Atul Gawande is a Harvard-based surgeon and staff writer for The New Yorker. He's made big news with an exposé of healthcare spending in a poor county on the US-Mexican border. McAllen, Texas is the poorest county in the US, with a per capita income of $12,000. Yet Medicare - the government program that pays most of the health bills for Americans 65 and over - costs $15,000 per senior per year, exceeded nationally only by Miami. McAllen has high rates of obesity and other dodgy health indicators. But out in the west Texas town of El Paso, 800 miles up the Rio Grande, the demographics and health status are similar, but Medicare costs are half.

Gawande pulls no punches: the higher costs are due to physicians and for-profit clinics exploiting every conceivable opportunity to maximize their incomes. He demolishes the claim that McAllen's cost and utilization patterns can be explained by "defensive medicine" (fear of malpractice) or superior care. Texas has capped malpractice awards at $250,000 and the doctors in McAllen acknowledge that lawsuits and premiums are way down. On 25 quality indicators, El Paso beats McAllen on 23.

As Gawande points out, clusters of doctors think and behave differently, producing major variations in practice patterns and costs. Just as greed has turned McAllen medicine into a testing-and-patient-recall factory, Grand Junction, Colorado and the Mayo Clinic in Minnesota have created a culture of quality and restraint that produces first class results for much less money.

But that's the US, and we like to believe that the sins of American healthcare stop at the 49th parallel. At the level of provinces or regions, that's probably true: governments allocate health care budgets and the politics of distribution tend to pre-empt two or threefold variation in total spending between Calgarians and Winnipeggers. The smugness should end here. Our McAllens come in smaller packages. We just need stronger lenses to see them.

Canadian researchers have long documented variations in the use of individual services, such as C-sections and hip replacements. Three decades ago, my home town of Moose Jaw, SK was observed to experience 8 times the hysterectomy rate of Regina, 45 minutes down the road. (There were rumours that the local gynaecologists were about to start rounding up men.) Practice atlases have reported three- and four-fold variations in surgical rates among Ontario counties. But even these provocative data may obscure even larger variations attributable to individual physicians or practice groups. If a county's propensity to invade your prostate is three times the provincial average, it is possible - even probable - that an individual urologist is ten times more likely to intervene.

To find our little corners of Texas, we have to analyse overall healthcare use and spending among thousands or tens of thousands of people, not the hundreds of thousands or millions typical of US analysis. US Medicare doesn't allocate funds on population-based formulas or seriously attempt to even out per capita spending. Governments in Canada do more than just process bills; they allocate funding, more or less on an age-sex adjusted population basis in most regionalized provinces and by organization elsewhere. Interprovincial variations in total spending are relatively small, and at the level of decent-sized cities, it is highly unlikely that there are Texas-sized gulfs.

Drill down another level or two and the real story emerges. Physicians, especially specialists, tend to cluster in large cities. There are gentrified enclaves of urban Canada that get their primary care from paediatricians, obstetricians, and internists. Some GPs will have triple the referral rates to specialists than others. Walk in one clinic with heartburn and you'll start with Tums; walk in another and you'll be slated for an endoscopy. Some doctors will do annual Pap tests on low-risk women while others will follow the three-year guideline. Some will tell sedentary and overweight patients to eat better and exercise; others will order periodic stress tests, prescribe a statin, and call in the cardiologist.

Lots of policy-makers, managers and practitioners agree that there is a lot of waste in healthcare. The problem is knowing exactly where and why it occurs. If we want to trim the fat from healthcare spending, we have to learn exactly where the avoidable spending takes place. It's not all physician-driven, but doctors are the gatekeepers to a huge swath of health care spending, and the cascade of costs that originates in the GP office accelerates through laboratories, diagnostic imaging centres, pharmacies, specialists, and hospitals.

A well-known concept in health economics is supply-induced demand. Patients tend to trust their doctors, submit to batteries of tests, and reappear when summoned to a follow-up appointment. Parts of large urban centres with high concentrations of specialists and abundant technology are potentially mini-McAllens. But so are parts of small-town Canada where a GP with a local monopoly and few peers in the vicinity can indulge in wildly skewed practices.

Gawande went to McAllen because readily available data pointed him there. For all the injustices in its healthcare, at least the US produces the evidence. President Obama has read Gawande's piece and has begun to use it as an example of why an overhaul of the system is essential. McAllen and El Paso may be this President's Harry and Louise.

Canada needs to map our McAllens and Mayos routinely and develop strategies for reducing the spread. Disaggregation is key since clinical autonomy is deeply embedded in Canadian health care culture and the one certainty is that groups with similar needs are treated differently. We don't know because we don't look hard enough, often enough. As long as we remain in the dark, not much will get better.