Fifteen problems in the Kansas City metro, rank-ordered by the effective altruism framework: impact × tractability × neglectedness. Impact is measured, in annual disability-adjusted life years built from county death certificates and CDC prevalence data. Tractability comes from verified intervention cost-effectiveness and Missouri's actual legal room to act. Neglectedness comes from mapping who currently funds what, in dollars. Each factor gets its own section below, a 10,000-draw Monte Carlo over the uncertainty turns the ranking into rank intervals, and every score expands into the evidence and citations behind it.
The impact figures are measured; the tractability and neglectedness scores are structured judgment over verified evidence, and each one states its reasoning where it is scored. Every claim cites its source at the bottom of the page.
1 · P(top 3) = 100%
Quitlines at $849–2,358/QALY meet the most extreme neglect in the study: Missouri funds tobacco control at 7.4% of the CDC-recommended level.
2 · P(top 3) = 93%
Naloxone at $438/QALY, a 2–5× mortality gap for people in vs out of MOUD treatment, and settlement money sitting undeployed while KC lags the national decline.
3 · P(top 3) = 65%
Cost-saving interventions, an evidence-based local program running at a deficit with no funder behind it, and the least-funded cause of death in America relative to its mortality.
4 · P(top 3) = 28%
$2,800–15,000/QALY levers against the region's single largest burden, deliverable through safety-net clinics with no policy fight.
One number that integrates all three ITN factors: if an additional $10M/yr showed up in Kansas City, how many disability-adjusted life years would it buy against each problem, through the best verified locally-feasible intervention, at current levels of crowding? Whiskers are the 10th–90th percentile of the Monte Carlo.
Each dot is a problem's median rank across 10,000 Monte Carlo draws; the line is its 10th–90th percentile rank range. Tight lines mean the ranking is confident; long lines mean reasonable people can land in different places. Violence has the longest line in the study.
Effective Altruism (or "EA") is a philosophy that tries to answer the question "How can we do the most good?"
Having good intentions does not mean you are doing good. In fact, lots of charities cause active harm without even realizing it. Furthermore, every time you decide to do one thing, you are actively choosing to not do everything else. This is called your "opportunity cost"; failure to do a better thing has a cost in all the good you could be doing and are choosing not to.
Having data is the only way to even begin getting around these problems. It is not a guarantee that you will make the right decision, but it's the only way you will ever be able to move from a worse decision to a better one.
Now, to be clear: that does not mean your life should be entirely in the service of doing the Most Good. The good life involves many different things, and I believe this is only one of them. I recommend treating EA as more of a tool. Use the tool to do good as far as you feel comfortable, and then stop. The most effective form of good is the one you will sustain, after all.
As for how EA approaches doing good, we first start with asking the question "What exactly are the most important problems?" We do not take for granted that we already know the answer to this. We grade all problems using Impact (how much of a burden the problem is, usually measured in Disability Adjusted Life Years), Tractability (how easy it is to start making headway on the problem), and Neglectedness (how few resources, usually measured in money, are currently going toward the problem). The higher a problem scores on any of these three metrics, the higher the priority it is, indicating where a marginal dollar will go the furthest. The cool thing about these metrics is that they change over time in response to our changing circumstances; as we start fixing a problem, its impact goes down. As we start picking the low hanging fruit, its tractability goes down. As we funnel more money and resources to the problem, its neglectedness goes down.
Ultimately, the best thing you can do is donate to developing nations. Malaria nets, unconditional cash transfers, deworming, etc. These are all extremely cheap, thoroughly researched, and do an outsized amount of good. Attempting to give money to people in first world nations tends to not go very far; we are already so wealthy, and our problems are extremely complex. But for those times when there is money earmarked for local charity, I want that money to go where it can do the most good.
The measured factor, and the one this whole ranking sits on. Every major problem in the metro quantified in annual disability-adjusted life years (DALYs) and in dollars, with 80% confidence intervals, for the five instrumented counties: Jackson, Clay and Platte in Missouri, Wyandotte and Johnson in Kansas. 1.88 million people, about 85% of the MSA.
Total annual health burden
~620K
DALYs per year, 5-county core (envelope 570–700K). The fifteen rows below are the largest endpoints and sum to 567K; the rest is causes too small to itemize.
Excess years of life lost before 75
62,160
every year, vs the Johnson County benchmark 20 minutes away
County life-expectancy gap
6.9 yrs
Johnson Co. 80.64 vs Wyandotte Co. 73.70 (CHR 2020–22)
Tangible economic damage
~$14B
per year, excluding VSL valuations. The sum of the non-overlapping rows in the damages chart below; the hatched rows are not added to it.
