| Title: | Data Package for Medical Datasets |
|---|---|
| Description: | Provides access to well-documented medical datasets for teaching. Featuring several from the Teaching of Statistics in the Health Sciences website <https://www.causeweb.org/tshs/category/dataset/>, a few reconstructed datasets of historical significance in medical research, some reformatted and extended from existing R packages, and some data donations. |
| Authors: | Peter Higgins [aut, cre] |
| Maintainer: | Peter Higgins <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 0.2.0.9001 |
| Built: | 2024-10-31 18:37:08 UTC |
| Source: | https://github.com/higgi13425/medicaldata |
This dataset is a subset of the data from a Duke University study of acute meningitis, as provided by Frank Harrell on his website here. The patients were found to have clinical symptoms of meningitis, which include a painful, stiff neck with a limited range of motion, headache, light sensitivity, high fever, confusion, and lethargy. This can progress rapidly, and prove fatal. Bacterial meningitis can be treated successfully with antibiotics, if given early. Antibiotics are unnecessary and unhelpful in viral meningitis, and can cause allergic reactions or other adverse events in some patients. Clinically, viral and bacterial meningitis have very similar symptoms, due to the common cause of inflammation of the meninges, the three membranous layers that protect the brain and spinal cord. A spinal tap (lumbar puncture) can be performed to obtain cerebrospinal fluid (CSF) for analysis to help classify cases as bacterial or viral in origin.
Timely classification and early treatment of bacterial meningitis can be life-saving.
abmabm
A data frame with 581 observations and 22 variables
case number, from 1 to 581; type: double
2 digit year, from 68 (1968) to 80 (1980), 72 missing; type: double
month, from 1 (January) to 12 (December), 81 missing; type: double
age in years, 81 missing; type: double
coded as 1 = Black, 2 = White, 85 missing; type: double
gender, coded as 0 = male and 1 = female, 81 missing; type: double
white blood cell count in blood, in thousands of cells per cubic millimeter, 141 missing; type: double
Percentage of white blood cells that are neutrophils, considered indicative of acute bacterial infection, in blood, 146 missing; type: double
Band cells, also known as immature neutrophils, considered indicative of acute bacterial infection, in blood, in percent of the total white blood cells, 153 missing; type: double
Blood glucose, in milligrams per deciliter, 258 missing; type: double
CSF (cerebrospinal fluid) glucose, in milligrams per deciliter, 129 missing; type: double
CSF (cerebrospinal fluid) protein, in milligrams per deciliter, 249 missing; type: double
CSF (cerebrospinal fluid) red blood cells, thousands of cells per cubic millimeter, 271 missing; type: double
CSF (cerebrospinal fluid) white blood cells, thousands of cells per cubic millimeter, 101 missing; type: double
CSF (cerebrospinal fluid) neutrophil percentage, percent of total white blood cells in CSF, 132 missing; type: double
CSF (cerebrospinal fluid) monocyte percentage, percent of total white blood cells in CSF, 165 missing; type: double
CSF (cerebrospinal fluid) lymphocyte percentage, percent of total white blood cells in CSF, 162 missing; type: double
gram stain result, values 0-6, 313 missing; type: double
result of csf culture for bacterial growth, values 0-6, 307 missing; type: double
result of blood culture for bacterial growth, values 0-6, 434 missing, 11; type: double
subset, training = 1 or test = 2; type: double
the outcome variable, acute bacterial meningitis, 0 = absent or 1 = present, 80 missing; type: double
Bacterial meningitis occurs in 1 in 100,000 people per year in the United States, most commonly in people between 16 and 23 years old, with additional age peaks in infants and young children, and the elderly. Early symptoms include headache, fever, and pinprick rash of reddish-purple tiny spots. The peak season for bacterial meningitis is in dry winter months in each hemisphere (December-February in the Northern Hemisphere). Bacterial meningitis has a fatality rate of 15-20%, which is higher in the elderly. Untreated bacterial meningitis due to Streptococcus pneumoniae or H. influenzae approaches 100% mortality.
Viral meningitis generally resolves with supportive care, but bacterial meningitis requires early use of antibiotic therapy. A subset of patients will require a head CT (computed tomography) scan before a lumbar puncture to obtain CSF, which must occur before antibiotics are given. The CSF (cerebrospinal fluid) is sent to the lab for cell count and differential, glucose, protein, gram stain, and culture. Normally, the meninges are part of the blood-brain barrier, which keeps assorted toxins we ingest out of the brain. Proteins and large molecules from the bloodstream are not able to reach the CSF normally, and even smaller molecules like glucose equilibrate slowly, so that CSF glucose lags behind changes in blood glucose. Normal CSF has some white blood cells, with around 70% lymphocytes and 30% monocytes, with very few neutrophils. When bacteria begin to multiply in the CSF, CSF protein levels go up, glucose is consumed, and white blood cells (largely neutrophils) arrive to fight the bacteria. Viral infections tend to attract more lymphocytes and monocytes. Characteristic findings in bacterial meningitis include low CSF glucose, a low CSF to serum glucose ratio, high CSF protein, and a high CSF white blood cell count, usually composed of neutrophils. However, the spectrum of CSF values in bacterial meningitis is so wide that any one of these findings is of little value. About 20% of lumbar punctures draw a bit of blood, which is seen when there are more than 100,000 RBCs per cubic mm. This blood will also increase the CSF WBCs by one WBC per every 500-1000 RBCs, and may require adjustment of the WBC number.
Empiric treatment with antibiotics is begun as soon as the CSF reaches the lab, often with a combination of ceftriaxone plus vancomycin, and ampicillin is added in the elderly to cover Listeria. Meropenem is often used in the immunocompromised for broader coverage including Listeria and Pseudomonas. The gram stain helps detect bacteria in the CSF, but is only ~60% sensitive, and will usually miss tuberculosis, toxoplasmosis, and fungal meningitis. CSF culture (growing the bacteria) is more sensitive, but takes several days.
The primary analysis task is to classify viral (abm = 0) vs. bacterial (abm = 1) cases, as published in Spanos A, Harrell FE, Durack DT (1989): Differential diagnosis of acute meningitis: An analysis of the predictive value of initial observations. JAMA 262: 2700-2707. There are data on 581 patients, with typical levels of missing data in clincial observational studies done in the days of paper charts, so you may choose to impute some of the missing data. Some of the variables in the published paper need to be derived, including the CSF to blood glucose ratio, and the number of months from peak of summer (in North Carolina, in the Northern Hemisphere). In addition to imputation, modern machine learning classification techniques can be applied and compared to logistic regression.
This data set is provided generously by Frank Harrell, from his website here, as the dataset "abm". These data were published as Spanos A, Harrell FE, Durack DT (1989): Differential diagnosis of acute meningitis: An analysis of the predictive value of initial observations. JAMA 262: 2700-2707, which can be found here.
This data set contains data on 316 men who had undergone radical prostatectomy and received transfusion during or within 30 days of the surgical procedure and had available prostate serum antigen (PSA) follow-up data. The main exposure of interest was RBC storage duration group. A number of demographic, baseline and prognostic factors were also collected. The outcome was time to biochemical (PSA) cancer recurrence. The dataset is cleaned and complete. There are no outliers or data problems (more details after variable information).
blood_storageblood_storage
A data frame with 316 observations and 20 variables
RBC.Age.GroupNA, numeric, range: 1.00- 3
Median.RBC.AgeNA, numeric, range:10.00- 25
AgeNA, numeric, range:38.40- 79
AANA, numeric, range: 0.00- 1
FamHxNA, numeric, range: 0.00- 1
PVolNA, numeric, range:19.40-274
TVolNA, numeric, range: 1.00- 3
T.StageNA, numeric, range: 1.00- 2
bGSNA, numeric, range: 1.00- 3
BN+NA, numeric, range: 0.00- 1
OrganConfinedNA, numeric, range: 0.00- 1
PreopPSANA, numeric, range: 1.30- 40
PreopTherapyNA, numeric, range: 0.00- 1
UnitsNA, numeric, range: 1.00- 19
sGSNA, numeric, range: 1.00- 4
AnyAdjTherapyNA, numeric, range: 0.00- 1
AdjRadTherapyNA, numeric, range: 0.00- 1
RecurrenceNA, numeric, range: 0.00- 1
CensorNA, numeric, range: 0.00- 1
TimeToRecurrenceNA, numeric, range: 0.27-104
Prostate cancer is the most common malignant neoplasm in men, and radical prostatectomy is among the primary therapies for localized prostate cancer. The biochemical recurrence-free survival rate 5 years after prostatectomy ranges from 70% to 90%. Improvements in the surgical technique have decreased the amount of intraoperative blood loss occurring during radical prostatectomy; however, substantial numbers of patients still require perioperative blood transfusions.
Blood transfusions are associated with adverse reactions, including postoperative infections and transfusion-related immune perturbations. Allogeneic leukocytes present in the transfused blood are thought to suppress host cellular immune responses. Furthermore, the immunodepressant effect is secondary to an imbalance of accumulated cytokines and proinflammatory mediators in the transfused blood against decreased production of lymphocyte stimulating cell-mediated cytokines, such as interleukin 2 and increased release of immunosuppressive prostaglandins in the patient undergoing transfusion.
In cancer patients, perioperative blood transfusion has long been suspected of reducing long-term survival, but available evidence is inconsistent. It is also unclear which components of transfused blood underlie the cancer-promoting effects reported by some studies. An important factor associated with the deleterious effects of blood transfusion is the storage age of the transfused blood units. It is suspected that cancer recurrence may be worsened after the transfusion of older blood.
This study evaluated the association between red blood cells (RBC) storage duration and biochemical prostate cancer recurrence after radical prostatectomy. Specifically, tested was the hypothesis that perioperative transfusion of allogeneic RBCs stored for a prolonged period is associated with earlier biochemical recurrence of prostate cancer after prostatectomy.
Patients were assigned to 1 of 3 RBC age exposure groups on the basis of the terciles (ie, the 33rd and 66th percentiles) of the overall distribution of RBC storage duration if all their transfused units could be loosely characterized as of 'younger,' 'middle,' or 'older' age. Although this approach resulted in the removal of certain patients with wide RBC age distributions, it has the advantage of defining an essentially random and clearly separable exposure.
Prostate-specific antigen (PSA) was used as a biochemical marker of prostate cancer recurrence after prostatectomy. A PSA value of at least 0.4 ng/mL (to convert to microg/L, multiply by 1.0) followed by another increase was considered biochemical cancer recurrence.
The initial population consisted of 865 men who had undergone radical prostatectomy and received transfusion during or within 30 days of the surgical procedure at Cleveland Clinic and had available PSA follow-up data. Of these patients, 110 were excluded from the analysis because they received a combination of allogeneic and autologous blood products. Of the remaining 755 patients, 405 (54%) received solely allogeneic and 350 patients (46%) received solely autologous RBC units. Of the 405 patients who received allogeneic RBC transfusion, 89 were excluded because their transfused RBC age distribution included more than one of the terciles. Thus, this dataset consists of the 316 patients who received solely allogeneic blood products and could be classified into an RBC age exposure group.
Cata et al. 'Blood Storage Duration and Biochemical Recurrence of Cancer after Radical Prostatectomy'. Mayo Clin Proc 2011; 86(2): 120-127.
This dataset is from the Duke University Cardiovascular Disease Databank, as provided by Frank Harrell at https://hbiostat.org. The patients were referred to Duke University Medical Center for chest pain. Cardiac catheterization (also known as a 'cath') is performed to diagnose and open blockages in these arteries, often followed by stenting to keep them open.
When I (PDRH) was an intern at Duke in 2001, it was a nightly occurrence for the 'cath bus' to arrive from Lumberton, North Carolina with 8 patients in chest pain (all at once), lighting up the pager of the cardiology intern with multiple requests for nitroglycerin.
cathcath
A data frame with 3504 observations and 6 variables
gender, coded as 0=male and 1=female; type: double
age in years; type: double
duration of symptoms of chest pain, in days; type: double
serum cholesterol level in milligrams per deciliter. Note that 35.56% of observations for cholesterol are missing; type: double
significant coronary artery disease found on cardiac catheterization: 0=no, 1=yes; type: double
three vessel or left main coronary artery disease found on cardiac catheterization. There are 3 missing observations for tvdlm: 0=no, 1=yes; type: double
Coronary artery disease, or atherosclerosis of the blood vessels supplying the heart, is the most common cause of death in the United States. Duke University Medical Center is located in the southeastern US, in a region of highly prevalent coronary artery disease (CAD), associated with frequent smoking, diabetes, and obesity, along with genetic risk factors for early-onset CAD in the local Lumbee Indian tribe.
