• What studies should I write about?

    From time to time I solicit suggestions on what to write about. Usually I solicit topics, but that often doesn’t generate very many ideas I can work with. The reason is that my writing isn’t driven by topics so much as studies.

    Of course, each piece is on a topic. But my entry is through the research. I start with a study that sparks an idea I think will resonate with people. That leads to other studies, and a post is born.

    With that, what studies do you think I should write about? Please send links or full enough citations so I can find them. You can reach me on Twitter, by email, or in the comments below, which are open for one week.


    Comments closed
  • Will we ever see Medicare Advantage encounter data?

    Charles Ornstein reports on the latest chapter in research access to Medicare Advantage encounter data:

    The government has collected [Medicare Advantage] data on patients’ diagnoses and the services they receive since 2012 and began using it last year to help calculate payments to private insurers, which run the Medicare Advantage plans. But it has never made that data public.

    Officials at the Centers for Medicare and Medicaid Services have been validating the accuracy of the data and, in recent months, were preparing to release it to researchers. Medicare already shares data on the 38 million patients in the traditional Medicare program, which the government runs. […]

    The grand unveiling of the new data was scheduled to take place at the annual research meeting of AcademyHealth, a festival of health wonkery, which just concluded in New Orleans.

    But at the last minute, the session was canceled.

    Go read the whole thing. I am quoted.


    Comments closed
  • Medicaid, a “broken program that harms its beneficiaries”

    I asked on Twitter for policymakers’ expressions of harm caused by Medicaid.

    By email, Andrew Goodman-Bacon came through in a huge way. The bullets below are a lightly edited version of what he sent me, shared with his permission. (For the record, Medicaid does not cause harm. More about that soon.)

    • The Sommers/Epstein paper surveyed governors and found that five of those who opposed expansion felt that Medicaid was a “broken program that harms its beneficiaries.”
    • Senator Ted Cruz has said that Medicaid hurts health care access
    • In one of Tom Price’s recent testimonies he said ,“Medicaid is a program that has, by and large, decreased people’s ability to access care.”
    • Speaker Paul Ryan comes close on pg 24 of “A Better Way
    • A Healthy Indiana report (notably produced by the Pence administration) cites Roy and LaPar, but doesn’t go all in on the “harms” claim
    • Here is the American Action Forum saying “harm”
    • See also, this brief from a policy shop in MI, this brief from a policy shop in NC, and this brief from a policy shop in PA
    • Here is ALEC citing that study, although not going so far as to say patients will be “harmed”

    To these, I will add this quote of Representative Bill Cassidy (via Aaron) and this op-ed by Seema Verma (via Adrianna). Note: I have not looked through everything in the above list. If you find errors or have more contributions, let me know.


    Comments closed
  • New collaboration for VA drug pricing

    Great news:

    The Institute for Clinical and Economic Review (ICER) has agreed to work closely with the Department of Veterans Affairs (VA) Pharmacy Benefits Management Services office (PBM) to support its use of ICER drug assessment reports in drug coverage and price negotiations with the pharmaceutical industry.

    More here. I am employed by the VA and serve on one of ICER’s evidence review panels.


    Comments closed
  • All the plan-provider vertical integration literature to date

    There aren’t a lot of studies and reports on the integration of health insurance plans with health care providers (meaning they’re owned by the same organization). Still, I want to keep track, so here’s what I’m aware of (I’m a coauthor on the first two):

    Peer-Reviewed Publications

    • Frakt, A.B., Pizer, S.D. and Feldman, R., 2013. Plan–Provider Integration, Premiums, and Quality in the Medicare Advantage Market. Health Services Research, 48(6pt1), pp.1996-2013. I wrote about this paper here and here.
    • Johnson, G., Lyon, Z.M. and Frakt, A., 2017. Provider-Offered Medicare Advantage Plans: Recent Growth And Care Quality. Health Affairs, 36(3), pp.539-547. I wrote about this paper here.
    • La Forgia, A., Maeda, J.L.K. and Banthin, J.S., 2017. Are Integrated Plan Providers Associated With Lower Premiums on the Health Insurance Marketplaces? Medical Care Research and Review.
    • Burns, L.R., McCullough, J.S., Wholey, D.R., Kruse, G., Kralovec, P. and Muller, R., 2015. Is the system really the solution? Operating costs in hospital systems. Medical Care Research and Review, 72(3), pp.247-272.


