on 2 oral agents and A1C>8%. Usually start with single daily dose of basal insulin. Dosage can be adjusted by the patient, with clear instructions, based on SMBG readings. If on basal insulin and inadequate control, consider adding GLP-1, SGLT-2 or DPP-4 active drugs [I personally advocate the GLP-1 ones, as below]. May need to decrease insulin dose when adding these drugs. Can also add mealtime rapid insulin bolus to cover the meals. Problems with insulin include weight gain (1-3 kg more than those on other agents) and hypoglycemia (7-15% have at least 1 episode/yr). [I also have some concern about potential cardiovascular effects of the commonly associated hyperinsulinemia, as per prior blogs]
  • In new onset diabetics with A1C<7.5, lifestyle plus metformin. But can use pretty much anything (GLP-1, SGLT-2 inhib, DPP-4 inhib, TZDs, AGIs, SFUs or glinides)
  • If A1C>7.5, use metformin withanother drug. [See below. I think this is too aggressive and potentially dangerous]
  • A1C>9 who are symptomatic might benefit from insulin. Asymptomatic, can use insulin or “may initiate therapy with maximum doses of 2 other medications” [See below. I think this is too aggressive and potentially dangerous]


  • Blood pressure: target <130/80 [lower than JNC-8 suggests]. Less stringent in elderly [and I would add my usual caveats: check orthostatics regularly anyway, since I have seen many patients with systolics of 150 range which fall rapidly to 110 on standing; and even 130+ may be associated with cognitive decline]. In younger patients, a goal of <120/80 is reasonable if achieved safely, which seems to have a protective effect against strokes in a few studies. Lifestyle therapy is also important here (weight loss, Na restriction, alcohol restriction [though they comment on the benefits of moderate drinking, which I have challenged in several recent blogs, such as], and exercise. They also suggest starting with 2 agents if BP>150/100 [which I still think is too aggressive/potentially dangerous, and has appropriately been abandoned by JNC-8]
  • Lipids: all diabetic patients should be classified as high or very high ASCVD risk, even those <40 yo. They recommend treating-to-goal (unlike the AHA recommendations) with LDL targets of <100 or <70 depending on their ASCVD risk. Lifestyle changes as above. They do bring up using ezetimibe to achieve further LDL reduction [though I am hesitant to endorse this; see ] and the PCSK9 inhibitors. They also mention colesevelam and fibrates as potential add-ons, and even high-dose niacin. Also, high dose of fish or omega-3 fish oil. In those with triglycerides>500, consider low-fat and low carbohydrate diet, fibrates, omega-3-fatty acid and/or niacin (though these combos are unproven in clinical trials)


  • They do acknowledge the importance of life style optimization. In my experience, treating diabetes is the hardest issue in primary care, since achieving optimal (or even marginal) control is contingent on motivating the patient to optimize their lifestyle issues: exercise is often difficult for patients (g., for medical reasons such as knee arthritis or underlying depression/other psych issues, etc; or for practical reasons such as unsafe neighborhoods to walk around for personal safety issues or specific neighborhood issues such as poorly maintained sidewalks, unsafe crosswalks, etc), and diet is often difficult for a variety of reasons (lack of access to healthy foods, traditional diets high in carbs, psych issues, just trouble  breaking the habit of eating unhealthy and very tasty foods for a long time, etc). and if someone does not eat a reasonably consistent diabetic die, blood sugars may range from 45-450 mg/dl, and simply cranking up the diabetic meds is potentially dangerous no matter what their A1C is (ie, lowering the 45 even further).
  • As noted in many of my blogs on diabetes meds, there are precious few studies looking at important clinical outcomes. The FDA many years ago allowed drug companies to use the A1C as a surrogate marker for advances in diabetic meds. I will certainly grant that there are enough older studies that have shown consistently that lowering the A1C does lower the risk for microvascular disease, and I do not want to understate the importance of that. BUT, diabetic patients, by far, die from macrovascular disease (heart disease, strokes…), on the order of 80% of their deaths. And, though studies suggest that lowering the A1c can be beneficial for that outcome, it really depends on the drug (g., rosaglitazone was pretty effective in lowering A1C levels, but was associated in several studies with increased cardiac outcomes). The DPP-4 drugs have also been associated with adverse cardiovascular events in some studies (see And there are other significant adverse outcomes which may override the benefits of lowering the A1C: the FDA just supplemented their warning on SGLT-2 inhibitors, which are also pretty good for lowering A1c, but are associated with severe urosepsis and ketoacidosis (for prior blog, see​ )
  • So, my general approach to drugs (again, always continuing to work on the lifestyle issues) is:​
    • Metformin is generally very well-tolerated. I have seen dramatic effects of low dose metformin, with my typical starting dose of 500mg once a day, and only raise the dose as needed. There is decreased marginal effectiveness with increasing doses, similar to statins, where the majority of effectiveness is with the lowest dose. Also, metformin is better tolerated from a GI perspective when taken with meals. I think metformin is the single most useful agent for diabetic control. also, see the curious/interesting blog on metformin and the intestinal microbiome, suggesting that much of its effect may be from positive changes in the microbiome (see: I am concerned with the above recommendations which advocate using pretty much any drug as a single agent, as opposed to other guidelines which make metformin #1, then adding others.
    • I really disagree with their aggressive approach to patients with new onset diabetes. Many have A1C in the 13-19 range. I think this is mostly due to slowly progressive increases over time, which in part is from glucotoxicity (that is: as the sugar increases, the endogenous insulin works less well with documented decreases in insulin-mediated glucose uptake, leading to higher blood sugars), and this progresses until the patient becomes very symptomatic. Physiologically, my approach, as with very high but not hypertension emergency, is to see the increased sugar or blood pressure typically as a slowly progressive, with some physiologic adaptation by the body, and that it is safer to lower the blood sugar or pressure more slowly with frequent follow-up (I have seen too many patients treated aggressively, leading to electrolyte disturbances in diabetics and strokes in hypertensives).  In terms of new onset diabetes with initial A1C >15%: my experience with at least 25 newly diagnosed diabetics, 100% do very well and for a long time with just metformin (which often is initially augmented with insulin, which can be tapered off over weeks to months). And this is typically with just 500-1000 mg metformin/day.
    • Lately, I have been using more GLP-1 agonists if metformin needs a boost. They work really well, are mostly very well tolerated, do not cause hyperglycemia, to me seem to be pretty targeted and physiologic agents (they restore the “incretin effect”, which is blunted in diabetics). I also prescribe older drugs with more of a track record and known adverse effects (g., insulin, sulfonylureas). Though I often do add them to metformin, there is not much support for their macrovascular efficacy. Pioglitazone, especially in the 15-30mg range, works well and seems to be cardioprotective, per the PROACTIVE study (Lancet2005; 366: 1279)
    • ​For the newer drugs, I really focus on those with the best efficacy/adverse effect ratio, and also those that have the least ubiquitous effects (g., the DPP4 inhibitors poison a prevalent enzyme system in the body, so I am hesitant to use them because their efficacy is not so great in lowering A1C levels, and their potential for longterm adverse events is high given the abundance of potential non-diabetic effects). These issues have become more concerning as the FDA is pushing through approvals more quickly, often with the intention of requiring post-marketing surveillance which the drug companies mostly do not do (see blogs below). And, of note, 19 of the 20 members of the committee developing these recommendations have drug company ties….

