Harm Scenario

You see a 73 year old man in your general medicine clinic who has a history of hypertension for which he has been taking nifedipine XL 30 mg PO per day for 5 years. In his retirement he has learned how to use the Internet, and he brings you a print-out of some newspaper headlines that he unearthed while surfing the net:

• LA Times – HYPERTENSION MED KILLS
• Washington Post – DRUG FOR BLOOD PRESSURE LINKED TO HEART ATTACKS: RESEARCHERS FEAR 6 MILLION ARE IMPERILLED

He is very worried and wants to know if he should stop taking nifedipine. Together, you form the clinical question, “In patients taking calcium antagonists for hypertension, do these calcium antagonists cause cancer?” You track down the Lancet article referred to in the scary newspaper article
Lancet 1996;348:493-7.

• Are the results of this harm study valid
• Are the results of this harm study important?
• Should these valid, important results of this study about a potentially harmful treatment change the treatment of your patient?

Completed Harm Worksheet for Evidence-Based Geriatric Medicine

Citation

Pahor M et al: Calcium-channel blockade and incidence of cancer in aged populations.
Lancet 1996;348:493-7. (also see 487-9 and 541-2)

Are the results of this harm study valid?

Were there clearly defined groups of patients, similar in all important ways other than exposure to the treatment or other cause?
Clearly defined, but heterogeneous. Exposed individuals different from non-exposed (more diabetes and cardiovascular disease, disability, hospitalisation, but lower diastolic pressure), but both groups cancer-free at the start of the study.
Were treatment exposures and clinical outcomes measured the same ways in both groups (e.g., was the assessment of outcomes either objective (e.g., death) or blinded to exposure)?
Yes. Asked to show their medications, and clinical outcomes measured the same way in both groups.
Was the follow-up of study patients complete and long enough?
Averaged 3.7 years, and long enough to show a positive relationship between CCBs and cancer. But there were only 47 cancers in 1549 person-years of CCB taking.
Do the results satisfy some “diagnostic tests for causation”?
Is it clear that the exposure preceded the onset of the outcome?
Probably; excluded everyone with known cancer at the start (still may have been some smouldering).
Yes (figure 2)
Is there positive evidence from a “dechallenge-rechallenge” study?
No
Is the association consistent from study to study?
No
Does the association make biological sense?
Whether interference with apoptotic destruction of cancer cells is sensible is hotly debated.
BUT: Was this a previously generated hypothesis, or was it one of several analyses carried out on a large data set of drugs and diseases? (We’ve written to the authors about this and Dr. Pahor has informed us that this hypotheses was generated prior to the study onset.)

Are the valid results of this randomised trial important?

Present (Case) Absent (Control)
Exposed to the treatment Yes (Cohort) 3.03%
a
b a + b
No (Cohort) 2.17%
c
d c + d
Totals a + c b + d a + b + c + d

In this study:
begin{align} qquad text{Relative Risk ($RR$)} &= 3.03%/2.17% \
&= 1.4 text{ ($P$ = 0.032)} end{align}

(and when adjusted for several baseline differences, RR ROSE (!) to 1.7 (P = 0.0005))

Should these valid, potentially important results of a critical appraisal about a harmful treatment change the treatment of your patient?

Can the study results be extrapolated to your patient?
Depends on whether you believe them. If you do believe them, they can be extrapolated to your patient.
To calculate the NNH for any Odds Ratio (OR) and your Patient’s Expected Event Rate for this adverse event if they were NOT exposed to this treatment (PEER)
$$qquad mathit{NNH} = frac{mathit{PEER}(mathit{OR}-1)+1}{mathit{PEER}(mathit{OR}-1)times(1-mathit{PEER})}$$
If we assume our patient is like the average individual in this study (the hazard ratios are like odds ratios and don’t differ in important ways between subgroups), then his Absolute Risk Increase in cancer over 3.7 years is 3.03% – 2.17% = 0.86% = and 1/0.86% gives an NNH of 116.
What are your patient’s preferences, concerns and expectations from this treatment?
Need to be determined.
What alternative treatments are available?
Lots of alternative treatments available for his hypertension (thiazides, beta-blockers). They have their own side-effects but are not reputed to cause cancer.

Other case-control and cohort studies vary in their conclusions about CCB risks, and a meta-analysis is awaited.

Until it is sorted out, you could describe NNHs (and possible NNHs) for alternative antihypertensive regimens with your patient and the two of you could collaborate in deciding on the most appropriate one for him.

Hypertension – Calcium-channel blockers may cause cancer

Clinical Bottom Line

• Until this gets sorted out properly, if your patient’s problem could be treated as well by some alternative drug (e.g., hypertension), it would be prudent to avoid using calcium-channel agents.
• If this result is true, the NNH to cause one additional cancer from taking CCBs for 3.7 years is 116.

Citation

Pahor M et al: Calcium-channel blockade and incidence of cancer in aged populations.
Lancet 1996;348:493-7.

(also see 487-9 and 541-2)

Clinical Question

In patients taking calcium antagonists for hypertension, are they at increased risk of cancer?

Search Terms

From the newspaper headline or from MEDLINE using “calcium antagonists” and “cancer”

The Study

Total or stratified random samples (>80% response rate) of 65+ y/o men and women in 3 sites in the USA. Showed their meds, were interviewed 90 minutes, and had their blood pressure, height and weight measured. Anyone with cancer in previous 3 years or on any cancer Rx was excluded and 94% of the remainder were followed for an average of 3.7 years by follow-up interview, hospital discharge info and the national death registry for the occurrence of new cancers.

The Evidence

Later Cancer Totals
Present Absent
Exposed to Calcium Channel Blockers Yes (Cohort) 3.03%
a
b a + b
No (Cohort) 2.17%
c
d c + d
Totals a + c b + d a + b + c + d

begin{align} qquad text{Relative Risk ($RR$)} &= 3.03%/2.17% \
&= 1.4 text{ ($P$ = 0.032)} end{align}

(and when adjusted for several baseline differences, RR ROSE (!) to 1.7 (P = 0.0005)).