Effect of General Practitioner Training in a Collaborative Child Mental Health Care Program on Children’s Mental Health Outcomes in a Low-Resource Setting

Key Points Question Is it possible to improve child mental health by adding a child and youth component to an adult-focused integrated care program in a middle-income country? Findings In this cluster randomized trial with 49 general practitioners (GPs) caring for 389 children and youths, child mental health training for GPs increased children’s receipt of mental health care (GPs more likely to counsel, parents more likely to report that their children had received mental health care) compared with GPs instructed only to refer. Across all GPs, children’s mental health improved similarly regardless of training; however, among GPs whose patients were predominantly children, training significantly improved child mental health status compared with instruction to refer. Meaning These findings suggest that it may be feasible to expand adult-focused GP collaborative care models to include children and youth; however, GPs with more experience with children may be best able to take advantage of a brief training that expands their mental health role.


152
TUCCP also regularly calls a 5% sample of enrolled patients to ask them about satisfaction with 153 the services they have received from GPs and their receptionists.

155
Physicians participating in the TUCCP are advised to refer identified children/youth with mental health problems to CMHCs for further care. In contrast to the volume of adults, only 192 child/youth patients were enrolled in TUCCP's first 30 months, substantially fewer than expected, and only 67% completed referral to the CMHC. TUCCP leaders attributed this underutilization in large part to the mandatory referral of child cases to CMHCs, which constituted a 161 disincentive for both parents and GPs. In preparation for this project, TUCCP surveyed a 162 sample (n=27) of participating GPs to explore their interest in additional child mental health 163 training; 78% said they would likely participate. GPs believed that a majority of parents would 164 welcome alternatives to referral to CMHCs. They said that parents declined CMHC referrals 165 because of distance, stigma, and concern that specialists would propose medication rather than 166 psychotherapy. Over 75% of patients GPs referred had conditions (ADHD and anxiety) that 167 could be treated in primary care.

169
Study procedures 170 Design, sequence and timing of procedures: In this phase of the work we will conduct a hybrid effectiveness-implementation trial (Curran 2012) to develop preliminary evidence of the feasibility of training GPs to address child/youth mental health problems and mechanisms leading to training uptake and outcomes. The trial is a staggered-start parallel design in which the recruited GPs will be randomly assigned to one of three groups, and then within these groups assigned to either intervention or control status. Intervention GPs will receive the child 176 mental health training that is the subject of the study; control GPs will receive a child mental 177 health "booster" (focusing on identification and referral) as an attention control. At the end of each group's study period the control GPs will have the opportunity to take the training received by the intervention arm. In the intervention arm, after training, if GPs believe a child/youth has a mental health problem, the GP will have the option to offer a) treatment him or herself, b) 181 treatment in the GP office but with a referral for a one-time evaluation or c) referral to the CMHC 182 as is currently the standard. For all cases in which the GP believes that the child/youth has a 183 mental health problem, whether in the intervention or control group, office receptionists will 184 register the child with the TUCCP, stay in contact with the family, re-enforce the GP's advice 185 (whether that involves an intervention directly prescribed by the GP or involves a referral to the 186 CMHC), and check on follow-through. All of these steps are routine parts of TUCCP care and 187 are documented in the TUCCPs routine on-line data system.

189
To examine the impact of the intervention, in the first 1-2 weeks following training, cohorts of 190 children 5-15 years old will be recruited and screened using the parent report version of the 191 Strengths and Difficulties Questionnaire (SDQ) that corresponds to the age of their child. The

192
SDQ has been validated in Iran and yields measures of symptoms, functioning, and family 193 burden plus overall scores and domain scores for hyperactivity and emotional, peer relationship, 194 and conduct problems. In studies in other countries, the SDQ has been found to have similar or 195 better sensitivity and specificity as the much longer CBCL (Goodman 1999). Screening results

196
will not be shared with the GP, and summary scores will not be discussed with parents. Parents 197 and children/youth will be free to discuss the screen with the GP or anyone else, and RAs 198 administering the screens will encourage parents to discuss any concerns they have with their 199 GP. Those children/youth who screen positive will be followed by telephone with re-assessment 200 at 3 and 6 months after screening (regardless of whether or not the GP has considered them to 201 have a mental health problem and entered them in the TUCCP registry).