Annual DALYs by problem, split into years of life lost to death (YLL) and years lived with disability (YLD). Whiskers are 80% confidence intervals. The KC/US column is the local rate divided by the national rate: the "is Kansas City distinctively bad here?" number. Hover or focus any row for details.
Excess premature mortality (years of potential life lost before age 75, YPLL-75) against two benchmarks. Against the US average the metro nets out to almost exactly zero. Against its own healthiest county it loses 62,000 years of life annually.
Tangible annual costs (medical, productivity, property, criminal justice) in 2025 dollars. Value-of-statistical-life and quality-of-life valuations are deliberately excluded: they are the dollar shadow of the DALY column, and including both counts the same suffering twice. Hatched bars overlap the disease rows above them (tobacco's cost lives inside CVD, cancer and COPD) and are never summed.
Child poverty's $5–7B/yr is a long-run future cost (NAS Roadmap apportioned to ~60K local poor children), not current cash flow, and is listed separately from the annual-damage rows.[17] Comprehensive VSL-inclusive figures, for comparison against advocacy numbers: violence ~$3.5B, overdose ~$6B, traffic ~$8.9B.
Deviation from the national baseline matters more than absolute size for local action, because it is the part a local actor can plausibly close. Eight places KC measurably departs from the country:
Read that list against the ranking and one thing stands out: the places KC is most distinctively bad (homicide, the missed overdose recovery) are not the places with the most total burden, and two of the eight (homelessness, evictions) are structural drivers that sit underneath the disease rows rather than inside them. Distinctiveness tells a local actor where the gap is; it does not by itself say the gap is closable, which is what the next section is for.
The first judgment factor. A 0–10 score built from four things: the verified cost-effectiveness of the best locally-feasible intervention, the strength of the evidence behind it (CPSTF- or RCT-verified beats contested beats absent), Missouri's legal room to act, and whether Kansas City has tried it before and what happened. High burden with nothing deployable scores low, which is the entire point of the factor.
Cost per quality-adjusted life year for the best locally-feasible intervention, on a log axis. Bands are the range the source publishes; a dot marks a published point estimate, and rows with a band alone are ranges the source gives without one. Dollar years differ by source and are not inflation-harmonized, so read the decade, not the digit. For scale, the conventional US willingness-to-pay benchmark sits around $100,000 per QALY, the threshold the diabetes evidence below is measured against.
Eight of the fifteen problems have no usable cost-per-QALY for a locally-deployable intervention. The reasons are not the same, and the difference matters more than the missing number.
Two states, one metro, and the same intervention can be legal on one side of a street and void on the other. This is the single largest structural constraint on tractability in Kansas City, and it is the reason policy-shaped effort belongs on the Kansas side or in Jefferson City rather than in a KCMO ordinance campaign.
| Lever | Missouri | Kansas |
|---|---|---|
| Local firearm regulation | Totally preempted. RSMo 21.750: the state "occupies and preempts the entire field of legislation touching in any way firearms."[33] | Not the binding constraint on the Kansas side of the metro. |
| Local tobacco tax | Void. RSMo §149.192 freezes existing local taxes at 1993 levels; the 17-cent state tax needs a ballot initiative.[37][38] | Constitutional home rule, proven when Topeka's Tobacco 21 ordinance survived court challenge.[39] |
| Syringe services | Not authorized. No SSP exemption in statute; legalization bills have sat in committee since 2021.[40] | Not separately verified. |
| Local minimum wage | Preempted by RSMo 290.528 (2017). The older RSMo 67.1571 is still widely cited but was struck down as a single-subject violation in Cooperative Home Care v. City of St. Louis.[41] | Not separately verified. |
| Paid sick leave | Repealed by HB 567 in 2025, nine months after voters passed it.[41] | Not separately verified. |
| Automated traffic enforcement | Constrained since Tupper; statewide ban bills pending.[34] | Not separately verified. |
| Fentanyl test strips, naloxone standing order | Legal since 2023.[40] | Not separately verified. |
Notice which way the asymmetry cuts. Every preempted lever above is a policy lever, and every intervention that survives into the top of the ranking is programmatic: a quitline, a naloxone supply, a home visit, a blood-pressure cuff. Preemption does not just remove options, it systematically selects for the kind of work a donor can fund and against the kind a voter can pass.
