by family physicians and obstetricians

I OriginalResearch

Differences in intrapartum obstetric careprovided to women at low riskby family physicians and obstetricians

Anthony J. Reid, MD, CCFPJune C. Carroll, MD, CCFPJames Ruderman, MD, CCFPMichael A. Murray, PhD

To determine differences in practice style and toexamine maternal and neonatal outcomes, wereviewed the hospital charts of 1115 womenadmitted by family physicians and 1250 womenadmitted by obstetricians who gave birth at oneof three teaching hospitals in Toronto betweenApril 1985 and March 1986. All the women inthe two groups were categorized retrospectivelyas being at low risk at the onset of labour on thebasis of their prenatal records and their admis-sion histories and physical examination results.There were higher proportions of youngerwomen and women of lower socioeconomicstatus in the family physician group than in theobstetrician group (p < 0.001). The rates ofinterventions, including artificial rupture of themembranes, induction, augmentation, low for-ceps plus vacuum extraction, episiotomy andepidural anesthesia, were all higher in theobstetrician group. The mean birth weight andthe cesarean section rate were the same in thetwo groups. Differences in labour and deliveryoutcomes between the two groups, including ahigher rate of spontaneous vaginal delivery forthe family physicians, reflected a more "expec-tanf" practice style by family doctors. However,there were no significant differences in the ratesof maternal or neonatal complications. A prac-tice style characterized by a higher rate ofinterventions was not associated with improvedmaternal or newborn outcome in this low-risksetting.

From the Department of Family and Community Medicine andthe Health Care Research Unit, University of Toronto

Reprint requests to: Dr. Anthony J. Reid, Family Practice Unit,Toronto General Hospital, 200 Elizabeth St., Toronto, Ont.M5G 2C4

Comparaison sur dossiers des pratiques obstetri-cales et des issues maternelles et neonatalesdans le cas de 1115 femmes hospitalisees par desmedecins de famille et 1250 par des obstetri-ciens, chacune d'entre elles ayant donne nais-sance dans un de trois hopitaux universitairesde Toronto d'avril 1985 a mars 1986. L'analyseretrospective du dossier prenatal, de l'anamneseet de l'examen k lentree fait placer chacuned'elles dans la categorie des grossesses a risquepeu eleve en dEbut de travail. Les patientes desmEdecins de famille sont, dans lensemble, plusjeunes et de niveau socio-Economique plus basque celles des obstEtriciens (p < 0,001). Maischez celles-ci on note plus d'interventions (rup-ture des membranes, provocation du travail,stimulation, application de forceps au detroitinferieur et de ventouse, Episiotomie, anesthEsieepidurale). Dans les deux groupes le poids natalmoyen et le taux de cEsarienne sont les mimes.Les differences observEes dans l'issue du travailet de l'accouchement, tel le taux plus eleve denaissances spontanEes par voie basse chez lespatientes des mEdecins de famille, temoignentchez ceux-ci d'une plus grande tendance k l'ex-pectative. Cependant, il n'existe aucune dif-ference significative dans les taux de complica-tions maternelles ou nEonatales. Aussi croyons-nous que dans le cadre que nous avons decrit,les pratiques obstEtricales comportant un forttaux d&interventions n'amEliorent nullement lepronostic de la mbre et de son nouveau-nE pourles grossesses k risque peu dlevE.

amily physicians' involvement in obstetricsF in Canada and the United States is de-

cliningl-4 at a time of marked change inobstetric practice. Two conflicting trends are

CMAJ, VOL. 140, MARCH 15, 1989 625- For prescribing information see page 756

emerging: increasing use of technology in anattempt to ensure optimal outcome of childbirthand the desire to "humanize" the birth experienceand avoid unnecessary technical intrusions into anatural process.5'6 Moreover, a "cascade" effect hasbeen described by which some interventions inex-orably lead to others and, hence, the use of severalprocedures with unproven beneficial effects onoutcome.7

Numerous studies have shown that familyphysicians practise obstetrics with fewer interven-tions than do obstetricians.8-8 However, the effectof this management style on quality of care and onmaternal and neonatal outcomes is not clear.18'19Noncomparability of the two patient populationshas been an important problem, with wide differ-ences in medical, obstetric and social characterist-ics. As well, differences in patients' preferences asto type of caregiver, which theoretically introducea selection bias, have not been described. Moststudies have had small samples and were unable toreach sufficient statistical power to compare certaininfrequent outcomes. In addition, process and out-come measures were not all recorded blindly orobjectively.