203
The following figure is the currently planned timeline for the study, assuming recruiting of GPs in   years of age (male and female) and one of their parents or guardians (whoever accompanies 232 the child/youth to their GP visit). These children will generally be coming to the GP for a non-233 emergent medical, emotional, or behavioral concern. Some children may have a chronic 234 medical or mental health condition, but we will not recruit any child/youth who reports being in

235
pain or who appears to be acutely ill. Children/youth who are actively being treated at the 236 CMHC at the time of recruiting will also not be eligible. We plan to recruit approximately 1500 237 children/youth and 1500 parents (screening step) and follow about 350 of each (children who 238 are SDQ+ and their parent) by telephone at 3 months and 6 months after the index visit.

240
The required patient sample size was estimated in four stages. We first used the "change"

255
In cluster-randomized studies, sample size is dependent on the extent to which outcomes are There are varying estimates of the prevalence of high/abnormal SDQ scores (=>17). We found 24% of those screened to be in this range in a study of children 6-16 in primary care in the US 266 (Wissow 2008); Alavi and colleagues (2010) found a prevalence of 26% using this scoring of the then need to screen about 1479 children/youth.

270
For screening, we will recruit consecutive patients as they come for clinic visits. Families will be 271 approached by a research assistant when they arrive for visits and asked if they are interested.

272
Families will not be approached if the child appears to be acutely ill or in pain. If the family is 273 interested, they will be taken to a nearby private space to discuss and consider consent/assent.

274
We anticipate that most families will complete baseline data collection by telephone to avoid

291
GPs will be approached by staff from the CMHC with whom they already interact regularly. If 292 they are interested in the study, one of the Iranian investigators will contact them, provide more 293 information and obtain written consent. GPs will receive a payment of about $200 to cover the 294 time that will be required for training in child/youth mental health skills. In addition, as in the 295 current Tehran University Collaborative Care Program (TUCCP), they will receive a care 296 coordination payment for patients identified as having a mental health problem who they 297 successfully follow for at least three months (presently this applies only to adults -for 298 participants in this trial it will apply children as well). Successful follow-up is defined as a)

299
monthly visits with the GP for the first three months after treatment initiation or substantial change in treatment, and then b) follow-up visits at no more than 3-month intervals for patients 301 who are stable. Patients for whom the GP is not certain of the diagnosis, or a) for whom there is 302 concern for self-harm, harm to others, psychosis, mania, or a clinically important drop in 303 function, or b) who cannot be stabilized after three months or may be having severe side effects 304 to medications, are referred to the CMHC.

306
Data collection

307
Baseline data collection will take place at the time of enrollment or by telephone after the GP 308 visit. Cognitive interviews and pilot testing will help assure that respondent burden is acceptable and achievable in a reasonably brief telephone call; should the plan proposed in this application require changes we will submit amendments to the protocol. The investigators have experience using the SDQ by telephone in Iran, and the GHQ (see below) by telephone in the US. The research assistant administering the instruments will enter the responses into a secure 313 electronic data base that will a) determine if the family is to be entered into the follow-up phase 314 of the study (ie, child "positive" at the "possible" or "probable" level on the SDQ) or if responses 315 to the parent self-harm question on the GHQ triggers notification of the GP. There will also be a 316 brief exit survey for the parent and for older children/youth after the index visit. This will be 317 administered by the research assistant in person at the GP office; however, as above, if lack of 318 privacy or time do not permit, the survey can be administered by telephone.

320
The GHQ (General Health Questionnaire) is a brief instrument (28 items) used to rate adult

328
Parents will be asked to completed the EuroQol EQ-5D-5L, a brief (6 item) instrument that 329 measures five areas of function (mobility, self-care, usual activities, pain/discomfort, 330 anxiety/depression) at five levels. It will offer us the opportunity to understand the impact of 331 parental disability which may or may not include mental health concerns disclosed on the GHQ.