At catheterization, a reduction in artery diameter by at least 75% is considered a significant reduction in blood flow that puts downstream heart muscle at risk of ischemia. Significant coronary disease can be treated with multiple medications, or by opening the partially blocked artery. This is done with a via a catheter through blood vessels in the wrist or the groin, and threaded through blood vessels to the heart. Once at the location of the blockage, the narrowed area can be opened with a balloon, and the newly opened artery kept open with a carefully placed coronary stent.
The sigdz variable is an indicator for the presence of a blockage of at least 75% in any one of the left main coronary artery or in any of the three distributing arteries - the LAD, LCA, and RCA.
The tvdlm variable is an indicator for one of two results from the catheterization. One is three vessel disease - having blockages of all three of the left anterior descendig (LAD) coronary artery, the left circumflex coronary artery (LCA), and the right coronary artery (RCA). This occurs most commonly in association with diabetes, and with longstanding severe CAD. The other is left main disease, which is blockage of the artery that feeds both the LAD and the LCA. Blockage in the left main coronary artery is frequently fatal, and is colloquially known as a "widowmaker".
Some interesting potential analyses include predicting the probability of significant (>= 75% diameter narrowing in at least one major coronary artery) coronary disease, and predicting the probability of severe coronary disease given that some significant disease is present. The first analysis would use sigdz as a response variable, and the second would use tvdlm on the subset of patients having sigdz=1.
This data set is provided generously by Frank Harrell, from his website here, as the dataset "acath".
Longitudinal results of a randomized, placebo-controlled trial of botulinum toxin type B (BotB) in 109 subjects across 9 sites. (more details available below the variable definitions).
cdystoniacdystonia
A data frame with 631 observations and 7 variables
week of measurement; type: numeric
study site, levels: 1-9; type: double
patient id at each site (not unique across sites); type: numeric 1-19
treatment assignment: levels: 1 = placebo, 2 = botox B 5000 units, 3 = botox B 10,000 units; type: double 1-3)
age in years, range 26-83; type: numeric
sex of participant, 1 = F, 2 = M; type: double
total score of Toronto Western Spasmodic Torticollis Rating Scale, range 0-87; type: numeric
Cervical dystonia, also called spasmodic torticollis, is a painful condition in which your neck muscles contract involuntarily, causing your head to twist or turn to one side. Cervical dystonia can also cause your head to uncontrollably tilt forward or backward.
A rare disorder that can occur at any age, cervical dystonia most often occurs in middle-aged people, women more than men. Symptoms generally begin gradually and then reach a point where they don't get substantially worse.
109 subjects across 9 sites were randomized to placebo (N = 36), 5000 units of botulinum toxin B (N = 36), or 10,000 units of botulinum toxin B (N = 37), injected once at baseline into the affected muscle to partially paralyze it and make it relax, releasing the spasmed side of the neck and head.
The response variable is the score on the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS-Total on a 0-87 scale), which measures the severity, pain, and disability of cervical dystonia (higher scores mean more impairment) at weeks 0 (baseline), 2, 4, 8, 12, and 16. It is expected that the single botox injection at week 0 may wear off over time.
This data set is from a study published in 1999 in the journal Neurology,
A. Brashear, M.F. Lew, D.D. Dykstra, C.L. Comella, S.A. Factor, R.L. Rodnitzky, R. Trosch, C. Singer, M.F. Brin, J.J. Murray, J.D. Wallace, A. Willmer–Hulme, M. Koller (1999), Safety and efficacy of NeuroBloc (botulinum toxin type B) in type A–responsive cervical dystonia. Neurology, 53(7), 1439.
This dataset has been passed through Statistical Methods for the Analysis of Repeated Measurements by Charles S. Davis, pp. 161-163 (Springer, 2002) and can also be found at Frank Harrell's website here.
A dataset containing details of SARS-CoV-2 testing in 2020 at CHOP
covid_testingcovid_testing
A data frame with 15524 observations and 17 variables
id number for each subject; type: numeric
an auto-generated fake first name; type: character
an auto-generated fake last name; character
anonymized Gender, levels: female, male; type: character
day after start of pandemic; type: numeric
test that was performed, levels: covid, xcvd1; type: character
Clinic or ward where the specimen was collected, 88 levels; type: character
result of test, levels: positive, negative, invalid; type: character
patient group, levels: patient, misc_adult, client, other adult, unidentified; type: character
Age of subject at time of specimen collection (Anonymized), units = years; type: numeric
Whether the specimen was collected via a drive-thru site, levels: 1: Collected at drive-thru site; 0: Not collected at drive-thru site; type: numeric
Cycle at which threshold reached during PCR, range: 14.05-45; type: numeric
Whether an order set was used for test order, levels: 1: Collected via orderset; 0: Not collected via orderset; numeric
Payor associated with order, levels: commercial, government, unassigned, medical assistance, self pay, charity care, other; type: character
Disposition of subject at time of collection, levels: inpatient, emergency, observation, recurring outpatient, outpatient, not applicable, day surgery, admit after surgery-obs, admit after surgery-ip; type: character
Time elapsed between collect time and receive time, range: 0 - 61370.2, units = hours; type: numeric
Time elapsed between receive time and verification time, range: -18.6 - 218.2, units = hours; type: numeric
...
Data on testing for SARS-CoV2 from days 4-107 of the COVID pandemic in 2020. CHOP is a pediatric hospital in Philadelphia, Pennsylvania, USA. These data have been anonymized, time-shifted, and permuted.
This data set is from Amrom E. Obstfeld, who de-identified data on COVID-19 testing during 2020 at CHOP (Children's Hospital of Pennsylvania). This data set contains data concerning testing for SARS-CoV2 via PCR as well as associated metadata. These data have been anonymized, time-shifted, and permuted.
This data set contains 64 consecutive patients who underwent T-cell replete, matched sibling donor reduced-intensity conditioning allogeneic hematopoietic stem cell transplant. The primary risk factor of interest was the number of activating killer immunoglobulin-like receptors (aKIRs: 1-4 vs. 5-6). (more details after variable information).
cytomegaloviruscytomegalovirus
A data frame with 64 observations and 26 variables
IDPatient ID, numeric, range: 1-64
ageRecipient age at transplant, numeric, range: 29-67
sexRecipient sex, numeric, range: 0 (female) - 1(male)
raceRecipient race, numeric, range: 0 (african-american) - 1 (white)
diagnosistype: character, levels: 13
diagnosis.typeCategory of cancer diagnosis, numeric, range: 0 (lymphoid) - 1 (myeloid)
time.to.transplantTime from cancer diagnosis to transplant (months), numeric, range: 1.84-173.8
prior.radiationPrior radiation therapy, numeric, range: 0 (no) - 1 (yes)
prior.chemoNumber of prior chemotherapy regimens, numeric, range: 0-8
prior.transplantPrior stem cell transplant, numeric, range: 0 (no) - 1 (yes)
recipient.cmvRecipient cytomegalovirus seropositive status, numeric, range: 0 (negative) - 1 (positive)
donor.cmvDonor cytomegalovirus seropositive status, numeric, range: 0 (negative) - 1 (positive)
donor.sexDonor sex, numeric, range: 0 (female) - 1 (male)
TNC.doseTotal nucleated cell dose (x 10^8/kg), numeric, range: 2.06- 21.0
CD34.doseTotal CD34+ (stem) cell dose (x 10^8/kg), numeric, range: 2.04- 12.5
CD3.doseTotal CD3+ (T) cell dose (x 10^8/kg), numeric, range: 1.08- 8.2
CD8.doseTotal CD8+ cell dose (x 10^8/kg), numeric, range: 0.16- 3.2
TBI.doseTotal body irradiation dosage (centiGrays), numeric, range:200.00-400.0
C1/C2HLA-Cw group, numeric, range: 0 (heterozygous) - 1 (homozygous)
aKIRsNumber of donor activating killer immunoglobulin-like receptors (hypothesized Predictor), numeric, range: 1.00- 6.0
cmvcytomegalovirus reactivation posttransplant (hypothesized Outcome), numeric, range: 0 (No) - 1 (Yes)
time.to.cmvTime to cytomegalovirus reactivation (months), numeric, range: 0.43- 84.5
agvhdAcute level 2-4 graft versus host disease, numeric, range: 0 (no) - 1 (yes)
time.to.agvhdTime to acute level 2-4 graft versus host disease (months), numeric, range: 0.66- 85.2
cgvhdChronic graft versus host disease, numeric, range: 0 (no) - 1 (yes)
time.to.cgvhdTime to chronic graft versus host disease (months), numeric, range: 0.82- 65.1
A number of demographic, baseline and transplant characteristics were also collected. The primary outcome is presence of and time to cytomegalovirus reactivation. The dataset is cleaned and relatively complete. There are no outliers or data problems.
Hematopoietic stem cell transplantation (HSCT) is the transplantation of multipotent hematopoietic stem cells, from bone marrow, peripheral blood, or umbilical cord blood. It is a medical procedure most often performed for patients with certain cancers of the blood or bone marrow, such as multiple myeloma or leukemia. Allogeneic HSCT involves two people: the (healthy) donor and the (patient) recipient. Allogeneic HSC donors must have a tissue (HLA) type that matches the recipient.
In myeloablative allogeneic HSCT, chemotherapy or irradiation is given immediately prior to a transplant (the conditioning regimen) with the purpose of eradicating the patient's disease prior to the infusion of HSC and to suppress immune reactions. The bone marrow can be ablated (destroyed) with dose- levels that cause minimal injury to other tissues. For many patients who are at high risk for transplant-related mortality with myeloablative allogeneic HSCT, reduced- intensity conditioning allogeneic hematopoietic stem cell transplant has proven effective. Although the reduced-intensity conditioning allogeneic HSCT may avoid many of the organ toxicities associated with myeloablative conditioning, the risk for developing graft-versus-host disease and infection including cytomegalovirus remains significant.
Cytomegalovirus (CMV) is a common virus that can infect almost anyone. Once infected, your body retains the virus for life. Most people don't know they have CMV because it rarely causes problems in healthy people. But if pregnant or having a weakened immune system, CMV is cause for concern. For people with compromised immunity, such as after allogeneic HSCT, CMV infection can be fatal. Natural killer (NK) and T cells provide protection against CMV reactivation. The reactivity of NK cells and some T-cell subsets are regulated by the interaction of killer immunoglobulin-like receptors (KIRs) with target cell HLA class 1 molecules. The donor activating KIR genotype has been implicated as a contributing factor for CMV reactivation after myeloablative allogeneic HSCT.
This study investigates whether donor KIR genotype influences reactivation of CMV after T-cell replete, matched sibling donor reduced-intensity conditioning allogeneic HSCT.
The study included 64 consecutive patients who underwent T-cell replete, matched sibling donor reduced-intensity conditioning allogeneic hematopoietic stem cell transplant between January 16, 2000 and April 24, 2007 at the Cleveland Clinic.
Human leucocyte antigen (HLA) typing on donors and recipients was performed to allow assessment of killer immunoglobulin-like receptor ligands (KIRs). To allow for comparison with previous studies, donors were categorized as having 1-4 or 5-6 activating killer immunoglobulin-like receptor genes (aKIRs).
CMV reactivation was defined as any detection of cytomegalovirus DNA in the blood; the lower detection limit for this assay was 600 copies/mL.
The initial population consisted of 865 men who had undergone radical prostatectomy and received transfusion during or within 30 days of the surgical procedure at Cleveland Clinic and had available PSA follow-up data. Of these patients, 110 were excluded from the analysis because they received a combination of allogeneic and autologous blood products. Of the remaining 755 patients, 405 (54%) received solely allogeneic and 350 patients (46%) received solely autologous RBC units. Of the 405 patients who received allogeneic RBC transfusion, 89 were excluded because their transfused RBC age distribution included more than one of the terciles. Thus, this dataset consists of the 316 patients who received solely allogeneic blood products and could be classified into an RBC age exposure group.
Sobecks et al. 'Cytomegalovirus Reactivation After Matched Sibling Donor Reduced-Intensity Conditioning Allogeneic Hematopoietic Stem Cell Transplant Correlates With Donor Killer Immunoglobulin-like Receptor Genotype'. Exp Clin Transplant 2011; 1: 7-13.
This dataset was collected with funding from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). The Pima Indian tribe located near Phoenix, Arizona (USA) has a very high rate of type 2 diabetes. This dataset includes a number of variables predictive of diabetes, and the outcome of a type 2 diabetes diagnosis within 5 years of the initial measurements. This dataset includes only females of at least 21 years of age, and of Pima Indian heritage, with at least 5 years of followup in a longitudinal study of diabetes.
