Critical Appraisal Worksheet - Harm - Sheehy et al. PDF

# Critical Appraisal Worksheet - Harm - Sheehy et al. ## PICO question * **Population:** Patients who underwent a PCI who filled a prescription for a non-restricted cardiovascular drug during the year following the PCI * **Intervention:** Delay or absence of clopidogrel fill after PCI * **Comparat...

# Critical Appraisal Worksheet - Harm - Sheehy et al. ## PICO question * **Population:** Patients who underwent a PCI who filled a prescription for a non-restricted cardiovascular drug during the year following the PCI * **Intervention:** Delay or absence of clopidogrel fill after PCI * **Comparator:** No delay in clopidogrel fill after PCI * **Outcome:** All-cause death * **Time:** 1 year **In patients that underwent PCI is a delay or absence of filling clopidogrel associated with an increase in all-cause death at 1 year compared to those who did not have a delay in filling clopidogrel?** ## Are the results valid? ### Aside from the exposure of interest, did the exposed and control groups start and finish with the same risk for the outcome? - No. The exposed and unexposed groups did not start and finish with the same risk for the outcome based on the differences in characteristics seen between our groups. Statistical adjustment was conducted but likely not adequate and immortal time bias introduces issues. ### Were patients similar for prognostic factors known to be associated with the outcome (or was statistical adjustment done)? - According to Table 1, 2 and 3 there were slight differences in characteristics between the comparison groups. - For example, those who filled a rx with a delay had a greater number of chronic conditions compared to those who did not fill a rx or those who filled without a delay. - These differences are difficult to qualify given that we are not provided with p-values for the differences between groups, however, in the results section the authors state these characteristics are similar across the three groups. - Statistical adjustment was conducted using covariates listed in the 3 tables (multivariable Cox Proportional Hazards modelling utilized) and this would have accounted for some differences in characteristics between the groups, however, we are not provided with information on the nature of PCI (type of stent, where it was placed, urgent v non-urgent stent), history of CV disease, patients lab values (like cholesterol which is a prognostic factor for CV outcomes), other CV drug therapies, etc. - Therefore, it is likely that adjustment was not adequate to account for differences in prognostic factors between the groups. Residual confounding is likely a problem. - Also, immortal time bias is an issue to varying extents in all three groups. - Patients had to survive post discharge in order to fill their rx for their cardiovascular agent as well as clopidogrel in the first place. - Patients not only needed to survive, but needed to be capable of filling their rx. - If patients were re-hospitalized post PCI this would not be possible. - Even if they had received rx for clopidogrel during hospital stay, this rx would not have been captured in admin database. ### Were the circumstances and methods for detecting the outcome similar? - Yes. The outcome variable of all cause death was measured from the demographics database of the RAMQ databases for each of the exposure groups of interest. ### Was the follow-up sufficiently complete? - Yes. Exposure and outcome information was provided for all 13,663 individuals included cohort; however, a lot of people were censored throughout the follow-up. - One year was also likely long enough for outcome of interest to occur. - One would expect an acute event to occur within a few months of leaving the hospital as the drug is not meant to be used long term. ### Did the exposed and control group have the same risk (chance) for being exposed in the past? - No. Exposed people were more likely to fill their rx than controls because of the way the study was designed; you had to remain alive in order to fill your rx for a non restricted cv agent and then clopidogrel. Moreover, the study design itself dictated if you were considered exposed or not exposed. ### Were exposed and controls similar with respect to the indication or circumstances that would lead to exposure? - No. Those who died soon after PCI were less likely to fill a rx for a non-restricted CV agent or clopidogrel compared to those who didn't die. - By extension, those people would then also be more likely to be placed into the no clopidogrel group which could bias the results. - Probability of exposure was not independent between exposed and controls. ### Were the circumstances and methods for determining exposure