SCI.help › Privacy

Privacy

People come here to read about bladder programs, pressure injuries, depression, and sex after injury. That is nobody else's business. This page says exactly what the site records, in plain language, and gives you a button to turn it off.

The short version

SCI.help counts how many people read each page, and groups a month's activity together so it can tell one person reading six pages from six people reading one. There are no cookies, no advertising trackers, no Google Analytics, no Facebook pixel, and no third-party analytics company of any kind. Nothing identifying is stored on your device. Your IP address is never written down. Nothing here is kept longer than 30 days: the grouping key is a one-way hash that is thrown away and replaced, and the records it points at are deleted.

Turn it off

If you would rather not be counted at all, this button stops the tracker on this device. It takes effect immediately and it sticks until you clear your browser storage.

Two notes on that button. It stores one value in your browser's local storage, the flag that tells the tracker to stay quiet, which is the only thing this site ever writes to local storage. And if your browser blocks storage entirely, in a private window or with strict blocking on, the button cannot save the setting, but browser-level tracking protection generally blocks the tracker anyway.

What is recorded when you read a page

Every page loads one small script, written and hosted here, that reports what happened on the page. Nothing is sent anywhere except sci.help. Here is every field, and why it exists.

WhatExampleWhy
Page address and title/articles/pressure-injuries/To know which guides get read and which get ignored
What you did on itClicked a link, ran a search, changed a filter, scrolled 75% down, submitted a form, hit a JavaScript error, left the pageTo find broken features, searches that return nothing, and pages people give up on
Seconds spent reading90To tell a page that helps from a page people bounce off
Where you came fromgoogle.com, chatgpt.com, a link someone shared, or DirectTo know whether this information is reaching people
The page you were on before/newly-injured/To see which paths through the site make sense
A session idA random value your browser makes, hashed before it is storedTo count one visit as one visit. Dies when you close the tab
A device keyA one-way hash, computed on the server, that rotates every 30 daysTo tell returning from new inside a 30-day window. Details below
Country and cityUS, Austin, TXCoarse geography, from the hosting edge. Never an address, never GPS
Device, browser, operating systemMobile, Safari, iOSTo catch layout and feature breakage on real devices
Screen size, language, time zone390x844, en-US, America/ChicagoSame reason

The device key, in detail

This is the part most sites are vague about, so here is the whole mechanism.

Your browser is not given an identifier and does not send one. Instead the server computes a key from things the request already carries:

device key = SHA-256( secret salt + your IP + your browser's user agent + sci.help )

A hash is one-way: the key cannot be turned back into your IP. The salt is a long random secret held only on the server, and it is replaced with a new random value every 30 days. The previous salt is kept alongside it so that a visit spanning the changeover is not split in half, which is why grouping can run 30 to 60 days rather than exactly 30. What is stored is capped at 30 days regardless, by the two rules below.

That rotation is what makes this different from a tracking id. Once a salt is discarded, the keys it produced cannot be recreated from any IP address by anyone, including me. Old records stop corresponding to a findable person.

Your IP address is an input to that calculation and is never itself written to storage. It is not in the daily files, not in the device records, and not in anything the dashboard can display.

What a device record holds, and for how long

Inside the window, a record for a device holds the pages it viewed in order, its sessions, counts of clicks and searches, the source it first arrived from, and the coarse location and device type already listed above. Two separate rules bound it:

  • Every write trims the record to the last 30 days. This holds no matter how often someone visits. A device that comes back every week for a year still never has more than a 30-day tail stored.
  • A daily sweep trims and deletes regardless of whether you come back. It removes anything past 30 days from records that are still active, and deletes outright any record untouched for 30 days. Not archived, not anonymized further, deleted. This runs on its own daily schedule rather than waiting for the salt to rotate, because waiting would let an abandoned record sit for up to 60 days.

So the honest answer to "how long do you keep it" is 30 days, with a sweep each day that enforces it. The aggregate daily counts, which are not tied to any device, are kept longer so year-over-year traffic comparisons are possible.

What this approach gets wrong

Because the IP is part of the key, people sharing one internet connection collapse into a single "device." Everyone on a rehab hospital's wifi looks like one device. So does a large share of mobile traffic, because phone carriers route many customers through shared addresses. The site's device counts therefore run low, and mobile runs lowest.

This is worth stating plainly for two reasons. It means the numbers on the dashboard understate real readership. And it means the record labeled as one device is sometimes several people's reading mixed together, which is a privacy property working in your favor, and an accuracy problem working against mine. The alternative, a permanent id stored in your browser, would count better and is what this site used until 18 September 2026. It was removed and the profiles it had built were deleted.

