Skip to content

Latest commit

 

History

History
453 lines (291 loc) · 14.8 KB

README.rst

File metadata and controls

453 lines (291 loc) · 14.8 KB
Raw

pypi actions codecov downloads

edc-consent

Add classes for the Informed Consent form and process.

Installation

Declare the consent model:

class SubjectConsent(
    ConsentModelMixin,
    SiteModelMixin,
    UpdatesOrCreatesRegistrationModelMixin,
    NonUniqueSubjectIdentifierModelMixin,
    IdentityFieldsMixin,
    PersonalFieldsMixin,
    SampleCollectionFieldsMixin,
    ReviewFieldsMixin,
    VulnerabilityFieldsMixin,
    SearchSlugModelMixin,
    BaseUuidModel,
):

    """A model completed by the user that captures the ICF."""

    subject_identifier_cls = SubjectIdentifier

    subject_screening_model = "edc_example.subjectscreening"

    objects = ConsentObjectsManager()
    on_site = CurrentSiteManager()
    history = HistoricalRecords()

    class Meta(ConsentModelMixin.Meta, BaseUuidModel.Meta):
        pass


class SubjectConsentV1(SubjectConsent):
    """A proxy model completed by the user that captures version 1
     of the ICF.
     """
    objects = ConsentObjectsByCdefManager()
    on_site = CurrentSiteByCdefManager()
    history = HistoricalRecords()

    class Meta:
        proxy = True
        verbose_name = "Consent V1"
        verbose_name_plural = "Consent V1"

The next step is to declare and register a ConsentDefinition. A consent definition is a class that represents an approved Informed Consent. It is linked to a proxy of the consent model, for example SubjectConsent from above, using the class attribute model. We use a proxy model since over time each subject may need to submit more than one version of the consent. Each version of a subject's consent is represented by an instance of the Each version is paird with a proxy model. The approved Informed Consent also includes a validity period (start=datetime1 to end=datetime2) and a version number (version=1). There are other attributes of a ConsentDefinition to consider but lets focus on the start date, end date, version and model for now.

ConsentDefinitions are declared in the root of your app in module consents.py. A typical declaration looks something like this:

from datetime import datetime
from zoneifo import ZoneInfo

from edc_consent.consent_definition import ConsentDefinition
from edc_consent.site_consents import site_consents
from edc_constants.constants import MALE, FEMALE

consent_v1 = ConsentDefinition(
    'edc_example.subjectconsentv1',
    version='1',
    start=datetime(2013, 10, 15, tzinfo=ZoneInfo("UTC")),
    end=datetime(2016, 10, 15, 23, 59, 999999, tzinfo=ZoneInfo("UTC")),
    age_min=16,
    age_is_adult=18,
    age_max=64,
    gender=[MALE, FEMALE])

site_consents.register(consent_v1)

add to settings:

INSTALLED_APPS = [
    ...
    'edc_consent.apps.AppConfig',
    ...
]

On bootup site_consents will autodiscover the consents.py and register the ConsentDefinition.

To create an instance of the consent for a subject, find the ConsentDefinitions and use model_cls.

cdef = site_consents.get_consent_definition(
    report_datetime=datetime(2013, 10, 16, tzinfo=ZoneInfo("UTC"))
)

assert cdef.version == "1"
assert cdef.model == "edc_example.subjectconsentv1"

consent_obj = cdef.model_cls.objects.create(
    subject_identifier="123456789",
    consent_datetime=datetime(2013, 10, 16, tzinfo=ZoneInfo("UTC"),
    ...)

assert consent_obj.consent_version == "1"
assert consent_obj.consent_model == "edc_example.subjectconsentv1"

Add a second ConsentDefinition to your consents.py for version 2:

class SubjectConsentV2(SubjectConsent):
    """A proxy model completed by the user that captures version 2
     of the ICF.
     """
    objects = ConsentObjectsByCdefManager()
    on_site = CurrentSiteByCdefManager()
    history = HistoricalRecords()

    class Meta:
        proxy = True
        verbose_name = "Consent V2"
        verbose_name_plural = "Consent V2"
consent_v1 = ConsentDefinition(...)

