Proof, Not Promises

A concern is raised about a young child, often by the other adult in the household. Police attend, weigh the file against the charging standard, and decline to lay a charge. The file closes as “no charges laid.”

Child welfare, reading the police outcome, records the concern as unsubstantiated and closes its own file. Family court, reading a closed child welfare file, treats the closure as an absence of risk. The parent’s assertion that conduct has ended now sits in a record containing no contradicting institutional finding, and unsupervised contact follows.

At no point in that sequence did anyone find the child safe. A prosecutorial threshold went unmet, and the not-meeting was converted, across three institutions, into an affirmative finding that none of them made.

A parent’s claim to have stopped is one of the most consequential sentences in a family file, and one of the least tested. It arrives in many forms - that they have stopped drinking, stopped using drugs, stopped misusing prescribed medication, stopped gambling, stopped viewing pornography, stopped a compulsive sexual or spending behaviour. It arrives in the same shape where the conduct was directed at an intimate partner: that the violence has stopped, that the threats and the surveillance and the humiliation have stopped, that the coerced or non-consensual sexual conduct has stopped, that the anger is under control now, that it was situational and belongs to the past.

The structure is identical in every version: an assertion by the person whose access to a child depends on it being believed, about conduct that took place where no one outside the household was watching.

The conduct itself is not equivalent, and nothing here suggests otherwise. Compulsive pornography use and strangulation sit at opposite ends of a severity range, and the verification proportionate to one is plainly disproportionate to the other. What follows concerns the evidentiary form of the claim - an untested assertion offered as proof of a changed household - not an equivalence of harms, and any response has to be scaled to the conduct actually at issue. It is nonetheless treated as evidence - recorded in an intake note, restated in an affidavit, relayed to a court, and used to justify unsupervised contact with a child too young to describe what happens when that adult is alone with them.

The child is not the only person the claim is made about. Where the conduct was directed at a partner, that adult usually remains inside the arrangement, exchanging the child and absorbing contact an order requires, and their safety turns on the same untested sentence. Harm to the child, meanwhile, does not depend on the child having been the target: physical, psychological, sexual, and financial abuse of one adult reaches the children in that household through what they see, hear, and register in the person caring for them - as Canadian family law expressly recognizes.

Canadian law does not actually require anyone to accept that statement. The federal Divorce Act asks what steps a person has taken to reduce risk and improve their ability to meet a child’s needs. Provincial child protection statutes ask whether a child’s survival, security, or development is at risk. Neither asks whether the parent has made a promise.

Nor does either require criminal-level proof. Criminal courts ask whether guilt in a past offence is established beyond a reasonable doubt. Child protection and family courts ask what is more likely than not, and what a child needs going forward. These are different questions answered to different standards, and a failure at the first says nothing about the answer to the second.

The gap is not in the legislation. It is in what institutions collect, count, and treat as proof - and in the fact that the person most likely to hold the relevant evidence, the other caregiver in the household, is frequently reclassified from witness to suspect.

The Assertion Problem

Recovery is a clinical process, measured through relapse, disclosure, honesty about lapses, and gradual behavioural change. It was never designed to answer the legal question, which is narrower: is a specific child safe in a specific environment during specific hours, without an adult present who can intervene?

The two get collapsed constantly. Treatment attendance becomes proof of safety. A completed program becomes proof of change. A negative test on a scheduled date becomes proof of a state that persists between test dates. Where the conduct was violence or coercive control, a stretch of time without a reported incident becomes proof of safety - when it may equally reflect separation, supervised contact, or a survivor who has concluded that reporting is not worth what it costs. An absent record is not evidence of an altered pattern. It is evidence of an absent record.

The most common input of all is self-report: a parent’s own account of their consumption, their triggers, their duration of abstinence, their version of an incident and how far they have come from it. Self-report is a reasonable clinical starting point, because a treating clinician’s job is to build a therapeutic relationship rather than to adjudicate. It is a poor evidentiary foundation, because the person supplying it is the person whose interests turn on the answer.

