Table 3 Challenges & Case Examples

Undeliverable Surveys

Definition: Surveys that were not delivered to the recipient as intended either because of a failed email delivery (Wave 1) and/or a mailed survey coming back as ‘return to sender’ (Wave2).

Wave 1: Delivery Failed Messaging

(1) The email was rejected by recipient email system

(2) The recipient mail server may be temporarily offline or temporarily unable to accept messages

(3) A problem occurred while delivering the message to the recipient

Wave 2: Return to Sender, Unable to Forward Messaging

(1) Not deliverable as addressed

(2) Insufficient address

(3) No mail receptacle

(4) No such number

(5) Attempted—not known

During Wave 1, our study team received a significant number (n = 257) of undeliverable recruitment emails sent via REDCap. Based on the delivery failed messaging, the study team attributed the undelivered emails to recipients’ servers blocking emails from an unknown email address that may be suspected as spam or alternatively, attributable to incorrect or outdated email addresses from the Compendium. However, those who did not receive the email in Wave 1 were recruited again in Wave 2. In Wave 2, we received a large number (n = 151) of ‘return to sender’ returned envelopes due to incorrect addresses. If a recipient had an undeliverable email and a ‘return to sender’ mailing response in Wave 2, we designated those contacts as unreachable (n = 14) by both recruitment methods. Recipients solely in Wave 2 who had a ‘return to sender’ response (n = 44) were also deemed unreachable. We attributed Wave 2 return to sender responses to incorrect physical mailing addresses. Unreachable contacts were removed from the response rate.

We acknowledge that leadership positions in jails are often times elected officials which could contribute to outdated contact information. However, inaccurate physical addresses lend to the notion that general information on jails is limited to the public who they purposely serve, a theme also highlighted in the creation of the Compendium. The undelivered surveys ultimately help us and others understand the difficulties with getting in contact with jails for research (and/or other purposes) and suggests that the existence of a Compendium would need to be updated regularly.

Addressee Not Here

Definition: Surveys that were received and returned [but not completed] indicating that the person the survey was addressed to was not at that physical location.

The study team addressed all mailed surveys in Wave 2 to the listed contact person provided in the Compendium; as noted, ‘Health Services Administrator’ was used if no contact person was listed. Unpredictably, some unopened surveys (n = 27) were returned with messaging indicating that the recipient was no longer there nor in custody.

Addressee not here Messaging:

(1) Addressee released (no forwarding address)

(2) Not at address

(3) Inmate no longer at [facility]

(4) No inmate [with] this spelling

(5) Inmate unknown

(6) Not in custody

In response, the study team resent all surveys to those facilities in new envelopes and addressed them to the attention of the ‘Health Services Administrator.’ Upon the second attempt, nine of the 27 completed the survey. The remaining 18 were also included in the overall response rate (but will excluded from data analysis) since the respondents returned the survey with a note.

The study team did not send any surveys to incarcerated people, however, whoever handled the mail did not know the contact person listed in the Compendium and assumed it was for a detainee. This, again, suggests there is either inaccurate reporting of who works in the facility or high turnover.

True Duplicate Responses

Definition: When respondents submitted more than 1 survey response from the same physical location according to the Compendium; surveys could have been completed by the same person or multiple people.

The study team and representatives from the Johns Hopkins University BEAD (Biostatistics, Epidemiology and Data Management) Core discussed the many complexities of the types of survey responses we received and decided to use the basis of reporting from the same physical location as the primary criterion for a ‘true duplicate.’ This helped us eliminate instances where there was geographical reporting overlap with neighboring counties, cities and regional facilities (further discussed in the ‘Location Challenges’ section).

We received duplicate responses (n = 33)* from the same facility despite a screener question in place to prevent duplicates demonstrating instances of miscommunication in jail settings. Duplicates included both fully completed and partially completed surveys. Some duplicate responses were submitted by the same person; in other instances, two jail leaders (e.g. the Warden and Captain) both submitted a response. In all but three instances, the respondents were custody personnel, not facility health care providers, which could contribute to the misinformation. The amount of discrepant information varied by facility and these facilities were not unique based on any characteristics we are aware of (i.e. geography, size, etc.). We preliminarily reviewed duplicate submissions to determine consistency in answers to help inform our resolution approach. Our initial review showed discrepancy in the reported data warranting us to contact the respondents to get an accurate depiction of that facility’s response. Only 2 of the 33 duplicates submitted the same information on both reports.