consent_v2 = ConsentDefinition(
    'edc_example.subjectconsentv2',
    version='2',
    start=datetime(2016, 10, 16, 0,0,0, tzinfo=ZoneInfo("UTC")),
    end=datetime(2020, 10, 15, 23, 59, 999999, tzinfo=ZoneInfo("UTC")),
    age_min=16,
    age_is_adult=18,
    age_max=64,
    gender=[MALE, FEMALE])

site_consents.register(consent_v1)
site_consents.register(consent_v2)
cdef = site_consents.get_consent_definition(
    report_datetime=datetime(2016, 10, 17, tzinfo=ZoneInfo("UTC"))
)

assert cdef.version == "2"
assert cdef.model == "edc_example.subjectconsentv2"

consent_obj = cdef.model_cls.objects.create(
    subject_identifier="123456789",
    consent_datetime=datetime(2016, 10, 17, tzinfo=ZoneInfo("UTC"),
    ...)

assert consent_obj.consent_version == "2"
assert consent_obj.consent_model == "edc_example.subjectconsentv2"

edc_consent is coupled with edc_visit_schedule. In fact, a data collection schedule is declared with one or more ConsentDefinitions. CRFs and Requisitions listed in a schedule may only be submitted if the subject has consented.

schedule = Schedule(
    name=SCHEDULE,
    verbose_name="Day 1 to Month 6 Follow-up",
    onschedule_model="effect_prn.onschedule",
    offschedule_model="effect_prn.endofstudy",
    consent_definitions=[consent_v1, consent_v2],
)

When a CRF is saved, the CRF model will check the schedule to find the ConsentDefinition with a validity period that contains the crf.report_datetime. Using the located ConsentDefinitions, the CRF model will confirm the subject has a saved subject_consent with this consent_definition.version.

The ConsentDefinitions above assume that consent version 1 is completed for a subject consenting on or before 2016/10/15 and version 2 for those consenting after 2016/10/15.

Sometimes when version 2 is introduced, those subjects who consented for version 1 need to update their version 1 consent to version 2. For example, a question may have been added in version 2 to allow a subject to opt-out of having their specimens put into longterm storage. The subjects who are already consented under version 1 need to indicate their preference as well by submitting a version 2 consent. (To make things simple, we would programatically carry-over and validate duplicate data from the subject's version 1 consent.)

To allow this, we would add update_versions to the version 2 ConsentDefinition.

consent_v1 = ConsentDefinition(
    'edc_example.subjectconsentv1',
    version='1', ...)

consent_v2 = ConsentDefinition(
    'edc_example.subjectconsentv2',
    version='2',
    update_versions=[UpdateVersion(consent_v1.version, consent_v1.end)],

site_consents.register(consent_v1)
site_consents.register(consent_v2)

As the trial continues past 2016/10/15, there will three categories of subjects:

  • Subjects who completed version 1 only
  • Subjects who completed version 1 and version 2
  • Subjects who completed version 2 only

If the report date is after 2016/10/15, data entry for "Subjects who completed version 1 only" will be blocked until the version 2 consent is submitted.

Features

  • base class for an informed consent document
  • data for models that require consent cannot be add until the consent is added
  • consents have a version number and validity period
  • maximum number of consented subjects can be controlled.
  • data collection is only allowed within the validity period of the consent per consented participant
  • data for models that require consent are tagged with the consent version

TODO

  • link subject type to the consent model. e.g. maternal, infant, adult, etc.
  • version at model field level (e.g. a new consent period adds additional questions to a form)
  • allow a different subject's consent to cover for another, for example mother and infant.

Usage

Declare the ModelForm:

class SubjectConsentForm(BaseConsentForm):

    class Meta:
        model = SubjectConsent

Now that you have a consent model class, declare the models that will require this consent:

class Questionnaire(RequiresConsentMixin, models.Model):

    report_datetime = models.DateTimeField(default=timezone.now)

    question1 = models.CharField(max_length=10)

    question2 = models.CharField(max_length=10)

    question3 = models.CharField(max_length=10)

@property
def subject_identifier(self):
    """Returns the subject identifier from ..."""
    return subject_identifier

class Meta:
    app_label = 'my_app'
    verbose_name = 'My Questionnaire'
  • report_datetime: a required field used to lookup the correct ConsentDefinition and to find, together with subject_identifier, a valid instance of SubjectConsent;
  • subject_identifier: a required field or may be a property that knows how to find the subject_identifier for the instance of Questionnaire.