Meanwhile, the household’s most continuous observer - a spouse, partner, co-parent, grandparent, or adult sibling - often holds the only longitudinal view of the pattern: the concealment and the cycle, the household finances, the hidden containers, the escalation and the apology that follows it, the rules that tighten and loosen without warning, the conduct that resumes once an institution stops watching. That account is routinely discounted precisely because the witness is close to the situation, on the assumption that proximity equals bias.

This is backwards. Proximity is what produces relevant evidence in the first place, and bias is managed the way courts manage it everywhere else - through corroboration, contemporaneous records, internal consistency, and cross-examination. Discounting inner-circle testimony wholesale does not remove the bias; it removes the information along with it.

The Child Who Cannot Testify

Early childhood is the period of maximum exposure and minimum evidentiary capacity. A child under five cannot report, cannot compare their household to another, cannot distinguish an impaired caregiver from a tired one, cannot leave, and cannot summon help. Any system that waits for a child’s account has, in these years, built a waiting period that outlasts the harm.

The physical exposure is documented, and two separate national datasets speak to it.

The first is the Public Health Agency of Canada’s sentinel surveillance through CHIRPP, which records injuries and poisonings presenting at participating emergency departments. Its opioid analysis found that among children under five, the most common factor contributing to an opioid-related poisoning was access to medication combined with an absence of supervision. This unsupervised child finds the pills. That is a finding about supervision, which is precisely the variable a claim of recovery is offered to resolve.

The second is the annual report of Canada’s poison centres, compiled by the Canadian Association of Poison Control Centres. Of the roughly 187,000 human exposure cases those centres managed in 2020, more than a third involved a child aged five or under. That figure covers all poison exposures, including household cleaners and pain-relief medication, not substance misuse specifically - but it establishes the underlying fact on which everything else here rests, which is that very young children are the population that finds what adults leave within reach.

The cannabis data is more directly on point. A peer-reviewed national analysis by Varin and colleagues, covering 2015 to 2021, found sharp and sustained increases in emergency department visits and hospitalizations for unintentional cannabis poisoning among Canadian children, concentrated in the youngest ages, with affected children averaging roughly three and a half years old and the steepest increases in provinces permitting the sale of edibles. Public Health Agency research published in 2026 reports the same pattern in British Columbia emergency department data, which suggests a standing feature of Canadian households rather than a transitional effect of legalization.

None of this requires that anyone intended harm. It is the ordinary consequence of an unsecured substance in a home where a toddler moves faster than an impaired adult.

The developmental exposure is less visible and better established. Canadian and international research consistently links parental addiction in childhood to elevated psychological distress in adulthood, and finds that the households in question are characterized by instability, unpredictability, and diminished parental availability - not only by discrete incidents.

That distinction matters enormously. Institutions are built to detect incidents. Environments shape children.

The Criminal-Charge Threshold and What It Was Never Built to Do

A recurring institutional reflex is to treat the absence of criminal charges as the absence of risk. Nothing in Canadian law supports this.

The criminal standard is deliberately difficult, because the consequence is loss of liberty: proof beyond a reasonable doubt of a defined offence, by an identified person, at an identifiable time, on admissible evidence, prosecuted at the Crown’s discretion. It is an accountability mechanism aimed at the past.

Child protection and family law serve a different purpose on a lower standard. Child protection is provincial, and the thresholds are not uniform: statutes differ in how they define a child in need of protection, in whether risk of future harm is expressly captured, and in the reporting duty they impose. Alberta’s Child, Youth and Family Enhancement Act frames intervention around whether a child’s survival, security, or development is at risk, and imposes a reporting duty on anyone with reasonable and probable grounds to believe a child needs intervention. The federal Divorce Act is more explicit still: its definition of family violence expressly applies “whether or not the conduct constitutes a criminal offence,” and it captures a child’s direct or indirect exposure to that conduct. Section 16(2) requires courts to give primary consideration to a child’s physical, emotional, and psychological safety, security, and well-being.

The statutory architecture is therefore already forward-looking and already risk-based. The failure is operational, and it follows the sequence this article opened with - a charging decision about the sufficiency of evidence for prosecution, and nothing more, read downstream as a finding about a child.