See: which critques the study on the SGLT2 inhibitor empaglifozin which purportedly shows cardiac benefit, though the FDA came out with an advisory about it causing severe urosepsis and ketoacidosis

But, as noted in prior blog , <20% of required postmarketing studies for medical devices were done within 3 years of the FDA requiring them, and no fines have been issued for noncompliance with the FDA mandates. which goes through some of the drug company malfeasance in nonreporting of severe drug adverse effects, lack of follow-up on postmarketing studies, and the poor enforcement by the FDA.​ promoting fructose restriction in diabetes management which questions advocating alcohol for diabetics with FDA warning about DPP-4 inhibitors and joint pain


Primary Care Corner with Geoffrey Modest MD: Antibiotic Overprescribing and Acute Respiratory Infections

22 Jan, 16 | by EBM

By Dr. Geoffrey Modest

In my never-ending pursuit of protecting the microbiome, and also decreasing further development of antibiotic resistance, there was a helpful clinical guideline from the American College of Physicians and the CDC (see doi:10.7326/M15-1840). Nothing new here, but it really reinforces both the remarkable overuse of anti-bacterial antibiotics for nonbacterial conditions and the very common conditions when this is happening (the lowest-hanging fruits), which if implemented should decrease bacterial resistance, decrease antibiotic-associated adverse events including lots of deaths, and save lots of money (and the microbiome).


  • Antibiotics are prescribed in >100M adult ambulatory care visits in the US/yr, with cost of $10.7 b(as in, billion)​ in 2009, with $6.5 b in the community setting
  • 41% of prescriptions are for respiratory conditions
  • There are >2M antibiotic-resistant illnesses and 23K deaths in US/yr, with cost of $30 b
  • Higher rates of multi-drug resistant pneumococcal disease occur in places where there is more prescribing of broad-spectrum antibiotics, esp extended-spectrum cephalosporins and macrolides
  • And, antibiotics as a group are responsible for the largest number of medication-related adverse events, including 20% of ER visits for adverse drug reactions
  • Adverse drug reactions range from mild to serious: an estimated 5-25% of patients on antibiotics have adverse events, and 1 in 1000 have a serious one. E.g.: c difficile causes 500K infections and 29,300 deaths/year in the US, with an estimated $1 billion in extra medical costs
  • An estimated 50% of outpatient antibiotics are considered unnecessary, with a direct cost of $3 b
  • Although (the good news) antibiotic prescriptions have decreased 18% in people>5 yo in the past decade, (the bad news) prescriptions for broad-spectrum fluoroquinolones and macrolides have increased >4-fold


  1. Acute uncomplicated bronchitis (self-limited inflammation of bronchi with cough, which may or may not be productive), lasting up to 6 weeks.
  • ​Has 100M outpatient visits/yr (10%!! of total), and >70% result in antibiotic prescription!! — the largest cause of inappropriate antibiotic prescriptions
  • ​Need to differentiate from pneumonia (in adults <70 yo, pneumonia unlikely in absence of all of: tachycardia with HR>100, tachypnea with RR>24, fever with T>38C, and abnormal findings on chest exam (rales, egophony, or tactile fremitus)
  • Meta-analysis of 15 RCTs found limited benefit of antibiotics but a trend to increased adverse events. and no clear decrease in days to cough resolution
  • There may be benefit from cough suppressants (dextromethorphan or codeine), expectorants (guaifenesin), first-generation antihistamines (diphenhydramine), decongestants (phenylephrine), b-agonists (albuterol), though data to support specific therapies are limited and b-agonists don’t seem to help unless there is underlying asthma or COPD [by the way, I think there are lots of scripts for the more selective antihistamines, such as loratadine, which have limited effectiveness for viral infections]
  • ​There are some cases of antibiotic-sensitive causes (mycoplasma, chlamydia, B pertussis), though these are uncommon and should be considered in cases where transmission in the community is reported [no comment here that it is hard to know of transmission in the community if these are not tested for….., but in broad strokes, treating people indiscriminately with antibiotics does not seem to help]
  • So, do not perform testing or give antibiotics in patients with bronchitis unless pneumonia is suspected


  1. Pharyngitis (benign self-limited illness with sore throat, worse when swallowing, with or without constitutional symptoms
  • 12M outpatient visits/yr (1-2% of total). 60% get antibiotics
  • No further testing indicated if likely viral: associated symptoms of cough, nasal congestion, conjunctivitis, hoarseness, diarrhea, oropharyngeal ulcers/vesicles
  • Those with symptoms suggestive of strep should get rapid strep test, throat culture or both: e.g. symptoms of persistent fever, rigors, night sweats, tender lymph nodes, tonsillopharyngeal exudates, scarlatiniform rash, palatal petechiae, and swollen tonsils.
  • ​The Centor criteria are often used (fever, tonsillar exudates, tender anterior cervical nodes, absence of cough), BUT because the positive predictive value for group A strep is so low, the IDSA (Infectious Diseases Society of America) only suggests using these as a means to do no further testing or prescribing antibiotics if there are <3 criteria.
  • Some patients present with severe signs/symptoms, such as difficulty swallowing, drooling, neck tenderness or swelling, and should be evaluated for peritonsillar abscess, parapharyngeal abscess, epiglottitis, Lemierre syndrome
  • If strep found on testing, then treat with appropriate narrow-spectrum antibiotics, which decreases duration of sore throat by 1-2 days, but may also decrease risk of acute rheumatic fever (more common in kids and teens), peritonsillar abscess, and spreading the infection.
  • No need to treat asymptomatic carriers of strep, since low likelihood of spreading it and low potential for complications
  • ​Not do tonsillectomy in adults just to reduce frequency of recurrent strep infections
  • Treat adults with analgesics (aspirin, acetaminophen, NSAIDs), throat lozenges. Little data on salt water, viscous lidocaine [or, I would add, tea with honey or lemon, as is preferred by many of my patients]
  • ​So, test patients with symptoms suggestive of Group A strep pharyngitis, and treat only if confirmed strep [and use narrow-spectrum antibiotics, my addition]


