OHDSI Home | Forums | Wiki | Github

Women of OHDSI - CALL FOR RESEARCH QUESTIONS

General
#1

The Women of OHDSI working group wants to run a study and are looking for clinical research questions! We’re calling on the OHDSI community to propose research questions which we will then use to prepare our study design.

If you would like to propose a question, please respond to this thread by Monday, May 20th with your research question. To help you formulate your question, we’ve provided simple ‘fill in the blanks’-style templates below:

Template for Estimation Questions:

To compare the risk of [insert outcome of interest here] between [insert target exposure here] and [insert comparator cohort here] we will estimate the population-level effect of exposure on the [Insert the metric of your analysis model here: hazards for Cox/ odds for logistic / rate ratio for Poisson] of the outcome during the period from [Insert time-at-risk start: e.g. 1 day after exposure start] to [Insert time-at-risk end: e.g. 30 days after exposure end] .

Example:
To compare the risk of angioedema between new users of levetiracetam and new users of phenytoin , we will estimate the population-level effect of exposure on the hazards of the outcome during the period from 1 day after exposure start to 0 days after exposure end

Estimation questions can focus on safety surveillance, comparing treatment options, and investigate treatment pathways more indepth. For more support formulating an estimation question, please check out Patrick’s forum post outlining possible research questions for the 2018 OHDSI F2F: OHDSI Face-to-Face at Columbia May2-3: Community study-a-thon

Template for Prediction Questions:

Among patients who [insert patient cohort], which patients will go on to have [insert outcome of interest] within [time window].

Example:
Of patients newly diagnosed with major depressive disorder, which patients will go on to have a suicidal event within 1-year of their diagnosis.

Prediction questions can focus on disease onset and progression, treatment choice, treatment response, treatment safety and treatment adherence.

OHDSI Community Call 7May2019
OHDSI Community Call 23Apr2019
#2

Because I like killing two birds with one stone. I want to promote @SCYou’s suggestion for a simple PLP study (Defining Cardiovascular related death in RWD).

Of patients who have myocardial infarction, heart failure or stroke and does not have any history of malignancy or other life-threatening condition, which patients will go on to die within 6-months or 1-year (up for debate the right timeframe) of their diagnosis.

The idea is the resulting PLP model could be validated against Chan’s NHIS-NCS cohort data and essentially used to create a phenotype for CV-related death that could be reused by future PLE studies.

1 Like
#3

Atrial fibrillation (AF) is a chronic progressive disease characterized by exacerbations and remissions. It remains the most common sustained arrhythmia seen in clinical practice, and represents a major burden to health care systems. Similar to other cardiovascular conditions, significant sex-specific differences have been observed in the epidemiology (lower rate of prevalence in women, women present at a later age), pathophysiology (sex-related differences in AF triggers and substrate), clinical presentation (women are more likely symptomatic, with relatively more severe symptoms), and natural history. Moreover, similar to other cardiovascular conditions there are substantial sex-specific differences in the management of AF, with women being significantly less likely to receive therapeutic anticoagulation, attempts at rhythm control, or undergo invasive cardiovascular procedures [quoted from Can J Cardiol. 2018 Apr].

So I want to investigate gender difference in management and its outcome of AF among various health-care systems with diverse genetic and national backgrounds.

To compare 1. therapeutic pathway (rhythm vs rate control, warfarin vs NOAC, regular dose vs under-dose, AF ablation, cardioversion), 2. the risk of ischemic and hemorrhagic event, and 3. disability-adjusted life year according to the therapy between women with atrial fibrillation and men with arial fibrillation we will estimate the population-level effect of exposure on the hazards for Cox of the outcome during the period from 1 day after exposure start to end of observation .

1 Like
#4

Just a reminder to post your research questions before the next Women of OHDSI WG call at 2pm on May 21st.

