SPIFA—A presentation of the Structured Psychiatric Interview for General Practice

© 2009 Informa UK Ltd. (Informa Healthcare, Taylor & Francis AS) DOI: 10.3109/08039480902874769

SPIFA—A presentation of the Structured Psychiatric Interview for General Practice

ALV A. DAHL, MARIT BJARTVEIT KRÜGER, NILS HÅVARD DAHL, HASSE KARLSSON, LARS VON KNORRING, EYSTEIN STORDAL

Dahl AA, Bjartveit Krüger M, Håvard Dahl N, Karlsson H, von Knorring L, Stordal E. SPIFA—A presentation of the Structured Psychiatric Interview for General Practice. Nord J Psychiatry 2009;63:443–453.

Background: The diagnostic ability of general practitioners (GPs) concerning mental disorders is not optimal, and could be improved by structured diagnostic interviews. Various aspects of the Structured Psychiatric Interview for General Practice (SPIFA) are examined. Aims: The inter-rater reliability of the SPIFA, the time used by GPs and specialists and the GPs satisfaction are examined. The properties of the SPIFA are compared with those of the Prime-MD and the MINI schedules. Methods: Inter-rater reliability of the SPIFA was tested in 336 patients in general practice. The patients were randomized to two interview strategies. Either both GPs and psychiatrists used the SPIFA, or GPs used the SPIFA and psychiatrists a modifi ed version of the SCID for Axis I disorders. The satisfaction was investigated by a questionnaire sent to 1000 GPs who had SPIFA training. Results: The SPIFA showed adequate inter-rater reliability for depression, anxiety disorders and increased suicidal risk for both interview strategies. In patients with more than two co-morbid disorders, the inter-rater reliability was poor. The mean duration of SPIFA was 21 min for SPIFA screening and 22 min for SPIFA manual. The 192 GPs responding to the questionnaire were mostly satisfi ed with the SPIFA. Conclusions: The SPIFA seems to be a reliable, valid and helpful instrument for GPs making diagnoses of mental disorders in their patients. Compared with the Prime MD and the MINI, the SPIFA seemed to have comparable psychometric properties but better feasibility in primary care. • General practitioners, Inter-rater reliability, Mental disorders, Structured diagnostic interview.

Alv A. Dahl, M.D., Ph.D., Radiumhospitalet, Rikshospitalet HF, Montebello, 0310 Oslo, Norway, E-mail [email protected]; Accepted 6 March 2009.

Background INCEPTION OF THE SPIFA At the end of the 1980s, the Gotland study in Sweden ( 1, 2 ) reported a marked reduction of suicides associated with a psychiatric educational programme offered to gen-eral practitioners (GPs). The study inspired the idea of doing a suicide preventative intervention as a sub-study of the Health Study of Nord-Trøndelag County in Norway 1995–97 (HUNT 2). The Intervention study against Depression and Anxiety in Nord-Trøndelag (IDANT study) thus was conceived by the Norwegian authors in 1995. Since high levels of anxiety and/or depression was closely associated with risk of suicide ( 3, 4 ), the IDANT study was based on the ratings the Hospital Anxiety and Depression Scale (HADS) by the participants of HUNT-2 ( 5, 6 ). In case of an extremely high HADS score (HADS

total score ! 25) both the participant and his GP got a warning note approximately 6 weeks after the HUNT-2 examination, in the same way as they got for somatic risk factors like high cholesterol. Details on the IDANT study have been published elsewhere ( 7 ).

In 1995, the World Health Organization published its international Study of Mental Illness in General Health Care Disorders ( 8 ) showing that GPs were able to identify a mean of 49% (range 16–75%) of patients with current mental disorders in their practices. This fi nding as well as the close collaboration with GPs in Nord-Trøndelag County over years by members of the IDANT group (MBK, NHD and ES) made it clear that the IDANT study had to offer training in diagnosing and treating common mental disorders to the GPs participating in HUNT 2. Since a positive HADS screening should lead to a

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.

AA DAHL ET AL.

444 NORD J PSYCHIATRY VOL 63 · NO 6 · 2009

diagnostic exploration of mental disorders and suicidal risk, the IDANT group developed a structured diagnostic interview for mental disorders for GPs [Structured Psy-chiatric Interview for General Practice (“Allmennpraksis” in Norwegian (SPIFA)]. The SPIFA was introduced to the GPs at the HUNT-2 training courses that were given by NHD, MBK and ES. The GPs continuously gave feedback based on their experiences with the SPIFA that lead to modifi cations and improvements of the interview.

BASIC IDEAS AND STRUCTURE OF THE SPIFA Based on their close collaboration with GPs, the IDANT group realized that any diagnostic tool offered to the GPs had to fulfi l two basic requirements: brief duration and simple structure. The IDANT group reasoned that useof the obligatory diagnostic criteria for various disorders in DSM-IV as screening items would fulfi l both these requirements. If the questions concerning obligatory cri-teria for a disorder were answered negatively, that disor-der could be excluded from further examination. Such criteria could be used for a screening part of the SPIFA.In contrast, if any of the obligatory diagnostic criteria were rated as present, further interviewing had to be made in order to check the other criteria needed for a defi nite diagnosis. These criteria were outlined in the manualpart of the SPIFA. The fi rst edition (1995) of SPIFA thus contained a screening part with obligatory criteria for 14 mental disorders and three conditions (suicidality, cognitive impairment and personality disorders in general) and the manual part contained the criteria interviewed for in order to make defi nite diagnoses in the case of positive screening.