Even though duplicate reporting was discouraged it yielded important insights about how jails operate and how they interpret and report their health care services [specifically for pregnant people with OUD]. In some cases, a duplicate report by the same person exhibited different responses to the same questions. In other instances, the two leaders reported fundamentally different responses, even about who they detain at the facility. Duplicates further emphasize the incongruences of the inner-workings of jail facilities.

*The duplicates were only counted once in the sum of total responses

Not Applicable

Definition: Surveys in which the respondent communicated that the survey does not apply to their facility and meets at least one of the criteria described below.

We largely operated in a blind recruitment strategy in the sense that we had no details about the facilities in the Compendium other than the information provided to us. However, the study team accounted for this by incorporating an initial screener question about whether the facility houses pregnant women to determine if a survey about pregnancy and opioid use disorder would apply. If the respondent selected ‘no,’ they were instructed to end the survey [and mail it back for Wave 2]. Of the 226 categorized as not applicable, the majority (n = 142) do not house females or pregnant females. We believe the subject may have contributed slightly to lower response rates. Additionally. Since the contact persons in the Compendium are jail administrators rather than clinicians, the topic area may not have resonated. The study team used the information respondents provided to group ‘not applicable’ situations into five non-exclusive main categories:

(1) Don’t house females and/or pregnant females (n = 142)

(2) No jail at this location (n = 76)

(3) Don’t operate/run a (the) jail (n = 6)

(4) Short-term holding facility/send elsewhere for medical care (n = 13)

(5) Don’t accept detainees that meet these criteria (pregnant with OUD) (n = 2)

(6) House detainees in another facility/county (n = 38)*

The surveys classified as ‘not applicable’ yielded valuable information about research involving jails and how they operate. The following case example highlight inconsistencies in communication and knowing which facilities provide what services.

Case Example #1: Don’t house females and/or pregnant females.

Two facilities in the same county were surveyed. Facility A responded requesting the survey be forwarded to Facility B. Facility B responded reporting they don’t house pregnant females.

Case Example #2: Reports of no jail at the surveyed location.

One respondent communicated to the study team that the county jail closed several years before but recommended we forward the survey to another facility in which they provided the address. When the forwarded survey was received, a different respondent replied with “We don’t have a jail. The jail is a regional jail.”

Case Example #3: Reports of we do not run/operate a jail.

Many facilities responded with reports that they either don’t run the jail or operate a jail which questions how they met the criteria to be included in the Compendium.

[County name] reported they do not run the jail, asked us to direct it to the prison and provided an address—the address is the same address this person responded from.

Case Example #4: Some respondents communicated with the study team to express that the survey was not applicable because they were short-term holding facilities.

“When I started into the survey I thought I could answer all the questions about the very small population that my office has incarcerated over the years. But I find myself not knowing the answers because we do not hold females in our facility. I have a 72 h holding facility. All females are transported to other facilities who have the programs in the survey. I am not aware which program or how they address any opioid addiction they might have. Your survey has helped me to investigate further into the jails I utilize on short term and longer term basis. I am sorry I cannot complete your survey, I would rather not answer questions I do not know the answer too.”

“I am not sure our Detention Facility applies for many of these questions since we are not a long-term housing facility. We do not provide medical care other than Paramedics when needed. If one of our female inmates is pregnant, and there would be any medical complications, we will transport that inmate to the court holding their warrant (as a county would have medical staffing). If the inmate is here for something local, we would release the inmate to the paramedics who would then transport him/her to the hospital.”

Case Example #5: Some facilities reported that they would not accept a detainee that was pregnant and had opioid use disorder.

“Not available in our facility. We are too small to deal with this problem. Female would be sent to a facility that is equipped.”

“We don’t accept inmates falling in this category. We are in the infancy stages of talking about a MAT program.”

*The study team encountered a number of facilities responding to the survey who indicated they house their detainees in another facility or county (n = 38). Because the survey is technically not applicable to them for this reason, the study team counted them towards the ‘not applicable’ total despite being described in the ‘Location Issues’ section. This, coupled with the responses indicated no jail at said location, questions why those facilities were included in the Compendium and what constitutes a jail, especially for research involving health services.

Location Challenges

Definition: Surveys that reveal complexities regarding the physical area of jurisdiction (i.e. county), have some physical overlap with other facilities in the Compendium and meet one or more of the criteria outlined below.