It would be wrong to read this as indifference by the people inside it. Charging standards exist because a charge means criminal jeopardy. Intake workers carry caseloads that make independent re-investigation of a police-closed matter impractical, and Alberta’s own advocate has reported that workers frequently decide with incomplete, inaccurate, or conflicting information. Privacy legislation and jurisdictional boundaries limit what one institution may lawfully tell another. Family courts see a fraction of the record and depend on what the agencies before them have gathered. The silos are the predictable output of three systems built to different standards, funded below their demand, and given no shared instrument for recording risk - which is what makes the problem structural, and also what makes it fixable.

Two features of the Divorce Act deserve particular attention here. Section 16(4)(g) directs courts to consider the steps a person has taken to prevent further family violence and to improve their ability to care for and meet the child’s needs. Steps taken. Not statements made, not programs enrolled in, not intentions expressed.

Section 16(5) works the other way round, and the distinction matters. It is drafted as a prohibition with an exception: a court shall not consider a person’s past conduct unless that conduct is relevant to the exercise of parenting time or decision-making responsibility. It does not invite a general inquiry into character. What it does mean is that past conduct is admissible where - and only where - it bears on parenting, and conduct that placed a young child at risk in the home plainly does.

That is a narrower gateway than protective caregivers are sometimes promised. It is also a wider one than they are usually told exists.

What Is Not Counted

Beneath the charge threshold sits a large volume of harm that no Canadian system systematically records.

• Environmental hazard. Substances stored in reach. Paraphernalia, medication, or alcohol accessible to a child in a crawling or grasping phase. Impaired supervision during bathing, feeding, sleeping, stair use, and transport. Impaired driving with a child restrained in the back seat, which is only counted if it is detected.
• Behavioural unpredictability. Not violence in the chargeable sense, but a household in which another person’s conduct sets the emotional weather. Cycles of remorse and repetition. Nights that cannot be planned around. A young child’s nervous system calibrates to this before language arrives.
• Financial instability. Money that disappears without explanation. Compulsive spending and gambling losses that do not appear in any disclosure document because they were never itemized. Utility arrears, missed rent, and food insecurity. The Canadian Incidence Study found households running out of money among the conditions recorded in substantiated maltreatment investigations; that is an association within investigated families, not a demonstration that poverty causes harm, and it should not be read as one.
• The protective caregiver’s unpaid labour. Continuous vigilance. Sleep loss. Silent risk management. Cleaning up hazards before a child finds them. Absorbing the household’s instability so a child does not.

None of these generate a record. All of them generate risk. And the last one generates a specific institutional injustice.

The Reclassification of the Protective Parent

There is a well-documented pattern in Canadian and international child welfare practice by which the non-offending caregiver - overwhelmingly, in the research literature, mothers - is held responsible for “failure to protect” a child from another adult’s conduct. Strega’s work on child welfare responses where men beat mothers set out the mechanism; Azzopardi’s 2021 discourse analysis traced how the attribution of blame is constructed and institutionally entrenched. The scale of the overlap is not small: the 2008 Canadian Incidence Study recorded the primary caregiver as a victim of domestic violence in 46 percent of substantiated maltreatment investigations, alcohol abuse in 21 percent, and drug or solvent abuse in 17 percent.

The critique is operational rather than sentimental. Where a protective caregiver believes reporting will trigger their own investigation, they report less; caseworkers then assess with less information, and safety planning weakens accordingly. The doctrine defeats the function it claims to serve.

It also produces a perverse asymmetry. The caregiver who leaves may lose housing stability, and housing stability is scored. The caregiver who stays may be assessed as complicit. The parent whose conduct created the risk is frequently assessed on their stated commitment to change, while the parent managing that risk is assessed on their demonstrated circumstances.

A perpetrator-pattern approach - assessing what the person creating the risk has actually done, and what the protective caregiver has actually done to mitigate it - is available, taught internationally, and unevenly adopted. Its absence is a design choice, not an inevitability.

The Verification Trap

The obvious remedy is objective testing. Canada has already learned, at great cost, that objective testing is only as good as the forensic standards behind it - and that testing regimes fall hardest on the poorest families.