    • Khanna, G., Smith, E. and Sutaria, S., 2015. Provider-Led Health Plans: The Next Frontier—or the 1990s All Over Again. McKinsey & Company Healthcare Systems and Services Practice.
    • Khanna, G., Narula, D. and Rao, N., 2016. The market evolution of provider-led health plans. McKinsey & Company Healthcare Systems and Services Practice.
    • Carpenter, E., 2016. Nearly 60 Percent of New Medicare Advantage Plans Are Sponsored by Healthcare Providers. Avalere.
    • Blumberg, L.J., Holahan, J. and Wengle, E., 2015. Marketplace Price Competition in 2014 and 2015. Urban Institute.
    • Pascaris, M. and Smith, K., 2015. Entrance of US not-for-profit hospitals into health insurance will continue to rise. Moody’s.
    • Goldsmith, J., Burns, L.R., Sen, A. and Goldsmith, T., 2015. Integrated delivery networks: In search of benefits and market effects. National Academy of Social Insurance.
    • McKinsey Center for U.S. Health System Reform. (2016). Hospital networks: Perspective from three years of exchanges.
    • Baumgarten, A., 2017. Analysis of Integrated Delivery Systems and New Provider- Sponsored Health Plans. Robert Woods Johnson Foundation.


    Comments closed
  • Get Well Sooner? A Healthier Roommate Could Help

    The following originally appeared on The Upshot (copyright 2017, The New York Times Company).

    Your speed of recovery in a hospital is related to many things. Among them is one you might not consider: the condition of your roommate.

    A recent study published in the American Journal of Health Economics found that hospital patients who are assigned healthier roommates require less care and are discharged more rapidly, with no negative effects on their health. For example, a patient who rooms with the healthiest roommate has a hospital stay that is about eight hours shorter, requiring 27 percent less medical attention, and costing about $840 less, compared with a patient with the sickest roommate. Female patients with healthier roommates are discharged in better condition and have a smaller chance of requiring re-hospitalization.

    The study examined a broad range of hospital patients, including those who had surgical procedures — like heart bypasses or joint replacement operations — as well as those admitted for medical conditions like pneumonia or cancer.

    At first glance, there is an apparently obvious explanation for these findings: Patients are typically assigned to room with other patients of similar condition. In particular, healthier people are assigned to rooms farther from the nursing station. Therefore, healthier patients — those who require less care and are discharged more rapidly — also tend to room with relatively healthier patients. So there seems to be a noncausal connection between the recovery speed of patients and the health of their roommates.

    But the author of the study, Olga Yakusheva, a University of Michigan economist, controlled for the factors that nurses at the Connecticut hospital she studied use to assign patients to rooms, including diagnosis and specific room assignment. She found that even in a particular room at the hospital and even among patients with a specific diagnosis, those who ended up with healthier roommates fared better. (The study did not include patients who had single rooms.)

    What’s more, “placing a sick and a healthy patient in one room benefited the sicker patient without ill effects for the healthier roommate,” Ms. Yakusheva said.

    Had that particular hospital taken fuller advantage of this phenomenon in patient room assignments, it could have reduced total inpatient days by 900 per year, saving about $1 million, for the sample of patients the study examined.

    In recent years, most American hospitals have gone another way, though. They’ve added private rooms and renovated shared ones to accommodate only one person. The amount of hospital room space per patient has doubled since the late 1980s. Naturally, that increases costs.

    Many patients prefer the privacy of a single room. And some studies indicate that single rooms reduce the spread of flu and other infections, though the evidence is not conclusive. Despite the potential risks and preferences, having a hospital roommate, and a healthier one in particular, may be better than having no roommate.

    There are several hypotheses for how roommate assignments affect patients’ health. A healthier roommate — particularly one with a similar condition — may be better able to transfer important self-care knowledge or even lend a helping hand, as a few studies have documented. Or, patients with healthier roommates may feel better when they observe other patients doing well, relative to those who observe patients doing poorly. Some studies have found that patient interactions can reduce anxiety.

    Other possibilities are indirect. If your roommate is healthier, she may draw on fewer nursing resources (time and attention), leaving more for you. Or, you may be better able to rest because nurses and doctors are entering the room less frequently when your roommate needs less care. Additional analysis by Ms. Yakusheva doesn’t support these indirect explanations, however.

