Telephone Counseling project/ Parent's Wisdom Project

• Randomized, controlled trial. Intervention is counseling by telephone. The control group didn't really get anything.
• There are three data collections: baseline and fu1 and fu 2.
• A number of instruments looking at psychosocial wellbeing were implemented to the mothers/parents. These instrument measures are the outcomes.
• Three sites.
• Total n is 47. 39 completed.
• There is only one respondent per family. It's usually the mother.
• There were several missing questionnaires.
• All patients are one year away from completion of therapy.

Data

• All data are in redcap.
• Kristen will send me all of the scoring algorithms. (She is still looking for a couple of them.)
• Kristen will put new variable names on much of the data.
• Most of the data I need is in the patient questionnaire. Also the demographics are in the consenting part.

Outcomes

The outcomes are measures of psychosocial wellbeing

• alttiq
  • alttiq_18_1 -- alttiq_18_4
  • sum them
• cahs
  • Each item has 5 answer choices which are a Likert scale from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicate more agreement with the appraisal item or scale.
  • Two journal articles are saved with the other scoring algorithms. They don't exactly give instructions for scoring. Based on what I read, I think the questions 1-5 are the threat subscale, 6-10 are the harmless subscale, and 11-13 are the challenge subscale. (Kathy is somewhat familiar.)
  • For now, just sum up all the items.
  • I don't know if there is an overall score. Is there an overall score that measures cognitive appraisal?
  • I don't know if there is a provision for missing items.
  • The items from the original 32-item scale that were used in the 13-item scale are: 1, 3, 4, 5, 8,11, 13, 17, 18, 19, 21, 23, 26. I matched these to our items 1-13 which follow a different order. Muayaad says that his items 23 and 5 need to be reversed, and these correspond to items 5 and 2 of our numbering. This is a little strange to me, since the last three items, 11, 12, and 13, corresponding to 11, 26, and 1, are really going the other direction.
  • Ranges: Total: 13-65; Threat: 5-25; Harmless: 5-25; Challenge: 3-15
• CSE (Coping self-efficacy scale) (not CESD)
  • ces_23_1 -- ces_23_26
  • sum all the items
• Communicating with medical team
  • not validated.
  • Questions cmt_19_1 -- cmt_20_10.
  • Make up scoring algorithm! I just summed these.
• ies (should be same one that Charles is doing.)
  • ies_21_1 -- ies_21_15
  • The numeric scores for the answer choices are 0, 1, 3, 5, corresponding to not at all, rarely, sometimes, often.
  • There are three scores for the ies: intrusion, avoidance, and total/overall.
  • intrusion involves items a, d, e, f, j, k, n. Take sum.
  • avoidance involves terms b, c, g ,h, i, l, m, o. Take sum.
  • Total score is the sum of the two subscores.
• POMS
  • poms_22_1 -- poms_22_37
• pat: patient assessment tool
  • This is really cumbersomely laid out. It looks kind of non-uniformly scored. Some of the demographic info is asked a second time on the pat.
  • Has several subscales and a total scale as well as a categorized risk.
  • Subscales are Family structure and resources, Social support, Child problems, Sibling problems, Caregiver problems, Caregiver stress reactions, and Family beliefs.
• Post traumattic growth inventory (ptgi)
  • Appears that an increase is the direction for improvement.
  • ptgi_27_1 --- ptgi_27_21
• Program evaluation
  • eval_28 -- eval_32, eval_33_a --eval_33_h, eval_34, eval_35, eval_36_a -- eval_36_o, eval_37 -- eval_40
• stress thermometer
  • stress_thermo_24, stress_thermo_25
• Travel items
  • pat_16
  • pat_17

Analysis tasks

• score instruments. poms and pat are remaining.
• compute vars: time since completion of therapy?
• Compare people who didn't fill out any fu by group. Get demographics of nonresponders.
• Preliminary analysis for grant submission: to compare groups wrt outcomes at all time points, also assess group comparability.
  • unadjusted comparisons
  • Table 1: type of dx, age at dx, parent's education levels, race (asked Kristen to ask Deb how to group the dx types)
  • Summarize travel distance. q17 and 18.
  • Program evaluation: for intervention group only, they were asked for feedback on the program. Need descriptions of these items.

Results and interpretation

Added linear regression models with group and baseline measure for all continuously measured outcomes. Significant effect fore ptgi factor 2.

Meeting notes

2015 May 19

• I showed them the mins and maxes of the PAT scores by group. Anne Kazak is the one who asked for it, and she wasn't on the call.
• Someone will send some cut offs for ies and .
• There are only 4 people above the cutoffs, so we wouldn't want to do that. They want to slice the scores more and "look for a signal."

2014 October 3 Conference call

• Is there a "third variable" that may impact how the treatment affects patients? Are patients who are closer to the end of treatment more likely to benefit from the treatment?
• Anne said there is an algorithm for scoring pat with missing items. later said they don't.
• We noticed that there is not much going on, but it could be because the patients, including the control group, were not high risk to begin with.