The scale of the Motherisk failure is the reason for that caution. Between 2005 and 2015, hair samples from more than sixteen thousand individuals were tested at the request of child protection agencies by the Motherisk Drug Testing Laboratory at Toronto’s Hospital for Sick Children.

In 2015, an independent review by Justice Susan Lang concluded the methodology was inadequate and unreliable for use in child protection and criminal proceedings, fell short of international forensic standards, and produced results that were frequently misinterpreted. A subsequent commission led by Justice Judith Beaman reviewed 1,271 cases and found the results had a substantial impact on the outcome in 56 of them, including cases in which families were permanently separated.

Her findings about process matter as much as her findings about numbers. Beaman described the reliance on the testing as manifestly unfair and harmful even where it did not change outcomes: results were routinely admitted without the checks that would apply to any other expert evidence, and the testing was imposed on some of the most vulnerable people in the country with scant regard for due process or bodily integrity. She raised comparable concerns about urine screening then in use. Indigenous families were markedly over-represented among the affected cases.

The lesson is not that biological verification is worthless. It is that verification must itself be verified.

Six Questions to Ask of Any Monitoring Proposal

These apply whether the proposal comes from a protective caregiver, an agency, or a court.

A monitoring regime that cannot answer those questions is not evidence. It is ritual - and Canadian history shows it is a ritual capable of dismantling families based on nothing.

The Boundary That Should Not Move

Two propositions can hold simultaneously, and any credible framework must hold both.

Treatment is a medical entitlement. Substance use disorder and compulsive behaviour disorders are health conditions. Access to evidence-based care, without moralizing, is a public good. Punitive frameworks reduce disclosure and worsen outcomes, and children benefit when a parent recovers.

Access to a young child is a safety privilege, not a therapeutic tool. A child under five is not an instrument of another person’s motivation. Reunification is a legitimate goal; it is not a legitimate reason to accept unverified conditions in the interim. Where the two conflict, the child’s safety governs - a principle that section 16(2) of the Divorce Act already states in terms.

The practical implication is graduated, condition-based contact, calibrated to verified change: supervised contact where risk is unresolved; defined and observable conditions rather than promises; reassessment on evidence rather than on elapsed time; and expansion of contact tied to demonstrated compliance.

It is equally important to say what this does not license. A framework that only ever tightens is not a safety framework at all - it is a removal framework operating under a safety justification.

Separation carries documented harm of its own - disrupted attachment, placement instability, loss of siblings, community, language, and cultural continuity, and elevated long-run risks across health, education, and justice outcomes. Those consequences are also distributed with severe inequity. Statistics Canada’s analysis of 2021 census data found that Indigenous children made up 53.7 percent of foster children aged 14 and under living in private households, while accounting for 7.6 percent of children in that age range. This share has risen across the 2011, 2016, and 2021 cycles rather than falling. Manitoba recorded the highest rate; Alberta recorded the widest disparity between Indigenous and non-Indigenous rates of any province. The Motherisk record showed the same skew among the families most damaged by unreliable evidence. A system that answers uncertainty by removing more children will not distribute that answer evenly, and the record indicates where the weight falls.

Reunification is a risk point rather than a resolution. Alberta’s Child and Youth Advocate has reported on deaths of young children following reunification and identified the transition period as one requiring sustained support rather than case closure. Returning a child without conditions, monitoring, and funded support is not the opposite of over-removal; it is the same failure to calibrate, running the other direction.

Rigour therefore has to cut both ways. The same standard that refuses to accept a promise must also refuse an unreliable test, an unexamined agency file, a closure mistaken for a finding, or a stereotype about which families are dangerous.

Documentation That Meets an Evidentiary Standard

Systemic reform is the objective. It is not available to anyone this week. Protective caregivers are navigating the system that exists, in which the burden of producing usable evidence falls on the person with the least institutional power and the most to lose - and the guidance that follows is offered on that basis, as a way of closing the distance between what a survivor knows and what a court can act on.

Those caregivers are frequently told that their observations are “just their word.” Often the difference between testimony that is discounted and testimony that is relied upon is not credibility. It is form.