Here’s a question I would love to run across the OHDSI network:

Among patients newly diagnosed with major depressive disorder, which patients will go on to have suicidal event within the 30-days following a primary care visit .

1 Like
Women of OHDSI Meeting - 21May2019
#5

Looking at a special issue in PLOS one: https://collections.plos.org/s/alzheimersdisease. I would like to predict the development of Alzheimer’s Disease.

Maybe something like:
Among patients with an initial outpatient visit that satisfies: no prior Alzheimer’s Disease, aged 65 to 70, which patients will go on the have Alzheimer’s Disease 1 year until 5 years following the outpatient visit

#6

#1- Among women with osteoporosis who have been on bisphosphonates for at least 5 yrs, for those who have been on a ‘drug holiday’ (off the bisphosphonate for at least 6 months) or not, what is the association between ‘a drug holiday’ and ‘length of drug holiday’ with an osteoporosis-related fracture.

#2- Among women with a osteoporotic-fragility fracture, what is the association with the length of time between ( index fracture and treatment start date) and repeat fracture, and other outcomes (‘back to normal activities’)

I’ll be on a flight so will miss the call, but would likely have missed anyway. I like all of these- AF, Alz, Depression!

1 Like
#7

Just wanted to let you know that at EMC we are working on a study on dementia prediction, which may have some overlap with the Alzheimer study proposed by you.

We are predicting dementia in patients with an outpatient visit occurrence and no prior dementia, and then stratify the target population into various age groups, with a 5 year TAR.

Should you move forward with that question it may be good to sit together some day and discuss how to avoid redundant work.

#8

Here is another special issue that might be suitable, but the deadline is end of June: https://jbhi.embs.org/special-issues/enabling-technologies-in-health-engineering-and-informatics-for-the-new-revolution-of-healthcare-4-0/

#9

WoO, I’m glad to see your workgroup coming together to support each other in moving forward a goal to generate reliable evidence.

I support whatever question you ultimately settle on and would be delighted to help in any way I can once you decide on a study.

To add some additional study ideas to the table, here’s a type of prediction question that could be informative, for which I think our OHDSI data network could usefully contribute:

The US Preventative Services Task Force recommends regular screening for women for a variety of conditions, including breast, cervical, colorectal (colon) cancers. For each of these screenings, there is some diagnostic procedure performed which can detect the presence of the condition at that time. If a person tests positive, some additional diagnostics and then treatment intervention can be considered; if a person tests negative, the person is recommended to return in some time interval to be retested.

  1. Woman aged 30 to 65 are recommended to been screened for cervical cancer every 3-5 years with cervical cytology and.or hrHPV testing. (https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/cervical-cancer-screening2)
  2. All women aged 50 to 74 are recommended to be screened for breast cancer every 2 years with mammography, but there remains debate about screening mammography when aged 40-49, as it can depend on patient preference toward the benefit-risk tradeoff (https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/breast-cancer-screening1)
  3. Adults (men or women) aged 50 to 75 are recommended for colorectal cancer screening through multiple methods under different frequency intervals. (https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/colorectal-cancer-screening2)

So, while a screening is designed to support immediate detection of disease, my thought it that is also offers a useful moment in time to consider the application of a Patient-Level Prediction model to discuss future risk. In this way, a patient can be educated not just on ‘do you have the disease today?’ but ‘what is the chance you will develop the disease in the next time horizon?’. Having this personalized knowledge may encourage greater adherence to the screening recommendations for followup care.

An example framing of the prediction problem to complement an existing USPSTF screening recommendation would be:

Amongst women aged 40-74 who are undergo a screening mammagraphy who do not have prior breast cancer and are screened negative, which patients will go on to develop breast cancer in the 90d to 3 years following the screening mammagraphy?

1 Like
#10

Hi WOO,

I have two ideas to share:

  1. An analysis of which covariates are associated with IUD extrusion or mechanical complications of IUD placement. We could compare IUD users by age, parity, and type of contraceptive device. I would also be interested in whether a 3D rendering prior to the IUD placement is associated with the incidence of an extrusion or a mechanical complication.