    The phenomenon is just one of many “peer effects” — the tendency for certain behaviors to spread through social interaction — that have been identified and studied by researchers. It’s probably no surprise that the nature of your social engagement with friends, family and colleagues influences your degree of cooperation and happiness, as studies have found. Research also suggests that it affects behaviors more closely tied to health. For example, obesity tends to spread in social networks. So does smoking behavior. Alcohol consumption follows similar patterns.

    Even if it’s plausible that healthier roommates improve hospital patients’ outcomes, we should acknowledge some limitations of the research in this area. There are very few studies of the subject beyond Ms. Yakusheva’s. Hers is a study of one hospital, with a sample that may not be representative of all hospital patients. Also, as with all observational studies (as opposed to randomized trials), there may be other important influences that could not be accounted for that affect the results.

    If there is a positive efect, of what use are the findings? If you’re hospitalized, could you increase your chances of being assigned to a room with a healthier roommate? The answer, according to Bradley Flansbaum, a hospitalist with the Geisinger Health System, is yes.

    “You could always ask the nurse in charge for a room change,” he said. “If asked why, and you say it’s for personal reasons, the nurse will probably accommodate.”

    But this might put too much onus on the patient, and there is no guarantee the new roommate will be any healthier. A better approach might be for hospital staff to systematically take the phenomenon into account when assigning patients to rooms.



    Comments closed
  • Maybe I was wrong about ACOs

    It was at the 2009 AcademyHealth Annual Research Meeting in Chicago that I learned about accountable care organizations (ACOs). The day after leaving the meeting, I was already concerned:

    [I]f the U.S. health care system moves toward an ACO model we will see greater provider consolidation, all other things being equal. Provider groups’ greater market power to negotiate higher payments will be a countervailing force against the promise of the ACO model to provide higher quality care at lower cost.

    I have not been alone in harboring this reservation about ACOs. For years, I and others have viewed ACOs as one of several legal and regulatory motivations for provider consolidation. The logic goes like this: the ACO model encourages the provision of more coordinated care for a population of patients. In addition, it rewards provider organizations for higher quality and lower costs. Historically, all of these are have been justifications for provider consolidation, including the horizontal integration of hospitals, the horizontal integration of physician practices, and the vertical integration of hospitals and physician groups.

    But, a recent paper by Hannah Neprash, Michael Chernew and J. Michael McWilliams suggests this logic may be flawed. By comparing changes in provider consolidation from 2008–2010 to 2011–2013 between markets with higher versus lower Medicare ACO penetration (as measured in 2014), they found no evidence of a change in the trend toward consolidation associated with ACO adoption.

    The chart, just below, summarizes the main findings. Though vertical integration of physicians with hospitals (left panel) and physician group size (right panel) both increased over the study period, the rate of growth for both was identical across quartiles of 2014 ACO penetration. Here, ACO penetration is measured at the metropolitan statistical level as the proportion of Medicare beneficiaries attributed to an ACO.

    Many other measures assessed by the authors — changes in physician market concentration, hospital market
    concentration, and commercial health care prices — are consistent with the hypothesis that the Medicare ACO model did not encourage consolidation. However, mean physician group size grew more rapidly for physicians participating in an ACO, relative to physicians who did not.

    From this evidence, it seems I was (and others were) wrong to blame ACOs for consolidation of health care providers … maybe.

    But, the authors also found an increase in hospital mergers after the Affordable Care Act became law alongside an inverse relationship between hospital market concentration and ACO penetration. This leads to a different hypothesis about how ACOs might influence market concentration after all:

    These findings suggest that new payment models may have triggered some consolidation as a defensive reaction to the threat these models could pose, rather than as a way to achieve efficiencies in response to the new incentives. Hospitals and specialists in particular might consolidate both horizontally and vertically to achieve sufficient market share to resist payer pressure to enter risk contracts or weaken ACOs’ ability to exploit competition in hospital and specialty markets, to compel reductions in prices and service volume.

    It takes a close read to understand what the authors are saying here. It’s not necessarily about Medicare ACOs, but about ACOs more generally (including commercial market ACOs). They’re saying that providers might consolidate to resist this broader ACO movement and/or its implications for pressure on prices and volume. However, it’s a reasonable hypothesis that commercial market ACO presence is correlated with Medicare ACO presence across markets. Therefore, observations of relationships between consolidation and Medicare ACO penetration may also be valid for informing inferences about broader relationships involving commercial market ACOs as well.

