“90% of marriages end in divorce when a child has cancer”

The short version

If someone has told you that most marriages don’t survive a child’s cancer, I want to give you the actual numbers. I went looking for the study behind the 90 percent figure and could not find one. What I did find were seven studies that measured it, three of them following every family in a country through its national records. None of them shows anything close to the saying. In Denmark and Norway, parents of children with cancer separated and divorced at about the same rate as other parents.^{1, 2, 3} The one American study of families like ours, followed through the first year of leukemia treatment, found that about one couple in eight had separated by the twelve-month survey.⁹

None of that means it is easy. In one clinic survey, about four in ten parents said the relationship had gotten worse since the diagnosis.⁴ A relationship can be under real strain without ending. Below I take the studies one at a time, say what each one measured, and say what I think it means for a family in our position.

Where the number comes from

I tried to trace the 90 percent figure and could not find where it started.^{5, 6} The oldest lead is a 1977 book for bereaved parents, The Bereaved Parent by Harriet Sarnoff Schiff. A 2015 article from a bereavement organization says the book described as many as 90 percent of couples having “serious marital difficulty” in the months after a child’s death, without naming a study.⁵ That is about strain, not divorce, and about bereavement, not a cancer diagnosis. I have not read the original passage, so even this is a lead rather than a fact. By 1998 a researcher named Reiko Schwab had published a review calling the high-divorce belief after a child’s death a myth.⁶ How it became a cancer statistic, nobody seems to know.

The seven studies, one at a time

The first three use national registers: the government records that log every cancer diagnosis, every marriage and every divorce in a country. Researchers can follow all the affected families for years and compare them with families whose children were well, and nobody has to fill in a survey. That makes them the strongest evidence on this page, with one limit worth keeping in mind: a register knows whether a couple split up, not whether they were happy.

How to read the ratio charts. The register studies don’t report a divorce percentage. They report a ratio: how often parents of children with cancer divorced compared with similar parents whose children were well. A ratio of 1.0 means the same. 1.5 would mean half again as often. 0.9 would mean slightly less often. The line through each dot is the 95 percent confidence interval, the range of ratios the study’s data fit reasonably well; a wide line means less certainty. If the line crosses 1.0, the study could not tell the two groups apart.²⁴ One thing that takes a moment: a ratio of 1.5 does not mean half the parents divorced. It means one small rate was one and a half times another small rate.

Study 1: Denmark, every child diagnosed from 1982 to 2014

Researchers in Denmark took every child diagnosed with cancer before age 20 over 33 years and found their parents in the national records. For separation, meaning a couple stopped living together, they compared 5,245 couples who had a child with cancer against 52,599 couples who did not. For legal divorce, they compared 4,371 married couples against 43,881. Each family was followed from the year before the diagnosis to as far as ten years after it.¹

Study 1 · Denmark, published 2020

Separation and divorce among parents of children with cancer, compared with other Danish parents

Each dot is the study's adjusted result. The line through it is the 95 percent confidence interval. 1.0 means the same rate as the comparison parents.

Separation (couple stops living together)0.95 (0.92 to 0.98) · 5,245 couples vs 52,599

Divorce (legal)0.91 (0.86 to 0.96) · 4,371 married couples vs 43,881

Adjusted hazard ratios with 95% confidence intervals. Source 1, Results, “Parental separation and divorce.”