332
The EQ-5D-5L can be administered by telephone and is sensitive to a wide range of chronic

340
Parent satisfaction with care will be assessed by asking parents if the visit involved discussion 341 of a mental health problem, and, if so, if the GP suggested a plan for treatment (Brown 2008).

342
Subsequently we will ask about satisfaction using questions initially developed by Zastowny and  Parent about self and child Baseline and 6 months later if SDQ+ * note that all completed by telephone except some baseline instruments may be completed in-Future studies may be able to address the costs and benefits of routine screening being part of the support offered by participation in TUCCP.
Child/parent follow-up data collection. An abbreviated data set (the key mental health outcome 400 measures -SDQ and GHQ) will be collected by telephone three months after the initial visit for 401 both the control and intervention children/parents; a longer final data collection will occur at six 402 months after the initial visit. We will only follow those families whose child screened positive on 403 the SDQ at the index visit, or for whom the SDQ was negative but the GP entered the child in 404 the TUCCP database. It will not be possible to guarantee that research assistants making 405 follow-up calls will be blind to the child's control-intervention status because assistants who 406 recruit families will, when possible, also make follow-up calls so that a known person is 407 maintaining contact. However, the assistants will have no other information about the family 408 other than what is reported on the study instruments. In particular, they will have no knowledge

409
of any data about the child from the TUCCP database, and they will administer the SDQ and 410 GHQ prior to instruments (at the baseline and final visit) that might reveal information about use 411 of any mental health services. We believe that the families' familiarity with the research 412 assistant, coming from an in-person contact at the time of consenting, will promote follow-up.

413
We believe that this advantage balances out concern about biased outcome assessment.

415
The literacy rate varies among the families to be recruited, and we anticipate that some parents 416 will require that the consent forms (and instruments if done in person) will be read to them. This 417 will take place in private, and we will have a sufficient number of female research assistants so 418 that there are no barriers to participation because of staff gender.

420
We will collect, and maintain separate from study data, basic contact information for child/youth 421 participants and their parent/guardian. Cell phone penetration is nearly 100% in the population 422 to be studied, but in the event that contact is not possible, research assistants will attempt 423 contact by mail and/or by leaving a message at the GPs office and the community mental health

428
From the TUCCP data system, we will be able to determine the number of mental health cases

433
Child/parent follow-up interview: A small sample (16-24) of those child/parent pairs being 434 followed in the intervention groups will be approached for qualitative interviews at three and six 435 months into their observation period (this possibility will be included in their initial consent). The 436 interviews will be scheduled at a time and place of the family's convenience; with parents' 437 consent, youth will be given the opportunity to be interviewed alone. The interviews will focus 438 on the experience of receiving care for emotional or behavioral problems from the GP.

439
However, the interview will not imply that the child/youth has such a problem. It will simply refer 440 to the issues the parent/youth have reported on the SDQ, which have been the subject of the 441 study follow-up. We will use purposeful sampling to select families for follow-up interviews, with 442 the sampling frame to be determined partly by child age, gender, and predominant symptoms 443 on the baseline SDQ. We may sample according to other characteristics that emerge as 444 important during the trial (for example, if there are many families that decline GP treatment).

445
These interviews will be conducted by a two-person team, one of whom is assigned to take 446 notes. They will not be recorded.

467
Training is planned for a two and a half day in-person session followed by an in-person booster 468 session a short while later. GPs will have continuous support from mental health center staff 469 who will provide them with feedback about registered cases and who are available for informal 470 telephone consultation. A "pocket guide" to key material in the training will also be developed 471 for use either on paper or on line. GPs will receive the equivalent of about $200 as 472 compensation for their time. As noted above, they will also receive an incentive payment from the collaborative care program (a standard part of the program, not part of the research) for successfully managed child cases (defined by the TUCCP).