The next (1996) edition of the SPIFA covered 18 dis-orders and conditions ( Table 1 ). The obligatory criteria for major depressive episode (depressed mood, loss of interest or pleasure and attenuated emotional reactions) were based on three items of the Montgomery–Åsberg Depression Rating Scale (MADRS) ( 9 ). One to three obligatory diagnostic criteria were used to screen for bipo-lar disorder, panic disorder, agoraphobia, social phobia, generalized anxiety disorder, obsessive–compulsive dis-order and somatoform disorder, adjustment disorder, post-traumatic stress disorder, eating disorders, substance use disorders (except those concerning alcohol), functional psychoses and organic mental disorders. Suicidality was screened for by three items recommended by the National Health Services ( 10 ) and personality disorders in general with fi ve items from the IOWA personality screener ( 11 ). The screening item for cognitive impairment concerned any sign of confusion, attention loss or memory impair-ment identifi ed during the interview. Alcohol problems were pre-screened for by asking if the patient was “never” drinking alcohol vs. “sometimes” or “often”. Those belonging to the two latter categories were screened

further by four questions (“intoxicated several days in a row”, “skipping meals because of alcohol”, “change of behaviour or becoming more amenable after drinking” and “do you need an eye-opener in the morning”) found effec-tive in a Norwegian trial of the AUDIT instrument to identify men who consumed ! 40 g of alcohol per day ( 12 ).

Any diagnostic criterion scored positive in the short screening part of SPIFA, was further examined in the manual part , which contained the other diagnostic criteria for major depressive episode (by the other seven items of the MADRS), bipolar disorder, adjustment disorder, the anxiety disorders, alcohol/substance abuse and depen-dence, eating disorders (anorexia, bulimia), somatoform disorder (somatization disorder and undifferentiated somatoform disorder), functional psychoses and organic mental disorders based on additional DSM-IV criteria. A diagnosis of cognitive impairment was based on the Mini Mental Status ( 13 ), and suicidality on the responses to 12 further questions recommended by the National Health Services ( 10 ). Finally, SPIFA contain the split version of the Global Assessment of Functioning Scale (Symptoms and Function separately) for a global assessment of mental health ( 14 ).

Table 1. Content of the Norwegian Structured Psychiatric Interview for General Practice (SPIFA) 1996 and 2003 versions.

SPIFA short version SPIFA longer version

• Major depressive episode (MDE) • Adjustment disorder• Bipolar disorder • Post-traumatic stress

disorder (PTSD)• [Recurrent depressive disorder] • Eating disorders• [Dysthymia] Anorexia nervosa• Panic disorder Bulimia nervosa• Agoraphobia • Suicidality• Social phobia • Substance abuse and

dependence• Generalized anxiety disorder (GAD) • Functional psychosis• Obsessive–compulsive disorder

(OCD)• Organic mental disorder

• Somatoform disorders Somatization disorder

• Mental disorder related to general somatic condition

Undifferentiated somatoform • [Personality disorder]• Alcohol abuse and dependence• Cognitive impairment

Rating scales• Montgomery–Åsberg depression rating scale (MADRS)a

• Mini Mental Statusb [Mini Mental Status 12 items versionc

(MMS-12)]• [SAD PERSON Scaled]• [Global Assessment of Functioning—Symptoms (GAS-S)e and

Function (GAS-F)e]

[—] diagnoses and rating scales not used in the SPIFA-336 study.aMontgomery & Åsberg (9); bFolstein et al. (13); cBrækhus et al. (20); dPatterson et al. (19); ePedersen et al. (14).

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.

A PRESENTATION OF THE SPIFA INTERVIEW

NORD J PSYCHIATRY VOL 63 · NO 6 · 2009 445

Aims The GPs of Nord-Trøndelag County were mostly satisfi ed with the SPIFA, and their positive experiences were soon spread through collegial networks to GPs all over Norway. Demands for SPIFA training courses were raised outside Nord-Trøndelag County, and the IDANT group trained more psychiatrists as teachers for SPIFA courses. In 1996, the Primary Care of Evaluation of Mental Disorders (PRIME-MD) was introduced in Norway. Prime-MD had documented inter-rater reliability and feasibility ( 15 ). In this situation research-based documentation of the SPIFA as an alternative diagnostic tool for primary care became mandatory. The IDANT group, therefore, in collaboration with Novo Nordic Pharma Ltd that had sponsored the HUNT-2 study, planned two scientifi c studies: the SPIFA-336 reliability study and the SPIFA questionnaire study.

The SPIFA-336 study examined patients consulting GPs and had three aims: 1) to examine the inter-rater reli-ability of diagnoses of mental disorders based on SPIFA interviews of the same patients carried out by pairs of GPs and psychiatrists, 2) to study the inter-rater reliability of diagnoses of mental disorders of the same patients inter-viewed with SPIFA by the GPs and by a modifi ed version of the Structured Clinical Interview for DSM-IV Axis I (SCID) ( 16 ) made by psychiatrists, and 3) to compare the time used by GPs and psychiatrists on the SPIFA screen-ing and manual as well as on the SCID, and to explore how helpful the GPs found the SPIFA.

The aim of the SPIFA questionnaire study was in GPs who had been to SPIFA training courses to 4) explore their use and experiences with the SPIFA, and eventually to make a comparison with their experiences with the Prime-MD.

Methods The SPIFA-336 reliability study In 1997, 19 pairs consisting of a GP and a psychiatrist (except for one clinical psychologist) were established in order to interview patients from the caseloads of the GPs. In all, 336 patients were included, which implied 18 patients evaluated by most pairs. The inclusion criteria for the SPIFA-336 study were: 1) age ! 18 years; 2) able to explain his/her situation without the help of an assisting person; 3) giving written informed consent. The only exclusion criterion was the need of emergency psychiatric treatment. Each GP should select nine patients who spon-taneously complained about mental symptoms, seven patients who they suspected had mental problems, and two patients free from mental problems.

The GPs always interviewed their patients with the SPIFA, while the specialists interviewed half the patients with the SPIFA and the other half with the SCID ( 16 ). The selection of SPIFA vs. SCID interviews for the specialists

was randomized by Medstat Ltd based on a statistical pro-cedure. For the SCID interviews, no screening procedure was used, and only current diagnoses were scored. Thus lifetime diagnoses and sub-classifi cations of disorders according to the SCID were omitted. For suicidality and cognitive impairment, the SPIFA diagnostic procedures were used also for the SCID cases. If a major depressive episode was diagnosed by the SCID, the MADRS was also fi lled in.

