OralCheck

Scoring Methodology

OralCheck assigns a risk weight to each answer based on published odds ratios. This page documents how that scoring works, which sources it's built on, and where the model has known limitations.

Weight Derivation

Each risk factor is assigned a weight proportional to the natural log of its published odds ratio (OR), scaled by a constant k:

weight = round( ln(OR) × k )

The scaling constant k = 4.47 was chosen so that daily tobacco use (the highest-weighted modifiable factor, OR 6.0× per Gandini et al. 2008) maps to a weight of 8. That anchors all other weights to a consistent ordinal scale from the same evidence base.

Where studies report a range of ORs, we use the midpoint or a conservative estimate. Symptoms like erythroplakia, leukoplakia, and non-healing ulcers work as clinical overrides rather than additive scores. They're better understood as possible in-situ pathology than as population-level risk exposures, so they're treated differently in the model.

Risk Factor Weights

FactorPublished ORWeightSource
Tobacco (daily)
Anchor for the scaling constant k = 4.47
2.5 – 6.0×8Gandini et al., Oral Oncology, 2008
Betel quid / paan / gutka (current)
IARC Group 1 carcinogen independent of tobacco
7 – 10×9IARC Monograph 85, 2004
Tobacco (occasional)~3.0×5Gandini et al., Oral Oncology, 2008
Alcohol (daily)
IARC Group 1 carcinogen
~3.0×5Bagnardi et al., Annals of Oncology, 2015
HPV-related history
Conservative estimate; OR ~15× for oropharyngeal specifically
3 – 5× (blended)5Gillison et al., JAMA, 2008
Age 65+
Median age at diagnosis is 62
~4.0× (adjusted)6SEER, NCI; multivariable-adjusted
Age 55 – 64~2.5×4SEER, NCI
Betel quid (past use)~2.5×4IARC Monograph 85, 2004
Alcohol (weekly)~2.0×3Bagnardi et al., Annals of Oncology, 2015
Family history (first-degree)~2.0×3General epidemiological consensus
Diet low in fruit/vegetables
Antioxidant-protective effect
~2.0×3Pavia et al., Oral Oncology, 2006
Age 35 – 54~1.5×2SEER, NCI
Tobacco (former)
Risk declines ~50% within 5 years of cessation
~1.5×2Gandini et al., Oral Oncology, 2008
HPV (not vaccinated)~1.5×2Gillison et al., JAMA, 2008

Tobacco + Alcohol Interaction Term

When both tobacco and alcohol use are present at meaningful levels (daily or weekly), the model adds an interaction bonus of +3 points.

The reason is that tobacco and alcohol together produce a multiplicative rather than additive increase in oral cancer risk. Combined regular use raises risk roughly 15× above baseline, while simply summing the individual weights only gets you to about 9×. The +3 captures that gap.

Source: Bagnardi V, et al. Annals of Oncology, 2015.

Risk Tier Thresholds

Low
≤ 4
Moderate
5 – 13
Elevated
14 – 22
High
≥ 23

The maximum possible score is approximately 53 (all highest-risk answers plus the interaction bonus). A few reference points: a daily smoker alone scores 8 (moderate); tobacco + alcohol + interaction lands at 16 (elevated); betel + tobacco + alcohol + interaction reaches 25 (high).

Regardless of score, any symptom lasting 2+ weeks (a persistent sore, red or white patch, unexplained lump, or difficulty swallowing) overrides the tier to High and triggers a prompt for immediate evaluation.

Known Limitations

  • 1OralCheck hasn't been validated against a clinical outcome dataset. The weights are grounded in published literature, but they haven't been calibrated against a prospective cohort.
  • 2The screener combines oral cavity and oropharyngeal cancer risk into a single score. These are epidemiologically distinct entities: HPV-16 is primarily a driver of oropharyngeal cancer, while tobacco and alcohol dominate in oral cavity cancer.
  • 3Biological sex isn't captured. Men have roughly 2× the oral cavity cancer incidence of women, and that's a gap in the current model.
  • 4Immunosuppression (HIV, organ transplant recipients, long-term corticosteroid use) is linked to significantly elevated oral cancer risk and isn't currently asked about.
  • 5The screener relies on self-reported data. Tobacco and alcohol use are commonly under-reported.
  • 6Dental visit frequency is included as a detection-delay proxy, not as a causal risk factor with a well-characterized OR. Its weight is based on clinical rationale rather than direct epidemiological derivation.

References

  1. 1.Gandini S, et al. Tobacco smoking and cancer: a meta-analysis. Oral Oncology. 2008;44(7):617–638.View on PubMed / publisher →
  2. 2.Bagnardi V, et al. Alcohol consumption and site-specific cancer risk: a comprehensive dose–response meta-analysis. Annals of Oncology. 2015;26(1):39–55.View on PubMed / publisher →
  3. 3.Gillison ML, et al. Distinct risk factor profiles for human papillomavirus type 16–positive and human papillomavirus type 16–negative head and neck cancers. JAMA. 2008;168(3):294–305.View on PubMed / publisher →
  4. 4.International Agency for Research on Cancer. Betel-quid and Areca-nut Chewing and Some Areca-nut Derived Nitrosamines. IARC Monograph 85. Lyon, France: IARC; 2004.View on PubMed / publisher →
  5. 5.Napier SS, Speight PM. Natural history of potentially malignant oral lesions and conditions: an overview of the literature. Journal of Oral Pathology & Medicine. 2008;37(1):1–10.Leukoplakia 5–17% and erythroplakia 14–50% malignant transformation ratesView on PubMed / publisher →
  6. 6.Pavia M, et al. Evidence-based medicine on the relationship between diet and cancers of the oral cavity and pharynx. Oral Oncology. 2006;42(1):15–25.View on PubMed / publisher →
  7. 7.National Cancer Institute. SEER Cancer Statistics Review 1975–2021. Surveillance, Epidemiology, and End Results Program.View on PubMed / publisher →