475
Content areas for child/youth mental health training Topic Brief interventions Patient-provider interaction Engaging patients in conversation, attentive listening, eliciting an agenda, empathy, giving advice, decision support, interacting with child and parent ("Common factors") Children with problematic behavior Assessing possible causes (anxiety, depression, trauma, developmental delay or school problems, attention problems), general parenting advice, help with schoolwork, psychoeducation for families about children with delays; linking to community resources for children with developmental delays; appropriate medication use. Depression (child and parental) Psychoeducation, behavioral activation, exploration of contributing stresses, exploration of suicidal ideation, safety for suicidal patients, medications for older adolescents and parents. Anxiety Psychoeducation, active coping (role models, supported exposure, relaxation), support and response to maladaptive cognitions following trauma, exploration of contributing stresses, indications for medication.

Substance use
Recognition of harmful use of tobacco, alcohol, and inhalants. Brief counseling regarding cessation.

476
GPs assigned to the control arm of the study will receive a one-day child mental health update that will include role plays and practice recognizing child mental health problems. This will be a currently TUCCP policy, GPs will be instructed to refer all detected cases to the CMHC. At the 480 end of the trial, all control GPs will be offered the intervention training.

482
GP data collection. The GPs will complete brief demographic and attitude surveys at the outset of the 483 study and again at its conclusion. Both control and intervention GPs will complete very brief post-visit 484 forms for enrolled patients indicating their assessment of whether a child has a mental health problem 485 and, if so, their working diagnosis and treatment plan. They will also interact with two standardized 486 patients portraying child mental health problems -one at the end of their training and once during the 487 patient follow-up period. 488 able to build GP confidence in treating these problems and willingness to do so. These 497 interviews will be recorded and transcribed for analysis.

499
Inclusion/exclusion criteria: 500 1. Children and youth: age 5-15 years of age (male and female) and one of their parents or 501 guardians (whoever accompanies the child//youth to their GP visit). These children will 502 generally be coming to the GP for a non-emergent medical, emotional, or behavioral concern.
503 Some children may have a chronic condition, but we will not recruit any child/youth who reports 504 being in pain or who appears to be acutely ill. Children actively treated in the CMHC at the time 505 of recruiting will also not be eligible.

507
2. GPs: Must be a current participant in the TUCCP collaborative care program and see children 508 and youth as part of their practice. GPs who function solely as specialists will be excluded 509 (some may have had a more general practice at the time they joined TUCCP).

511
Risks 512 Potential risks are listed below, with methods to address the risk detailed in the next section. being. However, unwanted disclosure to third parties could be socially damaging or traumatic.

517
Other than the loss of study data, the main threats to privacy are unwanted disclosure during 518 attempts to follow-up on SDQ+ children/youth and unwanted disclosure of youth data to parents 519 and vice-versa. We try to minimize these risks by training of staff regarding confidentiality and security of data. In addition, no data collection other than the small number of follow-up interviews will involve information collected directly from children/youth (and even then they will not be the primary informants in the interviews), so instances in which youth may say something 523 to a research staff member in confidence will be very limited. It is possible that some youth 524 participants could disclose information to their GP that they might not want disclosed to their 525 parents. In Iran, it is more common than in the US for teens or even adults to come for visits 526 with a family member who remains present in the examination room during treatment.

527
However, as in the US, GPs can choose to ask family members to step out to afford the patient 528 privacy. In the other direction, it is possible that responding to the GHQ could prompt parents to 529 want to discuss issues with the GP that they would also prefer to discuss in confidence.

531
The SDQ (for parents about their children) does not ask about self-harm or victimization, and it 532 includes questions in which the parent is able to report positive attributes as well as problems.

538
Though we will not ask about these things, when interviewed alone by their GP or at the CMHC 539 youth may make statements raising concerns for maltreatment, self-harm, or a need for medical 540 services about which they do not wish their parents to be aware. Except if it is discovered in the 541 course of hospital treatment, Iranian law does not mandate child abuse reporting. Assent forms 542 for youth will include a reminder that they may always ask to speak in confidence with their GP 543 or another provider, though the decision to keep that confidence will rest with those providers.

545
2. Privacy for GPs: GP performance with standardized patients could subject their skills and 546 knowledge to a level of evaluation that they would not ordinarily experience in the course of their 547 regular practice (although these assessments are a regular part of TUCCP and GPs will have experienced them previously). All GP data will be treated confidentially.