  2. The association between 3D rendering use prior to fibroid surgery and revision rates.

Please let me know if you would like to develop these ideas further.

Thanks,
Matt

#11

Hi I have an idea to share with you all:

To compare the risk of all-cause mortality between new users of opioids (codeine, morphine, tramadol) and new users of NSAIDs (diclofenac, naproxen, celecoxib, etoricoxib) among patients with osteoarthritis, we will estimate the population-level effect of exposure on the hazards of the outcome during the period from 1 day after exposure start to 365 days after exposure end.

We would like to estimate the risk for each drug and not just opioids class vs. NSAIDs class.

#12

Hi everybody,

My name is Talita Duarte Salles and I’m an epidemiologist working with a primary care database in Spain called SIDIAP (https://www.sidiap.org/index.php/en).

I think it is a great initiative to support women to become leaders within the community, and I would be happy to participate.

I’m not sure I’ll be able to make the call tomorrow, but I would support all the ideas presented here, I think they are all quite relevant. I think it is important that we work on topics of research that are relevant for women’s’ health. So I’d suggest a stratified analysis by sex for those proposed ideas for which a stratification haven’t already been suggested.

I also have an idea of a prediction study:

Cardiovascular diseases (CVD) and breast cancer are the main causes of morbidity and mortality in women living in developed countries. “Improvements in early detection and treatment of breast cancer have led to an increasing number of breast cancer survivors who are at risk of long-term cardiac complications from cancer treatments. Currently, there are ≈3 million breast cancer survivors in the United States; however, older women are more likely to die of diseases other than breast cancer, and CVD is the most frequent cause. In older, postmenopausal women, the risk of mortality attributable to CVD is higher in breast cancer survivors than in women without a history of breast cancer. This greater risk manifests itself ≈7 years after the diagnosis of breast cancer, which highlights the need to reduce the additional burden of CVD during this time frame with early recognition and treatment of CVD risk factors .” (From Mehta LS, Watson KE, Barac A, et al. Cardiovascular Disease and Breast Cancer: Where These Entities Intersect: A Scientific Statement From the American Heart Association. Circulation 2018;Feb 1.)

So given that the main cause of death among breast cancer survivors is CVDs, a good prediction of CVD events among these patients at the moment of cancer diagnosis could help clinicians to focus on specific CVD prevention strategies. So I think it would be interesting to look at:

  • Among women newly diagnosed with breast cancer , which women will develop a cardiovascular disease in the first to the 7th year after cancer diagnosis?
#13

To compare the risk of newly diagnosed auto-immune disorders within 18 months following child birth between new users of IUDs and those who stay on the pill. Perhaps stratify results by those who have had previous, but less severe auto-immune disorders, such as celiac.

It is often the practice for women who have decided they have had their last child to be recommended for using a IUD for contraception post child-birth. Furthermore, there is some evidence that women have an elevated risk of serious auto-immune diseases following child birth. However, there are no studies that have asked if it’s the inflammatory response to the IUD that is at least partially responsible for the occurrence of a new autoimmune response. Indeed clinicians believe that it’s only a few rare anecdotal claims on the Internets that those with autoimmune had their problems resolved, or at least made better (measured by reduced steroid usage), after removing the IUD. This second part could be another inquiry: for those women having an autoimmune and an IUD, was removal of the IUD correlated to a reduction of steroid usage?

Given that women, minorities and children have been traditionally disadvantaged with regard to studies, I see great hope that network studies could be far less biased.

#14

The CUMC team has drafted a proposal about the covariates associated with uterine perforation following IUD placement. Please let us know if you are interested in collaborating!

#15

Thanks for letting us know @mattspotnitz. We’ve already selected a research question, but will keep your proposal in mind for future collaboration

t