Decision Support for Intraoperative Low Blood Pressure

The purpose of this study is to determine whether a decision support system can improve the adherence to thresholds for low blood pressure by anesthesia providers, which in turn prevents their patients from having organ injury.

Blood pressure management is an important part of anesthesia. Many factors contribute to a change in blood pressure during a surgical procedure, such as blood loss, manipulation by surgeons, and there are several mechanisms through which anesthesia itself changes blood pressure. Although a high blood pressure also occurs during anesthesia, most of these factors lower a patient's blood pressure. When a patient's blood pressure becomes too low, the internal organs become at risk of receiving not enough blood (low perfusion or hypoperfusion). This low perfusion state can result in organ damage (ischemia) because of an insufficient supply of oxygen and glucose. Hence the important task of anesthesia providers to maintain the blood pressure of patients, using a wide range of drugs and other interventions. A big challenge in blood pressure management is to know when a low blood pressure indeed results in low perfusion of organs. There is a large variation between patients in how susceptible they are to low blood pressure, as well as a difference between the organs in how easily they are damaged because of low perfusion. Elder patients, or patients with preexisting hypertension, heart problems or other cardiovascular diseases are more prone to a low blood pressure and are more likely to develop organ ischemia when there is a low blood pressure. The kidneys, the heart and the brain are the organs that are most at risk of organ damage. As one cannot measure the perfusion states of individual organs in individual patients, it is very difficult to know 'how low to go' with a patient's blood pressure. Recent studies have used large datasets of patients to demonstrate that there is statistical association between low blood pressure during surgery and various types of organ injury. As patients are already treated for low blood pressure by anesthesia providers, this suggests that patients have low organ perfusion states despite the current treatment standards. A patient's blood pressure is not simply a dial that can be adjusted to a specific level. Finding the right level of interventions can be difficult in some patients. Consequently, lower blood pressures are common in anesthesia, even with the current standards of blood pressure management. In this proposed study the investigators will implement two forms of decision support to assist anesthesia providers in blood pressure management. The decision support aims to educate anesthesia providers about the risks of low blood pressures in direct relation to the patients that they treat. One form of decision support will provide automated notifications through pagers and through the anesthesia information management system. These automated notifications pop up when the patient's blood pressure drops below a level that is associated with a risk of organ injury, and thus alerts the anesthesia provider of the blood pressure and its associated risk. The second form of decision support will send a postoperative email the day after the procedure when the patient has had a low blood pressure for particular duration. This email then provides feedback to the anesthesia provider by informing them of the increased risks of organ injury that are associated with that low blood pressure. The study will look at both a change in patient outcome and a change in blood pressure management and will be performed at the Vanderbilt University Medical Center (VUMC). The change in patient outcome will primarily be studied through the occurrence of acute kidney injury in the first days following the procedure at the VUMC. The change in blood pressure management (provider behavior) will be studied by observing the depth and duration of low pressures during anesthesia, and the number of interventions that have been used to treat the blood pressure. Patient outcome will be studied by comparison of a baseline phase - before the decision support is implemented and uses historic data- and the intervention phase - the period during which the intervention is active. Only routinely collected clinical data will be used for these analyses: no additional data collection is required. As it is impossible to know which form of decision support will be the most effective, the first three months of the intervention period will be a 'nested cluster-randomized trial'. The anesthesia providers (not the patients) will be randomized to either the automated notifications or the feedback emails. After three months all anesthesia providers will receive both forms of decision support for the remainder of the intervention period. The reason why anesthesia providers are randomized only during the first three months is that cross-over or contamination between the two groups is expected. This contamination could make it impossible to study the effect of the decision support on patient outcome, as there will be no longer any difference between the study groups.

Decision support, Decision Support Systems, Clinical, Decision Support Techniques, Hypotension, Intraoperative Hypotension, Acute Kidney Injury, Postoperative Mortality, Blood Pressure

It is an interrupted time-series of a decision support package intervention that aims to study the the effect of the decision support package on patient outcome. The effects of the individual decision support components of the package on healthcare provider behavior will be studied in a nested cluster-randomized trial: in the first month the attending anesthesiologists will be randomized to near-realtime notifications or feedback emails; in the second month in-room providers will be randomized to near-realtime notifications or feedback emails. Starting month four all providers will receive both near-realtime notifications and feedback emails.

The intervention period. Several decision support elements are implemented to notify anesthesia providers: attending anesthesiologists and in-room anesthesia providers of intraoperative hypotension (threshold of a mean arterial pressure below 60 mmHg). Two types of decision support will be implemented: near real-time decision support and feedback emails. Near real-time decision support elements will notify the anesthesia providers of a blood pressure drop below the threshold and display the associated increased risk of acute kidney injury. The notification is presented through the pager system for attending anesthesiologists and through the anesthesia information management system for the in-room anesthesia provider. All providers will be notified through email within 24 hours after the end of an anesthetic case, when the patient had an episode of intraoperative hypotension that is associated with an increased risk of organ injury due to organ ischemia.

The 'before' period - or historic control group - during which no decision support for intraoperative hypotension was being used, also known as 'usual care'. This is the three year period prior to the intervention period (the 'Intraoperative hypotension decision support' arm).

Near real-time decision support elements will notify the attending anesthesiologists of a blood pressure drop below the threshold for intraoperative hypotension (mean arterial pressure below 60 mmHg). The notification is presented through the pager system. The page will also display the associated increased risk of organ injury due to organ ischemia.

Near real-time decision support elements will notify the in-room anesthesia provider of a blood pressure drop below the threshold for intraoperative hypotension (mean arterial pressure below 60 mmHg). The notification is presented through the anesthesia information management system. The decision support system will display the associated increased risk of organ injury due to organ ischemia.

Attending anesthesiologists will be notified through email within 24 hours after the end of an anesthetic case, when the patient had an episode of intraoperative hypotension (mean arterial pressure below 60 mmHg or lower for a particular duration) that is associated with an increased risk of organ injury due to organ ischemia.

In-room anesthesia providers will be notified through email within 24 hours after the end of an anesthetic case, when the patient had an episode of intraoperative hypotension (mean arterial pressure below 60 mmHg or lower for a particular duration) that is associated with an increased risk of organ injury due to organ ischemia.

The data warehouse that is used to gather perioperative data and create user reports. In this instance the PDW will be used to send the postoperative feedback emails.

Regional anesthesia, Spinal anesthesia, Intrathecal anesthesia, Subarachnoid anesthesia, Epidural anesthesia, Central neuraxial blockade

Regional anesthesia effectuated through the placement of local anesthetics around the nerves of the central nervous system, e.g. spinal anesthesia and epidural anesthesia.

Postoperative Acute Kidney Injury (AKI), Stage I or higher according to the KDIGO criteria (Kidney Disease: Improving Global Outcomes). The staging will be based on serum creatinine values, as documentation of urine output is probably not sufficiently accurate. This will be the primary outcome for the Vanderbilt University Medical Center. The creatinine measurements are part of routine clinical care. Therefore, absence of creatinine postoperative measurements are considered to be 'no suspicion of kidney injury'. KDIGO defines AKI as any of the following: Increase in serum creatinine by 0.3mg/dL or more within 48 hours or Increase in serum creatinine to 1.5 times baseline or more within the last 7 days or Urine output less than 0.5 mL/kg/h for 6 hours. Stage 1 is 1.5-9x baseline or >0.3 increase; Stage 2 is 2-2.9x baseline; Stage 3 is 3x baseline, or increase to > 4, or initiation of renal replacement therapy.

Vanderbilt University Medical Center: combination of in-hospital mortality and 'alive-index' (which checks for visits to the hospital in the electronic healthcare record as indication of being alive at 30 days)

Postoperative Acute Kidney Injury (AKI), Stage II or higher according to the KDIGO criteria (Kidney Disease: Improving Global Outcomes). The staging will be based on serum creatinine values, as documentation of urine output is probably not sufficiently accurate. The creatinine measurements are part of routine clinical care. Therefore, absence of creatinine postoperative measurements are considered to be 'no suspicion of kidney injury'. KDIGO defines AKI as any of the following: Increase in serum creatinine by 0.3mg/dL or more within 48 hours or Increase in serum creatinine to 1.5 times baseline or more within the last 7 days or Urine output less than 0.5 mL/kg/h for 6 hours. Stage 1 is 1.5-9x baseline or >0.3 increase; Stage 2 is 2-2.9x baseline; Stage 3 is 3x baseline, or increase to > 4, or initiation of renal replacement therapy.

Absolute values for serum creatinine before and after surgery will be compared. When multiple postoperative creatinine measurements are made, the maximum difference is reported.

Incidence of a mean arterial pressure (MAP) < 60 mmHg for a cumulative duration of all hypotensive episodes of more than 10 minutes during the anesthetic phase of the procedure.

Incidence of a mean arterial pressure (MAP) < 55 mmHg for a cumulative duration of all hypotensive episodes of more than 10 minutes during the anesthetic phase of the procedure.

Incidence of a mean arterial pressure (MAP) < 50 mmHg for a cumulative duration of all hypotensive episodes of more than 10 minutes during the anesthetic phase of the procedure.

Incidence of a mean arterial pressure (MAP) < 60 mmHg for a cumulative duration of all hypotensive episodes of more than 20 minutes during the anesthetic phase of the procedure.

Incidence of a mean arterial pressure (MAP) < 55 mmHg for a cumulative duration of all hypotensive episodes of more than 20 minutes during the anesthetic phase of the procedure.

Incidence of a mean arterial pressure (MAP) < 50 mmHg for a cumulative duration of all hypotensive episodes of more than 20 minutes during the anesthetic phase of the procedure.

Depth and duration of intraoperative hypotension will be modeled by calculating areas under the threshold for mean arterial pressures (MAPs). Thresholds will vary from 75 mmHg to 50 mmHg in 5 mmHg decrements. Together these variables represent the depth and duration of intraoperative hypotension. To optimize goodness of fit of these variables, the decremental steps may be increased to 10 mmHg and more restrictive lowest and highest thresholds may be chosen for the statistical analysis.

Depth and duration of intraoperative hypotension will be modeled by calculating areas under the threshold for mean arterial pressures (MAPs). Thresholds will vary from 75 mmHg to 50 mmHg in 5 mmHg decrements. Together these variables represent the depth and duration of intraoperative hypotension. To optimize goodness of fit of these variables, the decremental steps may be increased to 10 mmHg and more restrictive lowest and highest thresholds may be chosen for the statistical analysis.

Depth and duration of intraoperative hypotension will be modeled by calculating areas under the threshold for mean arterial pressures (MAPs). Thresholds will vary from 75 mmHg to 50 mmHg in 5 mmHg decrements. Together these variables represent the depth and duration of intraoperative hypotension. To optimize goodness of fit of these variables, the decremental steps may be increased to 10 mmHg and more restrictive lowest and highest thresholds may be chosen for the statistical analysis.

Depth and duration of intraoperative hypotension will be modeled by calculating areas under the threshold for mean arterial pressures (MAPs). Thresholds will vary from 75 mmHg to 50 mmHg in 5 mmHg decrements. Together these variables represent the depth and duration of intraoperative hypotension. To optimize goodness of fit of these variables, the decremental steps may be increased to 10 mmHg and more restrictive lowest and highest thresholds may be chosen for the statistical analysis.

Depth and duration of intraoperative hypotension will be modeled by calculating areas under the threshold for mean arterial pressures (MAPs). Thresholds will vary from 75 mmHg to 50 mmHg in 5 mmHg decrements. Together these variables represent the depth and duration of intraoperative hypotension. To optimize goodness of fit of these variables, the decremental steps may be increased to 10 mmHg and more restrictive lowest and highest thresholds may be chosen for the statistical analysis.

Depth and duration of intraoperative hypotension will be modeled by calculating areas under the threshold for mean arterial pressures (MAPs). Thresholds will vary from 75 mmHg to 50 mmHg in 5 mmHg decrements. Together these variables represent the depth and duration of intraoperative hypotension. To optimize goodness of fit of these variables, the decremental steps may be increased to 10 mmHg and more restrictive lowest and highest thresholds may be chosen for the statistical analysis.

The time from arriving at the postanesthesia care unit (PACU) until the time the patient is considered ready for discharge (in minutes).

A specific time frame on the day of surgery: from the start of admission to the PACU to discharge from the PACU, an expected average of 4 hours

Cardiovascular drugs as defined under interventions. Average use for each drug will be calculated. Cardiovascular drugs that were given in <1% of cases are not reported, as the average dosages would be meaningless.

Cardiovascular drugs as defined under interventions. Average use for each drug will be calculated. Cardiovascular drugs that were given in <1% of cases are not reported, as the average dosage would be meaningless.

Cardiovascular drugs as defined under interventions. Average use for each drug will be calculated. Cardiovascular drugs that were given in <1% of cases are not reported, as the average dosage would be meaningless.

Cardiovascular drugs as defined under interventions. Average use for each drug will be calculated. Cardiovascular drugs that were given in <1% of cases are not reported, as the average dosage would be meaningless.

Cardiovascular drugs as defined under interventions. Average use for each drug will be calculated. Cardiovascular drugs that were given in <1% of cases are not reported, as the average dosage would be meaningless.

Cardiovascular drugs as defined under interventions. Time of first administration of cardiovascular drug relative to the time at which the mean arterial pressure (MAP) drops below 60 mmHg. Per patient the average time to first administration of all hypotensive episodes was calculated. That average time is used as the outcome variable. A negative value indicates that administration occurred before the drop in MAP.

Cardiovascular drugs as defined under interventions. Time of first administration of cardiovascular drug relative to the time at which the mean arterial pressure (MAP) drops below 60 mmHg. Per patient the average time to first administration of all hypotensive episodes was calculated. That average time is used as the outcome variable. A negative value indicates that administration occurred before the drop in MAP.

Cardiovascular drugs as defined under interventions. Time of first administration of cardiovascular drug relative to the time at which the mean arterial pressure (MAP) drops below 55 mmHg. Per patient the average time to first administration of all hypotensive episodes was calculated. That average time is used as the outcome variable. A negative value indicates that administration occurred before the drop in MAP.

Cardiovascular drugs as defined under interventions. Time of first administration of cardiovascular drug relative to the time at which the mean arterial pressure (MAP) drops below 50 mmHg. Per patient the average time to first administration of all hypotensive episodes was calculated. That average time is used as the outcome variable. A negative value indicates that administration occurred before the drop in MAP.

Total amount (mL) of intravenous fluids (as defined under interventions) administered during the surgical procedure.

Inclusion Criteria: 60 years and older Inpatients Scheduled for a non-cardiac surgical procedure under general or central neuraxial anesthesia Exclusion Criteria: Pre-existing end-stage renal disease: operationalized as a preoperative need for dialysis The following surgical procedures: renal surgery, cardiac surgery, organ transplantation, ophthalmic surgery, endoscopic gastrointestinal procedures, and (interventional) radiologic procedures. small non-invasive or minimally-invasive procedures will also be excluded, operationalized as excluding procedures with a surgical time of less than twenty minutes.