In this study we sought to address some ofthese problems. The problem of noncomparabilityof the groups was reduced by including onlywomen classified as being at low risk medicallyand obstetrically. A larger number of subjects wasused than in previous studies, with statisticalpower calculated in the planning stage, and criteriafor interventions and outcomes were defined be-fore the data were recorded.

Our study was a retrospective, hypothesis-generating study whose primary purpose was toexamine the rates of obstetric intervention in twogroups of women at low risk, one cared for byfamily physicians and the other by obstetricians. Asecondary purpose was to determine whether anyobserved differences in management style werecorrelated with different maternal and newbornoutcomes.

Setting

Patients were chosen from three downtownteaching hospitals of the University of Toronto. Ineach setting family physicians and obstetricianspractised in the same hospital environment. Thefamily physician groups varied from a group of 30physicians doing approximately 1000 deliveries peryear (hospital 1) to a group of 14 performing about400 births per year (hospital 3). Two of thehospitals (1 and 2) were level 3 regional referralcentres, and the third had level 1 neonatal careavailable. (A three-tiered system exists in Ontario.Level 1 centres provide perinatal care for patientsat low risk and mildly ill patients [approximately85% of patients], level 2 centres provide care formoderately ill neonates and women [12%], andlevel 3 centres are sophisticated regional referralcentres providing ventilation and surgical facilities

[2% to 3%].2°) Both the family physician and theobstetrician groups included university-appointedphysicians as well as community-based, nonaca-demic practitioners.

In this setting obstetricians care for a largenumber of women at low risk. Women whosefamily physicians practise obstetrics usually remainwith these physicians for maternity care. Somewomen whose family doctors do not practiseobstetrics are referred to another family physicianpractising obstetrics. Some women directly seekobstetrician care, even in low-risk cases, believingit to be more highly skilled, whereas other womenchoose family physicians, feeling they will receivemore personal and less technically intrusive care.The actual number of women in each of thesegroups is not known, and it has not been reportedfor other practice sites. Since only about 30% ofOntario family doctors practise obstetrics,3 theproportion being smaller in the major cities, largenumbers of women at low risk receive their carefrom specialists. At the time of the study mostobstetricians were charging their patients a directfee that was not covered by the provincial healthinsurance plan or private insurance, a factor thatmay have discouraged some women from seekingspecialist care. The family physicians in the studygenerally did not charge this fee. It is clear,therefore, that the two groups of women may havechosen their caregiver with certain characteristicsin mind, and we did not attempt to quantify this,although it is a recognized source of bias in theanalysis.

There were some differences in obstetric prac-tice between the three institutions, such as proto-cols for use of oxytocin and the availability oftechnical aids such as electronic fetal monitors andvacuum extractors (the latter were only used inhospital 1). Guidelines for consultation were quitesimilar, however: basically, consultation was re-quired in any complicated cases, such as thoseinvolving induction, augmentation and complexoperative delivery. By combining data from thethree hospitals we attempted to reduce the influ-ence of specific institutions and individual physi-cians, which has been described as a factor in styleof obstetric practice.18

Methods

We reviewed the hospital records of womenwho gave birth at the three hospitals betweenApril 1985 and March 1986. We examined thepatients' prenatal records and their admitting his-tories and physical examination results and classi-fied the women as being at low risk at the onset oflabour if they satisfied the criteria for grade A(pregnancy at no predictable risk) of the OntarioAntenatal Record (no prior perinatal death orinfant of low birth weight, no significant medicaldisease, no complications in current or prior preg-nancy [e.g., bleeding, hypertension or premature

626 CMAJ, VOL. 140, MARCH 15, 1989

labour] and adequate fetal growth). Even womenwith significant tobacco, alcohol or drug intakewere excluded. Cases in which risk status couldnot be determined were excluded.

The charts were then analysed prospectively,the results being recorded by technicians who wereblind as to type of caregiver. The charts wereassigned to the family physician or obstetriciangroup on the basis of the admitting doctor's status,regardless of the type of personnel subsequentlyinvolved in management. This was to ensure thatthe data for patients admitted under the care of afamily physician and subsequently transferred toan obstetrician because of intrapartum complica-tions would be analysed as part of the familyphysician group.

The family physician group consisted of allwomen who were admitted by a family doctor andwho met the criteria for low risk. The obstetriciangroup was selected by means of random numbertables from all women admitted by an obstetrician.Each chart was then examined to determinewhether it met the criteria for low risk. Chartswere selected and classified until the number ofwomen at low risk admitted by an obstetrician wasapproximately equal to the number of womenadmitted by a family physician at that hospital.

