Fall 2021 NAACCR Narrative Leave a comment

Jane Barnhill, BA
Casefinding Coordinator, NC Central Cancer Registry

 

WHAT is Death Clearance?

Death Clearance (DC) is fundamentally a search. The DC caseload is the result of a linkage between the state’s annual Vital Records death list and the North Carolina Central Cancer Registry (NC CCR) database. People with a cancer-related condition or diagnosis on their death certificates are included in the linkage. The DC process pulls deaths within the state, including residents, visitors, travelers, and people who have relocated here. Many of the non- or new resident cases can be eliminated based on residence at diagnosis, but all must be investigated because of their connection to the state of death, however brief that connection may have been. DC is typically performed by the NC CCR a year after the year of death, so someone who died in December 2019 will be included in the DC process that begins in the Fall of 2020. Considering the timing rules for reporting, most cancer cases will have been reported to the cancer registry before the DC process even starts. Cases that are already in the registry are updated with Cause of Death (COD) information. Cases that are missing from the registry are added to the DC caseload list and researched.

 

WHY is it important?

It is important to remember that DC is performed for several reasons:

  1. To catch any cases that should have been reported by facilities required to report to the NC CCR. True missed cases can be red flags for casefinding issues.
  2. To be included in POPULATION-BASED studies and statistics. Just because a case isn’t analytic (reported by a provider who was directly involved in the patient’s diagnosis or first course of treatment) doesn’t mean that it isn’t reportable, relevant, or important. Non-analytic cases matter.

Once the linkage with the registry is complete, the DC caseload list is finalized, and the DC process begins. DC cases are needles in the haystack. For example, over 95,000 people died in NC in 2019, but less than 3,000 were included in the DC caseload list for that year. Many DC cases prove to be not reportable, but the reason for non-reportability must be found before the case can be excluded. Many will be non-analytic because the medical provider at the time of the patient’s death was not directly involved in the initial diagnosis and treatment, but they are still important cases for population-based studies and statistics. Many will be missed cases that should have been reported—these will be collected through the usual reporting processes and through follow back.

Follow back to facilities that may have knowledge of the patient’s cancer diagnosis is essential as the goal of death clearance is to confirm the diagnosis from the patient’s medical provider. COD on the death certificate may have been assigned by a physician with very little knowledge of the patient’s medical history. This introduces uncertainty as to the validity of the cancer diagnosis and critical details such as the originating primary site of the cancer. Any cases that cannot be obtained from a managing provider will be included in the NC Central Cancer Registry (CCR) database as a Death Clearance Only (DCO) case as part of the 24-month data submission. The objective is to confirm the diagnosis and have the fewest number of DCO cases possible. NC’s internal goal is to have a DCO percentage of < 1% of the total annual caseload. This is well below the NPCR and NAACCR requirement of < 3%.

 

HOW do we manage it in NC?

The NC CCR has a full-time, designated DC Coordinator. After linkage, approximately 3000 DC cases require research each year. NC also uses a database designed specifically for DC that we call the DC Database. The DC cases are imported into the DC Database for tracking. Follow-back efforts, letter generation, and final decisions are managed in the DC Database.

The starting point in the search to confirm the diagnosis is the death certificate and the information that it contains. Upon visual review, some cases can be excluded right away as non-reportable conditions. Most require further research. There are several sections of a death certificate that provide the first follow back contact (where we will go first in the search for more information). The most obvious one is often “Place of Death.” If the death occurred in a facility (hospital, nursing home, hospice center), then the search naturally begins there. If the death occurred at a residence, then the next option is to check for the certifying medical professional (the person who signed the DC). If the certifier is based at a medical practice, then that practice is contacted (doctor’s office, nursing home, hospice provider). Sometimes the certifier is the Medical Examiner’s office, an out-of-state provider, or a practice that has closed. Sometimes the first follow back contact doesn’t have the necessary information but does have another contact option, such as a specialist or the place of diagnosis. Wherever the trail leads, requests are sent until sufficient information is received, or until time runs out before the submission file is due.