similar for exposed and controls? - Yes. The exposure statuses for both exposed and controls was determined using the drug claims database of the RAMQ. ## What are the results? ### How strong is the association between exposure and outcome? - There is a strong and clinically relevant association between not filling a rx for clopidogrel post PCI or filling with delay compared to filling the rx without delay. - There is a 35% increased risk of death filing with delay and a 70% increased risk of death not filling a rx post PCI compared to filling without delay. - Both of these measures are above our 20% threshold of a relative risk reduction which we would deem clinically relevant. ### What is the risk ratio or odds ratio? - Patients who filled a rx for clopidogrel with a delay after filling a NRCD were at higher risk of mortality than patients without a delay: HR 1.35 - Patients who did not fill a clopidegrel prescription had a greater adjusted risk of mortality compared to those who filled rx without delay: HR 1.70 | | | |------------------------------------|-------------------| | Risk (no rx) | 108/1565= 6.9% | | Risk (delay) | 53/1180=4.5% | | Risk (no delay) | 315/10918=2.9% | | **RR delay vs no delay** | **4.5%/2.9%=1.55** | | **RR no rx vs no delay** | **6.9%/2.9%=2.38** | **Relative Risk increase (i.e. this is not a reduction): Delay vs no Delay** - 55% relative risk increase **Relative Risk increase (i.e. this is not a reduction): No RX vs no Delay** - 138% relative risk increase ### Is there a dose-response relationship between exposure and outcome? - Possibly. - One could look at the delay and no delay as a type of dose response. - We see an increasing risk of death with increasing delay of exposure to clopidogrel post PCI ### How precise was the estimate of the risk? - The effect estimates are not very precise. - Although neither cross the null, the CI for delay vs no delay is very close to 1. - Moreover, if we look at the lower bound of the CI for the HR comparing time to death in those who had a delay for their rx vs those who did not, there may not be a clinically important increased risk of death with the rx delay; however the no Rx would still likely be clinically important as the lower limit is still a 35% increase in risk. ### What is the confidence interval for the relative risk or odds ratio? - o delay HR 1.7, 95% CI: 1.35-2.15. ## How can I apply the results to patient care? - Lady Gaga was mystified when strong pain struck her left chest and left arm. - Gaga, who is a 36-year-old mother of one, didn't know it at the time what the issue was. - As the pain moved into her shoulder and back, Gaga took pain relievers and showered for relief. - But the next day, she was overwhelmed with nausea, sweating, vomiting, and chest pain. - An ambulance rushed her to the emergency room. - Gaga had difficulty convincing the attending physician that her symptoms were not related to her rock-n-roll lifestyle. - “They didn't take me seriously,” Gaga says. - She didn't fit the profile of a heart attack patient. - The doctors told her she was too young, she was not overweight, and there was no family history of heart disease. - Bedevilled by worsening pain and weakness and convinced she was dying, Gaga returned to the hospital several times over the days that followed, only to come home with no answers. - “I was angry and frustrated". As Gaga recalls, a cardiologist who had previously dismissed her complaints made the diagnosis. - “The doctor told me, ‘Well, it's a good thing you're persistent because you're having a heart attack." - After the diagnosis, Gaga is rushed into the cardiac cath lab where PCI is completed and a stent (little tube to keep arteries open) is placed to keep open the blocked artery. - The cardiologist, however is not sure if Gaga should receive clopidogrel or not as she does not have drug coverage in Canada. - Being an astute user of the evidence, you turn to the paper by Sheehy to decide whether or not delaying the use of clopidogrel until Gaga is approved for coverage will hurt her or not. - The cardiologist believes the delay could be detrimental. - Please use the scenario to answer the section on the JAMA guidelines. ### Were the study subjects similar to your patients or population? - Yes. We assume study population was similar to the general population undergoing PCI given that this is an observational study. ### Is your patient so different from those included in the study that the results may not apply? - Clopidogrel is an antiplatelet agent that reduces the risk of thrombosis following coronary stenting. Clopidogrel reduces atherothrombotic events in the year following percutaneous coronary interventions with stent implantation. - Most patients in the