One research assistant was responsible forcoordinating data collection and checking the reli-ability of the data extracted by the other techni-cians. All items extracted were checked beforehandto be sure they were recorded clearly on thehospital charts. Problems with data collection wereidentified in a pilot study of 170 charts andresolved before the actual study charts were exam-ined. We were involved in establishing rules forstandardized data extraction and verification ofaccuracy. There were regular checks of interraterand intrarater reliability for about 10% of thecharts during the data collection.

Information was recorded as it appeared in thecharts. Measures of maternal outcome were eitheravailable directly from the chart or had beendefined previously with expert colleague input(postpartum hemorrhage: loss of at least 500 ml ofblood plus treatment with extra oxytocin, ergome-trine or transfusion; postpartum fever: temperatureof 38.5°C or more on two occasions at least 6hours apart; urinary tract infection: positive cultureresult; and perineal infection: infection requiringtreatment with antibiotics or incision and drain-age). Neonatal outcome measures were availabledirectly from the labour records or the nurserynotes. All problems necessitating admission to aspecial care baby unit were tracked individuallyand classified into groups.

Sample size was calculated before the datacollection to give sufficient power to minimize thechance of a type II error. Power calculation isdifficult in a study with a large number of varia-bles. However, for events occurring at least 10% ofthe time in the group with the higher rate a sampleof 1200 per group can allow detection of a 40%

difference between the two groups with an alphavalue of less than p = 0.05 and a power of 0.8.Analyses were performed by the Health CareResearch Unit, University of Toronto, using theStatistical Package for the Social Sciences(SPSS/PC Plus, version 2)21 for the t-test andchi-squared test and PC-SIZE22 for the powercalculations. The validity of the t-tests was checkedby median (parametric) and Mann-Whitney (non-parametric) tests to compare measures of centraltendency.

The most important neonatal outcome wasadmission to a special care unit. We calculated thata sample size of 1200 per group would allowdetection of a 50% difference in rates (i.e., 2% to4%) with a power of 0.8.

Results

Of 1779 women admitted by family physi-cians during the study period 1115 (62.7%) wereclassified as being at low risk at the onset oflabour. The charts of 1250 women at low riskadmitted by obstetricians were examined.

Table I shows the characteristics of the 1115patients in the family physician group and the1250 patients in the obstetrician group. There wereno differences in mean diastolic blood pressure ormean gestational age between the two groups.However, there was a higher proportion of patientsaged 19 years or less in the family physician groupthan in the obstetrician group (6.2% v. 2.3%).There were also more women of lower socioeco-nomic status in the former group, as indicated bythe proportions requesting ward v. private accom-modation (extra fees were required for a privateroom) and the proportions receiving health insur-ance premium assistance. In addition, the familyphysician group had more primiparous patients(53.8% v. 47.8%).

Table II shows the rates of several intrapartummanagement procedures. Not shown are rates ofminor procedures, which were all more prevalentin the obstetrician group than in the family physi-cian group and included use of intravenous fluids(75% v. 67%), shaving (8% v. 5%) and enema (8%v. 4%). The rates of more invasive procedures suchas artificial rupture of the membranes, induction,augmentation and epidural anesthesia, were allsubstantially higher in the obstetrician group, par-ticularly among multiparous women. For example,the induction rate among primiparous women wastwice as high in the obstetrician group as in thefamily physician group (11.9% v. 6.7%), and therate among multiparous women was three times ashigh (15.2% v. 5.4%). Epidural anesthesia wasused almost exclusively for analgesia in this set-ting. The rates for continuous electronic fetalmonitoring (data available only from hospitals 2and 3) did not differ between the two groups.

The rates for delivery procedures show asimilar pattern (Table II). The rate of combined low

CMAJ, VOL. 140, MARCH 15, 1989 627

forceps and vacuum delivery was higher in theobstetrician group than in the family physiciangroup, particularly among multiparous women.The rate of midforceps delivery was also higher inthe obstetrician group. The rates of episiotomyamong primiparous women were similar in thetwo groups, but the rate among multiparous

d"able Characteristit-s owvt neen -ronsido be At Ioffarnily physician or a10 obstetFectr -

women was higher in the obstetrician group.Almost 33% of the women cared for by familydoctors received a consultation by an obstetricianduring labour or delivery.