As the deadline for submission approaches, cases are excluded or entered into the database. Any cases that cannot be resolved must be entered into the database as a DCO case. These are unconfirmed cases, and the NC CCR strives to minimize these. In NC, we are always looking for ways to streamline the DC process and trying new resources in order to minimize our DCO count. An overview of our current process is outlined below.

 

STEP 1: DOUBLE CHECK THE LIST

Cases come into the registry all year long. From the time that the DC linkage is performed until it is received by the DC coordinator, a significant number of “missing” cases will have been reported. In NC, our first step is to run the DC list through a “double check” with the registry database again, and to review any matches that are found. These could potentially be (and usually are) eliminated from DC follow up, which saves a lot of time and a lot of work! For the 2019 caseload, 3.1% of our original DC cases were eliminated with this step.

STEP 2: CATEGORIZE

Just as an assembly line in a factory saves time by streamlining efforts, categorizing death certificates into follow back source types makes the process run much more smoothly and efficiently. Our categories include hospital deaths, Hospital Discharge Dataset matches, hospice cases, nursing home cases, physician practice cases, Medical Examiner cases, non-reportable cases, VA Clinic cases, and out-of-state deaths. The appropriate category is determined upon visual review of the death certificates, with attention to these fields: place of death, certifying physician and address, referral to Medical Examiner/autopsy status. Diagnoses described with ambiguous terminology do not automatically exclude a case from follow back; we still attempt to gather more information before that determination is made. Visual review and categorization both take time, but it is worth the effort. 4.4% of all 2019 DC cases fell into the non-reportable by visual review category.

STEP 3: STREAMLINE AND SEND

In NC, once every case is categorized, our first dragnet is for hospital deaths. If the place of death listed on the DC is a North Carolina hospital, we request the case from that hospital. A broader spectrum of reporting criteria applies to these death clearance cases. If there is any mention of the patient having a reportable cancer/condition (currently or in the past), the case is to be reported regardless of residence, disease status, class of case or visit type. The case is abstracted using all information available in the medical record with the understanding that many data items may be coded to unknown for non-analytic cases due to lack of information.

This portion of the DC caseload is managed by using an Excel spreadsheet to track follow back efforts. Cases are sent out to our hospital registries for research and abstraction. Some cases are returned as non-reportable with details to explain the reason for non-reportability. Many are submitted to the registry through the usual reporting process. Over 17% of all 2019 DC cases fell into this group.

Our next batch comes from a linkage with the Hospital Discharge Dataset (HDD), which we limit to cancer-related admissions/discharges. Our statistical team performs a linkage with this massive dataset and pulls out matches to patients on the complete DC list (including those from the hospital deaths above). The results are then filtered down to only those visits with a cancer-related diagnosis code, then any duplicates from the hospital death list are excluded. Many cases have multiple admissions–from the same or from different hospitals; duplicates are excluded, but multiple facilities are often contacted for a single case. Keep in mind that at the state central registry, we are interested in gathering as much information as possible for each case, so we reach out to all facilities who may be able to contribute information to complete the case. The HDD linkage has proven to be an exceptionally valuable tool in our DC process, because it pinpoints many visits where a diagnosis was made or where the records hold valuable information, which leads to many successfully disposed cases. The HDD linkage list is also managed through an Excel spreadsheet which tracks follow back efforts. Again, the hospital registry will either abstract the case or non-reportability information is returned. 12.7% of all 2019 DC cases were included in the HDD follow back set.

Another large DC category is those cases associated with hospice providers. These are identified in multiple ways. First, some patients die in a hospice inpatient facility. Many hospice patients die at home, and the certifying provider uses the hospice address. In NC, we receive cancer reports from hospice providers on a quarterly basis, and these are checked for matches with the DC cases. Many patients who do not receive treatment have elected hospice care, so it makes sense that they coincide with the DC case list. Hospice cases are batched, and additional information is requested from the hospice providers on a reporting form to be returned to the central registry. Keep in mind that hospice providers have limited information available and many patients to serve, so the streamlined approach is especially helpful for these cases. It requires extra time at the registry but results in a greater response rate from the providers. There is considerable overlap of hospice cases with hospital deaths and HDD cases, but they are sent to the hospitals first because the information there is often better than what hospice has available in their records. Over 21% of 2019 NC DC cases included contact with hospice providers.