The Pima Indian tribe has participated in a longitudinal diabetes study since 1965 because of its high incidence rate of diabetes. Each community resident over 5 years of age was asked to undergo a standardized examination every two years, which included an oral glucose tolerance test. Diabetes was diagnosed according to World Health Organization Criteria; that is, if the 2 hour post-load plasma glucose was at least 200 mg/dl at any survey examination or if the Indian Health Service Hospital serving the community found a glucose concentration of at least 200 mg/dl during the course of routine medical care.
diabetesdiabetes
A data frame with 768 observations on 9 variables, with significant missing data for some predictor variables.
number of pregnancies from 0 to 17; type: double
Plasma glucose concentration at 2 hours after administration of an oral glucose tolerance test in mg/deciliter, from 44 to 199, 5 missing; type: double
diastolic blood pressure in millimeters of mercury (mm Hg), the second number reported in blood pressure, from 24 to 122, 35 missing; type: double
triceps skin fold thickness in mm, a measure of subcutaneous fat, from 7 to 99, 227 missing; type: double
serum insulin at 2 hours after administration of an oral glucose tolerance test in microIU per milliliter (IU is international units), from 14 to 846, 374 missing; type: double
body mass index, in kg of weight per meters of height squared, from 18.2 to 67.1, 11 missing; type: double
a diabetes pedigree score, with points added for each additional relative with diabetes, weighted for the closeness of their genetic relation to the participant, from 0.78 to 2.42. Zero missing; type: double
Age in years, from 21 to 81, Zero missing; type: double
Diagnosis of diabetes in the following 5 years - pos or neg, 268 pos, Zero missing; type: factor
Type 2 diabetes mellitus (DM) is associated with obesity and a diet high in sugars and low in vegetables. People with type 2 DM become less sensitive to insulin, so that after a glucose load, their blood glucose and insulin rise, but the glucose level does not fall as quickly as it should, leading to sustained elevations in glucose and insulin. The incidence of type 2 DM is rising in many western cultures, as increasingly unhealthy and calorie rich diets become common. A version of this dataset is available through the UCI (University of California-Irvine) Machine Learning Repository as "PimaIndiansDiabetes". This dataset was recoded with NA for zero values which were likely to be missing in the variables glucose_mg-dl, insulin_mIU-mL, dbp_mm-hg, triceps_mm, and bmi by Friedrich Leisch. The units of each predictor were added to variable names, and several variables renamed for clarity by Peter Higgins.
The primary analysis task is to classify in each participant whether diabetes developed within 5 years of data collection (diabetes_5y = pos), or the participant tested repeatedly negative for diabetes over the next 5 years (diabetes_5y = neg).
This data set was provided through funding from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) as the dataset "diabetes". The donor of dataset to UCI was Vincent Sigillito of Johns Hopkins.
Data from a case-control study of esophageal cancer in Ille-et-Vilaine, France, evaluating the effects of smoking and alcohol on the incidence of esophageal cancer. Smoking and alcohol are associated risk factors for squamous cell cancer of the esophagus, rather than adenocarcinoma of the esophagus, which is associated with obesity and esophageal reflux (more details available below the variable definitions).
esoph_caesoph_ca
A data frame with 88 rows and 5 variables, with 200 cases and 975 controls.
6 levels of age: "25-34", "35-44", "45-54", "55-64", "65-74", "75+"; type: ordinal factor
4 levels of alcohol consumption: "0-39g/day", "40-79", "80-119", "120+"; type: ordinal factor
4 levels of tobacco consumption: "0-9g/day", "10-19", '20-29", "30+"; type: ordinal factor
Number of cases; type: integer
Number of controls; type: integer
An original base R dataset, though of somewhat unclear origin. The statistical textbook source is clear, though it is not clear which of the original epidemiological papers on esophageal cancer in Ille-et-Vilaine is referred to by this dataset. The original authors of the medical study were not credited in the base R dataset. There are several possible papers in PubMed, none of which quite match up with this dataset. This could be from Tuyns, AJ, et al., Bull Cancer, 1977;64(1):45-60, but this paper reports 778 controls, rather than the 975 found here. A 1975 paper from the same group reported 718 cases (Int J Epidemiol, 1975 Mar;4(1):55-9. doi: 10.1093/ije/4.1.55.). There is also another possible source - a 1975 paper from the same group, Usefulness of population controls in retrospective studies of alcohol consumption. Experience from a case–control study of esophageal cancer in Ille-et-Vilaine, France, Journal of Studies on Alcohol, 39(1): 175-182 (1978), which is behind a publisher paywall.
Benign fine needle aspirate (FNA) of a breast lesion. Notice the regular size of cells and nuclei, which are organized in orderly spacing. The nuclei are homogeneously dark with few visible nucleoli.
Malignant (cancerous) fine needle aspirate (FNA) of a breast lesion. Notice the very irregular size of cells and nuclei, which are are disorganized and seem to be growing over each other. The nuclei are also less homogeneously dark and more granular, suggesting active transcription from the dark nucleoli within each nucleus.
Breslow, N. E. and Day, N. E. (1980) Statistical Methods in Cancer Research. Volume 1: The Analysis of Case-Control Studies. IARC Lyon / Oxford University Press. Originally in base R datasets.
Results of a randomized, placebo-controlled, prospective 2-arm trial of rectal indomethacin (100 mg) vs. placebo prevent post-ERCP pancreatitis in 602 participants, as reported by Elmunzer, Higgins, et al. in 2012 in the New England Journal of Medicine (more details available below the variable definitions).
indo_rctindo_rct
A data frame with 602 observations and 33 variables
subject id, first integer indicates center, integer, range:1001-4003
study site (center), factor, 1 = University of Michigan, 2= Indiana University, 3 = University of Kentucky, 4 = Case Western
age in years, numeric, range: 19-90
risk score, numeric, range: 1-5.5
male or female, factor, levels: 1_female, 2_male
sphincter of oddi dysfunction was present, a risk factor favoring post-ERCP pancreatitis, factor, levels: 0_no, 1_yes
previous post-ERCP pancreatitis (PEP), a risk factor for future PEP, factor, levels: 0_no, 1_yes
Recurrent Pancreatitis, a risk factor for future PEP, factor, levels: 0_no, 1_yes
a Pancreatic Sphincterotomy was performed, a risk factor for PEP, factor, levels: 0_no, 1_yes
a sphincter pre-cut was needed to enter the papilla, a risk factor for PEP, factor, levels: 0_no, 1_yes
Cannulation of the papilla was difficult, a risk factor for PEP, factor, levels: 0_no, 1_yes
Pneumatic dilation of the papilla was performed, a risk factor for PEP, factor, levels: 0_no, 1_yes
An Ampullectomy was performed for dysplasia or cancer, which could be a risk factor for PEP, factor, levels: 0_no, 1_yes
Contrast was injected into the pancreas during the procedure, a risk factor for PEP, factor, levels: 0_no, 1_yes
The pancreas appeared to have acinarization on imaging, which could be a risk factor for PEP, factor, levels: 0_no, 1_yes
Brushings were taken from the pancreatic duct, a possible risk factor favoring post-ERCP pancreatitis. factor, levels: 0_no, 1_yes
Aspirin was used at a dose of 81 mg per day, which may increase the risk of bleeding. factor, levels: 0_no, 1_yes
Aspirin was used at a dose of 325 mg per day, which may increase the risk of bleeding. factor, levels: 0_no, 1_yes
Aspirin was used (at a dose of 325 mg per day(at any dose), which may increase the risk of bleeding. factor, levels: 0_no, 1_yes
A pancreatic duct stent was placed at the end of the procedure per the judgement of the endoscopist (more often in high-risk cases), a potential protective effect against PEP, factor, levels: 0_no, 1_yes
A pancreatic duct stent was placed in order to treat a clinically significant narrowing of the pancreatic duct, a potential protective effect against PEP, factor, levels: 0_no, 1_yes
A pancreatic duct stent was placed at the end of the procedure for any reason, a potential protective effect against PEP, factor, levels: 0_no, 1_yes
Sphincter of oddi manometry was performed during the procedure for SOD, a risk factor for PEP, factor, levels: 0_no, 1_yes
A biliary sphincterotomy was performed, which could be a risk factor for PEP, factor, levels: 0_no, 1_yes
A biliary stent was placed to relieve significant biliary obstruction, factor, levels: 0_no, 1_yes
Choledocholithiasis (gallstones blocking the biliary duct) was present, factor, levels: 0_no, 1_yes
Malignancy of the biliary duct or pancreas was found, factor, levels: 0_no, 1_yes
A trainee participated in the ERCP, which could be a risk factor for PEP, factor, levels: 0_no, 1_yes
outcome of post-ercp pancreatitis, factor, levels: 0_no, 1_yes
outpatient status, factor, levels: 0_inpatient, 1_outpatient
Sphincter of Oddi dysfunction type/level - higher numbers are more severe with greater association with PEP, factor, levels: 0_no SOC, 1_type 1, 2_type 2, 3_type 3
treatment arm, factor, levels: 0_placebo, 1_indomethacin
A gastrointestinal bleed occurred (which could be a complication of indomethacin therapy), factor, levels: 1. no, 2. yes
ERCP, or endoscopic retrograde cholangio-pancreatogram, is a procedure performed by threading an endoscope through the mouth to the opening in the duodenum where bile and pancreatic digestive juices are released into the intestine. ERCP is helpful for treating blockages of flow of bile (gallstones, cancer), or diagnosing cancers of the pancreas, but has a high rate of complications (15-25%).
The occurrence of post-ERCP pancreatitis is a common and feared complication, as pancreatitis can result in multisystem organ failure and death, and can occur in ~ 16% of ERCP procedures.
The inflammatory cytokine storm that can result from this procedural complication can be quite severe. Several small randomized trials suggested that anti-inflammatory NSAID therapies at the time of ERCP could reduce the rate of this complication, but all were rather small single-center studies, and were not sufficiently convincing to change practice.
Elmunzer, Higgins, and colleagues performed a meta-analysis of these small trials, which suggested that this was a significant effect, and that indomethacin could result in a 64% reduction in post-ERCP pancreatitis.
The investigators took this as a possible over-estimate of the effect (due to publication bias), and designed a multicenter RCT of a planned 948 patients to see a reduction of 50% from a placebo rate of 10% to an indomethacin rate of 5%. Two interim analyses were performed, after 400 and 600 patients were enrolled, using an alpha spending function. The Data and Safety Monitoring Board stopped the study after 602 participants were enrolled because of the significantly positive effect of indomethacin, which reduced post-ERCP pancreatitis from 16% in the placebo group to 9% in the indomethacin group.
You can find the manuscript at Indomethacin to Prevent Post-ERCP Pancreatitis.
This data set is sourced from the authors of the 2012 manuscript in the New England Journal of Medicine, entitled, A Randomized Trial of Rectal Indomethacin to Prevent Post-ERCP Pancreatitis, pages 1414-1422 volume 366, in the April 12, 2012 edition, authored by the Elmunzer, BJ, Higgins PDR, et al. You can find the manuscript at Indomethacin to Prevent Post-ERCP Pancreatitis.
Results of a Cohort Study of the Pharmacokinetics of Intravenous Indomethacin, with plasma concentrations over time (more details available below the variable definitions).
indomethindometh
A data frame with 66 observations and 3 variables
subject id number for each participant; type: character
Time from initial dose in hours; type: double
Concentration of indomethacin in the plasma in micrograms per milliliter' type: double
This data set contains data on 6 healthy volunteer subjects who participated in a pharmacokinetic study of intravenous indomethacin. Indomethacin is an anti-inflammatory and pain-relieving non-steroidal medication. It can be administered by the intravenous, oral, or rectal suppository routes. Some of the indomethacin is excreted in the bile and reabsorbed by the intestine. This phenomenon, called enterohepatic circulation, keeps the drug around longer than would be expected otherwise.
Each subject in Study 1 (intravenous route) received a single 50 mg dose of radioactively labeled indomethacin (^14^-carbon-labeled, with each dose containing 25 microCuries of radioactivity). Subjects received a standard meal (one 8-oz can of Metrecal, 8 oz of whole milk, and one medium-size apple) 30 rain prior to medication and 8 oz of water every 2 hr throughout the waking hours to ensure adequate urine output.
Blood samples were taken at frequent intervals over the first 8 hours after dosing, and the quantity of indomethacin in the plasma (as well as stool and urine) at each time point was measured in micrograms per milliliter. This data set only contains the plasma measurements from Table 1 on page 258 of the manuscript. While this paper was published in 1976 (post-Tuskegee reveal), there is no mention of ethics review, IRB review, or consent of the healthy volunteers.