  1. Acute rhinosinusitis (self-limited viral infection, allergy, or irritant which causes inflammation in nasal or paranasal sinus cavity. Lasts 1-33 days with most resolving in a week. Can be associated with nasal congestion, purulent discharge, maxillary tooth pain, facial pain/pressure, fever, cough, hyposmia/anosmia, ear pressure, headache, hallitosis.
  • >4.3 M adults have diagnosis annually, >80% get antibiotics, mostly macrolides, and most unnecessarily so
  • No role for radiologic imaging — does not reliably differentiate bacterial from viral causes
  • Bacterial cause more likely if symptoms persist >10 days without improvement, symptoms are severe (fever >39C, purulent nasal discharge, or facial pain >3 days), or if new onset fever, headache, increased nasal discharge after a viral URI was improving
  • If treating, preferred agent per IDSA is amoxacillin-clavulanate,with doxycycline or respiratory fluoroquinolone as alternative; the Am Acad of Otolaryngology emphasizes initial management of watchful waiting regardless of symptom severity; some medical societies recommend amoxacillin (no direct evidence that amoxacillin-clavulanate is clinically superior). In fact rhinosinusitis is usually self-limited even if caused by bacteria. A meta-analysis found that the number-needed-to-treat was 18 for 1 patient to have a more rapid cure, but the number-needed-to-harmfrom adverse antibiotic effects was 8
  • Nasal saline irrigation and intranasal steroids may alleviate symptoms and decrease likelihood of antibiotic use. Other supportive therapies include analgesics for pain, systemic or topical decongestants, mucolytics, and antihistamines
  • ​So, reserve antibiotics unless patient has persistent symptoms >10 days, onset of severe symptoms/signs of fever > 39C and purulent nasal discharge or facial pain for >3 consecutive days, or worsening of symptoms after a viral illness that lasted >5 days and was initially improving


  1. URI
  • ​37 M ambulatory care visits (3%), 30% get antibiotics
  • Complications include acute bacterial sinusitis, asthma exacerbation, and otitis media. Antibiotics play NO ROLE in preventing these.
  • ​Best means to prevent transmission: handwashing
  • Symptomatic therapy, though advise patient that symptoms can last 2 weeks. antihistamine-analgesic-decongestants work. Zinc supplements help if given within 24 hours. no evidence for vitamins/herbal remedies
  • So, do not prescribe antibiotics for patients with the common cold

As many of you know, I have sent out many blogs on this. In particular:  highlights the worldwide emergence of antibiotic-resistant bugs includes a slew of blogs on antimicrobial resistance, including long-term changes in the gut microbiome after even a single dose of antibiotics, importance of antibiotic use in food industry in creating very threatening changes in microbial sensitivity, effects of international travel on changes in microbial sensitivities in the gut microbiome, and the real importance of using the narrowest-spectrum antibiotics in treating pneumonia/strep pharyngitis

So, I do realize that it can be difficult to dissuade some patients from getting antibiotics. I frequently hear “but my (cough, bronchitis, cold…) is bad and in the past whenever I get antibiotics it goes away right away”. Though, I have noted over the years that fewer patients are so insistent (perhaps relating to my sense that it is the dying breed of older patients who are more insistent than younger ones). In any event, I have had some success in stating over-enthusiastically that “the good news is that you do not have a bacterial infection, do not need antibiotics, but I can give you some medicines to help relieve the symptoms.” Or, if pushed, “antibiotics really don’t help this type of infection, which gets better on its own. And there is a very real chance you could get a very serious side-effect. So I really think it is important not to take antibiotics”. And in my experience (perhaps augmented by my gray hair), these are usually successful and lead to a satisfying encounter, without antibiotics being prescribed.

Primary Care Corner with Geoffrey Modest MD: Barretts Esophagus Guidelines

21 Jan, 16 | by EBM

By Dr. Geoffrey Modest

The American College of Gastroenterology updated their recommendations for screening for Barrett’s Esophagus (see doi: 10.1038/ajg.2015.322)


  • There is increasing prevalence of GERD in the US and worldwide
  • GERD is associated with 10-15% risk of Barrett’s esophagus (BE), i.e. in 1-2% of the population, with risk factors of male sex, central obesity, intensity/duration of GERD sx, and >50yo. The most profound risk factors are: those with early onset (<30 yo) and weekly symptoms (OR 31.4), and family history of BE or esophageal adenocarcinoma (EAC) (OR 12.23). White men with GERD have 2% chance of BE in the third decade of life, increasing to 9% by 6th decade. Men have twice the risk as compared to women for both BE and EAC. Alcohol does not increase risk of BE.
  • BE is associated with EAC, which has also been increasing in incidence. Risk factors for EAC in people with BE include: advancing age, increased length of BE segment, central obesity, tobacco use, lack of use of NSAIDs, lack of usage of PPIs, lack of usage of statins
  • The relationship with H pylori and BE is complex. Certain strains (e.g. cytotoxin-associated gene A, or Cag A+) may have a decreased risk. A VA study confirmed a lower risk of BE in those with H pylori, especially in those with gastric atrophy (which might be associated with the increased risk of stomach cancer, however) or use of antisecretory meds [see The American Journal of Gastroenterology109, 357-368 (March 2014)]. The Cag A+ strain is a more aggressive H pylori strain and decreases gastric acid production more, which may be the reason it may be somewhat protective for developing BE.
  • Risk of EAC:
    • For those without dysplasia, 0.2-0.5%/yr
    • With low-grade dysplasia: 0.7%/yr
    • With high-grade dysplasia: 7%/yr
    • ​90% of patients with BE die from causes other than EAC
  • Consider in men with >5 year history and/or frequent (weekly or more) symptoms of GERD, and 2 or more risk factors (age >50, white, central obesity with waist circumf >102 cm/40 inches or waist-hip ratio of >0.9, current or past smoking, confirmed family history of BE or EAC in a first-degree relative. (Strong recommendation, mod level of evidence)
  • In females (who have half the BE risk), screening not recommended. But, consider in individual cases if multiple risk factors (as above, though in women waist circumf of > 88cm/35 inches or waist-hip ratio of >0.8). (Strong recommendation, low level of evidence)
  • Do not screen general population [though it is important to remember that 40% of EAC occur in patients without history of GERD]
  • Consider life expectancy in decision to screen
  • Can do unsedated transnasal endoscopy instead of conventional upper endoscopy
  • If initial endoscopy is negative for BE, no need to repeat. If esophagitis, repeat endoscopy after 8-12 weeks of PPI to ensure healing and exclude underlying BE (conditional recommendation, low level of evidence)

So, a few points:

  • The Am Cancer Society’s estimated incidence of esophageal cancer in 2015 is: 16980 new cases, with 13570 in men and 3410 in women.
  • Of the common esophageal cancers, BE is associated with adenocarcinoma and not with squamous cell carcinoma (squamous cell ca in Western countries is largely associated with alcohol and smoking, and its incidence is decreasing in parallel to decreases in these risk factors)
  • EAC has an abysmal survival rate when detected late (which is common in those who are symptomatic): those with regional or distant disease have a 5-year survival <20%.
  • I personally think there has been supportive data for many years that men >50yo with chronic GERD symptoms should get a one-time endoscopy, so I welcome that recommendation (and in the past, when I have referred patients for EGD, some gastroenterologists would do them and others would not). But there are some unclear parts of the new recommendations:
    • Per the letter of their recommendation, a 35 year old white man with central obesity and a 5.1 year history of GERD symptoms once a week should get an endoscopy. I doubt there are data to support that
    • ​Though women seem to be spared the higher likelihood of BE or EAC, EAC is still a terrible disease, so I would like to see a clear risk/benefit analysis before not recommending this screen. Prior guidelines from the Am Gastroenterology Assn have just focused on risk factors overall, where male sex is a risk factor, but women would be included in the recommendation for endoscopy if they had other of the risk factors.
    • One overhanging concern with the recommendations is the lack of smoking gun evidence that doing surveillance and treating Barrett’s leads to fewer cases of EAC. Case-control studies are suggestive (though there are some showing no decrease in cases in those undergoing consistent monitoring and treatment of BE), despite the reasonable pathophysiology suggesting an orderly progression from BE to increasing levels of dysplasia to malignancy.
    • Bottom line: who knows?? To make an informed decision, we need more data (e.g., does screening really work? What are the real risk/benefit analyses for women, assuming screening helps?). At this point, it seems reasonable to me to screen men and women with lots of risk factors one time around age 50 (though I understand that the number of people >50 yo who have central obesity, chronic GERD symptoms, and present/past cigarette smoking is pretty staggering).

Primary Care Corner with Geoffrey Modest MD: 3 Heart Failure Articles of Note

20 Jan, 16 | by EBM

By Dr. Geoffrey Modest

There were 3 useful articles in the JACC-Heart Failure journal which I think are useful in primary care

  1. This article (see​) looked at the target for b-blocker (BB) use in patients with heart failure and reduced ejection fraction (EF, <35%) within the HF-ACTION study, the largest trial of BB use involving 2331 medically stable outpatients and showing benefit from a structured aerobic exercise program over 2.5 years (see JAMA 2009; 301(14): 1439). The issue addressed is titration of the BB dose: the HF-ACTION and other trials have found that just lowering the heart rate (HR) improves mortality but raise the question of whether that explains the full BB effect, or is the achieved dose of the BB more important.


  • 2320 patients (mean age 60, 28% female, 32% black, BMI 30, 62% NYHA Class II, 50% ischemic cardiomyopathy, EF 25%, SBP 110 mmHg, resting HR 70, 80% in sinus rhythm).
  • Divided into 2 groups: those on high dose carvedilol (>=25mg/d) vs low dose (<25 mg/d); and those with high heart rate (>=70 bpm) vs low (<70 bpm)


  • All-cause death or all-cause rehospitalization (all with adjusted HR values)
    • High vs low dose BB, hazard ratio 0.87 (0.77-0.99), p=0.03, i.e. 13% decrease in deaths/hospitalizations
    • High vs low HR, hazard ratio 11 (0.98-1.24), p=0.09, nonsignificant
    • And looking at the combination, adding HR to the analysis added no significant value
  • There was a significant 21% decrease in all-cause death as well as a 23% decrease in cardiovascular death or heart failure rehospitalization with high-dose BB. HR was not significant for any of the outcomes in either unadjusted or adjusted analyses. Part of the reason for the lack of significance is the relatively low number of events, under powering the study.
  • The association between BB dose and the above clinical outcomes was independent of the baseline HR
  • There was no significant difference in the biomarker NT-proBNP between these groups, though there was more of a decrease in the biomarker in the high BB dose group, though nonsignificant. Related to small numbers?

So, this study reinforces maximizing the BB dosage, even in those with HR<70bpm. I’m not sure what the BP was for this group (not reported), but the high numbers of patients in the community not on optimal BB doses (in the 80-90% range!!) reflect perhaps a less aggressive approach to therapy than seems warranted. I should add the caveat that this is a post-hoc analysis, and as such is potentially limited by confounding.


  1. A meta-analysis of 11 studies looking at the progression of asymptomatic heart failure (HF) to symptomatic HF, over an average of 7.9 years (see org/10.1016/j.jchf.2015.09.015), assessing both those with ALVSD (asymptomatic left ventricular systolic dysfunction), and ALVDD (asymptomatic left ventricular diastolic dysfunction).


  • 11 studies identified with 5,369 patients. 5 with ALVSD, 3 with ALVDD, and 4 with both
  • ALVSD somewhat variably defined, but typically ejection fraction (EF)<30-50%


  • ALVSD: absolute risk of progression to symptomatic HF was 8.4/100 person-years (4.0-12.8 per 100 person-years); adjusted relative risk was 4.6 (2.2-9.8)
  • ​ALVDD: absolute risk of progression to symptomatic HF was 2.8/100 person-years (1.9-3.7 per 100 person-years); adjusted relative risk was 1.7 (1.3-2.2)
  • In controls (no ventricular dysfunction): absolute risk of progression to symptomatic HF was 1.04/100 person-years (0.0-2.2 per 100 person-years)
  • ​Per a 1-SD change in EF, there is an overall 38% increased risk of developing clinical HF [RR=1.38 (1.14-1.67]
  • The main predictors of progression were: age, sex, blood pressure, diabetes, and BMI

So, it is pretty clear that asymptomatic LV dysfunction is associated with a significant risk of becoming symptomatic, more so in those with reduced systolic function, and with increasing risk as the EF decreases. Studies have shown that asymptomatic LV dysfunction is also associated with reduced survival, and interventions as with ACE inhibitors prevent or delay the development of symptomatic HF and decreases mortality.


  1. Looking at the CIBIS-ELD trial (Cardiac Insufficiency Bisoprolol Study in Elderly), researchers assessed the tolerability and efficacy of bisoprolol vs. carvedilol in symptomatic (at least NYHA class 2) patients > 65 yo with either HFrEF (heart failure with reduced ejection fraction) or HFpEF (heart failure with preserved EF) —​.


  • 626 patients with HFrEF (mean age 73, 26% women, 61% NYHA class 2/32% class 3, 15% with peripheral edema, mean HR 75, mean BP 134/80, BMI 27, EF 35%, NT-proBNP 968; CAD in 68%, dilated cardiomyopathy in 19%, hypertension in 80%, 9% smokers, 48% with MI) and 250 with HFpEF (mean age 73, 66% women, 76% NYHA class 2/23% class 3, 35% with peripheral edema, mean HR 71, mean BP 146/80, BMI 29, EF 59%, NT-proBNP 253; CAD in 33%, dilated cardiomyop in 2%, hypertension in 90%, 8% smokers, 18% with MI) were randomized to bisoprolol vs carvedilol, with a goal to up-titrate to the target or maximally tolerated dose
  • Carvedilol started at 6.25mg bid with target of 25 mg bid (or 50mg bid in those with body weight >85kg); bisoprolol was 2.5mg daily with target of 10mg daily, with biweekly dose doubling
  • Titration was to be over 6 weeks (8 weeks for those >85kg and on 50mg carvediol), then 4 week maintenance period