Table III shows the labour and delivery out-comes. A difference in delivery style is reflected inthe longer lengths of the second and third stages in

. .. .. . *...vi., r S;k n S -mtAi rO ons;-~ X{;f.t65 S44" hit,% .,v h; vv e, e i t i .

a;C-s 1tr

15 i-; t

t

5¼ jem!du1J .210 S11H52 ¼)

331 ¼.

V

'. I3" t i

* i.3 .LiOdi E ,U .z

.J.3::,I 3I , -: -r , ;. . ,: -

able It Rates of labou, ciieI8verv proc-eduv MS Cf the two ur3til.

Vo..l........

I .

.T rh¼331L-ef' T1.

* 1 3. -f>, t -.*. ! , " ;

;I rta t :i

'vi Ji3 015esf;.-1 d3otrr;,fce~. ¼-.. taT

v13iydxj|t1lo.-i; .1¼.1-333 "¼ - 313 18 -'*

'3

*3, 45¼.~;5;.4$,@ ..S

:*84 :", 8,;. i :.",

Vh 2~. ..- 83.

;:,)''-. f

..; ...

x,*'iI 1;

20. 39t

. I3r.3614 13 r?. .s

a¼.33¼11A..ff3 .¼3h9133.¼

628 CMAJ, VOL. 140, MARCH 15, 1989

ll V

i- ; 1" - : 'e

C.r-: :!

the family physician group. Spontaneous vaginaldelivery occurred more frequently in the familyphysician group than in the obstetrician group(68.2% v. 55.1%). The rates of cesarean section didnot differ significantly between the two groups.

Maternal outcomes are shown in Table IV.The family physician group had a higher rate of.*first-degree and second-degree perineal tears thanthe obstetrician group (37.1% v. 25.1%), but therewas no difference in the rates of third-degree andfourth-degree tears. The proportions of patientswith an intact perineum were similar in the twogroups. There was no difference in the rates ofmaternal postpartum complications, as defined be-fore data collection, between the two groups.Family physicians' patients had slightly shorterhospital stays than obstetricians' patients, andslightly more were breast-feeding on discharge.

Table V shows neonatal outcomes. The twogroups were similar in mean birth weight and inproportions of babies weighing less than 2500 gand weighing over 4000 g. Of particular note is theremarkable similarity in rates of admission to a

special care baby unit, intubation and Apgar scoreless than 6 at 5 minutes. There were no stillbirths.In the two cases with an Apgar score less than 6 at5 minutes in the family physician group theintrapartum record indicated no problems until latein the second stage, when there were fetal heartrate decelerations. After prompt delivery a tightcord around each infant's neck was discovered.Both infants were resuscitated.

Table VI shows the number of admissions to aspecial care baby unit and intubations, by diagno-sis. In every category the results were virtually thesame for the family physician and obstetriciangroups.

Discussion

Our study extends the findings from previouswork comparing maternity care delivered by fami-ly physicians and obstetricians.8-18 The lower ratesamong family physicans of specific interventionsnoted in previous studies are confirmed in our

Table Ill - Labour and delivery outcomes in the two groups

Group

Outcome Family physician Obstetrician p value*

Mean (and SD)

Amount of cervical dilation on admission, mmLength of second stage, minLength of third stage, mint

Length of second stage > 60 minPrimiparasMultiparas

Length of third stage > 10 mint

Spontaneous vaginal delivery (no forceps or vacuum

extraction)Cesarean section

3.76 (2.25) 3.55 (2.24)52.76 (46.15) 47.05 (42.71)7.56 (6.94) 5.22 (3.86)

No. (and %) of patients

271 (45.2)66 (12.8)130 (24.0)(n = 542)

760 (68.2)76 (6.8)

261 (43.6)49 (9.5)68 (11.7)

(n = 581)

689 (55.1)96 (7.7)

*In this table and Tables IV and V the significance of the difference in mean values was assessed with the t-test; all the otherdifferences were studied with the chi-squared test.

tData from hospitals 2 and 3 only.

Table IV - Maternal outcomes in the two groups

Group

Outcome Family physician Obstetrician p value

Mean length of hospital stay (and SD), d 4.05 (1.73) 4.30 (1.54) < 0.001

No. (and %) of patientsPerineal tear

First- and second-degree 414 (37.1) 314 (25.1) < 0.001Third- and fourth-degree 53 (4.8) 52 (4.2) NSIntact perineum 113 (10.1) 120 (9.6) NSPostpartum hemorrhage 96 (8.6) 108 (8.6) NSPostpartum fever 29 (2.6) 24 (1.9) NSPerineal infection 1 1 (1.0) 1 1 (0.9) NSUrinary tract infection 11 (1.0) 15 (1.2) NSBreast-feeding on discharge 1048 (94.0) 1101 (88.1) < 0.001

CMAJ, VOL. 140, MARCH 15, 1989 629

0.0310.003

< 0.001

NS0.003

< 0.001

< 0.001NS

study with larger numbers of more strictly compa-rable women.