The abstract from the original manuscript:
There are no discernible quantitative differences in the biotransformation and the excretion of indomethacin following oral, rectal, and intravenous administration of indomethacin-2-^14^C. Approximately 50% (range 24-115% for n = 6) of an intravenous dose undergoes enterohepatic circulation. Thus the bioavailability of indomethacin to the systemic circulation may exceed the administered dose. Relative to the intravenous dose, indomethacin is 80 and 100% bioavailable from suppositories and capsules, respectively. Absorption and/or reabsorption appears to be more rapid and uniform by the rectal route. Recognition of the attributes of biliary recycling also helps to explain the observed variability in apparent plasma half-life, while their neglect requires alternative explanations for anomalies between the disappearance rate from plasma and the corresponding appearance rate in urine.
Kwan, Breault, Umbenhauer, McMahon and Duggan (1976) Kinetics of Indomethacin absorption, elimination, and enterohepatic circulation in man. Pharmacokinetics and Biopharmaceutics. 1976 Jun;4(3):255-80. doi: 10.1007/BF01063617.
This data set contains 99 adult patients with a body mass index between 30 and 50 kg/m2 who required orotracheal intubation for elective surgery. Patient demographics, airway assessment data, intubation success rate, time to intubation, ease of intubation, and occurrence of complications were recorded. The dataset is cleaned and complete. There are no outliers or data problems (more details available below the variable definitions).
laryngoscopelaryngoscope
A data frame with 99 observations and 22 variables
ageAge (years), numeric, range: 20-77
genderGender, numeric, 0 = female; 1 = male
asaAmerican Society of Anesthesiologists physical status(1-4), range: 2-4
BMIBody Mass Index (kg/m^2), numeric, range: 31-61
MallampatiMallampati score predicting ease of intubation 1 = Full visibility of tonsils, uvula and soft palate (easy intubation); 2 = Visibility of hard and soft palate, upper portion of tonsils and uvula; 3 = Soft and hard palate and base of the uvula are visible; 4 = Only Hard Palate visible (difficult intubation), numeric, range: 1-4
RandomizationLaryngoscope randomized, numeric, range: 0 = Standard Macintosh #4, 1 = AWS Pentaz Video
attempt1_timeFirst intubation attempt time (seconds), numeric, range: 9-113
attempt1_S_FSuccessful intubation first attempt, numeric, 0 = no, 1 = yes
attempt2_timeSecond intubation attempt time (seconds), numeric, range: 11- 60
attempt2_assigned_methodSecond intubation attempt made with assigned laryngoscope, numeric, 0 = no, 1 = yes
attempt2_S_FSuccessful intubation second attempt, numeric, 0 = no, 1 = yes, numeric, range: 0 = no, 1 = yes
attempt3_timeThird intubation attempt time (seconds), numeric, range: 15- 30
attempt3_assigned_methodThird intubation attempt made with assigned laryngoscope, numeric, 0 = no, 1 = yes
attempt3_S_FSuccessful intubation third attempt, numeric, 0 = no, 1 = yes, numeric, range: 1-1
attemptsNumber of intubation attempts, numeric, range: 1-3
failuresNumber of intubation failures, numeric, range: 0-2
total_intubation_timeTotal Intubation time (second), numeric, range: 9-100
intubation_overall_S_FOverall successful intubation, numeric, 0 = no, 1 = yes
bleedingBleeding (trace), numeric, 0 = no, 1 = yes
easeEase of tracheal intubation, 0 = extremely easy to 100 = extremely difficult, numeric, range: 0-100
sore_throatSeverity of postoperative sore throat, 0 = none; 1 = mild; 2 = moderate; 3 = severe, numeric, range: 0- 3
viewCormack-Lehane grade of glottic view 0 = "not good" Cormack- Lehane grade 1 or 2; 1 = "good" Cormack-Lehane grade 3 or 4, numeric, range: 0- 1
The Laryngoscope dataset was contributed by Dr. Amy Nowacki, Associate Professor, Cleveland Clinic. Please refer to this resource as: Amy S. Nowacki, 'Laryngoscope Dataset', TSHS Resources Portal (2017). Available at https://www.causeweb.org/tshs/laryngoscope/.
Difficult and failed tracheal intubations are among the principal causes of anesthetic-related mortality and morbidity. Because a good laryngeal view facilitates successful tracheal intubation, new technologies have been introduced to improve visualization. Video laryngoscopes, for example, often use miniature cameras to facilitate visualization of the laryngeal inlet with no need to align the oral, pharyngeal, and tracheal axes.
The Pentax AWS is a novel video laryngoscope, available in Japan since 2006, which is designed to facilitate intubation by providing a video image of the glottis. It incorporates a miniature video camera and a battery-powered, built-in LCD monitor. A disposable blade is attached to the base system. Incorporation of an LCD display makes it possible to view the glottis simultaneously with insertion of the endotracheal tube (ETT). In this regard, it differs from some other video laryngoscope designs that use external monitors. The Pentax AWS also differs in having a side channel that positions and guides the ETT. Reports suggest that the Pentax AWS can help intubate, but randomized data remain sparse.cr
This study tested the hypothesis that intubation with the Pentax AWS would be easier and faster than with a standard Macintosh laryngoscope with a #4 blade.
These are data from a study by Abdallah et al. A Randomized Comparison between the Pentax AWS Video Laryngoscope and the Macintosh Laryngoscope in Morbidly Obese Patients. Anesthesia Analgesia 2011; 113: 1082-7.
This study enrolled 236 adult patients undergoing elective thoracic surgery requiring a double-lumen endotracheal tube. Gender, physical status, BMI, age, Mallampati score, smoking status, preoperative pain, surgery size, intervention and the outcomes (cough, sore throat and pain swallowing at various time points) are provided. The dataset is cleaned and complete (missing outcomes for 2 patients). There are no outliers or data problems (more details available below the variable definitions).
licorice_garglelicorice_gargle
A data frame with 235 observations and 19 variables
preOp_genderGender, numeric, 0 = Male; 1 = Female
preOp_asaAmerican Society of Anesthesiologists physical status, numeric, 1 = a normal healthy patient; 2 = a patient with mild systemic disease; 3 = a patient with severe systemic disease
preOp_calcBMIBody mass index (kg/m^2), numeric, range:16-36
preOp_ageAge (years), numeric, range:18-86
preOp_mallampatiMallampati score, with 1 = easy to intubate, 4= difficult intubation, numeric, 1 = soft palate, fauces, uvula, pillars visible; 2 = soft palate, fauces, uvula visible; 3 = soft palate, base of uvula visible; 4 = soft palate not visible at all
preOp_smokingSmoking status, numeric, 1 = Current; 2 = Past; 3 = Never
preOp_painPreoperative pain, numeric, 0 = No; 1 = Yes
treatIntervention, 0 = Sugar 5g; 1 = Licorice 0.5g
intraOp_surgerySizeSurgery size, numeric, 1 = Small (thoracoscopy); 2 = Medium (thoracotomy < 3 h); 3 = Large (thoracotomy > 3 h or blood loss > 1000 mL)
extubation_coughAmount of coughing immediately after extubation, numeric, 0 = No cough; 1 = Mild; 2 = Moderate; 3 = Severe
pacu30min_coughAmount of coughing at 30 minutes after arrival in PACU, numeric, 0 = No cough; 1 = Mild; 2 = Moderate; 3 = Severe
pacu30min_throatPainSore throat pain score at rest at 30 minutes after arrival in PACU (11 point Likert scale, 0=no pain, 10 = worst pain)
pacu30min_swallowPainSore throat pain score during swallowing at 30 minutes after arrival in PACU (11 point Likert scale, 0=no pain, 10 = worst pain), numeric, range: 0-10
pacu90min_coughAmount of coughing at 90 minutes after arrival in PACU, numeric, 0 = No cough; 1 = Mild; 2 = Moderate; 3 = Severe
pacu90min_throatPainSore throat pain score at rest at 90 minutes after arrival in PACU (11 point Likert scale, 0=no pain, 10 = worst pain), numeric, range: 0-6)
postOp4hour_coughAmount of coughing at 4 hours after surgery, numeric, 0 = No cough; 1 = Mild; 2 = Moderate; 3 = Severe, range: 0-2
postOp4hour_throatPainSore throat pain score at rest at 4 hours after surgery (11 point Likert scale, 0=no pain, 10 = worst pain), numeric, range: 0-6), numeric, range: 0- 7
pod1am_coughAmount of coughing on the first postoperative morning, 0 = No cough; 1 = Mild; 2 = Moderate; 3 = Severe, numeric, range: 0- 3
pod1am_throatPainSore throat pain score at rest on the first postoperative morning (11 point Likert scale, 0=no pain, 10 = worst pain), numeric, range: 0-6), numeric, range: 0- 6
The Licorice Gargle dataset was contributed by Dr. Amy Nowacki, Associate Professor, Cleveland Clinic. Please refer to this resource as: Amy S. Nowacki, 'Licorice Gargle Dataset', TSHS Resources Portal (2017). Available at https://www.causeweb.org/tshs/licorice-gargle/.
Postoperative sore throat is a common and annoying complication of endotracheal intubation. Intubation with double-lumen tubes, which are much larger than conventional single-lumen tubes, are especially likely to provoke sore throats, with a reported incidence up to 90%. Presumably, postoperative sore throats are a consequence of local tissue trauma, due to laryngoscopy and/or endotracheal intubation, leading to inflammation of pharyngeal mucosa.
Nonpharmacological methods for preventing an intubation-related sore throat include using smaller-sized endotracheal tubes, lubricating the endotracheal tube with water-soluble jelly, and careful airway instrumentation as examples. Pharmacological measures for attenuating postoperative sore throats include inhalation of beclomethasone or fluticasone propionate; gargling with azulene sulfonate, aspirin, or ketamine; and gargling or spraying benzydamine hydrochloride on the endotracheal cuff for example. Each of these approaches and others not listed, however, has limitations and variable success rates; thus none has become established or is in routine clinical use.
Recently, a study reported that gargling with licorice halves the risk of sore throat after intubation with conventional endotracheal tubes, based on a study of just 40 patients. A number of active ingredients have been isolated from licorice, including glycyrrhizin, liquilitin, liquiritigenin, and glabridin. The glycyrrhizin component reportedly has anti-inflammatory and antiallergic properties. Liquilitin and liquiritigenin have peripheral and central antitussive properties. Glabridin has significant antioxidant and ulcer-healing properties, which might help heal pharyngeal and tracheal mucosa after minor injuries that often complicate laryngoscopy, intubation, and endotracheal tube cuff inflation.
This study tested the hypothesis that gargling with licorice solution immediately before induction of anesthesia prevents sore throat and postextubation coughing in patients intubated with double-lumen tubes.
These are data from a study by Ruetzler et al. 'A Randomized, Double-Blind Comparison of Licorice Versus Sugar-Water Gargle for Prevention of Postoperative Sore Throat and Postextubation Coughing'. Anesth Analg 2013; 117: 614 – 21.