Both dots sit a little below 1.0, and both lines stay below it: 0.95 for separation and 0.91 for divorce. That is the study saying that parents of children with cancer separated and divorced slightly less often than other Danish parents over those years. I would not read that as cancer protecting a marriage; the authors don’t either. What I take from it is simpler. Across more than 5,000 families followed for up to ten years, nothing like the saying happened.¹

One more detail from the same study, because it bears on the first year of treatment. When the researchers looked at which cancer families were more likely to split, they found younger parents, parents with less schooling, unemployed parents and more recent diagnosis years. Those analyses began three years after diagnosis, so they describe the long haul, not the early months.¹

Study 2: Norway, 977,928 married couples from 1974 to 2001

Norway’s records let researchers follow almost a million married couples with children over 27 years. Among them, 4,590 couples had a child with cancer. The researchers compared divorce in those families with divorce in the rest, and then sliced the cancer group by diagnosis, by whether the child survived, and by the mother’s education.²

Study 2 · Norway, published 2010

Divorce among parents of children with cancer, compared with other Norwegian parents, overall and in subgroups

Same kind of chart. The first row is the whole group; the rows below it are slices of the same study, so they are smaller and their lines are wider.

All childhood cancers1.04 (0.95 to 1.13) · 4,590 couples

Acute lymphoblastic leukemia (ALL)0.99 (0.83 to 1.18) · 123 divorces

Kidney cancer, mostly Wilms tumor1.52 (1.12 to 2.05) · 43 divorces

The child with cancer died1.01 (0.88 to 1.15) · 213 divorces

Child died, mother college-educated1.31 (1.05 to 1.63) · vs cancer-free parents with the same education

Adjusted odds ratios with 95% confidence intervals. Source 2, Tables 1 to 4.

The top row is the whole group: 1.04, with a line that crosses 1.0 comfortably. Parents of children with cancer divorced at the same rate as everyone else. The second row is the one I care most about, because it is our diagnosis: for acute lymphoblastic leukemia the ratio was 0.99, from 123 divorces. The same as everyone else.²

The third row is the one exception in these studies. Parents of children with kidney cancer, mostly Wilms tumor, divorced more often, a ratio of 1.52. It rests on 43 divorces, and the authors themselves say it could be chance, since they ran many comparisons. I would not disregard it and I would not build anything on it.²

The last two rows are about families whose child died, and I will come back to them further down. The short version is that overall there was no increase, and in one subgroup, mothers with a college education, there was, compared with cancer-free families of the same education. That is one slice among many, reported because the authors reported it.²

Study 3: Denmark again, an earlier look published in 2012

Before the 2020 study, a Danish team had already asked the same question with an earlier slice of the same records: parents of 2,450 children with cancer compared with parents of 44,853 matched children who were well. The outcome was a couple ending their cohabitation. The ratio was 1.00, with an interval of 0.91 to 1.10. Exactly the same as other parents.³ It draws on the same country and overlapping years as Study 1, so I count it as a second look rather than an independent confirmation. It still points the same way.

Study 4: American families in the first year of treatment, published in 2014

This is the study closest to home, and the only one on this page that follows families through treatment itself. Between April 2005 and March 2009, 159 families joined a Children’s Oncology Group study at 30 US hospitals and one in Australia. The children were 2 to 9 years old, more than half of them under 5, all with average-risk B-ALL on AALL0331, an earlier COG protocol for the same group of kids that AALL1731 treats today. Families were enrolled about four weeks after diagnosis, and 82 percent of the eligible families took part. The main caregiver filled in a survey three times in the first year, the last one six months into maintenance, and that twelve-month survey is the one that matters here.⁹

Study 4 · United States, published 2014

Couples who separated or divorced during the first year of leukemia treatment

Two ways of counting the same families, out of those who answered the twelve-month survey.

Started out married or living together14 of 106 couples

Started out married9 of 91 couples (a subset of the row above)

Percentages as rounded in the paper. There was no comparison group of families without a sick child. Source 9, Table II, twelve-month assessment.

