PEMS Memo

RESPONSE TO OMB QUESTIONS CONCERNING PEMS 0920-0696

August 16, 2007

In comparing the estimated PEMS Average Burden per Response with the burden of prior data collections, it is important to keep several significant differences among the systems in mind. The first is that PEMS not only collects data previously collected under other OMB approved data collections (0920-0497, the Evaluation Reporting and Analysis System, “ERAS”; 0920-0525, the Community-based Organizations System, “CBOS”; and 0920-0208, the HIV Counseling and Testing System, “CTS.”), but also collects data in a standardized way that was previously submitted in non-standard, often narrative, formats. Thus, PEMS partially replaces (and therefore reduces the burden of) these other reports (e.g., State HIV Prevention Community Plans, grantee Annual Progress Reports and Interim Progress Reports to CDC project officers).

The second significant factor is the level at which the data are reported. While all these systems basically provide a way for the grantees to report data to CDC that they normally collect anyway, some of the prior systems reported data at the aggregate level, while PEMS requires data at the client level. Therefore previous system burden estimates only included the time required to enter totals of existing data, while the PEMS burden estimates includes the time to enter the client-level data, not just the total. For example, the ERAS burden estimate is based on the time to complete three forms showing aggregate totals without taking into account the time to collect or calculate aggregate data. The burden estimates for PEMS, at the client level, include the time to collect and input data for each client and to conduct a quality assessment review of data before submission. Client-level data are required in order to respond to the level of detail requested in Congressional and Executive Branch inquiries and to perform outcome monitoring analysis. The grantees also aggregated data different ways, making national-level data unobtainable. PEMS enables the collection of national-level data.

The third fact to take into account is that PEMS provides the capability, at the grantee’s option, of replacing other data collection systems. For example, the PEMS variables and the STD Management Information System (STD MIS) variables have been harmonized so that grantees can choose to use either the PEMS or the STD MIS to collect Partner Counseling and Referral Service (PCRS) data. While the burden estimates for PEMS include the PCRS data collection because some grantees will chose to use PEMS for PCRS reports, PEMS is actually reducing burden because double data entry systems are eliminated for grantees that use the same staff of Disease Intervention Specialists to conduct partner tracing and counseling for both HIV and STDs.

The fourth element to consider is that PEMS enables reporting of data that we have never been able to collect before. For example, none of the previous systems collected significant information on Health Communication/Public Information programs. While many grantees will not conduct this type of program, some will, and the PEMS data burden estimates must take into account the full range of data collection from grantees that elect to conduct a wide range of HIV prevention programs. In addition, PEMS enables the collection of agency and budget information that was not readily available before, thus allowing enhanced program monitoring by CDC project officers.

Perhaps most significantly, a fifth factor makes any comparison of burden hours invalid. Namely, what is included in the estimates of the burden hours is drastically different in PEMS compared to the previous systems. For example, for the CTS scan system burden, only the average time to transmit a file to CDC containing data on each test was considered in the burden calculation. However, the PEMS burden estimate included the time to gather the information from each client on the scan form, the time to actually scan each form, and the time to conduct quality assurance checks on the scanned data (each multiplied by the number of HIV tests conducted). Data gathered by the grantee, and the actual time expended, is basically the same (actually the time may actually be a little lower in PEMS because of simplified forms), but the estimated burden is drastically greater for PEMS because of differences in what is included in the estimated burden. Similar differences exist for the HERR and PCRS burden estimates for PEMS and for the other systems.

In addition, while the heart of PEMS is a set of standardized variables that all grantees must report (plus standardized optional variables for use on the state or local level for program evaluation or by CDC for special evaluation projects that will fund the collection of non-required data) we have gone to great expense and effort to provide an optional, state-of-the-art, web-based data collection and reporting software system to reduce the burden as much as technology allows. The system complies with all Federal regulations for use by grantee agencies, provides templates to streamline repetitive data entry, and automatically pre-populates any fields that can pull data from anywhere else in the system. Moreover, the software provides a data extract function and numerous reporting functions that will enable grantees to get their data back in formats that are immediately usable for local program monitoring and improvement. The system also provides for automated data quality assurance checks that will drastically improve the quality of the data provided to CDC. Also note that the PEMS burden includes hours for training grantee staff in data collection, while none of the other data reporting mechanisms took that burden into account.

These factors explained above are partially reflected in the table below. However, there is no way to fully indicate on the table the differences in the systems that significantly affect the estimation of the Average Burden per Response, which was used as the basis of comparison. This is because in addition to the factors described above, there are major differences in the number of respondents and responses among the various reporting mechanisms.

AVERAGE BURDEN HOURS PER RESPONSE
	PEMS	ERAS	CBOS	CTS	Community Plan	APR/IPR
Health Jurisdictions	137 (Agency, Budget, Community Planning, Program Planning, and HERR, PCRS, and HC/PI data at the client level)	3 (aggregate PCRS and some budget info only)			?	?
Health Jurisdictions (CTR-Scan)	509 (includes collection, scanning, and QA)			0.25 (scan file transmission only)
Health Jurisdictions (CTR non-scan)	165			2
Health Jurisdictions (training)	10
Community-based Organizations (Agency, Budget, Program Planning, HERR, PCRS, and HC/PI data at the client level)	84		4 (HERR data only, some at aggregate level only)			?
Community-based Organizations (CTR)	23
Community-based Organizations (Training)	10

4