The objective of this randomized controlled trial was to determine whether treatment of maternal periodontal disease can reduce risk of preterm birth and low birth weight (more details available below the variable definitions).
optopt
A data frame with 823 observations and 171 variables
PIDParticipant ID, First digit indicates enrollment center (1 = NY, 2 = MN, 3 = KY, 4 = MS); Next 4 digits are sequential; Sixth digit is a check digit; There are no missing data, numeric, range: 100034-402477
ClinicEnrollment Center, factor, NY = Harlem Hospital, MN = Hennepin County Center; KY = University of Kentucky; MS = University of Mississippi Medical Center; There are no missing data
GroupRandomized treatment assignment, factor, T = Intervention; C = Control; There are no missing data
AgeAge of participant at baseline (years), numeric, range: 16-44
BlackBlack participant (self-identified), factor; Yes, No
WhiteWhite participant (self-identified), factor; Yes, No
Nat.AmNative American participant, incl. Latin Americans with aboriginal origin(self-identified), factor; Yes, No
AsianAsian participant (self-identified), factor; Yes, No
HispHispanic participant (self-identified), factor; Yes, No
EducationEducation level of participant, factor; LT 8 yrs = Less than 8 years; 8-12 yrs = 8 to 12 years; MT 12 yrs = More than 12 yrs; blank = Missing
Public.AsstcePublic Assistance: Whether a government agency paid for the delivery, factor; Yes, No;
HypertensionWhether participant had chronic hypertension at baseline, factor; Yes, No
DiabetesWhether participant had diabetes at baseline (self-reported), factor; Yes, No
BL.Diab.TypeBaseline Diabetes Type: Type of diabetes, for participants having diabetes at baseline (self-reported), factor; Type I; Type II; Blank = No diabetes at baseline (variable 13 = No)
BMINA, numeric, range: 15.000-68.0
Use.TobSelf-reported participant history of tobacco use, factor; Yes, No; Blank = Missing
BL.Cig.DaySelf-reported number of cigarettes per day for those with tobacco use history, numeric, range: 1-30; Blank = Missing (variable 16= Yes or blank) or non-smoker (variable 16 = No)
Use.AlcSelf-reported participant history of alcohol use, factor; Yes, No; Blank = Missing
BL.Drks.Day, Blank = Missing (variable 18 = Yes or blank) or non-drinker (variable 18 = No)
Drug.AddSelf-reported participant history of drug addiction, factor; Yes, No; Blank = Missing
Prev.pregAny previous pregnancy, factor; Yes, No; No missing data
N.prev.pregNumber of previous pregnancies for those with any previous pregnancy, numeric, range: 1-11; Blank = Missing (variable 21 = Yes) or no previous pregnancies (variable 21 = No)
Live.PTBPrevious live preterm birth for those with any previous pregnancy, factor; Yes; No = No previous live preterm birth (variable 21 = Yes) or no previous pregnancies (variable 21 = No)
Any.stillbirthPrevious stillbirth, factor; Yes; No = No previous stillbirth (variable 21 = Yes) or no previous pregnancies (variable 21 = No)
Spont.abPrevious spontaneous abortion, factor; Yes; No; Blank = Missing (variable 21 = Yes) or no previous pregnancies (variable 21 = No)
Induced.abPrevious induced abortion, factor; Yes; No; Blank = Missing (variable 21 = Yes) or no previous pregnancies (variable 21 = No)
Any.live.ptb.sb.sp.ab.in.abAny previous live pre-term birth, stillbirth, spontaneous abortion, or induced abortion, factor; Yes; No = No live pre-term birth/stillbirth/abortion (variable 21 = Yes) or no previous pregnancies (variable 21 = No)
N.living.kidsNumber of living children the subject had at baseline, numeric, range: 0-9; Blank = Missing (variable 21 = Yes) or no previous pregnancies (variable 21 = No)
Tx.comp.Whether treatment plans were completed by participants in treatment group, factor, Yes = Completed; No = Not completed; Und = Some therapy (unknown whether completed); Blank = Withdrew from treatment (variable 3 = T) or no periodontal therapy (variable 3 = C)
Local.anesWhether any local anesthetic used during periodontal therapy for participants in treatment group, factor, Yes; No = No local anesthetic used or withdrew from treatment (variable 3 = T); Blank = No periodontal therapy (variable 3 = C)
Topical.AnestWhether any topical anesthetic used during periodontal therapy for participants in treatment group, factor, Yes; No = No topical anesthetic used or withdrew from treatment (variable 3 = T); Blank = No periodontal therapy (variable 3 = C)
Tx.timeTotal treatment time for participants in treatment group (hours), numeric, range: 0.117-5.8; Blank = Withdrew from treatment (variable 3 = T and variable 29 = blank) or no periodontal therapy (variable 3 = C)
EDC.necessary.Whether patient required1 essential dental care (EDC), factor, Yes; No; Blank = Missing
Completed.EDCDid patient complete EDC before 20 weeks gestational age?, factor, Yes; No; Blank = Missing
N.extractionsNumber of teeth extracted during EDC, numeric, range: 0-20; Blank = Missing
N.perm.restorationsNumber of permanent restorations carried out as a part of EDC, numeric, range: 0-18; Blank = Missing
N.qualifying.teethNumber of teeth meeting OPT (Obstetrics and Periodontal Therapy Study) criteria for having periodontal disease at baseline, numeric, range: 3.000-28.0
BL.GEWhole-mouth average gingival index at baseline, numeric, range: 0.429-3.0, Silness-Lowe Gingival Index: Higher value indicates more severe inflammation; 0 = Normal gingiva; There are no missing data
BL..BOPPercentage of sites bleeding on probing at baseline, numeric, range:33.951-100.0
BL.PD.avgWhole-mouth average pocket depth at baseline (mm), numeric, range: 1.851-7.0
BL..PD.4Percentage of sites with pocket depth greater than or equal to 4mm at baseline, numeric, range: 3.571-99.2
BL..PD.5Percentage of sites with pocket depth greater than or equal to 5mm at baseline, numeric, range: 0-91.7
BL.CAL.avgWhole-mouth average clinical attachment level at baseline (mm), numeric, range: 0.185-5.1
BL..CAL.2Percentage of sites with clinical attachment level greater than or equal to 2 mm at baseline, numeric, range: 2.381-100.0
BL..CAL.3Percentage of site with clinical attachment level greater than or equal to 3 mm at baseline, numeric, range: 0-94.9
BL.Calc.IWhole-mouth average calculus index at baseline, Simplified Oral Hygiene Index (OHI-S): Higher value indicates more calculus; 0 = No calculus present; numeric, range: 0-3.0
BL.Pl.IWhole-mouth average plaque index at baseline, Silness-Lowe Gingival Index:Higher value indicates more severe inflammation, 0= normal gingiva, numeric, range: 0.056-3.0
V3.GEWhole-mouth average gingival index at Visit 3, numeric, range: 0.030-3.0
V3..BOPPercentage of sites bleeding on probing at Visit 3, numeric, range: 0.725-100.0, Blank = Missing
V3.PD.avgWhole-mouth average pocket depth at Visit 3 (mm), numeric, range: 1.601-5.5, Blank = Missing
V3..PD.4Percentage of sites with pocket depth greater than or equal to 4mm at Visit 3, numeric, range: 0-83.9, Blank = Missing
V3..PD.5Percentage of sites with pocket depth greater than or equal to 5mm at Visit 3, numeric, range: 0-77.4, Blank = Missing
V3.CAL.avgWhole-mouth average clinical attachment level at Visit 3 (mm), numeric, range: 0.036-3.9, Blank = Missing
V3..CAL.2Percentage of sites with clinical attachment level greater than or equal to 2 mm at visit 3, numeric, range: 0-97.8, Blank = Missing
V3..CAL.3Percentage of sites with clinical attachment level greater than or equal to 3 mm at visit 3, numeric, range: 0-85.7, Blank = Missing
V3.Calc.IWhole-mouth average calculus index at visit 3, numeric, range: 0-2.6, Simplified Oral Hygiene Index (OHI-S): Higher value indicates more calculus; 0 = No calculus present; Blank = Missing
V3.Pl.IWhole-mouth average plaque index at visit 3, numeric, range: 0-2.6, Silness-Lowe Plaque Index: Higher value indicates more abundant plaque; 0 = No plaque in gingival area; Blank = Missing
V5.GEWhole-mouth average gingival index at visit 5, numeric, range: 0.190-2.7, Silness-Lowe Gingival Index: Higher value indicates more severe inflammation; 0 = Normal gingiva; Blank = Missing
V5..BOPPercentage of sites bleeding on probing at visit 5, numeric, range: 3.571-100.0, Blank = Missing
V5.PD.avgWhole-mouth average pocket depth at visit 5, numeric, range: 1.536-5.4, Blank = Missing
V5..PD.4Percentage of sites with pocket depth greater than or equal to 4mm at Visit 5, numeric, range: 0-83, Blank = Missing
V5..PD.5Percentage of sites with pocket depth greater than or equal to 5mm at Visit 3, numeric, range: 0-75.6, Blank = Missing
V5.CAL.avgWhole-mouth average clinical attachment level at visit 5 (mm), numeric, range: 0.018-4.3, Blank = Missing
V5..CAL.2Percentage of sites with clinical attachment level greater than or equal to 2 mm at visit 5, numeric, range: 0.000-99.2, Blank = Missing
V5..CAL.3Percentage of sites with clinical attachment level greater than or equal to 3 mm at visit 5, numeric, range: 0.000-85.0, Blank = Missing
V5.Calc.IWhole-mouth average calculus index at visit 5, numeric, range: 0.0-2.6, Simplified Oral Hygiene Index (OHI-S): Higher value indicates more calculus; 0 = No calculus present; Blank = Missing
V5.Pl.IWhole-mouth average plaque index at visit 5, numeric, range: 0.0-2.5, Silness-Lowe Plaque Index: Higher value indicates more abundant plaque; 0 = No plaque in gingival area; Blank = Missing
N.PAL.sitesNumber of sites for which attachment loss increased from baseline by greater than or equal to 3 mm, numeric, range: 0-33, 0 = No sites; Blank = Missing
Birth.outcomeBirth outcome, factor, Elective abortion; Live birth; Lost to FU = Lost to Follow-Up; Non-live birth = Stillbirth or spontaneous abortion; There are no missing data
Preg.ended...37.wkWhether the pregnancy ended before gestational age 37 weeks (259 days), factor, Yes; No; Blank = Lost to Follow-Up
GA.at.outcomeGestational age at end of pregnancy, or at mother's last follow-up visit if lost to follow-up, numeric, range: 103-302
BirthweightInfant birth weight at time of birth, abstracted from obstetrical records (grams), numeric, range: 101-5160, Blank = Missing
Fetal.congenital.anomalyFetal/congenital anomaly identified at birth or during pregnancy?, factor, Yes; No; There are no missing data
Apgar1Apgar score, a summary of a newborn infant's 'Appearance, Pulse, Grimace, Activity, Respiration' at 1 minute Score interpretation: less than or equal to 3: Critically low 4-6: Fairly low greater than or equal to 7: Normal, numeric, range: 0-10, Blank = Missing
Apgar5Apgar score at 5 minutes, numeric, range: 0-10, Blank = Missing
Any.SAE.Whether participant experienced any serious adverse events (e.g. lost pregnancies) factor, Yes; No; There are no missing data
GA...1st.SAEGestational age of first SAE (serious adverse event), integer, range: 96-467, 259 = No SAE (variable 76 must = No); There are no missing data
Bact.vagWhether mother had bacterial vaginosis during pregnancy, factor, Yes; No; Blank = Missing
Gest.diabWhether mother had gestational diabetes during pregnancy, factor, Yes; No; Blank = Missing
OligoWhether mother had oligohydramnios during pregnancy, factor, Yes; No; Blank = Missing
PolyhydWhether mother had polyhydramnios during pregnancy, factor, Yes; No; Blank = Missing
GonorrheaWhether mother had gonorrhea during pregnancy, factor, Yes; No; Blank = Missing
ChlamydiaWhether mother had chlamydia during pregnancy, factor, Yes; No; Blank = Missing
Strep.BWhether mother had strep B colonization during pregnancy, factor, Yes; No; Blank = Missing
Traumatic.InjWhether mother had a traumatic injury during pregnancy, factor, Yes; No; Blank = Missing
UTIWhether mother had a urinary tract infection during pregnancy, factor, Yes; No; Blank = Missing
Pre.eclampWhether mother had pre-eclampsia, a pregnancy condition characterized by high blood pressure and associated with fetal growth restriction during pregnancy, factor, Yes; No; Blank = Missing
Mom.HIV.statusHIV status of mother during pregnancy, factor, Yes = HIV-positive; No = HIV-negative or unknown (question answered but HIV status at delivery not recorded); Blank = Missing (question not answered)
BL.Anti.infDid participant report use of antiinflammatory medication at or less than 6 months before baseline?, integer, 0 = No; 1 = Yes; There are no missing data
BL.CorticoDid participant report use of corticosteroids at or less than 6 months before baseline?, integer, 0 = No; 1 = Yes; There are no missing data
BL.AntibioDid participant report use of antibiotics at or less than 6 months before baseline?, integer, 0 = No; 1 = Yes; There are no missing data
BL.Bac.vagDid participant report use of bacterial vaginitis treatments at or less than 6 months before baseline?, integer, 0 = No; 1 = Yes; There are no missing data