  • Mean daily dose of bisoprolol after titration was 4.93 mg in the HFpEF and 5.01 in HFrEF groups; for carvedilol it was 25.3 and 29.1 respectively) — nonsignificant differences (19% of HFpEF vs 27% with HPrEF reached target dose, regardless of treatment group, which was significant)
  • ​It took longer to uptitrate those with preserved EF (12%) vs reduced EF (5%)
  • Similar HR reductions, at 7 bpm in both those with preserved and reduced EFs. BP decreased more with the preserved EF group (14/5 mmHg) vs reduced (8/4)
  • NYHA functional class improvement:
    • Reduced EF group had 34% improvement vs preserved EF group (decrease of 31% vs 18%): 23% improvement (p<0.001  for comparison), and no diff if on carvedilol vs bisoprolol
  • 6-minute walk distances and physical quality of life measures improved in the HFrEF group but no change in HFpEF groups (20 vs 4 meters).
  • In 3 measures of quality of life (SF-36 PFS, SF-36 PCS and SF-36 MCS, reflecting physical functioning, physical component, and mental component): there was only a significant change in the HFrEF group only
  • ​Echo for systolic function: mean EF improved in those with reduced EF from 35% to 39% by week 12, no change in the preserved group.
  • ​Echo for diastolic function: mean E/e’ and left atrial volume index not change in either group, but E/A mitral flow did increase in those with preserved EF [i.e., BBs did not change any markers of diastolic function in either those with preserved or reduced EF), and no difference between the BBs]
  • ​Adverse effects: more in HFpEF patients (79%) vs HFrEF (58%), with p<0.001 for difference, mostly bradycardia (20% vs 11%), dizziness (15% vs 4%), and fatigue (18% vs 4%)

So, one of the major conundrums in treating HF is that about 50% have HFpEF (preserved EF), with increasing prevalence over time. This group in general has the same overall prognosis as those with reduced EF. And we have minimally effective treatments as compared with those with reduced EF. On the surface, BBs would seem to be reasonable drugs (they decrease BP, decrease LVH, decrease HR, and decrease ventricular arrhythmias — a leading cause of death in this group). This study did not find any really significant differences between using bisoprolol vs carvedilol, either in tolerability or efficacy. But there were real differences between those with reduced EF (easier ability to titrate dose up, more improvement in NYHA functional class and measures of quality of life, walking distance, EF, fewer adverse events) and much less in those with preserved EF (more decrease in HR, BP, and no real clinical improvement). This was a short-term study, and without a placebo group,


So, why do I bring up these 3 articles?

  • HF is really common (5.7 million US adults, lifetime risk in a young adult being 20%), has a substantial morbidity/mortality, and there are lots of people with asymptomatic HF (3-4x the number with symptomatic).
  • The first article found that for those with reduced EF, trying to push BB to the targeted dose (which, unfortunately, does not happen so often in community practice), seems to confer the most benefit
    • The treatment algorithm is pretty straightforward (diuretics as needed, then ACE/ARB, then BB, then aldosterone antagonists; though can use hydralazine/isosorbide dinitrate in African-Americans and those not tolerant of ACE/ARBs), but these meds all lower blood pressure, and in my experience the not-so-uncommon limiting factor is that the blood pressure can’t abide this many agents at full dose, leaving us some unanswered questions:
      • What are the priority agents (in general the studies have assumed this medication order, but if the blood pressure is low, which individual agent is the most effective?)
      • Is it better to have low doses of multiple agents vs maxing out the meds in a specific order (studies do suggest that low doses of meds are useful)?
    • And another issue is hyperkalemia. Though the diuretics help counter this, all the other meds make it worse. And this is especially an issue in diabetics (not-so-uncommon), where hyperkalemia is more often an issue.
      • Again, should we give a little of many agents or push individual agents in a particular order? (E.g. my experience is that BBs are less hyperkalemic than the ACE/ARB or aldo antagonists. I should add that the CIBIS III trial did find that starting BBs first seems to be equivalent in outcome to ACE/ARBs, but the advantage of ACE/ARB first is that they work faster, and BBs can cause worsening heart failure in the short-term, further supporting starting first with ACE/ARB to better optimized initial treatment)
    • The second article found that those with asymptomatic HF have a real risk of progression to symptoms, more-so in those with reduced EF (systolic dysfunction). And these asymptomatic people, a very common issue, are at much higher risk of morbidity and mortality, raising:
      • The potential possibility that we should be screening more routinely for asymptomatic LV dysfunction, then doing interventions to improve the function and potentially decrease both the likelihood of symptomatic HF and mortality. Seems like a useful study to do, since the prevalence, morbidity and mortality of HF is so profound. But at this point it is not clear that early identification and treatment changes any clinical outcomes.
      • And, it reinforces the likely benefit of preventing ventricular dysfunction in the first place: decreasing systolic dysfunction by decreasing cardiovascular risk factors, the development of nonischemic cardiomyopathies (e.g. decreasing stress, treating infections such as chagas, lyme early and aggressively, etc.); and decreasing diastolic dysfunction (also treating hypertension, stress, development of myocardial ischemia)
    • And the third article, though actually a short-term tolerability study, highlights the difficulty of treating those with preserved EF, though it should be pointed out that this was a short-term study and the mechanism of disease in those with preserved EF is different from those with reduced EF (i.e., it may take longer to reverse the cardiac remodeling in those with preserved EF)
      • ​But the bigger and more relevant/longer-termstudies do not find the same benefit with current treatment of those with preserved EF as in those with reduced EF, though there are some data that ARBs decrease hospitalizations (CHARM-Preserved trial) and that spironolactone decreases hospitalizations and (secondary analysis) deaths in those with high BNP levels (TOPCAT trial) though with a fair amount of hyperkalemia
      • In fact, treatment of those with preserved EF is pretty much several decades behind where we are with patients with reduced EF (where we were basically using loop diuretics and digoxin for symptom control, without improving mortality)
      • e., we really need better drugs (and reinforcing the above interventions to decrease the development of left ventricular dysfunction with preserved EF to the extent we can)
    • But my experience is that treating HF, esp. in those with reduced EF, is really easy to do in the primary care setting in the vast majority of patients and that there is really a very remarkable response to the meds with the above-noted algorithm (I have had several patients with shockingly low EFs in the 10-15% range who function totally normally for more than 10 years…). And we can also provide symptom control for the majority of those with preserved EF, using diuretics, ACE/ARB, ?spironolactone (which are more effective than BBs anyway in improving left ventricular size and shape), and BBs as needed to help control HR, anginal symptoms, and BP. Treating HF one of the more rewarding primary care interventions, since patients pretty regularly respond so well clinically to these interventions and feel much better so rapidly.