The GPs and the psychiatrists fi nally had a common discussion of each patient, and the utility of this discus-sion was rated by both of them on a 5-point Likert scale from 0: “no use” to 4: “very useful”. Duration of the inter-views in minutes was noted. The patients’ experience of being interviewed on psychic problems was asked for by both GPs and specialist, and rated on a 5-point Liker scale from 0: “very negative” to 4: “very positive”.

The SPIFA questionnaire study In the autumn of 1997, a questionnaire was mailed to a random sample of 1000 Norwegian GPs who had attended SPIFA training courses since their introduction in 1996. The questionnaire was developed by the IDANT group and covered experiences with the SPIFA and an eventual comparison with their Prime-MD experiences. The responses were anonymous and no reminders were sent out.

Statistical issues Descriptive statistics were used in both the SPIFA-336 and the questionnaire study. In the SPIFA-336 study, inter-rater reliability was calculated with the kappa coef-fi cient statistic ( 17 ). Since the kappa coeffi cient is infl u-enced by the prevalence of disorders, the prevalence indexes (PIs) were calculated. When PIs are high (preva-lence of a disorder is either very high or very low), chance agreement is also high. Kappa is diffi cult to interpret meaningfully unless PIs is taken into account. Bias effects occur when two observers differ on the proportion of positive or negative results. We calculated the bias indexes (BIs), and when BIs are high, kappas are higher than when BIs are low. The most valid kappa values are observed when both PI and BI are low. Kappa is not reliable if the base rate of the disorders observed is " 5%.

Ethical concerns The SPIFA-336 study was approved by the Regional Committee of Ethics in Medical Research, and all patients gave written informed consent. The SPIFA questionnaire study was considered as a quality assur-ance study and no ethical approval or informed consent was applied for.

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.

AA DAHL ET AL.


INTER-RATER RELIABILITY IN RELATION TO THE

SPECIFIC DIAGNOSES

A total of 164 patients were evaluated by SPIFA by both GPs and psychiatrist. Alcohol and substance abuse and dependence, functional psychosis, eating disorders and cognitive impairment showed a base rate " 5% accord-ing to any rating of presence ( Table 3 ), and their kappa values should not be considered. The kappa values for the other disorders (major depression, bipolar disorder, the anxiety disorders, somatoform disorders and for sui-cidality) ranged from 0.41 to 0.77, except for adjustment disorders with a kappa of 0.29. The PIs was rather high for most of these disorders, while the BIs consistently were low.

A total of 167 patients were evaluated by SPIFA by the GPs and SCID by the psychiatrists. Bipolar disorder, substance abuse and dependence, functional psychosis, eating disorders and cognitive impairment showed a base rate " 5% according to any rating of positive diagnoses ( Table 3 ). The kappa values for the other disorders (major depression, the anxiety disorders, somatoform disorders, alcohol abuse and dependence, and suicidality) ranged from 0.30 to 0.74, with the exception of adjustment disorder with a kappa of 0.07.

The PIs was rather high for most of these disorders, while the BIs were consistently low.

Some of the kappa values of the SPIFA/SPIFA ratings were somewhat higher than the SPIFA/SCID ones, while others were also quite similar ( Table 3 ). Taken together the kappa values varied between 0.41 and 0.70 for the disorders with ! 5% base rate, the only exception was adjustment disorders (kappa 0.21). Even in the total sample, substance abuse and dependence, functional psychosis, cognitive impairment and the eating disorders had to low prevalence of meaningful interpretation of the kappa coeffi cients.

For SPIFA/SCID, SPIFA/SPIFA and total the BIs con-sistently were low, while the PIs mostly were high in the SPIA/SCID and SPIFA/SPIFA combinations and thus also in the total sample.

INTERVIEW TIME USE

The time use of SPIFA screening was calculated according to the number of disorders present, and the minutes used increased with the number of disorders found positive on screening ( Table 4 ). The same pattern was also found when the specialists used SPIFA screening. A pattern of increased time use with more disorders present was also found for both GPs and specialists concerning the SPIFA manual and for the specialists when they used the SCID.

The mean duration of SPIFA in all patients was 21 min for SPIFA screening and 22 min for SPIFA manual. The

Results The SPIFA-336 reliability study For the SPIFA/SPIFA combination, 169 patients were randomized, but for administrative reasons only 164 patients had a SPIFA interview with a specialist. All 167 patients randomized to the SPIFA/SCID combination had both their interviews.

INTER-RATER RELIABILITY IN RELATION TO THE

NUMBER OF DIAGNOSES

The diagnoses indicated by the SPIFA screening could be confi rmed or disconfi rmed by the manual for both GPs and specialist. However, SCID-based diagnoses were only given by specialists without any screening.

The number of diagnoses given by the GPs and specialist taken together varied from 0 to 10, and their distribution and kappa values are given in Table 2 . Altogether 843 diagnoses were given to the 331 patient (mean 2.5 diagnoses/patient), and for 439 (52%) the GPs and the specialist were in agreement; 167 (20%) diagno-ses based on the SPIFA manual or the SCID were made only by the specialists and 218 (28%) based on the SPIFA manual only by the GPs. For 336 screening diagnoses, both GPs and specialists were in agreement that no men-tal disorder was present when checked by the SPIFA manual or the SCID.

In 70 patients, both members of the diagnostic pairs agreed that no mental disorder was present, and the inter-rater kappa coeffi cient was 1.0 (perfect agree-ment). Thus the specifi city for both the SPIFA and the SCID was 100% in this sample of patients selected from general practice. The inter-rater reliability expressed as kappa values became lower as more positive diagnoses were made. For the SPIFA/SCID combination, the kappa value was satisfactory only for the presence of one disorder (0.67), while for the SPIFA/SPIFA combi-nation a good kappa value also was observed for two disorders (0.66). For more than two disorders, the kappa values for the SPIFA/SCID became very low and also negative, indicating an inter-rater reliability worse than by chance ( Table 2 ). For the SPIFA/SPIFA combina-tion, the kappa value was between 0.21 and 0.28 when three to seven disorders were present, mostly with low PI and BI values. These values thus indicate fair inter-rater reliability for this combination. The kappa values for the total sample are coloured by the poorer values of the SPIFA/SCID combination, however the overall kappa for the SPIFA combination was 0.44. The BI was low for all numbers of diagnoses, while the PI was for one to fi ve diagnoses, and medium (0.28–0.44) six or more diagnoses.