V3.Anti.infDid participant report use of antiinflammatory medication between baseline and visit 3?,integer, 0 = No; 1 = Yes; There are no missing data
V3.CorticoDid participant report use of corticosteroids between baseline and visit 3?, integer, 0 = No; 1 = Yes; There are no missing data
V3.AntibioDid participant report use of antibiotics between baseline and visit 3?, integer, 0 = No; 1 = Yes; There are no missing data
V3.Bac.vagDid participant report use of bacterial vaginitis treatments between baseline and visit 3?, integer, 0 = No; 1 = Yes; There are no missing data
V5.Anti.infDid participant report use of antiinflammatory medication between visit 3 and visit 5?, integer, 0 = No; 1 = Yes; There are no missing data
V5.CorticoDid participant report use of corticosteroids between visit 3 and visit 5?, integer, 0 = No; 1 = Yes; There are no missing data
V5.AntibioDid participant report use of antibiotics between visit 3 and visit 5?, integer, 0 = No; 1 = Yes; There are no missing data
V5.Bac.vagDid participant report use of bacterial vaginitis treatments between visit 3 and visit 5?, integer, 0 = No; 1 = Yes; There are no missing data
X..Vis.AttVisit attendance: Number of study visits attended AFTER baseline, integer, Range: 0-5
X..Vis.EligNumber of visits for which participant was eligible (could become ineligible after miscarriage or early delivery), integer, Range: 0-5
X1st.Miss.VisFirst missed visit. No one missed the baseline visit, so this variable takes values 2, 3, 4, 5, 6, and 100 (no eligible visits missed), integer, Range: 2-6, 100
OAA1Serum IgG (immunoglobulin) antibodies to A. actinomycetemcomitans at baseline, factor (actually numeric or missing), dot(.) = Missing
OCR1Serum IgG (immunoglobulin) antibodies to C. rectus at baseline, factor (actually numeric or missing), dot(.) = Missing
OFN1Serum IgG (immunoglobulin) antibodies to F. nucleatum at baseline, factor (actually numeric or missing), dot(.) = Missing
OPG1Serum IgG (immunoglobulin) antibodies to P. gingivalis at baseline, factor (actually numeric or missing), dot(.) = Missing
OPI1Serum IgG (immunoglobulin) antibodies to P. intermedia at baseline, factor (actually numeric or missing), dot(.) = Missing
OTD1Serum IgG (immunoglobulin) antibodies to T. denticola at baseline, factor (actually numeric or missing), dot(.) = Missing
OTF1Serum IgG (immunoglobulin) antibodies to T. forsythus at baseline, factor (actually numeric or missing), dot(.) = Missing
OCRP1Serum measure for C-reactive protein (CRP) at baseline, factor (actually numeric or missing), dot(.) = Missing
O1B1Serum measure for Interleukin(IL)-1b at baseline, factor (actually numeric or missing), dot(.) = Missing
O61Serum measure for Interleukin(IL)-6 at baseline, factor (actually numeric or missing), dot(.) = Missing
O81Serum measure forInterleukin(IL)-8 at baseline, factor (actually numeric or missing), dot(.) = Missing
OPGE21Serum measure for Prostaglandin E2 at baseline, factor (actually numeric or missing), dot(.) = Missing
OTNF1Serum measure for tumor necrosis factor (TNF)-alpha at baseline, factor (actually numeric or missing), dot(.) = Missing
OMMP91Serum measure for gelatinase (MMP9) at baseline, factor (actually numeric or missing), dot(.) = Missing
ETXU_CAT1Serum endotoxin level at baseline, factor (actually numeric or missing), dot(.) = Missing
OFIBRIN1Serum measure for fibrinogen at baseline, factor (actually numeric or missing), dot(.) = Missing
OAA5Serum IgG (immunoglobulin) antibodies to A. actinomycetemcomitans at visit 5, factor (actually numeric or missing), dot(.) = Missing
OCR5Serum IgG (immunoglobulin) antibodies to C. rectus at visit 5, factor (actually numeric or missing), dot(.) = Missing
OFN5Serum IgG (immunoglobulin) antibodies to F. nucleatum at visit 5, factor (actually numeric or missing), dot(.) = Missing
OPG5Serum IgG (immunoglobulin) antibodies to P. gingivalis at visit 5, factor (actually numeric or missing), dot(.) = Missing
OPI5Serum IgG (immunoglobulin) antibodies to P. intermedia at visit 5, factor (actually numeric or missing), dot(.) = Missing
OTD5Serum IgG (immunoglobulin) antibodies to T. denticola at visit 5, factor (actually numeric or missing), dot(.) = Missing
OTF5Serum IgG (immunoglobulin) antibodies to T. forsythus at visit 5, factor (actually numeric or missing), dot(.) = Missing
OCRP5Serum measure for C-reactive protein (CRP) at visit 5, factor (actually numeric or missing), dot(.) = Missing
O1B5Serum measure for Interleukin(IL)-1b at visit 5, factor (actually numeric or missing), dot(.) = Missing
O65Serum measure forInterleukin(IL)-6 at visit 5, factor (actually numeric or missing), dot(.) = Missing
O85Serum measure forInterleukin(IL)-8 at visit 5, factor (actually numeric or missing), dot(.) = Missing
OPGE25Serum measure for Prostaglandin E2 at visit 5, factor (actually numeric or missing), dot(.) = Missing
OTNF5Serum measure for tumor necrosis factor (TNF)-alpha at visit 5, factor (actually numeric or missing), dot(.) = Missing
OMMP95Serum measure for gelatinase (MMP9) at visit 5, factor (actually numeric or missing), dot(.) = Missing
ETXU_CAT5Serum endotoxin level at visit 5, factor (actually numeric or missing), dot(.) = Missing
OFIBRIN5Serum measure for fibrinogen at visit 5, factor (actually numeric or missing), dot(.) = Missing
BL.DNATotal amount of bacterial DNA extracted from plaque as a measure of total bacterial concentration at baseline (ng/mL), numeric, range: 0-5750.0
BL.UnivCount of all bacteria detected by universal primer at baseline, numeric, range: 1,890,000-1,070,000,000, Blank = Missing
BL.AACount of A. actinomycetemcomitans bacteria at baseline, numeric, range: 0-7,970,000, Blank = Missing
BL.PGCount of P. gingivalis bacteria at baseline, numeric, range: 0-167,000,000, Blank = Missing
BL.TDCount of T. denticola bacteria at baseline, numeric, range: 0-50,500,000, Blank = Missing
BL.TFCount of T. forsythus bacteria at baseline, numeric, range: 0-40,200,000, Blank = Missing
BL.PICount of P. intermedia bacteria at baseline, numeric, range: 0-87,500,000, Blank = Missing
BL.CRCount of C. rectus bacteria at baseline, numeric, range: 0-32,600,000, Blank = Missing
BL.FNCount of F. nucleatum bacteria at baseline, numeric, range: 67,300- 152,000,000, Blank = Missing
BL.S7Sum of the 7 species-specific bacterial counts (variables 138-144) at baseline, rounded to 3 significant figures, numeric, range: 87,000-391,000,000, Blank = Missing
V5.DNATotal amount of bacterial DNA extracted from plaque as a measure of total bacterial concentration at visit 5 (ng/mL), numeric, range: 0-5750.0
V5.UnivCount of all bacteria detected by universal primer at visit 5, numeric, range: 1,890,000-1,070,000,000, Blank = Missing
V5.AACount of A. actinomycetemcomitans bacteria at visit 5, numeric, range: 0-40,200,000, Blank = Missing
V5.PGCount of P. gingivalis bacteria at visit 5, numeric, range: 0-40,200,000, Blank = Missing
V5.TDCount of T. forsythus bacteria at visit 5, numeric, range: 0-40,200,000, Blank = Missing
V5.TFCount of T. forsythus bacteria at visit 5, numeric, range: 0-40,200,000, Blank = Missing
V5.PICount of P. intermedia bacteria at visit 5, numeric, range: 0-87,500,000, Blank = Missing
V5.CRCount of C. rectus bacteria at visit 5, numeric, range: 0-32,600,000, Blank = Missing
V5.FNCount of F. nucleatum bacteria at visit 5, numeric, range: 67,300- 152,000,000, Blank = Missing
V5.S7Sum of the 7 species-specific bacterial counts (variables 138-144) at visit 5, rounded to 3 significant figures, numeric, range: 87,000-391,000,000, Blank = Missing
BL..AAPercent of A. actinomycetemcomitans out of total DNA (variable 146) at baseline, numeric, range: 0-8.9, Blank = Missing
BL..PGPercent of P. gingivalis out of total DNA at baseline, numeric, range: 0-37.3, Blank = Missing
BL..TDPercent of T. denticola out of total DNA at baseline, numeric, range: 0-13.2, Blank = Missing
BL..TFPercent of T. forsythus out of total DNA at baseline, numeric, range: 0-17.7, Blank = Missing
BL..PIPercent of P. intermedia out of total DNA at baseline, numeric, range: 0-46.3, Blank = Missing
BL..CRPercent of C. rectus out of total DNA at baseline, numeric, range: 0-10.5, Blank = Missing
BL..FNPercent of F. nucleatum out of total DNA at baseline, numeric, range: 0.330-63.2, Blank = Missing
BL..S7Sum of the percents for the 7 species (AA, PG, TD, TF, PI, CR, and FN) at baseline, numeric, range: 0.420-86.3, Blank = Missing
V5..AAPercent of A. actinomycetemcomitans out of total DNA at visit 5, numeric, range: 0-16.1, Blank = Missing
V5..PGPercent of P. gingivalis out of total DNA at visit 5, numeric, range: 0-59.7, Blank = Missing
V5..TDPercent of T. denticola out of total DNA at visit 5, numeric, range: 0-20.5, Blank = Missing
V5..TFPercent of T. forsythus out of total DNA at visit 5, numeric, range: 0-19.3, Blank = Missing
V5..PIPercent of P. intermedia out of total DNA at visit 5, numeric, range: 0-40.7, Blank = Missing
V5..CRPercent of C. rectus out of total DNA at visit 5, numeric, range: 0-14.6, Blank = Missing
V5..FNPercent of F. nucleatum out of total DNA at visit 5, numeric, range: 0-49.9, Blank = Missing
V5..S7Sum of the percents for the 7 species (AA, PG, TD, TF, PI, CR, and FN) at visit 5, numeric, range: 2.560-80.8, Blank = Missing
Randomized Clinical Trial on the Effect of Treatment of Maternal Periodontal Disease Can Reduce Preterm Birth Risk.
Maternal periodontal disease has been linked in observational studies to preterm birth (< 37 weeks) and low birth weight (< 2500 g) outcomes. The Obstetrics and Periodontal Therapy study was a multi-center randomized trial evaluating the effect of nonsurgical periodontal treatment intervention on preterm birth, comparing outcomes of women treated before 21 weeks gestation (treatment) to those treated after delivery (control).
Preterm birth, defined as delivery before 37 weeks of gestation, is a growing problem. In some cases, preterm birth can lead to infant death; in others, its consequences may include neurodevelopmental disabilities, cognitive impairment, and/or respiratory disorders in the child. Many risk factors for preterm birth have already been identified, including maternal age, drug use, and diabetes. However, such factors are exhibited in only about half of preterm birth mothers, highlighting a need to expand our understanding of what contributes to preterm birth risk.
Several observational studies have suggested an association between maternal periodontal disease and preterm birth. Periodontal disease is an inflammatory condition characterized by the destruction of tissue and/or bone around the teeth. A major component of periodontal disease is oral colonization by gram-negative bacteria; systemic release of cytokines and/or lipopolysaccharides from these bacteria may impact fetal condition.
Inoculation of the periodontal pathogen P. gingivalis into pregnant animals does have a dose-dependent effect on birth weight and preterm birth signaling, but no such causal link has been shown in humans, only some associations. Though not definitive, the possibility of a significant relationship raises the question of whether treatment of maternal periodontal disease can decrease preterm birth risk.
823 participants enrolled at 4 centers underwent stratified randomization, resulting in 413 women assigned to the treatment group and 410 to control. All participants were 13-16 weeks pregnant at time of randomization (baseline/visit 1) and went on to attend monthly follow-up visits defined as visits 2, 3, 4, and 5 corresponding to gestational age ranges of 17-20, 21-24, 25-28, and 29-32 weeks.
The treatment group received periodontal treatment, oral hygiene instruction, and tooth polishing at their follow-ups, while those assigned to control underwent only brief oral exams. Data collection occurred at visits 1 (baseline), 3, and 5. The primary outcome of interest is gestational age at end of pregnancy. Additional outcomes include birthweight, clinical measures of periodontal disease, and various microbiological and immunological outcomes.
Statistical analyses were carried out on an intent-to-treat basis. Gestational age can be thought of as 'time until end of pregnancy,' for which certain survival analysis methods would be appropriate. The study used a log-rank test stratified by center to compare time until end of pregnancy for treatment and control groups.
A semiparametric proportional hazards model was also used for this purpose and incorporated maternal risk factors as predictors. For the study's main analyses, gestational age was censored at 37 weeks (259 days) because the interest was in extending pregnancies that would otherwise end pre-term, not extending pregnancies generally.