Do

• Record contemporaneously. The same day, ideally within hours. Contemporaneous records carry weight that reconstructed accounts do not.
• Record observable facts. Time, date, location, who was present, and what was directly perceived. “Slurred speech, unsteady on stairs, 7:40 pm, child in bath” is evidence. “He was a mess again” is characterization.
• Separate observation from conclusion. Keep interpretations in a distinct column or section, or leave them out entirely.
• Preserve records in their original form. Keep original files and metadata intact. Do not retype, tidy, or consolidate an original log into a cleaner version and discard the first.
• Build the corroboration layer. Medical records, pharmacy records, daycare and school incident notes, bank and credit statements, insurance claims, repair invoices, and police occurrence numbers, even where no charge follows. Third-party records anchor a personal account.
• Conduct a written environmental audit. Where substances are stored, what a child can reach, what supervision exists at high-risk moments, transport arrangements, and sleeping arrangements.
• Obtain independent legal advice before disclosing, filing, or recording. Rules on admissibility, privacy, disclosure obligations, and audio or video recording differ by province and by forum, and a well-intentioned recording can damage a case.

Don’t

• Don’t editorialize, diagnose, or use clinical labels you are not qualified to apply. Describe conduct, not conditions.
• Don’t exaggerate or round upward. A single demonstrable overstatement can be used to discount an entire log.
• Don’t rely on memory to be reconstructed later. The evidentiary value of a log is created by when it was written.
• Don’t assume that no charge means no record. Occurrence reports, ambulance calls, and agency intake notes exist independently of charging decisions and can often be obtained.
• Don’t let documentation displace safety planning. If there is immediate danger, the log is not the priority.

If you are in Canada and currently at risk, safety planning comes first. Systemic Press maintains a survivor toolkit with practical steps and routing information.

A Blueprint for Transformation

Count exposure, not only incidents. Child protection data systems should record household environmental risk, caregiver impairment during supervision hours, and financial destabilization as measurable variables, not as context narrative that disappears at file closure.

Legislate forensic standards for testing used in child protection. Accreditation, chain of custody, confirmatory methodology, expert interpretation, and a right to challenge should be statutory preconditions to admissibility. The Motherisk record shows what the absence of these requirements costs.

Retire the failure-to-protect frame. Replace it with an explicit perpetrator-pattern standard that assesses the conduct creating the risk and credits the protective caregiver’s mitigation efforts. Several jurisdictions already prohibit holding a caregiver responsible for harm arising from their own victimization.

Close the inter-institutional laundering loop. No agency should be permitted to treat another institution’s declination as an affirmative safety finding. A police decision not to charge should be recorded, in child welfare and family court files, as what it is - a prosecutorial threshold not met - and nothing more.

Fund supervised contact. A graduated-access framework is unenforceable where supervised access capacity does not exist - condition-based orders without publicly funded supervision default, in practice, to unsupervised contact.

Give the youngest children a voice. Where a child cannot speak, independent representation, structured environmental assessment, and inner-circle testimony are the only channels available. Systems that recognize none of these have not protected the child’s voice. They have arranged not to hear it.

Glossary

Balance of probabilities - The civil standard of proof: more likely than not. Applies in family and child protection proceedings.

Beyond a reasonable doubt - The criminal standard of proof. Higher, and directed at establishing past guilt rather than future risk.

Chain of custody - Documentation tracking a sample from collection to analysis, establishing that it was not altered or substituted.

Confirmatory testing - A second, methodologically distinct analysis used to confirm a presumptive screening result.

Exposure - In the family law context, a child’s direct or indirect experience of conduct in the household, including hearing it or perceiving its effects on a caregiver.

Failure to protect - A child welfare doctrine attributing responsibility to a caregiver for harm caused by another adult.

Substantiation - A child welfare finding that maltreatment or risk of maltreatment is established on the applicable civil standard.

Sources

Legislation

• Divorce Act, RSC 1985, c 3 (2nd Supp), ss 2 (definition of family violence), 16, 16.1.
• Child, Youth and Family Enhancement Act, RSA 2000, c C-12.
• Child, Youth and Family Enhancement (Protecting Alberta’s Children) Amendment Act, SA 2019, c 10.

