Appendix A
Detailed Methods and Results for Expert Panel Review (Phase 2): Revising the existing ACSC definition
Methods
We sought to strike a panel of nine to twelve individuals1 who had clinical, administrative, and/or research experience in the areas of continuing care and/or potentially avoidable hospitalization of LTCF residents. Potential participants were nominated either by members of the study team or by invitees who declined. In total, 25 invitees were sent an introductory letter by mail, fax and email. Of these, 22 agreed to participate, two did not respond, and one declined due to a lack of recent experience in the area. Of the 22, nine were available for the selected meeting date, including two LTCF physicians, three LTCF nurses, one LTCF administrator, and three health service researchers who had published on avoidable hospitalizations among older populations. The panel included members from three Canadian provinces and three U.S. states.
We held the meeting over a day and a half in Calgary, Alberta. Two members of the research team facilitated. We tape-recorded the session, took written notes, and distributed a summary to attending panel members and the other experts who had agreed to participate but could not attend.
In advance of the meeting, we sent participants a one-page background document on the ACSC approach along with relevant references. Through discussion at the meeting, the group reached consensus on whether each of the original ACSCs2 listed should be retained, what modifications should be made to the specific ICD-9-CM codes included, and whether or not the most-responsible diagnosis field was sufficient to capture the condition. We also asked participants to identify and discuss other factors and conditions that were important in defining PAH in a Canadian LTCF context.
Participants rated each of the original ACSC diagnoses and the additional conditions that arose during their discussion on a 9-point scale where nine meant that the item was essential to a definition of PAH and one meant that it was irrelevant to our understanding of PAH. Disagreement on the 9-point ratings for each item was defined as having at least three panelists rating in 1-3 (“unnecessary”) range and at least three rating in the 7-9 (“necessary”) range. This approach was developed for 9-member expert panels for use in the RAND Appropriateness Method.3 Where there was no disagreement, the median rating was used to classify each item as necessary (median rating between 7 and 9), supplementary (4-6), or unnecessary (1-3). The panel recommended that diagnostic coding experts be consulted to assist with differentiating the specific ICD-9-CM codes to be included in the revised PAH definition. Consequently, two health records specialists with prior health services research experience were interviewed in Calgary, Alberta and two specialists with additional teaching and administrative database research experience were consulted in Toronto, Ontario.
Results
The expert panel members concluded that the ACSC methodology generally provided a helpful systems-level approach to identify PAH of residents residing in LTC settings in Canada. There was consensus on the potential utility of developing a revised list of diagnoses that could be used not only to define PAHs, but also as a quality of care indicator for the identification, prevention and effective management of select chronic and acute conditions (and their complications) in LTC residents.
A summary of the initial literature-derived ACSC list, the PAH list achieved by expert panel discussion, and the results of the panel’s rating process are included in Table A1. Lack of consensus on whether to include hypoglycaemia stemmed from panel members’ differing views on whether it should be considered distinct from diabetes rather than whether it was potentially avoidable or not. Thus, in the end, it was included in the overall list of recommended conditions.
The panel further recommended several modifications to the list of diagnostic ICD-9-CM codes used to define the original ACSC list. Specifically, a recommendation was made to add codes 402 and 404 for CHF. The panel also had several questions about Canadian diagnostic coding conventions (e.g., the selection of diagnostic types, common coding practices for selected conditions and the hierarchy of decision-making for selected ICD-9-CM codes) that were presented to the four hospital coding experts. The coding experts recommended the inclusion of type 1 diagnoses in addition to type M (excluding those that are also listed as type 2) diagnoses in order to avoid missing conditions that may have prompted a hospital admission but did not end up being most responsible for the length of stay or resource use. In addition, the coders assisted with the development of an approach to capture falls and fractures. The final list of ICD-9-CM diagnosis types and codes for the revised PAH definition is presented in Table 1 of the Brief Report.
Although there was panel consensus regarding the relevance and utility of the revised PAH definition as a system-level quality indicator for LTCFs, the panel also strongly recommended further research to examine its validity across LTC settings, such as detailed chart reviews. They emphasized that a PAH indicator must be interpreted within the confines of the limitations of this methodology (i.e., must be used at a system-level and not as an outcome measure for individual residents).
References
1. Pope C, Mays N, eds. Qualitative Research in Health Care. Second Edition. 2 ed. London: BMJ Books; 1999.
2. Billings J, Zeitel L, Lukomnik J, Carey TS, Blank AE, Newman L. Impact of socioeconomic status on hospital use in New York City. Health Aff 1993;12:162-73.
3. Fitch K, Dernstein SJ, Aguilar MD, et al. The RAND/UCLA Appropriateness Method User's Manual. Santa Monica: RAND; 2001.
Table A1: Summary of conditions included in the initial list of Ambulatory Care Sensitive Conditions (ACSC), the results of the expert panel rating, and the final revised Potentially Avoidable Hospitalization (PAH) list.*
Median ** / Disagreement***
Angina / Yes / 7 / No / Yes
Asthma / Yes / 6 / No / Yes
Cellulitis / Yes / 5 / No / Yes
COPD / Yes / 8 / No / Yes
Congestive Heart Failure / Yes / 8 / No / Yes
Dehydration / Yes / 8.5 / No / Yes
Dental Conditions / Yes / 6 / No / Yes
Diabetes / Yes / 8 / No / Yes
Gastroenteritis / Yes / 8 / No / Yes
Grand mal seizure disorders / Yes / 4 / No / Yes
Hypertension / Yes / 6 / No / Yes
Hypoglycemia / Yes / 7 / Yes / Yes
Immunization-preventable conditions / Yes / 1 / No / No
Urinary tract infection / Yes / 8 / No / Yes
Nutritional deficiency / Yes / 3 / No / No
Pneumonia / Yes / 9 / No / Yes
Severe ear, nose, throat infections / Yes / 1 / No / No
Tuberculosis / Yes / 1 / No / No
Fractures / No / 8 / No / Yes
Decubitus ulcers / No / 3 / No / No
Behaviour / Mental Health / No / 2 / No / No
Septicemia / No / 8 / No / Yes
Cardiovascular accidents / No / 3 / No / No
* Deletions from the initial ACSC list are highlighted in bold and additions to the final PAH list are highlighted in bold italics
** Rating score: 1 = “Unsuitable for PAH definition” to 9 = “Essential for PAH definition”.
*** Disagreement occurred when at least three panelists rated in the 1-3 region and at least three panelists rated in the 7-9 region.