Of the 106 couples who started out married or living together and answered at twelve months, 14 had separated or divorced. That is 13 percent, about one couple in eight. Counting only couples who were married at the start, it was 9 of 91, or 10 percent.⁹

What does that mean? First, it is a real number for a hard year, and I would not soften it. Second, the study had no comparison group of families without a sick child, so it cannot say whether one in eight is more than those same couples would have seen anyway. The authors set the 10 percent next to a national figure of 7.4 percent and called the difference not significant, but the Census table they used measures who is currently divorced, not how many couples split up in a year, so I do not lean on that comparison.^{9, 10} Third, the same families reported far more upheaval in every other part of life: two-thirds cut their work hours, nearly half had a parent lose or leave a job, and a quarter moved house.⁹

Put this next to the register studies and here is the picture I get. The first year shakes everything, work, money, where you live, and some marriages. Followed for up to ten years, families who went through it ended up separating and divorcing at the same rate as everyone else. The first year is the storm; it does not decide the outcome.

Study 5: Switzerland, parents of survivors about 24 years later, published in 2019

The Swiss study asks a different question: long after treatment, were the parents still together? Researchers surveyed 784 parents of 512 people who had survived childhood cancer, on average 24 years after the diagnosis, and compared them with 471 parents from the general population, weighted so the two groups matched in age, sex and background.⁸

Study 5 · Switzerland, published 2019

Parents who were divorced or separated at the time of the survey, about 24 years after diagnosis

One survey, two groups of parents, asked the same question at one moment.

Parents of long-term survivors784 parents of 512 survivors

Comparison parents from the general population471 parents, weighted for age, sex and migration background

Current status, not a count of marriages that ended after diagnosis. Source 8, Results, “Civil status.”

Nine percent of the survivors’ parents were divorced or separated when they answered, against 17.5 percent of the comparison parents. Fewer, not more, by about half.⁸ Two cautions before reading too much into that. The survey went to parents of survivors only, so families whose child died are not in it. And it counts who was divorced on the day they answered, not every marriage that ever ended; a parent who divorced and remarried would not be counted. Even so, I think the direction is hard to argue with. Decades on, the parents who answered were not more likely to be divorced than their neighbors. They were less.

Here is the saying next to those two numbers.

The saying next to the Swiss numbers

Out of every 100

What the saying claims
90 of 100 marriages end

Parents of survivors, Switzerland
9 of 100 divorced or separated

Other Swiss parents, same study
17.5 of 100 divorced or separated

The saying is about marriages ending; the survey counted who was divorced or separated on the day they answered. They are not the same measurement, and the gap is still not subtle. Source 8.

The claim and the survey are not measuring the same thing, and I have said so in the caption. I still think the picture is worth a look, because the gap between 90 and 9 is not a matter of definitions.

Studies 6 and 7: two smaller ones that point the same way

A German survey in 2024 asked 74 parents and found 12.2 percent divorced or separated, against a general-population figure of 14 percent. It is small, and I could not check how the comparison figure was built.¹¹ A 1978 study from Kansas and Missouri, the oldest on this page, followed the families of 191 children and found a divorce rate of 1.19 percent per year, against 2.03 percent per year among married parents in the same two states. It is old, and I could not verify its details beyond the abstract.⁷ Neither carries much weight on its own. Both land on the same side as the big studies.

After a child dies

Some families reading this have lost a child, and I want to treat that question separately and gently. The Norwegian study’s last two rows are the best evidence I found: overall, no increase in divorce among parents whose child with cancer died (1.01), and an increase in one subgroup, mothers with a college education (1.31), compared with cancer-free families of the same education.² A Swedish study found 449 parents four to nine years after losing a child to cancer and compared them with 457 matched parents; the bereaved parents were no less likely to still be with the child’s other parent.¹⁴

Looking beyond cancer, the picture is less reassuring. A 2013 Norwegian study of couples who lost a child to any cause found divorce 1.61 times as often, and a Finnish register study described a more modest effect.^{16, 17} Neither is specific to cancer, and I include them because leaving them out would suggest every study found nothing. A Finnish study from 2000 is sometimes cited for more divorce after a child dies in hospice-style care; the finding is not in its abstract and I could not verify it, so it stays a lead.¹⁵

The year of diagnosis itself

One more study deserves a mention. A 2024 Finnish working paper, listed as forthcoming in a demography journal, found that separations dropped sharply in the year a child was diagnosed with cancer, a ratio of 0.27, and then returned to the usual level afterward.^{18, 23} The paper’s counts don’t fully agree with each other yet, so I am waiting for the final version before leaning on it. It fits what I have seen in the hospital: in the first year, most couples close ranks.