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.



Table 2. Inter-rater reliablity according to the number of positive Structured Psychiatric Interview for General Practice (SPIFA) manual or Structured Clinical Interview for DSM-IV Axis I (SCID-I) diagnoses.

SPIFA/SCID, n# 167 SPIFA/SPIFA (n# 164) Total (n# 331)

Kappa Kappa KappaPI_BIa PI_BIa PI_BIa

SPb SPb SPb

$% $% $%n G $|$$|$%| n G $|$$|$%| n G $|$$|$%|

No. of disorders (n diagnoses) P %|$%|%%| (n diagnoses) P %|$%|%%| (n diagnoses) P %|$%|%%|

No disorders 41 1.00 29 1.00 70 1.00(41) (29) (70)

1 disorder 24 0.67 34 0.74 58 0.71(24) 0.40_0.01 (34) 0.28_0.01 (58) 0.33_0.01

15| 5 24| 5 39|10 4|41 5|49 9|90

2 disorders 26 0.02 37 0.68 63 0.45(52) 0.13_0.04 (74) 0.08_0.00 (126) 0.10_0.01

15|20 52|11 67|3117|25 11|65 28|90

3 disorders 27 0.25 24 0.28 51 0.27(81) 0.21_0.04 (72) 0.06_0.05 (153) 0.12_0.06

44|23 35|21 79|4414|22 16|30 30|52

4 disorders 25 0.11 19 0.26 44 0.18(100) (76) 0.07_0.01 (176) 0.11_0.05

0.15_0.10 37|20 82|5345|33 19|30 41|56

5 disorders 14 !0.2822|26 11 0.27 25 0.01(70) 0.21_0.01 (55) 0.06_0.03 (125) 0.14_0.01

23|23 27|15 50|3824|7 13|22 37|29

6 disorders 5 !0.07 5 0.21 10 !0.05(30) 0.14_0.01 (30) 0.44_0.09 (60) 0.28_0.01

11|8 16|9 27|1711|6 5|2 16|8

7 disorders 4 !0.05 3 0.21 7 0.08(28) 0.37_0.30 (21) 0.11_0.04 (49) 0.24_0.17

13|12 10|6 23|183|2 5|7 8|9

8 disorders – 2 0.11 2 0.11(16) 0.44_0.06 (16) 0.44_0.06

10|4 10|4 3|2 3|2

9 disorders – – –

10 disorders 1 0.27 – 1 0.27(10) 0.33_0.17 (10) 0.33_0.17

6|3 6|31|2 1|2

Total 167 0.21 164 0.44 331 0.33(436) 0.18_0.06 (407) 0.06_0.06 (843) 0.17_0.06

228|127 211|91 439|218 95|129 72|207 167|336

aPI_BI, prevalence index_bias index.bClassifi cation by general practitioners (GP) and specialist in psychiatry (SP).

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.

AA DAHL ET AL.


Table 3. Correspondence of Structured Psychiatric Interview for General Practice (SPIFA) diagnoses by GPs and specialists in the SPIFA-336 study.

SPIFA/SCID SPIFA/SPIFA Total

n# 167 n# 164 n# 331SP SP SP

G +|+ +|! +| G +|+ +|! +| G +|+ +|! +|P !|+ !|! !| P !|+ !|! !| P !|+ !|! !|

Disorder Kappa PI_BIc Kappa PI_BIc Kappa PI_BIc

Adjustment disorder 1|11 5|14 6|255|150 5|140 10|290

0.07 0.89_0.04 0.29 0.81_0.05 0.21 0.85_0.05PTSD 5|4 4|6 9|10

2|156 4|150 6|3060.61 0.90_0.01 0.41 0.89_0.01 0.41 0.89_0.01

Major depressionb MADRS & 15 55|8 53|9 108|1716|88 14|88 30|176

0.70 0.20_0.05 0.71 0.34_0.03 0.70 0.19_0.04Bipolar disorder 2|3 7|2 9|5

1|161 2|153 3|3140.49 0.95_0.01 0.77 0.90_0.00 0.68 0.92_0.01

Panic disorder 13|17 14|10 27|27 9|128 6|134 15|262

0.41 0.69_0.05 0.58 0.73_0.02 0.49 0.71_0.04Agoraphobia 16|16 16|9 32|25

12|123 15|124 27|2470.43 0.65_0.02 0.48 0.64_0.04 0.46 0.65_0.01

Social phobia 24|14 24|2 48|1611|118 9|129 20|247

0.56 0.56_0.02 0.77 0.64_0.04 0.66 0.60_0.01Generalized anxiety disorder 21|41 35|15 56|56