Primary Care Corner with Geoffrey Modest MD: Antithrombotic Therapy Guidelines

19 Jan, 16 | by EBM

By Dr. Geoffrey Modest

The American College of Chest Physicians just came out with updated guidelines on antithrombotic therapy (see doi 10.1016/j.chest.2015.11.026). Details of those recommendations that are important for primary care providers (i.e., not including those critically ill, use of thrombolytic therapy, catheter-based thrombus removal):

  • They suggest using dabigatran, rivaroxaban, apixaban, or edoxaban over vitamin K antagonists (VKA, such as warfarin) for those with venous thromboembolic disease (VTE) and no cancer (all grade 2B recommendations, where Grade 2 recommendations are “weak recommendations”, based on “moderate” evidence = Grade B)
  • For those with VTE and cancer: low molecular weight heparin (LMWH) preferred over dabigatran/rivaroxaban/apixaban/edoxaban – i.e. NOACs  (non-vitamin K oral anticoagulants)  (grade 2C, where C=”low quality evidence”)
  • For those with VTE and on anticoagulants, do NOT use IVC filter — Grade 1B (strong recommendation)
  • Recommends against using compression stockings routinely to prevent post-thrombotic syndrome (PTS) — grade 2B [a large study did not find stockings decreased PTS or reduce leg pain, though they may be useful for chronic symptoms of PTS]
  • For those with subsegmental PE and no proximal DVT, they suggest clinical surveillance over anticoagulation if a low risk of recurrent DVT (grade 2C), but anticoagulation if high risk (also grade 2C) [2 issues here: these are often picked up by CT scan and can be false positives; and if real, are likely from small DVT and risk of recurrence or progression is small, but there are NO RCTs on this. reasonable to do bilateral leg ultrasounds, and upper extremity one if there is a central venous catheter]
  • For those with recurrent DVT while on non-LMWH therapy, then give LMWH (grade 2C); and if already on LMWH, then increase the dose (grade 2C)

In the body of the article, they note:

  • For those with proximal DVT or PE and nonsurgical transient risk factor, anticoagulate for 3 months
  • For patients with distal DVT provoked by surgery or nonsurgical transient risk factor, anticoagulate for 3 months [they note that there is a 15% risk of extension to the popliteal vein, and note that the decision to anticoagulate is not clear, with some providers suggesting anticoagulation, others only if there is proximal propagation in 1-2 week followup ultrasound. They also comment that a risk factor for extension is a positive D-dimer, especially if very high (not specified further), if thrombosis is multiple or >5cm long, if close to proximal vein, if active cancer, if history VTE, also if DVT not in the muscular veins of the soleus or gastrocs).
  • In those with unprovoked DVT (isolated distal or proximal DVT/PE), they recommend 3 months of anticoagulation. Re-evaluate the patient after 3 months for risk-benefit of continued therapy. Later in the text of the article, they comment: in those with isolated acute distal DVT and without severe symptoms or risk factors for extension (see above), do serial imaging of deep veins. If the thrombus extends, even within the distal veins, then anticoagulate
  • But, in those with a first VTE which is an unprovoked proximal DVT/PE and low to mod bleeding risk, give extended anticoagulation without stop date. If high risk of bleeding, then 3 months of anticoagulants (they do comment that d-dimer levels 1 months after stopping therapy might influence the decision to continue therapy)
  • Those with a second unprovoked VTE and low bleeding risk, extended therapy without stop date; high risk = 3 months. Re-evaluate periodically, such as annually
  • Those with DVT/PE and active cancer really should get anticoagulant therapy without stop date, but re-evaluate annually
  • In those with unprovoked proximal DVT/PE and the decision is to stop anticoagulants, you should give aspirin if no contraindication
  • In those with upper extremity DVT involving axillary or more proximal veins, anticoagulate
  • In terms of the choice of NOACs vs VKAs, they comment that the efficacy is similar though the risk of bleeding is less with NOACs (esp. intracranial), though in patients with atrial fibrillation, there is more GI bleeding with NOACs
  • Of note, of the 54 recommendations in the 30 statements in the article, “20 were strong and none were based on high quality evidence highlighting the need for further research” [my highlight]
  • Their estimates of risks of recurrent VTEs are:
    • VTE provoked by surgery: 3%  at 5 years
    • VTE provoked by non-surgical transient risk factor (e.g. estrogen therapy, pregnancy, leg injury, flight>8hrs,: 15% at 5 years
    • Unprovoked VTE: 30% at 5 years
    • Cancer-associated thrombosis: 15%/year
  • In terms of determining length of therapy (and leading to above recommendations):
    • An isolated distal DVT has 1/2 the risk of a proximal one for PE
    • A second unprovoked proximal DVT/PE increases the risk by an additional 50%
    • Categories for bleeding risk: [bleeding risk, per them, is similar to the HAS-BLED algorithm, including age (esp>75, though some increase >65), cancer, liver/kidney dz, diabetes, antiplatelet drugs, NSAIDs, alcohol, frequent falls]
      • Low (no bleeding factors): 0.8%/yr
      • Moderated (one bleeding factor): 1.6%/yr
      • High (2 or more bleeding factors): >6.5%/yr
    • ​A common issue is the patient with an unprovoked proximal DVT/PE without high risk of bleeding. They suggest that men have 75% increased risk of recurrence (the risk of recurrence in women is about 15% at 5 years) and those with positive D-dimer measured 1 month after stopping anticoagulation doubles the risk [I would add that there also was a study looking at D-dimer prior to stopping, finding that a raised level also predicted further VTEs. So I have been using both: D-dimer prior to stopping, and if normal, then 1 month after stopping. And I have been prescribing aspirin therapy in those who stop the anticoagulation, which has been further validated by 2 trials, with 35% reduced risk].
  • Outpatient treatment of acute PE: there are studies (not involving patients treated outside the hospital) finding that NOACs seem okay without initial heparin therapy

So, because of newer studies finding that NOACs seem to work well in patients with VTE, including in the setting of active cancer or recurrent VTE, and with the reported decrease in bleeding, the main change in the new guidelines over the previous ones is the elevation of NOACs over VKA’s. However, I would add that they do hedge on the quality of this data (none of these recommendations are Grade A) as noted above, and I am very concerned that many of those involved in the guidelines were involved in the NOAC studies with drug company support, and that there have been very real concerns about serious adverse events with NOACs and with drug company malfeasance/withholding damning data in their reporting (I am appending old blogs, some predating their posting on  BMJ, with drug company data itself suggesting that drug levels for dabagratan for example be monitored (one of the major selling points of NOACs is the lack of necessity to monitor levels, as INRs with VKAs).


1/12/12: Recent article which found an increase in MI/ACS (acute coronary syndromes) in meta-analysis of dabagatran use (see doi:10.1001/archinternmed.2011.1666). This was large meta-analysis (30,514 pts) on dabagatran in a variety of trials (2 on afib, 1 acute dvt prophylaxis, 1 ACS, 3 on short-term prophylaxis of DVT) comparing dabigatran to warfarin/enoxaparin or placebo, which found a small but significant increase in MI/ACS (30% relative risk increase, though only about 0.17% absolute increase from 0.79% to 1.19%, OR=1.33 (1.03-1.71, p=0.03). Not sure how to explain this since it works in preventing emboli in afib, though there was a paper reporting increased inflammatory markers with the use of dabagatran.