What parents describe, and what helps

The numbers above count marriages. They say nothing about how a marriage feels, and the studies that ask are worth reading alongside them. In a survey of 192 parents at three cancer centers in the US and Canada, about 40 percent said the relationship had gotten worse since the diagnosis.⁴ Interview studies in Australia and Sweden, one of them recruiting parents specifically because they were struggling, describe exhaustion, lost intimacy and living like roommates.^{12, 13} Those studies cannot say how common that is, only that it is real.

On what helps, the evidence is thin and I will not oversell it. Parents in the same 192-parent study wrote that listening, sharing the load rather than one parent carrying it, and accepting help from other people were what kept them going.^{19, 4} In Sweden, 22 couples who went through a series of family conversations led by a social worker said they understood each other’s experience better afterward.²⁰ A small US pilot that randomized couples to relationship-skills phone sessions had mixed results and was not designed to test whether it prevented divorce.²¹

None of that proves counseling saves marriages. It does say that asking for help is a normal thing to do in this situation, and the National Cancer Institute’s advice to families is to ask the care team where to find it.²² Some of the programs listed under Find help offer counseling for parents.

What this can’t tell you about your family

A study counts thousands of families at once, and yours is one. A register can tell whether a couple stayed married; it cannot tell whether they felt close, or supported, or seen. The surveys can describe pain without showing that a divorce is coming.^{1, 2, 4, 13}

The thing I feel sure about is narrow: nothing in the research supports the 90 percent claim, and the strongest studies point the other way.^{1, 2, 3} If things at home feel strained, that is a common response to an uncommon situation, and the social worker on your child’s team is a reasonable first person to talk to.