550
3. Coercion to consent and involuntary participation for children/youth and parents:

551
Among the participants to be recruited, the children/youth by definition fall into the category of a 552 potentially vulnerable population. However, parents may also be vulnerable for reasons of 553 poverty, low literacy, or mental health problems of their own. We will make sure that the parents 554 and children clearly understand that whether or not they choose to participate in the study will 555 have no impact on the care they receive from the GP. We accomplish this by including clear 556 language in the consent and assent and by employing research staff that are not employees of 557 the practice.

559
In addition, this study uses a form of cluster-randomized design in which patients, who may 565 consent serves only as notification that they may be receiving a different treatment and allows them to agree to participate in an evaluation. As we describe below, trained GPs in this study will not be obliged to treat patients themselves (they can still make referrals to the CMHC) and patients will be able to access the CMHC or other mental health services regardless of the GPs decision.

Coercion to consent and involuntary participation by GPs: GPs might feel obligated to
572 participate in the new child mental health training in order to maintain their affiliation with 573 TUCCP. As noted below, we will make it clear that participation is voluntary. problems. In addition, as many as a third of children currently referred to CMHC's by TUCCP

580
GPs do not complete the referral. Still, there remains the risk, following GP training, that GPs 581 will attempt to treat children who should be referred, or will not realize that a child's condition 582 has not responded to treatment in primary care and now requires referral. We minimize this risk

603
The US PI will be in regular contact with the Iranian investigators and staff. Presently the US 604 and Iranian teams have joint calls every 1-2 weeks and frequent email exchanges during the 605 week. Once data collection is underway there will be one call a week for monitoring processes 606 and one every two weeks to discuss data quality. The US PI and co-investigator have made one 607 trip to Tehran for planning this phase of the study and plan subsequent trips to a) check on the 608 process of data collection during the baseline phase, and b) observe the GP training. These  2. Recruitment and informed consent: As noted above, children/youth will be recruited when 612 they come for GP visits with a parent or legal guardian. Parents/guardians will be asked for 613 written consent and children/youth for assent or co-consent as age-appropriate. Although the 614 age of majority in Iran is 18, Tehran University ethical guidelines call for co-consent for 615 children/youth ages 8 and older. If there is a concern about literacy, research assistants, will be 616 able to go over the consent verbally. The consent process will take place in a private area in 617 the clinic where the patient is being seen. Details about the consent process as it relates to 618 specific risks are discussed below.

620
GPs will be recruited by announcing, in the course of existing TUCCP activities, the availability 621 of new child mental health training. Opportunities for discussion about the training itself and its 622 evaluation will occur at both group and individual meetings. GPs will be individually and confidentially asked to provide consent. GPs choosing not to participate in the study will be able to continue in the TUCCP as at present but none of their patients will be recruited.
include making sure that no study data forms or files ever contain identifying information (except 628 those kept temporarily for follow-up tracking) and that linkage files are protected and destroyed 629 as soon as possible. If linkage files ever need to be transported they will move separately from data files. Identifiable data will never be stored on portable devices unless this is for the are capable of helping GPs review options for responding to concerns about maltreatment and other forms of family violence.
To manage risk of disclosure during follow-up, research assistants will use only means of research assistants will identify themselves only as calling on behalf of the GP until they are 674 certain that they are speaking to the enrolled parent and that he or she is able to speak in 675 confidence. Participants will have the right to leave the study at any point including declining to 676 participate in follow-up.

678
5. Privacy for GPs: All provider data will be treated as confidential and managed in the same 679 manner as patient data. In written accounts of the data for publication or internal reports care 680 will be taken to sufficiently disguise quotes or results that could because of "small cell" issues

686
6. Coercion and involuntary participation by children/youth and parents: Children/youth and 687 parents will not be able to chose whether or not their GP is participating in the study, but the 688 consent process will disclose that the GP is participating in a trial of training whose purpose is to

721
-Throughout the study both control and intervention GPs will continue to be members of 722 the TUCCP and have ready access to mental health consultation and referral for both 723 adults and children/youth. GPs will be free to refer any child for further evaluation or treatment at a CMHC regardless of study group assignment.
-In the Iranian system, while families can not directly access the CMHC without a GP there may be differences in cost depending on the family's insurance -referral to the elsewhere may be subject to fees of varying sizes. The consent process will remind parents and youth of these issues and that they may request a CMHC referral from their -As noted in the background section, the TUCCP involves regular monitoring of patient 733 status by CMHC personnel who review records and discuss care of all patients 734 registered with the system whether they are referred or cared for by the GP. This review is required before GPs can receive top-up payments for follow-up care. Thus, payment will be denied for follow-up care judged to be inappropriate.