Brings up the all-to-frequent medical scenario. New drug comes to market. Works in preliminary studies (pretty much all funded by drug company). Drug makes sense to us (using a direct thrombin inhibitor should decrease thrombotic events, much easier to use than warfarin — no need for INR monitoring). BUT…. Subsequent studies start to find problems, especially when looking at other than the marker initially studied (thromboemboli in afib, preventing thrombosis in pts with joint replacement, acute DVT)


8/9/12: Recent editorial in annals on large number of adverse effects of dabigatran (see Annals Intern Med 2012: 157, 66). In its first 6 months, 505 adverse events from hemorrhage with 65 deaths, vs 176 for warfarin. In initial 6 weeks of marketing, “the Institute for Safe Medication Practices reported that dabigatran was responsible for more serious adverse events than 98.7% of all medications”. Reinforces importance not to jump on bandwagon and use new meds (esp given that vast majority of studies done by the pharmaceutical companies who stand to benefit….).  A recurrent theme.​


9/25/12: meta-analysis of 7 prospective randomized controlled trials see doi:10.1001/archinternmed.2011.1666

“For the period January 1, 2000, through December 31, 2011, the researchers identified 7 prospective randomized placebo-controlled clinical trials that met the study criteria, involving 31 286 patients. Based on the pooled results, the use of new-generation oral anticoagulant agents in patients receiving antiplatelet therapy after an ACS was associated with a dramatic increase in major bleeding events (odds ratio, 3.03; 95% CI, 2.20-4.16; P < .001). Significant but moderate reductions in the risk for stent thrombosis or composite ischemic events were observed, without a significant effect on overall mortality. For the net clinical benefit, treatment with new-generation oral anticoagulant agents provided no advantage over placebo (odds ratio, 0.98; 95% CI, 0.90-1.06; P = .57).”


7/28/14 blog (for some reason not in my BMJ blogs): the BMJ’s last issue had 6 articles on dabigatran (Pradaxa), showing the malfeasance of the drug company in promoting this agent [dabigatran is a direct thrombin inhibitor, specifically marketed as a competitor of warfarin for nonvalvular atrial fibrillation, though promoted for other indications as well, with the specific selling points based on a single key trial (RE-LY, see critique below) finding slightly lower incidence of major bleeding (16.4% vs 18.15%) without the costly and inconvenient monitoring of INRs required for warfarin – i.e., the high cost of the drug was more than justified by the savings from office visits and laboratory monitoring for warfarin. the following issues were discovered from the drug maker (Boehringer Ingelheim), largely through internal previously confidential documents:

  • The RE-LY trial was felt to have poor design and oversight, with a Canadian agency calling for “an independent audit of RE-LY to check for irregularities in conduct, sources of bias (this was an open-labeled, not blinded trial) and the cause of the unusually high incidence of intracranial hemorrhage in the warfarin arm” (thereby exaggerating the benefit of dabigatran), requesting patient level data to be made public. The FDA specifically was concerned about the likely high frequency of errors in the data set (misreporting of events), with subsequent FDA-mandated review by the company finding a further 3848 events. Of note, the increase in new MIs (n=270) found in the dabigatran​ group was dismissed as due to chance because of small numbers, but the increase in warfarin-associated intracranial hemorrhages was highlighted, even though there were many fewer of such events (n=155). the FDA found that the “blinded adjudicator of events” in fact was unblinded in 20% of the cases (e.g., by seeing identifying patient information)
  • ​The drug company has settled for $650M in litigation because the evidence provided by the company was incomplete.  The lawsuit was from fatal hemorrhage cases in the RE-LY trial, cases which it turns out were not included in the RE-LY trial’s initial compilation of adverse events, nor in the FDA-mandated recalculation, which it turns out was conducted by company scientists!!  Previously unreleased documents came to light through this litigation and freedom of information act
  • The drug company noted in internal documents that the risk of bleeding could have been reduced by 30-40% if plasma levels of dabigatranwere in fact monitored (company’s own unreleased data), and with appropriate dosage adjustment (i.e. undercutting their main selling point for the drug that there didn’t need to be drug level monitoring). Internal emails in the company released in the litigation showed that the company did not want this information released. Both the FDA and EMA (European medicines agency, the European FDA equivalent) specifically inquired about monitoring drug levels to improve safety, and one cardiologist, an FDA advisor, expressed concern that there was a 5-fold variability in plasma levels on the 150-mg dose. With regard to this marked plasma variability, one of the company’s own scientists commenting on the potential safety issues from not checking plasma levels was rebuffed by a company official noting in a 2012 internal email that “the publication [of this result] will [do] more harm than be useful for us, neither in the market but especially harmful in the discussions in the regulatory bodies”, and further, “the world is crying for this information — but the tricky part is that we have to tailor the message smart.”  And “this will make any defense of no monitoring [of plasma levels] to [health authorities] extremely difficult and undermine our efforts to compete with other [new oral anticoagulants].”
  • ​And, one of the concerns with bleeding with dabigatran​ is that there is no antidote, unlike vitamin K with warfarin overdoses [there now is a reversal agent available]


For 2 of the most relevant BMJ articles, see:  doi: 10.1136/bmj.g4517, and doi: 10.1136/bmj.g4670

  • The big issue here, as mentioned in several prior blogs including a few on dabigatran​ is that we are increasingly propelled into an era of privatization and deregulation, beginning with the Reagan presidency and leading to the situation now where almost all of the studies in major medical journals are funded by the drug companies (in the 1970’s the vast majority were publicly-funded), a large percentage of guidelines now are from specialty societies (as opposed to the recent past when the NIH led most of the guideline development) and many of the leading specialists are directly supported by drug companies, and where the cash-strapped FDA is increasingly making drug companies do further studies instead of independent groups. My own bias is that we should be very hesitant in the current climate to be early-adopters of new medications, unless there is a compelling need (for example, I think there is a compelling case for the new oral hep c drugs, or new and better-tolerated hiv meds)
  • Also, a recent report from 2015 which looked at Medicare claims from 2010-2011, looking at those with newly diagnosed atrial fibrillation who were put on dabigatan or warfarin within 60 days of the diagnosis (see JAMA Intern Med. 2015;175(1):18-24). They found that dabigatran (vs warfarin) was associated with
    • Higher risk of any bleeding event, with HR 1.30 (1.20-1.41)
    • Higher risk of major bleeding, with HR 1.58 (1.36-1.83)
    • Higher risk for GI bleeding, with HR 1.85 (1.64-2.07)
    • Though, lower risk of intracranial hemorrhage, with HR 0.32 (0.20-0.50)  [the incidence of intracranial bleeds was about 1/11th that of GI bleeding, though intracranial bleeds have more morbidity/mortality; no clear explanation of why this difference. Though see above about RE-LY trial and reporting issues]
    • And, the results of increased bleeding were especially in African-Americans and those with chronic kidney disease