Where I read this

1. Mader L, et al. The impact of childhood cancer on parental separation, divorce, and family planning in Denmark. Cancer 2020. PubMed · author-posted paper. Where: Methods, Statistical analysis (printed p. 3332); Results, Parental separation and divorce (p. 3334); Figure 2 and Table 2 (pp. 3335 to 3337). Original-paper text was checked in an author-posted ResearchGate copy. The within-cancer risk-factor analyses start three years after diagnosis. Exact absolute-risk graph values were not extracted.
2. Syse A, Loge JH, Lyngstad TH. Does childhood cancer affect parental divorce rates? A population-based study. Journal of Clinical Oncology 2010. PubMed · author-posted paper. Where: Table 1 (overall divorce odds); Table 2 (ALL and renal cancer); Table 3 (child vital status); Table 4 and footnotes (maternal-education strata); Discussion (multiple comparisons). Original-paper text was checked in an author-posted ResearchGate copy. The 0.99 estimate and 123 divorces are for ALL, not all leukemia.
3. Grant S, et al. Parental separation and pediatric cancer: a Danish cohort study. Pediatrics 2012. PubMed. Where: PubMed abstract, Methods and Results (2,450 versus 44,853 children; adjusted rate ratio 1.00, 95% CI 0.91 to 1.10). Full methods were not retrieved. Exact diagnosis years, follow-up and absolute rates remain unverified.
4. Wiener L, et al. The perceived influence of childhood cancer on the parents' relationship. Psycho-Oncology 2017. free full text (PMC). Where: Patients and Methods; Results, Relationship quality. The retrieved indexed PMC text describes 192 parents and retrospective perceived change in a one-time questionnaire. Direct PMC reading was blocked during this review; this is not a prospective divorce estimate.
5. Frogge S. The Myth of Divorce Following the Death of a Child. Tragedy Assistance Program for Survivors, 2015. Article. Where: Myth Origins section. This is the secondary attribution to Schiff's 1977 book, not an independently inspected book passage or a verified first appearance of the cancer-specific claim.
6. Schwab R. A child's death and divorce: Dispelling the myth. Death Studies 1998. Abstract. Where: publisher abstract. Review of the bereavement-and-divorce literature, not a newly measured rate. The abstract was retrieved through the publisher's indexed text; full text was not retrieved.
7. Lansky SB, Cairns NU, Hassanein R, Wehr J, Lowman JT. Childhood cancer: parental discord and divorce. Pediatrics 1978. PubMed. Where: PubMed abstract, questionnaire sample and person-year divorce rates (1.19% versus 2.03%). Exact calendar years, event counts and comparator construction were not verified from full text.
8. Mader L, et al. Relationship status and quality of the partner relationship in parents of long-term childhood cancer survivors: the Swiss Childhood Cancer Survivor Study-Parents. Psycho-Oncology 2019. PubMed · free accepted manuscript. Where: accepted manuscript, Methods, Study population and Measurements (PDF pp. 2 to 3); Results, Civil status (PDF p. 4); Table 1 (PDF p. 7). Results and Table 1 were visually inspected. Comparison percentages are weighted; survivor-parent and comparison surveys were conducted in different years.
9. Lau S, et al. Family life events in the first year of acute lymphoblastic leukemia therapy: a Children's Oncology Group report. Pediatric Blood & Cancer 2014. free full text (PMC). Where: Participants and Procedures; Measures, Family life events; Results and Table II, time point 3; Discussion and reference 32. Indexed primary PMC text supplied the methods and table cells, but direct PMC reading and a complete paper PDF could not be opened during this review. The 46% job-loss denominator is internally ambiguous; the 82% figure is analyzed participation, not consent. See the review findings for access limits.
10. U.S. Census Bureau. American Community Survey 2011 one-year estimates, subject table S1201: Marital Status, variable definitions. Table metadata. Where: 2011 S1201 variable definitions, current marital-status columns, including Divorced (C04). The metadata were opened. They do not define new divorces over one year. The paper's reported 7.4% was not reproduced as a comparable annual incidence.
11. Gajda J, Johns C, Zimmermann T. Impact of pediatric cancer on parents' relationships. European Journal of Oncology Nursing 2024. PubMed. Where: PubMed abstract, Methods and Results (74 parents; 12.2% versus 14%). Full recruitment details, survey timing and the benchmark's definition were not verified. This is not plotted alongside the Swiss survey.
12. Chow WK, et al. “Like ships in the night”: a qualitative investigation of the impact of childhood cancer on parents' emotional and sexual intimacy. Pediatric Blood & Cancer 2022. PubMed · full publisher article. Where: full publisher article, Methods 2.1 and Results 3.1 (48 currently cohabiting parents; noncohabiting parents excluded), and themes in Results. Interviews after treatment describe experience, not divorce incidence.