There are several issues to be addressed re-garding the validity of comparing these two pop-ulations. The first is whether a retrospective exam-ination of hospital charts can adequately assess riskstatus. The Ontario Antenatal Record was therecognized standard in these hospitals. Informationas to risk category was obtained from threesources: the antenatal form itself, the risk gradeassigned, and the admission history and physicalexamination results. If any one of these indicatedrisk beyond grade A the patient was excluded fromthe study. Data collection was simplified by thefact that all physicians in the study used the sameantenatal and admission record format.

Various methods for predicting perinatal riskhave been devised, but none is ideal or universallyaccepted.23 Grade A of the Ontario AntenatalRecord excludes all known significant medical andobstetric risk factors. The retrospective nature ofthe evaluation is a problem, but it is unlikely that aprospective randomized trial will ever be possible.Women would not accept randomization, and theHawthorne effect would alter the practice patternof the providers.

Another issue in retrospective studies usingchart review is the accuracy of hospital records.The accuracy of recorded information could not beverified in our study, as in similar studies. Howev-er, it is unlikely that there would be much system-atic bias. Much of the record keeping consisted ofcompleting forms, and a large part of the actualdocumentation was done by nurses and housestaff, whose services are shared by the two groupsof physicians.

The validity of neonatal outcome measureshas recently been examined. Like previous inves-tigators,9-'3 we used birth weight, Apgar score andadmission to a special care baby unit as measuresof outcome, but their specificity and predictivevalue have been questioned.214 It has been pro-posed that outcomes such as convulsions, tube-feeding for more than 7 days and assisted ventila-tion for longer than 48 hours would be moreaccurate predictors of newborn well-being, al-though their predictive value for long-term prog-nosis has not been determined. Our outcomemeasure of admission to a special care baby unitwould have included all babies with these seriousconditions along with more mildiy affected neo-nates. The number of admissions to special care

able V Neoniatal outcorne! n-r the.- tw'AqiO

ci1 i e jxai ai

g;iar sia

rable VI Number of admissivLnrs 1c. a specsae ;.ave baby unit andtubations in the two groups, by diagnosis

J~~~~~~-i

-1 aS1I

H niiL

c--o e:-

630 CMAJ, VOL. 140, MARCH 15, 1989

,-Jbsetcr

units was the same for both groups, and althoughthe subset of seriously ill infants may have beengreater in one group this is statistically unlikely.Our study was not designed to examine long-termoutcome, but this is worth pursuing in a futurestudy.

The reliability of information extraction by ourresearch technicians was addressed in the planningstages and pilot study. Every effort was made touse indicators of risk recorded directly on thecharts. When interpretation was required, ruleswere drawn up and modified until a high degree ofaccuracy was obtained in the pilot study; accuracywas monitored throughout the study.

Another crucial factor in the validity of ourstudy was comparability of the two groups, espe-cially as there was recognized bias in the choice ofcaregiver. At present there is no reliable informa-tion on why women choose one caregiver overanother in this setting, and we had no way ofdocumenting reasons for this choice. Given theconstraint of a retrospective study, we tried tocontrol for the areas that were documented in therecords: medical and obstetric risk factors. Thefamily physician group had higher proportions ofyounger women and women of lower socioeco-nomic status than the obstetrician group. To exam-ine this difference more closely, we analysed therates of artificial rupture of the membranes, induc-tion, augmentation, assisted delivery, epidural an-esthesia and episiotomy according to socioeconom-ic status (ward, semiprivate and private accommo-dation) and age (19 years or less, 20 to 30 yearsand more than 31 years). The rates of the sixinterventions were thus each analysed in six sepa-rate subgroups, for a total of 36 separate compari-sons. Except in four specific instances the differ-ences in intervention rates observed for the entiregroup were maintained within each subgroup,although in some subgroups trends failed to reachstatistical significance because of small numbers.Among women aged 19 years or less, familyphysicians and obstetricians both had inductionrates of 3%. Similarly, in both groups 18%o ofpatients who requested ward accommodation hadaugmentation, and the episiotomy rate for patientsaged 19 years or less and for those who choseward accommodation did not differ significantly.In no subgroup were the intervention rates forfamily physicians significantly greater than thosefor obstetricians.