Though not used in the study itself, logistic regression is another method that could be applied: for example, to gestational age, dichotomized as 'preterm' or 'not preterm' according to a gestational age cutoff, or to birthweight dichotomized as 'low' or 'high' at the 2500 g or other cutoff (2500 g would be in keeping with the World Health Organization's definition for low birth weight). Changes in clinical measures of periodontal disease from baseline to visits 3 or 5 could be analyzed using mixed effects linear models. The dataset also features a number of baseline characteristics, which could be compared in treatment and control groups via Student t-tests, Wilcoxon rank sum tests, Fisher's exact tests or Pearson's chi-square tests, as appropriate.
The nonsurgical periodontal treatment involving scaling and root planing induced significant improvements in periodontal health. The study did not however find a significant relation between periodontal treatment and preterm birth risk. The results of this study were published in 2006 by Michalowicz et al., 'Treatment of periodontal disease and the risk of preterm birth', in The New England Journal of Medicine. The Obstetrics and Periodontal Therapy Dataset contains the data used in this study.
The obstetrics and periodontal therapy dataset was contributed by Dr. Ann Brearley, Assistant Professor, Division of Biostatistics, School of Public Health, University of Minnesota and her colleagues. Please refer to this resource as: Meredith Hyun, James S. Hodges and Ann M. Brearley, 'Obstetrics and Periodontal Therapy Dataset', TSHS Resources Portal (2019). Available at https://www.causeweb.org/tshs/obstetrics-and-periodontal-therapy/.
Michalowicz et al., 'Treatment of periodontal disease and the risk of preterm birth', N Engl J Med 2006; 355:1885-1894. DOI: 10.1056/NEJMoa062249
Results of a randomized, placebo-controlled trial of sulindac in the reduction of colonic polyps in Familial Adenomatous Polyposis (FAP) (more details available below the variable definitions).
polypspolyps
A data frame with 22 observations and 7 variables
id number for each participant; type: character
participant sex, levels: female, male; type: factor
age in years; type: numeric
number of colonic polyps at baseline; type: numeric
treatment assignment, levels: sulindac, placebo; type: factor
number of colonic polyps at 3 months; type: numeric
number of colonic polyps at 12 months; type: numeric
FAP is an inherited condition caused by mutations in the APC (Adenomatous Polyposis Coli) gene that leads to early and frequent formation of precancerous polyps of the colon at a young age, and invariably leads to the development of colon cancer at a young age.
Early, frequent surveillance colonoscopy and polyp removal is helpful, but this study examined whether there is a beneficial effect of preventive medical therapy with the nonsteroidal pain reliever, sulindac, versus placebo in a RCT vs placebo in 22 participants, with polyp number measured (via colonoscopy) at baseline, 3 months, and 12 months after starting the study drug. Note that one subject did not return for the 12 month colonoscopy.
This data set is from a study published in 1993 in the New England Journal of Medicine,
F. M. Giardiello, S. R. Hamilton, A. J. Krush, S. Piantadosi, L. M. Hylind, P. Celano, S. V. Booker, C. R. Robinson and G. J. A. Offerhaus (1993), Treatment of colonic and rectal adenomas with sulindac in familial adenomatous polyposis. New England Journal of Medicine, 328(18), 1313-1316.
This dataset is derived from and improved upon from the HSAUR package.
Results of a randomized, 6-arm comparator-controlled trial of 6 interventions to treat scurvy in 12 disabled seamen, as reported by James Lind in 1757 (more details available below the variable definitions).
scurvyscurvy
A data frame with 12 observations and 8 variables
invented id number for each participant; type: character
assigned treatment, levels: cider, dilute_sulfuric_acid, vinegar, sea_water, citrus, purgative_mixture; type: factor
details on daily dosing and schedule; type: character
rating of symptom of rotting of gums; type: factor, with levels: 0=none, 1=mild, 2=moderate, 3=severe
rating of symptom of skin sores; type: factor, with levels: 0=none, 1=mild, 2=moderate, 3=severe
rating of symptom of weakness of the knees (ability to stand); type: factor, with levels: 0=none, 1=mild, 2=moderate, 3=severe
rating of symptom of lassitude (generalized weakness); type: factor, with levels: 0=none, 1=mild, 2=moderate, 3=severe
dichotomous fitness for duty as a seaman; type: factor: 0_no, 1_yes
Scurvy was a common affliction of seamen on long voyages, leading to mouth sores, skin lesions, weakness of the knees, and lassitude. Scurvy could be fatal on long voyages. James Lind reported the treatment of 12 seamen with scurvy in 1757, in A Treatise on the Scurvy in Three Parts. This 476 page bloviation can be found scanned to the Google Books website A Treatise on the Scurvy. Pages 149-153 are a rare gem among what can be generously described as 400+ pages of evidence-free blathering, and these 4 pages may represent the first report of a controlled clinical trial.
Lind was the ship's surgeon on board the HMS Salisbury, and had a number of scurvy-affected seamen at his disposal. Many remedies had been described and advocated for, with no more than anecdotal evidence. On May 20, 1747, Lind decided to try the 6 available therapies at his disposal in a comparative study in 12 affected seamen. He selected 12 with roughly similar severity, with notable skin and mouth sores, weakness of the knees, and significant lassitude, making them unfit for duty. They each received the standard shipboard diet of gruel and mutton broth, supplemented with occasional biscuits and puddings. Each treatment was a dietary supplement (including citrus fruits) or a medicinal.
This data frame was reconstructed from Lind's account as recorded on these 4 pages, with his estimates of severity translated to a 4 point Likert scale (0-3) for each of the symptoms he described at his chosen endpoint on day 6. A fanciful study_id variable was added, along with detailed descriptions of the dosing schedule of each treatment.
Of note, there is some dispute about whether this was truly the first clinical trial, or whether it actually happened. See link about the historical debate. Lind reported that the seamen treated with 2 lemons and an orange daily did best, followed by those treated with cider. Those treated with elixir of vitriol only had improvement in mouth sores. One imagines that acidic substances (like dilute sulfuric acid, vinegar, cider, and citrus fruits) might have been rather painful on these mouth sores. Unfortunately, the burial of 4 valuable pages of data in 476 pages of noise, a publication delay of 10 years, and Lind's half-hearted conclusions, meant that it took until 1795 before the British Navy mandated daily limes for seamen.
This data set is faithfully reconstructed from a report published in 1757 as A Treatise on the Scurvy in 3 Parts, by James Lind, pp. 149-153, and you can find a scan of the source document that you can read yourself on Google Books here.
This study evaluated gastric emptying, small bowel transit time, and total intestinal transit time in 8 critically ill trauma patients. These data were compared with those obtained in 87 healthy volunteers from a separate trial. Data were obtained with a motility capsule that wirelessly transmitted pH, pressure, and temperature to a recorder attached to each subject's abdomen. Transit times were available for almost all patients, however, pH, pressure and temperature data is missing for all critically ill patients and sparsely missing for the healthy volunteers (more details available below the variable definitions)
smartpillsmartpill
A data frame with 95 obsrvations and 22 variables
GroupStudy group, numeric, 0 = Critically Ill Trama Patient, 1 = Healthy Volunteer
GenderGender, numeric, range: 0 = Female, 1 = Male
RaceRace, numeric, 1 = White, 2 = Black, 3 = Asian/Pacific Islander, 4 = Hispanic, 5 = Other
HeightHeight (centimeters), numeric, range: 132.1-193.0
WeightWeight (kilograms), numeric, range: 44.9-127.0
AgeAge (years), numeric, range: 18.0-72.0
GE.TimeGastric Emptying Time is time from ingestion to gastric emptying (hours), numeric, range: 1.7-74.3
SB.TimeSmall Bowel Transit Time is time from gastric emptying to ileocecal junction (hours), numeric, range: 1.8-13.8
C.TimeColonic Transit Time is time from ileocecal junction to body exit (hours), numeric, range: 0.7-118.9
WG.TimeWhole Gut Time is time from ingestion to body exit (hours), numeric, range: 6.0-816.0
S.ContractionsStomach contractions are counted if the peak amplitude of the contraction is over 10 mmHg and under 300 mmHg, numeric, range: 47.0-1665.0
S.Sum.of.AmplitudesStomach sum of amplitudes (mm Hg), numeric, range: 655.6-33800.3
S.Mean.Peak.AmplitudeStomach mean peak amplitude is the sum of amplitudes divided by number of contractions (mm Hg), numeric, range: 4.6-43.4
S.Mean.pHStomach mean pH is the average pH over the whole recording time in the stomach with normal ~ 1.5-3.5, numeric, range: 1.5-5.9
SB.ContractionsSmall Bowel contractions are counted if the peak amplitude of the contraction is over 10 mmHg and under 300 mmHg, numeric, range: 223.0-2375.0
SB.Sum.of.AmplitudesSmall Bowel sum of amplitudes (mm Hg), numeric, range:3899.4-41122.5
SB.Mean.Peak.AmplitudeSmall Bowell mean peak amplitude is the sum of amplitudes divided by number of contractions (mm Hg), numeric, range: 15.0-27.9
SB.Mean.pHSmall Bowel mean pH is the average pH over the whole recording time in the small bowel, normal ~ 6-7.4, numeric, range: 4.7-8.6
Colon.ContractionsColon contractions are counted if the peak amplitude of the contraction is over 10 mmHg and under 300 mmHg, numeric, range: 41.0-2672.0
Colon.Sum.of.AmplitudesColon sum of amplitudes (mm Hg), numeric, range:1872.6-117707.5
C.Mean.Peak.AmplitudeColon mean peak amplitude is the sum of amplitudes divided by number of contractions (mm Hg), numeric, range: 32.8- 64.2
C.Mean.pHColon mean pH is the average pH over the whole recording time in the colon, normal ~ 5-7-6.7, numeric, range: 3.9-8.1
The Smart Pill dataset was contributed by Dr. Amy Nowacki, Associate Professor, Cleveland Clinic. Please refer to this resource as: Amy S. Nowacki, 'Smart Pill Dataset', TSHS Resources Portal (2017). Available at https://www.causeweb.org/tshs/smart-pill/.
Delayed gastric emptying is a well-known problem in critically ill patients and is associated with feeding disturbances and inadequate nutrition. However, evaluating gastrointestinal function remains challenging in critically ill patients who are mechanically ventilated. Many tests that are practical and accurate under standardized, controlled conditions often fail in the critical care setting. For example, the consensus recommendations for gastric emptying scintigraphy are impractical in intubated patients because they recommend low-fat, egg white meal with imaging at 0, 1, 2, and 4 hours after meal ingestion. Another test, the lactulose hydrogen breath test, relies on prompt bacterial breakdown of lactulose in the colon; however, changes in bacterial flora - which are presumably common in critical care patients - can produce false transit times.
The 13C-octanoic acid breath test was reported as successful when used bedside to measure gastric emptying. However, manometry only assesses the upper gastrointestinal function, mainly esophagus, stomach, and proximal small bowel. Finally, video capsule technology has been used to determine small bowel transit time and pathomorphology in critically ill patients, although inadequate battery lifespan of the capsule (approximately 8-10 hours) could prevent complete examination in some cases.
An alternative technique, wireless capsule technology, may be useful for evaluating gastrointestinal motility in critical care patients. A newly developed motility capsule for assessing gastric emptying in patients with suspected gastroparesis has been available since 2006. It is a wireless capsule that transmits pH, pressure, and temperature.
This study describes the first use of a novel motility capsule to compare gastric emptying and small bowel transit times in critically ill trauma patients with intracranial hemorrhage with times recorded previously in healthy volunteers. Secondly, this study compares critically ill patients and volunteers on whole-gut transit time.
Rauch et al. 'Use of Wireless Utility Capsule to Determine Gastric Emptying and Small Intestinal Transit Times in Critically Ill Trauma Patients'. Journal of Critical Care 2012; 27(5): 534.e7-534.e12.