7|98 14|100 21|1980.31 0.46_0.20 0.58 0.40_0.04 0.44 0.43_0.11

Obsessive–compulsive disorder 7|2 7|5 14|75|153 4|148 9|301

0.65 0.87_0.02 0.58 0.86_0.01 0.61 0.87_0.01Somatoform disorder 9|5 11|10 20|15

6|147 6|137 12|2840.59 0.83_0.01 0.52 0.77_0.02 0.55 0.80_0.01

Alcohol abuse 3|8 5|1 8|94|152 1|157 5|309

0.30 0.89_0.02 0.83 0.93_0.00 0.51 0.90_0.01Alcohol dependence 6|4 5|0 11|4

6|151 1|158 7|3090.51 0.87_0.01 0.91 0.93_0.01 0.65 0.90_0.01

Substance abuse 1|1 4|0 5|12|163 0|160 2|323

0.39 0.87_0.01 1.00 0.95_0.01 0.77 0.95_0.01Substance dependence 3|2 3|0 6|2

0|162 1|160 1|3220.74 0.95_0.01 0.85 0.96_0.01 0.80 0.95_0.01

Suicidal behaviour 4|2 6|3 10|53|158 2|153 5|311

0.60 0.92_0.01 0.69 0.90_0.01 0.65 0.91_0.00Functional psychosis 2|1 1|1 3|2

1|163 1|161 2|3240.66 0.96_0.00 0.49 0.98_0.00 0.59 0.97_0.00

Cognitive impairment 0|0 0|0 0|06|161 1|163 7|324

0.00 0.96_0.04 0.00 0.99_0.01 0.00 0.97_0.01Anorexia nervosa 0|1 0|1 0|2

0|166 0|163 0|3290.00 0.99_0.01 0.00 0.99_0.01 0.00 0.99_0.01

Bulimia nervosa 0|0 1|0 1|01|166 1|162 2|328

0.00 0.99_0.01 0.66 0.98_0.01 0.50 0.98_0.01

aClassifi cation by general practitioners (GP) and specialist in psychiatry (SP).bOnly scored for the patients who fulfi lled the MADRS screening criteria in the SPIFA.cPI, prevalence index; BI, bias index (Sim et al., 2005).

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.



Table 4. Time use in minutes for Structured Psychiatric Interview for General Practice (SPIFA) screening and manual and Structured Clinical Interview for DSM-IV Axis I (SCID) interviews according to the number of diagnoses.

SPIFA screening SPIFA manual SCID

GPs Specialists GPs Specialists Specialists

(n# 336) (n# 164) (n# 336) (n# 164) (n# 167)

No. of disorders Mean (s) Mean (s) Mean (s) Mean (s) Mean (s)

0 16.1 (7.4) 17.8 (7.3) 7.8 (9.2) 5.2 (5.7) 34.8 (14.0) 1 18.5 (8.0) 17.3 (7.1) 13.8 (7.7) 16.0 (11.3) 50.8 (13.5) 2 20.7 (10.8) 23.4 (11.4) 16.9 (8.5) 20.6 (10.5) 50.7 (17.9) 3 23.4 (8.5) 21.0 (7.9) 19.2 (10.5) 23.7 (10.8) 58.2 (14.6) 4 21.5 (10.4) 22.2 (8.3) 20.9 (8.0) 26.7 (15.8) 59.2 (16.8) 5 20.4 (7.0) 20.7 (9.5) 23.9 (20.4) 38.3 (20.2) 55.0 (8.7) 6 24.7 (11.8) 26.4 (13.3) 23.8 (7.5) 19.0 (9.6) 51.7 (7.6) 7 22.4 (8.5) 31.9 (16.2) 26.7 (7.8) 37.5 (31.8) 54.3 (8.1) 8 32.5 (13.2) 34.4 (17.7) 33.3 (23.6) 39.3 (29.7) –10 20 – 30 – 65Total 21.0 (9.6) 21.7 (10.0) 15.4 (10.8) 17.6 (13.4) 48.5 (17.1)

s, standard deviation.

corresponding time use for the specialist using SPIFA was 15 and 18 min, respectively. The specialists used a mean of 49 min on a SCID interview ( Table 4 ).

SATISFACTION WITH THE SPIFA AND THE

REACTIONS TO INTERVIEWS BY PATIENTS

In a majority (85%), the cases the GPs found the SPIFA to be at least “moderately” or more useful ( Fig. 1 ). Very few patients had a distinctly negative attitude to the interviews about psychiatric matters and 50% had a neutral attitude ( Fig. 2 ). Among the GPs, 61% found the case discussions with the specialist at least “moderately helpful”, while that proportion was 72% among the specialists ( Fig. 3 ).

The SPIFA questionnaire study Among the 1000 GPs invited, 192 (19.2%) responded with valid questionnaires ( Table 5 ). Of the respondents, 59% reported regular use of SPIFA, and 83% considered

the need for a diagnostic tool like SPIFA from “quite large” to “very large”. The same proportion (83%) reported that SPIFA met their expectations to the extent from “quite much” to “very much”. Quite the same proportion (86%) felt that SPIFA fulfi lled their expectations among those (33%) who also had tried out the Prime-MD ( 15 ). The time use for SPIFA-1997 screening and manual reported, corresponded well to the time use reported by GPs in the SPIFA-336 study.

Discussion The SPIFA-336 studies showed that the SPIFA used by GPs and specialist in the same primary care patients showed moderate to good inter-rater reliability for com-mon mental disorders like depression and the anxiety disorders as well as for evaluation of suicidality. When GPs used the SPIFA and specialists the SCID, the fi ndings were essentially similar. The mean duration of SPIFA in all patients was 21 min for SPIFA screening and 22 min

Fig. 1. Usefulness of the Structured Psychiatric Interview for General Practice (SPIFA) according to the GPs based on 332 patients of the SPIFA-336 study.

Fig. 2. Patients reactions to being interviewed with the Structured Psychiatric Interview for General Practice (SPIFA) or the Structured Clinical Interview for DSM-IV Axis I (SCID).

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.

AA DAHL ET AL.


in cases of co-morbidity. A considerable proportion of the patients in the SPIFA-336 study had co-morbid disorders. The explanation could be that the GPs considered the study as an opportunity to get complicated patients evalu-ated by a specialist for free, and that they therefore over-sampled such patients compared with their usual caseload. Identifi cation of co-morbidity may be confusing for the GPs, but mostly will concern various depressions and anxiety disorders, which will hardly affect the choice of treatment.

The PRIME-MD had been developed with the same purpose as the SPIFA namely to improve the diagnostic accuracy of GPs. The PRIME-MD operates with a screen-ing by a 27-item questionnaire fi lled in by the patient cov-ering symptoms for fi ve depressive disorders, panic disorder, agoraphobia and anxiety disorder related to somatic illness or medication, alcohol abuse/dependence, eating disorders and somatoform disorders ( 15 ). Positive symptoms on the questionnaire are examined further by the GP doing a structured interview on the relevant disor-ders. Our development of the SPIFA was well under way, when the Prime-MD became known to us. We did not endorse the idea of a questionnaire since we preferred the direct contact between patient and doctor when symptoms are asked for. We also found that the focus of the Prime-MD was too much on depression, and that the total num-ber of mental disorders covered was insuffi cient. Development of the SPIFA was done in very close contact with GPs in Nord-Trøndelag County, so our instrument was well tuned in to the way of thinking common to these GPs. These were the main reasons for promoting the development of the SPIFA rather than supporting the use of the Prime-MD.