Because the differences observed for the entiregroup were also found in the stratification analysis,except in these four specific cases, it is clear thatthe somewhat higher proportions of youngerwomen and women of lower socioeconomic statusin the family physician group were not an impor-tant source of bias.

The success of assembling two comparablegroups of women at low risk is difficult to docu-ment completely, but certain variables were simi-lar. The gestational age at the onset of labour wassimilar, as were matemal diastolic blood pressure,

mean birth weight and proportions of newbornsweighing less than 2500 g and more than 4000 g.The exclusion criteria were very strict, and it seemsunlikely that women in a high-risk category at theonset of labour were included in our study.

Since the samples were large, small differ-ences between them readily reached statisticalsignificance. Hence, the clinical significance of thedifferences was examined in each case as well.Furthermore, given the multiple hypotheses beingtested, some differences may have been shown bychance alone, and the p values are not exact testsof inference. However, when the alpha value isless than p = 0.001 it is assumed that the individu-al test would pass adjustment maneuvers.

Some of the differences between the groupswere of minor clinical significance (e.g., use ofintravenous fluids, shaving and enema). However,differences in the rates of other, more invasiveinterventions, including artificial rupture of themembranes, induction, augmentation, assisted de-livery, epidural anesthesia and episiotomy, werelarger and more significant, particularly amongmultiparous women, who would be expected tohave fewer complications. Some of the differencesare directly linked; for example, epidural anesthe-sia necessitates the use of intravenous fluids.

Because the two groups of women were bothat low risk at the onset of labour, the differencesobserved in their intrapartum management arelikely the result of other factors. We speculate thatthis is a combination of patients' expectations andcaregivers' habits. Certainly selection bias was noteliminated by the study design, and some womenwere known to have chosen either a family physi-cian or an obstetrician on the basis of perception ofdifferences in style of practice. An obstetricianmight have felt an expectation to provide more"active" management of labour, whereas a familyphysician might have been affected by the oppo-site pressure.

All providers of obstetric care are aware ofextremely demanding patients who insist on natu-ral childbirth and detailed justification of interven-tions. It is possible that a higher proportion ofthese women seek out family physicians becauseof the perception that family physicians are lesstechnically oriented. On the other hand, the moreacquiescent patient may be subtly expressing anexpectation for the physician to undertake anactive or intervening course. Some patients mayfeel that an obstetrician offers a higher level ofexpertise and that they need this. As well, sinceobstetricians of necessity use more interventions inhigh-risk cases, this pattern of practice may wellcarry over to their care of patients at low risk. Thismay explain some of the differences in manage-ment style found in our study.

The argument may also be made that thehigher intervention rate in the obstetrician groupimplies a higher rate of complications necessitatingthe interventions. We have no evidence to supportthis view. Large groups of women at low risk are

CMAJ, VOL. 140, MARCH 15, 1989 631

likely to experience similar numbers and kinds ofintrapartum problems. The interventions are morelikely to be due to management style, particularlywhen indications for some of the interventions arenot as clearly defined as for procedures such ascesarean section.

The primary question raised here is whether astyle of care characterized by higher rates ofintervention is associated with better maternal andneonatal outcome. An initial intervention oftenleads to a cascade of several procedures, all linkedtogether.7 Recent reviews of a number of intrapar-tum management procedures have suggested thatmany have no proven efficacy in low-risk set-tings.18'19 Techniques of value in high-risk settings(e.g., electronic fetal monitoring) have often beenapplied to low-risk situations before their worthwas demonstrated in the latter setting.725 Althoughdiffering types of practice were demonstrated inour two groups, we could not detect differences inmatemal or neonatal outcome.

A lower episiotomy rate in the family physi-cian group was associated with more first-degreeand second-degree perineal tears, but the rate ofthird-degree and fourth-degree tears was similar inthe two groups. Interestingly, for both groups thecombined rate of episiotomy and first-degree andsecond-degree tears was the same, as was the rateof intact perineum. Recent work by Reynolds andYudkin26 indicates that the relatively high episioto-my rates in this setting may be reduced withoutincreased matemal or neonatal problems and thattraining to avoid tears may increase the rate ofintact perineum.