One tractability score deserves its own paragraph, because the history is specific and recent. Kansas City ran the celebrated focused-deterrence playbook once already: KC NoVA cut homicides in 2014, was declared a success, decayed to nothing by year three, and was formally rated Ineffective by the National Institute of Justice; KCPD withdrew in 2018.[42] SAVE KC, today's version, reports encouraging numbers from exactly the same early stage, with no independent evaluation. The expected-value ranking already discounts for this. What survives that history is a value-of-information play: fund rigorous independent evaluation of SAVE KC, and watch 2026–27, the window where a NoVA-style decay would first become visible.
The second judgment factor, and the one that most often decides the ranking. Scored at the prevention margin: annual local dollars aimed at stopping the problem, not at treating people who already have it, per DALY of burden. Treatment spending is reported but does not count as crowding, because a fully-staffed cardiology service tells you nothing about whether anyone is funding blood-pressure control.
Annual dollars visible against each problem, from verified budgets and 990 filings. This is deliberately not a like-for-like comparison and should not be read as one: the rows differ in geography (state, county, city, metro philanthropy) and in how much of the money is prevention rather than treatment or response. It is a map of where money exists at all, and the informative rows are the ones at zero.
Two things fall out of that chart. The largest single flow, the opioid settlement, is the one least constrained by scarcity and most constrained by deployment: the pool roughly doubled to ~$900M over fifteen years while 50-plus Missouri jurisdictions spent nothing at all in 2025 and Jackson County left ~$4M unspent.[43] And the two rows at zero are not small problems. Elderly falls is second in the factor product and third on the headline metric, and no Kansas City funder targets it.[44] Read that row carefully, though, because the follow-on charity vetting sharpened it: there is no dedicated falls funder, but there is a fundable vehicle, an evidence-based falls program running inside an organization posting three straight years of operating deficits.[44] Tobacco, the reverse case, has state program money in the chart above and no clean local giving vehicle at all. "Most neglected cause" and "best gift you can make this month" are not the same question.
State public health funding per capita. Missouri is last of all fifty states, at less than half of 49th-place Indiana and about a fifth of Kansas.
That is what neglectedness looks like from the inside: not an absence of money in Kansas City, but money pointed somewhere else, for reasons each institution can defend on its own terms.
Every problem gets the same three-part breakdown. Each factor carries a point estimate, the interval the Monte Carlo draws from, and its own contribution to the factor score (log10 I + T/2 + N/2), on a scale where any one factor can contribute at most 5 points. The list is ordered by that point score, which is not quite the same as the Monte Carlo median rank in the data table: musculoskeletal sits 6th here and ties for 7th there, because its interval overlaps mental illness's. "Evidence" opens all three: a separate paragraph of reasoning and citations for impact, for tractability, and for neglectedness. Watch the bar lengths across the fifteen rows: because impact enters as a logarithm it spans barely a single point across all fifteen (3.95 to 5.00), while tractability spans a full three (1.00 to 4.00) and neglectedness spans nearly that (1.75 to 4.50). Burden sets the scale of the prize; the two judgment scores do almost all of the ordering.
A sketch, not a plan. Splitting $10M across the top four: roughly $3M for a quitline and cessation surge (media plus free nicotine replacement, Missouri-side), $3M for naloxone saturation, low-barrier medication access, and settlement-deployment advocacy, $2M for falls programs (CAPABLE and Otago through senior centers and Medicaid waivers), and $2M for self-measured blood pressure and team-based hypertension care through the safety-net clinics. Central estimate on the model's own numbers: 1,500–2,500 DALYs averted per year at scale, with the falls and cessation lines partly cost-saving to Medicare and Medicaid. Every line is deliverable by organizations that already exist here, without a single preempted policy fight.
Impact (annual DALYs) is the burden model in the Impact section above[57]. Local DALYs per cause = the US cause-specific rate from WHO Global Health Estimates 2021, which publishes years of life lost and years lived with disability separately[68], × the five-county core population (Jackson, Johnson, Clay, Wyandotte, Platte; ~1.88M) × a local adjustment factor. The adjustment comes from a real local instrument wherever one exists: 2023 death counts by cause from Missouri and Kansas vital statistics[25], CDC PLACES prevalence with published CIs[24], KCPD's homicide analysis[18], and verified county overdose and traffic counts[20][26]. Where no instrument exists (musculoskeletal, digestive, neurological, falls) the adjustment defaults to 1.00 with a wider interval and a lower confidence grade, and each problem's impact paragraph above says which case it is. Intervals are 80%.