Among the PRIME-MD patients, 43% had a second SCID interview by telephone by two specialists, while 99% of our patients had a second face to face interview with altogether 19 specialists. The kappa for major depres-sion was 0.61 in the PRIME-MD/SCID examination and 0.70 MADRS ' 15/ MADRS ' 15 in our study. The corresponding kappa values for panic disorder were 0.60 and 0.41 and for generalized anxiety disorder 0.52 and 0.31 in the PRIME-MD/SCID vs. SPIFA/SCID combina-tions, respectively. In the PRIME-MD study, the GPs found the approach to be “very” or “somewhat” valuable in 61% of the cases, while such a consideration was given by 85% in the SPIFA-336 study. Summing up, hardly any major differences were found between the fi ndings of the PRIME-MD and the SPIFA-336 study in our view, except that the latter study goes more into detail concerning co-morbidity and time use.

In order to get a perspective on the SPIFA-336 fi nd-ings, we compared the concordance of our SPIFA/SCID diagnoses with those of the Mini-International Neuro-psychiatric Interview (MINI) combined with the SCID

for SPIFA manual. The time used by the GPs and the specialists on these procedures was about the same. In 50% of the interviews the GPs found the SPIFA to be “very useful” or “clearly useful” and in 35% “moderately useful”. The SPIFA Questionnaire study showed that the responding GPs were mostly satisfi ed with SPIFA and that their expectations were met, even if they had tried out the Prime-MD.

In the SPIFA-336 study, we investigated the infl uence of co-morbidity on the functioning of the SPIFA. As could be expected, the inter-rater reliability was reduced when the patients had more than two co-morbid disorders com-pared with when they had fewer or no disorders. The time use of the SPIFA screening and manual also increased

Fig. 3. The usefulness of the case discussions as rate by the GPs and the specialists.

Table 5. Responses to the Structured Psychiatric Interview for General Practice (SPIFA) Questionnaire study (n# 192).

Do you use SPIFA? Yes 114 (59%), No 78 (41%)How large is the need for a tool like SPIFA? Very large 13%, large 34%, quite large 36%, limited 15%, very

limited 2%How did you learn to use SPIFA? SPIFA course 76%, self-learning 8%, from a colleague 4%, other

ways (unspecifi ed) 12%Did SPIFA meet your expectations? Very much 7%, much 43%, quite much 33%, limited 15%, very

limited 2%How many minutes do you use on a SPIFA screening?

" 10 min 22%, 10–15 min 17%, 16–20 min 30%, 21–25 min 7%, 26–30 min 19%, ! 35 min 5%How many minutes do you use on the SPIFA manual?

" 10 min 13%, 10–15 min 21%, 16–20 min 21%, 21–25 min 8%, 26–30 min 22%, ! 35 min 15%Is there a need for a second SPIFA course? Yes 52%, no 23%, do not know 25%How many of the SPIFA users (n # 114) want another SPIFA course? Yes 75%Have you used the Prime-MD before? Yes 33%, no 67%Did SPIFA meet your expectations if you had used Prime-MD?(n # 63) Very much 15%, much 47%, quite much 26%, limited 12%, very

limited 0%

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.



way interfere with the planning and design of these stud-ies or the development of the SPIFA as carried out by the IDANT group.

Further development of the SPIFA The Norwegian version of the SPIFA had a fi nal modifi ca-tion in 2003 based on the experiences from the scientifi c studies, feedback from the GPs and experiences from Sweden and Finland. The suicide items of the SPIFA manual based on the National Health Services items were replaced by the SAD PERSONS items used in the Swed-ish version of the SPIFA ( 19 ). The 30-item version of the MMS was substituted with the 12-item version of Brækhus et al. ( 20 ). In addition, recurrent depressive disorder and dysthymia based on the recommendation of the National Institute of Mental Health ( 21 ) were added. Most impor-tantly the SPIFA was divided into a short version concern-ing mental disorders that are common in primary care, and a longer supplementary version including the disorders that are rare in that setting ( Table 1 ).

The Swedish SPIFA version In 1997, SPIFA was modifi ed and translated into Swedish by Lars von Knorring, Lisa Ekselius, Mia Ramklint and Karin Malmberg at the Department of Neuroscience, Psy-chiatry, Uppsala University. In the printed manual, SPIFA was combined with rating scales to measure changes in psychopathology, i.e. the Montgomery–Åsberg Depres-sion Rating Scale (MADRS) ( 9 ), the Clinical Anxiety Scale (CAS) ( 22 ), the Mini Mental State (MMS) ( 13 ) and the Global Assessment of Functioning (GAF) ( 14 ). To assess the suicidal behaviour, the SAD PERSONS scale was included ( 19 ). Although, the SPIFA is a diagnostic instrument and not a treatment guideline, the GPs were advised to consult the treatment guidelines presented by the Swedish Psychiatric Association. During the following years, several hundred GPs were educated and trained in the use of the SPIFA, and the instrument was well accepted by the GPs.

The SPIFA was also included in a study where the in- and outpatients at the Uppsala University Hospital as well as the patients visiting the GPs in Uppland were screened for depression ( 23 24 ). In a separate methodolog-ical study, 127 patients were interviewed by means of the SPIFA, they were rated by means of the MADRS, they completed the self-rating version of the MADRS and they completed a new diagnostic self-rating scale for anxiety and depression (DSRS) ( 25 ), based upon the Comprehen-sive Psychopathological Rating Scale (CPRS). The kappa between the SPIFA interview and the DSRS self-rating as concerns the presence of a depressive disorder was 0.59.