One of the surprising findings of our studywas the similarity of the proportions of patientswith electronic fetal monitoring and of patientswho underwent cesarean section, in contrast to thedifferences in the rates of other interventions. Thisfinding may reflect adherence by family doctors tothe mandatory guidelines for consultation (an ob-stetrician was consulted in 33% of cases). Once thespecialist became involved, the intervention ratewould reflect his or her style of care.

We feel that the high rate of consultation for agroup of women at low risk reflects the stringencyof the mandatory guidelines for consultation,which require obstetrician input on virtually anysituation not considered normal. This may be seenas a positive characteristic in that the familydoctors sought assistance readily and specialisthelp was available quickly, which undoubtedlycontributed to the excellent maternal and neonataloutcomes. It was also notable that 37.3% of thepatients admitted under the care of family physi-cians did not meet the low-risk criteria. We feelthat this reflects the confidence the family doctorshad in managing complicated cases with specialistcolleagues, whose assistance was readily available.In a rural area some of these patients might havebeen referred to a larger centre.

The slightly shorter hospital stay among thefamily physician group may have been due to the

fact that the family physicians were more comfort-able with earlier discharge, knowing that theycould see the woman and child together if neces-sary.

The most important measure of neonatal out-come was the rate of admission to a special carebaby unit. The rate was 3.9% in the familyphysician group and 3.4% in the obstetriciangroup; the difference is well below that initiallyspecified to avoid a significant type II error. A 4%rate for pregnancy at low risk was also recentlyreported by Moutquin and colleagues.27 Further-more, examination of the individual cases byindication and intubation rate showed that ourresults were remarkably similar for every category.We feel that it is unlikely that there were signifi-cant differences in neonatal complications thatwere not documented in this way.

Conclusion

In our study we found different styles ofobstetric management of low-risk cases character-ized by different rates of procedures for -familyphysicians and obstetricians. We were unable todemonstrate any significant improvement in ma-ternal or neonatal outcome among the womenmanaged in the generally more interventionistmanner. The results should prompt an examinationof practice style by all physicians. Since no inter-vention is risk free we suggest that physicians usean intervention only when it has been demonstrat-ed to have proven benefit in that setting.

Current obstetric research is generally aimedat areas of high risk in an effort to reduce perinatalcomplications and death. Our findings raise impor-tant questions as to why certain interventions areas frequent as they are in patients at low risk.Reappraisal of the current medical approach tolow-risk obstetrics is warranted, and more studiesspecifically developed for the low-risk setting areneeded. Outcome measures need to be refined, andinterventions must be assessed individually as totheir benefit in patients at low risk.

The purpose of our study was not to labeldifferent groups of obstetrics caregivers but, rather,to determine whether two distinct approaches tolow-risk maternity care led to different outcomes.Although in general obstetricians did use moreprocedures, it is recognized that within this medi-cal community there are practitioners in bothgroups who practise in a style characterized bymore or less intervention.

An encouraging finding was that, in thissetting, family physicians did appear to providematernity care with a degree of safety comparableto that of obstetricans for women at low risk. Thesimilar matemal and neonatal outcomes associatedwith a higher rate of spontaneous vaginal deliveryfor the family doctors may also be seen as apositive feature of their style of care. This isimportant for the family practice community,

632 CMAJ, VOL. 140, MARCH 15, 1989

which continues to withdraw from obstetric ser-vice.

Our findings should encourage both familyphysicians and obstetricians to carefully evaluatetheir approach when providing maternity care topatients at low risk.

We thank Dr. Michael Klein, the National Family Prac-tice Obstetrics Research Group in Canada and Dr. YvesTalbot for their support, Kaela Jubas, Anne Spitzer,Deena Ages and Bob Singh for data collection, SyrilinThompson and Karen Eveleigh for secretarial assistance,and Drs. Murray Enkin, John Frank and David White forcritical reviews of the paper.

This study was supported by the Department ofFamily and Community Medicine, University of Toron-to, the College of Family Physicians of Canada and theHealth Care Research Unit, University of Toronto.