2014 July 30

• Kristen reports remembering that when the patients did return surveys, they tended to fill out most all of the questions.
• Purpose of the travel questions is to establish feasibility for telephone counseling. In text, give % of patients with 60 or more minutes of travel time.
• IES, POMS, and CAHS are the most important outcomes.
• To do
  • Add group comparisons that take the baseline scores into account.
  • Add comparisons of the five who declined by those who did not
  • Report the communication with medical team questions separately._done_
  • Is there a variable for site? Do we want any of these results by site? Participation rates.
  • Get % of the following response patterns with 47 (all) as the denominator: BL, BL + FU1, BL + FU2, BL + FU1 + FU2 (Kristen also asked to do this)
  • Add direction of improvement for each of the outcomes
  • Get lengths of time between bl and fu1, bl and fu2, and fu1 and fu2.

2014 July 24

• Want to sent investigators results by Tuesday morning!!!
• Use an 80% rule to calculate scores.
• The CES is really CSE, coping self- efficacy scale.
• For CAHS, for now, just sum all items.
• Discussed how to present group comparisons of outcomes. Will need one table for baseline comparison. Also test group differences at f/u 1 and at f/u 2. Will check with TK whether we should use paired t-tests or just t tests. Also do test on fu 1 vs. fu 2. That might have to be a paired t-test. Or a regression.
• Kristin thinks that the number for having f/u 1 should be 32, not 33. She will double check.
• Combine the Solid tumor, Sarcoma, and CNS/ brain tumors into one group for table.
• There are three patients with time from end of tx to enrollment. We are first going to double check the dates and then see if they whether they even completed a f/u.
• Get Kristin the ids of the people with age at dx = 0 and 0.06, all lowest ages , and time SinceTxEnd > 12 months.
• Restrict the main analysis tables, including bl comparisons, to people who had bl. (should be n = 44)
• Make consort diagram:

2014 June 19

• Will have an investigator's meeting at end of July
• Kristin sat down with me, and we looked at the databases. Turns out that having everything in one redcap db was fine.
• They want preliminary analyses to use for grant submission
• Goals are: to compare groups wrt outcomes at all time points, also assess group comparability.
• Plus there will be at least one manuscript
• Sounds like main analyses for this will be unadjusted group comparisons.
• The things we want on Table 1: type of dx, age at dx, parent's education levels, race
• All patients are one year away from completion of therapy.
• Kristen will send me all of the scoring algorithms. (She is still looking for a couple of them.)
• Plus there were some additional questions that they have questions about, like travel distance. q17 and 18.
• Program evaluation: for intervention group only, they were asked for feedback on the program. Need descriptions of these items.
• Additional "communicating with medical team" piece that is not validated. Questions 19 and 20. Make up scoring algorithm!
• Kristen will put new variable names on much of the data.
• Most of the data I need is in the patient questionnaire. Also the demographics are in the consenting part.
• The pat is an instrument that is really cumbersomely laid out. It looks kind of non-uniformly scored. Some of the demographic info is asked a second time on the pat.

2014 January 15 Meeting with Kristin

• She wants us to look at the db she has made in redcap
• study is pilot to look at telephone counseling vs. standard care on survivors of childhood cancer.
• enrolled about 40 patients.
• Sources of data: questionnaire (3 times) data collection form, withdrawn caregivers form
• Questionnaire is administered at three time points. The two follow up questionnaires are slightly different. She has set it up as a longitudinal database.
• Recommendations:
  • For any dates, use a text box with a date validation.
  • Try to avoid checkboxes unless you really know that you want the person to select more than one of the choices.
  • We would much rather have the actual birth date than what ever the person filled out as the child's age currently.
  • The child's age now field should be a text box that is set to only accept numeric values. Also give yourself a note on the form to input the value in years (or any units, as long as all of them are in the same units).
  • Check with redcap personnel about whether you can omit the participants id, participant's initials, and any other fields that would be duplicated across forms (longitudinal or not). The simpler you can make it, the better.
  • Can use branching logic to make the question about how the things affect siblings.
  • Check if you can add branching logic for an entire matrix of questions at one time.
  • The way you can "test" your database is to go on the left menu under "data collection" and choose "add/edit records"
  • In the "longitudinal" set up, there may be a problem. We noticed this when looking at the event grid, that the event times should be across the top of the table, right? .
  • For every question, it would prevent entry errors if you add appropriate validation
  • Ask the redcap people if it is possible to cause a dialogue box to pop up if there are fields that have been left blank (even if they are not "required" fields). For example, a box that says "Questions 2, 24, and 54 have not been entered. Do you want to go back or save?"
  • In general, it's better for the statistician/analyst/programmer to calculate fields than to have it calculated within redcap or having either the research subject or the person entering the data to do it.

-- JoAnnAlvarez - 01 Jul 2014