Compared to the first three categories, nursing home follow back accounts for a relatively small portion of DC cases. NC receives cancer reports from a small number of nursing homes as well, and these are checked for matches to the DC case list. Matches are processed accordingly. If the place of death is a nursing home, the initial follow back attempt is directed there in batches to the individual nursing home as a request for more information. (Note: nursing homes can change ownership over time, and every year we struggle to match nursing homes that operated under a different name at the time of death with the current name and owner. Because of this, we initially batch nursing homes by city, then use the state licensing board’s list of current nursing homes to find the appropriate follow back contact.) Often the records are unavailable or inaccessible, so these cases are frequently recategorized into a different follow back type–most often to physician practices. 6% of our 2019 cases were sent to nursing homes for more information.

Our next category is individual physicians. Many requests for additional information are sent directly to the certifying physician’s office. We currently fax most of these letters, but sometimes the request has to be made through secure email or by phone. Additionally, any cases in the previous categories that cannot be resolved by follow back to the primary follow back source are then sent to the certifying provider’s office where applicable. (For example: A patient dies at hospital A, but the hospital has no information about the cancer. The certifying physician used the private practice address on the DC, so a follow back request is sent there). In an effort to streamline these letters, they are initially grouped by city, then by street address. If a physician has more than one case, then these are sent in a batch. (For example, if 5 death certificates are signed by Dr. Z, then we need to batch and send 5 follow back letters to Dr. Z. If we send them individually, the office might discard the 2nd, 3rd, 4th, and 5th letters that arrive, assuming that they are duplicates. If, however, those 5 letters are sent as one batch with a note that there are 5 patients included, then the chances are greater that the requests will be handled appropriately the first time.) Interestingly, almost 20% of the 2019 DC cases that were sent to physician practices ended up being non-reportable, mostly due to non-residency at time of diagnosis.

Our smallest categories include Medical Examiner cases, VA Clinic cases, and out-of-state deaths. ME cases are requested through the NC OCME’s website. Roughly half of these requests become autopsy cases; the rest often lead to other providers or reveal a non-reportable situation. VA Clinics receive individual requests in the same way as physician offices. For out-of-state deaths, if enough information is available to send a follow back request, then one is sent to the appropriate facility. Most have insufficient information, and most attempts are unsuccessful, but we still think the attempt is worthwhile. These cases may come in through the Interstate Data Exchange, but many are excluded from the DC process per the guidelines in the Death Clearance Manual. These smallest categories are typically handled last. Once these follow back attempts have been made, the process shifts to review of replies and a cycle of second attempts through all categories.

STEP 4: TRY AGAIN IF NO RESPONSE; TRY ELSEWHERE IF UNSUCCESSFUL; TRY OTHER SOURCES IF AVAILABLE

“If at first you don’t succeed….” Roughly half of our requests are handled quickly, and a response is received within a few days. We include “respond by” dates to encourage a quick response, but many providers will get a second request, and we handle these as a matter of routine. Generally, our process cycles through the different follow back types in order by size (i.e., we start with the larger batches and work through to the smallest). By the time the cycle restarts, the follow back contacts have had sufficient time to respond to the first request, so second requests are much fewer. In NC, we make at least two attempts for each case in the DC caseload. When a response is insufficient to resolve the case, we will call the provider for clarification if appropriate. When replies direct us to another provider, the case is recategorized if necessary and a new request is generated and sent to the alternate provider.

Some requests are never answered, no matter how many additional attempts are made. If an alternative follow back source is available from the information on the death certificate, we recategorize if needed and send a new request to the alternate provider. For example, a case from the HDD linkage is returned as “no information available,” and the DC was signed by a hospice provider. That case would be recategorized as a hospice case, and a request sent there. If a nursing home has changed ownership and the records have been archived, then the case might be sent to the signing physician’s private practice if applicable. If a hospital death case was referred to the Medical Examiner’s office, then the available reports are requested from the OCME’s office through their website.

STEP 5: MANY BATCHES MAKE LIGHT(ER) WORK

When sufficient information is received to confirm the case, it will need to be entered into the registry’s database. If the cases are to be entered manually, it helps to batch the cases by follow back source, primary cancer site, and date of diagnosis. This simple step helps to streamline the data entry process because the required fields are the same for all cases in the batch, which in turn saves time in looking up necessary information (especially for older cases that have different data field requirements).