737
-Receptionists, who provide case management, constitute a second set of "eyes" 738 interacting with families and assessing progress. Receptionists can trigger either return 739 visits to the GP or a request to the GP to initiate a referral.

740
-The study will have its own data safety management processes (see below). appointments. In addition, practices will receive top-up payments for patients they are able to 749 successfully follow and manage.

751
10: DSMP: As noted above, the TUCCP has safety and quality of care monitoring built into its 752 standard operating procedures; it also includes incentives for participating providers to 753 adequately follow-up treated patients. However, the study will have its own data monitoring 754 plan operating in parallel. The US and Iranian PIs will be primarily responsible for this second 755 plan. Incoming data will be cleaned and compiled on a regular basis and examined for unusual 756 or unexpected patterns. The PIs will be assisted in this task by a four-person DSMB that 757 includes both Iranian and US members. The present version of the protocol reflects feedback 758 from the DSMB. The DSMB will have internet meetings before study initiation, during the study 759 roll-out, and at its conclusion.

761
Site level research personnel will be thoroughly trained in good clinical practice, human subjects 762 procedures, and the importance of being alert for and documenting adverse events, protocol 763 deviations, and participant concerns. All of these will be immediately communicated to the in-764 country director and the PIs. Lists of possible expected and unexpected adverse events will be 765 developed during the initial stakeholder phase of the study and incorporated into training for 766 research assistants in Iran and the US.

768
Benefits 769 Patient participants may benefit by becoming more aware that their GP and/or her/his connections to mental health services are available for discussion and treatment of child and 771 adult mental health concerns. Clinician participants may receive feedback or instruction as part of the study that will be of interest to them and that could add to their skill level.
The general public may benefit if the study's interventions prove to be practical and effective, since 775 children's mental health services are currently difficult to obtain in Iran as in many other middle-income 776 countries. These benefits seem reasonable in comparison to the relatively small risk of involuntary 777 disclosure of information. 778 779

780
As noted above, GPs will receive about $200 for participation in the two-day child mental health 781 or control booster training and the trial. Families will receive a payment of about $2 to 782 compensate them for any extra time required at their index visit when they provide consent and 783 baseline data.

786
There are no direct costs associated with participation by patients. Patients asked to return to 787 the GP for mental health treatment might have to pay for the GP visit whereas their treatment, 788 had they been referred to the CMHC, would be provided without a charge. However, patients 789 are likely to live closer to the GP office than to the CHMC meaning that they would not have to 790 expend the cost of travel and the considerable time it could take to get to the CMHC, so costs 791 are likely to be equivalent.

793
There are no costs to GPs for participation in the child mental health training, but there could be 794 opportunity costs for participating if GPs were to be forgoing clinical income to attend. Incoming data from GP and patient instruments will be logged, examined for completeness, 938 nonsense, and ambiguous entries. Queries will be sent to the research assistants to examine 939 possibilities for re-contacting participants. Research assistants will use a messaging application 940 to report on their daily progress with recruiting or follow-up and to be able to clarify or report any 941 study issues immediately. Data collected on paper will be double entered. For each research 942 assistant, a 5% sample of the participant forms that they collect will be checked by having a 943 different member of the study team call the participant to verify key items. Supervisors of 944 research assistants will make random visits to GP offices while research assistants are present 945 to check on recruitment and data collection processes.

947
A trial data center in Tehran will be responsible for a) scoring baseline SDQ forms to determine Analysis will follow a pre-determined plan and take place under the direction of the trial 958 statistician. Analyses will be masked to study arm until data collection is finalized. Exploratory 959 analyses will confirm expected distributions and the prevalence and patterns of missing data.