    The upshot is that maybe the ACO model is influencing consolidation, just not in the way I and others had thought. Rather than consolidating to support the ACO model, providers may be doing so as a defense against it. It’s too soon to know for sure. More work needs to be done to gather evidence to support or refute this alternative hypothesis.


    Comments closed
  • Teaching Hospitals Cost More, but Could Save Your Life

    The following originally appeared on The Upshot (copyright 2017, The New York Times Company).

    Perhaps not evident to many patients, there are two kinds of hospitals — teaching and nonteaching — and a raging debate about which is better. Teaching hospitals, affiliated with medical schools, are the training grounds for the next generation of physicians. They cost more. The debate is over whether their increased cost is accompanied by better patient outcomes.

    Teaching hospitals cost taxpayers more in part because Medicare pays them more, to compensate them for their educational mission. They also tend to command higher prices in the commercial market because the medical-school affiliation enhances their brand. Their higher prices could even cost patients more, if they are paying out of pocket.

    To save money, insurers have started establishing hospital networks, and policy makers are considering ways to steer patients away from teaching hospitals. Those efforts may well save patients and taxpayers money. But how will that affect the quality of care?

    One answer is provided in a new study of over 21 million hospital visits paid for by Medicare in 2012 and 2013. Teaching hospitals save lives. For every 83 elderly patients seen by a major teaching hospital, one more is alive 30 days after discharge than if those patients had been admitted to a nonteaching hospital. This is a large mortality effect.

    “It’s about half the size of a breakthrough medical therapy like stenting for heart-attack patients,” said Amitabh Chandra, an economist with the Harvard Kennedy School and a longtime skeptic of the value of teaching hospitals, who wasn’t involved in this study.

    “Minor” teaching hospitals — which also have educational missions but are not members of the Council of Teaching Hospitals and Health Systems — also outperformed nonteaching hospitals, but by a smaller margin.

    The study, published in the Journal of the American Medical Association, adjusted for other factors that could have skewed the results, like demographics, patients’ diagnoses, hospital size and profit status. Because mortality rates differ geographically, it compared teaching with nonteaching hospitals within the same state. Even after such adjustments, it found mortality rates are lower at teaching hospitals for 11 of 15 common medical conditions and five of six major surgical conditions. The more doctors in training per bed a hospital had, the lower its mortality rate.

    Given the importance of this issue, you’d think we would already know the mortality differences between teaching and nonteaching hospitals. But the seminal studies on the subject are based on data at least two decades old. Other, more recent studies focus on only a few types of patients or offer conflicting results.

    “We thought the comparative performance of teaching and nonteaching hospitals was worth a fresh look because medicine has changed considerably since those older studies,” said Laura Burke, the lead author on the study and an emergency physician with the Harvard T.H. Chan School of Public Health. “And the more recent studies don’t settle the question.” (I am a co-author on the study, along with Dr. Burke and other Harvard colleagues Dhruv Khullar, E. John Orav and Ashish Jha. Dr. Khullar is also an Upshot contributor.) The study was funded by the American Association of Medical Colleges, which had no editorial control over analysis or publication.

    Though the study revealed mortality differences by teaching status, it could not illuminate the cause of those differences. Perhaps teaching hospitals attract higher-quality practitioners, more closely follow best practices, or use medical technology more effectively.

    Other studies suggest teaching hospitals do not offer higher quality more broadly. For example, an analysis led by Jose Figueroa, a physician with the Harvard T.H. Chan School of Public Health, found that teaching hospitals were more likely to be penalized by Medicare for low quality compared with nonteaching hospitals. Another study found teaching hospitals were more likely to be penalized for higher hospital readmission rates.

    An examination of Massachusetts hospitals found comparable quality performance at teaching and nonteaching hospitals. The state has a goal — codified in a 2012 state law — of bringing health care spending growth in line with overall economic growth. The Massachusetts Health Policy Commission has highlighted the high costs of teaching hospitals as part of this effort.

    The new study did not assess the cost of the benefits in mortality that teaching hospitals deliver.

