Accuracy of a Prescription Claims Database: RESULTS

14 Nov
2010

prescription databases

Of the 367 patients to whom letters were mailed inviting participation in the study, full medication history interviews were completed for 194 patients (52.9%) who had active PharmaNet profiles. The demographic characteristics of the study subjects are presented in Table 1. One hundred and thirty of the patients interviewed (67.0%) brought all of their prescription medications to the interview. On the basis of PharmaNet records and the patient interview, there was at least one difference in the list of medications currently consumed for 138 patients (71.1%; Table 2). Concordance between PharmaNet and the interview was 0.81 (95% confidence interval [CI] 0.77 to 0.84), and the proportion of medications from the interview that were active according to PharmaNet was 0.83 (95% CI 0.80 to 0.86). One hundred patients (51.5%) claimed to be taking at least one medication that was not listed in PharmaNet or was inactive according to the PharmaNet profile; 31 patients (16.0%) indicated that they were not taking at least one medication that appeared active on the PharmaNet profile. A higher number of currently consumed, regularly administered prescription medications identified during the patient interview was associated with a higher risk of discrepancies (odds ratio 1.31 per medication on the list, 95% CI 1.13 to 1.53; p < 0.001). The use of memory aids (pill boxes) was associated with fewer discrepancies (odds ratio 0.40, 95% CI 0.16 to 1.00; p = 0.05).


Table 1. Characteristics of 194 Patients Interviewed about Prescription Drug Use


Median age, years (range)


65


(27-94)


Recruitment site


Heart Function Clinic


162


(83.5)


Pre Heart Transplant Clinic


19


(9.8)


Community pharmacy


13


(6.7)

Of the 1457 medications reviewed from these lists, 353 (24.2%) were involved in a discrepancy. Most commonly (205 medications for 81 patients), these discrepancies involved medications that were listed somewhere on the 14-month PharmaNet profile and were currently being consumed by the patient, but appeared inactive (i.e., overdue for a refill) based on the number of days supplied at the most recent refill. Of these 205 medications, 72 (35.1%) were overdue by 1 week or less, 69 (33.7%) by 8 to 30 days, and 64 (31.2%) by more than 30 days. Patients’ reasons for being late in obtaining refills are listed in Table 3. While many patients simply “hadn’t got to it yet” or had doses remaining from a previous early refill, many patients described dosing changes relayed verbally by physicians that were not reflected in the PharmaNet profile. A total of 31 medications had been discontinued by patients, most commonly because of adverse effects and often (16/31 or 52%) on the advice of the physician. Twenty-three medications that did not appear anywhere on the PharmaNet profiles were being used by the patients interviewed. These were typically samples issued by physicians. eriacta 100 mg

Figure1. Bland–Altman agreement

Figure 1. Bland-Altman agreement plot of the number of prescription medications currently consumed, assessed by examination of PharmaNet data and a patient interview. The size of each circle is proportional to the number of data points, and the line at 0.00 represents no difference.

Table 2. Discrepancies between Medication Lists Obtained from PharmaNet and Information from Patient Interviews*


Any discrepancy


138


(71.1)


353


(24.2)


Different medications listedt


115


(59.3)


268


(18.4)


Patient taking medication overdue for refill


(according to PharmaNet)


81


(41.8)


205


(14.1)


< 7 days
overdue


37


(19.1)


72


(4.9)


8-30 days
overdue


34


(17.5)


69


(4.7)


> 30 days
overdue


42


(21.6)


64


(4.4)


Patient taking medication not recorded


in PharmaNet


19


(9.8)


23


(1.6)


Patient not taking medication recorded


as active in PharmaNet


15


(7.7)


31


(2.1)


Medication used “as needed” despite regular


schedule of use recorded in PharmaNet


9


(4.6)


9


(0.6)


Different doses listedt


64


(33.0)


85


(5.8)

There were 85 discrepancies in the doses of drugs that were listed on both PharmaNet and the interview list. According to the patients involved, the change in dosewas most commonly the result of adverse effects or lack of efficacy, or to simplify the drug regimen (Table 4). Adrenergic £-receptor blocking agents (^-blockers) and furosemide were the medications most commonly involved in identified discrepancies (Table 5). Viagra Professional

Table 3. Reasons Provided by Patients for Discrepancies between Medication Lists Obtained from PharmaNet and Patient Interviews (n = 268 Medications with Discrepancies)


Overdue for refill (according to


PharmaNet)


205


(76.5)


Patient “hadn’t got around to”


refilling prescription


40


(14.9)


Dose or frequency changed recently


38


(14.2)


Doses left from previous refill


30


(11.2)


Supply provided in hospital


20


(7.5)


Recently restarted a previous supply


18


(6.7)


Self-reported noncompliance


14


(5.2)


Borrowed from another person


5


(1.9)


Other reason


4


(1.5)


Unknown


36


(13.4)


Medication used “as needed” despite


regular schedule of use recorded


in PharmaNet


9


(3.4)


Medication appears active in PharmaNet


but not being taken by patient


31


(11.6)


Adverse effects


7


(2.6)


Changed by physician before patient


finished current supply


4


(1.5)


Symptoms resolved


2


(0.7)


Not working


1


(0.4)


Wrong medication supplied


1


(0.4)


Unknown


16


(5.6)


Patient taking medication not


recorded in PharmaNet


23


(8.6)


Using sample medication


16


(5.9)


Filled outside British Columbia


3


(1.1)


Using spouse’s supply


1


(0.4)


Not filled in
> 14
months


1


(0.4)


Using supply from hospital


1


(0.4)


Unknown


1


(0.4)

The total number of regularly administered prescription medications differed between the medication lists obtained from patients and those listed in PharmaNet for 109 (56.2%) of the patients. On average, PharmaNet list­ed one fewer medication than the list obtained during the interview (Figure 1). For 6 patients, an equal number of medications were listed on PharmaNet and by the patient, but the medications on the lists differed to some degree.
canadian cialis online

Table 4. Reasons Provided by Patients for Dosing Discrepancies between Medication Lists Obtained from PharmaNet and Patient Interviews (n = 85 Medications with Discrepancies)


Adverse effects


11


(13)


Lack of efficacy


10


(12)


Previous regimen too complicated


7


(8)


Condition improved


7


(8)


Dose varies with weight or renal function


6


(7)


Verbal instructions differ from prescription


6


(7)


To adjust for changes in other medications


2


(2)


Condition worsened


1


(1)


Mistake entering dose in PharmaNet


1


(1)


Cost


1


(1)


Unknown


33


(39)

Table 5. Medication Classes Involved in Discrepancies (n = 353 Medications with Discrepancies)


Drug Class


No. (%)
of


Discrepancies (n
= 353)


B-Blockers


38


(10.8)


Furosemide


38


(10.8)


Spironolactone


29


(8.2)


ACE inhibitors


28


(7.9)


Vitamins and supplements


24


(6.8)


Oral hypoglycemics


22


(6.2)


Lipid-lowering agents


20


(5.7)


Digoxin


17


(4.8)


Angiotensin receptor blockers


16


(4.5)


Antiarrhythmics


10


(2.8)


Other diuretics


10


(2.8)


Nitrates


9


(2.5)


Antiplatelet agents and anticoagulants


9


(2.5)


Calcium-channel blockers


6


(1.7)


Other


77


(21.8)

top