Results of a randomized, placebo-controlled, prospective 2-arm trial of streptomycin 2 grams daily (arm A2) vs. placebo (arm A1) to treat tuberculosis in 107 young patients, as reported by the Streptomycin in Tuberculosis Trials Committee in 1948 in the British Medical Journal (more details available below the variable definitions).
strep_tbstrep_tb
A data frame with 107 observations and 13 variables
invented id number for each participant; type: character
assigned treatment arm, Streptomycin or Control; type: factor
grams, dose of Streptomycin: numeric, 0, 1, or 2 grams
grams, dose of PAS (Para-Amino-Salicylate): numeric, 5, 10, or 20 grams. Note that no one in this intial study (study A) received PAS. This was added for combination therapy in studies B and C, as reported in 1952.
gender, dichotomous (this was in 1948); type: factor, with levels: M = Male, F= Female
Condition of the Patient at Baseline, 3 levels, 1_Good, 2_Fair, 3_Poor; type: factor
temperature at baseline in degrees fahrenheit or celsius, but categorized into 4 levels (afebrile level apparently were cases not measured with a thermometer): factor, with levels: 1_<99F/37.2C, 2_99-99.9F/37.3-37.8C, 3_100-100.9F/37.9-38.3C, 4_>=101F+/38.4C
Erythrocyte Sedimentation Rate in mm per hour, categorized into 4 levels, from 0-51+ mm per hour; type: factor, with levels: 1_0-10, 2_11-20, 3_21-50, 4_51+
dichotomous presence of cavitation on the baseline chest x-ray; type: factor: 0_no, 1_yes
streptomycin resistance after 6 months of therapy, measured on a 0-100+ scale, categorized into 3 levels - sensitive, moderate, and resistant; type: factor: 1_sens_0-8, 2_mod_8-99, 3_resist_100+
Likert score rating of radiologic response on chest x-ray at 6 months; type: factor: 1_Death, 2_Considerable_deterioration, 3_Moderate_deterioration, 4_No_change, 5_Moderate_improvement, 6_Considerable_improvement
Likert score numeric rating of radiologic response on chest x-ray at 6 months; type: numeric: 1-6, from Death to Considerable Improvement
Dichotomous outcome of improvement (equal to rad_num of 5-6); type: logical, TRUE or FALSE. 55 of the 107 participants were improved.
The Streptomycin for Tuberculosis trial in 1948 was considered the first modern randomized, placebo-controlled clinical trial, which could be done in part because there were very limited supplies of streptomycin in the UK after World War II.
This publication seems a bit primitive today, without standard features like a proper Table 1, and some creative use of graphs to display baseline characteristics of the study sample
More strikingly, there is no ethics committee approval, or consent.
You can read the pdf of the original journal article at Streptomycin in TB Study.
This was the first of a series of 3 trials, in which the initial effectiveness of Streptomycin was established, but rapid resistance developed, and significant side effects occurred at a dose of 2 grams of streptomycin. This type of resistance also occurred with another new anti-tubercular therapy at the time, PAS (Para-Amino-Salicylate). Subsequent trials B and C evaluated different doses and combinations of Streptomycin and PAS, and were published together in 1952 in the BMJ, with the pdf available here 1952 Three Streptomycin in TB Studies Summarized.
Commentary on the conduct of these trials from one of the MD investigators can be found at MD Clinical Trialist Commentary.
Commentary on the design and analysis of these trials from statistician A. Bradford Hill can be found at Statistican Commentary.
This data set is reconstructed to the best of my ability from the paper in the British Medical Journal from 1948, entitled, Streptomycin Treatment of Pulmonary Tuberculosis, pages 769-782 in the October 30, 1948 edition, authored by the Streptomycin in Tuberculosis Trials Committee. You can find the pdf at Streptomycin in TB.
This data set contains 103 patients who were scheduled to undergo an upper extremity procedure suitable for supraclavicular anesthesia. Patients were randomly assigned to either (1) combined group-ropivacaine and mepivacaine mixture; or (2) sequential group-mepivacaine followed by ropivacaine. A number of demographic and post-op pain medication variables (fentanyl, alfentanil, midazolam) were collected. The primary outcome is time to 4-nerve sensory block onset. The dataset is cleaned and relatively complete. There are no outliers or data problems (more details available below the variable definitions).
supraclavicularsupraclavicular
A data frame with 103 observations and 17 variables
subjectSubject ID, numeric, range: 1-103
groupAnesthetic group, numeric, 1 = Mixture; 2 = Sequential
genderGender, numeric, 1 = Male; 0= Female
bmiBody mass index (kg/m^2), numeric, range:19-43.5
ageAge (years), numeric, range:18-74
fentanylFentanyl pain medication (micrograms), numeric, range: 0-250.0
alfentanilAlfentanil pain medication (milligrams), numeric, range: 0-4.3
midazolamMidazolam hypnotic-sedative medication, numeric, range: 0-9.0
onset_sensoryTime to 4 nerve sensory block onset or, if onset_sensory block failed the observed worst outcome of minutes for any patient (50 minutes), numeric, range: 0-50.0
onset_first_sensoryTime to first sensory block in minutes, or if block failed, a value of 15 minutes, numeric, range: 6-15.0
onset_motorTime to complete motor block or, if motor block failed, the observed worst outcome of minutes for any patient (50 minutes), numeric, range: 1-50.0
nerve_block_censorblock failed, numeric, 0 = nerve block succeeded, 1 = block failed (censored)
med_durationTime from the onset of 4 nerve sensory block until the first request for an analgesic medication (hours), numeric, range: 0-48.0
med_censorPatients who did not take an analgesic were censored at 48 hours, numeric, 0 = nerve succeeded, 1 = block failed (censored)
vps_restMaximum postop verbal pain score (at rest), on 11 point Likert scale (0-10), numeric, range: 0-10
vps_movementMaximum postop verbal pain score (with movement), on 11 point Likert scale (0-10), numeric, range: 0-10
opioid_totalTotal opioid consumption in milligrams, numeric, range: 0-225.0
The choice of anesthetic technique combined with a suitable plan for postoperative analgesia can facilitate early discharge, improve patient comfort, and increase overall satisfaction. Patients having painful procedures who undergo general anesthesia have a 2- to 5-fold greater risk of unplanned overnight admissions compared with those having regional anesthesia. Regional anesthetic techniques and peripheral nerve blocks are especially favored for surgeries on the extremities. Both rapid onset of the block and prolonged postoperative analgesia are desired characteristics of regional anesthesia.
The choice of local anesthetics or combinations thereof can greatly influence the effectiveness of the block, onset time, duration of postoperative analgesia, need for opioid use, and patient satisfaction. Mepivacaine and ropivacaine are commonly used in peripheral nerve blocks, their drawbacks being a short duration with 1.5% mepivacaine and a delayed onset with 0.5% ropivacaine. An ideal local anesthetic with high potency, low toxicity, rapid onset, and prolonged duration does not exist yet. Investigators have therefore tried mixtures of local anesthetics in an attempt to combine their advantages with conflicting results. A potential problem is that mixing drugs dilutes the effects of each. Thus, a mixture of a rapid-onset drug such as mepivacaine with a long-acting one such as ropivacaine may well result in slower onset than mepivacaine alone and shorter duration of action than ropivacaine alone. In contrast, sequential administration of the same amounts of the same drugs may preserve the desirable features of each.
Objective: This study investigates whether sequential supraclavicular injection of 1.5% mepivacaine followed 90 seconds later by 0.5% ropivacaine provides a quicker onset and a longer duration of analgesia than an equidose combination of the 2 local anesthetics.
These are data from a study by Roberman et al. 'Combined Versus Sequential Injection of Mepivacaine and Ropivacaine for Supraclavicular Nerve Blocks'. Reg Anesth Pain Med 2011; 36:145-50.
Results of a Cohort Study of the Pharmacokinetics of Oral Theophylline, with plasma concentrations over time (more details available below the variable definitions).
theophtheoph
A data frame with 132 observations and 5 variables
subject id number for each participant; type: ordinal factor
Weight in kilograms; type: double
Dose in milligrams per kilogram; type: double
Time from initial dose in hours; type: double
Concentration of theophylline in the plasma in micrograms per milliliter' type: double
This data set is from a pharmacokinetic study of oral dosing of the anti-asthma medication, theophylline, in 12 subjects over 25 hours, published By Dr. Robert A. Upton around 1980. The original publication, if any, is unclear and not cited. These data were used in a package named nlme, and reported in Boeckmann, A.J., et al.Dr. Upton did publish several papers on theophylline pharmacokinetics around 1980-1984, and these data could have been from one of these.
Theophylline is an methylxanthine anti-asthma medication, which acts as a bronchodilator, with secondary effects to strengthen diaphragm contraction, reduce pulmonary artery pressures, and reduce mast cell release. It can be administered by the intravenous, oral, or rectal suppository routes.
Each subject in this Study (oral route) received a single oral dose of theophylline.
Blood samples were taken at frequent intervals over the first 25 hours after dosing, and the quantity of theophylline in the plasma at each time point was measured in micrograms per milliliter.
Unfortunately, the theophylline plasma level in blood varies considerably between patients, because of differences in drug clearance, which is affected by body mass, age, smoking, liver and heart function, and viral infections. To complicate this drug further, it has important interactions with a number of other common medicines which can increase or decrease the drug level. Each subject in this study received a single oral dose of 300 mg of theophylline, which has been converted to a milligrams per kilogram dose. Blood samples were taken at frequent intervals over the next 25 hours after dosing, and the quantity of theophylline in the plasma at each time point was measured in micrograms per milliliter of plasma.
Boeckmann, A. J., Sheiner, L. B. and Beal, S. L. (1994), NONMEM Users Guide: Part V, NONMEM Project Group, University of California, San Francisco. Note that the original data collector, Robert A. Upton, is not credited, nor is the original work cited.
A dataset containing laboratory data and outcomes of IBD patients on Thiopurine therapy at the University of Michigan
The variables in this data set are as follows:
days_of_lifeNumeric. Range: 1207-32356. 1 missing value.
pltPlatelet Count. Numeric. Range: 11-1114. 4 missing values.
mpvMean Platelet Volume. Numeric. Range: 5.3-13.5. 21 missing values.
unBlood Urea Nitrogen. Numeric. Range: 2-118. 53 missing values.
wbcWhite Blood Cell Count. Numeric. Range: 0.7-33.5. No missing values.
hgbHemoglobin. Numeric. Range: 4.5-18.6. 4 missing values.
hctHematocrit. Numeric. Range: 13.7-55.2. 3 missing values.
rbcRed Blood Cell Count. Numeric. Range: 1.57-7.04. 3 missing values.
mcvMean Corpuscular (RBC) Volume. Numeric. Range: 56.5-124. 3 missing values.
mchMean Corpuscular (RBC) Hemoglobin. Numeric. Range: 16.7-42.3. 7 missing values.
mchcMean Corpuscular (RBC) Hemoglobin per Cell. Numeric. Range: 28.2-38.0. 7 missing values.
rdwRed cell Distribution Width. Numeric. Range: 11.3-39.7. 3 missing values.
neut_percentPercent of Neutrophils in WBC count. Numeric. Range: 17-98.1. No missing values.
lymph_percentPercent of Lymphocytes in WBC count. Numeric. Range: 1-67.9. No missing values.
mono_percentPercent of Monocytes in WBC count. Numeric. Range: 0-30.3. No missing values.
eos_percentPercent of Eosinophils in WBC count. Numeric. Range: 0.5-29.3. 6 missing values.
baso_percentPercent of Basoophils in WBC count. Numeric. Range: 0.2-5.3. 6 missing values.
sodSodium. Numeric. Range: 116-151. No missing values.
potPotassium. Numeric. Range: 2.6-10.1. 1 missing value.
chlorChloride. Numeric. Range: 83-126. No missing values.
co2Bicarbonate (CO2). Numeric. Range: 12-40. 5 missing values.
creatCreatinine. Numeric. Range: 0.2-8.4. No missing values.
glucGlucose. Numeric. Range: 41-486. No missing values.
calCalcium. Numeric. Range: 6.5-11.8. 1 missing value.
protProtein. Numeric, range 2.9-10, 0 missing values
albAlbumin. Numeric, range 1.2-5.5, 0 missing values
astAspartate Transaminase. Numeric, range 5-7765, 0 missing values
altAlanine Transaminase. Numeric, range 1-10666, 18 missing values
alkAlkaline phosphatase. Numeric, range 13-1938, 0 missing values
tbilTotal Bilirubin. Numeric, range 0.09-27, 0 missing values
activeActive Inflammation despite Thiopurines for > 12 weeks. Numeric, range 0-1, 0 missing values
remissionRemission of Inflammation after Thiopurines for > 12 weeks. Numeric, range 0-1, 0 missing values
thiomonthiomon
A data frame with 5168 observations of 32 variables. All are numeric variables.
Data on laboratory values for a complete blood count and chemistry panel at least 4 weeks after start of thiopurine therapy in IBD patients. The University of Michigan Hospital is in Ann Arbor, USA. These data have been anonymized, and time-shifted. Age is reported in days of life. Random Forest approaches can work well in modeling Active or Remission status. As published in Clin Gastroenterol Hepatol. 2010 Feb;8(2):143-150.
This data set is from Akbar K. Waljee and Peter D. Higgins, who de-identified data on CBC and chemistry testing at the University of Michigan for development of a machine learning algorithm to predict response to thiopurine medications in IBD patients. This data set contains individual laboratory values, age in days, and outcome (active or remission). These data have been anonymized, and time-shifted. As published in Clin Gastroenterol Hepatol. 2010 Feb;8(2):143-150.