Reports and Inquiries

• Lang, Susan. Report of the Motherisk Hair Analysis Independent Review. Ontario Ministry of the Attorney General, 2015.
• Beaman, Judith C. Harmful Impacts: The Reliance on Hair Testing in Child Protection - Report of the Motherisk Commission. Ontario Ministry of the Attorney General, 2018.
• British Columbia Ministry of Children and Family Development. A Review of Motherisk Drug Testing Laboratory Hair Strand Testing in Child Protection in BC, 2017.
• Office of the Child and Youth Advocate Alberta. Voices for Change: Aboriginal Child Welfare in Alberta - A Special Report, July 2016.
• Office of the Child and Youth Advocate Alberta. Toward a Better Tomorrow: Addressing the Challenge of Aboriginal Youth Suicide, and successive investigative reviews of child deaths following reunification. Reports indexed at ocya.alberta.ca.
• Department of Justice Canada. The Divorce Act Changes Explained, ss 16(2), 16(3), 16(4), 16(5).

Data and Studies

• Trocmé, N., Fallon, B., MacLaurin, B., et al. Canadian Incidence Study of Reported Child Abuse and Neglect - 2008: Major Findings. Public Health Agency of Canada. See Table on primary caregiver risk factors in substantiated investigations.
• Fallon, B., et al. Canadian Incidence Study of Reported Child Abuse and Neglect - 2019.
• Taillieu, T. L., Cheung, K., Sareen, J., Katz, L. Y., Tonmyr, L., and Afifi, T. O. “Caregiver Vulnerabilities Associated With the Perpetration of Substantiated Child Maltreatment in Canada.” Journal of Interpersonal Violence, 2021.
• Public Health Agency of Canada. Sentinel Surveillance of Substance-Related Poisonings in Canada: Spotlight on Opioids. Public Health Infobase, eCHIRPP.
• Canadian Association of Poison Control Centres. Canadian Poison Centre Report, 2020 data year.
• Varin, M., Champagne, A., Venugopal, J., et al. “Trends in Cannabis-Related Emergency Department Visits and Hospitalizations Among Children Aged 0-11 Years in Canada from 2015 to 2021.” BMC Public Health, 2023.
• Public Health Agency of Canada. “The Impact of Recreational Cannabis Legalization on Pediatric Emergency Department Visits in British Columbia, Canada.” Health Promotion and Chronic Disease Prevention in Canada, vol. 46, no. 4, 2026.
• Statistics Canada. “Indigenous Foster Children Living in Private Households: Rates and Household Characteristics.” The Daily, 18 April 2024. Figures cited are for children aged 14 and under in private households.
• Myran, D. T., Tanuseputro, P., Auger, N., et al. “Pediatric Hospitalizations for Unintentional Cannabis Poisonings and All-Cause Poisonings Associated With Edible Cannabis Product Legalization and Sales in Canada.” JAMA Health Forum, 2023.
• Statistics Canada. “Trajectories of Psychological Distress Among Canadian Adults Who Experienced Parental Addiction in Childhood.” Health Reports.
• McCabe, S. E., Schepis, T., McCabe, V., et al. Parental substance use disorder prevalence among US children, JAMA Pediatrics, 2025.
• Azzopardi, C. “Gendered Attributions of Blame and Failure to Protect in Child Welfare Responses to Sexual Abuse: A Feminist Critical Discourse Analysis.” Violence Against Women, 2021.
• Strega, S. “Failure to Protect: Child Welfare Interventions When Men Beat Mothers,” in Alaggia and Vine, eds., Cruel but Not Unusual, 2006.

Journalism

• CBC News. “‘Intrusive and Stigmatizing’: Discredited Hair Tests Harmed Vulnerable Families in Ontario, Report Concludes.” 26 February 2018.

Related Systemic Press Articles

• “When There Are No Charges - Understanding the Gap Between the Criminal System and Child Protection in Canada”
• Survivor Toolkit, systemicpress.ca

Read Further

The full research and structural analysis are available in: