A computerized aid in ventilating neonates

Cornput. Biol. Med. Vol. 21. No. l/2. pp. 1%21,199l Printed in Great Britain

A COMPUTERIZED

00104325/91 s3.oo+.oo @ 1991 Pergamon Press plc

AID IN VENTILATING

NEONATES

Department of Paediatrics, University Hospital of Copenhagen, Hvidovre Hospital, DK-2650 Hvidovre, Denmark (Received 6 April 1990; in revised form 1 November 19!40;received for publication 14 December 1990) Abstract-A computer program for ventilating neonates using a volume controlled ventilator is presented. The program proposes directions for changes of ventilator settings decided from the actual arterial blood gas samples and ventilator settings. The program deals with up to six babies at the same time and contains a continuous evaluation of the last six values of pC0, and p02 resulting in statements and warnings in potentially harmful situations. The program is consistent with the written instructions of the department. The ventilator treatment of 30 premature babies is evaluated retrospectively using the program, showing a total agreement of 37.5%, lowest among the babies who died in respiratory insufficiency. The advantage of the use of the program is discussed. Computer program PC02 PO2

Ventilator treatment Arterial blood gas

Neonates

Trend

INTRODUCTION

During the later years an increasing proposition of very prematurely born babies have survived, many of them being dependent on ventilator treatment during days or weeks. Premature babies are very fragile and their primary lung disease is progressive during the first two to three days. The respiratory support adequate at birth will then often become insufficient during hours or days and their ventilator treatment has to be re-evaluated frequently [l]. Many premature babies survive due to ventilator treatment but much harm can be done if the treatment is delayed or mismanaged. Treating babies on a ventilator requires a good deal of experience and many physicians-especially the youngest-have problems with the ventilator settings and adjustments. To reduce this problem a computerized program has been constructed for ventilator treatment of neonates. The program has been worked out for a SERVO Ventilator (type 900A) which is a volume controlled ventilator. It refers primarily to babies with Respiratory Distress Syndrome (RDS). The principles are concordant with the instructions for ventilator treatment used in the Neontal Department, Hvidovre Hospital, Copenhagen. In this program it is possible to follow the settings and blood gas values of up to six babies at the same time and to switch between the babies. THE

PROGRAM

The computer program is written in Turbo-Basic@. It is designed for a volume controlled Servo 900A ventilator and is based on arterial blood gas values. It starts with an introduction, then a Main Menu where the user can choose submenus. Using subroutines it is possible to: (a) introduce new babies in the program; (b) exclude babies from the program; (c) choose intervals for the blood gases; * Current address: Department of Pediatrics, Viborg Sygehus Vest, DK-8800 Viborg, Denmark. 15

M. ARRBE

16

(d) input actual blood gases and ventilator settings. After finishing each subroutine or by using the function key “Es? Main Menu.

the user returns to the

(a) In starting the program the user can introduce one or more babies by writing its name. After each introduced baby the question “Would you like to introduce more babies? Y/N” has to be answered. The maximum input is six babies. (b) If the ventilator treatment has stopped or a child for some other reason has to leave the program he can be excluded using the subroutine “Exclusion of babies”. In doing this, all values of ventilator settings and blood gases for that child are deleted. (c) For each introduced baby it is necessary to choose the intervals which are considered optimal for the arterial blood gas values. This is done by putting in the MIN and MAX values for pC0, and pOZ respectively. The unit for blood gas values is kPa. The interval must be specified individually for every child and during the running of the program it is possible to change this interval. The user is free to choose intervals but remarks are passed on if the MIN or MAX values are exceptional, that means if MIN pCOZ is lower than 2.5 kPa or MAX pCOZ is higher than 6 kPa or if MIN pOZ is lower than 8 kPa or MAX p02 is higher than 12 kPa. If the user considers an exceptional value appropriate it will be accepted but a MIN value for pCOz or PO;, wrongly keyed in as higher than the corresponding MAX value will not. This feature of individual chosen MIN and MAX values and the possibility of changing the values during the treatment gives a flexibility of the program to treat babies with different categories of respiratory insufficiency in different phases of their disease. (d) To continue the program the user is asked to put in the actual or new values of blood gases and the corresponding ventilator settings. This is only possible after having chosen the intervals. The actual blood gas values have to be measured in kPa. The settings of the ventilator (FiOz, volume, frequency, maximum pressure, PEEP pressure, inspiration time and pause time) are then keyed in. For every setting the unit is specified. The program will only accept regular values. That means: 21SFiOz5100 for FiO,(%): 0 < volume < 10 for volume (Urnin): 0 < frequency I70 for frequency (cycles/minute): 0 < maximum pressure for maximum pressure (cm H,O): for PEEP pressure (cm HZO): 0 d PEEP pressure 5 20 15 5 inspiration time 5 33 for inspiration time (%): 0 I pause time S 20. for pause time (%): Having put in all values and settings the user is able to make changes if a fault has occurred. Having accepted all changes the actual values and settings of the ventilator are shown together with a table consisting of the chosen intervals and the last six values of pCOZ and p02 thus enabling the user to see the fluctuations of the blood gases (Fig. 1). The user is then asked to write down his own proposal for changing the ventilator settings. After clearing the screen this suggestion is repeated along with the computerized proposal (Fig. 2). The last six values of pC0, and p02 are used as basis for trend analysis to inform the user about both abrupt and more gradual deviations. Principles

The principle of the program is changing the ventilator settings in a manner first related to the value of the pCOZ then the ~0~. The program does not contain any authorized or ideal values of ventilator settings. The actual ventilator settings are taken into consideration in a strictly logical sequence based on knowledge of the function of the volume controlled ventilator and the physiology of the respiratory system of the neonates [l] to suggest a rational way of changing the settings.

Computerized

aid in ventilating neonates

17

LATEST VALUES FOR BLOOD GAS STATUS AND VENTILATORSETTINGS

:

You are still treatfng

Leyla

4.00 kPa

(1) PCO2=

11.50 kPa

(2) PO2= (3) FiO2=

70% 1.70 Litres

(4) VOLUME= (5) FREQUENCY=

50 per minute

(6) MAX. PRESSURE=

35 mbar

(7) PEEP PRESSURE=

4 mbar

(6) INSPIRATION TIME=

25%

(9) PAUSE TIME=

CHOSEN INTERVAL

10%

PC02 3.50 to 4.50 kPa

Last 6 values for PC02 Last 6 values for PO2

PO2 10.00 to 12.00 kPa

4.00

4.49

3.58

2.90

3.25

11.50

6.03

10.30

23.00

24.90

Which ventilator-setting(s)

4.09 8.31

would you suggest to change ?

Write your suggestion - and press ?

Fig. 1. Screen picture showing the latest values of blood gases and ventilator settings for a baby (Leyla) along with the chosen interval and the last 8~ values for pCOz and pOz.

The proposal for the changes are given not as numerical values but as direction of changes. Every proposal includes changes of the most essential settings, as few as necessary and not more than four. Of the many possible combinations the program is constructed to choose the proposal of changes in ventilator settings which will give the lowest risk of barotrauma to the baby. The magnitude of changes is left to the user. If pCOz is within the chosen interval the suggestion depends upon the ~0,: If pOz is in the lowest part of the chosen interval, the suggestion is “not doing anything”. If PO,! is in the upper part of the interval or up to 15 kPa the advice depends upon the settings: a high FiOz (>70%) results in the advice “reduce the FiO,“. If FiOz is less than 70% the succeeding advices are: (1) reducing the pause time if > 10% ; (2) reducing the inspiration time if ~25%; (3) reducing the PEEP pressure if more than 6 cm H,O; (4) reducing the Fi02 if over 50%; (5) reducing a PEEP pressure over 4 cm H,O; (6) reducing the Fi02 if over 35%; (7) reducing the PEEP pressure if over 2 cm H,O; and (8) reducing the FiOz to atmospheric level.

YOUR SUGGESTION

: INCREASE Fi02

MR. Me’s SUGGESTION

: INCREASE Fi02 AND VOLUME Press to continue

Fig. 2. An example of the users and the computers (Mr Me’s) proposals for changing ventilator settings.

18

M. ARRBE

When the FiOz is reduced to 35% or less and the PEEP pressure 4 cm HZ0 or less the computer will ask the question “Is there any reason to keep the baby on the ventilator? Y/N”. If “Y” the following advice will be in the order given above, if “N” the computer proposes reduction of volume and weaning from the ventilator. Whatever level of MAX pOZ chosen by the user an actual p02 higher than 15 kPa always results in some advice of new settings which will decrease ~0~. The advice given depends on the actual ventilator settings to reduce the risk of combined barotrauma and oxygen toxicity. If p0, is lower than the chosen interval the advices are: (1) increasing FiOz if less than 40% ; (2) increasing PEEP pressure if 5 4 cm I-I*0 combined with an increase in Fi02 if possible; (3) if the baby is not particularly hyperventilated (pCOZ in the upper half of the chosen interval) the proposal is to increase the ventilation. If the maximum pressure is S 38 cm H,O this is done by increasing the volume. If the maximum pressure is more than 38 cm HZ0 it is done by increasing both the volume and the frequency if this is not maximal. In both instances the FiOz is suggested elevated if possible. When the pCO* is reduced the following will be advised: (4) increasing the PEEP pressure if lower than 6 cm H,O; (5) increasing the FiOz if lower than 80%; (6) increasing the PEEP pressure; (7) increasing the FiOz up to 100% ; (8) increasing the inspiration time; (9) increasing the pause time; and (10) increasing the PEEP pressure. During these steps warnings are shown to point out that the baby is shunting, calling attention to the possibility of electrolyte problems, patent ductus arteriosus or persistent pulmonary hypertension and the possibility of given medical treatment with diuretics, indomethacin or tolazoline depending upon the clinical findings. If pCOZ is higher than the chosen interval the proposals will be as follows: (1) If the maximum pressure is 5 38 cm H20 the advice will be to increase the volume of the ventilator combined with a reduction in Fi02 if the pOZ is above the MAX value of the chosen interval for pOZ. If p02 is below the MIN value of the interval, Fi02 will be advised elevated and an increase in PEEP pressure will also be proposed if the PEEP pressure is lower than 4 cm H20. (2) If the maximum pressure is ~38 cm HZ0 the advice will be to increase the volume and the frequency up to the maximum frequency combined with a reduction in Fi02 if pOZ is higher than the MAX value of the chosen interval for ~0~. If p02 is below the MIN value of the interval the proposal will be to increase the FiO,, and to increase the PEEP pressure provided the PEEP pressure is lower than 4 cm H20. (3) if the maximum pressure is >38 cm HZ0 and the frequency >60 the advice will be to increase first the inspiration time then the pause time combined with a reduction or an increase in FiOz depending on whether the PO;! is above or below the chosen values of the interval for pOz. (4) If both the inspiration time and the pause time are maximal the advice will be to increase the volume combined with an increase or reduction of FiO, depending on the level of the pOz as described above. If pCOZ is below the chosen interval the proposals will be as follows: (1) Reduction of the pause time if it is >lO% combined with a reduction or increase of FiOZ depending on whether the p02 is above or below the chosen values of the interval for PO*. (2) Reduction of the inspiration time if it is >25% combined with a reduction or increase of FiOz depending on the level of the ~0~. (3) Reduction of the volume combined with a reduction or increase of FiO, depending on the level of the PO,;.

Computerized

aid in ventilating

neonates

19

The baby’s pO2 has fallen, consider --

The CIRCULATION and BLOOD PRESSURE - does the baby need transfusion ? Has the baby developed PNEUMOTHORAX ? Is the TUBE PLACEMENT correct 7 Is the endotracheal tube PATENT ? Does X-ray of the lungs show ATELECTASE or INFILTRATES 3 Are there any signs of INFECTION 3 DID YOU FIND ANYTHING WRONG WITH YOUR BABY ?

If YES : This is NOT the time for coffee drinking !! You have to treat your baby and take good care of It.

To continue : Press
Fig. 3. The warning given when p02 decreases

too much.

Trends

Trends for both pCOz and p02 are built into the program. Every pOz within the accepted interval is related to the previous five values to find a propensity. If p0, within the chosen interval decreases more than 1 kPa between two measurements a warning is given (Fig. 3) to draw attention to the possibility of complications, for example air leak or problems caused by the tube, circulation or infection. If p02 decreases more gradually but continuously within the chosen interval the computer will make a remark “The baby’s p02 is decreasing. Is that OK? Y/N”. The resulting proposal depends on the answer “Y” or “N” accepting or trying to rectify the trend. For pCOz trends are built in for both increases and decreases within the chosen interval. The reason for this double trend is the potential harm of both low and high pCOz and the importance of active early intervention. The resulting proposal depends, in a way similar to that just described, on the user’s acceptance or not of the trend. All advice and remarks are shown in contrasting colors. Validation

The advantage of computer assisted ventilator treatment has been tested by comparing the arterial blood gases and ventilator settings of 30 premature babies with RDS treated in the conventional way with the suggestions obtained by using the program. 745 blood gas samples were examined. Agreement was divided into total agreement and partial agreement. A total agreement defined as cases where all the suggestions given by the program were carried out was found in 37.5% of the babies. A partial agreement where at least one of the proposed changes was carried out and no contradiction in the remaining changes existed was found in 68.0%. No inconsistency between the written instructions and the program was revealed. Six of the 30 babies died of respiratory insufficiency. The average total agreement for these babies was l&7%, partial agreement was 60.7%. Of 10 babies where the total

M. ARRP)E

20

agreement was 45.0% or more the partial agreement was on average 81 .O% and only one

died of respiratory

insufficiency. DISCUSSION

A variety of computer programs have been constructed to guide decision-making and to provide assistance in different fields of medicine [2,3]. An expert system for ventilator treatment in neonates using a pressure-limited, timecycled ventilator has been described [4] resulting in a facilitated treatment of neonates although it did not take into consideration clinical findings and previous responses to the ventilator treatment. This study presents a knowledge based system for ventilating neonates using a volume controlled ventilator. The algorithm is based on clinical studies of respiratory insufficiency in neonates [l, 5,6] and the instructions for ventilator treatment in our neontal department. The program deals with up to six babies at the same time and saves the last six values of blood gases for each. Individual values for the desired intervals for pCO,? and pOz have to be chosen and can be changed during the running of the program. The program contains a continuous evalution of trends for pCOz and pOz resulting in a variety of statements to draw the user’s attention to potentially harmful events such as hypoxemia, hypo- and hypercapnia and warnings in situations where complications may have occurred. The advantage of using the program has been tested retrospectively on 30 babies showing a total agreement of 37.5% and a partial agreement of 68.0%. The agreement was lowest among the babies dying from respiratory insufficiency (18.7%). In this examination of a total number of 745 blood gas values no disagreement between the proposal of the program and the written instructions was found. The disappointingly low total agreement demonstrates the difficulties in optimizing a treatment as complicated as ventilator treatment without an easy access to qualified assistance. This knowledge based system may be of help in decision-making in ventilating neonates combined with continuous clinical evaluation of the baby. On request the program can be obtained in a compiled version on 5y diskettes. SUMMARY A knowledge based system for ventilating neonates using a volume controlled ventilator is presented. The algorithm is derived from clinical studies of respiratory insufficiency in neonates and the instructions used for ventilator treatment in our neonatal department. The program proposes directions for changes of ventilator settings decided from the actual settings and arterial blood gas samples. The program can deal with up to six babies at the same time. Individual values for the desired intervals for pC02 and pOz have to be chosen and can be changed during the running of the program. The program contains a continuous trend evaluation of the last six values of pCOz and pOz resulting in statements and warnings in potentially harmful situations. The program has been tested retrospectively on 30 babies showing a total agreement of 37.5%. The agreement was lowest among the babies who died in respiratory insufficiency (18.7%). This knowledge based system cannot be used without evaluation of the clinical course of the baby but may be of help in decision-making in ventilating neonates. Acknowledgement-‘The Hygiene, Copenhagen,

author would like to thank K. R. Ennow Denmark, for assistance in computerizing.

at the National

Institute

of Radiation

REFERENCES 1. J. P. Goldsmith and E. H. Karotkin, Assisted Venrilafion of the Neonate. Saunders, Philadelphia, PA (1988). 2. P. A. Ball, D. C. A. Candy, J. W. L. Puntis and A. S. McNeish, Portable bedside microcomputer system for management of parenteral nutrition in all age groups, Archs Dis. Childh. 60, 435-439 (1985).

Computerized

aid in ventilating neonates

21

3. W. Shoemaker, A patient care algorithm for cardiac arrest, Crit. Care Med. 4, 157 (1978). 4. W. A. Carlo, L. Pacifido, R. L. Chatburn and A. Fanaroff, Efficacy of computer-assisted management of respiratory failure in neonates, Pediatrics 78, 139-143 (1986). 5. G. Sedin, Positive-pressure ventilation at moderately high frequency in newborn infants with respiratory distress syndrome (IRDS), Actu Anaesth. &and. 30, 515-520 (1986). 6. M. K. Chakrabarti, R. M. Grounds, G. 0. Swenzen and J. G. Whitwam, Relationship between frequency of ventilation, airways and pulmonary artery pressures, cardiac output and tracheal tube deadspace, Actu Anaesrh. Stand. 30, 678-684 (1986). About the AU~~O~-METTE ARR(DE received her M.D. degree from the University of Copenhagen, Denmark, in 1974. Postgraduate training at several university hospitals in Copenhagen led to authorization as specialist of paediatrics in 1986. Since 1985 she has worked mainly as neonatologist at the University Hospital of Copenhagen, Hvidovre Hospital, with special interest in ventilator treatment of premature babies. She has a special interest in the application of computers in medical practice. Since 1990 she has been employed as paediatric consultant at Viborg County Hospital, Denmark.