References

1. Klein M, Reynolds JL, Boucher F et al: Obstetrical practiceand training in Canadian family medicine: conserving anendangered species. Can Fam Physician 1984; 30: 2093-2099

2. Scherger J: The family physician delivering babies: anendangered species. Fam Med 1987; 19: 95-96

3. Perkin RL: Obstetrics anyone [E]? Can Fam Physician 1985;31:1561

4. Klein M: Obstetrics is too important to be left to theobstetricians [E]. Fam Med 1987; 19: 167-169

5. Report of the Task Force on the Implementation of Mid-wifery in Ontario, Ont Ministry of Health, Toronto, 1987

6. Klein M: The Canadian family practice accoucheur. CanFam Physician 1986; 32: 533-540

7. Brody H, Thompson JR: The maximin strategy in modemobstetrics. IFam Pract 1981; 12: 977-986

8. Meyer BA: Audit of obstetrical care: comparison betweenfamily practitioners and obstetricians. Fam Pract Res J 1981;1: 20-27

9. Klein M, Uoyd L, Redman C et al: A comparison oflow-risk pregnant women booked for delivery in twosystems of care: shared care (consultant) and integratedgeneral practice unit [two parts]. BrJ Obstet Gynaecol 1983;90: 118-128

10. Wanderer MJ, Suyehira JG: Obstetrical care in a prepaidcooperative: a comparison between family practice resi-dents, family physicians, and obstetrcians. J Fam Pract1980; 11: 601-606

11. Franks P, Eisinger S: Adverse perinatal outcomes: Is physi-cian specialty a risk factor?JFam Pract 1987; 24: 152-156

12. Phillips WR, Rice GA, Layton RH: Audit of obstetric careand outcome in family medicine, obstetrics, and generalpractice. JFam Pract 1978; 6: 1209-1216

13. Caetano DF: The relationship of medical specialization(obstetricians and general practitioners) to complications inpregnancy and delivery, birth injury and malformation. AmI Obstet Gynecol 1975; 123: 221-227

14. Ely JW, Ueland K, Gordon MJ: An audit of obstetric care ina university family medicine department and an obstetrics-gynecology department. JFam Pract 1976; 3: 397-401

15. Taylor GW, Edgar W, Taylor BA et al: How safe is generalpractice obstetrics? Lancet 1980; 2: 1287-1289

16. Rosenblatt RA, Reinken J, Shoemack P: Is obstetrics safe insmall hospitals? Evidence from New Zealand's regionalisedperinatal system. Lancet 1985; 1: 429-432

17. Phillips WR, Stevens GS: Obstetrics in family practice:competence, continuity and caring. J Fam Pract 1986; 20:595-596

18. Rosenberg E, Klein M: Is maternity care different in familypractice? A pilot matched pair study. J Fam Pract 1987; 25:237-242

19. Mengel MB, Phillips WR: The quality of obstetric care infamily practice: Are family physicians as safe as obstetri-cians? JFam Pract 1987; 24: 159-164

20. Patrick J (chmn): Levels of perinatal care [appendix E]. InReproductive Care: Towards the 1990's. Second Report ofthe Advisory Committee on Reproductive Care, Ont Minis-try of Health, Toronto, 1988: 112-114

21. Norusis MJ: SPSS/PC+ Version 2.0. Base Manual for IBMPC, PC/XT/AT, PS/2, SPSS Inc, Chicago, 1988

22. Dallar G: PC-SIZE. A Program for Sample Size Determina-tion. Version 2.0, Tufts University, Boston, 1985

23. Wall E: Assessing obstetrical risk. I Fam Pract 1988; 27:153-163

24. Patrick J (chmn): Report of subcommittee on reproductivemorbidity [appendix G]. In Reproductive Care: Towards the1990's. Second Report of the Advisory Committee onReproductive Care, Ont Ministry of Health, Toronto, 1988:154-156

25. Klein M and Medical Subcommittee on Humanizing Ob-stetrical Care (Lefevre Y, chmn): Controversies in Obstetri-cal Management and Matemal Care, Direction des Commu-nications, ministere de la Sante et des Services sociaux,Quebec, 1987

26. Reynolds JL, Yudkin PL: Changes in the management oflabour: 2. Perineal management. Can Med Assoc J 1987;136: 1045-1049

27. Moutquin JM, Gagnon R, Rainville C et al: Matemal andneonatal outcome in pregnancies with no risk factors. CanMedAssocJ1987; 137: 728-732

For the birds!

It is bad enough to sleep in feathers in summer, when one lives far in the north; but tobe delivered over to such a fate in the latitude of 340 is deplorable. Every where in thisregion, the taverns, and, in general, the houses of the people, are furnished with featherbeds, forJune, July and August, not less than for January. The medical gentlemen of thiscountry should raise their voices against this absurd and enervating custom. A hard bedfor hot weather referring rather to health or comfort, should be the motto of the wholeSouth.

- Daniel Drake (1785-1852)

CMAJ, VOL. 140, MARCH 15, 1989 633