Thus, the Swedish version of the SPIFA has been easy to use, it has been well accepted by the GPs, it seems to

diagnoses published by Sheehan et al. ( 18 ). It should be noted however that the MINI/SCID diagnoses were based on a sample ( n# 370) from the psychiatric specialist services with a much higher base rate of mental disorders than found in general practice. The kappa values for the MINI/SCID vs. SPIFA/ SCID were similar for social pho-bia (0.51 vs. 0.56), OCD (0.63 vs. 0.65) and alcohol dependence (0.46 vs. 0.51), while the MINI/SCID had higher kappa values for PTSD (0.78 vs. 0.61), panic dis-order (0.76 vs. 0.41), agoraphobia (0.67 vs. 0.43) and GAD (0.70 vs. 0.31). Since the MINI/SCID had much higher base-rates of these disorders and since the study was carried in the specialist services, these fi ndings are not surprising, and do not invalidate the kappa values of the SPIFA/SCID study. Interestingly, the time use for the MINI was 18.7 ( ( standard deviation, s# 11.6) min, which was very close to that used by our GPs and special-ists on the SPIFA manual (15.4 ( 10.8 and 17.6 ( 13.4,respectively]. The specialists in the MINI study used a mean 43.0 ( 30.6 min on the SCID while our specialists used a mean of 48.5 ( 17.1 min. Although with somewhat lower kappa values for some disorders in the SPIFA-336 study, most probably explained by the higher base-rates in the specialist settings, the fi ndings of the SPIFA study were in accordance with the MINI ones. Besides the dif-ferent settings, a major difference between the studies, however, was that the SPIFA-336 study examined the infl uence of co-morbidity, while no such analyses were done in the MINI study.

To our knowledge, no follow-up studies of the experi-ences of the GPs have been published concerning the PRIME-MD study, as we have done in the SPIFA ques-tionnaire study. With only 19% response rate, there is obvious risk of selection bias. Probably, the responders are those being positive towards the SPIFA. Thus, the fi ndings of the SPIFA questionnaire study have to be taken cautiously.

As stated, brief duration of the evaluation is an essential demand in primary care, and that is secured through ascreening procedure for the mental disorders. Screening in the SPIFA consists of the GPs asking for key symptoms necessary for a diagnosis, while in the Prime-MD, the patients complete a questionnaire that is handed over to the GP. The MINI contains no screening, and it is therefore less applicable to primary care. In our view, screening by ques-tionnaire does not give the GPs the same information and contact with the patients as direct questioning. Question-naire responses often must be validated by the GPs in order to secure that the patients have understood the items cor-rectly. So we fi nd that the SPIFA approach best fulfi ls the demand of brief time use in getting essential information.

Although Novo Nordic Pharma Norway fully fi nanced the SPIFA the SPIFA-336 reliability study and the SPIFA questionnaire study, the company did not in any

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.

AA DAHL ET AL.


References 1. Rutz W, von Knorring L, Wålinder J. Frequency of suicide on

Gotland after systematic postgraduate education of general practi-tioners. Acta Psychiatr Scand 1989; 80: 151– 4.

2. Rihmer Z, Rutz W, Pihlgren H. Depression and suicide on Gotland. An intensive study of all suicides before and after a depression-training programme for general practitioners. J Affect Disord 1995; 35: 147– 2.

3. Bronisch T, Wittchen HU. Suicidal ideation and suicide attempts: Comorbidity with depression, anxiety disorders, and substance abuse disorder. Eur Arch Psychiatry Clin Neurosci 1994; 244: 93– 8.

4. Stordal E, Morken G, Mykletun A, Neckelmann D, Dahl AA. Monthly variation in prevalences of anxiety disorder and depression in the general population at 63–65° North: The HUNT Study. J Affect Dis 2008; 106: 273– 8.

5. Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the Hospital Anxiety and Depression Scale. An updated literature review. J Psychosom Res 2002; 52: 69– 77.

6. Mykletun A, Stordal E, Dahl AA. The Hospital Anxiety and Depres-sion Scale (HAD): Factor structure, item analyses, and internal con-sistency in a large population. Br J Psychiatry 2001; 179: 540– 4.

7. Bjerkeset O, Dahl AA, Stordal E, Dahl NH, Bjartveit Krüger M, Linaker O. Feasibility of mental health screening and intervention in the HUNT population study. Soc Psychiatry Psychiatr Epidemiol 2006; 41: 191– 8.

8. Üstün TB, Sartorius N, editors. Mental illness in general health care . An international study . Chichester : Wiley , 1995 .

9. Montgomery S, Åsberg M. A new depression scale designed to be sensitive to change. Br J Psychiatry 1979; 134: 382– 9.

10. NHS Health Advisory Service . Suicide prevention: The challenge confronted. A manual of guidance for the purchasers and providers of mental health care . London : Her Majesty’s Stationery Offi ce ; 1994 .

11. Langbehn DR, Pfohl BM, Reynolds S, Clark LA, Battaglia M, Bellodi L, et al. The Iowa Personality Disorder Screen. Develop-ment and preliminary validation of a brief screening interview. J Person Dis 1999; 13: 75– 89.

12. Aasland OG, Amundsen A, Bovim G, Fauske S, Mørland J. Identifi kasjon av pasienter med risiko for alkoholbetinget skade [Identifi cation of patients at risk of alcohol related damage]. Tidsskr NorLaegeforen 1990; 110: 1523– 7.

13. Folstein MF, Folstein SE, McHugh PR. Mini-Mental State”: A practical method for grading the cognitive state of patients for clinicians. J Psychiatr Res 1975; 12: 189– 98.

14. Pedersen G, Hagtvedt KA, Karterud S. Generalizability studies of the Global Assessment of Functioning—split version. Comprehen Psychiatr 2007; 48: 88– 94.

15. Spitzer RL, Williams JBW, Kroenke K, Linzer M, Verloin deGruy III F, Hahn SR, et al. Utility of a new procedure for diagnosing mental disorders in primary care. The Prime-MD 1000 study. JAMA 1994; 272: 1749– 56.

16. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clin-ical Interview for DSM-IV Axis I Disorders. New York: Biometrics Research Department, New York State Psychiatric Institute; 1995.