As previously explained in the section that describes the inclusion and exclusion criteria used when creating the National Jails Compendium, both county- and city-level jails were included. We also found a number of regional jails that house detainees from multiple counties. Our study team decided to survey all facilities in the Compendium to capture as many institutions of detention as possible which has revealed many challenges when accounting for geographical coverage areas. On the survey, we asked respondents to report their county for tracking purposes. We then followed by asking respondents to report if they oversee jails in more than one county, and if yes, identify those counties. We took common themes from returned surveys (n = 141) that exhibited location challenges and organized their responses into the following categories:

(1) City jail located within a county that was also surveyed (n = 19)

(2) Regional jail that covers multiple counties (n = 13)

(3) Facility reported the same county as another surveyed facility (n = 24)

(4) Facility (not identified as a regional jail) lists multiple counties (n = 33)

(5) House detainees in another facility/county (n = 28)

(6) Survey response accounts for more than one facility [different physical address but same contact person] (n = 25)

Case Example #1: Responses from city jails that are located in counties where a county facility was also surveyed.

Many respondents from city jails indicated they are short-term holding facilities or they don’t house pregnant females, information that could not be gleaned about the jails through publically available information. In other instances, they responded with complete surveys noting their county affiliation. We did not receive any correspondence from city facility that indicated any overlap or connection with the county facility and vice versa.

Case Example #2: Facilities that are recognized as Regional jails [based on the institution name] and hold detainees from multiple counties.

A number of regional jail facilities were included in the Compendium. Additionally, a number of jails in the compendium reported that they contract with a regional jail or house their detainees at a regional facility.

“[County name] does not currently have a jail. We contract with [regional jail name].”

“We are a law enforcement agency with no jail. We are partners with 6 other Sheriff’s Offices to form a regional jail, [regional jail name].”

“[County name] is part of a multi-county jail located in [City, State]. I do not oversee the jail, so I cannot give you the information that you are asking for. Please send a survey to the [facility name] at [facility address].”

“This agency does not have a jail. We use [regional jail name] located in [City, State].”

“[County name] does not have jail. All inmates are sent to [regional jail name] in [city, state].”

“[County name] does not operate a jail. We are a part of the [regional jail name] and it is operated by the Jail Authority—not the Sheriff of [county name].”

Case Example #3: Facilities with a different physical address and contact person but reported the same county of at least one other facility in the Compendium.

The study team intentionally cross-referenced county reporting when respondents submitted surveys to flag responses whose county overlapped with other facilities. This revealed a number of surveys who reported overseeing jails within the same county. Those surveys will be further evaluated for similarities in services provided.

Case Example #4: Respondents who reported overseeing jails in multiple counties, but are not identified as a regional jail.

Similarly, some facilities reported overseeing jails in multiple counties. Those surveys were also grouped so the study team could get a better understanding of their geographic coverage area and how it overlaps with other facilities in the Compendium.

Case Example #5: Respondents who indicated they house their inmates in another facility/county.

We received a number of notes from facilities reporting they house their detainees in other counties. In some instances, they identified those other facilities and/or counties. In almost every case, those identified facilities were also a part of the Compendium.

“We do not have a jail in our facility. All of our arrests go to the [jail name].”

“I just wanted to touch base with you and let you know that we do not have an incarceration facility in our county. We house all of our inmates at [detention facility name] in [city, state].”

“I would love to help you out with this but we don’t have a jail. We contract with [jail name] and [jail name].”

“The [county] Sheriff’s Office doesn’t house prisoners, we take them to the [detention facility name].”

“[County name] no longer has a correctional facility, closed in 2017. Combined facility with [county name].”

“All inmates for [county name] [city, state] are housed at [detention facility name and address]. They would have this information.”

“We no longer (as of 2004) run a city jail. [Jail name] now houses our prisoners. Please contact me if you have questions.”

Case Example #6: Survey response accounts for more than one facility with a different physical address but had the same contact person listed in the Compendium.

The Compendium listed multiple facilities with the same contact person. In electronic recruitment, that person was only emailed once, but in Wave 2 recruitment, a paper survey was sent to every address to maximize the chances of return. In some cases, those surveys were filled out multiple times. Those respondents are from the same county, have a different facility address but same contact person. However, the person filling out the survey could vary.

One jail with three different detention facilities (different addresses) but the same contact person responded to all three Wave 2 survey attempts. At least 2 different people filled out the surveys. The surveys were filled out differently and communicate different provision of health care services and treatment protocols.

Physical area of jurisdiction is essential in mass data collection and should be accounted for when surveying institutions that operate and are defined differently from state to state and even within states. Physical jurisdiction and operational oversight are important with research involving jails particularly because jail officials are elected. These factors influence how we are categorizing responses and the availability of health services. They also demonstrate the complexities of jail research and the usefulness of the Compendium.