What is not recorded

  • No cookies. The site sets none, for any purpose.
  • No stored IP address. It is an input to the device hash and is never written. The hosting provider sees it in transit, the way every web server does.
  • No identifier on your device. The only browser storage used is a session id that dies with the tab, plus the opt-out flag if you set one.
  • No permanent history. Nothing survives the retention above, and nothing links one window to the next.
  • No cross-site tracking. No advertising networks, no pixels, no fingerprinting scripts. Ordinary pages load no scripts from anyone else. Three features load something from another company only when you use them (a library video, a podcast episode, the Rehab Finder map), and the typeface comes from Google. All four are listed under Other companies involved.
  • No account, no email, no name unless you type one into a form yourself.
  • Nothing is sold or shared. No data broker, no advertiser, no partner, no exceptions.

Forms you fill in yourself

Two pages ask for information, and both are obvious about it because you type into them.

  • The feedback form takes your message, and optionally a name and email if you want a reply. If you leave the email blank there is no way to contact you, which is fine.
  • The equipment exchange takes a listing: what you are offering, its condition and price, your city and state, and a contact email or phone so a buyer can reach you. Listings are public by design, so treat that contact information as published. Do not put a home address in a listing.

Both are handled by Netlify Forms, the hosting provider's own form service. Submissions are emailed to the site owner and kept in the hosting account. They are not added to a mailing list and not used for anything but answering you and running the exchange. If you want a feedback message or a listing deleted, ask through the feedback form.

Other companies involved

  • Netlify hosts the site. Like every web host it sees request IP addresses and keeps its own short-term server logs, outside this site's control. Netlify also provides the country and city shown above, and stores the analytics files.
  • Google Fonts. The typeface (Inter) is fetched from Google's font service on every page. That request tells Google your IP address and browser type, the way any download does. Google says the font service sets no cookies and is not used for advertising. It is the one outside request an ordinary page makes.
  • YouTube. The videos in the Research Library are hosted by YouTube. Nothing loads from YouTube until you open a video. When you do, YouTube's player loads from Google, which may set its own cookies and applies its own privacy policy, exactly as if you had watched on youtube.com.
  • Podcast hosts. Playing a podcast episode in the library streams the audio file from that podcast's own host, which sees the request like any download.
  • Fourthwall. The Shop page loads its product list and pictures from Fourthwall, the company that prints and ships the merch, so Fourthwall sees that request like any download. If you add something to the cart, Fourthwall holds the cart and this site saves only the cart's ID in your browser so it is still there when you come back. Checkout and payment happen on Fourthwall's own site under its privacy policy. This site never sees your payment details.
  • OpenStreetMap and unpkg. The Rehab Finder map is off until you press "Show map." Pressing it loads the open-source map code (Leaflet, from the unpkg content network) and map images from OpenStreetMap. The "Near me" buttons in the Rehab and Van finders ask your browser for your location only to sort results on your own device. That location is never sent to this site or stored.
  • Amazon. Some equipment pages link to Amazon through the affiliate program, and a purchase there earns this site a small commission at no extra cost to you. Those are ordinary links: nothing loads from Amazon until you click one, and after you click you are on Amazon's site under Amazon's privacy policy.
  • Outbound links generally. ClinicalTrials.gov, CMS, PVA, MSKTC, hospital sites and the rest are independent sites with their own policies. A click is recorded here as a count of that link; where you go afterward is not visible to this site.

Children

This site is written for adults managing a spinal cord injury and is not directed at children. No account is required, and nothing here knowingly collects information from a child.

Your rights

Data protection laws such as the GDPR and the CCPA give people the right to see, correct, or delete personal information a site holds about them. There is not much here for those rights to reach: no account, no name, no stored IP, and a device key that cannot be worked backward to find you, which also means a request to produce "your" record cannot be honored, because there is no way to prove which record is yours. If you would rather not appear in the counts at all, the opt-out button above is the direct route. If you submitted a form, that message does exist and you can ask to have it deleted through the feedback form.

If this changes

This page describes what the code actually does, not an aspiration. The deploy process refuses to publish the site if the tracker starts writing a permanent identifier, if the retention windows stop matching the numbers written on this page, or if this page goes missing. That check is not a promise, it is a build step.

Last updated 14 November 2026, when the list of other companies was completed (Google Fonts, YouTube, podcast hosts and the map were already in use but were not listed).The permanent per-browser identifier was removed on 18 September 2026 and replaced with the rotating server-side key described above, and the profiles built under the old scheme were deleted. On 28 September the window moved from 7 days to 30, and the sweep that enforces it moved from running on salt rotation to running daily.