13. Kuylenstierna A, et al. “It is like he abandoned me in a war”: a qualitative study of relationship distress in parents of children diagnosed with cancer. European Journal of Oncology Nursing 2026 (online 2025). PubMed · free full text. Where: full paper, Abstract and Methods 2.2 to 2.4 (25 parents recruited for relationship distress; interviews November 2023 to March 2024); PDF p. 1 publication information. Online November 22, 2025; journal volume dated 2026. This selected sample cannot estimate prevalence.
14. Eilegård A, Kreicbergs U. Risk of parental dissolution of partnership following the loss of a child to cancer: a population-based long-term follow-up. Archives of Pediatrics & Adolescent Medicine 2010. PubMed · first-page preview. Where: original report's first page, p. 100, Methods, Results and table, retrieved in an author-posted ResearchGate preview. 449 bereaved and 457 comparison respondents, four to nine years after death. The outcome is remaining with the child's other parent at the survey. The complete report was not retrieved.
15. Sirkiä K, Saarinen-Pihkala UM, Hovi L. Coping of parents and siblings with the death of a child with cancer: death after terminal care compared with death during active anticancer therapy. Acta Paediatrica 2000. PubMed. Where: PubMed abstract (parents of 60 children after terminal care and 26 after active therapy; Helsinki, deaths in 1987 to 1992). The abstract does not report the divorce finding attributed to this study; the full result remains unverified.
16. Lyngstad TH. Bereavement and divorce: does the death of a child affect parents' marital stability? Family Science 2013. Abstract · author-posted paper. Where: Data and methods; Results 4.1, Overall effect of bereavement (adjusted odds ratio 1.61, 95% CI 1.49 to 1.73); Table 2. Original-paper text was checked in an author-posted ResearchGate copy. First marriages begun in 1970 to 2003; 4,170 bereaved couples. Cause of death was unavailable.
17. Finnäs F, Rostila M, Saarela J. Divorce and parity progression following the death of a child: a register-based study from Finland. Population Studies 2018. PubMed. Where: PubMed abstract, Methods and Results (Finnish registers, 1971 to 2003; more than 100,000 married couples; modest effect). All-cause child bereavement, not a cancer-specific analysis. Exact effect sizes were not verified.
18. Dierker P, et al. Parental separation risk before and after the diagnosis of a child physical health condition. MPIDR Working Paper WP-2024-011. Working paper (PDF). Where: indexed working-paper abstract and Table A3, diagnosis-year row in the life-threatening-condition category (cancer). The 0.27 point estimate was retrieved, but the complete PDF and its 0.10 to 0.71 interval could not be independently checked in this review. Count discrepancies are reported by the supplied research audit, not newly verified against the PDF. Source 23 supplies publication status.
19. de Arruda-Colli MNF, et al. In good times and in bad: what strengthens or challenges a parental relationship during a child's cancer trajectory? Journal of Psychosocial Oncology 2018. free full text (PMC) · PubMed. Where: PubMed abstract, Methods and Results; indexed PMC methods. Qualitative analysis of open-ended questionnaire responses from the same 192-parent sample as source 4, not an independent set of interviews. Direct PMC access was blocked.
20. Holm M, Lundberg T, Lövgren M, Ljungman L. Parenting a child with cancer and maintaining a healthy couple relationship: findings from the Family Talk Intervention. Pediatric Blood & Cancer 2024. PubMed · free full text. Where: full accepted paper, Methods 2.1 to 2.3 and Results; interview sample of 22 couples after the Family Talk Intervention. One Swedish center; recruitment September 2018 to August 2019. No controlled divorce outcome.
21. Porter LS, Baucom DH, Bonner M, Linardic C, Kazak AE. Parenting a child with cancer: a couple-based approach. Translational Behavioral Medicine 2019. free full text (PMC) · full publisher article. Where: full publisher article, Statistical analyses, Participants, and Enrollment and retention. There were 21 couples in the reported baseline sample; the flow narrative also reports 28 agreeing and seven withdrawing. This was an exploratory feasibility pilot without statistical tests of between-group differences or proof of divorce prevention.
22. National Cancer Institute. Support for Families: Childhood Cancer. Accessed September 23, 2026. Page. Where: Ways to help parents whose child has cancer, professional-help guidance. Describes asking the care team or primary doctor about referrals. Does not establish availability, eligibility or insurance coverage for a particular family.
23. Dierker P. Research publication list. Accessed September 23, 2026. Page. Where: Publications entry for Parental Separation Risk Before and After the Diagnosis of a Child Physical Health Condition, marked Demography, Forthcoming. Checked September 23, 2026. A final journal article and DOI were not verified.
24. NIST/SEMATECH. e-Handbook of Statistical Methods, section 1.3.5.2, Confidence Limits for the Mean. Handbook section. Where: opening explanation of interval uncertainty, precision and repeated-sampling coverage; Definition: Hypothesis Test. Used only for the general confidence-interval explanation, not to calculate or replace any study's intervals.