study were male and the median age was 70 years. - Most patients also had at least one chronic disease. - Although Gaga is significantly younger than the median age included in this study and female, there were young females included in the study and I do not believe that the mechanism of clopidogrel will be any different. - Therefore, I do not believe the study results should not apply. ### Was the follow-up sufficiently long? - Yes. 1 year of follow-up was likely adequate to detect differences in clopidogrel treatment on all-cause death. ### Were study participants followed-up long enough for important harmful effects to be detected? - Yes. The outcome in this study was all-cause death and we assume that any effects that clopidogrel treatment will have on this outcome will be relatively short-term, therefore, 1 year is likely adequate. ### Is the exposure similar to what might occur in your patient? #### Are there important differences in exposures (dose, duration, etc) for your patients? - Yes. Although it is likely that patients in the general population will be treated with clopidogrel in a similar way post PCI as seen in the study, the issue of immortal time bias brings up important issues related to differences in exposure probability based on death or hospital admission post PCI before individuals are able to fill a Rx for a non-restricted CV agent or clopidogrel. ## What is the magnitude of the risk? - The magnitude of the risk is moderate. - The baseline risk is the risk in the no treatment group. - The risk of death for those who had no Rx in filling their Rx over the 1 year follow-up period was 6.9%. - In those without delay was 2.9%; the ARR is 4%... a NNH of 25. - Similarly for the delay group the absolute even rate was 4.5% - ARR from no delay of 1.6% and a NNH of 63. ### What level of baseline risk for the harm is amplified by the exposure studied? - In our case it is difficult to determine as no subgroups according to baseline risk was completed. - However, imagine that in a very young female, the risk of a MI would be much smaller compared to a 90 year old male. - Thus, the NNH would not told true in each of these two types of patients because their baseline risk would be very different for the event. - If our average baseline risk was 6.9% in the no Rx group, one could easily imagine the risk in a younger female maybe only 3%, whereas in the older male it may be 10%. - Thus, the NNH would change as the ARR would be different. - For example, if we expect the RR increase to hold true across baseline risk (similar as we said in therapies) then the expected Absolute risk in the young female immediately treated with the drug would be 1.25% (i.e. 3% divided by 2.38 = 1.26). RR = EER/CER; the trick is to figure out where the numbers go. - The RR remains the same at 2.38. - So does the new baseline risk of 3% equate to the EER or the CER? - Try it both ways and see. - If we put the CER as 3% we get a value of 7.14%. - This 7.14% would be the expected event rate in those who received the drug without a delay. - This does not make sense as our immediately delivery of the drug is associated with a lower risk of the outcome. - Thus the 3% must be the EER in this calculation so that 3% divided by the RR of 2.38 equates to CER of 1.26%. - This makes sense as we expect our drug to lower the event rate. - The trick is understanding what are we calculating. - This study was set up as an increase in risk – not a reduction and so the numbers have to reflect this. - Our new ARR would be 0.0174 (0.03-0.0126 = 0.0174) or 1.74%. - Thus, the NNH would be would be 57. - So our risk of a bad outcome has changed from a NNH of 25 in the total population to now a NNH of 57 – the chance of a bad outcome is substantially lower in the young female because her baseline risk of the event was initially much lower. ### Are there any benefits known to be associated with the exposure? - Yes. Based on this study, filling a rx for clopidogrel on time post PCI decreases your risk of death. Previous studies have also evaluated the benefits of clopidogrel post PCI. ### What is the balance between benefits and harms for patients like yours? - The way the results are presented, there is a benefit of decreased mortality associated with filling a rx for clopidogrel on time post PCI. - Although other benefits or risks associated with this treatment are not presented, the benefits of clopidogrel post PCI have been studied previously and we would speculate that those who do not fill their rx are at decreased risk of harms or side effects associated with clopidogrel such as excessive bleeding. - The benefit of decreased mortality with timely receipt of clopidogrel treatment is likely to outweigh side effects of the drug.