    “The typical teaching hospital is at least 30 percent more expensive,” Mr. Chandra said. “Is 1 percent fewer deaths worth that price?” It’s a question few like to ask, but spending more on hospital care means less for other things we value — and that are known to improve health and welfare, too — like education and nutrition programs.

    About 26 percent of hospitals are teaching hospitals, accounting for just over half of all admissions. Unsure which hospitals in your area are teaching hospitals? It’s something most of them make a point of mentioning, so you can often find a hospital’s teaching status on its website. If not, an inquiry to the hospital should settle the matter. If you use one, the cost of your care will be higher, but it might save your life.


    Comments closed
  • Market power matters

    It’s the clash of titans.

    In January the Massachusetts the Group Insurance Commission (GIC) — the state agency that provides health insurance to nearly a half-million public employees, retirees, and their families — voted to cap provider payments at 160% of Medicare rates. Ignoring Medicare (~1M enrollees) and Medicaid (~1.6M enrollees), the GIC is the largest insurance group in the state.  According to reporting from The Boston Globe, the cap would be binding on a small number of concentrated providers, including Partners HealthCare, one of the largest hospital systems in the state.

    David Anderson summed the development up perfectly.

    The core of the fight is a big payer (the state employee plan) wants to use its market power to get a better rate from a set of powerfully concentrated providers who have used their market power to get very high rates historically.

    Anderson also pointed to a relevant, recent study that illustrates how a specific payer’s and provider’s market power jointly affect prices. In Health Affairs, Eric Roberts, Michael Chernew, and J. Michael McWilliams studied the phenomenon directly, which has rarely been done. Most prior work aggregate market power or prices across providers or payers in markets.

    Their source of price data was FAIR Health, which includes claims data from about 60 insurers across all states and D.C. In a county-level analysis, the authors crunched 2014 data for just ten of those insurers that offered PPO and POS plans and that did not have solely capitated contracts. These ten insurers represent 15% of commercial market enrollment. They then looked at prices paid by these insurers to providers in independent office settings for evaluation and management CPT codes 99213, 99214, and 99215. These span moderate length visits to longer visits for more complex patients and collectively represent 21% of FAIR Health captured claims.

    They computed insurer market share based on within-county enrollment. They computed a provider group’s market share as the county proportion of provider taxpayer identification numbers (TIN) associated with that group’s National Provider Identifier (NPI) — basically the size of group in terms of number of physicians.

    Some of the findings are illustrated in the charts below and are largely consistent with expectations. For all three CPT codes, insurers with greater market shares tend to pay lower prices. That’s shown just below. The biggest price drop occurs when moving from <5% to 5-15% market share. Greater market share than that is associated with still lower prices, but not by as much. For example, insurers with <5% market share pay an average of $86 for CPT code 99213; insurers with 5-15% market share pay 18% less and insurers with ≥15% just a few percent less than that. It’s roughly the same story for other CPT codes.

    That tells the expected story of payer market power, but what about providers? The authors also look at that, and one of their sets of results is illustrated in the chart below, for CPT code 99213. Again, greater insurer market power is associated with lower prices — you can see that within each of the three sets of colored bars.

    Each of those sets is for prices is for a different level of provider market share, 0 to 5%, ≥5 to 15%, and ≥15%. As one would expect, greater provider market share is associated with higher prices. Among providers with 0-5% market shares, the average CPT 99213 price is $88 for insurers with market shares of <5%, $72 for insurers with ≥5 to <15% market shares and $70 for insurers with market shares ≥15%. Among larger provider groups, with market shares ≥15%, price levels are higher: the average price for insurers with market shares of <5% is $97, for insurers with ≥5 to <15% market shares, it’s $86, and for insurers with market shares ≥15% percent it’s $76.

    The results are similar for the other two CPT codes: higher prices are independently associated with greater provider market share and lower insurer market share.

    From the research literature, we’ve long known that provider consolidation leads to higher prices. One of the arguments for insurer consolidation is to push prices downward. However, that doesn’t imply that the right policy approach to high health care prices is greater insurer market power, as there’s no guarantee that dominant insures will pass lower prices onto consumers. Minimum medical loss ratio regulations that would limit insurer relative profit margins don’t necessarily help because they could encourage a dominant insurer to keep price levels higher to extract higher absolute profits.