Tractability (0–10) anchors to the best locally-feasible intervention's verified cost-effectiveness on a log scale ($500/DALY ≈ 9, $5K ≈ 7, $50K ≈ 5, nothing proven ≈ 1–2), adjusted for evidence strength (CPSTF- or RCT-verified beats contested beats absent), Missouri's legal room to act, and KC-specific precedent.
Neglectedness (0–10) is scored at the prevention margin: annual local dollars aimed at reducing the problem, not treating it, per DALY of burden, from verified budgets (city, county COMBAT tax, settlement flows, HUD awards) and philanthropy (990-verified), benchmarked against national funding-versus-burden residuals. Treatment spending is reported but does not count as crowding for prevention.
Putting the three on one scale. The factor score is log10 I + T/2 + N/2, so each factor's contribution is directly readable and directly comparable: the bars in the list above are all drawn against the same 5-point ceiling. Tractability and neglectedness reach that ceiling by construction (10/2); impact would reach it at 100,000 DALYs, and the largest row in the study is 99,000. The compression is deliberate, not a bug: a logarithm says a problem ten times larger is worth about one extra point, not ten times the priority, because absolute burden is a weak guide to what a marginal dollar can move. The consequence is worth stating plainly, since it means the ranking mostly turns on the two scores I assigned rather than the one I measured. That is why tractability and neglectedness each carry an interval, a stated anchor, and their own cited paragraph.
Framework note. WHO GHE uses a frontier life table, so YLL-per-death runs higher than IHME GBD conventions. Everything here shares one framework, so ranks and ratios are internally consistent; absolute DALYs run ~10–15% above what an IHME-based version would report. IHME's own US ranking (CVD > cancer > musculoskeletal > mental disorders) matches this table's top four.[69]
Double-counting rules. The league table holds mutually exclusive disease and injury endpoints. Risk factors and structural drivers (smoking, obesity, alcohol-as-risk, air quality, poverty, housing) overlap those endpoints; they are shown hatched in the damages chart and never summed with it. Value-of-statistical-life and quality-of-life valuations are excluded from the dollar figures for the same reason: they are the dollar shadow of the DALY column.
Known limitations of the burden model. Kansas blocks county-level cancer-registry data by statute, so Wyandotte cancer leans on state rates. Kansas-side overdose and suicide counts are estimates. Musculoskeletal, digestive, neurological and falls have no local instrument. 2021 base rates embed COVID-era and peak-violence conditions, which makes the violence and drug KC/US ratios conservative. Scaling from the 5-county core to the full 2.2M MSA adds roughly 15–17% to most rows and changes no ranks.
The Monte Carlo draws every factor from a triangular distribution over its range, 10,000 times (fixed seed), and records each problem's rank per draw. The full simulation is a 60-line stdlib Python script that reproduces every number in the table above. Two scoring views run in parallel: marginal DALYs per $10M (the headline) and the factor product (the per-problem list above). Where they disagree, the disagreement is the finding: falls sits near the top of the factor product but its modest burden caps absolute gains; musculoskeletal disease scores high on burden-times-neglect but has no deployable intervention, making it a research bet, not a program bet.
Post-vetting update. A follow-on pass vetting Kansas City's actual charities fed back into three neglectedness scores, each moved by half a point, the granularity the whole scale is scored at: falls 9.5 to 9.0 (delivery infrastructure exists after all, small and deficit-funded), tobacco 8.5 to 9.0 (no local giving vehicle exists at all for the top cause), overdose 6.0 to 6.5 (the metro's joint federal harm-reduction grant lapsed in May). Re-running the simulation: View A is unchanged in every number, which is expected rather than reassuring, since View A ranks by marginal DALYs per $10M and never reads N at all. View B's only change is tobacco and falls swapping first and second, and every median stays inside the previously published rank intervals. For a donor, note the asymmetry the scores cannot hold: tobacco now ranks first in both views but has nothing to fund without building the channel, while falls has a starving, fundable vehicle today.
What this is not. Local institutions' own priorities entered only as crowdedness data, never as conclusions. The cost-per-QALY figures are not inflation-harmonized across sources, so they support comparisons by order of magnitude, not by digit. The T and N scores are my judgment over the verified evidence, with the reasoning for each one stated in full above so it can be argued with specifically. A state-level actor's ranking would differ (alcohol and tobacco taxes jump); so would a researcher's (musculoskeletal jumps).
Numbered as cited in the capsules and text. Flags mark anything that resisted primary verification; those claims are labeled in place.
Researched and modeled by Alex Hedtke. The interactive original, the published Monte Carlo code, and the companion giving guide live at becomingstronger.github.io. Corrections welcome at info@eakansascity.org.
