I picture a younger me on this article because the long line of thiazide trials stretches back decades in time to when I looked just like the photograph. The large picture that heads this article depicts my spreadsheet of all of the trials I have reviewed, and this link will take you to that sheetIt is a nice place to go because it has links to each of the trials in case you want to take a look at them yourself.

Of course the reason I am writing this article is the new trial by professor Fuster and his colleaguesYou can see at a glance it is much larger than the rest, and it is negative – meaning those who took the drug did no better than those who did not. 


Of all the branches of biomedicine trials should seem most intuitive and simple to almost anyone who lives in modern times. Medical trials are not different from trials of refrigerators, or cars, or sunscreen potions. They are what they say, practical demonstrations concerning the extent to which something in the real world accomplishes those tasks we desire accomplished. For good reasons modern trials use fancy statistics, fancy meaning beyond common intuition and using techniques that someone like myself, who is not at all unfamiliar with statistics, finds baffling – or, better said, dense enough in meaning one might hope someone else does the work.

Their valuable enhancement of precision tends to screen the coarse realities of a trial behind the gossamer of a shimmering mathematic. In service to the assumptions of the mathematics things about the trial are arranged as in a ballet – done this way, not that way, from start to end. At the end the magicians step before their shining curtain and pronounce their verdict. We are suitably awestruck.

But behind the curtain is just folks who form stones, some of whom take an active medication and some a lookalike placebo. After a while, we can count them – so many on the drug made a new stone, so many on the placebo made a new stone. From that fleshy and highly medical labor of counting the mathematicians can conclude if the difference departs in a meaningful way from what might occur by chance alone. For chance is the devil in the room. Unlikely things do happen. A useless drug might just by chance’s whim be taken by people whose stone disease was in a temporary decline and we are mislead. It is in the estimate against chance that we need our mathematicians, their glorious equations and punctilious manner of proceeding. They can save us from false belief.

Because I know nothing about these high affairs, and because we have had a lot of trials for thiazide over decades, going back to when I looked more young and winsome, I have taken what could be called a dumb view of the topic. I just accumulate all the people tried on the drug or off, and how many made another stone (relapsed is the common medical term). As for chance I do only a simple counting statistic. To my accumulation I have added the new numbers. They have a powerful effect being so large as they are. This seemingly witless approach is not bad if there are a lot of trials. For, after all, what is uncommon is – well – uncommon, so trial after trial will more or less give us a crude idea of how the world works. Is that not not untrue?



Like all science, we start with the objective: What did the investigators seek?

I would have thought is was merely if thiazide reduced new stone production, but it was more: “Thus both the efficacy of thiazide in the prevention of the recurrence of kidney stones and the dose-response effect remain unclear.” So they wanted two things. For the one it was just the usual – drug or not, new stones or not. For the other it was to use a range of doses to look for a dose response. Two objectives.

Research Plan

What They Did

The drug: hydrochlorothiazide, the doses 12.5, 25, and 50 mg once daily vs. placebo. So, four groups of roughly equal size, three for the drug doses, one for the placebo. The measure: number of patients in each group who passed a stone OR had one removed OR showed a new stone on CT, OR showed the enlargement of a prior stone on CT. Any ‘OR’ yes = ‘Recurrent Stone’. The final score is number of recurrent stone formers thus defined in each of the 4 groups during the 3 year trial. My little table included precisely these four numbers.

The analysis is far more sophisticated. Mathematicians abounded and I will not scruple to question or even try to describe their particular form of magic. We have to allow for time of exposure to the drugs, how many stones each person had formed (an estimate of their propensity to stone forming), the two major ‘OR’s were analyzed separately (passage OR removal) and radiographic new stone OR growth of old stone(s).

My Reservations

Their null hypothesis (no effects of drug) was odd: “that there would be no relation (i.e., no significant linear trend) between the hydrochlorothiazide dose and the symptomatic or radiologic recurrence of kidney stones”. As a physician this seems overly stringent. After all no one cares about the linear regression, just whether the drug reduces stones.

I have another worry. Those of us who work with patients know the hydrochlorothiazide is relatively short acting so we use longer acting drugs. We also know that at such a low dose as 12.5 mg once a day the drug is unlikely to work – experiments are planned a bit from practical wisdom even if analyzed with the scrupulous eye of a mathematician. So the lowest dose had little a priori likelihood of doing any good.