    However, other approaches that push prices downward can help consumers as well. One example is reference pricing. Other approaches include price transparencyhigher consumer cost sharing, and narrowing of networks. (Not all of these have been shown to consistently achieve their intended results.) These are not mutually exclusive approaches. The GIC, in addition to flexing its market power to push prices downward, has also offered narrow network plans that have saved enrollees substantially.

    For lowering health care prices, there are a lot of tools in the toolbox. But history suggests we haven’t settled on ones that both work and that patients and other stakeholders are willing to tolerate for long. Continued experimentation and analysis is essential.


    Comments closed
  • Allergies are even worse than you think

    The following originally appeared on The Upshot (copyright 2017, The New York Times Company)

    This is the time of year my kids and I have seasonal allergic rhinitis, better known as hay fever. I’d always thought it was merely a nuisance, but it turns out it also degrades cognitive performance, at least a little.

    Hay fever affects at least 10 percent of the population, and a higher percentage of children. The most obvious signs of allergic response include sneezing, itching and a runny nose. These can disrupt sleep, leading to fatigue, and the allergy can cause neurocognitive deficits we may not notice in ourselves or in our children. Medications used to treat the allergy can also induce sleepiness in some people.

    In the United States, school-age children collectively lose about two million school days because of pollen allergies. Even when they attend school, allergy-suffering students may perform a bit worse than their nonallergic counterparts.

    Using data from Norway, a recent study shows that when pollen levels rise, students’ test scores fall. The study used data from nearly 70,000 high school exit exams, which Norwegian students must pass to graduate and are used for higher education placement. Students take exams at different locations, and each student takes several at different times of year, providing multiple data points per student.

    The study’s author, Simon Bensnes, a Ph.D. candidate in the Department of Economics at the Norwegian University of Science and Technology, combined these with pollen count data linked to the location and time at which each student took each exam, as well as other demographic and air-quality data used to control for potentially confounding factors.

    Pollen counts are measured in grains of pollen per cubic meter of air and can be as high as the 100s at the height of pollen season in Norway. For students allergic to pollen, Mr. Bensnes found that a pollen count increase of 37 — large enough to cause symptoms in highly allergic people — is associated with a drop of about one-tenth of a point in exam scores. The scores range from one (worst performance) to six (best performance).

    Does such a seemingly small effect matter in the long run? Other results suggest they do. The study also finds that higher pollen counts correlate with a slightly lower likelihood of enrolling in a university and a lower probability of going into a STEM field. However, though the statistical methods to analyze test scores are rigorous enough to reasonably infer they’re causal, the ones for these longer-term results are less so.

    Still, Mr. Bensnes said, “it would be surprising if there were no effects in the longer run.” This is particularly likely in countries where exams are weighed more heavily than in Norway toward entrance to institutions of higher education. There, exams count only for about 15 percent of entrance determinations.

    Norway is not the only setting where a pollen-exam relationship has been found. In Britain, students take an exit exam at the end of secondary education in the spring or summer, when pollen counts are high. They also take a practice test the prior winter. Researchers found that compared with those with no allergy symptoms, British students who report allergies or take allergy medications during their secondary education exit exams are 40 percent to 70 percent more likely to score a full grade lower than they did on their practice test.

    A study in the United States found that a doubling of the pollen count is associated with about 1 or 2 percent drop in the proportion of third graders passing English and math achievement tests.

    Clinical studies have examined the cognitive effects of hay fever more directly. For example, a study found that people with hay fever experienced slower speeds of mental processing during ragweed season than at other times of year. Another study exposed allergic people to pollen in a controlled setting. It found that they exhibited slower mental function, decreased memory, and poorer reasoning and computation abilities compared with nonallergic test subjects.

    More generally, what we breathe affects how well we perform at school or work. Several studies found a link between air pollution and school absences, as well as labor supply and worker productivity. Worse air quality can cause or worsen respiratory problems, like asthma, reducing some children’s ability to attend school and adults’ entry into the work force. It can also harm job performance.

    One study found that higher concentrations of certain air pollutants hurt test scores of Israeli students and the chances of passing a high school exam necessary for higher education.

    Individually, we may not be able to do much about air pollution, but we can try to reduce the impact of pollen on school and work performance. Finding allergy medication that doesn’t induce drowsiness is an obvious approach. When it comes to high-stakes exams, it may be worth choosing test dates outside the allergy season, if possible. Hay fever is rarely debilitating, but its small effects can put us off our best game.


    Comments closed