Once all is declared you need to:

  • define the consent version and validity period for the consent version in ConsentDefinition;
  • add a Quota for the consent model.

As subjects are identified:

  • add a consent
  • add the models (e.g. Questionnaire)

If a consent version cannot be found given the consent model class and report_datetime a ConsentDefinitionError is raised.

If a consent for this subject_identifier cannot be found that matches the ConsentDefinition a NotConsentedError is raised.

Specimen Consent

A participant may consent to the study but not agree to have specimens stored long term. A specimen consent is administered separately to clarify the participant's intention.

The specimen consent is declared using the base class BaseSpecimenConsent. This is an abridged version of BaseConsent. The specimen consent also uses the RequiresConsentMixin as it cannot stand alone as an ICF. The RequiresConsentMixin ensures the specimen consent is administered after the main study ICF, in this case MyStudyConsent.

A specimen consent is declared in your app like this:

class SpecimenConsent(
    BaseSpecimenConsent, SampleCollectionFieldsMixin, RequiresConsentMixin,
    VulnerabilityFieldsMixin, AppointmentMixin, BaseUuidModel
):

    consent_model = MyStudyConsent

    registered_subject = models.OneToOneField(RegisteredSubject, null=True)

    objects = models.Manager()

    history = AuditTrail()

class Meta:
    app_label = 'my_app'
    verbose_name = 'Specimen Consent'

Validators

The ConsentAgeValidator validates the date of birth to within a given age range, for example:

from edc_consent.validtors import ConsentAgeValidator

class MyConsent(ConsentQuotaMixin, BaseConsent):

    dob = models.DateField(
        validators=[ConsentAgeValidator(16, 64)])

    quota = QuotaManager()

    class Meta:
        app_label = 'my_app'

The PersonalFieldsMixin includes a date of birth field and you can set the age bounds like this:

from edc_consent.validtors import ConsentAgeValidator
from edc_consent.models.fields import PersonalFieldsMixin

class MyConsent(ConsentQuotaMixin, PersonalFieldsMixin, BaseConsent):

    quota = QuotaManager()

    MIN_AGE_OF_CONSENT = 18
    MAX_AGE_OF_CONSENT = 64

    class Meta:
        app_label = 'my_app'

Common senarios

Tracking the consent version with collected data

All model data is tagged with the consent version identified in ConsentDefinition for the consent model class and report_datetime.

Reconsenting consented subjects when the consent changes

The consent model is unique on subject_identifier, identity and version. If a new consent version is added to ConsentDefinition, a new consent will be required for each subject as data is reported within the validity period of the new consent.

Some care must be taken to ensure that the consent model is queried with an understanding of the unique constraint.

Linking the consent version to added or removed model fields on models that require consent

TODO

Infants use mother's consent

TODO

By adding the property consenting_subject_identifier to the consent

Patient names

If patient names need to be removed from the data collection, there are a few helper attributes and methods to consider.

settings.EDC_CONSENT_REMOVE_PATIENT_NAMES_FROM_COUNTRIES: list[str]

If given a list of country names, name fields will be removed from any admin.fieldset.

See also edc_sites.all_sites

ConsentModelAdminMixin.get_fieldsets

def get_fieldsets(self, request, obj=None):
    fieldsets = super().get_fieldsets(request, obj)
    for country in get_remove_patient_names_from_countries():
        site = getattr(request, "site", None)
        if site and site.id in [s.site_id for s in self.all_sites.get(country)]:
            return self.fieldsets_without_names(fieldsets)
    return fieldsets

This method could be added to any ModeLadmin with names.

using

Other TODO

  • Timepoint model update in save method of models requiring consent
  • handle added or removed model fields (questions) because of consent version change
  • review verification actions
  • management command to update version on models that require consent (if edc_consent added after instances were created)
  • handle re-consenting issues, for example, if original consent was restricted by age (16-64) but the re-consent is not. May need to open upper bound.