Well, I live and work in Malmö at a hospital, the only hospital in the city, which happens to have a registry of film in the radiology department going further back in time than any other hospital in the world. It’s an extraordinary resource for doing epidemiological studies on osteoporosis. I have also spent some time working at the Mayo Clinic, where they also have extremely good film data going back long enough that it could be used to establish trends over time.

  Has there been a significant trend in the incidence of osteoporosis?

Here in Sweden, we have seen an increase in the number, and maybe also in the incidence, of osteoporotic fractures up until the middle of the 1980s, and then it leveled off. That’s an age-independent increase, which means it’s not just due to the aging of the population. The United States has seen a similar increase, but it began leveling off in the 1960s and 1970s. There are other countries, however, where the trend is still going up from a low level.

  What prompted the research that led to your highly cited 1996 BMJ paper "Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures"?

I was working with a government agency in Sweden known as SBU, which is the Swedish Council on Technological Assessment and Health Care. We were doing evidence-based medicine, and so we gathered all the available evidence for bone mineral density to assess how well it predicted fractures. We realized that while we had all this data, we could also do a proper meta-analysis.

  Why do you think that paper has been so highly cited over the years?

Well, bone mineral density (BMD) had been highlighted by the specialists as very probably predictive in osteoporosis. This was the first article to gather all the data and discuss how BMD could be used in this context. We also made some comparisons and put the findings in the perspective of other diseases. For instance, we said the ability of BMD to predict fractures was similar to the ability of blood pressure to predict stroke and better than the ability of cholesterol measurements to predict myocardial infarction, although cholesterol is less expensive to measure.

  Have you gone back since then to see if later data changed the results?

We are doing a follow-up at the moment, but we probably won’t publish the analysis. The results seem to be very much the same so there’s no use in publishing a new meta-analysis. We will know for sure within a few months.

  What about your 1992 BMJ paper on the question of whether drugs that affect bone metabolism can prevent future hip fractures ("Evidence for efficacy of drugs affecting bone metabolism in preventing hip fracture," BMJ 305[6862]: 1124-8, 7 November 1992)—that’s been cited over 200 times? How did that research come about?

That came out of the Mediterranean Osteoporosis Study, known as MEDOS. It was a case-control study of 8,000 men and women in the Mediterranean that was started in the 1980s. We wrote several papers on risk factors for osteoporosis based on data from this study. One part of the questionnaire these patients filled out was devoted to how they had been treated over the years. So this allowed us to look from an epidemiological point of view at what happens if you treat individuals with different drugs. The reason this paper is cited so much is that there had been very few randomized controlled drug trials that used fractures as an end point. So this gave us some of the first meaningful information on the efficacy of different drugs. This study then led to a lot of large randomized controlled trials testing the use of drugs to prevent fractures.

  And the data from your study were significant?

Yes, for several of the drugs.

  Both of your highly cited papers were published in the British Medical Journal. Why did you choose BMJ?

We wanted the widest possible distribution and we wanted it to be read by general practitioners as well as researchers, and BMJ made that possible.

  How do you see the treatment of osteoporosis changing in the next five years?

I think the major change will be a shift from only using a BMD measurement as an estimate of risk to more sophisticated risk assessment. We will use absolute 10-year risk as is done, for instance, with cardiovascular disease. We will also have more data on male osteoporosis. We have started a large study in the U.S., Sweden, and China that should give us much more information on male osteoporosis. We will have new treatments, of course, but at the moment the main thing in the pipeline is improvement of the delivery system. We’re testing bisphosphonate, for example, as an injection once per year. But we won’t know if it works, of course, until the trials are finished. There’s also a lot of research being done in the genetics of osteoporosis.

  What would the risk factors for osteoporosis be?

We don’t know yet. There are several candidate risk factors, though: age, of course, and gender. Then there’s low body bone mineral density, previous history of fractures, heredity, and low body mass index or low body weight. Also secondary osteoporosis, which is osteoporosis that comes from the presence of other diseases, is a risk factor.

  What are the biggest challenges in doing research in osteoporosis?

The biggest challenge is to create studies large enough to give meaningful results that can be used reliably in clinical practice. You can’t just do small studies. You have to do large studies and large international studies to be able to transfer the results into clinical practice. This has been the challenge for establishing risk factors and for establishing the efficacies of interventions.

  What are you concentrating on now in your research?

My main goal now is to figure out how to get data that will allow us to do risk assessment and calculate absolute risk and how best to transfer that knowledge to clinical factors. We also want to know if interventions based on risk factors will work. My other interest is establishing when it is cost-effective to treat osteoporosis and how best to do it. All these factors are connected with each other; they are all linked, both in the research itself and in the eventual treatment.

Dr. Olof Johnell
University of Lund
Malmö, Sweden