Methods
Funded by a grant from the Agency for Health Care Policy and Research 		 (AHCPR), a multidisciplinary study team from a major Midwest academic medical 		 center conducted a systematic evaluation of care provided to mechanically 		 ventilated patients in rural ICUs. The study team consisted of a pulmonologist, 		 research nurse, ICU nurse, respiratory therapist, dietician and pharmacist, all 		 with critical care experience.
Seventy-eight hospitals met eligibility criteria: short-stay hospitals 		 located in a Health Care Financing Administration (HCFA)-designated rural 		 county with a critical care unit. Twenty of the 78 eligible rural Iowa 		 hospitals were randomly selected and contacted. All hospitals contacted agreed 		 to participate in the study and share patient and cost information. Enrollment 		 was limited to 20 hospitals because of the extensive burden of data collection 		 and multiple visits required to each participating facility during the 3-year 		 study. The statistical power of the sample size was approved by the AHCPR.
On the basis of HCFA definitions, two major categories of rural 		 hospitals were identified. A hospital qualified as a rural referral hospital 		 (RRH) if it was in a rural area and met specific criteria concerning bed size, 		 referral patterns or case mix intensity. Seven of the 20 participating 		 hospitals qualified as RRH and represented all the rural referral facilities in 		 the state. The remaining 13 facilities enrolled were termed rural hospitals 		 (RH), which had more limited bed capacities and resources.
Hospital demographic characteristics were supplied by each 		 participating hospital. These included ICU size, average daily census, 		 equipment availability, ICU staff, physician mix, and the availability of 		 specialists and support personnel, including respiratory, pharmacy and dietary 		 professionals.
The data collection tool was based on objective indicators established 		 by the Task Force on Guidelines, Society of Critical Care Medicine [10]. Using these indicators as a guide, the university team of 		 ICU specialists developed more specific standards based on current practice 		 (Table 1). Standards were made more specific by 		 establishing strict criteria defining the dose, frequency and time frame within 		 which initiation of treatment was expected; for example, task force guidelines 		 recommended measures such as nutritional support, stress ulcer prophylaxis and 		 deep vein thrombosis prophylaxis. The team clarified these measures and 		 included additional criteria. Team standards required that a complete 		 nutritional assessment including the patient's protein and calorie 		 requirements be documented within 72 h of admission to the ICU. Stress ulcer and 		 deep vein thrombosis prophylaxis was required to be initiated by day 2 in the 		 ICU, and minimum therapeutic levels of treatment were established. The tool 		 reflects basic processes of ICU care that should be delivered regardless of 		 available technological resources. The tool allowed evaluation of processes of 		 care in seven major categories: laboratory assessments, nursing care, stomach 		 ulcer protection, thrombosis protection, dietary management, ventilator 		 management and ventilator weaning.
Patient records were selected for team review by International 		 Classification of Diseases (ICD)-9 procedure codes that reflected the presence 		 of mechanical ventilation (96.72, ventilated for longer than 96 h; 96.71, 		 ventilated for less than 96 h; 96.70, period of ventilation unspecified). The 		 patients most desired for this assessment were those ventilated for longer than 		 96 h. This longer ventilation period allowed sufficient time to evaluate overall 		 patient care management techniques and practice patterns and assured a more 		 homogeneous group of patients between facilities and across time. In rural 		 hospitals with under 100 beds, few patients met the criteria of ventilation for 		 longer than 96 h. Consequently, in these facilities, the medical records of all 		 patients with ventilation codes were reviewed.
Three categories of patients that would be likely to provide limited 		 evidence of ventilator management techniques were excluded from the study: (1) 		 patients on a home ventilator admitted for pulmonary exacerbation or respite 		 care; (2) postoperative patients requiring fewer than 6h of ventilation while 		 anesthetic agents were reversed; and (3) patients shown to be brain dead 		 shortly after admission but ventilated while treatment and organ donation 		 discussions could be conducted with the family. For patients requiring 		 re-admission to the ICU for ventilator support within the same hospital stay, 		 only the first ventilated period was reviewed.
Data were collected from medical records of 224 patients requiring 		 mechanical ventilation while treated in 20 rural Iowa ICUs between 1992 and 		 1994. One hundred and eleven patients were managed at RRHs, whereas 113 were 		 managed at RHs. Patient variables included age, sex, primary and secondary 		 ICD-9 diagnoses, severity of illness as measured by Acute Physiology and 		 Chronic Health Evaluation (APACHE) II, medical conditions resulting in 		 difficult or complicated ventilation, admission source (home, emergency room, 		 nursing home, hospital ward, or acute care facility), do not resuscitate (DNR) 		 status, pay class or insurance coverage and discharge disposition (expired, 		 home, skilled nursing facility, intermediate nursing care facility, other 		 hospital and tertiary facility). For those patients transferred to the tertiary 		 setting, the discharge disposition was also evaluated. A list of 10 medical 		 conditions resulting in difficult or complicated ventilation was developed 		 through consensus of a team of critical care specialists. This variable was 		 monitored because the decision to select appropriate patients for tertiary 		 transfer was considered a process of care. The conditions and related 		 definitions are as follows.
(1) Adult respiratory distress syndrome (ARDS) required diffuse 		 bilateral infiltrates, PO2 divided by FiO2 < 200 (both 		 required) and, if pulmonary artery catheter was in use, a wedge pressure 		 < 18.
(2) Status asthmaticus with hypercapnea despite adequate ventilation 		 required diagnosis by the contact physician.
(3) Neurologic catastrophe was defined by an acute deterioration in 		 Glasgow coma score without a specific diagnosis for the decline.
(4) Pneumothorax complicating ARDS or status asthmaticus required 		 diagnosis by the contact physician or radiologist.
(5) Multiple organ failure required diagnosis by the contact physician 		 that two or more organs were in failure (respiratory failure was assumed in all 		 patients requiring mechanical ventilation).
(6) Sepsis syndrome with disseminated intravascular coagulopathy (DIC) 		 or coagulopathy required sepsis as defined in Appendix A in nosocomial events 		 and DIC/coagulopathy defined as a drop in platelet count by 25% from baseline 		 and an increase in prothrombin time (PT) or presence of fibrin degradation 		 products.
(7) Ventilation with peak pressures > 50 and positive end-expiratory 		 pressure (PEEP)> 15 was defined by these parameters.
(8) Complex chest trauma involved documentation of flail chest or 		 multiple rib fractures and cardiac or pulmonary contusions or extensive 		 subcutaneous emphysemaor hemothorax.
(9) Failure to wean required diagnosis by the contact physician.
(10) Complex overdose was defined as overdose requiring treatment by 		 dialysis.
Outcome variables included length of stay, ventilation days, 		 nosocomial events, discharge disposition, and survival. Patients with acute 		 respiratory failure are at risk for a number of nosocomial events. Eighteen of 		 these events were described by Pingleton [11] in her 		 work on complications occurring in patients with acute respiratory failure and 		 were incorporated into the data collection tool. A nosocomial event was defined 		 as an event that occurred in the ICU that was not present or incubating at the 		 time of admission. Definitions for each nosocomial event were developed from 		 several sources and established by the review team. Definitions for infectious 		 events were based on the Center for Disease Control (CDC) criteria. Definitions 		 for mechanical events were based on Pingleton's publication, supporting 		 references, criteria utilized by the risk management division at our 		 institution or criteria developed through consensus of the critical care team. 		 Events and associated definitions are provided in Appendix A.
The study protocol was approved by the Internal Review Board of our 		 academic medical center and deemed exempt from the need for informed 		 consent.