17. Sim J, Wright CC. The kappa statistic in reliability studies: Use, interpretation, and sample size requirements. Phys Ther 2005; 85: 257– 68.

18. Sheehan DV, Lecrubier Y, Harnett Sheehan K, Janavs J, Weiller E, Keskiner A, et al. The validity of the Mini International Neuropsy-chiatric Interview (MINI) according to the SCID-P and its reliabil-ity. Eur Psychiatry 1997; 12: 232– 41.

19. Patterson WM, Dohn HH, Bird J, Patterson GA. Evaluation of suicidal patients: The SAD PERSONS scale. Psychosomatics 1983; 343– 5 : 348– 9.

have some validity and a reasonable agreement with other diagnostic instruments.

The Finnish SPIFA version In 1999–2000, SPIFA was translated into Finnish by Patrik Järvinen and Hasse Karlsson at the Department of Psychiatry, University of Turku. In Finland, about 2500 GPs and 500 psychiatrists were trained to use the SPIFA. There was a core group of 26 psychiatrists and one GP who fi rst were educated by Hasse Karlson, and who then started to conduct training courses for the GPs. These training courses took place between years 2000 and 2006. The related treatment guidelines (BAS) were translated year in year 2004 and after that the training for GPs included both SPIFA and BAS.

A small questionnaire was also distributed to about 30 GPs approximately 1 year after the training asking them about their experiences in using SPIFA. A clear majority of the GPs were satisfi ed with the training and they thought they had learned a lot about modern structured psychiatric diagnostics. However, only a minority of the GPs used SPIFA in their everyday work.

Conclusion A major aim of the IDANT group was fulfi lled namely to develop a tool that got the interest of the GPs and was considered useful in their daily work. A spin-off was that many GPs learned to question about key diagnostic criteria for several common mental disorders, thereby sharpening their diagnostic acumen compared with before they were trained in SPIFA. Although many GPs put SPIFA away after a while, a carry-over effect of improved repertoire for making more precise diagnoses of mental disorders could be a lasting effect of the SPIFA experience.

The SPIFA-336 study showed that the SPIFA had adequate inter-rater reliability for mental disorders com-monly seen in general practice. The time used for SPIFA screening and manual was within the time frame for a mental health consultation of GPs, and the SPIFA was mostly considered as a useful instrument for identifi cation of mental disorders in primary care.

Acknowledgements — The SPIFA-336 study was fully fi nanced by Novo Nordic Pharma Norway Ltd. However, the company did not in any way interfere with the development of the SPIFA or the design of the SPIFA-336 reliability study or the SPIFA questionnaire study as planned by the IDANT group. Medstat Research Ltd did initial quality control and pre-liminary data analyses. The IDANT Ltd has for the period 1999 to 2008 received an annual leasing fee for the SPIFA fi rst from SmithKline Beecham Scandinavia Ltd and from 2001 from Glaxo SmithKline, Norway. From 1999 to 2006 the shareholders of IDANT Ltd were M Bjartveit Krüger, AA Dahl, NH Dahl and E Stordal, and from 2006 and onwards Dahl, Dahl and Stordal.

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.



20. Brækhus A, Laake K, Engedal K. The Mini-Mental State Examina-tion: Identifying the most effi cient variables for detecting cognitive impairment in the elderly. J Am Geriatr Soc 1992; 40: 1139– 45.

21. Gwirtsman HE, Blehar MC, McCullough JP, Jr, Kocsis JH, Prien RF. Standardized assessment of dysthymia: Report of a National Institute of Mental Health conference. Psychopharm Bull 1997; 33: 3– 11.

22. Snaith RP, Baugh SJ, Clayden AD, Husain A, Sipple MA. The clinical anxiety scale: An instrument derived from the Hamilton anxiety scale. Br J Psychiatry 1982; 141: 518– 23.

23. Stålenheim G, von Knorring L, Penayo U, Bakall D, Dyster-Aas J, Eriksson T, et al. Odiagnostiserade depressioner fi nns överallt i vården. Somatisk öppen och sluten vård och primärvård i Uppland screenad (Undiagnosed depressions are everywhere in the world. Somatic outpatient and inpatient care and primary health care in Uppland screened). Läkartidningen 2003; 100: 2760– 3.

24. von Knorring AL, Cederberg-Byström K, Nyberg AM, Hedlund M, Stålenheim G, von Knorring L.. Depressioner vanligt hos unga som söker för kroppsliga besvär. Trötthet, minskat känsloengagemang,ökat sömnbehov varningssignaler (Depression common among young people with somatic disorders. Fatigue, lack of emotional

engagement, increased need of sleep are some of the warning signals). Läkartidningen 2004; 101: 365– 8.

25. Svanborg P, Åsberg M. A new self-rating scale for depression and anxiety states based on the Comprehensive Psychopathological Rating Scale. Acta Psychiatr Scand 1994; 89: 21– 8.

Alv A. Dahl, M.D., Ph.D., Department of Clinical Cancer Research,The Norwegian Radiumhospital, Rikshospitalet University Hospital,Oslo, and Faculty Division, The Norwegian Radiumhospital,University of Oslo, Oslo, Norway.Marit Bjartveit Kruger, M.D., Department of Psychiatry, UllevaalUniversity Hospital, Oslo, Norway.Nils Håvard Dahl, M.D., Department of Psychiatry, HospitalLevanger, Levanger, Norway.Hasse Karlsson, M.D., Ph.D., Department of Psychiatry, University of Helsinki, Helsinki, Finland.Lars von Knorring, M.D., Ph.D., Department of Neuroscience,Psychiatry, Uppsala University, Uppsala, Sweden.Eystein Stordal, M.D., Ph.D., Department of Psychiatry, HospitalNamsos, Namsos, Norway, Department of Neuroscience, NationalUniversity for Science and Technology, Trondheim, Norway.

Nor

d J P

sych

iatry

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity O

f Upp

sala

on

03/0

9/12

For p

erso

nal u

se o

nly.

SPIFA—A presentation of the Structured Psychiatric Interview for General Practice

Documents