See for more info

So, bottom line: I’m just not sure I trust the NOACs given all of the above. I have continued with warfarin, with no real issues/adverse events (with the one exception of a patient controlled on 1mg of warfarin, admitted to the hospital for chest pain, seen by cardiology, but patient had missed a dose of warfarin that day and her INR was a little low, and the cardiol fellow increased the warfarin to 2mg, noting it was “only 1 mg more”, and one week later she bled out from a GI hemorrhage. Now, I always check in with my warfarin patients when discharged from the hospital just to make sure things were not changed too much, and provide very close follow-up)

Primary Care Corner with Geoffrey Modest MD: PPIs and Chronic Kidney Disease

15 Jan, 16 | by EBM

By Dr. Geoffrey Modest

An article just came out looking at the relationship between PPI (proton-pump inhibitor) use and chronic kidney disease (CKD) (see doi:10.1001/jamainternmed.2015.7193).


  • 10,482 patients in the ARIC study (Atherosclerosis Risk In Communities, in 4 US communities) who had baseline GFR of >60 ml/min/1.73 m2 in 1996-9 were followed until 2011, mean 13.9 years
    • Mean age 63, 44% male, 80% white, 80% with education >=12th grade, mean eGFR 88, urinary albumin/creatinine ratio 4, 12% smokers, BMI 29, systolic BP 127, 50% hypertensive, 15% diabetic, 30% on NSAIDs, 15% on ACE inhibitors, 60% on aspirin
  • Replication study in the Geisinger Health System database with 248,751 patients followed mean of 6.2 years
    • Mean age 50, 43% male, 95% white, mean eGFR 95, 25% smokers, BMI 30, systolic BP 127, 33% hypertensive, 10% diabetic, 12% on NSAIDs, 30% on ACE inhibitors, 11% on aspirin
  • Assessed the occurrence of a diagnostic code for CKD in the ARIC study, and sustained GFR <60 in the Geisinger group, comparing PPI users, nonusers, and H2-blocker users


  • ARIC:
    • 56 incident CKD events among 322 baseline PPI users (14.2/1000 person-years) vs 1382 among 10,160 baseline nonusers (10.7/1000 person-years)
    • Unadjusted incidence of CKD in PPI users: HR 1.45 (1.11-1.90, p=0.006)
    • Adjusted for demographic (age, sex, race), socioeconomic (health insurance, education level) and clinical variables (baseline eGFR, urinary albumin/creatinine ratio, smoking, systolic BP, BMI, diabetes, cardiovasc disease, use of antihypertensives or anticoagulants): HR 1.50 (1.14-1.96, p=0.0013). They also considered annual household income, use of NSAIDs, aspirin, diuretics, statins, but these did not affect the adjusted HR results, so were not formally included.
    • Given that PPI use escalated dramatically after the baseline in years of 1996-9, they did an analysis of PPIs ever-used as a time-varying variable, with HR=1.35 (1.17-1.55, p<0.001)
    • In comparing PPI use vs H2-blocker use: HR 1.39 (1.01-1.91, p=0.05)  [Also, no association found between H2 blocker use vs non H2-blocker use and CKD]
    • In comparing PPI use to propensity-score matched non-users: HR 1.76 (1.13-2.74)
    • 10-year absolute risk of CKD among the 322 baseline PPI users was 11.8% vs 8.5% in nonusers
  • Geisinger:
    • 1921 incident CKD events among 16,900 baseline PPI users (20.1/1000 person-years), vs 28,226 events among 231,851 nonusers (18.3/1000 person-years)
    • Unadjusted incidence of CKD in PPI users: HR 1.20 (1.15-1.26, p<0.001)
    • For adjusted analysis HR 1.17 (1.12-1.23, p<0.001) (adjusted for age, sex, race, baseline eGFR, smoking, BMI, systolic BP, diabetes, history cardiovac disease, antihypertensive med use, anticoagulatnts, statins, aspirin and NSAIDs)
    • ​For time-varying ever-use model HR 1.22 (1.19-1.25, p<0.001)
    • Once-daily PPI use HR 1.15 (1.09-1.21, p<0.001)
    • Twice-daily PPI use HR 1.46 (1.28-1.67, p<0.001)
    • In comparing PPI use vs H2-blocker use: HR 1.29 (1.19-1.40, p<0.001=0.05) [again, no association between H2 blocker use vs non H2-blocker use and CKD)]
  • Also, the incidence of acute kidney injusry (AKI) was somewhat higher than CKD in both cohorts

So, a few points;

  • CKD is really common in the US (13.6% of adults, and increasing over time); not only is CKD associated with end-stage renal disease but also with increased risk of cardiovascular disease and death; there are clear relationships with many meds and CKD, an issue in the setting of increasing polypharmacy; PPIs are one of the most prescribed meds in the US (>15 million Americans had scripts in 2013) and are available OTC; they are increasingly prescribed to kids; and estimates are that 25-70% overall are not for appropriate indications, and that 25% of those on long-term PPIs could discontinue them without getting any symptoms.
  • This was a large observational study from 2 databases, with consistent results and even a dose-response relationship (at Geisinger, the more PPI taken, the more CKD). But, as an observational study, one cannot conclude that there is a causal relationship. Although mathematical attempts were made to control for many of the suspect variables (e.g., in the ARIC study, PPI users were more often white, obese and on antihypertensives), there still may be unknown or unaccounted variables (e.g., were those on twice-daily PPIs sicker in other ways which predispose them to CKD?, Does this modeling really apply to patients very under-represented in the cohort, such as non-whites?).
  • This study adds to the list of potential adverse effects associated with chronic PPIs: hip fracture, community-acqured pneumonia, c diff invections, acute interstitial nephritis, etc.
  • And, as mentioned in several prior blogs, the issue is that PPIs are often used as first-line therapy for gastritis or GERD (since they work so well, and not only make patients more reliably happier with their therapy but also give us a better diagnostic sense of what is going on), stepping-down therapy to an H2-blocker or antacid doesn’t happen often (much easier to continue the PPI and move on to dealing with the patient’s other concerns, easier to avoid a prolonged discussion and potentially ineffective move to the less powerful therapies…), and if the patient ever makes it to the ER or to a GI appointment, in my experience, they pretty much inevitably are given PPIs, often at maximal doses (which also makes it more difficult for the primary care provider to talk the patient into a less aggressive therapy). But, as mentioned in prior blogs and reinforced in the above study, although the short-term effectiveness of PPIs is pretty dramatic, they are really overused and the long-term sequelae may well be profound…

For other possible adverse events associated with PPI use, see

EBM blog homepage

Evidence-Based Medicine blog

Analysis and discussion of developments in Evidence-Based Medicine Visit site

Creative Comms logo

Latest from Evidence-Based Medicine

Latest from EBM