I have a third worry. Stone passage or removal can be from old stones or new ones. The latter requires on read the CT and count to show that the stone that left the person was not one already in the kidney. They attempted to correct for this by analyzing separately patients with no stones at baseline, and those events that occurred within 6 or 12 months of entry into the trial – present stones might washout in those early intervals. But I am not so sure they need to do this – why not pass at any random time during the trial. Why not count the stones?

Finally, a forth. Stone growth differs from new stone appearance. The latter requires nucleation of crystals the former merely addition of atoms and molecules to a completed crystal/protein matrix ensemble. The saturation needed for nucleation exceeds that needed for growth. The drug, so far as we know, acts by lowering urine calcium and thence saturation, and thence stone forming propensity. The two are different so their merge is to add dissimilars.

I Am Not Ungracious

This could seem carping and negative, but not so.

Quite opposite.

I know some of these investigators and think so highly of them that I am offering my best thoughts. They have performed a very large trial in a sophisticated and rigorous manner, and deserve the best I one offer intellectually.


The drug did not reduce stone events (passage OR surgery) but the two higher doses did reduce the CT recurrence estimates of combined growth OR new stone; odds ratio vs placebo were 0.49 (CI=0.27 to 0.87) and 0.54 (CI=0.29 to 0.98) for 25 and 50 mg/d doses respectively. In a fancier ‘corrected’ model this effect disappeared.

Urine calcium was not reduced beyond chance by any of the drug doses (Table S15). So the main action of thiazide to prevent stones was not in evidence in the 24 hour urine samples. But although not significant as against chance values did fall by about 40-50 mg/d at the two higher drug doses and by about 25 mg/d in the lowest dose group.

As expected from the weak effect on urine calcium, hydrochlorothiazide did not lower urine saturation with respect to calcium oxalate or calcium phosphate. Even so, the highest dose did lower SS CaOx from 8.1 to 6.8 a change of -1.23 (CI=-2.49 to 0.02) nearly significant as against chance. For CaP SS the lowest dose was least ineffective, but none of the doses lowered CaP SS. Although stones were predoinantly calcium oxalate, initial CaP phases may play in important role in nucleation and stone disease.

What Does it Mean?

We Need More Trials of this Fine Quality

Certainly this study is the first really large and scrupulous trial for thiazide in stone formers, and the only proper response is for the stone research community to get itself together and do more trials with more promising agents like chlorthalidone or indapamide. Both seemed effective in trials that had many inherent defects and I suspect fine trials with them will be stunningly positive.

I would hope future trials jettison dose response ranging and use their available power to compare placebo to a common clinical dose of 25 or 50 mg (I favor the former) chlorthalidone or the equivalent for indapamide. I would also be more conservative concerning clinical recurrence. Passage or removal of a stone is not recurrence, and stone counting by CT is very difficult so perhaps one would want to focus on patients with low stone burdens.

Hydrochlorothiazide is Not A Good Drug Against Stones

I have shunned hydrochlorothiazide for stone prevention life long, and will certainly continue to do so. It is short acting and just as much a nuisance as longer acting drugs with regard to potassium loss and changes in glucose tolerance. I never saw any good in it.

I Will Not Change My Practice

I will continue to use long acting thiazide drugs to prevent calcium stones and use them in the way that seems most effective. The weight of all trials combined makes me doubt profoundly that thiazide does not reduce stones depsite the evident weaknesses of number and design that abound among them. It is too precious to contend that although the summary of all cases ever tried points to an effect the drug is nevertheless without an effect. There is just too much imbalance on the side of stone reduction. I have added the present study numbers but only for the 25 and 50 mg doses as the lowest dose had little chance of working. The sum total of all the data remain biased in favor of a thiazide effect.

I do not use thiazide by itself but only with diet management. I have always advocated for diet first, meaning lower diet sodium and sugar as a way of lowering urine calcium, high fluids, and adequate diet calcium to reduce urine oxalate and for bone health. To me, thiazide is an addition to diet. In that context this trial is no reason to stop using long acting thiazide like drugs.

But withal, should trials of this quality discredit my thinking and prove the longer acting drugs ineffective I will drop them on the moment. Likewise for diet in that we have only one find diet trial after all these years of NIH funding for stone research.

As I was long a beneficiary of such NIH funding some might ask why I did nothing to advance the cause of stone trials. I would answer that I have no taste for such work and a poor kind of mind as well. Trials are for people with certain talents I lack utterly, being too caught up in imagination’s whorles. Too unworldly, impractical, and easily confused.

Scroll to Top