Typically, DC case abstraction does not start in NC until the Spring, after all first attempt follow back requests have been sent out. The more time that providers have to complete the requests, the lower the final DCO number will be. For this reason, getting the requests out as early as possible is a key factor in the success of our DC efforts. Once that first rounds of “asks” is out, the workflow shifts to receiving. Reviewing replies, excluding cases, answering phone calls and emails, calling providers whose responses were insufficient—all of these will drive the workflow. Usually by the time the influx slows, it will be time to send a second request for those cases that haven’t been resolved. The influx of replies repeats, and the cycle continues until time runs out. What remains becomes our DCO batch of cases for the year. 16.5% of our original DC caseload ended up as 2019 DCOs. This number will continue to drop slightly as late replies come in, but at this point we consider the DC process to be complete. As long as our overall percentage of DCOs in the registry is less than 3% of all annual cases, we channel our efforts towards other tasks until the next DC cycle begins.

 

MAJOR TURNING POINT IN DC FOLLOW-BACK SUCCESS

In 2014, for hospice and nursing home cases, we turned our focus from following back to the physician who signed the death certificate as the first follow back source to the facility where the patient died. An earlier study of DCO cases in NC found that over 30% were for patients who died in a hospice or nursing home. In many cases, this physician is on contract with that facility and may not have in-depth medical knowledge of the patient. As a result, we were getting a high percent of unsuccessful follow back. We implemented a process for these hospice and nursing home facilities to report cases to us proactively, while the patients were still alive. They are provided a template in spreadsheet format that they complete and send through secure email once a quarter. These spreadsheets are batched in a SAS program that can be used to conduct linkages with annual death data. With the medical record notes on file with the hospice or nursing home facility, successful follow back for these patients increased substantially. There has been a notable shift from DCO cases in the registry database to hospice and nursing home cases. This, along with the streamlining and categorizing mentioned above, allowed us to reduce our annual DCO percentage and set an internal goal of < 1%.

 

CHALLENGES

There are challenges to the DC process that recur every year. It is always necessary to educate some providers and their staff members about the registry—who we are, what we are doing, and why it is important that they return the information requested. It is surprising how many providers are unaware of the cancer registry and the reporting requirements, so it is also rewarding to be able to share that information. Our hope is that every year will see an improvement in the response rate because of prior years’ efforts. Another challenge is that most of our requests outside of the hospital system are sent by fax machine. This is a time-consuming way to manage the task, so we do what we can to streamline this process as well until a better alternative can be implemented. The sheer volume of the DC caseload can be overwhelming. In NC, shifting the focus to follow back categories breaks that volume into less daunting portions that are easier to manage. Finally, the challenges to our healthcare system that came with the COVID-19 pandemic in 2020 have affected even the 2019 DC caseload. Our DC process was delayed at some points, some cases became DCOs simply because the only available contact was no longer in business, and hospice and nursing home reporting slowed due to conflicting priorities brought on by the pandemic. These effects are sure to be worse for the 2020 death year, so we will have to find creative new ways to manage them as well.

The Death Clearance process has challenged us in NC to find ways to improve our processes and our results. The result of these continuous improvements is evident in the gradual drop in DCO percentages at time of submission in recent years:

 

YearDCO % at time of submission
20151.22
20161.16
20170.93
20180.96
20190.84
20200.72

 

Hopefully the information in this article will help other registries as well. In summary, here are a few quick tips that have made the greatest difference for us. May they serve you well!

QUICK TIPS

  1. Connect with hospice and nursing home facilities directly.
  2. Use your state’s Hospital Discharge Dataset (HDD) if available!
  3. Categorize and batch follow back requests by the different follow back source types.
  4. Don’t be afraid to pick up the phone, send 2nd requests, and try alternate contacts.
  5. Streamline data entry by batching cases by source, site, and diagnosis date.

 

 

Facebooktwitterlinkedinmail

Copyright © 2016 NAACCR, Inc. All Rights Reserved | See NAACCR Partners and Sponsors